PROCEEDINGS OF THE INTERIM MEETING OF THE ICOM-CC METALS WORKING GROUP

EDITED BY CLAUDIA CHEMELLO, LAURA BRAMBILLA, AND EDITH JOSEPH

PROCEEDINGS OF THE INTERIM MEETING OF THE ICOM-CC METALS WORKING GROUP

SEPTEMBER 2–6, 2019

NEUCHÂTEL, SWITZERLAND

EDITED BY CLAUDIA CHEMELLO, LAURA BRAMBILLA, AND EDITH JOSEPH

 

© 2019 International Council of Museums – Committee for Conservation (ICOM-CC)

The Metals Working Group is part of the Committee for Conservation (ICOM-CC), a committee of the International
Council of Museums (ICOM) network.

HE-Arc CR has been granted by ICOM/ICOM-CC a non-exclusive license for the Publication.

Publishers: International Council of Museums – Committee for Conservation (ICOM-CC) and Haute Ecole Arc
Conservation-restauration (HE-Arc CR)

All rights reserved. No part of this book may be reproduced in any form or by any means, whether electronic or
mechanical, including photocopying, recording, or otherwise, without the prior permission in writing from the
copyright holder.

ISBN 978-92-9012-458-0

Proceedings and user interface design and production: Eduardo Pulido (epulido@sapo.pt)

Copy-editing: Carla Nunes (carlarnunes@gmail.com) and Per Christopher Foster (chris.foster@netcabo.pt)

Paper-flow contribution platform production and management: use.it® – Virott e Associados, Lda. (www.useit.pt)

Cover image: Detail of the clock from Château de Morges, Henri Golay, 1773. Vaud Military Museum, Château de
Morges. © Nicolas Moret – HE-Arc CR
Table of contents
Foreword. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Preface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii

TECHNICAL STUDIES
A Diagnostic Study of the Cast-Bronze Great Buddha Statue in the Kotoku-in Temple, Japan
A. Fujisawa, N. Hemuki  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Technical Study and Conservation of a Large-Scale Metal Buddha Statue
K.A.A. Kasthuri  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Training a Database and Convolutional Neural Network for Automated Recognition of Hallmarks on (Dutch) Silver
E. van Bork, R. Erdmann, D.J. Biemond  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Metal Threads in 17th-Century Western European Textiles: Characterization of Metal Threads from the Burgzand
North 17 Shipwreck (YCP)
S. Savelli, E. van Bork, S. Meijer, I. Joosten  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
The Production of Medieval Gilt Silver Threads (YCP)
M.F. Gómez Núñez, I.K. Jiménez Cosme, J.L. Ruvalcaba Sil  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Iron Piano Wire Production at the Beginning of the 19th Century
M.G. Durier, L. Villiaume, R. Pires, Y. Cotrebil, E. Vega, A. Houssay, J.-C. Battault, M. Vion, C. Clarke,
I. Guillot, P. Dillmann, S. Vaiedelich  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Transparent Varnishes on Copper Alloys Dating from the 19th Century: Characterisation and Identification
Strategies
J. Schröter, A. Michel, C. Paris, S. Mirabaud, L. Bellot-Gurlet, L. Brambilla  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

ANALYTICAL INVESTIGATION
Study of Ancient Metal Artefacts Using Neutron Imaging Techniques, Visual Examination and Chemical Analysis
A. Duvauchelle, M. Krieg, D. Mannes, E.H. Lehmann, V. Hubert, L. Andrey, S. Gillioz  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Evidencing Risk to Inform Best Practice: The Limitations of Detecting Tinning Layers by Visual Analysis
During the Removal of Corrosion from Archaeological Iron
D. Watkinson, M. Crepeau, N.J. Emmerson  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
An Overview of the Use of EIS Measurements for the Assessment of Patinas and Coatings in the Conservation of
Metallic Cultural Heritage
B. Ramírez Barat, P. Letardi, E. Cano  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
A Multi-Analytical Approach to Evaluate Surface Treatments on Copper-Alloy Artefacts: A Case Study Applied to
the Restoration of the Memorial of Redipuglia
T. Bruni, B. Mariani, B. Salvadori, P. Letardi  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
A Non-Destructive Toolkit for Silver Plate Identification
R. Grayburn, A. Wagner  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Conservation and Examination of an Iconic 17th-Century Pewter Object: The Hartog Plate’s History Revealed
through Synchrotron Analysis
T. Davidowitz, D. Thurrowgood, R.G. Erdmann, I.D. MacLeod, D. Creagh, D.L. Howard, D. Hallam  . . . . . . . . . . . . . . . 108

PREVENTIVE CONSERVATION
ER-Based Sensors for Monitoring Atmospheric Corrosivity on Corroded or Patinated Copper-Based Objects
C. Petiti, M. Faifer, M. Zanoni, S. Goidanich  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
iv  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND

Metal Conservation, Cellulose Nitrate and the Oddy Test

G. Eggert, R. Kuiter, J. Ziegler, S. Bette, J. Stelzner  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Propellants and Primers: A Striking History with a Possibly Explosive End
F. Davidson  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
El Anatsui: Conservation Concerns Beyond Installation
J. Bloser, L. Zycherman, A. Haddad, A. Martins, C. McGlinchey  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

CORROSION STUDIES
Study on the Corrosion Mechanisms and Morphology of Archaeological Bronze Objects from a Bronze Age
Graveyard in Southwestern Iran
O. Oudbashi, R. Naseri, B. Heidarpour, A. Ahmadi  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
When Glass and Metal Corrode Together, VII: Zinc Formates and Further Unknown Zinc Compounds
A. Fischer, G. Eggert, J. Stelzner, S. Bette, R.E. Dinnebier  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
A Dark Discoloration on Aristide Maillol’s Outdoor Lead Sculptures: Identification, Formation, and Further Research
M. Toro, T. Beentjes, I. Joosten, J. Bloser, L. Zycherman  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Pilot Study Examining Tarnishing Caused by Nitrile Glove–Solvent Interaction on Metals
C. Southwick, K. van Lookeren Campagne, E. van Bork, J. van Bennekom, T. Davidowitz  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
Effect of Reburial Conditions on the Corrosion of Marine Iron Artifacts
A. Yanagida, Y. Ikeda, K. Matsuda, S. Wakiya, Y. Kohdzuma  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Bluing Steel: An Investigation into the Thermal and Chemical Methods of Colouring Mild Steel (YCP)
Júlia Tauber  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Behaviour of Lead Coated with Benzotriazole Solution in an Organic Acid Environment (YCP)
A. Nagar, M.R. Singh  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198

PROTECTION: CORROSION INHIBITORS

From Mono-oxime to Tri-oxime: The Development of a New Group of Corrosion Inhibitors for Copper Alloys
A.N. Abu-Baker  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Assessing the Interaction between Corrosion Inhibitors and the Historical Corrosion Layer on Copper Claddings
Based on Global and Micrometric-Scale Analysis
E. Ferrari, E. Apchain, D. Neff, M. Bayle, P. de Viviés, J.-B. Memet, P. Dillmann  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212

PROTECTION: COATINGS
Protection of Outdoor Bronzes with Eco-Friendly and Non-Hazardous Coatings Based on Silane and
Fluoropolymers: Results from the B-IMPACT Project
M. Aufray, J. Esvan, A. Balbo, V. Grassi, C. Monticelli, F. Zanotto, F. Benetti, W. Sperotto, E. Tedesco,
E. Bernardi, M.C. Bignozzi, G. Masi, C. Chiavari, N. Gartner, T. Kosec, L. Škrlep, E. Švara Fabjan, C. Josse,
C. Martini, L. Robbiola  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
The Use and Mis-use of Renaissance Wax
D. Thickett, B. Stanley  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232

DEVELOPMENT OF CONSERVATION PRACTICES

The Potential of Microorganisms for the Conservation-Restoration of Iron Artworks
L. Comensoli, W. Kooli, M. Monachon, M. Albini, P. Junier, E. Joseph  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Desalination of Archaeological Iron Objects: Comparing the Effectiveness of Sodium Hydroxide Treatments
B. Schmutzler, Z. Revay, C. Stieghorst  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Stabilisation of Archaeological Copper-Alloy Objects from Chloride-Induced Active Corrosion with Beauveria
bassiana (YCP)
N. Gutknecht, E. Joseph  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
 ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  v

Design of a Suitable Cleaning Procedure for the Tarnished Silver Elements of the Reliquary Bust of Santa
Vittoria, Diocesan Museum of Agrigento, Sicily (YCP)
N. Ricotta, C. Degrigny, A. Cagnini, C. Ortolani  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Equipment Identification and the Development of Dry-Ice Blasting Parameters for Cleaning Archaeological
Wrought Iron, Copper Alloys, and Gray Cast Iron
W. Hoffman, L. King  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Does Dry-Ice Blasting Allow Safe and Effective Cleaning of Either Coated or Corroded Bronze Surfaces?
S. Agnoletti, S. Basile, A. Brini, A. Cagnini, S. Porcinai, A. Santagostino Barbone, M. Galeotti  .. . . . . . . . . . . . . . . . . . . . . . . . 273
Flame Cleaning of Historic Wrought Iron: Practitioner Methods and Their Impact on Oxide Morphologies and
Post-Treatment Corrosion Rates
N.J. Emmerson, D.E. Watkinson, J.C. Thunberg  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Burnishable and Conductive: A New Filling Possibility for Losses in Gilded or Metallised Surfaces (YCP)
C. Huynh, D. Robcis, F. Urban  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Gap Fillers for Brittle Metal Artifacts and Their Suitability in High Temperature and High Relative Humidity
Conditions (YCP)
A. Elmarazky, A.B. Paterakis  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Conservation of Saint George Killing the Dragon, a 16th-Century Gilt Copper Statue from Gdansk:
A Collaborative Approach
A. Baranowska-Fietkiewicz, C.V. Wesolowska, K. Schaefer, B. Dajnowski  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297

CONSERVATION TREATMENTS
Redressing the Past: The Examination and Treatment of an Ancient South Arabian Censer
J.-F. de Lapérouse, F. Carò, A. Rizzo, D. Jovanovic  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 306
Conservation and Investigation of Cometh the Sun: A Monumental Weathering Steel Sculpture
C. Chemello, E. Cano, A. Crespo, P. Mardikian, C. Patterson  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
The Metal Patina and Surface Layer of El Caballito: Calling Things by Their Name
J. Contreras-Vargas, J. Cama-Villafranca  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322
Study and Conservation of Lead Curse Tablets
L. Rossetti, J.G. Aubert, N. Stephant, N. Mélard, E. Guilminot  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
Revealing the H.L. Hunley: The Strategic Planning and Deconcretion Process of an American Civil War Submarine
J. Rivera, P. Mardikian, M. Scafuri, L. Näsänen, N. González-Pereyra  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338

COMPOSITE OBJECTS
Challenges and Solutions for Conserving Non-Separable Marine Archaeological Composite Objects
P. Mardikian  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
Conservation Treatment and Stabilization of Marine Composite Medieval Swords from Rhodes, Greece, Using
Ammonium Hydroxide Solutions
V. Argyropoulos, A.-A. Papanikou, P. Mardikian  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
Stabilization of Archaeological Composites: A Complex Challenge Requiring Compromise (YCP)
T. Wiinblad  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Glass-Induced Metal Corrosion: Study and Conservation of an Enamelled Altarpiece (1954–56) from the
Collection of the University of Applied Arts Vienna (YCP)
M. Holzleitner, M. Hietz, E. Lenhart, M. Anghelone, G. Krist  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
New Strategies for the Conservation of Paintings on Metal
J. Gordon, L. Normand, A. Genachte-Le Bail, M.-A. Loeper-Attia, R. Catillon, A.-L. Carré, M. Saheb,
A.-M. Geffroy, C. Paris, L. Bellot-Gurlet, S. Reguer  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Preserving Fort Erie’s CN 6218. A Case Study
S.L. Maltby, J. Davies  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
vi  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND

Metal Textile Composites and Improved Treatment Outcomes through Application of In-Situ Corrosion Data
I.D. MacLeod, R. Car  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
Cloth of Metal (Indian Shamiana): Scientific Characterisation of Metal Thread Corrosion
W.A. Mohamed, M. Rifai, F.E.-Z. Sadat  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Physical and Chemical Interactions between Metallic Instruments and Plastic Threads
M.-A. Loeper-Attia  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399

MODERN METALS
Aluminium Alloys in Swiss Public Collections: Identification and Development of Diagnostic Tools to Assess
Their Condition
C. Degrigny, J. Schröter  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
Aluminium Surfaces in Heritage Collections: Evaluating Surface Appearance Modifications
C. Andraud, A. Azéma, P. Betinelli, C. Boust, M. Brunet, M. Florescu, A. Genachte-Le Bail, N. Melard,
A. Michelin, M. Page, G. Seine, A. Texier  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
Preserving Aviation Heritage: The Need for an Approach Based on Materials Analysis
M. Brunet, L. Robbiola, P. Sciau, M. Drieux-Daguerre  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 425

POSTER SESSION
Conserving Conservation: The Critical Application of Traditional Metalworking Skills and the Importance of
Knowledge Transfer to New Generations of Conservators
J. Saunders, H. Jacob, P. Willett †  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
Two Iron Age Barrel Bracelets – Investigation, Technology and Conservation
I.M. Huller, K. Schmidt, M. Anghelone, G. Krist  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
Gilding Iron in the 16th Century
D. Robcis, E. Plé  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
Two Bronze Statuettes from the Estense Gallery: Materials, Casting Technology and Conservation Intervention
C. Chiavari, C. Martini, L. Morigi  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Bells in Use: From Living Heritage to Legend
D. Lira, G. Peñuelas  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
Preliminary Results of Research on 17th–19th-Century Coffin Portraits from the Collection of the National
Museum in Krakow
M. Labut, M. Goryl, J. del Hoyo-Melendez, K. Stefanczyk, A. Stepien  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
All That Glitters: Treatment and Technical Study of an Ormolu Surtout de Table
S. Barack, J. Walthew, J. Godfrey  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
Marine Precious Metals After Excavation
J. van der Stok, M. van Bommel, I. Joosten, T. Beentjes  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
Developing a Sustainable Approach for Screening and Studying Functional Metal Objects
S. Russo, E. Balliana, F. Thaler, A. Cipolato, E. Zendri  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
Microstructural Examination of Kaushambi Copper-Alloy Coins
P. Verma, D. Rajpoot, G. Lal, B.V. Kharbade  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
Examining Metal Corrosion Using Electrochemistry
H.M.A. Hunter, D. Thickett, J. Najorka, C.G. Jones  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
A Project with Potential: Evaluating a New Monitoring System for Artifacts Undergoing Electrochemical
Treatment
W. Hoffman, M.K. McGath  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
Diagnostics and Monitoring of Historical Vehicle Engines by Acoustic Emission Testing
L. Brambilla, A. Roda-Buch, E. Cornet, G. Rapp, B. Chalançon  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 446
 ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  vii

Use of the MiCorr Application for the Local Diagnosis of Metal Artefacts: The Conservator’s Point of View
M. Hovind, C. Degrigny  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447
Classification of Indoor Air Quality with Respect to Lead Corrosion
M. Kouril, K.C. Strachotova, S. Msallamova, Z. Bartak, K. Kreislova  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
The Patina on Amilcar de Castro’s Weathering Steel Sculpture: The Influence of the Museological Environment
T. Russo dos Reis, T.C. Carneiro, M. de A. Rizzutto, A.A. Couto, P.H.O.V. Campos  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
Display Cases for Metal and Composite Objects at the National Museum in Warsaw, Poland
A. Mistewicz  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
Study of a Preventive Conservation Procedure for Paintings on Iron Plates
E. Menart, N. Nemeček, I.K. Cigić, H.P. Krajnc  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Appearance and Corrosion of Weathering Steel at an Urban-Coastal Site: Influence of Environmental Conditions
during the First Months of Exposure
E. Bernardi, L. Nobili, I. Vassura, C. Chiavari, C. Martini, D. de la Fuente, M. Morcillo  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
Atmospheric Alteration of α/δ Bronze Bells: Micro-Infiltrating Corrosion
A. Petitmangin, A. Chabas, M. Saheb, I. Guillot, B. Laik  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Whiskers and Strands: Unexpected Corrosion on Early Medieval Metals on Display at the Museum of London
H. Ganiaris, R. Lang, A.P. Cure  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
Corroding Lead Cames on Stained Glass Windows: A Case Study of Two 16th-Century Panels in Their Display Case
A. Gerber  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
Study of Dark-Red Staining on Lead Rooftops and Sculptures
D. Lopez, A. Azema, S. Joiret, S. Reguer, A. Texier  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
The Use of Industrial Vapor-Phase Corrosion Inhibitors on Large and Composite Objects
J. Alexander  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 457
Application of Self-Assembled Monolayers of Long-Chain Organic Acids in the Corrosion Protection of Bronze
D. Mikic, H.O. Curkovic  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Assessing the Performance of Protective Treatments on Outdoor Sculptures: To Be or Not To Be a
Representative Model Sample?
E. Joseph, P. Letardi, M. Albini, M. Monachon  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 459
Protection and Meaning: A Reflection on the Role of Painted Surfaces on Outdoor Metal Sculptures and Their
Conservation
A. Camoli, A. Carrieri, A. Ervas  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
The Mystery of Ancient Metal Amulets
H. Kupershmidt  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
Non-Destructive Investigation of the Conservation and Restoration of Ductile Metals in Cultural Heritage Artifacts
S. Borrós, P. Giménez-Barrera  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
Freshly Excavated Coins: Best Practice for Rapidly Obtaining Numismatic Information
S. Brechbühl  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
Comparing Different Deconcretion Strategies for the Ballast Tanks of the H.L. Hunley Submarine
A. Funke, J. Rivera-Diaz, G. Porteous  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
Custom-Made Desalination Structures for Iron and Copper-Alloy Composite Objects
A. Dittus, M. Schüch  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465
Use of a Heated Rinse to Facilitate the Thorough Removal of Sodium Sesquicarbonate from USS Monitor’s
Copper Alloys After Desalination
E. Sangouard, R. Spohn  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
viii  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND

Passive Removal of Silver Sulfide Layers from Metal Threads on Textiles by Storage in an Intercept Bag
I.D. MacLeod, R. Car  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 467
Conservation Strategies for Cans in Collections
A. Michel, R. Bertholon, A. Vecten, L. Brambilla  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 468
Conservation of Damaged Architectural Aluminum Elements
M. Raedel, M. Buecker, I. Feldmann, M. Reimann  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 469
The INTERFACE Project: Conservation of Painted Metal Artefacts
M. Albini, G.M. Ingo, C. Riccucci, M.P. Staccioli, C. Giuliani, G. Di Carlo, E. Messina, M. Pascucci  . . . . . . . . . . . . . . . . . 470
Conserving Clocks: Thoughts and Challenges
R. Hatahet  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
Exploring Suitable Parameters for Electrochemical Cleaning/Consolidation of Museum Objects and Practical Use
of the Pleco Electrolytic Pencil
N. Nemeček, E. Menart, D. Lavriha, M. Jeran, T. Lazar, C. Degrigny  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472
Use of Gels for the Treatment of Metals
E. Guilminot, A. Gomez, A. Raimon, M. Leroux  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Finding the Way Forward with Atomic Layer Deposited Coatings on Silver Heritage Objects: Moving from the
Laboratory to Conservation Practice
E. Breitung, S. Creange, T. Davidowitz  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
Conservation and Stabilization of Iron Objects from the Archaeological Site of La Tène (Neuchâtel) in View of
New Investigations and a New Exhibition in the Laténium
E.D. Beuret, C. Cevey, G.R. Servais  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
Alkaline Sulfite Desalination of Formerly Restored Damascened Merovingian Belt Buckles: A Possible Challenge
F. Lagger  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476
Conservation Treatment of the Six-Inch, Thirty-Caliber Mark III Gun and Carriage on the USS Maine (ACR-1)
J. Schwebler  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
Backyard Treasures: A Case Study from Kulautuva Cemetery, Lithuania
A. Vasiliauskaite, M. Daubaras  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
Conservation of a Restored Late Iron Age Strainer Bowl
S. Naidorf, M. van Bellegem, D. Hook  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
Restoration of a Decorative Baroque Door Lock from ca. 1720–40
J. Růžička  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 480
Conservation of 8th–9th-Century Curonian Women’s Jewellery
S. Garšvienė, V. Lukšėnienė  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
Just When You Thought You Knew How to Polish Silver: An Analysis of Silver Polish Applicators
T. Gessler, K. McEnroe  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482
Restoration and Conservation in Unfavorable Working Conditions: A Case Study
Z. Kirchhoffer, S. Kirchhoffer  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Treatment of a Composite Sock-Knitting Machine Using Dry-Ice Cleaning
P. Mardikian, C. Chemello, M. Bynon  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484
Spectacular Spectacles: Corrosion Cases
J. Huber, A. Fischer  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
Author index  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486
Keyword index  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 489
 ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  ix

Foreword
We are very happy to welcome colleagues from all over worked tirelessly from the inception of this project until
the world who share a great interest in the conservation, its conclusion to make this conference a huge success.
display, and heritage use of metal objects to Metal 2019, Published under the auspices of ICOM-CC, the Metal
the 9th Interim Meeting of the International Council of 2019 proceedings were coordinated by Claudia Chemello,
Museums–Committee for Conservation (ICOM-CC) coordinator of the ICOM-CC Metals Working Group,
Metals Working Group. for whose huge hard work we are very grateful. Thanks
For the first time, this conference was held in Switzerland, also to Carla Nunes, Per Christopher Foster, and Eduardo
in the heart of the watchmaking region. It was a great Pulido for copy-editing and designing the proceedings,
opportunity for participants to visit major collections of and to Miguel Mertens, Guy Silva, and team from use.it
objects of this type and to discover other aspects of Swiss for producing and managing the Paper-flow platform.
heritage and current manufacturing activities.
We warmly welcome you all to Neuchâtel!
In addition, and also for the first time, this triennial event
was hosted by a conservation training school—the Haute
Ecole Arc Conservation-restauration (HE-Arc CR) in
Neuchâtel. The study and conservation of metals are two
key areas of its programs, and it has carried out many
Brigitte Bachelard
research projects over the last 20 years. Their results have
Director General, HE-Arc
been incorporated into bachelor and master curricula and
are thus continuously passed on to future generations of
conservators. For many students, coming into contact
with professionals from all over the world and learning
about the up-to-date outcomes of metal heritage research
Dr. Régis Bertholon
are a wonderful opportunity to enrich their knowledge
Head of Studies and Research, HE-Arc CR
and develop professional links.
We would like to thank all those who have presented
papers and posters as well as other participants—we
are confident that the fruits of Metal 2019 will be rich
and productive and will lead to further international
exchanges and to the development of ideas, specific
knowledge, and techniques. We hope that members
of the ICOM-CC Metals Working Group and Swiss
professionals will find opportunities to build strong
relationships for further collaboration.
This program would not have been possible without the
collaboration of local institutions, the support of our
sponsors, and the active participation of many individ-
uals, colleagues, teachers, and students who have joined
the Organizing Committee: we are deeply grateful to all
of them.
We would especially like to acknowledge the contribution
of the Metal 2019 Steering Committee—Dr. Christian
Degrigny, Dr. Laura Brambilla, and Dr. Edith Joseph,
all HE-Arc CR teachers and researchers—who have
x  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Preface
This publication contains the proceedings of Metal
2019, the Interim Meeting of the International Council
of Museums – Committee for Conservation (ICOM-
CC) Metals Working Group. Metal 2019 was held in
Neuchâtel, Switzerland, from September 2 to 6, 2019, and
was hosted by the Haute Ecole Arc Conservation-res-
tauration of the University of Applied Sciences and Arts
Western Switzerland.
The Metals Working Group Interim Meetings are held
every three years and are organized by a different local host
institution each triennial. Metal 2019 was the 9th inter-
national meeting of the Metals Working Group. Previous
conferences have been held in France, Chile, Australia,
The Netherlands, USA, Scotland, and India. In 2019, the
conference was hosted for the first time by a University
featuring a preeminent conservation training program.
The Metals Working Group is the conservation commu-
nity’s foremost body dedicated to all aspects of the
conservation, preservation, technical analysis, and
study of objects made of metal and is one of the largest
working groups within ICOM-CC. Members are leading
experts in the conservation and scientific investigation
of metals from archaeological, historic, indigenous,
and artistic origins. The Metals Working Group works
to advance the field of metals conservation, research,
and scientific investigation, through the well-attended
and highly regarded Interim Meetings, by encouraging
cooperation and communication among members,
disseminating information on current conservation
practice and education, and facilitating the exchange of
ideas and experiences between all members.
At Metal 2019, authors from 21 countries presented 55
papers (chosen from more than 120 submitted abstracts),
providing an overview of the current state of research
in the field of metals conservation worldwide. Subjects
covered include: technical studies of ancient and contem-
porary objects; analytical investigation of metals and
their surface treatments using techniques ranging from
the most accessible to the most sophisticated; new devel-
opments in the field of preventive conservation; corro-
sion studies on archaeological, historic, modern, and
composite artifacts; innovative and traditional protec-
tion systems; new research into sustainable options for
stabilizing corroding objects; new methods of cleaning,
including dry-ice blasting and flame cleaning; and issues
raised by conservation treatments on prominent artifacts.
In addition to the papers, authors from 20 countries
presented 53 posters at the conference, expanding the
wide-ranging program. Subjects were equally diverse and
included the conservation of architectural aluminum,
use of gels for the treatment of metals, the conservation
of clocks, research into the use of vapor-phase corrosion
inhibitors, and individual case studies of large and small
metallic objects.
Many of the papers and posters were co-written by
multiple authors of national research laboratories,
cultural institutions, private practices, or joint research
programs, thus illustrating the cross-disciplinary nature
of our field. The proceedings gathered here are truly
international in their scope and represent an ongoing and
vibrant dialogue between conservation practice, research,
and science. Members of the Program Committee as well
as the editors of this volume subjected papers and posters
to a rigorous double-blind peer review.
The program for Metal 2019 introduced two unique pres-
entation styles to engage with attendees in different ways
and continued a successful off-site session began at Metal
2016. Ten papers by Young Conservation Professionals
were presented among their peers during the conference,
offering a chance for emerging professionals to share their
work with a global audience and receive feedback from
colleagues. Flash Poster Presentation sessions created a
lively and alternative way for poster authors to engage
with the audience and share their research. These sessions
were presented in addition to the dedicated poster session
at the conference.
The seven off-site sessions were held in leading museum
institutions in or around Neuchâtel and provided an
opportunity for conference participants to discover Swiss
heritage and discuss conservation issues raised by the
staff of these institutions.
Claudia Chemello
Terra Mare Conservation, LLC
Charleston, South Carolina, United States
Coordinator, ICOM-CC Metals Working Group
Christian Degrigny (Program Chair), Laura Bram-
billa, and Edith Joseph
Haute Ecole Arc Conservation-restauration
Neuchâtel, Switzerland
Steering Committee
 ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  xi

Acknowledgments
These proceedings were edited by Claudia Chemello, Valentin Boissonnas
Laura Brambilla, and Edith Joseph. Haute Ecole Arc Conservation-restauration,
Neuchâtel, Switzerland
The Organizing Committee for Metal 2019 comprised
Claudia Chemello, Tonny Beentjes, Emilio Cano Díaz, Laura Brambilla
Stavroula Golfomitsou, David Thickett, and Cátia Haute Ecole Arc Conservation-restauration,
Wesolowska from the Metals Working Group. Neuchâtel, Switzerland
Christian Degrigny from Haute Ecole Arc Conserva- Claudia Chemello
tion-restauration (HE-Arc CR) served as Program Chair. Terra Mare Conservation, LLC,
The Steering Committee in Neuchâtel comprised the Charleston, South Carolina, USA
following individuals from HE-Arc CR: Christian
Cristina Chiavari
Degrigny, Laura Brambilla, and Edith Joseph.
University of Bologna, Bologna, Italy
The local Organizing Committee comprised Régis Bert-
holon, Emilie Cornet, Hortense de Corneillan, Alexis James Crawford
Domjan, Nathalie Ducatel, Agnès Gelbert-Miermon, Sustainable Microclimates Ltd,
Gilles Hoffmann, Thierry Jacot, Romain Jeanneret, Birmingham, United Kingdom
Aline Michel, Manon Rais, Andrea Ramirez Calderon, Sara Creange
Guillaume Rapp, Isabelle Rérat, and Julie Schröter, as Rijksmuseum, Amsterdam, The Netherlands
well as the following students from HE-Arc CR: Martin
Barretta, Krystelle Besson, Inès Biscarel, Arthur Bozzi, Tamar Davidowitz
Sophie Braillard, Brice Chalançon, Victorine Dam, Sabine University of Amsterdam, Amsterdam,
de Raemy, Augustin Duc, François Dujardin, Agathe The Netherlands
Dumont, Laura Flückiger, Alice Gerber, Morgane Giran, Christian Degrigny
Elodie Granget, Frédéric Hsani, Joane Latty, Clara le Haute Ecole Arc Conservation-restauration,
Bail, Alexandra Lefebvre, Janine Meier, Zoé Meyer de Neuchâtel, Switzerland
Stadelhofen, Nicolas Moret, Marie-Paule Mottaz, Noémie
Nydegger, Céline Pacher, Emeline Perret-Gentil, Marion Emilio Cano Díaz
Petit, Audrey Regnault, Mathieu Rothenbühler, Delphine Centro Nacional de Investigaciones Metalúrgicas,
Rubin, Valentin Ruiz, Annick Simon, Mathilde Sneiders, Madrid, Spain
Quentin Valarcher, Théophile Vergnaud, Etienne von Nicola Emmerson
Gunten, Christophe von Kanel, Julie Vuignier, Camille Cardiff University, Cardiff, United Kingdom
Vuillème, and Zoé Snijders.
Stavroula Golfomitsou
We thank the Program Committee and supplemental
University of Gothenburg, Department of Conservation
external reviewers for volunteering their time to review
Gothenburg, Sweden
and discuss conference papers:
Elodie Guilminot
Vasilike Argyropoulos Conservation and research laboratory Arc’Antique,
Technical Education Institute of Athens, Nantes, France
Athens, Greece
David Hallam
Tonny Beentjes Metallic Heritage Consultant, Tasmania, Australia
University of Amsterdam, Amsterdam,
The Netherlands Edith Joseph
Haute Ecole Arc Conservation-restauration,
Régis Bertholon Neuchâtel, Switzerland
Haute Ecole Arc Conservation-restauration,
Neuchâtel, Switzerland
xii  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND

Milan Kouril We thank the Swiss National Science Foundation for

University of Chemistry and Technology, generously funding HE-Arc CR and allowing 18 individ-
Prague, Czech Republic uals to attend Metal 2019 with travel and accommodation
Paola Letardi grants, under the Scientific Exchange program (grant
Institute of Marine Sciences, National Research Council, IZSEZ0_189494).
Genoa, Italy
We thank the following authorities, institutions, and
Ian MacLeod sponsors who have made this conference possible:
Western Australian Maritime Museum, – Canton of Neuchâtel
Perth, Western Australia, Australia
– City of La Chaux-de-Fonds
Paul Mardikian – City of Neuchâtel
Terra Mare Conservation, LLC, – Federal Office of Culture
Charleston, South Carolina, USA – Haute Ecole Arc
Delphine Neff – International Council of Museums—Committee for
Laboratoire Archéomatériaux et Prévision de l’Altération, Conservation (ICOM-CC)
Gif-sur-Yvette, France – OrigaLys
Achal Pandya – Parmigiani Fleurier
Indira Gandhi National Centre for the Arts, – ThermoFischer Scientific
New Delhi, India – UBS Foundation
Céline Rémazeilles We thank the institutions which hosted off-site visits:
University of la Rochelle, La Rochelle, France – Art and History Museum, Neuchâtel
Luc Robbiola –A  ssociation pour la création d’un Musée de la
University of Toulouse Jean Jaurès, Toulouse, France Science et de la Technique (AMSTN), Neuchâtel
Katarina Schmidt-Ott – I nternational Museum of Horology,
Swiss National Museum, Zürich, Switzerland La Chaux-de-Fonds
Lyndsie Selwyn –L  aténium Archaeological Park and Museum,
Canadian Conservation Institute, Hauterive-Neuchâtel
Ottawa, Ontario, Canada – Renaud Foundation, car museum, Cortaillod
– S wiss Center for Electronics and Microtechnology
Carol Snow
Yale University Art Gallery, (CSEM), Neuchâtel
New Haven, Connecticut, USA We thank Miguel Mertens, Guy Silva, and team from
David Thickett use.it, Portugal, for their hard work and unwavering,
English Heritage, London, United Kingdom continued support in designing and maintaining Paper-
Lore Troalen flow, the paper management platform used for the
National Museums Scotland, Edinburgh, Scotland conference.
Philippe de Viviés We thank Carla Nunes and Eduardo Pulido, Lisbon,
A-CORROS Expertise, Arles, France Portugal, who were responsible for the layout and design
David Watkinson of this beautiful volume. In addition, Carla Nunes and
Cardiff University, Wales, United Kingdom Per Christopher Foster provided copy-editing for all of
Cátia Wesolowska the papers and the poster abstracts.
National Museum, Gdansk, Poland For more information about ICOM-CC and the Metals
We thank David Thickett, Paola Letardi, Blanca Ramírez Working Group, visit the webpage at
Barat, Emilio Cano Díaz, and Bartosz Dajnowski for http://www.icom-cc.org/31/working-groups/metals
presenting pre-conference workshops.

Introduction
This volume presents the proceedings of Metal 2019,
the Interim Meeting of the ICOM-CC Metals Working
Group, which was hosted by the Haute Ecole Arc Conser-
vation-restauration, Neuchâtel, Switzerland.
I am very pleased to have served as the Coordinator for
the Metals Working Group for this triennium, my last as
Coordinator. The conference was held for the first time
in Switzerland and was hosted by a University featuring
a preeminent conservation training program, another
significant achievement for the group. Bringing the
Metals Working Group conferences to different parts
of the world and new audiences is a goal that the group
strives to achieve each triennial, and one that I hope
will continue.
Bringing people together for a face-to-face dialogue at
the Interim Meeting is arguably an outdated method of
conference presentation. However, this form of dialogue
is the best way to learn from each other, break down
barriers that may exist from a distance, blur a profes-
sional and personal narrative, and bring our community
together for meaningful interactions. For more than 30
years—one of the longest running spans for a working
group of ICOM-CC—the Metals Working Group has
strived to be a forum for the exchange of best practices,
research, and new ideas, organized under the umbrella
disciplines of conservation practice and conservation
science. The numbers of attendees at Metal 2019 proved
that this type of interaction is still relevant and important
to members.
The research presented by conference participants was
exceptionally diverse, reflecting current trends, questions,
and issues in the field. I would like to thank all of the
Metal 2019 authors. This conference could not continue
to thrive without the high-quality contributions we
receive for the Interim Meetings.
Metal 2019 continued and expanded on several changes
implemented to the program at our last meeting in Delhi.
The diversification of the conference program has been
a much-loved change for the group. For Metal 2019,
this included a commitment to the inclusion of papers
and posters from Young Conservation Professionals,
encouraging emerging professionals to present their
work among their peers.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  xiii
Another addition to the program was a lively Flash
Poster Session where poster authors were challenged to
present a brief two-minute summary of their work in
the conference session in addition to the regular poster
session. A record-breaking 53 posters were presented at
the conference, more than at any other Interim Meeting
of our group.
We continued the popularity of the pre-conference
workshops by adding five workshops to the program,
greatly expanding the educational opportunities offered
to attendees of the conference.
Seven off-site sessions were proposed to the partici-
pants: held in leading museum institutions in or around
Neuchâtel, these provided opportunities to discuss
conservation issues on Swiss heritage objects.
A post-conference day trip to Val-de-Travers added
another great educational and social option to the
program.
We continued our successful partnership with Paper-
flow for paper and poster submission management, a
system that was created for the Metals Working Group
by use.it in Portugal. This system has transformed the
way submissions are managed and greatly streamlines
the publication process, from abstract submission to the
final published paper.
I would especially like to thank our hosts for the confer-
ence, the Haute Ecole Arc Conservation-restauration in
Neuchâtel, Switzerland. Organizing the Interim Meeting
is an extensive amount of work and financial invest-
ment. Our hosts, particularly Christian Degrigny, Laura
Brambilla, and Edith Joseph, members of the conference
Steering Committee, not only have generously provided
a wonderful venue, but we have also been privileged to
spend some time in the company of a teaching institution
training the next generation of young minds that will
shape our field.
Metal 2019, like previous conferences of the Metals
Working Group, is the result of a vast amount of unseen
work and preparation. I was assisted in these efforts by
a great team of Assistant Coordinators, namely Tonny
Beentjes, Emilio Cano Díaz, Stavroula Golfomitsou,
David Thickett, and Cátia Wesolowska.
xiv  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND

Please consider joining the ICOM-CC Metals Working

Group so that we continue to diversify our group and
learn from each other. Diversity and inclusion are not
merely buzzwords, they are real objectives that make an
organization successful and innovative. I encourage all
of you to continue the collaboration and connection that
is such an essential part of the Metals Working Group
conferences.

I hope you will find these proceedings educational,

informative, and energizing. I look forward to meeting
with you at our next gathering.

Claudia Chemello
Coordinator, ICOM-CC Metals Working Group
TECHNICAL STUDIES
A Diagnostic Study of the Cast-Bronze Great Buddha
Statue in the Kotoku-in Temple, Japan
Akira Fujisawa* Naomi Hemuki
Research Institute of Cultural Properties The National Museum of Western Art
Teikyo University Tokyo, Japan
Yamanashi, Japan hemuki@nmwa.go.jp
afujisawa@main.teikyo-u.ac.jp
*Author for correspondence
Abstract
The huge cast-bronze Great Buddha statue at the Kotoku-in
Temple in Kamakura, Japan, is a designated national treasure.
It was built in the 13th century and has been exposed to
the elements since the collapse of the building where it was
originally housed in the 15th century. Continued maintenance
work over the years has ensured that the statue remains in
a relatively stable condition, but as the last restoration was
in 1961 and the most recent scientific research in 1995, its
re-examination was considered necessary. This research
used empirical observation and scientific analysis in the aim
of acquiring the fundamental data necessary for the future
preservation of the statue. Observation included visual
condition checking, mapping and annotating of the condition
onto reproducible images, and digital imaging of the statue.
Scientific analysis used three non-destructive methods: digital
microscopy, portable x-ray fluorescence spectrometry, and
portable x-ray diffraction. Condition checks showed that the
statue is generally stable, although there are some areas of
concern, with the exterior exhibiting bird droppings, puddle
tidemarks, and corrosion rings from the accumulation of
Introduction
Kotoku-in Temple, located in the coastal city of Kamakura,
contains an enormous bronze seated statue of Buddha
measuring 11.3 m high and weighing 121 tons. It has
been designated one of Japan’s national treasures and a
historical chronicle describes that its casting began in
1252 (Kotoku-in 2018). Bronze casting of this type is one
of the most distinctive of Buddhist arts and the unique-
ness of this statue lies in its extraordinary dimensions
and near-original appearance.
The Great Buddha was originally housed indoors but
has been exposed to the elements since typhoons in 1334
and 1369 and a severe earthquake in 1498 destroyed the
building that housed it. Although the statue has been
unprotected for over 500 years, continuous maintenance,
rainwater around coin offerings. Dirt and dust was noticeable,
especially inside the statue, which also has grafitti and chewing
gum deposits. Water ingress was evident around the cast joints,
which has resulted in areas of chloride corrosion. Analysis
showed that the statue is mainly made of leaded bronze, but
also includes gold, silver, and brass. Investigations focused
on the distribution of chlorine, phosphorus, and sulfur, which
demonstrated that sea water, bird droppings, and air pollutants
have all had particular influence on the deterioration observed.
Other surface phenomena include a black layer consisting
mainly of cuprite and murdochite, white-green areas composed
of atacamite and easily affected by erosion and seawater, and
yellow-green areas consisting of pyromorphite owing to bird
droppings. While the corrosion found inside the statue was
mainly a stable cuprite, atacamite – which causes bronze
disease – was also partially detected.
Keywords
diagnosis, casting, outdoor bronze, Kamakura, corrosion, x-ray
fluorescence, x-ray diffraction
including the current biannual “spring cleaning” using
water-jet sprays, has ensured that the statue is in relatively
good condition.
The inside of the statue is hollow, which the public are
permitted to view. It is clear from its lattice patterns that it
was made in 40 separate castings, with evidence of three
different variations of a flash cast welding technique,
Ikarakuri, used for strengthening the joints.
Surprisingly, the statue is rather thin considering its
excessive size. Previous β (beta)-ray radiography esti-
mated it at between ~1 cm and 12 cm thick. As a result,
there was some anxiety about its possible collapse if
corrosion continues due to the perceived lack of wall
strength.
A DIAGNOSTIC STUDY OF THE CAST-BRONZE GREAT BUDDHA STATUE IN THE KOTOKU-IN TEMPLE, JAPAN
Figure 1. The Great Buddha statue: (a) front and (b) back
Between 1959 and 1961, a general investigation was
followed by conservation repairs to damaged parts. The
pedestal was reinforced with concrete and a seismic isola-
tion system engineered underneath. An arc-shaped metal
support was inserted into the hollow body to support the
upper body load, and cracks to the neck were reinforced
using fiberglass and epoxy resin (Committee for the
Conservation of the Great Buddha 1961). However, no
other details of previous conservation treatments had
been recorded.
These restoration projects and the later scientific analyses
in 1995 were focused on extant fragile parts and did not
include an overall evaluation of the statue’s condition
(Matsuda and Aoki 1996). No detailed data seems to
have survived except for that presented in the official
report, and this has made it difficult to accurately plot any
changes in condition since then. It was thus considered
time to conduct another survey, including investigating
17
the past restoration work, and this national research
project was carried out in early 2016 to acquire the
fundamental data necessary for designing a program for
the statue’s future care and preservation.
Research methods
Examination was required under all weather conditions
and so scaffolds were constructed both inside and outside
the statue with the latter covered with protective sheeting
to provide a workplace and to prevent the scattering of
debris and wastewater.
Inspections were made to record the statue’s current
condition as well as to detect traces of the foundry tech-
niques used during the statue’s creation. Three methods
were employed: firstly, condition-checking using sheets
made especially for the project; secondly, data-mapping
onto A3 printouts provided by a commercial imaging
company, Toppan Printing Co. Ltd.; and thirdly, extensive
imaging using digital single-lens reflex cameras.
Scientific analysis was aimed at diagnosing the current
status of the statue’s corrosion using non-destructive
methods. Surface color and texture were recorded
using a digital microscope (Microadvance DS-100) and
elemental analysis was conducted using a portable x-ray
fluorescence spectrometer (XRF) (Innov-X Systems,
Delta Premium DP-4000). This was used to detect
lighter elements by automatically switching the Ta bulb
between 40 kV and 15 kV in the “2-beam mining plus”
analytical mode. The detector area was approximately
10 mmφ and a 90 second beam time was applied. By
employing a fundamental parameters method (FP) for
the XRF, a semi-quantitative value could be calculated
for each element more readily than by comparing XRF
spectra, although the results do not show an element’s
exact concentration. Crystalline structures were iden-
tified with a portable x-ray diffractometer (XRD) by
TechnoX of Japan. This was custom made and allowed for
in-situ non-destructive analysis. It combined an Oxford
Instruments radiation shielded x-ray tube XTF 5011(Cu
bulb), an Amptek XR-100CR-0.5-BE-S x-ray detector and
an Amptek PX2CR preamplifier. The Cu bulb voltage
was 25 kV and the electric current was 0.2 mA with a
measurable diffraction angle of 2θ between 20° and 70°
according to the different object surface shapes. Meas-
urement conditions were set at a 0.1° step angle and a
1 second time interval and took ~15 minutes per spot
analyzed.
18  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Areas for in-situ analysis were chosen with reference
to the 1995 scientific analysis, but in order to quantify
the condition better, new analytical areas were evenly
plotted both vertically and horizontally across the statue.
Although it was impossible to examine several spots
because of the scaffolding or due to the complicated
surface shape, 105 areas on the outside of the statue
and 17 on the inside were investigated using a digital
microscope and XRF, and 30 outer and 7 inner areas
using XRD. All data from this research was digitized
with high-resolution images in cooperation with Toppan.
Figure 2. During diagnosis: (a) detail of deterioration/condition checking;
(b) detail of in-situ analysis using portable XRD
Results and discussion
Diagnosis of current condition and fabrication
techniques
Condition checking showed that, on the whole, the statue
was in a good state. The examination revealed some areas
of concern including accumulations of dirt and dust
both inside and outside, and the exterior exhibited bird
droppings, puddle tidemarks, and corrosion rings from
TECHNICAL STUDIES
the accumulation of rainwater around coin offerings.
Although chloride corrosion was not extensive, different
parts of the statue had various corrosion products along
with noticeable color changes. Furthermore, the interior of
the statue had been vandalized with graffiti and chewing
gum. Water ingress was also apparent around the casting
joints, which had provoked some chloride corrosion.
Figure 3. Types
of damage: (a)
bird droppings,
(b) chewing gum
hardened around
a coin offering
and a pebble, and
(c) water-marks
and coin-shaped
 corrosion
Because the project presented a rare opportunity to
improve the statue, conservation work was divided into
three tasks: the resetting of an inlaid panel that had dropped
down to the left knee in order to reduce water ingress, inner
and outer surface cleaning to remove any harmful surface
deposits (other than the chroride corrosion being used in
our research), and, lastly, thorough documentation of the
statue and the conservation work carried out.
After mechanically removing the deteriorated adhesive
from the dropped panel’s recess, an epoxy resin was
evenly applied to the panel’s verso and it was gently
A DIAGNOSTIC STUDY OF THE CAST-BRONZE GREAT BUDDHA STATUE IN THE KOTOKU-IN TEMPLE, JAPAN
hammered back into place using a wooden batten to
protect the surface. In order to prevent further water
ingress, members of the temple staff were advised to mop
up any puddles on the knee and to monitor the condition
of the inner wall.
Surface cleaning was executed using dry-cleaning
methods for the inside and both dry- and wet-cleaning
for the outside of the sculpture. Dry-cleaning with natural
bristle brushes was followed by high-pressure air cleaning
of any dirt and dust especially in unreachable parts and
then removal using a wet-dry vacuum cleaner.
After dry-cleaning the exterior, wet-cleaning was
performed using water and a non-ionic surfactant. A
high-pressure water jet spray was used to moisten the
surface of the statue, and it was then washed with a
3% (v/v) Triton X-100 solution in water applied with
bristle brushes before being thoroughly rinsed off. This
process was repeated twice with the wastewater removed
using brushes and the wet-dry vacuum cleaner. The
surface was then dried with cotton dustcloths.
To remove any stubborn surface deposits after
wet-cleaning, localized chemical cleaning was used: bird
Figure 4. XRF spectra for metals identified in (a) a white-green area on the cheek, (b) a gold area on the cheek, (c) a black area on the urna, and (d)
a green area on the ushnisha
19
droppings were cleaned with cotton swabs and ethanol,
graffiti was removed with propan-2-one or ethyl acetate
on cotton swabs, and chewing gum was mechanically
removed until very thin and then dissolved with ethyl
acetate. After cleaning, the statue was noticeably clearer
and brighter.
Scientific examination
In order to evaluate the corrosion, the statue’s constituent
materials were examined and the distribution of elements
affecting corrosion was investigated. Corrosion products
and likely corrosion mechanisms were determined based
on these results.
Materials used in the statue
Analysis confirmed that the statue was made of bronze
containing a high lead content with copper and tin
detected in many of the analyzed areas (Figure 4a). This
confirmed analyses carried out in 1935 and 1957, espe-
cially the latter, which suggested, based on ten samples,
that the average composition was copper with the addi-
tion of 9.26% of tin and 19.58% of lead alloy, although the

20  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

amount of deviation was significant (Committee for the Bird droppings were found on the upper area of the
Conservation of the Great Buddha 1961, Ishino 1977). ushnisha and large peaks of phosphorus and calcium,
as well as copper and zinc, were confirmed (see
Some areas of the statue exhibit gold, especially around
the face, and as the gold is only found in recesses it Figure 5b). The phosphorous is likely to be derived
is assumed that it was eroded from exposed surfaces from the droppings.
through weathering. XRF peaks for gold were detected Areas with over 2 mass% of phosphorus detected using
in addition to bronze elements (Figure 4b), but mercury FP had yellowish-brown streaks and represented 6% of
was not detected. Because the XRF peaks for gold and
all the areas analyzed. It is likely that the phosphorus is
mercury are similar and the equipment used cannot iden-
present in the streaks because it is dissolved in rainwater
tify small amounts of mercury, the method for applying
which then runs down the statue. In contrast, there was
gold to the statue cannot be further discussed .
little calcium content detected in the streaks, presum-
Parts of the statue exhibiting other metals are the urna,
a spiral projection from the forehead, and the ushnisha,
a button-shaped detail on top of the head. Large peaks
of silver were detected in the urna (Figure 4c), which
corresponds with the description of its renewal in 1735
in Daibutsu-son saikoki (“Records on the Restoration
of the Great Buddha”). Copper and zinc were detected
on the ushnisha but not tin (Figure 4d), and it was
identified as brass, which was probably imported as a
description exists in the 1735 records stating that “the
ushnisha was made of Chinese brass.” When different
metals are used together, galvanic corrosion can occur,
but nothing remarkable was observed around the joint,
suggesting that the statue had had non-conductive
corrosion products on the surface when these later
parts were attached.

Distribution of elements affecting corrosion

As elements heavier than magnesium are detectable
with the portable XRF equipment, chlorine, phosphorus,
and sulfur that affect corrosion were focused on for
semi-quantative analysis using the FP method.

Although chlorine was detected in all the analyzed areas,

those over 30 mass% using FP were mainly white and
represented 11% of the total areas analyzed. In contrast,
areas with relatively low chlorine intensity tended to be
exposed and more convex-shaped. Figure 5a shows the
results from an area with high chlorine assumed to be
due to wind and rain deposition from the sea, some 850
m to the south. As these salts are likely to be washed away
by rain and in the statue’s spring cleaning, relatively low
chlorine content was detected on the more exposed areas.
The area most affected by sea salts was the upper front Figure 5. XRF spectra showing elements affecting corrosion: (a) white-
part of the statue, including the face and neck because green area on the right armpit, (b) bird droppings on the ushnisha, and
they face seawards, to the south. (c) white-black area on the left chin
A DIAGNOSTIC STUDY OF THE CAST-BRONZE GREAT BUDDHA STATUE IN THE KOTOKU-IN TEMPLE, JAPAN
ably because the calcium compound remains relatively
insoluble in rainwater.
Areas where over 7 mass% of sulfur was detected using
FP were the back of the head and some brown streaks.
These represented 18% of all the areas analyzed and
Figure 5c shows a characteristic analytical result. Sulfur is
likely to be derived from SOx substances in air pollution
and deposited by rainwater. In a wind-tunnel test, it was
confirmed that wind blowing from the town towards the
north of the temple tends to blow onto the back of the
Buddha’s head (Miura et al. 1998).
State of corrosion
On the exterior, black, white-green, and yellow-green
areas are mainly visible. The majority of the back and
hollows on the statue’s front have black surfaces, char-
acterized in Figure 6a. Murdochite (Cu6PbO8, JCPDS:
07-0028) was detected as well as cuprite (Cu2O, JCPDS:
00-001-1142). Murdochite is a compound which appears
in copper ore deposits with a high lead content. Its
occurrence on the statue is thought to be due to a lead
content of less than 20%. Further investigation is needed
Figure 6. XRD patterns with optical microphotographs from the exterior surface of the statue: (a) black area, (b) white-green area, (c) yellow-green
area, and (d) black area from behind the statue
21
to confirm whether the cuprite is present in or under the
murdochite layer.
In the whitish areas where a high percentage of chlorine
was detected, atacamite (Cu2Cl(OH)3, JCPDS: 00-023-
0948) and cuprite were detected, as shown in Figure 6b.
Cuprite is a red-brown compound, and it was thought
that it was present under the atacamite because it was
unobservable under the microscope. Where the white-
green atacamite products were found, no active pitting
corrosion was detectable.
Yellow-green compounds were found mainly on the
statue’s front, such as on the shoulders, the knees, and the
chest. Figure 6c shows a characteristic XRD pattern and
the presence of a white-green compound on the yellow-
green compound was observed under the microscope.
Pyromorphite (2Pb5(PO4)3Cl, JCPDS: 00-006-0389),
which has a yellow appearance, was detected. As pyro-
morphite is produced by the reaction of the phosphorus
and the lead in the statue, it is found on the shoulders
and knees where birds perch, or on the chest area where
droppings run down from the head.
22  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Compounds containing sulfur were detected on some
areas of the statue’s proper left and proper left back, as
shown in Figure 6d. Copper sulfate (CuSO4, JCPDS:
00-001-1081) on the upper facing areas and copper
sulfide (Cu1.96S, JCPDS: 00-029-0578, Cu2S, JCPDS:
01-089-2670) on the lower areas were also detected.
These results indicate the influence of airborne pollut-
ants from the town and in the 1995 survey sulfur
compounds and sulfates were detected in 68% of all the
analyzed areas (Matsuda and Aoki 1996). In contrast,
in this study they were only detected in 13% of the
analyzed areas, showing a significant reduction in the
interim years, suggesting that the level of air pollution
has lowered and that sulfates have been washed away
by rainwater.
Most of the interior appears black with white-green
corrosion found in some areas. Figure 7a shows the
analytical results from one black area with a compound
identified as cuprite (Cu2O, JCPDS: 00-001-1142), an
inert corrosion product thought to be produced by
high temperature oxidation during casting because
the original casting sand was present on the interior’s
surface. Using elemental analysis, high chlorine content
was confirmed in the white-green spots, along with
Figure 7. XRD patterns with optical microphotographs from the interior
of the statue: (a) black area and (b) white-green area
TECHNICAL STUDIES
atacamite (CuCl2(Cu(OH)2)3, JCPDS: 01-077-0116) and
metallic copper, as shown in Figure 7b. The white-green
spotted surfaces were swollen by pitting corrosion and the
chemical composition, form, and color of the corrosion
product indicated that bronze disease had occurred.
This is assumed to have been caused by salts carried in
rainwater because streak marks can be seen around the
affected areas from leaks due to damaged or open joints.
Any chrolide ions (Cl-) in the rainwater will transfer and
accumulate in crevices due to the difference in electrical
potential across the resulting oxygen concentration cell
on the metal. As a result, pH is reduced and the protective
corrosion layer is destroyed, leading to localized crevice
corrosion. However, there were relatively few areas where
bronze disease was confirmed.
Presumed corrosion mechanisms
While atacamite was detected on the statue’s exterior and
interior, their different corrosion forms indicate that the
reaction mechanisms differ according to where they take
place. On the exterior, the upward facing surfaces gener-
ally exhibited a white-green color while other surfaces
tended to be black. On the front elevation, convex areas
were white-green and any hollows were black, with any
yellow-green areas almost overlapping with the white-
green ones.
Figure 8 represents the possible corrosion mechanisms
according to the distribution of the detected compounds.
Bronze surfaces that have a high lead content corrode
such that they form black layers of cuprite and murdo-
chite. If the areas with the black layer are hollow or on
the statue’s back and sheltered from the wind, they tend
to remain as they are (Figure 8a). If they are convex
or located on the front elevation, and more affected
by wind, erosion by blown sand occurs and the black
layer becomes thin or cracked (Figure 8b). At this stage,
atacamite is produced by contact with airborne salts
(Figure 8c). Pyromorphite is also produced in any areas
where bird droppings are deposited (Figure 8d). After
the inert black layer has formed, therefore, atacamite is
produced in exposed areas through the combination of
cuprite/murdochite erosion and airborne salts from the
sea, while pyromorphite is produced by contact with
bird droppings. The different colors seen on the surface
thus depend on the compound produced by the reaction
mechanisms described.
A DIAGNOSTIC STUDY OF THE CAST-BRONZE GREAT BUDDHA STATUE IN THE KOTOKU-IN TEMPLE, JAPAN 23

Inside the statue some areas exhibit typical symptoms

of both bronze disease and pitting corrosion and these
relate to the ingress of water. Considering that the statue
has been exposed to the elements for over 500 years, and
that bronze disease is only evident in very localized areas,
the statue’s condition is relatively stable. It is thought
that this is in part because the bronze surfaces have been
chemically stabilized by the layer of cuprite.

Figure 8. Schematic images detailing corrosion mechanisms: (a) structure

of the stable area, (b) damage to the surface by erosion, (c) formation of
atacamite by wind and airborne salts, and (d) formation of pyromorphite
by bird droppings

Preservation of research records

The condition report from this research was made in an
easily accessible PDF format, and all the information,
including scientific analyses, was mapped onto high-defi-
nition images to support future research in helping to
evaluate the condition and re-examination of the current
areas of analysis. DVDs with the data and reports were
distributed to various institutions to allow public access
to the work. As shown in Figure 9a, 13 manufacturing
techniques and 15 condition descriptions were annotated
in different colored layers on the images, which can all
be displayed individually. By clicking the tags for each
analyzed area (Figure 9b), the results of all the particular
analyses can be seen (Figure 9c) (Committee for the
Conservation of the Great Buddha 2018). This format
Figure 9. Integrated research records: (a) map of manufacturing
does not include any past analytical results because the techniques and condition descriptions, (b) areas of scientific analysis,
raw data used were not recorded. and (c) display of all analytical results for a designated area
24  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Conclusion project (高徳院国宝銅造阿弥陀如来坐像平成27年度

The conservation research project for the cast-bronze
Great Buddha statue in Kotoku-in Temple set out to inves-
tigate its current condition and obtain the scientific data
necessary to determine any future conservation work.
As the results show, it was found to be relatively stable,
with no serious threats discernible, although graffiti and
chewing gum deposits were noted in some areas.
Analysis confirmed that the statue was made of bronze
with a high lead content and with gold, silver, and brass
used in some areas. Chlorine derived from the nearby
seawater, phosphorus from bird droppings, and sulfur
from air pollutants were all confirmed as present on
its surfaces. Corrosion to the interior was confirmed
as being stable overall considering the statue’s outdoor
exposure for over 500 years. Exterior areas affected by
erosion and airborne salts had a white-green layer of
atacamite, whereas areas affected by bird droppings had
a yellow-green layer of pyromorphite. One conservation
measure would be to reduce the production of atacamite
by mitigating the effects of erosion by installing trees or
sand-banks as a windbreak. Compared with earlier anal-
yses, the percentage of areas where copper compounds
containing sulfur were detected were greatly reduced,
suggesting that the quality of the environment had
improved in the intervening years.
保存修理報告書). Kamakura: Kotoku-in Temple.
Ishino, T. 1977. Cast metals: The origin and the history of
techniques (鋳物: 技術の源流と歴史). Tokyo: Industrial
Research Center.
Kotoku-in. 2018. “An overview of the Great Buddha.”
Kotoku-in Temple website: www.kotoku-in.jp/en/char-
acteristic.html (accessed 15 August 2018).
Matsuda, S. and S. Aoki. 1996. Analysis of corrosion prod-
ucts formed on the Great Buddha image of Kotoku-in
Temple in Kamakura. Science for Conservation 35: 1–20.
Miura, S., S. Aoki, W. Kawanobe, O. Nakamura, and K.
Miyashita. 1998. Climate of the Great Buddha image of
Kotoku-in Temple in Kamakura. Science for Conservation
37: 1–14.
Authors
Akira Fujisawa is head of the Scientific Research Center
for Cultural Properties at Teikyo University and has a
Ph.D. in conservation science from Tokyo University
of the Arts.
Naomi Hemuki is head of conservation at the National
Museum of Western Art and has an M.A. in museum
studies from the University of Leicester.

In conclusion, it is suggested to conduct regular moni-

toring of the statue using this research as a guide to help
detect any potential change in condition at an early stage.

Acknowledgments
Special thanks are due to Prof. Takao Sato, the Chief Priest
of Kotoku-in Temple, and Dr. Noriyuki Morii and Dr.
Noriko Hayakawa of the Tokyo National Research Insti-
tute of Cultural Properties for allowing us to participate
in this project. We are also grateful to all team members
for their important contributions to this project.

References
Committee for the Conservation of the Great Buddha of
Kotoku-in Temple. 1961. Conservation report of the Great
Buddha of Kotoku-in Temple (高徳院国宝銅造阿弥陀如
来坐像修理工事報告書). Kamakura: Kotoku-in Temple.
Committee for the Conservation of the Great Buddha of
Kotoku-in Temple. 2018. Preservation and interdiscipli-
nary research report of the National Treasure subsidized
Technical Study and Conservation of a Large-Scale Metal
Buddha Statue
Kasthuri Arachchilage Anusha Kasthuri
Department of Archaeology of Sri Lanka
Colombo, Sri Lanka
anusha.kasthuri@gmail.com
Abstract
This paper presents a technical investigation on the archaeological
significance and conservation of a large-scale amalgam gilded
bronze standing Buddha statue found accidently in 2011 in
the village of Ankenda Wawa, Kurunegala District, Sri Lanka. In
order to date the statue, a sample of core material was removed
and radiocarbon-dated to AD 385–535. The manufacturing
technique and overall condition of the statue were examined
by magnification, film radiography, and elemental analysis using
x-ray fluorescence spectroscopy. The Buddha statue, which is
one hundred centimeters tall (even if lacking its Siraspatha
ornament at the top of the head), was probably cast using
the indirect lost-wax method without a base but with tangs
protruding from the bottoms of the feet. Due to later damage,
which perhaps occurred during discovery, a view was afforded of
the inside of the cavity containing what appeared to be an intact
hard black core material and the typical placement of an iron
armature. The wall thickness of the casting varies from 1.2 mm
to 5.5 mm. The statue weighs approximately 59 kg. There are
Introduction
A great deal is known about Sri Lankan statuary from
the perspectives of historical, stylistic, and iconographic
attributes, but very little is known about it in technical
terms, including material and technological aspects. Only
a few analyses have been undertaken and published by
scholars, and most of these are of objects of unknown
provenance in Sri Lanka.
This aesthetically appealing standing bronze Buddha
figure stands 100 cm tall and presents a curiously lively
representation of Buddhist society from over 1,500 years
ago. Considering the size, quality of figure, and the level
of craftsmanship, the casting probably reflects a very high
level of patronage during this time period. Image vener-
ation as a prominent feature of Buddhism is a common
and ancient practice in Sri Lanka. As stated in The Maha-
vamsa, the Great Chronicle of Sri Lanka, the kings made
metal statues for the great Bodhi tree in Anuradhapura
iron core pins, filled and unfilled sprues and vents, and casting
flows located on the nose, front of the body, and at the end of
the robe. Despite partial deterioration of the gilding, the statue
appears to be in a structurally stable condition, even though
casting defects, some corrosion, and signs of damage such as
separation and losses are evident. It is clear that the exposed
iron armatures are problematic, since their further corrosion
may threaten the condition of the casting. A combination of
mechanical and chemical methods was applied during the
conservation process. A mixture of 5% benzotriazole in ethanol
was applied by a brush as a stabilizer and Incralac was chosen
as a coating. Restoration of the neck area is being carried out in
preparation for a forthcoming exhibition at the Archaeological
Museum in Anuradhapura.
Keywords
bronze, hollow-casting, Buddha, cleaning, stabilization,
restoration, lost-wax, Sri Lanka
in different time periods, including King Vasabha
(AD 127–171), King Voharaka Tissa (AD 269–291) and
King Mahasena (AD 334–361) (The Mahavamsa, XXXV
and XXXVI). However, the earliest direct evidence of
the production of copper-alloy sculptures is uncertain
(Kasthuri 2016). A study by Reedy and Harlacher (2007)
places the commencement of bronze sculpture fabrication
in Sri Lanka in the late Anuradhapura period, i.e., from
ca. AD 432 to 993, while Brown (2012) gives the date of
AD 6–7 to the earliest images in bronze.
Provenance and date of the statue
The Buddha figure was accidently discovered in the
village of Ankenda Wawa, in the area of Galgamuwa,
of Kurunegala District,1 Sri Lanka, during land devel-
opment. A farmer, using machinery to excavate the
ground, uncovered the sculpture and donated it to the
26  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Department of Archaeology in 2011. Since the statue’s
immediate provenance was therefore definitely verified,
a sample of core material was removed to determine its
date. The result of C-14 dating conducted on this sample
indicated a date of AD 385–535 (Figure 1) which, as
assumed by John Twilley (personal communication,
November 2018), will typically be older than the statue’s
actual date.2
Iconography of the standing Buddha figure
The base of the Siraspatha (flame of enlightenment) is
visible at the top of the head, but the Siraspatha itself
was lost before discovery. The right hand is making the
asis mudra hand gesture, with the palm turned partially
to the figure’s left. The pleated outer robe evident on
the front does not continue on the reverse side of the
statue. The lower robe can be seen at the bottom of the
figure. The hem crossing the chest is raised and the robe
gently falls on both sides of the left arm to the lower
part of the body. The lack of clothing on the back of the
figure clearly indicates that it was not meant to be seen
Figure 1. Calibration of radiocarbon age to calendar years
TECHNICAL STUDIES
(Figure 2). This would be appropriate if the figure was
used in worship or teaching, despite it being hollow cast.
Although large images were usually hollow cast during
the early Anuradhapura period (ca. 200–432) (Listopad
2003), they were solid cast during the late Anuradhapura
period. The likely reason for this change in practice
was the Sariputhra Silpa text, which strictly forbade
the making of hollow images for worship. This figure,
which originally stood on a separately cast base that is
now missing, was likely connected to the royal courts
or monastic environments, where public worship would
not have taken place.
Figure 2. Iconography
TECHNICAL STUDY AND CONSERVATION OF A LARGE-SCALE METAL BUDDHA STATUE 27

Experimental details separation of an indirect section of an already existing

wax model and radiography indicates the evidence of
The casting technology for this statue continues to
sinking marks, suggesting the possibility that a segment
be investigated at the Department of Archaeology’s
was added to the wax model (Figure 4). The thickness of
conservation laboratory in Colombo, where the statue
the metal is also varied. Caliper measurements along the
is also being conserved since it will be included in the
opening at the head and body established that the metal
forthcoming exhibition at the Archaeological Museum
in Anuradhapura in 2019. Film x-radiography, which is
only available at Sri Lanka’s Atomic Energy Authority,
was undertaken together with careful visual examination
of the surface. In addition, qualitative x-ray fluorescence
spectroscopy (XRF) analysis was performed on the
surfaces of the object to identify the alloy and gilding.

Results and discussion

Fabrication technique
The statue is very heavy, weighing approximately 59 kg
before cleaning. Qualitative surface analysis by XRF
showed that the statue is essentially made of a leaded
bronze alloy with amalgam gilding on the surface.
Indeed, close visual inspection suggested that the head
of the statue consists of two layers of wax: one for the
scull and another for the hair. There are tool marks as
well as scoring, strongly implying that these effects were
achieved in a wax model. In addition, visual inspections
and use of a hand magnet revealed the presence of five
corroded iron chaplets that were used during casting
to hold the core in position. These are most evident at
the front of the body and at the left shoulder (Figure 3).
All these characteristics are typically associated with
lost-wax casting.
Figure 3. (A) Five chaplets: at the right and left shoulder, on the right
of the stomach, on the drapery, and below the knee; (B) a corroded
chaplet at the left shoulder; (C) tool mark and hair modelled into the
wax layer; (D) hair delineated in the wax; (E) wrinkles at the end of the
channel marked on wax; (F and G) separately molded and attached ear
The entire left side between the drapery and the body
has casting patches that run from the elbow to the end
of the robe. Also under the robe can be seen the metal
Figure 4. (A) Cast in place between the body and drapery; (B) x-ray
showing the position of casting patches; (C) separate wax segment
and sealed sprue
Figure 5. (A and C) Wax drip on interior metal and sealed sprue; (B, D,
and E) metal wall thickness
thickness varies, with the front edges being the thinnest,
at 1.2 mm, and the back wall the thickest, at 5.5 mm
(Figure 5). Due to damage that may have occurred
during discovery, the inside of the cavity containing what
appears to be an intact core is visible. This reveals a typical
placement of two flanking flat armature segments with
rectangular cross sections of approximately 7 × 12 mm.
These are assumed to be wrought iron. A similar-sized
third armature runs horizontally and is likely to extend
into both arms. The current suggestion is that the vertical
armatures start in the head and may have continued down
to the bottom of the torso, without reaching the feet.
Additionally, there would have been two parallel iron
rods, larger in cross section, that once protruded from the
bottom of the torso. The wax model was formed around
a core that was held upright by the rods and then, once
the torso was completed in wax, the wax ankles, feet,
and tangs were added over the iron rods. The chaplets,
sprues, vents, and casting repairs are mainly found on
28  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
the front of the statue. There are several unfilled square,
rectangular, and round cavities at both hands and at the
face that functioned as minor sprues to allow the flow
of molten metal and/or as vents to allow gases to escape.
All of these ensured uniform casting. Around five main
sprues exist: on the left armpit, on the elbow of the left
hand and beginning of the drapery, and at the right corner
of the lower robe. In addition, there are casting repairs
on the underside of the torso around and between the
ankles. The flat underside of the feet also exhibits casting
porosities. The location of two cut and filed-away major
sprues was determined by the convex swollen appearance
of the metal in a generally circular area, measuring about
1.7 cm in diameter. One is visible at the left corner at
the end of the back of the torso and a second, which is
TECHNICAL STUDIES
a 2 cm diameter circle, reappears in the right corner of
the lower front of the robe. A filled sprue is visible on
the side of the drapery running from the front of the
left arm to the back of the body. The two tangs may also
have served as in-gates (Figure 6). Unfilled undercuts
may have served as minor sprues or vents. For example,
one is square-shaped and located on the side of the left
arm (8 × 8 mm) and another, which is a rectangular
undercut, is visible on the front of the raised right hand
(10 × 0. 5 mm) (Figure 7). The square-shaped closed
sprue appears inside the face below the jaw (see Figure
8). This sprue faced forward and the molten metal was
possibly poured horizontally. There is also a waxy drip
close to the interior metal (see Figure 5). The casting flaws
at the nose are only revealed by radiographic examina-
Figure 6. (A) Merged metal on the back of the robe;
(B and C) a closed sprue; (D, E, F, and I) casting
porosities; (G) a cast-on repair; (H and K) a sealed
  circular sprue; (J) sealed sprues on tangs
  Figure 7. Minor sprues or vents
Figure 8. (A and B) Location of a square-shaped
sealed sprue; (C) the seat of a sprue; (D) the back
of the armpit (with no casting repair) and buttock
(with a casting repair); (E) direction of metal flow (i)
cast-on metal or melted-wax into the core; (ii) a
closed minor sprue); (F) no indication casting
repairs in the stomach area; (G) cast-on repairs in
  the armpit
TECHNICAL STUDY AND CONSERVATION OF A LARGE-SCALE METAL BUDDHA STATUE
tion and were not visible during the visual examination
(see Figure 5). Seen in the radiograph of Figure 8E are
arrow-shaped flow lines (possibly molten wax) which
radiate from the armpit and around the right side of the
body. The hypothesis is therefore that the mold was cast
at a slight angle off the horizontal position and casting
damage on the nose occurred at what would have been
the highest point of the mold in the horizontal casting
position. The molten metal would have been poured
from front to back, accumulating toward the center of
the buttocks. The surface of the buttocks was repaired
using cast patches for an unknown purpose (Figure 8).
Inlays
The presence of mercury within the gilding is indicated
in the XRF spectrum by the shoulders on the gold peaks
(Figure 9). Originally the eyes would have been inlaid
with crystals, stones or semi-precious gems, a common
feature of historic Sri Lankan statues. There appears to be
a rough surface inside both eye cavities that would have
provided good adhesion for the inserted eyes.
Figure 9. Traces of mercury (red arrow) next to a gold peak on the XRF
spectra
Condition and deterioration
The overall impression was that the object was well
preserved. However, deterioration of the gilding, casting
29
defects, corrosion of the bronze, and later damage, such as
separation and losses, are evident. The head of the sculp-
ture has been broken and separated from the body, and
there is drastic physical damage due to cuts and abrasions
that expose the bare metal on the back of the shoulder
and on the hair, likely due to the use of machinery during
the sculpture’s discovery. This damage has disturbed
the aesthetic value of the statue. Even though physical
damage is observed over the entire statue, there are large
areas where the gilding is intact. The presence of several
corrosion products, including hygroscopic compounds
at the gold/bronze interface, makes fire-gilded bronze
quite unstable (Masi et al. 2016). The statue also suffers
from active copper chloride corrosion which results in a
pale-green product known as paratacamite, the so-called
“bronze disease” (Figure 10). Degradation of gilded
bronze is generally initiated by galvanic coupling due
to the presence of pores and defects in the gilding layer.
This may lead to the growth of corrosion products at the
gold/bronze interface, inducing blistering or break-up of
the gold layer (Masi et al. 2016).
Conservation
The choice of treatment methods was determined by the
materials and technology of the statue, the sculpture’s
condition, and the desired treatment outcome.
Cleaning
Initial, investigative cleaning was carried out to identify
fabrication, location of sprues and chaplets, and any
inlays, as well as to establish surface condition. At the
beginning, a covering layer of loose soil was, to some
extent, removed mechanically with wooden tools and
plastic spatulas. However, the surface of the statue
remained obscured by the clay. To remove the clay,
Figure 10. (A) Separation and exposed iron
armature; (B) corroded iron armature on top
of the head; (C) loss of inlays; (D) cracked
and damaged places; (E) a corroded chaplet;
(F) deteriorated gilding; (G) corrosion pits
and bronze disease; (H) bronze disease; (I)
abrasion; (J) the cracked and exposed iron
  armature; (K) physical damage by machinery
30  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
initial mechanical cleaning was cautiously carried out
incrementally and under magnification using scalpels,
taking special care not to damage the gilded surface.
A thin layer of re-deposited copper corrosion on the
gilt layer was removed by a combination of mechanical
and chemical methods. A thin but tenaciously adhered
layer of corrosion was difficult to remove but was
successfully quantitatively reduced through mechanical
cleaning. Some basic solvent tests were carried out to
determine potential methods of removing corrosion. It
was clear that the copper corrosion could be detached
and was ultimately removed by a 3% solution of ethyl-
enediaminetetraacetic acid disodium salt (Na2EDTA) in
ethanol applied using cotton swabs and soft brushes. The
minimum amount of chemicals was used and only when
necessary. To clean the reverse side of the statue, it was
turned upside down and placed on a support made of
polyethylene wadding to avoid point loading of the hands,
fingers, and feet. Mechanical and chemical cleaning
was undertaken, as per the front for the casting. The
exposed iron armatures, including broken pieces, were
mechanically cleaned using scalpels and a glass-bristle
brush without using any chemical treatments.
Stabilization and coating
Although there are several methods for removing chlo-
ride ions, the author decided to apply a benzotriazole
(BTA) treatment given the remarkable corrosion resist-
ance it imparts to copper alloy objects. A 5% BTA solu-
tion in ethanol was brushed on the metal surface. Bare
metal, which had been exposed in a few places, and
areas of damage developed a greenish color due to the
BTA treatment. Incralac was chosen as a coating since
it has been widely used in conservation and meets the
requirements for chemical stability, adhesion to the metal
surface, and sufficient elasticity. For the uncovered iron
armatures in both the head and the body (as well as the
broken piece), protection was achieved with a coating of
Renaissance microcrystalline wax to produce a barrier
that would exclude moisture from the internal environ-
ment of the statue.
Restoration
Since there are no resources to control the exhibition
environment in the historic Anuradhapura museum
building, which is located in a dry zone of Sri Lanka, it
was decided to “seal” the core and iron armatures on the
TECHNICAL STUDIES
inside of the statue in a dryer-controlled environment
in the laboratory, in the hope that this would contribute
to the protection of the armature and thus the casting.
Although some additional corrosion of the iron is likely
to occur in the future, it is likely to be relatively small
since the fixed joint provides some degree of buffer
between the internal space and the external environment.
The restoration work was conducted under the general
guidance of Jerry Podany (personal communication,
November 2018).
The proposed restoration procedure was as follows: to
clean around the interior neck and near the joint; to cut
an Ethafoam plug to the shape of the interior cavity in the
neck area; and to provide a central hole to accommodate
the armature. The subsequent stages were to adhere the
recessed plug in place with Paraloid B-72 on either the
body or the head side of the joint; to coat the joint edges
of one segment of the joint (perhaps the head) with
20% B-72 in acetone solution and allow the coating to
completely dry (48 hours); to apply a structural adhesive
(Epofix hardener and resin 1:1) on several spots along the
other edge where the joint seems to fit well (only 9.5 cm
from 23.5 cm); to assemble the head and body and allow
the epoxy to set; to fill the gaps with epoxy mixed with a
microballoon filler; and, finally, to seal it with a coating
of 10% B-72 in acetone.
Conclusion
The iconographic, scientific, and technical study of the
Buddha statue led to a number of discoveries, including
its age, function, the qualitative composition of the alloy
and gilding, and a partial understanding of the casting
technology. The conservation and restoration of this
statue, comprising copper alloy, iron armatures, gilding,
and the unidentified contents of the core cavity, was a
challenge not taken lightly.
The hollow-cast mercury-gilded figure bears two circular
perforations – subsequently patched – on its front and
rear surfaces which indicate the possible locations
of main sprues near its extremities. Vents and minor
sprues can be observed on the upper part of the figure
that was cast lying at a slight angle from the horizontal
and facing upwards. Unfortunately, the x-radiography
equipment available to the author in Sri Lanka was not
powerful enough to penetrate the thicker metal parts and
mineralized core material, leaving the question open as
to whether the sculpture was fully hollow-cast, how thick
TECHNICAL STUDY AND CONSERVATION OF A LARGE-SCALE METAL BUDDHA STATUE
the walls are, the extent of casting porosity, and the nature
and number of armature and wax segments present, as
well as other internal structural features.
Composite materials like this present difficult and
complex conservation challenges, with the deterioration
of one material (iron armature) causing the deterioration
of others (bronze fabrication and inlays). Corresponding
exposed iron armatures are problematic and threaten
the physical integrity of the statue. The iron corrosion
product would ultimately disrupt the core and the casting
if not stabilized.
In terms of conservation, mechanical methods were
mainly applied during cleaning and 3% Na2EDTA only
used as an additional chemical method, where neces-
sary. Since immersion in BTA was not possible given
the adverse effects on the iron and core, as well as the
retention of the solution in the core, the 5% BTA solu-
tion in ethanol was brushed on for stabilization. Incralac
lacquer was used as a coating after the BTA treatment,
providing the metal with enduring protection from
atmospheric moisture.
Since there are no resources to control the environment
in the historic Anuradhapura museum building, it was
decided to seal the core and iron armature within the
statue by repairing the separation joint between the
head and the body and sealing the gap with an epoxy
filler and acrylic coating while the sculpture was in a
dry, controlled environment.
Acknowledgments
I would like to thank the US Ambassador’s Fund for
Cultural Preservation (AFCP 2015), Colombo Embassy,
for assistance with restoration and the forthcoming exhi-
bition that will include this statue. I would also like to
thank many of the members of staff at the Department of
Archaeology, Sri Lanka, including the conservation team:
M.A.D.R. Athukorala, Anuradha Fernando, Priyankara,
and the late Athula Kusumsiri, Dinushi, Jayasena and
Wasantha Liyanarachchi for assistance with handling the
statue. I am also grateful to Jerry Podany (retired Senior
Conservator of Antiquities, J. Paul Getty Museum), for
his generous guidance with restoring the statue and
for assistance in editing and interpreting the analytical
results; John Twilley, who was the conservation scientist
and provided valuable technical information on C-14
dating; and Palitha Herath, Arjuna Samaraweera, Ayesha
31
Fuentes, and Inoka Dissanayake for the photo credits
and photo editing. My thanks also to Sanjaya Perera, the
Science Officer and x-radiographer for the Sri Lankan
Atomic Energy Authority.
Notes
1 Kurunegala was one of Sri Lanka’s historic Divided
Kingdoms (AD 1293–1341).
2 As explained by John Twilley, “The routine outcome
of the radiocarbon testing charred mixed-source of core
material with the act of casting the sculpture will typically
be older than the true date and entirely dependent upon
the exclusion of soot derived from older organic matter.
Unless charcoal from plant matter can be physically
separated for testing, no amount of sample pretreat-
ment or cleaning can eliminate the influence of older
decayed plant matter resident in the clay. After charring,
the elemental carbon derived from contemporary plant
matter and earlier degraded plant matter will behave
the same under typical acid/base pretreatment. The
firing process converts all of those things that should
be removed during pretreatment into inert, elemental
carbon that will proceed through the sample cleanup
and contribute to the mixed result.”
References
Brown, R.L. 2012. Carrying Buddhism: The role of
metal icons in the spread and development of Buddhism.
Amsterdam: Royal Netherlands Academy of Arts and
Sciences.
Kasthuri, K.A. 2016. Preliminary investigation of Sri
Lankan copper-alloy statues. STAR: Science & Technology
of Archaeological Research 2(2): 159–76. DOI:10.1080/2
0548923.2016.1209055.
Listopad, J. 2003. Buddha images of the early
Anuradhapura period (c. 200 B.C.–432 A.D.). In Guardian
of the flame: Art of Sri Lanka, ed. J. Listopad. Phoenix:
Phoenix Art Museum.
The Mahavamsa (or The Great Chronicle of Ceylon). 1950.
English translation by W. Geiger, Colombo. The Ceylon
Government Information Department.
Masi, G., C. Chiavari, J. Avila, J. Esvan, S. Raffo, M.C.
Bignozzi, M.C. Asensio, L. Robbiola, and C. Martini.
2016. Corrosion investigation of fire-gilded bronze
involving high surface resolution spectroscopic imaging.
Applied Surface Science 366: 317–27.
32  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Podany, J., former Senior Conservator of Antiqui-

ties, J. Paul Getty Museum, personal communication,
November 2018.
Reedy, C. and S. Harlacher. 2007. Elemental composition
of Sri Lankan bronzes: Technological style and change.
In Scientific Research on the Sculptural Arts of Asia.
Proceedings of the Third Forbes Symposium at the Freer
Gallery of Art, eds. J.G. Douglas, P. Jett, and J. Winter.
Washington, DC: Archetype.
Twilley, J., Department of Materials Science and Chemical
Engineering, Stony Brook University, personal commu-
nication, November 2018.

Author
Kasthuri Arachchilage Anusha Kasthuri joined the
Central Cultural Fund as an Artifact Conservator in
1993. Since 2005, she has worked at the Department of
Archaeology of Sri Lanka as an Archaeological Research
Officer and Conservator.
Training a Database and Convolutional Neural Network
for Automated Recognition of Hallmarks on (Dutch) Silver
Ellen van Bork* Robert Erdmann
Rijksmuseum Rijksmuseum
Amsterdam, The Netherlands University of Amsterdam
e.van.bork@rijksmuseum.nl Amsterdam, The Netherlands
*Author for correspondence
r.erdmann@rijksmuseum.nl
Abstract
Currently, the available information regarding Dutch hallmarks
on silver is limited to printed books in which a short description
or drawing of the mark has been produced. Aside from these
minimalistic reproductions, archival information on guilds
and workshops has been lost. As a result, there is still much
uncertainty concerning hallmarks, and false attributions
are often made. For this research, a database will be built
containing all known and confirmed marks from surviving
insculpation plates and from objects with an unchallenged
provenance. The first phase of this project, reported here,
created a sample database and trained a deep ResNet-50
convolutional neural network (CNN) to perform classification
of images from the database to report the known identities of
those hallmarks. The network can be used to perform inference,
Introduction
From the 1880s onwards, many people have taken on
the challenge of reconstructing the complex system
of hallmarks on silver. As a result, numerous refer-
ence works have been published and databases built.
In the earlier works, idealised representations of the
marks are given, making the interpretation a matter
of connoisseurship, thus leaving a large margin for
error. In later publications, photos of marks, as well as
information on the objects on which they were found,
were added. Unfortunately, this information is still
rather scattered and no central database for (Dutch)
hallmarks exists. Over the last decades, scientific
research has become increasingly integrated into the
field of conservation and restoration, and our job as
conservators is no longer limited to safeguarding the
condition of objects. Our tasks have broadened: we
look at deterioration processes and critically assess
(traditional) conservation treatments, but we also
answer questions on authenticity and dating. Recent
developments in computer vision and artificial intelli-
Dirk Jan Biemond
Rijksmuseum
Amsterdam, The Netherlands
d.biemond@rijksmuseum.nl
during which hallmark images from new objects can be input
to the network, which outputs an estimate of the possible
identities of the hallmark along with estimates of the certainty
of those predictions. The proof-of-concept network was trained
on a database of 483 silver mark photographs of 82 distinct
types, resulting in an overall top-1 accuracy exceeding 99%.
This tool will aid in linking marks found on objects to known
marks, including their metadata, and could therefore help with
attribution and dating of silver works of art.
Keywords
hallmarks, silver, database, machine learning, automated
recognition, convolutional neural network, augmentation
gence have made it possible to make use of a database
of hallmarks on silver for automated identification,
creating a potentially invaluable tool to assist with
hallmark identification.
History of hallmarks
The basis of each investigation into silver objects lies
in the marks with which the master and the guild took
responsibility for the alloy used in the object. These guilds
were established in the 14th century, and from the middle
ages until the end of the 18th century, Dutch gold- and
silversmiths were obliged to become members of the
guild of their city or region (Kolderweij and Van Schaik
2012). To become a member, they were required to create
a masterpiece, giving proof of having an appropriate level
of skills. Once accepted to the guild, they would punch
their maker’s mark onto a so-called insculpation plate,
engraving their name and date of accession next to it.
An example of a 17th-century insculpation plate can be
seen in Figure 1.
34  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Figure 2. Syndics of the Amsterdam gold- and silversmiths’ guild, by

Thomas de Keyser, 1627, inv. no. 1960.11. The syndic on the proper left is
holding plyers with a cupel used for assaying. © The Toledo Museum of Art

Figure 1. Insculpation If the alloy proved to be of the right fineness, a syndic of

plate of the Delft assay the assay office (Figure 2) stamped the city mark onto
office, 1591–1511, inv. the silver, which at that point already held the personal
no. PDK 95-1. © Museum
maker’s mark of the gold- or silversmith. In case the
  Prinsenhof
assaying proved to be invalid, a letter was punched
to register which syndic had tested the material. The
The assaying and marking of silver
responsible syndic changed annually, taking a new letter
When a gold or silversmith had completed a piece, into use and thereby giving rise to the name year-letter
the guild’s assay office of which they were a member (Luijt 2001). Every guild worked with its own alphabet
would test the pureness of the silver. Silver is alloyed in a uniform font and, as the assaying was arranged per
with copper to increase the hardness, as pure silver is city, the letters would differ per guild. This changed in
soft and unsuitable for utensils. The material would be 1807, when the government approved a new law intro-
rubbed over a touchstone leaving a streak. The colour ducing a national year-letter system for all assay offices
(Kolderweij and Van Schaik 2012). Examples of the city
of this streak would be compared with known alloys
mark, year-letter and maker’s mark can be seen on the
provided by the government, giving an indication of
ewer by Adam van Vianen (Figures 3 and 4).
the alloy of the object. If more certainty on the exact
composition was needed, cupellation was used. A small
sample was taken with an engraving tool, leaving a
mark known as an assay groove—a zig-zag line most
commonly found on the bottom of an object. The
sample was weighed precisely, wrapped in an excess
of lead and placed in a cupel, which is a tray-shaped
porous refractory bowl made from bone ash (Figure 2)
(Agricola 1950). The cupel was subsequently placed
into a furnace of ca. 1100°C and, upon melting, all the
metals except the silver were absorbed by the porous
cupel. After cooling, the sphere of now pure silver was Figure 3. Ewer,
weighed again, revealing how much copper it contained gilded silver, Adam
van Vianen, 1614,
previously, and thereby revealing the fineness of the Rijksmuseum
silver (The Goldsmiths’ Company Assay Office n.d., collection, inv. no.
Van Laer 1967).   BK‑1976-75
TRAINING DATABASE AND CONVOLUTIONAL NEURAL NETWORK FOR AUTOMATED RECOGNITION OF HALLMARKS ON (DUTCH) SILVER
Figure 4. Examples of hallmarks commonly found on Dutch silver objects.
From left to right: a maker’s mark AV, year letter T and city mark Utrecht
as found on the ewer in Figure 3
From 1661 onwards, one can find a stamp of a Hollandse
Leeuw, a crowned lion, from the province of Holland to
indicate that the alloy was of the right fineness. In 1813,
after the founding of the kingdom of the Netherlands,
it was used nationally: the lion rampant for first grade
silver and a lion passant for second grade (Figure 5)
(Kolderweij and Van Schaik 2012). Despite this well-or-
ganised system, much uncertainty remains today as
many archives and insculpation plates no longer exist.1
Figure 5. Detail of a
drawing published
with the passing of
the new law in 1813
of a crowned lion
rampant assay mark
and lion passant
assay mark (Voet
  1963)
Overview of automated image recognition using
convolutional neural networks
The traditional way of identifying hallmarks is to look
for matches of the mark in the numerous publications
on this subject (Voet 1963, Gans 2014). However, the
documentation on marks is scattered, and many images
and information have not been published. Therefore,
we propose to create an open-access online database on
the basis of photographs of hallmarks. Such a database,
which can potentially include all Dutch hallmarks, can
be combined with automated image recognition to
become a very useful reference tool for the identification
of hallmarks.
Until recently, this would have been impracticable, but
the advent of deep convolutional neural networks (CNNs)
makes this feasible. Traditionally, high performance with
neural networks was only possible with huge training sets
containing millions of images. However, recent advances
with so-called transfer learning (Pan and Yang 2010)
have enabled a network trained on a different corpus to
35
be modified so that it will work well with a specialised
dataset such as the mark set herein proposed (Razavian
et al. 2014). Thus, for this work the construction and
training of a deep CNN was undertaken to identify hall-
marks from photographs. The CNN is provided with a
labelled training set consisting of images paired with their
associated class in the first training phase. For our current
purposes, the class is the identity of the mark. This is
especially challenging due to the numerous sources of
variation among photographs of silver marks. First, for
any given mark on an object, variation can arise from
differences in camera position, in lighting and, given that
silver is reflective, from differences in the environment.
Second, differences in the process of stamping from
object to object could result in any given mark being
applied with variable force or angle. Third, differences
in the wear of objects and punches could mean that even
marks that were originally indistinguishable look very
different. Any automated identification of these marks
must be considerably robust to all these variations.
The training phase of the network is complete when the
network is effectively as good as it can get at identifying
known marks in the training set. A portion of the pre-la-
belled training data is held out of the training process to
use as a validation set, so that a trained neural network’s
performance can be tested on images that it did not see
during training but for which the correct answers are
known, thereby ensuring that ‘overfitting’, in which the
network simply memorises its training images without
being able to generalise, is avoided.
The second, useful phase of the network is inference, in
which new images of silver marks can be input to the
network, which outputs best guesses for the identity of
the mark along with estimates of the certainty of those
predictions. Thus, the network does not simply output
a single guess; it outputs its confidence that the given
mark belongs to each of the different known marks on
which it was trained. If a given mark does not appear
to match any of the known classes, the network may
output an erroneous guess, since by construction the
mathematics of classification schemes with CNNs
require the total probability estimate across all classes
to sum to unity. This problem is universal to all comput-
er-based classification schemes. Thus, any predictions
of the network must be hand-checked, and the network
should always provide an array of guesses to the user
for a final check.
36  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Methodology
Preparation of training set
All computer investigations were carried out using the
Python programming language and the PyTorch neural
network framework (Paszke et al. 2017). The details of
each step of the process are given below.
The training of a CNN requires an annotated training set
of example images of each type of mark. The performance
of CNNs increases with the quantity and quality of the
training data. In particular, the network must be trained
to separate the relevant features of an image (the shape
of the mark) from the irrelevant aspects (lighting angle,
colour of surroundings, image size), so ideally it would
be trained with many images of the same mark in which
the irrelevant aspects were intentionally varied.
A database is currently being built containing all known
and confirmed marks from surviving insculpation plates
and from objects with an unchallenged provenance.
The present work seeks to demonstrate the feasibility
of using CNNs to assist in identification of hallmarks.
For this initial test phase, a database was used that was
built for a catalogue on the Rijksmuseum silver collec-
tion to be published in 2020. It consists of 483 images
of known marks drawn from 82 distinct classes. Marks
in the database fall broadly into five types: assay marks
(AM), city marks (CM), date letters (DL), maker’s marks
(MM) and tax marks (TM), and our class designations
start with these abbreviations. Images collected from
a HIROX KH-8700 3D microscope at magnifications
ranging from 20× to 60× were used. Each image was
then manually cropped to extract an approximately
square bounding box around each mark. A sample of
five high-quality examples of the 82 classes of mark is
shown in Figure 6.
Figure 6. Exemplars from 5 of the 82 classes used during the training of
the proof-of-concept neural network. From left to right: crowned lion
rampant (AM), Delft (CM), 1746 Dordrecht (DL), Dirk van de Goorbergh
(MM), Script J (TM)
Some of the classes, such as the crowned lion rampant
assay mark or the The Hague city mark, are very common
in the database, while others, such as certain year letters
or maker’s marks, have only a few single exemplars in the
TECHNICAL STUDIES
database. This imbalance in the frequency of the training
classes is a problem addressed below.
Network architecture
Once a training set is assembled, a CNN must be
constructed. Using transfer learning, we opted to modify
an existing neural network to adapt it to the given
problem. For the proof-of-concept network, a ResNet-50
architecture (He et al. 2016) pre-trained to win first-
place performance in the 2015 ILSVRC competition was
selected (Russakovsky et al. 2015). The network archi-
tecture and pre-trained weights are included in the free
open-source PyTorch package. This very deep network
contains 50 convolutional layers to detect patterns of
features in photographs and has a final fully connected
‘head’ layer to convert those detected patterns into prob-
ability estimates for each of the 1,000 classes present in
the original competition. The pre-training imbues the
convolutional layers with the ability to detect complex
patterns found in natural images, so the weights in those
layers are retained by freezing them during training since
they are presumed to be applicable to the patterns found
in hallmarks. The 1,000-output head layer is discarded
from the ResNet-50 architecture and replaced with a
fully connected layer with 82 outputs corresponding to
the 82 classes in the database.
Regularisation: Data augmentation
A typical classification CNN uses thousands of training
images to obtain high performance, but at this stage
of the project there are only 483 images. To avoid the
problem of ‘overfitting’, in which the network simply
memorises each of the training images without learning
to extract the features relevant for mark identification,
several strategies are adopted. First, the network itself
is trained using weight decay (gradual forgetting) and
dropout (Srivatsa et al. 2014), in which random neurons
are disabled during each step of training (p=0.5).
Second, we aggressively use a technique known as
‘augmentation’, in which each training image is dupli-
cated and artificially modified in ways that change its
pixel values but which retain its class identity. Examples
include coordinate transformations (rotation, scale
changes, random crops and perspective tilts) as well
as colour changes (e.g. brightness, contrast and satu-
ration). We note that the appearance of silver marks is
such that if the position of the light source is changed
TRAINING DATABASE AND CONVOLUTIONAL NEURAL NETWORK FOR AUTOMATED RECOGNITION OF HALLMARKS ON (DUTCH) SILVER
dramatically, previously light areas of the image may
become dark and vice versa. To increase the robustness
of the network to this kind of variation, the images
are also augmented by training on the original as well
as on a photonegative of each one since negation has
the effect of swapping light and dark areas. Figure 7
shows sample augmentations generated from a single
input image.
To tackle the problem of having a severely unbalanced
training set in which some marks appear one or two
orders of magnitude more frequently than others,
we follow the recommendations of Buda, Maki and
Mazurowski (2018), who found that, unlike in other
machine learning settings, systematic oversampling
in CNNs did not result in overfitting. Thus, each class
is oversampled to be perfectly balanced, yielding a
training set of 1,000 augmented images of each of the
training classes.
Training
To further regularise the network, a technique called
mixup (Zhang et al. 2017) was used in which the
network was trained with linear mixtures of images.
The modified ResNet-50 network was trained using
stochastic gradient descent with restarts (SGDR)
(Loshchilov and Hutter 2017) using cyclical learning
rates (Leslie 2017) and the standard cross-entropy
loss to obtain super-convergence during training. The
training was performed for three epochs on an NVIDIA
GeForce GTX 1080 Ti graphics processing unit with
3,584 compute cores and 11 GB of memory. Each epoch
consisted of 82,000 augmented images, each downsized
to 299 pixels width. A batch size of 16 images, weight
decay of 0.01, and a learning rate that ramped up from
37
Figure 7. Augmented images
generated from a single input
image. Augmentations randomly
vary the geometry, camera pose
and lighting. Training images are
inverted with 50% probability to
approximate changes in lighting
  position
0.001 to 0.01 and back down during each epoch was
used. Each epoch required approximately 9 minutes
to complete.
Validation
As mentioned above, the input images are augmented
and oversampled to produce a large (85,000 image)
perfectly balanced dataset. Of these images, 68,000
(80%) are used for training, while 17,000 (20%) are
held out from training and used for validation. While
balancing the validation set is non-standard practice,
this approach was chosen for two reasons: first, an
accuracy metric measured on a true sample from the
training data would be misleading; if 25% of the marks
are lion rampant assay marks, a random guess that every
image was of this class would automatically result in
25% accuracy; second, the value of the network as a
tool is primarily as an aid to experts, who will easily
identify the most common marks. In other words, the
prevalence of the examples in the training set does not
match the expected frequency of inputs to the network
in daily use. Additionally, the common approach of
test-time augmentation (TTA) is used, in which an
image is augmented multiple times at inference time
and the predictions averaged in order to increase the
robustness of the network to chance irregularities in the
input image, thereby boosting the accuracy.
Results
After three epochs of training, the network obtained
an accuracy of 99.84% on the validation set. Typical
examples showing known marks from the validation set
and an unknown mark are shown in Figure 8 a–b. This
demonstrates that the network is robust to large changes
38  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
in lighting, to severe wear patterns, to orientation differ-
ences and to brightness patterns in the background.
The top-5 guesses are shown for the input image; note
how the network is able to pick marks having similar
overall contours or inner features. Figure 8a shows a
classification in which the network has high confidence,
while Figure 8b shows a classification with low confi-
dence since there are several candidates with similar
probabilities. The exemplars of top candidates shown
there demonstrate that they share several geometric
features with the input image such as letter shapes or
repetitive features.
 
Figure 8. Visualisation of (a) a successful classification during validation,
in which a badly damaged image of an Engelbart Joosten maker’s mark
is input to the network, and (b) results for an image for which the mark is
not in the training set. The displays show the five most likely candidate
classes with an exemplar from each, along with blue bars on the right
of the image to indicate the probability of class membership
Future work
While the feasibility of using CNNs to assist with silver
mark identification has been demonstrated, much work
remains to be done. In the future, the small training
database will be further extended with marks from the
Rijksmuseum and from as many reliable sources as we
TECHNICAL STUDIES
can access. In addition, we intend to build a generative
model which will (1) estimate the 3D profile of a mark
from its photograph and then (2) simulate photography
of that geometry with moving light positions and
camera pose and under different degrees of wear or
damage. This would dramatically increase the robust-
ness of the network to these changes and would also
enable it to input pencil rubbings or other renderings
of images such as the idealisations found in books of
silver marks.
Online portal
In order to maximise the utility of this tool, an online
portal hosted and maintained by the Rijksmuseum will
be designed where users, including other museums or
collectors, can upload their own images and be shown
the network’s predictions of the mark’s classification or
possible matches to other marks. The user-input images
can later be used as additional training data in cases
where the user provides confirmation of the correct
classification and provenance of the object, thereby
leading to a tool with growing utility. Additionally, the
source code of the network and portal will be provided
as open-source software so that it can be validated,
adapted to other uses and improved by others. The
design of the online portal and related considerations
of its costs, maintenance and security will be the subject
of a future paper.
Conclusion
It has been successfully demonstrated that transfer
learning can be used with a modern convolutional neural
network to make a high-reliability automatic classifier
for silver marks despite dramatic variation in lighting,
camera pose, mark wear and mark completeness. The
network is based on completely free open-source soft-
ware, can be run on any consumer PC and can run at
dramatically accelerated speed using a consumer-grade
graphics processing unit costing less than €500. It is
anticipated that with more training data, the tool will
grow to be invaluable, especially at recalling obscure
marks so that an expert can make a final assessment.
The options for further extension of this database are
boundless, and it is the overarching aim of this project
to establish a centre of excellence at the Rijksmuseum in
the study of (Dutch) silver.
TRAINING DATABASE AND CONVOLUTIONAL NEURAL NETWORK FOR AUTOMATED RECOGNITION OF HALLMARKS ON (DUTCH) SILVER
Acknowledgements
The authors would like to extend sincere thanks to Robert
van Langh for his leading role in the discussions that
inspired this project and for his tireless fundraising which
provided the necessary financial support. Joosje van
Bennekom, Sara Creange, Tamar Davidowitz, Benjamin
Rous and Arie Pappot are thanked for their thoughts and
comments on this research project and article.
Notes
1 The marking system is more complex than stated
above as each province, region or city had their own
rules and laws. The given information was generalised
and simplified for this article, ensuring, however, that
the provided information is correct and relevant.
References
Agricola, G. 1156, trans. 1950. De re metallica, trans. H.C.
Hoover and L.H. Hoover. New York: Dover Publications.
Buda, M., A. Maki, and M.A. Mazurowski. 2018. A
systematic study of the class imbalance problem in convo-
lutional neural networks. In Neural Networks 106: 249–59.
Gans, L.B. 2014. Goud- en zilvermerken van Voet.
Amsterdam: Premsela en Hamburger.
He, K., X. Zhang, S. Ren, and J. Sun. 2016. Deep residual
learning for image recognition. In Proceedings of the 29th
IEEE Conference on Computer Vision and Pattern Recog-
nition – CVPR 2016, Las Vegas, Nevada, 26 June–1 July
2016, 770–78. The Institute of Electrical and Electronics
Engineers, Inc.
Kolderweij, A. and A.M. van Schaik. 2012. Een keur aan
zilver. Amersfoort: Bekking en Blitz Uitgevers B.V.
Laer, W. van. (1730) 1967. Weg-wyzer voor aankoomend
goud- en zilversmeden, pref. B. Dubbe. Lochem: De
Tijdstroom.
Leslie, N. 2017. Cyclical learning rates for training
neural networks. In 2017 IEEE Winter Conference on
Applications of Computer Vision (WACV), Santa Rosa,
CA, USA, 24-31 March 2017, 464–72. Piscataway, NJ:
Institute of Electrical and Electronics Engineers. https://
doi.org/10.1109/WACV.2017.58.
Loshchilov, I. and F. Hutter. 2017. SGDR: Stochastic gradient
descent with warm restarts. Paper presented at ICLR 2018:
6th International Conference on Learning Representations,
Vancouver, BC, Canada, April 30–May 3, 2018.
39
Luijt, J.J. 2001. De jaarletters van het Koninkrijk Holland.
In Zilverstukken. Tien opstellen over zilver en zilvermerken.
Utrecht.
Pan, S. and Q. Yang. 2010. A survey on transfer learning.
In IEEE Transactions on Knowledge and Data Engi-
neering 22(10): 1345–59. https://doi.org/10.1109/
TKDE.2009.191.
Paszke, A., S. Gross, S. Chintala, G. Chanan, E. Yang,
and Z. De Vito, Z. Lin, A. Desmaison, L. Antiga, and A.
Lerer. 2017. Automatic differentiation in PyTorch. Paper
presented at the 31st Conference on NIPS 2017 Workshop
Autodiff, Long Beach, CA, 9 December 2017.
Razavian, A., H. Azizpour, J. Sullivan, and S. Carlsson.
2014. CNN features off-the-shelf: An astounding base-
line for recognition. In Proceedings of the 2014 IEEE
Conference on Computer Vision and Pattern Recognition
Workshops (CVPRW 2014), Columbus, OH, 23–28 June
2014, 512–19. Piscataway, NJ: The Institute of Electrical
and Electronics Engineers, Inc. https://doi.org/10.1109/
CVPRW.2014.131.
Russakovsky, O., J. Deng, H. Su, J. Krause, S. Satheesh,
S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein,
A.C. Berg, and L. Fei-Fei. 2015. ImageNet large scale
visual recognition challenge. In International Journal of
Computer Vision 115(3): 211–52. https://doi.org/10.1007/
s11263-015-0816-y.
Srivastava, N., G. Hinton, A. Krizhevsky, I. Sutskever,
and R. Salakhutdinov. 2014. Dropout: A simple way to
prevent neural networks from overfitting. In Journal of
Machine Learning Research 15: 1929–58.
The Goldsmiths’ Company Assay Office. n.d. The
hallmarking process. www.assayofficelondon.co.uk/
media/1426/hallmarking-process.pdf.
Voet, E. 1963. Nederlandse goud- en zilvermerken. The
Hague: Martinus Nijhoff.
Zhang, H., M. Cisse, Y.N. Dauphin, and D. Lopez-Paz.
2017. Mixup: Beyond empirical risk minimization. In
ICLR 2018: Sixth International Conference on Learning
Representations Preprints, Vancouver, Canada, 30 Apr–3
May 2018.
Authors
Ellen van Bork is a metals conservator at the Rijksmu-
seum.
40  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Robert Erdmann is a senior scientist at the Rijksmu-

seum and professor in the Department of Conservation
and Restoration at the Institute of Physics, University of
Amsterdam.

Dirk Jan Biemond is a metals curator at the Rijksmu-

seum and an acknowledged specialist in the field of
Dutch silver.
 Young Conservation Professional

Metal Threads in 17th-Century Western European

Textiles: Characterization of Metal Threads from the
Burgzand North 17 Shipwreck
Sandra Savelli* Ellen van Bork Suzan Meijer Ineke Joosten
University of Amsterdam Rijksmuseum Rijksmuseum The Cultural Heritage Agency of the
Amsterdam, The Netherlands Amsterdam, The Netherlands Amsterdam, The Netherlands Netherlands
sandrasavelli@gmail.com e.van.bork@rijksmuseum.nl S.Meijer@rijksmuseum.nl National Heritage Laboratory
*Author for correspondence
Amsterdam, The Netherlands
I.Joosten@cultureelerfgoed.nl

Abstract
The embroidered metalwork on a 17th-century toilet service
found on the Burgzand North 17 shipwreck was analysed by
microscopic inspection and scanning electron microscopy/
energy-dispersive x-ray spectroscopy to identify the alloys
and manufacturing methods. The research concluded that the
decorations, which in current condition are fully mineralised,
were mainly composed of gilt silver threads, present in both
a round and a rectangular cross section, cut gilt silver strips
wound around an organic fibrous core and spangles exhibiting
a copper core.
Keywords
marine silver, metal threads, archaeological metal

Introduction
In 2014, a 17th-century shipwreck, known as the Burg-
zand North 17 (BZN 17) or the Palmhout (boxwood)
shipwreck, was discovered close to the Dutch island of
Texel. Among the finds brought ashore by local divers
was a luxurious toilet service, consisting of a clutch
purse (Figure 1a), mirror (Figure 1d), toiletry brush
(Figure 1c), loose textile fragment (possibly once part of
a pin cushion) (Figure 1b) and a rectangular tablecloth.
All five objects show the use of red silk velvet decorated
with elaborate metal embroideries.
the mirror were suitable for elemental analysis as they
had not undergone any conservation treatment aside
from rinsing with tap water. Thus, the toilet service was
subjected to research at the University of Amsterdam.
Materials and methodology
Examination
The toilet service was analysed by visual inspection
and detailed examination with a stereomicroscope and
Hirox digital microscope to determine the shape of
the metal elements. Object mapping was carried out to
document the arrangement. Hirox measurements of the
metalwork were gathered from all five objects, chiefly on
areas where cross sections of the metalwork had been
exposed (Figure 2).

Figure 1. Four objects from the toilet service: (a) a clutch purse, (b) a
loose textile fragment, (c) a brush and (d) a mirror. Photographs a, b
and c by Margareta Svensson Figure 2.
Photomicrograph
A mineralised concretion layer covering the metalwork of split round wires
from the mirror
complicated the work of determining the shape and (118 and 124 μm in
materials of its elements. However, all objects except   diameter)
42  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Samples indicates the use of a different type of metalwork. The

construction of the metalwork is exemplified by graphic
This paper discusses the elemental analysis of four samples
renderings in Figure 4.
from decorations on the purse (S1, round wire), tablecloth
(S2, spangle, and S4, flat metal strip) and the loose textile
fragment (S3, rectangular wire). Three samples (S1, S2
and S3) were cold-mounted for cross-sectional analysis.
Sample S4 was subjected to surface analysis.

Analysis
All samples were examined by scanning electron micros-
copy with energy-dispersive x-ray spectroscopy (SEM-
EDX). The primary aim of the elemental analysis was to
establish which materials were used for manufacturing
different types of metalwork. The equipment used was
a Nova NanoSEM 450 from FEI and an UltraDry silicon
drift detector with Pathfinder software from Thermo
Scientific. The analysis was performed in a high and
low vacuum (50–90 pa) and an accelerating voltage of
20 keV. Backscattered electron (BSE) images were taken
with the gaseous analytical detector (GAD).

Results
Figure 3. Pattern mapping of the loose textile fragment. Each colour
Metalwork types and arrangement indicates the use of a different type of metalwork

The decorations of the five research objects follow the Dimensions

same logic both stylistically and in the materials used,
except for the tablecloth, which has lost the majority of
its decorative pattern. Figures 3 and 4 demonstrate the
placement of round and rectangular wires, metal strips
and flat disks on the loose textile fragment. Each colour
The wires, strips and spangles are of similar dimensions
on all five objects, with the exception of the tablecloth,
where no data was collected for the round and the rectan-
gular wires. The calculated average for the dimensions

Figure 4. Types of metalwork found on the BZN 17 research objects. Graphic renderings by Toms Lucans
METAL THREADS IN 17TH-CENTURY WESTERN EUROPEAN TEXTILES: CHARACTERIZATION OF METAL THREADS FROM THE BURGZAND NORTH 17 SHIPWRECK
is as follows: 118 μm in diameter for the round wires,
166 μm in width for the rectangular wires, 248 μm in
width for the flat metal strips and 2600 μm in diameter
for the spangles. However, the dense mineralised crust
complicated the taking of precise measurements as it
was difficult to establish to what extent the wires had
been distorted by the corrosive processes. Therefore,
these measurements cannot serve to define the exact
dimensions of the original metalwork.
Alloys and gilding
The cross-sectional analysis showed that several metal-
work types consisted of gilt silver. Figure 5 shows a
magnified BSE image of the cross section of a rectangular
wire (S3) where the outline of the original silver wire is
clearly distinguishable. Remains of the original gilding
layer were also detected in the bulky deposition layer. The
Figure 5. BSE image of a cross section of rectangular wire (S3): (1)
Gold-rich phase, marking the remains of the original gilding layer; (2)
mineralised silver-rich phase, marking the original shape of the wire; (3)
silver-rich mineralised crust
Figure 6. BSE image of the flat metal strip (S4): (1) EDX data showing
the presence of gold on the surface, referring to gilt material; (2) no
gold was detected
43
round wire (S1) also demonstrated the same phenom-
enon, showing silver as the main component and gold
present at the edges of the cross section. For the metal
strips, gold was detected on the top surface, indicating
that the material is gilt (Figure 6, point 1). No gold was
found on the short edges (Figure 6, point 2), suggesting
that the strips are cut from a gilded silver sheet.
The flat disk-shaped spangles demonstrated visual and
compositional differences when compared to other
metalwork in the toilet service. During initial optical
observation, a colour difference between corrosion prod-
ucts on the spangles of the purse (Figure 7) and the brush,
in comparison with the rest of the metalwork, raised
questions about possible alloy variances. Supporting this
hypothesis, the elemental analysis of the sampled spangle
of the tablecloth (S2) revealed significantly high amounts
of copper (Figure 8), whereas all other sampled metal-
work (S1, S3 and S4) demonstrated a silver-rich core.
Figure 7. Photomicrograph illustrating the colour difference between
spangles and the rest of the metalwork on the clutch purse
Figure 8. BSE image of a cross section of the spangle (S2): (1) Copper-
rich core; (2) silver-rich phase; (3) gold detected on the edges of the
cross section
44  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Discussion
As described in the literature (Biringuccio 1540, trans.
1990), it is likely that for manufacturing the round wires
of the BZN 17 toiletry service, cast ingots were hammered
into rods, gilded by heat welding and then drawn through
a series of die holes to reduce the material to the desired
dimensions. The rectangular wires are likely to have been
made by flattening round wires. As seen from the BSE
image of the sample S3 cross section (Figure 5), the short
sides of the wire show a rounded outline resulting from
compression. Flattening of these wires, common in the
17th century, was most likely carried out by rolling (Járó
and Tóth 1991).
The examination of the metal strips suggested a signif-
icantly different manufacturing method than that for
the rectangular wires. Firstly, the edges of the strips are
burred and the width varies significantly. Secondly, the
surface of the metal strips appears relatively smooth,
exhibiting small irregular dents and cuts that indicate
rolling or hammering. Lastly, no marks from a draw-
plate were found on the surface. This shows that these
elements were made from cut thin metal sheets instead
of flattened wire.
It is surprising that metalwork manufactured from gilt
wire both with a silver and a copper core was found on
the same object. Biringuccio (1540, trans. 1990) refers
to gilt copper wires as a ‘great fraud’ already in the 16th
century. Gilded or silvered copper threads were used
only in embroideries commissioned by less wealthy
people (Járó and Tóth 1991) or manufactured for theatre
costumes (Glover 1979), and are therefore not expected to
be the material of choice for adorning luxurious artefacts
such as the toilet service discussed in this paper.
Acknowledgements
The authors would like to thank Huis van Hilde, the
Rijksmuseum Amsterdam, Dr Tonny Beentjes (UvA)
and Dr Rene Peschar (UvA).
References
Biringuccio, V. 1540 (trans. 1990). The pirotechnia of
Vannoccio Biringuccio. The classic sixteenth-century trea-
tise on metals and metallurgy, trans. and eds. M. Gnudi
and C.S. Smith. New York: Dover Publications.
Glover, E. 1979. The gold & silver wyre-drawers. London
and Chichester: Phillimore & Company.
Járó, M. and A. Tóth. 1991. Scientific identification of
European metal thread manufacturing techniques of the
17–19th centuries. Endeavour 15(4): 175–84.
Authors
Sandra Savelli has an MA in Conservation and Resto-
ration of Cultural Heritage from the University of
Amsterdam. Her thesis was on 17th-century metal
threads.
Ellen van Bork is a metals conservator at the Rijksmu-
seum Amsterdam and lecturer at the University of
Amsterdam.
Suzan Meijer is head of textile conservation at the
Rijksmuseum Amsterdam.
Ineke Joosten is a conservation scientist at the Cultural
Heritage Agency of the Netherlands.

Conclusion
Sample analysis of the metalwork of the BZN 17 toilet
service allowed the original major alloy constituents
of the metal elements to be detected. Since the metal-
work was covered by a crust of corrosion products, this
information would have remained unknown if only
surface analysis techniques had been used. The research
concluded that the different types of metalwork, found to
consist mainly of gilt silver, fitted into the 17th-century
context.

The Production of Medieval Gilt Silver Threads
María Fernanda Gómez Núñez* Ingrid Karina Jiménez Cosme
National School of Conservation, Restoration and IG Restoration
Museology “Manuel del Castillo Negrete” (ENCRyM) Mexico City, Mexico
Mexico City, Mexico ig.restauracion.contacto@gmail.com
fernanda_gomez_n@encrym.edu.mx
*Author for correspondence
Abstract
This is a study of gilt silver threads based on the treatises
Schedula diversarum artium by Theophilus Presbyter and De
la pirotechnia by Vannoccio Biringuccio. The research used
the anthropology of technology, applying the concepts of
Lemonnier (1992) to create a methodology to help reproduce
the manufacturing process described in the treatises. Four
samples of metal threads were made and analyzed by scanning
electron microscopy with energy-dispersive x-ray spectroscopy
at the National Laboratory of Sciences for the Research and
Introduction
An understanding of manufacturing processes is crucial
for creating a conservation proposal. A conservator needs
to know the material and the “what” and “why” of what
she/he is conserving. This research into gilt silver threads
consists in reproducing the manufacturing techniques
in the 12th-century treatise Schedula diversarum artium
by Theophilus (trans. 2013) and De la pirotechnia by
Biringuccio (trans. 1990), published in 1540. This is the
first step in an ongoing research project whose objective
is to observe the characteristics of each process. Four
samples are presented in this paper made and analyzed
by a multidisciplinary team.
Manufacturing of gilt silver threads
The techniques for making gilt silver threads in the medi-
eval treatises by Theophilus and Biringuccio were studied.
The process consisted in extracting and refining silver
and gold usually with lead or mercury, followed by the
gilding of the silver by soldering or diffusion. Theophilus
(trans. 2013, 156) describes soldering, which requires a
filler material of gold or silver fused with copper, while
Biringuccio (trans. 1990, 381–82) refers to diffusion, the
term used to describe the joining of metals through heat
Young Conservation Professional
José Luis Ruvalcaba Sil
National Laboratory of Sciences for the Research
and Conservation of Cultural Heritage (LANCIC-UNAM)
Mexico City, Mexico
sil@fisica.unam.mx
Conservation of Cultural Heritage (LANCIC-UNAM) in Mexico
City. The analysis allowed the characteristic markings made by
each tool used during manufacturing to be observed. This study
is a first approach in an ongoing investigation.
Keywords
metal threads, gilt silver, Vannoccio Biringuccio, Theophilus
Presbyter
and pressure. After obtaining silver plate gilded only on
one side or both, it is transformed into flat threads by
hammering and cutting. These threads were then later
spun into silk or linen thread. If the artisan wanted to
make long, thin wires, only gold or silver was drawn
through a drawplate (Biringuccio, trans. 1990, 377–78).
There are other goldsmithing processes described in the
treatises that are now considered thread manufacturing
techniques. However, neither Biringuccio nor Theophilus
make a visible connection between these methods and
metal threads. For example, amalgamation gilding is
described by Theophilus (trans. 2013, 113–14) as a
method used for applying gold to the handles of a chalice,
but Biringuccio (trans. 1990, 367) does not mention its
use in making metal threads. The use of parchment and
red pigment (probably bole) for hammering is mentioned
also by Theophilus (trans. 2013, 29) as a method for
obtaining gold leaf.
The concepts of Lemonnier (1990) were used to compre-
hend better the manufacturing of gilt silver threads. He
suggested that production techniques have five compo-
nents that interact with each other to form an object:
46  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

matter; energy; objects, artifacts, or tools; gestures; and

specific knowledge.

Samples
From the treatises and the concepts of Lemonnier (1990),
it was possible to unravel the manufacturing of metal
threads and to understand the use of materials and tools.
As part of the research, and with the help of a goldsmith,
four samples were made:

• Following the descriptions of Theophilus (trans. 2013):

–– MSM: gilt silver soldered on one surface. The sample
was hammered using red clay bole and parchment, Figure 1. MSM, textured surface, remains of clay bole and gold flake

and cut into narrow strips.

–– MSH: gilt silver soldered on one side and drawn
through a drawplate.

• Following the procedures of Biringuccio (trans. 1990):

–– MDM: gilt silver diffused on one surface by heat
and pressure only and hammered and cut into
narrow strips.
–– MDH: gilt silver diffused on one side and drawn
through a drawplate.

Results and discussion

The samples were analyzed using a ZEISS Stemi 508 Figure 2. MDH, gold surface
microscope and a HITACHI TM3030Plus scanning
electron microscopy with energy-dispersive x-ray spec-
troscopy (SEM-EDX) at the National Laboratory of
Sciences for the Research and Conservation of Cultural
Heritage (LANCIC-UNAM). Different results were
obtained depending on the type of gilded process used.
In the soldered samples (MSM and MSH), the resulting
union of Au and Ag was not strong enough to resist the
force and pressure exerted when hammered or drawn,
therefore causing cracks, gold flakes (Figure 1), and
an irregular surface with traces of silver; these areas
are prone to corrosion due to differential aeration. On
the other hand, the samples gilded by diffusion (MDM
and MDH) presented a more homogeneous and evenly
distributed gold surface. Furthermore, the gold and silver Figure 3. MSM, hammered surface
were better fused with only minimal visible detachment
and flaking (Figure 2). grains (Figure 3). The scissors used to cut the strips also
left uneven edges (Figure 4).
The tools used in shaping metal leave patterns on the
surface that can help to identify the techniques used. On Theophilus (2013) describes a method for making gold
the hammered samples (MSM and MDM), the tool left an leaf by hammering with parchment and red pigment. The
uneven surface with irregular marks and flattened metal MSM sample made following this method presented a
THE PRODUCTION OF MEDIEVAL GILT SILVER THREADS
Figure 4. MSM, uneven edges
rough surface due to the hammering and the texture of
the parchment (Figure 3). In this case, goat parchment
was used with a different surface finish to that used by
Theophilus. Byzantine or calf vellum would probably
give the metal a different print. The size of the sample
obtained with this process was 6 cm long, relatively small
for a thread.
As far as the drawn samples were concerned (MSH
and MDH), it was possible to obtain long threads/
wire. Analysis by SEM showed the surface consisted of
straight and parallel thin lines running along the thread,
which were attributed to passing the metal through the
drawplate (Figure 5). Other lines were also seen that
were thicker and shorter and attributed to the polishing
tool used during diffusion (Figure 5). The tweezers
used to pull the metal through the drawplate also left
a mark (Figure 6).
Figure 5. MDH, drawplate lines and polishing tool marks
47
Figure 6. MSH, tweezer marks
Some of the materials involved in the different produc-
tion stages can be detected on the finished product
by analysis. In some cases, the researcher may find
elements, such as C, Hg, and Pb, that may be associ-
ated with the solder or alloy used, the gilding or metal
extraction method, or metal refining, respectively. In
the samples, gold and silver were found as the main
components. The sodium detected by SEM-EDX was
associated with the flux used to lower the melting point.
In samples with soldered gilding, small traces of copper
were found due to the filler material (Ag-Cu). In the
sample with hammered red pigment as described by
Theophilus, earth elements (Fe, Mg, Si, and Al) were
found due to its composition.
The characteristics present in the experimental samples
can be found in historical metal threads, such as silver
gilded on only one side, problems in the fusing of silver
and gold (Figure 7), and a homogeneous gilt surface
Figure 7. Metal thread on a 17th-century chasuble
48  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

with uneven borders due to cuts (Figure 8) (Jimenez Jimenez, I. 2015. Produción orfebre en vestimenta litur-
2105). gica virreinal, caracterización de técnicas y materiales de
los textiles de la Catedral de México en el MNV. M.A.
dissertation, Escuela Nacional de Antropología e Historia
(ENAH), Mexico.
Lemonnier, P. 1992. Elements for an anthropology of
technology. Michigan: University of Michigan.
Theophilus (Presbyter). trans. 2013. On divers arts. The
foremost medieval treatise on painting, glassmaking and
metalworking, trans. and eds. J.G. Hawthorne and C.S.
Smith. New York: Dover Publications Inc.

Authors
María Fernanda Gómez Núñez is a recent graduate
Figure 8. Gilt silver strip on a 17th-century chasuble from the National School of Conservation, Restoration
and Museology (ENCRyM) in Mexico City.
Conclusion
Ingrid Karina Jiménez Cosme graduated from the
Studying the treatises on the manufacturing of metal National School of Conservation, Restoration and Muse-
threads involves analyzing sources and comparing infor- ology (ENCRyM) in Mexico City and manages the private
mation. Experiments can test hypotheses regarding the conservation practice IG Restoration.
knowledge and manual skills of the craftsmen involved
in the production of an object. José Luis Ruvalcaba Sil is head professor and research
scientist at LANCIC-UNAM in Mexico City.
The tool marks observed in the experimental samples
are similar to those seen on metal threads found in
historical textiles in Mexico, allowing an insight into the
manufacturing technique. Further research, experiments,
and comparison with case studies will follow.

Acknowledgments
The authors would like to thank ENCRyM for accepting
and supporting this research. We are particularly grateful
to the Seminar/Workshop in Conservation of Metallic
Heritage (STM-ENCRyM) for providing academic
guidance and material resources, as well as allowing
us to use its facilities. This research was supported by
grants from CONCACYT LANCIC LN 293904 and CB
239609, as well as PAPIIT UNAM IN112018. Special
thanks to Alejandro Mitrani at LANCIC-UNAM for his
assistance and knowledge in operating the SEM-EDX,
and to Adriana Vega for her expertise in goldsmithing.

References
Biringuccio, V. trans. 1990. The pirotechnia of Vannoccio
Biringuccio: The classic sixteenth-century treatise on metals
and metallurgy, trans. and eds. C.S. Smith and M.T.
Gnudi. New York: Dover Publications Inc.
Iron Piano Wire Production at the Beginning of the
19th Century

M.G. Durier* Y. Cotrebil J.-C. Battault I. Guillot

Laboratoire de Recherche et de
Restauration
Musée de la musique, affiliation
CNRS USR 3224
Paris, France
mgdurier@cite-musique.fr
L. Villiaume
Laboratoire Identités, Cultures,
Territoire
Université Paris 7 Diderot
Paris, France
leslie.villiaume@laposte.net
R. Pires
Institut de Chimie et des Matériaux
Paris-Est
CNRS, Université Paris Est
Thiais, France
pires@icmpe.cnrs.fr
*Author for correspondence
Institut de Chimie et des Matériaux Laboratoire de Recherche et de Institut de Chimie et des Matériaux
Paris-Est Restauration Paris-Est
CNRS, Université Paris Est Musée de la musique CNRS, Université Paris Est
Thiais, France Paris, France Thiais, France
cotrebil@glvt-cnrs.fr jcbattault@cite-musique.fr guillot@glvt-cnrs.fr
E. Vega M. Vion P. Dillmann
Laboratoire archéomatériaux et Former keyboard instrument maker Laboratoire archéomatériaux et
prévision de l’altération, LAPA-IRAMAT and restorer prévision de l’altération, LAPA-IRAMAT
NIMBE, CEA, CNRS, Université Paris- Uchizy, France NIMBE, CEA, CNRS, Université
Saclay matthieu.pierre.vion@gmail.com Paris-Saclay
Gif-sur-Yvette, France Gif-sur-Yvette, France
enrique.vega@cea.fr C. Clarke philippe.dillmann@cea.fr
Early keyboard instrument maker
A. Houssay and restorer S. Vaiedelich
Laboratoire de Recherche et de Donzy le National, La Vineuse sur Laboratoire de Recherche et de
Restauration Frégande, France Restauration
Musée de la musique christopherclarke@sfr.fr Musée de la musique
Paris, France Paris, France
houssay.am@gmail.com svaiedelich@cite-musique.fr

Abstract
Between 1780 and 1855, pianoforte makers used strings with
higher and higher tensile strength, in increasingly powerful
instruments played in larger concert halls, in order to satisfy the
musical and cultural demand of European society. While changes
to the piano are well known, there have been few analytical
studies on historical iron and steel piano wires, even though
innovative metallurgical processes occurred in the same period.
Understanding the materials constituting piano wire helps
establish the “history of piano wire.” Here, a multidisciplinary
study was conducted in close relationship between specialists
in material science, early keyboard instrument makers, and
historians of technology. This framework enabled the gathering
of historical written sources and the acquisition of physico-
chemical and metallurgical analyses on an extensive corpus
of early European piano strings from museum and private
collections. Based on these analyses, three categories of strings
were identified: phosphorus iron strings, steel strings with
globular carbides, and harder steel strings with higher carbon
content. It can be seen that most of the collected historical piano
wires are divided into the two first categories following a likely
technological breakthrough around 1830. This period of change
seems to coincide with the beginning of the use of the steam
machine in wire mills (Horsfall 1971, 114). Prior to this date, the
wires show a high phosphorus content. After 1830, the presence
of globular carbides corresponding to a low carbon content and
an absence of phosphorus is observed. Our research aimed to
meet the needs of museums and keyboard instrument makers
and restorers by providing guidance on the specification of the
most suitable wires for restoration and reconstruction of early
pianofortes (facsimiles).
Keywords
piano strings, historically informed music, Industrial
Revolution, metallurgy

Introduction
The pianoforte, invented ca. 1700, gradually supplanted
the harpsichord in Europe from the end of the 18th
century onwards. The earliest pianos had a delicate,
silvery sound and a light architecture associated with
low string tension. In the early 19th century, however, in
a span of less than 50 years, pianos acquired a powerful
sound and a heavier structure. With the industrial revo-
lution, a succession of innovations in string metallurgy
significantly modified production (Houssay 2004). Piano
makers upgraded the instrument design to include a
cast-iron frame capable of withstanding high tensions
and the repeated strokes applied to the strings (Haine
2006). Paris, like London, had become a main center for
piano production with numerous makers, including the
50  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Érard, Pleyel, and Pape workshops, that would become
references during the century.
Driven by technical advances, the production of piano
strings changed radically in Europe in the early 19th
century. As the strings are a major component in the
production of sound, changing them also transformed the
way pianos were manufactured and gave them a different
sonority. This is evidenced, for example, by comparing the
power of the 1791 piano by the brothers Johann Gottfried
and Johann Wilhelm Gräbner (E. 2002.7.1) with that of
the 1842 Pleyel piano (E.991.16.1) known to have been
played by Frédéric Chopin. Since the 1970s, analyses of
early strings (Piat et al. 1982, 88) and research on harp-
sichord string production processes have emerged (Gug
1977, Goodway 1987, Birkett and Poletti 2004). Several
studies have approached the subject of piano strings
produced during the pivotal early 19th-century period
(Gug 1977, Rose and Law 1991, Birkett and Poletti 2004,
Paulello 2006). The time period covered by our research
begins in 1780 with Sébastien Érard opening his piano
workshop in Paris, on Rue du Mail, at a time when he was
still using harpsichord strings. It ends in 1855 with the
Exposition Universelle in Paris, when the piano, which
had become widespread in Europe, became the flagship
instrument of the 19th century, a symbol of social success
and technical progress.
The Musée de la musique project
This project initiated by the Musée de la musique in
Paris is part of the general effort to ensure the preventive
conservation and maintenance of the playing condition
of certain instruments within the museum’s collection.
The aim here is to provide makers and restorers of early
pianos with new information on historical strings.
A multidisciplinary team was created in 2017 under
the scientific direction of the Musée de la musique,
bringing specialists in material sciences from the LAPA
(CEA, Saclay) and ICMPE (Université Paris Est Créteil)
metallurgy laboratories together with piano-making
specialists. A corpus of string samples was compiled from
the collections of the Musée de la musique, Musée des
Instruments de Musique in Brussels, and private collec-
tions. Physico-chemical and metallurgical analyses of this
corpus were carried out using the following methods:
optical micrography (OM), scanning electron microscopy
with energy-dispersive x-ray spectroscopy (SEM-EDS)
and electron backscatter diffraction (EBSD), electron
TECHNICAL STUDIES
probe microanalysis (EPMA), and Vickers microhardness
testing. An ongoing historical research enabled more
information to be obtained for the interpretation of the
experimental results.
Materials and sampling
The sampling campaign compiled 148 early strings
made of iron and steel (Table 1) from pianos built in
Europe and dated between 1769 and 1858. The relia-
bility of the samples collected and the authenticity of
the strings were evaluated by referring to knowledge
of the material history of each instrument (Goodway
1987) and the associated stringing practices. The
piano-making experts, after careful examination of
the historical conservation and state of restoration of
the piano, could authenticate its original state. The
way the strings were wound around the tuning pegs
depended on the working methods of a given maker
or workshop (see Figure 1 for examples of sampling).
The physico-chemical results obtained demonstrated
the relevance of this methodology. The corpus included
16 samples of strings currently available on the market
sold as reproductions of early strings. The length of the
samples ranged from 2 meters to 1 cm.
Figure 1. (a) Pleyel pianoforte, Paris, France, 1842, Musée de la musique,
Paris, E.991.16.1; (b) strings on the soundboard of Pleyel pianoforte, No.
11980, 1844; (c) wire winding on a Fritz pianoforte, Vienna, ca. 1820,
private collection
Experimental results
Three classes of strings
Three categories of strings were identified in our corpus:
category 1 strings made of phosphorus iron; category 2
strings made of steel with globular cementite (Fe3C); and
category 3 strings made of a harder steel. The classified
samples represent 80% (75 wires) of the wires analyzed
using both OM and SEM-EDS, leaving 15% (14 wires),
which could not to be categorized.
IRON PIANO WIRE PRODUCTION AT THE BEGINNING OF THE 19TH CENTURY 51

Category 1 consisted of strings made of iron with a low On the two others, carbides were not visible after Nital
phosphorus content, i.e., below the SEM-EDS detection etching or under SEM observations.
threshold (0.3 wt%). Ten pianos dating from 1769 to 1822
have this type of strings. Oberhoffer’s etching revealed
elongated lengthwise white strips in the strings with a
thickness of approximately 1 µm (Figure 2, A2). The average
phosphorus content of these strips found using EPMA
analyses ranged between 0.06 wt% and 0.11 wt%. No iron
carbide was detected. Nital etching revealed a cold-worked
microstructure characterized by elongated ferrite grains
appearing lengthwise in the string (Figure 2, A1).
Figure 2. Microstructure of the category 1 strings (A1) with elongated
Category 2 is characterized by a phosphorus-free iron
grains (A2) and white strips of phosphorus; microstructure of the category
matrix containing globular carbides (Figure 2, B 1). 2 strings (B1) with elongated grains and (B2) carbides surrounded by
These samples were taken from 21 pianos dating from holes; microstructure of the category 3 strings (C1) with elongated ferrite
1815 to 1855. The carbides are surrounded by elongated grains (C2) in cross section with lamellar carbides
holes (Figure 2, B2). SEM-EBSD analyses revealed pref-
erential crystallographic orientation in the elongated Typical inclusions
ferrite grains, indicating that the wire was cold drawn.
For the category 1 strings, OM revealed numerous inclu-
The carbon content of these strings was estimated to
sions of varying shapes and sizes, organized longitudi-
average 0.25 ± 0.07 wt% according to the surface ratio of
nally along the string (Figure 3, α1). The inclusions found
cementite (3.38 ± 1.06 wt%) obtained using SEM-EBSD
were primarily aluminosilicates (Si-Al-O), phosphorus
phase mapping and assuming that the thermodynamic
oxide with or without silicon (Si-P-O), and iron oxides
equilibrium is reached during globularization tempering.
(Fe-O, Fe-O-Si) (Figure 3, α2).
Category 3 consists of five strings made of steel: three
For the category 2 strings, inclusions were less frequent
strings from Érard pianos dated 1854, 1855 and 1858,
than in the category 1 strings (Figures 3, β1 and 4).
and two others from an 1838 Rogez piano, and an 1835
SEM-EDS analyses revealed the following typical
Roller & Blanchet piano. These five strings had a fine,
inclusions: (a) SiO2 beads and (b) elongated copper
cold-worked microstructure (Figure 2, C1). Only one
sulfide inclusions appearing lengthwise in the string
string from an Érard piano (ca. 1858) contained lamellar
(Figure 3, β2).
micrometric carbides (Figure 2, C2). On two strings from
the 1854 Érard piano, LECO carbon analysis indicated For the category 3 strings, inclusions were less frequent
a respective carbon content of 0.51 wt% and 0.67 wt%. than in the category 1 strings (Figures 3, γ1 and 4). Those

Figure 4. Amounts of porosity and

inclusions in cleanliness of piano
  strings
52  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
observed were typically angular inclusions of alumi-
nosilicates containing 1–2 wt% of titanium (Si-Al-Ti)
(Figure 3, γ2). Some strings also contain SiO2 beads and
elongated copper sulfide inclusions.
Figure 3. Typical inclusions: (α1) distribution and (α2) composition in
the category 1 strings; (β1) distribution and (β2) composition in the
category 2 strings; and (γ1) distribution and (γ2) composition in the
category 3 strings
Microhardness
The measured Vickers microhardnesses were very similar
for the category 1 and 2 strings, and were highest for the
category 3 strings (Figure 5). The average values for each
category were, respectively, 318 ± 32 Hv, 336 ± 32 Hv,
and 469 ± 37 Hv.
Findings from analysis of written sources
“The softest iron is the best for wire making,” wrote Bouchu
(1756, 168). A careful selection of the best iron is chosen to
be drawn into wire (Duhamel du Monceau 1768). Duhamel
describes the various stages of production:
1. selection of ores
2. processes and temperatures for smelting and mixing
in crucibles
TECHNICAL STUDIES
3. introduction of other materials during the rolling and
wire-drawing stages
4. heat and chemical treatments
All of these stages influence string composition, and
attempts were made to identify in the analyses the
features that correspond to each stage. Moreover, different
qualities of strings coexisted at a given point in time.
For example, Giorgio di Roma (1834, 37) refers to three
types of strings then available on the market: English
steel strings by Webster, iron strings from Berlin, and
brass strings from Nuremberg.
Discussion
Mechanical and acoustic performance
An attempt was made to understand the string compo-
sition and quality that allowed piano string tension
to increase in the 19th century as pianos moved from
reinforced wooden frames to bolted frames and then to
iron frames cast in a single piece, as invented in 1825 in
Boston by the piano maker Alpheus Babcock.
The Vickers microhardnesses are very similar between
the category 1 and 2 strings that coexisted in the 1820s
and 1830s. However, they are higher for the category
3 strings, which can be explained by a higher carbon
content. In principle, these strings are capable of with-
standing higher tensions without breaking.
The literature notes that the best strings available on
the European market had an increasing tensile strength
(Birkett and Poletti 2004). Initially they came from
Nuremberg, and from 1830 began to be produced in
Figure 5. Vickers microhardness
  measurements of early strings
IRON PIANO WIRE PRODUCTION AT THE BEGINNING OF THE 19TH CENTURY
Birmingham. The Birmingham producer Webster &
Horsfall had a monopoly on piano strings until the
introduction of Muller strings in 1840 (Vienna), which
then faced new competition from Poehlmann strings in
1855 (Nuremberg) (Dolge 1972, 123–24). The question
arises: to which manufacturer do the categories of strings
analyzed correspond?
The French piano making industry—a booming business
in Paris, notably with the development of the Érard and
Pleyel workshops—contrasts with the development of
string production in France: Pleyel (1810) registered
an import patent for a German process in 1810, during
the Continental Blockade, and Érard had a workshop
in London allowing it to procure high-quality strings
produced in England. During restoration of the grand
piano by Ignace Pleyel & Co, No. 1555, Paris, 1829,
MM E. 985.1.1, it was observed that, in the five Pleyel
pianos of the same model studied (Nos. 1429, 1555,
1559, 1619, and 1737), the string gauge numbering on
the wrest planks is different. The hypotheses are that
the Pleyel factory changed its stringing system (perhaps
using strings from other suppliers) or that the gauges
were changed due to modifications in the string manu-
facturing processes.
Metallurgical processes, ores, and fuel
The refining processes used to improve the level of inclu-
sion cleanliness are essential for the quality of strings,
which could break due to a cluster of inclusions under
the effect of tension. Category 1 strings contain more
inclusions than the others, suggesting that the refining
processes for the latter were more effective.
The three categories of strings have a microstructure
marked by cold wire-drawing, such as a ferritic matrix
with elongated grains in the wire-drawing direction and
holes formed during the iron stretching process around
inclusions that remain undistorted.
For the category 1 strings, our findings are consistent
with a phosphorus content of between 0.22 wt% and
0.1 wt%, revealed by Goodway and Odell (1988, 36–7).
It seems this is due to the metal melting directly in the
charcoal, which was prepared with bark, and to high
phosphorus iron ores (Wysor 1908, 31). The refining
process would lead to disparate phosphorus levels in the
ferrite grains, which explains the formation during cold
wire-drawing of strips with a higher phosphorus content
53
in their microstructure. The inclusions are organized in
lines during the wire-drawing (Ehrenreich, Hamilton,
and Nash 2005). Goodway and Odell (1988, 38) report
that the carbon content of harpsichord and early piano
strings must have been lower than 0.01 wt% for the phos-
phorus in the iron to conserve its hardening properties,
while maintaining ductility and ease of drawing. By
comparison, modern reproductions of early strings used
for the 1802 Érard facsimile at the Musée de la musique
have numerous inclusions aligned in the wire-drawing
direction containing mainly silicon and manganese,
unlike the phosphorus oxide inclusions found in the
category 1 strings. The iron matrix probably does contain
phosphorus, as indicated by the manufacturer (~0.1 wt%)
(Birkett 2006, 337–38).
In the category 2 strings, the small amount of carbon
could be due to carburization (Paulello 2006) and
indicates the use of phosphorus-free ores. Goodway
(1987) concluded that carbon replaces phosphorus as
a hardening element with the use of processes such as
annealing, quenching, and tempering in the first steel
piano strings. Carbides were obtained by metallurgical
annealing operations prior to wire-drawing. Their size
depends on the annealing temperature and duration:
the longer the annealing, the bigger the carbides. These
carbides in the strings are considered large, measuring
on average 1 µm across. The mechanical properties of
iron containing globular carbides facilitate cold drawing,
which became more ductile (Barralis 2005, 70). The iron
matrix was distorted during cold drawing, creating holes
in the elongated denser globular carbides. The presence
of copper sulfide may come from the introduction of
sulfur due to using coal rather than charcoal. An addition
of copper may have formed sulfides to counteract the
adverse effects of the sulfur in the iron. By comparison,
one type of modern “early” string contains globular
carbides with an average diameter of 0.3 µm instead of 1
µm; the ferrite grains are also smaller and without pref-
erential crystallographic orientation, indicating higher
mechanical and thermal stresses. Another type of modern
string contains the same SiO2 beads as early strings with
globular carbides, but has no globular carbides in its
iron matrix.
For category 3 strings, the steel may contain nanometric
lamellar carbides similar to the micrometric carbides
detected in the string from the ca. 1858 Érard piano,
as they have similar inclusions, microhardnesses, and
54  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
microstructures. This hypothesis remains to be verified
by additional TEM analyses. Si-Al-Ti angular inclusions
may come from the material of the crucibles used in a
carburization process. The crucibles in Sheffield were
reputed to be of a particular clay-based composition
(Dillmann, Hilaire-Pérez, and Verna 2011, 50).
Emergence and obsolescence of types of strings
Harpsichord wires and early pianoforte wires
before 1830
Several authors agree that phosphorus iron, as analyzed in
the category 1 strings, was used for 17th- and 18-century
harpsichord strings and for piano strings in the 1780s.
Goodway (1987) refers to this process of string produc-
tion as the “Westphalie process.” This refining process
was highly effective in decarburizing cast iron in which
phosphorus remained the only hardening element
present, producing an iron with few inclusions and
sufficient ductility (Goodway and Odell 1988, 41–3). Iron
ores had high phosphorus content in France, Germany,
and England (see oolitic and Cumberland iron ores in
Albitreccia 1939). At that time, hydraulic force was used
for the wire-drawing and hammering stages (Duhamel
Du Monceau 1768). In our analyses and according to
Montal (1865), these strings became obsolete after 1830
(Figure 6).
Webster wires and the pianoforte between 1830
and 1850
Can we say that category 2 strings made of steel with
globular carbides were produced from phosphorus-free
TECHNICAL STUDIES
blister steel and came from England? The English
industries are known for having produced blister steel
in a crucible and for using phosphorus-free Swedish
iron ores, notably in Sheffield. Webster’s English strings
dominated the European market from 1820 to 1840 and
replaced German strings for a twenty-year period (Dolge
1972). Due to a shortage of wood in England, charcoal
was replaced by coal, which was increasingly used as
steam engines developed. The category 2 strings emerged
around 1830 and were still used in 1855 (Figure 6). The
Horsfall factory in Birmingham, which produced piano
strings during this period, was the first wire maker to
use a steam engine in the 1810s and 1820s (Horsfall
1971, 114). Current analyses do not allow us to directly
associate the differences in microstructure with increases
in speed and tensile force on the strings as a result of the
power of the steam engines that replaced the hydraulic
mills (Figure 7).
Figure 7. Different examples of driving power used for wire-drawing:
(a) Nuremberg, ca. 1489; (b) France, Encyclopédie, Diderot, 1758; (c)
England, 1848
From 1850 onwards
James Horsfall exhibited his strings in London at the
first World’s Fair in 1851, registering his Steel Wire
Patent in 1854, then selling it to the Webster metal works
and forming the partnership Webster & Horsfall. The
Figure 6. Emergence and
  obsolescence of types of strings
IRON PIANO WIRE PRODUCTION AT THE BEGINNING OF THE 19TH CENTURY 55

patented process consisted in an innovative heat treat-
ment with quenching and wire-drawing, producing steel
strings with a tensile strength twice that of the strings of
the day. According to Borchers and Kirchheim (2016), the
metallurgical processes involved are known for producing
cold-drawn pearlitic steel strings with lamellar carbides,
as seen in the strings of an Érard piano from ca. 1858. The
string sampling campaign must be continued to confirm
or disprove the hypothesis that category 3 strings can be
associated with this process (Figure 6).

Table 1. Corpus of iron and steel piano strings

Number of Number of
Instruments Instruments
sampled sampled
(maker, series no., city/country, date, collection, inventory no.) (maker, series no., city/country, date, collection, inventory no.)
wires wires
Hellen, Bern, Switzerland, 1769. Private collections 4 Dammekens Jean, Gand, Belgium, 1838. MIM, Brussels, 3289 1
Poelmann, England, ca. 1780. Musée de la musique, Paris 1 Rogez, Paris, France, 1838. MIM, Brussels, 3345 1
Longman & Broderip, England, 1788. MIM, Brussels 2 Lichtenthal H. & Cie, Brussels, Belgium, 1839. MIM, 1
(dubious attribution) Brussels, 2955
Broadwood, England, ca. 1800. MIM, Brussels, 1629 1 Erard, Paris, France, 1840. MIM, Brussels, 1973.014 1
(dubious attribution)
Baumann, No. 9943, Rhineland, Germany, ca. 1800. Private 2 Persyn Joseph, Belgium, 1840. MIM, Brussels, 2010.014 1
collections
Erard, No. 5308, Paris, France, 1802. Private collections 12 Pleyel, Paris, France, 1842. Musée de la musique, Paris, 1
E.991.16.1
Erard, Paris, France, 1804. MIM, Brussels, 2000.002 1 Pape, Paris, France, 1843. MIM, Brussels, 3537 1
Lindholm, Stockholm, Sweden, 1808. Private collections 5 Pleyel, Paris, France, 1843. MIM, Brussels, 3814 1
Erard, Paris, France, 1809. Musée de la musique, Paris, 2 Pleyel, No. 11980, Paris, France, 1844. Private collections 5
E.2001.5.1
Broadwood, London, England, 1810. MIM, Brussels, 1636 1 Kiffer, Paris, France, 1845. MIM, Brussels, JT305 1
Dieudonné & Schiedmayer, Stuttgart, Germany, ca. 1815. 2 Pape, No. 5544, Paris, France, 1845. Private collections 1
Private collections
Fritz, Vienna, Austria, ca. 1815. Private collections 6 Pleyel, No. 12323, Paris, France, 1845. Private collections 1
Erard, Paris, France, 1818. Musée de la musique, Paris, 2 Roller & Blanchet, Paris, France, 1845. MIM, Brussels, 1
E.979.3.1 1995.27
Erard, Paris, France, 1820. MIM, Brussels, 1998.011 1 Vogelsangs Jacques-François, 1845 1
Fritz, Vienna, Austria, ca. 1820. HEM, Genève, Suisse 26 Hesselbein, No. 284, Paris, France, ca. 1845. Private 3
collections
Broadwood, No. 26612, London, England, ca. 1822. Private 1 Erard, Paris, France, 1846. MIM, Brussels, JT0446 2
collections
De Frey Père & Fils, Paris, France, 1826. MIM, Brussels, 1 Pleyel, Paris, France, 1846. MIM, Brussels, 3649 1
1986.002
Pleyel, No. 1475, Paris, France, 1829. Private collections 1 Boisselot & Fils, Paris, France, 1848. MIM, Brussels 1
Pleyel, Paris, France, 1831. MIM, Brussels, 3963 1 Boisselot & Fils, no. 2632 Paris, France, ca. 1848. Private 1
Collections
Pleyel, No. 2381, Paris, France, 1832. Private collections 8 Pleyel, Paris, France, 1851. MIM, Brussels, 1973.012 1
Pape, Paris, France, 1833. MIM, Brussels, 1976.030 2 Preiss F., Paris, France, 1853. MIM, Brussels, 1994.001 1
Mercier, Paris, France, 1834. Musée de la musique, Paris, 2 Erard, No. 26594, Paris, France, 1854. Private collections 13
E.01497
Broadwood, 1835. MIM, Brussels, 3291 1 Erard, London, England, 1855. MIM, Brussels, 2013.074 1
Debain, Paris, France, 1835. MIM, Brussels, 1998.018 1 Pleyel, No. 21975, Paris, France, ca. 1855. Private 1
collections
Roller & Blanchet, Paris, France, 1835. MIM, Brussels, 1 Gebauhr, Paris, France, 1855. Musée de la musique, Paris, 1
2007.11 E.2005.4
Schiedmayer, London, England, ca. 1835. Private collections 1 Erard, Paris, France, ca. 1858. MIM, Brussels, 1987.001 1
Pape, Paris, France, 1836. Musée de la musique, Paris, 1 Schmidt, No. 338. Private collections 2
E.973.2.1
Pleyel, No. 5610, Paris, France, 1837. Private collections 12
56  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conclusion and prospects
Based on our physico-chemical and metallurgical anal-
yses of early European piano strings, three types of strings
were identified: phosphorus iron strings, steel strings with
globular carbides, and harder steel strings with higher
carbon content. Our research confirms a major change
in the production of piano strings, with a shift away
from phosphorus iron strings in favor of strings with
a low carbon content in around 1830. A new sampling
campaign for post-1850 strings is needed to support the
hypothesis of another technological change around 1850.
In 19th-century Europe, during the industrial revolution,
piano string production might have been influenced
by technical innovations in metallurgy, including the
introduction of crucible steel in 1740, the use of phospho-
rus-free ores, and changes in driving force and fuels with
the transition from hydraulic power to steam engines, and
from charcoal to coal, etc. Characterizing the mechanical
properties of each class of strings through further tests
will be undertaken in a subsequent study.
Acknowledgments
This work is funded by the Fondation des Sciences du
Patrimoine (EUR-17-EURE-0021). The authors would
like to acknowledge the editors and reviewers for the
quality of their work.
References
Albitreccia, A. 1939. La production du minerai de fer en
Europe. Annales de géographie 48(272): 194–99.
Barralis, J. 2005. Précis de métallurgie. Paris: Editions
Nathan.
Birkett, S. and P. Poletti. 2004. Reproduction of authentic
historical soft iron wire for musical instruments. In
Instruments à claviers – expressivité et flexibilité sonore.
Actes des Rencontres Internationales «harmoniques»,
Lausanne, April 2002, ed. T. Steiner, 259–72. Bern: P.
Lang. http://fortepianos.com/iron%20wire.pdf.
Birkett, S. 2006. The physical characteristics of historical
iron music wire and a report on its replication as a viable
modern product. In Bowed and Keyboard Instruments in
the Age of Mozart. Actes des Rencontres Internationales
«harmoniques», Lausanne, 2006, ed. T. Steiner, 327–46.
Bern: P. Lang.
Borchers, C. and R. Kirchheim. 2016. Cold-drawn pearl-
itic steel wires. Progress in Materials Science 82: 405–44.
TECHNICAL STUDIES
Bouchu, M. 1757. Forges (Grosses). In L’Encyclopédie
Diderot & d’Alembert, vol. VII, 135–68. Paris: Briasson,
David, Le Breton, Durand.
Clarke, C. 2017. Searching and squirrelling: Sourcing
materials for early pianos. In 4th Annual Conference
COST FP1302 WoodMusICK, 5–7 October 2017, Brus-
sels, 66–7. http://woodmusick.org/wp-content/uploads/
Proceedings_WoodMusICK_BRS.pdf.
di Roma, G. 1834. Manuel simplifié de l’accordeur ou l’art
d’accorder les pianos. Paris: Librairie encyclopédique de
Roret.
Dillmann, P., L. Hilaire-Pérez, and C. Verna. 2011. L’acier
en Europe avant Bessemer. Toulouse: CNRS-Université
de Toulouse-Le Mirail.
Dolge, A. 1972. Pianos and their makers: A comprehensive
history of the development of the piano from the monochord
to the concert grand player piano, vol. 1, 123–26. New
York: Dover Publications.
Duhamel du Monceau, H.-L. 1768. L’Art de réduire le
fer en fil connu sous le nom de fil d’archal, reissued 1781.
Ehrenreich, R.M., E. Hamilton, and S.K. Nash. 2005.
Far from Barbaric: Re-assessing the sophistication of
Merovingian metalworking. JOM 57(8): 51–5.
Goodway, M. 1987. Phosphorus in antique iron music
wire. Science 236(4804): 927–32.
Goodway, M. and J.S. Odell. 1988. The historical harpsi-
chord – Volume two.The metallurgy of 17th- and 18th-cen-
tury music wire, ed. H. Schott. New York: Pendragon
Press.
Gug, R. 1977. Über Analysen alter Cembalosaiten.
Colloquium Ruckers Klavecimbels en Copieën, Anvers,
Ruckers Genootschap. In The Brussels Museum of Musical
Instruments 7: 125–28.
Haine, M. 2006. Les facteurs de piano du temps de Chopin.
Interpréter Chopin, ed. J.-J. Eigeldinger, Les cahiers du
Musée de la musique, No. 8. Paris: Cité de la Musique.
Horsfall, J. 1971. The Iron Masters of Penns: 1720-1970.
Kineton: The Roundwood Press. http://atlantic-cable.
com/CableCos/Webster-Horsfall/history.htm.
Houssay, A. 2004. La fabrication des cordes harmoniques
entre 1790 et 1960 : le rôle des artisans, des ingénieurs,
des facteurs d’instruments et des musiciens. DEA disser-
tation, Centre d’Histoire des Techniques, CNAM, Paris,
France.
IRON PIANO WIRE PRODUCTION AT THE BEGINNING OF THE 19TH CENTURY 57

Montal, C. 1865. Cordes de piano; manière de les remettre

et outils nécessaires à cette opération. In L’art d’accorder
soi-même son piano, 3rd edition, 65. Paris: J. Meissonnier,
Éditeur de Musique.
Paulello, S. 2006. La réalization de cordes pour instru-
ments anciens. In Actes de le journée d’étude La facture
du piano en France entre 1780 et 1820, Paris, 27 February
2006, 43–8. http://www.citedelamusique.fr/pdf/insti/
recherche/facture/realisation.pdf.
Piat, I., P. Henry, A. Gonthier, B. Dormard, and D. Bic.
1982. Étude des cordes métalliques anciennes d’instru-
ments à clavier pour le Musée Instrumental du Conserv-
atoire National Supérieur de Musique de Paris, École
Centrale de Lyon, France.
Pleyel, I. 1810. Procédés de fabrication de cordes en
cuivre et en fer, pour le forte-piano et autres instruments.
Brevet d’importation de 15 ans, No. 630, cote 1BA601.
Paris: INPI.
Rose, M. and D. Law, eds. 1991. A handbook of historical
stringing practice for keyboard instruments: 1671–1856.
Lewes, East Sussex: Malcolm Rose &​David Law.
Wysor, H. 1908. Metallurgy. Easton: The Chemical
Publishing Co.

Author
Marie-Gabrielle Durier is a materials science engi-
neer (CNRS, dr4, Paris) currently working on research
programs at the Musée de la musique (Paris) funded by
Patrima and DIM “Matériaux anciens et patrimoniaux”
of the Ile-de-France region.
Transparent Varnishes on Copper Alloys Dating from
the 19th Century: Characterisation and Identification
Strategies
Julie Schröter* Aline Michel Sigrid Mirabaud Laura Brambilla
Haute Ecole Arc Conservation- Haute Ecole Arc Conservation- Institut National d’Histoire de l’Art Haute Ecole Arc Conservation-
restauration restauration (INHA) restauration
HES-SO University of Applied HES-SO University of Applied Paris, France HES-SO University of Applied
Sciences and Arts Western Sciences and Arts Western sigrid.mirabaud@inha.fr Sciences and Arts Western
Switzerland Switzerland Switzerland
Neuchâtel, Switzerland Neuchâtel, Switzerland Ludovic Bellot-Gurlet Neuchâtel, Switzerland
julie.schroter@he-arc.ch aline.michel@he-arc.ch MONARIS (UMR 8233), Sorbonne
laura.brambilla@he-arc.ch
www.he-arc.ch University/CNRS
Celine Paris Paris, France
*Author for correspondence MONARIS (UMR 8233), Sorbonne ludovic.bellot-gurlet@sorbonne-
University/CNRS universite.fr
Paris, France
celine.paris@sorbonne-universite.fr

Abstract
Many varnished copper-based artefacts dating from the
19th century can be found in museum collections. Although
conservators deal on a daily basis with these surface finishes, few
studies on this topic have been conducted during the last few
decades on European objects. A more thorough understanding
of these historical coatings would be helpful to establish accurate
diagnoses and to contextualise the results with respect to the
historical written sources. This paper presents the first results
of the LacCa (Lacquered Copper Alloys) project currently being
undertaken by an international multidisciplinary team. LacCa
aims to develop a methodology that conservators can apply to
identify and preserve these coatings. To this end, both historical
objects and ancient varnish recipes have been studied. In a
first step, the identification and characterisation protocol was
validated using mock-ups, the results of which are presented
in this paper.
Keywords
varnishes, copper alloys, UV-induced fluorescence,
FTIR spectroscopy, eddy current, confocal microscopy,
spectroscopic ellipsometry

Introduction
A significant number of copper-based artefacts in
museum collections were originally covered by a
transparent, often tinted, varnish. This technology was
applied to various categories of objects such as scientific
instruments and decorative bronze items, including
horological and liturgical objects. The varnish had a
primary role of protecting the metallic surface from
corrosion. However, for the last two categories, the
purpose was also aesthetic, often applied as a low-cost
method for imitating gold.
These historical surface finishes present several diffi-
culties for conservators. When gold-imitating varnishes
are well preserved, they can be difficult to distin-
guish from authentic gilding by visual inspection.
This is especially true if the varnished parts cannot
be dismantled to reveal characteristic traces on the
reverse. Furthermore, there does not appear to be
any convenient published conservation methodology
despite the fact that these finishes are common on
objects in museum collections.
This paper presents the first results of the LacCA
(Lacquered Copper Alloys) project, a study aimed
at identifying the composition of the varnishes and
examining their morphology by researching both
historical objects and ancient varnish recipes from
the 19th century. The results will help identify and
create diagnosis criteria for conservators based on
simple and affordable methods. A methodology was
developed using mock-up coupons. The aim was to
validate an identification and characterisation protocol
for varnished copper alloys. The focus was on different
portable and non-invasive techniques, used within the
framework of on-site campaigns in Swiss and French
museums on historical artefacts.
TRANSPARENT VARNISHES ON COPPER ALLOYS DATING FROM THE 19TH CENTURY: CHARACTERISATION AND IDENTIFICATION STRATEGIES
State of the art
Only one systematic analysis campaign on varnished
copper-alloy objects has been published during the last
few decades (Lanterna and Giatti 2014). The analyses
were performed by Fourier transform infrared (FTIR)
spectroscopy on a corpus of 16 scientific instruments
from the Museo Galilei and Fondazione Scienza e Tecnica
(Italy), dating mostly from the 19th century. Binders
including shellac, sandarac and mastic were detected.
In addition, only a few case studies concerning gold
varnishes applied to copper alloys in decorative arts
are available. Long (2000) studied the varnished brass
parts of a pair of Argand lamps (1839–48). The lamps
were examined under UV radiation, and a natural resin
(‘garnet lac’) was detected by FTIR.
Thomson mentioned the case of a Grecian couch with
varnished copper-alloy elements (ca. 1820–40) which
were analysed by x-ray fluorescence (XRF) spectros-
copy and examined under ultraviolet (UV) radiation.
In addition, she included in the article a short review
of varnishing manuals, the majority from English or
American written sources and dating from the 18th and
19th centuries (Thomson 1991). A more extensive review
of gold varnish recipes is available, but it covers several
centuries and different kinds of organic and inorganic
substrates (Wellmer 1999).
A parameter that appears to be important and potentially
influential in the final result of the coating is the varnish
thickness. However, none of the studies previously cited
considered this parameter and no data have been found
in the literature.
In addition to analytical laboratory techniques, such
as the previously cited FTIR, other techniques were
applied to identify the varnishes. Of particular interest
for the LacCA project are techniques easily accessible
to conservators. UV examination and electric resistance
tests are mentioned in literature for identifying varnishes
on copper alloys and distinguishing them from genuine
gilding. Thomson (1991) mentions weak fluorescence
on the brassware of the Grecian couch under UV light.
In particular, examination under UVA radiation has
the advantage of being an affordable and quick way to
detect characteristic fluorescence from organic coatings
(Cosentino 2015). Unfortunately, no published reference
data specifically for copper-alloy substrates were found.
This is an issue, as the varnishes might be very thin and
59
the reflective metallic surface might influence the type
and colour of UV fluorescence.
Rogge and Lough (2016) showed that UVA-induced
visible fluorescence and false-colour reflected UVA
images of tintype varnishes are not sufficient to identify
the different varnish materials. The varnishes all emitted
a visible fluorescence but were applied on a white metal
and not on copper alloys.
Other techniques – like solubility tests or other chemical
tests – have been cited (Thomson 1991, Long 2000) but
have the disadvantage of being destructive. This is an
issue on varnished metallic substrates, as a partial varnish
removal might induce local anodic corrosion.
Experimental
Three series of mock-up samples were prepared with
different varnishes selected from 60 recipes from 23
ancient manuals, dating from 1803 to 1893 (Table 1). As
spirit varnishes are over-represented (72%) and shellac
is widely used as at least one binder, it was decided to
prepare the varnishes summarised in Table 1. As it was
recommended for gold varnishes to avoid interaction with
added colourants (Lami and Tharel 1881–91), bleached
dewaxed shellac was used for varnish C and D. For
comparison with varnish A (thickness and UV imaging),
a recipe (B) with coloured shellac was also prepared
(Table 2). Mirror-polished and satin-finished coupons,
made from CuZn37 alloy (60 × 60 mm), a common brass
alloy, were employed. Prior to varnishing, coupons were
degreased with ethanol. An additional aluminium sample
was used to compare imaging produced on a white metal.
Non-fluorescent glass substrates were used as a reference
for UV imaging.
A first series of coupons was produced to determine the
range of measurable thickness of the varnishes. Manual
dip coating and brush application were chosen because
they are mentioned in 19th-century literature (Tingry
1803, Dessaignes 1861). On decorative elements, the
surface geometry may be complex, with a thinner layer
on the reliefs. Therefore, it was chosen to simulate
the thinnest possible varnish by applying only one
layer of a simple shellac varnish (A), although several
layers might be necessary to replicate ancient recipes
(Nosban 1843, Londe 1893). In the case of dip coating,
no indication about the removal speed was available in
literature so the coupons were immersed vertically and
60  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Table 1. Type of produced (brass) series and characterisation techniques

Characterisation techniques
Substrate Application Varnish
Series Code Spectroscopic Confocal
finish technique recipe Eddy current UV FTIR
ellipsometry microscopy
Smd_A-Al
×
aluminium Satin manual dip
Smd_A coating
Series 1 A
Mmd_A × ×
Mirror-polish
Mb_A
brushing
Sb_A Satin × ×
Smd_B Satin manual dip
B ×
coating
Sb_C Satin
Series 2 C ×
Mb_C Mirror-polish
brushing
Sb_D Satin × ×
D
Mb_D Mirror-polish ×
Mid_A2 A × × × ×
Mid_C2 C × ×
Mid_D2 D × × ×
Mid_A11 A × × ×
industrial
Series 3 Mirror-polish
Mid_C11 dip coating C × ×
Mid_D11 D × ×
CMid_A
(cylinder) A × ×
9–10 µm

Table 2. Recipes and varnish viscosities

Varnish D
Varnish A Varnish B
Varnish C Vernis mutatif ‘gold
Blonde shellac Orange shellac
Recipe Vernis Anglais varnish’
(based on (based on
(Valicourt 1872) (Tingry 1803)
Riffault et al. 1862) Londe 1893)
(ounces converted to grams)
Shellac 125 g* 225 g 30.57 g**
Sandarac 250 g 61.14 g
Elemi ‘big as a hazelnut’ 61.14 g
Mastic 30 g
Camphor ‘big as a walnut’
Turmeric 22.92 g
Ingredients Saffron 636.9 mg
Gummi-gutta 22.92 g
Dragon’s blood 30.57 g
Alcohol 1 kg 611.43 g
2 kg 1000 mL
‘wine spirit’ ‘pure alcohol’
Turpentine 30 g
Crushed glass 91.71 g
Measured viscosity (cP)
1.81 2.08 3.34 4.40
according to Bingham model
* In the original recipe, shellac in flakes was used.
** In the original recipe, seed lac was used.
TRANSPARENT VARNISHES ON COPPER ALLOYS DATING FROM THE 19TH CENTURY: CHARACTERISATION AND IDENTIFICATION STRATEGIES
removed quickly to be hung up and left to dry at room
temperature (Figure 1b).
Figure 1. Varnishing of the samples: (a) automated dip coating, (b)
manual dip coating and (c) brushing
The brush application (Figure 1c) was performed
according to the procedure indicated in a varnish
manual (Nosban 1843): ‘Heated in a way it doesn’t burn
the upper side of the hand…’ Prior to varnishing, heating
of the substrate in an oven at around 30°C proved to be
convenient. This makes it possible to correlate a precise
temperature to more general recommendations in the
ancient literature, allowing to eliminate blanching
or retraction of the varnish which may occur during
drying, especially for varnish D. Prior to varnishing,
the excess was removed from the brush and the varnish
applied in juxtaposed layers using a 10 mm large squir-
rel-hair brush.
Series 2 was prepared in the same way with more complex
recipes: varnish C and varnish D (the latter is mentioned
in five French manuals dating from different periods).
These were prepared with one of the recommended water
bath procedures (Tingry 1803) at around 50°C and the
alcohol grade adapted to fit the historical period. All
varnishes were filtered using a vacuum pump.
In order to validate the thickness measurement as well
as the FTIR protocol, series 3 was produced using an
automated dip coating after adjusting the viscosity of the
varnish solution. According to the preliminary measure-
ments performed on the first two series, a thickness of
around 2–3 µm was selected for all recipes. As thicker
varnishes can sometimes be observed on scientific
objects, 11–12 µm thicknesses were also produced. In
addition to the flat coupons, one brass cylinder (diameter
50 mm) was coated with a 9 µm layer.
61
A multi-analytical approach was used for characterising
the three series of coupons. Cutting-edge laboratory
techniques, such as spectroscopic ellipsometry, confocal
microscopy and FTIR, were used together with tech-
niques more easily accessible to conservators, such as
UV imaging and eddy current (Table 1).
Thickness measurements
Depending on the thickness range, spectroscopic ellip-
sometry and confocal microscopy were used. Reference
zones were determined in order to perform comparative
measurements with an eddy-current probe and vali-
date the on-site measurement protocol. Ellipsometry
can measure thicknesses smaller than 2 µm locally on
polished substrates and was done on a straight line from
the top to the bottom of the coupon (perpendicular to the
brush application). Mapping with confocal microscopy
was used for thicknesses greater than 2 µm.
For series 3, homogeneous zones were located on the
samples with a mask and comparative measurements
performed with an eddy-current probe (Phynix Surface
Pro S gauge and dual FN 1.5 probe). The probe was
applied freehand five times at a 90° angle after the
recommended calibration process using a brass substrate
with an identical geometry and an 8 µm calibration foil.
According to the manufacturer, the gauge has a range of
0–200 µm with an accuracy of ± 0.7 µm or 1%.
Infrared spectroscopy in specular reflectance mode
Infrared analyses were performed by specular reflectance
(SRefl) using an Alpha portable spectrometer (Bruker)
with an R-Alpha module. This measurement configu-
ration permits non-invasive, contactless measurement,
but requires a reflective surface in order to obtain mean-
ingful signatures. The samples were placed in front of the
spectrometer head at 15 mm working distance. Focusing
and selection of areas was performed with an internal
video camera. The analysed areas had a diameter of
approximately 5 mm.
UV imaging
An unfiltered Canon CN 750 camera with a 60 mm
macro lens, a Baader UV/IR cut and an X-Nite CC1
filter were used. The images were taken against a black
cardboard background. Two UV lamps1 (Dominique
Dutscher, main emission peak at 365 nm) were mounted
62  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

on custom-made stands at 45° and at a distance of
13 cm. The reproducibility of the imaging conditions
was ensured by locating all elements on the cardboard
and measuring the tilt angle and working distance of
the lamps (Figure 2a).
the colour sampling tool was used for each coupon to
obtain average L*a*b* values from three zones and to
compare the colorimetric data (Figure 2b).
Results and discussion
Thickness measurements
The results of the thickness measurements are summa-
rised in Tables 3 and 4.

Table 3. Series 1 and 2: Thickness measurements and results

Manual dip coating/brushing

Thickness (µm)
Varnish Average (full sample)
Code
recipe Spectroscopic Confocal
ellipsometry microscopy
Series 1 Smd_A - < 1.6
Mmd_A 1.30 -
A Mb_A 1.0 -
Sb_A - < 1.6
Sb_C - < 1.6
C
Mb_C 1.8 -
Series 2
Sb_D - 2.8
D
Mb_D 2.87 -
Figure 2. (a) Setting, (b) after image processing and (c) after readjusting
exposure time on sample 11
Due to the low average thickness of the varnishes on
An image of the UV Innovations™ target was taken series 1 (< 1–2 µm), the majority of the samples were
separately for each series to avoid any reflection of the measured using spectroscopic ellipsometry. Satin-fin-
fluorescent target on the metallic surface. The exposure ished coupons could only be measured, to a limited
time (2 seconds) was adapted to the lowest fluorescence extent, by confocal microscopy, since the varnish was
level of the target. Image processing was performed too thin on the edges. This means that on satin-finished
later with Photoshop. As the fluorescence level of some surfaces, a thicker layer of varnish may be necessary to
provide sufficient protection.
samples appeared to be too low for the target (leading to
systematic under-exposure of the samples or systematic Results confirmed that the more viscous the varnish,
over-exposure of the target), exposure was increased for varnishes A, C and D (Table 2), the thicker the layer
after the usual image processing by +2. In the last step, when using the same application technique.

Table 4. Series 3: Thickness measurements and results

Industrial dip coating

Varnish Thickness (µm)/average in reference zone
Code
recipe Spectroscopic ellipsometry Confocal microscopy Eddy current
Mid_A2 2.52 ± 0.1 - 2.10 ± 0.3
Mid_A11 - 11.7 ± 0.6 11.20 ± 0.6
A
Mid_A
Series 3 - 9.47 ± 0.2 9.0 ± 1.2
(cylinder)
Mid_C2 2.67 ± 0.1 - 2.3 ± 0.5
C
Mid_C11 - 10.3 ± 0.4 9.5 ± 0.8
Mid_D2 3.17 ± 0.3 - 2.5 ± 0.6
D
Mid_D11 - 11.7 ± 0.7 11.70 ± 0.2
TRANSPARENT VARNISHES ON COPPER ALLOYS DATING FROM THE 19TH CENTURY: CHARACTERISATION AND IDENTIFICATION STRATEGIES 63

Both application techniques induced a similar thickness

on identical substrates. On the brushed coupons, the
partial overlapping of two layers gave maximum values
of around 2 µm.

The comparative tests showed that the eddy-current

measurements performed manually (freehand without
a stand) with adequate calibration ensured the precise
measurement of the selected varnishes and metal
substrates, especially for flat surfaces. The test performed
on the cylinder was also conclusive, but the possibilities
on smaller convex forms might be limited.

UV imaging
As shown by UV imaging and the similar a* b* values
(Figures 3 and 4) in comparison to the reference brass
satin sample, the fluorescence of bleached shellac can
be very limited when the coating is extremely thin
(< 2–3 µm). As expected, increasing the thickness of
Figure 3. Results of the UV imaging
the same varnish (A) generates stronger fluorescence,
which is evidenced by higher L* and b* values. The to the reference colour on the glass slides (∆E 3.34) in
bleaching process of shellac can decrease the quantity comparison to the aluminium substrate (∆E 17.80).
of fluorophores initially present (Sutherland 2010).
Fluorescence was only slightly influenced by the treat-
As a result, varnish A is much more difficult to detect
ment of the metal. The polished sample was more reflec-
than the orange reference shellac (B) at a comparable
tive and appeared reddish under the lamp, which was
thickness. Varnishes C and D generate a more visible
confirmed by the colorimetric values tending towards
chromatic response even at a lower thickness, which
red on the green-red axis. However, ∆E is low (3.98) for
can be explained by the presence of more fluorescent
the samples coated with varnish A whatever the metal
ingredients (organic colourants).
treatment. Additional comparative tests conducted with
Based on the results presented in Figures 3 and 4, the varnish D (naturally more fluorescent) on two different
fluorescence colour obtained on brass was more similar metal finishes also did not show any difference.

Figure 4. Colorimetric data for the

  UV imaging
64  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

FTIR spectral regions. Finally, the presence of saffron was not

detectable in the SRefl-IR spectra of samples Sb_D and
SRefl-IR signatures obtained for some of the samples in
Mid_D2. This was doubtless because of the very low
Table 1 were compared with reference spectra from their
proportion of saffron in the recipe (less than 1 gram).
ingredients. Measurements were repeated for each sample
but no significant differences were noted between each
analysis. As showed in Figure 5 for varnish A (Mid_A2)
and shellac IR signatures, all spectral features of shellac
are recognisable in the varnish spectrum, confirming
the ability to characterise single component varnishes.

Figure 6. Representative SRefl-IR spectra obtained for varnish D on

mirror-polished and satin-finished brass compared with the reference
spectra of the ingredients used for varnish D (baselines have been
subtracted). The bands cited in the text are marked by *

Complementary analyses were carried out on the same

samples (Mid_A2 and Mid_D2) by gas chromatog-
raphy coupled to mass spectrometry (GC-MS). The
Figure 5. Representative SRefl-IR spectrum from the varnish A sample
compared with the reference spectrum for shellac, used for varnish A results obtained accorded with the recipes. Indeed, for
(baselines have been subtracted). varnish D, the compounds detected could be attributed
to elemi, sandarac and dragon’s blood. However, these
The reliability of SRefl-IR measurements on metal
compounds may also have come from a variety of other
surfaces appears to be independent of metal finish. This
resins. Therefore, it would be very difficult to attribute
is highlighted in Figure 6 by the similarity of spectra
them properly without the original recipe. At that stage,
between samples Sb_D and Mid_D2. Therefore, the
no gain in sensitivity was obtained by GC-MS, but this
finish of the object’s metal does not have a significant
will be further investigated on artificially aged samples.
influence on the effectiveness of IR reflection analysis.
In addition, it can be noticed that results are satisfying
Conclusion
despite the thinness of the varnishes (2–3 µm).
The tests allowed the chosen analytical techniques
Figure 6 presents the spectra of the components of
(eddy current, UV imaging and FTIR) to be validated
varnish D. The common features between compounds
for on-site characterisation of varnished copper alloys.
and varnishes in the SRefl-IR spectra can be observed.
The experiment was conducted on mock-ups allowing
The main bands of the principal ingredients, such as
possible varnish thickness parameters, particularly
elemi and sandarac, are distinguishable in the varnish
possible minima, to be taken into account. Eddy-current
spectra (bands at 1034, 1375 and 1460 cm-1 for elemi,
measurements proved to be an adequate non-invasive
and 888 and 1690 cm-1 for sandarac). The main bands
method to measure films even as thin as 2 µm on flat
of the compounds present in minor proportions, such
surfaces. However, the accuracy of similar measure-
as shellac and dragon’s blood, with the presence of bands
ments on more complex surface geometries needs
at 1242 and 1613 cm-1, respectively, were also identified.
further investigation.
Despite the small amounts involved (a few dozen grams),
the main bands of gummi-gutta and turmeric were also UV imaging of the samples confirms that this simple
identifiable in the varnish spectra, at 1633 and 1518 cm-1, tool is useful and efficient to detect the presence of
thanks to the absence of other interfering bands in these varnish on copper alloy. The resulting colour seems to
TRANSPARENT VARNISHES ON COPPER ALLOYS DATING FROM THE 19TH CENTURY: CHARACTERISATION AND IDENTIFICATION STRATEGIES
be less influenced by brass substrate than white metal.
Nevertheless, the varnish thickness may play a crucial
role in limiting detection and generates very weak fluo-
rescence. In this case, the imaging protocol needs to be
adapted. FTIR analysis shows that the components of
both simple and more complex varnish mixtures can
be identified in a conclusive way regardless of the type
of metal finish.
This paper presents the preliminary results of the LacCA
project, which will include analytical campaigns on real
varnished objects from four collections (Musée des Arts
Décoratifs, Paris; Lausanne Historical Museum, Inter-
national Museum of Horology, La-Chaux-de-Fonds;
and Musée des Arts et Métiers, Paris). Furthermore, the
parameters and effects of the substrate preparation and
composition will be investigated to determine the effect
of the alloy composition and the chemical etching of
the surface prior to varnishing. Finally, the varnished
samples will be aged artificially in order to mimic the
state of historical varnished objects.
Acknowledgements
The authors would like to thank the HES-SO University
of Applied Sciences and Arts of Western Switzerland
for funding the LacCA project and those in charge
of the museum collections who enabled the on-site
campaigns. We are also grateful to David Grange and
Stephan Ramseyer (Laboratory of Surface Engineering)
as well as Laure Jeandupeux (Laboratory of Micro and
Nanosystems) from Néode, HEI Arc.
Notes
1 Window size: 150 × 50 mm.
Materials list
Brass coupons
Kaufmann & Fils
www.kaufmann-fils.ch/
Shellac flakes, orange, no. 60410
Shellac flakes, decolourised, wax-free, no. 60450
Sandarac, no. 60100
Gum elemi, no. 62050
Mastic, no. 60050
Camphor, no. 78710
Turmeric, no. 37220
Saffron, no. 37110
65
Gummi-gutta, no. 37050
Dragon’s blood, no. 37000
Alcohol, no. 70800
Turpentine, no. 62000
Kremer Pigmente
www.kremer-pigmente.com/de
References
Cosentino, A. 2015. Practical notes on ultraviolet
technical photography for art examination. Conservar
Património 21: 53–62.
Lami, E.O. and A. Tharel. 1881–91. Dictionnaire ency-
clopédique et biographique de l’industrie et des arts indus-
triels, S–Z. Paris: Lami, Tharel & Cie.
Lanterna, G. and A. Giatti. 2014. Caratterizzazione non
invasiva delle vernici da ottone degli strumenti scientifici:
ricette storiche, realizzazione di provini verniciati, ricerca
analitica e applicazioni “in situ” su strumenti storici. OPD
restauro: rivista dell’Opificio delle pietre dure e laboratorio
di restauro di Firenze 26:165–180.
Londe, A. 1893. Aide-mémoire pratique de photographie.
Paris: J.-B. Baillière et Fils.
Long, D. 2000. The treatment of false gilding: A case
study. In Gilded metals: History, technology and conser-
vation, ed. Terry Drayman-Weisser, 319–27. London:
Archetype Publications Ltd.
Nosban, M. 1843. Nouveau manuel complet du menuisier,
de l’ébéniste et du layetier, tome 2. Paris: Roret.
Riffault, J.R.D.A., A.-D. Vergnaud, and C.-J. Toussaint.
1862. Nouveau manuel complet du fabricant de couleurs et
de vernis : contenant les meilleures formules et les procédés
les plus nouveaux et les plus usités dans ces différents arts,
vol. 2. Paris: Roret.
Rogge, C.E. and K. Lough. 2016. Fluorescence fails:
Analysis of UVA-induced visible fluorescence and false-
color reflected UVA images of tintype varnishes do not
discriminate between varnish materials. Journal of the
American Institute for Conservation 55(2): 38–147.
Sutherland, K. 2010. Bleached shellac picture varnishes:
Characterization and case studies. Journal of the Institute
of Conservation 33(2): 129–45.
Thomson, C. 1991. Last but not least – Examination and
interpretation of coatings on brass hardware. In AIC
Wooden Artifacts Group Postprints, Albuquerque, New
Mexico, 1991.
66  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND TECHNICAL STUDIES

Tingry, Y. 1803. Traité théorique et pratique sur l’art de
faire et d’appliquer les vernis, vols. 1 and 2. Geneva: G.
J. Manget.
Valicourt, E. de. 1872. Nouveau manuel complet du
tourneur, ou Traité théorique et pratique de l’art du tour,
contenant la manière de tourner les bois, les pierres et les
métaux, vol. 1. Paris: Roret.
Wellmer, E.-C. 1999. Goldfarbene Lacke in den Quellen-
schriften zur Maltechnik: Ein Überblick zur Geschichte,
Verwendung und Material. In Firnis: Material, Ästhetik,
Geschichte. Internationales Kolloquium, Braunschweig,
15–17 Juni 1998, ed. A. Harmssen, 43–62.
Her research focuses on conservation science, on both
the developments of new conservation protocols and on
the material study of cultural heritage.
Laura Brambilla has a PhD in chemical sciences from
the Università degli Studi di Milano, Italy. Since April
2013, she has worked in the R&D team at HE-Arc CR,
first as a scientific collaborator and from 2017 as a
professor. She is currently project leader for different
projects and one of the organisers of the ICOM-CC
Metal 2019 conference.

Authors
Julie Schröter has a master’s in conservation of cultural
heritage from the University of Paris 1 Panthéon-
Sorbonne where she specialised in metal conservation.
Since 2010, she has worked as a freelance conservator.
In 2014, she joined the research unit at Haute Ecole Arc
Conservation-restauration (HE-Arc CR) and is currently
executive project leader of the LacCA project, which
includes aspects of her PhD thesis dedicated to the histor-
ical and technological study of varnished copper alloys.

Aline Michel has a master’s in conservation with a

specialisation in technical objects from HE-Arc CR. Since
2014, she has worked as a self-employed conservator and
as a part-time research assistant at the Applied Research
and Development Division at He-Arc CR. She is currently
working on the CANS and LacCA projects.

Céline Paris is a CNRS engineer at the Monaris labora-

tory (CNRS-SU). Her main area is the characterisation
of organic-based materials from cultural heritage objects
by developing new non-destructive analytical method-
ologies to identify and differentiate materials.

Ludovic Bellot-Gurlet is a professor at the Sorbonne

University (SU, Paris) at the Monaris laboratory (CNRS-
SU). He researches the physico-chemical composition
of ancient materials and objects by using non-invasive
or non-destructive analytical techniques: Raman and
infrared spectroscopies.

Sigrid Mirabaud has a PhD in analytical chemistry from

the University of Lille. Until 2019, she was the head of
the research laboratory at the department of conservators
at the Institut national du patrimoine and is currently a
resident scholar at the Institut national d’histoire de l’art.
 ANALYTICAL
INVESTIGATION
Study of Ancient Metal Artefacts Using Neutron Imaging
Techniques, Visual Examination and Chemical Analysis

A. Duvauchelle D. Mannes E.H. Lehmann L. Andrey

Site et musée romains d’Avenches
Avenches, Switzerland
atduvauchelle@vd.ch
https://aventicum.org
M. Krieg*
Site et musée romains d’Avenches
Avenches, Switzerland
myriam.krieg@vd.ch
*Author for correspondence
Neutron Imaging & Activation Group
Laboratory for Neutron Scattering &
Imaging
Paul Scherrer Institute
Villigen, Switzerland
david.mannes@psi.ch
https://www.psi.ch
Neutron Imaging & Activation Group Site et musée romains d’Avenches
Laboratory for Neutron Scattering & Avenches, Switzerland
Imaging laura.andrey@vd.ch
Paul Scherrer Institute
Villigen, Switzerland S. Gillioz
eberhard.lehmann@psi.ch Site et musée romains d’Avenches
Avenches, Switzerland
V. Hubert sandra.gillioz@vd.ch
Schweizerisches Nationalmuseum
Affoltern am Albis, Switzerland
Vera.Hubert@nationalmuseum.ch
https://www.nationalmuseum.ch

Abstract
This paper reports on the study of two brass objects discovered
at the À la Montagne necropolis in Avenches, Switzerland. The
first one, a toiletry kit, is a composite object (metal and wood)
with a complex design. The second, a capsule bracelet, raises the
question of the function of the capsule and its content. Various
analyses (scanning electron microscopy/energy-dispersive
X-ray spectroscopy, Raman-spectroscopy and laser ablation-
inductively coupled plasma-mass spectrometry), imaging
techniques (neutron and X-ray tomography) and experts’ views
provided the possibility not only to identify the content of the
capsule bracelet, but also to determine the composition of both
objects, to distinguish the multiple pieces that constitute them
and to understand how they were designed.
Keywords
Aventicum, Roman archaeology, neutron and X-ray
tomography, brass, scanning electron microscopy/energy-
dispersive X-ray spectroscopy (SEM-EDS), Raman spectroscopy,
laser ablation-inductively coupled plasma-mass spectrometry
(LA-ICP-MS)

Introduction: The objects and the context of their discovery

The two brass objects presented in this paper were The toiletry kit was unearthed in a 1st-century AD pit.
discovered in 2001, during archaeological investi- Identified as a refuse or garbage pit, its fill contained
gations conducted in the À la Montagne necropolis unburned faunal bones and burned pottery sherds
(Figure 1). (Sauteur 2017, 206–07).

The fragmentary capsule bracelet comes from a crema-

tion burial in the same necropolis. This grave, dated
to the third quarter of the 1st-century AD, contained
an urn with the cremated bones of a young individual
of unknown sex, as well as different artefacts (pottery,
glass, nails, etc.) and the capsule bracelet fragment
(ibid., 211–12).

It seemed relevant to conduct a thorough study on these

Figure 1. Location of Avenches (Switzerland) and the À la Montagne
necropolis (red dot) (SMRA, D. Castella)
two objects. Indeed, not only does the toiletry kit belong
to a rare variant of this type of object, but it is also well-
enough preserved to allow careful and detailed observa-
tion. Regarding the bracelet, its main feature is a swollen
body hiding a compartment which could be opened.
STUDY OF ANCIENT METAL ARTEFACTS USING NEUTRON IMAGING TECHNIQUES AND VISUAL AND CHEMICAL ANALYSIS 69

Given the numerous hypotheses about its content and
the impossibility of physically opening it, neutron and
X-ray imaging were done at the Paul Scherrer Institute
(PSI) in Villigen (Switzerland).
Furthermore, both objects underwent non-destructive
and minimal invasive analyses to investigate the nature
of the materials that constitute them and to understand
their design.
Dina Klimentyeva and Dr Marcel Guillong at the Depart-
ment of Earth Sciences, Institute of Geochemistry and
Petrology, ETH Zurich. The black matter visible in the
upper part of the toiletry kit, previously documented by
EFI, was moreover subjected to Raman spectroscopy by
Dr Pierre Brodard and Samuel Roth at the Haute École
d’ingénierie et d’architecture Fribourg to determine its
molecular composition.4

Results and discussion

Materials and methods
Toiletry kit
Imaging
Macroscopic observation and comparisons
After macroscopic investigation, microscopic imaging
with optimised depth of field was done with an opto-dig- The toiletry kit consists of different parts (from top
ital microscope1 with extended focal imaging (EFI) to bottom): a ring, a decorated cubic support and two
by Valentine Brodard at the Haute École of conserva- rods with originally six sanitary instruments (only one
tion-restauration (Neuchâtel). This method was used preserved) (Figure 2). The ring is extended by a rod
to document the black matter in the upper part of the crossing a small black block (organic matter) caught
toiletry kit. between two copper alloy sheets (Duvauchelle and
Krieg 2017a).
In order to understand the design of both objects and
to discover the content of the capsule bracelet, a series
of X-ray images were taken by Markus Detmer at the
Archäologischer Dienst Kanton Bern. These images
– resulting from a two-dimensional data projection –
provided several answers, but more information was still
needed to grasp the design of the objects.

With X-ray and neutron tomography, the interior struc-

ture as well as the objects’ outer shape were visualised in
a three-dimensional reconstruction. The scanning was
carried out at the PSI using the thermal neutron radi-
ography beamline NEUTRA at the spallation neutron
source SINQ. The spatial resolution achieved was about Figure 2. Toiletry kit from Avenches. Brass and wood, length 77 mm, 17.3
g, inv. 01/11548-01 (SMRA, A. Schneider)
50 µm and permitted satisfactory interpretation of the
X-ray and neutron images, which often provide comple-
This toiletry kit matches a variant of the F type, according
mentary information (Deschler-Erb et al. 2004, Lehmann
to the classification established by A. Miron (1989).
et al. 2005).
Examples of this type are characterised by a decora-
tive element made of bone placed between two plates,
Chemical analyses
surmounting four balusters and four instruments,
To be able to describe the composition of the materials, hanging on two axes. Like the example from Avenches,
the objects underwent micro-analysis using scanning these types are dated to the second half of the 1st or
electron microscopy with energy-dispersive X-ray beginning of the 2nd century AD.
spectroscopy (SEM-EDS)2 by Christoph Neururer at the
The kit unearthed in the À la Montagne necropolis differs
Department of Earth Sciences, University of Fribourg.
from this corpus in two aspects. It has six rather than four
In order to get more precise information about the instruments, but most of all it does not possess the four
alloy compositions, laser ablation-inductively coupled balusters. Its closest parallel was discovered in Aosta in
plasma-mass spectrometry (LA-ICP-MS)3 was done by northern Italy (Framarin, Pinacoli, and Ronc 2014, 106).
70  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Determination of the matter
Copper alloy: SEM-EDS and LA-ICP-MS
Micro-SEM-EDS measurements were conducted on 13
different spots (square sheet, axe and double-eyebolt) of
the more or less corroded surface. The results are very
heterogeneous due to the condition of the surface and the
poor penetration depth. With the LA-ICP-MS method,
the alloy of the same pieces was analysed (15 spots). The
results, still quite inhomogeneous, confirm the pres-
ence of copper and zinc, which are brass con­stituents
(Figure 3). The zinc level, in most cases, is less than
10 weight% (LA-ICP-MS analysis). All of these varia-
tions are probably due to a dezincification of the alloy
(Papadopoulou et al. 2016; Scott 2002, 27–32).
Not all of the pieces of the kit have been analysed.
However, two clues may indicate that the same alloy was
used. Firstly, neutron tomography (cf. infra) shows that
they all have the same grey level. The colour scale is a
result of the attenuation coefficient and of the density of
Figure 3. Brass analyses. (a) Location of the LA-ICP-MS measurements;
(b) table summing up the brass elements according to the LA-ICP-MS
analyses (weight%, without the trace elements; Cu values are calculated
from a total of 100%; analysed spot size: 43 µm) (SNM, V. Hubert); (c)
one of the spectra obtained by SEM-EDS on the upper square sheet (400
µm × 200 µm, on the opposite side of spot α) (University of Fribourg,
Ch. Neururer)
ANALYTICAL INVESTIGATION
the matter (Mannes et al. 2015) and, in the present case,
it is most unlikely that the grey level matches different
associations of chemical elements. Secondly, the appear-
ance of the corrosion products, resulting from the metal
composition and the nature of the soil, is similar all over
the objects (Papadopoulou et al. 2016).
Black matter: EFI images, Raman spectroscopy and
the experts’ view
The determination of the black matter preserved between
the square sheets under the ring is more delicate. The
images created by EFI show lines on the matter, recalling
a very dense wood or bone (Figure 4).
Figure 4. Decorative element on the toiletry kit. Details of sides A and B
and location of the Raman analyses (EFI, HE-Arc C-R, V. Brodard)
Raman spectroscopy and the experts’ view
Three main substances were detected in the black cube:
amorphous carbon, represented by two broad peaks
around 1340–1360 cm-1 and 1580–1590 cm-1, phosphate
(PO43-) by a peak at 963 cm-1 and traces of carbonate
(CO32-) with peaks showing at 1070 cm-1 and 737 cm-1.
These results would thus indicate that the black matter
consists of bone, teeth or ivory, as the mineral part of
these substances is mostly composed of phosphate
(>70%) and calcium carbonate (~10%) (Morris and
Mandair 2011). However, the Raman analyses do not
allow these three materials to be differentiated as the
information contained in the organic parts that could
enable their identification (Paris 2004) was destroyed
during incineration. The amorphous carbon stems
precisely from the combustion of the organic matter,
be it vegetal or animal.
Finally, the toiletry kit was examined by optical micro-
scope at the Institut für Integrative Prähistorische
und Naturwissenschaftliche Archäologie (IPNA),
University Basel, by Dr Sabine Deschler-Erb (archae-
ozoologist) and the archaeobotanists Dr Örni Akeret
and Dr Angela Schlumbaum. This visual analysis
proved that the black matter does indeed derive from
STUDY OF ANCIENT METAL ARTEFACTS USING NEUTRON IMAGING TECHNIQUES AND VISUAL AND CHEMICAL ANALYSIS
a broad-leaf tree. Moreover, the presence of horizontal
lines on three sides of this block indicates that it was
probably assembled.
The dichotomy between these two different results can
be explained a posteriori. Raman spectroscopy was
conducted on the surface of the object. The toiletry kit
was discovered with faunal bones and was probably
burned on a funeral pyre. This could explain the traces
of phosphate.
The kit’s design: X-rays and neutron imaging
2D X-rays
A first series of X-ray images enabled the rejection of
the hypothesis of four balusters crossing the angles of
the wooden block. Instead, they show a fluted structure
in the centre, but it is hard to determine the number of
pieces composing it.
X-ray and neutron tomography
The visualisation of the toiletry kit by X-ray and neutron
tomographies finally allowed a 3D image to be created.
This provided a better distinction between the different
pieces constituting the toiletry kit and allowed a good
understanding of its assembling. Originally, it consisted
of at least 25 brass and wood elements (Figure 5).
Figure 5. Pieces composing the toiletry kit and assembly diagram (SMRA,
B. Raymond)
71
The assembly of the toiletry kit started with the upper
part. On the rod of the suspension ring (a) was placed
the decorated cubic support (b–e). The suspension axes
(f–g) were placed during the second assembly phase.
Upon them, three double-eyebolt shaped suspension
elements (i) and two spacers (h) were placed. The toiletry
instruments (j) might have been attached during this
process or later through slots in the eyebolts.
Most of these pieces were observed during the macro-
scopic stage. However, some details or pieces could only
be seen and understood thanks to the tomographies.
For instance, the hammered end of the ring rod is only
perceptible on the tomographic images. This implies a
conical hole in the bottom sheet. During hammering, the
matter caulked into that space created a small non-pro-
tuberant head whose flared shape assured the cohesion
of the pieces.
Figure 6. Fluted ring and wooden element of the toiletry kit. (a) 3D
view of one half; (b) horizontal section through the centre of the cubic
support. Neutron tomography (left) and X-ray tomography (right) (PSI,
E. Lehmann, D. Mannes)
The tomographies also revealed the existence of a tubular
structure, a kind of fluted ring, in the centre of the
wooden block (Figure 6). Judging by its grey level on the
neutron images, its material is probably brass. Hidden on
the example from Avenches, a fluted ring with a finely
incised horizontal band is visible on the kit from Aosta,
because no wooden block was preserved. However, the
function of this ring could be different on each kit. On
the example from Avenches, it could maintain a space
between the sheets and prevent the wooden block from
getting crushed during hammering of the suspension rod.
72  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND ANALYTICAL INVESTIGATION

It could also ensure a better adhesion of the wooden plates
composing the block, probably assembled with glue.
Adhesives like wax, resin or pitch containing hydrogen
are visible on neutron tomographies (Lehmann et al.
2005, 74). However, the fire would probably have erased
all traces of glue on the toiletry kit from Avenches.
Another issue is the structure of the fluted ring. In
longitudinal sections, it looks like a ball with a diameter
of about 5 mm marked by three horizontal lines (Figure
6a). The horizontal sections show a square perforation
(about 2 mm) but also other geometrical – square and
circular – shapes visible in the metallic mass, which
remain a mystery (Figure 6b).
Copper alloy determination: SEM-EDS and LA-ICP-MS
As for the toiletry kit, the nature of the copper alloy
was first approached using SEM-EDS. Measurements
were taken on the surface of the capsule (cup, bottom
sheet) and on the axe (22 spots in total). The indicative
percentages of copper and zinc (and sometimes tin) are
highly variable. The LA-ICP-MS analyses on 10 spots
(bottom sheet of the capsule, axe) confirm the presence
of copper and zinc (Figure 8). The dezincification of the
alloy probably explains the variability in the level of zinc
(Papadopoulou et al. 2016; Scott 2002, 27–32).

Capsule bracelet
Macroscopic observations and comparisons
A kind of calotte, extended by a rod, is the only preserved
fragment of this bracelet (Figure 7). The two parts of this
hinged box are connected by a globular-head rivet and
kept closed by a clasp (Duvauchelle and Krieg 2017b).

Figure 7. Capsule bracelet

from Avenches. Brass,
length 58 mm, 21.7 g. Inv.
01/11387-01 (SMRA, A.
 Schneider)
Some rare parallels can be found around Milan (Italy),
where they were probably manufactured. Unearthed in
funerary contexts – when the circumstances of discovery
are known – they are dated to around the middle of the
1st century AD.
Many hypotheses have been formulated as to the function
of these capsule bracelets (Giorgi, Martinelli, and Butti
Ronchetti 2010, 253–55). They could have carried coins,
amulets, vegetal substances, cloths impregnated with
perfume or even an unguent. Metallic elements making
a rattling noise could also be related to its function.
Figure 8. Analyses of the copper alloy. (a) Location of the LA-ICP-MS
measurements; (b) table summing up the alloy elements according to
the LA-ICP-MS analyses (weight%, without trace elements; Cu values
calculated from a total of 100%; analysed spot size: 43µm) (SNM, V.
Hubert); (c) one of the SEM-EDS spectra on the rim of the bottom sheet
(1 mm × 1 mm, on the opposite side of spot α) (University of Fribourg,
Ch. Neururer)
Design and content of the bracelet: X-ray and
neutron imaging
2D X-ray
A first series of X-ray images revealed two elements.
Firstly, the cup was shaped by hammering, leaving a
characteristic ‘cloud’ structure (Armbruster 2012, 385)
(Figure 9a). Secondly, two circular points close to the
STUDY OF ANCIENT METAL ARTEFACTS USING NEUTRON IMAGING TECHNIQUES AND VISUAL AND CHEMICAL ANALYSIS
clasp indicate the presence of a matter that strongly
reduces the effect of the X-rays. X-rays do not, however,
show if any organic substances are present.
Neutron and X-ray tomography
The design of the bracelet is easily understood with
the tomographies. The preserved fragment is made of
five pieces: the bangle that also comprises the cup of
the capsule, the sheet that closes it, the rivet ended by a
‘pearl’ that constitutes the hinge axis and finally a pin for
the clasp mechanism. As for the toiletry kit, the hinge
axis comprises a hemispherical-head rivet completed
by a riveted ‘pearl’ on the opposite side (Figure 9c).
Composed of a short rod, the clasp mechanism is very
simple (Figure 9d).
Figure 9. Technical characteristics of the capsule bracelet. (a) 2D X-ray
image (Archaeological Service, Canton Bern, M. Detmer); (b) 3D view
of the bottom part; (c, d) horizontal section at the level of the hinge (c)
and at the level of the clasp’s rod (d). Neutron tomography (b, c, d) (PSI,
E. Lehmann, D. Mannes)
Technical defects are perceptible on the neutron
tomography images. Microcracks, generated during the
manufacturing of the object, are visible in the cup and
the head of the rivet (Figure 9c). A small blister can be
seen on the sheet closing the capsule (Figure 9b). This
‘defect’ is probably the result of intense heat (over 500°C)
and would confirm its deposit on a pyre.
The main issue to resolve was the content of the capsule
at the time it was placed on the pyre. Neutron tomog-
raphy images proved that the capsule bracelet from the
À la Montagne necropolis in Avenches did not contain
a coin or an amulet. The hypothesis of a container for
scented herbs, cloth or unguent cannot be confirmed.
Any organic matter would have disappeared in the fire.
Finally, the images revealed the presence of four small
beads, two of which were caught in an indeterminate
matter stuck at the bottom of the cup (Figure 10).
73
Figure 10. Interior of the capsule bracelet. (a) Semi-transparent and
colourised reconstitution. In green, four metallic beads; in brown-yellow,
probably corrosion products. Two beads close to the cup and two
others at the bottom can be seen; (b) section and perspective. Neutron
tomography (PSI, E. Lehmann, D. Mannes)
These beads have a diameter of 1 to 2 mm and could
be composed of a copper alloy similar to the bracelet,
judging by their grey level. The two bigger ones are
nearly fused to their support. This may be the result of
the heat of the pyre. The unidentified matter sticking
to the cup appears porous and spongy on the neutron
tomography pictures (Figure 10b). In this case as well,
the grey level seems to indicate that it could be a copper
component similar to the one found in the bracelet.
These two features – appearance (Bertholon 2000, 262)
and grey level – allow this matter to be interpreted as a
cluster of corrosion products.
The beads contained in the capsule, although especially
small, would have made this jewel jangle, thus protecting
its owner like rattles or small bells frequently unearthed in
funerary contexts. The hypothesis of it being a sounding
bracelet would seem to be corroborated by this discovery
made in Avenches.
Conclusion
The thorough and multidisciplinary study of the toiletry
kit and of the capsule bracelet of Avenches/À la Montagne,
both made of brass, combined a traditional archaeological
approach with the expertise of specialists, chemical analyses
and very efficient imaging techniques, thus enabling the
skilful and meticulous work, as well as the ingenuity of
the craftsmen who created them, to be brought to light.
The main goal of the study was to resolve several issues
raised by the archaeologists, in particular the ‘innards’,
the fabrication technique and the material composition.
From the conservator-restorers’ point of view, the
analyses contributed to a better understanding of the
preservation state of the objects. Some damage could
be detected thanks to the exertion of advanced imaging
methods, which offered information invisible during
OM investigation.
74  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Currently, the objects have undergone only mechanical
cleaning and minimal local consolidation work. In total,
35 spots were analysed by SEM-EDS and only two – on
the toiletry kit – showed chlorides. So far, no chemical
treatment to isolate the chlorides has been conducted.
The objects are stored in a humidity-controlled depot
and show no sign of active corrosion.
Acknowledgements
Collaboration is the essence of this type of work. There-
fore, we would like to thank all the numerous institutions
and people who contributed to this study. In addition to
those already mentioned above, thanks are due to the
following persons: Alte Schmiede, Schliern b. Köniz:
Markus Binggeli, a craftsman specialised in the reproduc-
tion of ancient techniques; MAR - Museo archeologico
regionale, Aosta (Italy): Dr Maria Cristina Ronc, curator;
UMR 6298 ARTEHIS, Dijon (France): Michaël Brunet,
associate researcher; INRAP Hauts-de-France, Passel
(France): Dominique Canny, archaeologist; Site et musée
romains d’Avenches: Cécile Matthey, Bernard Reymond,
Daniel Castella, Aurélie Schenk, Wendy Margot and
Andreas Schneider, illustrators, archaeologists, worked
bone specialists and a photographer. We would like to
thank Natasha Hathaway (Archeo-access) for the trans-
lation of this paper.
Notes
1 Olympus DSX100 Opto-digital Microscope.
2 Scanning electron microscope from FEI XL 30 Sirion
FEG; electron gun: Schottky field emitter/detector;
microanalytical system (EDS) from EDAX equipped
with a lithium-doped silicon detector.
3 Laser ablation performed with an Australian Scien-
tific Instruments Resolution ArF excimer laser; ICP-MS
analyses with a Thermo Fischer Scientific Element XR
sector field ICPMS; data reduction and evaluation with
in-house developed software ‘sills’.
4 LabRAM HR800 (Horiba Jobin Yvon); microscope:
Olympus BX41; camera: Andor EMCCD DU970P
(PELTIER); laser: 632.7 nm, ‘confocal hole’ aperture: 800
μm, ‘entry slit’ aperture: 200 μm.
References
Armbruster, B. 2012. Goldgefässe der Nordischen
Bronzezeit – eine Studie zur Metalltechnik. Prähistor-
ische Zeitschrift 87: 370–432.
ANALYTICAL INVESTIGATION
Bertholon, R. 2000. La limite de la surface d’origine des
objets métalliques archéologiques: caractérisation, locali-
sation et approche des mécanismes de conservation. PhD
dissertation, Université Paris I Panthéon-Sorbonne,
U.F.R. d’art et d’archéologie, France.
Deschler-Erb, E., E.H. Lehmann, L. Pernet, P. Vontobel,
and S. Hartmann. 2004. The complementary use of
neutrons and X-rays for the non-destructive investiga-
tion of archaeological objects from Swiss collections.
Archaeometry 46(4): 647–61.
Duvauchelle, A. and M. Krieg. 2017a. Un nécessaire de
toilette de la nécropole d’Avenches/À la Montagne. L’ap-
port des analyses. Bulletin de l’Association Pro Aventico
57: 139–48.
Duvauchelle, A. and M. Krieg. 2017b. Un bracelet-étui
de la nécropole d’Avenches/À la Montagne. L’apport des
analyses. Bulletin de l’Association Pro Aventico 57: 149–56.
Framarin, P., S.P. Pinacoli, and M.C. Ronc, eds. 2014.
MAR, Musée Archéologique Régional Vallée d’Aoste,
Guide, Quart.
Giorgi, M., S. Martinelli, and F. Butti Ronchetti. 2010.
La necropoli romana di Rovello Porro (CO). Rivista
Archeologica dell’antica Provincia e Diocesi di Como
191–192: 53–288.
Lehmann, E.H., P. Vontobel, E. Deschler-Erb, and M.
Soares. 2005. Non-invasive studies of objects from
cultural heritage. Nuclear Instruments and Methods in
Physics Research A 542: 68–75.
Mannes, D., F. Schmida, J. Frey, K. Schmidt-Ott, and E.
Lehmann. 2015. Combined neutron and X-ray imaging
for non-invasive investigations of cultural heritage
objects. Physics Procedia 69: 653–60.
Miron, A. 1989. Toilettebestecke mit Scharnierkonstruk-
tion. Archaeologia Mosellana 1: 41–65.
Morris, M.D. and G.S. Mandair. 2011. Raman assess-
ment of bone quality. Clinical Orthopaedics and Related
Research 469(8): 2160–69.
Papadopoulou, O., P. Vassiliou, S. Grassini, E. Angelini,
and V. Gou. 2016. Soil-induced corrosion of ancient
Roman brass – A case study. Materials and Corrosion
67: 160–69.
Paris, C. 2004. Méthodologies spectroscopiques pour
l’étude de matériaux : Objets du patrimoine de la fin
du 19ème au début du 20ème siècle. PhD dissertation,
University Paris 6, France.
STUDY OF ANCIENT METAL ARTEFACTS USING NEUTRON IMAGING TECHNIQUES AND VISUAL AND CHEMICAL ANALYSIS 75

Sauteur, E., ed. 2017. À la Montagne. Une nécropole du
Ier siècle après J.-C. à Avenches. Cahiers d’archéologie
romande 167/Aventicum XXI. Lausanne.
Scott, D.A. 2002. Copper and bronze in art. Corrosion,
colorants, conservation. Los Angeles: Getty Publications.
Sandra Gillioz has a master’s in conservation-restora-
tion from the Haute Ecole Arc, Neuchâtel, Switzerland.
She specialises in archaeological objects and joined the
SMRA team in 2014. She also works freelance for insti-
tutions in Switzerland and Italy.

Authors
Anika Duvauchelle has a master’s degree equivalent
(Diplom) in Roman provincial archaeology from the
University of Lausanne, Switzerland. She has specialised
in the study of metal artefacts and debris. She works
freelance for several institutions in Switzerland, France
and Italy. Since 2011, she has been responsible for metal
artefacts at the Site et musée romains d’Avenches (SMRA).
Myriam Krieg has a master’s degree equivalent (Diplom)
in biology from the University of Fribourg, Switzerland,
and a master’s in conservation-restoration from the Bern
University of Applied Sciences, Switzerland. She has
worked at SMRA since 1999 where she has been head
of the laboratory for archaeological objects conservation
since 2013.
David Mannes has a master’s degree equivalent (Diplom)
in forest sciences from the Albert-Ludwigs-Universität
Freiburg, Germany. He finished his doctoral thesis in
the department of building engineering at ETH Zurich,
Switzerland. Since 2010, he has worked as a researcher in
the neutron imaging and applied materials group at PSI.
Eberhard Lehmann has a master’s degree equivalent
(Diplom) in physics from the University of Leipzig,
Saxony, Germany. He completed his doctoral thesis at the
Nuclear Research Centre, Rossendorf (near Dresden), on
the physics of fast breeder nuclear reactors. In 1991, he
joined PSI as a reactor physicist and later moved to the
Spallation Neutron Source department where he initiated
the neutron imaging techniques and created the Neutron
Imaging and Activation Group.
Vera Hubert has a PhD in chemistry from the Free
University in Berlin, Germany. Since 2005, she has
worked for the Swiss National Museum in the field of
conservation science. She specialises in materials analysis
for archaeological and metal cultural heritage objects.
Laura Andrey has a master’s in conservation-restoration
from the Haute Ecole Arc, Neuchâtel, Switzerland. She
specialises in archaeological objects and worked at the
Ufficio dei Beni culturali in Bellinzona for four years,
joining SMRA in 2011.
Evidencing Risk to Inform Best Practice: The Limitations
of Detecting Tinning Layers by Visual Analysis During the
Removal of Corrosion from Archaeological Iron
David Watkinson* Michelle Crepeau
School of History, Archaeology and School of History, Archaeology and
Religion – Cardiff University Religion – Cardiff University
Cardiff, UK Cardiff, UK
watkinson@cf.ac.uk mcrepeau00@gmail.com
Abstract
Evidence for tinning on corroded archaeological iron is visible in
x-radiographs as distinct thin, high-density white lines. These are
used to guide the removal of overlying iron corrosion to reveal
the tin layer. Micro-abrasive blasting with 53 µm aluminium
oxide, at low pressure and powder flow, and guided by optical
microscopy was used to remove iron corrosion incrementally
from a tinned medieval key. At selected points during the
cleaning, SEM-BEI imaging with EDX spot analysis was used to
determine if the tinning layer had been reached. The presence
of tinning was confirmed by SEM-EDX spot analysis. Comparing
optical assessment with SEM-BEI images revealed that the
occurrence of tinning could not be detected with the naked eye
aided by optical microscopy. No metallic tin was observed and
Introduction
Context for study
Dissimilar metal coatings such as silver, tin and copper
have been applied historically to iron to decorate and/or
protect objects (Oddy 1980, Meeks 1993). Construction
methods such as brazing may also leave thin coatings
of metal on iron. On archaeological iron, these thin
coatings are normally obscured by overlying corro-
sion products and x-radiography is required to detect
them. The qualitative nature of radiography means
the coatings are identified by the position, density and
thickness of the lines they produce on the x-ray plate.
Potential misidentification can occur, as dense corrosion
product layers can induce similar thin dense lines on
radiographs.
When removing overlying corrosion, identification of
the dissimilar metal is by eye, often with the aid of a
microscope. Expectation is that tinning will appear as a
thin metallic silver coloured layer or a distinct grey tin
dioxide (SnO2), copper or brazing as an orange/yellow
Nicola J. Emmerson *Author for correspondence
School of History, Archaeology and
Religion – Cardiff University
Cardiff, UK
emmersonnj@cf.ac.uk
tin is presumed to be present as tin dioxide (SnO2). Inability to
see the tinning layer meant parts of it were lost during micro-
abrasive blasting. This evidences the risks associated with
mechanically removing overlying iron corrosion products to
expose tinning layers, and these risks should be incorporated
into developing conservation strategies for preserving tinning
layers on iron, where elemental surface analysis is required to
confirm its presence.
Keywords
iron, tinning, archaeological, corrosion, micro-abrasive
blasting, radiography, SEM-BEI
metallic layer or as green basic copper carbonates or
orange copper (I) oxide corrosion product layers, silver
as a metallic layer or black silver sulfide, and gilding as
shiny metallic gold. If it is not possible to detect these
layers visually during removal of the overlying iron
corrosion, they may be removed accidentally, leading
the conservator to conclude they were never present
and the white lines on the radiograph were in fact
dense corrosion products, misidentified as a dissimilar
metal. Since no study has attempted to align incremen-
tally what is visible on the object during cleaning with
semi-quantitative analysis of the surface, the question
remains whether these layers could be readily missed
and inadvertently removed by the conservator during
cleaning. Insight into how difficult it is to identify
layers will inform best practice on cleaning iron objects
thought to have tinning layers and will identify the level
of risk associated with the cleaning process. This study
reports cleaning and analysis of a medieval key that was
identified as being tinned by x-radiography.
EVIDENCING RISK TO INFORM BEST PRACTICE: THE LIMITATIONS OF DETECTING TINNING LAYERS BY VISUAL ANALYSIS DURING THE REMOVAL OF CORROSION FROM ARCHAEOLOGICAL IRON
Tinning in antiquity
Application of tin coatings to copper alloys is mentioned
in the writings of Pliny the Elder in his Naturalis
historia, but there is no parallel recording of tinning of
iron, although it is identified as occurring in Europe
from 500 BC and from the fifth century AD in Britain
(Meeks 1993, 249). Fusion plating of tin to wrought iron
uses heat to create an intermetallic phase that binds it
to the iron surface (Dionisio et al. 1985; Meeks 1993,
261). Application can involve either applying rosin as a
flux to clean the iron then wiping tin or tin/lead solder
filings melted in-situ over the metal surface, or hot-dip-
ping the iron in a liquid tin bath. The low solubility
of tin in other metals produces a very thin two-phase
intermetallic layer (<1 μm); FeSn (η = 68% Sn) and the
predominant thermodynamically stable phase FeSn2
(θ = 81% Sn) (Dionisio et al. 1985). The excess layer of
metallic tin sits on top of this thin intermetallic phase.
Adhesion to the metal can be improved by scoring of
the metal to key the tin onto the surface.
Corrosion of tin layers on wrought iron
The electrode potential of tin (-0.136 V) makes it
cathodic to iron (-0.44 V), consequently it might be
expected to survive in burial contexts sandwiched
between iron corrosion products. However, once
isolated in this way, it can corrode, provided oxygen and
moisture can access to it, with grey/white SnO2 as the
thermodynamically favoured product (Turgoose 1985).
Since SnO2 is highly insoluble, except in strong acids
and alkalis, it remains immobile where it was formed.
Studies of corrosion on archaeological bronzes show
that this immobility allows it to act as a marker layer for
original surface on these objects (Piccardo et al. 2007).
Consequently, oxidised tinning layers may be expected
to provide clear markers for the original surface on an
archaeological iron object. Lead (MP = 327.5°C) and
tin (MP = 231.7°C) may be mixed to produce tinning
layers. Its electrode potential (-0.126 V) is close enough
to that of tin for each metal to corrode independently
when mixed together (Turgoose 1985). Although
the lead compounds are slightly more soluble than
SnO2, a dense corrosion product layer is expected to
occur because of the immobility of tin. This is readily
detectable by x-radiography, as the atomic weights of
tin (118.7) and lead (208.2) greatly exceed that of iron
(56). Corrosion products on archaeological iron have
77
been characterised extensively (Neff et al. 2007). If the
tinning does not migrate from its original position
during corrosion, it will sit on the dense corrosion
product layer (DPL) (Bertholon 2001) and will be over-
laid with the less dense, porous and soft transformed
corrosion matrix (TM) that incorporates soil particles
within it. The DPL likely comprises mostly α-FeOOH,
Fe3O4 and γ-Fe2O3, while the altered layer will mostly be
α-FeOOH, but can contain many other iron corrosion
products. The DPL is normally grey/black due to the
Fe3O4 distributed within it.
Radiography of iron and micro-abrasive blasting
of corrosion products to expose tinning layers
X-radiography is commonly used to examine all iron
from archaeological excavations to determine its shape,
assess its condition, detect decoration and identify
technological processes used in its manufacture (Fell
et al. 2006). This qualitative technique relies on the
knowledge and skill of the operator to detect features
in objects. Differential adsorption and filtering of
the x-rays occurs within the object according to their
wavelength and the density and thickness of materials
that comprise the object (O’Connor 2007). A fine dense
white or light grey line will appear due to tinning,
brazing, copper, silver and gilding layers isolated from
the iron surface by corrosion products, provided their
atomic weight significantly exceeds that of metallic iron
and its corrosion products. Although they are often
thicker and less intense than tinning lines, misiden-
tification may occur if lines occur due to very dense
corrosion product layers forming at the original metal
surface. The skill of the conservator is essential for the
identification process. Where tinning has been lost
from a surface, traces frequently survive in corners
and undercuts where it has pooled into a thicker layer
during the application process.
A frequently applied procedure for revealing dissimilar
metal layers is the removal of overlying corrosion prod-
ucts using micro-abrasive blasting, which has been shown
to be a very sensitive and controllable tool (Farooq et al.
2011). This technique uses small particles of aluminium
oxide or other abrasives expelled under pressure from
a nozzle to cut the corrosion product from the object
surface. Detection of the tinning layer relies upon the
machine operator referencing the radiograph of the
object, while using a microscope to examine the surface
78  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
of the object and their expertise to detect the layer as it
is uncovered. With such thin coating layers, this can be
challenging, even when they retain metallic tin; when
they exist as corrosion products, it is exceptionally diffi-
cult, as the SnO2 is similar in colour to the α-FeOOH/
Fe3O4/γ-Fe2O3 DPL matrix that would normally underpin
it. Consequently, it would be easy to remove the SnO2
layer without detecting a colour or physical change in
the corrosion matrix. Morphologically, there is little
information on the nature of the SnO2 layer that might
aid in its detection. It might be expected to be softer and
less coherent than an iron corrosion matrix containing
dense Fe3O4, but infusion of dense iron corrosion prod-
ucts into the layer could contribute to improved hard-
ness. Currently, there is no research available to identify
whether a SnO2 layer is infused with iron corrosion, nor
is there data on its hardness and colour.
Experimental aims and objectives
Aim:
• to provide an evidence-based assessment of the effec-
tiveness of visual inspection as a method for detecting
the presence of tinning layers on archaeological iron
objects;
Objectives are to:
• use radiography to tentatively identify tinning layers
on a medieval key;
• remove overlying corrosion products incrementally
by micro-abrasive blasting and record the surface as
cleaning progresses using visual inspection, micro-
photography, SEM-BEI imaging and SEM-EDX spot
analysis to detect elemental composition;
• compare the results from visual interpretation to those
from elemental composition;
• assess the degree of certainty attached to using visual
inspection for determining the end point when
removing corrosion products overlying tinning layers.
Method
Sample selection
The sample is a late (14th–15th century) medieval key
from the deserted medieval village of West Whelpington,
Northumberland, UK (Evans and Jarrett 1987). Radiog-
raphy (Faxitron 438505, 110 kV) identified a tubular shaft,
designed to accept the guide pin in a lock, with extensive
mineralisation evidenced by the low overall density of the
ANALYTICAL INVESTIGATION
bow, tube walls and teeth (Figure 1). Lamellar structures
in the bow evidence slag planes from forging. There
are multiple areas where a dissimilar metal (henceforth
referred to as tinning) appears to be present as a coating
and potentially as decoration in incised areas of the key.
It is overlaid with a thick corrosion product layer and
records from the 1970s revealed this had received some
smoothing by mechanical cleaning, evidenced by marks
from a rotating micro-abrasive wheel, which revealed its
general shape but no evidence of tinning. The use of a
‘mini-drill’ with rotary burrs was a popular method for
removing corrosion in the past; it was applied prior to
the introduction of the more sensitive micro-abrasive
blasting technique to reveal detail.
Figure 1. Medieval key from the West Whelpington deserted medieval
village before micro-abrasive blasting, and radiograph showing hollow
barrel construction with tinning residues. Areas 1–6 were selected for
micro-abrasive blasting to expose tinning layers. The dotted line for area
6 indicates it is an edge, as the tooth was broken at that point and was
removed to reveal a cross section
The West Whelpington site is published and permission
was obtained for destructive investigation. The ethical
rationale behind this decision is clear: information
gained from this study can be used to guide protocol for
the mechanical cleaning of similar objects. Improving
understanding of the relationship between corrosion
processes, radiographic imaging and visual detection of
dissimilar metals will contribute to devising corrosion
removal strategies that limit loss of evidence and inform
decisions to clean objects or use only radiographs for
interpretation and publication due to the risk arising
from cleaning.
EVIDENCING RISK TO INFORM BEST PRACTICE: THE LIMITATIONS OF DETECTING TINNING LAYERS BY VISUAL ANALYSIS DURING THE REMOVAL OF CORROSION FROM ARCHAEOLOGICAL IRON 79

Investigative method since it would have to be removed before recommencing

micro-abrasive blasting. The data was ZAF corrected
Using a Faxitron 43805 x-ray machine, a series of side
and normalised. No calibration standard was used,
and dorsal radiographs of the key were taken from
but acquisition rates and dead time were maintained
0­–5 minute exposure values, at 30-second intervals
at 3000 and 20–50, respectively. Since the SEM was
for 85, 95, 100 and 110 kV to determine the location
employed in spot analysis mode as a detection tool for
of the tinning (Figure 2). Results identified optimum
presence or absence of tinning, the 1–10 wt% detection
exposure values for kV and time to record the tinning
was acceptable. The SEM detects elements rather than
(90 kV, 3 min, 3 mA constant current) and informed
compounds, so any confirmation that the tin was still
selection of six areas of tinning where the key would
metallic was determined by stereomicroscopy. EDX spot
be mechanically cleaned (Figure 1). This employed a
Texas Airsonics AJ-1™ Spirofeed Airbrasive® machine analysis of tinning sites was guided by the BEI images.
to deliver 53 µm diameter aluminium oxide particles
at pressures ranging from 20 to 60 psi at powder flow Results
level 4 (full scale 1–10). The powder was fed at a The key is extensively mineralised, particularly in the
working distance of 2 mm–6 mm and an angle of 45° areas of the bow and teeth (Figure 2 and 3). There is
for optimum effect (Farooq et al. 2011) and the operator radiographic evidence of a dissimilar metal presence as
used a ×10 stereo microscope to view the object during a coating, within incised lines and as pooling in right-an-
the cleaning process. After micro-abrasive blasting, the gled areas where the teeth meet the shaft, caused during
key was washed in baths of benzyl alcohol to remove its application process. There are also incised cuts visible,
aluminium oxide particles lodged on the surface of particularly on the tooth closest to the main body of the
the key, as their presence reflects light that mimics the shaft. These may be deliberate additions to aid keying
appearance of a metallic coating under the microscope. onto the surface of the iron. The abrasion technique
Removal of iron corrosion was guided by known corro- employed in this study effectively cut ‘trenches’ into
sion and technological criteria, therefore mimicking real
practice. Its progress was determined at selected inter-
vals by visual inspection and recorded by radiography,
photomicroscopy, scanning electron microscopy (SEM)
and energy dispersive x-ray analysis (EDX).

Radiography used the optimum parameters identified

in the radiography calibration previously described.
Stereomicroscopy, over a magnification range of ×4 to
×40, was carried out using a Nikon SMZ1000 equipped Figure 2. Radiographs of key ventral and left profile at 110, 100, 95 and
with a Nikon Coolpix 4500™ camera supported by NIS 85 kV (5-minute exposure)
Element D 3.0™ software for digital image recording.
SEM employed a CamScan 2040 SEM with Oxford
Link Pentafet 5518 dispersive x-ray spectrometer
and Maxim™ imaging and INCA analysis software
packages. Backscattered electron imaging mode (BEI)
was used to image cleaned areas, and spot analysis
with EDX determined elemental composition at the
surface. Since higher atomic weight elements produce
greater contrast, tin or SnO2 would be visible against
the iron oxides around them. The use of carbon coating
for SEM imaging was avoided by use of EnVac mode Figure 3. Radiographs recording incremental removal of iron corrosion
by micro-abrasive blasting in designated investigation areas 1–6 (see
with a 20 keV beam current. Although carbon coating
Figure 1). Image A records the key before micro-abrasive blasting, with the
increases conduction, lessens charge and improves broken end tooth in place on the bit. Images B to E record progressively
image quality, its use would have been problematic the removal of corrosion overlying the suspected tinning areas
80  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 4. SEM-BEI images and EDX spectra of tinning in areas 1, 3, 4, 5
and 6 (see Figure 3) following completion of micro-abrasive blasting. All
images are recorded at the same magnification (1 mm scale in image 1)
Figure 6. Optical microscopy, SEM-BEI images and radiographs of
site 1 (decorative lines on the shaft). Image A is the area to be micro-
abrasive blasted marked in china pencil. Images B and C record
progress of micro-abrasive blasting. Image D is a radiograph of site 1
after micro-abrasive blasting, showing tinning lines have been lost
during removal of corrosion. Image E confirms that the remains of
the three tinning lines visible in BEI image F are undetectable using
optical microscopy
the corrosion layers in search of the tinning (Figures
3—7). BEI images and EDX analysis of selected areas
of the key confirm the layer seen on the x-radiograph is
tinning (Figure 4). Optical microscopy of tinned areas,
BEI imaging and radiographs confirm the location of
tinning layers (Figures 5–7).
ANALYTICAL INVESTIGATION
Figure 5. Optical microscopy, SEM-BEI images and radiographs of
site 5 (flat face of broken teeth on the key bit). Image A is the optical
microscopy image, where the circle delineates area 5 after micro-
abrasive blasting and the rectangle identifies the area recorded in optical
microscopy (image C) and BEI (image D). Image B is a BEI image of the
teeth. The rectangles within the radiograph in images E and F identify
in detail area 5 before and after micro-abrasive blasting, respectively;
no evidence of tinning on the face of the teeth is visible either before
or after micro-abrasive blasting
Figure 7. Optical microscopy (A), SEM-BEI images (B) and radiographs (C) of
site 3 (shaft) and sites 4 and 6 (bit teeth) following micro-abrasive blasting
Discussion
Although micro-abrasive blasting was carried out with
the aid of a microscope, guided by radiographs and using
low pressure and powder feed for controlled corrosion
removal, identifying an end point that exposed tinning
EVIDENCING RISK TO INFORM BEST PRACTICE: THE LIMITATIONS OF DETECTING TINNING LAYERS BY VISUAL ANALYSIS DURING THE REMOVAL OF CORROSION FROM ARCHAEOLOGICAL IRON
layers was difficult, as no tinning was seen by eye during
this process, either as a shiny silver-coloured metal or
a grey oxide. Subtle differences between a grey SnO2
layer and the iron corrosion products in the DPL and
TM would be difficult to observe, especially if the SnO2
contains iron corrosion within its matrix.
Even when using optical microscopy guided by SEM-BEI
images, it was not possible to detect the tinning layer
visible on the SEM-BEI images that was confirmed as
tin by EDX spot analysis (Figure 4). The tinned areas
are discontinuous and very small (<1 mm), although
more tinning may lie undetected below overlying iron
corrosion; alternatively, the gaps in continuity may be
due to tinning having been inadvertently lost during
micro-abrasive blasting due to invisibility of SnO2 to the
naked eye. The thinness of tinning layers makes it easy
to pass through them during micro-abrasive blasting,
even if they are metallic and readily detectable by eye.
Spot analysis detected the presence of small amounts of
lead in all the tinning layers.
Comparing SEM-BEI and optical microscopy images, it
is clear that the corroded tinned areas are undetectable
by eye, despite the aid of a microscope (Figures 5–7).
Figure 5 records the flat face of the teeth that is area 5
on the key and, while tin is detected in BEI images (5.B
and C), no metallic tin is visible after micro-abrasive
blasting (5.A and C) and the SnO2 present cannot be
differentiated from the iron corrosion. Radiographs
(5.E and F) do not show the tinning as a distinct line to
guide the conservator during micro-abrasive blasting
to remove corrosion, since they are taken at 90° to the
surface, recording any tinning at its thinnest point and
effectively making it ‘invisible’ on the radiograph. In
contrast, dense white lines occur at object edges because
their thickness is effectively increased by the depth of
the object. In Figure 6, the three distinct tinning lines on
the shaft (area 1, 6.D) are marked out with grease pencil
(6.A) for micro-abrasive blasting (6.B, C and E). The three
lines are invisible to the eye in the micro-abrasive blasted
area, but traces of them are detected in BEI (6.F), where
it appears that micro-abrasive blasting had inadvertently
cut through them without this being detected by eye
during removal of iron corrosion. Invisibility to the eye
while being detected by BEI is repeated for tinning layers
across the object (Figure 7).
The study reveals the potential difficulty of identifying
tinning by visual means as an object is cleaned. The
81
radiographs were highly informative, yet the corroded
tinning was impossible to see with the aid of optical
microscopy. For some samples, two conservators with
a combined 40 years of experience and a student could
not detect evidence of tinning using optical microscopy.
Even with the aid of a BEI image to confirm the location
of the tinning, it was rare that observers could identify
colour or morphological differences that differentiated
the areas of tinning from the general surface of the object.
That the tinning may not be visible to the eye has to be
taken into account when attempting to reveal tinning
layers on iron. A second publication based on this key
will report morphology and composition of the corroded
tinning layers.
Conclusion
Evidence from this study indicates that using micro-abra-
sive blasting to reveal tinning layers on archaeological
iron, guided by radiography and optical microscopy,
is a high-risk exercise. Without tinning surviving in a
metallic state, its visual detection was not possible. It is
almost certain that evidence for tinning will be lost during
removal of iron corrosion because it cannot be seen and
the layers are thin. SEM-BEI imaging can be used to
detect tin and confirm technological processes, but its use
does not remove the risk of tinning loss during cleaning.
Management discussions should include consideration
about whether attempts to reveal tinning layers should
be part of the conservation plan. Localised cleaning and
SEM-BEI spot analysis to confirm the presence of tin for
reporting purposes will be a less destructive approach,
leaving the remainder of the object uncleaned for future
reference.
Acknowledgements
Thanks to Dr David Evans for releasing the West Whelp-
ington key for research purposes.
References
Bertholon, R. 2001. Characterisation and location of
the original surface of corroded archaeological objects.
Surface Engineering 17(3): 241–45.
Dionisio, P.H., Jr. B.A.S. de Barros, and I.J.R Baumvol.
1985. Comparative study of intermetallic phases formed
by direct ion implementation and radiation enhanced
diffusion of tin in two kinds of steel. Journal of Applied
Physics 58(2): 773–78.
82  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND ANALYTICAL INVESTIGATION

Evans, D.H. and M.G. Jarrett. 1987. The deserted medi- career highlights include the Çatalhöyük Archaeological
eval village of West Whelpington, Northumberland, Research Project, The Colonial Williamsburg Foundation
third report, part one. Archaeologia Aeliana (5th Series) and the Queen Anne’s Revenge Project out of Beaufort,
16: 199–308. North Carolina, USA.
Farooq, I., Z. Imran, and U. Farooq. 2011. Air abrasion: Nicola Emmerson is a lecturer in conservation at
Truly minimally invasive technique. International Journal Cardiff University, with research interests in the corro-
of Prosthodontics and Restorative Dentistry 1(2): 105–7. sion and conservation of ferrous metals within heritage
Fell, V., Q. Mould, and R. White. 2006. Guidelines on the contexts.
X-radiography of archaeological metalwork. Swindon:
English Heritage.
Meeks, N. 1993. Surface characterisation of tinned
bronze, high-tin bronze, tinned iron and arsenical bronze.
In Metal plating and patination – Cultural, technical and
historical developments, eds. S. La Niece and P. Craddock,
247–75. London: Butterworths.
Neff, D., E. Vega, P. Dillmann, and M. Descotes. 2007.
Contribution of iron archaeological artefacts to the
estimation of average corrosion rates and the long-term
corrosion mechanisms of low carbon steel. In Corrosion
of metallic heritage artefacts: Investigation, conservation
and prediction of long-term behaviour, eds. P. Dillmann,
G. Beranger, P. Piccardo, and H. Mathiesen, 41–74.
Cambridge, UK: Woodhead Publishing.
Oddy, A. 1980. Gilding and tinning in Anglo-Saxon
England. In Aspects of early metallurgy, ed. W.A. Oddy,
129–34, British Museum Occasional Paper 17. Londn:
British Museum.
Piccardo, P., B. Mille, and L. Robbiola. 2007. Tin and
copper oxides in corroded archaeological bronzes. In
Corrosion of metallic heritage artefacts: Investigation,
conservation and prediction of long-term behaviour, eds.
P. Dillmann, G. Beranger, P. Piccardo, and H. Mathiesen,
239–62. Cambridge, UK: Woodhead Publishing.
Turgoose, S. 1985. The corrosion of lead and tin: Before
and after excavation. In Lead and tin: Studies in conser-
vation and technology, eds. C.E. Miles and S.C. Pollard,
15–26, United Kingdom Institute for Conservation,
Occasional Paper No. 3.

Authors
David Watkinson is Professor of Conservation at
Cardiff University. His work on the corrosion rate of iron
underpins the preservation of Brunel’s iron steamship
SS Great Britain.

Michelle Crepeau is a graduate of the MSc programme

in Conservation Practice at Cardiff University. Her
An Overview of the Use of EIS Measurements for the
Assessment of Patinas and Coatings in the Conservation
of Metallic Cultural Heritage
Blanca Ramírez Barat* Paola Letardi
CENIM–CSIC IAS–CNR
Madrid, Spain Genoa, Italy
blanca.ramirez@csic.es paola.letardi@cnr.it
Abstract
Electrochemical impedance spectroscopy (EIS) is a technique
widely used in the field of corrosion science and for coatings
evaluation. Its application to the field of cultural heritage started
in the early 1990s. Since then, around a hundred references have
appeared in conservation and scientific literature addressing
different topics such as corrosion processes, field measurements,
conservation assessment, and protective coatings evaluation. All
of these research results constitute highly valuable information
Introduction
In recent decades, a large number of studies have been
published assessing the application of protective treat-
ments to the surfaces of metal artworks (Price et al. 1995,
Brostoff 2003, Argyropoulos et al. 2007, Joseph et al.
2007, Masi et al. 2019). Because of the very wide range
of metal alloys, surface finishes, and exposure history, a
general view is still missing. Moreover, some products
used in the past may disappear from the market and need
to be replaced, while several studies on coatings for use
in cultural heritage failed to consider that protective
treatments may work differently on different substrates
and/or in different environments.
Electrochemical impedance spectroscopy (EIS) has been
recognized as a powerful technique for testing coatings
since the second half of the 20th century. Its introduction
into the field of heritage science has brought a great
improvement in research and testing on coatings for
metal artwork surfaces, allowing an assessment based
on the comparison of impedance values and opening
the way to the monitoring of their evolution over time.
EIS is based on the application of a small sinusoidal
voltage (V) around the equilibrium potential (Eoc) and
the measurement of the corresponding current (I) over a
wide range of frequency; the ratio Z=V/I is the imped-
ance, a complex function of frequency. EIS spectra can
be represented in Cartesian coordinates by the modulus
Emilio Cano *Author for correspondence
CENIM–CSIC
Madrid, Spain
ecano@cenim.csic.es
for metal conservators. This paper addresses the state of the art
on available setups for EIS measurements in heritage science
and presents an overview of the results on different coatings
and substrates.
Keywords
electrochemical impedance spectroscopy, metal, coating,
patina, experimental setup
|Z| and phase angle Φ (Bode plot), or in polar coordinates
with real and imaginary parts (Nyquist plot). While for
more in-depth analysis of the spectra equivalent circuits
(EC) are proposed and the values for the relevant electro-
chemical parameters derived by best fit, the maximum if
the impedance modulus is at low frequency (|Z|max) can
be adopted as a rough indication of the corrosion rate
(higher |Z|max corresponds to a lower corrosion rate). For
further reading, the basics of the technique with special
focus on cultural heritage can be found in Letardi (2000
and 2013) and Cano, Lafuente, and Bastidas (2010).
Since the early history of EIS in heritage science, two
main approaches have been followed: measurements
on coupons to address conservation issues (mainly
protective treatment testing) (Price et al. 1995); and
development of a measurement setup for direct use on
metal artworks (Letardi et al. 1998). Research activity
in this field has been growing exponentially, as pointed
out in the published reviews (Cano, Lafuente, and
Bastidas 2010, Ramírez Barat and Cano 2018), and
more recently field studies on monuments have been
published (Sansonetti et al. 2015, Arceo-Gómez and
Zambrano-Rengel 2016, Ramírez Barat et al. 2017)
with a few comparisons of results between coupons and
heritage objects (Bruni 2018, Letardi, Albini, and Joseph
2018). EIS has also been applied to compare cleaning
84  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
methods on bronze outdoor monuments (Sansonetti et
al. 2015, Letardi et al. 2016).
This paper addresses the state of the art of EIS measure-
ments in heritage science and gives an overview of field
measurement setups and results on different substrates
and protective treatments. A compilation of EIS studies
on different coatings is presented at the end of the
document with a brief overview of waxes and acrylics.
Field measurement setups
The equipment used for EIS measurements includes an
instrument to apply and measure the electrical signals
(potentiostat), an electrochemical cell (comprising
the electrolyte and electrodes), and a system to fix the
electrochemical cell on the rough, curved, irregular and
non-horizontal surfaces of heritage objects during in-situ
measurement (Figure 1).
The choice of electrolyte should provide a balance
between a reasonable reproduction of the corroding
Figure 1. Overview of the general setup used for in-situ EIS measurements,
including the potentiostat, cells, and holding devices
ANALYTICAL INVESTIGATION
condition of the object under investigation, an aggres-
siveness towards the object’s surface that is as low as
possible, and reasonable conductivity, allowing a good
signal-to-noise ratio and stable measurement. Artifi-
cial rain is a widely adopted solution (Letardi 2013,
Ramírez Barat, Cano, and Letardi 2018) that fulfills
these requirements.
Commercial portable potentiostats are nowadays avail-
able with a floating ground that are capable of taking
field measurements (Ramírez Barat, Cano, and Letardi
2018), and several cell setups have been designed and
tested, mainly based on a three-electrode configuration.
This configuration consists of a working electrode
(WE), which is the system under study, a reference
electrode (RE) to measure the Eoc, and a counter elec-
trode (CE) to collect the current flow (Figure 2). This
allows the Eoc to be measured and the measurement
to be controlled in order to avoid any alteration to the
system under study (NDT conditions). Among other
proposed setups, a contact probe (CP) and gel-polymer
electrolyte (G-PE) cell (Figure 2) are––at present––
more widely used, and therefore our discussion will
focus on these two setups for field measurements. A
more general overview may be found in Ramírez Barat
and Cano (2018).
Figure 2. EIS measurement setup used in field measurements on
artworks: (a) the contact probe (CP) consists of 316 L stainless-steel
cylinders embedded in PTFE (section on the left); the electrolyte wets
the WE surface by means of a soaked cloth in contact with an electrolyte
reservoir (right image). (b) The gel-polymer electrolyte (G-PE) cell
consists of a transparent cylinder which houses CE and RE in a fixed
position by means of nylon screws (schematic view on the left); the
electrolyte is gelled by means of agar (right image)
AN OVERVIEW OF THE USE OF EIS MEASUREMENTS FOR THE ASSESSMENT OF PATINAS AND COATINGS IN THE CONSERVATION OF METALLIC CULTURAL HERITAGE 85

The CP setup keeps the electrolyte in contact with the

measurement area by means of a cloth soaked with the
electrolyte and kept wet thanks to capillarity (Figure 2a).
The G-PE setup adopts a “solid electrolyte” approach:
the aqueous electrolyte is gelled with agar (Figure 2b).
The two setups have been compared with the same
artificial rain electrolyte, on stainless steel and bronze
(Letardi et al. 2017): while almost the same results have
been obtained for the |Z|max, slight differences appear
on the overall spectra, which are under investigation to
properly model with equivalent circuits. The CP setup
embeds a stainless-steel pseudo-reference electrode,
which allows the measurement to be fixed around Eoc
but not to measure Eoc on an absolute scale. The G-PE
setup allows either a standard or pseudo-reference
electrode to be used, for which a detailed study has
been conducted to address the influence of the choice
and position of the reference electrode (Ramírez Barat,
Cano, and Letardi 2018).

Several methods were tested to connect the artwork to

the WE input of the potentiostat (Figure 3). Sometimes
it can be easy to access a hidden part of the artifact, to
clean it down to the bare metal and to attach an alligator
clip (Figure 3a). This works fine with high conductivity
copper alloys once electrical continuity from the contact
point to the measurement area has been verified. As
an alternative, when small casting defects are present, Figure 3. In-situ WE contact: (a) alligator clips on polished surface in
microdrilling inside the defect may allow a copper wire hidden position; (b) alligator clips on a pin inserted in a microdrilled
or pin to be firmly inserted to which an alligator clip can defect; (c) stainless steel tip or contact-protected test probe for a
be attached (Figure 3b). A hard metal tip can also be printed circuit

used (Figure 3c) as quite an effective solution on bare
and patinated surfaces. On coated metals, it should be
considered that this kind of contact makes a pinhole in the
coating, which can be easily repaired after measurement.
EIS evaluation on patinated metals
Patinas play a relevant role from an aesthetic point of
view and their treatment/cleaning during restoration
is an issue to be carefully addressed. In addition, they
influence the effectiveness of protective treatments
(Otieno-Alego 1998).
Several papers address EIS measurements on a number
of natural and artificial patinas on copper and iron alloys
using quite different approaches, mainly on laboratory
coupons. An extended discussion is reported in a recent
review (Ramírez Barat and Cano 2018), which mainly
focused on EC adopted to fit the data. Weathering steel
is largely used in architecture and civil engineering and
there is more extensive and systematic literature with
examples on naturally weathered steel. Therefore, a
general consensus exists about models to be used to fit
EIS spectra. The situation seems less clear for copper
and its alloys. A detailed study has been published for
natural copper patinas (Zhang et al. 2002), and a local
EIS study on bare and artificially patinated copper
coupons has pointed out the quite complex behavior
of copper corrosion (Mennucci et al. 2012). Generally,
EIS spectra on copper and copper-alloy patinas indicate
various electrochemical processes with a different rela-
tive role on the different patinas considered. The shape
of EIS spectra on the Nyquist plots mainly suggests
the presence of three electrochemical processes with a
behavior attributable to the uneven nature of the patina,
but there is no commonly accepted model and parameter
interpretation (Ramírez Barat and Cano 2018).
86  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
It is worth mentioning that the role of alloying elements
in copper-alloy corrosion (i.e., patina formation) has
been pointed out (Chiavari et al. 2007). An overview
of EIS spectra from several copper and bronze patinas,
both on coupon and monuments, showed corrosion rate
results spanning a wide range of values (Letardi and
Luciano 2007).
EIS evaluation of protective coatings
EIS has been employed to study different coatings for
the conservation of metallic heritage. This paper does not
aim to be a detailed review of this topic. A brief overview
of waxes and acrylics is presented below, complemented
by Table 1 at the end of the document.
Waxes
The first application of EIS to cultural heritage materials
was the evaluation of waxes for the protection of bronze
sculptures (Price et al. 1995). Waxes are very popular
coatings for conservation, as they provide a natural
surface finish. Different waxes, including natural, micro-
crystalline (MC), and polyethylene (PE) types, have been
used to coat heritage metals. Microcrystalline waxes such
as Cosmolloid H80, Renaissance wax (a mixture of MC
and PE waxes), or Soter (MC waxes with benzotriazole)
are the most common.
In general, waxes offer poor corrosion protection if
compared to acrylic coatings. In most examples, the |Z|max
value increases by one or two orders of magnitude with
respect to bare metal. Only exceptionally do they provide
higher values, depending on the number or thickness of
the layers. In addition, their resistance to weathering is
also very limited. Most results show quick impedance
reduction in a very short period of outdoor exposure
(1–6 months), which can even result in the complete
failure of the coating in less than a year (Ramírez Barat
and Cano 2015, Swartz and Clare 2015). Waxes seem to
perform better on patinated surfaces, where they continue
to offer some protection after 1–2 years of natural
exposure (Letardi 2004, Joseph et al. 2007, Joseph et
al. 2013). This is probably because wax adheres better
on a porous patina than on a clean and polished metal
surface, penetrating and sealing the pores.
Combinations of wax and benzotriazole (BTA) have
shown to be more effective than wax alone, either as
a combined treatment or as a formulation (Soter wax),
ANALYTICAL INVESTIGATION
enduring at least 23–48 months of outdoor exposure
(Price et al. 1995, Letardi 2004).
More recent studies have considered the use of poly-
ethylene waxes, especially for iron and steel protec-
tion. Poligen ES91009 has shown promising results
on bronze and steel for indoor applications (Paterakis,
Lafuente, and Cano 2010, Cano et al. 2010). However,
under atmospheric exposure, yellowing and rust
stains appear after only six months (Shashoua and
Matthiesen 2010).
Acrylics
Incralac is probably the most popular acrylic coating for
bronze, and therefore the most studied. Different results
have been obtained for this coating, as several factors
may contribute to degradation. In general, it is estimated
to last between 3 and 5 years in outdoor environments
(Brostoff 2003, Ramírez Barat et al. 2017). In a marine
atmosphere, Incralac alone has shown limited efficiency,
offering no advantage with respect to wax coatings,
though the addition of a wax layer over the coating
improves its duration (Letardi 2004).
Paraloid B-72 and B-44 also give high impedances
(increasing |Z|max by a factor of ~106) and good stability
(Ramírez Barat and Cano 2015). Paraloid B-67 and
B-48N give slightly lower |Z| max values and worse
stability. Paraloid B-67 increased its |Z| max when
exposed to atmosphere for short periods, which was
attributed to undesirable cross-linking reactions.
This fact has been demonstrated by photodegradation
studies of acrylics (Chiantore and Lazzari 2001). On
the other hand, Paraloid B-48N seems to be a more
effective barrier against ammonia and acetic acid than
Paraloid B-72 (Švadlena and Stoulil 2017) for indoor
applications.
The lifespan of coatings may be extended if combined
with inhibitors or a wax topcoat. Some studies have
explored the use of Paraloid B-44, B-72, and other coat-
ings with different inhibitors. In some cases, the results
showed no great improvement to the performance of the
coating alone (Cano, Lafuente, and Bastidas 2010, Kosec
2010), led to an unacceptable whitish surface after aging
(Ćurković 2010), or were promising but no aging studies
were performed (Salvadori et al. 2017).
As an alternative to traditional acrylic coatings, other
formulations have been tested, such as acrylic urethanes
AN OVERVIEW OF THE USE OF EIS MEASUREMENTS FOR THE ASSESSMENT OF PATINAS AND COATINGS IN THE CONSERVATION OF METALLIC CULTURAL HERITAGE 87

or polyvinylidene fluorides (PVFD). Acrylic urethanes ering (Clare and Swartz 2014), though weathering studies
have shown slightly higher impedances and better weath- have been too short to draw definitive conclusions.
ering resistance than Incralac (Bierwagen, Shedlosky,
Recent research has focused on the development of new
and Stanek 2001, Ellingson et al. 2004), though limited
coatings, especially using sol-gel technology, and on the
studies have been carried out. More interest has been
analysis of natural products such as carboxylates, which
raised by fluoropolymers, such as PVFD. These coatings
will be discussed in future work.
have interesting properties; unfortunately, they lack good
adherence to the metal surface and so have been used
Conclusion
in mixtures with acrylics or as a topcoat. When used as
a topcoat over Paraloid B-44, they have been shown to Over the two last decades, the application of EIS to heri-
provide better resistance to natural and artificial weath- tage science has gained growing interest both in terms

Table 1. Summary of studies on different coatings

Coating Composition Substrates References

Carnauba Natural wax Bronze, Bronze + artificial (Price et al. 1995, Ćurković et al. 2010, Kosec, Legat, and
patinas Miloev 2010)
Butcher’s Boston Carnauba and microcrystalline Bronze (Swartz and Clare 2015)
Polish Amber waxes
Paste Wax
BeSq195 Microcrystalline wax Bronze, Bronze + artificial (Price et al. 1995, Otieno-Alego et al. 1998)
patinas
Cosmolloid_H80 Microcrystalline wax Bronze (Joseph et al. 2013, Ramírez Barat and Cano 2015, Albini et
al. 2016, Albini 2017)
HiMic3080 Microcrystalline wax Bronze (Price et al. 1995)
Microwax_80a Microcrystalline wax Bronze (Price et al. 1995, Otieno-Alego et al. 1998)
R21 Microcrystalline wax Bronze (Letardi 2000 and 2004)
TeCero_3534F Microcrystalline wax Bronze (Letardi 2000 and 2004, Joseph et al . 2007)
TWA_2095(*) Microcrystalline wax Bronze, Bronze + artificial (Price et al. 1995, Otieno-Alego et al. 1998)
patinas
Soter_201 LC Microcrystalline waxes with BTA Bronze (Letardi et al. 1998, Letardi 2000 and 2004)
Renaissance wax Cosmolloid 80 and BASF-A Bronze, steel, (Argyropoulos et al. 2007, Cano et al. 2010, Swartz and
waxes precorroded steel Clare 2015)
Wax 1847 Branched microcrystalline wax Bronze (Jáuregui-Gónzalez et al. 2016)
M420 Polyethylene wax Bronze (Price et al. 1995)
MP-22 Polyethylene wax Bronze + artificial patina (Jáuregui-Gónzalez et al. 2016)
Poligen™ ES-91009 Polyethylene wax emulsion Bronze, steel, precorroded (Paterakis, Lafuente, and Cano 2010, Cano et al. 2010)
steel
Ozokerite 1899 Paraffin Bronze + artificial patina (Jáuregui-Gónzalez et al. 2016)
Paraloid B-44 Methyl-methacrylate and ethyl- Bronze, steel (Bierwagen, Shedlosky, and Ellingson 2003, Ćurković et al.
acrylate copolymer 2010, Kosec, Legat, and Miloev 2010, Hosbein, Swartz, and
Clare 2014, Ramírez Barat and Cano 2015)
Paraloid B-48N Methyl-methacrylate and butyl- Copper, bronze, iron (Ramírez Barat and Cano 2015, Švadlena and Stoulil 2017)
acrylate copolymer
Paraloid B-67 Isobutyl methacrylate polymer Bronze (Ramírez Barat and Cano 2015)
Paraloid B-72 Ethyl-methacrylate and methyl- Copper, bronze, iron, steel (Cano et al. 2010, Ramírez Barat and Cano 2015, Švadlena
acrylate copolymer and Stoulil 2017)
Incralac Paraloid B-44 + BTA Bronze (Bierwagen, Shedlosky, and Stanek 2001, Ellingson 2004
Bronze + natural patina et al., Letardi 2004, Ramírez Barat and Cano 2015, Ramírez
Barat et al. 2017, Salvadori et al. 2017, Mihelčič et al. 2019)
Acrylic urethanes Polished cast bronze (Bierwagen, Shedlosky, and Stanek 2001, Brostoff 2003,
Ellingson et al. 2004)
Fluoropolymers Bronze + natural patina (Bierwagen, Shedlosky, and Ellingson 2003, Clare and Swartz
2014, Swartz, Price, and Clare 2016, Salvadori et al. 2017)
(*) no longer produced
88  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
of the development of more refined field measurement
setups and in laboratory application to characterize
patinas and protective treatments. The lack of generally
accepted standards often makes the comparison of results
from different studies far from straightforward. On the
other hand, the wealth of approaches adopted highlights
the many relevant parameters to be considered when
dealing with the conservation of metallic artifacts. More
comparative studies of different coatings and combina-
tions, outdoor natural exposure, and systematic long-term
studies are needed. The behavior of the same coating on
different substrates and environments is also a key issue
that needs further investigation.
Based on the published results, EIS, as an NDT method
to characterize corrosion behavior, may be a relevant tool
in the near future to define better the taxonomy of metal/
patina surfaces of interest to metallic cultural heritage.
It will allow test methodologies to assess treatment
effectiveness on weathering on different surfaces and
in different environments to be better defined.
Acknowledgments
This work was funded by project HAR2017-89911-R
(AEI-MCIU) and IPERION-CH (E.U. H2020, G.A.
654028) and supported by the CNR-STM program.
References
Albini, M. 2017. Fungal biogenic patina: Optimization of
an innovative conservation treatment for copper-based
artefacts. Ph.D. dissertation, Université de Neuchâtel,
Switzerland.
Albini, M., L. Comensoli, L. Brambilla, E. Domon Beuret,
W. Kooli, L. Mathys, P. Letardi, and E. Joseph. 2016.
Innovative biological approaches for metal conservation.
Materials and Corrosion 67(2): 200–06.
Arceo-Gómez, D.E. and G. Zambrano-Rengel. 2016.
Electrochemical characterization of patinas formed on
a historic bell from the Cathedral Museum of Campe-
che-México, world heritage site. International Journal of
Electrochemical Science 11(11): 9379–93.
Argyropoulos, V., D. Charalambous, A. Kaminari, A.
Karabotsos, K. Polikreti, A. Siatou, E. Cano, D.M. Bastidas,
I. Cayuela, J.-M. Bastidas, C. Degrigny, D. Vella, J. Crawford,
and S. Golfomitsou. 2007. Testing of a new wax coating
Poligen ES 91009® and corrosion inhibitor additives used for
improving coatings for historic iron alloys. In Metal 2007:
ANALYTICAL INVESTIGATION
Proceedings of the Interim Meeting of the ICOM-CC Metals
Working Group, Amsterdam, 17–21 September 2007, eds. C.
Degrigny, R. van Langh, I. Joosten, and B. Ankersmit, vol.
5, 10–15. Amsterdam: Rijksmuseum.
Bierwagen, G., T.J. Shedlosky, and K. Stanek. 2001. Electro-
chemical studies of the protection of bronzes from corro-
sion by organic coatings. In Metal 01: Proceedings of the
International Conference on Metals Conservation, Santiago
de Chile), eds. I. MacLeod, J.M. Theile, and C. Degrigny,
291–96. Fremantle: Western Australian Museum.
Bierwagen, G., T.J. Shedlosky, and L.A. Ellingson. 2003.
Developing and testing a new generation of protective
coatings for outdoor bronze sculpture. Progress in Organic
Coatings 48(2–4): 289–96.
Brostoff, L.B. 2003. Coating strategies for the protection
of outdoor bronze art and ornamentation. PhD disserta-
tion, Van ’t Hoff Institute for Molecular Sciences (HIMS),
The Netherlands.
Bruni, T. 2018. Evaluation of the treatment methodology
of bronze alloy artistic surfaces: The case of the restora-
tion of the plates from the Redipuglia military memorial.
Ph.D. dissertation, Florence University, Italy.
Cano, E., D. Lafuente, and D.M. Bastidas. 2010. Use of EIS
for the evaluation of the protective properties of coatings
for metallic cultural heritage: A review. Journal of Solid
State Electrochemistry 14(3): 381–91.
Cano, E., D.M. Bastidas, V. Argyropoulos, S. Fajardo,
A. Siatou, J.M. Bastidas, and C. Degrigny. 2010. Elec-
trochemical characterization of organic coatings for
protection of historic steel artefacts. Journal of Solid State
Electrochemistry 14(3): 453–63.
Chiantore, O. and M. Lazzari. 2001. Photo-oxidative
stability of paraloid acrylic protective polymers. Polymer
42(1): 17–27.
Chiavari, C., K. Rahmouni, H. Takenouti, S. Joiret,
P. Vermaut, and L. Robbiola. 2007. Composition and
electrochemical properties of natural patinas of outdoor
bronze monuments. Electrochimica Acta 52(27): 7760–69.
Clare, T.L. and N.A. Swartz. 2014. Characterization of
high performance protective coatings for use on cultur-
ally significant works. In Intelligent coatings for corrosion
control, eds. A. Tiwari, J. Rawlins, and L.H. Hihara, chap.
19, 641–71. Woburn, MA: Butterworth-Heinemann.
Ćurković, H.O., T. Kosec, A. Legat, and E. Stupnišek-Lisac.
2010. Improvement of corrosion stability of patinated
AN OVERVIEW OF THE USE OF EIS MEASUREMENTS FOR THE ASSESSMENT OF PATINAS AND COATINGS IN THE CONSERVATION OF METALLIC CULTURAL HERITAGE
bronze. Corrosion Engineering, Science & Technology
45(5): 327–33.
Ellingson, L.A., T.J. Shedlosky, G.P. Bierwagen, and E.R.
De la Rie. 2004. The use of electrochemical impedance
spectroscopy in the evaluation of coatings for outdoor
bronze. Studies in Conservation 49(1): 53–62.
Hosbein, K.N., N.A. Swartz, and T.L. Clare. 2014. Elec-
trochemical identification and categorization of the
protective quality of intact and damaged coatings. Elec-
troanalysis 26(9): 1935–44.
Jáuregui-Gónzalez, D.A., F.J. Rodríguez-Gómez, J.
Contreras-Vargas, P. Roncagliolo-Barrera, and M.
López-Arriaga. 2016. Influence of the application and
preparation of wax coatings on artificially patinated
bronze surfaces. In Metal 2016: Proceedings of the Interim
Meeting of the ICOM-CC Metals Working Group, New
Delhi, 26–30 September 2016, eds. R. Menon, C. Chemello,
and A. Pandya, 170–77. New Delhi: International Council
of Museums-Committee for Conservation (ICOM-CC)
and Indira Gandhi National Centre for the Arts (IGNCA).
Joseph, E., P. Letardi, L. Comensoli, A. Simon, P. Junier,
D. Job, and M. Wörle. 2013. Assessment of a biological
approach for the protection of copper-alloy artefacts.
In Metal 2013: Proceedings of the Interim Meeting of the
ICOM-CC Metals Working Group, Edinburgh, 16–20
September 2013, eds. E. Hyslop, V. Gonzalez, L. Troalen,
and L. Wilson, 203–07. Edinburgh: ICOM-CC and
Historic Scotland.
Joseph, E., P. Letardi, M. Rocco, and S. Prati. 2007.
Innovative treatments for the protection of outdoor
bronze monuments. In Metal 2007: Proceedings of the
Interim Meeting of the ICOM-CC Metals Working Group,
Amsterdam, 17–21 September 2007, eds. C. Degrigny, R.
van Langh, I. Joosten, and B. Ankersmit, vol. 5, 71–77.
Amsterdam: Rijksmuseum.
Kosec, T., A. Legat, and I. Miloev. 2010. The compar-
ison of organic protective layers on bronze and copper.
Progress in Organic Coatings 69(2): 199–206.
Letardi, P. 2000. Electrochemical impedance measure-
ments in the conservation of metals. In Radiation in art
and archeometry, eds. D.C. Creagh and O.A. Bradley,
15–39. Amsterdam: Elsevier Science B.V.
Letardi, P. 2004. Laboratory and field tests on patinas and
protective coating systems for outdoor bronze monu-
ments. In Metal 2004: Proceedings of the Interim Meeting
of the ICOM-CC Metals Working Group, Canberra, 4–8
89
October 2004, eds. J. Ashton and D. Hallam, 379–87.
Canberra ACT: National Museum of Australia.
Letardi, P. 2013. Electrochemical measurements in the
conservation of metallic heritage artefacts: An overview.
In Corrosion and conservation of cultural heritage metallic
artefacts, European Federation of Corrosion Series, eds.
P. Dillmann, D. Watkinson, E. Angelini, and A. Adriaens,
126–48. . Cambridge: Woodhead Publishing.
Letardi, P. and G. Luciano. 2007. Survey of EIS meas-
urements on copper and bronze patinas. In Metal 2007:
Proceedings of the Interim Meeting of the ICOM-CC Metals
Working Group, Amsterdam, 17–21 September 2007, eds.
C. Degrigny, R. van Langh, I. Joosten, and B. Ankersmit,
vol. 3, 44–50. Amsterdam: Rijksmuseum.
Letardi, P., A. Beccaria, M. Marabelli, and G. D’Er-
coli. 1998. Application of electrochemical impedance
measurements as a tool for the characterization of the
conservation and protection state of bronze works of art.
In Metal 1998: Proceedings of the International Conference
on Metals Conservation, Draguignan–Figanières, France,
27–29 May 1998, eds. J. Ashton and D. Hallam, 303–08.
London: James & James.
Letardi, P., B. Ramírez Barat, and E. Cano. 2017. Anal-
ysis of the influence of the electrochemical cell setup for
corrosion measurements on metallic cultural heritage.
In European Corrosion Congress (EUROCORR 2017)
and 20th International Corrosion Congress and Process
Safety Congress 2017: Corrosion Control for Safer Living,
Prague, 3–7 September 2017, vol. 2, 1136–42. Prague:
Czech Association of Corrosion Engineers.
Letardi, P., B. Salvadori, M. Galeotti, A. Cagnini, S.
Porcinai, A. Santagostino Barbone, and A. Sansonetti.
2016. An in situ multi-analytical approach in the resto-
ration of bronze artefacts. Microchemical Journal 125:
151–58.
Letardi, P., M. Albini, and E. Joseph. 2018. EIS meas-
urements for treatments testing: The case of a bio-based
method applied on outdoor bronze statues in Switzerland.
In 69th Annual International Society of Electrochemistry
Meeting, Bologna, 2–7 September 2018, poster S13-006.
Masi, G., C. Josse, J. Esvan, C. Chiavari, E. Bernardi, C.
Martini, M.C. Bignozzi, C. Monticelli, F. Zanotto, A.
Balbo, E. Svara Fabjan, T. Kosec, and L. Robbiola. 2019.
Evaluation of the protectiveness of an organosilane
coating on patinated Cu-Si-Mn bronze for contemporary
art. Progress in Organic Coatings 127: 286–99.
90  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Mennucci, M.M., M. Sanchez-Moreno, I.V. Aoki, M.C.
Bernard, H.G. De Melo, S. Joiret, and V. Vivier. 2012.
Local electrochemical investigation of copper patina.
Journal of Solid State Electrochemistry 16(1): 109–16.
Mihelčič, M., M. Gaberšček, G. Di Carlo, C. Giuliani,
M. Salzano de Luna, M. Lavorgna, and A.K. Surca.
2019. Influence of silsesquioxane addition on polyu-
rethane-based protective coatings for bronze surfaces.
Applied Surface Science 467–468: 912–25.
Otieno-Alego, V., D. Hallam, A. Viduka, G. Heath, and
D. Creagh. 1998. Electrochemical impedance studies of
the corrosion resistance of wax coatings on artificially
patinated bronze. In Metal 98: Proceedings of the Interna-
tional Conference on Metals Conservation, Draguignan–
Figanières, France, 27–29 May 1998, eds. J. Ashton and
D. Hallam, 315–19. London: James & James.
Paterakis, A.B., D. Lafuente, and E. Cano. 2010. The
corrosive influence of acetic acid emissions on bronze
and the efficacy of two protective coatings. In Metal
2010: Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, Charleston, South Carolina,
11–15 October 2010, eds. P. Mardikian, C. Chemello,
C. Watters, and P. Hull, 178–84. Clemson, SC: Clemson
University.
Price, C., D. Hallam, G. Heath, D. Creagh, and J. Ashton.
1995. An electrochemical study of waxes for bronze
sculpture. In Metal 95: Proceedings of the International
Conference on Metals Conservation, Semur-en-Auxois,
France, 25–28 September 1995, eds. I.D. MacLeod, S.L.
Pennec, and L. Robbiola, 233–41. London: James &
James.
Ramírez Barat, B. and E. Cano. 2015. Evaluación in situ
de recubrimientos protectores para patrimonio cultural
metálico mediante espectroscopía de impedancia elec-
troquímica. Ge-conservación 8(8): 6–13.
Ramírez Barat, B. and E. Cano. 2018. In situ electro-
chemical impedance spectroscopy measurements and
their interpretation for the diagnostic of metallic cultural
heritage: A review. ChemElectroChem 5(19): 2698–716.
Ramírez Barat, B., A. Crespo, E. García, S. Díaz, and E.
Cano. 2017. An EIS study of the conservation treatment of
the bronze sphinxes at the Museo Arqueológico Nacional
(Madrid). Journal of Cultural Heritage 24: 93–9.
Ramírez Barat, B., E. Cano, and P. Letardi. 2018. Advances
in the design of a gel-cell electrochemical sensor for
ANALYTICAL INVESTIGATION
corrosion measurements on metallic cultural heritage.
Sensors and Actuators B: Chemical 261: 572–80.
Salvadori, B., A. Cagnini, M. Galeotti, S. Porcinai, S.
Goidanich, A. Vicenzo, C. Celi, P. Frediani, L. Rosi, M.
Frediani, G. Giuntoli, L. Brambilla, R. Beltrami, and S.
Trasatti. 2017. Traditional and innovative protective
coatings for outdoor bronze: Application and perfor-
mance comparison. Journal of Applied Polymer Science
135(2): 46011.
Sansonetti, A., M. Colella, P. Letardi, B. Salvadori, and
J. Striova. 2015. Laser cleaning of a nineteenth-century
bronze sculpture: In situ multi-analytical evaluation.
Studies in Conservation 60(1): S28–S33.
Švadlena, J. and J. Stoulil. 2017. Evaluation of protective
properties of acrylate varnishes used for conservation of
historical metal artefacts. Koroze a Ochrana Materialu
61(1): 25–31.
Swartz, N. and T.L. Clare. 2015. On the protective nature
of wax coatings for culturally significant outdoor metal-
works: Microstructural flaws, oxidative changes, and
barrier properties. Journal of the American Institute for
Conservation 54(3): 181–201.
Swartz, N.A, C.A. Price, and T.L. Clare. 2016. Minimizing
corrosion of outdoor metalworks using dispersed chemi-
cally stabilized nanoclays in polyvinylidene fluoride latex
coatings. ACS Omega 1(1): 138–47.
Zhang, X., W. He, I. Odnevall Wallinder, J. Pan, and C.
Leygraf. 2002. Determination of instantaneous corrosion
rates and runoff rates of copper from naturally patinated
copper during continuous rain events. Corrosion Science
44(9): 2131–51.
Blanca Ramírez Barat has a B.Sc. in Chemistry,
B.A. in Fine Arts, and Ph.D. in Materials Science and
Engineering. After several years in R&D management,
she joined the “Corrosion and protection of metal in
cultural heritage” research group at CENIM-CSIC. She
is currently working on electrochemical techniques for
the study and conservation of metallic cultural heritage.
Paola Letardi has a degree in Physics and works in the
field of material science and surface spectroscopy, with
particular interest in the development of methodologies
and instrumentation. She has been active in national and
international projects on diagnostics and monitoring for
the conservation of cultural heritage. Her research is
focused on the study of metal corrosion in the marine
AN OVERVIEW OF THE USE OF EIS MEASUREMENTS FOR THE ASSESSMENT OF PATINAS AND COATINGS IN THE CONSERVATION OF METALLIC CULTURAL HERITAGE 91

environment and on specific applications of electrochem-

ical and spectroscopic techniques in the field of artifacts
of historical interest.

Emilio Cano has a Ph.D. in Fine Arts and is a tenured

scientist at CENIM-CSIC. His fields of expertise include
corrosion and protection of metallic cultural heritage,
electrochemical techniques applied to conservation
science, and corrosion inhibitors. He has published more
than 80 research papers in international scientific journals
and given talks at more than 100 scientific conferences,
both on corrosion science and conservation science. He
is assistant coordinator of ICOM-CC Metals Working
Group.
A Multi-Analytical Approach to Evaluate Surface Treatments
on Copper-Alloy Artefacts: A Case Study Applied to the
Restoration of the Memorial of Redipuglia
Teresa Bruni Bruna Mariani
Università di Firenze Fonderia Artistica Battaglia
Florence, Italy Milan, Italy
teresa.bruni@yahoo.it restauro@fonderiabattaglia.com
*Author for correspondence
Abstract
The Redipuglia First World War Memorial accommodates
thousands of copper-alloy (Cu76-Zn11-Sn5-Pb2, other6) plates
with the names of 39,857 identified soldiers who fell in battles
in the surrounding areas. Within the framework of a restoration
intervention at the site, a test project was planned a few
months prior to the beginning of the full restoration in order
to compare surface treatments on coupons with the same alloy
composition as the 80-year-old plates. Surface morphology,
aesthetic appearance and corrosion rate were characterised by
non-destructive techniques both on coupons and on a test plate
from the memorial after sandblasting, black artificial patination
and application of a protective coating. A widely used micro-
crystalline wax (Soter) and a micro-crystalline wax blend (CTS)
applied in one or two layers were compared on coupons and the
test plate. A silane treatment from the B-IMPACT project was also
tested on coupons only. The results obtained were compared
all along the treatment steps in order to check for the degree
of similarity between the treatment effect on the original plate
and on the coupons. A higher surface roughness characterised
the test plate compared to the coupons after sandblasting, even
Introduction
The Memorial of Redipuglia is Italy’s largest and most
majestic memorial to the soldiers who fell in the Great
War.1 It was opened in 1938 and it underwent extensive
restoration for the occasion of the centenary of the First
World War. Built on the slopes of Mount Sei Busi and
designed by architect Giovanni Greppi and sculptor
Giannino Castiglioni, it was opened on 18th September
1938 after ten years of construction. This massive monu-
ment, also known as the Memorial ‘of the Hundred Thou-
sands’, accommodates the remains of 100,187 soldiers who
fell in battle in the surrounding areas. Strongly supported
by the fascist regime, this monument was intended to cele-
brate the sacrifice of the fallen soldiers as well as provide
a dignified resting place for the dead who could not be
buried in the cemetery of the Undefeated. It has a three-
Barbara Salvadori Paola Letardi*
ICVBC-CNR IAS-CNR
Sesto Fiorentino, Florence, Italy Genoa, Italy
barbara.salvadori@cnr.it paola.letardi@cnr.it
though the colour difference was negligible, and very similar
results were obtained for the corrosion rate, which was just
slightly lower on the coupons. The change in colour on the
artificial patina was only just perceivable on the coupons and
the test plates, with an increase in the corrosion rate. The surface
colour on more or less all the tested coatings did not change in
any perceivable way with respect to the patinated surface. The
improvement in corrosion protection was systematically lower
for the test plate with respect to the coupons but identical results
were obtained for the wax treatments, the best being the CTS
wax blend applied in two layers. The highest protection was
obtained on the coupons using the B-IMPACT silane coating.
The treated coupons were exposed to natural weathering at the
Genoa Experimental Marine Station. The results obtained helped
and supported the choice of treatment to be adopted for the
Redipuglia plates.
Keywords
copper alloys, NdT measurements, coatings, microcrystalline
waxes
tier structure symbolising the army descending from the
sky led by its Commander towards the Path of Heroes. At
the top, three crosses evoke Mt. Golgotha and the cruci-
fixion of Christ. Twenty-two large steps (2.5 m high by
12 m wide), containing the remains of 39,857 identified
soldiers in alphabetical order, are tiled with copper-alloy
(Cu76-Zn11-Sn5-Pb2, other6) 55 × 150 cm plates with the
names of the soldiers. Each burial niche is surmounted by
the wording ‘Present’ and can be reached via the lateral
stairs leading to the top (Figure 1). In the centre of the first
large step, is the niche of the only woman buried here, a
nurse named Margherita Kaiser Parodi Orlando, while the
22nd step accommodates the remains of 72 soldiers from
the Navy and 56 from the Customs Corps (Gualandra
1995, Fiore 2003).
A MULTI-ANALYTICAL APPROACH TO EVALUATE SURFACE TREATMENTS ON COPPER-ALLOY ARTEFACTS: A CASE STUDY APPLIED TO THE RESTORATION OF THE MEMORIAL OF REDIPUGLIA
The above-mentioned copper-alloy plates were produced
by the Fonderia Artistica Battaglia (Milan)2 and the
Vincenzo Perani (Brescia) foundry. They were cast
between 1935 and 1939, except for those on the last four
steps, which were cast and installed in the 1950s.
Over the years, the surface of the plates had been painted
with various layers of grey synthetic products, distorting
the perception of the rationalist aesthetics of the archi-
tecture. Because of the poor state of conservation of
the layers of paint applied to the surface of the plates
(Figure 2), the treatment selected was to sandblast the
surfaces until the bare metal was exposed, apply an arti-
ficial ammonium-sulfide-based patina by the foundry
and protect them with a wax coating.
Figure 2. Redipuglia plates before restoration: (a) failure of the painted
layer on the 22nd large step; (b) damages on the 14th large step
93
Figure 1. The Memorial of Redipuglia (Italy).
Twenty-two large steps, containing the
remains of 39,857 identified soldiers in
alphabetical order, are tiled with copper-
alloy plates with the soldiers’ names.
Each burial niche is surmounted by the
wording ‘Present’ and can be reached via
the lateral stairs leading to the top. At the
top, three crosses evoke Mt. Golgotha and
  the crucifixion of Christ
Because of the huge number of plates to be treated
(covering thousands of square metres) and their peculiar
features, the restorers were faced with several challenges.
On three-dimensional works, the discontinuities of the
materials and the possible flaws in application may go
unnoticed, while on a plain surface like that of the Redi-
puglia plates, any error gains macroscopic visibility. Even
though the conservation problem is quite peculiar, the
huge number of plates to be treated and their relatively
simple surface meant they were considered quite a good
case study for comparing surface behaviour on metal
artworks for conservation and on coupons. The same
foundry in charge of the original copper-alloy plates was
asked to repatinate them after sandblasting and to protect
them with a properly selected transparent coating. A
parallel characterisation plan was then designed on a test
plate and on a number of coupons especially made by the
same foundry with the same nominal composition and
fusion technique as the original plates. The project started
only a few months prior to the full plate restoration at
Redipuglia. Several NdT measurements were taken to
characterise the surface properties after sandblasting,
artificial patination and protective treatment both on the
coupons and test plate. This comparison was adopted to
evaluate the representativeness of the results obtained on
coupons with different protective treatments with respect
to the plates to be treated in Redipuglia. Two waxes and
two different application methods were selected that
could be of interest for the restoration of the memorial
plates. Moreover, a Silane polymer transparent coating
developed within the framework of the B-IMPACT
Project for black patinated modern bronzes was consid-
ered. This paper reports on the project plan, the results of
the surface characterisation on the coupons and the test
plate after sandblasting/artificial patination/protective
treatment application and provides some preliminary
results on the coupons exposed to natural weathering at
the Genoa Experimental Marine Station (GEMS) marine
exposure site (Stifanese et al. 2018).
94  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND ANALYTICAL INVESTIGATION

Materials and methods

Samples
To characterise the treatments adopted and to choose
between the two waxes and two different application
methods for one of them, sixty 5 × 5 cm coupons
(Figure 3a) of the same copper alloy as the Redipuglia
plates (composition in weight: copper 76%, zinc 11%,
tin 5%, lead 2%, minor alloying elements 6%) were cast
by Fonderia Artistica Battaglia. In addition, an original
plate (Figure 3b) was moved from the Memorial to the
foundry to provide a reference test surface.

Surface treatments
The labels adopted to identify the different surface
finishing under evaluation are listed in Table 1. The
surfaces of the coupons and the original plate were
prepared for patination by sandblasting with an Ibix
machine equipped with a helical-vortex ‘Helix’ nozzle
Figure 3. Samples for treatment testing: (a) coupons after sandblasting;
and Almandine Garnet abrasive (120 mesh) at 1.5 atm (b) the test plate moved from Redipuglia to Fonderia Artistica Battaglia
pressure. (Milan, Italy) (after partial sandblasting)

Table 1. Labels for the surface finishing considered

Label Surface finishing
S sandblasted
P S + artificial patina
C1 P + one layer of CTS wax
C2 P + two layers of CTS wax
CS P + one layer of Soter wax
BI P + B-IMPACT silane-based coating for black patinated
modern bronzes
After sandblasting, the coupons and the original plate
were hot-patinated (130–160°C) by brush with a 10
wt% solution of ammonium sulfide to provide the same
chromatic effect originally imparted to the surfaces by
the application of liver of sulfur (potassium sulfide).
Coatings were then applied by brush. Soter 202/LS wax
is a commercial blend of microcrystalline waxes and
benzotriazole in turpentine oil and ethyl acetate (see
Materials list for details) and it was used as a reference.
One layer of wax was cold-applied and gently polished
with a soft cloth after 48 hours. CTS WH, a blend of
microcrystalline wax (Reswax WH) and benzotriazole
in D40 white spirit, was applied in one or two layers
on different sets of coupons to test the effectiveness
of different amounts of protective layer: a first layer
was cold-applied, followed by a second application on
some coupons after one hour. Both groups of coupons
were gently polished with a soft cloth after 48 hours.
B-IMPACT is a 3-mercapto-propyl trimethoxysilane
(PropS-SH) in non-toxic solvents which has been devel-
oped within the framework of the European B-IMPACT
(bronze-improved non-hazardous patina coatings)
project (www.b-impact.cloud) and which proved a
promising protective treatment for black-patinated
modern bronze (Monticelli et al. 2017, Masi et al. 2018).
According to the application instructions, the first layer
was cold-applied, followed by drying at ambient temper-
ature with the aid of a low-temperature phon air flux,
and the second layer applied after two hours. C1, C2 and
CS wax coatings (cf. Table 2) were applied on groups of
seven coupons and on one third of the test plate surface
(left, centre and right, respectively). The silane-based
coating BI was tested on a group of five coupons only,
as it is not yet commercially available and could not be
used for the restoration of the Redipuglia plates. After
application, surfaces were allowed to stabilise for 15 days
prior to the characterisation measurements.
Characterisation techniques
Mainly non-destructive techniques (NdT) were adopted
to characterise the surface properties in order to compare
A MULTI-ANALYTICAL APPROACH TO EVALUATE SURFACE TREATMENTS ON COPPER-ALLOY ARTEFACTS: A CASE STUDY APPLIED TO THE RESTORATION OF THE MEMORIAL OF REDIPUGLIA 95

the results on the coupons and on the test plate under (Letardi et al. 2016). Each measurement consisted of
the same experimental conditions. 1900 s Eoc monitoring, a linear polarisation resistance
(LPR) measurement from -10 mV to 10 mV around Eoc
A Scalar DG-2A portable optical microscope equipped
with a 0.167 mV/s scan rate, 300 s Eoc monitoring, an
with a 25–200× optical zoom was used to study the
EIS measurement in the range of 1 MHz–10 mHz, with
surface morphology. After sandblasting, five roughness
5 pnt/decade and Vrms = 10 mV, and a final 200 s Eoc
measurements were taken on each coupon and sixteen
monitoring. At least three measurements were taken
measurements on the test plate using a Pocket Surf III
for each surface finishing both on the coupons and the
portable device (measuring range 0.03–6.35 µm).
test plate. The low frequency (10 mHz) value of the
Colour measurements were taken according to European impedance modulus |Z|lf was used to compare roughly
Standard EN 15886 (European Committee for Standardi- the corrosion rate of the different surface finishing on
zation 2010) with a Minolta d2600 spectrophotometer in the coupons and the test plate.
SCI/SCE mode, 100% UV, illuminant D65 and observer
The results obtained on the coupons and the Redipuglia
10°. The CIELAB 1976 colour coordinates L*, a* and
test plate (Figure 3b) were compared at every treatment
b* were recorded for each selected point (Ø~8 mm)
step in order to check for the degree of similarity between
after each treatment step (sandblasting, patination
the treatment effects on the 80-year-old plates and the
and coating). To follow the evolution of the aesthetic
coupons.
appearance through the different surface treatments,
three equally spaced measuring points along one diag-
Weathering
onal were selected for each coupon and localised using
a transparent plastic mask. On the test plate, 16 evenly The treated coupons were exposed to natural weathering
spaced points were selected. DE colour differences were at GEMS (Letardi et al. 2016, Stifanese et al. 2018) in
computed according to the CIELAB 1976 formula. April 2018 in order to compare the behaviour of the
protective treatments under test and to support the choice
The spectral features of the coatings on representative
to be adopted for the Redipuglia plates. To accelerate
coupons were characterised with a portable Alpha
the degradation, three coupons for each wax treatment
Bruker Fourier transform infrared (FTIR) spectrometer
were exposed on a different rack and sprayed twice a
operating in total reflection mode and equipped with
week with a NaCl solution for the first 140 days. Colour
a camera. The measuring spot was 6 mm in diameter.
measurements were taken after 1, 2, 3 and 6 months
Spectra were acquired at a resolution of 4 cm−1 in the
of weathering to monitor the evolution of the surface
7500–375 cm−1 range, collecting 128 scans. Five spectra
appearance.
were collected on each coupon and processed using
OPUS 7.2 software.
Results and discussion
Electrochemical measurements were taken with a Gamry
Surface characterisation before protective
Ref600 potentiostat with a CP contact probe electro-
treatments
chemical cell setup (measurement area of 1.77 cm2)
(Letardi 2013). A tenfold concentrated artificial rain Observation by microscope (Figure 4) and roughness
(conductivity 130 µS/cm, pH6) was used as an electrolyte measurements highlighted some differences between the

Table 2. Average colour values on the test plate and coupons for the different surface finishing (SF, labels in Table 1)

Plate Coupon ∆E
SF
L* a* b* ∆E(coated) L* a* b* ∆E(coated) (plate-coupon)

S 64 ± 3 5.3 ± 0.2 22.0 ± 0.7 64 ± 2 4.5 ± 0.3 23 ± 1 1±3

P 28 ± 2 2.0 ± 0.5 3±1 36 ± 3 4.1 ± 0.7 8±2 9±6
C1 29.6 ± 0.4 1.0 ± 0.3 1.2 ± 0.6 3±3 36 ± 2 3.8 ± 0.7 8±2 0±6 10 ± 4
C2 30.3 ± 0.5 1.2 ± 0.4 1.9 ± 0.6 2±4 34 ± 2 3.2 ± 0.6 6±2 3±6 6±4
CS 31 ± 1 1.0 ± 0.2 1.3 ± 0.4 4±4 35.5 ± 0.9 2.4 ± 0.6 4±2 4±4 5±3
BI 37 ± 1 4.1 ± 0.8 7±2 1±5
96  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
coupons (roughness 2.2 ± 0.2 µm) and the plate (rough-
ness 5.8 ± 0.4 µm). Trials to increase the coupons’ surface
roughness by further sandblasting or acid attacks were
ineffective. Several factors may be responsible for this
difference (such as technological evolution of foundry
techniques, or surface corrosion upon exposure) which
have yet to be looked into.
Figure 4. Microscopic images of the surfaces after sandblasting: (a)
coupon 50×; (b) coupon 200×; (c) plate 50×; (d) plate 200×
On the other hand, colour measurements gave an
unperceivable (ΔE < 5) difference in colour between the
coupons and the test plate (Table 2). EIS measurements
gave similar results for coupons (|Z|lf = 29 ± 7 KΩcm2)
and the test plate (|Z|lf = 16 ± 9 KΩcm2).
The artificial patina appeared as incoherent with variable
thicknesses (from 0.5 to 40 µm as characterised on cross
sections for a couple of coupons). The patina applica-
tion turns the surface shade to red with a halving of L*
(Table 2) and a perceivable colour difference between the
test plate and coupons (ΔE = 9 ± 6). EIS measurements
showed an increase in corrosion rate after patination
for both the coupons (|Z|lf = 5 ± 1 KΩcm2) and the test
plate (|Z|lf = 2.9 ± 0.5 KΩcm2). This is a reasonable result
considering it was obtained with an acidic attack, which
produces a powdery patina (Salvadori et al. 2015).
Comparison of protective treatments
The coatings did not change the colour in a perceivable
way either on the test plate or the coupons (Table 2).
The colour difference between the plate and coupons
remained roughly the same as after patination, with a
slight decrease for C2 and CS.
The homogeneity of the coatings on the coupon was
estimated by the relationship between the intensity of
the peaks and the quantity of the coatings on the surface,
ANALYTICAL INVESTIGATION
referring to ν (CH) bands (3000–2800 cm-1) and ν (CH)
peaks (1500–1400 cm-1) (Griffiths and de Haseth 2007,
Boyatzis et al. 2009). The comparison of five spectra
acquired in different areas of each coupon suggests that
the films were applied homogeneously.
EIS measurements on the coupons showed higher values
of |Z|lf, i.e. better protection, for C2 and BI (Figure 5).
Systematically lower values were registered for the test
plate surface with respect to coupons. The best results
were also obtained for the C2 wax on the test plate, even
though the results were quite uneven on the different
areas measured.
Figure 5. EIS data (low-frequency impedance modulus) on the coupon
(circles) and the plate (squares) for different surface finishing (see Table 1
for surface treatment labels)
First results after weathering
Colour measurements after weathering (Figure 6)
showed a larger variation for sprayed coupons, as
expected, with an initial increase in brightness and a
general shift in the a* and b* parameters to negative
values, indicating a variation in colour towards green
and blue shades. The sprayed coupons generally show
a larger decrease in a* with the CS characterised by a
smaller colour variation.
Future work
EIS measurements on 6-month-weathered coupons
are underway to compare the evolution of corrosion
resistance. The coupons will then be further weathered
to characterise the behaviour of the different coatings
A MULTI-ANALYTICAL APPROACH TO EVALUATE SURFACE TREATMENTS ON COPPER-ALLOY ARTEFACTS: A CASE STUDY APPLIED TO THE RESTORATION OF THE MEMORIAL OF REDIPUGLIA
Figure 6. Colour data on the coupon exposed to natural weathering at
GEMS. Labels for the applied coatings are listed in Table 1. The added ‘s’
is for samples sprayed twice a week with a 5% NaCl solution for the
first 5 months
Figure 7. The restored plates newly mounted on their large step
under evaluation. The restoration of the plates at the
Memorial of Redipuglia began in summer 2018; the C2
coating was chosen because of its better cost/effectiveness
ratio. Field measurements are planned for spring 2019
to compare in-situ treatment with the results obtained
on the coupons.
Conclusion
The study showed that the test plate had greater rough-
ness (5.8 ± 0.4 µm) than the coupons (2.2 ± 0.2 µm) after
97
sandblasting. Applying coatings did not seem to change
the colour with respect to the patinated-only surfaces,
either on the plate or the coupons. The colour difference
between the coated surfaces on the plate and the coupons
remained the same as the patina, with a slight decrease
for the C2 and CS waxes.
The best protective results were obtained with C2 and
BI. Systematically lower values were measured on the
test plate surface, meaning that the protection efficiency
was higher for the coupons. Nonetheless, the tested
treatments had the same rating on the coupons and on
the test plate.
The colour measurements after 6 months of weathering
showed a larger aesthetic change for sprayed coupons
with a variation in colour towards green and blue shades.
The CS proved to be less prone to colour change than
the C1 and C2.
As a final result, taking into account the cost and perfor-
mance, the C2 coating was chosen for application on the
plates (Figure 7). Further work is in progress to compare
coating performance on the coupons and the Memorial
copper-alloy plates in greater detail after weathering.
Acknowledgements
We are grateful to Cecilia Monticelli, who provided the
B-IMPACT coating with proper guidelines for application,
and Roberto Stifanese and Pierluigi Traverso, for their
support with spraying the samples at GEMS. We would
also like to thank Marilena Ricci (University of Florence),
Emma Cantisani and Silvia Vettori (ICVBC-CNR) for their
help with characterising the coupons.
Notes
1 www.itinerarigrandeguerra.it/en/33055/Memori-
al-of-Redipuglia .
2 www.fonderiabattaglia.com/index.php?node=stat-
ic&static=storia-gli-anni-trenta&lang=en .
References
Boyatzis, S.C., V. Argyropoulos, A. Siatou, K. Polikreti, D.
Charalambous, and A.M. Douvas. 2009. An evaluation of
reflectance-absorption infrared spectroscopy for in situ
investigation of organic coatings on corroded metals. In
Proceedings of the International Conference TECHNART
2009: Non-destructive and Microanalytical Techniques
in Art and Cultural Heritage, Athens, 27–30 April 2009.
98  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
European Committee for Standardization. 2010. Conser-
vation of cultural property. Test methods – Colour meas-
urement of surfaces (EN 15886:2010).
Fiore, A.M. 2003. La monumentalizzazione dei luoghi
teatro della Grande Guerra: Il sacrario di Redipuglia di
Giovanni Greppi e Giannino Castigioni. In Annali di
architettura: Rivista del Centro internazionale di Studi di
Architettura “Andrea Palladio” 15: 234–47.
Griffiths, P.R. and J.A. de Haseth. 2007. Specular reflec-
tion. In Fourier transform infrared spectrometry, chapt.
13, 277–301. Hoboken, New Jersey: John Wiley & Sons.
Gualandra, G. 1995. Redipuglia, vecchio e nuovo sacrario.
Reana del Rojale: Chiandretti Editore.
Letardi, P. 2013. Electrochemical measurements in the
conservation of metallic heritage artefacts: An overview.
In Corrosion and conservation of cultural heritage metallic
artefacts, eds. P. Dillmann, D. Watkinson, E. Angelini,
and A. Adriens, no. 65, 126–48. EFC Publications.
Letardi, P., B. Ramírez-Barat, M. Albini, P. Traverso, E.
Cano, and E. Joseph. 2016. Copper alloys and weathering
steel used in outdoor monuments: Weathering in an
urban-marine environment. In Metal 2016: Proceedings
of the Interim Meeting of the ICOM-CC Metals Working
Group, New Delhi, 26–30 September 2016, eds. R. Menon,
C. Chemello, and A. Pandya, 320–28. New Delhi: Interna-
tional Council of Museums-Committee for Conservation
(ICOM-CC) and Indira Gandhi National Centre for the
Arts (IGNCA).
Masi, G., A. Balbo, J. Esvan, C. Monticelli, J. Avila, L.
Robbiola, E. Bernardi, M.C. Bignozzi, M.C. Asensio,
C. Martini, and C. Chiavari. 2018. X-ray photoelectron
spectroscopy as a tool to investigate silane-based coatings
for the protection of outdoor bronze: The role of alloying
elements. Applied Surface Science 433: 468–79.
Monticelli, C., V. Grassi, C. Martini, C. Chiavari, G.
Mavilia, F. Zanotto, G. Masi, E. Bernardi, and A. Balbo.
2017. Protectiveness of different silane coatings on
patinated ancient and modern bronzes: Part 1. In EURO-
CORR 2017 & 20th International Corrosion Congress
(ICC) & Process Safety Congress 2017, 3–7 September
2017, Prague, Czech Republic, European Federation of
Corrosion Event No. 417, paper no. 82171.
Salvadori, B., E. Cantisani, T. Bruni, A. Cagnini, M.
Galeotti, S. Porcinai, A. Santagostino Barbone, and C.
Bonzanigo. 2015. Approccio multi-analitico per la carat-
terizzazione di patinature artistiche su opere in bronzo di
ANALYTICAL INVESTIGATION
epoca contemporanea. In Metalli in architettura. Conos-
cenza, conservazione, innovazione, Atti del 31° convegno
scienza e beni culturali, 30 June–3 July 2015), Bressanone,
Italia, eds. G. Biscontinand G. Driussi editors), , 241–52.
Stifanese, R., P. Letardi, and P. Traverso. 2018. A tool for
research activity and technology transfer in corrosion
and protection of materials: Genoa Experimental Marine
Station (GEMS). Pitture e Vernici 2(2018): 36–45.
Materials list
B-IMPACT – Black Patinated Modern Bronze
A silane transparent polymer coating developed within
the framework of the B-IMPACT project (M-ERA.
NET Transnational Call 2013; www.b-impact.cloud )
CTS WH
A ready-to-use blend based on Reswax WH micro-
crystalline wax and benzotriazole by CTS, Altavilla
Vicentina, Italy (www.ctseurope.com )
Soter 202/LS
A ready-to-use microcrystalline-wax-based coating
with a corrosion inhibitor by Bresciani s.r.l.
(www.brescianisrl.it )
Authors
Teresa Bruni has a bachelor’s and master’s degree in
Conservation Science with a thesis on natural and arti-
ficial patinas and metal conservation. She undertook
an Erasmus Plus Traineeship at the Cultural Heritage
Agency of Amsterdam where she worked at the Iperion
Ch European project.
Bruna Mariani is a restorer with a degree in Conser-
vation of Cultural Heritage from the School of Higher
Education of the Opificio delle Pietre Dure in Florence.
She specialises in the restoration of metal artefacts
(monuments, ancient weapons, archaeological mate-
rial, coins, ethnographic objects and jewellery) and has
worked for public and private institutions internationally
since 2004. She is responsible for the Restoration Depart-
ment at the Battaglia Artistic Foundry and teaches at the
OPD School in Florence, the Academy of Fine Arts in
Como and the Academy of Fine Arts in Naples.
Barbara Salvadori has a master’s degree in Chem-
istry and a PhD in Science for Conservation of Cultural
Heritage. Her major research interests deal with the
application of spectroscopic techniques (FTIR, XRF) and
A MULTI-ANALYTICAL APPROACH TO EVALUATE SURFACE TREATMENTS ON COPPER-ALLOY ARTEFACTS: A CASE STUDY APPLIED TO THE RESTORATION OF THE MEMORIAL OF REDIPUGLIA 99

optical and electronic microscopy to investigate metal and

stone artworks, evaluate the performance and durability
of conservation treatments, and monitor artistic surfaces
in situ. She is involved in several European and national
projects in the field of cultural heritage conservation. She
also works on specific projects with private conservators
and public institutions in the area of restoration and
archaeometric studies.

Paola Letardi has a degree in Physics and has worked

in the field of Material Science and Surface Spectroscopy,
with a particular interest in the development of meth-
odologies and instrumentation. She has been active in
national and international projects on diagnostics and
monitoring for the conservation of cultural heritage. Her
research is focused on the study of corrosion of metals
in the marine environment and on specific applications
of electrochemical and spectroscopic techniques in the
field of artefacts of historical interest.
A Non-Destructive Toolkit for Silver Plate Identification

Rosie Grayburn* Ann Wagner *Author for correspondence

Winterthur Museum, Garden and
Library
Winterthur DE, USA
rgrayb@winterthur.org
http://www.winterthur.org/
Winterthur Museum, Garden and
Library
Winterthur DE, USA
awagner@winterthur.org
http://www.winterthur.org/

Abstract
How can we effectively differentiate fused plating from
electroplating in decorative arts? When silver plating is intact
or restored, differentiating is not always clear. This is a crucial
question in silver conservation, as the plating method will likely
determine the treatment of the surface. Using a combination of
visual aids and non-destructive analysis, a selection of objects
were studied to inform effective and suitable treatment for
delicate electroplated surfaces and other forms of silver plate.
The thickness of silver plate was calculated using the ratio of
copper Kα and copper Kβ peak heights in energy dispersive
x-ray fluorescence (ED-XRF) spectra of the presentation surface;
results compared favorably to 3D microscopic measurements
and differences were clear between plating techniques. It was
therefore found that ED-XRF used in combination with expert
visual analysis can be a powerful tool in determining possible
manufacture methods and ascertaining treatment steps.
Keywords
silver plate, decorative arts, visual analysis, non-destructive
analysis, metal cleaning, electroplate, fused plate

Introduction
The Conservation of Silver and Copper Alloy Objects
project is part of a multiphase initiative to treat objects in
Winterthur Museum’s collections prioritized by greatest
conservation need. In October 2016, the Institute of
Museum and Library Services (IMLS) awarded funding
to support a two-year project to remove aged or failed
lacquer coatings, then polish and recoat approximately
500 silver objects, continue analytical research related
to silver surface corrosion (Pouliot et al. 2013), and
commence new research on corrosion and coating issues.
As the project progressed, inaccurately identified plating
techniques were discovered in the museum catalogue:
this accounted for approximately 5% of objects. Fused
plate (also called Sheffield or silver plate) was mistaken
for items with electroplated or brush-plated surfaces.
Restoration techniques, fabrication methods such as elec-
troforming, and misleading marks further complicated
visual identifications. These objects presented their own
unique treatment problems.
Tarnish is easily removed from a silver surface with a
mild abrasive, such as calcium carbonate. This polishing
technique permanently removes the most superficial layer
of the precious metal. Unlike sterling or coin silver, silver-
plated objects have a thin display surface of silver and a
more substantial substrate of a base metal. Electroplated
objects are particularly prone to mistreatment, as the
precious metal display surface is microns thick and can
be removed during a polishing campaign.
Identification to date has focused on destructive and
non-destructive methodologies (Carl and Young 2016)
not always accessible to those in smaller institutions or
private practice. This paper aims to show how visual
aids and portable energy dispersive x-ray fluorescence
(ED-XRF) equipment can be used to identify plating
methods and thus inform treatment.
A brief summary of silver-plating methods
Silver-plated objects included in this study illustrate
typical plating methods used since the late 1700s. Early
European application of silver plating employed steps of
mechanically burnishing layers of extremely thin leaves
of pure silver onto a completed base metal object.1 By
the 1700s, “French plating” described silver leaf applied
A NON-DESTRUCTIVE TOOLKIT FOR SILVER PLATE IDENTIFICATION
onto copper alloy (brass) objects, and “close plating”
described silver applied (often with tin) to iron/steel
items.2
Innovations in rolled sheet metal production contrib-
uted the new method of fusing, via heat and pressure,
a surface layer of silver onto copper. In the 1740s such
“fused plate” became a patented industrial manufac-
turing technique to chemically fuse a silver layer to a
prepared copper core, sometimes on both sides. The
method was also used infrequently for silver on copper
alloy. To create completed objects, the fused metal could
be rolled, hammered, stamped, and joined mechanically
or with silver solder.
In Sheffield and other parts of England and the United
States, some manufacturers applied narrow sterling
silver edges to cover joins, reinforce edges, and inhibit
loss to the surface from wear and polishing during the
first generation of use. Inset sections of sterling or more
thickly-plated fused plate were used for hand-engraved
ornaments. The fused plate body of this Argand lamp
(Figure 1) bears owner’s initials engraved on a silver inset
that is much more obvious now due to two centuries of
polishing.
Figure 1. Argand lamp, attributed to Matthew Boulton, probably
Birmingham, UK, fused plate, brass, glass, tinned sheet iron; Winterthur
Museum, Bequest of Henry Francis du Pont, 1965.1374
101
Beginning in the 1840s, faster and much more efficient
methods using Faraday’s principles of electrolysis were
employed to deposit a nearly pure silver layer on fully-
formed base metal objects and components. This fish
server (Figure 2) now exhibits plating loss, particularly on
the handle. The thickness of electroplating could vary by
the duration of time for application and by manufacturer’s
intent.3 In 1878, one company’s standard to measure the
silver thickness was maintained by weighing the finished
white metal object (often an alloy of tin, copper, and
antimony or German/nickel silver) prior to and again
after plating.4 During the same era, electroplating was
used to silver-plate electroformed objects (see Case
Study 4). As this brief summary indicates, verifying an
object’s construction and plating techniques is essential
for pre-treatment queries.
Figure 2. Patented fish server or fish knife, marked by John and Joseph
Cox, England, ca. 1848–51, nickel silver and electroplated silver;
Winterthur Museum, Gift of Mr. and Mrs. C. Thomas Attix, Jr., 1999.0023
A non-destructive toolkit
Throughout the project, ED-XRF spectroscopy deter-
mined object elemental composition and helped deduce
object manufacture in conjunction with visual examina-
tion methods. The following “toolkit” outlines the meth-
odologies used for both techniques in the examination
of objects to be treated.
Visual examination
Visually confirming plating methods and materials is
assisted by identifying (or ruling out) techniques as
well as recognizing evidence of loss of plating. Plating
loss occurs from use and commensurate polishing, as
well as corrosion of the exposed base metal substrate.
Fused plate is easy to confirm today through observa-
tion of the copper substrate, visually thick plating, and
manufacturing methods (i.e., not sand cast). Plated
white metals are more challenging to identify without
manufacture-specific stamped identifications or a firm’s
known preferences such as thin, spun Britannia metal,
stamped copper, or strong, cast or drop press-formed
nickel silver.5 Adding complexity is surface restoration
102  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
by electroplating or French plating, practiced from the
mid-1800s until today. Thinly electroplated surfaces are
sometimes difficult to distinguish from reduced original
fire gilding or fused plating. As ever, seek expertise and
references as well as ED-XRF spectroscopy.
Recommended steps for visual examination
• Step one
Surface: Examine the entire surface and patina to
discover character of plating method(s).
Ask: How thick is the plating? Burnished, smooth
surfaces? Brushy looking application? Detectable
granular deposits or layers? Blistering? Any solid silver
fills? Plating on all surfaces? Other plating (gold, tin)
present? Corrosion color? Loss commensurate with
function?
• Step two
Ornament and Decoration: Ornamenting techniques
using hammering, die-rolled sections, cast elements,
or even drop press-formed elements can help confirm
less visible materials.
Ask: Applied by hand with solder? Integral to body
construction? Chased or stamped? Pierced with fly
press or by hand? Engraved on body or an inlay?
• Step three
Construction techniques: Historically plated object
components were formed by combining many
construction techniques, most machine-aided. Manu-
facturing signs are visible on fused plate. Electroplating
typically is applied to a completed object, thus overlays
solder joins and hand-finishing techniques.
Ask: Hand-raised/hammered? Rolled sheet metal?
Spun or lathe skimming lines? Stamped or swaged?
Seamed and soldered joins? No joins visible? Cast?
Hammer marks, but superficial (not from forming)?
Electroformed?
• Step four
Weight and “feel”: Fused plate is hand-raised and/
or stamped and joined, thus thin-walled and vulner-
able to dents and deformation. Spun, stamped, or
hammered silver-plated white metal objects are often
thicker gauge, stronger, and heavier in the same form.
Electroplated Britannia metal can be an exception.
Ask: Gauge consistent between all elements? Base or
components filled or weighted (pitch, sand, plaster)?
Resonance or density?
ANALYTICAL INVESTIGATION
Analysis by ED-XRF spectroscopy
Analysis was performed with a handheld Bruker Tracer
III-SD XRF spectrometer using a rhodium tube (40 kV
high voltage, 9.6 μA anode current, 25 μm Ti/305 μm Al)
for 60 seconds live time irradiation. The spot size is
oblong in shape, approximately 1 cm × 0.5 cm. Spectra
were interpreted using ARTAX software.
Additional information can be gleaned from spectra to
semi-quantify plate thickness. Due to the phenomenon of
attenuation from different emission lines in the ED-XRF,
the thickness of a silver “coating” atop the substrate metal
was calculated using the ratio of copper Kα (8.04 keV)
and Kβ (8.90 keV) peak heights (Cu(Kα /Kβ)) in terms of
counts per second, using the following equation:
Cu(Kα /Kβ) = Cu(Kα /Kβ)0 exp -560dAg
where Cu(Kα /Kβ)0 was measured experimentally from
pure copper (6.986), dAg is the thickness of the silver
“coating” (in cm) and “560” refers to the difference in
theoretical attenuation coefficients of silver at the energies
of the copper Kα (8.04 keV) and Kβ (8.90 keV) emission
lines (Cesareo et al. 2009). Errors in the results derive
from alloying elements in the base metal or plate.
Identification case studies
The following case studies demonstrate how the toolkit
was developed and used during the project.
Case study 1: Fused plate spoon with original plating
Figure 3. Tablespoon, Birmingham, UK, 1808-23, silver plate on copper
(fused plate); Winterthur Museum, Gift of Philip E. Toussaint, 2018.0009
A recent acquisition allowed us to examine the plating
technique microscopically and to validate ED-XRF
thickness calculations. An English fused plate table-
spoon (Figure 3) showed some worn surfaces, exposing
the base metal, and a crack in the bowl near the handle.
ED-XRF of the exposed base metal showed pure copper,
without alloying elements. ED-XRF of the silver pres-
A NON-DESTRUCTIVE TOOLKIT FOR SILVER PLATE IDENTIFICATION 103

entation surface showed copper and silver with trace ED-XRF analysis of the shade showed the presence
amounts of lead and gold, suggesting a non-electro- of both silver and copper. The lamp’s fuel reservoir
plated object. also showed the presence of zinc, which is a common
alloying element for copper used in the fused plate
The crack exposed a cross section of the plating
process (Crosskey 2011). Comparing results from the
(Figure 4a). The base and plate metals are distinguished
shade versus areas of suspected fused plate, the shade
by their coloration and texture. However, the profile of
shows a much lower signal from silver, which implies
the crack (Figure 4 b) showed a lip in the approximate
a smaller concentration of silver and possibly a thinner
location of the plate/base metal boundary. Further
plate layer.
analysis of the ED-XRF data from the silver presenta-
tion surface semi-quantified the thickness of plate to In order to confirm this, further analysis of the data
5 µm. This is similar to the plate thickness measured was required (Table 1). Alpha and β emission lines are
by 3D microscopy. absorbed differently by surrounding elements after
emission, so the deeper the element or the thicker the
layer on top, the more the ratios are affected. Silver L
lines are less penetrating than K, so they only reveal
what is happening at the uppermost surface of the
object. Table 1 shows that AgL ratios for all regions of
the lamp are similar, which implies that silver exists
in the uppermost layer across the entire object. Silver
Figure 4. (a) HIROX digital microscope image (credit: Steve Buck, K lines are more penetrating, so they provide infor-
HIROX USA) of fused plate spoon (2018.0009) at a crack in the bowl
mation about silver beneath the plate: electroplated
near the handle; (b) 3D image of cross section with integrated height
measurement. The red box indicates the location of the silver plate/ samples have lower ratios due to low absorption by
base metal boundary, the height of which was measured to be 8 µm the thin electroplate, whereas thicker plate will show
higher ratios due to larger amounts of absorption from
Case study 2: Argand lamp with a misidentified the plate. Therefore, Table 1 shows that the shade has
surface clearly lower ratios compared to other regions of the
lamp, implying a very thin, possibly electroplated layer.
This case study demonstrates how ED-XRF can be used
to enhance visual identification and inform treatment of
Table 1. K and L ratios and plating thickness for various sampled parts
an object containing mixed techniques. This oil-burning
of the Argand lamp (Figure 5)
lamp’s metal components were previously visually
Plate
identified as fused plate, but the lampshade (Figure 5) Lamp part
CuK AgL AgK
thickness
ratios ratios ratios
raised questions about the accuracy of this conclusion. (µm)
A cleaning test revealed the silver easily abraded and Shade (DT) spot 1 6.84 1.87 2.62 0.37
therefore was too thin for fused plate. Shade (DT) spot 2 6.93 1.82 2.89 0.14
Shade support (DT) 5.69 1.61 6.36 3.67
Shade rim (DT) 3.87 1.76 5.91 10.6
Fuel reservoir (AT) 4.20 1.68 5.85 8.71

In order to semi-quantify the thickness of the plate,

Figure 5. Argand lamp, Sheffield or Birmingham, UK, 1785-95, silver plate
on copper (fused plate) and silver-plated copper (electroplate), iron,
glass; Winterthur Museum, Gift of Henry Francis du Pont, 1961.1692.001.
Left to right: before treatment (BT), during treatment (DT) (shade
exhibiting thin-film interference), and after treatment (AT)
CuK ratios were used to measure the absorption of
these emission lines by the silver layer. The results are
shown in Table 1. The shade shows a silver thickness
of much less than 1 µm, which is clearly different in
from the rim, the shade support, and the fuel reservoir,
which was visually identified as fused plate. In fact,
the shade was found to have been brush-plated in its
treatment history.
104  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND ANALYTICAL INVESTIGATION

Case study 3: Sugar bowl with interior/exterior
differences
Visual examination of this sterling hallmarked sugar
bowl (Figure 6) revealed worn areas having a copper
substrate, suggesting silver plate, and a yellow color on
the interior, suggesting gilding or preferential tarnish.
ED-XRF data collected from the exterior of the bowl
showed a similar composition of Ag and Cu with traces
of Zn and Ni, while data obtained from the interior
showed Ag and Cu only. Small amounts of gold were
also detected in the interior, suggesting the silver-plated
interior had been gilded.
Figure 6. Sugar bowl, marked for “R.P.,” London, UK, 1927, fused plate,
electroplated gold; Winterthur Museum, Gift of Mr. and Mrs. James A.
Drain, 1982.0272.002. Left to right: BT and AT
As discussed previously, further analysis of the ED-XRF
data can provide more information about the plating
and layer structure (Table 2). In general, the lower the
α/β ratio, the more attenuation: either self-attenuation
by a thick layer, or attenuation by a covering element.
Table 2 demonstrates that AgL ratios for all exterior
surfaces are similar, indicating that silver is present in
the uppermost layer across the entire object. However,
the AgL ratios for the interior are higher, suggesting a
different surface treatment, e.g. gilding. Because silver
K lines are more penetrating, they detect silver beneath
the uppermost layer: gilded interiors show lower ratios
due to the attenuation by the gilded layer of the silver
emission K lines from the underlying silver.
To semi-quantify the thickness of the plate, CuK ratios
were used to measure the absorption of these emission
lines by the plating or gilding (Table 2). The results from
the gilded interior are not accurate because the copper is
silver-plated and gold-plated. However, the exterior plate
is of a uniform thickness, approximately 3 µm.
Case study 4: Electroformed Kettle
This hot water kettle (Figure 7) is designed to rest on
a pierced base, which holds a small lamp. The copper
substrate is visible in some areas, but not around the
curve of the body, where abrasion typically occurs.
The body and lid are very thin, with no clear signs of
hammering, repousse tool work, or joins from assem-
bled stamped components. The rigid upright handle
with the appearance of a riveted swing handle confirmed
the visual assessment that the vessel was not fused
plate but entirely electroformed. The interior of the
electroplated body contains a rectangular tab to which
a diode was attached during electroforming. ED-XRF
results show copper and silver only, with trace amounts
of mercury and arsenic (Figure 8), and plate thickness
was calculated as 4.5 µm, confirming the object’s visual
identification.
Figure 7. Hot water kettle, probably made in England or the United
States, 1850–1900, electroformed and silver-plated copper, rattan;
Winterthur Museum, Bequest of Henry Francis du Pont, 1958.1965. Left
to right: BT and AT

Table 2. K and L ratios and plating thickness for various sampled parts
of the bowl

Plate
CuK AgL AgK AuL
thickness
ratios ratios ratios ratios
(µm)
Side 5.77 1.70 5.94 3.41
Side 5.74 1.71 5.86 3.51
Base 5.95 1.78 5.42 2.85
Handle 5.81 1.68 5.62 3.28
Figure 8. Spectrum of an exterior section of electroformed kettle in Case
Interior 4.94 3.30 4.72 1.93 3.64 study 4; Winterthur Museum, 1958.1965
A NON-DESTRUCTIVE TOOLKIT FOR SILVER PLATE IDENTIFICATION 105

Treatment implications at Winterthur Table 3. Materials used for cleaning study and results summary from
photomicrographs shown in Figure 9. Solutions 1–4 were applied
The primary goal of the IMLS grant-funded project is using an unbleached cotton flannel (Polidori et al. 2018)
to remove aged or failed lacquer coatings, polish, and Cleaning solution Ingredients Result
recoat approximately 500 of the museum’s most vulner- 1. Calcium Precipitated calcium Somewhat reduces
able silver objects. In general, aged coatings are removed carbonate slurry carbonate in a scratches, but also
slurry paste with 1:1 adds new scratches to
with pressurized steam or solvent. The surface is then denatured alcohol: the surface.
lightly polished with a calcium carbonate slurry. Finally, deionized water
a minimum of two coats of Agateen #27 are sprayed or 2. Dilute calcium 2 g precipitated Significant reduction
brush-applied. carbonate calcium carbonate in of scratches observed
50 mL deionized water under magnification.
Some new scratches
While these steps remain the same for most types of
may be present.
silver plate, electroplating has been determined to be 3. Acidified 2 g xanthan gum, No alteration in
too thin to polish with calcium carbonate. For example, thiourea gel 100 mL deionized the scratch pattern
the Argand lampshade (Case study 2) was washed with water, 5 mL observed. Thiourea
concentrated sulfuric creates a yellowed,
a very dilute solution of calcium carbonate instead of acid, 8 g thiourea matte surface.
being abrasively polished with a paste. Further investi- 4. Dilute micro- 2 g 0.3 μm α-alumina in Reduces surface
gation used photomicrography to explore the effects of alumina 50 mL deionized water scratches by
burnishing the silver.
five different polishing materials on electroplated copper No new scratches
coupons. Five cleaning solutions were chosen based on observed.
common silver-cleaning practices (Figure 9, Table 3) 5. Selvyt® Rouge The manufacturer Photomicrographs
(Polidori et al. 2018). Cloth indicates that the indicate significant
Selvyt® Rouge Cloth scratch reduction.
is comprised of 100% Note that SEM-EDS
cotton impregnated revealed that the
with a “rouge polishing Selvyt® Rouge Cloth
compound.” contains no Fe or
rouge compound.

Figure 9. Photomicrographs taken before and after polishing coupons
with a variety of cleaning solutions. Top to bottom: BT and AT (Polidori
et al. 2018)
Since silver sulfide can act as an additional abrasive,
results of the cleaning test were limited to an untar-
nished silver surface. As a result, tarnish was removed
from the electroformed kettle (Case study 4) with dilute
micro-alumina (Table 3). For this particular object, the
potential for a highly reflective, burnished surface was
acceptable.
Conclusion
ED-XRF can be used to quickly and non-destructively
semi-quantify plate thickness. Results have shown
that thin electroplate can be identified, but thicker
electroplate (5 µm) can be a similar thickness to fused
plate. However, the determination of manufacturing
technique by non-destructive analysis is not straight-
forward. The main compositional difference between
electroplate and fused plate is that the silver layer on
fused plate is sterling standard instead of the pure
silver on electroplate. In addition, the most prevalent
substrate of fused plate is copper, with often up to 5%
zinc. There are also electroplate imitations of fused
plate using a copper substrate. Therefore, the presence
of elements other than silver and copper in the plate
and base metal, respectively, adds a margin of error to
these calculations. Specifically, this error derives from
the attenuation of the x-rays by the base metal. There-
fore, this technique should always be used alongside
informed visual inspection and expert consultation to
differentiate between plating types.
All tested treatment methods for polishing silver plate
altered the silver surface to varying degrees. The most
effective cleaning method for silver-plated copper was
found to be dilute micro-alumina, which burnishes
the display surface while removing the least amount of
precious metal. Burnishing occurs when metal transfers
from high to low areas on the surface; this may be objec-
tionable to curators or collectors given that it produces
a mirror-like finish. However, since re-plating an object
is generally considered unethical, cleaning methods that
106  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
preserve the silver plate may be preferred despite the risks
of creating a highly reflective surface.
Acknowledgements
Deep thanks go to our project partners and colleagues
Bruno Pouliot, Lauren Fair, William Donnelly, Tia
Polidori, Katie Rovito, Joy Gardiner, Catherine Matsen,
Jim Schneck, and Giovanna Urist. Thanks also go to Tina
Gessler at Colonial Williamsburg Foundation. Project
funding: Conservation of Silver and Copper-Alloy
Objects, Phase II; IMLS MA-30-16-0154-16.
Notes
1 English manufacturers describe the plating as “pure,”
meaning the best standard achievable, but we do not have
ED-XRF data today to quantify the percentage of silver
(Bradbury 1912, 96; Crosskey 2011, 19).
2 Applying gold and silver onto base metals is summa-
rized by Vitali (1997, 19–21) and by British plating
experiments in Glanville (2006, 90–97) and Crosskey
(2011, 14–62).
3 John Hodges, who worked for Matthew Boulton, cites
a proportion of 24:1 copper to silver for candlesticks
(Crosskey 2011, 24 and note 37). Approximately 12
plated finishes were available in early flatware electro-
plating (Glanville 2006, 98). In the United States, silver-
smiths sought recipes and replicated plating methods,
using British terms like “heavy,” “best,” and “Albata,” as
well as “triple” and “quadruple” plate and symbols “A-1”
for superior quality and “EP” for electroplated (Hogan
1972, 10).
4 Percy (1878, 490–91). The author describes switching
from a galvanic battery power source to a “magneto-elec-
tric machine,” thus removing mercury from the plating
workshop. Site-specific “washing” or “brush plating”
techniques placed a thin layer of silver or gold on an
isolated location, replacing fire gilding.
5 Early electroplate was self-regulated in the United
States, thus hundreds of names were created, sometimes
omitting the word “plate.” Typical abbreviations: “EP”
for electroplate, “triple” or “quadruple plate” and EPNS
for electroplated nickel silver. British marks include:
EPNS and EPGS for electroplated German silver, EPCA
for electroplated copper alloy, EPBM for electroplated
Britannia metal.
ANALYTICAL INVESTIGATION
References
Bradbury, F. 1912. History of old Sheffield plate. London:
Macmillan.
Canadian Conservation Institute. 2017. The science of
silver tarnish and its cause. Understanding how silver
objects tarnish. www.canada.ca/en/conservation-insti-
tute/services/workshops-conferences/regional-work-
shops-conservation/understanding-silver-tarnish.
html#a3 .
Carl, M. and M.L. Young. 2016. Complementary analyt-
ical methods for analysis of Ag-plated cultural heritage
objects. Microchemical Journal 126 (May): 307–15.
Cesareo, R., M.A. Rizzutto, A. Brunetti, and D.V. Rao.
2009. Metal location and thickness in a multilayered
sheet by measuring Kα/Kβ, Lα/Lβ and Lα/Lγ x-ray ratios.
Nuclear Instruments and Methods in Physics Research
Section B: Beam Interactions with Materials and Atoms
267(17): 2890–96.
Crosskey, G. 2011. Old Sheffield plate: A history of the
18th century plated trade. Sheffield, England: Treffry
Publishing.
Glanville, P. 2006. Manufacturing and marketing in
Europe, 1600‒2000. Feeding desire: Design and the tools
of the table, 1500‒2005, eds. S. D. Coffin et al., exhib. cat.:
76–101. New York: Smithsonian Institution, Cooper-He-
witt, National Design Museum.
Hogan, E.P. 1972. Victorian silverplated holloware.
Princeton, New Jersey: The Pyne Press.
Percy, R.T.. 1878. The American at work. Among the
silverplaters. Appletons’ Journal 5(6): 481–94.
Polidori, T., K. Rovito, and R. Grayburn. Electroplated
silver during a re-lacquering campaign at Winterthur:
Treatment and analytical insights. Poster presented
at the AIC’s 46th Annual Meeting, Houston, 29 May–3
June 2018.
Pouliot, B., J. Mass, C. Matsen, W. Donnelly, K. Andrews,
and M. Bearden. 2013. Three decades later: A status
report on the silver lacquering program at Winterthur.
Objects Specialty Group Postprints, Volume Twenty, eds.
L. Kaplan, K. Dodson, and E. Hamilton, 32–48. Wash-
ington, DC: The American Institute for Conservation of
Historic & Artistic Works.
A NON-DESTRUCTIVE TOOLKIT FOR SILVER PLATE IDENTIFICATION 107

Vitali, U. 1997. A quest for the domus aurea in the

resurgence of gilding. In Antiquity revisited: English and
French silver-gilt from the collection of Audrey Love, eds.
A. Phillips and J. Sloane, exhib. cat., 17–27. New York:
Christie’s.

Authors
Rosie Grayburn, Ph.D., is an associate scientist at
Winterthur Museum, where she manages the Scientific
Research and Analysis laboratory.

Ann Wagner is curator of decorative arts at Winterthur

Museum, where she is responsible for metalwork and
related organic collections.
Conservation and Examination of an Iconic 17th-Century
Pewter Object: The Hartog Plate’s History Revealed
through Synchrotron Analysis
Tamar Davidowitz* Robert G. Erdmann
Rijksmuseum Amsterdam Rijksmuseum Amsterdam
University of Amsterdam University of Amsterdam
Amsterdam, The Netherlands Amsterdam, The Netherlands
t.davidowitz@rijksmuseum.nl r.erdmann@rijksmuseum.nl
David Thurrowgood Ian D. MacLeod
Queen Victoria Museum and Art Gallery Western Australian Museum
Launceston TAS, Australia Fremantle WA, Australia
david.thurrowgood@launceston.tas.gov.au ian.macleod@museum.wa.gov.au
*Author for correspondence Dudley Creagh
University of Canberra
Bruce ACT, Australia
dudley.creagh@canberra.edu.au
Abstract
The Hartog Plate is the earliest object of European origin found
in Australia. It has immeasurable significance to maritime history
and exploration. To commemorate the 400th anniversary of
its placement on the Western Australian coast, an extensive
research and conservation project has been undertaken in
the Netherlands and in Australia. The plate is extremely fragile
and suffers from many conservation issues. Analytical and
visualisation techniques such as high-resolution photography,
UV-light photography, x-radiography, microscopy, and 3D
structured-light and CT scans were conducted at the Rijksmuseum
to understand aspects such as geometry, adhesion, degradation
processes, and crack propagation. In 2017, the Hartog Plate was
analysed at the x-ray fluorescence microscopy beamline at the
Australian Synchrotron using GeoPIXE software in combination
with the Maia 384 detector array. This system integrates the
sample stage motion with continuous fly scanning, providing
high-resolution elemental maps of the surface. This has led
Introduction
As the first known object of European origin on the
continent of Australia, the symbolic and historical
significance of the Hartog Plate is unsurpassed. It is of
singular importance to Australia and the Netherlands,
and a fascinating object in terms of material complexity.
The plate is incredibly fragile as a result of its corrosion
and conservation history. Now in the collection of the
Rijksmuseum, the significance of the plate requires that
it remains accessible, both to the public and for materials
research. This presents the conservator with the challenge
of how to responsibly preserve an object which is so
fragile, but from which much is required.
Daryl L. Howard
Australian Synchrotron (ANSTO)
Clayton VIC, Australia
daryl.howard@synchrotron.org.au
David Hallam
University of Canberra
Bruce ACT, Australia
david.hallam@canberra.edu.au
to invaluable insights into the production, degradation, and
conservation history of the plate. The data acquired in this
project have been integrated into a single self-consistent
model of the Hartog Plate developed at the Rijksmuseum,
which encompasses all of the imaging modalities. This enables
real-time interactive visualisation through a web-based engine,
allowing for statistical analysis and multimodal visualisations
which help to reveal interesting or anomalous areas of the object
for further study and documentation. The technologies and
tools applied to the analysis, documentation, and conservation
of the Hartog Plate demonstrate what is already technically
possible and indicate potential further developments. This paper
describes the first steps in this project and heralds future reports.
Keywords
Hartog Plate, x-ray fluorescence microscopy, epoxy,
visualisation, De Vlamingh, synchrotron
The need for conservation was urgent, largely due to
the plate’s last restoration treatment in 1999 with inap-
propriate conservation materials, which was causing
significant internal stress (Figure 1). The 400th anni-
versary of its placement in 1616 was the motivation for
an extensive documentation, research, and conservation
project. Contingent on the success of the conservation
was a request by the Australian government for the plate
to travel to Australia for the duration of the historic year;
a tour which would be launched in Fremantle, Western
Australia, by the King and Queen of the Netherlands as
a symbol of diplomatic links and shared history.
CONSERVATION AND EXAMINATION OF AN ICONIC 17TH-CENTURY PEWTER OBJECT: THE HARTOG PLATE’S HISTORY REVEALED THROUGH SYNCHROTRON ANALYSIS 109
Figure 1. The Hartog Plate before (top) and after (bottom) the 2016
conservation treatment
Historical context
On October 25, 1616, Dirk Hartog, skipper of the
Eendracht of the Dutch East India Company (Vereenigde
Oostindische Compagnie, or VOC), landed on the island
that now bears his name, just off the Western Australian
coast. In Europe, a great southern landmass had been
postulated since ancient times, but this was the first
definitive proof of its existence. Hartog was one of the first
skippers to use the new Brouwers Route, taking advantage
of the strong winds and currents in the southern Indian
Ocean now known as the Roaring Forties. His ship was
bound for Bantam in the Indonesian archipelago, and
this route could potentially hasten the journey by several
months (Halls 1964, 33).
Figure 2. The inscription visible on the plate
Hartog was aware that their arrival was a significant
event. In the Dutch nautical tradition of the time, Hartog
and his crew left tangible evidence behind on the most
robust and conspicuous material they could find aboard
their ship to ensure their presence was documented: they
flattened a pewter dish and chiselled an inscription into
its surface detailing the date, the name of their ship, their
destination, and the names of their most important crew
members (Figure 2). They nailed the plate to a wooden
post, and fixed it into a narrow crevasse on a cliff at the
northern tip of the island.
In 1697, Willem de Vlamingh, another skipper in the
service of the VOC, replaced it with his own pewter
plate. He copied the original inscription and added his
own addendum, taking the original plate back with him
to Batavia, the capital of the Dutch East Indies at the
time. It was immediately seen as an important historical
document by the Governor General, who sent it back
to VOC headquarters in Amsterdam later that year. He
cited the logs of the ship in his accompanying letter: ‘a
pole nearly decayed but still standing upright, with a
common middle-sized tin plate, which had been beaten
flat and attached to the pole, and which was still lying
near it. On this plate the following engraved words
were still legible . . . look and marvel how it remained
through such a number of years unaffected by air,
rain or sun’ (Major 1859, 115). The plate remained in
the archives of the VOC until 1799, when the Dutch
government took custody of it following the dissolution
of the VOC. Twenty years later it was transferred to
the Royal Cabinet of Rarities (Koninklijk Kabinet van
110  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Zeldzaamheden), and ultimately to the Rijksmuseum in
1883, where it can be found on display today (Sigmond
2003, 255).
Condition
Although the Hartog Plate suffers from complex degra-
dation issues, the famous inscription is still clearly visible.
It endured decades of exposure to salt water and extreme
variations in temperature and humidity, as well as phys-
ical strain due to strong winds and surface abrasion by
sand and other materials. The forced flattening and the
deep chiselling of the inscription caused significant
deformation of the plate’s metallic structure, rendering it
susceptible to corrosion. This has led to severe oxidation
of the surface, embrittlement, and crack propagation in
the bulk of the plate.
As a consequence, the plate has been fragmented for
quite some time. Most of the fragment edges present
oxide layers with a similar stratigraphy and volume to
those on the surface, and it is clear that the fragmen-
tation and plastic deformation occurred before these
slow-forming oxide layers developed. A metallic core is
still present, the thickness of which varies over the plate.
The larger fragments cannot carry their own weight, and
need continuous support due to the malleability of the
metallic core. This makes the brittle oxide layers on the
surface prone to cracking and delamination during any
kind of handling or treatment. As a result, the upper layer
containing the most surface information, or the limit
of the original surface, has been almost completely lost
(Bertholon 2001, 11).
Conservation history
Before any intervention could be considered, it was
essential to gain an understanding of the plate’s past
treatments. Documentation is almost exclusively limited
to the 20th century and is largely unpublished. Only two
previous conservation treatments were recorded: (i) one
carried out in the early 1960s at the Institut Royal du
Patrimoine Artistique (IRPA) in Brussels; (ii) the other
in the late 1990s at the Rijksmuseum. However, the plate
has a long and complicated history of interventions,
most of which have had a discernible influence on its
condition (Table 1).
Photographic and radiographic images, dating back
to the end of the 19th century, have proved invaluable
ANALYTICAL INVESTIGATION
in understanding the plate’s degradation processes. By
using precise pixel-scale (up to 20 µm/pixel) registration
across the various high- and low-resolution images of
the plate, a self-consistent model was constructed as
part of an interactive visualisation of all the known
historical images of the Hartog Plate, thereby allowing
the user to visually track the state of the object over time.
Cross-referencing this with the existing documentation
gave a much more complete picture of its deterioration
and the effects of the various interventions on its current
condition.
The Hartog Plate has always been presented as a whole
to maintain readability of its inscription. Previous treat-
ments concealed its fragmented state using adhesives,
fill materials, and extensive retouching. Detachment and
misalignment of the fragments over time necessitated
intermittent disassembly, re-adhesion, and handling.
As the corroded join edges cannot support the adjacent
fragments, handling of the plate when the fragments are
adhered causes significant structural stress, resulting
in the propagation of cracks and local surface damage,
especially around the inscription.
Methodology
X-ray diffraction (XRD) analysis performed during the
conservation treatment at IRPA in 1960 and a separate
study carried out at the Centraal Laboratorium voor
Onderzoek van Voorwerpen van Kunst en Weten-
schap (CL) in Amsterdam in 1973, both confirmed
no sign of active corrosion on the plate (Table 1). As
there were no visual indicators to suggest otherwise
at the commencement of this project, the initial
priority was mechanical stabilisation. To alleviate the
stress imposed by previous treatments, all unoriginal
materials were removed and the decision was made
to disassemble the fragments permanently without
subsequent re-adhesion. Another consideration for
this approach was the authentic presentation of the
plate. The plate’s condition is inextricably bound to its
provenance, and to conceal the damage it has suffered
would be to conceal its history. Considering the value
and significance of the inscription, it is vital that the
plate remains readable as a whole. To this end, a new
method of presentation was developed to fully support
the fragments while also preserving the legibility of
the inscription.
CONSERVATION AND EXAMINATION OF AN ICONIC 17TH-CENTURY PEWTER OBJECT: THE HARTOG PLATE’S HISTORY REVEALED THROUGH SYNCHROTRON ANALYSIS 111
Material analysis
To inform proper removal procedures, all foreign mate-
rials found on the surface were analysed before treatment.
Gas chromatography-mass spectrometry (GC-MS)
analysis confirmed the prior retouching was done with
methyl acrylate, or acrylic paint. The adhesive joining the
fragments proved to be an epoxy resin (glycidyl ether)
containing bisphenol-A and isopropylphenol by-prod-
ucts, consistent with Araldite®, the resin described in the
1999 conservation notes (Table 1). Fourier-transform
infrared spectroscopy (FTIR) analysis identified the white
residues as paraffin oil, and also indicated the presence
of silicone on the surface associated with past methods
of support and replication procedures.
Visualisation
Ultraviolet light (UV) photographs were taken of
the exposed epoxy to identify it visually for removal
(Figure 3). This only gave information from the surface,
not the geometry of the fragment edges within the
Figure 3. Ultraviolet light photograph showing fluorescence of the epoxy
adhesive used to join the plate fragments before removal
Figure 4. Radiograph (200 kv, 3 mA, 20 sec) of the plate
opaque, sometimes overlapping joins. Computed tomog-
raphy (CT) scans failed, owing to the absorption of the
x-rays by the plate. Although it was possible to extract
a complete image, the degree of processing ultimately
made the model unreliable and provided insufficient
resolution to be an effective tool for treatment. Radiog-
raphy was more successful (Figure 4), and it was possible
to integrate these images into a self-consistent digital
document for use during conservation. Registering the
data in this way ensured that any point in the plate could
be referenced, containing all of the signals from all of the
imaging modalities.
Treatment
As the mechanical stability of the Hartog Plate had been
significantly compromised by handling, insufficient
support, and previous conservation interventions, it was
paramount that the treatment itself would not exacerbate
the problem. The first priority was to remove the epoxy.
To minimise potential damage, the plate was moved back
onto its 1999 silicone mould support. This kept the plate
horizontal and provided a fitted and radiolucent backing
during treatment.
The retouching paint was removed with minimal amounts
of ethanol and cotton swabs, revealing the condition of
the surface surrounding the epoxy joins. Impregnation
with paraffin in the early 1960s (Table 1) had lessened
adhesion of the epoxy in many areas, occasionally causing
failure of the bond and aiding epoxy removal.
Risk analyses of potential removal methods eliminated
solvent-based removal of the epoxy and the use of
heating/cooling treatment, and confirmed that mechan-
ical removal was the only safe option for both the object
and the conservator. The joins needed to be removed in
a very specific order, determined by the fragility, orien-
tation, and weight of the fragments. Edges which had
been filed down to bare metal in the past to achieve a
better fit were the most difficult to disassemble. Epoxy
was removed mechanically using customised micro-scal-
pels and micro-hand drills varying from 0.2 to 0.4 mm
in width. Following removal of all the epoxy resin, each
fragment was thoroughly mechanically cleaned using fine
tweezers, soft brushes and wooden, ivory, or steel tools
to avoid surface residues being absorbed into the porous
surface, or creating local irregularities in the distribution
of the paraffin. Sawdust from the 1960s treatment could
still be found inside the letters, along with deposits of
112  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND ANALYTICAL INVESTIGATION

Table 1. Inventory of past interventions

Type of Type of
Date Institute Results Consequences
intervention documentation
1953 Analysis BM Published (1953) - Analysis requested due to concerns regarding tin pest, disproven by - Samples taken for metallographic
Published (1956) metallographic analysis (Plenderleith and Organ 1953, 67). analysis.
Correspondence - Sealed Perspex® case and impregnation with a synthetic resin is
suggested to limit any further exposure to oxygen (Plenderleith and
Organ 1953, 67).
- Consolidation with microcrystalline wax is suggested (Plenderleith
1962, 276).
- Note from A. van Schrendel claims no records of previous
treatments exist, but object has been coated with a ‘resinous matter’
sometime within the 20th century (Plenderleith and Organ 1953,
66).
1959– Analysis IRPA Published (1963) - Superficial coating confirmed, removed with xylene (Thomas- - Sample material removed using a
1963 Treatment Goorieckx 1963, 76). jeweller’s saw (Thomas-Goorieckx
- XRD confirms absence of soluble salts and tin pest. 1963, 75).
- Heat treatment performed to relieve tension in the crystal lattice - Surface oxides lost during
(Thomas-Goorieckx 1963, 75): 115°C in an oven for four days, treatment.
submerged in sawdust for one day and in molten paraffin for the - 1999 RMA report: plasticine fills
last three days (correspondence between directors A. van Schendel and silicone support have shrunk
and R. Lefève, February 11, 1963). and are no longer functional.
- Silicone support produced by casting a mould of the reverse. Fragments are no longer adhered.
- Placed in a wooden frame sealed with Perspex® sheet.
1966 Production of RMA Annual Report - Multiple replicas produced. - Remains of silicone found in plate
replicas 1966 - Mould made using Silopren K1, Bayer (silicone rubber). during 2016 treatment.
- Replicas made with Araldite® D (CY 230) HY 956 epoxy resin. Fill
materials: tin powder and CIBA epoxy dyes. Surface finish: pigments
and acrylate as a binder.
1973 Analysis CL Report - XRD analysis showed minimal amounts of tin(IV) oxide (Sn02), and - Sample material removed from
a significant amount of tin(II) oxide (SnO). No evidence of tin pest plate.
detected.
1987 Transport RMA Correspondence - Transport to Australia for exhibition. - 1997 ICN report: nitrogen capsule
Exhibition - Nitrogen capsule developed for safe transport and exhibition. no longer functional and does not
seal properly.
1997– Analysis ICN Report - Plate still exhibited on the 1963 support, consisted of 11 fragments - Sample material removed from
1998 Correspondence when the capsule was opened. plate.
- XRD, FTIR, and Oddy testing done on black support material and
grey modelling material. Identified as silicone latex and plasticine,
respectively. Silicone contained SiO2, TiO2, and BaS, while plasticine
contained CaCO3 as fill materials. Both determined to be chemically
stable but aged.
- Testing of three epoxies for application, a two-component epoxy
resin from Windsor was selected for adhesion with no sulfur
additive.
- Original layer of silicone latex determined to be in good condition
and re-used in the next capsule.
1999 Treatment RMA Conservation notes - Before intervention, the plate was exhibited at a 30° angle, causing - Epoxy resins cause significant
Annual Report it to slide off of its support. internal stress.
1999 - Paraffin, adhesive, paint, plasticine, and grey and white silicone - Glass capsule with nitrogen over-
rubber residues removed. pressure ruptures in 2006.
- All edges cleaned and degreased with white spirit and acetone.
Fragments adhered using Araldite® epoxy resin in combination with
fumed silica to increase viscosity. Joins retouched with acrylic paint.
- Glass capsule with a slight over-pressure of nitrogen designed by
Fokker Aerospace (collaboration with the University of Nijmegen;
capsule made by the firm Coelen).
- Medical-grade silicone support for both obverse and reverse for
transport.
2012 Analysis RCE Report - FTIR analysis of white residue identifies paraffin oil and silicone on - Sample removal.
the surface.
2012 Exhibition RMA Conservation notes - Development of a customised display case with a minimised air - Transfer and handling.
Correspondence exchange rate/vibration sensitivity and a milled aluminium support.
CONSERVATION AND EXAMINATION OF AN ICONIC 17TH-CENTURY PEWTER OBJECT: THE HARTOG PLATE’S HISTORY REVEALED THROUGH SYNCHROTRON ANALYSIS 113
Figure 5. A representative image showing the order and manner of epoxy
removal (Join 11, 12.5× magnification)
Figure 6. Digital reconstruction of the plate fragments (in collaboration
with Panoptes Heritage)
Figure 7. The largest plate fragment (A) resting on its milled polyethylene
transport/temporary exhibition support
plasticine, silicone, and paraffin. These materials were
also removed mechanically. Finally, the obverse and
reverse of each fragment was cleaned using minimal
amounts of ShellSol® T to remove any excess paraffin
and homogenise the surface. Treatment procedures were
carried out by hand using a binocular microscope at a
magnification of 12.5× to 50× and documented before,
during, and after removal (Figure 5). All micrographs,
samples, and areas of interest were recorded using a
consistent coordinate system, allowing for precise loca-
tion identification and cross-referencing.
Transport and exhibition
After the fragments were disassembled, it became
clear that if light-weight, chemically inert, and fully
fitted supports could be produced to transport them,
the plate would be able to travel to Australia for exhi-
bition. To achieve this, three-dimensional structured
light scans were made of the obverse and reverse of
each fragment using a calibrated HDI Advance scanner
(45 μm resolution). A flight case with custom-milled
dense polyethylene foam packaging for each separate
fragment was produced using data from the scans
(200 μm resolution). The fragment scans were subse-
quently digitally reconstructed to create a complete
model of the plate with the optimal orientation and fit,
without requiring any handling (Figure 6). This digital
model was used to design a single fitted support for
exhibition milled from the same foam, which simul-
taneously served as the transport packaging for the
largest fragment (Figure 7).
Synchrotron investigation
In 2013, the De Vlamingh Plate was analysed at the
Australian Synchrotron’s x-ray fluorescence microscopy
(XFM) beamline, providing extremely high-definition
elemental surface mapping. This led to many insights
into the history of the De Vlamingh Plate and helped
identify the best method for studying the active chem-
ical deterioration. The project was also much more
successful than anticipated at revealing markings
on the object’s surface, both directly and indirectly;
for example, copper ions accumulated at engraving
sites enabled surface details to become decipherable
(MacLeod et al. 2014, 4).
The greater structural stability of the Hartog Plate after
treatment enabled new opportunities for research.
114  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Following the 400th anniversary exhibitions, an extensive
research project by Dutch and Australian scientists and
conservators was undertaken at the Australian Synchro-
tron. It was anticipated that this would make legible
poorly preserved markings on the surface of the Hartog
Plate, and that elemental distributions on the surface
would assist in deciphering the chemistry of its deterio-
ration, as was achieved in the De Vlamingh experiment,
facilitating comparison of the datasets.
The XFM beamline requires a vertical scanning position,
but uses a sample stage motion with continuous fly scan-
ning. This ensures the object experiences no vibrations
or shocks during analysis, and leads to zero data readout
overhead. This, in combination with a customised
mounting system featuring a series of carefully placed
apertures, allowed for the fragments to be analysed safely
over the course of multiple scans.
The x-ray fluorescence data was acquired with the
Maia 384 detector array, which has an energy sensi-
tivity down to approximately 2 keV. The finest scans
were recorded with a 70 μm pixel-size with dwell times
of approximately 2 ms to 3 ms per pixel. An incident
excitation beam energy of 18.5 keV was used as this
was the best compromise for the fluorescent yields of
the elements being examined (Figure 8). The custom-
ised support set-up required a larger distance from the
detector than that of the De Vlamingh Plate, making
lighter elements more difficult to detect and reducing
potential insight into the Hartog Plate’s chemical
stability. Furthermore, the strongly deformed surface
of the plate, with a height difference of more than 20
millimetres, limits the quantitative value of the data.
While lab-based elemental composition may be used
to calculate possible elemental ratios in the image to
improve the level of estimated accuracy, the qualitative
information alone proved invaluable.
Figure 8. An indicative XRF spectrum extracted from a 2.2 mm × 1.8 mm
region within the triangle of the second ‘A’ in ‘AMSTERDAM’
ANALYTICAL INVESTIGATION
Full spectrum x-ray fluorescence (XRF) data were decom-
posed into elemental maps using the dynamic analysis
method implemented in the GeoPIXE software suite.
Output high-resolution image files were post-processed
using custom software developed by the research partic-
ipants to further extract surface detail and to develop
multi-element overlays of surface features.
Results
The resolution of the elemental maps highlighted the
extensive degree of porosity and micro-cracks on the
surface of the Hartog Plate not visible to the naked eye.
The state of degradation also renders important surface
features imperceptible to the naked eye or even under
magnification. One of the primary aims of this exper-
iment was to discover or refine surface details, and the
data surpassed expectation. The elemental distribution
of lead is especially effective in revealing original surface
deformations. For example, the letter p found in the word
schip, was no longer visible, but became obvious during
scanning (Figure 9). In addition, two touchmarks were
identified: one of a crowned rose on the obverse, just
below the date (Figure 10), and the second along the lower
edge of the reverse of the plate, possibly in the form of an
angel. While this would be a surprisingly early example
of this combination, it can be found on other surviving
VOC tin tableware, also in the Rijksmuseum collection
(Dubbe 1978, 454).
Chemical information, such as the presence of iron
corrosion products in the area surrounding the nail
holes as well as in severely degraded areas where similar
Figure 9. PbL element map of Fragment A showing a detail of the letter
‘p’ on the upper proper left corner (70 μm/pixel resolution)
CONSERVATION AND EXAMINATION OF AN ICONIC 17TH-CENTURY PEWTER OBJECT: THE HARTOG PLATE’S HISTORY REVEALED THROUGH SYNCHROTRON ANALYSIS 115
Figure 10. PbL element map of the upper edge of Fragment B revealing
a crowned rose touchmark (100 μm/pixel resolution)
holes were suspected, shows the plate was fastened
to its post with iron nails. Identification of new areas
where the plate was fastened, in combination with the
manipulation of the fragment positions in the digital
three-dimensional reconstruction, help to explain how
the plate has plastically deformed and fragmented over
time. While these are just some examples of the infor-
mation gained, the results of the synchrotron experiment
and their importance in understanding the plate’s history,
its deterioration, and the complex chemical interactions
which have led to its current condition, will be published
in a more detailed report.
The XFM data has also been incorporated into a self-con-
sistent model of the plate developed by the Rijksmuseum,
further assisting statistical analyses and multimodal
visualisations to reveal interesting or anomalous areas
of the plate for further study and documentation.
Conclusion
The decision to remove all unoriginal material from the
Hartog Plate and keep the fragments disassembled on a
fully fitted support has proven to be a successful approach
for handling, analysing, exhibiting, and transporting it
safely. It has led to the elimination of stress and to an
honest presentation of the plate that is faithful to its
condition and history. The plate has incurred no observ-
able loss or damage throughout the project, and it is now
back in its display case in the Rijksmuseum. Current
efforts are focused now on developing a sustainable
solution for the plate’s exhibition, including designing a
permanent support and minimising its exposure to any
corrosive gases.
As the Hartog Plate is nearly flat, it presents an inter-
esting challenge in regard to analytical and visualis-
ation techniques, as it is effectively a ‘hybrid’ two- and
three-dimensional object. Scanning and visualisation
methods originally designed for two-dimensional objects
have been applied, but needed to be modified to allow
for such a severely deformed surface. This has initiated
interesting discussions concerning the adaptability of
these techniques and how to develop them further for
truly three-dimensional objects.
The collection and integration of various image modal-
ities into a self-consistent model, including high-defini-
tion elemental maps gathered at the Australian Synchro-
tron, has played an integral role in the conservation of
the Hartog Plate and led to a better understanding of its
degradation. It has also added invaluable new insights
into its production and history, including several unex-
pected discoveries. A balance has been achieved between
respecting the extreme physical limitations of the plate
and greatly improving accessibility for researchers and the
public. The technologies and tools applied in the study
and treatment of the Hartog Plate serve as an example
of what is already possible and what direction further
developments might take.
Acknowledgements
The authors would like to thank Gijs van der Ham, Robert
van Langh, Maarten Sepers (Panoptes Heritage), David
Patterson, Henk van Keulen, Bas Hesselink (IDAP),
Richard Mulvaney, Dutch Embassy in Canberra, XFM
beamline at the Australian Synchrotron (part of ANSTO),
Lucinda Timmermans, Joosje van Bennekom, Ellen van
Bork, and Sara Creange.
References
Bertholon, R. 2001. To get rid of the crust or not: Emer-
gence of the idea of ‘original surface’ in the conservation
of metal archaeological objects in the first half of the 20th
century. In Past practice – Future prospect, eds. A. Oddy
et al., 5–11. London: British Museum Press.
116  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND ANALYTICAL INVESTIGATION

Dubbe, B. 1978. Tin en tinnegieters in nederland. Lochem: David Hallam is a senior conservator and director at
De Tijdstroom. RM Tait and Associates, as well as honorary research
Halls, C. 1964. Two plates. Westerly 9(1): 33–40. associate at the Queen Victoria Museum & Art Gallery
in Launceston, Tasmania.
MacLeod, I.D., D. Thurrowgood, G. Pohl, D. Howard,
and D. Patterson. 2014. Centuries of decay revealed by Daryl Howard is a chemist at the x-ray fluorescence
synchrotron analysis of the de Vlamingh 1697 pewter microscopy beamline of the Australian Synchrotron,
plate. In ICOM-CC 17th Triennial Conference Preprints, part of ANSTO.
Melbourne, 15–19 September 2014, ed. J. Bridgland, art.
0903, 6 pp. Paris: International Council of Museums.
Major, R.H. 1859. Early voyages to Terra Australis, now
called Australia. London: The Hakluyt Society.
Plenderleith, H.J. and R.M. Organ. 1953. The decay
and conservation of museum objects of tin. Studies in
Conservation 1(2): 63–72.
Plenderleith, H.J. 1962. The conservation of antiques and
works of art: Treatment, repair, and restoration. Oxford:
Oxford University Press.
Sigmond, P. 2003. Two pewter plates. In het Kielzog.
Maritiem-historische studies aangeboden aan Jaap R.
Bruijn bij zijn vertrek als hoogleraar zeegeschiedenis aan
de Universiteit Leiden. Amsterdam: De Bataafse Leeuw.
Thomas-Goorieckx, D. 1963. Le plat de Hartogsz du
Rijksmuseum: examen et traitement particulier d’un
étain. Bulletin IRPA VI: 69–79.

Authors
Tamar Davidowitz is a metals conservator at the
Rijksmuseum as well as lecturer and program coordinator
at the Metals Conservation Program at the University of
Amsterdam.

David Thurrowgood is a conservator at the Queen

Victoria Museum and Art Gallery and director at Applied
Conservation Science Pty Ltd.

Robert Erdmann is currently a senior scientist at the

Rijksmuseum and also a professor at the University of
Amsterdam, where he holds the chair in Conservation
Science.

Ian D. MacLeod has forty years of experience in

conservation of metals recovered from marine envi-
ronments.

Dudley Creagh is an emeritus professor in Cultural

Heritage Studies and author of definitive papers on the
scientific use of synchrotron radiation.
 PREVENTIVE
CONSERVATION
ER-Based Sensors for Monitoring Atmospheric Corrosivity
on Corroded or Patinated Copper-Based Objects
Chiara Petiti* Marco Faifer
Politecnico di Milano Politecnico di Milano
Milan, Italy Milan, Italy
chiara.petiti@polimi.it marco.faifer@polimi.it
*Author for correspondence
Abstract
Corrosivity sensors are typically material representations of
works of art, allowing the corrosion rate to be monitored and
information on environmental corrosivity to be obtained.
Several types of sensors have been proposed for real-time
monitoring in the industrial field, but few solutions have
been reported for cultural heritage applications. Within the
framework of the European MUSECORR project, automated
real-time electrical resistance (ER) sensors were developed
with very interesting results (Prosek et al. 2013). However, the
sensors proposed so far for bronze and copper-alloy artefacts
are basically made of uncorroded alloys, while the real surfaces
of cultural heritage objects are usually more complex in
stratigraphy due to the presence of corrosion layers, which
can heavily affect the corrosion behaviour of the material. Thus,
an improved version of ER-based corrosion sensors, which
should be pre-corroded or artificially patinated, is proposed. It
has to be underlined that the production of artificial corrosion
layers on a metallic track is a critical issue, because it requires
Introduction
The monitoring of environmental corrosivity plays a
crucial role in the preservation of metallic artworks. It
provides the possibility to not only verify whether the
environment in which an artefact is stored or exposed to
could be harmful to their conservation, but also select the
optimal conservation conditions. In fact, when possible,
the selection of the optimal environmental conditions
represents the first step in preventive conservation.
Different approaches to the monitoring of environmental
corrosivity on the conservation of metallic artworks
have been proposed. Besides the monitoring of the
environmental parameters that can affect the corrosion
behaviour of metallic surfaces (e.g. temperature, relative
humidity, pollutants, etc.), one possible approach is to
expose a metallic coupon made of the same materials as
the examined artworks to the same environment. The
Michele Zanoni Sara Goidanich
Politecnico di Milano Politecnico di Milano
Milan, Italy Milan, Italy
michele.zanoni@polimi.it sara.goidanich@polimi.it
the track to have a higher thickness, which would affect the
sensitivity of the ER measurements. In order to find the optimal
compromise between sensitivity of the sensor, dimensions and
representativeness of the surface analysed, a new model of
metallic track was designed to maximise the exposed surface,
thus increasing the sensor’s sensitivity. In order to reduce costs
and ensure flexibility, a printed circuit board (PCB) solution
was selected. Temperature compensation and measurement
uncertainty were also considered. The aim of the work was
to develop new customised sensors for which the alloy of
the track and the nature of the artificial corrosion layers can
be chosen according to the specific characteristics of each
application. The influence of the presence of the corrosion
layers on the corrosion rate was evaluated.
Keywords
ER sensors, environmental corrosivity, monitoring, copper,
corrosion layers
corrosion rate is then evaluated by performing a series
of measurements on the coupons. This approach also
provides the possibility of performing invasive measure-
ments directly on the coupons (La Gennusa et al. 2005,
Ryhl-Svendsen 2008, de la Fuente et al. 2013, Thickett,
Chisholm, and Lankester 2013, Ingo et al. 2015).
Another approach to environmental corrosivity moni-
toring is the employment of environmental corrosivity
sensors. While for industrial applications a great number
of these types of sensors have been reported (Shitanda
et al. 2009, Mizuno et al. 2014, Nishikata, Zhu, and Tada
2014), few solutions have been specifically designed
for the cultural heritage field (Sjögren and Bozec 2004,
Prosek et al. 2013, Agbota et al. 2014, Goidanich et al.
2014). In addition to this, most of the sensors proposed
so far for monitoring environmental corrosivity, as well
ER-BASED SENSORS FOR MONITORING ATMOSPHERIC CORROSIVITY ON CORRODED OR PATINATED COPPER-BASED OBJECTS
as the European standard (International Organization
for Standardization 2012) used for its evaluation, do
not take into account the possible presence of corrosion
products that could significantly accelerate the corro-
sion processes and make the surfaces more sensitive to
even low relative humidity and specific pollutants. This
is of particular importance when dealing with historic
metallic surfaces, which usually present quite complex
stratigraphies of natural and/or artificial corrosion layers
that can significantly affect their corrosion behaviour.
By using galvanic sensors for the study of gilded bronzes,
it was verified (Goidanich et al. 2014) that the corrosion
behaviour of copper alloys can change significantly
depending on the composition of the corrosion layers,
particularly in the presence of chlorides. The effect of
corrosion layers may become particularly relevant when
dealing with archaeological artefacts, especially in the
presence of chlorides (Dracott 2014).
ER sensors are based on electrical resistance measure-
ments that are performed periodically. ER variations are
proportional to the instantaneous corrosion rate.
For this work, a new type of ER-based sensor for the
monitoring of environmental corrosivity on copper-based
surfaces was proposed. These sensors have to be pre-cor-
roded and pre-aged before exposure in order to ensure it
is representative of the real conditions affecting metallic
artefacts. For this, a custom geometry was designed and
developed in order to find the optimal compromise
between sensitivity, durability and representativeness
of real historic surfaces. In fact, the artificial patination
and ageing should only be performed on quite thick
metallic tracks in order to have sufficient durability,
with a consequent decrease in the sensitivity of the
sensors. The sensors were therefore developed in order
to maximise the exposed surface of the metallic track
to increase sensitivity. The aim of the work was, thus, to
develop new customised sensors for which the alloy of
the track and the nature of the artificial corrosion layers
can be chosen according to the specific characteristics
of each type of application.
Materials and methods
Sensors: Structure, geometry and measurements
In order to define the requirements of the measurement
system and the optimal dimensions for the new sensor,
119
a lab prototype was developed. The prototype was
designed in consideration of the needs of pre-corrosion,
cost and testing flexibility. The implemented solution is
based on printed circuit board (PCB) technology so as
to reduce the cost. The sensor size was defined by taking
into account the typical corrosion rates, the measuring
capability (in terms of accuracy and resolution) of a
multi-channel high-resolution multimeter (Keithley
3706) and the production constraints of PCB technology.
Under these constraints, the implemented solution is a
double-layer PCB with a copper thickness of 17 µm and
the layout shown in Figure 1.
The sensor is basically composed by four sections, two
for each face of the PCB, allowing testing of different
corrosion rates and introducing temperature compensa-
tion. For this experimentation, the two sections of each
face were connected (terminals E and E’ in Figure 1) so
that one full PCB face was exposed to corrosion (Rcorr)
while the other was protected and used for the temper-
ature compensation (Rref ). In fact, considering that the
ER values are heavily reliant on the temperature, one of
the two faces was protected and worked as a reference
resistance (Rref ). A four-wire ER measurement method
was adopted to estimate both Rcorr and Rref. For each face
of the PCB, the current was introduced via terminal A
and collected at terminal B, while the voltage drop was
measured between terminals C and D. In this way, no
contact resistance was included in the ER measurement.
The ER measurements were performed with a high-pre-
cision Keithley 3706 multimeter allowing four-wire
Figure 1. Sensor
layout
 
120  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PREVENTIVE CONSERVATION

ER measurements. The data acquisition and analysis Electrochemical measurements

were done with a virtual instrument (VI) application
Measurements of linear polarisation resistance (LPR)
controlling the high-resolution multimeter.
and electrochemical impedance spectroscopy (EIS) were
All sensors were initially cleaned with ethanol and performed with an Ivium Technologies CompactStat
exposed without corrosion layers on the surface. Four portable potentiostat with Ivium software. A three-elec-
of them (S01, S02, S03 and S06) were then treated in trode contact probe (CP) cell (Letardi 2002) configura-
order to produce artificial corrosion layers, while two tion was used comprising an AISI316L stainless steel
others (S04 and S05) were not, in order to act as refer- counter and pseudo-reference electrodes embedded in
ence sensors. a polytetrafluoroethylene (PTFE) case. Oligo-mineral
water with a pH of around 7.5 and conductivity of around
The sensors were placed in a climatic chamber and
200 µS/cm was used as an electrolyte. LPR measurements
monitored for more than 100 days at a temperature of
were performed after 10 minutes of open-circuit poten-
25°C–30°C and an RH of 50%–85%.
tial monitoring, varying the potential by ± 10 mV with
respect to Ecorr, and at a scan rate of 10 mV/min. For
Procedures for the realization of corrosion layers
the EIS measurements, the following setup was adopted:
The selection of the patination procedures represented a frequency range of 100 kHz–10 mHz at ± 10 mV with
a crucial step in the setup of the sensors. The aim was to respect to Ecorr. Polarisation resistance (Rp) values were
produce representative artificial patinas of the composi- obtained both from LPR and EIS measurements (Zhang
tion of a series of corrosion products commonly found et al. 2002).
on the surfaces of cultural heritage objects. In addition,
some technical and practical constraints were considered. Results and discussion
Due to the presence of a polymeric support for the PCB
Temperature compensation is obtained by calculating
metallic track, procedures requiring high temperatures
the ratio between Rcorr of the face that is exposed to
were discarded. In addition, highly controllable patina-
the atmosphere and Rref of the track on the coated side,
tion procedures were selected in order to promote the
because it must not be affected by corrosion processes
growth of corrosion products only on the metallic track
in order to have proper temperature compensation.
surface, avoiding the formation of conductive products
Corrosion is expected instead to cause a reduction of
between the ‘arms’ of the track.
the thickness of the copper track on the Rcorr side and
Therefore, two patination methodologies were selected: therefore an increase in the Rcorr/Rref ratio. To help the
on one side, a chemical method was adopted for the comparison between the different tested sensors, the
production of cuprite, while an applied paste method was ΔRatio% was calculated and used as follows:
employed for the production of a copper chloride and
sulfate-based patina (Hughes and Rowe 1991, Goidanich
et al. 2014). The cuprite patina was selected because it is
the first corrosion product that normally forms on copper
alloys. It was obtained after a 30–60 min immersion
of the sensor in a boiling solution containing 6.5 g/L
CuSO4·5H2O, 1.25 g/L Cu(CH3COO)2, 2 g/L NaCl and
where the subscript t and 0 indicate quantities measured
1.25 g/L KNO3 (Hughes and Rowe 1991, Goidanich et al.
respectively at generic time t and at the starting point of
2014). The chloride and sulfate-based patina was instead
the monitoring. With a series of assumptions, it is also
selected in order to reproduce the surface conditions of
possible to calculate the variation in thickness from the
highly unstable artefacts from a corrosion point of view.
ΔRatio%. ΔRatio% was used in the following graphs,
It was obtained by mixing CuSO4·5H2O, CuCl2 and CuCl
since it allows the obtained results to be represented
(ratio 4:3:1) finely ground in a mortar and adding water
without making assumptions or approximations.
(at a ratio of 2:1 with respect to the copper salt mixture)
in order to obtain a diluted paste to be applied over the In Figure 2, the values of the ΔRatio% for a selection of
copper surface. the tested sensors are reported for the first monitoring
ER-BASED SENSORS FOR MONITORING ATMOSPHERIC CORROSIVITY ON CORRODED OR PATINATED COPPER-BASED OBJECTS 121

week, during which none of the sensors had been arti- a)

ficially pre-corroded. The temperature ranged between
25°C–30°C and the RH between 50%–60%. There is
apparently no significant variation in the ΔRatio% and
the observed fluctuation can be mainly ascribed to
noise. In terms of thickness reduction, the noise can be
evaluated at about 1 nm.

The lack of significant variation in the ΔRatio% could

be associated both with the low corrosion rate of copper
in such conditions (Prosek et al. 2014) and the fact that
the selected geometry for the sensors aimed to monitor
the environmental corrosivity of surfaces with corrosion b)
layers that are expected to heavily accelerate their corro-
sion rate. Further tests are planned for longer monitoring
periods and sensors with different thicknesses, to be
defined based on surface conditions.

Figure 2. ΔRatio%
monitoring
of selected
sensors before
production of the
artificial corrosion
layers
Production of artificial corrosion layers
In Figure 3, the different phases of production and
stabilisation of corrosion layers can be easily recognised
by following the changes in ΔRatio% for sensors S01,
S02, S03 and S06. Sensors S02, S03 and S06 were all
pre-corroded by using the applied paste method in order
to obtain corrosion layers rich in copper chlorides and
sulfates. The differences between them are due to slightly
different procedures for applying the paste. The sudden
increase observed in ΔRatio% for sensor S03 after one
week is related to the production of a corrosion layer
of copper chlorides and sulfates by repeating the paste
method twice an hour. A lower increase in ΔRatio%
was recorded on the same day on S02 and S06 when the
paste method was applied only once. When the corrosion
layers dried, S02 and S03 were returned to the climatic
chamber, while S06 was placed in a beaker containing
demineralised water.
The ΔRatio% increase recorded on S02 and S06 after
13 days is related to the application of a second layer
Figure 3. ΔRatio%
monitoring of
sensors (a) before
and (b) after
production of the
artificial corrosion
 layers
by the applied paste method. On the same day the
cuprite layer was produced on sensor S01. Once the
corrosion layers dried, S01 and S02 were returned to the
climatic chamber, while S06 was again placed in a beaker
containing demineralised water.
The sudden increase in ΔRatio% recorded on S06 on
day 15 corresponds to its removal from the solution
and placing in a climatic chamber. This increase should
not be seen as a sudden reduction in the section of the
sensor but related to the change in environment, since
the conductibility of the solution contaminated by the
corrosion products probably reduced the measured
resistance of sensor S06 while it was in the solution.
The slow increase in ΔRatio% recorded on all sensors
on day 19 was related to a significant increase in relative
humidity inside the chamber. This was raised to about
85% and led to an increase in the corrosion rate.
The total loss in thickness of the tracks was about
1.7 microns for S03, about 1.3 microns for S02, about
1 micron for S06 and about 0.3 microns for S01. It can
be concluded that the chemical procedure followed to
produce the cuprite layer was less aggressive compared to
122  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
the applied paste method which caused deeper corrosion of
the sensor with a drastic reduction in the thickness of the
copper track (about 10% for S03 after about one month).
The chemical method therefore looks like a more suitable
and less damaging approach for the sensors during the
production phase of the artificial corrosion layers.
Among the three procedures tested based on the applied
paste method, the application of two layers in a short time
period resulted in the most critical procedure leading to
the highest increase in ΔRatio% and therefore reduction
in the thickness of the copper track. The least aggressive
procedure was the one followed for sensor S06 that
included immersion in demineralised water after the
application of the paste.
Monitoring of stabilised corrosion layers
Considering the high increase in ΔRatio% recorded for
all sensors, they were all left to stabilise in a climatic
chamber with relative humidity ranging between 50 %
and 60% for about one month after the production of the
second corrosion layer. After that period, the ΔRatio%
was recalculated for all pre-corroded sensors to take day
45 as time=0 (Figure 4a). Unfortunately, sensor S02 had
to be discarded due to the loss of an electrical connection
with Rref.
It was confirmed that the selected geometry did not
allow detection of small reductions in thickness caused
by corrosion on the non-pre-corroded sensors (S04 and
S05) due to the short exposure time. The very small fluc-
tuations were not correlated to temperature or relative
humidity variations (Figure 4b) and must be ascribed to
noise, which has to be investigated.
The geometry is instead clearly suitable for the moni-
toring of environmental corrosivity for all pre-corroded
sensors since ΔRatio% always increased when RH% was
higher than 20%. No increase of ΔRatio% was detected
for RH% below 20% which may be simply due to the
fact that at such a low RH, corrosion is expected to be
negligible. The slope of the curve increases when RH
increases, suggesting, as expected, an increase in the
corrosion rate. However, if RH is kept constant, the
slope of the curve slowly decreases, suggesting that the
corrosion rate tends to reduce with time.
The S03 sensor had the most corrosive corrosion layers.
S01 and S03 initially had quite similar behaviour after
PREVENTIVE CONSERVATION
a)
b)
Figure 4. (a)
ΔRatio%
monitoring of
non-pre-corroded
sensors and of
sensors after
stabilisation
of the artificial
corrosion layers;
(b) monitoring of
temperature and
 relative humidity
stabilisation. However, after 70 days, S01 showed a series
of instability phenomena similar to those recorded for
sensor S02, and therefore had to be discarded. Tests with
cuprite patina (S01) need, therefore, to be repeated in
order to confirm the corrosion behaviour of pre-corroded
specimens with this chemical procedure.
Further experimentations are planned on sensors made
of different copper alloys, such as bronze. However,
considering the results presented and those reported
in the literature (Stöckle et al. 1998, Chang et al. 2019),
it is expected that the presence of corrosion layers may
affect the corrosion behaviour of copper-alloy substrates
much more than the alloy composition. In this sense, the
procedure followed for the production of artificial patinas
or corrosion layers definitely represents a critical step.
The use of chemicals, which can partially remain among
the corrosion products, may indeed affect the corrosion
behaviour compared to naturally developed layers. It is
therefore planned to perform tests with sensors after
natural and accelerated ageing.
LPR and EIS measurements were performed on bare-
copper specimens and on copper specimens that were
ER-BASED SENSORS FOR MONITORING ATMOSPHERIC CORROSIVITY ON CORRODED OR PATINATED COPPER-BASED OBJECTS
subjected to the same treatments as sensors S01, S03 and
S06. As shown in Figure 5, the polarisation resistance
(Rp) obtained, which is inversely correlated to the corro-
sion rate, confirms that the highest value is recorded for
non-pre-corroded surfaces (35 ohm/m2) and that the
surface with the cuprite layer presents a significantly lower
Rp (0.2 ohm m2). The surfaces with copper chlorides and
sulfates were characterised by extremely low Rp values,
which confirms the high corrosion caused by this kind of
corrosion layer on a copper substrate. The low quality of
Nyquist plots (not shown here) obtained by EIS measure-
ments also suggests the instability of this kind of patina.
a)
b) Figure 5. (a) Rp
values measured
on non-pre-
corroded and
artificially
patinated copper
specimens;
(b) Bode
plots from EIS
measurements on
non-pre-corroded
and artificially
patinated copper
 specimens
Conclusion
The proposed geometry was shown to be suitable for
the implementation of pre-corroded sensors to monitor
environmental corrosivity on the corroded surfaces of
copper-alloy objects.
It was confirmed that the presence of corrosion layers on
the surface can heavily affect the corrosion behaviour of
copper objects, confirming the need for ER pre-corroded
sensors for more precise monitoring.
The proposed methods for the production of artificial
corrosion layers were observed to be suitable. However,
123
the applied paste method caused severe corrosion to
the sensors and it would be advisable to test further
procedures with a lower impact on the service life of
the sensors.
References
Agbota, H., J.E. Mitchell, M. Odlyha, and M. Strlič. 2014.
Remote assessment of cultural heritage environments
with wireless sensor array networks. Sensors 14: 8779–93.
Chang, T., G. Herting, S. Goidanich, J.M. Sánchez Amaya,
M.A. Arenas, N. Le Bozec, Y. Jin, C. Leygraf, and I. Odne-
vall Wallinder. 2019. The role of Sn on the long-term
atmospheric corrosion of binary Cu-Sn bronze alloys in
architecture. Corrosion Science 149: 54–67.
de la Fuente, D., J.M. Vega, F. Viejo, I. Díaz, and M.
Morcillo. 2013. Mapping air pollution effects on atmos-
pheric degradation of cultural heritage. Journal of Cultural
Heritage 14(2): 138–45.
Dracott, J.P. 2014. Piezoelectric printing and pre-cor-
rosion: Electrical resistance corrosion monitors for the
conservation of heritage iron. PhD dissertation, School
of Materials, University of Manchester, UK.
Dubus, M. and T. Prosek. 2012. Standardized assessment
of cultural heritage environments by electrical resistance
measurements. e-PRESERVATION Science 9: 67–71.
Goidanich, S., D. Gulotta, L. Brambilla, R. Beltrami, P.
Fermo, and L. Toniolo. 2014. Setup of galvanic sensors for
the monitoring of gilded bronzes. Sensors 14: 7066–83.
Hughes, R. and M. Rowe. 1991. The colouring, bronzing
and patination of metals: A manual for fine metalworkers,
sculptors and designers. London: Thames and Hudson.
Ingo, G.M., E. Angelini, C. Riccucci, T. de Caro, A. Mezzi,
F. Faraldi, D. Caschera, C. Giuliani, and G. Di Carlo. 2015.
Indoor environmental corrosion of Ag-based alloys in
the Egyptian Museum (Cairo, Egypt). Applied Surface
Science 326: 222–35.
International Organization for Standardization. 2012.
Corrosion of metals and alloys – Corrosivity of atmospheres
– Classification, determination and estimation (BS EN
ISO 9223:2012, last reviewed and confirmed in 2017).
La Gennusa, M., G. Rizzo, G. Scaccianoce, and F. Nico-
letti. 2005. Control of indoor environments in heritage
buildings: Experimental measurements in an old Italian
museum and proposal of a methodology. Journal of
Cultural Heritage 6(2): 147–55.
124  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Letardi, P. 2002. Outdoor bronze protective coatings:
Characterisation by a new contact-probe electrochemical
impedance measurements technique. In Proceedings of
the First Meeting the Science of Art, Bressanone, Italy, 26
February–1 March 2001, ed. P.A. Vigato, 173–78.
Mizuno, D., S. Suzuki, S. Fujita, and N. Hara. 2014. Corro-
sion monitoring and materials selection for automotive
environments by using Atmospheric Corrosion Monitor
(ACM) sensor. Corrosion Science 83: 217–25.
Nishikata, A., Q. Zhu, and E. Tada. 2014. Long-term
monitoring of atmospheric corrosion at weathering
steel bridges by an electrochemical impedance method.
Corrosion Science 87: 80–8.
Prosek, T., M. Kouril, M. Dubus, M. Taube, V. Hubert, B.
Scheffel, Y. Degres, M. Jouannic, and D. Thierry. 2013.
Real-time monitoring of indoor air corrosivity in cultural
heritage institutions with metallic electrical resistance
sensors. Studies in Conservation 58: 117–28.
Prosek, T., M. Taube, F. Dubois, and D. Thierry. 2014.
Application of automated electrical resistance sensors for
measurements of corrosion rate of copper, bronze and
iron in model indoor atmospheres containing short-chain
volatile carboxylic acids. Corrosion Science 87: 376–82.
Ryhl-Svendsen, M. 2008. Corrosivity measurements of
indoor museum environments using lead coupons as
dosimeters. Journal of Cultural Heritage 9(3): 285–93.
Shitanda, I., A. Okumura, M. Itagaki, K. Watanabe, and
Y. Asano. 2009. Screen-printed atmospheric corrosion
monitoring sensor based on electrochemical impedance
spectroscopy. Sensors and Actuators B: Chemical 139:
292–97.
Sjögren, L. and N.L. Bozec. 2007. On-line corrosion
monitoring of indoor atmospheres. In EUROCORR
2004 Proceedings: Long-Term Prediction & Modelling of
Corrosion, Nice, France, 10–14 September 2004. European
Federation of Corrosion (EFC).
Stöckle, B., A. Krätschmer, M. Mach, and R. Snethlage.
1998. Report no. 23: Corrosion attack on copper and cast
bronze. Evaluation after 8 years of exposure, Convention
on Long-Range Transboundary Air Pollution. UN/ECE
International Co-operative Programme on Effects on
Materials, Including Historic and Cultural Monuments.
Munich, Germany: Bavarian State Conservation Office.
Thickett, D., R. Chisholm, and P. Lankester. 2013.
Reactivity monitoring of atmospheres. In Metal 2013:
PREVENTIVE CONSERVATION
Proceedings of the Interim Meeting of the ICOM-CC Metals
Working Group, Edinburgh, Scotland, 16–20 September
2013, eds. E. Hyslop, V. Gonzalez, L. Troalen, and L.
Wilson, 129–36. Edinburgh: ICOM-CC and Historic
Scotland.
Zhang, X.Y., W.L. He, I.O. Wallinder, J. Pan, and C.
Leygraf. 2002. Determination of instantaneous corrosion
rates and runoff rates of copper from naturally patinated
copper during continuous rain events. Corrosion Science
44: 2131–51.
Materials list
Chemicals were provided by Sigma-Aldrich
(https://www.sigmaaldrich.com/portugal.html ):
Cu(CH3COO)2 ACS Reagent grade
CuCl and CuCl2 Reagent grade
CuSO4∙5H2O BioReagent
KNO3 ≥ 99% pure
Na2SO4 ReagentPlus grade
NaCl ≥ 99.5% pure
Authors
Chiara Petiti is a PhD student in Materials Engineering
from the Chemistry, Materials and Chemical Engineering
Department at the Politecnico di Milano.
Marco Faifer is an associate professor in the Electronics,
Information and Bioengineering Department at the
Politecnico di Milano.
Michele Zanoni is a PhD student in Electrical Engi-
neering in the Electronics, Information and Bioengi-
neering Department at the Politecnico di Milano.
Sara Goidanich is an assistant professor in the Chem-
istry, Materials and Chemical Engineering Department
at the Politecnico di Milano.
Metal Conservation, Cellulose Nitrate and the Oddy Test
Gerhard Eggert* Capucine Korenberg
State Academy of Art and Design British Museum
Stuttgart, Germany London, UK
gerhard.eggert@abk-stuttgart.de CKorenberg@britishmuseum.org
Rebekka Kuiter Julia Ziegler
Museum und Park Kalkriese Museum of Ethnology
Bramsche-Kalkriese, Germany Dresden, Germany
rebekka.kuiter@kalkriesevarusschlacht.de julia.ziegler@skd.museum
Abstract
Conservation materials such as coatings, adhesives and
consolidants are often applied to museum objects and, as such,
are present in stores or display cases. It is necessary to check
they do not emit volatiles that are harmful to object materials
and one way this can be done is with the Oddy test – a material
screening carried out at an elevated temperature. Cellulose
nitrate (CN) is routinely used by conservators as an adhesive and
a lacquer for silver objects. A commercial ready-to-use variant of
dissolved CN – HMG – was Oddy tested by the State Academy of
Art and Design Stuttgart (ABK) (where it failed) and the British
Museum (BM) (where it passed), the only difference being that
at the ABK the sample had been left to dry for two months,
while at the BM only three days of drying were allowed, as per
the standard BM protocol. When the HMG samples were dried
for two months they consistently failed the Oddy test at both
institutions. Traces of solvent left in the resin might act as a
Introduction
It has become common practice to test the suitability of
all materials used for the storage and display of museum
objects with the Oddy test. In this test, a sample of mate-
rial is exposed to 60 °C at 100% RH for four weeks in a
test tube containing three freshly abraded metal coupons
(copper, lead and silver). Corrosion products forming
on the coupons can be indicative of the emission of
harmful pollutants from the sample. Two decades after
the development of the Oddy test in the 1970s to assess
all materials used in display cases at the British Museum,
Green and Thickett (1995, 145) stated that ‘no occur-
rences of corrosion due to outgassing of appropriately
tested display materials [had] been noted’. However, in
recent decades during the development of the Oddy test,
it was noted that to obtain reproducible results, strict
adherence to a standardised protocol has been found
necessary (Thickett and Lee 2004, 12).
Sebastian Bette Jörg Stelzner
Max Planck Institute for Solid State State Academy of Art and Design
Research Stuttgart, Germany
Stuttgart, Germany joerg.stelzner@abk-stuttgart.de
s.bette@fkf.mpg.de
*Author for correspondence
stabiliser (radical scavenger) hindering CN decay. Further Oddy
tests on Paraloid B-72 showed that, when dissolved in ethyl
acetate and cured, traces of solvent in the resin corroded the
lead coupons. This demonstrates that resins must be fully cured
before their volatile emissions are assessed and it is necessary
to re-consider the drying of solvent-based products in the BM
standard Oddy test protocol. In the Oddy test for CN, silver
cyanide nitrate was detected on the silver coupons. Even tests at
room temperature showed cyanide compounds formed on silver
when exposed to HMG. CN products could pose a risk to museum
artefacts and conservators should carefully consider their use
and select alternative tested products as much as possible.
Keywords
conservation materials, cellulose nitrate, cyanide, Frigilene,
HMG, Oddy test, silver
Conservation materials are added to museum objects
so are also present in stores or display cases. Volatile
emissions from them might be harmful to the object to
which they have been applied or to objects in the vicinity.
Therefore, it is necessary to ensure that conservation
materials do not emit harmful volatiles. But this is not
done systematically. Recently, products based on cellulose
nitrate (CN) used in conservation were tested at the State
Academy of Art and Design Stuttgart (ABK) using the
Oddy test and failed (Ziegler et al. 2014).
The present study focuses on a CN commercial product,
HMG, commonly used in Great Britain as an adhesive
in metal conservation. While it passed the Oddy test at
the British Museum (BM) (Korenberg et al. 2018, 12) in
2014, it dramatically failed the test at the ABK. The Oddy
tests were repeated at both institutions in the summer of
126  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
2016 using the same product purchased from the same
supplier at the same time, but different results were still
obtained. The aim of this subsequent study was to clarify
the reasons for this striking disparity, carefully comparing
the Oddy test protocols followed by each institution.
Experiments investigating some of the factors in the Oddy
test protocols were carried out. Their methods are given
below and the results are discussed jointly.
Comparison of the Oddy test protocols at ABK
and BM
If not stated otherwise, tests used the standard protocol
for the ‘3-in-1’ version (Lee and Tickett 2004, 15–16).
A blank test was prepared for every batch of Oddy tests
as a control.
Sample preparation
At the ABK, a silicone mould was initially used to cast
small blocks (76 × 26 mm) of samples of 2 g (Ziegler et
al. 2014), while the BM used a thin layer applied on a
6 × 12 cm film of Melinex without controlling the mass
(Korenberg et al. 2018, 12). Following the discovery of
the disparity between the test results, the tests at the ABK
were repeated using thin films of HMG on Melinex and,
although less corrosion was formed, the HMG still failed
the test (Figure 1). The freshly cast silicone (Elastosil
M 4600) was tested on its own and found to cause slight
corrosion in the Oddy test. This may explain why more
corrosion was initially observed with the blocks of HMG.
Figure 1. Silver, lead and copper coupons from an Oddy test on HMG, a
commercial cellulose nitrate product. Left: reference trio not exposed to
HMG. Right: trio exposed to HMG (2 g, two months' drying)
Amount of water
In the Oddy test protocol, water is inserted into the test
tube to create 100% RH inside. In the standard Oddy test
PREVENTIVE CONSERVATION
protocol of 1995 (Thickett and Lee 1995), the amount of
water was set to 0.5 mL for a 50 mL test tube and later
reduced to ca. 0.17 mL for tests of non-hygroscopic mate-
rials (Korenberg et al. 2018, 3). They found that using
more water caused the formation of corrosion on the lead
coupon, even in the control test with no material, most
likely due to water condensation. In the original tests at
the ABK, 0.5 mL of water was used. To check whether the
larger amount of water was responsible for the formation
of corrosion, tests with lead only were conducted using
0.15 mL of water. Contrary to expectations, even more
corrosion was formed on the lead coupon. When the
test tubes were opened at the end of the tests, a pungent
smell was detected and the HMG sample had become
brittle. It was concluded that variations in the amount
of water were not responsible for the corrosion observed
in the tests at the ABK.
Amount of sample
As BM and ABK used different sample preparation
methods, different amounts of sample were used in their
tests, i.e. 6 × 12 cm (which corresponded to approximately
0.5 g) and 2 g respectively. When 2 g of HMG was spread
as a thin film on Melinex, it was found to cover a large
area (ca. 20 × 30 cm) and it was difficult to place it into the
test tube because of lack of space and electrostatic forces.
Tests with lead coupons and different amounts of HMG
(0.5, 1.0 and 2.0 g on Melinex) found, as expected, that
more corrosion corresponded to more material. Never-
theless, even 0.5 g HMG caused considerable corrosion
on lead and was also rated as ‘unsuitable’ according to the
Oddy test protocol. Therefore, the increased size of the
sample compared to the BM does not explain the forma-
tion of corrosion in the Oddy test observed at the ABK.
Drying time
To avoid any influence from retained solvent on the
performance of HMG, samples were dried for two
months at the ABK, while the BM dried for two days or,
in subsequent tests, for a week. However, when samples
applied on Melinex were left to dry for two months at
the BM, they failed the Oddy test. Samples of HMG
on Melinex were also left to dry for one, two or three
days at the ABK (using 0.5 mL water for the Oddy test)
and the amount of corrosion on the coupons decreased
dramatically. These tests showed the drying time was the
main factor responsible for the formation of corrosion.
METAL CONSERVATION, CELLULOSE NITRATE AND THE ODDY TEST 127

A difference in drying time is likely to cause a difference Mechanism of corrosion reaction on silver
in the amount of solvent retained in the film. CN is
As Ziegler et al. (2014, 6) reported, the most unusual
prone to decay by radical chain reactions (Selwitz 1988)
result from our initial tests was observed for the silver
and solvent molecules left in the samples might act as
coupons, on which extreme corrosion was documented
stabilisers (free radical scavengers), as plasticisers do.
(Figure 1). The product was identified as silver cyanide
The effect of the drying time on solvent-cured resins was nitrate (AgCN·2AgNO3) using X-ray diffraction and
further investigated using Paraloid B-72, which is known µ-Raman spectroscopy. Thickett and Hockey (2003,
for its high retention of solvent molecules (Horie 2010, 158) had already reported the occurrence of a corrosion
155). Solid pellets of Paraloid B-72 passed the Oddy test. product on the silver coupons of their Oddy tests when
Paraloid B-72 dissolved in ethyl acetate and left to dry they tested two CN products – Frigilene and Ercalene –
for two months (blocks) or five weeks (film on Melinex) and assumed it was silver nitrate with another material,
caused corrosion on the lead coupon. Only when left to possibly silver carbonate, using Fourier transform infrared
dry for one year did samples pass the Oddy test. (FTIR) spectroscopy (Figure 2). However, it is more likely
The corrosion product on the lead coupon was identified that silver cyanide nitrate formed under the highly acidic
as lead acetate. The acetate can only originate from the conditions produced by the decay of CN than carbonates.
hydrolysis of solvent traces present in the resin sample. To ascertain this, silver cyanide nitrate was synthesised at
In fact, the decomposition of ester into ethanol and acetic ABK by adding AgCN to a concentrated aqueous solution
acid is known to occur in ethyl acetate at higher humidity of AgNO3 and leaving the solution to dry (Guo et al. 2000).
and temperatures similar to Oddy test conditions (see This reference compound was then analysed using FTIR
The Merck Index, 14th ed., s.v. ‘Ethyl acetate’). and compared with the spectra collected by Thickett. Silver
cyanide nitrate is a good match for the corrosion product
Discussion of the implications for the Oddy test (Figure 2, Table 1). Depending on the structural environ-
protocol ment, cyanide ions exhibit relatively sharp bands in the
2250–2050 cm-1 region for the stretch vibration of the triple
These findings show that long drying times (i.e. up
C≡N bond and a band at 2119 cm-1 was observed both for
to two months) are crucial when conducting Oddy
the reference compound and the corrosion product, but
tests on solvent-based resins to avoid false positive or
not for the other cyanide peak at 2163 cm-1. The reason
negative results, for both coatings and adhesives. This
for this is unclear. Raman spectroscopy can also be used
is in contrast with current experimental protocols. For
for identification (Figure 3).
instance, Green and Thickett (1995, 149) stipulate that
‘paints, varnishes and adhesives should be tested as
freshly cast films’ and the most recently published test
protocol (Korenberg et al. 2018, 7) requires four weeks
of drying for coatings and just two days for adhesives.
This needs to be reconsidered. A drying time of four
weeks, though not always sufficient, is often impractical
if the Oddy tests results are needed at short notice.
Accelerated drying (e.g. at 60 °C in the same heating
cabinets used for the Oddy test) before starting the
Oddy test (with 100% RH) might be suitable for accel-
erating the drying of solvent-based resins. This would
cause some additional thermal ageing of the resin and
the time for the Oddy test might need to be shortened
accordingly. A preliminary test with Paraloid B-72 in
ethyl acetate showed that even drying a film for one
Figure 2. Comparison of ATR-FTIR spectra of the corrosion product
week at 60 °C was not totally sufficient, but two weeks
present on the silver coupon in the Oddy test on Frigilene (open-air
worked. Modifying the drying in the Oddy test protocol IR microscopy, Thickett, personal communication) and on synthesised
will require further experimentation. AgCN·2AgNO3
128  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PREVENTIVE CONSERVATION

Table 1. Tentative assignment of the observed bands (Figure 2) in the Previous accelerated ageing tests have shown that silver
ATR-FTIR spectrum of synthetic AgCN·2AgNO3 by using the reported cyanide nitrate can form on silver coupons exposed to
band positions of solid AgCN (Bowmaker et al. 1998, Reckeweg and
HMG under relatively mild conditions, i.e. 40 °C and
Simon 2002) and AgNO3 (Martina et al. 2012)
75% RH for 60 days, and 20 °C and 100% RH for 45 days
Band no. Position/cm-1 Assignment (Ziegler et al. 2014). Another test was then conducted
(1) 2163, m ν(CN)
under ambient conditions to investigate this phenomenon
(2) 2119, m
further. Ceramics glued with HMG were covered with a
(3) 1352, sh νas(NO)
glass bowl with metal coupons attached, which was then
(4) 1291, m
glued with the same adhesive onto a glass plate to mimic
(5) 1036, s νs(NO)
display cases (Figure 4). These samples were then stored
(6) 806, s γ(NO) out-of-plane
for five years under ambient conditions (i.e. 20–30 °C
(7) 726, s γ(NO) in-plane
with no RH control or measurement). Although there
(8) 715, s
was no visible corrosion, when the silver coupons were
s: strong, m: medium, sh: shoulder
analysed using Raman spectroscopy,1 a band at 2168
cm-1 indicative of cyanide was observed (Figure 5) which
was not present on the control display test set up with
Paraloid B-72 (known to be stable) and on the blank
silver foil itself stored in the laboratory. This shows that
if an object treated with CN is displayed alongside silver,
cyanide could form on the silver surface in the long term.

Figure 3. Raman spectrum of synthesised AgCN·2AgNO3

The formation of silver cyanide nitrate on the silver

coupon in the Oddy test is unusual and implies that
nitrogen was reduced from the oxidation state +V (in
nitrate) to –III (in cyanide). AgCN·2AgNO3 has been Figure 4. Model display case used to expose an HMG CN treated object
for five years
reported to form when ethanol and an acidic silver nitrate
solution are mixed together (Britton and Dunitz 1965).
Such a reaction might also occur during the Oddy test:
as cellulose nitrate degrades thermally, the weak O-NO2
bonds of the ester groups split and NO2 is released. NO2
then reacts with nitric acid, which attacks silver and
forms silver nitrate. Carbon necessary for the formation
of cyanide can come from the breakdown of cellulose in
the CN (Selwitz 1988, 19).

Indeed, Pouliot et al. (2013) reported silver cyanide on

silver objects lacquered with CN. It is not clear if this was
restricted to the uppermost atomic layers or if a real 3D
AgCN phase had formed. However, it cannot be ruled
out that the objects might have been treated with cyanide Figure 5. Raman spectrum of the silver coupon stored for five years with
solutions in the past. HMG CN-treated object
METAL CONSERVATION, CELLULOSE NITRATE AND THE ODDY TEST
Given the widespread use of CN products in conserva-
tion, it is surprising that there have not been more reports
of silver cyanide (nitrate) corrosion on silver artefacts.
This could be because the amount of CN products used
as adhesives is relatively small compared to the volume
of display cases or stores. Alternatively, it is possible that
silver cyanide nitrate corrosion has formed on many silver
surfaces, but has not been identified as such since it can be
difficult to distinguish visually from the common sulfide
tarnish. It is recommended that more silver surfaces
treated with CN products are analysed with Raman or
ATR-FTIR spectroscopy for the presence of cyanide.
As CN products fail the Oddy test when fully cured, their
use should be avoided in conservation. Many adhesives
could be used instead of CN adhesives. For instance,
Stephen Koob (1986) showed more than 30 years ago that
Paraloid B-72 is well suited as an adhesive for glass and
ceramics and can be used as a replacement for unstable
cellulose nitrate (Koob 1982).
However, finding an alternative to CN products as a
lacquer for silver surfaces is more difficult. The tarnishing
of silver should be prevented as much as possible since
cleaning silver, whichever method is used, is a severe
intervention. It is possible to use scavengers in showcases,
but when silver artefacts are on open display the only
method to slow down tarnishing is the application of a
barrier coating. CN products, like Agateen and Frigilene,
are often used as silver lacquers ‘for [their] good leveling
qualities, ease and reliability of application, quick drying
and lack of color’ (Selwitz 1988, 53). Unfortunately, there
have been very few studies that have investigated the
effectiveness of alternative silver lacquers. De Witte’s
(1973–74) early study is summarised by Selwitz (1988, fig.
13). Reedy et al. (1999) found the CN product Agateen to
be a much better barrier to hydrogen sulfide (a pollutant
responsible for the tarnishing of silver) than Paraloid B-72
(first used in 1950 as silver lacquer (Horie 2010, 165)) and
Paraloid B-48N. Grabow et al. (2007) and Grissom et al.
(2013) tested a number of products, including Agateen
and HMG, and found an aqueous dispersion of Acrylsol
WS24 (0.6 % ammonia) promising. To assess whether
this product could be a suitable replacement for CN,
a film of Acrylsol WS24 was dried for one month and
Oddy tested. Unfortunately, corrosion formed on the
copper and lead coupons and was identified as tenorite,
massicot and litharge respectively using Raman spec-
troscopy. Apparently, WS24 favours metal oxidation. The
129
polyvinyl acetate AYAT was found to be a good barrier
against hydrogen sulfide, but it has a ‘dull appearance’
(Grabow et al. 2007, 47), which makes it unsuitable for
museum objects on display, and the low glass transition
temperature of 26 °C would cause the collection of dirt
on its surface. Therefore, until there is a better alternative
to CN products, the risk of using a CN lacquer on silver
surfaces and possibly causing some surface change needs
to be compared to the risk of exposing the silver surface
to pollutants and causing severe tarnishing.
Conclusion
HMG® failed the Oddy test when it had been left to dry for
two months before being tested, but passed when left to
dry for only three days. The mechanism of the corrosion
reaction was investigated for the silver coupons and it was
postulated the residue of solvent, still present in the resin
when it was left to dry for three days, acts as a stabiliser
of CN polymers (like plasticisers). For long drying times
(e.g. two months), it is likely that the solvent has fully
evaporated and the CN resin emits harmful volatiles.
Tests at room temperature showed cyanide compounds
also formed on silver when exposed to CN. Additional
tests on Paraloid B-72 showed that, when dissolved in
a solution of ethyl acetate and left to dry for up to two
months, traces of solvent in the resin caused corrosion to
lead coupons. Therefore, resins must be tested once they
are fully cured and it is recommended that the length of
the drying time of solvent-based resins in the Oddy test
protocol be reassessed, e.g. by testing stable solids (like
B-72) dissolved in corrosive solvents (like ethyl acetate).
Finally, CN products could pose a risk to museum arte-
facts, including silver, and conservators should carefully
consider their use and select alternative products when-
ever possible. To evaluate better the risk of CN products
to silver, it is recommended that corrosion products
on silver be systematically analysed for the presence of
cyanide compounds (for instance using FTIR or Raman
spectroscopy).
Notes
1 The micro-Raman spectra were recorded with a
Renishaw inVia Raman spectrometer (grating 1800L/mm,
resolution 1 cm−1) equipped with a Leica DMLM micro-
scope and a RenCam CCD detector. The spectra were
taken from 100–3700 cm−1 using a He−Ne laser operating
at 632.8 nm. The power of the laser was less than 400 μW
130  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
focused onto a 0.7 μm spot through a 50× microscope
objective.
References
Bowmaker, G.A., B.J. Kennedy, and J.C. Reid. 1998.
Crystal structures of AuCN and AgCN and vibrational
spectroscopic studies of AuCN, AgCN, and CuCN.
Inorganic Chemistry 37: 3968–74.
Britton, D. and J.D. Dunitz. 1965. The crystal structure
of AgCN·2AgNO3. Acta Crystallographica 19: 815–20.
de Witte, E. 1973–74. The protection of silverware with
varnishes. Bulletin de l’Institut royal du patrimoine artis-
tique 14: 140–51.
Grabow, N., C. Smith, C. Grissom, and L. Brostoff. 2007.
Effectiveness of organic coatings on silver exposed to
hydrogen sulfide. In Metal 07: Interim Meeting of the
ICOM-CC Metal Working Group, Amsterdam, 17–21
September 2007, Volume 5: Protection of Metal Artefacts,
eds. C. Degrigny, R. van Langh, I. Joosten, and B. Anker-
smit, 44–50. Amsterdam: Rijksmuseum.
Green, L.R. and D. Thickett. 1995. Testing materials for
use in the storage and display of antiquities: A revised
methodology. Studies in Conservation 40(3): 145–52.
Grissom, C.A., N. Grabow, C.S. Riley, and A.E. Charola.
2013. Evaluation of coating performance on silver
exposed to hydrogen sulfide. Journal of the American
Institute for Conservation 52(2): 82–96.
Guo, G.-C., Q.-M. Wang, and T.C.W. Mak. 2000. Variation
of [AgCN] chain configuration in silver(I) double salts:
Crystal structures of AgCN·2AgNO3 and AgCN·2Ag-
ClO4·2H2O. Inorganic Chemistry Communications 3:
313–15.
Koob, S.P. 1982. The instability of cellulose nitrate adhe-
sives. The Conservator 6: 31–34.
Koob, S.P. 1986. The use of Paraloid B-72 as an adhesive:
Its application for archaeological ceramics and other
materials. Studies in Conservation 31(1): 7–14.
Korenberg, C., M. Keable, J. Phippard, and A. Doyle.
2018. Refinements introduced in the Oddy Test meth-
odology. Studies in Conservation 63(1): 2–12.
Martina, I., R. Wiesinger, D. Jembrih-Simbürger, and M.
Schreiner. 2012. Micro-Raman characterisation of silver
corrosion products: Instrumental setup and reference
database. e-PRESERVATIONScience 9: 1–8.
PREVENTIVE CONSERVATION
Pouliot, B., C. Matsen, J. Mass, W. Donnelly, K. Andrews,
and M. Bearden. 2013. Three decades later: A status
report on the silver lacquering program at Winterthur.
In Objects Specialty Group Postprints, Volume Twenty,
eds. L. Kaplan, K. Dodson, and E. Hamilton, 32–48.
Washington, DC: American Institute for Conservation
of Historic & Artistic Works (AIC).
Reckeweg, O. and A. Simon. 2002. X-ray and Raman
investigations on cyanides of mono- and divalent metals
and synthesis, crystal structure and Raman spectrum
of Tl5(CO3)2(CN). Zeitschrift für Naturforschung B 57:
895–900.
Reedy, C.L., D.L. Long, R.A. Corbett, R.E. Tatnall, and
B.K. Krantz. 1999. Evaluation of three protective coatings
for indoor silver artifacts. In Objects Specialty Group
Postprints, Volume Six, comp. V. Greene and E. Kaplan,
41–69. Washington, DC: American Institute for Conser-
vation of Historic & Artistic Works (AIC).
Selwitz, C. 1988. Cellulose nitrate in conservation,
Research in Conservation 2. Los Angeles: The Getty
Conservation Institute.
Thickett, D. and M. Hockey. 2003. The effects of conser-
vation treatments on the subsequent tarnishing of silver.
In Conservation Science 2002, eds. J. Townsend, K.
Eremin, and A. Adriaens, 155–61. London: Archetype.
Thickett, D. and L.R. Lee. 2004. Selection of materials for
the storage or display of museum objects, rev. ed., British
Museum Occasional Papers 111.
Ziegler, J., C. Kuhn-Wawrzinek, M. Eska, and G. Eggert.
2014. Popping stoppers, crumbling coupons – Oddy
testing of common cellulose nitrate ceramic adhe-
sives. In ICOM-CC 17th Triennial Conference Preprints,
Melbourne, 15–19 September 2014, ed. J. Bridgland, art.
0505, 8 pp. Paris: ICOM.
Authors
Gerhard Eggert FIIC has a PhD in chemistry from the
University of Bonn. He was head of the Conservation
Department at Rheinisches Landesmuseum Bonn from
1985 to 1998 and has been chair of Objects’ Conserva-
tion at the State Academy of Art and Design Stuttgart
since 1998.
Rebekka Kuiter is an objects conservator at the
Kalkriese Museum. She has a BA and MA from the
State Academy of Art and Design Stuttgart.
METAL CONSERVATION, CELLULOSE NITRATE AND THE ODDY TEST 131

Capucine Korenberg has worked at The British

Museum as a conservation scientist since 2003. She has
a PhD from Imperial College London.

Julia Ziegler is an objects conservator at the Museum

of Ethnology Dresden. She has a diploma from the State
Academy of Art and Design Stuttgart.

Sebastian Bette has a PhD in chemistry from TU

Bergakademie Freiberg. He has been a researcher on the
structure of corrosion products on museum objects at the
Max Planck Institute for Solid State Research Stuttgart
(DFG funded) since 2017.

Jörg Stelzner is an objects conservator. He has a PhD

from the State Academy of Art and Design Stuttgart and
has researched glass-induced metal corrosion (funded
by DBU) since 2016.
Propellants and Primers: A Striking History with a
Possibly Explosive End
Flora Davidson
Parks Canada
Indigenous Affairs and Cultural Heritage Directorate
Collections, Curatorial and Conservation Branch
Ottawa ON, Canada
flora.davidson@canada.ca
Abstract
Objects such as unfired percussion caps and cartridges from
military and naval sites are often found in museum collections.
However, existing guidelines and protocols may not be
addressing the human risks posed in handling archaeological
objects that may have undergone considerable chemical change.
Black powder has long been used as a propellant in firearms. When
ignited, the pressure of the gas produced provides propulsion to
fire a projectile. By the 19th century, primers, which are highly
sensitive to shock, were introduced and used to set off an explosive
train resulting in more rapid firing capability. While this changed
the course of warfare, it now presents a particular challenge when
these munitions are found in an archaeological context.
Guidelines for handling firearms and safety protocols for explosive
objects found on archaeological sites are common, but information
regarding the specific risks posed by deteriorated examples are
scarce. This gap in information may lead to human health risks not
being properly assessed and objects being destroyed without due
consideration or mitigation of the risks involved.
Introduction
The identification and characterization of materials is an
important step in any conservation assessment. In the
case of explosive compounds, characterization of the state
of the compound and the type of reactions it has under-
gone is essential before any intervention or handling.
A process for evaluating objects in military ammunition
collections that uses artifact history, chemistry of the
original explosive agent and determination of possible
interactions, and formation of deterioration products
which could pose human risks has been well laid out
by Frame and Odegaard (2011). Their work filled the
gap left in sources of information about weaponry care,
which tend not to consider unfired ordnance. This
protocol to identify possible risks proves useful for
determining appropriate safety measures and treatment
This paper illustrates how risk-management protocols, together
with site-specific details and period knowledge of munitions,
can provide a greater understanding of the human risks involved
and a unique opportunity to gather contextual information from
an object. This may ultimately contribute to the interpretation of
the site. It should be noted, however, that this approach is not
appropriate for munitions dating towards the end of the 19th
century, as the increased stability of explosive compounds then
in use poses risks that are beyond the ability of most heritage
institutions to manage.
A concretion containing percussion caps from the recent marine
excavation of one of the two 1845 Franklin Expedition ships,
HMS Erebus, provided the opportunity to employ this deductive
assessment process on objects from an archaeological context.
Keywords
propellants, primers, archaeological munitions, unexploded
ordnance, 19th-century weaponry
and may have further implications for the study of the
original compounds’ chemistry and their susceptibility to
change due to environmental reactions which may also
provide insight as to how weapons may have performed
in historic context.
Military historians have noted that the effectiveness of
weaponry is often difficult to assess despite meticulous
records of weapons, mechanisms and usage (Hughes
1974). Often, incalculable factors, including human
conditions and circumstances, influence the outcome.
In the case of black powder, the oldest source of fire
power, stability and performance are generally well
understood. This is not so for more modern chemical
formulations which appeared and developed rapidly in
the 19th century.
PROPELLANTS AND PRIMERS: A STRIKING HISTORY WITH A POSSIBLY EXPLOSIVE END
The study of the effects of deterioration and stability of
historic explosives used as primers and propellants on
the performance of munitions in historic context could
add to the interpretation of sites. A group of percussion
caps found on HMS Erebus were examples of recent
technological advances in weaponry during the first half
of the 19th century. The Royal Navy was one of the early
adopters of the percussion musket, just six years before
the Franklin Expedition (Blackmore 1961, Winant 1970).
The caps provided an opportunity to research both the
use of this technology and deterioration of the explosive
contents after 160 years in extreme arctic conditions.
A protocol for handling, conserving and deriving infor-
mation from these finds would also be important for the
excavation. A note found in a cairn by a search party—one
of the scant clues left behind—revealed that HMS Erebus,
a naval ship, was abandoned after only two years into a
multiyear expedition (Cyriax 1939). The potential for this
wreck site to yield quantities of firearms and associated
materials is high (Figure 1).
Figure 1. Section of HMS Erebus near the bow at the wreck site (Nunavut,
Canada)
Collections and excavation safety
The first step with hazardous objects, including firearms
and munitions, is to ensure that they are handled with
due regard for human safety. Parks Canada, like many
other institutions, has hazardous materials policies and
procedures in place (Parks Canada 2011, Parks Canada
1986 and 1991). These policies provide a helpful struc-
ture to navigate the investigation and actions to follow.
In Canada, firearms are closely regulated by legislation.
The Canadian Firearms Program is administered by the
Royal Canadian Mounted Police, who are consulted
when suspected explosive munitions are found on site.
133
The Canadian Government’s Department of National
Defence has rated former military land and underwater
sites in Canada according to risk of finding unexploded
explosive ordnance (UXO). The information available
is site specific and includes historical data and type of
UXO that may be present (Government of Canada 2017).
These are valuable resources, yet the outcome of both
is that discovery of any suspected explosives mandates
disposal, often by controlled detonation. Suspected
explosives cannot be recovered unless it is possible to
prove their safety.
Proving firearms and munitions safe is one of the
objectives of contemporary firearm training under the
Canadian Firearms Program; good collections policy
should aim to achieve the same. Recognition of the
risk posed by historic objects can be complicated. This
was perhaps evident when two objects associated with
weaponry—a 5’10” cannon and a concretion of approx-
imately 6.6 cm × 4.5 cm × 2.5 cm comprised of percus-
sion caps—were retrieved from HMS Erebus (Figures 2
and 3). The risks associated with a cannon weighing
700 lb (300 kg), which could be loaded with up to two
pounds (0.9 kg) of gunpowder and a six-pound (2.7 kg)
projectile (McConnell 1988), appear more obvious than
Figure 2. Cannon from HMS Erebus
Figure 3.
Concretion
consisting of
percussion caps
 from HMS Erebus
134  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
with a group of percussion caps, each of which weighing
less than a small coin. However, although the risks may
be different, any explosive-containing object should not
be assumed benign. In hindsight, although the possible
presence of explosive contents was recognized for both,
the investigation of the risks associated with the percus-
sion caps was unfortunately considered with much less
sense of immediacy.
Since there is information available in the conservation
literature on precautions and safe removal of black
powder, the investigation and prospect of cleaning the
cannon was relatively uncomplicated compared to that
of the percussion caps (Figure 4).
Figure 4. During safety and cleaning of the cannon’s bore
Historical use and development of fire power
Early examples of black powder used as a propellant
date from the 14th century (Cocroft 2000, Crocker
2002). Empirical observations would later lead to some
improvements in weaponry and refinement of black
powder, but it was not until the 18th century that firing
trials would provide data on performance. This would
eventually hasten the further development of firepower
to improve reliability and rate of fire (Hughes 1974). By
the 18th and early 19th centuries, the industrialization of
manufacturing processes produced a product of higher
stability on an industrial scale. This revolutionized
warfare (Hughes 1974, Cocroft 2000).
Black powder has a long history as perhaps the widest
known and documented explosive in historical and
archaeological records. The 19th century brought rapid
development of new propellants, primers and more
efficient firearms. While black powder (classified as a
low explosive) relied on ignition, later primers (primary
high explosives) could be detonated by shock, friction
PREVENTIVE CONSERVATION
and heat. Minute quantities of these new compounds
could ignite a powerful charge.
Like black powder, earlier primers such as mercury
fulminate are subject to deterioration and change in
sensitivity depending on circumstances. Their stability
and reaction to environmental influences such as mois-
ture and heat can be useful in both assessing risk in
managing these objects and providing insight into the
performance of certain munitions according to the type
of compound used.
As technology advanced, synthetic explosives became
more stable and less prone to desensitization. Nitrocel-
lulose, which can be detonated or ignited by combustion,
began to be used as guncotton in ammunitions in 1865,
years after its discovery, following the development of
a procedure to remove acidic residues from the manu-
facturing process which could otherwise lead to spon-
taneous combustion (Department of the Army 1984,
Cocroft 2000). Damp guncotton ignites with a primer
such as mercury fulminate, and in compressed form can
still be detonated. Even wet guncotton can be detonated
if a small amount of it is dried or if a large amount of
mercury fulminate is used, yet it will only smolder if
ignited (Admiralty 1880). Unlike black powder, which
had to be stored dry, guncotton is safest when kept wet
and away from detonators. It is reactive in various states,
but can be manageable if given specialized housing.
During the 19th century, the stability of newly developed
compounds improved, rendering these less reactive to
environmental or situational circumstances. Metal azides,
developed as detonators, are extremely stable to heat and
show no decrease in sensitivity if stored in water. They
will, however, react with copper to produce a metallic salt
which is even more sensitive (Department of the Army
1984). Consequently, the liability of housing objects with
such detonators requires a high degree of expertise which
is likely beyond the ability of museum professionals.
Percussion caps: Significance to the interpreta-
tion of the site
It was hoped that the first group of excavated artifacts
would provide a glimpse of life aboard HMS Erebus.
The fate of the ships under the command of Sir John
Franklin, HMS Erebus and HMS Terror, had remained
a mystery that was only partially solved with the
discovery of the wreck. During the 160 years between
PROPELLANTS AND PRIMERS: A STRIKING HISTORY WITH A POSSIBLY EXPLOSIVE END
disappearance and rediscovery, few objects surfaced.
There were no survivors and few clues left from the
expedition to inform historians, archaeologists and a
host of interested Franklin enthusiasts. It is highly hoped
that findings from the wreck will provide details of how
such an expedition could fail so spectacularly, despite
being well provisioned and equipped with the latest of
Victorian technologies.
The majority of the artifacts removed from the site to date
pertain directly to the ship’s fittings. Few personal effects
of those aboard have been retrieved. The two objects,
the cannon and the concretion of percussion caps,
spoke directly of the ship’s provisions for survival. That
naval ships were equipped with weaponry for defensive
purposes is no surprise. However, on a mid-19th-cen-
tury expedition in the High Arctic, weapons could also
provide a means of signaling for vital supplies and rescue.
Percussion-fired muskets were also used for hunting,
providing an active diversion during long winters and
supplementing the diet when rations were limited. Fresh
meat was necessary to ward off the effects of scurvy. In
such ways, both artifacts would have played a significant
part in the lives of the crew.
The cannon and the percussion caps are representative
of the Victorian era. Older technology was still in use,
but newer technologies would begin to shape and define
the period. Cannons had remained little changed from
previous centuries, yet percussion-fired muskets were
a very recent innovation to replace, and improve upon,
the known failings of previous muskets. How well the
technologies, both developed for military tactical use and
defence, would perform in the extreme climate of arctic
winters would be crucial to the comfort and possibly the
survival of the expedition.
Assessment of the percussion caps
Assessment of the percussion caps followed the order
of materials identification through historical research
and a risk assessment based on the materials and their
interaction with the burial environment.
The corroded green percussion caps appeared likely to
be copper alloy. Brass caps had replaced earlier copper
versions by the time the Royal Navy adopted percussion
muskets in 1839 (Bailey 1972), shortly before Franklin’s
last Arctic expedition. Copper had proven too brittle and
could break upon detonation, risking injury to the user.
135
Percussion muskets replaced flintlock muskets, which
were in standard use for land and sea service until the
19th century but tended to misfire in wet weather due to
the hygroscopicity of the loose priming powder (black
powder) (Blackmore 1965, Hughes 1974, Bryce 1977).
Development of a pressure-sensitive priming compound
able to ignite a propellant without a spark was considered
more reliable. This would increase the rate of fire—a
tactical advantage. This compound, a mercury fulminate
mix, was applied inside each cap (Blackmore 1961). The
striking agent, as it was called, was not visually apparent
on the caps from HMS Erebus, nor would it have been
in its original state.
The caps in the concretion appeared unfired, since
typical deformation observed after firing was not found
(Figure 5). Back pressure, created when the main charge
in the barrel of the gun is ignited, travels up the hollow
nipple on which the caps are set. This leads to physical
deformity or splaying of the caps after firing (Martin,
personal communication, 2016).
Figure 5. Percussion caps: (left) unfired and (right) fired
A radiograph of the concretion confirmed that it appeared
to be comprised solely of approximately 40 unfired
percussion caps held together by corrosion products.
Only a thin metal core remained between mineralized
exterior surfaces (Figure 6). This suggested that the
explosive was likely present during deposition and the
chemistry of the mercury fulminate would need to be
closely considered.
Figure 6. Radiograph of percussion caps concretion
136  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Normal manufacturing processing of mercury fulminate
yields a gray crystalline product of 98%–99% purity. Only
a few grains are required since it is highly sensitive to
percussion (Department of the Army 1984). It can also
be detonated by electric spark, heat and flame. Predicting
the sensitivity and deterioration state of the compound
depends on environmental factors.
The concretion was kept wet after excavation from marine
burial. Mercury fulminate is desensitized but not deacti-
vated in water; it has low solubility in water. Deterioration
leading to reduction to 95% purity seriously impairs its
function. At 92% purity, the initiating efficiency would
be practically destroyed, although it will still explode
when ignited (Department of the Army 1984). There is a
correlation between temperature and reduction in purity,
with higher temperatures leading to decreased purity in
shorter intervals (Table 1). While there is no data on the
temperatures sustained during the long burial, it is likely
to have been generally slightly below zero Celsius up to
single-digit positive numbers.
Table 1. Deterioration of mercury fulminate. United States Depart-
ment of the Army Technical Manual
Time required to reduce purity to:
Storage
temperature 95% 92 %
(°C) Days Months Years Days Months Years
80 0.5 1 concretion according to information given in the Safety
50 8 11
30–35 1.7 5.8
20 7 9
10 8 10
The products of deterioration of mercury fulminate are
non-explosive solids (Department of the Army 1984,
Frame and Odegaard 2011). Though the rate of deteri-
oration in these cold temperatures was not found, the
association with the brass cases would likely have caused
deterioration and reduction in purity since mercury
fulminate reacts rapidly with brass when wet (Depart-
ment of the Army 1984, Frame and Odegaard 2011).
The caps from HMS Erebus would be desensitized while
in water but very likely have reacted with the brass of the
cap leading to reduction in purity. Risk of explosion was
assessed to be low. Archaeological ammunition containing
mercury fulminate has been similarly assessed elsewhere
(Frame and Odegaard 2011). Further, the difficulty in
maintaining high purity has led to its disuse by the United
States Military (Department of the Army 1984).
PREVENTIVE CONSERVATION
Material analysis was undertaken to detect mercury to
confirm if the original compound or residues from it
could be found. Archaeologists allowed one cap to be
removed from the edge of the concretion for analysis.
Scanning electron microscopy with energy dispersive
x-ray analysis was not able to detect or confirm the
presence of elements associated with the primer from the
sample (Kavousanaki 2016). In addition to the manufac-
turing process, the corrosion and reaction of the metal
with mercury from the primer had embrittled the caps,
rendering them fragile and a challenge to sample with
precision. This, combined with the limitations of the
technique used as it analyses only a few cubic microm-
eters, made it difficult to draw conclusions from this
analysis. Follow-up with other analytical techniques was
requested; however, due to the circumstances at the time,
this was not addressed.
Summary of assessment
The assessment determined that the concretion could
be carefully conserved. Since the risks in handling and
storing the object were low, a deactivation step was not
warranted. However, because residues of mercury or
mercuric compounds may be present, a precautionary
recommendation was made to store and handle this
Data Sheets in accordance with the Globally Harmonized
System of Classification and Labelling of Chemicals
(GHS)1 for mercury fulminate, mercury and mercury
chloride. It was not felt that this would prove difficult
to follow or prohibitive to handling and storage and,
in addition, it could create awareness for the types of
hazards that may be associated with these objects. The
recommendations were as follows:
• Possibility of mercury residue. Handle with well-fitting
nitrile gloves (polyvinyl chloride or rubber are also
acceptable choices).
• Treatment to deactivate mercury fulminate, if present,
should be considered and, if required, in accordance
with collections policies and management directives.
Alternatively, the concretion of caps could be stored
and handled as per Parks Canada’s policies and guide-
lines for live ammunition.
• Although risk of explosion is low for archaeolog-
ical ammunition containing mercury fulminate,
protocols for handling mercury fulminate should
be considered.
PROPELLANTS AND PRIMERS: A STRIKING HISTORY WITH A POSSIBLY EXPLOSIVE END
Recommendations for avoidance of fire, explosion and
deterioration are as follows:
• Protect from friction, impact and static electricity.
• Keep away from heat, sparks and open flame.
• Keep away from strong oxidizing sources.
The implication that the effect of moisture on the percus-
sion caps leads to desensitization of the primer and, in
addition, can also lead to loss of purity due to reaction
with the cap metal may be interesting to archaeologists.
Accounts contemporary with Franklin’s time such as
those of Miertsching (1967), who spent several winters
aboard HMS Investigator while in search of the Franklin
ships, spoke of the persistent and extreme damp winter
conditions aboard ships while these were frozen in ice
in the Arctic Archipelago. In such conditions, would the
reliability of the percussion caps be reduced, thus making
them as likely to fail as the flintlocks they replaced?
Treatment of the percussion caps
While a request for the object for an exhibition precluded
further and more extensive analysis, it was felt that the
research undertaken was strong enough to support going
ahead with treatment. It was determined that the risks
associated with this object could be managed and a suit-
able treatment could be found without jeopardizing the
integrity of the object. The treatment goal was to consol-
idate and stabilize the further corrosion of the metal.
Treatments aimed at stabilizing marine copper alloys
which focus on the efficient removal of soluble cuprous
chloride compounds risk dissolving copper from sound
metallic areas and alteration to the patina. Instead, an
inhibitor that would form a barrier between the cuprous
chloride at the metal surface and the atmosphere in order
to prevent further corrosion, benzotriazole (BTA), was
chosen. Since mercury fulminate is only very slightly more
soluble in ethanol than water, neither of these would be
likely to affect the primer. Mercury is also unlikely to react
with BTA given the higher affinity of copper to react and
form copper-BTA compound. Therefore, the object was
immersed in aqueous BTA solution after a thorough rinse
in deionized water. This was followed by a succession of
ethanol rinses and, lastly, by immersion in Paraloid B-72
in ethanol to consolidate and coat. It was then air-dried.
After treatment, the concretion appeared well consoli-
dated and stable. To ensure preservation, safety and mini-
mize the possibility of the formation of free mercury, envi-
137
ronmental recommendations for storage and handling
were placed in the artifact dossier. The artifact’s packaging
was labelled to advise of and to identify the potentially
hazardous contents and appropriate handling protocols. It
was felt that recommendations for environmental control
during a two-year travelling exhibit following treatment
should balance good preservation with conditions easily
feasible for the duration of the exhibit. Conditions stipu-
lated included low light, maximum temperature of 21°C
and a maximum of 40% relative humidity.
Conclusion
Through a deductive process, it was possible to identify
possible human risks associated with this object and work
within collections and firearm policies to responsibly care
for a potentially hazardous object from an archaeological
19th-century naval wreck site.
Historical and more contemporary sources proved useful
to gain important insight into the state of the original
explosive compound. For a site prior to the end of the
19th century, this framework was useful in choosing
appropriate treatment, handling and housing of the
artifact within the collection.
In the case of the Franklin ships, reviewing the literature
of munitions dating to the Victorian period was useful in
forming a risk profile for excavated munitions that may
be encountered on the ships. Lastly, during the process,
a potential avenue for further archaeological research
was highlighted.
It was noted that for newer explosive compounds with
exceptional stability the associated risks appear beyond the
ability of anyone but an explosives technician to manage.
Acknowledgements
The author would like to thank Parks Canada and former
Parks Canada Cultural Resources Material Specialist,
Charles Bradley, for their support with this project.
Notes
1 See www.ccohs.ca/oshanswers/chemicals/whmis_ghs/
hazard_classes.html.
References
Admiralty (Great Britain). 1880. Manual of gunnery for
Her Majesty’s fleet. 1880. London: printed by George
138  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PREVENTIVE CONSERVATION

E. Eyre and William Spottiswoode for H.M. Stationery
Office.
Bailey, D.W. 1972. British military longarms 1815–1865.
London: Arms and Armour Press.
Blackmore, H.L. 1961. British military firearms 1650–
1850. London: Herbert Jenkins Ltd.
Blackmore, H.L. 1965. Guns and rifles of the world.
London: B.T. Batsford Ltd.
Parks Canada. 2011. Firearms policy, approval date 21
February 2011.
Parks Canada. 1986. Management directive 2.1.23:
Collection management system: Archaeological services
(under revision).
Parks Canada. 1991. Management directive 2.3.3: Archae-
ological explosives: Archaeological discovery (under
revision).

Bryce, D. 1977. A guide to the cataloguing of historic firing Winant, L. 1970. Early percussion firearms. London:
mechanism. Ottawa, ON: Parks Canada. Spring Books.

Cocroft, W. 2000. Dangerous energy: The archaeology of

Author
gunpowder and military explosives manufacture. Swindon:
English Heritage. Flora Davidson worked at several institutions across
Canada before joining Parks Canada in 2004. She special-
Crocker, G. 2002. The gunpowder industry. Aylesbury,
izes in the conservation and protection of submerged
Bucks: Shire Publications Ltd.
cultural heritage and is currently the lead conservator
Cyriax, R. 1939. Sir John Franklin’s last Arctic expedi- for the Franklin ships HMS Erebus and HMS Terror.
tion: A chapter in the history of the Royal Navy. London: Prior to working at Parks Canada, she earned under-
Methuen & Co, Ltd. graduate degrees in cell and developmental biology and
Department of the Army (USA). 1984. Military explosives: fine arts from the University of British Columbia before
TM 9-1300-214. Technical manual. Washington, D.C.: completing a master’s in conservation of historical and
Headquarters, Department of the Army. archaeological objects at Durham University (UK).
Frame, L.D. and N. Odegaard. 2011. Patinas, powders
and primers: Safety with a museum collection of small
arms ammunition. In Metal 2010: Proceedings of the
Interim Meeting of the ICOM-CC Metals Working Group,
Charleston, South Carolina, USA, 11–15 October 2010,
eds. P. Mardikian, C. Chemello, C. Watters, and P. Hull,
437–44. Clemson, SC: Clemson University.
Government of Canada. 2017. How we protect Cana-
dians from UXO, modified 06 November 2017. www.
canada.ca/en/department-national-defence/services/
uxo/uxo-program.html.
Hughes, B.P. 1974. Firepower: Weapons effectiveness on the
battlefield, 1630–1750. London: Arms and Armour Press.
Kavousanaki, D. 2016. Analysis of HMS Erebus artefacts.
Parks Canada, Ottawa, ON, Canada.
Martin, R., Historic Weapons Advisory Committee, Parks
Canada, personal communication, 2016.
McConnell, D. 1988. British smooth-bore artillery: A
technological study. Ottawa: Canadian Government
Publishing Centre.
Miertsching, J.A. trans. 1967. Frozen ships: The Arctic
diary of Johann Miertsching 1850–1854, trans. L.H.
Neatby. New York: St. Martin’s Press.
El Anatsui: Conservation Concerns Beyond Installation

Joy Bloser* Lynda Zycherman Ana Martins

The Museum of Modern Art
New York NY, USA
joy.bloser@gmail.com
*Author for correspondence
The Museum of Modern Art
New York NY, USA
lynda_zycherman@moma.org
Abed Haddad
The Museum of Modern Art
New York NY, USA
abedhaddad123@gmail.com
The Museum of Modern Art
New York NY, USA
ana_martins@moma.org
Chris McGlinchey
The Museum of Modern Art
New York NY, USA
chris_mcglinchey@moma.org

Abstract
Bleeding Takari II (2007) is a large-scale sculptural hanging by El
Anatsui held in the collection of the Museum of Modern Art, New
York. The artist manages a robust studio practice to construct
his works, which are assembled from thousands of aluminum
bottle tops recovered from used liquor bottles and connected
with copper wire. Bleeding Takari II is predominantly red and
silver and can be variably draped with each installation. Existing
research on Anatsui’s oeuvre has been largely concerned with
installation methods, while long-term preservation and display
parameters have not yet been addressed. Industry research and
materials analyses were carried out to better understand the
object’s composition and to help inform treatment, installation,
and preservation recommendations. Infrared (FTIR) and Raman
spectroscopy identified various coatings and pigments, x-ray
fluorescence (XRF) confirmed the metal as an aluminum alloy,
and microfadeometry (MFT) revealed potential light sensitivity
of the red bottle tops.
Keywords
El Anatsui, aluminum, coatings, display recommendations

Introduction
Bleeding Takari II (2007) is a large-scale sculptural
hanging by El Anatsui (b. 1944, Ghana) held in the
collection of the Museum of Modern Art, New York
(Figure 1).1 Thousands of cut, folded, and perforated
aluminum bottle tops recovered from used liquor bottles
are connected together with copper wire. The large-scale
work (376 × 557 × 78 cm) can be hung, draped, and folded
in various arrangements with each new installation. In
preparation for loan, long-term preservation concerns
were investigated to establish display and treatment
parameters that align with the artistic intent and physical
material. Research included a literature review of the
artist’s process, industry research and technical analysis
of the bottle tops, assessment of light stability, methods
for mitigating stress during installation, and improved
protocols for crating and installation.

Artist’s process and materials

Figure 1. El Anatsui, Bleeding Takari II, aluminum bottle caps and copper
wire, 376 × 557 × 78 cm, 2007. Gift of Donald L. Bryant, Jr. and Jerry
Speyer, 201.2008. © The Museum of Modern Art
Artist’s process
Living and working in the vibrant town of Nsukka,
Nigeria, El Anatsui manages an active studio, producing
large, variable hanging sculptures. The colorful composi-
tions are assembled from bottle tops sourced from local
Nigerian liquor bottles. His material choice is culturally
loaded, carrying the histories of colonization, the rum
and slave trade, and alcohol-fueled economic develop-
ment of West Africa. Nigeria has a robust liquor industry,
140  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
and we identified more than 65 different brands of local
Nigerian liquor in Bleeding Takari II, a large proportion
of which originate from distilleries or distributors in
Anatsui’s home Enugu State (e.g., Obisco Industries,
I.G. Chemical & Allied Products, Sussycent Distillers,
and Chigin Industries) and neighboring Anambra State
(e.g., Bosotex).
Anatsui purchases the used bottle tops from distillers and
aluminum scrap suppliers. In the studio, assistants sort
them by color, separate the “caps” from the “skirts,” and
assemble composition blocks from specified parameters
of color and shape. They pierce holes in multiple loca-
tions with an awl, and finally join pieces together with
pre-cut lengths of uncoated copper wire threaded through
adjacent holes and twisted closed. Anatsui and his studio
manager, Onyishi Uchechukwu, have developed a set
terminology for the composition blocks that is further
designated by dominant color (Figure 2). The many
hands that make the artwork are visible in the different
twists of copper – loose or tight, twisted to the left or the
right, tails folded back or clipped short. Anatsui inspects
completed composition blocks for dimensional flexibility
and to ensure they can be folded in any direction (Fold
crumple crush 2011).
 closures. ROPP closures have been the industry standard
 Anatsui’s work. Lord’s Dry Gin, Regal Dry Gin, Seaman’s
Figure 2. Composition block
terminology: (A) plain; (B) plain
round; (C) folded; (D) crushed;
 (E) crumpled
The composition blocks are arranged in various permuta-
tions on the studio floor. Studio assistants join the compo-
nents together with additional wire, but if the arrange-
ments shift, the copper wires are cut and rejoined anew.
The traces of old arrangements are left in the dangling
PREVENTIVE CONSERVATION
cut wires on the piece. Once connected together, Anatsui
experiments with draping and folding the hangings on a
large plywood framework outside his studio. The work
is then photographed, named, and folded into compact
dimensions for storage or transport (Fold crumple crush
2011, Hurt 2011).
Anatsui values the haptic involvement of the many
people who create and install each piece. He encour-
ages those installing his work to change folds, drapery
patterns, or even orientation with every new installation:
“I don’t believe in artworks being fixed” (Fold crumple
crush 2011). Aware that encouraged handling will affect
the stability of the work, Anatsui provides replacement
copper wire and aluminum pieces with each piece.
Bleeding Takari II is composed entirely of plain rectan-
gular and plain round composition blocks in silver, red,
orange, and black. Additionally, 13 red plain round
composition blocks are separately arranged on the
ground below the piece. The artwork hangs on five metal
brackets; folds and pleats are secured with monofilament
and L-pins. The artwork has been installed eight times
and exhibited three times. Previous treatment has only
included broken copper wire replacement. Display spec-
ifications were not stipulated prior to 2018.
Liquor bottle tops
The primary bottle tops used in Bleeding Takari II are
from single-use, aluminum roll-on pilfer-proof (ROPP)
for liquor packaging since the 1960s and are gaining
popularity in the wine industry. Nigeria has a robust
distillery market both locally and as a foreign investment
market for bottling and distribution. Market trends
are visible in the prominence of certain bottle caps in
Aromatic Schnapps, and Dark Sailor Blended Dark Rum
are produced by Nigeria Distilleries Limited, based in
Lagos, which is a market leader in liquor volume sales;
Eastern Distilleries & Food Industries and Intercon-
tinental Distillers are other major distributors in the
country (Euromonitor International 2018). There are
three large-scale ROPP manufacturing facilities located
in Nigeria who supply caps and closures to the West
African market; their closures are also used throughout
Anatsui’s works (e.g., Chelsea, Intercontinental Dry
Gin, Sabrina).
EL ANATSUI: CONSERVATION CONCERNS BEYOND INSTALLATION
ROPP manufacturing is largely standardized through
specifications of the capping machinery and global
market, with only slight variation between large- and
small-scale production. Metal ROPP closures are
produced from wrought aluminum alloy sheeting,
typically AA3105 and AA8011, and other 3xxx series
aluminum alloys (Al-Mn alloys) strain-hardened at
various H tempers.2 Roll-on applications of organic
coatings are applied to the interior surface to protect
the consumer, and a base color, often referred to as the
“lacquer” layer, is drawn down the exterior. Optional
printed designs and protective coatings can then be
applied over the lacquer layer. Bottle caps from Bleeding
Takari II show examples of off-set lithography, screen
printing, and relief printing (Figure 3). Coatings are
added to improve corrosion resistance, processing, and
adhesion. In order to withstand the aggressive forming
process and be food-safe, the coatings require a high
flexibility and cohesion strength (Forrest 2007).
Figure 3. Details of printing techniques: (left) (P-JS) screen-printed;
(center) (R5) off-set lithography; (right) (R2) relief print
A cupping press draws blanks from the prepared
aluminum sheeting into shallow cups that are redrawn
in succession to achieve the finished depth of the closure
(Figure 4). The caps can be embellished with embossing
and rotary off-set lithography on the “skirt” or side
walls. At the capping facility, capsules are fitted over
filled bottles and, using the bottle profile as a mandrill,
rubber rollers conform the aluminum to the threading
of the bottle under pressure (Langley 2003). Once the
consumer opens them, the cap separates from the skirt of
the closure, which remains on the bottle neck, indicating
the bottle has been opened or “pilfered with.”
ROPP closures are produced in set dimensions corre-
sponding to their diameter and height. Measurements
141
of cap diameters and skirt heights across Bleeding Takari
II largely correspond to standard capsule sizes of 25/29
(25.80 mm in diameter/29 mm in height) and 28s
(28 mm in diameter/either 18.10 mm or 28.90 mm in
height). Anatsui constructed Bleeding Takari II with plain
composition blocks made from the skirt of the closures
with the short ends folded to create elongated hexagonal
shapes. The plain round composition blocks were made
from the top caps of the closures cut away from the skirts.
Materials analysis
A representative set of bottle tops from Bleeding Takari
II and a replacement bottle top (ca. 2009) provided
by Jack Shainman Gallery were analyzed. Red is the
dominant color in Bleeding Takari II, thus red was prior-
itized for analysis. Analysis goals were to characterize
the aluminum alloy, printing layers and coatings, and
quantify light sensitivity, in order to inform treatment
and display specifications.
Methods
Eleven samples were removed from the artwork (P1–P5,
R1–R6) and one sample was donated by the gallery (P-JS)
for testing (Table 1). The eleven samples were reattached
to the work after analysis was complete. Samples were
observed under magnification up to 400× using a
Wild Heerbrugg TYP (376788) microscope and images
captured using a Canon EOS T3i mounted with an Optem
1.6× T-Mount Adapter using EOS Utility Canon software
and processed with Adobe Photoshop.
X-ray fluorescence (XRF) was performed using a hand-
held Bruker Tracer III-SD spectrometer equipped with a
rhodium x-ray tube at 40 KeV, 9.7 µamps, 75 s; spectra
were analyzed with Bruker P1 XRF and ARTAX softwares.
Scanning electron microscopy and electron dispersive
spectroscopy (SEM-EDS) was performed using a Hitachi
TM3000 scanning electron microscope with a Bruker
Xflash MIN SVE using Quantax 70 software, courtesy of
New York University.
Figure 4. Diagram of ROPP closure
 redraw process
142  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Table 1. Sample descriptions
Image
Sample Description
exterior/interior
P-JS Red plain; Castello® brandy; red and
gold off-set litho printing on silver
aluminum
P1 Red plain; red lacquer on silver
aluminum
P2 Red plain; red lacquer on silver
aluminum
P3 Red plain; red lacquer on silver
aluminum
P4 Red plain; red lacquer on silver
aluminum
P5 Gold plain*; Squadron dark rum;
red and gold off-set litho printing
on gold-colored aluminum, letters
embossed and filed
R1 Red plain round; red lacquer on silver
aluminum
R2 Red plain round; UC Natural local gin;
gold-colored stamp on red lacquer
on silver aluminum
R3 Red plain round; red lacquer on silver
aluminum
R4 Red plain round; red lacquer on silver
aluminum
R5 Yellow plain round; Edon’s Beverages
Nig. Ltd.; Screen-printed blue on
yellow lacquer on silver aluminum
R6 Yellow plain round; yellow lacquer
on silver aluminum, logo embossed
and filed
*Aluminum interior surface faces out when attached to the
artwork, making it a gold plain, rather than a red plain.
Fourier transform infrared spectroscopy (FTIR) was
carried out using a Nicolet iS50-FTIR coupled with a
Smart Orbit Real Type IIa Single-reflection Diamond
ATR crystal equipped with a DTGS KBr detector. Spectra
(64 scans) were acquired at a 4 cm−1 resolution in the
PREVENTIVE CONSERVATION
4000–600 cm−1 range. Spectra were analyzed in absorb-
ance using Omnic 9 software by Thermo Scientific.
Normal Raman spectra were collected using a Renishaw
In-via Raman system equipped with a 785-diode laser,
1800 lines/grating, and a Leica confocal microscope with
a 100× objective. In-situ spectra illustrate 30-second
acquisitions and pigments identified using OMNIC with
established spectral libraries.
Microfadeometry (MFT) illumination was achieved with
an Oriel Instruments LIK-LMP Light Intensity Controller
Kit equipped with a Si Detector Head and Xe-arc lamp
controlled by a Spectra-Physics power supply coupled
to a thermo-Oriel Digital Flux controller. Spectra were
registered using a Control Development spectrometer in
the range of 380–900 nm. Blue wool standards 1, 2, and 3
were used for comparison. Integration time of 0.36 s was
used for a course of 44 data points. Delta E was calculated
using the CIE94 equation.
Aluminum
Both interior and exterior surfaces of the aluminum
samples were analyzed using XRF, and SEM-EDS was
completed on select samples. Results were compared
with industry standards.
XRF analysis identified aluminum (Al), iron (Fe),
gallium (Ga), and manganese (Mn) on the interior of all
but one sample, P2. Results suggest all of the samples,
except P2, are made from sheets of wrought aluminum
alloyed with Mn. Al-Mn alloys are favored in the pack-
aging industry for their improved tensile strength over
wrought aluminum, good formability, ability to withstand
deep-drawing, and improved resistance to intergranular
and stress corrosion (Nam and Lee 2000). Quantitative
analysis is beyond the scope of research, but would be
helpful in characterizing alloy compositions. XRF results
are summarized in Table 2.
Aluminum alloy surfaces were assessed with SEM-EDS;
surface defects are visible near the punctured holes where
the lacquer layer has delaminated (Figure 5). No filiform
corrosion was observed.
Printing and coatings
Printing media and coatings were analyzed with both
FTIR and Raman spectroscopy. Results were compared
with XRF findings.
EL ANATSUI: CONSERVATION CONCERNS BEYOND INSTALLATION 143

Table 2. Summary of XRF results. Al, Fe, and Ga peaks present in all promote adhesion. Samples P3 and P4 of interior coatings
samples were identified as a phenoxy resin (with possible source
of fatty acids). FTIR spectra are shown in Figure 6.
Sample Mn Cl Cr Pb Mo Ti Ba Ca Cu Zn
P-JS interior X X
P-JS exterior red X X X X X
P1 interior X X X X
P1 exterior red X X X X X
P2 interior ? X X X X
P2 exterior red X X X X X
P3 interior X
P3 exterior red X X
P4 interior X X X X
P4 exterior red X X X X
P5 interior X X X
P5 exterior red X X X X X
P5 exterior gold X X X
R1 interior X X X X
R1 exterior red X X X X X X Figure 6. FTIR spectra of interior coatings (P4, R4)
R2 interior X X X X
R2 exterior red/
X X X X X X X X
gold
R3 interior X X X X
R3 exterior red X X X X X X
R4 interior X X X X
R4 exterior red X X X X X
R5 interior X X X
R5 exterior
X X X X X
yellow/blue
R6 interior X X X
R6 exterior
X X X X X
yellow

Figure 7. Raman spectral identification of pigments

Raman analysis was performed on all samples (Figure 7).

Figure 5.
SEM image
of aluminum
surface defect
 (R2)
XRF spectra of the interior surfaces showed strong
chlorine (Cl) peaks on nine samples. Further analysis
with FTIR identified the interior coatings as a copol-
ymer of poly(vinyl chloride-vinyl acetate) (PVC/PVAc)
with a small percentage of maleic acid; a polymer blend
common in closure coatings with maleic acid added to
See Table 3 for identified pigments with corresponding
chemical compositions. Samples P1, P2, P5, and R1–R4
are printed with a mixture of Molybdate Orange and
Permanent Red BB (Chua et al. 2016); P3 and P4 are
printed with Napthol Red AS; and R5 and R6 are printed
with a mixture of Chrome Yellow Deep and unknown
pigments unique to each sample. In addition, the blue
imprint on R5 was identified as a beta modification of
Phthalocyanine Blue that prevents flocculation of printing
ink (Lambourne and Strivens 1999, Charvat 2005).
XRF serves as a complementary technique to both Raman
and FTIR spectroscopies. In the former, the detection of
Cr, Pb, and Mo is associated with lead chromate (Chrome
Yellow Deep) and molybdate (Molybdate Orange) in the
144  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PREVENTIVE CONSERVATION

Table 3. Samples and their identified pigments with chemical compo- (Liao et al. 1998). Filiform corrosion (FFC), another
sitions form of localized corrosion, propagates in a thread-like
Pigment formation between a coating and a treated metal surface.
Sample
(chemical composition) Pitting and FFC are concerns because they can form
P1 from defects in the coating when in humid (> 70% RH),
P2 chloride-rich environments (Leth-Olsen and Nisancioglu
P5 Molybdate Orange (PR104) 1997). Galvanic corrosion potential also exists between
7PbCrO4·2PbSO4·PbMoO4
the aluminum alloy and copper wire. In all cases, main-
R1
Permanent Red BB (PR48:2) taining a stable climate below 60% RH, avoiding aqueous
R2 C18H13ClN2O6S·Ca applications, and minimizing exposure to chlorides
R3 reduces corrosion potential (Selwyn 2004).
R4
Although chlorine is found in the coatings, it is strongly
P3 Naphthol Red AS (PR170) bound in the conjugated structure of the PVC/PVAc
P4 C26H22N4O4
copolymer. Ultraviolet (UV) and thermal degradation
R5 Chrome Yellow Deep (PO21) still cause dehydrochlorination of PVC and deacetylation
R6 PbCrO4.PbO of PVAc at higher activation energies (Kaczmarek et al.
Phthalocyanine Blue (PB15:4) 2002). Both degradation products are known to induce
R5- imprint
C32H16CuN8 corrosion in aluminum and copper (Selwyn 2004).
Benzidine Orange (PO13) Keeping the object protected from UV exposure and high
P-JS
C32H24Cl2N8O2 temperatures will help the coatings remain intact, and
maintaining a moderate relative humidity and temper-
red color. In the latter, the detection of Cl confirms the ature will reduce corrosion potential.
vinyl chloride polymer coating. Observed surface defects may point towards corrosion
beyond the expected formation of aluminum oxide or
Light sensitivity may also be a result of mechanical deformation. A larger
To quantify light sensitivity of the red bottle tops, samples sample set for analysis is necessary, but beyond the scope
were submitted to MFT. Results suggest a change in the of this paper.
surface appearance occurring at a rate generally equiv- MFT results suggest the surface of red bottle tops equiv-
alent to Blue Wool 2. To complement MFT analysis, alent to Blue Wool 2 will undergo a perceptible change
spectrophotometer readings were taken in 13 locations in color (ΔE=1) after approximately 20,000 hours of
before exhibition. Exhibition light levels and hours light exposure at 50 lux (Tétreault 2003). The identified
were recorded and spectrophotometer readings will be pigments on these samples are generally considered to be
repeated once the artwork returns from loan. Data will light stable in product literature. PVC/PVAc, however, has
be compared with MFT results. been shown to darken with UV exposure (Pereira 2015)
and MFT results of samples with phenoxy resin coatings
Discussion corroborate light sensitivity of the surface. Thus, it is
likely that light exposure of the coatings may be causing
Analysis revealed factors that may affect long-term pres-
the detected color change on the surface.
ervation efforts: corrosion stability of the aluminum alloy
near the puncture sites, deterioration products from the
Condition, treatment, and installation
PVC/PVAc coatings, and light stability.
Condition
Al-Mn wrought alloys are generally considered to have
good resistance to corrosion; however, certain conditions Before loan, Bleeding Takari II had many loose or missing
will promote pitting, filiform, and galvanic corrosion. copper wires and torn aluminum where the copper wire
Pitting, a highly localized corrosion phenomenon, ripped through. Gaps in the structure of the artwork were
depends on alloy composition, intermetallic particles, exacerbated each time the work was hung and pleated.
the presence of chlorides, and environmental conditions Localized areas of waxy green accretions were visible on
EL ANATSUI: CONSERVATION CONCERNS BEYOND INSTALLATION 145

plain aluminum pieces near connecting copper wires.
Lacquer and print layers showed loss primarily from
abrasion and mechanical deformation. Housing was inad-
equately sized and there were no formalized installation
instructions, placing the artwork at risk of further damage
with each improvised installation. Condition was mapped
in a grid (Figure 8) and compared to earlier installations.
While the severity of condition correlates with the drapes
and folds of previous installations, damage also occurred
from poor housing and handling.
the aluminum piece, and finally the copper wires were
reinserted through the layered aluminum and mesh and
re-twisted to connect the lined aluminum piece back to
the artwork (Figure 9). The mesh liner provides struc-
tural support to the torn holes and is flexible, removable,
and archival.

Figure 9. Polyester mesh liner, verso

To alleviate the localized stress of hanging 15 kg on

approximately 30 twisted copper wires (AWG 24), we
Figure 8. Condition map of Bleeding Takari II, repair of torn aluminum
(red) and copper wire (pink) mechanically joined a polypropylene (PP) mesh header to
the top verso of the piece. PP mesh is translucent, flexible,
Treatment strong, and archival. The mesh was cut into four strips
and attached with twisted copper wire through existing
The aim of treatment was to repair and reinforce the holes in the aluminum pieces (Figure 10).
structure of the artwork for hanging while retaining
maximum dimensional flexibility. Treatment included
repairing broken wires and torn aluminum, strength-
ening the upper area where it hangs from the metal bars,
designing new housing, and improving documentation.
We sought solutions that were sensitive to the artistic
process and materials, and prioritized qualities of the
artwork we felt were most important – reflectivity and
color, dimensional flexibility, installation variability, and
the mechanical approach to its construction.

We vacuumed the surface with low suction and noted

locations of broken wires. We replaced lost or broken
Figure 10. Polypropylene mesh header, verso: (left) overall; (right) detail
wires with uncoated copper wire of similar gauge, leaving
vestigial hanging wires in place to preserve any compo- Repairs were documented with annotated images of
sition rearrangements made in the studio. each grid square. Treatment required approximately
238 labor hours.
For the torn aluminum, we mechanically joined polyester
(PET) monofilament mesh liner to the back of each
Installation and storage
torn aluminum piece (a non-adhesive solution). Torn
aluminum pieces were unwired and removed from the Shared installation experiences from other institutions
artwork, mesh was cut to size and placed on the back of were particularly valuable in improving our own proto-
146  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
cols (Hurt 2011). We designed a reinforced, telescoping
hanging-assist bar that hooks onto the front of the
artwork. It is raised on two lifts enabling the work to
be manually transferred from the hanging bar to the
wall mounts. The retracted telescoping bar is later used
as the internal support for the Sonotube on which the
artwork is rolled with a Tyvek (high-density polyethylene)
interleaving. The large dimensions of the work require it
either be rolled or folded for shipping and storage; rolling
causes less deformation and provides an easier transition
between crating and installation. The object is handled
without gloves with freshly washed hands because the
copper wires snag on gloves and clothing, and it is stored
in the museum’s climate-controlled storage at 21° ± 3°C,
50 ± 5% RH. We documented our installation protocols
with illustrated instructions and produced new storage
and handling recommendations.
Recommendations and conclusion
Research into the artist’s practice, industrial manufac-
turing of his supplies, technical analysis of the mate-
rials, and assessment of light stability revealed complex
conservation concerns for treatment and long-term
preservation. Within the museum context, the domi-
nant agent of damage to Anatsui’s bottle top hangings
is an installation with its concomitant wire bending. It
is critical to document installation and change the folds
and pleats with each presentation to better distribute
stress over the artwork’s lifetime. Wash hands prior to
handling without gloves. Any installation should mini-
mize visible light levels and filter for UV. Chloride-laden
environments are discouraged. The hanging should be
kept in a climate-controlled environment.
The MoMA sculpture conservation and science depart-
ments analyzed, treated, and conserved El Anatsui’s
Bleeding Takari II. We assessed materials, methods of
manufacture, and aging properties, treated the work by
repairing broken elements and reinforcing the hanging
structure, and preserved the work by implementing
preventive conservation with improved packing, instal-
lation, documentation, and display recommendations.
We hope this contributes to the preservation of Anatsui’s
immense body of work.
Acknowledgements
Thanks to El Anatsui and Jack Shainman Gallery for
providing samples and generous assistance; to our
PREVENTIVE CONSERVATION
colleagues at MoMA for their support and contributions,
to Dan Hanford and Atelier, Janelle Borig, Lisa Bruno,
Jakki Godfrey, Hélène Gilette-Woodard, Perry Hurt,
Gisela Cartwright, Tixo Augustus, and special thanks
to Tomoyo Yoshioka.
Notes
1 The term “hanging” is used as established by Robert
Storr (see pp. 51–62 in Binder 2010). “Tapestry” and
“cloth” are often used to describe these works, but can
block heterotemporal understanding of Anatsui’s practice
(see Sanyal 2014).
2 Temper refers to the degree of hardness or strength of
aluminum alloys. H temper is a designation that identifies
how an alloy is strain-hardened and to what degree.
References
Binder, L.R., ed. 2010. El Anatsui: When I last wrote to
you about Africa. New York: Museum for African Art.
Charvat, R.A. 2005. Coloring of plastics: Fundamentals.
Hoboken: John Wiley & Sons.
Chua, L., C. Hoevel, and G.D. Smith. 2016. Characteri-
zation of Haku Maki prints from the ‘Poem’ series using
light-based techniques. Heritage Science 4(25). https://
doi.org/10.1186/s40494-016-0096-z .
Euromonitor International. 2018. Spirits in Nigeria.
https://www.euromonitor.com/spirits-in-nigeria/report
(accessed 1 October 2018).
Fold crumple crush: The art of El Anatsui. 2011. Directed
by Susan Vogel. Brooklyn, NY: Icarus Films. DVD.
Forrest, M.J. 2005. Coatings and inks for food contact
materials. Rapra Review Reports 16(6): 18–28.
Hurt, P. 2011. El Anatsui wall sculpture: Adventures
in hanging, installation, and display. www.paccin.org/
content.php?188-El-Anatsui-Wall-Sculpture-Adven-
tures-in-Handling-Installation-and-Display-Part-1
(accessed 1 October 2018).
Kaczmarek, H., R. Drag, M. Świątek, and D. Ołdak. 2002.
The influence of UV-irradiation on poly(vinyl chloride)
modified by poly(vinyl acetate). Surface Science 507:
877–82.
Lambourne, R. and T.A. Strivens, eds. 1999. Paint and
surface coatings: Theory and practice. Cambridge:
Woodhead Publishing.
EL ANATSUI: CONSERVATION CONCERNS BEYOND INSTALLATION 147

Langley, J., A. Yoxall, and J. Yates. 2003. Understanding Abed Haddad is the David Booth Fellow in Conserva-
the roll-on-pilfer-proof process. In 21st IAPRI Sympo- tion Science at the Museum of Modern Art, New York
sium, Valencia, 18–21 May 2003. Sheffield: Sheffield (2019–2021).
Hallam University.
Ana Martins is a Conservation Scientist at the Museum
Leth-Olsen, H. and K. Nisancioglu. 1997. Filiform corro- of Modern Art, New York.
sion morphologies on painted aluminum alloy 3105 coil
material. Corrosion 53(9): 705–17. Chris McGlinchey is the Sally and Michael Gordon
Senior Conservation Scientist at the Museum of Modern
Liao, C.-M., J.-M. Olive, R.P. Wei, and M.-M. Gao. 1998. Art.
In-situ monitoring of pitting corrosion in AA2024.
Corrosion 54(6): 451–58.
Nam, S.W. and D.H. Lee. 2000. The effect of Mn on the
mechanical behavior of Al alloys. Metals and Materials,
6(13). https://doi.org/10.1007/BF03026339.
Pereira, A.I. 2015. The perfect paint in modern art
conservation: A comparative study of 21st century vinyl
emulsions. Ph.D. dissertation, Faculdade de Ciências e
Tecnologia, Universidade Nova de Lisboa, Portugal.
Sanyal, S.K. 2014. Critiquing the critique: El Anatsui and
the politics of inclusion. World Art 4(1): 89–108.
Selwyn, L. 2004. Metals and corrosion: A handbook for
the conservation professional. Ottawa: Canadian Conser-
vation Institute.
Té t reault, J. 2003. Airborne pollutants in museums,
galleries and archives: Risk assessment, control strategies
and preservation management. Ottawa: Canadian Conser-
vation Institute.

Materials list
Copper wire, AWG 23/24 (0.56 mm diameter) by
Eisco Labs (www.eiscolabs.com)

Polyester (PET) mesh, Pe-Cap 7-60, plain weave by

Talas (www.talasonline.com )

Polypropylene mesh, WP0800 – 800 Micron PP

Woven Mesh, 0.016” Thickness
(www.industrialnetting.com )

Authors
Joy Bloser is the David Booth Fellow in Sculpture
Conservation at the Museum of Modern Art, New York
(2018–2020).

Lynda Zycherman is the Conservator of Sculpture at

the Museum of Modern Art, New York.
CORROSION STUDIES
Study on the Corrosion Mechanisms and Morphology
of Archaeological Bronze Objects from a Bronze Age
Graveyard in Southwestern Iran
Omid Oudbashi* Reza Naseri
Art University of Isfahan University of Zabol
Isfahan, Iran Zabol, Iran
o.oudbashi@aui.ac.ir rezanaseri@uoz.ac.ir
*Author for correspondence
Abstract
Archaeological bronze artefacts exhibit a range of interesting
corrosion features, including various corrosion products and
morphologies. This paper reports on an investigation of the
corrosion mechanisms that have occurred in recently excavated
bronze objects from southwestern Iran. These objects, excavated
from the graveyard site of Deh Dumen dating from the third
millennium BC, Early to Middle Bronze Age (E/MBA), are broken
vessels made of a binary Cu-Sn alloy shaped and manufactured
by thermo-mechanical processes. In this study, the corrosion
layers were observed by optical microscopy and analysed by X-ray
diffraction. The physical and chemical properties of samples of
the Deh Dumen burial soil were investigated to identify corrosive
factors (soluble salts, pH, water and organic content) influencing
Introduction
The corrosion product composition and the morphology
of the archaeological copper alloys may reflect the chem-
ical and physical conditions of the environment in which
they have been buried for long periods of time. Linked to
this is the metallurgy of the copper alloys, which may also
influence the corrosion events observed in this type of
objects (Oudbashi 2018). Although a significant number
of studies on the corrosion of archaeological copper alloy
objects have been undertaken in recent decades (e.g.
Robbiola, Blengino and Fiaud 1998, Oudbashi 2015,
Pronti et al. 2015, Oudbashi, Hasanpour and Davami
2016, Imbardelli, Gallucci and Chidichimo 2017), the
further study of corrosion mechanisms and their correla-
tion with burial conditions can reveal more details about
the long-term corrosion of copper alloys.
It has been mentioned in the literature that archaeological
bronzes show two main corrosion morphologies based
on the influence of the burial environment (soil), namely
type I and II, depending on the presence or absence of
the original surface (Robbiola, Blengino and Fiaud 1998,
Banafsheh Heidarpour Ali Ahmadi
Art University of Isfahan Art University of Isfahan
Isfahan, Iran Isfahan, Iran
banafsheheidarpour@gmail.com ali.ahmadi.maramat@gmail.com
corrosion of the bronzes. The results indicate that the dominant
corrosion mechanism that has been occurring is the selective
dissolution of copper leading to the formation of a noble, or type
I, patina. Malachite is identified as a main corrosion product that
has formed on the surface, and there is visual evidence of active
corrosion on some objects. The results of burial soil analysis
imply a rational relationship between soil conditions (corrosivity)
and the corrosion mechanism that has occurred in bronzes from
the Deh Dumen graveyard.
Keywords
archaeological bronzes, burial environment, type I corrosion,
malachite, soil analysis
Piccardo, Mille and Robbiola 2007). In many archaeo-
logical copper alloys buried in soil, a thin layer of cuprite
(Cu2O) may be formed on the metal that may transform
to malachite on the surface. Malachite, or basic copper
carbonate (CuCO 3.Cu(OH) 2), is the product of the
reaction between copper/copper oxides and carbonate/
bicarbonate anions dissolved in soil water, and its uniform
growth can lead to the formation of a green patina on
the surface of archaeological copper alloys (Scott 2002,
Selwyn 2006). The presence of chloride ions in the soil
may lead to the formation of nantokite (CuCl), which
transforms to atacamite or paratacamite (Cu2(OH)3Cl,
basic copper chlorides) in the presence of moisture and
oxygen. This reaction is known as ‘bronze disease’, a cyclic
reaction that will continue in the presence of moisture
until copper transforms into basic copper chlorides (Scott
1990, Scott 2002, Selwyn 2006).
Deh Dumen is a recently excavated Bronze Age graveyard
in southwestern Iran that has produced many bronze
artefacts among other burial goods, including vessels,
STUDY ON THE CORROSION MECHANISMS AND MORPHOLOGY OF ARCHAEOLOGICAL BRONZE OBJECTS FROM A BRONZE AGE GRAVEYARD IN SOUTHWESTERN IRAN
swords and dishes (Oudbashi, Naseri and Malekzadeh
2016, Sołtysiak and Naseri 2017). Previous analysis of the
Deh Dumen bronze vessels has identified that they were
manufactured largely from tin bronze, with parts of the
vessels being made of arsenical copper (Oudbashi, Naseri
and Malekzadeh 2016, Oudbashi and Naseri 2017). The
aim of this study was to investigate correlations between
burial soil conditions and corrosion mechanisms in this
collection of archaeological bronze objects using micro-
scopic, chemical and analytical methods.
Materials and methods
To identify the corrosion mechanisms and corrosion
product morphologies, seven broken vessels from previ-
ously studied objects were chosen (Figure 1) (Oudbashi,
Naseri and Malekzadeh 2016, Oudbahi and Naseri 2017),
with one sample being taken from the sheet metal body
of each vessel. One small part from each sample was cut
Figure 1. Seven broken vessels excavated from the Deh Dumen graveyard
151
and mounted in two-part epoxy resin, ground with abra-
sive papers and polished with diamond paste to prepare
metallographic cross sections (Scott 1991). The corrosion
products of the second part of each sample were removed
and powdered in preparation for chemical analysis.
The prepared cross sections were studied using an MR-11
metallographic microscope with bright- and dark-field
illumination. To identify the corrosion products, the
powdered samples were analysed by X-ray diffraction
(XRD) using a Bruker D8 ADVANCE diffractometer.
Three soil samples from the site were selected for anal-
ysis of the burial environment. They include the soil of
a grave near its floor (grave no. G-15, Oudbashi, Naseri
and Malekzadeh 2016) as well as soils found inside two
studied bronze objects (objects D.P.283 and D.P.287).
To remove free water from the soil, all three samples
were dried in the open air (Theocharopoulos, Mitsios
and Arvanitoyannis 2004). The dried samples were then
sieved with mesh no. 10 (2 mm) (Pansu and Gautheyrou
2006) and used for chemical tests, including character-
isation of soluble ions (soluble salts), pH measurement,
electrical conductivity test (EC) and measurement of
soil water and organic matter. Water and soil organic
matter (SOM) content of the soil samples was calculated
by heating the sieved samples in an oven at 105°C (12 h),
then burning off in a furnace at 360°C (4 h) and meas-
uring weight loss during these processes. The weight
change after heating (two stages) was considered to be the
percentage of SOM content of the soil samples (Oudbashi
2018). The pH of the soil samples was analysed using
a Metrohm 744 pH meter in a procedure based on the
ASTM standard D4972-01 (ASTM International 2007).
A water extract was prepared from the dried and sieved
soil samples with 1/1 proportion of soil and water. From
these, the EC of each sample was determined using a
WTW inoLab Terminal 740 conductometer (Pansu
and Gautheyrou 2006) and soluble anions measured by
ion chromatography using a Thermo Scientific Dionex
ICS-2100 ion chromatography system.
Results and discussion
Corrosion analysis
The results of the XRD analysis of the corrosion prod-
ucts from the seven samples are presented in Table 1
and Figure 2. The main corrosion products detected
were cuprite (Cu2O) and malachite (CuCO3.Cu(OH)2).
152  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Figure 2. Diffractogram of the

corrosion products of the seven
studied vessels showing the
presence of cuprite and malachite as
 the main corrosion products

Table 1. Results of the XRD analysis of the corrosion products in seven bronze vessels from the Deh Dumen graveyard

Cuprite Malachite Tenorite Paratacamite Brochantite Copper Calcite Quartz

Cu2O CuCO3.Cu(OH)2 CuO Cu2(OH)3Cl Cu4SO4(OH)6 Cu CaCO3 SiO2
PDF Number 05–0667 41–1390 05–0661 25–1427 43–1458 04–0836 05–0586 46–1045
D.P.102     
D.P.144    
D.P.150   
D.P.190  
D.P.222     
D.P.283    
D.P.287     
STUDY ON THE CORROSION MECHANISMS AND MORPHOLOGY OF ARCHAEOLOGICAL BRONZE OBJECTS FROM A BRONZE AGE GRAVEYARD IN SOUTHWESTERN IRAN 153

Tenorite (CuO), paratacamite (Cu 2(OH) 3Cl) and this layer varies from black to dark green and red in
brochantite (Cu4SO4(OH)6) were detected in samples dark-field optical microscopy.
D.P.222 and D.P.287. Quartz (SiO2) and calcite (CaCO3) • The inter- and intragranular corrosion formed under
are likely to be soil minerals from the burial environment. the internal corrosion layer and of significant thickness
Cuprite and malachite are commonly reported corrosion in some cases.
products on archaeological copper alloys (McNeil and
The limit of the original surface is located between the
Little 1992, Scott 2002, Selwyn 2006), whereas tenorite is
external and internal layers and can be determined clearly
rare in natural patinas and has been observed in buried
in the bright-field illumination (Figure 3). The structure
copper alloys exposed to heat (Scott 2002). The presence
of the external layer is completely different from the
of paratacamite, an isomer of copper trihydroxychloride,
internal layer because of the presence of the metallurgical
may indicate bronze disease or active corrosion (Scott
evidence of annealing in the internal corrosion layer.
1990). Copper trihydroxychlorides are a product of the
hydrolysis of nantokite (CuCl) and show that the burial
environment has been contaminated by soluble chloride
salts (Scott 1990 and 2002). Brochantite, or basic copper
sulfate, is a rare corrosion product in buried copper alloys
and has been observed as the main corrosion product in
monuments exposed to urban environments (Graedel,
Nassau and Franey 1987).

The metallography of cross sections of the Deh Dumen

samples shows interesting features of the bronze corro-
sion stratigraphy (Figure 3). The limit of the original
surface is completely retained within the corrosion
layers and a compact corrosion layer, whose thickness
varies from a few to more than 100 microns, has formed
under the original surface in all samples. Inter- and
intragranular attacks are visible under the internal
corrosion layer that can be considered as a specific
corrosion layer in some cases. The internal attacks
have revealed slip lines and grain shapes, proving the
use of thermo-mechanical processes in the manufac-
Figure 3. Metallographic micrographs of some bronze samples from the
ture of the vessels (Oudbashi, Naseri and Malekzadeh
Deh Dumen graveyard in bright- and dark-field illumination. The limit
2016, Oudbashi and Naseri 2017). A corrosion layer of the original surface is as well as internal and external corrosion layers
deposited on the limit of the original surface is variable are indicated by the arrows in the bright-field micrographs
in thickness across and between samples. Differences
between the microstructure of the internal and external
Soil analysis
corrosion layers establish the location of the original
surface in the bronze samples. In summary, three Table 2 shows the results of the determination of different
different corrosion layers can be characterised in the soil factors in three samples. The pH of the soil samples
Deh Dumen samples: varies from 7.94 to 8.08, showing that the soil of Deh
• The external layer formed on the limit of the original Dumen is slightly alkaline, and their EC lies between
surface that is green, black or red in colour, with 185.5 and 233 μs/cm, evidencing the low salinity of the
different thicknesses and a multi-layered structure site soil. As demonstrated by an earlier analysis of another
in some cases. archaeological site in Iran, a low EC (ca. 500 μs/cm or less)
• The internal layer formed under the limit of the orig- may be attributable to low salinity of the soil environment
inal surface and of variable thickness which retains (Oudbashi 2018). The water and SOM content is very
grain characteristics of the bronze as the ‘ghost struc- low in all samples, measuring 0.86%–1.52% (water) and
ture’ visible in bright-field illumination. The colour of 0.54%–1.1% (SOM) in the samples.
154  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Table 2. Results of different analyses of soil samples from the Deh Dumen graveyard including pH, water and SOM, EC and soluble anions
pH EC Water
μs/cm % ppm
G.15 8.08 185.5 0.86
283 7.94 227 1.23
287 8.03 233 1.52
The soluble anions in the soil samples were measured
to trace their content, with soluble sulfate and chloride
having the highest detected concentrations in the soil
composition (Table 2). Figure 4 presents a columnar
diagram showing comparisons of different soluble
anions measured in the three soil samples from the
Deh Dumen site. In all samples, chloride is measured
at 14–18 ppm, while sulfate was determined at ca.
20 ppm; soluble nitrate was also detected, as were soluble
carbonate and phosphate in very low amounts.
Figure 4. Columnar diagram showing the soluble anion content of three
soil samples from the site
From these results, it is likely that the soil of the Deh
Dumen graveyard can be categorised as non- or partially
corrosive towards archaeological copper alloys (Wagner et
al. 1997, Fernandes 2009, Oudbashi 2018). The low soluble
salt content, partially alkaline pH and low water and SOM
content support this conclusion. Other soil content factors
are important (Oudbashi 2018) and should be considered
but, based on those measured here, the burial environment
of the Deh Dumen graveyard can be considered safe for
the long-term reburial of archaeological objects.
Corrosion mechanism in the studied bronzes
Investigation of the corrosion of archaeological bronzes
from the Deh Dumen graveyard indicates that these
objects exhibit type I corrosion, or noble patina (Robbiola,
Blengino and Fiaud 1998). In this type of corrosion,
CORROSION STUDIES
SOM Cl- NO3- CO32- SO42- PO43-
0.54 14.1 1.07 0 19.4 0.14
1.1 13.91 2.58 0 20.64 0.08
0.95 18.07 2.4 0.54 19.63 0
a tin-rich patina forms under the limit of the original
surface of the bronze that is the result of the selective disso-
lution of copper and internal oxidation of tin (Robbiola,
Blengino and Fiaud 1998, Piccardo, Mille and Robbiola
2007). In many cases, an external corrosion layer including
copper (II) compounds can occur over the original surface
(Robbiola, Blengino and Fiaud 1998, Oudbashi, Hasan-
pour and Davami 2016). The three-layered corrosion
stratigraphy observed in the Deh Dumen bronzes fits very
well with the model presented by Robbiola et al. (1994
and 1998) as type I corrosion. The corrosion products
identified in these artefacts are observed commonly in
bronzes with noble patina, including cuprite and mala-
chite. Cuprite can form as a corrosion product in both
internal and external layers, while malachite is one of the
main corrosion products of copper that deposits over
the original surface (Oudbashi, Hasanpour and Davami
2016). The presence of copper trihydroxychlorides in
bronzes with type I corrosion is unusual, but some local
chloride attacks may have occurred in these bronzes due to
a concentration of chloride ions in some areas leading to
the formation of chloride compounds as corrosion prod-
ucts (Robbiola, Blengino and Fiaud 1998). The presence
of malachite as the main corrosion product is interesting
as no significant amount of soluble carbonate has been
found in the soil of the graveyard. The pH of the soil
samples is approximately pH 8 and, in these conditions,
carbonate will transform into bicarbonate because more
alkaline conditions are necessary for the stabilisation of
soluble carbonates in the soil (Boxma 1972, Brautigan
2010). Thus, the probable presence of large amounts
of soluble bicarbonate anion has caused the formation
of malachite as the main corrosion product in the Deh
Dumen graveyard.
The presence of cuprite and malachite as the main corro-
sion products in the studied bronzes can be explained
with the following equations:
STUDY ON THE CORROSION MECHANISMS AND MORPHOLOGY OF ARCHAEOLOGICAL BRONZE OBJECTS FROM A BRONZE AGE GRAVEYARD IN SOUTHWESTERN IRAN
in which it can be seen that these copper compounds
form in the presence of oxygen, water and CO2 (McCann
et al. 1999). High concentrations of CO2 (or carbonate/
bicarbonate in soil) may, of course, lead to the formation
of another isomer of basic copper carbonate, azurite
(McNeil and Little 1992).
The presence of paratacamite as a minor corrosion
product in two samples shows that, despite the low
concentration of soluble chloride in the soil, active
corrosion has occurred. This is the reaction of copper
and chloride in the presence of oxygen and moisture.
The following equations show this corrosion process
(Scott 1990 and 2002):
The presence of brochantite is also interesting. This
corrosion compound is common in bronze statuary
located in urban environments but the presence of
copper sulfates as corrosion products in archaeological
copper alloys is a rare phenomenon. Brochantite has been
observed as a corrosion product in some archaeological
bronzes and has been attributed to the oxidation of
copper sulfides which result from bacterial activities in
the burial environment (Muras and Scott 2018). Sulfate
and chloride are the main soluble anions in the soil of
the Deh Dumen graveyard, although they have been
measured in trace amounts. The presence of these soluble
salts may have led to the formation of paratacamite and
brochantite in some samples.
There is a very good correlation between corrosion
mechanisms and stratigraphy in the bronze objects
and soil conditions in the Deh Dumen site. The soil of
the graveyard can be classified as a non- or partially
corrosive soil that has caused formation of type I corro-
sion in the objects by retaining the limit of the original
surface, forming noble patina and depositing non-active
corrosion compounds, such as malachite, as external
corrosion layers. The presence of cuprite and malachite
as the main corrosion products in all samples along with
the retention of the limit of the original surface and
surface details in these bronze objects allows classifica-
tion of the site burial environment as non- or partially
corrosive. This is in line with a previous study of archae-
ological bronzes from another site in Iran (Oudbashi
2018). Paratacamite (Cu2(OH)3Cl) was detected as a
155
minor corrosion product in two Deh Dumen bronze
objects, indicating the occurrence of bronze disease
in these objects. The soluble chloride identified in the
graveyard soil may have led to the formation of basic
copper chlorides during burial or as a post-excavation
product (Scott 1990 and 2002).
Conclusion
Study of the corrosion mechanism of some excavated
bronze vessels and parameters of the burial environ-
ment in the Bronze Age graveyard of Deh Dumen was
performed to investigate the correlation between soil
conditions and corrosion events. The results show that the
corrosion phenomenon occurred in the bronze objects
is type I corrosion with formation of noble patina and
with copper oxides and carbonate as the main corrosion
products. The soil environment of the graveyard was
determined to be non- or partially corrosive with slightly
alkaline pH and small amounts of some soluble salts,
which is in line with the type I corrosion identified. Some
evidence of active corrosion is observed in two objects,
including the presence of basic copper chloride as a
minor corrosion product which may be attributable to the
presence of small amounts of soluble chloride in the soil.
The corrosion mechanism of the Deh Dumen bronzes
will be the subject of further studies with a wider range
of analytical methods to gain a better understanding of
other aspects of the corrosion mechanisms occurring
in these objects.
Acknowledgements
The authors are grateful to Mehri Ghobadi, Art Univer-
sity of Isfahan, for her valuable help in performing soil
analyses, and to Claudia Chemello, the coordinator of
the ICOM-CC Metals Working Group, for reading and
editing the paper. This study was undertaken with the
financial support of the Deputy for Research and Tech-
nology, Art University of Isfahan, Iran.
References
ASTM International. 2007. Standard test method for pH
of soils (ASTM D4972-01). West Conshohocken, PA,
USA. https://doi.org/10.1520/d4972-01r07 .
Boxma, R. 1972. Bicarbonate as the most important soil
factor in lime-induced chlorosis in the Netherlands. Plant
and Soil 37: 233–43.
156  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Brautigan, D. 2010. Chemistry, phytotoxicity and reme-
diation of alkaline soils. PhD dissertation, The University
of Adelaide, Australia.
Fernandes, R. 2009. Study on Roman and Merovingian
copper alloyed artefacts. In soil corrosion processes and
recycling practices. Master’s thesis (O-variant), Vrije
Universiteit Amsterdam, The Netherlands.
Graedel, T.E., K. Nassau, and J.P. Franey. 1987. Copper
patinas formed in the atmosphere–I. Introduction.
Corrosion Science 27: 639–57.
Imbardelli, D., M.C. Gallucci, and G. Chidichimo. 2017.
SEM and electrochemical characterization of bronze
artifacts from the Francavilla archaeological site. Archae-
ological and Anthropological Sciences 9: 567–78.
McCann, L.I., K. Trentelman, T. Possley, and B. Golding.
1999. Corrosion of ancient Chinese bronze money trees
studied by Raman microscopy. Journal of Raman Spec-
troscopy 30: 121–32.
McNeil, M.B. and B.J. Little. 1992. Corrosion mechanism
for copper and silver in near-surface environments.
Journal of the American Institute for Conservation 31:
355–66.
Muras, V. and D.A. Scott. 2018. The occurrence of
brochantite on archaeological bronzes: A case study from
Lofkënd, Albania. Studies in Conservation 63: 113–25.
Oudbashi, O. 2015. Multianalytical study of corrosion
layers in some archaeological copper alloy artefacts.
Surface and Interface Analysis 47: 1133–47.
Oudbashi, O. 2018. A methodological approach to
estimate soil corrosivity for archaeological copper alloy
artefacts. Heritage Science 6(2).
Oudbashi, O., A. Hasanpour, and P. Davami. 2016. Inves-
tigation on corrosion stratigraphy and morphology in
some Iron Age bronze alloys vessels by OM, XRD and
SEM-EDS methods. Applied Physics A 122: 262.
Oudbashi, O. and R. Naseri. 2017. Archaeometallurgy
of tin bronze in the Deh Dumen Bronze Age graveyard,
southwest of Iran. In Archaeometallurgy in Europe IV,
Bibliotheca Præhistorica Hispana, vol. XXXIII, eds. I.
Montero Ruiz and A. Perea, 105–14. Madrid: Consejo
Superior de Iinvestigaciones Científicas.
Oudbashi, O., R. Naseri, and M. Malekzadeh. 2016.
Technical studies on the Bronze Age metal artefacts from
the graveyard of Deh Dumen, south-western Iran (third
millennium BC). Archaeometry 58(6): 947–65.
CORROSION STUDIES
Pansu, M. and J. Gautheyrou. 2006. Handbook of soil
analysis: Mineralogical, organic and inorganic methods.
Berlin: Springer.
Piccardo, P., B. Mille, and L. Robbiola. 2007. Tin and
copper oxides in corroded archaeological bronzes. In
Corrosion of metallic heritage artefacts: Investigation,
conservation and prediction for long-term behaviour, eds.
P. Dillmann, G. Béranger, P. Piccardo, and H. Matthiesen,
European Federation of Corrosion Publication 48,
239–62. Cambridge: Woodhead Publishing.
Pronti, L., A.C. Felici, M. Alesiani, O. Tarquini, M.P. Brac-
ciale, M.L. Santarelli, G. Pardini, and M. Piacentini. 2015.
Characterisation of corrosion layers formed under burial
environment of copper-based Greek and Roman coins
from Pompeii. Applied Physics A 121: 59–68.
Robbiola, L., C. Fiaud, and A. Harch. 1994. Characteri-
zation of passive layers of bronze patinas (Cu-Sn alloys)
in relation with tin content of the alloy. In Proceedings
of the European Symposium on Modifications of Passive
Films, Paris, 15–17 February 1993, Paris, eds. P. Marcus,
B. Baroux, and M. Keddam, 150–54. London: The Insti-
tute of Materials.
Robbiola, L., J.M. Blengino, and C. Fiaud. 1998.
Morphology and mechanisms of formation of natural
patinas on archaeological Cu-Sn alloys. Corrosion Science
40(12): 2083–111.
Scott, D.A. 1990. Bronze disease: A review of some chem-
ical problems and the role of relative humidity. Journal
of the American Institute for Conservation 29: 193–206.
Scott, D.A. 1991. Metallography and microstructure
of ancient and historic metals. Los Angeles: The Getty
Conservation Institute.
Scott, D.A. 2002. Copper and bronze in art: Corrosion,
colorants, conservation. Los Angeles: The Getty Conser-
vation Institute.
Selwyn, L.S. 2006. Corrosion of metal artifacts in buried
environments. In Corrosion: Environments and indus-
tries, ASM Handbook, vol. 13C, eds. S.D. Cramer and
B.S. Covino, Jr., 306–22. Materials Park, OH: ASM
International.
Sołtysiak, A. and R. Naseri. 2017. Human remains from
Deh Dumen, Iran, 2013–2016. Bioarchaeology of the
Near East 11: 70–75.
Theocharopoulos, S.P., I.K. Mitsios, and I. Arvanitoy-
annis. 2004. Traceability of environmental soil meas-
urements. Trends in Analytical Chemistry 23: 237–51.
STUDY ON THE CORROSION MECHANISMS AND MORPHOLOGY OF ARCHAEOLOGICAL BRONZE OBJECTS FROM A BRONZE AGE GRAVEYARD IN SOUTHWESTERN IRAN 157

Wagner, D., F. Dakoronia, C. Ferguson, W.R. Fischer,

C. Hills, H. Kars, and R. Meijers. 1997. ‘Soil archive’
classification in terms of impacts of conservability of
archaeological heritage. In Metal 95: Proceedings of the
International Conference on Metals Conservation, Semur-
en-Auxois, 25–28 September 1995, eds. I.D. MacLeod, S.L.
Pennec, and L. Robbiola, 21–26. London: James & James.

Authors
Omid Oudbashi has a BSc, MA and PhD in conserva-
tion of cultural and historical properties and is currently
assistant professor in the Department of Conservation
of Cultural and Historical Properties, Art University of
Isfahan, Iran. His research interests are the deterioration
of archaeological objects, especially the corrosion mecha-
nism of ancient metals in burial environments, preventive
and archaeological conservation, archaeometallurgy and
metal working study of the ancient world and metallog-
raphy of ancient and historic metals.

Reza Naseri has a BA, MA and PhD in archaeology

and is currently assistant professor in the Department
of Archaeology, University of Zabol, Iran. His research
interests are the archaeology of the prehistoric and early
historic periods of Iran (from the Bronze to the Iron Age).

Banafsheh Heidarpour has a BSc and MA in conser-

vation of cultural and historical properties from the Art
University of Isfahan, Iran. She has worked on desalina-
tion and inhibition of active corrosion in archaeological
copper alloys, and has experience in the study of corro-
sion and conservation of archaeological metals.

Ali Ahmadi has a BSc in conservation of cultural and

historical properties from the Art University of Isfahan,
Iran.
When Glass and Metal Corrode Together, VII: Zinc
Formates and Further Unknown Zinc Compounds

Andrea Fischer* Jörg Stelzner Sebastian Bette Robert E. Dinnebier

State Academy of Art and Design
Stuttgart, Germany
andrea.fischer@abk-stuttgart.de
Gerhard Eggert
State Academy of Art and Design
Stuttgart, Germany
gerhard.eggert@abk-stuttgart.de
State Academy of Art and Design
Stuttgart, Germany
joerg.stelzner@abk-stuttgart.de
State Academy of Art and Design Max Planck Institute for Solid State
Max Planck Institute for Solid State Research
Research Stuttgart, Germany
Stuttgart, Germany r.dinnebier@fkf.mpg.de
s.bette@fkf.mpg.de
*Author for correspondence

Abstract
Decaying glass induces corrosion of adjacent brass in heritage
objects containing both materials. Indoor air pollution by
formaldehyde leads to the precipitation of formates. While
copper corrosion products have been studied intensively, zinc
compounds need further characterisation. In comprehensive
surveys of collections, zinc formate dihydrate was the most
frequent product in contact with glass, whereas occurrences
without glass were rarely reported in conservation science.
The crystal structure and formula of a novel product, the
basic zinc copper formate Zn(10-6x)Cu(3+6x)(HCOO)8(OH)18∙6H2O
(‘Zinc C’), could be proposed from powder diffraction data.
Both compounds also occurred in laboratory simulation
experiments. A number of other zinc corrosion products
containing sodium and/or potassium and/or copper need
further study in order to provide a rational basis for preventive
and remedial conservation.
Keywords
basic zinc copper formate, brass, GIMME, glass, metal
corrosion, zinc formate

Introduction
Due to their composition, a great number of historic glass
objects are unstable. Alkaline liquid films on glass formed
in reaction with water molecules may serve as an electro-
lyte, inducing the corrosion of adjacent metals (glass-in-
duced metal corrosion of museum exhibits, GIMME).
These films contain ions such as OH-, Na+ and/or K+ from
the ion exchange reactions occurring on the glass surface
and CO32- from the absorption of atmospheric carbon
dioxide. Therefore, the basic carbonates NaOH·2PbCO3
(Fischer and Eggert 2013) and KOH·2PbCO3 (Bette et
al. 2017) have been identified on lead in contact with
glass, chalconatronite, Na2[Cu(CO3)]2∙3H2O (Fischer,
Eggert and Stelzner, forthcoming) and sodium copper
acetate carbonate (Fischer 2016, 111f.) on copper alloys.
Indoor carbonyl pollutants emitted, for example, from
wood products play an important role. In alkaline films,
formaldehyde can directly react to formate (HCOO-) via
the Cannizzaro reaction without intermediate adsorp-
tion and oxidation steps: 2H2CO + NaOH → CH3OH +
Na+HCOO-. Therefore, formates rather than carbonates
are prevalent in the salts on historic glass (Verhaar
et al. 2016) and as GIMME corrosion products on
copper alloys. Most frequently, the copper formates
Na4Cu4O(HCOO)8(OH)2∙4H2O (Fischer et al. 2018) and
Cu(OH)3HCOO (Eggert et al. 2011) are detected.
GIMME was often found on objects made of ‘golden’
brass, which contains zinc. Zinc corrosion products were
detected on 58 samples from eleven collections. The
aim of this study is to characterise these compounds by
analysing samples from brass objects in contact with glass
and coupons from laboratory simulation experiments to
understand better the corrosion taking place and to form
a rational base for conservation.
Experimental
Examination methods
Combined glass-metal objects were examined in
numerous museum collections and alterations and
damage on the glass and metal surface characterised
while gathering samples of the corrosion products. First,
the corrosion products were investigated by µ-Raman
WHEN GLASS AND METAL CORRODE TOGETHER, VII: ZINC FORMATES AND FURTHER UNKNOWN ZINC COMPOUNDS
spectroscopy. Being able to use this technique to focus
on individual particles under the microscope is an
advantage, as in many instances only small quantities
of corrosion products were available. Particles could
be easily differentiated according to their colour and
shape. All spectra were recorded using a Renishaw inVia
Raman spectrometer (grating 1800 l/mm, resolution
1 cm-1) equipped with a Leica DMLM microscope and a
RenCam CCD detector. The spectra were taken from 100
to 3700 cm-1 using a He–Ne laser operating at 632.8 nm.
The power of the laser was less than 400 μW focused
onto a 0.7 μm spot through a 50× microscope objective.
Element analyses were carried out on unknown
compounds using scanning electron microscopy/ener-
gy-dispersive X-ray spectroscopy (SEM-EDX). A Zeiss
EVO 60 microscope with a Bruker X Flash 6130 detector
was used. Field-emission scanning electron microscopy
(FE-SEM) images were obtained using a Zeiss SUPRA
40 VP Microscope and a Thermo Scientific detector.
If there was a sufficient amount of an unknown corro-
sion product, X-ray powder diffraction (XRPD) meas-
urements were carried out. High-resolution XRPD
patterns were collected using a Stoe Stadi-P high-reso-
lution laboratory powder diffractometer (primary beam
Johann-type Ge monochromator for Cu-Kα1-radiation)
using a linear position sensitive silicon strip detector
(Mythen-Dectris) with an opening of approximately
12° 2θ. XRPD patterns for crystal structure determination
were recorded on a Bruker D8 Advance diffractometer
(Cu-Kα1 radiation from primary beam Ge(220) Johans-
son-type monochromator and Bruker LynxEye position
sensitive detector) with Debye-Scherrer geometry. The
sample was sealed in a 0.5 mm diameter borosilicate glass
capillary (Hilgenberg glass no. 0140), which was spun
during the measurement. The pattern was measured in
a 2θ range from 5° to 90° applying a total scan time of
20 hours.
Laboratory simulation experiments
Essentially, GIMME is metal corrosion in the presence of
alkali carbonate solutions exposed to carbonyl pollutants
via the air. For simulation experiments, brass coupons
with zinc contents of 10% (CuZn10) and 37% (CuZn37)
were dipped into alkali carbonate solutions (1 mol/l),
mimicking the alkaline films on soda and potash glass.
The coupons were then exposed in desiccators (2 litre
volume) to vapours of aqueous formaldehyde solutions
159
of 4% or 37%, yielding atmospheric concentrations of
roughly 100 ppm or 925 ppm by volume (Tétreault et
al. 2003). The relative humidity was kept at 75% by a
saturated sodium chloride solution and at 85% by potas-
sium chloride. After six months, the corrosion products
grown on the coupons were analysed using µ-Raman
spectroscopy.
Results
Zinc formate dihydrate
GIMME corrosion products on copper alloys are usually
green to green-blue and cannot be distinguished without
chemical analysis. But the bezel of a pocket watch
(Table 1, no. 1) in contact with ‘weeping’ glass (Figure 1)
showed not only green-blue (identified as sodium
copper formate), but also white corrosion products,
which contained Zn as the only metal (SEM-EDX). The
object had been stored in a mobile cabinet partly made
of chipboard for more than 30 years. Micro-Raman
spectroscopy and XRPD identified the product as zinc
formate dihydrate, Zn(HCOO) 2·2H 2O. The Raman
spectrum displayed sharp bands at 1355 and 1372 cm-1
(O-C-O-symmetric stretch), a shoulder at 1386 cm-1
(C-H deformation) and a weak band at 1397 cm-1 (O-C-O
symmetric stretch) (Forslund et al. 2013).
Figure 1. Detail of a pocket watch with droplets on the inside of the cover
glass. White particles visible in the green GIMME corrosion. Museum für
Hamburgische Geschichte, no. 1899.101b (Table 1, no. 1)
This product has so far been rarely detected on heritage
metal objects, mostly by scientists at the British Museum
in connection with wood-based materials for drawers,
cupboards and display cases – Werner (1972, 60), Oddy
and Bradley (1989, 237), Green and Thickett (1993,
263f.). The latter detected zinc formate dihydrate on a
160  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Table 1. Corrosion samples from heritage objects with zinc compounds (detected by Raman spectroscopy and SEM-EDX) from glass-induced metal
corrosion. Figures given in the second column correspond to the SEM-EDX elemental ratio

No. Sample no. Corrosion products No. Sample no. Corrosion products
Museum für Hamburgische Geschichte Glass button, SNM- Chalkonatronite
22
Pocket watch Sodium copper formate LM-92631-1 White cpd., Zn:Cu:Na ca. 10:3:5
1
MHG-1899.101b-1 Zinc formate dihydrate Glass button Blue green cpd. with
23
Earring Sodium copper formate SNM-LM-92645-1 Zn:Cu:Na:K ca. 5:9:2:2
2
MHG-2015-278-1 Zinc formate dihydrate Glass button Sodium copper acetate carbonate
24
Earrings Sodium copper formate SNM-LM-92645-2 Greenish cpd., Zn:Cu:Na ca. 3:2:1
3
MHG-2015-275-1 Zinc formate dihydrate Glass button Unknown compound
25
Earring Sodium copper formate SNM-LM-92647-1 White cpd., Zn:Na ca. 7:2, Cu low
4
MHG-2015-277-1 Zinc formate dihydrate Glass button Whitish cpd. with
26
Earrings Sodium copper formate SNM-LM-93875-1 Zn:Cu:Na:K ca. 10:3:3:2
5
MHG-2015-279-1 Zinc formate dihydrate Glass button Sodium copper acetate carbonate
27
Earring Sodium copper formate SNM-LM-93887-1 Green cpd.
6 MHG-2015-244-0-1 Copper formate trihydroxide Miniature painting Copper formate trihydroxide
Zinc C 28
SNM-LM-4587-1 White cpd.
Earring Sodium copper formate Miniature painting Copper formate trihydroxide
7 MHG-2015-247-0-1 Zinc formate dihydrate 29
SNM-LM-4587-2 Green cpd.
Zinc C
Bridal crown Copper formate trihydroxide
Earrings Sodium copper acetate carbonate 30
SNM-AG-2416-2 Green cpd., Zn:Cu:K ca. 3:5:1
8
MHG-2015-281-3 Greenish cpd., Zn:Cu:Na ca. 7:6:3
Museumslandschaft Hessen Kassel
Musée d’Ethnographie de Neuchâtel
Hat needle (Sch 47) Sodium copper formate
Necklace Sodium copper formate 31 MHK-Sch-58b-2 Sodium copper acetate carbonate
9
MEN-IIIA.564-1 Zinc formate dihydrate Zinc A, Zn:Na ca. 1:1, 2% Cu
Swiss National Museum, Zurich Germanisches Nationalmuseum, Nuremberg
Hat needle Sodium copper formate Necklace Sodium copper formate
10 32
SNM-ohne-3-1 Zinc formate dihydrate GNM-Kl-2022-2 Zinc B, Zn:K:Na ca. 4:5:3,
Christmas tree bauble Zinc formate dihydrate Earrings Sodium copper formate
11 33
SNM-LM-72958-1 GNM-Kl-2011-1 Zinc B, Zn:K:Na ca. 3:4:2
Thermometer Sodium copper formate Belt buckle Copper formate trihydroxide
12 34
SNM-LM-94198-1 Zinc formate dihydrate GNM-T-3197-1 Zinc B, Zn:K ca. 1:1.5-2
Lorgnette with Sodium copper formate Pocket microscope Copper formate trihydroxide
13 tortoiseshell frame Zinc formate dihydrate 35
GNM-WI-1452-1 Green cpd., Zn:Cu ca. 2:1
SNM-LM-71249-1
Historisches Museum Basel
Christmas tree decoration Copper formate trihydroxide
SNM-LM-71220-1 Spectacles Zinc C, Zn:Cu:Na ca. 11:5:2
Sodium copper acetate carbonate 36
HMB-1931.111-1
14 Zinc A, Zn:Cu:Na ca. 10:2:5
White cpd., Zn:Cu:Na ca. 10:2:5 Institute of Plant Biology, Zurich
Greenish cpd., Zn:Cu:Na ca. 10:3:4 Sample tube Zinc C, Zn:Cu:Na ca. 10:4:3
37
Glass button Sodium copper acetate carbonate UZH.PB-ohne-1-1
15 SNM-LM-93887-1 Zinc A, Zn:Cu:Na ca. 10:2:5 Sample tube Copper formate trihydroxide
Zn-Cu-carboxylate cpd. 38
UZH.PB-ohne-1-2 Zinc C
Glass button White cpd., Zn:Cu:Na ca. 10:3:4 Landesmuseum Württemberg, Stuttgart
16
SNM-LM-92626-1
Miniature painting Sodium copper formate
Glass button Greenish cpd., Zn:Cu:Na ca. 13:4:2 39 Inv.-Nr. 22,110 Copper formate trihydroxide
17
SNM-LM-92626-2 White cpd., Zn:Cu:Na ca. 5:2:2 Zinc A, Zinc C
Glass button Greenish cpd. Miniature painting Sodium copper formate
18 40
SNM-LM-92627-1 White cpd., Zn:Cu:Na ca. 7:1:2 Inv.-Nr. 1927-433 Zinc formate dihydrate
Glass button Chalconatronite Miniature painting Copper formate trihydroxide
19 41
SNM-LM-92629-1 White cpd., Zn:Cu:Na ca. 5:1:3 Inv.-Nr. 1940-11 Zinc C
Glass button, SNM- White cpd., Zn:Cu:Na ca. 5:1:3 Miniature painting Zinc A, Zinc C
20 42
LM-92629-2 Inv.-Nr. 9264
Glass button, SNM- Blue-green cpd. Miniature painting Zinc B
21 43
LM-92630-1 White cpd., Zn:Cu:Na ca. 6:2:1 Inv.-Nr. E1383
WHEN GLASS AND METAL CORRODE TOGETHER, VII: ZINC FORMATES AND FURTHER UNKNOWN ZINC COMPOUNDS 161

No. Sample no. Corrosion products bezel of the pocket watch (Table 1, no. 1). Tabular sodium
Private Collection Elisabeth Krebs, Vienna copper formate microcrystals can be distinguished from
Glass pendant with hanger, Sodium copper formate hexagonal zinc formate dihydrate crystals. Even using
44 sample no. S.K.4 Copper formate trihydroxide a binocular, the different appearance of the very fine
Zinc A
green-blue sodium copper formate and large white zinc
Glass pendant with brass Sodium copper formate
45 hanger, sample no. S.K.6 Copper formate trihydroxide
formate particles is obvious. The small particle size of
Zinc A the sodium copper formate results in a matt appearance.
46
Glass pendant with hanger, Sodium copper formate Under these corrosion products, the surface of the gilded
sample no. S.K.7 Zinc A
copper alloy has been severely damaged.
Glass pendant with hanger, Sodium copper formate
47
sample no. S.K.15 Zinc A
Glass pendant with hanger, Sodium sulphate
48
sample no. S.K.27 Copper zinc compound
Deutsches Bergbau-Museum Bochum
Mining theodolite Copper zinc compound
49
Inv.-Nr. 030150369000
Magnet collimator with Zinc formate dihydrate
50 theodolite
Inv.-Nr. 030150332000 Figure 2. FE-SEM image of corrosion in Figure 1: (a) zinc formate
dihydrate in addition to sodium copper formate; (b) hexagonal zinc
Levelling instrument Zinc formate dihydrate
51 formate dihydrate crystals
Inv.-Nr. 30015071000 Zinc C
Levelling instrument Zinc formate dihydrate
52
Inv.-Nr. 30015067000 Zinc C ‘Zinc C’ = Basic zinc copper formate
Compass with aiming circle Zinc C
53
Inv.-Nr. 030015043000 An unknown compound, labelled ‘Zinc C’ in the course
Security oil lamp Chalconatronite of our investigations, was found on twelve objects in
54
Inv.-Nr. 030140115000 Copper zinc compound five collections (Table 1). Its colour can vary from case
Accu hand lamp Copper zinc compound
55 to case, appearing as whitish, greenish or even bluish.
Inv.-Nr. 030140463000
Firedamp test lamp Copper formate trihydroxide
Sodium copper formate and copper formate trihydroxide
56
Inv.-Nr. 030001852309 Zinc B are usually more intensely coloured than Zinc C, thus
57
Lamp Copper zinc compound the corrosion phases can be well distinguished visually
Inv.-Nr. 030001852930
from each other. The compound was detected on two
Kunstgewerbemuseum, Berlin
brass mountings for glass earrings (Table 1, no. 6 and
Reliquary Zinc B, Zinc C
58
Inv.-Nr. 1896,975
7 ), a metal spectacle frame (no. 36), a metal clamp for
a reagent tube (no. 37 and 38, Figure 3), three brass-
sculpture by Beuys comprising a zinc box containing framed miniature paintings with glass covers (no. 39,
fat, in a (re-)painted wooden showcase, and on German 41, 42), three mining survey instruments (no. 51–53)
zinc coins from World War I and II stored in tight-fitting and a reliquary (no. 58).
wooden cabinets. Knight (1994, 174) identified it on the
brass cuirass of lead soldiers in the Waterloo diorama
at Dover Castle. Our frequent identification of the
compound on artefacts (15 examples in five collections:
Table 1, no. 1–5, 7, 9–13, 40, 50–52) and in our simulation
experiments (Table 2) indicates the role of glass decay Figure 3. (a) Display of plant remains with reagent tubes held by metal
in the formation of formates by providing the necessary clamps; (b) corrosion on metal clamp. Institute of Plant Biology Zurich,
no number (Table 1, no. 37 and 38). © HE-Arc CR
alkaline conditions for the Cannizzaro reaction. This
effect has already been observed for copper corrosion. The Raman spectra showed characteristic bands at 1340,
Zinc formate dihydrate was often accompanied by 1386, 1628 and 2846 cm-1 and a weak hydroxyl band >
sodium copper formate, well known as GIMME copper 3500 cm-1 (Figure 4), indicating a basic formate with
corrosion product (Fischer et al. 2018). The FE-SEM copper and zinc as the main cations (SEM-EDX). The
image (Figure 2) shows the corrosion products on the compound also occurred in simulations (Table 2a and
162  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Table 2. Corrosion products formed during exposure of impregnated brass coupons to formaldehyde

a) Formaldehyde concentration of 4%
Brass + salt Corrosion products, 75% RH Corrosion products, 85% RH
CuZn10 Sodium copper formate Sodium formate
+ Na2CO3 Zinc formate dihydrate Chalconatronite
Zinc C Copper formate trihydroxide
CuZn10 Copper formate trihydroxide Zinc C
+ K2CO3 Zinc C
CuZn10 Copper formate trihydroxide Copper formate trihydroxide
+ Na2CO3 + K2CO3 Zinc C
CuZn10 Zinc C Copper formate trihydroxide
+ NaHCO3 Copper zinc compound
CuZn10 Zinc C Copper formate trihydroxide
+ KHCO3 Zinc C

CuZn37 Zinc formate dihydrate Sodium formate

+ Na2CO3 Zinc C Copper formate trihydroxide
Zinc formate dihydrate
CuZn37 Copper formate trihydroxide Copper formate trihydroxide
+ K2CO3 Zinc formate dihydrate Zinc formate dihydrate
Zinc C Zinc C
CuZn37 Zinc formate dihydrate Zinc formate dihydrate
+ Na2CO3 + K2CO3 Zinc C Zinc C
CuZn37 Zinc formate dihydrate Copper formate trihydroxide
+ NaHCO3 Zinc C Zinc C
CuZn37 Copper formate trihydroxide Copper formate trihydroxide
+ KHCO3 Zinc formate dihydrate Zinc formate dihydrate
Zinc C Zinc C
b) Formaldehyde concentration of 37%
Brass + salt Corrosion products, 75% RH Corrosion products, 85% RH
CuZn10 Sodium copper formate Sodium copper formate
+ Na2CO3 Zinc formate dihydrate Copper formate trihydroxide
Zinc formate dihydrate
CuZn10 Copper formate trihydroxide Copper formate trihydroxide
+ K2CO3 Potassium copper compound Zinc formate dihydrate
Zinc formate dihydrate
CuZn10 Sodium copper formate Sodium copper formate
+ Na2CO3 + K2CO3 Copper formate trihydroxide Copper formate trihydroxide
Zinc formate dihydrate Potassium copper compound
Zinc formate dihydrate
CuZn10 Sodium copper formate Sodium copper formate
+ NaHCO3 Zinc formate dihydrate Copper formate trihydroxide
Zinc formate dihydrate
CuZn10 Copper formate trihydroxide Copper formate trihydroxide
+ KHCO3 Zinc formate dihydrate Potassium copper compound
Zinc formate dihydrate

CuZn37 Sodium copper formate Copper formate trihydroxide

+ Na2CO3 Zinc formate dihydrate Zinc formate dihydrate
CuZn37 Potassium copper compound Copper formate trihydroxide
+ K2CO3 Zinc formate dihydrate Zinc formate dihydrate
CuZn37 Copper formate trihydroxide Zinc formate dihydrate
+ Na2CO3 + K2CO3 Zinc formate dihydrate
CuZn37 Sodium copper formate Copper formate trihydroxide
+ NaHCO3 Zinc formate dihydrate Zinc formate dihydrate
CuZn37 Copper formate trihydroxide Copper formate trihydroxide
+ KHCO3 Zinc formate dihydrate Zinc formate dihydrate
WHEN GLASS AND METAL CORRODE TOGETHER, VII: ZINC FORMATES AND FURTHER UNKNOWN ZINC COMPOUNDS
b) with sodium as well as potassium salts. Therefore, it
might contain no alkali cations at all.
Figure 4. Raman spectrum of corrosion products in Figure 3b: Zinc C
Due to the multiphase character of the samples and
the limited amount of material, classical elemental
analysis does not allow the formula/composition of
such compounds to be determined. No single crystals
were visible, even under magnification, which would
allow conventional crystal structure solution from 3D
diffraction data. But powder diffraction, which reduces
3D reciprocal space to one-dimensional d-spacing, still
contains enough information provided that high preci-
sion measurements and sophisticated methods of data
processing are used to derive the structure, even for large
unit cells (see Bette et al. 2017 and Fischer et al. 2018
for applications in GIMME research). Using TOPAS 6.0
software, atomic positions were localised in the unit cell
and assigned to the metal cation, formate and hydroxide
anions by considering both their scattering power and
coordination geometries. Finally, a phase composition of
Zn(10-6x)Cu(3+6x)(HCOO)8(OH)18∙6H2O could be derived,
supported by Raman and SEM-EDX data.
The Zinc C corrosion phase exhibits a complex crystal
structure. During the process of crystal structure determi-
nation, four distinct metal sites were identified (Figure 5).
Two different sites are exclusively occupied by zinc cations
(yellow polyhedra) and another by copper ions (blue poly-
hedra); the fourth site exhibits an occupational disorder
between zinc and copper cations (green octahedra) yielding
the formula Zn4Cu3(Zn1–xCux)6(HCOO)8(OH)18·6(H2O).
Hence, the corrosion phase shows a variable zinc-copper
ratio that can vary from Zn10Cu3(HCOO)8(OH)18∙6(H2O)
to Zn4Cu9(HCOO)8(OH)18∙6(H2O). For the investigated
sample, a ratio of 2:1 was determined by EDX anal-
ysis, which equates to x = 0.22 and a composition of
163
Zn8.68Cu4.32(HCOO)8(OH)18∙6(H2O). Two distinct formate
ions are present in the crystal structure and they exhibit
monodentate coordination behaviour. One formate ion
bridges three metal cations, whereas the other formate
anion occupies the apical edge of the Zn(HCOO)(OH)3/3
tetrahedron. All hydroxide ions are situated in the coor-
dination spheres of the metal cations and bridge three
of them.
Figure 5. Packing diagrams of Zn(10-6x)Cu(3+6x)(HCOO)8(OH)18∙6H2O with
x = 0.22 in c- (a) and a-direction (b). Yellow polyhedra indicate a metal
position that is purely occupied by zinc, blue polyhedra indicate pure
copper sites and green polyhedra indicate a mixed metal site
In the crystal structure of the corrosion phase, edge-
sharing Zn(HCOO)3/3(OH)3/3 (Figure 5; yellow octa-
hedra), Cu(HCOO)2/3(OH) 4/3 (blue octahedra) and
Zn 1-xCu x(HCOO) 1/3(OH) 5/3 (green octahedra) form
layers perpendicular to the c-axis. Within the layers of
the cation sites are vacant. Each vacancy is surrounded
by six Zn1-xCux(HCOO)1/3(OH)5/3 octahedra (Figure 5a,
green octahedra) within the layer plane and decorated
by two Zn(HCOO)(OH)3/3 tetrahedra (Figure 5a and
b, yellow tetrahedra) above and below that layer plane.
Interlayer interactions are mediated by possible hydrogen
bonds between non-coordinate carboxylate related
oxygen, non-coordinated water molecules and coordi-
nated hydroxide ions. Therefore, the water molecules
are only loosely bound in the crystal structure of the
corrosion phase. Details of the structure determination
are published elsewhere (Bette et al. 2019). Table 3 gives
the eight strongest diffraction lines, allowing diffracto-
grams to be checked for the presence of this compound.
Results of simulation experiments are listed in Table 2a
for 4% formaldehyde and Table 2b for 37%. At the lower
formaldehyde concentration, Zinc C occurs regularly –
independent of the alkali cation – accompanied by zinc
formate dihydrate on CuZn37. At the high formaldehyde
concentration, only zinc formate dihydrate was found as
164  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Table 3. Eight strongest diffraction peaks for Zinc C The Raman spectra show sharp bands at 933, 1375,
d-spacing/Å rel. intensity/% reflection indices 2865 and 2936 cm-1 (Figure 7). The compound often
9.17 100 003 occurred in connection with sodium copper acetate
4.58 11.6 006 carbonate, indicating the emission of acetic acid. This
2.73 12.5 -240 would match an assignment of the 933 cm-1 to the C-C
2.62 20.9 223 stretch in acetate. Zinc A was not found in simulation
2.35 15.5 226 experiments (Table 2), but here no acetic acid or acet-
2.04 9.4 229 aldehyde were used. Three semi-quantitative SEM-EDX
1.56 10.3 060 measurements found relations of Zn:Na:Cu of 1:1:0.01
1.55 6.1 -663 (no. 31) and 2:1:0.4 (no. 14 and 15). Whether copper is
present at all in the lattice can only be seen when the
a zinc corrosion product. Differences in relative humidity
only play a minor role, but larger amounts of Zinc C were crystal structure is solved successfully.
found at 75% RH. The pure sample from the CuZn10
coupon dipped in potassium hydrogencarbonate was
used to determine the crystal structure.

The experiments proved the importance of zinc formate

dihydrate and basic zinc copper formate for brass corro-
sion in the presence of formaldehyde at pH > 7. (A
saturated sodium hydrogencarbonate solution formed
by water adsorption on the dried coupon has a pH of
8–9, while sodium carbonate yields pH ≈12.)

Unknown zinc compounds

The composition of other zinc-containing compounds
detected in a survey of some 300 GIMME objects is still Figure 7. Raman spectrum of white corrosion products in Figure 6: Zinc A

unknown.
Zinc B
Zinc A
Zinc B, another white corrosion product, is rich in
The compound labelled Zinc A forms distinct white potassium, containing more potassium than zinc.
crystalline needles, but XRPD found no match in the Sodium might be a minor element, but in one sample
International Centre for Diffraction Data (ICDD) data- it was totally missing. These conspicuous white
base. It occurred on a hat needle with transparent green corrosion products had been formed on the setting
glass (Table 1, no. 31, Figure 6), a horn-shaped Christmas of a heavily deteriorated black glass stone (Table 1,
tree decoration (no. 14), a button of black glass (no. 15), no. 34; Figure 8). XRPD measurement of a sample
two miniature paintings (together with Zinc C, no. 39 found it to be crystalline but again without a match
and 42) and four glass pendants (no. 44–47). in the ICDD reference database. Zinc B also occurred
on two other glass imitations of gems in the Germa-
nisches Nationalmuseum (no. 32 and 33), on a reli-
quary (no. 58), a miniature painting (no. 43) and a
miner’s lamp (no. 56). The Raman spectrum (Figure 9)
displays intensive peaks at 285, 379, 779, 1333, 1618
Figure 6. Detail and 2867 cm-1 that are useful for the identification
of a hat needle, of this compound. Acetate might not be present, as
Museumslandschaft
Hessen Kassel, no.
there is no band around 940 cm-1 for the C-C stretch.
Sch-58b-2 (Table 1, The compound often occurs together with the usual
 no. 31) GIMME copper formates.
WHEN GLASS AND METAL CORRODE TOGETHER, VII: ZINC FORMATES AND FURTHER UNKNOWN ZINC COMPOUNDS
Figure 8. Black glass
stone on a belt
buckle, Germanisches
Nationalmuseum, no.
 T-3197 (Table 1, no. 34)
Figure 9. Raman spectrum of white corrosion products in Figure 8: Zinc B
Other copper zinc compounds
Whitish to greenish corrosion phases were observed on
the brass shanks of glass buttons in the Swiss National
Museum, collectively termed copper zinc compounds.
In our investigations, a distinction was made here
between white compounds and greenish compounds
(Table 1, no. 15–27). Similar products were found in the
same museum on a Christmas tree decoration (no. 14),
a miniature painting (no. 28) and a bridal crown (no.
30), and on an earring from the Museum für Hambur-
gische Geschichte (no. 8). Products contain variable
amounts of copper, zinc and sodium, with zinc as the
main component.
Corrosion on buttons no. 23 and 26 and the bridal crown
no. 30 also contained some potassium. The Raman
spectra display weak to medium bands at 933, 1353
and 2935 cm-1, pointing to the presence of acetate and
formate as in GIMME copper compounds. The most
intensive peak at 1072 cm-1 might be interpreted as a
symmetric C-O stretch of carbonate as in aurichalcite
(Frost, Hales and Reddy 2007, 5). In some samples this
peak is weaker and lies at 1060 cm-1 as in white hydroz-
incite, Zn5(CO3)2(OH)6 (Bouchard and Smith 2003, 2265).
165
As found for Zinc C, Cu2+ and Zn2+, which have similar
ionic radii, might replace each other, as is known for
a number of minerals. Together with the variability in
anions and alkali cations (from glass) and their ratio, this
points to a corrosion chemistry rich in compounds, yet
nearly unexplored.
Conclusion
When unstable glass induces the corrosion of adjacent
brass, both copper and zinc compounds are formed. This
is what the authors detected in 58 samples from eleven
collections. They were collected from glass imitations
of gems (folk jewellery, etc.), glass baubles, buttons with
brass shanks, glass-framed miniature paintings, a pocket
watch, spectacles, a reagent tube held by metal clamps
and scientific instruments, illustrating the wide range of
heritage object types with glass/brass contact.
The two fully identified corrosion products are both
zinc formates emphasising the enrichment of formate
(from formaldehyde) on the alkaline surfaces of glass.
Other compounds might contain acetate. Like dosim-
eters, corrosion products can thus give a clue as to the
long-term storage conditions of objects. Unfortunately,
these aggregates cannot be distinguished clearly under
the binocular. Instrumental methods are needed for
identification. The aim of this project is to provide the
necessary reference data for this.
The aim of ongoing research in this field is to determine
crystal structures and formulas of other crystalline zinc
compounds using high-resolution XRPD in order to
understand better the corrosion processes required for
informed decision-making in the field of (preventive)
conservation. The ultimate objective is to create a set
of guidelines for conservators on how to deal with the
GIMME phenomenon (Where does it occur? What does
it look like? Prevention, relative humidity, absorbers,
mechanical or chemical removal of corrosion products,
etc.).
Acknowledgements
We gratefully acknowledge the assistance of many
colleagues during our surveys of the collections (Table 1).
Prof. Dr. Christoph Krekel provided access to Raman
and scanning electron microscopy in his Archaeometry
and Conservation Science Lab at the Academy. The
field-emission scanning electron microscope images
166  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
were performed by Dirk Kirchner, Deutsches Berg-
bau-Museum Bochum. Christine Stefani took the XRPD
measurements at the Max Planck Institute for Solid State
Research, Stuttgart. The surveys of the Landesmuseum
Württemberg, Krebs Collection Vienna and Deutsches
Bergbau-Museum Bochum, as well as the simulation
experiments, are part of an ongoing project on glass/
metal corrosion on heritage objects funded by DBU (Az.
33255/01). The crystal structure determination of Zinc
C is part of the ‘In search of structure’ project funded by
DFG (grant EG 137/9-1).
References
Bette, S., G. Eggert, A. Fischer, and R.E. Dinnebier.
2017. Glass-induced lead corrosion of heritage objects:
Structural characterization of K(OH)·2PbCO3. Inorganic
Chemistry 56: 5762−70.
Bette, S., A. Fischer, J. Stelzner, G. Eggert, and R.E.
Dinnebier. 2019. Brass and glass: Crystal structure solu-
tion and phase characterisation of the corrosion product
Zn4Cu3(Zn1–xCux)6(HCOO)8(OH)18·6(H2O). European
Journal of Inorganic Chemistry 2019(7): 920–27.
Bouchard, M. and D.C. Smith. 2003. Catalogue of 45 refer-
ence Raman spectra of minerals concerning research in
art history or archaeology, especially on corroded metals
and coloured glass. Spectrochimica Acta Part A: Molecular
and Biomolecular Spectroscopy 59(10): 2247–66.
Eggert, G., S. Haseloff, H. Euler, and B. Barbier. 2011.
When glass and metal corrode together, III: The forma-
tion of dicoppertrihydroxyformate. In ICOM-CC 16th
Triennial Conference Preprints, Lisbon, 19–23 September
2011, ed. J. Bridgland, art. 901, 9 pp. Almada: Critério
Produção Gráfica, Lda.
Fischer, A. 2016. Glasinduzierte Metallkorrosion an
Museums-Exponaten. PhD Thesis, Staatliche Akademie
der Bildenden Künste, Stuttgart, Germany.
Fischer, A., G. Eggert, D. Kirchner, H. Euler, and B.
Barbier. 2014. When glass and metal corrode together,
IV: Sodium lead carbonate hydroxide. In Metal 2013:
Proceedings of the Interim Meeting of the ICOM-CC Metals
Working Group, Edinburgh, 16–20 September 2013, eds.
E. Hyslop, V. Gonzalez, L. Troalen, and L. Wilson, 13–19.
Edinburgh: ICOM-CC and Historic Scotland.
Fischer, A., G. Eggert, R. Dinnebier, and T. Runčevski.
2018. When glass and metal corrode together, V: Sodium
copper formate. Studies in Conservation 63(6): 342–55.
CORROSION STUDIES
Fischer, A., G. Eggert, and J. Stelzner. Forthcoming. When
glass and metal corrode together, VI: Chalconatronite.
(Submitted for publication.)
Forslund, M., C. Leygraf, P.M. Claesson, C. Lin, and J.
Pan. 2013. Micro-galvanic corrosion effects on patterned
copper-zinc samples during exposure in humidified air
containing formic acid. Journal of the Electrochemical
Society 160(9): C423–31.
Frost, R.L., M.C. Hales, and B.J. Reddy. 2007. Aurichalcite
– An SEM and Raman spectroscopic study. Polyhedron
26(13): 3291–300.
Green, L.R. and D. Thickett. 1993. Modern metals
in museum collections. In Proceedings of the Sympo-
sium’91: Saving the Twentieth Century. The Degradation
and Conservation of Modern Materials, Ottawa, 15–20
September 1991, ed. D.W. Grattan, 261–72. Ottawa:
Canadian Conservation Institute.
Knight, B. 1994. Passive monitoring for museum show-
case pollutants. In Preventive conservation: Practice,
theory and research. Preprints of the Contributions to
the Ottawa Congress, eds. A. Roy and P. Smith, 174–76.
London: International Institute for Conservation.
Oddy, W.A. and S.M. Bradley. 1993. The corrosion of
metal objects in storage and display. In Current Problems
in the Conservation of Metal Antiquities: International
Symposium on the Conservation and Restoration of
Cultural Property, Tokyo, 4-6 October 1989, ed. S. Aoki,
225–44. Tokyo: Tokyo National Research Institute of
Cultural Properties.
Tétreault, J., E. Cano, M. van Bommel, D.A. Scott,
M. Dennis, M.-G. Barthés-Labrousse, L. Minel, and
L. Robbiola. 2003. Corrosion of copper and lead by
formaldehyde, formic and acetic acid vapors. Studies in
Conservation 48(4): 237–50.
Werner, A.E. 1972. Conservation and display, 1: Envi-
ronment control. Museums Journal 72: 58–60.
Authors
Andrea Fischer is an objects conservator. She has a PhD
from the State Academy of Art and Design Stuttgart and
is a senior lecturer in objects’ conservation at the same
institution.
Gerhard Eggert FIIC has a PhD in chemistry from the
University of Bonn. He was head of the Conservation
Department at Rheinisches Landesmuseum Bonn from
WHEN GLASS AND METAL CORRODE TOGETHER, VII: ZINC FORMATES AND FURTHER UNKNOWN ZINC COMPOUNDS 167

1985 to 1998 and has been chair of Objects’ Conserva-

tion at the State Academy of Art and Design Stuttgart
since 1998.

Jörg Stelzner is an objects conservator. He has a PhD

from the State Academy of Art and Design Stuttgart
and has researched glass-induced metal corrosion
(DBU-funded) since 2016.

Sebastian Bette has a PhD in chemistry from TU

Bergakademie Freiberg. He has been a researcher on the
structure of corrosion products in museum objects at the
Max Planck Institute for Solid State Research, Stuttgart
(DFG-funded) since 2017.

Robert E. Dinnebier is a mineralogist. He has a PhD

from the University of Heidelberg and a Habilitation
from the University of Bayreuth. He is head of Scientific
Facility Diffraction at the Max Planck Institute for Solid
State Research, Stuttgart.
A Dark Discoloration on Aristide Maillol’s Outdoor
Lead Sculptures: Identification, Formation, and Further
Research
Manuela Toro* Ineke Joosten
University of Amsterdam National Heritage Laboratory
Amsterdam, The Netherlands Rijksdienst voor het Cultureel
mtsilvad@gmail.com Erfgoed (RCE)
Amsterdam, The Netherlands
Tonny Beentjes I.Joosten@cultureelerfgoed.nl
University of Amsterdam
Amsterdam, The Netherlands
T.P.C.Beentjes@uva.nl
Abstract
Two lead sculptures by Maillol, one which exhibits a dark red
discoloration of the surface and one which does not, are held
in the collections of the Kröller-Müller Museum, Otterlo, and
the Museum of Modern Art (MoMA), New York, respectively.
In an effort to understand better the formation of the
corrosion layer, the two works were analyzed and compared
in a collaborative effort between authors at the University
of Amsterdam and MoMA. Technical analysis using scanning
electron microscopy with energy-dispersive x-ray spectroscopy
(SEM-EDX) and x-ray fluorescence (XRF) was performed, and
Introduction
The French sculptor Aristide Maillol (1861–1944) cast a
selection of his sculptures in lead together with founder
Eugène Rudier throughout the early 20th century. In
recent years, an unexplained, dark red discoloration has
been noted on a cast edition of Maillol’s L’Air (The Air)
displayed outdoors, while another cast edition of Maillol’s
La Rivière (The River), also displayed outdoors, has
retained the expected blue-gray appearance of an oxidized
lead surface. The physical change in surface appearance
suggests a chemical change is occurring on the surface
of some objects outside the expected behavior of lead
oxidation in an outdoor environment. Research on lead
discoloration of outdoor sculptures in Queluz, Portugal,
supported by the World Monuments Fund suggests this
discoloration is a mixture of lead(IV) oxide and a lead
oxycarbonate. The complete corrosion mechanism,
however, was not identified (Bernard et al. 2010).
La Rivière (no. 2, cast in 1948 by Alexis (Eugène) Rudier
Foundry), held at the Museum of Modern Art (MoMA),
New York, has been on display in the outdoor sculp-
ture garden since its acquisition in 1948 and does not
Joy Bloser Lynda Zycherman
The Museum of Modern Art The Museum of Modern Art
New York NY, USA New York NY, USA
joy_bloser@moma.org lynda_zycherman@moma.org
*Author for correspondence
research into the sculptures’ casting history, past treatment,
outdoor weathering patterns, and collections care practices
was compared. XRF and SEM-EDX analysis did not point out
substantial differences between the objects. Their treatment
histories and environmental conditions revealed possible
future paths of research.
Keywords
Maillol, lead, outdoor sculpture, lead dioxide
exhibit any dark discoloration. L’Air (3/6, cast in 1962
by Georges Rudier Foundry), held at the Kröller-Müller
Museum (KMM) in Otterlo, the Netherlands, has also
been on display in the outdoor sculpture garden since
its acquisition in 1962, and a dark red discoloration
appears to follow its water runoff patterns. The casting
histories, past treatments, outdoor weathering patterns,
and ongoing preventive care of the two sculptures were
compared and both submitted to technical analysis by
scanning electron microscopy with energy-dispersive
x-ray spectroscopy (SEM-EDX) and x-ray fluorescence
(XRF). X-ray diffraction (XRD) analysis on samples from
L’Air identified a mixture of two polymorphs of lead(IV)
oxide as the corrosion products present in the discolored
areas (Joosten and van Hoesel 2016).
This investigation studied the corrosion mechanisms of
the discoloration forming on the surface of select Maillol
lead sculptures and considered how their different casting
histories, varying preventive care, and treatment proto-
cols may affect or possibly prevent the discoloration of
lead over time.
A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH 169

Foundry history a natural oxidation layer in lead, but a dark red color
can be seen on most of the surface of L’Air in the form
Alexis Rudier set up his foundry in Paris in 1874 and
of stains and drips (Figure 1). Its thin composition has a
after his death in 1897, his son Eugène ran the foundry.
patina-like effect on several areas of the sculpture. Upon
Maillol formed a close working relationship with Eugène
close examination, white powdery spots can also be seen
Rudier that lasted more than 39 years. After Maillol’s
in the darkened areas (Figure 2).
death in 1944, Eugène continued to work with Maillol’s
estate until his death in 1952. Despite Eugène’s successful L’Air was cared for in the same way as outdoor bronzes
foundry, he inscribed all of his work under his father’s until the early 1990s: with light cleaning with soapy
name, “Alexis Rudier.” water and applications of a beeswax coating once a year.
Treatment documentation does not indicate exactly
After Eugène Rudier’s death, his cousin, Georges Rudier,
what kind of soap, water, beeswax, or tools were used
started a foundry despite Eugène’s wish that the Rudier
or in what way. The period in which these treatments
foundry close upon his death. Georges retained Eugène’s
stopped seems to correspond with the approximate date
workers and began to produce work for the Maillol
on which the dark red stains were first spotted. Since
estate and the Musée Rodin and for other contemporary
the wax coating would have isolated the sculpture from
sculptors of the time. However, both the Maillol Museum
water during most of its life, this further confirms the
and Musée Rodin ceased working with Georges around
impact of water on the object, as do the drip-like stains.
1976 among rumors of suspicious activity at the foundry
At this point, there seem to be no traces of beeswax left
(Lebon 2014, 133–41).
on the sculpture.

History of the sculptures, their condition, and
treatments
Maillol’s L’Air depicts a reclined female nude first real-
ized in stone in 1938 for a monument commissioned by
the city of Toulouse. The second version was cast after
his death in six lead editions and a number of bronze
editions. The lead edition 3/6 was cast in 1962 and is
installed outdoors in the KMM’s gardens.
Condition reports begin to mention the dark corrosion
stains around 1993, but no further descriptions are
provided until 2001. Photographic documentation has
provided some indication of how and when the red stains
started to appear and how quickly they developed. A
major increase was observed between 2001 and 2005
(Mahu 2016).
The corrosion products in Maillol’s L’Air are not only
distracting to the eye, but also pose risks to the sculp-
ture’s structure and integrity if the corrosion is active.
Examples of dark red corrosion on lead sculptures
continue to be found elsewhere, yet other lead sculp-
tures, such as MoMA’s La Rivière, do not yet exhibit
this condition.

Figure 1. Aristide Maillol, L’Air, 3/6, 1939, cast 1962. KMM, 2018

Early documentation of the KMM’s L’Air shows a dark

gray surface polished to an even sheen. As it is installed
in the museum’s outdoor garden, the object is often
touched by visitors. Today, the sculpture is of a dull gray-
blue color overall, which is the expected appearance of Figure 2. Condition detail of L’Air. KMM, 2018
170  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Maillol’s La Rivière was originally commissioned as
a monument to Henri Barbusse in late 1938. Eugène
Rudier cast the work in an edition of two, plus the artist’s
proof, in 1948. Two subsequent editions of six, in bronze
and in lead, were cast in the 1960s by Georges Rudier.
MoMA acquired La Rivière no. 2 from the original set of
casts in early 1949, and subsequently installed the work
outdoors, where it has been on view almost continuously
since 1953 (Figure 3).
Figure 3. Aristide Maillol, La Rivière, no. 2, 1938–43, cast 1948. Installation
view in the garden. Image by Soichi Sunami, MoMA, prior to 1953
Photographic documentation of MoMA’s La Rivière just
after acquisition shows the sculpture with a middle-gray
tone and the surface polished to an even sheen (Figure 4).
While it largely retains this middle-gray tone overall,
dark discoloration where water drips from the sculpture’s
undercut surfaces is becoming more prominent (Figure 5).
Figure 4. Aristide Maillol, La Rivière, MoMA. Image by Thomas Griesel,
MoMA, 2014
A complete treatment history for MoMA’s La Rivière
(697.1949) is not available, but notes in the object’s record
provide a general scope of treatment. The most interven-
CORROSION STUDIES
Figure 5. Condition detail of La Rivière. MoMA, 2018
tive treatment on the sculpture occurred in 1952: it was
washed with a mild soap and water, rinsed with benzine,
and polished with a proprietary metal polish. The sculp-
ture was then installed over the reflecting pool in 1953
and was regularly rinsed with water, occasionally with a
mild detergent, and waxed periodically into the 1970s.
Treatment protocol since the early 1980s has been to
rinse the sculpture twice a year with water, and wash
with a mild detergent and light mechanical action once
every few years. It is not within touching distance of the
public. The sculpture has developed dark black accretions
in between the water run-off patterns on the sculpture.
There is no indication of dark red or purple staining in
these runoff areas.
Overview of lead corrosion
When exposed outdoors, lead usually corrodes until it
develops an adherent film of insoluble lead compounds
that protect the surface from further attack. This behavior
differs from what has been observed on L’Air. The sculp-
ture started to show brown/red stains on the surface that
developed and darkened in the relatively short time span
of approximately twenty years. The corrosion products
that formed on it were identified in 2016 as plattnerite
(β-PbO2) and scrutinyite (α-PbO2) by SEM-EDX and
XRD analysis (Joosten and van Hoesel 2016). These
compounds are polymorphs of lead(IV) oxide (also
known as lead dioxide, PbO2) which have been found
on lead water-supply pipes. The corrosion mechanisms
by which plattnerite and scrutinyite form at the surface
of the sculpture are yet to be fully explained.
Other cases of outdoor lead sculptures showing dark red
stains include that of the Palace of Queluz in Portugal,
A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH
where plattnerite was also found (Bernard, Costa, and
Joiret 2010). The sculptures at Queluz were treated in
2012 and have since developed the dark red corrosion
layer again, indicating that the process is still active,
possibly as a cyclic or continuous mechanism.
Extensive research has been done regarding the behavior
of lead in water-supply systems, some including the
formation of lead dioxides. In recent years, it has been
proposed that lead dioxide forms from the oxidation
of lead carbonate hydrocerussite (Bernard, Costa, and
Joiret 2010); however, opposing research suggests “. . .
PbO2 formation does not require lead(II) carbonates as
precursors or intermediate phases” (Wang et al. 2010).
Other influential factors in the formation of lead dioxide
include the effect of dissolved inorganic carbon (DIC)3,
pH fluctuations, and free chlorine concentration (Wang
et al. 2010).
Environmental factors
Since an oxidative environment must be present for
formation of the red corrosion layer to occur, it is worth
comparing the outdoor climates of the sculptures with
and without staining. Queluz, Otterlo, and New York
City are similar in some respects and quite distinct
in others. New York City has the most extreme high
and low temperatures of the three cities, the highest
humidity levels, regular snowfall, and the highest
average rainfall. Queluz has the warmest climate and
lowest average rainfall of the three cities.1 Air quality in
New York City has been historically poor, but has been
improved to levels roughly equivalent with Otterlo and
Queluz for the past two decades.2 Although specific
particulate matter was not investigated in this paper as
a possible cause of corrosion, it may shed more light
on the differences in corrosion behavior between these
three locations.
Local microclimates were also noted. Neither MoMA
nor Queluz cover their sculptures during the winter
months, leaving them fully exposed to the weather all
year round. The KMM builds an open air shed around
its sculptures to shield the work from November to
April. While it is difficult to attribute lead dioxide
formation to a single environmental scenario, it is
interesting to consider why the dark red stains do not
appear to form on the sculpture left uncovered in New
York City. Water dwell time on the sculptures may be
an important factor.
171
Tests
Visual analysis: The sculptures were observed to under-
stand the patterns in which the stains occur (when
present), how water runoff behaves, and the patterns of
sunlight exposure.
Drip experiment: In order to confirm water runoff
patterns on the sculptures’ surface, drops of deionized
water were released using a pipette and observed as they
flowed, pooled, and evaporated from the surface.
XRF analysis: XRF analysis of the KMM’s L’Air was
performed in 2016 by Arie Pappot (Rijksmuseum
Amsterdam).4 XRF measurements of MoMA’s La Rivière
were taken by Joy Bloser.5 Both collected spectra were
analyzed by Arie Pappot with the same software cali-
brated with lead standards. Quantitative analysis was
not performed on the MoMA’s La Rivière.
Sampling and microscopy: L’Air had seven samples
taken on site according to accessibility and appearance,
embedded in a polyester casting resin, and polished.
Polyester was chosen as an embedding material due to
its minimal shrinkage upon drying, minimizing stress on
the soft metal samples. The cross sections were observed
and photographed under a Leica DM2500 M optical
microscope with ZEN Lite 2.3 software. La Rivière had
three samples taken from non-visible locations exhibiting
drip patterns. Samples were taken at the edge of drip
formations to include both the black and white crust
formation on the lead substrate. Samples were embedded
in a polyester resin and polished to match protocols
used for analysis of the KMM’s L’Air. Cross sections were
imaged using a Leica DM IRM research microscope with
Phase One H20 and Capture One software.
SEM-EDX: This technique allowed the investigation
of the morphology and elemental composition of the
cross sections, and made it possible to assess possible
stratigraphies. L’Air samples were analyzed by Ineke
Joosten (RCE).6 La Rivière embedded cross sections
were analyzed by Joy Bloser (MoMA), courtesy of the
Conservation Center at the Institute of Fine Arts, New
York University.7
Test results
Visual analysis and dripping experiment: Observations of
L’Air confirmed the red stains follow the pattern in which
water falls on the sculpture. However, they increasingly
172  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

develop in the areas where the water runs, not where

the water pools. White spots (identified in 2016 as lead
carbonates) are only visible within the red stains (lead
dioxides) (Joosten and van Hoesel 2016), and not in the
gray surface (lead oxides). It was also observed that, given
its outdoor location, the sculpture receives sunlight for
long periods of time at different angles throughout the
year. The drip experiment also showed that the white
spots concentrate in the areas where the water pools.
For La Rivière, drips of water flowed down the white
channels of the drip striations on the undercut surfaces
of the sculpture. The sculpture receives direct sunlight
primarily only at midday due to its location in the garden
and shadows cast by the surrounding buildings. It does
receive reflected light from the surrounding building
surfaces and reflecting pool below it.
Figure 6. Diagram of microscopic, SEM, and EDX analysis of L’Air
XRF: Qualitative analysis confirmed the two sculptures (sample 2), 2018
are both cast from lead alloys containing lead (Pb), anti-
mony (Sb), and tin (Sn). This is a common casting alloy.
Comparison of percent composition between the two
sculptures was not possible given experimental limita-
tions. As the two objects were cast by different founders
at different times, it is expected their compositions are
not identical. The 2016 measurements of L’Air were
compared with lead standards, and quantitative analysis
concluded the lead alloy composition contained 4–6%
of antimony (Pappot 2016).

Sampling and microscopy: During the sampling process

for L’Air, the dark red layer appeared to be harder and
more brittle than the underlying lead alloy. Microscopic
analysis did not allow discerning the morphologies of the
corrosion products, although differences in color within
the corrosion layers were evident. The red and white
products were randomly mixed in a thin, well-adhered
layer at the surface (Figure 6). Sampling for La Rivière
revealed the white and black drip channels to be slightly
more brittle than the lead substrate. Observed under Figure 7. Diagram of microscopic, SEM, and EDX analysis of La Rivière
magnification, the white and black corrosion products (sample A), 2018
appear as thin, well-adhered crusts (Figure 7).

SEM-EDX: L’Air sample analysis showed that there was
no evident difference in microstructure between the
red (lead dioxides), white (lead carbonates), and dark
gray (lead oxide) areas. However, it was possible to
distinguish the oxide layer from the bare metal. The
elemental analysis confirmed that there was an increase
of carbon and oxygen in the corroded layers, and no trace
elements were present at significant levels. It was possible
to observe antimony-rich areas in the bare metal which
are not linked to the oxide layers. Important to note is
that the sulfur and lead peaks overlap at nearly 2,340 eV8,
making it very difficult to determine whether sulfur is
present. Overall, distinguishing the red and white areas
was extremely difficult, both chemically and visually.
The main difference in morphology was that the red
areas looked more fractured than the white. Whisker-like
A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH 173

formations on the samples were first noticed seven days surface of L’Air. At which speed these processes may
after the polishing date, suggesting that they may have be occurring, or whether one is predominant over the
developed during that time. While a few whiskers are other, is still under investigation.
found on the bare metal, most are highly concentrated
A surprising find from L’Air’s analysis was the obser-
near the surface, at the interface between the metal and
vation that white powdery spots appeared first on the
the corrosion layer (Figure 6). La Rivière sample anal-
fresh metal of the sculpture after the samples were
ysis confirmed XRF findings of an antimony-lead alloy.
taken (Figure 8). Based on the previous documentation
Elemental mapping showed discrete particles of antimony
of L’Air, it was not possible to judge if lead carbonates
surrounded by a lead-rich matrix. No discernible differ-
appeared on the artwork before the red layer formed.
ence between the metal surface and the oxide layer was
Whether the white spots of lead carbonate will degrade
detected using SEM-EDX (Figure 7).
into lead dioxide can only be confirmed in the future.
The current condition of La Rivière provides a good
Discussion
baseline for the future observation of lead carbonate
The drip-like pattern of the oxidation layers confirmed and lead dioxide development on lead-antimony alloys
that both sculptures are reacting to the runoff of precip- displayed outdoors.
itation, one more drastically than the other. In the case
of L’Air, the difference in color between the places where
water runs and those where it sits indicates that lead
reacts differently in flowing water versus stagnant water
conditions. This observation coincides with research
suggesting flow destabilizes the corrosion scales of lead
pipes mechanically or chemically within a few hours (Xie
2010). Mechanical destabilization of the surface due to
flow could also be linked to the formation of whiskers
observed in L’Air’s samples, allowing corrosion to develop
further in a continuous manner by creating cracks and
exposing fresh surfaces. However, whisker formation has
not been observed on the object itself, and it may not be
possible given their scale. Figure 8. Sample #3 location a month after sampling showing white
corrosion (Dino-Lite digital microscope at approx. 50×), 2018
Visual examination and SEM-EDX analysis suggest a
layer of white lead carbonates is present in the areas
XRF analysis did not single out any substantial differ-
of water runoff on the MoMA’s La Rivière, similar in
ences that could explain the formation of lead dioxide.
appearance to L’Air, although no red corrosion layer or
However, additional research into the percent compo-
staining is visible on the sculpture. As La Rivière did
sition of trace elements could provide more detailed
not show clear traces of lead dioxide, XRD analysis
information. Past waxing treatments in both sculptures,
was not deemed essential at this stage of the research.
more recently at the KMM, may have influenced the
Raman analysis of cross-section samples is scheduled
development of lead dioxide in L’Air; nevertheless, this
in order to understand any surface changes better. The
does not explain why that was not the case in La Rivière.
assorted red and white corrosion layers on L’Air suggest
Evaluation of the treatment history of other examples
that the corrosion build-up is not layered, but rather
would be beneficial.
a well-combined mixture of products. Bernard, Costa,
and Joiret (2010, 341–44) proposed the red color may So far, the results seem to circle back to environmental
be an oxidation product of hydrocerussite, while Wang conditions and water chemistry. Freeze-thaw cycles,
et al. (2010, 8950–56) concluded both lead carbonates snow, and winter exposure may stunt the formation
and lead oxides can be a precursor of the formation of of lead dioxide according to the weather patterns
lead dioxide. It can therefore be suggested that more observed in New York. Whether water temperature
than one mechanism is likely to be taking place at the also has a role requires further research. While pH
174  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
is certainly a part of lead dioxide formation in lead
water-supply pipes, pH experiments on the surface of
objects remain problematic in the field of conservation.
Electrochemical research could also shed more light
on the subject.
Conclusion and suggestions for further
research
In summary, the results show that the surfaces of L’Air
and La Rivière are reacting in various ways to oxygen and
water in their environment. XRF and SEM-EDX analysis
did not point out substantial differences between the
objects. Their treatment histories and environmental
conditions suggest that the application of a protective
coating such as wax may prevent the formation of some
corrosion layers by limiting the sculptures’ contact with
water. Further research into the removal of the corrosion
when desired is also suggested, including mechanical,
chemical, and electrochemical methods.
Acknowledgments
The authors would like to thank Arie Pappot (Rijksmu-
seum Amsterdam), Susanne Kensche (KMM), Virginia
Costa (Institut National du Patrimoine, France), Luc
Megens (RCE), René Peschar (University of Amsterdam),
Roger Griffith, Ellen Moody, and Megan Randall (MoMA)
for their advice and support.
Notes
1 “Comparison of the Average Weather in New York
City, Queluz, and Otterlo.” Weather Spark. https://
weatherspark.com/compare/y/23912~32002~52680/
Comparison-of-the-Average-Weather-in-New-York-
City-Queluz-and-Otterlo (accessed 28 October 2018).
2 “New York City trends in air pollution and its health
consequences.” NYC Health, September 26, 2013. https://
www1.nyc.gov/assets/doh/downloads/pdf/environ-
mental/air-quality-report-2013.pdf (accessed 20 October
2018).
3 Dissolved inorganic carbon (DIC): the sum of inor-
ganic carbon species in a solution; these include carbon
dioxide, carbonic acid, and carbonates. https://en.wiki-
pedia.org/wiki/Total_inorganic_carbon (accessed 06
October 2018).
4 XRF analysis was performed on eight spots using a
handheld Olympus Delta X professional with a two-beam
CORROSION STUDIES
measurement protocol (at 40 kV and 10 kV) and treated
with PyMca software.
5 XRF analysis was performed using a handheld Bruker
Tracer III-SD XRF spectrometer at 40 KeV, 1.5 μAmps,
and 75 seconds with P1 XRF software; files were exported
as .csv. Analysis was performed on six locations.
6 Analysis was performed using SEM (NovaNano 450
FEG from FEI) combined with EDX (Thermo Fisher
Scientific UltraDry silicon drift detector), and analyzed
with Pathfinder software.
7 Analysis was carried out using a Hitachi TM3000
scanning electron microscope; electron dispersive spec-
troscopy was performed with a Bruker XFlash MIN SVE
using Quantax 70 software.
8 X-Ray Data Booklet Table 1–3. http://xdb.lbl.gov/
Section1/Table_1-3.pdf .
References
Bernard, M.-C., V. Costa, and S. Joiret. 2010. On unex-
pected colour of lead sculptures in Queluz: Degradation
of lead white. Corrosion Engineering, Science and Tech-
nology 45(5): 341–44.
Joosten, I. and A. van Hoesel. 2016. L’Air, Aristide
Maillol, 1930, Kröller Müller Museum. Onderzoek
naar corrosieproducten. Rijksdienst voor het Cultureel
Erfgoed, Amsterdam.
Lebon, É. 2014. Dictionary of art bronze founders: France
1890–1950: English translation. London: Sladmore
Editions; Perth, Australia: Marjon Editions.
Mahu, I. 2016. KM 127.576 Maillol – L’Air. Conser-
vation report. Kröller-Müller Museum, Otterlo, The
Netherlands.
Pappot, A. 2016. Corrosion phenomenon on L’Air by
Aristide Maillol (KM 127.576). Conservation report.
Rijksmuseum, Amsterdam.
Wang, Y., Y. Xie., W. Li., Z. Wang, and D. Giammar. 2010.
Formation of lead(IV) oxides from lead(II) compounds.
Environmental Science & Technology 44(23): 8950–56.
Xie, Y. 2010. Dissolution, formation, and transformation
of the lead corrosion product PbO2: Rates and mecha-
nisms of reactions that control lead release in drinking
water distribution systems. Ph.D. dissertation, School of
Engineering & Applied Science, Washington University in
St. Louis, USA. http://openscholarship.wustl.edu/etd/38  .
A DARK DISCOLORATION ON ARISTIDE MAILLOL’S OUTDOOR LEAD SCULPTURES: IDENTIFICATION, FORMATION, AND FURTHER RESEARCH 175

Authors
Manuela Toro is a post-master’s metal conserva-
tor-in-training at the University of Amsterdam. Her MA
thesis focuses on the theme of this paper.

Joy Bloser is the David Booth Fellow in Sculpture

Conservation at the Museum of Modern Art, New York
(2018–2020).

Lynda Zycherman is the Conservator of Sculpture at

the Museum of Modern Art, New York.

Ineke Joosten is a senior researcher in conservation

and restoration at the Cultural Heritage Laboratory of
the Cultural Heritage Agency of the Netherlands.

Tonny Beentjes is program leader in metal conservation

and researcher at the University of Amsterdam.
Pilot Study Examining Tarnishing Caused by Nitrile
Glove–Solvent Interaction on Metals

Caitlin Southwick*
University of Amsterdam
Amsterdam, The Netherlands
c.c.southwick@uva.nl
caitlin.southwick@gmail.com
*Author for correspondence
Kate van Lookeren Campagne Joosje van Bennekom Tamar Davidowitz
University of Amsterdam Rijksmuseum Amsterdam Rijksmuseum Amsterdam
Amsterdam, The Netherlands Amsterdam, The Netherlands University of Amsterdam
K.E.vanLookerenCampagne@uva.nl J.van.Bennekom@rijksmuseum.nl Amsterdam, The Netherlands
T.Davidowitz@uva.nl
Ellen van Bork
Rijksmuseum Amsterdam
Amsterdam, The Netherlands
e.van.bork@rijksmuseum.nl

Abstract
Nitrile gloves are used in conservation for protection of people
and objects. While testing recyclable gloves at the Rijksmuseum
Amsterdam, tarnishing was observed on a silver object when
using Kimtech Purple Nitrile gloves with acetone during a
treatment. This observation led to further investigations
regarding the interaction of gloves and metals with solvents.
Previous studies have shown that in accelerated aging conditions
a reaction occurs between gloves and metals and that nitrile
gloves contain sulfur and chloride compounds. This pilot study
further investigates solvents as a factor in tarnishing metals
when used with nitrile gloves and the time frame of this reaction.
A modified Oddy test was performed testing five types of nitrile
gloves on three types of metals in five environments. Progress
was measured at various time intervals and after one week.
Results revealed that within realistic working times reactions
between metals, gloves and solvents are minimal to non-
existent. Reactions occurred on almost all experimental coupons
after one week (excluding control coupons). Accelerated reaction
conditions would be required to make tarnishing observable
within short time periods. Further studies to investigate the
reaction mechanisms and identify the types of tarnishing are
recommended.
Keywords
nitrile gloves, Oddy test, metal conservation, solvents,
tarnishing, sustainability

Introduction
Gloves are a common tool used by conservators in
everyday practice. Different types of gloves are recom-
mended for particular purposes (i.e., various types of
solvents) but the standard in most studios is a disposable
nitrile glove (Washington Conservation Guild 2001, 69).
There are still many questions regarding the proper
use of gloves in conservation, including sustainability.
In order to address the waste produced by disposable
gloves, the University of Amsterdam and the Rijksmu-
seum Amsterdam initiated a trial of the Kimberly-Clark
Professional RightCycle program using Kimtech recy-
clable gloves. Three types of gloves eligible for recycling
were put into various studios at the conservation lab of
the University of Amsterdam and the Rijksmuseum for
a trial period of six months. The program was halted at
the Rijksmuseum when it was suspected that the Kimtech
Purple Nitrile gloves were causing tarnishing when used
in combination with acetone on silver. This concern led
to further investigations to determine if Kimberly-Clark
recyclable gloves are suitable for use in conservation and
to assess if nitrile gloves are safe to use in combination
with solvents on metals.
Background
The reported tarnishing occurred when an intern was
using Kimtech Purple gloves in combination with acetone
applied by cotton swab on a silver object. Within a short
working time, approximately 30 to 60 minutes, tarnishing
was noticed on the silver.
Gloves in conservation
Gloves are used in conservation to protect the object
and to protect the conservator. Salts and acids from
hands cause tarnishing on metals and gloves act
as a barrier to protect objects during handling and
treatment (Ankersmit et al. 2017). Gloves protect the
PILOT STUDY EXAMINING TARNISHING CAUSED BY NITRILE GLOVE–SOLVENT INTERACTION ON METALS
conservator from harmful residues of past treatments
or direct contact with solvents. Nitrile gloves are
usually recommended, as they offer greater protection
from oil transfer than cotton gloves and have better
traction (Conservation Resources 2004). However,
very few studies have been conducted to determine if
these gloves are actually safe for use in conservation
(Hoffman 2009). Nitrile gloves are not recommended
for use with acetone and little is known about the
composition of gloves or their residues. Concerns
about gloves causing tarnishing have been raised but
no full-scale studies have been conducted to determine
the actual reactions or products.
Literature and previous studies
In previous studies, both gloves and glove residues
have shown reactions with metals. Hoffman (2009)
identified sulfur and chloride compounds in nitrile
gloves, which were presumably the source for silver
tarnishing observed during contact tests. A follow-up
study tested accelerator-free nitrile gloves (designed
to have a lower sulfur content) to determine if they
would be a viable alternative. In this study, Imamura
(2016) ran Oddy tests on coupons with glove residues
in accelerated aging conditions and found tarnishing
occurred in three of four types of gloves. As sulfur was
detected in only two, it cannot be assumed that the
sulfur compound alone was responsible for tarnishing.
Other aspects must have contributed to the tarnishing;
possibly other corrosive elements were present, or the
reaction was catalyzed/accelerated by heat and high
relative humidity.
Additional tests on the emission behavior of accelera-
tor-free nitrile gloves using traditional Oddy tests were
conducted by Stanek (2018). Corrosion was observed on
copper and lead but not on silver. These results indicate
the possible presence of another corrosive reactant,
as it would be surprising if no reaction occurred on
silver in the presence of a sulfuric component. It is
also possible that corrosion occurred on the silver, but
the layer was too thin to be observed. Environmental
conditions and catalysts may also affect the initiation
and rate of the reaction.
Studies have not been conducted to investigate the
reactions or products formed. Since little information
is known about this problem, nitrile gloves continue to
be used and alternatives are hardly considered.
177
To improve the understanding of this phenomenon, a
study was designed to investigate solvents and various
types of nitrile gloves as sources for corrosion. An initial
experiment was conducted to determine if a reaction
takes place between nitrile gloves and metal in the
presence of solvents. Results showed that after one week
corrosion does occur, and that none of the gloves tested
performed satisfactorily in combination with solvents for
extended contact (Southwick 2018). This pilot study was
conducted as a follow-up to evaluate if a reaction takes
place within realistic working times.
The aim of the study was to gain further information
about the type of reaction taking place to see if the type
of glove affects corrosion and if solvents influence the
reaction, as well as to explore the time frame of these
reactions. Modified Oddy tests (direct contact) were
designed to evaluate different types of gloves and solvents.
Control coupons were included to see if gloves are safe
to use when handling objects without solvents.
Methods
Five different gloves were tested in contact with three
types of metals and four different solvents. A control
was set up for comparison. The control group consisted
of the same set up as the other desiccators, but without
solvents or water.
Materials tested
Variables were chosen based on discussions with metal
conservators from the Rijksmuseum Amsterdam.
Solvents were selected based on those most commonly
Table 1. Variables tested. For the full list see Materials list
Control
Deionized water
Solvents Acetone
Ethanol
Shellsol T
Copper
Metals Sterling silver (925/1000)
Lead
TouchNTuff Green Gloves
N-Dex Blue Gloves
Gloves Kimberly-Clark Professional Purple
Kimberly-Clark Professional Sterling
Kimberly-Clark Professional Comfort Nitrile
10 minutes
30 minutes
Times 60 minutes
2.5 hours
1 week
178  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

used and times were chosen based on realistic working

times as described by conservators.

Experimental protocol
Coupons of the three types of metals were prepared
according to the Rijksmuseum Amsterdam Oddy test
protocol (Southwick 2018). The copper and silver
coupons were polished with Sianor 1600 B 7/0 polishing
paper, then rinsed with acetone to remove residue and
dried with clean paper towels. The coupons were then
polished three times with chalk. The chalk was prepared
by mixing CaCO3 powder with deionized water to reach
a yoghurt consistency and applied with a cotton swab.
After the third application, the chalk was rubbed off
using a clean paper towel and the edges wiped. The Figure 1. Set up of experiment. Each glove sample was placed onto the
coupons were then cleaned with acetone (applied by respective coupons, according to Table 2
spray bottle) and blotted two more times. A final rinse
was conducted using deionized water. The coupons
were then dried using Kimtech wipes and placed into
a new plastic zip bag.

The lead coupons were polished on all sides with

a stainless-steel rotary tool. They were then rinsed
three times with acetone, dried immediately after each
rinsing with a Kimtech wipe and then placed in a new
plastic zip bag.

Square samples of the gloves to be tested were cut out and

handled exclusively with clean tweezers. Five desiccators
were cleaned with acetone and ethanol and dried. Each
desiccator was designated to one solvent and one was left
as a control (no solvent). Seventy-five ml of each solvent
was poured into four glass containers and then placed
in the bottom of the specified desiccator. Six silver and
copper coupons and five lead coupons were placed on
the shelf of each of the five desiccators (except Shellsol Figure 2. Glass microscope slides were then placed on top of the
samples to secure them in place and ensure contact. The samples, on
T, where only silver was tested). The sample test was run
the desiccator plate, were then placed into the prepared vapor chamber
using Shellsol T to determine if polarity was a factor.

The prepared samples of each of the gloves were placed

in contact on top of each coupon using tweezers
(Figure 1). A glass microscope slide was then placed
on top of the glove to ensure contact (Figure  2). 1
One coupon of each metal was left without a glove
as a control. The shelves were then placed into the
desiccators, which were sealed and left for set amounts
of time.
Each desiccator contained 17 coupons and 14 glove
samples as seen above and described in Table 2.
Evaluation of results
This experiment was conducted based on clear visible
alterations observed during a treatment; therefore, visual
assessment was used for this pilot study and data recorded
photographically. This study aimed to establish if a reac-
tion between glove and solvents in contact with metals
occurs; further testing to determine exact time frames
when tarnishing occurs and to what degree should be
conducted using microscopic analysis.
PILOT STUDY EXAMINING TARNISHING CAUSED BY NITRILE GLOVE–SOLVENT INTERACTION ON METALS 179

Table 2. Layout of coupons with gloves

Solvent groups: Control, ethanol, deionized Solvent groups: Control, ethanol, deionized Solvent groups: Control, ethanol,
water, acetone and Shellsol T water, acetone deionized water, acetone
COUPON GLOVE COUPON GLOVE COUPON GLOVE
Silver N-Dex Blue Copper N-Dex Blue Lead N-Dex Blue
Silver TouchNTuff Copper TouchNTuff Lead TouchNTuff
Silver KC Purple Copper KC Purple Lead KC Purple
Silver KC Sterling Copper KC Sterling Lead KC Sterling
Silver KC Comfort Copper KC Comfort Lead KC Comfort
Silver Control (no glove) Copper Control (no glove) Lead Control (no glove)

Method
Tests were run simultaneously, initiating each group in a
staggered sequence (Table 3). Sequencing was determined
by comparative evaporation rates of the various solvents
to allow for chamber saturation to be reached. Times
for formation of a vapor chamber were only allotted at
the beginning of the experiment. Times were based on
comparative rates of evaporation amongst the various
solvents used.
Table 3. Evaporation times of solvents allotted for formation of an
initial vapor chamber
Time allotted for vapor
Desiccator Solvent
chamber to form
1 Control group - Results
2 Ethanol 2.5 hours
3 Deionized water 4 hours
4 Shellsol T 4.5 hours
5 Acetone 2 hours
Photographs were taken using a Nikon D750 digital
camera with an AF-S Micro NIKKOR 60mm 1:2.8 G ED
lens. Before and after photographs were taken using
two 45-degree-angled LED Video Light boxes at
5600 K.
Samples were left for a total of 2.5 hours and checked at set
intervals: 10 minutes (+10 min), 30 minutes (+20 min),
60 minutes (+30 min) and 2.5 hours (+1.5 h). At each
interval, the lid was removed from the desiccator, the
glass microscope slides and glove samples were taken off
the metal coupons with clean tweezers and the coupons
photographed in situ. The gloves and glass slides were
then replaced onto the coupons (same side down) and
the desiccator resealed. The experiment continued to
run until the next time interval, when the procedure was
repeated. After 2.5 hours, the coupons were removed
from the desiccator and final photographs taken.
Removing the lid from the desiccator released vapor from
the chamber. This may have affected the results of the
experiment by decreasing the solvent concentration for
the next time interval and initiating solvent evaporation
(gradient environment instead of saturated). Further
testing should be conducted with multiple desiccators,
each designated to a specific time interval to allow each
set interval to run for the full length in a saturated envi-
ronment without disruption.
In-situ photography resulted in different lighting condi-
tions, making subtle comparisons in overall color difficult
to assess. However, distinctive, localized discoloration
was able to be recorded.
Initial results show no change to the coupons in each
of the intervals up to and including 60 minutes. After
2.5 hours, four coupons from the acetone group showed
some change.
Control
Visual observations: No visible changes were observed.
Considerations: The lack of visible changes in the control
group indicates that in the absence of solvents and other
variables (such as heat, high relative humidity or sweat)
all gloves tested should not cause tarnishing on metals.
Acetone
Visual observations: Four coupons showed some small
discolorations after 2.5 hours. These discolorations
appear as small, darker spots.
Considerations: This was the only group which showed
any visible alterations, indicating that the reaction is most
severe with acetone. While the exact mechanisms of the
reaction are not explored, it is possible to infer that the
180  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
nitrile gloves are reacting with the acetone to leach the
sulfur compound from the gloves and cause tarnishing.
Ethanol
Visual observations: No visible changes were observed.
Considerations: Results indicate that within realistic
working times, nitrile gloves should not cause visible
discoloration on silver, copper or lead when used with
ethanol.
Deionized water
Visual observations: No visible changes were observed.
Considerations: The results indicate that water or
humidity alone do not cause a corrosion reaction.
However, humidity may act as an accelerator if other
species are present.
Shellsol T
Visual observations: No visible changes were observed.
Considerations: The results showed no changes in the
first stage of the experiment; however, as none of the
other solvents caused any drastic changes, it is too early to
conclude if polarity is a factor in the reactions occurring.
After one week
The experiment was left for a week to determine if
results were consistent with previous tests (Southwick
2018). More extensive tarnishing was visible and patterns
emerged on the coupons where the gloves had been in
direct contact. Results were rated as displayed in Figure 3.
CORROSION STUDIES
Results were ranked from 0 (no change) to 3 (severe
tarnishing). Table 4 shows examples of the various
gradients of tarnishing.
Figure 4. Silver coupons from acetone chamber after one week. From
left to right: Control, KC Comfort, KC Sterling, KC Purple, TouchNTuff,
N-Dex Blue
The most noticeable discoloration took place in the
acetone chamber. Tarnishing was present on most
coupons. Several observations were made:
• Various gloves performed differently in various
solvents and with various metals.
• No one glove performed well in all variables (metals
or solvents).
• No tarnishing was observed in the control group: all
gloves performed equally satisfactorily.
Silver
After one week, there was little discoloration observed
on the silver coupons. Some yellow, grey and dark grey
patterns emerged on the surface in correspondence
with contact of the gloves (Figure 4). The discoloration
was consistent in hue, but varied in tone. These results
should be compared to the corrosion observed at the
Rijksmuseum to evaluate if the reactions are the same.
Copper
The copper coupons showed distinctive, vein-like patterns
on all coupons except the control (Figure 5). The patterns
ranged in color from red to light copper to orange to
Figure 3. The degree of visible discoloration
observed on the various test coupons after
one week of exposure in each environment.
Discoloration was measured using visual
assessment on a scale of 0 to 3: 0 = no visible
difference (no bar is shown); 0.5 = barely visible,
discoloration possibly occurring; 1 = slight
noticeable visual discoloration; 2 = medium
discoloration; 3 = severe discoloration; 3.5
represents the most severe tarnishing observed

PILOT STUDY EXAMINING TARNISHING CAUSED BY NITRILE GLOVE–SOLVENT INTERACTION ON METALS 181

Table 4. Levels of tarnishing. Photographs taken from acetone chamber after one week unless denoted by *

Metal 0 0.5 1 2 3 3.5

Copper N/A

Control TouchNTuff N-Dex Blue KC Sterling

KC Comfort
Silver N/A

Control TouchNTuff KC Purple KC Sterling KC Comfort

Lead N/A N/A

Control – DW* TouchNTuff N-Dex Blue

Control
*Deionized water

brown to dark blue. Some of the corrosion patterns have ations correlated to specific gloves, but more research
strong outlines, whereas others have blurry edges. would be needed to investigate exact reactions with
various solvents. Possible reactions could include leaching
of plasticizers or emission of volatile acids.

Figure 5. Copper coupons from acetone chamber after one week. From
left to right: Control, KC Comfort, KC Sterling, KC Purple, TouchNTuff,
N-Dex Blue

Lead
The lead coupons all showed significant corrosion
(Figure  ). All coupons significantly darkened and black,
gold, silver and opalescence colors appeared on the
surface. The silver areas resemble pitting, which could
indicate a different reaction.

Figure 7. All coupons and gloves from acetone chamber after one week.
Tarnishing is visible on some of the coupons and clear deterioration of
the gloves is present

It is known that acetone is not recommended for use

Figure 6. Lead coupons from acetone chamber after one week. From left
to right: KC Sterling, KC Purple, TouchNTuff, N-Dex Blue
Gloves
The most significant changes occurred to the gloves
themselves in the form of crinkling, shrinking, bubbling,
discoloration and condensation (Figure 7). Certain alter-
with acetone, so it was unsurprising that the most severe
deformation and discoloration observed was seen in the
acetone group (Figure 8).
Aspects such as reactivity with solvents, leaching of
compounds and permeability should be evaluated to
further assess the suitability of nitrile gloves for use in
conservation.
182  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 8. Detail of deterioration of glove samples from the acetone
chamber during testing
Discussion
The absence of tarnishing on the control coupons
indicates that neither solvents nor gloves alone cause
tarnishing. However, earlier testing showed that when
both materials are used in combination a reaction does
occur, but is only visible after extended times and is
not expected to be observed within realistic working
periods.
The tarnishing observed at the Rijksmuseum may not
have been caused by a simple solvent-glove reaction
with silver. Variables such as heat, humidity and sweat
(acids, oils, salts) which were not examined during this
study may have influenced or accelerated the reaction.
While definitive conclusions about tarnishing reactions
cannot be drawn based on this pilot study, speculations
regarding the source and the type of reaction can be
considered. Based on results after one week, variations
in intensity and color of tarnishing indicate that the
type of solvent and type of glove, as well as the metal, all
have an effect on the reaction. Lead coupons tarnished
over the entire surface, possibly indicating simultaneous
reactions of oxidation and sulfur compounds resulting
in multiple corrosion products. Copper and silver
showed patterns of discoloration only in correspond-
ence with direct contact of the gloves.
In most cases, the observed tarnishing could be consistent
with expected reactions with sulfur compounds. As
Hoffman (2009) identified sulfur compounds in nitrile
gloves, and the control coupons showed no corrosion,
CORROSION STUDIES
it is likely that the solvents act as leaching agents. The
gloves appear to deteriorate in a solvent environment,
releasing sulfur compounds which then react with the
metal to cause corrosion. If this is the case, it appears
accelerators are still needed to instigate a noticeable
reaction within a short time frame.
The reaction could also be a result of solvents influ-
encing the permeability of gloves, which could lead
to sweat penetration during use. This could explain
why tarnishing occurs with some conservators and
not others, as it has been established that the effect of
sweat varies from conservator to conservator (Haver-
kamp 2017).
Further research
This data can be used as preliminary information for
understanding where the reaction is stemming from,
what kind of reaction may be taking place and possibly
the rates or time frames of these reactions and estab-
lishing parameters for further investigation. Corrosion
product analysis, such as x-ray diffraction, would be
needed to determine what reactions are occurring and to
better understand the reactive elements involved (sulfur,
oxygen and/or chloride compounds).
Research should be conducted to determine what type
of reaction is occurring, what variables (or combinations
thereof) are responsible for the reactions, the source and
amount of sulfur, the time frame of the reactions and what
kind of corrosion products are present.2 Another inter-
esting aspect was the lack of reaction seen in the Shellsol
T group, which could indicate that solvent polarity may
also be a factor and should be further investigated.
Additional testing which mimics working conditions
would give more information about reactions in realistic
situations. Key factors, such as heat, humidity, acids/
oils from sweat and even glove residues were not exam-
ined here but probably played a significant role in the
corrosion observed at the Rijksmuseum.3 Measuring
temperatures of conservators’ hands during work and salt
presence in sweat and examining how these aspects affect
corrosion reactions, independently and in combination
with solvents and gloves, would be recommended.4
Conclusion
This study has contributed to further understanding
of the issues regarding the use of nitrile gloves in the
PILOT STUDY EXAMINING TARNISHING CAUSED BY NITRILE GLOVE–SOLVENT INTERACTION ON METALS
conservation of metal objects. Several conclusions can
be drawn:
• There is currently no glove that has proven safe for
use on metals and can be recommended for use with
all solvents.
• None of the gloves tested caused tarnishing in direct
contact with metals without solvents (or any other
variables) present.
• More testing should be conducted to further under-
stand the mechanisms behind gloves and corrosion
on metals during conservation treatments.
Even though it is inadvisable to use nitrile gloves with
certain solvents, nitrile is still the standard in conser-
vation laboratories. This practice may be called into
question now that tarnishing has been observed. Further
research is recommended to find safest practices or
alternative options.
Acknowledgements
The authors would like to thank Nicolas Verhulst
( PET R AC ON ) , S ar a C re ange ( R ij k s mus e u m
Amsterdam), Rene Peschar and Mandy Slager (Univer-
sity of Amsterdam), Lisa Imamura (Queen’s University),
Sabine Stanek (Naturwissenschaftlichen Labor des
KHM Wien) and Roy van der Wielen (University of
Amsterdam).
Notes
1 Kimberly-Clark Comfort Nitrile was not tested on
lead. The glass slides used were too light to press the
glove flat onto the metal coupon. For future studies, a
heavier weight should be used to provide better contact.
2 Multiple corrosion products may be forming. Corro-
sion products formed through a gloves-solvents-metal
reaction accelerated by heat differ from products resulting
from sweat. Permeability tests of various types of gloves
should also be conducted to evaluate penetration of sweat
during use.
3 Future studies should include data-loggers in desic-
cators to measure temperature and RH to evaluate this
variable further.
4 Time of year and the individual are other variables
to consider with sweat (Haverkamp 2017). Individuals’
sweat differs and therefore some conservators are more
at risk to tarnish objects depending on the composition
and amount of sweat their hands produce.
183
Materials list
Kimberly-Clark Kimtech™ Comfort Nitrile Gloves
https://nl.vwr.com/store/product/14645408/
nitril-handschoenen-kimtech-science-comfort-nitrile
Kimberly-Clark Kimtech™ Purple Nitrile Gloves
https://nl.vwr.com/store/product/2174224/nitril-hand-
schoenen-kimtech-science-purple-nitrile
Kimberly-Clark Kimtech™ Sterling Nitrile Gloves
https://nl.vwr.com/store/product/2322089/nitril-hand-
schoenen-kimtech-science-sterling-nitrile
N-Dex™ Blue Nitrile Gloves 7005PFL
https://www.fishersci.com/shop/products/best-n-dex-
original-n-dex-plus-nitrile-gloves-8/19050269a
TouchNTuff® Disposable Nitrile Gloves 92-600
https://nl.vwr.com/store/product/559448/nitril-hand-
schoenen-touchntuff-92-600-92-605
References
Ankersmit, B., M. Griesser-Stermscheg, L. Selwyn, and
S. Sutherland. 2017. Basic care – recognizing metals
and their corrosion products. Canadian Conservation
Institute. https://www.canada.ca/en/conservation-in-
stitute/services/care-objects/metals/basic-care-recog-
nizing-metals-corrosion-products.html#a2b (accessed
12 April 2019).
“Corrosion.” CAMEO (Conservation and Art Materials
Encyclopedia Online). Museum of Fine Arts, Boston.
Last edited 29 April 2016. http://cameo.mfa.org/wiki/
Corrosion .
“Corrosion.” Encyclopædia Britannica. Published 7
February 2019. https://www.britannica.com/science/
corrosion .
Haverkamp, N. 2017. Handling of glass objects. Research
project, University of Amsterdam, The Netherlands.
Hoffman, W. 2009. Silver tarnishing properties of gloves
used in conservation. Research project, Queen’s Univer-
sity, Canada. https://www.queensu.ca/art/sites/webpub-
lish.queensu.ca.artwww/files/files/WilliamHoffman.pdf .
Imamura, L. 2016. Excerpts from investigation of nitrile
gloves for the handling of silver objects. Research project,
Queen’s University, Canada. https://www.queensu.ca/
art/sites/webpublish.queensu.ca.artwww/files/images/
Slide1(4).jpg .
184  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Logan, J. 2007. Recognizing active corrosion. CCI Notes

9(1): 1–3. Ottawa: Canadian Conservation Institute.
“Nitrile Gloves.” Conservation Resources. http://www.
conservationresources.com.au/html/home/products/
product.php?pid=284 (accessed 12 April 2019).
Podany, J.C., J.H. Larson, A.L. Rosenthal, N.S. Brom-
melle, D.W. Insall, and F. Zuccari. 2018. Art conservation
and restoration. Encyclopedia Britannica. https://www.
britannica.com/art/art-conservation-and-restoration/
Prints-and-drawings-on-paper#ref755128 .
Selwyn, L. 2004. Metals and corrosion: A handbook for
the conservation professional. Ottawa: Canadian Conser-
vation Institute.
Southwick, C. 2018. Glove experiment report. Research
project, University of Amsterdam, The Netherlands.
Stanek, S. 2018. Investigation Report Oddy Material Tests
83/17-85/17. Investigation Report, Naturwissenschaftli-
chen Labor des KHM, Vienna, Austria.
“Tarnish.” CAMEO (Conservation and Art Materials
Encyclopedia Online). Museum of Fine Arts, Boston. Last
edited 10 May 2016. http://cameo.mfa.org/wiki/Tarnish .
Washington Conservation Guild. 2001. Conservation
resources for art and antiques. Washington, DC: Wash-
ington Conservation Guild.

Author
Caitlin Southwick is a stone conservator at the Univer-
sity of Amsterdam with experience working in Italy, the
United States and the Netherlands. She has a particular
interest in sustainability and is a professional member of
the AIC Sustainability Committee, the Secretary for the
ICOM Working Group on Sustainability and the founder
of Sustainability in Conservation (SiC), an international
organization aimed at building awareness and promoting
sustainable practices in cultural heritage.
Effect of Reburial Conditions on the Corrosion of Marine
Iron Artifacts
Akinobu Yanagida* Kazutaka Matsuda
Nara National Research Institute for Nara National Research Institute for
Cultural Properties Cultural Properties
Nara city, Japan Nara city, Japan
yanagida@nabunken.go.jp matsuda@nabunken.go.jp
Yoshifumi Ikeda Soichiro Wakiya
University of the Ryukyus Nara National Research Institute for
Okinawa, Japan Cultural Properties
y-ikeda@ll.u-ryukyu.ac.jp Nara city, Japan
s-wakiya@nabunken.go.jp
Abstract
To evaluate the corrosion mechanism of iron artifacts under
reburial conditions at underwater sites, corrosion tests were
conducted under different reburial conditions using carbon
steels and carbon steels with wood. Results of the corrosion
tests suggest that it is important to suppress mass transfer
of ions such as Fe2+ and OH− or H2O, in addition to dissolved
oxygen (DO) depletion, to controll the corrosion of iron
artifacts. Although carbon steel with wood corroded intensively
via microbiologically influenced corrosion involving sulfate-
reducing bacteria, there tended to be less corrosion in sediments
Introduction
The UNESCO Convention of 2001 (UNESCO 2001)
recommends that archaeological objects excavated at
sites located in coastal zones are preserved in situ. As it is
difficult from a practical perspective to perform conser-
vation treatment on all objects, it is necessary to develop
effective reburial methods. In general, from a theoretical
viewpoint, it is essential to maintain dissolved oxygen
(DO)-deficient conditions to preserve iron artifacts in
underwater sites. However, considering the efficiency
of re-excavation and reburial operations, it is essential
to investigate the effectiveness not only of sediments or
sandy soil layers but also of sandbags and DO-imperme-
able sheets as reburial materials. Furthermore, because
it is well known that the corrosion of iron artifacts
with wood is accelerated because of microbiologically
influenced corrosion (MIC) caused by sulfate-reducing
bacteria (SRB) under a reducing atmosphere (Fors and
Sandström 2005), it is important to investigate how MIC
of iron artifacts under different reburial conditions can
be controlled.
Yohsei Kohdzuma
Nara National Research Institute for
Cultural Properties
Nara city, Japan
y-kouzuma@nabunken.go.jp
*Author for correspondence
or sandy soil layers as compared to that under reburial using
sandbags with DO-impermeable sheets. The results of this study
indicate that covering iron artifacts with porous media like
sandy soil is an effective means of ensuring their preservation
in underwater sites.
Keywords
iron artifact, corrosion, reburial, in-situ preservation,
microbiologically influenced corrosion, underwater site
Therefore, to analyze the corrosion mechanism of iron
artifacts under different reburial conditions and propose
an effective reburial method, corrosion tests under sedi-
ments and various reburial conditions were conducted
by using carbon steels and carbon steels with wood at
the Takashima underwater site.
Description of the site
Studies were conducted at the Takashima underwater
site, Matsuura city, Japan (Figure 1). Takashima island
is located in Imari Bay. The environment in the bay
is eutrophicated by scattered chum from fish farming
(Suzuki et al. 2010).
Many archaeological artifacts dating back to the Moko
shurai,1 the 13th-century Mongol invasion of Japan, were
excavated from the Takashima underwater site; these
artifacts are mainly comprised of wooden objects, metal
artifacts, and pottery. Owing to interdisciplinary research
that combined underwater archaeological techniques
186  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
with geophysical exploration in recent years, two ship-
wrecks were discovered in 2012 and 2015 (Ikeda 2016).
After investigation of the shipwrecks, attempts are being
made to preserve them in situ.
Figure 1. Location of Takashima underwater site and investigation area
Experimental methods
Investigation area
The experiment was carried out on the south side of
Takashima island at a depth of approximately 13.5 m
(Figure 1). According to a study by Kagohara and Takino
(2016), sediments do not show distinct vertical changes
in particle size and consist of relatively coarse particles,
comprising silt and approximately 30 wt% of mud.
Environmental investigations
The temperature and DO at the bottom of the sea, the
vertical change in pH, oxidation-reduction potential
(ORP), and the porosity of the sediments were measured.
The temperature and DO meters were set at 20 cm from
the seabed and measured at 15 min intervals. Sediment
cores (10 cmφx, 100 cm long) were collected from an
undisturbed area on May 15, 2016, and the pH and ORP
were measured using a soil pH/ORP meter. A platinum
electrode and a saturated Ag/AgCl electrode serving as
the reference electrode were used to measure the ORP.
Corrosion test
Samples
Samples of carbon steel and carbon steel with wood
(SW) were placed at the site. Carbon steel (SS400 –
C: 0.1%, Si: 0.21, Mn: 0.58%, P: 0.02%, and S: 0.02%)
(JIS G 3101:2015, Japanese Standards Association 2015a)
was cut to a size of 30L × 30W × 3T mm, polished (final
CORROSION STUDIES
polishing: #800 emery paper), washed in distilled water,
and degreased in acetone using an ultrasonic cleaning
machine. In the SW samples, pine wood was used and
cut to a size of 30L × 30W × 3T mm, and placed in contact
with carbon steel on one side. Five samples of each were
used under the same conditions.
Experimental systems
The conditions and schematics of each experimental
system are presented in Table 1 and Figure 2, respectively.
Experiments were conducted twice in different periods;
the experimental systems are referred to as experimental
system 1 (E.S.1) and experimental system 2 (E.S.2).
Test samples in E.S.1 were mounted on the outside of a
polyvinyl chloride pipe (φ125 mm) at different depths
of 1.5, 20, 60, and 100 cm and placed under sediments.
In addition to the above depths, samples in E.S.1-1 were
placed in seawater 5 cm above the sediment surface. Each
sample unit was set under sediments (E.S.1-1) to simulate
pre-excavation conditions and placed in a piled sandbag
100 cm in thickness (E.S.1-2). Sandbags containing sandy
soil were used in these experiments.
In E.S.2, test samples were mounted on a polyvinyl
chloride plate at different depths of 5, 15, 25, 35, 45, and
55 cm; these sample units were placed in a polyvinyl
chloride pipe (φ612 mm) with its bottom end covered
by a DO-impermeable sheet (BALLITEC P-3).2 E.S.2
comprised three experimental systems: E.S.2-1 was
packed with a pile of sandbags by placing DO-imper-
meable sheets between sandbags; E.S.2-2 was filled using
a sandy soil layer 60 cm in thickness, and E.S.2-3 was
packed similar to E.S.2-1 up to a 40 cm depth; sandy
soil packing was done at a 40–60 cm depth in E.S.2-3.
DO meters were set in the gap between the sandbags of
E.S.1-2 and within the polyvinyl chloride pipe in E.S.2-1
at 10, 30, and 40 cm depths (Figure 2).
Analytical methods
After observation with an optical microscope, the local
structures of the corrosion products formed on the
surface of samples were identified using micro x-ray
diffraction. Furthermore, cross sections of some samples
were observed to analyze the fine structure. Samples
were mounted in an epoxy resin, cut, and polished (final
polishing: polycrystalline diamond paste 6 µm) under
ethanol. First, the morphology was observed via optical
microscopy and scanning electron microscopy (SEM).
EFFECT OF REBURIAL CONDITIONS ON THE CORROSION OF MARINE IRON ARTIFACTS
Table 1. Experimental period, burial depth of samples, and experimental conditions
E.S. Experiment period Depth (cm)
1-1 26-Jun-15~14-May-16
Seawater, 1.5, 20, 60, 100
1-2 (for 323 days)
2-1
10-Dec-16~9-Dec-17
2-2 5, 15, 25, 35, 45, 55
(for 364 days)
2-3
—: not measured
The composition and elemental distribution of the cross
sections were analyzed using an energy-dispersive spec-
trometer coupled with SEM.
The corrosion rate of the samples was calculated
according to weight loss during the experimental
period. To remove the corrosion products, samples
were soaked in 6 mol/L hydrochloric acid including
hexamethylenetetramine (0.35 wt%) for 10 min (JIS
Z 2371:2015, Japanese Standards Association 2015b).
The weight loss of the samples was determined after
brushing their surfaces and removing the corrosion
products completely.
Figure 2. Schematic of the corrosion test setup
Results and discussion
Temperature and DO of seawater and pH, ORP,
and porosity of sediments
As indicated in Figure 3, the temperature of the bottom
seawater decreased to approximately 12°C in the winter
and increased to approximately 27°C in the summer.
In contrast, DO showed the opposite behavior; in the
bottom seawater, it remained in saturation state during
the winter and decreased to approximately 40% during
the summer. Because the amount of DO remained at
4 mg/L even in summer, the corrosion of iron artifacts
exposed to seawater was presumed to be caused by the
reduction of DO in the Takashima underwater site.
187
Environment
Experimental condition
measurement
In sediment (simulated before excavation) pH, ORP, porosity
Reburial using sandbags DO
Reburial using sandbags and DO-impermeable sheets DO
Reburial using sandy soil layer —
Reburial similar to E.S.2-2 and sandy soil layer at 40~60 cm —
The pH values of the sediments corresponded to that
of a weak base (7.3–8.5), and the ORP of the sediments
was between +113 mV and +289 mV (vs the standard
hydrogen electrode). Clear vertical changes in the pH
and ORP were not observed. Based on the ORP values
of the sediments (Jørgensen 2006), the DO in the pore
water was presumed to be almost consumed. The porosity
of the sediments was 49.5% at the surface and gradually
decreased with increasing depth to 46.1% at 60 cm.
Figure 3. Changes in the DO of bottom seawater and between sandbags
Corrosion of carbon steel in each E.S.
Corrosion products detected in the samples placed in
E.S.1 and E.S.2 are listed in Tables 1 and 2, respectively.
Figures 4 and 5 show the state of samples removed from
E.S.1 and E.S.2, respectively. Figure 6 shows the vertical
change in the corrosion rate.
In E.S.1-1 (in sediments), the corrosion rate reached
approximately 0.14 mm/y in seawater and dramatically
decreased depending on the depth of immersion in the
sediment, reaching 0.0039 mm/y at 100 cm (Figure 6a-1).
It is suggested that the samples showed high corrosion
rates equivalent to steady state rates in seawater because
the corrosion rate is limited by the diffusion of DO
(Winston Revie and Uhlig 2008). Considering both
the ORP of the sediments and the DO in the seawater,
and given that the corrosion rate was relatively high at
188  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

the surface of the sediments owing to the DO, it can be
presumed that the corrosion rate decreased owing to the
depletion of the DO at increased depth.
In contrast, the corrosion rate in E.S.1-2 (reburial using
sandbags) was much higher than that in E.S.1-1 and
reached 0.079 mm/y at a 60 cm depth with some scatter
(Figure 6a-1). Although samples placed at a 100 cm
depth in E.S.1-2 showed relatively low corrosion rates, it
appears that these results were influenced by the exper-
imental conditions. Samples placed at the lower end of
the polyvinyl chloride pipe were in contact under the
sediments during the experiment. The DO between the
sandbags changed from approximately 0 mg/L to that of
the bottom seawater when the position of the DO meter
was changed (Figure 3). The DO level is considered to
have been influenced by the contact between the DO
meter and the sandbags.3 These results indicate that the
reburial of iron artifacts using sandbags causes consid-
erable corrosion owing to the influx of the DO present
between the sandbags.
Based on the results of E.S.1-2, in E.S.2-1 sandbags and
DO-impermeable sheets were used to create DO-defi-

Table 2. Corrosion products detected in E.S.1

Experimental system
Depth
1-1 1-2
(cm)
Carbon steel Carbon steel with wood Carbon steel Carbon steel with wood
Seawater Magnetite
Magnetite
(+5 cm) Green rust 1 (CO32−)
Magnetite Magnetite
1.5 Mackinawite Magnetite
Akaganeite Green rust 1 (CO32−)
Magnetite Magnetite
20 — Mackinawite
Lepidocrocite Siderite
Magnetite Magnetite
60 Goethite Magnetite
Akaganeite Akaganeite
Mackinawite
100 — * Magnetite
Green Rust1(CO32−)
—: not detected, *: could not take up

Table 3. Corrosion products detected in E.S.2

Experimental system
Depth 2-1 2-2 2-3
(cm) Carbon steel Carbon steel Carbon steel
Carbon steel Carbon steel Carbon steel
with wood with wood with wood
Magnetite Magnetite
Magnetite
5 Green rust 1 (Cl−) Green rust 1 (CO32−) — Akaganeite Green rust 1 (Cl−)
Green rust 1 (Cl−)
Green rust II Green rust 2 (OH−)

Green rust 1 (CO32−) Mackinawite Magnetite Mackinawite

15 Green rust 1 (CO32−) —
β-Fe(OH)3Cl Fougerite Green rust Green rust
Mackinawite
Magnetite Mackinawite Magnetite Fougerite
25 — Green rust (Cl−)
Green rust Green rust 1 (CO32−) Akaganeite β-Fe2(OH)3Cl
Green rust 1 (CO32−)
Mackinawite
35 — Green rust 1 (CO32−) Akaganeite Mackinawite β-Fe(OH)3Cl —
β-Fe2(OH)3Cl
Green rust
Mackinawite Magnetite Mackinawite
45 Green rust 1 (CO32−) Green rust (Cl−) Green rust 1 (CO32−)
Green rust 1 (CO32−) Green rust 1 (CO32−) Magnetite
Fougerite
Magnetite
Mackinawite Green rust 1 (CO32−) Mackinawite
55 β-Fe2(OH)3Cl) Green rust 1 (CO32-) Green rust 1 (Cl-)
Green rust Green rust 2 (OH−) Magnetite
β-Fe2(OH)3Cl
—: not detected
EFFECT OF REBURIAL CONDITIONS ON THE CORROSION OF MARINE IRON ARTIFACTS
cient conditions. Figure 7 shows the behavior of DO in
E.S.2-1: DO at a 10 cm depth increased intermittently
in the winter and decreased during the summer, and at
30–50 cm was lower throughout the year. This behavior
indicates that DO-impermeable sheets are effective in
suppressing the influx of DO. The corrosion rate in
E.S.2-1 was lower than that in E.S.1-2, which indicates
the effect of DO depletion.
In E.S. 2-2 (reburial with a sandy soil layer), the corrosion
rate decreased slightly at a depth of 5 to 15 cm, reaching
0.013–0.011 mm/y below 15 cm. The corrosion rate in
E.S.2-2 was found to be one-fourth of that in E.S.2-1,
indicating that reburial using a sandy soil layer is more
effective than reburial using sandbags and DO-imperme-
able sheets. The corrosion behavior observed in E.S.2-3,
wherein the corrosion rate dramatically decreased at a
depth of 35 to 45 cm (reburial using a sandy soil layer
rather than sandbags), supports this conclusion.
Figure 4. Samples in E.S.1
Figure 5. Samples in E.S.2
189
These results imply that although reburial using sandbags
and DO-impermeable sheets can create DO-deficient
conditions and decrease the corrosion rate compared
to just using sandbags, the effectiveness of corrosion
suppression was lower than in reburial using a sandy
soil layer or in sediments.
Figure 6. Changes in the corrosion rates in each experimental system.
a-1: Carbon steel in E.S.1; a-2: carbon steel with wood in E.S.1; b-1: carbon
steel in E.S.2; and b-2: carbon steel with wood in E.S.2. Small plots show
each measured value and large plots show the mean value
Figure 7. Changes in DO in E.S.2-1
Corrosion of carbon steel with wood (SW)
In the case of SW, in E.S.1-1 (in sediments), mackinawite
(FeS) was formed at a depth below 20 cm from the sedi-
ment surface, which indicates that samples were corroded
by SRB growing under anaerobic conditions (Table 2).
Wood is a food source for microorganisms and thus,
on SW, the microbial activity increased and changed
the atmosphere around the samples to a reducing one.
In comparison with carbon steel samples in E.S.1-1,
although the corrosion rate of SW was lower in seawater
and at a 2.5 cm depth in sediments, the corrosion rate
was higher than that of carbon steel samples at 20 cm
below the sediment surface. Owing to the formation of
a reducing atmosphere and the availability of wood as
190  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
a food source, corrosion was promoted by SRB below a
20 cm depth in sediments.
In E.S.1-2 (reburial using sandbags), magnetite was
mainly formed and iron sulfide could not be detected
at any depth,5 except at 100 cm. The corrosion rate of
SW was comparable to that of carbon steel samples in
E.S.1-2 (Figure 4a-2). These results indicate that MIC
could not occur in E.S.1-2 because of the abundant DO
between sandbags.
In contrast, in E.S.2-1 (reburial using sandbags and
DO-impermeable sheets), mackinawite was detected,
and the corrosion rate of SW increased to approximately
more than twice that of carbon steel. Because DO-de-
ficient conditions were maintained (Figure 7) in the
polyvinyl chloride pipe and iron sulfides were formed on
the samples, it is presumed that MIC by SRB occurred
actively under the reduction conditions in E.S.2-1.
Detection of mackinawite below a10 cm depth suggests
that MIC by SRB also occurred in E.S.2-2 (reburial
with sandy soil layer). Figure 8 shows the SEM and
elemental distribution images of the cross section of
the SW samples placed in E.S.2-1 and E.S.2-2. Both the
SEM and elemental distribution images agree in terms
of the distribution of Fe and S in the corrosion layer;
this indicates that corrosion by SRB and the thinning of
carbon steel in E.S.2-1 were more intense than in E.S.2-2.
Despite the maintenance of reduction conditions during
the experiments in E.S.2-1-3, the corrosion rate in E.S.2-2
was lower than that in E.S.2-1 and E.S.2-3 even in the
case of SW samples, except for samples placed in the
sandy soil layer in E.S.2-3.
Figure 8. SEM and elemental distribution images of the cross section of
carbon steel with wood: (a) sample placed at a 55 cm depth in E.S.2-1;
(b) sample placed at a 45 cm depth in E.S.2-2
Environmental conditions for preserving iron
artifacts in underwater burial sites
Table 4 summarizes the results of the corrosion tests.
As observed, corrosion suppression was not as effective
under reburial using only sandbags because corrosion
CORROSION STUDIES
occurred owing to the DO present between the sandbags.
When using sandbags and DO-impermeable sheets, the
corrosion rate was still higher than that of samples placed
in sediments or under a sandy soil layer.
Figure 9 shows the SEM and elemental distribution
images of the cross sections of carbon steel samples
placed in E.S.1-1 (in sediments) and E.S.2-1 (sandbags
and DO-impermeable sheets). A dense corrosion layer
was observed for samples placed in sediments (E.S.1-1).
In contrast to E.S.2-1, a rough corrosion layer composed
of iron oxide was observed. The relationship between
the porosity of sediments and the corrosion rate is
shown in Figure 10. Figure 10 indicates that the corro-
sion rate decreases with the decreasing porosity of the
sediments, suggesting that the diffusion of the chemical
species causing corrosion depends on the porosity of
the sediments and this controls the corrosion rate of
iron artifacts. In theory, the cathode reaction changes
from the reduction in DO to the reduction in H2O when
DO in aqueous environments decreases and becomes
depleted. Therefore, carbon steel continues to corrode
as the thermodynamic behavior and corrosion rate are
controlled by the diffusion of chemical species causing
corrosion. In the case of E.S.1-1 and E.S.2-2, samples were
in contact with porous media like sediments or sandy
soil; the corrosion rate decreased with a decrease in the
porosity of the porous media because mass transfer via
diffusion depends on the porosity of porous media. In
contrast to E.S.1-1 and E.S.2-2, Fe2+ eluted from carbon
steel moved rapidly between the sandbags in E.S.2-1.
Therefore, a dense corrosion layer could not form, thus
promoting the corrosion of carbon steel in comparison
with E.S.1-1.
As observed in Table 4, the corrosion of carbon steel was
promoted in a reducing atmosphere with wood because
of the effect of SRB. Because the corrosion rate in E.S.2-1
was much higher than that in E.S.2-2, it is presumed that
corrosion influenced by SRB was also suppressed owing
to contact with the porous media. Considering that SRB
are heterotrophic (Hamilton 1985), it is possible that the
corrosion rate decreased in E.S.1-1 and E.S.2-1 owing to
the low diffusion of chemical species like SO42− utilized
for SRB metabolism or as nutrient salts in porous media.
These results imply that it is important to preserve iron
artifacts by maintaining an environment in which there
is slow transfer of species causing corrosion reactions
(e.g. Fe2+, OH−, H2O, etc.) and depleted DO. Using close
EFFECT OF REBURIAL CONDITIONS ON THE CORROSION OF MARINE IRON ARTIFACTS
Table 4. Summary of the corrosion tests results: environment vs corrosion development under different reburial conditions
Environment
Reburial method
DO
Sediment (E.S.1-1), sandy soil (E.S.2-1) Depletion 5
Sandbags (E.S.1-2) Contain
Sandbags + DO-impermeable sheet (E.S.2-1) Depletion
 corrosion is very slow;  corrosion is slow;  corrosion progresses rapidly
porous media with low porosity is therefore effective for
the preservation of iron artifacts under reburial condi-
tions. From a practical viewpoint, covering the iron arti-
facts with porous media like sandy soil and suppressing
the influx of DO using sandbags and DO-impermeable
sheets is suggested, as in E.S.2-3.
Figure 9. SEM and elemental distribution images of the cross section of
carbon steel: (a) sample placed at a 20 cm depth in E.S.1-1; (b) sample
placed at a 55 cm depth in E.S.2-1
Figure 10. Relationship between the porosity of sediments and corrosion
rate of carbon steel
Conclusion
In the present study, corrosion tests were conducted to
evaluate the corrosion mechanism of iron artifacts under
different reburial conditions and to propose an effective
reburial method. The following are the main conclusions
drawn from this study: (1) For preserving iron artifacts,
it is necessary to maintain an environment in which
there is slow mass transfer of species causing corrosion
191
Preservation of iron artifacts
Mass transfer Carbon steel Carbon steel + wood
Low  
High  
High  
reactions and depleted DO. (2) From an MIC viewpoint
as well, it is necessary to create conditions that suppress
mass transfer. (3) Practically, it is difficult to sufficiently
suppress corrosion by using sandbags and DO-imperme-
able sheets but corrosion can be effectively controlled by
using porous media like sediments or a sandy soil layer.
Acknowledgments
We would like to thank the board of education of
Matsuura city and the divers from the Nagasaki office
of KUNITOMI Co., Ltd. for making this work possible.
We are also grateful for the support provided by JSPS
KAKENHI grant no. JP 18H05220 and JP18K12570.
Notes
1 Two shipwrecks discovered at the test site date back
to the invasion in 1281.
2 The DO-impermeable sheet consists of seven layers
–VMPET (vacuum metalized polyethylene terephtha-
late), PE, aluminum foil, PE, PE cloth, PE, LLDPE (linear
low-density polyethylene) –and was sold as a gas barrier
sheet.
3 There is a possibility that DO penetrated through a gap
between the DO meter and sandbags when the DO meter
was replaced. Laboratory experiments will be conducted
to confirm the amount of DO between sandbags in the
future.
4 Because mackinawite was formed only on the wood
side, the effect of wood is presumed to be topical.
5 Although mackinawite was detected in the sample
placed at 100 cm in E.S.1-2, the formation of mackinawite
is attributed to contact with under sediments.
6 The environmental measurements in the sandy soil
layer were not conducted; however, according to Glud’s
research (Glud 2008), DO is present up to depths of
only a few centimeters at most in sediments. Laboratory
experiments will be conducted to confirm the amount
of DO in sandy soil layers in the future.
192  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
References
Fors, Y. and M. Sandström. 2005. Sulfur and iron in ship-
wrecks cause conservation concerns. Chemical Society
Reviews 35(5): 399–415.
Glud, R.N. 2008. Oxygen dynamics of marine sediment.
Marine Ecology Research 4: 243–89.
Hamilton, W.A. 1985. Sulphate-reducing bacteria and anaer-
obic corrosion. Annual Review of Microbiology 39: 195–217.
Ikeda, Y. 2016. The research and study of the shipwrecks
at the Mongol invasions using the underwater archae-
ological method. Report of Grant-in-Aid Scientific
Research. www.jsps.go.jp/j-grantsinaid/12_kiban/ichi-
ran_23/e-data/e28_ikeda.pdf .
Japanese Standards Association. 2015a. Rolled steels for
general structure (JIS G 3101:2015), Tokyo, Japan.
Japanese Standards Association. 2015b. Methods of salt
spray testing (JIS Z 2371:2015), Tokyo, Japan.
Jørgensen, B.B. 2006. Bacteria and marine biogeochem-
istry. In Marine geochemistry, 2nd ed., eds. H.D. Schulz
and M. Zble, 173–207. Berlin: Springer.
Kagohara, K. and Y. Takino. 2016. Characteristics and
deposition process of marine sediments covering the
Mongolian sunken ships. In Proceedings of the General
Meeting of the Association of Japanese Geographers. Japan
(in Japanese).
Suzuki, S., S. Tada, T. Taira, M. Sakaguchi, K. Matsu-
moto, and W. Nishida. 2010. Study of the characteris-
tics of flow structure and water quality in IMARI Bay
by means of field observation and numerical model.
Journal of JSCE B2 Ser. Coastal Engineering 66(1):
1021–25 (in Japanese).
UN Educational, Scientific and Cultural Organisation
(UNESCO). 2001. Convention on the protection of the under-
water cultural heritage, Paris, 2 November 2001. https://
unesdoc.unesco.org/ark:/48223/pf0000126065.locale=en .
Winston Revie, R. and H.H. Uhlig. 2008. Corrosion and
corrosion control. An introduction to corrosion science
and engineering, 4th ed., 9–19. New Jersey: John Wiley
& Sons, Inc.
Yanagida, A, Y. Ikeda, S. Wakiya, and Y. Kohdzuma. 2017.
Effects of seasonal fluctuation and sediment condition
on corrosion of iron artifacts at Takashima-kozaki
Archeological site. Archaeology and Natural Science 74:
45–55 (in Japanese).
CORROSION STUDIES
Materials list
BALLITEC P-3
Kosumo Chemical Industry Co., Inc
Koshigaya, Saitama prefecture, Japan
http://kosumokasei.co.jp/
Authors
Akinobu Yanagida is a researcher at the Nara National
Research Institute for Cultural Properties. He has a PhD
in the corrosion mechanism of metal artifacts under
burial conditions from Kyoto University. He specializes
in the conservation of archaeological metal artifacts and
is currently studying the corrosion mechanism under
burial and storage conditions.
Yoshifumi Ikeda is a professor at the University of the
Ryukyus. He has a master’s degree in archaeology from
Kokugakuin University, a field in which he specializes. He
is the project leader for the investigation of the Takashima
underwater site by JSPS KAKENHI.
Kazutaka Matsuda is a researcher at the Nara National
Research Institute for Cultural Properties. He has a
master’s degree in conservation science and specializes
in the conservation of waterlogged wooden objects. He
is currently studying temporary storage of waterlogged
wood excavated from archaeological sites.
Soichiro Wakiya is a senior researcher at the Nara
National Research Institute for Cultural Properties and
has a PhD in preservation of soil structure remains by
environmental control from Kyoto University. His study
interests include the preservation of soil structure remains
from the viewpoint of soil physics and the thermal envi-
ronment.
Yohsei Kohdzuma is the director of the Center for
Archaeological Operations at the Nara National Research
Institute for Cultural Properties. He has a PhD in conser-
vation of waterlogged wooden objects, a field in which
he specializes. He is also a guest professor at Kyoto
University.
 Young Conservation Professional

Bluing Steel: An Investigation into the Thermal and

Chemical Methods of Colouring Mild Steel

Júlia Tauber
Landesamt für Denkmalpflege
Esslingen am Neckar, Baden-Württemberg, Germany
tau.julis@gmail.com

Abstract
National Museums Scotland holds a triple-expansion steam
engine that has suffered corrosion damage leading to a
disfiguring orange patina on its surfaces. An experiment for
bluing mild steel was carried out to replace parts of the engine.
Late 19th- and early 20th-century recipes comprising nontoxic,
environmentally innocuous ingredients such as sodium nitrate,
potassium nitrate, manganese dioxide, tannic acid and tartaric
acid were used to protect both the conservator and the
environment. The experiment tested hot and cold patination
recipes, including a ready-made cold patination liquid, against
the original recipes. In addition, finishing techniques were
compared, including abrasive and polishing methods, on 22
mild-steel tokens. Other finishing protocols including immersion,
vapour colouring, and brushing and flaming techniques were
also investigated. Moreover, the reconstituted recipes to produce
gels with poultice techniques using laponite and fumed silica
were used. The blue patinas obtained were evaluated using
different analytical techniques: (i) stereo and optical microscopy
to analyse the colour, texture and homogeneity; and (ii) scanning
electron microscopy with energy-dispersive x-ray spectroscopy
to analyse texture, surface composition and surface distribution
using elemental mapping. Analysis revealed typical corrosion
features on the steel which appeared to be most severe on
patinas created with gels. In contrast, samples patinated using
heat only developed minor corrosion, which on some samples
was undetectable. These results indicate that the heating
technique provides a homogeneous patina with favourable
colouration. However, further studies are needed to determine
how to achieve the best results when patinating large metal
sheets.
Keywords
bluing, environmentally friendly chemicals, National
Museums Scotland, triple expansion steam engine, ferrous
metal corrosion, scanning electron microscopy with energy-
dispersive x-ray spectroscopy

Introduction
National Museums Scotland (NMS) owns a working
model of a triple-expansion high-speed steam engine
which has developed active corrosion on most surfaces.
The blued metal surfaces had a patchy, disfigured appear-
ance after conservation treatment, resulting in the need
to replace some parts of the engine. Testing was carried
out to replicate the exact colour of the engine after bluing.

The aim was to find techniques that could provide the

metal surface with a protective homogeneous patina.
Based on reviews of the literature, original 19th- and
20th-century recipes containing innocuous ingredients
were selected in order to protect both the conservator
and the environment. Figure 1. Triple-expansion steam engine

Bluing or browning is a chemical process applied to

ferrous metals, namely on arms and armour, in addi- inlaid surface, but also to create a protective layer on
tion to machinery and tools. It can be carried out for metalwork (Hughes 1993) and to prevent dazzling and
aesthetic reasons, to create a patina or a contrast on an reflection (Angier 1936). The purpose of colouring metals
194  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

(Table 1). A commercially available non-environmentally

friendly cold patination liquid, Curator Antiquing Fluid
Black, was also tested for comparison. Following applica-
tion, each token was thoroughly washed with water, dried
with paper towels and stored in the same environment.

Results and discussion

Figure 2. Ferrous metal cylinder during conservation treatment,
highlighting how disfigured the surface looks after corrosion treatment
is to accelerate the naturally occurring phenomenon
of corrosion and patination and thus create the desired
ageing effect (Untracht 1969).
Test design and overview of methods
Different application techniques were tested on 22 mild-
steel tokens of 0.7 mm thickness. The protocols included
immersion, brushing and vapour and torch techniques.
Prior to testing, all tokens were manually and chemically
cleaned. Various surface finishes were produced during
the cleaning process on an arbitrary basis. Observations
showed that these made no difference to the outcome of
the tests and were therefore not used as a control or vari-
able for evaluation (Table 2). The recipes were adapted by
adding gelling agents to allow testing of cold patination
techniques as this is the preferred method in conservation
The patinas were evaluated using optical and stereo
microscopy to compare colour, thickness and coverage.
Scanning electron microscopy with energy-dispersive
x-ray (SEM-EDX) spectroscopy was used to analyse the
patinas on eight tokens in greater detail. Terms such as
‘good colour or ‘good patina’ refer herein to the analysis
and visual comparison of the colours. The colour of the
tokens was not measured scientifically; it was purely
judged by the human eye of the conservators at NMS.
This experiment did not aim to produce scientifically
measurable colour results as museum visitors will also
judge the colours on the restored object by eye.
Cold patination technique
The data from stereo microscopy, optical microscopy
and electron microscopy indicates that none of the cold
patination techniques gave good results. The patinas were
uneven and discoloured. Interestingly, the control patina
consisting of the ready-made cold patination liquid also
failed to give a satisfactory result as, despite the patina
being homogeneous, it is too thin and active corrosion
is visible on its surfaces.

Table 1. Recipes used

Recipes Source
1 Recipe no. 1: Mix equal amounts of sodium nitrate and potassium nitrate and then melt. When the solution Metal coloring and finishing
reaches 560–580°F/293.5–304°C, the colour of the metal changes. Transfer to boiling water once the metal methods, 1914, Industrial Press
reaches the desired colour to remove salts from the surface.
2 Parker process: Dilute 25 g of phosphoric acid + 1.5 g of manganese dioxide in 1 L of water (six parts Field and Bonney 1925
phosphoric acid to one part manganese dioxide). Place the metals in the solution and boil for 2–4 hours.
3 Nitre Blue: Mix potassium nitrate and 8–10% manganese dioxide and heat to 200°C. The metals should be Angier 1936, 123
immersed in this solution for 5 minutes.
4 Fruit-acid black: Mix 4 grams of tannic acid with tartaric acid in 1/4 pints of water (150 mL). The solution is Angier 1936, 127
applied to the cold metal surface until the desired colour appears.
5 Heat treatment: Butane gas torch. Field and Bonney 1925
6 Parker process gel: Dilute 3.12 mL of phosphoric acid and 0.25 mL of manganese dioxide in 125 mL of water.
Add laponite synthetic clay to the solution to create a gel.
7 Recipe no. 1 gel: Mix equal amounts of sodium nitrate and potassium nitrate and dilute in water. Add fumed
silica to the solution to create a gel.
8 Nitre Blue gel: Dilute potassium nitrate and 8–10% manganese dioxide in water. Add fumed silica to the
solution to create a gel.
9 Fruit-acid black gel: Mix 4 g of tannic acid with tartaric acid in 1/4 pints of water (150 mL). Add laponite
powder to the solution to create a gel.
BLUING STEEL: AN INVESTIGATION INTO THE THERMAL AND CHEMICAL METHODS OF COLOURING MILD STEEL 195

Table 2. Summary of cleaning techniques, application methods and results for each sample piece

Surface finish and Token

Results/Information on topography Discussion/Notes
patination technique no.
Abrasive papers, pumice Colour: surfaces are rather black instead of blue, with purple appearing in No. 22 The aim was to see what type of patina the
powder, acetone the middle. ready-made patination liquid produced. The
gained patina appeared to be homogeneous
Antiquing Fluid Black (10% Topography:
under optical microscopy, but needed
diluted in water) - the patina appears to be homogeneous; however, SEM imaging showed
confirmation from high-resolution imaging.
- Immersion technique/less visible scratch marks from surface finishing, which means the created
than 1 minute patina is thin. Small areas of the surface are beginning to show corrosion.
- Micro mesh (grade Colour: green, greyish-black towards the edges. No. 1 SEM was carried out only on this sample
1800–8000), pumice powder out of 6 tokens which were patinated with
Topography:
and acetone the same chemical compound. Both visual
- uneven patina.
and optical microscopy confirmed that this
Fruit-acid black – brushing - built-up thick deposits of patina towards the edges of surfaces.
process did not produce a good patina,
- the rest of the surface is uneven with a few visible brush marks.
therefore only one of the samples was looked
SEM-EDX highlighted that the green area is homogeneous but covered with at in more detail.
small particles. It was not possible to see chemical differences between the
patina and the particles.
- Abrasion unit. Cooked in Colour: bluish grey. No. 10 Prior to patination, the surface was cleaned
baking soda for 20 minutes with an abrasion unit (using particles of glass),
Topography:
prior to colouring. which gave it a matte look.
- homogeneous patina at first glance; however, SEM imaging highlighted
Recipe no. 1 – that the patina looks heterogeneous, its morphology reflects the abrasion
Immersion technique/20 and numerous porosities are visible.
minutes - surface has developed pitted corrosion.
Abrasive papers, polishing Colour: blue. No. 13 This patina was visually the most appealing
mop, acetone of the chemically patinated samples. It looked
Topography:
homogeneous and well adhered.
- Recipe no. 1 -Immersion - patina appears to be homogeneous.
technique/20 minutes - SEM-EDX imaging suggested the presence of a very thin patina.
Acetone Colour: shades of blue, purple and pink. No. 15 A homogeneous, good patina appeared on
the surface of the token despite the lack of
Recipe no. 1 – Immersion Topography:
surface finish prior to colouration.
technique/20 minutes - homogeneous patina.
- corrosion has started to form on surfaces.

The second SEM-EDX images showed longitudinal lines, probably from

the manufacture of the coupon as the surface was not particularly treated
mechanically prior to patination.
Abrasive papers, acetone Colour: light purple to grey. No. 18 A type of salt crystal growth was present on
the surface of the specimen. Etched parts of
Parker process gel Topography:
the mild-steel surface could also be detected.
Immersion technique/5 days - thick, uneven patina
The combination of the above-mentioned
- salt crystal growth on surfaces
formations produced an interesting and
- etched surfaces
unusual texture, to be investigated in more
- surfaces are actively corroding
detail by SEM-EDX.
The information of the chemical composition of the hexagonal and compact-
looking crystalline structure provided by SEM-EDX analysis showed Mg, Fe
and P (with minor Si, Al, Na and Ca); this appears to be something created
during the application of the laponite-based gel.
Garryflex blocks, acetone Colour: greyish - black. No. 19 The token was analysed by SEM-EDX to
examine and compare its surface to token
Parker process gel Topography:
no. 18. The patina formed differed in both
Immersion technique/10 - uneven, thick patina
colour and thickness. A formation of salt
hours - does not cover all surfaces
crystals was expected, but it was not visible
- some salt crystal growth on surfaces
in optical microscopy images. SEM analysis
- active corrosion has formed on surfaces
then confirmed the crystal particles on the
SEM-EDX analysis detected round, small particles covering the surface of surfaces.
the steel. Their diameter can be variable, but the larger ones are around 20
microns and the smaller around 10 microns.
0000 grade fine wire wool, Colour: blue, very similar to the colour of the triple-expansion steam engine. No. 16 This patina was the most appealing of all
acetone test samples. The heat treatment provided a
Topography:
visually pleasing, homogeneous patina.
Flaming Heat treatment - homogeneous patina
using butane gas torch/few
SEM imaging highlighted that although the surface of the patina appeared
seconds
to be homogeneous, it was not completely flat or smooth.
196  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Gels
The results obtained show that the gels do not produce a
good patina. Most are too acidic and cause active corro-
sion and etching on the surfaces.
Thermal treatments
In total, three recipes were tried (Table 1), of which Recipe
no. 1 worked best. Looking at the data gained from the
scientific instruments used, the results show that this
patination method can work well. The colours produced
are good, identical or very similar to the colour of the
triple-expansion steam engine. The patinas appear to be
homogeneous in general. The different surface finishes
did not make a huge difference on the final patina.
The issue with this patination method is that a few of the
samples developed active corrosion. Further testing on
tokens that are thoroughly degreased and subsequently
monitored following patination will allow this issue to
be further evaluated.
Heat treatment/torch technique
Analysis indicates that this patination method provides
the best results. The patinas are homogeneous, of satis-
factory colouration and have good coverage (Figures 3,
4 and 5). SEM-EDX images highlighted that only oxygen
and iron elements are present, meaning no cross-con-
tamination or corrosion occurred.
Figure 3. Detail of
sample 1 (badly
developed patina)
 at 1.0–4.0×
Figure 4. Surface
of sample 16 (best
 patina) at 1.6–2.0×
CORROSION STUDIES
Figure 5. Sample 1
(badly developed
patina), SEI 100×
high vacuum at
 30 mmwd
Figure 6. Surface
of sample 16 (best
patina) SEI 100×
high vacuum
SEM-EDX at
 25 mmwd
This experiment demonstrated that most chemically
produced patinas showed active corrosion reactions
approximately two weeks after the patination was carried
out. The reason for the corrosion is unknown, but it
may be due to the solutions being highly acidic (pH 2),
which can lead to the rapid formation of corrosion on
the surface. The laboratory environment may have also
contributed to corrosion development due to the relative
humidity and temperature.
Conclusion
Cold patination methods were the least successful of
the techniques tested. Treatment with gels resulted in a
patchy, uneven appearance on the tokens. Furthermore,
corrosion on the surface of the objects was exacerbated
by immersion in the gels. The best chemical treatment
was achieved using Recipe no. 1. The patinas produced
with this method were the most homogeneous, providing
a satisfactory result.
However, in comparison, the heat treatment provided
superior results. By using a torch, a homogeneous,
attractive blue colour was achieved that closely matched
the original object and also provided a protective layer
on the steel surface.
Acknowledgements
This paper was produced with the help of Dr Lynda
Skipper (University of Lincoln), who supervised my
BLUING STEEL: AN INVESTIGATION INTO THE THERMAL AND CHEMICAL METHODS OF COLOURING MILD STEEL 197

work and made suggestions during the research process,

and Dr Lore Troalen (National Museums Scotland), who
helped me with the analytical techniques used to evaluate
the created patinas.

Materials list
Tannic acid (C76H52O46), tartaric acid (C4H6O6),
sodium nitrate (NaNO3), potassium nitrate (KNO3),
phosphoric acid (H3PO4), manganese dioxide (MnO2),
Laponite™ - Synthetic Smectite Clay, Fumed Silica-Syn-
thetic Amorphous Silicon Dioxide (SiO2), Curator©
Antiquing Fluid Black - Selenious acid (H2SeO3), solu-
tion CAS No. 7446-08-4

References
Angier, R.H. 1936. Firearm blueing and browning.
London: Arms and Armour Press.
Field, S. and S.R. Bonney. 1925. The chemical colouring
of metals. London: Chapman and Hall Ltd.
Goodway, M. 1993. The patination of iron by bower-
barffing. In Metal plating and patination. Cultural, tech-
nical and historical developments, eds. S. La Niece and P.
Craddock, 155–60. Oxford: Butterworth-Heinemann Ltd.
Hughes, R. 1993. Artificial patination. In Metal plating
and patination. Cultural, technical and historical develop-
ments, eds. S. La Niece and P. Craddock, 1–18. Oxford:
Butterworth-Heinemann Ltd.
Untracht, O. 1969. Metal techniques for craftsmen. A
basic manual for craftsmen on the methods of forming
and decorating metals. London: Robert Hale Limited.

Author
Júlia Tauber is a metals conservator at Landesamt für
Denkmalpflege Baden-Württemberg, Germany. She
has a master’s in historic object conservation from the
University of Lincoln (UK) and a graduate diploma
in metalwork conservation from West Dean College
(UK). She was assistant conservator in the Engineering
Conservation Department at National Museums Scot-
land, completed an internship in metalwork conservation
at the National Maritime Museum and has worked as a
freelance conservator in the private sector.
 Young Conservation Professional

Behaviour of Lead Coated with Benzotriazole Solution in

an Organic Acid Environment

Aditi Nagar* Manager Rajdeo Singh

Institute for Sustainable Heritage
University College London
London, UK
aditi.nagar.1993@gmail.com
aditi.nagar.18@ucl.ac.uk
National Museum Institute
New Delhi, India
manager.singh.nmi@gov.in
*Author for correspondence

Abstract
The National Research Laboratory for Conservation of Cultural
Heritage in Lucknow, India, researched a method to inhibit the
corrosion of lead objects using a benzotriazole (BTA) solution. The
formation of a protective film of BTA on the lead surface and the
sensitivity of BTA to organic acid concentration at varied levels
of high relative humidity was studied. Samples were analysed
using scanning electron microscopy-energy dispersive x-ray
spectroscopy and x-ray diffraction. The samples were exposed
to acetic acid, formic acid and hydrochloric acid vapours and
subjected to electrochemical impedance spectroscopy in ASTM
D 1384-87 test solution.
Keywords
lead, benzotriazole, acetic acid, formic acid, hydrochloric acid,
polarisation resistance

Introduction
All metals, except precious metals like gold and silver, Experimental details
exist in mineral and ore form. Once a metal comes into
Experiments were conducted on 86%-pure lead coupons
contact with the environment it has a tendency to revert
(Table 2). Eight square-shaped lead coupons (4 × 4 cm)
to its mineral form. For example, lead is extracted from
were polished with fine grade emery paper to achieve a
the ore galena but when it is exposed to the environment
scratch-free surface, followed by degreasing with acetone.
it forms lead oxide, carbonates and other products. Lead
An aqueous, neutral BTA solution (1%) was prepared
also reacts with acids such as acetic acid and formic acid
and four coupons were coated while four remained
to form readily soluble lead acetate and lead formates.
uncoated. The coated coupons were immersed overnight
Lead is continuously corroded by organic acids to create
in the BTA solution to allow the formation of a Pb-BTA
a uniform corrosion pattern (Costa and Urban 2005).
layer. Coupon A1 was analysed using energy-dispersive
This type of corrosion can be seen when lead comes into
x-ray spectroscopy (EDX) after coating in the BTA to
contact with wood, varnishes and paints (Blackshaw and
analyse the percentage composition of lead in the sample
Daniels 1979).
(Table 2). Two coupons, A1 (with BTA) and B1 (without
Benzotriazole (BTA) has been used as a corrosion BTA), were then subjected to impedance analyses to
inhibitor for the conservation of copper-based artefacts gauge the protective property of the BTA coating and
(Sease 1978). Sharma, Lal and Singh (2003) introduced a the patinated lead surface (Table 1).
method using BTA to stabilise leaded bronze. The effect
Electrochemical impedance spectroscopy (EIS) was
of neutral BTA solution in inhibiting the corrosion of lead
performed using a three-electrode system: a working
is due to the formation of a crystalline and polymeric
electrode (lead coupon), a counter electrode (graphite)
Pb-BTA complex on its surface.
and a calomel electrode as the reference. The experiments
This research was conducted with the aim of testing were performed at room temperature (25°C) with a
the efficacy of the proposed method, and to analyse sinusoidal voltage of 10 mV at the open circuit potential
the action of corrosive species such as carbonates, (varying the frequency between 10 mHz and 100 kHz
chlorides and sulfates on the Pb-BTA and patinated using a potentiostat). An ASTM D 1384-87 test solution
lead surface. was used composed of sodium sulfate (148 mg/L), sodium
BEHAVIOUR OF LEAD COATED WITH BENZOTRIAZOLE SOLUTION IN AN ORGANIC ACID ENVIRONMENT 199

chloride (165 mg/L) and sodium bicarbonate (138 mg/L). This suggests that environmental oxygen formed a layer
The test solution was used to provide an aggressive of lead oxide before the application of the BTA.
environment for the BTA-protected and patinated lead
surfaces. Lead makes different corrosion products with Table 2. EDX results on the lead sample, BTA-coated lead sample (A1)
and samples AA, FA and HcA (after exposure to an acidic environ-
carbonates, and the sodium chloride and sodium sulfate ment)
were used to observe the action of chloride and sulfate
EDX results
ions on the BTA-coated and patinated lead surfaces. A
Sample name Weight Weight Weight Weight Weight
constant phase element (CPE) model was used to analyse % % % % %
the impedance data. C N O Pb Cl
Lead coupon
uncoated and 5.9 - 3.5 86.2 -
exposed
A1 12.4 2.3 10.2 74.9 0.00
AA 23.0 3.4 24.3 46.0 0.00
FA 22.5 5.0 23.4 48.9 0.00
HcA 15.6 0.00 0.00 62.0 23.3

EDX analysis on sample AA was similar to sample A1

(Table 2). There was a considerable increase in carbon and
Figure 1. Equivalent electrochemical circuit of the CPE model used to
extract the impedance data from the plots oxygen after the lead-BTA surface was exposed to acetic
acid (Table 2). This is corroborated by the XRD spectra,
Six lead coupons were simultaneously exposed to which show a strong peak in lead acetate (Figure 2).
different acids (Table 1) in a desiccator to accelerate the
degradation of the BTA-coated and patinated surfaces.
After two weeks of exposure, the surfaces were analysed
using x-ray powder diffraction (XRD) and EDX to iden-
tify the corrosion products. The lead coupons were then
subjected to further electrochemical analyses to compare
the difference in protective property of the BTA-coated
surface and the patinated surface after exposure to
different acidic environments.

Results and discussion

Figure 2. XRD spectra of the surface of sample AA
EDX analysis on Sample A1 showed the presence of C,
N, O and Pb. The amount of oxygen detected was greater A similar trend was observed for sample FA with EDX
than for the uncoated or unexposed surfaces (Table 2). (Table 2) and XRD analysis (Figure 3).

Table 1. Nomenclature of lead samples used in the experiment. Samples A1 and B1 were directly subjected to EIS, while samples AA, AA1, FA, FA1,
HcA and HcA1 were first exposed to an acidic environment and then subjected to EIS

Specimen Inhibitor coating Given environment Exposure time After exposure

A1 With BTA EIS - -
B1 Without BTA EIS - -
AA With BTA Acetic acid 14 days EIS
AA1 Without BTA Acetic acid 14 days EIS
FA With BTA Formic acid 14 days EIS
FA1 Without BTA Formic acid 14 days EIS
HcA With BTA Hydrochloric acid 14 days EIS
HcA1 Without BTA Hydrochloric acid 14 days EIS
200  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CORROSION STUDIES

Figure 5. Bode plot of samples A1 and B1

Figure 3. XRD spectra of sample FA

In contrast to samples AA and FA, a different corrosion

product formed on the surface of sample HcA. Interest-
ingly, no nitrogen was detected but there was a significant
amount of chloride ion concentration (Table 2). These Figure 6. Bode plot of samples AA and AA1
results indicate that the BTA had completely decomposed
during exposure to the hydrochloric acid (Table 2). The
presence of carbon is probably due to the BTA residue
that decomposed during the period of exposure. To
identify the phases present on the surface, the sample
was analysed by XRD and showed a strong peak of lead
chloride and carbon (Figure 4).

Figure 7. Bode plot of samples FA and FA1

Figure 8. Bode plot of samples HcA and HcA1

Figure 4. XRD spectra of sample HcA
After EDX and XRD analysis of samples AA, FA and HcA,
EIS was performed in an ASTM D 1384-87 test solution
to assess the performance of the corrosion inhibitor on
lead surfaces. A Bode plot for samples A1 and B1 showed
impedance values at the lowest frequency (left Y-axis)
(Figure 5). This indicated the lowest impedance provided
by the specific sample in an environment. This figure for
A1 (the BTA-protected lead surface) was considerably
higher than for B1 (patinated lead sample). The phase
shift value (right Y-axis) showed that both samples had
only one time constant at an intermediate range of about
100 Hz. The Bode plot for samples AA, FA and HcA also
showed the same pattern (Figures 6–8).
Table 3 shows impedance parameters α (CPE) and Rp
(polarisation resistance). As expected, polarisation resist-
ance shows that sample A1 exhibits about three times
higher Rp as compared to B1 (Table 3).
When the samples were exposed to a corrosive envi-
ronment (samples AA, FA and HcA), their properties
deteriorated considerably. Table 3 shows the decrease in
Rp values compared to sample A1, which demonstrates
the deterioration of the protective layer for both BTA
and the patinated lead surface.
In order to access the quantitative inhibitive performance
of the BTA treatment on the surface of lead samples,
percentage inhibition efficiencies (IE%) were calculated
BEHAVIOUR OF LEAD COATED WITH BENZOTRIAZOLE SOLUTION IN AN ORGANIC ACID ENVIRONMENT 201

Table 3. Electrochemical impedance parameters taken from the CPE It is evident from the results that BTA is not a promising
model of equivalent circuits in tests performed in ASTM D 1384-87 solution for the conservation of lead-based metal arte-
solution shown in Figure 8
facts. The Pb-BTA layer is sensitive to acids and does
Y0 not resist the action of chloride ions effectively. Further
Sample Ru (ohm.cm2)
(S.sa/cm2) ααα Rp (ohm.cm2)
studies are required to check the longevity of the BTA
A1 805 17.2*10-6 0.601 6970
layer for the long-term protection of lead-based artefacts.
B1 1376 59.72*10 -6
0.702 2150
AA 637 424*10 -6
0.741 487
Acknowledgements
AA1 469 1870*10-6 0.791 72
FA 520 2240*10-6 0.790 317 We would like to express our major gratitude to Dr Achal
FA1 701 455*10 -6
0.780 97 Pandya and his staff at Indira Gandhi National Centre
HcA 543 293*10 -6
0.785 82 for the Arts, New Delhi, for donating their time and
HcA1 524 34.5*10 -6
0.780 100 resources to conduct the electrochemical impedance
spectroscopy. Our thanks also go to Dr D.D.N Singh for
valuable discussions on the analytical data.
using Rp (Table 4) for the different treated samples using
the following formula: References

( )
corrosion rate of untreated sample –
corrosion rate of BTA treatead sample
%E = 100
corrosion rate of untreated sample
Table 4. Percentage inhibition efficiency of BTA on lead samples
exposed to different environments
No. Sample name %E
1 A1 68.84
2 AA 85.21
3 FA 69.40
4 HcA -21.95
The data recorded in Table 4 shows that the BTA gives
higher inhibition efficiency in acetic acid, followed by
formic acid. On the other hand, HCl vapour accelerated
corrosion and imparted negative efficiency, which is
also evident from the EDX and XRD results on sample
HcA, which shows the presence of lead chloride on the
BTA-protected lead surface (Table 2).
Conclusion
The EDX results (Table 2) and the Rp values (Table 3)
obtained in the experiments showed that acidic environ-
ments caused the deterioration of the Pb-BTA layer. The
EDX and XRD results showed the formation of different
corrosion products on the Pb-BTA surface when exposed
to different acids, such as lead acetate and lead chloride.
The results suggest that HCl vapour has the most corro-
sive effect on Pb-BTA and the patinated lead surface,
followed by formic acid and acetic acid.
Blackshaw, S. and V. Daniels. 1979. The testing of mate-
rials for use in storage and display in Museums. The
Conservator 3(1):16–19.
Costa, V. and F. Urban. 2005. Lead and its alloys: Metal-
lurgy, deterioration and conservation. Studies in Conser-
vation 50(sup1.): 48–62.
Sease, C. 1978. Benzotriazole: A review for conservators.
Studies in Conservation 23(2): 76.
Sharma, V., U. Lal, and T. Singh. 2003. Method for stabi-
lization of leaded bronzes affected by corrosion of lead.
Studies in Conservation 48(3): 203–09.
Authors
Aditi Nagar has a master’s in Art Conservation from the
National Museum Institute of History of Art, Conserva-
tion and Museology, New Delhi. She is currently studying
for a master’s in Research Science and Engineering in
Arts, Heritage and Archaeology from the Institute for
Sustainable Heritage at University College London.
Manager Rajdeo Singh is a professor and head of the
Department of Conservation at the National Museum
Institute of History of Art, Conservation and Museology,
New Delhi.
 PROTECTION:
CORROSION INHIBITORS
From Mono-oxime to Tri-oxime: The Development of a
New Group of Corrosion Inhibitors for Copper Alloys
Ahmad N. Abu-Baker
Yarmouk University
Irbid, Jordan
ahmad.abubaker@yu.edu.jo
Abstract
This study aims to synthesize and investigate a trioxime
compound—2,4,6-trihydroxybenzene-1,3,5-tricarbaldehyde
trioxime (TBTT)—as a corrosion inhibitor for corroded bronze
artifacts. The chloride-based corrosion of archaeological
bronze artifacts was synthetically simulated on bronze samples
(Cu94/Sn6) by anodic polarization measured by linear sweep
voltammetry (LSV) in NaCl solution, followed by exposing
the samples to high relative humidity and temperature in an
environmental chamber to promote the oxidation of the initially
formed corrosion products. Scanning electron microscopy/
energy dispersive spectroscopy (SEM/EDS) and x-ray diffraction
(XRD) analyses were used to investigate the morphology and
composition of the synthetically corroded bronze samples. A
solution containing TBTT and Paraloid B-72 was used as a single
Introduction
Archaeological bronze artifacts excavated from chlo-
ride-containing environments and stored in aerated
storage areas with high relative humidity are susceptible
to an active corrosion process commonly known as
bronze disease (MacLeod 1981, Scott 1990, Robbiola,
Blengino, and Fiaud 1998, Casaletto et al. 2006,
Abu-Baker 2008). The use of corrosion inhibitors has
been a common treatment for such artifacts (Falter-
meier 1997, Argyropoulos et al. 2007, Mezzi et al.
2012, Hassairi et al. 2013, Albini et al. 2018); it has the
benefit of preserving the inhibited corrosion layer on the
surface of the artifact, which protects its historical and
aesthetic value (Cano et al. 2010, Ćurković et al. 2012).
Nevertheless, only few corrosion inhibitors are actually
used in conservation practice, and they are not always
effective for heavily corroded artifacts with a mixture
of corrosion products (Cano and Lafuente 2013). This
paper aims to synthesize and investigate a trioxime
compound––2,4,6-trihydroxybenzene-1,3,5-tricarbal-
dehyde trioxime (TBTT)––derived from the structure
of salicylaldoxime as a corrosion inhibitor for bronze.
conservation treatment for the synthetically corroded bronze
samples. Electrochemical impedance spectroscopy (EIS) in 0.5 M
NaCl solution showed that the inhibition effectiveness increased
with increasing the concentration of TBTT in the treating solution.
SEM/EDS and Fourier transform infrared (FTIR) spectroscopy
showed that TBTT coordinated with metal ions in the corrosion
products and formed a protective film on the bronze samples.
The treated samples showed no signs of corrosion re-occurrence
after exposure to a high relative humidity (84% RH) at 38°C for
one week. It was concluded that the treatment was effective in
inhibiting the active chloride-based corrosion of bronze.
Keywords
oxime, bronze, corrosion, inhibitor, conservation
The study represents a sequential development from
previous studies investigating salicylaldoxime and its
dioxime derivative (4,6-dihydroxy benzene-1,3-dicar-
baldehyde dioxime [DBDD]) as corrosion inhibitors for
copper alloys (Abu-Baker et al. 2013, Abu-Baker and
Al-Qudah 2016, Abu-Baker and Al-Qudah 2017). Since
salicylaldoxime was found to be a reasonably safe, effec-
tive, easy to apply, and environmentally friendly corro-
sion inhibitor and to form an aesthetically acceptable
inhibited corrosion layer, it was proposed that increasing
the number of oxime and hydroxyl groups around
the benzene ring in the compound would improve its
interaction and coverage on the corroded surface. The
interaction of TBTT with the chloride-based corrosion
of bronze was investigated by scanning electron micros-
copy/energy dispersive spectroscopy (SEM/EDS) and
Fourier transform infrared (FTIR) analyses. Electro-
chemical impedance spectroscopy (EIS) measurements
in NaCl solution and accelerated corrosion testing in
an environmental chamber were used to evaluate the
effectiveness of a mixture of TBTT and Paraloid B-72
FROM MONO-OXIME TO TRI-OXIME: THE DEVELOPMENT OF A NEW GROUP OF CORROSION INHIBITORS FOR COPPER ALLOYS
as a single treatment to inhibit the chloride-based
corrosion of bronze.
Materials and methods
The trioxime compound TBTT was prepared by
reacting 2,4,6-trihydroxybenzene-1,3,5-tricarbal-
dehyde (Hangzhou Yuhao Chemical Technology),
hydroxylamine hydrochloride (Loba Chemie), and
sodium hydroxide (Fisher Chemical) according to the
procedure mentioned in Premužić, Filarowski,  and
Hołyńska (2017). TBTT was characterized by nuclear
magnetic resonance (NMR) spectroscopy using a Bruker
DRX-400 spectrometer. DMSO-d6 was the solvent, and
TMS was the internal standard. Spectra of 1H-NMR and
13
C-NMR were recorded at 400 MHz and 100 MHz,
respectively.
Bronze samples (50 × 50 × 0.4 mm) with Cu94/Sn6
composition (Goodfellow) were abraded with 2000-grit
silicon carbide paper, then washed with distilled water,
degreased in acetone in an ultrasonic bath, rinsed with
distilled water, and dried. A synthetic corrosion layer
was formed by anodic polarization measured by linear
sweep voltammetry (LSV) in 0.5 M NaCl from the open
circuit potential (Eoc) reached after immersion for one
hour in +1V vs Ag/AgCl/sat’d KCl (0.198 vs SHE) as the
reference electrode, with a scanning rate of 10 mV/s,
using an Ivium Vertex potentiostat. For each sample,
a 250 mL solution of 0.5 M NaCl was used as the elec-
trolyte, and a T-316 stainless-steel mesh was used as
the counter electrode. After the anodic polarization,
the sample was removed from the electrolyte, left to
dry in air to allow the corrosion layer to develop, then
rinsed in acetone. The samples were then moved to
a climatic chamber (Xenotes 150 S+) and exposed to
high relative humidity and temperature (84% RH and
38°C) for 24 hours.
The surface morphological features and elemental
compositions of the initial corrosion products formed
on the bronze samples after anodic polarization in
0.5  M NaCl solution, and then the final corrosion
products were examined after exposure to 84% RH at
38°C for 24 hours by SEM/EDS (FEI-Quanta 200). The
mineralogical compositions of the corrosion products
were determined using a Shimadzu LabX, XRD-6000
x-ray diffractometer under voltage U = 40 kV, current
I = 30 mA, wavelength CuKα = 1.54178 Å, test range
10–90° 2θ, step size 0.02 in 2θ, and 0.6 s time constant.
205
Samples of the synthetically corroded bronze were
immersed in 200 mL solutions containing mixtures of
3% Paraloid B-72 (Museum Services Corporation) and
0.001 M, 0.005 M, or 0.01 M TBTT dissolved in acetone
for 24 hours, then removed, and allowed to dry. EIS meas-
urements in 0.5 M NaCl solution were then conducted
on the treated samples to evaluate the effectiveness of
the treatment in blocking the chloride ions, therefore
inhibiting the corrosion of bronze. The measurements
were carried out in a cell consisting of a glass tube (50 mL)
mounted on a nylon base, the test area exposed to the
solution was 1 cm2, Ag/AgCl/sat’d KCl was the reference
electrode, and a graphite rod was the counter electrode.
SEM/EDS and FTIR spectroscopy were used to inves-
tigate the surface morphological and compositional
changes and the bonding mechanism of the inhibitor after
treating the corroded samples with a solution containing
0.01 M TBTT and 3% Paraloid B-72. FTIR spectra were
measured with a Bruker Tensor 27 spectrometer using a
KBr pellet, and Opus software was used to collect the data.
Triplicate samples of the corroded bronze treated with
the 3% Paraloid B-72/0.001 M, 0.005 M, or 0.01 M TBTT
solutions were exposed to 84% RH and 38°C for one week
in the Xenotes 150 S+ climatic chamber to evaluate the
treatment’s effectiveness. The method was adapted from
the ASTM D1735–04 and ASTM D2247–02 standards
(ASTM International 2002 and 2004) for testing the water
resistance of coatings in an enclosed chamber, with slight
modification in the RH% value to a more realistic value
experienced in poor storage conditions (Abu-Baker et
al. 2013). After the test, the samples were evaluated by
visual and microscopic examination using a Cooling
Tech digital microscope to spot any corrosion eruption.
Results and discussion
Characterization of TBTT
Figure 1a shows the structure of TBTT and the number
assignment for its carbon atoms. The 1H-NMR spectrum
(Figure 1b) shows a proton signal at δ 8.34 ppm for the
proton attached to the double bond (H7, H8, and H9).
The broad signals at δ 11.53 ppm and 11.74 ppm indicate
the presence of hydroxyl groups, where the down field
signal at δ 11.74 ppm is due to intra-molecular hydrogen
bonding, which is in accordance with the proposed
structure of the compound. The 13C-NMR spectrum of
the compound in Figure 1c shows three signals identified
206  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
at δ 97.66 ppm, 145.54 ppm, and 158.03 ppm. The peak
at δ 97.66 ppm refers to C1, C3, and C5, and the peak
at δ 158.03 ppm refers to C2, C4, and C6 of the TBTT
compound. The peak of the olefinic carbons (C7, C8,
and C9) of the oxime groups is located at δ 145.54 ppm.
a)
b)
c)
Figure 1. (a) The structure and number assignment for carbon atoms,
(b) 1H-NMR spectrum, and (c) 13C-NMR spectrum of TBTT compound
Synthesizing and characterizing the corrosion
layer on bronze samples
Figure 2 shows the anodic polarization curve measured
by LSV for a bronze sample in 0.5 M NaCl. It shows three
anodic regions that correspond to the dissolution of the
alloy’s metal ions and the formation of corrosion species
PROTECTION: CORROSION INHIBITORS
over the polarization range. The first region starts from
the Eoc and increases in its current density to reach a
maximum point marked “a” equaling 783 mA at 0.098 V.
This region corresponds to the simultaneous oxidation
and dissolution of tin and copper ions into the electrolyte.
This is followed by a decrease in the current density at
higher potential to form a lower current plateau marked
“b”, and corresponds to the formation of tin hydroxy
chlorides and cuprous chloride corrosion products. The
last region, marked “c”, corresponds to a low steady state
current region resulting from the blocking behavior of
the formed copper and tin corrosion products (Souissi
et al. 2007, Robbiola et al. 2008).
Figure 2. Anodic polarization curve measured by linear sweep
voltammetry (LSV) for a bronze (Cu94/Sn6) sample in 0.5 M NaCl solution
Figure 3 shows a corroded bronze sample after the anodic
polarization process in the 0.5 M NaCl solution, and
its SEM/EDS analysis. The micrograph shows a porous
corrosion layer covering the surface, and the EDS analysis
indicates that copper chlorides are the main corrosion
products, with the presence of a small amount of tin-con-
taining corrosion products. XRD analysis (Figure 4)
shows that the mixture of corrosion products contains
mainly nantokite (CuCl) and mushistonite (CuSn(OH)6).
The exposure of this electrochemically corroded sample
to 84% RH and 38°C for 24 hours increased the thickness
and porosity of the corrosion layer, and the EDS analysis
indicates a decrease in the amount of tin and chloride
on the surface (Figure 5). The XRD analysis shows that
the main corrosion products on the surface are atacamite
and paratacamite (Cu2(OH)3Cl) (Figure 6). It is suggested
that the electrochemical corrosion process of the bronze
alloy in the NaCl solution involved an internal oxidation
and preferential dissolution of copper (decuprification),
while tin compounds remained in the corrosion layer and
FROM MONO-OXIME TO TRI-OXIME: THE DEVELOPMENT OF A NEW GROUP OF CORROSION INHIBITORS FOR COPPER ALLOYS 207

formed a network through which copper ions migrated a) b)

(Souissi et al. 2007, Robbiola et al. 2008). The initially
formed nantokite (CuCl) oxidized in the high humidity
and high temperature atmosphere to the powdery basic
cupric chloride polymorphs (Cu2(OH)3Cl) on the outer
surface, and the released chloride ions promoted further
corrosion in the underlying alloy, therefore starting a
new corrosion cycle. This corrosion process resem-  
bles the corrosion of archaeological bronze artifacts c)
commonly known as bronze disease (MacLeod 1981,
Scott 1990). The two-step accelerated corrosion proce-
dure allowed the corrosion behavior of archaeological
bronze artifacts to be reproduced on a modern bronze
alloy, therefore avoiding sampling from ancient artifacts
for research experiments and improving the reproduci- Figure 5. (a) Corroded bronze sample after the anodic polarization in
bility of samples for comparative purposes. the 0.5 M NaCl solution, followed by exposure to 84% RH and 38°C for
24 hours; (b) SEM image of the corroded surface; (c) EDS analysis of the
corrosion products
a) b)

 
c)

Figure 6. XRD analysis of a corroded bronze sample after the anodic

Figure 3. (a) Corroded bronze sample after the anodic polarization
process in the 0.5 M NaCl solution, (b) SEM image of the corroded surface,
(c) EDS analysis of the corrosion products
Figure 4. XRD analysis of a corroded bronze sample after the anodic
polarization process in the 0.5 M NaCl solution
polarization process in 0.5 M NaCl, followed by exposure to 84% RH
and 38°C for 24 hours
Evaluating the effectiveness of TBTT by electro-
chemical impedance spectroscopy (EIS)
The Bode plots of EIS in 0.5 M NaCl solution for the
untreated and treated corroded bronze samples are
shown in Figure 7a and demonstrate that as the corrosion
inhibitor concentration increases, the resistance increases
and the capacitance decreases. The sample treated with
the solution containing 0.01 M TBTT has the highest
impedance at the low frequency end; its inhibitor-con-
taining film increases the difficulty for charge transfer
and adsorption of the electrolyte, which indicates an
increase in the effectiveness of the treatment with the
increase in the concentration of TBTT. The Nyquist plots
in Figure 7b show that increasing the concentration of
208  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
TBTT leads to an increase in the resistance, which is
manifested by the increase in the radius of the semi-circle.
a)
a)
b)
Figure 7. (a) Bode magnitude and phase angle plots and (b) Nyquist plots
of the EIS in 0.5 M NaCl for synthetically corroded bronze untreated and
treated with TBTT and Paraloid B-72
Investigating the Nyquist plots at low frequency showed
the presence of a second time constant, so an equivalent
circuit with two time constants was used to fit the results
(Figure 8). In this equivalent circuit, Rs represents the
solution resistance, Rf and Rct are the film resistance and
charge transfer resistance, and their sum is the polari-
zation resistance (Rp). CPEf is a constant phase element
PROTECTION: CORROSION INHIBITORS
for film capacitance (Cf ) and CPEdl is a constant phase
element for the double layer (Cdl) (Hassairi et al. 2013).
Figure 8. Equivalent circuit used to fit the EIS parameters in 0.5 M NaCl
for synthetically corroded bronze untreated and treated with TBTT and
Paraloid B-72
The fitted equivalent circuit parameters for bronze are
listed in Table 1. The table shows that CPEf and CPEdl
decreased with the increase in the concentration of TBTT.
The decrease in CPEf was attributed to increasing inhib-
itor adsorption, which reduces the exposed electrode
surface area, while the decrease in CPEdl was attributed to
replacing water molecules at the alloy/solution interface
by the adsorbed inhibitor molecules, which results in a
decreased local dielectric constant and a thicker electric
double layer. The values of CPE-P became closer to 1 with
the increase in TBTT, which indicates a better insulating
layer on the surface of the treated sample. The increase
in Rf and Rct following the increase in the concentration
of TBTT indicated the formation of a film of adsorbed
inhibitor molecules on the synthetically corroded bronze
samples, which hinders the chloride-induced corrosion
from the EIS electrolyte and inhibits the interfacial charge
transfer. The sum of the film resistance (Rf ) and charge
transfer resistance (Rct) was considered the polarization
resistance (Rp) and used to calculate the inhibition effi-
ciency according to the following equation (Tansuğ et
al. 2014):
Rp — Rpo
IE% =  100%
Rp
where Rpo and Rp are the polarization resistances for
untreated and treated samples, respectively. The esti-
mated inhibition efficiencies are listed in Table 1. The
maximum inhibition efficiency was found to be 91% for
the samples treated with 0.01 M TBTT with 3% Paraloid
B-72.
Investigating the treated samples by SEM/EDS and
Fourier transform infrared (FTIR) spectroscopy
The SEM image of the synthetically corroded bronze
sample after treatment with 0.01 M TBTT and 3% Para-
loid B-72-containing solution is shown in Figure 9a.
FROM MONO-OXIME TO TRI-OXIME: THE DEVELOPMENT OF A NEW GROUP OF CORROSION INHIBITORS FOR COPPER ALLOYS 209

Table 1. EIS in 0.5 M NaCl solution parameters for synthetically corroded bronze untreated and treated with TBTT and Paraloid B-72

CPE- Rp
CPE- Rf CPE-T-2 CPE- Rct IE
Treatment T-1 (Rf + Rct)
P-1 (Ω cm2) (F cm-2) P-2 (Ω cm2) (%)
(F cm-2) (Ω cm2)
Untreated 1.97E-04 0.4 9.12E+01 2.55E-03 0.6 1.08E+03 1.17E+03
0.001 M TBTT + 3% 1.54E-04 0.5 2.05E+02 7.56E-04 0.6 1.59E+03 1.80E+03 35
Paraloid B-72
0.005 M TBTT + 3% 1.40E-04 0.5 2.09E+02 1.81E-04 0.8 5.99E+03 6.20E+03 81
Paraloid B-72
0.01 M TBTT + 3% 1.35E-04 0.5 2.40E+02 1.70E-04 0.8 1.28E+04 1.30E+04 91
Paraloid B-72

Comparing the micrograph of the treated sample with products of bronze. There is also a decrease in intensity
that of the untreated sample shown in Figure 5b shows the in the C=N, C-O, and N-O bands, which indicates the
formation of a surface-covering layer by the chemisorp- involvement of nitrogen and deprotonated oxygen in
tion of inhibitor molecules and coordination with metal the complex formation with metal ions in the corrosion
ions in the corrosion products. The EDS analysis in products of bronze. The hydrogen atom of the =N-OH
Figure 9b confirms the presence of TBTT as its C and group may have also been involved in the intermolecular
N appear in the result. hydrogen bonding of the formed complex (Abu-Baker and
Al-Qudah 2017). It is suggested that the three aldoxime
a) groups and three hydroxyl groups may interact with the
metal ions in the corrosion products to form a protective
film on the corroded surface. No clear, distinctive bands
for Paraloid B-72 appear in the FTIR spectrum of the
Figure 9. SEM image (a)
and EDS analysis (b) of treated sample in Figure 10; however, the presence of a low
a synthetically corroded concentration of Paraloid B-72 increased the adherence
bronze sample treated of the inhibited corrosion layer to the underlying alloy. It
with a 0.01 M TBTT and 3%
may also have improved surface coverage and protection
 Paraloid B-72 solution
against water absorption on the underlying surface, which
b)
increased the effectiveness of the treatment.

The FTIR spectrum for TBTT (the black line in Figure 10)

showed bands at 2850–3550 cm-1 for O-H stretching
vibration in the =N-OH and C-OH groups. C=C and C=N
stretching vibration was observed between 1390 cm-1 and
1680 cm-1, and C-O stretching and in-plane O-H bending
were at 1100–1370 cm-1. N-O stretching vibrations were
observed in the range of 915–1050 cm-1, out-of-plane
C-C bending vibration for the benzene ring was at
700–840 cm-1, and the out-of-plane O-H bending vibra- Figure 10. FTIR spectra of TBTT and the synthetically corroded bronze
tion was at 570–660 cm-1 (Cicileo et al. 1999). treated with a 0.01 M TBTT and 3% Paraloid B-72 solution

The FTIR spectrum for the synthetically corroded bronze

treated with the TBTT-containing solution (red line in
Figure 10) shows that the O-H stretching bands became
weaker as a result of the ionization of the OH phenolic
group and complexation to the metal ions in the corrosion
Evaluating the effectiveness of TBTT in an envi-
ronmental chamber
Comparing the status of samples before and after expo-
sure to 84% RH and 38°C for one week (Table 2) shows
210  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
that no corrosion occurred on the treated samples with
various concentrations of TBTT; however, a slight color
darkening occurred on the treated surface, that is, a
loss of hue in the L*a*b* system. The samples were also
weighed before and after the test; the change in the
weight of treated samples was negligible in comparison
with the untreated sample, which increased in weight by
0.04%—this suggests a continuation of the active corro-
sion processes in the alloy of the untreated samples. The
microscopic images of the samples after the test show
not only an increase in surface coverage by increasing
the concentration of TBTT in the treating solution but
also the absence of active corrosion blisters on the treated
samples, which indicates a complete inhibition of the
corroded products on the bronze alloy.
Table 2. A set of untreated and treated synthetically corroded bronze
samples before (a) and after (b) exposure to the high relative humid-
ity and temperature test (84% RH and 38°C) for one week, and their
microscopic images after the test (c)
Treatment a) Before the b) After the c) Microscopic
test test images
Untreated
0.001 M
TBTT + 3%
Paraloid
B-72
0.005 M
TBTT + 3%
Paraloid
B-72
0.01 M
TBTT + 3%
Paraloid
B-72
Conclusion
This study aimed to synthesize and investigate a trioxime
compound––2,4,6-trihydroxybenzene-1,3,5-tricarbal-
dehyde trioxime (TBTT)––as a corrosion inhibitor for
corroded bronze artifacts. The interaction of TBTT with
the metal ions in the chloride-based corrosion of bronze
was confirmed by SEM/EDS and FTIR analyses. EIS
measurements in NaCl solution and accelerated corro-
sion testing in an environmental chamber showed that
PROTECTION: CORROSION INHIBITORS
a mixture of TBTT and Paraloid B-72 was effective as a
single treatment to inhibit the chloride-based corrosion
and protect the aesthetic appearance of the synthetically
treated corroded samples. The results of this research
support the previously achieved results of introducing
and testing oxime compounds as corrosion inhibitors
for copper alloys.
Acknowledgments
I would like to thank the Deanship of Scientific Research
and Graduate Studies at Yarmouk University for funding
this research. I would also like to thank Dr. Mahmoud
Al-Qudah and Ayat Foudeh for the NMR analysis, Sana
Al-Khasawneh for the XRD and FTIR analyses, and
Mohammad Khawaja for the SEM/EDS analysis.
References
Abu-Baker, A. 2008. Investigating the corrosion and
microstructure of five copper-based archaeological
artefacts from Tell el-Ajjul. AICCM Bulletin 31(1): 87–96.
Abu-Baker, A.N., I.D. MacLeod, R. Sloggett, and R.
Taylor. 2013. A comparative study of salicylaldoxime,
cysteine and benzotriazole as inhibitors for the active
chloride-based corrosion of copper and bronze artifacts.
European Scientific Journal 9(33): 228–51.
Abu-Baker, A.N. and M.A. Al-Qudah. 2016. A new
dioxime corrosion inhibitor for the protection and
conservation of copper: Synthesis, characterization and
evaluation in acidic chloride solution. Applied Physics
A 122: 765.
Abu-Baker, A.N. and M.A. Al-Qudah. 2017. A novel
dioxime compound for protecting copper in neutral
chloride solutions and to treat bronze disease in archae-
ological artefacts. AICCM Bulletin 38 (2): 94–102.
Albini, M., P. Letardi, L. Mathys, L. Brambilla, J. Schröter,
P. Junier, and E. Joseph. 2018. Comparison of a bio-based
corrosion inhibitor versus benzotriazole on corroded
copper surfaces. Corrosion Science 143: 84–92.
Argyropoulos, V., M. Giannoulaki, G.P. Michalakakos,
and A. Siatou. 2007. A survey of the types of corrosion
inhibitors and protective coatings used for the conser-
vation of metal objects from museum collections in the
Mediterranean basin. In Strategies for saving our cultural
heritage, eds. V. Argyropoulos, A. Hein, and M.A. Harith,
1–5. Athens: TEI of Athens.
FROM MONO-OXIME TO TRI-OXIME: THE DEVELOPMENT OF A NEW GROUP OF CORROSION INHIBITORS FOR COPPER ALLOYS
ASTM International. 2002. Standard practice for testing
water resistance of Coatings in 100% relative humidity
(ASTM D2247–02), West Conshohocken, PA, USA. www.
astm.org/cgi-bin/resolver.cgi?D2247 .
ASTM International. 2004. Standard practice for testing
water resistance of coatings using water fog apparatus
(ASTM D1735–04), West Conshohocken, PA, USA.
www.astm.org/cgi-bin/resolver.cgi?D1735 .
Cano, E., D.M. Bastidas, V. Argyropoulos, S. Fajardo,
A. Siatou, J.M. Bastidas, and C. Degrigny. 2010. Elec-
trochemical characterization of organic coatings for
protection of historic steel artefacts. Journal of Solid State
Electrochemistry 14: 453–63.
Cano, E. and D. Lafuente. 2013. Corrosion inhibitors for
the preservation of metallic heritage artefacts. In Corrosion
and conservation of cultural heritage metallic artefacts, eds.
P. Dillmann, D. Watkinson, E. Angelini, and A. Adriaens,
570–94. Cambridge: Woodhead Publishing Limited.
Casaletto, M.P., T. De Caro, G.M. Ingo, and C. Riccucci.
2006. Production of reference “ancient” Cu-based alloys
and their accelerated degradation methods. Applied
Physics A 83(4): 617–22.
Cicileo, G.P., B.M. Rosales, F.E. Varela, and J.R. Vilche.
1999. Comparative study of organic inhibitors of copper
corrosion. Corrosion Science 41(7): 1359–75.
Ćurković, H.O., T. Kosec, K. Marušić, and A. Legat. 2012.
An electrochemical impedance study of the corrosion
protection of artificially formed patinas on recent bronze.
Electrochimica Acta 83: 28–39.
Faltermeier, R. 1997. Testing corrosion inhibitors for the
conservation of archaeological copper and copper alloys.
Journal of Conservation and Museum Studies 3: 1–5.
Hassairi, H., L. Bousselmi, S. Khosrof, and E. Triki. 2013.
Evaluation of the inhibitive effect of benzotriazole on
archeological bronze in acidic medium. Applied Physics
A 113(4): 923–31.
MacLeod, I.D. 1981. Bronze disease: An electrochemical
explanation. ICCM Bulletin 7(1): 16–26.
Mezzi, A., E. Angelini, T. De Caro, S. Grassini, F. Faraldi,
C. Riccucci, and G.M. Ingo. 2012. Investigation of the
benzotriazole inhibition mechanism of bronze disease.
Surface and Interface Analysis 44(8): 968–71.
Premužić, D., A. Filarowski, and M. Hołyńska. 2017.
Structure and properties of a new rigid tripodal oxime
ligand. Journal of Molecular Structure 1136: 100–106.
211
Robbiola, L., J.M. Blengino, and C. Fiaud. 1998.
Morphology and mechanisms of formation of natural
patinas on archaeological Cu-Sn alloys. Corrosion Science
40(12): 2083–111.
Robbiola, L., T.T.M. Tran, P. Dubot, O. Majerus, and
K. Rahmouni. 2008. Characterisation of anodic layers
on Cu–10Sn bronze (RDE) in aerated NaCl solution.
Corrosion Science 50(8): 2205–215.
Scott, D.A. 1990. Bronze disease: A review of some chem-
ical problems and the role of relative humidity. Journal of
the American Institute for Conservation 29(2): 193–206.
Souissi,  N., E. Sidot, L. Bousselmi,  E. Triki,  and L.
Robbiola. 2007. Corrosion behaviour of Cu–10Sn bronze
in aerated NaCl aqueous media – Electrochemical inves-
tigation. Corrosion Science 49(8): 3333–47.
Tansuğ, G., T. Tüken, E.S. Giray, G. Fındıkkıran, G.
Sığırcık, O. Demirkol, and M. Erbil. 2014. A new corro-
sion inhibitor for copper protection. Corrosion Science
84: 21–9.
Author
Ahmad N. Abu-Baker is an associate professor at
the Department of Conservation and Management of
Cultural Resources at the Faculty of Archaeology and
Anthropology at Yarmouk University in Irbid, Jordan.
He has qualifications in applied chemistry (B.Sc.),
conservation and management of cultural resources
(MA), and cultural materials conservation (Ph.D.).
His research interests include corrosion and corrosion
inhibition of archaeological metals and alloys, electro-
chemical techniques in metal conservation, innovative
analytical techniques applied to conservation science,
and archaeometallurgy.
Assessing the Interaction between Corrosion Inhibitors
and the Historical Corrosion Layer on Copper Claddings
Based on Global and Micrometric-Scale Analysis
Erika Ferrari* Delphine Neff
LAPA NIMBE – IRAMAT, CEA/CNRS, LAPA NIMBE – IRAMAT, CEA/CNRS,
UMR 3685 UMR 3685
Gif-sur-Yvette, France Gif-sur-Yvette, France
erika.ferrari@cea.fr delphine.neff@cea.fr
Emilande Apchain Marine Bayle
LPPI – Université de Cergy-Pontoise A-CORROS Expertise
Cergy-Pontoise, France Arles, France
emilande.apchain@u-cergy.fr mbayle@a-corros.fr
Abstract
In this paper, some of the physico-chemical properties of the
patina on naturally corroded copper samples before and after
treatment with organic corrosion inhibitors are examined. The
interaction of the brochantite layer is assessed in terms of color
change, penetration depth of the treatment, change in surface
wettability, and reduction of permeability to liquid water. The
aim of the study was to evaluate the characteristics of the
Introduction
Corrosion inhibitors are different compounds (inorganic,
organic) that reduce the corrosion rate by acting on the
corrosion reaction (Trabanelli and Carassiti 1970, Cano
and Lafuente 2013). They have many applications in the
industrial field and also in the protection of metallic
artifacts.
Inhibitors in the industrial field are often applied in
solution and in direct contact with the metal surface.
Inhibition efficiency is assessed mainly by immersing the
metal under study in a solution containing the inhibitor
and subjecting it to electrochemical tests. Specific stand-
ards corresponding to laboratory methodologies have
to be followed when determining inhibition efficiency
in industrial applications (Fateh, Aliofkhazraei, and
Rezvanian 2017).
The testing methodology as well as the commercially
available corrosion inhibitors need to be developed
and adjusted from industrial application to the cultural
heritage field. From the methodological point of view, a
common standardized way to evaluate the protectiveness
of inhibitors and coatings does not yet exist in cultural
heritage. Electrochemical tests are widely employed to
Philippe de Viviés Philippe Dillmann
A-CORROS Expertise LAPA NIMBE – IRAMAT, CEA/CNRS,
Arles, France UMR 3685
devivies@a-corros.fr Gif-sur-Yvette, France
philippe.dillmann@cea.fr
Jean-Bernard Memet
A-CORROS Expertise *Author for correspondence
Arles, France
jbmemet@a-corros.fr
transformed patina as a possible innovative methodology to
approach the study and role of corrosion inhibitors in cultural
heritage.
Keywords
corrosion inhibitor, atmospheric corrosion, natural patina,
copper, multi-scale analysis
understand the effectiveness of coatings and inhibitors
and have the advantage of being non-destructive and able
to be performed in situ (Letardi 2004; Cano, Lafuente, and
Bastidas 2010). When laboratory studies are performed, a
certain effort is devoted to reproducing patinas naturally
formed in specific environmental conditions. Different
methods for producing artificial patinas are available
in the literature (Ropret and Kosec 2012, Di Carlo et
al. 2017, Masi et al. 2017). In most cases, the composi-
tion of the corrosion layer on these coupons represents
that found on monuments and artifacts corroded in an
outdoor environment. In an urban environment, for
instance, brochantite, a cupric hydroxyl sulfate, is the
main phase identified in corrosion layers (Mazzeo et al.
2008). Nevertheless, the morphological aspects of natural
patinas (porosity, thickness), despite being considered
relevant in the corrosion process, are usually neglected
because it is hard to reproduce them in the laboratory.
As an alternative, it is possible to use naturally corroded
coupons, like, for example, copper roofs (Mazzeo et al.
2008, Joseph et al. 2011, Albini et al. 2018).
In this study, two derivatives of benzotriazole (BTA)
are investigated. Azoles represent a relevant part of
ASSESSING THE INTERACTION BETWEEN CORROSION INHIBITORS AND THE HISTORICAL CORROSION LAYER ON COPPER CLADDINGS BASED ON GLOBAL AND MICROMETRIC-SCALE ANALYSIS 213

organic inhibitors for copper. Their action is related to and the possibility of performing destructive analysis. A
the presence of nitrogen atoms, which can share free disadvantage was the limited number of samples, which
pairs of electrons with Cu(I) and Cu(II). The presence did not allow the production of replicas. The copper
of heterocycles in the molecule is also recognized to be plate was cut into 1.5 × 3 cm samples and each sample
responsible for enhancing chemisorption on copper embedded in polyurethane resin, allowing penetration
surfaces (Antonijevic and Petrovic 2008). Among the of the solution from the upper corroded surface only and
azoles, BTA is largely employed as an inhibitor for the avoiding penetration from the sides.
protection of artworks. BTA is classified as an anodic
inhibitor, which acts by both chemisorption (forming
the complex Cu(I)- and Cu(II)-BTA) and physisorption
on copper surfaces. In cultural heritage, it is used to
stabilize archaeological artifacts (Angelucci et al. 1978)
and to inhibit corrosion, as both pre-treatment (Brostoff,
Shedlosky, and de la Rie 2000, Kosec, Legat, and Milošev
2010) or as an additive in a coating system, such as
Incralac (Wolfe and Grayburn 2017).
Figure 1. The city of Metz. To the right, one of the slabs employed in
The formation of Cu(II)-BTA and Cu(I)-BTA insoluble the study
complexes was assessed by reacting BTA with different
copper phases (Brostoff 1997). The protectiveness of BTA
Corrosion inhibitors
on corroded surfaces is usually related to the formation
of these complexes. Despite numerous studies, there Three corrosion inhibitors were used (Table 1) whose
is no agreement on the effectiveness of BTA in inhib- molecular structure is shown in Figure 2. The compounds
iting corrosion of heritage artifacts. BTA has also been chosen included BTA, considered a reference since it has
extensively investigated concerning its toxicity (Beltoft, been studied both for industrial and cultural heritage
Nielsen, and Ladefoged 2013). It is often defined as applications. The two BTA derivatives are supposed to act
toxic and carcinogenic and its use in cultural heritage like BTA, although their efficiency and, hypothetically,
is discouraged. Many BTA derivatives are available for the effect of their interaction with the patina might vary
experimentation, but information concerning their due to the different substituents on the benzene and the
toxicity is sometimes missing. azole ring, respectively:
• benzotriazole (BTA)
The goal of the study was to understand the changes the
• benzotriazole-5-carboxylic acid (5-CBT)
patina undergoes after interaction with three inhibitors
• methyl 1H-benzotriazole-1-carboxylate (Me-CBT)
by coupling global information (surface wettability,
color change) and micrometric-scale information
(interaction with brochantite, penetration of liquid
water), and evaluating the properties before and after
treatment.

Figure 2. Molecular structure of the inhibitors

Materials and methods
Copper samples Table 1. List of the compounds and concentration used for the exper-
iments
The set of samples comprised copper plates from the clad-
dings of Saint Martin’s church in Metz, France (Figure 1). Name used in Molar concentration
Compound
the publication (mol/L)
This allowed for work to be carried out on patinas
Benzotriazole BTA 0.25
formed over a long period of time (ca. 100 years) and
Benzotriazole-5-carboxylic 5-CBT 0.25
thus representative of copper corrosion in an unsheltered acid
urban environment. Other advantages were the overall Methyl-benzotriazole Me-CBT 0.06
homogeneity of the patina, the flatness of the samples, carboxylate
214  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Treatment
Hydroalcoholic solutions (50% v/v ethanol/water) with
the three different inhibitors were prepared (Table 1).
The choice of concentration is related to the literature
available on BTA (Scott 2002) and related to the low
solubility in the case of Me-CBT.
Samples were immersed for 2 hours. After immersion,
each sample was rinsed with a 50% v/v hydroalcoholic
solution and air-dried, then subsequently oven-dried at
60°C for 48 hours.
Influence of the treatment on the patina
The samples were investigated by non-destructive and
destructive techniques. The effect of the interaction
between the patina and the different corrosion inhibitors
was assessed both on the surface (color change, contact
angle) and in cross section (penetration of the treatment
and penetration of liquid water).
Surface characterization
Visual examination can assess whether there has been any
color change. For this purpose, pictures of the samples
were taken under the same lighting conditions and
referred to a color scale.
The static contact angle was also measured by following
the recommendations of UNI EN 15802:2010 (Ente
Nazionale Italiano di Unificazione 2010). A drop of
distilled water was placed on the surface and lateral
pictures of the drop determined the contact angle. Here,
due to the limited size and surface roughness of the
samples, 5 10 µL drops were deposited on the surface
instead of 15 as recommended by UNI EN 15802:2010.
Standard deviation was measured based on the values
for the 5 drops.
Brochantite-inhibitor interaction
Cross sections of the copper samples were prepared. An
embedding system using a “barrier” film of Paraloid B-72
avoided penetration by the embedding resin (Epofix) into
the porous patina, as the presence of the resin can hinder
the possibility of distinguishing the treatment from the
resin itself when performing µ-Raman spectroscopy.
Cross sections prepared following this methodology were
then polished with SiC paper (800, 1200, 2400, 4000) and
diamond paste (3 and 1 µm).
PROTECTION: CORROSION INHIBITORS
Cross sections were analyzed with an Olympus light
microscope. A JEOL JSM-7001F scanning electron
microscope (SEM) coupled with an energy-dispersive
x-ray spectroscope (EDS) ; accelerating voltage varied
from 5 to 15 kV. EDS data were collected and treated
with Aztec software.
Raman spectra and mappings were collected using
a Renishaw µ-Raman spectrometer equipped with a
532 nm Nd:YAG laser and a CCD detector. Resolution of
the spectra obtained was 2 cm-1 with a spot size of 1 µm.
Laser power was set at 0.5–1 percent of LP.
Change in water permeability
An immersion test in D2O allowed treated and untreated
samples to be compared in terms of liquid water intake.
Untreated and treated samples were embedded in polyu-
rethane resin to leave a surface of approximately 1 cm2 in
contact with the D2O. They were immersed in 20 mL of
D2O for 30 days. Samples were oven-dried at 60°C until
a stable weight was reached, then embedded in resin
and cut and polished to 1 µm to study the cross section.
Detection of deuterium was performed using an IONTOF
TOF-SIMS 5 equipped with Surface Lab software. The
Bi+ primary ion beam produces an ionization on the
surface and a beam of secondary ions, among which the
deuterium, is collected and analyzed. The enrichment
in deuterium within the corrosion layer is calculated as
follows (Equation 1):
(—HD )x — 1
( )
δ = 100 ∙ (1)
(—HD )mat
Prior to analysis, roughly 500 µm2 of the surface was
sputtered in three areas of the cross- section using a
Cs+ ion beam to eliminate any surface pollution. Within
this area, three regions of interest (100 µm2 each) were
analyzed: the corrosion layer, part of the resin, and the
metal. A flood gun was used during the analysis for
charge compensation.
Results
The corrosion layer before treatment
Stratigraphic observations of the corrosion layer before
applying any treatment were performed with optical
microscopy (OM), µ-Raman spectroscopy and SEM-EDS.
ASSESSING THE INTERACTION BETWEEN CORROSION INHIBITORS AND THE HISTORICAL CORROSION LAYER ON COPPER CLADDINGS BASED ON GLOBAL AND MICROMETRIC-SCALE ANALYSIS
Average thickness was estimated by OM at 50–60 µm. In
Figure 3a, the SEM picture of the cross -section shows
the characteristic porosity of the patina: perpendicular
cracks of a few µm running from the surface down to
the metal with micro, and macro pores.
Figure 3. (a) SE image of the untreated corrosion layer (CL) and (b) OM
bright field image with superimposed µ-Raman mapping
As previously observed in corrosion studies (see discus-
sion section), two main phases are identified by µ-Raman
spectroscopy. Figure 3b shows the µ-Raman mapping
overlapping the corresponding OM picture. In contact
with the metal, there is a thin, irregular layer of cuprite
(in red on the Raman map) evidenced by the peak at
216 cm-1; in general, this layer is less than 5 µm thick.
The main mineral phase found in the corrosion layer is
brochantite (green on the Raman map), which is iden-
tified by a strong peak corresponding to symmetrical
stretching vibration of (SO4)2- at 970 cm-1. SEM-EDS anal-
ysis identified the presence of chlorine, but at very low
concentration. As shown in Figure 4, the EDS linescan,
showing chlorine (blue) and copper (pink), evidences
how a higher amount of chlorine, of up to 6 percent in
weight, is located at the interface between cuprite and
brochantite.
Figure 4. SE image and linescan corresponding to the white rectangle
for copper (pink) and chlorine (blue). The scale in wt% for chlorine is
on the right
215
After verifying the overall compositional and morpho-
logical homogeneity of the patina, the copper plates were
immersed in each of the inhibitor solutions. Two non-de-
structive tests (color change and contact angle test) were
performed to evaluate the effect of the treatment.
Surface characterization of treated coupons
In the absence of a colorimeter, color change was eval-
uated by visual examination (Figure 5); the results are
summarized in Table 2. The shift to a dark-green color
can be noticed for both BTA and 5-CBT. Almost no
change is registered for Me-CBT.
Table 2. Color change observations
Treatment Color change observations
BTA Dark green, white crystals present on the surface
5-CBT Dark green
Me-CBT No change observed
Figure 5. Color
of treated and
untreated
 samples
Figure 6. Result of the drop test
Surface wettability was assessed by the contact angle
test. The untreated copper surface was characterized by
high surface wettability. A drastic change in volume was
observed when a water drop was placed on the surface,
thus allowing no contact angle measurement. On the
contrary, as shown in Figure 6, all tested compounds
216  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PROTECTION: CORROSION INHIBITORS

Figure 7. Micro-Raman
spectra of pure inhibitors
(red) and spectra of the
treated brochantite layer
(black). The marked peak in
each spectrum corresponds
to the one used in the
mapping to identify the
interaction with each
inhibitor. This observation
is the result of a previous
study on the interaction
among mineral phases and
 inhibitors

provided a certain hydrophobicity to the surface layer
(θ>90°).
Interaction with the brochantite layer
Cross-sections of the treated samples were analyzed
to determine the penetration of the inhibitor within
the patina. For all compounds, no clear interaction
could be identified with the cuprite layer or the metal.
Nevertheless, a clear interaction with the brochantite
layer was detected by µ-Raman spectroscopy. Raman
mapping corresponding to the complex is based on the
shift of the strong benzene ring peak when the azole
ring interacts with Cu(II) (Figure 7). This conclusion
is the result of previous experiments (not presented
here) on the synthesis of Cu(II) complexes and further
comparison with existing Raman and IR spectra (Rubim
et al. 1983, Brostoff 1997, Brostoff 2003). Figure 8 shows
the cross-section and Raman mapping for (a) BTA, (b)
5-CBT, and (c) Me-CBT, respectively.
The Raman mapping pointed out the presence of cuprite
(red), brochantite (light blue), and the BTA/BTA deriv-
ative complex. Evidence of the presence of the complex
Cu(II)-BTA and Cu(II)-Me-CBT was found throughout
the brochantite layer, while Cu(II)-5-CBT could only be
detected in the first half of the corrosion layer.

Permeability to liquid water

The D2O experiment results are shown in Figure 9.
Enrichment in deuterium (δ%) is represented by the
columns in the histogram, each corresponding to one
of the treated samples. The first column (UT) shows δ%
for the untreated sample. For the treated samples, δ%

Figure 9. Results
of the Tof-SIMS
analysis expressed
Figure 8. OM image and µ-Raman mapping in cross section for the as deuterium
samples treated with (a) BTA, (b) 5-CBT, and (c) Me-CBT  enrichment
ASSESSING THE INTERACTION BETWEEN CORROSION INHIBITORS AND THE HISTORICAL CORROSION LAYER ON COPPER CLADDINGS BASED ON GLOBAL AND MICROMETRIC-SCALE ANALYSIS
decreases. Comparable values for 5-CBT (δ%=479) and
Me-CBT (δ%=478) are registered. None of the inhibitors
completely prevented liquid water from penetrating the
layer after a one-month immersion.
Discussion
The compositional features of the corrosion layer exam-
ined here are in agreement with observations previously
done on similar samples exposed to an urban environ-
ment (Graedel, Nassau, and Franey 1987, FitzGerald et
al. 2006, Morcillo et al. 2017). The higher stability of
brochantite when compared to other hydroxyl-sulfates
explained its presence as the main corrosion product.
The layer of cuprite in contact with the metal is the other
typical corrosion species described in the literature. The
localization of chlorine inside the corrosion layer is
discussed in previous studies about atmospheric copper
corrosion (Graedel, Nassau, and Franey 1987, Opila
1987, Watanabe, Higashi, and Ichino 2003, Watanabe et
al. 2009), although different explanations are suggested.
Since no Cl-containing mineral was detected, we can
suppose that chlorinated phases are present in very low
concentration.
The compounds tested interacted with the corrosion
layer by changing some of its properties.
1. Color change is an important parameter for cultural
heritage applications. It can also give an idea of the
formation of a chemical interaction. This was the
case with the BTA and 5-CBT: in both compounds, a
nitrogen atom from the azole ring bonded with copper
ions, causing a color shift to dark green.
2. A decrease in surface wettability was observed for all
the inhibitors, suggesting the formation of a hydro-
phobic layer on the surface. However, considering the
limits on measuring the contact angle (irregularity of
the surface, limited number of drops), an evaluation
of the treatment based exclusively on this parameter
was not possible.
3. The compounds penetrated the brochantite layer to
different degrees. BTA and 5-CBT were present in the
same concentration in solution and showed a similar
chemical structure, but their behavior was different
in terms of penetration: the Cu(II)-BTA complex
penetrated to the cuprite layer, while 5-CBT is present
on the surface and penetrated to within roughly the
first 20 µm. This could be explained by the speed of
formation of the 5-CBT complex compared to the BTA
217
complex. The complex itself could therefore have a role
in preventing further penetration of the solution into
the patina during the chosen immersion time. Me-CBT
penetrated the corrosion layer in a similar way to BTA,
despite being present at a lower concentration.
4. The difference in deuterium enrichment in treated and
untreated samples can be used to estimate permeability
to liquid water and to compare the sample before and
after treatment. A drastic decrease in water intake was
observed for 5-CBT and Me-CBT treated samples,
despite the difference in concentration in the testing
solutions. No complete blockage of liquid water was
registered (Figure 9).
In general, comparing the properties of the treated and
untreated samples, surface hydrophobicity increased,
permeability to liquid water decreased, and penetra-
tion depth for the different inhibitors was uneven. No
connection between penetration depth and water intake
could be drawn from the data available. However, over
time, in the case of an interaction limited to the surface,
it is expected that the protective action of the inhibitor
will be easily subjected to the action of rain runoff, thus
limiting the protective efficiency of the treatment.
The validity of this last observation needs to be verified
by long-term studies.
Conclusion
According to the definition by Lorenz (1985), some
inhibitors are defined as “interphase” as they “co-operate”
with the corrosion products to form a protective layer.
In this study, the interaction of different corrosion
inhibitors with a naturally corroded corrosion layer was
investigated using a multiscale physico-chemical char-
acterization approach. Results showed the interaction
with the brochantite layer at various depths, conferring
hydrophobic properties to the corrosion layer and
decreasing liquid water penetration, two properties that
are usually attributed to coating systems. The (partly)
modified treated corrosion layer was therefore seen to
be protective, while the role of the inhibitor directly on
the bare copper could not be evaluated and will be the
subject of further research.
References
Albini, M., P. Letardi, L. Mathys, L. Brambilla, J. Schröter,
P. Junier, and E. Joseph. 2018. Comparison of bio-based
218  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
corrosion inhibitors versus benzotriazole on corroded
copper surfaces. Corrosion Science 143: 84–92.
Angelucci, S., P. Fiorentino, J. Kosinkova, and M. Mara-
belli. 1978. Pitting corrosion in copper and copper alloys:
Comparative treatment tests. Studies in Conservation
23(4): 147–56.
Antonijevic, M.M. and M.B. Petrovic. 2008. Copper
corrosion inhibitors. A review. International Journal of
Electrochemical Science 3: 1–28.
Beltoft, V., E. Nielsen, and O. Ladefoged. 2013. Benzotri-
azole and tolyltriazole. Evaluation of health hazards
and proposal of health-based quality criteria for soil and
drinking water, Environmental project no.1526. Copen-
hagen: The Danish Environmental Protection Agency.
Brostoff, L.B. 1997. Investigation into the interaction
of benzotriazole with copper corrosion minerals and
surfaces. In Metal 95: Proceedings of the International
Conference on Metals Conservation, Semur-en-Auxois,
25–28 September 1995, eds. I.D. MacLeod, S.L. Pennec,
and L. Robbiola, 99–108. London: James & James (Science
Publishers) Ltd.
Brostoff, L.B., T.J. Shedlosky, and E.R. de la Rie. 2000.
External reflection study of copper-benzotriazole films
on bronze in relation to pretreatments of coated outdoor
bronzes. Studies in Conservation 45(sup. 1): 29–33.
Brostoff, L.B. 2003. Coating strategies for the protection
of outdoor bronze art and ornamentation. Ph.D. disser-
tation, University of Amsterdam, The Netherlands.
Cano, E., D. Lafuente, and D.M. Bastidas. 2010. Use of EIS
for the evaluation of the protective properties of coatings
for metallic cultural heritage: A review. Journal of Solid
State Electrochemistry 14(3): 381–91.
Cano, E. and D. Lafuente. 2013. Corrosion inhibitors
for the preservation of metallic heritage artefacts. In
Corrosion and conservation of cultural heritage metallic
artefacts, eds. P. Dillmann, D. Watkinson, E. Angelini, and
A. Adriaens, 570–94. Cambridge: Woodhead Publishing.
Di Carlo, G., C. Giuliani, C. Riccucci, M. Pascucci, E.
Messina, G. Fierro, M. Lavorgna, and G.M. Ingo. 2017.
Artificial patina formation onto copper-based alloys:
Chloride and sulphate-induced corrosion processes.
Applied Surface Science 421: 120–27.
Ente Nazionale Italiano di Unificazione. 2010. Conserva-
tion of cultural property – Test methods – Determination
of static contact angle (UNI EN 15802:2010).
PROTECTION: CORROSION INHIBITORS
Fateh, A., M. Aliofkhazraei, and A.R. Rezvanian. 2017.
Review of corrosive environments for copper and its
corrosion inhibitors. Arabian Journal of Chemistry
(in press, available online). https://doi.org/10.1016/j.
arabjc.2017.05.021 .
FitzGerald, K.P., J. Nairn, G. Skennerton, and A. Atrens.
2006. Atmospheric corrosion of copper and the colour,
structure and composition of natural patinas on copper.
Corrosion Science 48(9): 2480–509.
Graedel, T.E. 1987. Copper patinas formed in the atmos-
phere-II. A qualitative assessment of mechanisms. Corro-
sion Science 27(7): 721–40.
Graedel, T., K. Nassau, and J.P. Franey. 1987. Copper
patinas formed in the atmosphere-I. Introduction.
Corrosion Science 27(7): 639–57.
Joseph, E., A. Simon, S. Prati, M. Wörle, D. Job, and R.
Mazzeo. 2011. Development of an analytical procedure
for the evaluation of the protective behavior of innova-
tive fungal patinas on archaeological and artistic metal
artefacts. Analytical and Bioanalytical Chemistry 399:
2899–907.
Kosec, T., A. Legat, and I. Milošev. 2010. The compar-
ison of organic protective layers on bronze and copper.
Progress in Organic Coatings 69(2): 199–206.
Letardi, P. 2004. Laboratory and field tests on patinas
and protective coating systems for outdoor bronze
monuments. In Metal 04: Proceedings of the International
Conference on Metals Conservation, Canberra (Australia),
4–8 October 2004, eds. J. Ashton and D. Hallam, 379–87.
Canberra, ACT: National Museum of Australia.
Lorenz, W.J.M.F. 1985. Interface and interphase corrosion
inhibitors. Electrochimica Acta 31(4): 467–76.
Masi, G., J. Esvan, C. Josse, C. Chiavari, E. Bernardi,
C. Martini, and L. Robbiola. 2017. Characterization of
typical patinas simulating bronze corrosion in outdoor
conditions. Materials Chemistry and Physics 200: 308–21.
Mazzeo, R., S. Bittner, G. Farron, R. Fontinha, D. Job,
E. Joseph, P. Letardi, M. Mach, S. Prati, M. Salta, and
A. Simon. 2008. Development and evaluation of new
treatments for outdoor bronze monuments. In Conser-
vation Science 2007, eds. J. Townsend, L. Toniolo, and F.
Cappitelli, 40–48. London: Archetype Publications Ltd.
Morcillo, M., T. Chang, B. Chico, D. de la Fuente, I.
Odenvall Wallinder, J.A. Jiménez, and C. Leygraf. 2017.
Characterisation of a centuries-old patinated copper
ASSESSING THE INTERACTION BETWEEN CORROSION INHIBITORS AND THE HISTORICAL CORROSION LAYER ON COPPER CLADDINGS BASED ON GLOBAL AND MICROMETRIC-SCALE ANALYSIS 219

roof tile from Queen Anne’s Summer Palace in Prague.
Material Characterization 133: 146–55.
Opila, R.L. 1987. Copper patinas: An investigation by
Auger electron spectroscopy. Corrosion Science 27(7):
685–94.
Ropret, P. and T. Kosec. 2012. Raman investigation of arti-
ficial patinas on recent bronze – Part I: Climatic chamber
exposure. Journal of Raman Spectroscopy 43(11): 1578–86.
Rubim, J., I.G.R. Gutz, O. Sala, and W.J. Orville-Thomas.
1983. Surface enhanced Raman spectra of benzotriazole
adsorbed on a copper electrode. Journal of Molecular
Structure 100: 571–83.
Scott, D.A. 2002. Copper and bronze in art. Los Angeles:
The Getty Conservation Institute.
Trabanelli, G. and V. Carassiti. 1970. Mechanism and
phenomenology of organic inhibitors. In Advances in
corrosion science and technology, eds. M.G. Fontana and
R.W. Steahle. Boston, MA: Springer.
Watanabe, M., Y. Higashi, and T. Ichino. 2003. Surface
observation and depth profiling analysis studies of corro-
sion products on copper exposed outdoors. Journal of the
Electrochemical Society 150(2): B37–B44.
Watanabe, M., T. Handa, T. Ichino, N. Kuwaki, and J.
Sakai. 2009. Characterization of patinas that formed on
copper exposed in different environments for one month,
Zairyo-to-Kankyo 58(4): 143–57.
Wolfe, J. and R. Grayburn. 2017. A review of the devel-
opment and testing of Incralac lacquer. Journal of the
American Institute for Conservation 56(3–4): 225–44.
iaux et Prévision de l’Altération (LAPA) NIMBE, CNRS/
CEA Saclay, France, on the protection of copper artifacts
with carboxylate compounds. She currently works
as a researcher at the Laboratoire de Physicochimie
des Polymères et des Interfaces (LPPI – Université de
Cergy-Pontoise).
Delphine Neff is a research scientist at LAPA. She
develops research projects on understanding the trans-
formation of ancient corrosion layers during desalina-
tion treatments as well as their interaction with organic
compounds applied as protective coatings.
Marine Bayle is a research and development scientist at
A-CORROS Expertise (located at the Pôle économique
et technologique ARCHEOMED©) in Arles, France, a
position she has held since 2016. She completed her
Ph.D. at LAPA on the transformation mechanisms that
occur during subcritical stabilization of archaeological
iron artifacts.
Philippe de Viviés is co-director of A-CORROS Exper-
tise. He specializes in the conservation of archaeological
artifacts from submarine environments.
Jean-Bernard Memet is co-director of A-CORROS
Expertise. He has a Ph.D. in material science and special-
izes in marine corrosion.
Philippe Dillmann is director of LAPA. He has a Ph.D.
in material science and undertakes research on the long-
term alteration of metals and glasses.

Materials list
Benzotriazole 99% (Sigma-Aldrich)
Benzotriazole-5-carboxylic acid 99% (Sigma-Aldrich)
Ethanol 98% (VWR)
Metyl 1H-benzotriazole-1-carboxylate 97% (Sigma-
Aldrich)

Authors
Erika Ferrari is a Ph.D. student at the Université
Paris-Saclay. She has a master’s degree in science applied
to Conservation of Cultural Heritage from Bologna
University.

Emilande Apchain carried out her Ph.D. thesis (Univer-

sité de Cergy-Pontoise) at the Laboratoire Archéomatér-
PROTECTION: COATINGS
Protection of Outdoor Bronzes with Eco-Friendly
and Non-Hazardous Coatings Based on Silane and
Fluoropolymers: Results from the B-IMPACT Project
M. Aufray, J. Esvan
CIRIMAT-ENSIACET (CNRS-INP
Toulouse)
Université Fédérale de Toulouse
Toulouse, France
maelenn.aufray@ensiacet.fr
jerome.esvan@ensiacet.fr
A. Balbo, V. Grassi,
C. Monticelli, F. Zanotto
Centro di Studi sulla Corrosione e
Metallurgia “A. Daccò”,
Tecnopolo di Ferrara, University of
Ferrara
Ferrara, Italy
andrea.balbo@unife.it
vincenzo.grassi@unife.it
cecilia.monticelli@unife.it
federica.zanotto@unife.it
F. Benetti, W. Sperotto,
E. Tedesco
Ecamricert SRL
Monte di Malo (VI), Italy
f.benetti@ecamricert.com
w.sperotto@ecamricert.com
e.tedesco @ecamricert.com
E. Bernardi
Dept. of Industrial Chemistry “Toso
Montanari”
University of Bologna
Bologna, Italy
elena.bernardi@unibo.it
M.C. Bignozzi, G. Masi
Dept. of Civil, Chemical,
Environmental, and Materials
Engineering
University of Bologna
Bologna, Italy
maria.bignozzi@unibo.it
giulia.masi5@unibo.it
C. Chiavari
Interdepartmental Centre for
Industrial Research
“Advanced Applications in
Mechanical Engineering and
Materials Technology”
University of Bologna
Bologna, Italy
Dept. of Cultural Heritage
University of Bologna
Ravenna, Italy
cristina.chiavari@unibo.it
N. Gartner, T. Kosec,
L. Škrlep, E. Švara Fabjan
National Building and Civil
Engineering Institute
Ljubljana, Slovenia
nina.gartner@zag.si
tadeja.kosec@zag.si
luka.skrlep@zag.si
erika.svara@zag.si
B-IMPACT@zag.si
C. Josse
Centre de Microcaractérisation
Raimond Castaing
Université Fédérale de Toulouse
Toulouse, France
claudie.josse@ensicacet.fr
C. Martini
Dept. of Industrial Engineering
University of Bologna
Bologna, Italy
carla.martini@unibo.it
L. Robbiola*
Laboratoire TRACES (CNRS)
Université de Toulouse
Toulouse, France
robbiola@univ-tlse2.fr
*Author for correspondence

Abstract
A two-year M-ERA.NET project – the Bronze-IMproved non-
hazardous PAtina CoaTings project (B-IMPACT) – focused on this
conservation issue with the aim of promoting innovative, eco-
friendly, and non-hazardous protection treatments for outdoor
patinated bronze. Two types of sand-cast bronzes (traditional
Cu-Sn-Zn-Pb and modern Cu-Si-Mn foundry alloys) were artificially
patinated and several protective treatments were selected by
preliminary electrochemical tests in artificial acid rain. After
patination, accelerated aging was applied to test their protective
efficiency. The visual aspect of the coated surfaces was assessed
both before and after aging, including the aesthetic impact, which
was one of the selection parameters. The health impacts of selected
coatings were also investigated by occupational hazard tests.
The behavior of these innovative treatments was compared to
uncoated samples and to a reference coating: the benzotriazole
(BTA)-containing acrylic polymer Incralac widely used for the
protection of outdoor bronze monuments. A full assessment
of the protectiveness of the selected coatings was performed
using focused ion beam-scanning electron microscopy (FIB-SEM)
cross sections down to the nanometric scale and top-surface
investigation by x-ray photoelectron spectroscopy (XPS) analysis. A
specifically designed adhesion 3-point bending test was also used.
For the patinated Cu-Sn-Zn-Pb bronze, a hydrophobic
fluoropolymer was identified as the best-performing coating in
comparison to several other candidates, such as sol-gel polymers
or organosilane coatings. For the modern Cu-Si-Mn alloy, black-
patinated according to the 19th-century traditional recipe,
a silane-based coating (PropS-SH) was selected from different
options. These included silane-based coatings with different
functional groups, such as amino, thiol, and tetrasulfide groups, or
long aliphatic chains. Finally, guidelines for the application of the
best-performing coatings on bronze statues were also developed
and issued during the project, thanks to interaction with restorers
involved in the optimization of application procedures on the real
cast bronze objects produced for this purpose.
Keywords
bronze, organic coatings, atmospheric corrosion,
electrochemical techniques, x-ray photoelectron spectroscopy
(XPS), coating process, focused ion beam/field emission
gun-scanning electron microscopy (FIB/FEG-SEM),
occupational hazard tests

Introduction
Bronze exposed outdoor is sensitive to pollutants but also Robbiola et al. 2014, Masi et al. 2016a and 2017a,). Until
to rainfall exposure geometry, changing surface compo- now, protective treatments have mostly been developed
sition and forming different patinas (Chiavari et al. 2011, by considering only the original metallic surfaces and
PROTECTION OF OUTDOOR BRONZES WITH ECO-FRIENDLY AND NON-HAZARDOUS COATINGS BASED ON SILANE AND FLUOROPOLYMERS: RESULTS FROM THE B-IMPACT PROJECT 223

not the composition of the natural or artificial patinas. to coating application), and (iii) an end user test aimed
Furthermore, the most widely used coatings, such as at optimizing the application method as well as real
Incralac (whose formulation may include toxic solvents or environment exposure and preparation of the coating
compounds such as BTA) (Health Council 2000, Pillard et formulation for industrial production.
al. 2011), do not meet national and European regulations
which increasingly demand the use of environmentally
sustainable products. Few steps have yet been taken in
this regard in the conservation field.

To that end, within the framework of the two-year

European B-IMPACT project (www.b-impact.cloud/ ),
new bronze-improved non-hazardous patina coatings
were developed and assessed specifically with regard to
improving the conservation of outdoor bronzes. Two
representative substrates were considered in particular:
(i) a traditional cast bronze (Cu-Sn-Zn-Pb) “pre-pat-
inated” by aging techniques that accurately simulated
Figure 1. B-IMPACT project: global procedure
natural corrosion processes involving decuprification, as
demonstrated in real exposure conditions (Chiavari et al.
Substrate preparation
2011, Masi et al. 2016a, 2017a, and 2017b); (ii) a modern
silicon bronze (Cu-Si-Mn) used for contemporary art Casting
and black-patinated following a standard recipe still
A traditional quaternary Cu-Sn-Zn-Pb and a modern
in use today (K2S aqueous solution) (Masi et al. 2016a,
ternary Cu-Si-Mn bronze (Si-bronze), representative
2017b, 2017c, and 2018). For each coating/substrate
of statuary alloys, were obtained by sand casting from
system, a series of tests was conducted to evaluate a wide
Livartis d.o.o foundry (Slovenia) as 6 mm thick bronze
range of coating properties: protective efficiency (both
plates. Their composition is shown in Table 1. The two
by electrochemical tests and accelerated aging, so as to
as-cast alloys had a typical dendritic pattern of α-Cu solid
fully understand degradation mechanisms), aesthetic
impact (by color measurements), practical adhesion solution, with (α + d) eutectoid within the interdendritic
and cohesion, and occupational hazard associated with spaces in the case of the quaternary bronze (as detailed
coating application (by analyzing the volatile organic in Masi et al. 2017a and 2018).
compounds (VOCs) emitted and evaluating the exposure
Table 1. Bronze composition in weight% (balance Cu)
limit). Incralac was used in each step of the assessment
as a reference for performance comparison. The best Bronze Cu Sn Zn Pb Si Mn Others
(wt%)
two coatings for each of the representative substrates
Traditional base 6.9 ± 3.1 ± 2.0 ± Al, P, Mn, and
were then selected. A silane-based coating for modern quaternary 0.6 0.4 0.9 Si (< 0.1%)
K2S-patinated bronze was sent to end users (i.e., restorers, Modern base 3.1 ± 0.9 ± Sn and Zn (< 0.1%)
art foundries, sculptors) for feedback. Both selected ternary 0.4 0.1 P (0.02%)

coatings were also applied to real bronze sculptures and

are still undergoing outdoor exposure as part of long-
term testing.
Methods of study
The B-IMPACT operational scheme is depicted in
Figure 1. It involves the following steps: (i) substrate
preparation (including casting, patination, and coating
application), (ii) coating assessment (anticorrosion effi-
ciency, aesthetic impact, and occupational hazard due
Patination
Traditional Cu-Sn-Zn-Pb bronzes were aged to simulate
natural patinas: (i) a “pale green patina” (Robbiola et al.
2008) by Dropping testing (runoff rain, time of wetness
(ToW): 37.3 days, Figure 2) and a “dark patina” by wet
& dry (W&D) testing (stagnant rain, ToW: 87 days).
The composition of the artificial acid rain (AR) solution
used for both aging methods (Table 2) is representative
of winter acid rains in Bologna (Italy).
224  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PROTECTION: COATINGS

Table 2. Characteristics and composition of artificial acid rain

solution (AR)

pH 4.35
SO 4
2-
1.90 mg L-1
Cl
-
1.27 mg L-1
NO3- 4.62 mg L-1
CH3COO- 0.23 mg L-1
HCOO -
0.05 mg L-1
NH4+ 1.05 mg L-1
Ca 2+
0.34 mg L-1
Na+ 0.53 mg L-1

Dropping test is representative of more severe conditions,

thus only the results obtained by this aging method are
reported and discussed here.

Figure 3. Surface of patinated bronzes (optical and SEM surface

observation, followed by FIB-SEM cross section with EDS maps): (a) aged
Cu-Sn-Zn-Pb after dropping test and (b) K2S-black patinated Cu-Si-Mn

the patinated Cu-Sn-Zn-Pb (Masi et al. 2017a) showed

Figure 2. A) General view of the implemented Dropping test system
developed specifically for large-scale sample production, with a detail
(bottom right) of the bronze coupon position (45°) exposed under
dropping nozzles coupled to an automatic drain system; B) Surface
aspect of cross-cut coated specimens after 14 days in an alternated acid
rain spray cabinet: (a) PropS-SH and (b) PropS-SH + β-CD-MPT complex
For the modern Si-bronze, the black patina used in
modern art and produced by the “liver of sulfur” tech-
nique was applied in the foundry using a standard proce-
dure. After sand-blasting, the bronze surface was heated
using a torch, reaching temperatures of up to 50°C. An
aqueous solution of K2S (3 wt%) was then applied by
brushing and rinsed off with water.
Both bronzes had a patina which was a few micrometers
thick (Figure 3) and whose structure and composition
are detailed by Masi et al. (2016a, 2017a, 2017b, 2017c,
and 2018). In-situ FIB-SEM cross-section images of
decuprification of the bronze surface, which is in line
with natural patinas found on ancient outdoor statues
(Robbiola et al. 2008). This mainly affects the α-Cu solid
solution (dendrite core), anodic towards the eutectoid,
forming a thin layer of cuprous oxide (including poorly
crystallized tin (hydroxy-)oxide) together with posnjakite,
identified by Raman spectroscopy and grazing angle
x-ray diffraction (GXRD) (Masi et al. 2016a and 2017a).
As regards the K2S-patinated Si-bronze, the patina was
rough and irregular. It consisted of a multilayer structure
with locally decuprified areas (relatively Si-O enriched),
including thin layers of cuprous oxide and copper sulfide
(Cu1.96S and Cu2S) (Masi et al. 2016a, 2017b, and 2018).
Coating selection and application
Promising candidate coatings were first selected from
different types of polymers: silanes with/without an
inhibitor or particle additions, sol-gel spray, fluoropol-
ymer, and multilayer polyelectrolyte coatings. Commer-
cial Incralac (thermoplastic methyl methacrylate copol-
PROTECTION OF OUTDOOR BRONZES WITH ECO-FRIENDLY AND NON-HAZARDOUS COATINGS BASED ON SILANE AND FLUOROPOLYMERS: RESULTS FROM THE B-IMPACT PROJECT 225

ymer with BTA inhibitor) was used as a reference for
comparison (Masi et al. 2018).
Selected coatings were applied to bronze surfaces (both
Cu-Sn-Zn-Pb and Si-bronze) patinated (cf. §2.1.2) by
immersion, spraying, or brushing (Table 3).
Coatings assessment
Electrochemical tests
The protective efficiency of coatings was assessed by
both short (1 h–24 h) and long (30-day) electrochemical
tests in concentrated artificial acid rain (tenfold concen-
tration at 30°C). The electrochemical cell consisted of a
polymethylmethacrylate tube with an internal diameter
of 20 mm and height of 80 mm, which was glued onto
50 × 50 × 5 mm uncoated or coated patinated bronze
coupons by an epoxy adhesive. Inside this tube, filled with
ARx10, a Pt wire and a saturated calomel electrode (SCE)
were used as counter and reference electrodes, respectively,
while the bronze specimen was the working electrode. The
tests were performed by a PAR/AMETEK PARSTAT2273
potentiostat/galvanostat. The inhibiting efficiency (IE%)
was calculated with the following equation (1):
icor0 — icor
IE% = x 100 (1)
icor0
where icor0 is the value for the uncoated sample and icor
is the corrosion current density for coated surfaces. The
corrosion current densities are obtained by Tafel extrap-
olation from ohmic-drop-compensated I-E polarization
curves. The IE% values after a 30-day immersion in
ARx10 are reported in Table 3 for coatings revealing a
tangible corrosion inhibition.
Accelerated aging
Accelerated aging was carried out with the same Drop-
ping test used for the patination of Cu-Sn-Zn-Pb (§2.1.2).

Table 3. Inhibiting efficiencies (IE%) from corrosion current values obtained by polarization curves in ARx10 after 30 days of immersion for coatings
on dropping-patinated traditional Cu-Sn-Zn-Pb bronze (1) and K2S-black patinated Si-bronze (2)

IE % IE%
Additional information Application method
1 2
INCRALAC® (INC) Acrylic polymer (with BTA and anti-UV) Single dip coating / spray 89 / - 83 / 89
Commercial reference
Fluoropolymers FA: Fluoroacrylate
2MS-FA + MS: copolymer methyl methacrylate + methacryloxy- Brushing 92
propyl-trimethoxysilane
LbL-FA-2MS Idem but LbL: Layer-by-layer application Layer by layer 100
FA-MS FA + MS Brushing 100 40
FA-MS-BS + BS: Triethoxysilylpropyl-5-benzotriazole)urea Brushing 100 10
LbL-FA-2MS-BS Idem but LbL Layer by layer 98
Sol-gel coatings Silica particle pretreated with
HMFS-GPTMS-8 HMS: hexamethyldisilazane Spray 69 40
+ GMPTS: 3-glycidyloxypropyl trimethoxysilane (GPTMS)
HMFS-GPTMS-3 + Zirconium (IV° butoxide) Spray 55 80
HMFS-GPTMS-MTMS-2 + MTMS: methyltrimethoxy silane + Zr(IV) butoxide Spray 69 80
Organosilanes Room temperature reticulating siloxane polymers
PropS-SH 3-mercapto-propyl-trimethoxysilane Single dip coating / spray 62 / 98 82 / 97
PropS-SH + Fa + FA: Class F Fly ashes Single dip coating / spray 68 / 99 75 / 83
PropS-SH + La2O3 La2O3: Lanthanum oxide Single dip coating 50 82
PropS-SH-DT DT: Double dip coating Double dip coating 94
PropS-SH* 3-mercapto-propyl-trimethoxysilane Single dip coating 67
PropS-TS 3-(triethoxysilyl)propyl] tetrasulfide Single dip coating 70
OctadecS Trimethoxy (octadecyl)silane Single dip coating 90
PropS-SH + MPT + MPT: 5-Mercapto-1-phenyl-tetrazole (inhibitor) Spray 78
PropS-SH + MPT+β-CD + β-CD: β-cyclodextrin complex (inhibitor) Spray 99 80
Multilayer coatings
POLI 3 combination Chitosan + Poly(acrylic acid sodium salt) Layer by layer ≈ 50
226  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PROTECTION: COATINGS

Table 4. Selected candidate coatings tested by accelerated aging on (1) quaternary bronze patinated by dropping and (2) K2S-black patinated
Si-bronze

Coating reference 1 2
Incralac® Incralac® Acrylic polymer (with BTA and anti-UV) X X
Commercial reference
Organosilanes
3-Mercapto-propyl-trimethoxy-silane PropS-SH OCH3 X
+Addition of inhibitors to improve + Fa X
anticorrosion behavior + β-CD + MPT Si OCH3 X
HS
OCH3

Trimethoxy(octadecyl)silane OctadecS X

Sol-gel coating
HMFS-GPTMS-MTMS-2 Sol-gel X

Fluoropolymers
HMFS-GMPTS-MTMS-2 FA-MS F F F CH3 X
F F F F H3C O
F F F F
F F O Si O
F F
F F F F
F F CH3
F F F F
F F
F F
F F F
O
O
O O
O O
O O

9n n
O O O O
CH3 CH3 CH3
+ (primer MS) CH3

Only the best-performing coatings (Table 4) were used,
selected on the basis of electrochemical tests.
The Dropping test consisted of dropping AR for periods
of time onto the samples placed at a 45° angle on plastic
supports and periods of sample drying. In particular,
cycles of 3 days of dropping/1 day of drying and 2 days
of dropping/1 day of drying were alternated for two
weeks, for a ToW of 10 days per sample. The runoff
solution was separately collected from each sample and,
at the end of each dropping period, volume, pH, and
concentration of released metals were determined. The
concentration of alloying metals in the solutions was
analyzed by atomic absorption spectrometry (AAS). The
IE in terms of Cu release (hCu) was calculated using the
following equation (2):
Msol,NC — Msol,C
( )
ηCu = × 100
Msol,NC
where Msol is the amount of Cu in the aging solution
(µg cm-2), NC is the uncoated sample, and C is the coated
sample.
A range of techniques was applied on coated samples
before and after dropping: optical microscopy (OM); field
emission gun scanning electron microscopy (FEGSEM)
with energy-dispersive x-ray spectroscopy (EDS) and
FIB milling for cross-section observation; micro-Raman
spectroscopy; XPS; and GXRD. Full experimental details
can be found in Masi et al. (2017a and 2018).
Color measurements
Coated samples were characterized before and after
accelerated aging by measuring color coordinates in
the CIELab colour space (Datacolor D400 spectropho-
tometer, D65 illuminant, 10° observer, d/0 measuring
geometry, specular component excluded, measured area
(2)
of ∅ 6.6 mm). Total color change (∆E*) in the range
between 3 and 5 is generally considered acceptable in
PROTECTION OF OUTDOOR BRONZES WITH ECO-FRIENDLY AND NON-HAZARDOUS COATINGS BASED ON SILANE AND FLUOROPOLYMERS: RESULTS FROM THE B-IMPACT PROJECT 227

the field of cultural heritage, while ∆E* > 5 is considered tion and the corresponding limit for each VOC detected.
clearly perceivable (Graziani et al. 2014). To be compliant, the EL of a mixture must be < 1.

Adherence tests Results

Additionally, in order to gain further information on Electrochemical tests
coating protection limits, before and after a 28-day
The final IE% values for all the tested coatings are
immersion in ARx10, adherence (practical adhesion)
reported in Table 3. Regarding the patinated Cu-Sn-
tests were performed on coated bare or K2S-patinated
Zn-Pb substrate, charged silane coatings (especially of fly
Cu-6Sn bronze samples. For this reason, a tailored
ash (Fa) and β-CD-MPT complex, in both cases IE% = 99)
method was applied based on 3-point bending (Aufray
and all tested fluoropolymers (IE% = 100) performed
et al. 2019).
better than Incralac (IE% = 89%). In particular, the
addition of the β-CD-complex can induce a self-healing
Occupational hazard tests
effect in coating defects (as assessed by exposing cross-cut
Occupational exposure tests for the candidate coatings coated specimens to an acid rain spray test, Figure 2B).
(Table 4) were performed to guarantee the health and
For coated K2S-black patinated bronze, electrochemical
safety of restorers. The potential toxicity of VOCs
tests showed that, under the same thickness, sprayed
released during the coating application was assessed in
plain PropS-SH is more efficient than Incralac: IE% = 97
particular. The same amount (500 mL) of each coating,
and 89, respectively. On this substrate, no additive can
together with 30% diluted Incralac in nitro diluent, was
improve the performances of PropS-SH coating. Other
applied by brushing on a 70 × 70 cm inert glass plate
silanes (e.g. OctadecS) and two sol-gel coatings (HMFS-
on both surfaces, for a total exposure time of 1 h (test
GPTMS-3 and HMFS-GPTMS-8) offer good corrosion
in triplicate).
protection (IE% = 90 and 80, respectively) comparable
VOCs released during the test were sampled through to that offered by Incralac (IE% = 83 or 89, according to
a device positioned at operator-collar-bone height the application method).
(sampling rate 0.5 L/min, for a total volume of 30 L
of air) and analyzed using gas chromatography (GC) Accelerated aging
with flame ionization detection following standardized
The evolution of Cu release in the weathering solutions
methodologies (UNI EN 689:1997 “Workplace atmos-
as a function of ToW and the ability of the coatings to
pheres – Guidance for the assessment of exposure by
limit metal release (hCu, eq. (2)) are shown in Figure 4
inhalation to chemical agents for comparison with limit
for traditional Cu-Sn-Zn-Pb and for patinated Si-bronze.
values and measurement strategy” and UNI EN 16017-
1:2002 “Indoor, ambient and workplace air – Sampling
and analysis of volatile organic compounds by sorbent
tube/thermal desorption/capillary gas chromatography
– Pumped sampling”). Then, considering a standard
working day, the 8 hour-weighted average concentra-
tion for each VOC was calculated and compared with
the corresponding threshold limit value-time weighted
average (TLV-TWA), i.e., the airborne concentration
of a compound to which workers can be exposed on
a daily basis for a working lifetime without adverse
effect, assuming an average exposure of 8 hours/day,
40 hours/week) established by the American Confer- Figure 4. Cumulative Cu release as a function of time of wetness (ToW)
and inhibiting efficiency (hCu) in terms of Cu release at 10 days of ToW
ence of Governmental Industrial Hygienists (ACGIH)
for uncoated and coated patinated (a) Cu-Sn-Zn-Pb and (b) Si-bronze
(Tedesco, Sperotto, and Benetti 2017). As coatings release
a mixture of VOCs, the exposure limit (EL) was calculated All tested coatings showed a great improvement in
by summing up the ratios among calculated concentra- corrosion protection, reducing Cu release by roughly
228  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
two orders of magnitude with respect to uncoated
samples. After 10 days of wetness, the uncoated patinated
Cu-Sn-Zn-Pb and Si-bronze samples showed a total
Cu release of about 1600 mg/cm-2 and 1400 mg/cm-2,
respectively, indicating the marked impact of dropping
on uncoated surfaces. For Incralac, a slightly higher but
more scattered Cu release value was obtained. Thus, a
slight improvement in corrosion protection was observed
for the selected coatings compared to Incralac.
Detailed information on the protection efficiencies of
plain and inhibited silane and fluoropolymer coatings
are available in Masi et al. (2016b, 2017b, and 2018),
Monticelli et al. (2017), and Kosec et al. (2019) and in
Masi et al. (2017c) and Kosec et al. (2018), respectively.
Color measurements
Color variation (∆E*) values are shown in Figure 5. Both
uncoated patinated substrates exhibited a marked change
in appearance after aging (∆E* = 12 for Cu-Sn-Zn-Pb
and ∆E* = 9 for Si-bronze).
Figure 5. Color variation (∆E*) induced by the accelerated aging on
coated patinated (a) Cu-Sn-Zn-Pb and (b) Si-bronze
Conversely, all the coated samples showed a very slight
color variation (∆E* < 3), indicating no appreciable
visual alteration of the coatings due to accelerated aging.
The exception was OctadecS on patinated Si-bronze
(∆E* > 10), which, in spite of its high level of protection,
was unsuitable on this substrate.
Adherence
Figure 6 shows the machine-specific assembly for prac-
tical adhesion tests. Different types of failure (adhesive
before and cohesive after aging) were characterized,
as illustrated for the PropS-SH coating in Figure 6
(bottom).
PROTECTION: COATINGS
Figure 6. (Top) Specimen device diagram (3-point test) and machine-
specific assembly for practical adhesion measurement. (Bottom) Optical
(surface observation) and microscopic observation (FEGSEM section) of
the cohesive failure of silane coating (PropS-SH) on black patinated Si-
bronze after aging for 28 days in concentrated artificial acid rain
In the case of the K2S-patinated Cu-6Sn bronze test speci-
mens, cohesive failure within the patina always occurred.
The full methodological approach and results are given
in Aufray et al. (2019). Carrying out accurate practical
adhesion measurements, taking into account the full
“coating-patina-alloy” system, is fundamental in order
to obtain a proper evaluation of the protective behavior
of selected coatings for the conservation of ancient and
historic metals.
Occupational hazard tests
As regards occupational hazard for restorers in term of
VOC inhalation, all the new selected coatings were safe
(EL<1) and performed better than Incralac, whether
undiluted or 30% diluted (EL = 1.98 or 1.1, respectively).
In summary, according to the EL values (Table 5), tested
coating formulations can be ranked as follows, from the
safest to the most hazardous:
PropS-SH > (PropS-SH + b-CD + MPT) ~ (PropS-SH +
Fa) > OctadecS > FA-MS >> Incralac
Table 5. Exposure limit values obtained for the tested coating formu-
lations
Coatings EL
Incralac® 1.982
Incralac® (30%) 1.108
PropS-SH 0.019
PropS-SH+Fa 0.050
PropS-SH+β-CD+MPT 0.045
FA-MS 0.301
PROTECTION OF OUTDOOR BRONZES WITH ECO-FRIENDLY AND NON-HAZARDOUS COATINGS BASED ON SILANE AND FLUOROPOLYMERS: RESULTS FROM THE B-IMPACT PROJECT 229

End user test Feedback was collected using an online questionnaire

(Figure 9). Evaluation with grading was performed.
The main results obtained in the coating assessment step
are summarized in Figures 7 and 8.

Figure 9. The B-IMPACT Test Kit (bronze object, kit box, online
questionnaire) sent to selected end users

Very good feedback was received on the coating perfor-

mance. Only a few comments were registered regarding
the smell during application and about the visual appear-
Figure 7. Comparison of coating performance on patinated quaternary
bronze
ance. During the two-year project, therefore, the develop-
ment of the B-IMPACT coatings reached a breakthrough,
advancing from technology readiness level (TRL) 2 to 5.

While the B-IMPACT project has finished, the two

selected coatings continue to be monitored for long-
term durability in real environments. A “test park” was
established in front of the Slovenian National Building
and Civil Engineering Institute (ZAG). Protected sculp-
tures were provided by Livartis d.o.o. and it is planned
to expose them to an outdoor environment for at least
a two-year period after completion of the project. The
preliminary results observed in urban outdoor conditions
after one year are very positive (Figure 10).

Figure 8. Comparison of coating performance on black patinated Si-

bronze

Taking into account all the assessment parameters,

the results identified the best coating for each pati-
nated substrate: FA-MS for aged traditional bronze and
PropS-SH for K2S-patinated modern bronze. Application
tests were then performed by the end users (i.e., conser-
vators, art foundries, and sculptors) on two scultures
exposed to an outdoor environment. The feedback from
these tests was included in the project as a preliminary
step in preparing the coatings for industrial production.

An application kit, which was sent to selected end users

in different countries, contained the silane-based coating, Figure 10. Coating application for long-term exposure to urban outdoor
a small bronze object and instructions for application. conditions
230  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
As regards the removability of the coatings, there are some
suggestions that fluoropolymers can be easily removed
by non-polar solvents. A complementary study has to be
performed on this topic as well as on re-applicability of
the coatings, in order to fulfill all the requirements for
full acceptance in conservation applications.
Conclusion
All of the selected coatings in the B-IMPACT project
can be considered non-health hazardous for operators.
During the project, according to a methodology based
on a set of complementary investigation methods, two
protective coatings were selected for their protective
efficiency and cultural heritage compatibility: FA-MS
fluoropolymer for traditional naturally prepatinated
quaternary bronze, and PropS-SH organosilane for
black-patinated Si-bronze. The study identified two very
efficient coatings for the protection of outdoor patinated
bronzes at a pre-industrial level of readiness. Some inves-
tigations are still underway, regarding both long-term
in-situ surface evolution monitoring and removability
of these coatings on bronze statues.
Acknowledgments
We would like to thank the support provided by national
funding organizations (MIZS-Slovenia, MIUR-Italy,
RMP-France). We would also like to express our gratitude
to the following for their contribution to the B-IMPACT
project: B. Kamšek from Livartis d.o.o. (Slovenia); David
Perez and Frédéric Aymard from C2M Aurochs Industrie
(France); Marie-Laure Desse and Loïc Marchin from
PYLOTE SAS Company (France); Marija Babnik, Marko
Kete, Teja Koršič, Irena Oblak, and Urška Lavrenčič
Štangar from GEIDA Company (Slovenia); and Federico
Benetti, Andrea Camporese, Francesco Fusco, Maria
Girotto, Giorgia Merlin, and Lucia Valente from ECAM-
RICERT S.R.L. and LIVARTIS (Italy). Our thanks also to
the end users for their professional interest and helpful
comments: Anne-Marie Geffroy, Clarisse Fabre and
Nicolas Parc (Fonderie de Bronze Lauragaise), Lorenzo
Morigi, and Antoine Amarger.
References
Aufray, M., et al. 2019. Practical adhesion measurements
of protective coatings on bronze by 3-point bending test.
Journal of Coatings Technology and Research (accepted
for publication).
PROTECTION: COATINGS
Chiavari, C., E. Bernardi, C. Martini, L. Morselli, F.
Ospitali, and L. Robbiola. 2011. Predicting the corrosion
behaviour of outdoor bronzes: Assessment of artificially
exposed and real outdoor samples. In Metal 2010:
Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, Charleston, South Carolina,
11–15 October 2010, eds. P. Mardikian, C. Chemello,
C. Watters, and P. Hull, 218–26. Clemson, SC: Clemson
University.
Graziani, L., E. Quagliarini, F. Bondioli, and M. D’Orazio.
2014. Durability of self-cleaning TiO2 coatings on fired
clay brick façades: Effects of UV exposure and wet & dry
cycles. Building and Environment 71: 193–203.
Health-based recommended occupational exposure limit.
2000. The Hague: Health Council of the Netherlands,
Publ. No. 2000/14OSH.
Kosec, T., L. Škrlep, E. Švara, F. Adrijana, S. Škapin, G.
Masi, E. Bernardi, C. Chiavari, C. Josse, J. Esvan, and
L. Robbiola. 2019. Development of multi-component
fluoropolymer based coating on simulated outdoor
patina on quaternary bronze. Progress In Organic Coatings
131(6): 27–35.
Masi, G., J. Esvan, L. Robbiola, and C. Chiavari. 2016a.
Corrosion protection of outdoor bronze: Historical
quaternary alloy and modern alloy with artificial patina-
tion. In EUROCORR 2016. Advances in Linking Science to
Engineering, Montpellier, France, 11–15 September 2016,
volume: Proceedings (USB device), paper no. 57297.
Masi, G., C. Chiavari, J. Avila, J. Esvan, S. Raffo, M.C.
Bignozzi, M.C. Asensio, L. Robbiola, and C. Martini.
2016b. Corrosion investigation of fire-gilded bronze
involving high surface resolution spectroscopic imaging.
Applied Surface Science 366: 317–27.
Masi, G., J. Esvan, C. Jossec, C. Chiavari, E. Bernardi,
C.Martini, M.C. Bignozzi, N. Gartner, T. Kosec, and
L. Robbiola. 2017a. Characterization of typical patinas
simulating bronze corrosion in outdoor conditions.
Materials Chemistry and Physics 200: 308–21.
Masi, G., C. Martini, E. Bernardi, M.C. Bignozzi, C.
Monticelli, A. Balbo, F. Zanotto, M. Aufray, C. Josse, J.
Esvan, L. Robbiola, and C. Chiavari. 2017b. Protective-
ness of different silane coatings on patinated ancient
and modern bronzes: Part 2. In EUROCORR 2017 and
20th International Corrosion Congress and Process Safety
Congress 2017. Corrosion Control for Safer Living, Prague,
Czech Republic, 3–7 September 2017, paper no. 95722.
PROTECTION OF OUTDOOR BRONZES WITH ECO-FRIENDLY AND NON-HAZARDOUS COATINGS BASED ON SILANE AND FLUOROPOLYMERS: RESULTS FROM THE B-IMPACT PROJECT 231

Masi, G., C. Josse, J. Esvan, E. Bernardi, C. Martini, M.C.

Bignozzi, L. Škrlep, E.S. Fabjan, T. Kosec, M. Aufray, L.
Robbiola, and C. Chiavari. 2017c. Micro-characterisation
of innovative organic coatings applied for the protection
of outdoor bronze monuments. In EUROCORR 2017 and
20th International Corrosion Congress and Process Safety
Congress 2017. Corrosion Control for Safer Living, Prague,
Czech Republic, 3–7 September 2017, paper no. 95587.
Masi, G., C. Josse, J. Esvan, C. Chiavari, E. Bernardi, C.
Martini, M.C. Bignozzi, C. Monticelli, F. Zanotto, A.
Balbo, E. Švara Fabjan, T. Kosec, and L. Robbiola. 2018.
Evaluation of the protectiveness of an organosilane
coating on patinated Cu-Si-Mn bronze for contemporary
art. Progress In Organic Coatings 127: 286–99.
Monticelli, C., V. Grassi, G. Mavilia, F. Zanotto, and
A. Balbo. 2017. Entrapment of corrosion inhibitors in
silane coatings to improve bronze corrosion protection.
In EUROCORR 2017 and 20th International Corrosion
Congress and Process Safety Congress 2017. Corrosion
Control for Safer Living, Prague, Czech Republic, 3–7
September 2017, paper no. 82386.
Pillard, D.A., J.S. Cornell, D.L. DuFresne, and M.T
Hernandez. 2001. Toxicity of benzotriazole and benzotri-
azole derivatives to three aquatic species. Water Research
35: 557–60.
Robbiola, L., K. Rahmouni, C. Chiavari, C. Martini, D.
Prandstraller, A. Texier, H. Takenouti, and P. Vermaut.
2008. New insight into the nature and properties of pale
green surfaces of outdoor bronze monuments. Applied
Physics A 92: 161–69.
Robbiola, L., C. Chiavari, E. Bernardi, M.C. Bignozzi,
and C. Martini. 2014. General understanding of outdoor
bronze corrosion – An overview. In Open Air Metal.
Outdoor Metallic Sculpture from the XIXth to the early
XXth century. Identification, conservation, restoration,
15th SFIIC Study Days, Paris, 4–5 December 2014, eds.
A. Azéma and A. Tixier, 146–53.
Tedesco, E., W. Sperotto, and F. Benetti. 2017. B-IMPACT
Deliverable 3.3 – Assessment of TLW/TWA and STEL.

Author
Luc Robbiola is a CNRS research engineer in cultural
materials science. He has a Ph.D. in metallurgy and an
M.A. in conservation-restoration.
The Use and Mis-use of Renaissance Wax
David Thickett* Bethan Stanley
English Heritage English Heritage
London, UK London, UK
david.thickett@english-heritage.org.uk bethan.stanley@english-heritage.org.uk
www.english-heritage.org.uk *Author for correspondence
Abstract
A large ongoing study of coating protective performance has
identified issues with using Renaissance Wax in some situations.
Previous research has determined that Renaissance Wax used
in saline, high relative humidity (RH) internal environments can
actually increase steel corrosion rates. Cold Renaissance Wax
coated steel has been exposed to different RH environments
and surfaces doped with a salt mixture. Enhanced corrosion was
observed at high RH values and with the salt mixture present.
Introduction
Renaissance Wax was formulated at the British Museum
for indoor use. It avoided the corrosion issues with the
plant- and insect-based waxes it was designed to replace;
several instances of copper stearate associated with using
these waxes were analysed. Its use was extensively tested
for the museum’s environment, generally with relative
humidity (RH) levels only occasionally reaching 70% for
short periods and little airborne salt, due to its central
London location. As with many ‘accepted’ conservation
materials, its use has spread to other situations. A large
study of coating protective performance for iron and
steels is being undertaken. Initial results have indicated
some issues with the use of Renaissance Wax in certain
circumstances, and these are reported in this work.
Previous research has determined that Renaissance
Wax used in saline, high RH internal environments can
actually increase steel corrosion rates. Initially there is
a protective effect, but in the environment studied this
was lost after three months and then steel corroded at
a greater rate. Analysis showed the wax did not provide
any kind of barrier against chloride solutions and actu-
ally appeared to hold liquid water at the metal surface
(Thickett 2015).
Other authors have reported poor performance in
uncontrolled maritime environments (Argyropoulos
Analysis showed the wax held liquid water at the metal surface.
Its use over alkyd paints outdoors was assessed and the wetting
and contact angle measured. Steel under alkyd paint coated with
the wax corroded more rapidly than uncoated paint. Heating
above 70°C was found to interfere with the alkyd steel bond.
Keywords
Renaissance Wax, indoor, outdoor, steel
et al. 2005). Little systematic research has been reported
otherwise.
Good levels of protection have been reported anecdo-
tally in more ambient indoor environments without salt
present. The effect of RH and salt contamination has
been assessed.
Renaissance Wax is frequently recommended for outdoor
use, although no reported research to underpin this could
be found. English Heritage cares for a large number of
cannon in maritime environments. Most have had alkyd
paints applied. Many of the cannon are on loan and the
owners, along with many contractors, specify annual
coating with Renaissance Wax. The protective effect of
the wax applied cold on bare metal and painted steels
has been investigated.
Improved properties have been reported with hot wax
application and this is sometimes recommended for
outdoor painted surfaces. However, thermal shock is a
recognized mechanism for paint failure.
The corrosion rate of clean and salt-doped cold Renais-
sance Wax-coated steel exposed to different RH envi-
ronments has been measured. The protective effect of
the cold wax coatings outdoors on bare steel and paint
systems was investigated. The impact of heating paint
THE USE AND MIS-USE OF RENAISSANCE WAX
surfaces to apply the wax was measured with pull-off
testing.
Methods and materials
Corrosion under high RH and salt conditions
Low-carbon steel coupons (20 by 30 by 0.5 mm) were
embedded in epoxy (Araldite MY753/ hardener HY951)
resin blocks, with the coupon exposed at one surface. A
wire was soldered to the back of the coupon and extended
out of the block. After the resin had set, the coupon
surface was abraded with silicon carbide paper (1200
grade), rinsed with acetone, hot air dried and stored
above dry silica gel. When dry, the coupons were coated
with Renaissance Wax applied with a hogs’ hair brush
and then burnished with a lint-free cloth. Some of the
wax-coated pieces were doped with a modified artificial
seawater solution.
Levels of salt deposition were determined after the anal-
ysis of salt deposition onto glass slides for 12 months at a
range of English Heritage sites. The slides were extracted
with 18.2 MΩ/cm water and the ion concentrations
analysed with a Dionex DX-600. For anions, a AS14A
column with 8 mM sodium carbonate and 18 mM sodium
bicarbonate eluent was used. Cations were analysed with
a IonPac CS16 column with 30 mM methanesulfonic acid
eluent. The range and interquartile ranges are included
in Figure 1 for maritime (within 2 km of the coast) and
non-maritime sites.
Figure 1. Salt ion deposition rates
Vanishingly low levels of potassium, fluoride and stron-
tium were detected in the deposited material from
maritime English Heritage sites. It is likely the particle
formation, transport and deposition mechanisms exclude
these species. The standard artificial sea water solution
(ASTM International 2013) was then modified to exclude
potassium chloride, potassium bromide, sodium fluoride
and strontium chloride. The surface chloride level was
set to match the upper interquartile range of the marine
233
sites measurements shown in Figure 1. The solution
concentrations (g/kg) were: 23.926 sodium chloride,
4.008 sodium sulfate, 0.196 sodium bicarbonate, while
magnesium chloride and calcium chloride were added as
solutions to give 0.05327 and 0.01033 mol/kg. The solu-
tion was pipetted onto the surfaces with a micropipette.
The coupon blocks were exposed in well-sealed glass
vessels (250 ml, Bernardin Mason) above glycerol solutions
producing RH values of 50–95% in 5% intervals for six
months. There were four vessels per RH environments,
one each for the control, coated, doped, coated and then
doped coupons, with five coupons in each vessel. After
exposure, the coupons were analysed with XRD, with a
Phillips 1830/1840 diffractometer with Co tube operating
at 40 V and 40 mA, to minimise fluorescence. Three of the
coupons were then stripped of the wax and cathodically
stripped with a Uniscan PG580 potentiostat in a solution
of 7.5% sodium hydroxide, 2.5% sodium sulfate and 7.5%
sodium carbonate (ASTM International 1999). The fourth
and fifth coupons were analysed with FTIR, using both
direct reflection and a Germanium ATR head on a Nicolet
Inspect IR microscope to determine water on, and below,
the surface (Thickett 2015). One of these coupons was
then placed in a sealed container with silica gel (Prosorb)
at 40% for three months. This RH was selected, as it is the
minimum RH normally experienced in historic house
environments with conservation heating.
Further attempts were made to better localise the water
present below the surface of the wax using FTIR micros-
copy and UV laser ablation. Wax layer depth is also
critical for performance and a calibration was developed
for FTIR microscopy. Renaissance Wax was compacted
loosely onto a steel coupon, at different depths. Multiple
sheets of copper foil (Goodfellow) with a nominal 0.5 µm
thickness were stacked (Figure 2).
After six weeks drying, the wax films were checked with a
profilometer. The nominal copper sheet depth minus the
profilometer reading gave the wax depth at each point.
The films were analysed with direct reflectance on the
Nicolet FTIR microscope. A calibration for wax depth
was developed using the absorption band at 1450 cm-1.
The waxed coupons exposed to higher RH values,
and also those then dried at 40%, were ablated with
an Excimer UV laser. Each pulse was found to remove
approximately 1.1 µm of wax. A series of ablations were
undertaken to produce craters at interval depths through
234  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND PROTECTION: COATINGS

This particular sensor is well suited to coating trials as

the reference track (for temperature compensation) is
buried in the epoxy body, leaving only the measurement
track on the surface. This makes coating much easier. The
coated ECRM sensors, along with a reference uncoated
sensor, are exposed on the cliff face at Dover Castle. The
relative resistance is read quarterly, and the amount of
corrosion on the steel sensor under the coating calculated.
A Signatrol SR006 logger allows continuous (30-minute
interval) monitoring of the resistances of three sensors.
Renaissance Wax was applied cold, as above, over the
alkyd coating Cromadex 200. The uncoated reference
Figure 2. Schematic for wax layer generation and thickness measurement
electrode, the Cromadex 200 and wax-coated Cromadex
200 were monitored with the SR006 logger. The relative
the whole depth of the wax films. Each depth crater was resistances of the reference track and measurement track
then analysed by direct reflection with the Nicolet FTIR were recorded, and the corrosion loss determined, using
microscope to determine the depth of wax, and then with the equations given by Cai and Lyon (2005). The number
the Germanium ATR head for a surface analysis. of days to onset of corrosion, when the depth increased
beyond the manufacturers stated detection limit, as well
Application onto painted metal outdoors as the initial rate (over 30 days after this point) and final
English Heritage is presently running a long-term perfor- rate (over the last 30 days of the 44-month recorded
mance trial of several paint and wax systems. The systems exposure period) were calculated.
are shown in Table 1.
Surface wetness sensors (Starlog 60506) were coated
Table 1. Materials being tested
with cold Renaissance Wax. The sensors were exposed
in triplicate at an angle of 45° facing north, with resist-
Product Supplier
ance data collected with Signatrol SR002 loggers. There
PAINTS
is uncertainty in the actual resistance that corresponds
Cromadex 200 over AQ58 primer Cromadex
to wetting of a surface (Norberg 1993). Wetting was
*Hempthane HS BS 632 over Hempel
Hempadur Mastic primer calculated when the resistance dropped below 0.1, 0.2,
*Cromadex 400 Topcoat BS 632 over Cromadex 0.5 and 1.0 MΩ. For each value, the time of wetness was
395 Anti-Corrosive Primer calculated over the three-month exposure period.
*Acrolon C237 semi-gloss to BS Sherwin Williams
138C/632 Dark Admiralty Grey over The wetting angle was determined on the surface by
Macropoxy C400V3 Light grey photography. The sensors were placed horizontally on
WAXES
a rig made from microscope stage to allow fine position
Renaissance Wax Picreator Enterprises Ltd
adjustment. The rig was levelled and had vertical bars.
WaxOyl Amazon.co.uk
A Sony A600 camera was used parallel to the surface
Ankor Morris Lubricants
with a fixed protractor viewed behind the sensor. A high
Rocol Z30 Bearing Transmission and
Lubricant Supplies Ltd purity (18.2 MΩ/cm) water droplet was applied with a
Shell Ensis DW 6055 www.smithandallan.com micropipette. The surface wetting angle was estimated
* Applied by Nicola Emmerson, Cardiff University from a series of images taken with the camera every five
seconds.
Part of the trials involve coating electrical resistance
Rainfall was measured with a graduated tube, read daily.
corrosion monitoring (ERCM) based low C steel
sensors (Metal Systems ER0500 with element of G10100 Cromadex 200 was applied to pieces (approximately 500
low-carbon steel, 0.080–0.130% carbon, 0.30–0.60% by 50 by 4 mm) of a 1920s cast-iron drainpipe, blast-
manganese, < 0.050% sulfur, < 0.040% phosphorous) cleaned to Sa 2.5 with glass beads. After curing for 14
with the coating product, to manufacturers’ instructions. days, the pieces were heated in an oven to 50, 60, 70, 80
THE USE AND MIS-USE OF RENAISSANCE WAX 235

and 90°C for ten minutes and allowed to cool to room with the salt mixture, akaganeite was the major corrosion
temperature. The pull-off strength was determined with product. The wax-coated coupons give a similar spread
a Positest AT-A pull-off adhesion tester using 20 mm of corrosion products.
dollies attached with epoxy (DeFelsko AT adhesive)
FTIR spectra from the direct reflection and ATR head
according to ISO 4624 (International Standards Organ-
are shown in Figure 4.
isation 2016). The pull-off tests were undertaken in a
room heated to 20°C, as temperature has been shown
to have a strong effect on the value measured (Fletcher
and Barnes 2018).

Results
Corrosion under high RH and salt conditions
The metal loss results are shown in Figure 2. The metal
loss was calculated as microns of surface recession using
the density, the area of the coupons and mass loss deter-
mined by stripping. Figure 4. ATR spectrum of wax plus water

The differences in spectra are consistent with the two

sampling methods (Drude 1902). ATR shows stronger
relative absorption at lower wavenumbers. The direct
reflectance spectra, which penetrates the whole depth of
the wax, and is reflected from the metal surface, shows
an additional strong, broad absorption peak at 1640 cm-1.
This is characteristic of water. The ATR spectrum has an
information depth of approximately 1.77 µm at 1640 cm-1
(Harrick and du Pre 1966). This indicates water is mainly
present below the surface of the wax. A very small absorp-
tion band, probably due to water, was observed on the
ATR spectrum. The differences in interaction between the
two techniques mean the amounts cannot be determined.
Figure 3. RH and salt results
The calibration obtained from the wax and feeler gauges
The critical RH (the RH when the corrosion rate
is shown in Figure 5.
increases dramatically and corrosion is observed) for
the low-carbon steel is between 65 and 70%. Corrosion The absorption band at 1450 cm-1 was used as there are
is detected at lower RH values, but very low values – no absorption bands in this region from the common
<0.001 µm – during the 180 days of the tests. When iron corrosion products. Hematite and maghemite have
the salt mixture is present on the bare surface this is
lowered to between 55 and 60%. Wax protects the clean
metal surface up to 70% RH with no significant corro-
sion detected (<0.001 µm). The exact RH has not been
determined, but the experiments show a low corrosion
rate below 70%. At 75% and above, more metal was
corroded when the wax was present. The increase in
corrosion rate was at least 250% and up to 800%. The wax
had less effect when the salt mixture was on its surface.
Enhanced corrosion occurred from 60% RH. The corro-
sion products found on the steel control coupons were
mixtures of Goethite and lepidocrocite. When doped Figure 5. Depth calibration
236  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
bands at 1382 cm-1, and lepidocrocite and maghemite
at 1628 cm-1, but these are clearly distinguishable. Iron
carbonate does have an absorption band at 1417 cm-1,
which would interfere, but this corrosion product is rare
and was not detected by XRD (Thickett 2012).
A good calibration was obtained, validating the direct
reflection as a suitable and sensitive method to determine
Renaissance Wax depth on steel surfaces.
The Excimer laser ablation was found to remove approx-
imately 1.1 µm of wax per pulse. Analysis with the ATR
head indicated that water was mainly present in the lower
1.4–2.1 µm of the wax only. Analysis of the coupons
further exposed to 40% RH for 90 days did not reduce this
depth or the apparent overall amount of water retained.
Application onto painted metal outdoors
The results calculated from the coatings trials on ECRM
sensors are shown in Table 2.
Table 2. Corrosion rate of resistance-based sensors
Corrosion rate Corrosion rate
Onset time initial, 30 days final, last 30 days
Sample
(days) after onset of 44-month tests
(µm/year)
Control, C steel 21 + 2 50.4
Cromadex 200 598 + 43 11.2
Cromadex 200 plus 241 + 21 14.8
wax
The uncoated sensor responds relatively rapidly to the
environment. The initial onset period could be under-
stood as the time during which chloride (and other)
ions are accumulating on the steel surface. Once a crit-
ical threshold is reached, corrosion accelerates and is
detected by the sensor (Santana Rodriguez et al. 2002).
Thinner steel tracks would increase sensitivity, but seri-
ously limit the usable period of the sensors. Even with
increased sensitivity, the reported threshold may lead to
a similar onset time. The Cromadex alkyd paint system
both retards the onset time and reduces the initial and
final corrosion rates. With cold-applied Renaissance Wax
present over the paint, the onset time drops dramatically.
Although, the overall system is still quite protective
compared to bare metal, the bare metal corrosion rates are
extremely high. The initial corrosion rate is approximately
similar to that of the Cromadex. However, the corrosion
rate at the end of the monitoring period was far higher.
This may indicate the paint is failing. Further investiga-
tion is planned when the exposure tests are completed.
PROTECTION: COATINGS
The times of wetness calculated from the surface wetness
sensors are shown in Table 3.
Table 3. Surface wetness
Wetness threshold Time of wetness (hours)
(MΩ) Control Wax-coated
1.0 538.75 1065.75
0.5 500.00 1065.75
0.2 451.50 1065.75
0.1 451.75 1065.50
Varying the wetness resistance threshold gave different
values for the control (uncoated sensor). Inspection of
the data showed the initial drop in resistance was rapid
on wetting and most events dropped below the 0.1 MΩ
threshold. The major difference in the integrated values
was caused by the slow drying rate of the surface, taking
time for the resistance to rise above the various thresh-
olds. The resistance of the wax-coated sensors dropped
(but not as rapidly as the uncoated surface) as soon as
rainfall occurred, and then stayed wet for the whole expo-
sure period. The slight retardation of wetting explains
the very slightly shorter time of wetness for the lower
resistance thresholds.
(µm/year)
The wetting angle results are shown in Figure 6.
347
23.1
135.6
Figure 6. Wetting angle
The wax had an initial effect, but this had dropped to a
constant level after approximately 60 days. This value,
67°, is indistinguishable from that of the sensor surface.
The estimated error is 2°. The sensors are gold grids set
in an epoxy body and the wetting angle is mainly deter-
mined by the epoxy with has a greater surface area. This
behaviour differs from that determined with the surface
wetness sensors, which showed very rapid initial wetting
of the surface. The difference is probably attributable to
the pore structure formed in the applied wax. There are
open pores down to the metal surface, but none above
500 µm (Cano et al. 2007). So, although the wax can take
THE USE AND MIS-USE OF RENAISSANCE WAX
up some water quickly, the amount is limited due to the
very thin wax coating. This does not significantly impact
the droplet size on the surface or the apparent contact
angle. As the wax degrades, the contact angle decreases.
The tests on Cromadex 200 and wax-coated Cromadex
200 possibly showed an apparently slightly higher wetting
angle on the wax, but the difference was less than the
error. The observed differences disappeared after 30
days exposure.
Trials on cannon at Pendennis Castle confirmed the
loss of wetting angle of Renaissance Wax after 90 days
exposure.
The pull-off testing results are shown in Figure 7. The
boxes mark interquartile ranges, and the bars, the
minimum and maximum value.
Figure 7. Pull-off strength
There is a dramatic decrease in pull-off strength on
heating to 70°C. Further loss of adhesion strength is
observed when heating to higher temperatures.
Discussion and conclusion
The corrosion tests clearly show that whilst cold applied
Renaissance Wax provides a degree of protection at
lower RHs, once the RH exceeds 75%, the corrosion rate
actually increases with the wax, compared to uncoated
steel. The presence of soluble salts on the surface lowers
the threshold. As previously reported, Renaissance Wax
provides almost no protection against chloride ions. It
also appears to hold water at, or near, the metal surface.
The FTIR experiments showed a high water concentra-
tion within the lower 1.4–2.1 µm of the wax. Once this
was present, long exposure to low RH values did not
remove the water, as it would on an uncoated surface.
These experiments used only a single composition of
steel and it is possible that different responses would be
237
observed on steels of other compositions and certainly on
copper alloys. However, the results indicate Renaissance
Wax should be used with extreme caution in indoor envi-
ronments that reach 75% RH, or maritime environments
with high soluble salt concentrations.
In this work, the use of Renaissance Wax outdoors has
mainly been investigated for use on paint surfaces. On
carbon steel, it reduced the time in which corrosion
occurs under one composition of alkyd paint. It also
significantly increased the final corrosion rate, after 44
months. Such corrosion would be expected to signifi-
cantly interfere with the paint metal bond and reduce the
paint’s effective lifetime. Again, significant evidence of
water being held under the wax was shown. On a macro
level, the cold-applied Renaissance Wax did increase
the wetting angle. However, this effect was eliminated
by exposure within four months. Other authors have
reported similar results with electrochemical impedance
spectroscopy of Cosmoloid 80H wax over bronze stat-
uary, the major component of Renaissance Wax (Cano
et al. 2014). This period is well under the frequently
recommended maintenance period of 12 months and
re-application every three months is unlikely to be
feasible in most situations. It would appear that Renais-
sance Wax has at least some detrimental effects to some
paint systems and its use cold, over outdoor painted
steels, has to be questioned.
Hot application of the wax is reported to give better
results (Jáuregui-Gónzalez et al. 2016). Heating an alkyd
paint to 70°C begins to interfere with the paint metal
bond, with a significant drop in pull-off strength. It is
very difficult to control the temperature when applying
a wax. Application of molten Renaissance Wax (circa
100°C) may not cause the surface temperature to exceed
70°C for anything beyond a very short period of time.
The results indicate caution should be applied, especially
with gas torches or infra-red heaters.
The interaction of the environment with metal is often
more complex outdoors than indoors, and the reserva-
tions about a single metal composition and paint type
are even more relevant about generalising the work. The
protective nature of a coating can rely on many factors,
including:
• barrier properties against RH, rain, fog, condensation,
chloride, other pollution and oxygen;
• the working properties of the coating and how
complete a barrier layer it forms;
238  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
• the degradation and wear characteristics of the coating;
• the coating’s tolerance to surface contamination and
existing corrosion products, or remains of previous
coatings;
• the coating’s ability to physically fill porosity in mate-
rials and their corrosion products.
Only a small portion of these have been investigated
in the research presented here. The larger project will
investigate these other factors for the most promising
coatings. Good results have been reported anecdotally
with Renaissance Wax on copper alloys and further
research would be required to provide relevant advice.
References
Argyropoulos, V., Z. Sakki, A.G. Karydas, C. Zarkadas,
P. Pouli, K. Melessanaki, A. Giakoumaki, D. Anglos, and
M. Giannoulaki. 2005. Problems in stabilizing historic
iron-bladed weapons displayed in an uncontrolled envi-
ronment in the Criminal Museum of Athens, Greece. In
ICOM-CC 14th Triennial Meeting Preprints, The Hague,
12–16 September 2005, ed. I. Verger, vol. 1, 293–300.
London: James & James.
ASTM International. 1999. Standard practice for preparing,
cleaning and evaluating corrosion test specimens (ASTM
G1 E1.1). West Conshohocken, PA, USA.
ASTM International. 2013. Standard practice for the
preparation of substitute ocean water (ASTM D1141-98).
West Conshohocken, PA, USA.
Bracco, G. and B. Holst. 2013. Surface science techniques.
Berlin: Springer.
Cai, J.P. and S.B. Lyon. 2005. A mechanistic study of initial
atmospheric corrosion kinetics using electrical resistance
sensors. Corrosion Science 47(12): 2956–73.
Cano, E., D.M. Bastidas, V. Argyropoulos, and A.
Siatou. 2007. Electrochemical techniques as a tool for
testing the efficiency of protection systems for histor-
ical steel objects. In Strategies for Saving our Cultural
Heritage. Proceedings of the International Conference
on Conservation Strategies for Saving Indoor Metallic
Collections, Cairo, 25 February–1 March 2007, eds. V.
Argyropoulos, A. Hein, and M. Abdel Harith, 121–26.
Athens: TEI of Athens.
Cano, E., D. Lafuente, A. Crespo, and B. Ramirez Barat.
2014. A novel gel polymer electrolyte cell for in-situ
application of corrosion electrochemical techniques.
Electrochemistry Communications 41: 16–19.
PROTECTION: COATINGS
Drude, P. 1902. The theory of optics, 299–301. London:
Longmans Green.
Fletcher, J.F and D.J. Barnes. 2018. Pull-off adhesion
testing of coatings. https://www.elcometer.com/images/
stories/PDFs/Pull-off_Adhesion_Testing_of_Coat-
ings_Improve_Your_Technique.pdf (accessed 17 October
2018).
Harrick, N.J. and F.K. du Pre. 1966. Effective thickness
of bulk materials and of thin films for internal reflection
spectroscopy. Applied Optics 5: 1739–43.
International Standards Organisation. 2016. Paints and
varnishes – Pull-off test for adhesion (ISO 4624:2016),
Geneva, Switzerland.
Jáuregui-Gónzalez, D.A., M. López-Arriaga, F.J.
Rodríguez-Gómez, and J. Contreras-Vargas. 2016.
Influence of the application and preparation of wax
coatings artificially patined bronze surfaces. In Metal
2016: Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, New Delhi, 26–30 September
2016, eds. R. Menon, C. Chemello, and A. Pandya,
170–79. New Delhi: Indira Gandhi National Centre
for the Arts.
Norberg. P. 1993. Evaluation of a new surface moisture
monitoring system. In Durability of Building Materials
and Components. Proceedings of the 6th International
Conference, Omiya, Japan, 26–29 October 1993, eds S.
Nagataki, T. Nireki, and F. Tomosawa, 2 vols., 72–84.
London: E. & F.N. Spon.
Plenderleith, H.J. and A.E.A. Werner. 1971. The
conservation of antiquities and works of art. Treat-
ment, repair and restoration, p. 374. Oxford: Oxford
University Press.
Santana Rodriguez, J.J., F.J. Santana Hernandez, and J.E.
Gonzalez Gonzalez. 2002. XRD and SEM studies of the
layer of corrosion products on carbon steel in various
different environments in the province of Las Palmas.
Corrosion Science 44: 2425–38.
Thickett, D. 2012. Post excavation changes and preventive
conservation of archaeological iron. PhD dissertation,
School of Biological and Chemical Sciences, Birkbeck
College, University of London, UK.
Thickett, D. 2015. Corrosion protection for complex large
equipment. In Big Stuff 2015. https://bigstuff.omeka.net/
items/show/1 .
THE USE AND MIS-USE OF RENAISSANCE WAX 239

Authors
Bethan Stanley ACR is senior collections conservator
at English Heritage, UK.

David Thickett is senior conservation scientist at

English Heritage, UK.
 DEVELOPMENT OF
CONSERVATION PRACTICES
The Potential of Microorganisms for the Conservation-
Restoration of Iron Artworks

Lucrezia Comensoli Mathilde Monachon Monica Albini Edith Joseph*

Swiss Federal Laboratories for
Materials Science and Technology
(Empa)
Dübendorf, Zurich, Switzerland
lucrezia.comensoli@empa.ch
Wafa Kooli
University of Neuchâtel
Laboratory of Microbiology
Neuchâtel, Switzerland
wafam.kooli@gmail.com
University of Neuchâtel
Laboratory of Technologies for
Heritage Materials
Neuchâtel, Switzerland
mathilde.monachon@unine.ch
Institute for Nanostructured Material
(ISMN-CNR)
Monterotondo Stazione, Rome, Italy
monica.albini@ismn.cnr.it
Pilar Junier
University of Neuchâtel
Laboratory of Microbiology
Neuchâtel, Switzerland
pilar.junier@unine.ch
University of Applied Sciences
Western Switzerland, Haute Ecole Arc
Conservation-restauration
Laboratory of Technologies for
Heritage Materials, Institute of
Chemistry, University of Neuchâtel
Neuchâtel, Switzerland
edith.joseph@he-arc.ch
edith.joseph@unine.ch
*Author for correspondence

Abstract
Archaeological iron artefacts encounter serious post-excavation
problems when contaminated with salts. Once excavated,
exposure to a higher oxygen concentration and lower relative
humidity renders the corrosion crust formed during burial no
longer stable. In particular, the process is induced by chloride
ions and leads to flaking, cracking and the final loss of shape
of the object. The MAIA project (Microbes for Archaeological
Iron Artefacts) studied microbial metabolisms to explore their
potential for the development of innovative and sustainable
methods for the stabilisation of corroded iron archaeological
objects. Two different approaches were investigated. First,
bacterial reduction of iron solid-phases and biogenic mineral
formation were studied as a way to replace unstable corrosion
products. Several bacterial strains were compared. Spectroscopic
investigations with Raman and Fourier transform infrared
spectroscopy on iron coupons, nail surfaces and cross sections
demonstrated the conversion of the outermost part of the
corrosion layer into more stable biogenic minerals, such as
vivianite and siderite. The second approach was to study fungi
and their metabolic ability with iron. In particular, alkaliphile
fungi that tolerate chlorine were studied for their ability to
produce biogenic minerals and to adsorb metals in their biomass.
Colorimetric investigation and evaluation of the thickness of the
corrosion layer demonstrated that fungi are good candidates for
developing bio-cleaning methods for corroded iron, permitting
the selective removal of the unstable and powdery corrosion
layer without damaging the original metal surface. This study
details these approaches and explores the possibilities of their
exploitation for development of innovative and more sustainable
treatments for the conservation-restoration of corroded iron.
Keywords
cultural heritage, iron, corrosion, microorganisms, biogenic
minerals, stabilisation, bio-cleaning, green chemistry

Introduction
The preservation of archaeological iron artefacts faces
severe obstacles after excavation when different chlo-
rine-containing salts contaminate the corrosion crust
surrounding the object (Réguer, Dillmann and Mirambet
2007, Eggert and Schmutzler 2009, Rimmer and Wang
2010). In fact, this anion easily penetrates the corro-
sion layer reaching the interface between the corrosion
compounds and the metal core (Scott and Eggert 2009).
Its reaction with oxygen, water and iron leads to the
formation of an acidic FeCl2 solution (Selwyn 2004).
This latter, through oxidation and hydrolysis, creates
iron oxyhydroxides and hydrochloric acid. Both of these
compounds are detrimental to the objects. Indeed, iron
oxyhydroxides are three times more voluminous than
elemental iron and produce a mechanical pressure that
leads to the formation of severe cracks and ruptures.
Among iron oxyhydroxides, the formation of akaganeite
[FeO0.833(OH)1.167Cl0.167] (Turgoose 1982 and 1985, Ståhl
et al. 2003) is particularly detrimental. This mineral is
considered an unstable corrosion compound as it can
release chloride ions in the presence of humidity and
instigate further corrosion (Ståhl et al. 2003, Selwyn
2004). Hydrochloric acid, on the other hand, enhances the
corrosion of the object by decreasing the pH and reacting
with iron ions producing iron chlorides (FeCl2) (Selwyn
2004). Chlorine is then responsible for a cyclic corrosion
process that involves chemical as well as mechanical
damage to the iron objects.
THE POTENTIAL OF MICROORGANISMS FOR THE CONSERVATION-RESTORATION OF IRON ARTWORKS
In order to enhance the extraction of chloride ions, two
key factors have to be considered: whether to slow down
or stop the corrosion of the object or whether to increase
the porosity of its corrosion crusts (Scott and Eggert
2009). The former can be achieved by passivating the
iron surface with an alkaline treatment solution (Selwyn
and Argyropoulos 2005), by removing dissolved oxygen
from the treatment solution (Watkinson 1996) or by using
electrochemical methods (Carlin, Keith and Rodriguez
2001). The porosity of the corrosion products may be
increased by heating the treatment solutions, dissolving
extraneous material in alkaline solutions or reducing
some of the iron oxyhydroxides FeO(OH) to magnetite
Fe3O4 or some other lower oxidation state iron oxide or
hydroxide (Schmidt-Ott and Boissonnas 2002, Scott and
Eggert 2009).
To this end, the MAIA project proposed an innovative
alternative treatment whereby iron is stabilised by the
metabolic activity of naturally occurring microorganisms.
Two different strategies were adopted either leading to
the formation of stable iron compounds or to the removal
of chloride ions, instigators of post-excavation corro-
sion (Réguer, Dillmann and Mirambet 2007). The first
approach involves the use of iron-reducing bacteria able,
under anoxic conditions, of inducing the precipitation of
stable iron oxides of low molar volume, such as magnetite
Fe3O4 (Bharde et al. 2006, Bazylinski and Schübbe 2007,
Javaherdashti 2008).
The second approach involves the use of halogenophilic
fungi that need chlorinated species to grow and are
naturally found in extremely saline environments (Ma
et al. 2010). Within the framework of the MAIA project,
iron accumulation by fungi was exploited to remove
powdery corrosion layers from the iron objects selectively
(bio-cleaning).
In both approaches – bacterial biomineralisation or
fungal accumulation – a careful assessment of the meth-
odology was carried out over iron- and chloride-rich
phases, respectively. The validation of these approaches
was carried out on steel coupons weathered for 12 months
and iron nails from the late Roman period (3rd century
CE). The study aimed to evaluate bacteria’s capacity
to reduce iron species from iron(III) compounds, in
particular akaganeite, present as solid phases and corro-
sion layers on coupons/nails. The idea was to produce a
homogeneous layer of chemically stable minerals while
also allowing the diffusion of the potentially present
243
chlorinated species outside the corrosion layer. In parallel,
fungal strains were tested for the removal of powdery
iron corrosion compounds with the aim of applying them
during maintenance programmes on outdoor iron-based
(i.e. steel) sculptures.
Material and methods
Microbes and growth conditions
The following bacteria were evaluated: Desulfitobacterium
hafniense strains LBE and TCE1, Shewanella loihica PV-4
and Aeromonas sp. strains CA23 and CU5. Their growth
conditions are described in previous studies (Comensoli
et al. 2017b, Comensoli et al. 2018, Kooli et al. 2019).
Concerning fungal species, Beauveria bassiana was
investigated for the production of biogenic minerals a)
after one-month incubation in a solid medium amended
with malt extract 1.2 %, agar 1.5 % and iron(III) citrate
10 mM or iron(II) chloride 10 mM in deionised water,
at neutral pH or buffered at pH 9.2 (NaHCO3 0.765 % +
NaCO3 0.106 %), and b) on corroded coupons after one,
two and four weeks of incubation in a liquid medium
(malt extract 1.2 %, buffered at pH 9.2). Alternaria sp.
was employed for the bio-cleaning assay with the same
liquid medium. Abiotic controls were performed with
immersion of coupons in a liquid medium without fungi.
Corroded iron substrates
Corroded steel coupons were obtained from iron plates
of 50 × 50 × 2–3 mm size presenting either typical urban
or marine corrosion. The two corroded surfaces were
obtained after exposing the plates for one year to an
outdoor environment in Zürich and at the French Insti-
tute of Corrosion (Brest, France), respectively (Table 1).
Archaeological iron nails dating from the late Roman
period (3rd century BCE, Table 1) were excavated from
chalky soil in the Champagne region, France. Before
treatment, non-destructive Fourier transform infrared
(FTIR) spectroscopy and Raman spectroscopy investi-
gations were performed directly on the surface of the
coupons and nails to characterise the molecular compo-
sition of the corrosion layer.
Samples were then sterilised by spraying with ethanol
70% and UV exposure (20 minutes each side) followed
by autoclaving (120°C for 20 minutes). After treatment,
the coupons and nails were washed with ethanol 70%.
244  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Table 1. Corroded iron substrates used for bio-stabilisation and as vivianite Fe2+3(PO4)2·8H2O. However, the culture
bio-cleaning methods. The chemical composition was determined media required for working with these strict anaerobic
by FTIR and Raman spectroscopy analyses performed directly on the
surface before treatment
bacteria are complex and their composition directly
affects the chemistry of the biogenic minerals (Roh
Iron Strains used Dimensions Chemical
substrates (f-fungi) (mm) composition
et al. 2003). Thus, in addition to bacterial iron reduc-
D. hafniense LBE lepidocrocite tion, an abiotic reaction with Na2S (added to maintain
Urban-aged D. hafniense TCE1 goethite anoxic conditions) was also observed and resulted in
12.5 × 25 × 2-3 lipids
coupons B. bassiana (f) the production of poorly crystallised mackinawite
(contamination)
Alternatia sp. (f) (Fe 2+/Fe 3+S) and mineral sulfur (α-S 8) (Comensoli
S. loihica 10 × 10 × 2-3 lepidocrocite et al. 2017b). Therefore, a further development of
Marine-aged goethite
Aeromonas sp. CA23 the method used facultative anaerobes and a reduc-
coupons 10 × 5 × 2-3 chlorinated
Aeromonas sp. CU5
aerosols tion in the controlled chemical matrix. In this case,
D. hafniense TCE1 lepidocrocite marine-aged corroded coupons presenting chlorine-rich
Roman-period
Aeromonas sp. CA23 5 × 5 × 25-40 goethite corrosion products were treated with the halophilic
nails
Aeromonas sp. CU5 bacterium S. loihica. The treatment was effective after
one week as more stable Fe(II) minerals were formed
Equipment without any by-products from abiotic reaction. After
two weeks of treatment, the colour of the coupons
An iS5 Thermo Scientific spectrometer with a diamond
turned progressively greyish green. Two crystalline
attenuated total reflectance (ATR) crystal plate (iD5 ATR
phases were identified as vivianite and siderite FeCO3,
accessory) was used to obtain spectra of 4000–650 cm-1,
respectively. Vivianite is considered a chemically stable
4 cm-1 resolution, 32 scans. Data collection and post-run
iron compound (Volkland et al. 2000). Siderite has a
processing was conducted with Omnic software.
smaller molar volume. This particularity can facilitate
A Minolta CM-508D spectrophotometer was used as the removal of chlorides from the objects through an
follows: specular component included (SCI), illuminant enhanced diffusion in a treatment solution (jellified or
D65, d/8° geometry, 10° observer, 8 mm diameter (Ø) not) (Kooli et al. 2018). Nevertheless, using a halophilic
measurement area, illumination with Xe flash light source bacterium is not ideal as it requires the addition of NaCl
100% or 0% UV, CIELab 1976 colour space, duplicated for optimal growth. A third trial was then performed
measurements. using two non-halophilic environmental bacteria
(Aeromonas sp. strains CA23 and CU5). The colour of
A Surfix Pro S coating thickness gauge with an FN1.5/90°
the coupons changed from a rusty aspect to dark grey
right angle connected probe was employed as follows:
after two weeks of treatment and crystalline phases of
5 mm Ø measurement sensor, 1± 0.1 μm precision,
vivianite and siderite were identified. In addition, there
triplicated measurements.
was a promising stabilisation of the archaeological iron
nails using a gel delivery system (Kooli et al. 2019).
Results and discussion
The long-term behaviour of the newly formed Fe(II)
Stabilisation of corroded iron surfaces with bacteria biogenic minerals was unsatisfactory as a regrowth of
corrosion was observed after 15 days of storage in an
Reductive dissolution of iron by strict and facultative
uncontrolled environment. Further optimisation studies
anaerobic bacteria was successfully achieved to convert
for the application and/or adaptation of the storage
unstable corrosion of iron surfaces into more stable
conditions is needed.
biogenic minerals (Comensoli et al. 2017a, Comensoli et
al. 2017b, Kooli et al. 2018). A comparative overview of
Stabilisation of corroded iron surfaces with fungi
the results obtained with the different species is presented
in Table 2. The halotolerant and alkalitolerant fungus B. bassiana
produced biogenic minerals only from iron(III) citrate at
With D. hafniense, the corroded coupons turned grey-
a neutral pH (Figure 1). These minerals were assumed to
blue after one week of treatment. Crystalline phases
be iron oxalates (FeOx) due to their rosette-like habitus
(~ 100–150 µm Ø, rosette-like habitus) were identified
(similar to copper oxalates) and to the fact that the same
THE POTENTIAL OF MICROORGANISMS FOR THE CONSERVATION-RESTORATION OF IRON ARTWORKS 245

Table 2. Comparison of the results obtained with different bacterial treatments for the stabilisation of corroded iron surfaces (black scale represents
10 mm and red scale 5 mm)

D. hafniense LBE S. loihica Aeromonas sp. CA23

D. hafniense TCE1 Aeromonas sp. CU5
Application
Immersion Immersion Immersion or gel patch
method
Treatment
1 week 2 weeks 4 weeks
duration
Colour after Grey – blue Greyish green Grey
treatment

before after

before after
before after
Biogenic minerals Vivianite Vivianite and siderite Vivianite and siderite

Remarks Abiotic reduction of iron due to Na2S Need of chloride ions supply. No abiotic reduction
present in the growth medium. No abiotic reduction. with CU5, mineral formation already
Vivianite already observed after 1 week observed after 2 weeks.
(green colour of coupons). No chloride ions supply.

fungal species is known to produce hyphae sheathed with source was needed as B. bassiana was not able to form
iron oxalate crystals (Joseph et al. 2011). FeOx from iron(II) chloride.

Nonetheless, B. bassiana was incubated with corroded

iron coupons in alkaline conditions. Treated coupons
had a similar colour to the untreated coupons and abiotic
controls. However, small differences were observed as
black spots on the abiotic controls and treated coupons
(Figure 2).

Figure 1. Biogenic minerals produced by B. bassiana after one month

of incubation in a solid malt extract (1.2%) and agar (1.5%) medium
amended with iron(III)-citrate at neutral pH. Red arrows: crystals with
rosette-like habitus and brown colour, black arrows: fungal mycelium

Interestingly, crystal formation was not observed in the
medium buffered at pH 9.2. This was probably due to
the presence of Na ions in NaHCO3 and Na2CO3 which
interacted with the secreted oxalic acid, avoiding the
formation of iron oxalates. In addition, an iron(III)
Figure 2. From left to right: visual appearance of the corrosion layer
of untreated coupons, abiotic controls and coupons treated with B.
bassiana
FTIR investigations on all the abiotic control samples
revealed residues of culture medium with the presence
246  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
of malt (maltose at 2980 and 2900 cm-1, carbohydrates at
1060 cm-1) and the production of siderite (peak observed
at 1394 cm-1), likely as the result of the interaction with
the carbonate-containing buffer (Figure 3a). On the other
hand, FTIR spectra recorded on treated coupons differed
depending on the treatment time (Figure 3b). After one
week, malt (as in the abiotic control) and lepidocrocite
(mainly 891 and 787 cm-1) were observed. The analyses
carried out after four weeks revealed the presence of resi-
dues of growth medium (2970, 2914 and 1057 cm-1) as
well as fungal proteins (amide I at 1645 cm-1, amide II at
1539 cm-1) and lepidocrocite (892 cm-1). The presence of
proteins suggested the production of extracellular poly-
meric substances (EPS), as reported in the literature (Jiao
et al. 2010). This is supported by the fact that experimental
evidence from previous work (Comensoli et al. 2017a)
reported that this fungus is able to produce small aggre-
gates of EPS-chelating iron when exposed to 10 mM FeCl3.
Figure 3. ATR-FTIR spectra recorded on untreated coupons (black) as well
as on coupons a) immersed in medium without fungi (abiotic control)
and b) treated with B. bassiana, for one week (blue), two weeks (green)
and four weeks (red). Principal peaks attributed to lipids, lepidocrocite
(Le), carbohydrates (medium) and proteins.* Absorbance peak can be
attributed to siderite
Finally, contrary to the previously published report on
B. bassiana (Joseph et al. 2011), it was not possible to
identify the presence of iron oxalates (1627 cm-1) with
certainty, as the broad absorbance peak of amide I (at
1645 cm-1) may overlap the one for oxalates. Further
investigations on this aspect are required.
Bio-cleaning of corroded iron surfaces with fungi
The potential of Alternaria sp. for bio-cleaning of
corroded iron was assessed on iron coupons after one,
two and four weeks. After treatment, abiotic control
DEVELOPMENT OF CONSERVATION PRACTICES
coupons presented a colour similar to untreated coupons
(Figure 4a). In contrast, the colour of the coupons incu-
bated with Alternaria sp. changed drastically. In fact,
the corrosion layer was almost entirely removed after
four weeks, revealing the metallic substrate (Figure 4b).
This was confirmed by the shifting of the colour coor-
dinates towards green (less red), as well as a higher
lightness L* value (towards white) for the treated surfaces
(Figures 4c–d).
Figure 4. Visual appearance of the coupons a) immersed in a medium
without fungi (abiotic control) and b) treated with Alternaria sp. c) and
d) respective colorimetry plots representing lightness L* (bright/dark)
and colour coordinates a* (red/green) and b* (yellow/blue). Symbols
used: untreated samples (black), 1-week (blue), 2-week (green) and
4-week (red) treatments
The removal of corrosion products was probably the
consequence of the iron uptake/adsorption abilities of
Alternaria sp. Eddy current measurements on treated
samples revealed a decrease in thickness of the corro-
sion layer by more than half with respect to the abiotic
controls (Figure 5).
This encouraging result revealed the potential of fungi for
the development of bio-cleaning methods for corroded
iron surfaces, i.e. contemporary sculptures with bare
iron-based alloys but also historical objects where the
thickness of the corrosion layer needs to be reduced
down to the original surface. In particular for archaeo-
logical objects, the cleaning methods currently used are
sandblasting or electrolytic reduction. These techniques
remove part of the corrosion layer mechanically, but can
also alter engravings present on the metallic surface.
In contrast, fungal melanin will smoothly capture iron
THE POTENTIAL OF MICROORGANISMS FOR THE CONSERVATION-RESTORATION OF IRON ARTWORKS
Figure 5. Thickness of the corrosion layer measured on untreated, abiotic control and treated coupons (Alternaria sp. for one, two and four weeks).
The mean value for three different measurements is shown
from the corrosion layer (oxidised forms of iron). Fungal
bio-cleaning could also be useful with fragile objects
with a thin corrosion layer and/or fragile metallic parts
presenting a non-cohesive structure (when little of the
metal is still present).
Conclusion
During the last decade, different model organisms as well
as new isolated strains have been investigated for their
potential to develop stabilisation methods for corroded
iron surfaces. Our results demonstrated that bacterial
iron reduction of the unstable iron corrosion compounds
into more stable biogenic minerals is a promising strategy
that needs to be investigated in more detail. Fungus B.
Bassiana, when exposed to iron, produced preferentially
EPS-chelating iron. However, this could have another
application as a bio-cleaning method. This study presents
the first bio-cleaning assay carried out with the fungus
Alternaria sp. that allowed the removal of corrosion
products from iron surfaces. This encouraging result
revealed the potential of fungi for bio-cleaning, allowing a
softer approach compared to mechanical cleaning of thin
corrosion layers and/or fragile non-cohesive structures.
Acknowledgements
We would like to thank the Swiss National Science Foun-
dation (Ambizione grant PZ00P2_142514, P.I. Dr Edith
Joseph) for funding the MAIA project (2013–2016). The
authors are also grateful to the Swiss National Museum
for providing the urban-aged iron coupons and to Art
Metal Conservation S.A.R.L. for the archaeological iron
nails, as well as to the French Institute of Corrosion for
the exposure of steel coupons in a marine environment.
References
Bazylinski, D.A. and S. Schübbe. 2007. Controlled
biomineralization by and applications of magnetotactic
247
bacteria. In Advances in applied microbiology, eds.
A. Laskin, S. Sariaslani, and G. Gadd, vol. 62, 21–62.
Amsterdam: Elsevier.
Bharde, A., D. Rautaray, V. Bansal, A. Ahmad, I. Sarkar,
S.M. Yusuf, M. Sanyal, and M. Sastry. 2006. Extracel-
lular biosynthesis of magnetite using fungi. Small 2(1):
135–41.
Carlin, W., D. Keith, and J. Rodriguez. 2001. Less is
more: Measure of chloride removal rate from wrought
iron artifacts during electrolysis. Studies in Conservation
46: 68–76.
Comensoli, L., S. Bindschedler, P. Junier, and E. Joseph.
2017a. Iron and fungal physiology: A review of biotech-
nological opportunities. In Advances in applied microbi-
ology, eds. A. Laskin , S. Sariaslani, and G. Gadd, vol. 98,
31–60. Amsterdam: Elsevier.
Comensoli, L., J. Maillard, M. Albini, F. Sandoz, P. Junier,
and E. Joseph. 2017b. Use of bacteria to stabilize archae-
ological iron. Applied and Environmental Microbiology
83(9): e03416–e03478.
Comensoli, L., J. Maillard, W.M. Kooli, P. Junier, and E.
Joseph. 2018. Soluble and solid iron reduction assays
with Desulfitobacterium hafniense. Bio-protocol 8(17):
e3002.
Eggert, G. and B. Schmutzler. 2009. Lässt sich die
Konservierung von Eisenfunden auf Standard bringen?
kulturGUTerhalten: Standards in der Restaurierung-
swissenschaft und Denkmalpflege. U. Peltz and O. Zorn.
Mainz, Zabern, V.: 91–95.
Javaherdashti, R. 2008. Microbiologically influenced
corrosion: An engineering insight. Springer.
Jiao, Y., G.D. Cody, A.K. Harding, P. Wilmes, M. Schrenk,
K.E. Wheeler, J.F. Banfield, and M.P. Thelen. 2010.
Characterization of extracellular polymeric substances
from acidophilic microbial biofilms. Applied and Envi-
ronmental Microbiology 76(9): 2916–22.
248  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Joseph, E., S. Cario, A. Simon, M. Wörle, R. Mazzeo, P.
Junier, and D. Job. 2011. Protection of metal artifacts with
the formation of metal-oxalates complexes by Beauveria
bassiana. Frontiers in Microbiology 2(270): 1–8.
Kooli, W.M., T. Junier, M. Shakya, M. Monachon,
K.W. Davenport, K. Vaideeswaran, A. Vernudachi, I.
Marozau, T. Monrouzeau, C.D. Gleasner, K. McMurry,
R. Lienhard, L. Rufener, J.-L. Perret, O. Sereda, P.S.
Chain, E. Joseph, and P. Junier. 2019. Remedial treat-
ment of corroded iron objects by environmental Aero-
monas isolates. Applied and Environmental Microbiology
85(3): e02042–18.
Kooli, W.M., L. Comensoli, J. Maillard, M. Albini, A. Gelb,
P. Junier, and E. Joseph. 2018. Bacterial iron reduction
and biogenic mineral formation for the stabilisation of
corroded iron objects. Scientific Reports 8(1): 764.
Ma, Y., E.A. Galinski, W.D. Grant, A. Oren, and A.
Ventosa. 2010. Halophiles 2010: Life in saline environ-
ments. Applied and Environmental Microbiology 76(21):
6971–81.
Réguer, S., P.E. Dillmann, and F. Mirambet. 2007. Buried
iron archaeological artefacts: Corrosion mechanisms
related to the presence of Cl-containing phases. Corrosion
Science 49(6): 2726–44.
Rimmer, M. and Q. Wang. 2010. Assessing the effects of
alkaline desalination treatments for archaeological iron
using scanning electron microscopy. British Museum
Technical Research Bulletin 4: 79–86.
Roh, Y., C.L. Zhang, H. Vali, R.J. Lauf, J. Zhou, and T.J.
Phelps. 2003. Biogeochemical and environmental factors
in Fe biomineralization: Magnetite and siderite forma-
tion. Clays and Clay Minerals 51(1): 83–95.
Schmidt-Ott, K. and V. Boissonnas. 2002. Low-pressure
hydrogen plasma: An assessment of its application on
archaeological iron. Studies in Conservation 47: 81–87.
Scott, D.A. and G. Eggert. 2009. Iron and steel in art:
Corrosion, colorants, conservation. London: Archetype
Publications.
Selwyn, L. 2004. Overview of archaeological iron: The
corrosion problem, key factors affecting treatment, and
gaps in current knowledge. In Metal 04: Proceedings of
the International Conference on Metals Conservation,
National Museum of Australia, Canberra, 4–8 October
2004, eds. J. Ashton and D. Hallam, 294–306. Canberra,
ACT: National Museum of Australia.
DEVELOPMENT OF CONSERVATION PRACTICES
Selwyn, L.S. and V. Argyropoulos. 2005. Removal of
chloride and iron ions from archaeological wrought iron
with sodium hydroxide and ethylenediamine solutions.
Studies in Conservation 50(2): 81–100.
Ståhl, K., K. Nielsen, J. Jiang, B. Lebech, J.C. Hanson,
P. Norby, and J. van Lanschot, 2003. On the akaganéite
crystal structure, phase transformations and possible role
in post-excavational corrosion of iron artifacts. Corrosion
Science 45(11): 2563–75.
Turgoose, S. 1982. Post-excavation changes in iron antiq-
uities. Studies in Conservation 27(3): 97–101.
Turgoose, S. 1985. The corrosion of archaeological iron
during burial and treatment. Studies in Conservation
30(1): 13–18.
Watkinson, D. 1996. Chloride extraction from archae-
ological iron: Comparative treatment efficiencies. In
Archaeological conservation and its consequences, eds. A.
Roy and P. Smith, 208–12. London: International Institute
for Conservation.
Authors
Lucrezia Comensoli is a post-doctoral associate at the
Swiss Federal Laboratories for Materials Science and
Technology (Empa), Zurich, Switzerland, where she is
studying the potential of microbes for the degradation
of extremely resistant synthetic polymers. She has a PhD
from the University of Neuchâtel, Switzerland. She has
studied the interaction of microbes with metals (copper,
iron and silver) and has contributed to the development
of several biotechnological approaches for the stabilisa-
tion and/or cleaning of copper-based surfaces as well as
corroded iron. Her greatest interest is to find innovative
solutions to problems by using microbial metabolisms.
Wafa Kooli is a post-doctoral collaborator at the labo-
ratory of microbiology of the University of Neuchâtel.
She has a PhD in microbiology, having investigated the
use of bacteria for the stabilisation of corroded iron. Her
interests are mainly focused on the study of the interac-
tions between microbes and metals from a fundamental
and applied point of view.
Mathilde Monachon has a master’s in material chem-
istry from the University of Rennes, France. She is
currently a PhD student at the University of Neuchâtel
where she is developing a green extraction method for
iron and sulfur species to preserve waterlogged archae-
ological wood and is working on the conservation of
THE POTENTIAL OF MICROORGANISMS FOR THE CONSERVATION-RESTORATION OF IRON ARTWORKS 249

metal artefacts. Her main skills are in analytical chemistry

related to conservation science.

Monica Albini is a conservation scientist specialising in

the study of metal artefacts. After working with Italian
institutions and museums, she moved to Switzerland
where she obtained a PhD in Sciences from the Univer-
sity of Neuchâtel. Her work focused on biotechnology
applied to the development of eco-friendly conserva-
tion treatments for copper-based alloys. In 2018, she
received funding from the European Union’s Horizon
2020 research and innovation programme under the
Marie Sklodowska-Curie Action on the conservation
of oil paintings on metal substrates at the Institute for
Nanostructured Material (ISMN-CNR) in Rome, Italy.

Pilar Junier is a professor at the University of Neuchâtel.

She has a PhD in microbial sciences from the University
of Chile. Her main research interest is microbial ecology,
using terrestrial and aquatic ecosystems as model systems.
She is working simultaneously on basic and applied
science. She has published over 60 peer-reviewed articles.

Edith Joseph is an assistant professor at both the

University of Neuchâtel and the University of Applied
Sciences Haute Ecole ARC Conservation-restauration in
Neuchâtel. She has a PhD in chemistry from the Univer-
sity of Bologna, Italy. Her main research activities are
the application of green chemistry for the preservation
of artistic and archaeological objects. The characteri-
sation of heterogeneous matrixes and the interaction
between organic substances and inorganic compounds,
in particular micro-organism-metals, are some of her
research interests. She has published some 50 papers on
analytical chemistry and conservation science in inter-
national journals and books.
Desalination of Archaeological Iron Objects: Comparing
the Effectiveness of Sodium Hydroxide Treatments
Britta Schmutzler* Zsolt Revay
Erfurt University of Applied Sciences Technical University of Munich
Erfurt, Germany Garching, Germany
Britta.schmutzler@fh-erfurt.de zsolt.revay@frm2.tum.de
Abstract
The results of the research project “Besonderes Eisern bewahren”
are outlined. Batches of archaeological iron nails (100 g) were
desalinated using aqueous sodium hydroxide solutions:
0.5 mol/L at 70°C; 0.5 mol/L aided by an ultrasonic field (80 kHz)
at room temperature; 0.5 and 2.0 mol/L in a polyethylene (PE)
box with a tight fitting lid at 25°C; and 0.5 mol/L in an unlidded
PE box covered with PE film at 25°C. Prompt gamma activation
analysis was used to estimate the residual chloride within
nail batches following their desalination (n = 128). Statistical
analysis of the data set is provided. All desalination strategies
reduced the chloride ion concentration of the iron to mean
Introduction
Role of akaganeite in the corrosion of iron
Archaeological iron objects are unstable following exca-
vation due to the presence of chloride ions within them,
atmospheric oxygen and humidity. Chloride ions diffuse
into the objects during their burial (attracted by Fe2+ due
to corrosion) and are often concentrated in spot areas
(Zucchi et al. 1977). After excavation, the local concen-
tration of chloride and Fe2+ ions is then high enough to
favor akaganeite formation (Rémazeilles and Refait 2007).
Akaganeite causes severe macroscopic damage such as
spalling and cracking of the objects’ corrosion layers
due to its needle-shaped crystal growth (Selwyn 2004).
Akaganeite contains chloride ions inside its tunnel struc-
ture to stabilize its crystal and adsorbs chloride ions onto
its surface (Ståhl et al. 2003). This leads to variable chloride
content according to growth conditions (Keller 1969,
Cornell and Giovanoli 1990). Cornell and Schwertmann
(2003) reported that akaganeite can be dissolved in acids
and transformed by dissolution-precipitation reactions in
alkaline solutions heated to 70°C. The adsorbed chloride
can be washed away by water, leaving a stable crystal
structure containing chloride not lower than 1 wt%)
(Réguer et al. 2009, 2801). Desalination experiments
Christian Stieghorst *Author for correspondence
Technical University of Munich
Garching, Germany
christian.stieghorst@frm2.tum.de
values of 200–300 ppm. Post-treatment oxygen consumption
measurements of selected nails were carried out at 100% and
0% humidity to determine their responsiveness to corrosion. The
mean oxygen consumption of nail batches treated with sodium
hydroxide was lower than that of untreated nails.
Keywords
archaeological iron, desalination, sodium hydroxide
solution, prompt gamma activation analysis (PGAA), oxygen
consumption, residual chloride
performed by Watkinson and Emmerson (2017) using
water showed rapid removal of the adsorbed chloride
in akaganeite (chloride content after desalination of
2–4 wt%), leading to a significantly reduced ability to
corrode iron since less adsorbed chloride can attract less
humidity from the atmosphere to create an electrolyte
(Turgoose 1982, Watkinson and Lewis 2005, Thickett
et al. 2008). Dialysis with water of an akaganeite sample
lowers its chloride content even further, down to 0.34 wt%
(Snow et al. 2011). However, complete desalination of iron
objects requires—beneath the extraction of water-soluble
compounds—the dissolution of akaganeite, achievable by
alkaline solutions (Keller 1969, Cornell and Giovanoli
1990). The chloride-containing phase β-Fe2(OH)3Cl is a
precursor for akaganeite formation and is more readily
desalinated from objects (Kergourlay et al. 2010) than
akaganeite, which is not decomposed by immersion
in 0.5 mol/L sodium hydroxide at room temperature.
Whereas Keller (1969) reported heated 0.5 mol/L sodium
hydroxide solutions dissolved akaganeite, Schmutzler
(2012) found that, following immersion of akaganeite
in 2 mol/L sodium hydroxide solution at 70°C, Raman
spectra analysis identified mainly goethite (the same
results were obtained with 0.1 and 0.5 mol/L NaOH
DESALINATION OF ARCHAEOLOGICAL IRON OBJECTS: COMPARING THE EFFECTIVENESS OF SODIUM HYDROXIDE TREATMENTS 251

solutions at 70°C). Akaganeite powder samples containing
chloride (ca. 2–5 wt%) cause corrosion of iron coupons at
100% RH after 36 days (Schmutzler 2012). This synopsis
reveals contrasting data on the stability and desalination
of akaganeite, leading to the conclusion that its dissolution
is a preferred outcome of a desalination treatment.
Aim of the research project
Since desalination treatments ideally aim to extract all
chloride from archaeological iron, the measurement
of residual chloride inside the treated objects is the
linchpin for the correct evaluation of a desalination
treatment. While it is possible to determine destruc-
tively the chloride content of iron after its digestion
(Schmutzler and Eggert 2011), a non-destructive test
would make it possible to measure chloride as often as
needed during treatment. Selwyn et al. (2006) suggested
using neutron activation analysis (NAA) to monitor the
chlorine content of iron objects, and Watkinson et al.
(2014) produced a promising pilot study of ten Roman
nails using prompt gamma activation analysis (PGAA)
to record the location and distribution of chloride along
the length of each nail.
This study used PGAA to detect chlorine before and
after desalination of archaeological iron nails using
a range of aqueous NaOH treatment solutions. To
analyze the corrosivity of the chloride ion residues,
oxygen consumption tests were conducted by exposing
each treated object to a high humidity atmosphere in
a closed container and monitoring the oxygen within
the vessel to provide an indicator of the level of stability
achieved by the treatment (Matthiesen and Wonsyld
2010, Watkinson et al. 2019).
Experimental
Desalination experiments
Sample material
Archaeological nails from the Roman settlement site
Güglingen in southwest Germany were used as samples.
Since desalination is a diffusion-controlled process,
the corrosion crust determines the speed and extent of
chloride ion removal (Réguer et al. 2007). The nails were
x-rayed for documentation purposes and to identify the
presence of a metallic core then cleaned by air abrasive
to remove loose corrosion crust. They were packed into
100 g batches according to their archaeological record
number assigned on site, ready for batch desalination
treatment.
Desalination treatments
The variables in the experiments were temperature,
ultra-sonic waves, covering and concentration of the
solution. Table 1 records the treatment conditions for
each batch of nails. The reference solution in this study
is 0.5 mol/L sodium hydroxide (NaOH) at 25°C in a
polyethylene (PE) container with a plastic lid. This
treatment set up was earlier reported as an effective
desalination treatment (Schmutzler and Eggert 2011,
Schmutzler 2012). For every experimental condition,
batches of 100 g iron objects were desalinated in 1 L
solutions. A three weeks’ treatment per bath at room
temperature was adopted as standard (Schmutzler

Table 1. Experimental conditions with case number, concentration of treatment solution, temperature, covering, number and duration of baths and
desalination treatment.

Obj. Variable and treatment No. of Time per

Conc. OH- Temp. Duration of desalination
no. conditions baths bath
20 2 mol/L 25°C Plastic box with lid 3 3 weeks 9 weeks
(airtight)
17 0.5 mol/L 70°C Stainless steel container 3 24 h 3 days
(non-airtight) 5-hour heating each day, cooling down overnight
24 0.5 mol/L 70°C Ultrasonic agitation open container 3 24 h 3 days
(non-airtight) 5-hour ultrasonic treatment each day, cooling
down overnight
20 0.5 mol/L 25°C PE film stretched over open plastic 3 3 weeks 9 weeks
box
(airtight)
35 0.5 mol/L 25°C Plastic box with lid 3 3 weeks 9 weeks
(airtight)
252  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
2012). The treatment solution was monitored for
chloride content until there was little (5 ppm) or no
chloride in the third bath solution, at which point the
desalination was stopped. After desalination, the objects
were rinsed in four baths of deionized water until the
pH was neutral and were then dried for 48 hours at
80°C in a climate chamber.
Ultrasonic support of desalination has been suggested
by Burshneva and Smirnova (2010), in whose study a
0.5 mol/L NaOH solution was agitated with ultrasonics
of 80 kHz frequency combined with a sweep function,
which raised the bath temperature to 70°C after two
hours. Additionally, heated solutions of 0.5 mol/L NaOH
at 70°C without ultrasonic waves were investigated for
comparison. In daily work, sometimes archaeological
objects do not fit into standard boxes, so an airtight
cover cannot be produced. Consequently, covering with
PE film is a common practice, but it is unclear whether
this has any influence on the ingress of oxygen or effec-
tiveness of the desalination. Consequently, this covering
method was tested for 0.5 mol/L NaOH solution.
Data acquisition in the neutron beam and
chloride analysis
The PGAA facility at the Heinz Maier-Leibnitz Zentrum
was used to measure the local chlorine concentration in
discrete areas of individual iron objects before and after
treatment. PGAA uses gamma radiation for the acti-
vation of the neutrons inside an element’s atomic core.
During and due to this energy input the neutrons emit
element-specific radiation whose intensity is propor-
tional to the mass of the element; the data is measured in
wt% and given as ppm chlorine. The detection limit was
calculated at 0.1–10 µg for chlorine (Heinz Maier-Leib-
nitz Zentrum et al. 2015). The pre-measurement took a
minimum of 30 minutes/sample and the post-measure-
ments 180 minutes and more. Révay (2009) shows the
high precision and reproducibility of the PGAA method
for qualitative and quantitative analysis in general. In
the chosen experimental design, reproducible data
related to relocating the area of the nail that was initially
analyzed before treatment is reported.
Ideally, the nails were max. 10 cm long and up to 0.5 cm
thick. To focus the neutron beam on a nail (activated
area is ca. 2 × 3 cm), it was held in an aluminum frame
using PTFE threads and its mounting position docu-
mented photographically. The areas chosen for analysis
DEVELOPMENT OF CONSERVATION PRACTICES
on an individual nail typically had a corrosion spot at
their center. For the post-treatment measurement, the
photograph was used to mount the nail in the same
position.
Mostly, one measurement was taken on each object;
only in few cases two corrosion spots were analyzed.
Therefore, this measurement of chloride records the
impact of the desalination treatment on a specific
area of a nail. All in all, 128 measurements were taken
(n = 128). Since PGAA delivers single values for each
nail, these values were summarized by calculating the
mean chloride content for each treatment batch and
then comparing the means to evaluate the relative
desalination effectiveness of treatments.
To compare the pre- and post-treatment chloride
concentration, the mean of the residual chloride content
(RCC) in mg/kg of a treated batch was compared to the
mean of a control group with no treatment. Since the
deviations of the single RCC values around the mean
were very high due to use of archaeological material
leading to an unclear picture, a significance test was
carried out with a significance level of 5% probability
of error. Statistical analysis was conducted by ANOVA
analysis (mixed design for pretest-posttest comparison
with different desalination treatments including control
group), which takes the deviations of the single values
into account. Outliers were defined as samples with
a chloride concentration of less than 200 ppm prior
to desalination or more than two standard deviations
(1660 ppm) above the recorded mean. This led to a
total number of n = 103 cases with 12 or more nails per
experimental condition.
Before the analysis of the treated material, a reproduc-
ibility test was conducted to determine if the same area
could be located for retesting chloride concentration
using the neutron beam. For this purpose, a group of
12 nails (n = 12) was analyzed three times as control
group without any treatment in between. A nail was
mounted in the sample holder, photographed and
PGAA was carried out in a chosen location on it. In the
second and third measuring cycles, the procedure was
repeated, using the photographs to remount the nails
and relocate the area analyzed initially. In this way, three
sets of 12 measurements were recorded, which could
be used to record three mean values for each group of
12 measurements, allowing comparison of means for
“batch” treatments.
DESALINATION OF ARCHAEOLOGICAL IRON OBJECTS: COMPARING THE EFFECTIVENESS OF SODIUM HYDROXIDE TREATMENTS 253

Data revealed heterogeneously distributed chloride infes- Results

tation, at a maximum of 1580 ppm and a minimum of
Figure 1 shows all desalination variations reduced chlo-
360 ppm. The mean of the three means was 776 ppm, with
ride concentration. Compared against the untreated
a standard deviation (SD) of 373 ppm (single means/SD
control group, all treatment variations were significant
were: 803/413 ppm; 733/307 ppm; 792/400 ppm). A
(p < 0.01). No desalination strategy achieved zero
significance test showed that there was no statistically
chloride in the end, since all retained chloride in the
significant difference between the mean value for each
200–300 ppm region. The 2 mol/L solution appeared
measurement (F = 1.208, df = 2, p = 0.318; all pair-
to desalinate more effectively if relative starting values
wise comparisons p > 0.10). Therefore, the mounting
for chloride in the regions of the objects measured are
strategy did not bias the results of PGAA, despite spalling
considered. However, absolute values for chloride after
off of parts of the corrosion crust during mounting
desalination treatment indicated that it did not perform
and handling since the analyzed areas were corrosion
better than the other solutions in relation to the residual
spots which had already lost the surface. This indicates
chloride within nail batches calculated as mean values.
reproducibility of the PGAA measurements including
mounting.

Experiments with oxygen consumption

The nails of one batch were analyzed by PGAA to
determine their chlorine content, desalinated, again
measured by PGAA and afterwards investigated as
batch to determine their oxygen consumption using
the methodology for archaeological iron reported by
Matthiesen and Wonsyld (2010). This records oxidation
reactions in closed environments and hence oxidation
of iron. The glass vessels had a volume of 250 mL and
were closed airtight by a silicone lid, additionally sealed
with silicone grease. The experimental conditions
around the jar were: 25°C ± 2 (climate chamber, checked Figure 1. Development of the average chloride concentration due to
by internal and external thermometer) and inside the different desalination treatments measured against the control group.
Pretest (1) are the means before desalination treatment. 0.5 is synonym
vessels 100% RH, whereby a highly corrosive system was
with 0.5 mol/L NaOH, all solutions at 25°C, exclusive 0.5 mol/L hot (70°C)
created. The humidity, created by water in a small glass
reservoir, was recorded by data loggers inside the vessels. The desalinated nails consumed less oxygen than the
For comparison, untreated nails were stored together untreated ones (Figure 2). Further, there is no or little
with dried silica gel at 0% RH. Oxygen concentration oxygen consumption under dry conditions (0% RH).
was measured in vol% of air using a Fibox 4 by PreSens
(Regensburg, Germany). The oxygen-sensitive sensor Oxygen consumption of treated and untreated objects
spots were glued with silicone adhesive inside the vessel 25,0 Linear (0.5 M 0 % RH)
and read out non-invasively by optical fibers through the
Linear (mean untreated
wall of the container. Additionally, the oxygen consump- 20,0 0 % RH)
Linear (2M 0 % RH)
tion inside the alkaline solutions during treatment
Vol-% Oxygen

15,0 Log. (mean 0.5 M 100 %

was monitored. This was a point of interest regarding RH)
Log. (mean 2 M 100 %
10,0
the de-aeration of an alkaline desalination solution RH)
Log. (mean 0.5 M

by ferrous objects either due to their initial corrosion 5,0

Ultrasonic 100 % RH)
Log. (mean 2 untreated
100 % RH)
and oxidation of Fe2+ to Fe3+ in the bath solution or to 0,0
Log. (mean 1 untreated
100 % RH)
oxidation of metallic iron until oxygen is used up and 1 31 61 91 121 151 181 211
days
passivation achieved. In daily work, it is impossible to
Figure 2. Treated and untreated iron objects compared regarding oxygen
make large treatment baths airtight, which is why they consumption at different relative humidities. Dotted lines are raw data,
are sheeted with PE films. continuous lines are trendlines
254  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
The oxygen content of the desalination solutions is slightly
different according to whether the solutions are covered
or uncovered — especially the first bath (Figure 3).
Figure 3. Oxygen content of desalination solutions of sodium hydroxide
– dependent from covering
Discussion
Chloride removal
All treatments were effective at removing chloride and
none of the physical variables of temperature, agitation
and covering to control oxygen ingress appeared to
have a statistically significant effect on the desalination
effectiveness of 0.5 mol/L NaOH solutions (Figure 1). In
other words, all 0.5 mol/L NaOH solutions showed the
same desalination effectiveness. In contrast, the 2 mol/L
NaOH desalinated significantly better statistically
(F = 4.260, df = 1, p = 0.042) than the averaged desal-
ination variations with 0.5 mol/L NaOH. The 2 mol/L
NaOH began at a higher starting mean for concentra-
tion of chloride and ended at the 200–300 ppm level
similar to the other treatment solutions, resulting in a
statistically significantly higher pretest-posttest differ-
ence. On the other hand, the end mean concentration
of chloride was not the lowest recorded, which might
be expected if it were the most effective treatment from
the range of treatments tested. It might be thought that
since the higher pH of the 2 mol/L NaOH is reported
to increase solvation of akaganeite (Cornell and Schw-
ertmann 2003, 217; Kuhn et al. 2011), it therefore
improves desalination; however, a counter argument
could be put forward asking why it does not produce
the lowest end mean value for chloride if this is the
case. This question remains unresolved. Also, the high
pH of the 2.0 mol/L NaOH solutions risks dissolving
oxides or oxyhydroxides in the corrosion crust, which
should be preserved.
The average end mean values for all treatments fall in
the region of 200–300 ppm, which aligns with North
and Pearson (1978), who defined 200 ppm as the lowest
end point and the threshold for corrosion stabilization.
DEVELOPMENT OF CONSERVATION PRACTICES
Another interesting outcome of this study is the option
of reducing treatment time for the same end point to a
few days by application of heat or an ultrasonic field.
Although the nails did not appear to be physically
affected by this treatment and did not disintegrate, the
risk of using these techniques requires investigation. It
might be inferred that covering with the PE film permits
oxygen access more readily and the oxidation of the iron
is completed earlier than in the more airtight lidded solu-
tions (Figure 3). However, there is no difference in the
effectiveness of the end point between the airtight and
the loosely covered solutions (0.5 M vs. 0.5 M PE film).
Overall, to develop a pattern of understanding regarding
treatment effectiveness and chloride removal, more data
must be obtained, especially regarding 2 mol/L NaOH
solutions.
In keeping with other data (Watkinson et al. 2019), the
oxygen consumption tests (Figure 2) confirm treated and
untreated nails corrode little or not at all in low relative
humidity conditions, irrespective of chloride content. In
addition, nails treated with 2 mol/L and 0.5 mol/L NaOH
solutions produce a similar consumption end point at 140
days, spanning less than 2 vol/% saturation of air (max.
oxygen saturation of air is 21 vol%), making it impossible
to confirm whether one treatment is more effective than
another at reducing corrosion rate and leading to the
conclusion that 200–300 ppm chloride produces approx-
imately the same reduction in corrosion rate.
Suggestions that PGAA could potentially be employed
to determine chloride within iron (Selwyn and Argy-
ropoulos 2006, Watkinson et al. 2014) seem to be
supported by the experimental work undertaken here.
However, this requires confirmation by a study involving
complementary methods to determine chloride values in
iron objects and assess the accuracy and reproducibility
of PGAA for this application.
Conclusion
In the desalination tests carried out here, sodium
hydroxide solutions reduced chloride levels in archae-
ological iron objects, irrespective of the treatment
variables, to values in the region of 200–300 ppm for
an end point of less than 5 ppm in a treatment bath
with a ratio of 100 g object/1 L NaOH. Since the oxygen
consumption of treated objects recorded a similar
measure for corrosion rate over 230 days, and given the
DESALINATION OF ARCHAEOLOGICAL IRON OBJECTS: COMPARING THE EFFECTIVENESS OF SODIUM HYDROXIDE TREATMENTS
lower concentration of 0.5 mol/L NaOH and consequent
reduced risk of oxide dissolution, this concentration was
found to be preferable to using 2 mol/L NaOH. Where
time is a constraint, ultrasonic and 70°C treatment solu-
tions of 0.5 mol/L NaOH can be used with caution and
stringent observation to control risk of physical damage
to objects. PGAA appears to be a promising method-
ology for non-destructive detection of chloride in iron,
albeit with two limitations: the chloride content is deter-
mined only locally for discrete areas by the chosen test
design, which means that total chloride concentrations
within objects will be unknown; and access to neutron
reactors is limited as a routine procedure. However,
the opportunity to monitor non-destructively the local
chloride concentration before and after treatment is
essential for the analysis of cultural heritage objects to
be conserved.
Acknowledgments
I wish to thank the Deutsche Bundesstiftung Umwelt
(DBU) for funding the research project “Besonderes
Eisern bewahren.” I would also like to thank Gerhard
Eggert, my Ph.D. supervisor and mentor. Furthermore,
the cooperation with Landesamt für Denkmalpflege
Baden-Württemberg was essential for the success of this
project, so I am grateful to Nicole Ebinger-Rist for her
assistance and support. The PGAA data were conducted
and processed by experts at the Technical University of
Munich (TUM) and the Heinz Maier-Leibnitz Zentrum
(MLZ) in Garching, Germany, to whom I am deeply
indebted for their expertise and support, especially to
Dr. Petra Kudejova.
References
Burshneva, S. and N. Smirnova. 2010. Some new advances
in alkaline sulfite treatment of archaeological iron. In
Archaeological Iron Conservation Colloquium 2010,
State Academy of Art and Design Stuttgart, 24–26 June
2010, eds. G. Eggert and B. Schmutzler, 21–23. Stuttgart:
Staatliche Akademie der Bildenden Künste Stuttgart.
Cornell, R.M. and U. Giovanoli. 1990. Transformation of
akaganeite into goethite and hematite in alkaline media.
Clays and Clay Minerals 38: 469–76.
Cornell, R.M. and U. Schwertmann. 2003. The iron oxides,
2nd ed. Weinheim: VCH.
Heinz Maier-Leibnitz Zentrum, Universität zu Köln, and
Technische Universität München. 2015. PGAA: Prompt
255
gamma and in-beam neutron activation analysis facility.
Journal of large-scale research facilities 1(A20). http://
dx.doi.org/10.17815/jlsrf-1-46 .
Keller, P. 1969. Vorkommen, Entstehung und Phasenum-
wandlung von β-FeOOH in Rost. Werkstoffe und Korro-
sion 20(2): 102–08.
Kergourlay, F., E. Guilminot, D. Neff, C. Rémazeilles, S.
Réguer, P. Refait, F. Mirambet, E. Foy, and P. Dillmann.
2010. Influence of corrosion products nature on dechlo-
rination treatment: Case of wrought iron archaeological
ingots stored 2 years in air before NaOH treatment.
Corrosion Engineering, Science and Technology 45(5):
407–13.
Kuhn, C.K., C.H. Wunderlich, G. Eggert, and T. Schleid.
2011. Another base, another solvent? e-PS 8: 81–85.
Matthiesen, H. and K. Wonsyld. 2010. In-situ meas-
urement of oxygen consumption to estimate corrosion
rates. Corrosion Engineering, Science and Technology
45(5): 350–56.
North, N.A. and C. Pearson. 1978. Methods for treating
marine iron. In ICOM-CC 5th Triennial Meeting Preprints,
Zagreb, 1–8 October 1978. 78/23/3/1–28/23/3/9. Paris:
International Council of Museums.
Réguer, S., P. Dillmann, and F. Mirambet. 2007. Buried
iron archaeological artefacts: Corrosion mechanisms
related to the presence of Cl-containing phases. Corrosion
Science 49(6): 2726–44.
Rémazeilles, C. and Ph. Refait. 2007. On the formation
of β-FeOOH (akaganéite) in chloride-containing envi-
ronments. Corrosion Science 49: 844–57.
Révay, Z. 2009. Determining elemental composition
using prompt γ activation analysis. Analytical Chemistry
81: 6851–59.
Schmutzler, B. 2012. Rettung vor dem Rost – Die
Weiterentwicklung der Eisenentsalzung nach der Alka-
li-Sulfit-Methode zur Erhaltung großer Fundmengen,
Internationale Archäologie: Naturwissenschaft und
Technologie Bd. 7. Rahden Westfalen: VML Verlag
Marie Leidorf.
Schmutzler, B. and G. Eggert. 2011. Chloride calamities:
Assessment of residual chloride analysis to compare iron
desalination methods. In Metal 2010: Proceedings of the
Interim Meeting of the ICOM-CC Metals Working Group,
Charleston, South Carolina, 11–15 October 2010, eds. P.
256  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Mardikian, C. Chemello, C. Watters, and P. Hull, 24–31. Zucchi, F., G. Morigi, and V. Bertolasi. 1977. Beta iron
Clemson: Clemson University. oxide hydroxide formation in localized active corrosion
Selwyn, L. 2004. Overview of archaeological iron: The of iron artefacts. In Corrosion and metal artefacts: A
corrosion problem, key factors affecting treatment, and dialogue between conservators and archaeologists and
gaps in current knowledge. In Metal 2004: Proceedings corrosion scientists, eds. B.F. Brown, H.C. Burnett, W.T.
of the Interim Meeting of the ICOM-CC Metals Working Chase, M. Goodway, J. Kruger, and M. Pourbaix, NBS
Group, Canberra, 4–8 October 2004, eds. J. Ashton and Special Publication 479, 103–105. Washington, DC:
D. Hallam, 294–306. Canberra: National Museum of National Bureau of Standards.
Australia.
Author
Selwyn, L. and V. Argyropoulos. 2006. Chlorine determi-
nation in archaeological wrought iron by instrumental Britta Schmutzler is professor of objects’ conservation
neutron activation analysis. Journal of the Canadian at the Erfurt University of Applied Sciences (Fachhoch-
Association for Conservation 31: 3–12. schule Erfurt) in Erfurt, Germany. She studied objects’
Snow, C.L., S.J. Smith, B.E. Lang, Q. Shi, J. Boerio-Goates, conservation at the Stuttgart State Academy of Art and
B.F. Woodfield, and A. Navrotsky. 2011. Heat capacity Design, from which she obtained her Ph.D. (Dr. Phil.)
studies of the iron oxyhydroxides akaganéite (β-FeOOH) with a thesis on the research project “Saving from Rust.”
and lepidocrocite (γ-FeOOH). The Journal of Chemical She also led the research project “Non-destructive anal-
Thermodynamics 43: 190–99. ysis methods for testing corrosion stability of iron objects”
(funded by the DBU).
Ståhl, K., K. Nielsen, J. Jiang, B. Lebech, J.C. Hanson,
P. Norby, and J. van Lanschot. 2003. On the akaganéite
crystal structure, phase transformations and possible role
in post-excavational corrosion of iron artifacts. Corrosion
Science 45: 2563–75.
Thickett, D., S. Lambarth, and P. Wyeth. 2008. Deter-
mining the stability and durability of archaeological
materials. In Art 2008: 9th International Conference on
NDT of Art, Jerusalem, Israel, 25–30 May 2008, 1–10.
Turgoose, S. 1982. Post-excavation changes in iron antiq-
uities. Studies in Conservation 27: 97–101.
Watkinson, D., M. Rimmer, Z. Kasztovszky, Z. Kis, B.
Maróti, and L. Szentmiklósi. 2014. The use of neutron
analysis techniques for detecting the concentration and
distribution of chloride ions in archaeological iron.
Archaeometry 56(5): 841–59.
Watkinson, D. and M. Lewis. 2005. Desiccated storage
of chloride-contaminated archaeological iron objects.
Studies in Conservation 50: 241–52.
Watkinson, D. and N.J. Emmerson. 2017. The impact of
aqueous washing on the ability of βFeOOH to corrode
iron. Environmental Science and Pollution Research Inter-
national 24(3): 2138–49.
Watkinson, D.E., M.B. Rimmer, and N.J. Emmerson.
2019. The influence of relative humidity and intrinsic
chloride on post-excavation corrosion rates of archae-
ological wrought iron. Studies in Conservation. https://
doi.org/10.1080/00393630.2018.1565006 .

Stabilisation of Archaeological Copper-Alloy Objects
from Chloride-Induced Active Corrosion with Beauveria
bassiana
Naïma Gutknecht
Haute Ecole Arc Conservation-restauration
University of Applied Sciences Western Switzerland
Neuchâtel, Switzerland
naima.gutknecht@gmail.com
*Author for correspondence
Abstract
Recent studies have revealed the possibility of using a fungal
strain of Beauveria bassiana on copper-alloy artefacts to
convert chloride-induced corrosion products into more stable
copper oxalates. This study aims to determine the modifications
resulting from the green treatment of active corrosion, focusing
on archaeological copper surfaces with pitting corrosion and/or
tinning. Results showed that synthesised copper chlorides were
transformed first into copper hydroxychlorides and then copper
oxalates. A development of copper oxalates was also observed
Introduction
In recent years, there has been growing interest in devel-
oping methods which do not harm the health of conser-
vators and the environment. One example is the biomin-
eralisation process induced by a fungal strain of Beauveria
bassiana that allows reactive corrosion products to be
converted into more stable passivating copper oxalates
(CuOx). This green approach was performed on outdoor
monuments and archaeological objects (Joseph et al. 2011,
Joseph et al. 2013, Domon-Beuret et al. 2015). The present
work aims to assess the performance of this bio-based
treatment on active pitting corrosion on copper-based
archaeological objects and to evaluate if it could be applied
on tinned bronze surfaces. Synthesised corrosion products,
artificially corroded coupons and archaeological objects
were treated with B. bassiana. The chemical and physical
modifications induced were ascertained through visual
observations and analytical methods. Finally, artificial
ageing procedures were carried out on treated objects to
assess the long-term behaviour of the treatment.
Materials and methods
Synthetic copper(I) chloride was purchased from
Sigma-Aldrich (CAS No. 7758-89-6). Eight bronze
Young Conservation Professional
Edith Joseph*
University of Applied Sciences Western Switzerland, Haute Ecole Arc
Conservation-restauration
Laboratory of Technologies for Heritage Materials
Institute of Chemistry, University of Neuchâtel
Neuchâtel, Switzerland
edith.joseph@he-arc.ch
edith.joseph@unine.ch
on artificially corroded coupons. Moreover, the treatment
demonstrated the stabilisation of the active corrosion present
on archaeological objects. On tinned surfaces, copper oxalates
formed mainly where the tin layer was lacking.
Keywords
bronze, tinned surface, chloride-induced corrosion,
sustainability in conservation, stabilisation treatment
(Cu 90/Sn 10) coupons (60 × 60 × 2–3 mm) cast at the
Swiss Art Foundry in St-Gallen (two of which were
given additional tinning) were corroded to simulate
the effects of terrestrial chlorinated and carbonated
environments according to recipes in the literature
(Leyssens 2006). The composition of the corrosion on
the coupons was assessed by FTIR/Raman analyses (data
not shown). A corpus of ten Roman-era archaeological
objects with pitting corrosion and/or tinned surfaces
was lent by various Swiss heritage institutions. Table 1
provides information on all of the samples (synthesised
corrosion products, artificially corroded coupons and
archaeological objects).
The powdered copper(I) chloride, coupons and objects
were treated with gelled cultures of B. bassiana in deion-
ized water according to usual procedures (Albini et al.
2015). The objects were then exposed to an aggressive
artificial atmosphere according to ISO 16701 (Interna-
tional Organization for Standardization 2015), minus
the salt-spraying step, to simulate a non-controlled
storage environment: 60 cycles of 12 h (4 h at 95% RH,
4 h at 50% RH and intervals of 2 h).
258  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Table 1. Summary of sample preparation, treatment and cross sec- scope with a slide-on ATR objective and a germanium
tions performed tip (FTIR correlation map of malachite and 1365 cm−1
Treatment peak intensity map [CuOx]). Spectra were obtained at
Cross
Sample name Preparation duration
section a range of 4000–650 cm−1 (4 cm−1 spectral resolution,
(days)
Synthetic corrosion compounds
16 scans, 1.56 µm spatial resolution).
CuCl powder - 7 -
Raman spectroscopy was performed using a Renishaw
Coupons
inVia spectrometer equipped with a Nd:YAG laser of
B2-B3-B4 7 x
633 nm at less than 1 mW (600 gr/mm). Spectra were
recorded at a range of 4000–400 cm−1 (100× objec-
tive, 1000 mm hole, 100 mm slit, 4 accumulations of
1-hour immersion in CuCl2
10 s). Nantokite was identified using reference spectra of
saturated bath
B5-B6-B7 7 x
synthetic copper(I) chloride (Sigma-Aldrich, CAS No.
7758-89-6).

Secondary electron images were obtained with a Zeiss

1-hour immersion in CuCl2 EVO MA 10 environmental scanning electron micro-
saturated bath
14-days immersion in 10%
scope (environmental SEM).
sodium sesquicarbonate bath
B8-B9 7 x Results and discussion
The conversion from synthetic copper(I) chloride into
a mixture of copper hydroxychlorides and CuOx was
Electrolytic tin plating +
electrolytic corrosion (6 hours,
observed after seven days with FTIR (Gutknecht 2018).
OCP/+1V in NaCl 1M) Hydrolysis probably occurred first due to the use of the
Archaeological objects aqueous gel followed by a reaction with the oxalic acid
07/14448-07 Sediments removed 10 (selected x secreted by the fungi.
area)
14/16481-03 Sediments removed 10 - On all the coupons, the formation of CuOx was assessed
14/16711-07 Sediments removed 10 (selected -
(Figures 1 and 2). A strong visual impact of the treatment
area)
14/16478-11 Sediments removed 10 -
was observed in a change of colour. Even if the chemical
14/16478-13 Sediments removed 21 - composition of the coupons was representative of that
14/16475-09 Sediments removed 7 - found on archaeological objects (Gutknecht 2018), the
outermost corrosion removed
visual appearance was not reproduced with the recipes
14/16538-29 Sediments removed 10 -
outermost corrosion removed
used due to the difficulties in reproducing long-term
14/16475-08 Sediments removed 10 - archaeological corrosion artificially. This major change in
outermost corrosion removed the coupons’ appearance was not observed on the objects.
V.011.1/1383.1 Sediments removed 10 (half of x
outermost corrosion removed the object)
V.011.1/2458.4 Sediments removed 10 (half of -
the object)

Before and after treatment samples were taken using

Struers SpeciFix-20 resin and prepared as cross sections
by polishing with Micromesh tissues to 6000 grit.

Attenuated total reflectance-Fourier transform

infrared spectroscopy (ATR-FTIR) measurements
were performed with a Thermo Scientific Nicolet iS5
spectrometer (iD5 diamond accessory). Mapping on
the section of the tinned archaeological object was Figure 1. Coupons before and after treatment. The upper left edge of the
performed with a Nicolet iN10 MX imaging micro- coupons delimited by a dotted line was left untreated for comparison
STABILISATION OF ARCHAEOLOGICAL COPPER-ALLOY OBJECTS FROM CHLORIDE-INDUCED ACTIVE CORROSION WITH BEAUVERIA BASSIANA
Figure 2. ATR-FTIR spectrum obtained on the B9 tinned coupon
after treatment on the surface. The characteristic vibrational bands
corresponding to copper oxalates are indicated as Ox. The other peaks
correspond to traces of copper hydroxychlorides
After treatment, the archaeological objects gained a
lighter green tonality, suggesting the presence of CuOx,
as confirmed by FTIR analyses. On tinned surfaces, the
CuOx film gained a more matte and greener surface
after treatment (Figure 3A) with typical rosette-like
aggregates (Figure 3B), which also filled the gaps in the
tin layer (Figure 3C).
Figure 3. (A) Tinned object no. V.011.1/1383.1 (Canton Aargau
Archaeological Service) after treatment; (B) SEM secondary electron
image of formed rosette-like aggregates; (C, from left to right) SEM
image of cross section with red box indicating mapping area, FTIR
correlation map of malachite and 1365 cm−1 peak intensity map (CuOx).
Peak intensity scale from blue to red
The sample taken from the treated object no. 07/14448-07
(Figure 4) showed that copper chlorides (CuCl) were still
present deep in the corrosion layer (2 mm).
However, no corrosion regrowth was observed on the
same object after ageing (Table 2). This suggests that
the CuOx film formed on the outer part of the corrosion
layer isolated the potentially detrimental compounds and
hence protected the object from further corrosion as a
protective organic coating helps to isolate from external
humidity. The inhibition of active corrosion was effec-
259
tive on nine of the ten objects (no. 14/16481-03 showed
regrowth of corrosion).
Figure 4. (A) Object no. 07-14448-07 (Site et Musée romains d’Avenches)
after treatment; (B) cross-section micrograph with Raman measurements
indicated by a red arrow; (C) corresponding Raman spectrum showing
shifts in CuCl. A shift at 300 cm−1 is indication of the transformation of
CuCl into copper hydroxychlorides
Conclusion
This study provides a deeper understanding of the process
occurring during green treatment. Protection seems to
be achieved by converting reactive compounds into more
stable CuOx and forming a film isolating any remaining
CuCl. In addition, stabilisation was confirmed for objects
exposed to either aggressive or uncontrolled atmospheres.
Acknowledgements
The authors are grateful to the Site et Musée romains
d’Avenches and the Canton Aargau Archaeological
Service in Brugg for their support in providing the
archaeological objects for this study. They would also
like to thank Nadim Scherrer and Hochschule der Kunst
Bern for the SEM-EDS and Raman analyses.
References
Albini, M., L. Comensoli, L. Brambilla, E. Domon Beuret,
W. Kooli, L. Mathys, P. Letardi, and E. Joseph. 2015.
Innovative biological approaches for metal conservation.
Materials and Corrosion 67: 200–206.
Domon-Beuret, E., L. Mathys, L. Comensoli, L. Bram-
billa, M. Albini, C. Cevey, R. Bertholon, P. Junier, and E.
Joseph. 2015. Biopatines : des champignons au service des
alliages cuivreux. ARAAFU Cahier technique 22: 45–48.
Gutknecht, N. 2018. La corrosion active sur les alliages
cuivreux archéologiques – Evaluation de la stabilisation
par biopassivation fongique. Unpublished master’s
260  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Table 2. Archaeological objects before and after treatment, and after thesis, HES-SO University of Applied Sciences Western
ageing according to ISO 16701 (except no. 14/16478-13 for which Switzerland.
100% RH was maintained for 15 days)
International Organization for Standardization. 2015.
Before treatment After treatment After ageing
Corrosion of metals and alloys – Corrosion in artificial
Site et Musée romains d’Avenches
07/14448-07
atmosphere – Accelerated corrosion test involving expo-
sure under controlled conditions of humidity cycling and
intermittent spraying of a salt solution (ISO 16701:2003
revised by ISO 16701:2015).
Joseph, E., P. Letardi, L. Comensoli, A. Simon, P. Junier,
14/16481-03 D. Job, and M. Wörle. 2013. Assessment of a biological
approach for the protection of copper alloys artefacts.
In Metal 2013: Proceedings of the Interim Meeting of the
ICOM-CC Metals Working Group, Edinburgh, 16–20
14/16711-07 September 2013, eds. E. Hyslop, V. Gonzalez, L. Troalen,
and L. Wilson, 203–08. Edinburgh: International Council
of Museums and Historic England.
Joseph, E., A. Simon, S. Prati, M. Wörle, D. Job, and R.
14/16478-11
Mazzeo. 2011. Development of an analytical procedure
for evaluation of the protective behaviour of innovative
fungal patinas on archaeological and artistic metal
14/16478-13 artefacts. Analytical and Bioanalytical Chemistry 399:
2899–907.
Leyssens, K. 2006. Evaluation of corrosion potential
measurements as a means to monitor the storage and
14/16475-09
stabilization processes of archaeological copper-based
artifacts. Analytical Chemistry 78(8): 2794–801.

Authors
14/16538-29 Naima Gutknecht has an MA in conservation-resto-
ration from the Haute Ecole ARC Conservation-restau-
ration (Neuchâtel, Switzerland) where she specialised in
the preservation of metal objects.

14/16475-08
Edith Joseph is assistant professor at both the Univer-
sity of Neuchâtel and the Haute Ecole ARC Conserva-
tion-restauration (Neuchâtel, Switzerland). She has a
PhD in chemistry from the University of Bologna, Italy.
Her main research activities are the application of green
Canton Argau Archaeological Service chemistry for the preservation of artistic and archaeo-
V.011.1/1383.1 logical objects. She has published around 50 papers in
international journals and books on analytical chemistry
and conservation science.

V.011.1/2458.4

Design of a Suitable Cleaning Procedure for the Tarnished
Silver Elements of the Reliquary Bust of Santa Vittoria,
Diocesan Museum of Agrigento, Sicily
Nicola Ricotta* Christian Degrigny
Opificio delle Pietre Dure Haute Ecole Arc Conservation-restauration
Florence, Italy HES-SO University of Applied Sciences and
nico.ricotta@hotmail.it Arts Western Switzerland
www.opificiodellepietredure.it Neuchâtel, Switzerland
*Author for correspondence
christian.degrigny@he-arc.ch
www.he-arc.ch
Abstract
The restoration of the Santa Vittoria reliquary conserved in the
Diocesan Museum of Agrigento in Sicily at the Opificio delle
Pietre Dure (OPD), Florence, was an opportunity to test various
cleaning processes on its tarnished bust made of partly gilded
silver plates. The mechanical/chemical protocols used at the
OPD – calcium carbonate (CaCO3) or sodium bicarbonate
(NaHCO3), and, more recently, erasers – were compared to
local electrolytic cleaning techniques recently developed by
the Conservation Research Unit at Haute Ecole Arc (UR-Arc
Introduction
The reliquary bust of Santa Vittoria dating from 1593
and conserved in the Diocesan Museum of Agrigento,
Sicily, is a composite object comprising a finely embossed,
chiselled and engraved tunic made of a silver-based
alloy (around Ag-10%Cu), with face and hands made of
polychrome copper-based alloys. The bust is mounted
on a partially gilded and polychrome copper-alloy base
(Figure 1). Its relatively well-preserved surface suggests
that it has been regularly cleaned.
Figure 1. The reliquary bust of Santa Vittoria on its base
Young Conservation Professional
Andrea Cagnini Cinzia Ortolani
Opificio delle Pietre Dure Opificio delle Pietre Dure
Florence, Italy Florence, Italy
andrea.cagnini@beniculturali.it cinzia.ortolani@beniculturali.it
CR), Neuchâtel, Switzerland. This article presents the different
cleaning processes, describes the tests carried out and shows
the results which led to the selection of electrolytic cleaning. As
electrolytic treatments are new to OPD, the way in which these
technologies have been implemented is discussed.
Keywords
tarnished silver, reliquary bust, electrolytic cleaning, the Pleco,
Santa Vittoria, Sicily
The tunic is covered with a thin, homogeneous tarnish,
probably silver sulfide (Ag2S) from sulfur-based pollut-
ants and silver chloride (AgCl) from the handling of the
objects (Degrigny, Jeanneret and Witschard 2015). Small
darker spots are distributed heterogeneously on the metal
surface (Figure 2). As the reliquary was a good candidate
for applying electrolytic cleaning (ibid.), a collaboration
was set up to compare the mechanical/chemical proto-
cols used traditionally at the Opificio delle Pietre Dure
Figure 2. Detail of the tunic: the darker spots on the silver surface are
indicated with a white arrow
262  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

(OPD) and the selective electrolytic treatment with a suited for the cleaning of tarnished silver from previous
Pleco recently developed by the Conservation Research studies (Basilissi 2015), were tested in a third zone: the
Unit at Haute Ecole Arc (UR-Arc CR) (ibid.). Faber-Castell Perfection 7057 and the Staedtler Mars
Plastic 528 55 (Figure 3b). With regard to electrolytic
The cleaning techniques tested processing, cathodic LSV was first performed on differ-
ently tarnished areas to identify the corrosion products
Calcium carbonate and sodium bicarbonate, both
(Figure 3c). The blue circle in the inserted picture
in aqueous suspension, have an abrasive action (like
below the LSV plots of Figure 3d refers to the homoge-
erasers) used to remove surface corrosion products
neous black tarnish. The corresponding plot shows the
(Palomar et al. 2016). In addition, sodium bicarbo-
presence of a peak of silver sulfide with a maximum at
nate can produce a chemical action. The matrix of
-1.6 V/VC preceded by a small peak of silver chloride
erasers reduces the mechanical action of their abrasive
with a maximum at -0.4V/VC and followed by the wave
compounds (calcium carbonate). All of the processes
of bubbling of H2 (Degrigny, Jeanneret and Witschard
leave residues which have to be cleaned by rinsing in
2015). The red plot referring to one of the dark spots on
water or in ethanol/acetone, in the case of erasers.
Figure 3d similarly shows a silver chloride peak (max.
On the other hand, electrolytic cleaning does not remove at -0.48 V/VC) followed by a larger silver sulfide peak
surface corrosion products but converts them into (max. at -1.9 V/VC).
metallic silver through reduction processes. The Pleco
equipped with a three-electrode electrolytic cell allows
local processes to take place (ibid.). The electrolyte
(NaNO3 1% by weight) in contact with the metal through
a microporous polyvinyl formal (PVFM) pad is continu-
ously renewed using diaphragm pumps, preventing any
restaining of the metal surface by the polluted electrolyte.
The definition of the treatment parameters (potentials to
apply) requires carrying out linear sweep voltammetry
(LSV) with a potentiostat. As the available instrumenta-
tion (PalmSens) at the OPD had limitations (a few mA),
the shape of the pad was sharpened to limit the size of
the contact point to a minimum. As a result, the flow of
the electrolyte had to be adjusted to prevent any leakage.
The voltammetric curves carried out in static mode (the
Pleco fixed securely) were plotted from the corrosion
potential (Ecorr about -0.2 V versus a vitreous carbon
electrode /VC - 0.06 V versus the standard hydrogen Figure 3. Cleaning tests with (a) CaCO3 (or NaHCO3), (b) eraser and (c)
electrode) down to -2 V/VC, with a scanning rate of the Pleco. Cathodic LSV plots of different areas in NaNO3 1% in weight,
scanning rate of 10 mV/s and potentials are given in V/VC (d)
10 mV/s.

The duration of the reduction of silver sulfide (about

Conditions of application and evaluation of
cleaning tests
After discussion with the authority responsible for the
object, a discreet but representative surface on the back
of the bust was selected and divided into four distinct
zones to carry out cleaning tests. The whole area was
first degreased with a mixture of solvents LE2 (80%
ligroin and 20% ethanol). Calcium carbonate and sodium
bicarbonate were applied as a paste in two zones with
cotton swabs (Figure 3a). Two types of erasers, best
30 s) was determined in chronoamperometry mode
(monitoring of the current versus time at the poten-
tials indicated previously – Degrigny, Jeanneret and
Witschard 2015.) with the Pleco being moved incremen-
tally (dynamic mode, Figure 4). Applying -1.9 V/VC on
the dark spots did not fully remove them. Instead, the
surrounding metal surface looked overcleaned (reduced
silver particles dislodged by hydrogen bubbling). Further
cathodic voltammetries on the dark spots did not show
new peaks.
DESIGN OF A SUITABLE CLEANING PROCEDURE FOR THE TARNISHED SILVER ELEMENTS OF THE RELIQUARY BUST OF SANTA VITTORIA, DIOCESAN MUSEUM OF AGRIGENTO, SICILY
Figure 4. Chronoamperometry in dynamic mode. The areas locally
cleaned were rinsed with a cotton swab impregnated with deionised
water
Both calcium carbonate and sodium bicarbonate
allowed a gradual thinning out of the tarnished layer.
Thanks to its chemical action, the cleaning with sodium
bicarbonate seemed to be easier to perform. With the
Faber-Castell Perfection 7057 eraser, unlike the Staedtler
Mars Plastic 528 55 eraser, the gradual removal of the
tarnished silver was difficult to manage. In both cases,
the metal surface was immediately reached. In fact, a
major drawback of mechanical cleaning techniques is
their subjectivity and the absence of scientific parame-
ters that define the completion of the cleaning process.
The Pleco, on the other hand, appeared to be a versatile
tool allowing an analysis of the constituents of the silver
tarnish, an adjustment of the electrolytic parameters to
reduce silver-based compounds while preventing over-
cleaning, and the possibility of improving the surface
appearance by additional burnishing of the reduced
silver particles.
Given these considerations and the problems to be
solved, it was decided to use the Pleco to clean the silver
tarnish on the whole bust.
Local electrolytic cleaning with the Pleco
The potentiostat used previously was replaced by a
conventional power supply and a voltmeter to continue
monitoring the potential to apply to the metal surface.
The nozzle of the Pleco was changed from a conical to a
cylinder shape to allow larger areas to be cleaned in one
263
go. The pad was also shaped to have a 2 cm diameter.
As a result, the flow of the electrolyte was adjusted.
Based on preliminary tests and to prevent any over-
cleaning, the potential used to clean the bust was
-1.6 V/VC applied locally for 30 s, after which the Pleco
was moved to another tarnished area (dynamic mode).
The remaining black spots were removed afterwards with
a Staedtler Mars Plastic 528 55 eraser.
The surface appearance was homogenised by light
burnishing with a cotton cloth, and a nitrocellulose resin
(Zapon Sopraffina from Lechler) diluted in a mixture of
isopropyl alcohol (20%), isothane (45%) and methyl ethyl
ketone (35%) was applied as a protective layer.
Conclusion
The level of cleaning achieved fully satisfied the objectives
set during the design phase. It was possible to recover a
homogeneous surface appearance. Once the necessary
skills needed to use the Pleco were acquired, it proved
to be an invaluable tool that allows full control of the
operations of electrolytic cleaning.
Acknowledgements
The authors wish to thank the Diocesan Museum of Agri-
gento and the Superintendence BB.CC.AA of the same
city for authorizing the project. Additional thanks go to
the Scuola di Alta Formazione e di Studio of the OPD
for having supported the study and application of the
Pleco electrolytic pencil; the Conservation Department of
Haute Ecole Arc in Neuchâtel for providing both a Pleco
and its diaphragm pumps; and the Università degli Studi
of Florence for lending us their portable potentiostat.
References
Basilissi, G. 2015. Il restauro di tre manufatti islamici
in metallo ageminato del XIII–XIV secolo. Ricerca
e sperimentazione di nuove metodologie di pulitura
dell’argento e messa a punto di idonei formulati protet-
tivi. Bachelor’s dissertation, Opificio delle Pietre Dure,
Florence, Italy.
Degrigny, C., R. Jeanneret, and D. Witschard. 2015.
Local cleaning with the Pleco electrolytic pencil of
tarnished Saint Candide reliquary head at the Treasury of
Saint-Maurice Abbey, Valais (Switzerland). e-preservation
science 12: 20–27.
264  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Palomar, T., B. Ramírez Barat, E. García, and E. Cano.

2016. A comparative study of cleaning methods for
tarnished silver. Journal of Cultural Heritage 17: 20–26.

Authors
Nicola Ricotta is a graduate student at the Scuola di
Alta Formazione e di Studio, Opificio delle Pietre Dure,
Florence, Italy.

Christian Degrigny is an electrochemist and professor

at the Haute Ecole Arc Conservation-restauration,
Neuchâtel, Switzerland.

Andrea Cagnini is a chemist and professor at the Scuola

di Alta Formazione e di Studio, Opificio delle Pietre
Dure, Florence, Italy.

Cinzia Ortolani is a conservator and professor at the

Scuola di Alta Formazione e di Studio, Opificio delle
Pietre Dure, Florence, Italy.
Equipment Identification and the Development of
Dry-Ice Blasting Parameters for Cleaning Archaeological
Wrought Iron, Copper Alloys, and Gray Cast Iron
William Hoffman* Laurie King *Author for correspondence
The Mariners’ Museum and Park The Mariners’ Museum and Park
Newport News VA, USA Newport News VA, USA
whoffman@marinersmuseum.org lking@marinersmuseum.org
www.marinersmuseum.org www.marinersmuseum.org

Abstract
In 2013, conservators within The Batten Conservation Complex at
The Mariners’ Museum and Park in Newport News, Virginia, began
investigating the application of dry-ice blasting to assist in the
cleaning of wrought-iron artifacts recovered from the wreck site
of the American Civil War ironclad USS Monitor. The technique
had been used in industry for several decades, but research for
its use within conservation was relatively recent. Information
was limited concerning effective blasting equipment as well
as procedures to clean archaeological wrought iron. This paper
provides an overview of the identification of a suitable dry-ice
blasting system and a description of the method developed to
determine safe cleaning parameters for wrought-iron artifacts.
It then describes how the procedure was modified to identify
cleaning parameters to detach thin concretion from copper-alloy
artifacts and remove paint from gray cast iron without causing
damage.
Keywords
dry-ice blasting, marine archaeological conservation, wrought
iron, gray cast iron, copper alloys

Introduction
The Mariners’ Museum and Park (TMMP), in Newport trial and marine sites predominantly develops along
News, Virginia, has undertaken conservation within the slag stringers embedded in the metal (Cronyn 1990).
Batten Conservation Complex (BCC) of 182 metric tons This results in deep, linear corrosion pockets inter-
of artifacts recovered from the wreck site of the Amer- mixed within the metal structure, which retain chlorides
ican Civil War ironclad USS Monitor in partnership with (Rimmer and Wang 2010).
the National Oceanic and Atmospheric Administration
(NOAA), which maintains stewardship over the vessel.
Dry-ice blasting was researched as a potential cleaning
technique for wrought-iron objects when traditional
mechanical methods proved insufficient. Once a testing
methodology, effective blasting settings, and appropriate
equipment were identified, the research was expanded to
include copper-alloy artifacts from the Monitor and gray
cast-iron objects from the Museum’s collection.

Background: Wrought-iron artifacts

Wrought-iron objects make up the majority of the
Monitor Collection, with the most significant being the
2.7 m tall, 6 m in diameter revolving gun turret (Figure 1).
As bulk concretion was removed manually, the surviving
metal structures of the objects were exposed, revealing
a surface similar in appearance to waterlogged wood.
Corrosion on wrought-iron artifacts from both terres-
Figure 1. Monitor’s turret after recovery – eight overlapping layers of
2.5 cm thick wrought-iron plate forming a ring covered by a roof of
additional wrought-iron components
To desalinate the artifacts, conservators within the BCC
utilize electrolytic reduction (ER) using a solution of
1% (w/v) sodium hydroxide (NaOH) in reverse osmosis
or deionized water. In addition, mechanical cleaning
266  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
campaigns are undertaken. Early on in the conservation
project, corrosion removal was kept to the object’s orig-
inal surface, and even with the use of electrochemistry,
many wrought-iron artifacts corroded post-treatment.
Chlorides had remained trapped in embedded corrosion
products. Due to the possibility that low humidity levels
could not be guaranteed for all objects, and that the
majority are structurally intact, it was decided to remove
as much corrosion as possible.
On small artifacts, inset corrosion products are removed
using scalpels, dental tools, and pneumatic chisels,
which is laborious and time consuming. When time
came to clean large objects, such as the turret, it was
clear that hand cleaning was impractical. The conser-
vation staff began looking into abrasive techniques.
They experimented with a Farrow system which blasts
high-pressure water mixed with fine-glass media as an
abrasive. Although the technique was quick and effec-
tive at cleaning out the corrosion pockets, it was ruled
out as a long-term approach, along with other blasting
methods, due to the accumulation of used media. Much
work would take place inside large treatment tanks and
the collection and disposal of blasting material would be
difficult, a potential environmental hazard, and costly.
Furthermore, there were safety concerns about the staff
accidentally breathing in media dust and the possibility
of clogging plumbing equipment. However, testing results
showed the potential of large-scale abrasive cleaning.
Conservators next looked for a method that was effec-
tive, time efficient, safe for both staff and objects, and
left behind minimal by-product. Dry-ice blasting using
solid carbon dioxide (CO2) media was explored to fit
the criteria.
Dry-ice blasting
Solid dry-ice blasting is an air-powered cleaning method
using frozen carbon dioxide (CO2) aggregate at -78.5°C
in place of other abrasive media such as aluminum oxide
(Al2O3) or walnut shell. Unlike traditional blasting tech-
niques, dry-ice sublimates, meaning no residue is left
behind other than the material removed during cleaning
(Spur et al. 1999). Dry ice is noncorrosive, nonconductive,
dry, and nontoxic, making it ideal for many applications
(Cold Jet 2018). Furthermore, dry ice is a soft material
with a Mohs hardness of 1.5 to 2, limiting its ability to
abrade a variety of materials (Wilson 2013). Solid dry
ice is produced in two forms: pellets with 1–6 mm in
DEVELOPMENT OF CONSERVATION PRACTICES
diameter (average of 3 mm) and 5–15 mm in length, and
varying-sized solid blocks (Spur et al. 1999).
Blasting systems
Solid dry-ice blasting systems can be found in fixed
or mobile setups. A common system consists of an air
compressor connected to an air-drying unit connected
to a blasting controller. Within the controller, dry ice
mixes with compressed air, accelerating the media out
through an abrasion-resistant hose, which remains
flexible under cryogenic conditions and where friction
is limited during particle movement (McNeil 2017).
Ultimately, the particles pass out through a nozzle affixed
to a gun-like applicator.
Dry-ice blasting controllers primarily come in two forms:
full-pellet blasting and shaved-block systems. Both
machine types include a hopper where dry-ice media is
added. This compartment is connected to a feeder rotor,
which introduces the dry ice to incoming compressed air.
In a shaved-block system, a push arm mounted within
the hopper presses a dry-ice block against a bladed rotor
(Watson 2017). As the rotor spins, its blades shave the end
of the block creating fine 0.3 mm particles (Wilson 2013).
Cleaning mechanism
The mechanism for dry-ice cleaning has been described
with three factors: thermal shock effect, gas expansion
effect, and mechanical (kinetic) effect (Spur et al. 1999).
However, Krieg (2008) showed that thermal and gas
expansion effects play only a minor role in the overall
cleaning power. As stated by Van der Molen et al. (2011),
this means that the mechanical effect is the driving force
behind the process, i.e., the impact strength of the dry-ice
particles hitting the material for removal. The stronger
the dry-ice particles collide, the greater the ability to
break up surface debris, coatings, etc. This cleaning
power is tied to the velocity (speed and direction) the
dry-ice particles are traveling and the number and size
of particles in the system.
Particle speed is primarily controlled by the supplied
air pressure (in bar), whereas the direction of media
movement is determined by the shape and design of the
blasting nozzle and blasting angle. Another contributing
factor is working distance: as particles leave the blasting
nozzle, they deaccelerate and begin to sublimate, losing
mass.
EQUIPMENT IDENTIFICATION AND THE DEVELOPMENT OF DRY-ICE BLASTING PARAMETERS FOR CLEANING ARCHAEOLOGICAL WROUGHT IRON, COPPER ALLOYS, AND GRAY CAST IRON
The number and size of dry-ice particles in the system are
tied to the volume of air flow/consumption expressed in
cubic meters per minute (m3/min), and the rate at which
the dry-ice media is added to the airstream, measured in
kilograms per minute (kg/min). The greater the volume
of air going through the system, the larger the amount
and/or size of media particles that can be introduced and
accelerated at a given pressure, resulting in a stronger
cleaning effect.
Initial experimentation
When BCC personnel researched dry-ice blasting specif-
ically for corrosion removal within the structure of
archaeological wrought iron, no literature existed. Most
research focused on surface coatings (Weston et al. 2005,
Brush 2010, Van der Molen et al. 2011) and debris removal
(Spur et al. 1999, Silverman and Irwin 2009, Wilson 2013).
Therefore, conservators began test cleaning using a Cold
Jet i3 MicroClean shaved-block system (Cold Jet 2018)
together with visual inspection on a cut section of Monitor’s
wrought-iron propeller shaft. Operating parameters tested
included the full range of media flow rate (0−0.6 kg/min)
with air pressure up to 8.6 bar and an air consumption rate
of 1 m3/min (the air-compressor system limit). Blasting
distance and angle were experimented with using different
nozzles. Observations during and after cleaning showed
the parameters used were ineffective.
To access a wider range of blasting settings, conservators
experimented with a Cold Jet SDI Select 60 (Cold Jet
2018), whose blasting controller is a hybrid machine that
can be set to blast pellets or shave ice. Similar to block-
shaving systems, the controller has a bladed rotor and
a push arm. However, the rotor can also shave pellets,
creating a medium-sized material between full-sized
pellets and shaved-block media. The SDI Select 60 has a
media flow rate of 0−2.7 kg/min and can blast at pressures
of 4.5−17.2 bar with the ability to consume air at over
6 m3/min. To supply air to the blasting unit, the BCC
rented a Doosan HP375 diesel-powered air compressor
(supports air consumption up to 10.6 m 3/min and
pressure up to 12.1 bar) attached to a Van Air Systems
Blast-Pak portable desiccant air dryer (Figure 2).
Additional tests were conducted on wrought-iron
stanchions from Monitor’s turret roof, initially using
pellets assessed by visual inspection. It was observed
that the pellets caused deflection on thin metal edges
and conservators switched the blasting controller to
267
shaved pellets instead. Results identified that 10.3 bar of
dried air pushing 1.8 kg/min of shaved pellets through
a medium-flow fan nozzle at a 90° angle and a distance
of 15–30 cm removed almost all embedded corrosion.
No damage to the metal could be observed. Several
nozzles were tried, with the 312S1 nozzle (30.5 cm long)
working best and producing a 2.5 cm blasting swath.
The nozzle was connected to SDI Select 60’s standard
applicator and a 6 m long, 2.5 cm diameter silicone
blasting hose.
It is hypothesized that these settings were successful
because the increase in volume of media and particle
speed enabled the breakup of the corrosion, but adjusting
the particle size and cleaning distance resulted in less
impact on thin metal edges.
Figure 2. Blasting setup used during testing
Additional experimentation
Conservation staff next wanted to determine if dry-ice
blasting caused microstructure damage unapparent on
visual inspection. If so, could a damage threshold be
determined, and were the successful cleaning parameters
identified for corrosion product removal higher or lower?
Therefore, cleaning tests were conducted on sample
material with microscopic analyses at 350× magnification
before and after cleaning, using a Hirox KH-7700 Digital
Optical Microscope.
Methodology
A wrought-iron fragment from Monitor’s propeller shaft
was cut into 27 samples (approximately 0.5 cm3 each).
The samples were mounted in Buehler EpoKwick epoxy
resin and polished using a Buehler Phoenix 4000 sample
preparation system. The sample group was divided into
three sets of nine, with one set to be blasted at 6.9 bar,
one at 8.6 bar, and one at 10.3 bar. Within each set, the
samples were further divided into two sub-sets according
to a blasting distance of 7.5, 15, and 30 cm, and the other
by volume of shaved-pellet media used at 0.9, 1.8, or
2.7 kg/min. Each sample thus represented a variation
268  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Table 1. Wrought-iron sample-testing parameters. The sample in the top left reads pressure of 6.9 bar using 0.9 kg/min of media at a distance of 7.5
cm. The highlighted areas indicate the working parameters found for corrosion removal

Parameters 6.9 bar 8.6 bar 10.3 bar

0.9 kg/min 7.5 cm 7.5 cm 7.5 cm 7.5 cm 7.5 cm 7.5 cm 7.5 cm 7.5 cm 7.5 cm
1.8 kg/min 15 cm 15 cm 15 cm 15 cm 15 cm 15 cm 15 cm 15 cm 15 cm
2.7 kg/min 30cm 30cm 30 cm 30 cm 30 cm 30 cm 30 cm 30 cm 30 cm

in pressure setting, volume of media, and measured Results and application

distance (Table 1).
Results showed little change to the metal surface on all
Sample tests were conducted using a mounting platform, samples regardless of setting, including those blasted
which allowed for three samples to be held while blasted. at the empirically identified corrosion-removal param-
Along with the platform, three “guide arms” were fabri- eters (Figure 4). The only alterations observed to the
cated to attach to the 312S1 nozzle and hold it at the polished samples occurred within some slag inclusions
predetermined distances (Figure 3). During blasting, where material had been removed, and, in a few cases,
the base of each guide arm was kept in contact with the metal grains were lost around the edges of the affected
mounting platform and the nozzle was kept perpendic- inclusions. Based on the positive results, conservation
ular to the samples. staff applied dry-ice blasting to several wrought-iron
bulkheads using the successful cleaning parameters
identified during the initial cleaning tests. This resulted
in most of the embedded corrosion being removed
from a 1 m × 3.5 m keelson plate in approximately one
hour compared to the weeks it would have taken by
hand (Figure 5). To date, over 400 artifacts have been
cleaned. Through experience gained using the blasting
equipment, and maintaining control over working
distance, the deflection of thin metal edges has been
avoided overall.
Figure 3. Mounting
platform and guide
arms; the 30 cm guide
arm attached to the
blasting nozzle and
applicator held in the
 testing position

The testing procedure took place in three phases based on

the successive attachment of each guide arm. Starting with Figure 4. Wrought-iron sample blasted at 10.3 bar using 1.8 kg/min
the 7.5 cm arm, the pressure was regulated to 6.9 bar and of shaved-pellet media at 15 cm. (A) Photograph taken at 350×
the first series of samples were loaded onto the mounting magnification before cleaning; (B) after cleaning
platform. The SDI Select 60 was set to produce 0.9 kg/min
of media and the corresponding sample was hit with a short
back-and-forth pass of dry ice to simulate cleaning. Once
the first sample had been blasted, the volume of media was
turned up to 1.8 kg/min and the next sample was blasted,
followed by the last sample at 2.7 kg/min. After the series
was complete, the pressure was turned up to 8.6 bar and
the next batch of samples were blasted followed by the
10.3 bar samples. After all 7.5 cm distance samples were
finished, the 15 cm guide arm was attached and the process Figure 5. Keelson plate during cleaning. (A) After cleaning; (B) before
repeated, followed by the 30 cm arm. cleaning
EQUIPMENT IDENTIFICATION AND THE DEVELOPMENT OF DRY-ICE BLASTING PARAMETERS FOR CLEANING ARCHAEOLOGICAL WROUGHT IRON, COPPER ALLOYS, AND GRAY CAST IRON 269

Project expansion yellow brass or 60:40 brass (Selwyn 2004). Modern

copper-alloy sample materials were analyzed, with those
In 2015, the BCC purchased all required machinery,
matching the artifacts’ metallurgy closest selected for
which was made possible through funds from a National
testing. While the resulting data on the elemental content
Maritime Heritage Grant from the Maritime Admin-
was qualitative, enough information was provided for a
istration overseen by the National Park Service. After
rough alloy match. Samples were prepared with the same
conducting additional research into compressor and
method as wrought iron.
air-dryer systems, and conversations with Cold Jet
over operational requirements of the SDI Select 60, Considering the hardness of wrought iron compared
the BCC selected a fixed-air system consisting of an to copper alloys, it was expected that gentler blasting
exterior-mounted Ingersoll Rand R55i rotary screw air settings would be required. Research by Van der Molen
compressor (supporting air consumption at 6.46 m3/min et al. (2011) identified successful cleaning parameters
at a pressure of 14 bar) connected to an NVC300 refrig- for bronze between 1.4–6.9 bar, using Cold Jet’s i 3
erant air dryer. MicroClean system. Conservators sought to emulate a
similar pressure range for testing. However, the hose,
Historical background: Copper-alloy artifacts applicator, and nozzle currently being used with SDI
Select 60 did not enable the system to function efficiently
In addition to wrought-iron artifacts, the Monitor
at lower pressures. Fortunately, due to the versatility of
Collection contains a significant number of copper-
the blasting unit, it was possible to swap out compo-
alloy objects, including elements associated with the
nents and attach the hose, applicator, and nozzle of the
ironclad’s two Dahlgren guns. These artifacts are covered
i3 MicroClean (Figure 6). Marketed as the Precision Kit
in a thin calcium-based concretion that is difficult to
for the SDI Select 60, the attachments allowed the unit
remove mechanically or chemically without damaging
to operate down to 1.4 bar, with a maximum media feed
the objects’ surfaces. Dry-ice blasting was explored as a
rate of 0.7 kg/min (Wright 2017). The setup includes the
possible cleaning method.
MC88 nozzle (10 cm long) producing a blasting swath of
2.2 cm, and a gun-like applicator connected to a 9.5 mm
Experimentation
diameter and 3.7 m long blasting hose.
The objects selected for trial cleaning were copper-alloy
plating from the vessel’s port gun carriage. The compo-
nent parts were chosen due to flat surfaces, uniform
patina, and concretion cover, all optimal conditions for
evaluating cleaning performance. Due to the perceived
softness of the metal, before applying dry-ice blasting to
the artifacts, conservators identified safe working param-
eters on metal samples, using the procedure developed
for wrought iron.
Figure 6. Precision
kit blasting
Methodology applicator and
 nozzle
To find appropriate testing material, the artifacts were
analyzed via portable x-ray fluorescence (PXRF) using Eighteen brass samples were blasted with shaved-block
a Bruker Tracer III-V+. Examination revealed that the media at a 90° angle at varying pressure (4.1, 5.5, and
objects were composed of copper and zinc, similar to 6.9 bar), working distance (2.5, 7.5, and 15 cm), and

Table 2. Copper-alloy sample-testing parameters. Areas highlighted in green show parameters that caused no surface damage; blue areas show
minor damage

Parameters 4.1 bar 5.5 bar 6.9 bar

0.5 kg/min 2.5 cm 7.5 cm 15 cm 2.5 cm 7.5 cm 15 cm 2.5 cm 7.5 cm 15 cm
0.7 kg/min 2.5 cm 7.5 cm 15 cm 2.5 cm 7.5 cm 15 cm 2.5 cm 7.5 cm 15 cm
270  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
media volumes (0.5 and 0.7 kg/min). To ensure working
distances were maintained during testing, new guide arms
were constructed for the smaller nozzle.
Results and application
Results identified safe working parameters at 4.1 bar
using 0.5 kg/min of media at 7.5 to 15 cm, and also at
4.1 bar using 0.7 kg/min of media at 15 cm (Table 2).
All other settings caused severe pitting to the sample
surfaces (Figure 7).
From the identified working parameters, effective cleaning
parameters were found at 4.1 bar using 0.7 kg/min of
media at 7.5 cm. When applied to the gun carriage
siding, concretion deposits were easily removed without
damaging the patina (Figure 8).
Figure 7. Copper-alloy samples, taken at 350× magnification. (A) Before
cleaning; (B) after cleaning, using shaved block at 4.1 bar with 0.5 kg/min
of media at 7.5 cm; (C) before cleaning; (D) after cleaning, using shaved
block at 6.9 bar with 0.7 kg/min of media at 7.5 cm
Figure 8. Gun carriage side before (A) and after (B) dry-ice blasting
DEVELOPMENT OF CONSERVATION PRACTICES
Historical background: Gray cast iron
In addition to the Monitor Collection, TMMP is a reposi-
tory for other archaeological maritime objects, including
a gray cast-iron four-pounder British naval gun. The
artifact was recovered from the York River, adjacent
to Yorktown, Virginia, from a shipwreck associated
with an American Revolutionary War battle (Ferguson
1939). It has been displayed both indoors and out, and
has lost its graphitized surface. During recent exterior
display, its paint coating failed, leading to exposure to
marine salts and formation of corrosion and surface
pitting. The artifact was brought to the BCC for evalu-
ation. It was determined that the object would receive
conservation treatment involving paint and corrosion
removal followed by desalination. Traditional methods
to remove the surface deposits could have been time
consuming and difficult; dry-ice blasting was explored
to accelerate cleaning.
Initial experimentation
To establish if dry-ice blasting was suitable, initial spot
testing was conducted using the equipment setup and
cleaning parameters tried with corrosion removal on
wrought iron. Conservators started with less aggressive
settings and worked higher. Results revealed that dry-ice
blasting alone was not effective at removing the paint.
However, in combination with the solubilizing effect of
Savogran Strypeeze varnish and paint remover, removal
was successful at a pressure of 10.3 bar using 1.8 kg/min
of shaved pellets at a distance of 7.5 cm.
Additional experimentation
To determine if the empirical cleaning parameters caused
microstructure damage, or to identify alternate safe
working parameters, the sample testing procedure was
applied using modern gray cast-iron samples.
Methodology
Eighteen samples were blasted using the 312S1 nozzle
at a 90° angle and a standard hose, with a range of vari-
ables including aggregate size (shaved pellet and shaved
block), pressure (5.5, 6.9, 8.2, and 10.3 bar), and volume
of dry-ice media (0.9, 1.8, 2.7 kg/min). All samples were
cleaned at a distance of 7.5 cm.
EQUIPMENT IDENTIFICATION AND THE DEVELOPMENT OF DRY-ICE BLASTING PARAMETERS FOR CLEANING ARCHAEOLOGICAL WROUGHT IRON, COPPER ALLOYS, AND GRAY CAST IRON 271

Table 3. Gray cast-iron sample-testing parameters. All samples were tested at 7.5 cm distance. Areas highlighted in green show safe parameters;
blue areas show minor surface damage

Parameters 5.5 bar 6.9 bar 8.2 bar 10.3 bar

0.9 kg/min shaved pellet shaved block shaved pellet shaved block shaved pellet shaved block shaved pellet shaved block
1.8 kg/min shaved pellet shaved block shaved pellet shaved block shaved pellet shaved block shaved pellet shaved block
2.7 kg/min shaved pellet shaved block shaved pellet shaved block shaved pellet shaved block shaved pellet shaved block

Results and application
Results showed safe working parameters at 5.5 bar using
0.9 and 1.8 kg/min of both shaved pellets and block media
(Table 3). All other settings caused pitting damage to the
was found that 5.5 bar using 1.8 kg/min of shaved-
pellet media at 7.5 cm was the most effective cleaning
setting. Dry-ice blasting in combination with the paint
remover took approximately five hours and eliminated
the paint layer and corrosion without damaging the object
(Figure 10). The gun was then desalinated.

Figure 9. Gray cast-iron samples, taken at 350× magnification. (A)
Before cleaning; (B) after cleaning, using shaved pellets at 5.5 bar with
1.8 kg/min of media at 7.5 cm; (C) before cleaning; (D) after cleaning,
using shaved pellets at 10.3 bar with 1.8 kg/min at 7.5 cm
Conclusion
Dry-ice blasting is a useful and practical mechanical
cleaning method that can be adapted for use on a
variety of materials. While it is not the solution for
every treatment conundrum, settings and accessories
can be modified to fit various cleaning situations.
Using a self-produced standardized testing method-
ology, the BCC found acceptable blasting parameters
for cleaning archaeological marine wrought iron,
copper alloys, and gray cast iron. Using dry-ice blasting
as a cleaning technique decreased working time and
allowed more effective removal of concretion, corro-
sion, and paint.

Acknowledgements
Special thanks to Machella Wright and Kevin McNeil
from Cold Jet® LLC for their support and expertise.
Thanks to Olga Trofimova and Amy Wilkerson from the
College of William & Mary’s Applied Research Center
for assisting in microscopic analysis.

References
Brush, M.B. 2010. Using dry ice for spray-paint removal
on weathering steel. Journal of Preservation Technologies
(APT Bulletin) 41(1): 1–5.
Cronyn, M.J. 1990. The elements of archaeological conser-
vation. New York, NY: Routledge.

Figure 10. Naval gun before (A) dry-ice blasting and after (B) treatment
sample surfaces (Figure 9). An exception was 6.9 bar
using 0.9 and 1.8 kg/min of media, which produced
minor pitting. Utilizing the identified parameters, it
Cold Jet 2018. www.coldjet.com/en/information/what-
is-dry-ice-blasting.php (accessed October 5, 2018).
--. “SDI Select 60.” www.coldjet.com/en/products/
product-catalog.php?id=13&type=machine (accessed
5 September 2018).
272  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

--.“I 3 MicroClean.” www.coldjet.com/en/products/ Wright, M., Cold Jet, LLC, personal communication (16
product-catalog.php?id=6&type=machine (accessed 6 February 2017).
September 2018).
Ferguson, H.L. 1939. Salvaging revolutionary relics from Authors
the York River. The William and Mary Quarterly 19(3): William Hoffman holds bachelor’s degrees in Anthro-
257–71. pology and Fine Arts from Buffalo State College, USA.
Krieg, M. 2008. Analyse der Effekte Beim Trockeneis- He has a master’s degree in Art Conservation from
strahlen. Ph.D. dissertation, Technische University Berlin, Queen’s University, Canada, specializing in objects. He
Fraunhofer IRB Verlag, Stuttgart, Germany. is presently Director of Conservation and Chief Conser-
vator at The Mariners’ Museum and Park, overseeing all
McNeil, K., Cold Jet, LLC, personal communication (18
conservation-related activities.
September 2017).
Rimmer, M. and Q. Wang. 2010. Assessing the effects of Laurie King has an undergraduate degree in Art History
alkaline desalination treatments for archeological iron from the University of Virginia, USA and a M.Sc. in
using scanning electron microscopy. The British Museum Conservation Practice from Cardiff University, UK.
Technical Research Bulletin 4: 79–86. She is currently Assistant Conservator at The Mariners’
Museum and Park, where she conserves artifacts from
Selwyn, L. 2004. Metals and corrosion: A handbook for USS Monitor.
the conservation professional. Ottawa, Canada: Canadian
Conservation Institute.
Silverman, R. and S. Irwin. 2009. Fire and ice revisited:
A comparison of two soot removal techniques for book.
International Preservation News 49: 31–35.
Spur, G., E. Uhlmann, and F. Elbing. 1999. Dry-ice
blasting for cleaning: Process, optimization and appli-
cation. Wear 233–235: 402–11.
Van der Molen, R., I. Joosten, T. Beentjes, and L. Megens.
2011. Dry ice blasting for the conservation cleaning of
metals. In Metal 2010 Proceedings of the Interim Meeting
of the ICOM-CC Metals Working Group, Charleston, 11–15
October 2010, ed. P. Mardikian et al., 96–103. Clemson:
Clemson University.
Watson, D. 2017. Environmentally responsible surface
preparation with dry ice. In WJTA-IMCA Conference
Proceedings, New Orleans, Louisiana, 25–27 October
2017. https://www.wjta.org/images/wjta/Proceedings/
Papers/2017/C3.pdf .
Weston, D.P., P.H. Shipway, and S.J. Harris. 2005. Coating
removal from an industrial polypropylene blend by cryo-
genic blasting: The development of substrate damage.
Wear 258: 392–401.
Wilson, S. 2013. Sustainable in-machine mold cleaning
and part deburring & deflashing using dry ice. In ANTEC
Technical Conference Proceedings, Cincinnati, Ohio, 22–24
April 2013. https://www.coldjet.com/media-2014/en/pdf/
ANTEC_2013_Paper.pdf.
Does Dry-Ice Blasting Allow Safe and Effective Cleaning of
Either Coated or Corroded Bronze Surfaces?
Stefania Agnoletti Annalena Brini
Opificio delle Pietre Dure Opificio delle Pietre Dure
Florence, Italy Florence, Italy
stefania.agnoletti@beniculturali.it annalena.brini@beniculturali.it
Silvia Basile Andra Cagnini
Università di Parma Opificio delle Pietre Dure
Parma, Italy Florence, Italy
Silvia150692@hotmail.it andrea.cagnini@beniculturali.it
Abstract
The invasiveness and effectiveness of bronze cleaning by dry-
ice blasting under different work parameters and on different
types of surfaces (bare alloys, corrosion, decorative patinas,
gilded bronze and protective coatings) were monitored. A multi-
method approach was used to test the effectiveness of dry ice
for removing unwanted layers and its effects on the morphology
and visual characteristics of surfaces. The results show that, while
bare alloys are not changed even under the strongest settings,
surfaces with either natural or artificial patinas are affected by
Introduction
This paper reports on the effects of dry-ice blasting on
copper-based artefacts. The effectiveness of the cleaning
treatment and the possible unwanted modifications
to the original surface were evaluated as a function of
parameters such as pressure, time, distance and mass
flow. Bronzes with various characteristics – protective
coatings, decorations and corrosion – were considered.
Cleaning operations on bronzes may have different
aims: complete removal of protective coatings and
decorative, non-original patination layers; complete or
partial removal of corrosion layers; or removal of dust
and encrustations on an original decorative patination.
Artistic bronzes are usually treated with protective
coatings to block or slow down the alteration processes.
The most common coatings are based on natural or
synthetic wax and on synthetic polymers, such as acrylics.
Ageing may lead to a reduction in protection efficiency,
with yellowing and embedding of dust and other alter-
ation products. Patination of artistic surfaces, whether
historical or modern, is conducted for both aesthetic
and protective purposes. Patination layers of variable
thickness may be produced using chemicals. Sometimes,
Simone Porcinai Monica Galeotti*
Opificio delle Pietre Dure Opificio delle Pietre Dure
Florence, Italy Florence, Italy
simone.porcinai@beniculturali.it monica.galeotti@beniculturali.it
Alessandra Santagostino *Author for correspondence
Barbone
Opificio delle Pietre Dure
Florence, Italy
santagostinob@yahoo.it
the treatment even within the mildest work parameters. On the
other hand, work parameters can be regulated in such a way
that both organic protective coatings and chemical patinas are
removed in a satisfactory way with no damage to the underlying
metal.
Keywords
dry-ice blasting, metal artworks, cleaning, patination,
corrosion, gilded bronze
a new patination layer is found to have been re-applied
over the original surface. The interaction of bronze
surfaces with the environment leads to the formation of
corrosion layers that exhibit a broad range of thicknesses,
morphologies and chemical compositions. Their exact
features depend on the characteristics of the surrounding
environment, both outdoor and burial. Thus, the original
surface of the bronze may or may not still exist (Robbiola,
Blengino and Fiaud 1998).
For all the above situations, a range of cleaning techniques
is available and often combinations of them are used to
fit specific requirements. The cleaning operation may
aim at full removal of unoriginal layers, such as protec-
tive coatings and re-applied patinas, or removal (full or
partial) of dust or corrosion.
Dry-ice blasting presents many advantages over other
cleaning methods, such as the absence of residues and
aqueous solutions, non-toxicity and environmental
sustainability (Shockey 2009). For these reasons, it is
used in several cleaning processes, including cleaning
of metal artworks (Brush 2010, Catulle and Kim 2015).
274  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Cleaning by dry-ice blasting is based on different effects Table 1. Naming, description and scope of the samples
due to the interaction of solid carbon dioxide (CO2) Protective Decoration and
Bare
pellets with the surface: mechanical impacts, thermal metal
coatings on corrosion on bronze
bronze Cu-Sn Cu-Sn
shock and sublimation of carbon dioxide (Sherman and

Simulated corrosion
Adams 1995, Uhlmann and Hollan 2015, Schröter and

Wax on acrylic

Green patina
Black patina
Domjan 2016). The possibility of regulating the operating

Gilding
Acrylic
Cu-Sn

Cu-Si

Wax
Description
parameters (mass flow, time, pressure, distance, mass
and size of pellets) makes this technique promising in
terms of providing a tunable system which combines
A1 B1 C1 D1 E1 F1 G1 H1 I1
effectiveness of cleaning with the safety of the surface Sample A2 B2 C2 D2 E2 F2 G2 H2 I2
under the materials removed. name A3 B3 C3 D3 E3 F3 G3 H3
A4 B4 C4 D4 E4 F4 G4 H4
For this study, an experiment was set up that included Damage Effectiveness Effectiveness Damage
many variables, both in terms of machine parameters on bare of removal of removal of on
and surface characteristics. Tests were carried out on metal of protective patinas and gilding
Scope
coatings damage on
nine kinds of bronze samples. Some parameters such as and surface patinas
machine pressure, mass flow, nozzle-to-surface distance condition
and time of blasting were modified, while the angle
of incidence, shape of the nozzle and the CO2 pellets Eight samples were used to monitor damage on bare
remained unchanged. metals of different composition. Two types of bronze
were used: A samples (four samples) of 90% Cu-10% Sn
On the one hand, the effectiveness of the technique composition, representative of historical alloys; and B
for removing protective coatings, rather than original samples (four samples) consisting of a silicon bronze
decorative patinas and corrosion layers, was tested. On representative of modern alloys. Twelve samples (all
the other, unwanted surface changes possibly produced of 90% Cu-10% Sn) were coated with three different
on the original surface by removing dust or corrosion protective materials: microcrystalline wax (C – four
with dry-ice blasting were monitored on bare metals, samples), acrylic (D – four samples) and a layer of
decorative patinas and gilded surfaces. The invasiveness wax on a coating of acrylic (E – four samples). These
of the dry-ice blasting on the various types of surfaces were used to check the effectiveness of ice blasting for
was assessed in terms of abrasions and visual changes, removal of protective coatings. Twelve samples had
such as variation in colour, gloss and roughness. artificial patinas on their surface: decorative black
(F – four samples), decorative green (G – four samples)
Methods
and a corrosion layer simulation (H – four samples).
Samples These were used to monitor both the effectiveness
Nine bronze sample types were studied. For each type, of removal of patina layers and the effects of dry-ice
four samples were prepared, except for the gilded blasting on the patina in cases where dust and corrosion
samples where only two samples were prepared, for a has to be cleaned and the object preserved. Finally, two
total of 34 samples (Figure 1). Each sample measured bronze coupons (I) were fire-gilded with an amalgam of
4.5 cm × 4.5 cm × 0.5 cm. The gilded samples measured mercury and gold and the mercury allowed to evaporate
5 cm × 2.5 cm × 0.7 cm. The characteristics of the samples by heating. The effect of removing dust and corrosion
are reported in Table 1 along with their purpose. from the delicate, thin gold layer with dry-ice blasting
was evaluated.

The decorative black and green patinas consisted, respec-

Figure 1. Samples prepared for the tests
tively, of liver of sulfur (K2S) and iron(III) chloride (FeCl3)
(sample F) and copper nitrate Cu(NO3)2 (sample G).
Chemicals were applied by heating the surface. On the
samples from group H, an atacamite-rich patina was
artificially created by the application of liver of sulfur
and ammonium chloride. The presence of atacamite
DOES DRY-ICE BLASTING ALLOW SAFE AND EFFECTIVE CLEANING OF EITHER COATED OR CORRODED BRONZE SURFACES? 275

was assessed with Fourier transform infrared (FTIR) high) and 2.5 cm (maximum). Finally, the high setting
spectroscopy and Raman spectroscopy. features intermediate values for pressure (6 bar) and
mass flow (0.4 kg/min) at 5.5 cm distance. Low, High,
Work parameters VHigh and Max were applied to bare samples (A, B)
to test the effects of the stronger conditions. Patinated
In this experiment, only the most relevant parameters,
surfaces (F, G, H) were tested at VLow, Low, High and
namely pressure, mass flow and distance, were varied. A
VHigh to evaluate invasiveness and efficacy. Gilded
Cold Jet i3 MicroClean was used with a fragmenting fan
samples (I) were tested at Low and High settings, while
nozzle with a 20 mm × 2 mm opening. A standard flow
the removal of protective coatings (samples C, D, E)
of dry ice was maintained on each area of the samples
was tested at VLow and Low, but increasing the time
for five seconds. However, longer times were required
of application, as specified above.
to achieve complete removal of the protective coatings
and subsequent steps of five seconds were applied when
Analysis
necessary. Cylindrical carbon dioxide pellets were used
(length 2.5–10 mm and diameter 2–3 mm). The angle Evidence of the effect of the treatment was monitored
of incidence of the flow was 90° (Spur, Uhlmann and by stereomicroscopy and variable pressure scanning
Elbing 1999) and the nozzle-to-surface distance was electron microscopy (VPSEM, EVO MA 25) before
kept fixed during application with two sticks (Figure 2). and after blasting to determine whether abrasion or
Five combinations of the three parameters were selected flaking had occurred. Both backscattered-electron
(Table 2). Low and very low (VLow) conditions had the (BSE) imaging and secondary-electron (SE) imaging
same pressure and mass flow (3 bar and 0.1 kg/min), were acquired.
while the nozzle-to-surface distance was changed (10 cm
Visual features of the surface and its changes were
in very low, 5.5 cm in low). These are the minimum
evaluated by measuring roughness, colour and gloss,
values for pressure and mass flow to obtain a stable jet
while FTIR was used to check the effectiveness of the
of pellets.
method for reducing or eliminating organic coatings.
An ALPHA Bruker spectrometer, equipped with a
reflectance module, was used for reflectance FTIR.
One hundred scans on gold were acquired at a spec-
tral range of 500–4000 cm-1 at a resolution of 4 cm-1.
The measuring spot was 1 cm in diameter and was
positioned in the centre of the sample with the aid of
Figure 2. The nozzle- a plastic mask.
to-surface distance
and inclination were Roughness measurements were carried out by means
kept fixed during
of a Surftest SJ-210 (Mitutoyo, Japan) roughness meter.
blasting with two
An arithmetical mean deviation parameter (Ra) was
 sticks
measured in two orthogonal directions. Gloss meas-
Table 2. List of conditions for blasting
urements were carried out with an ARW E-20/60/85
glossmeter, averaging five gloss (G) readings on the
Pressure Mass flow Nozzle-to-surface
Setting
(bar) (kg/min) distance (cm) same central area with the aid of a plastic mask. The
VLow 3.0 0.1 10 size of the measured area was 1.2 cm × 0.5 cm (for
Low 3.0 0.1 5.5 measurements at 60°) and 1 cm × 5 cm (for meas-
High 6.0 0.4 5.5 urements at 85°). For each sample, a suitable reading
VHigh 6.8 0.6 5.5 angle was selected according to the gloss value meas-
Max 6.8 0.6 2.5 ured in the screening test at 60°. Ultraviolet-visible
spectrophotometry (Konica Minolta CM-2600) was
Very high (VHigh) and maximum (Max) settings are used for the colour tests. These are reported as the
characterised by higher pressure and mass flow (6.8 bar total difference, Delta E (ΔE*), of the three CIE L*a*b*
and 0.6 kg/min) with a working distance of 5.5 cm (very coordinates.
276  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Results and discussion
Table 3 shows a summary of the results of the tests
performed.
Invasiveness
Tests were performed on bare metals in order to monitor
the effects produced by an extra flow of pellets striking
the surface in an uncontrolled way just after encrusta-
tions, dust and/or old protective coatings were removed.
Observations by microscope, VPSEM and roughness
measurements showed that ice blasting did not affect the
surface morphology of either the 90% Cu-10% Sn (A)
and silicon bronze (B) bare samples, even at the most
aggressive setting (Max). Colour and gloss measure-
ments indicated no changes to the visual characteristics.
The only effect of the application on bare metal was the
removal of incoherent material formed in the early stages
of corrosion or through deposition.
In the samples with a black patina (sample F), some
changes on the surface were also already evident at the
mildest setting (VLow). In sum, results showed that for
the samples with decorative patination (F, G), dry-ice
treatment at a Low setting had little influence on the
visual gloss, colour and roughness parameters. However,
monitoring of the surface with the microscope high-
lighted the formation of uncovered areas, while VPSEM
showed pellet marks. This indicated that the method is
not completely safe when used to remove dust or corro-
sion from original patination.
The stereomicroscope image in Figure 4 shows the
effects of dry-ice cleaning at VLow and VHigh settings
on sample H (with an artificially induced corrosion
patina). The mildest setting resulted in the full removal
of the corrosion layer. No significant marks on the alloy
underneath were produced even at the VHigh setting,
as confirmed by the VPSEM observations (Figure 5).

Invasiveness was also monitored on samples with patinas

and gilding. Dust and deposits may need to be cleaned
on patinated and gilded bronzes while preserving the
original surface. Figure 3 shows the effect of dry ice appli-
cation on the sample with a decorative green patina (G) at
the mildest settings (VLow and Low). In both cases, CO2
pellets left clear marks after impact on the patina surface.
The two settings differed depending on the operating
distance: the effects of the closer distance between the
nozzle and the surface at the Low setting is evident in the
higher depth of the indentations in Figure 3d. For sample
Figure 4. Stereomicroscope images of two H samples before and after
G treated with the Low setting, the following changes treatment at VLow (a and b) and VHigh (c and d) settings
after cleaning were observed: gloss ΔG = 6.28 (± 0.34),
colour ΔE = 3.57 (± 0.04) and roughness ΔRa = 0.87
(± 0.60)-vertical and 1.14 (± 0.60)-horizontal.

Figure 3. Stereomicroscope images of two G samples before and after Figure 5. VPSEM-SE images of H samples before and after blasting at
treatment at VLow (a and b) and Low (c and d) settings VLow (top) and VHigh (bottom) settings
DOES DRY-ICE BLASTING ALLOW SAFE AND EFFECTIVE CLEANING OF EITHER COATED OR CORRODED BRONZE SURFACES?
On sample I (gilded), the invasiveness of the cleaning
with CO2 was evaluated at Low and High settings. Figure
6 shows the VPSEM images before and after application.
While the Low setting did not produce any significant
effect on the surface, the High setting removed a flake
of gold about 80 μm wide. As for the variation in colour
on the gilded surface, the change of ΔE* after dry ice
application for both settings was measured (Table 4).
This was due to the cleaning effect of blasting, which
removes deposits in the porosity of the gilding layer. The
roughness and gloss measurements, shown in Table 4,
indicated a neat change in both parameters for the High
setting (an increase in Ra and decrease in glossy G),
while roughness did not change in a significant way at
Low setting.
Figure 6. VPSEM-SE images of I samples before and after blasting at Low
(top) and High (bottom) settings
Effectiveness
On patinated samples F, G and H, the process of removing
a non-original patina layer was simulated. For sample F
(black patinated), the increase in impact in the VLow
setting led to the gradual removal of the patina, with
changes in colour and roughness (Table 4). In High, the
green patina (G) was almost fully removed, as macro-
scopic observations show. The process left a thin (circa 10
µm) unaltered layer of black patina beneath the green, as
shown in Figure 7. The figure shows cross sections taken
from sample G before and after treatment. Although
this method required cutting fragments from sample G
and prevented a direct comparison for the same area, it
indicated that dry-ice blasting removed the green layer
and kept the black layer. The VPSEM image showed
micro-cracks running horizontally along the black layer,
277
a possible outcome of the impact and thermal shock
created by the CO2 pellets. The effectiveness of removing
corrosion layers (sample H) was reported in the previous
section: good results were achieved even at VLow setting.
Figure 7. Optical microscope photographs of cross sections of fragments
taken from the G samples before (left) and after (centre) treatment at
High setting. The centre and right (VPSEM-BSE) images show a thin black
patina after removal of the green layer
Figure 8. FTIR spectra of samples C (a) and D (b) before and after five
seconds of blasting at Low setting
Figure 8a shows the FTIR spectra for sample C, coated
with wax. No peaks were detected after five seconds of
application at VLow setting.
Nevertheless, stereomicroscope images (not included)
showed some residual wax on the surface. However, wax
was not visible on another replica of sample C treated for
five seconds at Low. Table 5 shows changes to ∆E* and
gloss on the same areas prior to coating application and
after application and removal with blasting for samples
from groups C and D. After the coating application, an
expected increase in gloss was measured. After blasting,
gloss returned to its initial level, showing that the removal
of the wax layer with dry-ice blasting did not significantly
278  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
affect the gloss of the metal surface. There was no signif-
icant change in Ra between the initial bare surface and
that after protection and blasting.
Figure 8b shows the FTIR spectra for sample D (coated
with acrylic) before and after blasting at the same Low
setting as in sample C in Figure 8a, but applied in four
bursts of five seconds each. The spectra showed that this
setting is not suitable for removal of acrylic coating. Only
at the High setting (three applications of five seconds
each) were the FTIR bands for the acrylic no longer
visible. The VPSEM images in Figure 9 show changes
to the surface before and after the application of acrylic
coating (a, b), and after cleaning at High (c). For this
very sample, no significant changes in Ra were detected
between the initial bare surface and after application
of the coating and blasting. In Table 5, ∆G between the
bare surface and after application of the coating and
its removal is shown: the level of gloss is similar before
coating and after the cleaning process.
Figure 9. VPSEM-SE images of sample D before and after application of
acrylic coating (a, b) and after cleaning at High setting (c). Circles point
out the position of the reference mark
These data show that the Low setting is enough to
remove the wax coating, while more aggressive settings
are needed for acrylic. This observation matches
the conclusions reported by Cutulle and Kim (2015)
that wax is the easiest material to remove from metal
surfaces. Spur, Uhlmann and Elbing (1999) have
commented that wax is more affected by the mechan-
ical impact of the pellets than other materials, owing
to its softness. This kind of coating needs low thermal
energy (higher mass pellets) and low mass flow (larger
intervals between pellets) to achieve maximum cleaning
effectiveness. Hard coatings need smaller pellets, higher
mass flow and high speed.
Figure 10 shows the FTIR spectra for sample E (acrylic
coating + top layer of wax). The spectrum of the sample
with the coating shows distinct signs of both the acrylic
and the wax. After application of mild settings of
dry-ice blasting (Low) for five seconds, changes in the
spectrum showed evidence of the removal of wax with
an increase of the acrylic C=O band at about 1745 cm-1
DEVELOPMENT OF CONSERVATION PRACTICES
Figure 10. Reflectance FTIR spectra for E samples after blasting for
different lengths of time
and disappearance of the wax bands at 2929 cm -1,
2850 cm-1 and 1480 cm-1. The same spectral profile
was obtained after four cycles of five seconds each.
This suggests that the Low setting is not suitable for
acrylic removal. The full removal of the two overlapping
layers was achieved only at the High setting and after
an overall time of 25 seconds, as demonstrated by the
FTIR spectrum (not shown here), where no signs of
acrylic or wax appear. The VPSEM images (not shown
here) confirm the full removal of the coating. Tests on
sample E matched the results of tests on samples C and
D (only wax and only acrylic coating). In fact, for the
cleaning of these coatings, a five-second application
of the Low setting led to full removal of the wax on
sample C, while to remove the acrylic from sample D,
the High setting and a longer time were needed. The
removal of the overlapping coatings of acrylic and wax
therefore requires an overall time, which is the sum
of the time to remove the wax coating and the acrylic
coating, while the possibility of selectively removing
wax is assured by the use of a Low setting and short
application time.
The visual characteristics of the metal surface after
removal of the acrylic coating with a wax top layer
were unchanged with regard to the bare surface. Ra
and gloss were monitored before and after application
of the double coating and after its removal. Noticeably,
the difference in gloss before and after application of
the coating (20.3 ± 0.2) was the same as between the
surface blasted at High setting for 25 seconds and the
coated surface (-20.2 ± 0.2). Ra was significantly lower
between the bare surface and after coating application.
Ra returned to the same value as the bare surface after
application at High setting for 25 seconds.
DOES DRY-ICE BLASTING ALLOW SAFE AND EFFECTIVE CLEANING OF EITHER COATED OR CORRODED BRONZE SURFACES? 279

Table 3. Overview of the results of the tests

Sample Scope Setting Effects Assessment

A INVASIVENESS No changes detected with Suitable for cleaning historical and
Max stereomicroscope, SEM, Ra, or gloss and modern bronzes in a way safe for
B INVASIVENESS colour measurements the bare metal
C REMOVAL OF WAX Low Full removal after short application (5 s) Suitable
Full removal only after long application Suitable
D REMOVAL OF ACRYLIC High
(15 s)
Selective removal of top layer of wax at Suitable for selective and full
E REMOVAL OF WAX + ACRYLIC Low & High Low setting and full removal of the double removal of the two coatings
layer after High and long (25 s) application
REMOVAL OF BLACK PATINA High & VHigh Partial to full removal Suitable
F Some visual and morphological changes to Not suitable for cleaning deposits
INVASIVENESS ON BLACK PATINA VLow & Low
the patina on the decorative green patina
Good removal with no changes to the Suitable
REMOVAL OF GREEN PATINA High & VHigh
surface
G Changes detected by stereomicroscope, Not suitable for cleaning deposits
INVASIVENESS ON GREEN PATINA VLow & Low SEM, colour, gloss and Ra measurements on the decorative green patina
Partial removal of patina at Low setting
REMOVAL OF CORROSION VLow Full removal Suitable
H No changes to the substrate observed by Suitable
INVASIVENESS ON METAL VHigh
optical microscope and SEM
Changes to morphology and roughness Suitable with caution
I INVASIVENESS ON GILDING Low & High
observed after application at High setting

Table 4. Variations in colour parameters L*, a* and b*, ∆E*, in roughness Ra and in gloss G on groups F and I samples treated at different settings

Colour (specular component excluded – SCE) Roughness Gloss

Sample-Setting
ΔL* Δa* Δb* ΔE* ΔRa (µm) ΔG
0.58 ± 0.60 - vertical
F-VLow 0.32 ± 0.01 -0.06 ± 0.01 0.25 ± 0.02 0.41 ± 0.02 -2.3 ± 0.2
0.51 ± 0.60 - horizontal
0.32 ± 0.60 - vertical
F-Low 3.07 ± 0.04 -0.83 ± 0.01 -0.49 ± 0.04 3.22 ± 0.04 -3.3 ± 0.2
0.33 ± 0.60 - horizontal
1.05 ± 0.60 - vertical
F-High 4.20 ± 0.03 -0.76 ± 0.01 -0.56 ± 0.01 4.30 ± 0.03 -7.7 ± 0.7
1.10 ± 0.60 - horizontal
0.93 ± 0.60 - vertical
F-VHigh 7.47 ± 0.11 -0.56 ± 0.06 -0.74 ± 0.03 7.52 ± 0,11 -5.5 ± 0.3
0.87 ± 0.60 - horizontal
I-Low 2.70 ± 0.11 0.16 ± 0.04 2.09 ± 0.11 3.42 ± 0.16 No variation -17.8 ± 2.2
0.8 ± 0.3 - vertical
I-High 8.03 ± 0.18 0.53 ± 0.05 4.60 ± 0.14 9.27 ± 0.22 -30.4 ± 6.2
0.8 ± 0.3 - horizontal

Table 5. Variations in colour parameter ∆E* and gloss G measured on the bare surface of samples C and D and after application of coatings and
blasting

ΔG
Sample-Setting ΔE*
Coated-bare surface Blasted-coated surface Blasted-bare surface
C-Low 0.67 ± 0.04 21.4 ± 0,2 -19.5 ± 0.1 1.9 ± 0.3
C-High 2.10 ± 0.03 10.1 ± 0.1 -10.3 ± 0.1 -0.2 ± 0.2
D-Low 0.48 ± 0.09 12.6 ± 0.2 0.7 ± 0.2 13.3 ± 0.4
D-High (4 applications) 2.78 ± 0.05 21.2 ± 0.2 -21.6 ± 0.1 -0.4 ± 0.3
280  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Conclusion C. Riccardelli, 13–24. Washington, DC: The American

Institute for Conservation of Historic & Artistic Works.
This study provides an insight into the use of dry-ice
blasting for cleaning metal surfaces with different charac- Spur, G., E. Uhlmann, and F. Elbing. 1999. Dry-ice
teristics. Spectroscopic and optical tests were conducted blasting for cleaning: Process, optimization and appli-
on samples that provided scientific evidence of the cation. Wear 233–235: 402–411.
effects of the treatment on the surface and helped to Uhlmann, E. and R. Hollan. 2015. Blasting with solid
define working parameters for selective removal of carbon dioxide – Investigation of thermal and mechanical
coatings, corrosion, dust and deposits in a way that was removal mechanisms. Procedia CIRP 26: 544–47.
safe for the original surface underneath. In particular,
dry-ice blasting is a suitable method for cleaning bronze Authors
surfaces of dust and corrosion, as well as non-original
Monica Galeotti, Andrea Cagnini, Simone Porcinai
decorative patinas and protective coatings. It allows
and Alessandra Santagostino Barbone are conserva-
controlled removal of overlapping layers while leaving
tion scientists at the Scientific Laboratory of the Opificio
the underlying surface of the alloy unaltered. However,
delle Pietre Dure (OPD) in Florence, Italy. Silvia Basile
dry-ice blasting is not recommended for removal of dust
received a master’s degree in 2017 from the University
or corrosion from decorative patinas and gilded surfaces
of Parma for her thesis on the use of dry-ice blasting on
as it produces visual changes and indentations even at
copper alloys. Stefania Agnoletti and Annalena Brini are
the mildest settings.
conservators in the Department of Metal Conservation
at OPD.
References
Brush, M.B. 2010. Using dry ice for spray-paint removal
on weathering steel. Association for Preservation Tech-
nology Bulletin 41(1): 1‒5.
Cutulle, C. and S. Kim. 2015. Dry ice blasting in the
conservation of metals: A technical assessment as a
conservation technique and practical application in the
removal of surface coatings. In Objects Specialty Group
Postprints, Volume Twenty-Two, eds. E. Hamilton and
K. Dodson, 77–100. Washington, DC: The American
Institute for Conservation of Historic & Artistic Works.
Robbiola, L., J.M. Blengino, and C. Fiaud. 1998.
Morphology and mechanisms of formation of natural
patinas on archaeological Cu-Sn alloys. Corrosion Science
40(12): 2083–111.
Schröter, J. and A. Domjan. 2016. Le nettoyage cryo-
génique des métaux : Évaluation du procédé appliqué
au patrimoine. Conservation-Restauration des Biens
Culturels 34: 19–25.
Sherman, R. and P. Adams. 1995. Carbon dioxide snow
cleaning – The next generation of clean. In Precision
Cleaning ‘95 Proceedings, 15–17 May, Rosemont, Illinois,
271–300.
Shockey, L.H. 2009. Blow it off: Moving beyond
compressed air with carbon dioxide snow. In Objects
Specialty Group Postprints, Volume Sixteen, eds. H. Alten,
C. Del Re, P. Griffin, E. Hamilton, K. Kipper, and
Flame Cleaning of Historic Wrought Iron: Practitioner
Methods and Their Impact on Oxide Morphologies and
Post-Treatment Corrosion Rates
Nicola J. Emmerson* David E. Watkinson
School of History, Archaeology and School of History, Archaeology and
Religion – Cardiff University Religion – Cardiff University
Cardiff, UK Cardiff, UK
EmmersonNJ@Cardiff.ac.uk Watkinson@Cardiff.ac.uk
Abstract
Flame cleaning followed by wire brushing is a common
treatment for wrought iron: the flame combusts existing
coatings and spalls oxides while wire brushing removes any
debris, producing a sound surface for recoating. Although
frequently applied, little is known about the effects of the
treatment on the substrate material and its post-treatment
corrosion rate. This study reports scanning electron microscopy-
backscattered electron imaging and oxygen consumption
corrosion rate testing of wrought iron flame-cleaned by three
practitioners. Wrought iron samples treated by two practitioners
corrode up to four times faster than uncleaned control samples.
Samples cleaned by one practitioner exhibit no increase in
corrosion rate. Torch fuel type and temperature attained by the
Introduction
Conservation of historic wrought iron
Effective management of heritage assets requires that
decision-making is underpinned by empirical evidence
of the impact of treatments on the long-term survival
of materials. Historic ironwork, which occupies a niche
position between heritage and engineering, is frequently
exposed to outdoor atmospheric corrosion and, in
the case of bridges, gates and similar structures, may
be required to perform a distinct structural function.
Sector guidance to direct practice is based largely on
anecdotal evidence and established methods (Ashurst
and Ashurst 1988, Blackney and Martin 1998, Mitchell
2005, Watkinson et al. 2005, Davey 2007, Schütz and
Gehrke 2008, Davey 2009, Wilson et al. 2010). Inter-
national and British Standards relate to modern steels
(ASTM 2008, British Standards Institute 2000 and 2007),
their application to historic ferrous metals being compli-
cated by differences in metallurgy and lack of conces-
sion to conservation ethics (International Council of
Museums 1984, American Institute for Conservation of
Johanna C. Thunberg *Author for correspondence
School of History, Archaeology and
Religion – Cardiff University
Cardiff, UK
ThunbergJC@Cardiff.ac.uk
iron are identified as parameters potentially contributing to
differences in corrosion rates. Comparing oxide morphology of
treated and untreated samples reveals extensive cracking and
fragmentation following flame cleaning. This offers multiple
pathways for ingress of oxygen and water to the metal core to
support corrosion. This data simultaneously calls into question
the ethics of flame cleaning and offers evidence that a ‘safe’
method exists, prompting further research into this popular
treatment.
Keywords
iron, corrosion, wrought iron, flame cleaning, treatment,
practitioner, oxygen consumption
Historic and Artistic Works 1994, Canadian Association
for the Conservation of Cultural Property 2000).
Recognition that corrosion of iron degrades its functional
and aesthetic properties, leading to failures of structures
and objects, is long standing with treatments applied to
iron historically to increase its useful lifespan. In theory,
reducing the rate of atmospheric corrosion of metals is
achievable by: modifying the environment to make it less
corrosive; exploiting electrochemical principles to suppress
anodic or cathodic reactions; or by separating the metal
from the corrosive environment and introducing a large
resistance to impede ion transport between anode and
cathode (Mayne 1954, Scully 1990). In practice, modern
techniques for corrosion prevention of historic iron
continue to centre on the application of organic coatings,
effectively separating the metal from its exposure environ-
ment and exploiting electrochemical principles.
Adhesion is crucial for successful anti-corrosion perfor-
mance of organic coatings and can be maximised by
thorough surface preparation, with clean metal surfaces
282  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
or sound metal oxides ideal for wetting due to their
high surface energies (Bierwagen and Huovinen 2010).
Surface preparation techniques are therefore integral to
conservation treatments for historic ironwork and aim
to remove existing paint layers and corrosion products
prior to recoating, optimising performance and longevity
of protective coatings. Methods of surface preparation
include: blasting at high pressure with abrasive media
or dry ice; immersion in alkaline solution baths to
dissolve paint layers; wire brushing or needle-gunning
to abrade flaking coatings and corrosion products; and
flame cleaning.
Flame cleaning
Flame cleaning uses a torch flame to combust paint layers
and heat the surface of the object. With the increased
temperature, the metal core expands more than the
overlying corrosion products, causing these to spall from
the surface. Wire brushing removes any remaining loose
corrosion products. Increasing focus on ethical and envi-
ronmentally friendly practices has sparked a resurgence
in popularity of flame cleaning thanks to its perceived
advantages over alternative techniques. These include:
minimal loss of original historic material; retention of
protective oxides; ease of use on-site; controllability;
ready availability of equipment; minimal detritus of
operation; and cost-effectiveness.
Despite anecdotal treatment successes, prior research
identified increased corrosion rates of wrought iron
following flame cleaning (Emmerson and Watkinson
2016) as measured via oxygen consumption (Emmerson
and Watkinson 2014, Watkinson and Rimmer 2014,
Emmerson and Watkinson 2016). Laboratory flame-
cleaned iron corroded up to four times faster than
untreated, abrasive-blasted or chemically stripped iron.
The widespread use of flame cleaning and its potential to
drive corrosion rates of ironwork, ultimately leading to
loss of heritage structures, makes empirical investigation
of the treatment an important issue for conservation
science.
Aim and objectives
This research investigated the impact of flame cleaning on
the post-treatment corrosion of historic wrought iron by:
• recording flame cleaning of historic wrought iron
samples by three historic ironwork conservation
practitioners;
DEVELOPMENT OF CONSERVATION PRACTICES
• imaging oxide morphology pre- and post-treatment
by scanning electron microscopy;
• measuring oxygen consumption of individual flame-
cleaned and uncleaned samples at high relative
humidity (RH) as a proxy corrosion rate; and
• correlating experimental results to practitioner treat-
ment methods.
Method
Sample material
Historic wrought iron was used as the sample material to
avoid the pitfalls of analogous samples and ensure direct
applicability of results to heritage practices. Mid-19th
century rolled wrought iron plate from the Kings Cross/
St Pancras gasometer was sourced. Optical and scanning
electron (CamScan Maxim 2040) microscopy confirmed
this to be wrought iron due to the presence of stringers of
slag. X-ray diffraction (PANalytical X’Pert Pro (Cu Kα))
of the corrosion products on untreated samples iden-
tified magnetite, goethite and lepidocrocite consistent
with corrosion products reported on iron subjected to
atmospheric corrosion (Bouchar et al. 2014).
The gasometer plates are of consistent thickness (4 mm)
and samples (30 mm × 40 mm × 4 mm) were cut using
a water-cooled cutting process to minimise temperature
increase and associated changes in the microstructure
of the iron. Historic wrought iron is by nature an inho-
mogeneous material (Dillmann et al. 2004) and local
differences in microstructure and slag distribution are
possible. Producing all samples from the same rolled sheet
minimised likelihood of large compositional variations.
Flame cleaning
Eight samples were flame-cleaned by each of the three
practitioners according to their preferred method. All
practitioners applied the flame to the surfaces of the
samples and intermittently brushed them with a steel
wire brush. Duration of application of the flame varied
between practitioners. The colour of the metal during
cleaning was recorded as an indicator of temperature. The
specifics of these methods were recorded for comparison.
Examining corrosion product morphology
Samples of uncleaned and practitioner-cleaned samples
were embedded in Struers Epofix resin and cross-sec-
tioned using a Struers Minitom precision cut-off machine.
FLAME CLEANING OF HISTORIC WROUGHT IRON: PRACTITIONER METHODS AND THEIR IMPACT ON OXIDE MORPHOLOGIES AND POST-TREATMENT CORROSION RATES 283

Sections were ground and polished (P600–0.25 μm),
carbon-coated and imaged in backscattered electron
imaging (BEI) mode using a CamScan Maxim 2040 scan-
ning electron microscope (SEM) equipped with Oxford
Instruments energy and wavelength dispersive x-ray spec-
trometers (Oxford Link Pentafet 5518 Caesium 7.2.17).
and individual methods (Table 1). There were notable
differences which centred on the variables of torch fuel,
temperature of the metal and duration of flame appli-
cation which is linked to determination of end point.
Prepared samples show a distinct colour change to red
oxides after treatment (Figure 1).

Oxygen consumption corrosion rate measurement

Flame-cleaned and uncleaned control samples were
enclosed individually in airtight reaction vessels (250 ml
Mason Ball glass jars with plastic-coated brass sealing
discs tightened with threaded outer sealing rings of
brass) containing 160 g of silica gel conditioned to 
Figure 1. Samples after preparation by practitioner A (a), practitioner
90% RH. To the interior wall of each vessel was adhered
B (b) and practitioner C (c), and uncleaned sample (d)
an oxygen-sensitive spot (World Precision Instruments
(WPI) part #503090 adhered with Radio Spares RTV Post-treatment oxide morphologies
silicone rubber compound). A watch glass separated
SEM-BEI images of cross-sectioned treated and untreated
each sample from the silica gel. The reaction vessels were
samples (Figure 2) evidence a reduction in oxide layer
stored in a Binder KBF240 climate chamber at a constant
thickness for all treated samples relative to untreated.
20 ± 0.5°C to avoid RH fluctuations within vessels due
Reduction in thickness is least evident in the sample
to temperature change. MadgeTech RHTemp 101A data-
prepared by Practitioner B (Figure 2c and g) and most
loggers recorded internal vessel conditions confirming
evident for Practitioner A (Figure 2b and f). Oxides remain
90 ± 3% throughout the test period for all vessels.
in pits on all treated samples. Whilst a small degree of oxide
Oxygen concentration within each vessel was measured
at regular intervals over a 108-day period using a WPI
OxyMini meter with fibre optic cable (WPI OXY-MI-
NI-AOT with cable #501644). The accuracy of the oxygen
measurements is ± 2 mbar at 210 mbar (atmospheric oxygen
pressure) and increases with decreasing oxygen pressure.
Negligible ingress of oxygen has been shown in control
vessels filled with nitrogen over this time period, indicating
a low leakage rate (Watkinson and Rimmer 2014).

Results
Practitioner methods and prepared samples
Examining the video recordings of practitioner flame
cleaning allowed identification of parameters of operation
Figure 2. SEM-BEI of untreated (a, e), and practitioner A (b, f), practitioner
B (c, g) and practitioner C (d, h) cleaned samples at 50× (a–d) and 150×
(e–h) magnification

Table 1. Parameters of flame cleaning by practitioners A, B and C

Estimated max. Duration of flame

Metal colour
Practitioner Torch temperature of application to Determination of end point
during cleaning
the metal (°C) surface (seconds)
Oxypropane (oxygen Dull red 500 c. 20–30 Visual cleanliness
A
cutting boost)
Oxypropane (oxygen Grey <150 <10 Cessation of luminescence of coatings
B
cutting boost) and oxide vestiges under flame
Oxyacetylene (no oxygen Dull/ cherry red 700 c. 50–60 No further corrosion removable or
C
boost) coating visible
284  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Figure 3. Oxygen consumption of wrought iron samples flame-cleaned Figure 5. Oxygen consumption of wrought iron samples flame-cleaned
by practitioner A over the test period of 108 days. The shape of the plot by practitioner C over the test period of 108 days
for sample A2 indicates failure of the reaction vessel seal

Figure 4. Oxygen consumption of wrought iron samples flame-cleaned Figure 6. Oxygen consumption of uncleaned wrought iron samples over
by practitioner B over the test period of 108 days the test period of 108 days

layer micro-cracking can be observed for the uncleaned comparable oxygen consumption ranges if outlier A8 is
sample (Figure 2a and e), cracking and fragmentation is included (sample numbers argue for this), with practi-
extensive for oxide layers on all treated samples. Cracks tioner B samples having a slightly larger range around a
have increased in both number and size post-treatment. similar median value and particularly above it.

Oxygen consumption measurements

All practitioner flame-cleaned and uncleaned samples
consumed oxygen during the test period (Figures 3–6).
Consumption is characterised by a faster initial rate which
slows over time for all flame-cleaned samples (Figures
3–5). Uncleaned samples exhibit a more constant oxygen
consumption rate over the test period (Figure 6).

Examining boxplots of oxygen consumption calcu-

lated by sample mass (to minimise influence of minor
discrepancies in sample dimension) over the test period
for all flame-cleaned and uncleaned samples (Figure 7)
indicates that uncleaned samples and those prepared by Figure 7. Boxplot showing the oxygen consumption rate of the wrought
iron samples flame-cleaned by practitioners A, B and C and of uncleaned
practitioner C share similar, lower oxygen consumption
samples over the test period of 108 days calculated as mbar yr-1 g-1 for
rates, with smaller consumption ranges between samples. all samples. Sample A2 has been removed from the dataset as leakage
Samples prepared by practitioners A and B exhibit of the reaction vessel was identified (Figure 3)
FLAME CLEANING OF HISTORIC WROUGHT IRON: PRACTITIONER METHODS AND THEIR IMPACT ON OXIDE MORPHOLOGIES AND POST-TREATMENT CORROSION RATES
Discussion
Corrosion product morphology
Macroscopically, the extent and form of corrosion
products on wrought iron surfaces is not altered greatly
by flame cleaning, a colour change (to brighter orange/
red) being the most noticeable aspect (Figure 1). At high
magnification, an extensive fragmentation of corrosion
product layers is evident relative to uncleaned samples
(Figure 2). While differential expansion of oxides and
metal substrate describes the mechanism by which
loosely adhering corrosion products are spalled from
the surface of wrought iron, it has likely also caused
this fracturing and cracking of closely adhering oxide
layers. This then offers pathways for ingress of oxygen
and water to the metal surface, which allows corrosion
to occur. Capillaries of small diameter increase corrosion
risk by reducing the RH at which condensation can occur
(down to 50% RH in capillaries of 1.5 nm) (Garverick
1994), hence increasing the number of micro-cracks may
increase corrosion in treated samples.
Oxygen consumption and corrosion rates
As tests have shown that nothing in the experimental
set-up consumes oxygen (Emmerson and Watkinson
2014, Watkinson and Rimmer 2014, Emmerson and
Watkinson 2016), any oxygen consumed during the test
period may be attributed to corrosion. Thus, the rate of
oxygen consumption is a proxy measure for corrosion
rate. Samples cleaned by practitioners A and B showed
a considerably higher corrosion rate than uncleaned
samples and those cleaned by practitioner C (Figure 7).
This indicates that the flame cleaning methods employed
by practitioners A and B have increased the corrosion
rate of the wrought iron relative to leaving the material
untreated. Corrosion rates of uncleaned samples and
those cleaned by practitioner C are not significantly
different, with ranges overlapping and practitioner C
cleaned samples having a slightly lower median value.
The clear implication is that the flame cleaning method
employed by practitioner C has not increased the corro-
sion rate of the wrought iron samples beyond their
uncleaned corrosion rate.
Influence of practitioner methods on post-treat-
ment oxide morphologies and corrosion rates
Temperature attained during cleaning is likely more
important than duration of flame application. No practi-
285
tioner could state with confidence the metal temperature
during cleaning, but there were noticeable differences
between methods. Samples cleaned by practitioner B did
not exceed 150°C and oxides are unlikely to have been
affected beyond the influence of differential thermal
expansion. Samples cleaned by practitioner A and C
reached approximately 500 and 700°C, respectively. It
is unlikely that changes in the microstructure of this
wrought iron would arise from heating to these temper-
atures (North et al. 1976).
Corrosion product transformations may occur at elevated
temperatures: goethite to haematite has been reported
at 240°C–250°C (Ruan et al. 2002) and lepidocrocite
to maghemite at c. 200°C which is metastable and
transforms again to haematite at 500°C (Gehring and
Hofmeister 1994). Transformation of oxides may explain
the characteristic bright red/orange colour of the corro-
sion products observed within c. 60 seconds of flame
cleaning a sample (Figure 1). Decomposition of chlo-
ride-containing iron oxides at temperatures above 380°C
has been reported (North and Pearson 1977, Kanungo
and Mishra 1996). It is not expected that these samples
contain large quantities of chloride and the duration
of high-temperature exposure during flame cleaning
may not be sufficiently long to support decomposition.
Should this be a factor influencing corrosion rates of these
samples, the increased corrosion rate of practitioner A
samples suggests that the temperature required is perhaps
towards the 700°C attained by practitioner C samples.
Clear differences in the extent of oxide remaining
post-treatment are evident between practitioners
(Figure 2). Least reduction in oxide thickness is evident
on samples treated by practitioner B, likely due to low
treatment temperatures and short duration of cleaning.
Higher temperatures and longer treatment durations for
practitioners A and C led to greater reduction in oxide
thickness. There is no direct correlation between these
parameters and the extent of oxide removal (being greater
for practitioner A than practitioner B despite slightly
reduced treatment time and temperature). It is worth
noting that the treatment variable not considered here
is length and vigour of wire brushing, which probably
influences removal of oxides.
The treatment parameter which separates practitioners
A and B from C, possibly accounting for corrosion rate
differences, is the nature of the torch and its fuel. Prac-
titioners A and B used oxypropane torches with oxygen
286  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

boost settings to increase flame temperature. Practitioner through training that a synergy of practical experience
C used an oxyacetylene torch with no oxygen boost. and understanding of chemical and physical processes
Oxyacetylene flame temperature is generally higher than can be produced to improve both practice and researcher
oxypropane, although the temperature of the oxypro- knowledge. Practitioners are an invaluable source of
pane torches with oxygen cutting boost is not known. information regarding wrought iron, its properties and
The combustion reaction equations for acetylene [1] responses to treatment.
and propane [2] support anecdotal reports from practi-
Flame cleaning is a precursor to application of protective
tioners that propane burns with a ‘wetter’ flame, which
coatings. Although it is important to understand the
is feared to introduce water to oxide layers and increase
corrosion rate of the substrate material, what is missing
post-treatment corrosion.
from this study is investigation of the performance of
2H2C2 (g) + 5O2 (g) → 4CO2 (g) + 2H2O (g)[1] coatings applied to the flame-cleaned substrates. Differ-
ences in adhesion of coatings to these sample surfaces
C3H8 (g) + 5O2 (g) → 3CO2 (g) + 4H2O (g)[2]
pre- and post-treatment, plus measurement of their
corrosion rates once coated, is required.
Flame cleaning in practice
Examining practitioner methods reveals that flame Conclusion
cleaning is carried out in an ad hoc manner with a
The popularity of flame cleaning is due to numerous
range of methods employed and different philosophical
practical and ethical advantages over other surface
approaches. Torch choice may be based on availability and
preparation techniques for historic wrought iron. This
cost; oxyacetylene is more expensive than oxypropane.
study has demonstrated that practitioner methods are
Use of the oxygen boost flame with the oxypropane torch
variable and are highly influential when considering
was deliberate to increase flame temperature and burn off
post-treatment oxide morphology and corrosion rates
paint layers more quickly. Flame application time must
of uncoated material. It has been shown that there is a
relate to the extent of corrosion product and amount
method by which flame cleaning can be applied without
and nature of existing coatings, but also to the philos-
increasing the corrosion rate of the iron, but as yet the
ophy of the practitioner. Practitioner B was concerned
parameters of this method cannot be identified. Results
with conservation ethics, preserving historic evidence
reported here indicating an increased post-treatment
within the metallographic structure and the concept
corrosion rate must introduce a note of caution to
that the metal last attained red heat during forging.
recommendations for use of flame cleaning. With the
This concern led to the shortest treatment duration and
support of the practitioners, work continues at Cardiff
lowest metal temperature. Practitioners A and C did not
University to identify optimal treatment parameters
express concern over the heating of the historic iron in
for flame cleaning and produce guidelines for the safe
relation to the material or conservation principles. This
treatment of historic wrought iron.
divergence – one practitioner using ethics to dictate end
points and two others using appearance – apparently
Acknowledgements
ignores any interpretation or conception of physical or
chemical outcome for the metal and relation to subse- The authors are grateful to Historic Scotland, Dorothea
quent application of coatings. Restorations, Heritage Blacksmith Partnership, Ironart
of Bath and Peter Meehan for their support throughout
A high level of experience and knowledge of the mate-
this research.
rials and their properties does not prevent a ‘technician’
approach to work. Unless an agreement is in place for
References
maintenance work by the same practitioner, an individual
may never be in situ to judge long-term implications of American Institute for Conservation of Historic and
decisions and practices. Training directs practitioner Artistic Works. 1994. Code of ethics and guidance for
thought processes; conservation trained practitioner B practice.
was more thoughtful regarding impacts of decisions on American Society for Testing Materials (ASTM). 2008.
retention of historic evidence, for example. It may be Standard practice for use of pictorial surface preparation
FLAME CLEANING OF HISTORIC WROUGHT IRON: PRACTITIONER METHODS AND THEIR IMPACT ON OXIDE MORPHOLOGIES AND POST-TREATMENT CORROSION RATES
standards and guides for painting steel surfaces (ASTM
D2200-17). West Conshohocken, PA, USA: ASTM
International.
Ashurst, J. and N. Ashurst. 1988. Practical building conser-
vation, Vol. 4: Metals. Aldershot: Gower Technical Press.
Bierwagen, G.P. and A.M. Huovinen. 2010. Paint Formu-
lation. Shrier’s Corrosion 4: 2643–55.
Blackney, K. and B. Martin. 1998. Development of long-
term testing of methods to clean and coat architectural
wrought ironwork located in a marine environment:
The maintenance of railings at the Garrison Church,
Portsmouth. In English Heritage Research Transactions.
Research and Case Studies in Architectural Conservation.
Volume 1: Metals, ed. J.M. Teutonico, 103–16. London:
James & James.
Bouchar, M., P. Dillmann, and D. Neff. 2014. Atmos-
pheric corrosion of iron-based reinforcement of gothic
cathedrals – Overview of possible corrosion systems
and influence of the structure and composition of the
corrosion layers on the corrosion diagnosis. In Metal
2013: Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, Edinburgh, 16–20 September 2013,
eds. E. Hyslop, V. Gonzalez, L. Troalen, and L. Wilson,
215–21. Edinburgh: ICOM-CC and Historic Scotland.
British Standards Institute. 2000. Preparation of steel
substrates before application of paints and related products.
Surface preparation methods (BS EN ISO 8504). London:
British Standards Institute.
British Standards Institute. 2007. Paints and varnishes.
Corrosion protection of steel structures by protective paint
systems (BS EN ISO12944). London: British Standards
Institute.
Canadian Association for the Conservation of Cultural
Property. 2000. Code of ethics and guidance for practice,
3rd ed. Ontario: CAC/ CAPD.
Davey, A. 2007. The maintenance of iron gates and railings.
INFORM – Information for Historic Building Owners.
Edinburgh: Historic Scotland.
Davey, A. 2009. Iron: The repair of wrought and cast
ironwork. Dublin: Stationary Office.
Dillmann, P., F. Mazaudier, and S. Hoerlé. 2004. Advances
in understanding atmospheric corrosion of iron. I. Rust
characterisation of ancient ferrous artefacts exposed to
indoor atmospheric corrosion. Corrosion Science 46(6):
1401–29.
287
Emmerson, N. and D. Watkinson. 2014. Preparing
historic wrought iron for protective coatings: Quantitative
assessment to produce evidence-based protocols. In Metal
2013: Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, Edinburgh, 16–20 September 2013,
eds. E. Hyslop, V. Gonzalez, L. Troalen, and L. Wilson,
119–27. Edinburgh: ICOM-CC and Historic Scotland.
Emmerson, N.J. and D.E. Watkinson. 2016. Surface
preparation of historic wrought iron: Evidencing the
requirement for standardisation. Materials and Corrosion
67(2): 176–89.
Garverick, L. 1994. Corrosion in the petrochemical
industry, 2nd edition. ASM International.
Gehring, A.U. and A.M. Hofmeister. 1994. The trans-
formation of lepidocrocite during heating: A magnetic
and spectroscopic study. Clay and Clay Minerals 42(4):
409–15.
International Council of Museums. 1984. The Conserva-
tor-Restorer: A Definition of the Profession. www.icom-cc.
org/47/about/definition-of-profession-1984/ .
Kanungo, S.B. and S.K. Mishra. 1996. Thermal dehy-
dration and decomposition of FeCl3.xH2O. Journal of
Thermal Analysis 46: 1487–1500.
Mayne, J.E.O. 1954. How paints prevent corrosion.
Anti-Corrosion Methods and Materials 1(8): 286–90.
Mitchell, D.S. 2005. Boundary ironwork: A guide to re-in-
statement. INFORM – Information for Historic Building
Owners. Edinburgh: Historic Scotland.
North, N., M. Owen, and C. Pearson. 1976. Thermal
stability of cast and wrought marine iron. Studies in
Conservation 21(4): 192–97.
North, N. and C. Pearson. 1977. Thermal decomposition
of FeOCl and marine cast iron corrosion products. Studies
in Conservation 22(3): 146–57.
Ruan, H.D., R.L. Frost, J.T. Kloprogge, and L. Duong.
2002. Infrared spectroscopy of goethite dihydroxylation:
III. Spectrochimica Acta Part A 58: 967–81.
Schütz, A. and J. Gehrke. 2008. Modern corrosion
protection on historic buildings. Materials and Corrosion
59(3): 228–31.
Scully, J.C. 1990. The fundamentals of corrosion, 3rd ed.
Oxford: Pergamon Press.
Watkinson, D. and M. Rimmer. 2014. Quantifying
effectiveness of chloride desalination treatments for
288  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

archaeological iron using oxygen consumption measure-

ment. In Metal 2013: Proceedings of the Interim Meeting
of the ICOM-CC Metals Working Group, Edinburgh,
16–20 September 2013, eds. E. Hyslop, V. Gonzalez, L.
Troalen, and L. Wilson, 95–101. Edinburgh: ICOM-CC
and Historic Scotland.
Watkinson, D., M. Tanner, R. Turner, and M. Lewis. 2005.
SS Great Britain: Teamwork as a platform for innovative
conservation. The Conservator 29: 73–86.
Wilson, L., A. Davey, D.S. Mitchell, and A. Davidson.
2010. Traditional architectural ironwork: Scientific
approaches to determining best conservation practice
and the Bute Canopy study. In Metal 2010: Proceedings
of the Interim Meeting of the ICOM-CC Metals Working
Group, Charleston, South Carolina, 11–15 October 2010,
eds. P. Mardikian, C. Chemello, C. Watters, and P. Hull,
248–55. Charleston, SC: Clemson University.

Authors
Nicola Emmerson lectures in conservation science at
Cardiff University. Her research focuses on the manage-
ment of corrosion prevention strategies for metallic
cultural heritage artefacts.

David Watkinson teaches and researches conservation

at Cardiff University. His work on the corrosion rate of
iron underpins the preservation of Brunel’s iron steam-
ship SS Great Britain.

Johanna Thunberg has held conservation and teaching

roles at Cardiff University, where she is undertaking an
AHRC-funded PhD. Her research centres on controlled
microclimate storage protocols for archaeological metals
and identification of instability in copper alloy artefacts.

Burnishable and Conductive: A New Filling Possibility for
Losses in Gilded or Metallised Surfaces
Clara Huynh* Dominique Robcis
CH Conservation Centre de recherche et de
Paris, France restauration des musées de France
CH.Conservation@gmail.com Paris, France
*Author for correspondence
dominique.robcis@culture.gouv.fr
Abstract
This work focuses on filling solutions for losses in metallic
coatings. Research in the field of metal conservation is often
concerned with treatments to improve stability of objects or with
consolidation options for corroded metals. However, the increase
in design-, industry- and contemporary art-related heritage
artefacts in collections presents new problems for conservators.
Losses in metallised surfaces or mirror-like finishes are a
challenge to repair: effective treatments for the consolidation
or reattachment of loose fragments or sheets have been found,
but successful retouch methods for metallic substrates remain
difficult to establish. This paper reports on a graphite-based
fill which acts as a traditional gilding bole but is suitable for a
metal substrate, in this case ferrous. After solvent evaporation,
it can be worked easily, burnished to a shine and covered with
Introduction and context
This project started with the restoration of a cast-iron
stove. Produced in the 1930s by the French manufacturers
Godin, it is displayed in Le Familistère de Guise museum
in France (Figure 1).
Figure 1. The Godin 700D model (© INP/A. Dequier)
Young Conservation Professional
Françoise Urban
Independent metal conservator
Brussels, Belgium
michiels.urban@gmail.com
a metal layer to imitate more closely the original appearance of
the surface. This is achieved either by activating the acrylic resin
with solvent to create a size on which gold foil or metallic leaf
can be applied, or by exploiting the conductivity of graphite to
perform an electroplating treatment. The commercially available
products used here were also investigated to achieve a better
understanding of their performance. Samples were artificially
aged to confirm the stability and compatibility of the systems
with their oxidised ferrous substrate.
Keywords
graphite, gilding, filling, loss compensation, nickel, chrome,
plated
The nickel-plated fittings came at extra cost to the buyers
when ordering the stove from the factory. Some damage
to the nickel had occurred, likely due to accidents or poor
storage. The missing layer was thin, leaving the conser-
vator very little depth with which to work. In order to
preserve the aesthetic value of this luxury range stove, it
was decided to fill some of the main losses.
Retouch on metal with mica has delivered good results
(Caru 2007), but it can produce a glittery effect on losses
larger than ca. 1 mm2 that is far from the shine of metal.
The nickel of the stove was still glossy, so reproducing
this aesthetic was important. Gilding techniques using
burnished white coloured metallic foil resulted in a
comparable shine.
A technical limitation was then encountered: tradi-
tional gilding boles are made of gesso and proteinic
glues (Saulo 1886/2008) and are inappropriate for use
on metal objects because of their hygroscopicity. A
metallic substrate such as cast iron reduces the range
of compatible products to mostly synthetic, non-wa-
290  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

ter-based ones. Unfortunately, acrylic media cannot be

burnished fully (Sawicki 2007).

There are few publications concerned with this issue

(Anastasow 1999) and the research reported here aimed
to establish and refine a new filling technique by deter-
mining the products, parameters and methods capable
of producing the best outcome within the ethics of
conservation.

Formulation
The putty is an acrylic resin loaded with a large propor-
tion of graphite powder. It can be metallised to obtain
the right colour and finish by two methods: either a thin
metallic foil is applied after activating the resin with
solvent in a process similar to gilding with size, or a
layer of metal is added by localised electroplating which
exploits the electrical conductivity of graphite.
Choice of the resin
Paraloid B-44, a methyl methacrylate and ethyl acrylate
copolymer, was used at a 25% w/v concentration in a
mix of ethyl acetate and diacetone alcohol (70:30). A
solvent with slow evaporation was mandatory to avoid
bubbles becoming trapped in the fill during drying. The
solution had to be sufficiently fluid; the higher viscosity
of a greater resin concentration created application
difficulties.
Figure 2. The macro and SEM images showing the variation of particle
size and shape between the three graphite powders tested (D. Robcis/C.
Huynh)
Electrolytic metallisation
Two commercially available solutions for nickel electro-
plating were used. The application surface was approx-
imately 1 cm2 and the optimal potential difference was
found to be 6–7 V. The deposits were both relatively
free of impurities and of good thickness, with 1–3 μm of
metal being deposited in three minutes (Figure 3). The
satin finish of the electroplating had less shine than foil.
All three graphite powders tested were equally suitable
for electroplating.

Choice of the quantity of graphite powder

After several tests, the optimum balance was found to be
45% graphite powder in the fill, which was burnishable
from 30% graphite but produced a better finish with a
higher content. With more than 45% graphite powder
Figure 3. XRF spectra of electrodeposited nickel (#882 Dalic; # 883
the fill became too viscous to be applied successfully.
Newminal) on ferrous substrate, showing very few impurities (© D.
Robcis)
Characterisation and study
Graphite quality Palladium foil

Three commercially available graphite powders were
used. All showed good results and could be burnished,
but noticeable differences were observed. Study under
scanning electron microscopy (SEM) revealed that the
particle size of those ‘fine’ powders varies widely from
ca. 1 µm to more than 200 µm (Figure 2). The finer the
graphite, the slower the solvent evaporation and the
denser the resulting fill. All fills were sufficiently soft to
be marked with a fingernail and thus easily workable.
Palladium was chosen as it offered a darker tone and
greater stability than silver-based foils such as white gold.
Two palladium foils were investigated and both behaved
similarly. They were applied after activation with a 2:1
ethanol:water solution and could be burnished following
solvent evaporation. X-ray fluorescence surface analysis
showed no conclusive modification of foil thickness
before (#879) and after (#880) burnishing (Figure 4).
Adhesion of the foil to the fill was good and followed the
BURNISHABLE AND CONDUCTIVE: A NEW FILLING POSSIBILITY FOR LOSSES IN GILDED OR METALLISED SURFACES
contours of the surface of the fill very closely (Figure 5).
Test samples of this fill on ferrous alloy were artificially
aged through light exposure and humidity fluctuations,
as reported previously (Huynh 2017).
Figure 4. XRF comparison of unburnished (#879) and burnished (#880)
foil. Similar thicknesses are indicated by comparable peak heights for
palladium and iron before and after burnishing (© D. Robcis)
Figure 5. Microscope image of palladium foil applied to the abraded fill
showing close adhesion to fill surface morphology (© D. Robcis)
Results and discussion
The system did not favour oxidation of the substrate,
acting rather like a thick varnish. It is not sufficiently
hard-wearing to be applied to ‘in-use’ objects but is
suitable for metallic cultural heritage.
Resulting fills with palladium foil or nickel electrolysis
were visually close to the smooth warm-white coloured
metallic coating that adorned some parts of the stove.
Losses in the nickel were restored successfully with this
method. Figure 6 illustrates the process using a palla-
dium foil.
Further work and conclusion
Further testing with graphite as a load but with different
resin and solvent combinations should be carried out.
291
Figure 6. Four steps of the treatment process (left to right): nickel losses,
the dry and adjusted fill, after applying a palladium foil and the result
after surface finishing (© INP/C. Huynh)
This could allow the use of this method for other mate-
rials with high gloss surfaces, e.g. plastics. Application
of a coloured veil of varnish, using Microlith pigments
to adjust the final shade, has been tested.
The fill investigated here – a mix of acrylic resin and
graphite powder – is a burnishable alternative to commer-
cially available acrylic-based fills. It appears to be revers-
ible with solvent, stable to ageing and compatible with
a ferrous substrate, requiring material and equipment
common in conservation workshops. Through working,
a large range of textures can be obtained, from gritty to
almost mirror-like. It is hoped that this research contrib-
utes a treatment option for conservators with a range of
applications.
Acknowledgements
The authors wish to acknowledge the following persons
for their support during the first phase of the project:
Virginia Costa, Anne Genachte-Le Bail, Anne-Cécile
Viseux, Martine Bailly, Marie-Anne Loeper Attia and
Frédéric Panni. For their precious input, we thank Leslie
Villiaume, Nelly Koenig, Aurélie Compère and Isabelle
Cuoco. We also thank the Fondation pour les Monuments
Historiques and the Crédit Agricole d’Ile de France for
their help.
References
Anastasow, S. 1999. Ergänzung metallischer Überzüge –
Nickel und Chrom. Master’s thesis, University of Applied
Sciences (HTW) Berlin, Germany.
Caru, L. 2007. Réemploi et transformation : Conserva-
tion-restauration d’un ensemble d’éléments métalliques
dorés et argentés de style classique. Recherche d’un
médium de réintégration des argentures lacunaires,
Mémoire INP.
Huynh, C. 2017. Conservation-restauration d’une cuisin-
ière Godin (1931) en fonte de fer émaillée et nickelée –
292  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Familistère de Guise (Aisne). Mise au point et évaluation

de systèmes de comblement-retouche pour revêtement
métallique brillant, Mémoire INP.
Saulo, M.J. (1886) 2008. Nouveau manuel complet de la
dorure sur bois suivi de Fabrication des peintures laquées,
coll. Manuels Roret, reprint. Aix-en-Provence: Éditions
Émotion Primitive.
Sawicki, M. 2006. Practical implications of research into
non-traditional in-gilding techniques: Loss compensa-
tion in conservation of gilded objects. AICCM Bulletin
30(1): 63–69.

Materials list
Dauvet Palladium (pure)
Laverdure Palladium (pure)

Plating solution Dalic “N°EP 11020”

“Nickel” Newminal plating pen

Niton XL3t portable XRF spectrometer: configuration Ag

(147-00X) anode, 40 µA max., tension 50 kV max; high
performance Si-PIN semiconductor sensor; Hirox KH
8700 3D digital microscope; Optic: triple Revo 35-2500

Rohm and Haas Paraloid B-44®

Merck (supplier Millipore Sigma) 1042062500
Kremer Silver 47700
Kremer Black 47710

Authors
Clara Huynh is a conservator in private practice at CH
Conservation. She graduated from the Institut National
du Patrimoine in 2017 with a main focus on the conser-
vation of metals. Prior to graduation, she studied and
worked in Italy and Germany, and completed internships
at the Grassi Ethnographic Museum (Leipzig) and Die
Neue Sammlung design museum (Munich). Her research
interests include creative filling and retouching tech-
niques, and conservation pedagogy.

Gap Fillers for Brittle Metal Artifacts and Their
Suitability in High Temperature and High Relative
Humidity Conditions
Abdelaziz Elmarazky* Alice Boccia Paterakis
The Grand Egyptian Museum – Japanese Institute of Anatolian
Conservation Centre Archaeology
Giza, Egypt Kaman, Kirsehir, Turkey
abdelaziz.elmarazky@yahoo.com alicepaterakis@yahoo.com
Abstract
Archaeological metal artifacts often suffer from embrittlement
and require physical support to protect them from further
damage caused by injudicious handling. The extreme climatic
conditions in Middle Eastern countries call for conservation
materials that withstand high temperature and high relative
humidity. Four polymer solutions with glass microballoons were
tested as gap fillers for thin and fragile archaeological metal
artifacts in the conservation laboratory of the Japanese Institute
Introduction
Museum collections in Middle Eastern countries are
often subjected to large yearly fluctuations in ambient
conditions since their indoor environments are uncon-
trolled. The delicate nature of their objects dictates a
lightweight and readily soluble fill material with a glass
transition temperature (Tg) high enough to withstand
such temperature extremes. A literature search in the
Abstracts of International Conservation Literature
(AATA) database revealed very little research and
testing has been done on fill materials for weakened
archaeological metals.
Aging tests exposing fill materials to elevated temperature
and relative humidity (RH) under fluctuating conditions
were run in the conservation laboratory of the Japanese
Institute of Anatolian Archaeology (JIAA) in Kaman,
Turkey, in 2017.
Methods of study
Only polymers with glass transition temperatures (Tg)
that exceed the high temperatures encountered in the
Middle East were considered. Based on the equipment
available in the conservation laboratory of the JIAA, it
was decided to run accelerated aging tests using Paraloid
B-48N, Paraloid B-44, Butvar B-98, and Butvar B-76
Young Conservation Professional
*Author for correspondence
of Anatolian Archaeology in Turkey. It was concluded that
Paraloid B-44 with glass microballoons was the most resistant
fill material for delicate metal artifacts.
Keywords
gap filler, microballoon, Butvar, Paraloid B-44, relative
humidity, acetic acid, Japanese Institute of Anatolian
Archaeology
due to their proven long-term solubility and successful
application with glass microballoons.
Ten percent solutions (w/v) of Paraloid B-48N and
Paraloid B-44 in acetone and 10% solutions (w/v) of
Butvar B-98 and Butvar B-76 in ethanol were mixed
with 3M Scotchlite glass microballoons (density of 0.25
g/cc, isostatic crush strength of 750 psi) in a ratio of one
part polymer solution to two parts fill material (v/v), as
shown in Table 1.
The test samples were formed into inert plastic molds
of equal size (3.2 × 3.2 × 0.3 cm) and allowed to reach
gravimetric equilibrium before proceeding with acceler-
ated aging. Aging tests were carried out in high relative
humidity and at high temperature under fluctuating
conditions.
Fluctuating temperature and relative humidity
Three test samples of each of the four polymer groups
were exposed to high humidity in a vacuum desiccator
that was heated with a 500-watt tungsten lamp for 12
hours each day, for a period of 448 hours (19 days), as
presented in Table 2. The elevated humidity was obtained
by placing water in the base of a bell jar, and a Hobo
294  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Table 1. Polymers used in the fluctuating temperature and relative humidity test

Brand Polymer Tg (°C) Mw Viscosity (cP) Solvent used

Butvar B-76 polyvinyl butyral 62–72 90,000–120,000 18.0–28.0* ethanol
Butvar B-98 polyvinyl butyral 72–78 40,000–70,000 6.0–9.0** ethanol
Paraloid B-44 methyl methacrylate and ethyl acrylate copolymer 60 140,000 460*** acetone
Paraloid B-48N methacrylate methyl methacrylate and butyl acrylate 50 250,000 3000*** acetone
* 5% solids in ethanol @ 25°C; ** 6% solids in methanol @ 20°C; *** 40% solids in acetone @ 25°C

Table 2. Test parameters

Test parameters
Temperature RH Absolute humidity
Max. Min. Average
Test
(heat & Max. Min. Average Max. Min. Average absolute absolute absolute
duration
moisture) (°C) (°C) (°C) (%) (%) (%) humidity humidity humidity
(hours)
(mmHg) (mmHg) (mmHg)
All polymers 39 23 32 ± 6 100 48 83 ± 16 52 10 30 448

H08-003-02 datalogger was used to monitor the relative
humidity and temperature.
A Gastec 81D passive dosimeter for acetic acid (detection
limit of 0.2 ppm/10 hours) was used to measure vola-
tile acetic acid off-gassing from the polyvinyl butyral.
The maximum acetate content (% polyvinyl acetate)
is reported to be 2.5% for Butvar B-98 and 1.5% for
Butvar B-76 (see Talas, Product Information, accessed
05 August 2017).
Results
Gravimetric analysis
The test samples were assessed gravimetrically. Most
samples showed a weight gain in the fluctuating
temperature and relative humidity test, with Paraloid
B-48N giving the greatest weight gain, as shown in
Table 3.
Table 3. Gravimetric analysis of samples in the fluctuating tempera-
ture and relative humidity test
Gravimetric analysis
Average
Average (%)
beginning
Volatile acetic acid analysis
The Gastec 81D passive dosimeter for acetic acid
(detection limit of 0.2 ppm/h) measured 0.09 ppm
of volatile acetic acid off-gassing from the poly-
vinyl butyral after 448 hours. The maximum acetate
content (% polyvinyl acetate) is reported to be 2.5%
for Butvar B-98 and 1.5% for Butvar B-76. The total
weight of Butvar (3.48 g) multiplied by 2% (median
acetate content of both Butvars) and divided by the
5.8-liter bell jar volume gives a potential concentra-
tion of 0.835 ppm. The measured value of 0.09 ppm of
volatile acetic acid represents 10.8% of the theoretical
maximum concentration that could potentially off-gas
from the test samples.
Conclusion
The weight gain of the test samples was likely a result
of moisture absorbed by the polymers, a phenomenon
that could result in hydrolytic degradation. The weight-
gain values coincide with the molecular weights of the
polymers, as shown in Graph 1.
9
Molecular 8
Average weight gain (%)

Polymer weight gain 7

weight (g) weights (mw)
after testing 6
for all samples 5
4
Paraloid B-44 0.57 4% ± 0.01 140,000 3
2
Paraloid B-48N 0.60 8% ± 0.02 250,000 1
0
Butvar B-76 0.58 3% ± 0.01 90,000–120,000 40.000 90.000 140.000 250.000

Butvar B-98 0.58 2% ± 0.03 40,000–70,000 Molecular Weight (mw)

GAP FILLERS FOR BRITTLE METAL ARTIFACTS AND THEIR SUITABILITY IN HIGH TEMPERATURE AND HIGH RELATIVE HUMIDITY CONDITIONS
The use of Butvar cannot be recommended for the
conservation treatment of objects that are kept in well-
sealed display cases or storage cabinets over the long term
since the volatile acetic acid could increase to unaccept-
able levels. There is an inverse relationship between rela-
tive humidity and carbonyl concentrations. The higher
the relative humidity, the lower the acid concentration
required for carbonyl corrosion to occur and vice versa
(Paterakis 2010). Recommended safe levels of volatile
acetic acid are < 0.4 ppm/year for unleaded bronze and
< 0.02 ppm/year for leaded bronze at 52%–86% RH, and
< 0.4 ppm/year for leaded bronzes at 32% RH (Paterakis
2016).
The acetic acid target levels must be adjusted according
to the length of time the objects will be in storage or exhi-
bition. If the use of Butvar resins cannot be eliminated,
then the acetic acid vapor issue can be managed through
the use of adsorption trays or ventilation systems for the
display or storage cases. The fills made with Paraloid
B-44 and 3M Scotchlite glass microballoons were the
easiest to sand after filling gaps in copper-alloy vessels.
It is recommended that further testing be carried out
with different concentrations and ratios to assess the
suitability of these polymers for use in fill materials for
copper-alloy artifacts.
Acknowledgments
The authors would like to thank Dr. Ian Macleod for his
support and interpretation of the analytical results, and
Dr. Sachihiro Omura, director of the JIAA, for permission
to carry out this research project.
References
Dow Coating Materials. n.d. Paraloid™ Thermoplastic
Resins. www.palmerholland.com/News/file.axd?-
file=%2F2017%2F01%2FDow+PARALOID+Thermo-
plastic+Resin+Chart.pdf (accessed July 2018).
Paterakis, A.B. 2016. Volatile organic compounds and the
conservation of inorganic materials. London: Archetype
Publications.
Paterakis, A.B. 2010. The formation of acetate corrosion
on bronze antiquities: Characterisation and conservation.
Doctoral thesis, University College London, UK.
TALAS. n.d. 3M Scotchlite™ Glass Bubbles. Product Infor-
mation. www.talasonline.com/images/PDF/ProductIn-
295
formation/glass_bubbles_info.pdf (accessed 05 August
2017).
TALAS. n.d. Butvar® B-76. Product Information.
www.talasonline.com/images/PDF/ProductInforma-
tion/B-76INFO.pdf (accessed 05 August 2017).
TALAS. n.d. Butvar® B-98. Product Information.
www.talasonline.com/images/PDF/ProductInforma-
tion/B-98INFO.pdf (accessed 05 August 2017).
TALAS. n.d. Paraloid B-48N 100%. Product Information.
www.talasonline.com/images/PDF/ProductInformation/
paraloid_b-48n_100_pct.pdf (accessed 05 August 2017).
TALAS. n.d. Paraloid™ B-44 100%. Product Information.
www.talasonline.com/images/PDF/ProductInformation/
paraloid_b-44.pdf (accessed 05 August 2017).
Materials list
All polymers and 3M Scotchlite™ glass microballoons:
TALAS
330 Morgan Ave
Brooklyn, NY 11211, USA
Phone: 212-219-0770
http://talasonline.com
Gastec 81D passive dosimeter:
Gastec Corporation
8-8-6 Fukayanaka, Ayase-Shi
Kanagawa, 252-1195, Japan
Tel: +81(0)467-79-3911
www.gastec.co.jp/files/user/asset/pdf/en/detector_
tube/81D.pdf
Hobo H08-003-02 datalogger:
Onset Computer Corporation
470 MacArthur Blvd.
Bourne, MA 02532, USA
www.onsetcomp.com/products/data-loggers/
h08-003-02
Authors
Abdelaziz Elmarazky graduated from the Faculty of
Archaeology, Conservation and Restoration Department
at Cairo University. He has worked as a conservator in
the Grand Egyptian Museum – Conservation Centre,
Egypt, since 2015. He is a specialist in restoration and
conservation of ancient Egyptian metal artifacts whose
interests include conservation and restoration of brittle
296  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

metals as well as identification of unusual corrosion

products on copper and bronze objects.

Alice Boccia Paterakis has a MAC in Conservation

from Queen’s University and a Ph.D. in Conservation
from the University College London Institute of Archae-
ology. She has served as director of conservation of the
Kaman-Kalehöyük Excavation for the Japanese Institute
of Anatolian Archaeology (www.jiaa-kaman.org/en/ ),
Turkey, since 2008. She was also a lecturer in the Art
Conservation Program at Scripps College, Claremont,
California.
Conservation of Saint George Killing the Dragon,
a 16th-Century Gilt Copper Statue from Gdansk:
A Collaborative Approach
Anna Baranowska-Fietkiewicz Cátia Viegas Wesolowska*
The National Museum in Gdansk The National Museum in Gdansk
Gdansk, Poland Gdansk, Poland
a.fietkiewicz@mng.gda.pl c.wesolowska@mng.gda.pl
Abstract
The National Museum in Gdansk (NMG) houses a unique figure
of Saint George Killing the Dragon made of chased and gilded
copper sheet. The figure was once displayed on the rooftop of the
Court of the Brotherhood of Saint George, a building in the centre
of the historic town of Gdansk built in 1487–94. The museum’s
recent efforts to return the figure to displayable conditions led to
a collaborative project between the NMG, the National Maritime
Museum (NMM) and Chicago-based company G.C. Laser Systems.
The NMG envisages having the figure on display in the museum
by the end of 2019. The question of provenance, materials and
Historical background
The statue of Saint George Killing the Dragon (museum
number MNG/SD/417/Mt) originally stood on the roof
of a late-Gothic building in Gdansk erected in 1487–94
by Hans Glottau. The building was the seat of an elite
brotherhood of knights founded in the mid-14th century
whose members came from the richest noble families in
the city. These associations provided their members with
training in chivalry and organised shooting competitions
Figure 1. Detail of the helmet from right, showing the crest of the town
of Gdansk; the date 1566 is engraved on both sides. Photograph by the
National Museum in Gdansk
Katarzyna Schaefer Bartosz Dajnowski
The National Maritime Museum in G.C. Laser Systems Inc.
Gdansk Chicago IL, USA
Gdansk, Poland bartek@gclasers.com
k.schaefer@nmm.pl *Author for correspondence
conservation methods are described, focusing on the use of
laser cleaning as a viable conservation treatment method. Laser
cleaning is a very promising conservation approach and further
work on its advantages and disadvantages in the field of metals
conservation would be a benefit to the profession.
Keywords
copper, gilding, laser, x-ray fluorescence spectroscopy, statue,
gold leaf, Saint George
and tournaments. They were especially popular in the
territory of the State of the Teutonic Order (German
Deutschordensland or Deutschordensstaat) between
1226 and 1525 in those areas previously inhabited by
the Prussians (a Baltic people who, in the Middle Ages,
occupied an area that currently covers northern and
north-eastern Poland, Russia’s Kaliningrad region and
ancient Livonia – modern Latvia and Estonia).
The statue of Saint George Killing the Dragon was probably
made in 1566, a date engraved on the knight’s helmet next
to the coat of arms of Gdansk and the Polish heraldic
eagle (Figure 1).
As far as the artist is concerned, documents related
to the Gdansk Brotherhood of Saint George are only
fragmentarily preserved and the account books from
the period of interest are missing. However, a few years
earlier, in 1561, the new rooftop of the Main Town Hall
in Gdansk was decorated with a gilded copper figure of
the Polish king Sigismund II Augustus (Simson 1918–24,
357–58; Zarębska 1998, 43–50, 56, 62, illus. 61 and 62).
The author of the project was Dirk Daniels, a Dutch
architect from Zeeland who worked in Gdansk, and
two Gdansk armourers – Agatius Grabow and Johann
298  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND DEVELOPMENT OF CONSERVATION PRACTICES

Clus. The gilding of the statue was most probably the

work of the goldsmith Arndt. The figure of the king was
destroyed during the bombing of the city in March 1945;
currently, a copy made in 1950 on the basis of preserved
photographic and measurement documentation is located
on the tower of the town hall.
The making of both figures, King Sigismund II Augustus
and Saint George, was separated by a five-year period.
The authors can only suppose that wealthy noblemen
associated with the Brotherhood of Saint George wanted Figure 2. Detail of the
to decorate their headquarters in an equally effective way poleyns (knee guards)
as the city authorities. Similarly, it is impossible to know on the St. George statue.
Photograph by the
for sure whether the order for the figure was given to the
National Museum in
same craftsmen.
 Gdansk
Stylistically, the statue as a whole is characterised by a
certain stiffness. The contours of the figure of the saint
and especially the elements of the dragon’s body are
treated almost graphically and there is little of the fluidity Figure 3. Etching
of the Court of the
which characterises contemporary monumental statues Brotherhood of
of St. George created in Germany. It is worth paying Saint George by
attention to the fact that while the dragon’s silhouette J.C. Schultz, 1868,
is characterised by a certain primitiveness of form, the museum number
MNG/SD/3663/G/29.
individual elements attached to the body of the male Photograph by the
figure could be a testimony to the good work of the National Museum in
armourer (Figure 2). However, it should be remembered  Gdansk
that in contrast to the figures decorating fountains, placed
slightly above the viewer’s eyes and perfectly visible from From the autumn of 1944, as the Germans began to
all sides, the figure in question was conceived as the retreat and the front approached, a mass evacuation of
culmination of the building and, thus, was to be visible works of art began in nearby villages and deeper into
only in general outline and from a great distance. Germany. The statue survived thanks to its evacuation
to Bielkowo near Gdansk, where it was hidden in a
The armour is typically equestrian in form, a style in use
water pump station. In 1945, after the end of hostilities,
around the mid-16th century in German countries, with
Polish conservation services began the campaign of
certain elements of earlier origin such as the fluted faulds.
transporting surviving works of art back to Gdansk. As
The shape of the breastplate, the form of the helmet – a
the former Court of the Brotherhood of Saint George lay
Burgonet – and the design of the sabatons correspond to
in ruins after the air raids in March 1945, the figure was
the period of about 1550. The couters and poleyns (elbow
stored at the Gdansk Pomeranian Museum (the current
and knee guards) are carefully rounded with decoration
in the form of a twisted rope, with the edges thickened National Museum) and was entered into the museum
by curls and decorated with small, slanting incisions. registry. The building was rebuilt between 1951 and
The couters and pauldrons have additional decorative 1953 and a copy of the statue made by Marian Ogorzeja
besagues in the form of rosettes cut from metal sheet. was placed on the roof. The original figure remains to
this day in the collection of the National Museum in
The history of Saint George Killing the Dragon is not easy
Gdansk (Figure 4).
to trace (Rolland and Strzelecka 2006, 61–63, 477, and
previous literature). The building of the Brotherhood
Materials and later interventions
of Saint George changed functions many times over the
centuries, but the statue was preserved on the roof, as The statue is made of 2 mm chased rolled copper sheet
can be confirmed by iconographic sources (Figure 3). and measures approximately 3.35 meters in height (Figure
CONSERVATION OF SAINT GEORGE KILLING THE DRAGON, A 16TH-CENTURY GILT COPPER STATUE FROM GDANSK: A COLLABORATIVE APPROACH
Figure 4. Archival
photo from the
1950s when the
statue was on
display in the
museum cloisters.
Photograph by the
National Museum in
 Gdansk
Figure 5. Saint George Killing the Dragon, currently made up of 11
separate parts. Photograph by the National Museum in Gdansk
5). Individual elements are mounted with rivets and the
figure has an internal steel frame, which strengthens
and secures the upright form. All of the elements of the
composition (St. George, the dragon and the globe) are
mounted on a thick steel angular rod.
Gilding on copper is well described by Scott (2002). The
gilding methods available to craftsmen in the mid-1500s
consisted of applying gold leaf and fire gilding. Only traces
of the original gilded surface remain, with the majority of
what exists due to later interventions. Records describe
repairs being carried out in 1832 (Gliński 1997, 148).
After 1870, the turret crowning the roof was dismantled
and rebuilt in 1882 in a different form, only returning
to its previous shape in 1898. In the meantime, in 1888,
the statue was again renovated by a group of Gdansk
enthusiasts (Gliński 1997, 175). A guide to Gdansk from
1894 describes the ‘richly gilded’ figure, which led the
authors to believe that during the last intervention it was
completely re-gilded (Püttner 1894, 94).
In 2009, the statue was subject to analysis for the first
time by the Pomerania Laser Laboratory at the Polish
Academy of Sciences. Examination by x-ray fluorescence
spectroscopy (XRF), laser-induced plasma spectroscopy
(LIPS) and scanning electron microscopy with ener-
gy-dispersive x-ray spectroscopy (SEM-EDS) revealed
299
two gilding materials: Au, considered the original gilding,
and a Zn-Cu alloy applied during one of the restorations
in the 19th century (Iwulska et al. 2011). Hg was not
identified in either analysis. It is probable that the method
applied was that of oil gilding, mostly used for statues in
an outdoor environment. The thick, brushed wax coating
was also analysed by µ-Raman and identified as beeswax.
Soot (C), CaCO3, particles from the environment and
the corrosion product antlerite (Cu3SO4(OH)) were
also identified. The laboratory also performed test laser
cleaning on the surface with irradiation by single laser
pulses (0.3 J/cm2). This method detached the wax layer
which was then mechanically and chemically removed,
but it also damaged the gold layer in some of the tested
areas. After the tests were carried out, the figure was
placed in storage until 2019 (Figure 6) and no further
work was carried out.
Figure 6. The head
of the statue after
laser cleaning tests
(right side of the face)
carried out in 2009.
Photograph by the
National Museum in
 Gdansk
Condition before conservation
The statue was exposed to harsh outdoor conditions for
almost 400 years. The original gilding would have acted
as a protective layer that prevented corrosion of the
underlying copper provided the gilding was consistently
applied. However, where moisture and oxygen had pene-
trated cracks and porosity in the gold foil, the substrate
metal became exposed and started to react chemically
with the elements. The abrasive action of wind-blown
material would have accelerated the surface and struc-
tural damage of the gold layer. Once the blue-green
copper corrosion products formed underneath, they
spread and expanded to cause further cracking, lifting
and loss of the gilt layer.
300  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Copper corrosion products are not uniform in formation
and differ locally in thickness and composition in visually
distinctive areas. This is due to the shape of an object,
which differs in terms of the exposure of the materials
to individual environmental factors in different areas,
resulting in environmental components being transported
to the surface at varying speeds. Corrosion appears to
be more advanced in crevices and areas where particles
and water were trapped. Antlerite, the copper corrosion
product identified in 2009 (Iwulska et al. 2011), is common
on objects in outdoor environments. It is particularly
common in areas where there are atmospheric deposits.
Antlerite is a basic copper sulfate that forms on exposed
copper alloys in polluted urban environments where pH
levels are higher than 4 (Scott 2002).
During one of the 19th-century renovations, the statue
was re-gilded and a thick (and uneven) coating of
beeswax (Iwulska et al. 2011) was applied by brush.
Beeswax is hydrophobic, an advantage in protecting
outdoor copper alloys. However, its protective effect in
these environments is relatively short-lived and further
reduced by the impact of light. There are visible signs of
physical repairs all over the statue, with different types
of rivets and screws having been added, together with
new copper sheeting formed into the original shape.
There are also remains of tin soldering on the helmet
(Figure 1), but there are no records of these repairs or
any further conservation.
Since 2007, the statue has been stored in the metals
storage room at the NMG where conditions are fairly
stable, with temperatures ranging between 15 and 19°C
and an RH lower than 30%. The object was found to be
stable with no active corrosion.
Conservation programme
In 2018, the museum decided to re-establish a conser-
vation programme for the statue with the purpose of
placing it once again on display in the museum galleries.
The decision to apply laser cleaning to this object came
about after considering different possible treatments. The
factors that led to the choice of treatment were:
• the size of the object and the time factor;
• the successful use of laser cleaning in the past (at
certain pulse measurements);
• the fact that it is a safe alternative to mechanical
methods (ensuring secure removal of insoluble corro-
sion products); and
DEVELOPMENT OF CONSERVATION PRACTICES
• the fact that no chemicals are involved.
Laser cleaning is a promising method for cleaning fragile
and fine layers such as gilded copper surfaces. One of
the main problems associated with it is identifying the
correct length of the laser pulse duration in order to
avoid removing original layers. G.C. Laser Systems, a
Chicago-based company, work intensively on conserva-
tion of heritage projects and have built their own laser
equipment specifically for this purpose. As the experience
of the user plays an important role in effective cleaning,
collaboration with G.C. Laser Systems appeared to be the
perfect choice. Initial tests were carried out on sample
coupons of gilded copper sheet, as described in point 7.
Laser cleaning is intended to remove surface encrusta-
tions as well as corrosion layers and will hopefully reveal
original gilding layers. Further conservation work, which
is not described in this paper, includes creating protective
coatings and structurally stabilising the object for display.
Portable x-ray spectroscopy
XRF measurements were obtained using a Bruker III
Tracer SD with a 4W Rhodium anode and a Bruker
XFlash SDD with 2048 channels. Spectra were acquired
using a voltage of 40 keV, an anode current of 12 µA and
a yellow filter (12 mil Al, 1 mil Ti). The spot size was
an 8 mm wide oval on the short side. An accumulation
time of 15 s for each spectrum was applied. Data were
analysed qualitatively with Bruker ARTAX software using
Bayesian deconvolution.
Figure 7 presents the spectra for two areas analysed: first,
the surface cleaned by laser (2009); second, an uncleaned
area. The data in both figures indicate the presence of
Ca, Cr, Fe, Ni, Cu, Zn, Pb, Sr and Sb. Elements such as
Rh and Pd originated from the x-rays and the internal
collimator of the detector, respectively. In both cases, Au
was identified (Figure 7). This would point to the original
gilding, most probably leaf applied with an oil. However, a
higher intensity of Au can be observed on the surface on
the left side of the forehead of the St. George figure, which
was laser-cleaned, as opposed to the right side, which was
not. The wax and corrosion compounds may have had
an influence on the analysis results, hence the lower Au
intensity found (Figure 7, lower spectra). Alongside the Au,
small intensities of Ag were detected. Both spectra showed
the presence of a combination of Cu and Zn which could
possibly relate to the 19th-century gilding using Dutch
metal, a Cu-Zn alloy. However, due to the fact that at least
CONSERVATION OF SAINT GEORGE KILLING THE DRAGON, A 16TH-CENTURY GILT COPPER STATUE FROM GDANSK: A COLLABORATIVE APPROACH
Figure 7. The XRF spectra acquired (top) from the forehead of Saint
George (left side) that was potentially gilded and laser-cleaned; and
(bottom) from the forehead (right side). Photograph by the National
Maritime Museum in Gdansk
two conservation interventions were made over a long
period of time and the complexity of the protective layers
that might have been applied, more research is required
to identify the possible compositional differences between
the different parts of the St. George figure. This would help
to further support the hypothesis regarding the type of
gliding applied. In addition, further spectroscopic methods
could be used to identify copper corrosion products on the
surface of selected parts of the statue.
Laser cleaning samples: Tests
There are three basic effects that can occur on a surface
when a laser pulse interacts with a material: photo-
mechanical, photo-thermal and photochemical. Laser
parameters such as pulse duration directly influence
the balance between these three effects. A shorter pulse
duration results in relatively more photomechanical
shock to a surface than a longer pulse of equal fluence/
energy density. The goal with laser cleaning is to identify
the ablation threshold of the unwanted contaminant
while staying safely below the damage threshold of the
substrate being cleaned.
A 70W GC-1 laser cleaning system was used for the
tests. This is a 1064 nm air- cooled pulsed laser system
301
with a tuneable pulse duration range of 10–520 ns, a
pulse frequency range of 10 kHz–1 MHz and a hot-spot-
free circular beam delivery scan with an rpm range of
3,000–12,000.
Figure 8. The test grid on gilded copper showing the results of changes
in pulse duration and fluence. Detailed 3D map (140×) of damaged
gilding and laser delaminated oil size. Photograph by G.C. Laser Systems
In order to begin to identify safe laser parameters for
cleaning the gilding on the figure, a series of systematic
tests were performed on freshly prepared oil gilded copper
sheet to identify damage thresholds for the gilding. The
copper sheeting that was used was intentionally very thin,
around 0.37 mm thick, so that it would offer less potential
heat dissipation than a thicker piece of copper. One single
layer of 23 ¾ carat gold leaf that is only 1.5 microns thick
was brush-applied without burnishing. Ripples in the
gold were visible under magnification and the samples
were allowed to sit for four days before testing. The gold
layer was purposely applied thinly and not tightly bonded,
which makes it more susceptible to damage to help define
a safer approach towards cleaning a more hardened gilding
application which has lasted for many years.
Figure 9. Detail (35×). Ten ns pulses with a fluence of = .15 J/cm^2.
Laser-cleaned top half is glossier. Photograph by G.C. Laser Systems
302  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Each area was exposed to at least 10 passes of the laser
scan to ensure maximised exposure and reveal any
potential damage risks at given settings. During actual
treatment of a historic object, any unnecessary laser
exposure would be avoided. After the cleaning tests, the
samples were analysed under 140× magnification and
the surface was subject to 3D mapping with a Hirox 3D
microscope.
The laser was operated at various pulse durations and
fluence values while scanning at 12,000 rpm. Figure 8
shows the grid of laser tests with one parameter change
along each axis of the test grid. Along the X axis, the
fluence (F = J/cm^2) doubled in value with each consec-
utive cell to give the following: F = 0.0375, 0.075, 0.15,
0.30, 0.60 and 1.2. Along the Y axis, the pulse duration
doubled in length with each consecutive cell to give: 10
ns, 20 ns, 40 ns, 80 ns, 160 ns and 320 ns. The test grid
clearly shows that the results varied from no noticeable
change to the surface to an increase in the gloss levels
of the gilding and very obvious damage and removal of
the gold layer and spalling/delamination of the oil size
due to laser-induced spallation.
Figure 10. The top half was exposed to 80 ns pulses with a fluence of
0.3 J/cm^2 (140×). There is no alteration to the surface at these settings.
Photograph by G.C. Laser Systems
The test grid indicates that laser pulses with pulse
durations shorter than 40 ns and a fluence of around
0.3 J/cm^2 appear to burnish the gold leaf as the treated
area is noticeably glossier. The stronger photomechan-
ical effect of the shorter pulses could be responsible for
flattening out some of the micro-ripples in the gold leaf,
thus increasing the gloss. This phenomenon is observed
with 10–40 ns pulses and decreases noticeably with longer
pulse durations. In fact, it does not appear to occur at all
with 80 and 160 ns pulses with the same fluence value of
0.3 J/cm^2 (Figure X3).
DEVELOPMENT OF CONSERVATION PRACTICES
These tests suggest that a fluence level of 0.3 J/cm^2 could
be used as a relatively safe maximum energy density to
work with, and that longer pulses, such as 80–160 ns
pulses, should be explored further for this object as they
exert less photomechanical pressure on the gilding.
Further work
Further laser cleaning tests are required until a safe
parameter for cleaning gilded copper surfaces is obtained.
But promising results mean that laser cleaning is a
possible solution for the conservation of the figure of
Saint George Killing the Dragon. Gold strongly reflects
1064 nm laser light, which makes laser cleaning of gilded
surfaces a reasonable technique to be explored. Further
historical research and scientific analysis to map the
original gilding and determine later interventions are
also planned.
Acknowledgements
The authors would like to thank the Pomerania Laser
Laboratory at the Polish Academy of Sciences for
providing data from analysis carried out in 2009.
References
Gliński, K. 1997. Gliński Mirosław, Kukliński Jerzy. In
Kronika Gdańska, vol. I, 997–1997. Gdansk.
Iwulska, A., R. Jendrzejewski, M. Sawczak, I. Zmuda-Tr-
zebiatowska, M. Sawczak, G. Śliwiński, and A. Krieg-
seisen. 2011. Pomerania Laboratory – A solution for
the cultural heritage research and conservation. In
Proceedings of the International Conference on Lasers
in the Conservation of Artworks (LACONA VIII), Sibiu,
Romania, 21–25 September 2009, eds. R. Radvan, J.F.
Asmus, M. Castillejo, P. Pouli, and A. Nevin, 161–66.
Boca Raton, FL: CRC Press.
Knotkova, D. and K. Kreislova. 2007. Atmospheric corro-
sion and conservation of copper and bronze. In Environ-
mental Deterioration of Materials, ed. A. MoncMonova,
chap. 4, 107–42. Ashurst, UK: WIT Press.
Püttner, E. 1894. Nordostdeutsche Städte und Land-
schaften. No. 2. Danzig, ehemalig Freie Reichs- und Hans-
estadt, jetzt Hauptstadt der Provinz Westpreussen. Danzig.
Rolland, B. and I. Strzelecka, eds. 2006. Katalog Zabytków
Sztuki w Polsce, vol. VIII, Miasto Gdańsk, Część 1. Główne
Miasto. Warsaw.
CONSERVATION OF SAINT GEORGE KILLING THE DRAGON, A 16TH-CENTURY GILT COPPER STATUE FROM GDANSK: A COLLABORATIVE APPROACH 303

Scott, D.A. 2002. Copper and bronze in art: Corrosion,

colorants, conservation. Los Angeles, CA: Getty Publi-
cations.
Simson, P. 1918–24. Geschichte der Stadt Danzig, Bd. 2:
1517–1626. Danzig.
Zarębska, T. 1998. Przebudowa Gdańska w jego złotym
wieku. Oficyna Wydawnicza Politechniki Warszawskiej.

Authors
Anna Baranowska-Fietkiewicz is the metals curator
at the NMG where her field of interest is base metals.
She lectures on the topic of Pomeranian goldsmiths and
silversmiths. She is currently working on the new pewter
gallery and is also writing the catalogue on the museum’s
clock collection.

Cátia Viegas Wesolowska is deputy head of conserva-

tion at the NMG. She is currently working on the conser-
vation of clocks and is conducting PhD research on the
work of the enamellist Stefan Knapp at Birmingham City
University. She has previously worked as a senior metal
conservator at the V&A Museum in London.

Katarzyna Schaefer is deputy head of conservation

at the NMM. Her current focus is the scientific analysis
of underwater archaeological artefacts. She has a PhD
in iron corrosion.

Bartosz Dajnowski is an objects conservator who has

worked with lasers for over 15 years and has an MS in
art conservation science from Winterthur/UD. He is the
vice director of his family business, the Conservation of
Sculpture & Objects Studio Inc., inventor of the patented
GC-1 laser cleaning system and founder and president of
G.C. Laser Systems Inc., a company that manufactures
lasers for art and architecture conservation in the USA.
CONSERVATION
TREATMENTS
Redressing the Past: The Examination and Treatment
of an Ancient South Arabian Censer
Jean-François de Lapérouse* Federico Carò
The Metropolitan Museum of Art The Metropolitan Museum of Art
New York NY, USA New York NY, USA
J-F.DeLaperouse@metmuseum.org federico.caro@metmuseum.org
*Author for correspondence
Abstract
A technical and scientific examination of an ancient South Arabian
cast copper-alloy censer acquired by The Metropolitan Museum
of Art (The Met) in 1949 revealed aspects of its manufacture
that shed further light on the distinctive casting technology of
ancient Yemen. This study also uncovered extensive damage
resulting from its treatment before arrival at the museum. A
technical analysis and retreatment of this object allowed for an
Introduction
The ongoing conflict in Yemen—a war that has caused
so much human suffering—has also exerted a consider-
able toll on that country’s cultural heritage. To curb the
illegal trade in antiquities, ICOM published the Yemen
Red List in 2018 that includes a range of representative
objects found in collections around the world intended to
assist border control agents in identifying contraband. A
request made to The Metropolitan Museum of Art (The
Met) to include a censer in its collection (49.71.2) on this
list prompted its first technical examination since it was
acquired as a gift in 1949.
History of the Met’s censer
Few of the antiquities from Yemen found in museums
around the world have a known excavation history.
Our censer is first attested in the collection of Kaiky
Muncherjee (1925), who owned a successful shipping
company in Aden in the early 20th century. Muncherjee
amassed a large collection of ancient artifacts that was
offered for sale in the 1920s. Although individual objects
were purchased subsequently, the bulk of this collection
remained intact until his death in 1955 and formed the
basis of the holdings of the National Museum in Sanaa.
The Met’s censer consists of a circular main body (approx-
imately 16.5 cm wide and 7.5 cm tall) on a conical stand
Adriana Rizzo Danijela Jovanovic
The Metropolitan Museum of Art Central Institute for Conservation
New York NY, USA Belgrade, Serbia
adriana.rizzo@metmuseum.org art.artefacts@gmail.com
accurate restoration of its most salient decorative features while
uncovering remains of the original core on the interior.
Keywords
copper alloy, archaeology, conservation, corrosion, Yemen,
Fourier transform infrared microspectroscopy, scanning
electron microscopy, Raman spectroscopy, x-ray fluorescence
(8 cm tall and 14.5 cm wide) with a tall front panel
(19.5 cm tall and 12.8 cm wide) decorated with a disk
and crescent between a pair of snakes in low relief. The
seven spires that originally rose from the top edge of the
tall front panel evoke an architectural context—perhaps
the sacred building in which this censer would have been
used. The fully modelled ibex that projects from the tall
side (8.5 cm long, 14.5 cm tall and 5.7 cm wide) served
both as a handle and an important cultic symbol.
Avanzini has examined the important role that ibex
figures played in South Arabian art of the early first
millennium BCE, particularly in the Jawf and the region
of Marib, where a blend of Sabaean and Madhabaean
cultures eventually resulted in a standardized form of
writing and artistic representation (Avanzini 2002).
Several carved stone inscriptions feature representations
of ibexes within framing elements. As these inscriptions
can be dated, the style of their ibex representations—
which include an emphasis on large curving horns, a
voluminous body and a blunt muzzle with a flat beard—
can be used to date ibex representations on uninscribed
objects. Based on these comparisons as well as a related
Minaean stone incense burner from the Jawf, Avanzini
dates the Met’s censer as well as examples in the British
Museum and the University of Pennsylvania Museum
to the 8th to 7th centuries BCE.
REDRESSING THE PAST: THE EXAMINATION AND TREATMENT OF AN ANCIENT SOUTH ARABIAN CENSER
A photograph taken in the early 1920s when the censer
was still owned by Muncherjee reveals that it was covered
overall with a rough layer of burial accretions and corro-
sion. Three of the spires were extant, the ibex was held
in place with knotted wire and large losses had occurred
around much of the foot. It was purchased by a Bombay-
based art dealer who in turn sold it to Joseph Brummer, a
prominent dealer based in Paris and New York, in 1932. A
thumbnail image on Brummer’s inventory card indicates
it was received broken into many pieces. By the time it
was acquired by Dr. Charlat at auction following Brum-
mer’s death in 1947 and given to The Met in memory
of his parents, the censer’s exterior corrosion had been
mechanically cleaned down to an even surface, the spires
were lost and the breaks repaired. The uneven interior
surface appeared to be painted overall (Figure 1).
Figure 1. Censer (49.72.2) in the late 1920s (left) and in 1949 (right)
Recent examination of this censer under natural and
ultraviolet light (Reskolux UV 365 LED; 365 nm =/- 5 nm
UV-A) and x-radiography (Phillips MCN 321 Industrial
x-ray unit with exposures at 300 Kv, 3.2 mA, 30 seconds)
revealed that it had been extensively fractured and reas-
sembled using both an insoluble organic adhesive and
lead-tin solder. Paint and pigmented fill materials had
been liberally applied to disguise losses and fill missing
sections while the interior was partially filled with plaster
and coated overall with a thick layer of shellac.1 Two
irregularly shaped metal strips were imbedded in the
shellac to bridge a large loss in the raised panel while the
lower edge of the foot was replaced with painted plaster
and a rim of modern copper (Figure 2).
Radiographs also revealed the extent of the damage and
restoration of the ibex handle. Fractures had occurred
through the neck, ears and at numerous points along the
two horns. The horn sections, which had been drilled
for pinning, were joined with soft solder while the body
307
and platform were reattached to the burner using new
metal posts and adhesive. Comparing the restoration
of this ibex with the carved stone examples as well as a
comparable figure in the Museum’s collection that may
have served a similar function (53.47.2) (Muscarella
1988) indicated that the head had been poorly restored
with no articulation of the ears while the muzzle and
beard were rounded off with fill material (Figure 3).
Figure 2. Front of censer in natural light (left), UV light (center) and x-ray
radiograph (also facing forward, 300 Kv, 3.2 mA, 30 seconds) (right)
Figure 3. Ibex handle on censer in natural light (left), x-ray radiograph
(center) and the comparable ibex figure in The Met’s collection (53.47.2)
(right)
The Met’s censer within the context of South
Arabian metal casting
To date, Mille et al. (2010) have published the only
in-depth study of South Arabian large-scale casting
centering on the late 7th to early 6th century BCE cast
copper-alloy sculpture of Hawtar’athat, now in the
National Museum of Sanaa. One of the most striking
aspects of this sculpture and the other examples he exam-
ined is the thinness of their metal walls—as little as 3 mm
in the case of the Hawtar’athat statue. He proposes that
this characteristic as well as the close conformity between
interior and exterior surfaces indicates the use of indirect
casting in which wax was applied in liquid or sheet form
to the face of molds taken off a model. After casting,
308  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CONSERVATION TREATMENTS

ceramic cores were left in place to provide support for vated on the back of the tall panel for the attachment of
the thin metal and, in the case of large castings, were the ibex handle (Figure 4). The rather crude treatment
made hollow to reduce overall weight. of the core found on all three metal censers may reflect
the fact that the interiors were covered by separate metal
Evidence supporting the use of wax sheets in creating
pans which have not survived. They may also have only
the Met’s censer is found at the right-angled corner
been meant to be viewed from the front. A sense of their
of the main body of the censer, where a line seems to
use in ancient sacred contexts is described in a roughly
indicate the use of separate sheets as well as the uniform
contemporary passage in Leviticus 16:12–13: “The priest
thickness of the walls. In this case, it is likely that sheets
filled the censer with live coal from the sacred fire on
were applied to a core that had been modeled into shape.
the altar of burnt-offering, and having carried it into
After the wax was applied and the low relief elements
the sanctuary, there threw upon the burning coals the
were added to the tall side, the entire form was covered
sweet incense which sent up a cloud of smoke, filling the
with an investment of clay before firing to remove the
apartment with fragrance.”
wax. The ibex handle was modelled separately and cast
with an internal core.

In lost-wax casting the core normally is kept in place

after the wax is removed by the use of metal nails or
rods that bridge the gap between the core and the outer
investment. In the case of the statue of Hawtar’athat,
these core supports were removed after casting and
replaced by small copper-alloy rods hammered into place.
Radiographs of the two South Arabian bull statuettes
in The Met’s collection (47.100.85 and 2002.34) reveal
the presence of such supports, although their number
and arrangement differ. No evidence of the use of core
supports, however, is found in the main body of the
censer. This may be due to the fact that the censer’s core Figure 4. Interior
was not enclosed and could be held in place by the outer of censer showing
investment. No core supports were found in radiographs  extant core

of the ibex handle either, but this may be due to damage

and subsequent restoration. Core examination

Casting objects as thinly walled as this censer—the
uncleaned bottom edge of the foot indicating an orig-
inal thickness of 2–3 mm—presented a considerable
technical challenge in ensuring that all parts of the mold
received metal before solidification of the alloy began.
Presumably, the mold was well vented to ensure that
entrapped gases not absorbed by the porous core could
escape without causing “shorting.” It is also possible that
Yemeni craftsmen preheated molds immediately prior to
the metal pour—a longstanding practice in Asian metal
casting (Pillai et al. 2002).
The core was rather roughly cut back on the interior
after the censer was freed from the investment. The hole
through to the base may have been made at this time,
although the deep gash on one side of this central hole
appears to have been made recently. The core was exca-
Samples of the core, collected from the interior of the
bowl and from the back of the raised side, were prepared
as polished thin sections and studied by petrographic
microscopy (PM) and scanning electron microscopy
with energy-dispersive spectroscopy (SEM-EDS).1 Both
core samples display similar textural and compositional
characteristics (Figure 5) consisting of a very abundant
framework fraction, with moderate to well-sorted
particles ranging from about 20 μm to 150 μm and a
median grain-size of about 80–100 μm, in a subordinate,
fine marly clay matrix. The fine matrix represents the
minor fraction of the core, so that the structure is almost
grain-supported.
The framework grains are sub-angular to well rounded
and include quartz, altered feldspar, amphiboles, mostly
hornblende; small ortho- and clinopyroxenes, finely
REDRESSING THE PAST: THE EXAMINATION AND TREATMENT OF AN ANCIENT SOUTH ARABIAN CENSER 309

archaeological metal that has received different post-ex-

cavation treatments that could strongly affect the accu-
racy of the analysis. From a qualitative standpoint, all of
the XRF spectra were characterized by the dominant K
lines of copper and tin, and by the L lines of lead. Other
less intense lines are those of iron, cobalt, nickel, zinc,
arsenic, silver and antimony. No other elements were
found above the detection limits of the instrument.

The alloy compositions of these objects, calculated as

normalized weight percentages, are presented in Table 1.
It is important to note that these compositions might not
reflect the true compositions of original alloys. Some
differences should be expected because of segregation,
depletion and enrichment of selected elements at the
object’s surface and because of contaminations from
other materials, such as soil embedded in corrosion
Figure 5. Micrographs of the core sample from the interior of the bowl
products, solders and polishing compounds. Despite
taken at 5× (top) and 10× (bottom), showing the abundant silicate
framework characterized by scattered hornblende grains. The left these known limitations, the collected data indicates that
images are taken using plane polarized light, the right images using all the analyzed objects are leaded bronzes of medium to
crossed polars high tin contents, a general composition that parallels
dispersed iron and iron-titanium oxides, calcite parti- that found in Mille’s analyses.
cles, and less abundant biotite, ilmenite and rare zircon.
Investigation of the two metal cross sections taken
Rock fragments are also quite rare and mostly consist
from the circular body of the censer and the foot by
of round, altered micritic limestone and small aphanitic
means of optical microscopy (Figures 6 and 7) and
volcanic rocks.
SEM-EDS-EBSD revealed the high degree of mineraliza-
The core is tempered with a moderately abundant (about tion of the sampled metal walls (Figures 7 and 8), a fact
15–20 %) fraction of broken vegetal fibers that preserve that underscored the limited reliability of XRF analyses.
on occasion their cellular structure. These fibers are of The corrosion products mostly consist of cuprite, parat-
varying sizes with jagged, truncated ends. The presence
of carbonized vegetal fibers accounts for the gray to black
coloration of the core and for the high macro-porosity
of the material. In composition and texture this core is
similar to that found in the statue of Hawtar’athat as
reported by Mille et al. (2010) and tentatively associated
to alluvial or colluvial deposits of the Sanaa basin.

Metal analysis
Figure 6. Color micrographs in crossed polars of cross sections from the
Given curatorial preference for non-destructive analysis, foot (left) and wall (right) of the censer with the exterior sides down
open-architecture x-ray fluorescence (XRF) analysis
was performed on the censer and the four other South
Arabian copper-alloy sculptures previously noted.2
Permission also was granted to take two samples from
the censer to perform SEM-EDS and electron backscat-
tered diffraction (EBSD) analysis of their cross sections.

The surface condition of the five objects was found to be Figure 7. Backscattered electron images of same cross-section samples
extremely variable and characteristic of highly corroded as in Figure 6
310  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CONSERVATION TREATMENTS

Table 1. Alloy compositions of five South Arabian sculptures in the Met’s collection. The composition of a reference leaded copper alloy is also
included (*). The large variability of these calculated compositions reflects the highly corroded state of this object

Object Location Fe Co Ni Cu Zn As Ag Sn Sb Pb
2002.34 PL head 0.10 0.04 0.03 68.93 0.01 0.97 0.05 11.90 0.16 17.81
2002.34 PL head 0.15 0.02 0.00 71.93 0.01 0.60 0.05 9.52 0.13 17.58
47.100.85 PR ear 1.45 0.02 0.12 68.36 0.01 0.51 0.07 17.81 0.20 11.45
47.100.85 PR front leg 0.99 0.03 0.12 60.69 0.01 0.65 0.14 18.33 0.17 18.87
47.100.85 PR front leg 2.87 0.04 0.15 53.57 0.01 0.59 0.11 22.73 0.27 19.67
53.47.2 PL base 0.28 0.02 0.02 71.70 0.01 0.54 0.03 12.89 0.23 14.27
53.47.2 PL base 0.20 0.02 0.02 71.60 0.01 0.55 0.07 12.79 0.22 14.53
53.47.2 PR front leg 0.14 0.03 0.03 61.44 0.01 0.47 0.04 14.41 0.25 23.18
2002.457* PL ear break inner 0.27 0.05 0.03 61.72 0.01 0.24 0.03 11.69 0.13 25.85
2002.457* PL ear break middle 0.19 0.06 0.02 60.47 0.05 0.19 0.01 10.52 0.12 28.38
2002.457* PL ear break outer 0.11 0.02 0.00 89.72 0.01 0.29 0.04 2.53 0.01 7.27
49.71.2 PL horn 0.17 0.10 0.03 42.36 0.17 0.39 0.11 13.46 0.04 43.17
49.71.2 PL horn 0.08 0.09 0.02 54.79 0.07 0.25 0.08 9.85 0.04 34.74
49.71.2 PR horn 0.29 0.10 0.02 67.26 0.03 0.00 0.03 9.05 0.02 23.13
49.71.2 PR horn 0.13 0.11 0.02 55.27 0.17 0.47 0.02 10.30 0.06 33.46
49.71.2 Ibex PL front leg 0.10 0.05 0.01 79.30 0.01 0.85 0.00 9.07 0.08 10.51
49.71.2 Ibex PL front leg 0.12 0.03 0.00 67.94 0.01 0.38 0.01 9.05 0.06 22.37
49.71.2 Vessel rim 0.26 0.12 0.03 53.12 0.01 0.53 0.04 15.11 0.11 30.67
49.71.2 Vessel rim 0.34 0.10 0.03 40.98 0.02 0.29 0.02 12.13 0.12 45.98
49.71.2 Vessel rim 0.20 0.10 0.03 47.34 0.02 0.35 0.03 11.46 0.05 40.44
32XLB14 Calculated 0.10 0.00 0.25 79.51 0.01 0.05 0.05 5.37 0.05 13.41
32XLB14 Reference 0.06 0.00 0.31 78.08 0.04 0.05 0.06 5.41 0.06 14.62

acamite and atacamite3 admixed in variable proportions
and arranged in a characteristic layered microstructure
(Figures 6–8).4 The inner core of the metal is rather
porous and consists of predominant, well-developed
cuprite and less abundant paratacamite, atacamite and
tin oxide crystallites. Unspecified copper-containing
sulfates are scarce and only indicated by EDS. The lead is
mobilized and concentrated towards the outer corrosion
layers above the original metal surface, and it appears to
be present mostly as undetermined chlorides containing
lead and copper.5 Cuprite is abundant in the outer corro-
sion layers, where it embeds abundant silicates particles,
among which the most common are quartz, alkali feldspar
and plagioclase. Paratacamite sometimes mixed with
atacamite forms rather large pockets close to, and above
the limit of, the original metal surface6 (Figure 8).

Previous removal of corrosion layers

Figure 8. X-ray elemental maps showing the distribution of the major
elements in the cross section of the foot’s wall
Mechanical cleaning can lead to damage unless the nature
of the corrosion strata is fully understood—a position
strongly advocated in the work of Bertholon (2000).7 In
this case, it appears that the copper and lead in the orig-
inal alloy migrated outwards during the corrosion process
forming the relatively thick layers of cuprite and pockets
of copper chlorides found on the internal and exterior
surfaces. Apparently, the intention of the previous restorer
was to remove the rough exterior corrosion down to a
REDRESSING THE PAST: THE EXAMINATION AND TREATMENT OF AN ANCIENT SOUTH ARABIAN CENSER 311

flat surface that would reflect the original surface of the

object—what Bertholon refers to as the limitos. In this
case, however, the restorer did not fully appreciate the
thinness and extensive mineralization of the metal walls.
As a result, in many areas the cleaning went down into an
underlying layer which consists of a porous mixture of
cuprite with less abundant paratacamite, atacamite and
tin oxide crystallites. This mechanical cleaning created
numerous additional fractures and exposed pockets of
powdery green corrosion.

If this object had entered our collection today with its

archaeological corrosion intact, it would have been
repaired and stabilized as required, but not cleaned due
to the thinness of its walls and extensive mineralization.
It is likely the lost spires were completely mineralized
and may simply have crumbled when cleaning was
attempted. In addition, the interior core would not have
been obscured with modern materials.
Treatment
Given that the modern restorations were somewhat
disfiguring and covered original surfaces, modern mate-
rials were removed as much as possible although some
joins had to be retained due to the fragility of the miner-
alized metal. The thick shellac layer on the interior was
removed by carving away after softening with heat, and
modern plaster was removed mechanically. The horn and
ears of the ibex were restored to their correct configura-
tion—the latter reconstructed with Magic Sculpt epoxy
putty with a separating layer of Acryloid B-72. Fragments
were reassembled using Acryloid B-72 while Araldite
epoxy resin was used for the joins between the horns
and the head and between the tall side and the body of
the censer. Fills were made using glass micro-balloons
in acrylic resin, faced with a spackling putty and painted
to match surrounding original surfaces (Figure 9). Given
the absence of active corrosion, no stabilization was
required at this time.
Conclusion
Although The Met’s South Arabian censer differs in size
and function from the castings studied by Mille (2012),
it corroborates many of his findings and extends back by
at least a century the distinctive characteristics of copper
alloy casting in ancient Yemen. Its treatment history also
illustrates the risks of cleaning archaeological metals
without a thorough knowledge of their art historical
Figure 9. Ibex handle
after reconstruction of
 horns and ears
context and manufacturing process as well as subsequent
alteration during burial. While burial conditions often
differ, the instrumental analysis of the metal samples
provided here may be useful when assessing the condition
of similar alloys and possible treatment strategies when
such analysis is not available.
Notes
1 Fourier transform infrared microspectroscopy (micro-
FTIR) and pyrolysis-gas chromatography/mass spec-
trometry (Py-GC/MS) were used for the identification
of the restoration fills. Restoration materials consisted of
epoxy- and shellac-based fills bulked with crystalline and
amorphous silica, respectively. Off-white and pink resto-
ration fills were calcium sulfate hemihydrate, sometimes
admixed with calcium carbonate. FTIR was performed
on samples crushed on a compression diamond cell
(Spectra tech) using the 15× objective of a Hyperion 3000
microscope, featuring an MCT detector and interfaced
to a Tensor 27 (Bruker Optics). Acquisition was a sum
of 64 scans, at 4 cm-1 resolution in the range 4000–550
cm-1. Scanning electron microscopy with energy-dis-
persive x-ray spectrometry analysis (SEM-EDS) was
performed with a Zeiss Sigma field-emission scanning
electron microscope, equipped with an Oxford Instru-
ments X-MaxN 80 SDD detector. Backscattered electron
(BSE) images were acquired and EDS analyses were
performed on ion-milled, carbon-coated cross sections
in high vacuum with an accelerating voltage of 20 kV.
Electron backscattered diffraction (EBSD) was carried
out at 20 kV with a 70° incident angle using a Nordlys
Nano detector.
312  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
2 The XRF analysis was performed with a Bruker Artax
instrument using unfiltered Rh radiation at 50 kV,
700 μA, with a 1 mm collimator, and live-time acquisition
of 60 seconds. Elemental composition of the alloys was
calculated using fundamental parameter-based quanti-
fication method, whose accuracy was verified against a
copper alloy reference standard of known composition
(copper alloy 32X LB14 produced by MBH).
3 As identified by the combination of EDS, EBSD,
Raman and FTIR spectroscopy.
4 See Bertholon (2000) for a comprehensive description
of these corrosion phenomena.
5 These mixed copper-lead chlorides remain unde-
termined. Although EBSD suggests diaboleite [Pb2Cu-
Cl 2(OH) 4] as one of the most probable candidates,
attenuated total reflection-FTIR and Raman spectra (785
and 488 nm excitation) did not match those of reference
diaboleite or of other basic copper-lead chloride reported
in literature (Frost and Williams 2004). Nevertheless, the
presence of the -OH functional group could be verified
by the bands at 3426 cm-1 (ATR-FTIR) and 3439 cm-1
(Raman). Characteristic Raman bands detected were 52,
72, 89, 116, 135, 217, 296, 396, 654, 747 and 3439 cm-1.
Characteristic bands in the ATR-FTIR spectrum were
3426, 700 (or 694, 693) and 603 cm-1.
6 Cuprite was confirmed also by Raman analysis, which,
in addition, indicates variable intensity of bands at 150,
219 and 640 cm-1, possibly suggesting other oxide phases.
Atacamite, together with paratacamite, were also identi-
fied both by Raman and ATR-FTIR analyses.
7 Bertholon—building on L. Robbiola’s studies of
corrosion microstructures—has combined a compre-
hensive review of past literature on the examination
and cleaning of archaeological metals with an in-depth
study of corrosion strata with the aim of developing a
system that conservators could use to identify the closest
approximation of the original surface that could be
used to guide their cleaning protocol. In this case, a full
characterization of the original corrosion layers is not
possible given their alteration by past treatments.
Acknowledgements
The authors are grateful to Kim Benzel, Daira Szostak and
Shawn Osborne of The Met’s Ancient Near East Depart-
ment for their permission and assistance in studying
this censer and South Arabian related objects. We also
CONSERVATION TREATMENTS
thank Katy Blanchard at the University of Pennsylvania
Museum for facilitating a visual examination of their
South Arabian censer.
References
Avanzini, A. 2005. Some thoughts on ibex plinths in early
South Arabian art. Arabian Archaeology and Epigraphy
16: 144–53.
Bertholon, R. 2000. La limite de la surface d’origine des
objets métalliques archéologiques. Charactérisation, locali-
sation et approche des mécanismes de conservation. Paris:
Université Paris 1 Panthéon-Sorbonne.
Frost, R.L. and P.A. Williams. 2004. Raman spectros-
copy of some basic chloride containing minerals of lead
and copper. Spectrochimica Acta Part A: Molecular and
Biomolecular Spectroscopy 60(8–9): 2071–77.
Mille, B., I. Gajda, F. Demange, C. Pariselle, Y. Coquinot,
E. Porto, O. Tavoso, and A. Zink. 2010. Hawtar'athat, fils
de Raḍaw'il, du lignage de Shalalum. Une grande statue
de bronze du royaume de Saba' (Yémen). Monuments et
mémoires de la Fondation Eugène Piot 89: 5–68.
Mille, B. 2012. The casting techniques of antique South
Arabian large bronze statues. In Scientific Research on
Ancient Asian Metallurgy: Proceedings of the Fifth Forbes
Symposium at the Freer Gallery of Art, eds. P. Jett, B.E.
McCarthy, J.G. Douglas, 225–41. Washington, DC:
Archetype Publications in association with the Freer
Gallery of Art, Smithsonian Institution.
Muncherjee, K. ca. 1925. A Marvelous Collection of Rare,
Authentic, Genuine, Ancient, Hymarites and Sabeans
Antiques, of Ancient Southern Arabia. In the possession of
Kaiky Muncherjee. Unpublished sales catalogue.
Muscarella, O. 1988. Bronze and iron: Ancient Near
Eastern artifacts in The Metropolitan Museum of Art.
New York: The Metropolitan Museum of Art.
Pillai, R.M., S.G.K. Pilhii, and A.D. Damodaran. 2002.
The lost-wax casting of icons, utensils, bells and other
items in South India. The Journal of the Minerals, Metals
& Materials Society 54(10): 12–16. www.tms.org/pubs/
journals/JOM/0210/Pillai-0210.html (accessed 26
October 2018).
Authors
Jean-François de Lapérouse is a conservator in the
Objects Conservation Department at The Met working
primarily with the Ancient Near East and Islamic collections.
REDRESSING THE PAST: THE EXAMINATION AND TREATMENT OF AN ANCIENT SOUTH ARABIAN CENSER 313

Federico Carò is a research scientist in the Department

of Scientific Research at The Met where he investigates
inorganic materials and techniques employed in artistic
production. His research interests focus on the mineral-
ogical, petrographic and geochemical characterization of
stone and other geological materials in provenance and
conservation studies.

Adriana Rizzo is a research scientist in the Department

of Scientific Research at The Met. She carries out analyt-
ical work for the study and conservation of a variety
of objects from the collection, mainly using molecular
spectroscopy and mass-spectrometry techniques. She
is interested in the study of organic materials and their
degradation as well as conservation-related issues.

Danijela Jovanovic is an objects conservator based in

Belgrade who specializes in the study and treatment of
archaeological metal.
Conservation and Investigation of Cometh the Sun:
A Monumental Weathering Steel Sculpture

Claudia Chemello* Ana Crespo Curtis Patterson

Terra Mare Conservation, LLC
Charleston SC, USA
claudia@terramareconservation.com
Emilio Cano
Centro Nacional de Investigaciones
Metalúrgicas (CENIM)
Consejo Superior de
Investigaciones Científicas (CSIC)
Madrid, Spain
ecano@cenim.csic.es
Centro Nacional de Investigaciones Atlanta GA, USA
Metalúrgicas (CENIM) curtisrpatterson@comcast.net
Consejo Superior de http://cpattersonsculptor.com
Investigaciones Científicas (CSIC)
Madrid, Spain
*Author for correspondence
a.crespoibanez@cenim.csic.es
Paul Mardikian
Terra Mare Conservation, LLC
Charleston SC, USA
paul@terramareconservation.com

Abstract
In 2016, Terra Mare Conservation, LLC, was contracted by the
City of Atlanta, Georgia, to undertake conservation of Cometh
the Sun, a monumental sculpture fabricated in weathering
steel by artist Curtis Patterson. The collaborative project
involved conservators, metal fabricators, metal trades, city
public art officials, and the artist. The sculpture has a combined
weight of 5.7 tons, and is 15.2 meters in length, 6 meters in
width, and 4.5 meters in height. Conservation efforts involved
disassembly, replacement of 40 severely corroded areas of steel,
artificial patination, sealing of moisture-prone areas, drainage
enhancement through the installation of new anchors and
footers and a concrete slab, and reassembly.
Samples of the original weathering steel patina and new
artificially patinated samples of A588 weathering steel were
assessed by electrochemical impedance spectroscopy (EIS)
using a gel-polymer electrolyte cell. Composition of the natural
and artificial patinas was assessed by Raman spectroscopy,
and thickness and color were measured. Analytical results
demonstrated that the poor performance of the natural patina
was not attributable to pollutants but to improper development
in areas prone to water retention. These findings highlight
the need to follow manufacturers’ specifications when using
weathering steel to prevent failure of the protective patina.
Keywords
Corten steel, weathering steel, monumental sculpture,
conservation, EIS, Raman, patina

Introduction
The conservation and long-term preservation of weath-
ering steel sculptures is an evolving subject. The use of
weathering steel to fabricate large sculptures began in
the mid-1960s as the material was discovered by artists
and sculptors. There is now a considerable body of infor-
mation within the conservation community, as well as
ongoing research in academia and industry, contributing
to a better understanding of several key aspects about this
material from a preservation perspective. Studies have
examined the atmospheric corrosion of weathering steel
in different environments (for example Kamimura et al.
2006), characterized the rust-like patina that develops
(Scott 1991, Crespo et al. 2017), and explained how the
patina development is influenced by the length of expo-
sure and a sculpture’s environment (Chiavari et al. 2012).
Recent work examines whether an artificially applied
patina, often favored by artists to accelerate patina devel-
opment, may not provide the same protective properties
as a naturally developed patina (Ramírez Barat et al. 2016,
Crespo et al. 2017). These studies were important to this
project as an artificial patina was applied to Cometh the
Sun, thus presenting an opportunity to contribute to
research evaluating the durability of artificial patination
of weathering steel.
The conservation of Cometh the Sun was a complex and
challenging project involving a large group of stake-
holders including conservators, the artist, city public
art staff, metal fabricators, a concrete contractor, and an
engineer. Numerous issues had to be taken into consider-
ation, including the preservation of the artistic integrity
of the sculpture, the safety and feasibility of the work,
CONSERVATION AND INVESTIGATION OF COMETH THE SUN: A MONUMENTAL WEATHERING STEEL SCULPTURE 315

the complexity of the work needed to arrest the visible

corrosion, the costs, and the unidentified risks that could
become apparent as the work unfolded.

After 40 years of urban exposure, approximately 80% of

the exterior surface of Cometh the Sun was in fair to good
condition and 20% had suffered severe corrosion and was
in need of stabilization. The areas of severe corrosion
were chiefly due to the geometry and placement of the
sculpture and included recurrent water pooling, poor
drainage, accumulated dirt and organic debris against the
metal surface, the proximity of the steel to the concrete
slab, and other problems such as broken welds. Corrosion
of the footers was acute, with several having completely
failed. In addition, some areas of the surface had been
repeatedly tagged by graffiti, leaving an uneven patina
and a scarred surface (Figures 1 and 2).
Background and history
Born in Shreveport, Louisiana, artist Curtis Patterson
attended Georgia State University in Atlanta and became
the first African American to receive a Master of Visual
Arts in Sculpture from the university. His fascination
with the manipulation of three-dimensional space and
objects is clearly evident in many of his massive sculptural
works, whose symbolism and elements serve as indelible
references to the historical contributions of people of
African descent (Patterson, accessed October 10, 2018).
Cometh the Sun was commissioned by the City of Atlanta’s
Bureau of Cultural Affairs in 1977 and was the first public
art commission awarded to Patterson. The sculpture was
installed in Gordon White Park at a major intersection
that serves as one of the gateways to Southwest Atlanta,
an historically black neighborhood. The sculpture was
fabricated from 4.8 mm weathering steel and comprises
five monolithic conjoined components. The central
element of the piece is a disk which symbolically refer-
ences the sun. This component is undergirded by four
monolithic elements that are visually and structurally
supported by the central element. Although each of the
five elements is very large in scale, none of them has a
sub-structure. Each element has several internal angled
brackets fabricated from 4.8 mm weathering steel and
welded to the joints for additional strength.
Figure 1. Cometh the Sun before treatment. The numbering system used
to track components is seen in the image on the right
exposure of weathering steel to the elements, as seen in
Figure 1, with streaks and drip marks from water run-off.
Numerous areas of green algae staining were visible on
the north side and insect nests and accumulated organic
debris filled gaps between components and the broken
footers. The degree of corrosion varied from minimal to
severe, depending on the orientation of each component,
its geometry, and whether the surfaces were directly or
indirectly wetted, subject to pooling water, or protected
from wet/dry cycles. Severe corrosion was noted between
conjoined components, which were filled with “pack rust”
to a thickness of several centimeters (Figure 2), especially
in areas closest to or in contact with the concrete slab
where water pooled.
Four areas of graffiti were visible, created by spray paint
and scratching into the surface. Additional surface
damage was caused by mechanical abrasion of the metal
by people climbing and walking on the sculpture and
from three partially drilled holes where an attempt was
made by the city to take metal samples. An opening in
one of the components was the entry point made by the
artist during installation in 1977; this was welded closed
after installation.

Condition of the sculpture in 2016

Figure 2. Details showing deterioration problems: unevenly corroded
The sculpture was corroded overall with an uneven surfaces (top left), thick pack rust (bottom left), and delaminating
dark brown to dark purple patina typical of long-term surface layers (right)
316  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conservation treatment
Disassembly
A disassembly plan was finalized on site on January 16,
2017, which included a numbering system attached with
magnets by conservators to track the treatment of each
component (Figure 1). During the site meeting, access
to the interior of the sculpture was achieved by cutting
a manhole on top of component 2. Penetrating oil was
applied to all accessible bolts to facilitate removal. Picking
eyes were welded onto both ends of components 1, 2, and
3, and on top of 4 and 5 for lifting, and a metal bracket
was welded to the side of components 4 and 5, straddling
both pieces to keep them together during lifting.
On January 17, 2016, the sculpture was dismantled as
seen in Figure 3. Disassembly of the five components
required removing the original bolts, and/or cutting
them off with an oxyacetylene torch. Each component
was individually rigged with a separator bar and slings,
and placed onto a flatbed truck for transportation to the
metalworking facility.
Figure 3. Disassembly of the sculpture
The most difficult part of the disassembly was the sepa-
ration of the lower components (1 and 3) from those
on top (2, 4, and 5), as they were joined by welding and
bolts or a combination of both. Components 4 and 5
were separated using an oxyacetylene torch. The bolts
holding component 2 to component 1 were cut off, but
the components would not separate. One of the project’s
“unknowns” was realized at this point––the original
methods of joining used more than 40 years previously
could not be readily recalled by Patterson. A hole was
cut in the northeast wall of component 2 to access two
plug welds securing component 2 to component 3.
CONSERVATION TREATMENTS
These were cut out and steel bars used to separate the
two components.
Fabrication
Corroded sections of weathering steel were removed
with an abrasive cut-off wheel. New pieces of A588
weathering steel were welded into place with a Lincoln
SuperArc LA-75 MIG filler wire (ER80S-NI1, 0.9 mm
diameter) with 75% Ar/25% CO2 shielding gas. Discus-
sions between the conservators, fabricators, engineer, and
the artist resulted in agreement that severely corroded
areas, where corrosion had compromised the thickness
of the weathering steel by more than 30%, were to be
replaced. Approximately 40 new pieces of weathering steel
were welded onto the sculpture: eleven on component 1,
eight on component 2, four on component 3, four on
component 4, and thirteen on component 5 (Figure 4).
Figure 4. During treatment to replace new sections of deteriorated steel
After new weathering steel was welded into position,
each component was blasted with Black Beauty––a low
dusting and low free silica coal slag abrasive––in 20/40
mesh size to achieve a homogeneous surface profile. A
chemical patina called Auburn Rust was spray-applied
by Patterson to achieve an even color and speed up the
natural rusting process. After application, the surface
was repeatedly rinsed with water over several days to
help build the color, according to the usual practice of
the artist (Figure 5).
Reassembly
Prior to pouring a new sloped concrete pad, an anchoring
system for the cast-in-place anchors was installed. The
footers were made of type 304 stainless steel and the
anchor bolts of ASTM F1554-07a grade 55, S1, and A307
CONSERVATION AND INVESTIGATION OF COMETH THE SUN: A MONUMENTAL WEATHERING STEEL SCULPTURE
grade 55 modified weldable mild steel. The anchors
were leveled and welded onto the bottom channel of
the footer assembly and the assemblies positioned and
welded to the rebar.
The sculpture was reinstalled on October 25 and 26,
2017 (Figures 5 and 6). Each component was individu-
Figure 5. During treatment and installation: artificial patination (top left),
caulking between components (top right), installation (bottom left), and
after treatment (bottom right)
Figure 6. Cometh the Sun as originally installed in 1977 (top) and after
treatment in 2017 (bottom). Top photo credit: http://cpattersonsculptor.com
317
ally rigged and lifted into position. The upper channel
of the footer was positioned on top of the lower channel,
and components 1 and 3 were individually positioned
and leveled with hydraulic jacks. Once the position was
correct, the remaining three components were positioned,
and components 1, 2, 3, and 5 were bolted together from
the interior. Components 4 and 5 were joined prior to
arrival on site. No welding was used to join components
for the reinstallation. All bolts were new and made of
A325 weathering steel. After joining all components,
the hatch used to gain access to the interior was welded
closed with a Lincoln SuperArc LA-75 MIG filler wire
(ER80S-NI1, 0.9 mm diameter) with 75% Ar/25% CO2
shielding gas.
To prevent water intrusion and water pooling between
components 1 and 2 and between 1 and 3, Dow Corning
756 SMS Building Sealant, a silicone-modified sealant
(dark gray in color), was used as a gasket and applied
with a caulking gun. The sealant is a medium-modulus
elastomeric product designed for weatherproofing sensi-
tive porous stone and metal substrates. Silicone products
are flexible, offer excellent weatherability, and have a low
friction surface that reduces dirt accumulation.
The upper channel of each footer was welded to the
bottom of the sculpture and to the lower channel, and
each end of the footer was capped with a small plate of
stainless steel, welded into position. The footers were
coated by Patterson using Rust-Oleum Flat Protective
Enamel black paint, applied by spray.
Investigation of the patina
Two samples of weathering steel were obtained for
comparison: one exhibiting the natural patina (NP) of
the sculpture developed since installation in 1977 by
exposure to the atmosphere (removed from one of the cut
off corroded areas), and the second exhibiting a patina
artificially developed (AP) by chemical treatment as
applied to the new replacement steel. The artificial patina
was developed by spraying two coats of a commercial
patination product, Auburn Rust, followed by spraying
with tap water over several days, the usual practice by
Patterson.
According to the manufacturer, this product contains
amidosulfonic acid and cupric sulfate and is “used to
quickly and attractively rust steel and ferrous metals”
(Metal Finishes PLUS, accessed October 29, 2018). Recent
318  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CONSERVATION TREATMENTS

research has suggested that chemical patinas applied in different ratios. The Raman result from the AP sample
to weathering steel affect the protective properties and (Figure 9) shows mainly lepidocrocite, alone or combined
color of the patina, resulting in an ineffective or unstable with goethite in some areas.
method of achieving surface color (Ramírez Barat et
al. 2016, Crespo et al. 2017). The aim of this research
was to characterize the composition and the protective
properties of both the damaged sections and the restored
areas for comparison.

The protective properties of each patina were assessed by

electrochemical impedance spectroscopy (EIS) using a
gel-polymer electrolyte (G-PE) cell specifically developed
for analysis of cultural heritage (Ramírez Barat and Cano
2015, Ramírez Barat and Cano Díaz 2015, Ramírez Barat,
Cano, and Letardi 2018). The G-PE cell uses a classical
three-electrode configuration with an AISI 316 stain-
less-steel counter and Ag/AgCl reference electrode. The
electrolyte was artificial rainwater simulating an urban
Figure 8. Raman obtained from an NP sample with different goethite/
atmosphere (conductivity 5.4 µS, pH 5.53), gelled with
lepidocrocite ratios: (a) spectrum with less intensity in the lepidocrocite
3% agar. EIS was obtained from 100 kHz to 10 mHz, signal; (b) spectrum with more intensity in the lepidocrocite signal
using a ±10 mV RMS signal at the OCP and 10 points
per decade. Figure 7 shows the EIS results for the NP and
AP samples. The EIS results demonstrate that both the
natural and the artificial patinas are poorly protective,
with an impedance of about 700 Ω at low frequencies
(0.01 Hz) and barely 200 Ω at medium frequencies.

Figure 9. Raman obtained from an AP sample with different goethite/

lepidocrocite ratios: (a) spectrum with less intensity in the lepidocrocite
signal; (b) spectrum with more intensity in the lepidocrocite signal

Figure 7. (EIS) Electrochemical impedance spectra (Bode plots) obtained
on a sample from the original naturally patinated sculpture (natural
patina) and an artificially patinated sample of A588 steel (artificial patina)
The composition of the patinas was characterized
by Raman spectroscopy using a Renishaw RM 1000
Raman with a 634 nm laser and a 50× microscope lens,
10 seconds of integration time, and 5 accumulations.
Figure 8 shows representative Raman spectra from
different points on the outer surface of the NP sample.
The patina is mainly composed of goethite (α-FeOOH),
alone or in combination with lepidocrocite (γ-FeOOH)
The thicknesses of the patinas were measured using an
Elcometer 456 thickness gauge equipped with a probe
for ferrous metals. Several measurements (20 for the
NP, 30 for the AP) were taken across the surface of the
samples. The NP sample shows a thicker patina than the
AP (average 34.5 µm vs. 24.3 µm) but also more irregu-
larity (standard deviation 12.7 µm vs. 6.4 µm).
Lastly, as the aesthetic properties are one of the most
important features of a sculpture, colorimetry tests
were performed in order to measure the color differ-
ences between the two patinas using a Konica Minolta
CONSERVATION AND INVESTIGATION OF COMETH THE SUN: A MONUMENTAL WEATHERING STEEL SCULPTURE
CM-700d spectrophotometer with an 8 mm diameter
mask, CIE standard illuminant D65, and 10° observer.
The colorimetric results are reported in CIE L*a*b* color
space. Twenty points evenly distributed on the surface
were measured in the AP sample, while for the NP the
number of samples were limited to 7 due to its size. Table
1 shows the average and standard deviation (in brackets)
of the color coordinates. Both samples show an orange
color (positive and similar values for a* and b*), but
slightly more yellow for the AP. While the lightness is
similar in both samples, the chromaticity of the AP shows
much higher values, indicating a more saturated color.
This is typical of steel patinas, which show brighter colors
in the initial stages, moving towards less saturated and
darker colors upon aging. The difference in color ΔE is
33.8. When ΔE is greater than 1, the differences between
two colors are noticeable (Ghelardi et al. 2015). There-
fore, there is a big difference between the appearance of
the original and the artificial patina. Nevertheless, the
color of the artificial patina will probably evolve over
time, approaching the appearance of the original surface.
Table 1. Average and standard deviation of CIE L*a*b* colorimetry
results for NP and AP samples
L* (10°/D65) a* (10°/D65) b* (10°/D65)
NP 32.3 (2.3) 6.9 (1.0) 7.1 (1.6)
AP 32.7 (4.1) 16.9 (1.7) 21.7 (4.8)
These analytical results help to understand the acceler-
ated corrosion process of some areas of the sculpture.
EIS shows that the naturally developed patina is poorly
protective, with impedance values barely above the newly
patinated one. It is known that the protective patinas
of weathering steel evolve over several years (typically
5–10, depending on the environment), increasing their
protective properties (Crespo et al. 2017). This has not
been the case for Cometh the Sun in the areas analyzed.
The values obtained demonstrate that the patina has not
developed properly. Raman spectroscopy shows goethite
and lepidocrocite, in different ratios, in both cases. More
goethite is present in the original sample, which is, in
principle, positive for protection. No other compounds
that might have a detrimental effect on the protective
properties of the patina, such as akageneite (β-FeOOH),
were found. Therefore, the poor performance of the
patina cannot be attributed to its composition but to
the morphology of the corrosion layers: goethite has not
formed a uniform, continuous, or adherent layer, and
therefore it is not protective.
319
According to Kamimura et al. (2006), when a patina has
properly developed on weathering steel, the corrosion
rate does not exceed 0.01 mm/year. For Cometh the Sun,
considering that in some areas almost complete corro-
sion of the whole thickness of the plates had occurred,
the corrosion rate was 0.12 mm/year. The thinness of the
patina can also be explained by its low compactness and
adherence. While a properly developed patina exceeds
100 µm (Asami and Kikuchi 2002), the thinness of the
patina in the samples analyzed is probably the result of
a corrosion layer that delaminates and loses material
as it grows, and is unable to reach a greater thickness.
The fact that no other compounds apart from goethite
and lepidocrocite were found in the corrosion layers
suggests that the poor performance of the patina is
not related to pollutants but to other conditions less
favorable to its formation. It is known, and strongly and
repeatedly advised by the manufacturers,1 that correct
formation of the protective patina on weathering steels
requires repetitive wet/dry cycles. The use of weathering
steel buried in soil, submerged in water, or in areas
where water is retained precludes the formation of the
protective rust. For Cometh the Sun, the failure of patina
development in areas where poor drainage and installa-
tion issues were identified resulted in severe corrosion.
Conclusion
This challenging project accomplished the conser-
vation of a monumental work of contemporary art
fabricated from a material that requires a pragmatic
and nuanced approach to its care and long-term pres-
ervation. Working with an interdisciplinary team from
industry helped conservators identify appropriate
solutions that corrected problems inherent in the initial
design and installation. Investigation of the artificial
patina confirmed that the accelerated corrosion that
takes place to provide rapid color development does
not protect the surface of the steel in the same way as
a naturally developed patina. Conversely, analysis of
the naturally developed original patina also revealed
that it was poorly protective and not well developed
due to water retention in some areas, which prevents
the wet/dry cycles necessary for the development of a
protective patina.
The use of weathering steel in contemporary sculpture
comes with a number of strict specifications from
manufacturers which may be restrictive and difficult
320  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
for artists to implement. This case study illustrates the
critical importance of following the specifications for
fabrication and use of weathering steel, particularly
regarding the need to conform to wet/dry cycles and
avoid water entrapment for optimal protective patina
formation. If these are not observed, dramatic failure
may occur, causing accelerated deterioration and the
need for significant conservation intervention.
Acknowledgments
The authors would like to thank the artist Curtis
Patterson, the City of Atlanta, Georgia, Jesse Slagle and
the team from Holland LP, Tim Foecke, and Jon Tirpak.
The analytical work for this project was partly funded
by the project GEOMATERIALES 2-CM Program Ref.
S2013/MIT-2914 (Comunidad de Madrid). A. Crespo
acknowledges MINECO for the doctoral grant BES-2015-
071472 and Sagrario Martinez for her help with the
Raman analysis.
Notes
1 See, for example, www.nssmc.com/product/catalog_
download/pdf/A006en.pdf .
Materials list
Abrasive media
Black Beauty Abrasive
www.blackbeautyabrasives.com
Auburn Rust (amidosulfonic acid and cupric sulfate)
Metal Finishes PLUS
http://metalfinishesplus.com
Paint
Rust-Oleum Flat Protection Enamel
www.rustoleum.com
Silicone caulk
Dow Corning® 756 SMS Building Sealant
www.pksupplies.com/
Welding rods
Lincoln SuperArc LA-75, .035” diameter
www.lincolnelectric.com
References
Asami, K. and M. Kikuchi. 2002. Characterization of rust
layers on weathering steels air-exposed for a long period.
Materials Transactions 43: 2818–25.
CONSERVATION TREATMENTS
ASM International. 1990. Properties and selection: Iron,
steels, and high-performance alloys. ASM Handbook, vol.
1, 398–423. Materials Park: ASM International.
Chiavari, C., E. Bernardi, C. Martini, F. Passarini, A
Motori, and M.C. Bignozzi. 2012. Atmospheric corrosion
of Cor-Ten steel with different surface finish: Acceler-
ated ageing and metal release. Materials Chemistry and
Physics 136: 477–86.
Crespo, A., B. Ramírez Barat, I. Diaz, and E. Cano Diaz.
2017. Assessment of the protective properties of patinas
on contemporary sculpture made out of weathering
steel. In ICOM-CC 18th Triennial Conference Preprints,
Copenhagen, 4–8 September 2017, ed. J. Bridgland, art.
0802. Paris: International Council of Museums.
Ghelardi, E., I. Degano, M.P. Colombini, J. Mazurek,
M. Schilling, H. Khanjian, and T. Learner. 2015. A
multi-analytical study on the photochemical degradation
of synthetic organic pigments. Dyes and Pigments 123:
369–403.
Kamimura, T., S. Hara, H. Miyuki, M. Yamashita, and
H. Uchida. 2006. Composition and protective ability of
rust layer formed on weathering steel exposed to various
environments. Corrosion Science 48: 2799–812.
McGannon, H.E. 1964. High-strength low-alloy steels.
In The making, shaping and treating of steel, 1098–2004,
8th ed. Pittsburgh: United States Steel Corporation.
Metal Finishes PLUS. http://metalfinishesplus.com
(accessed October 29, 2018).
Patterson, C. 10 October 2018. Curtis Patterson, Sculptor.
http://cpattersonsculptor.com (accessed 10 October
2018).
Ramírez Barat, B. and E. Cano Díaz. 2015. Evaluación
in situ de recubrimientos protectores para patrimonio
cultural metálico mediante espectroscopía de impedancia
electroquímica. Ge-Conservacion 8(8): 6–13.
Ramírez Barat, B. and E. Cano. 2015. The use of agar
gelled electrolyte for in situ electrochemical measure-
ments on metallic cultural heritage. Electrochimica Acta
182: 751–62.
Ramírez Barat, B., T. Palomar, B. Garcia, D. de la Fuente,
and E. Cano. 2016. Composition and protective proper-
ties of weathering steel artificial patinas for the conserva-
tion of contemporary outdoor sculpture. In Metal 2016:
Proceedings of the Interim Meeting of the ICOM-CC Metals
Working Group, New Delhi, India, 26–30 September 2016,
CONSERVATION AND INVESTIGATION OF COMETH THE SUN: A MONUMENTAL WEATHERING STEEL SCULPTURE 321

eds. R. Menon, C. Chemello, and A. Pandya, 314–19. New

Delhi: ICOM-CC and Indira Gandhi National Centre
for the Arts.
Ramírez Barat, B., E. Cano, and P. Letardi. 2018. Advances
in the design of a gel-cell electrochemical sensor for
corrosion measurements on metallic cultural heritage.
Sensors and Actuators B: Chemical 261: 572–80.
Scott, J. 1991. Conservation of weathering steel sculpture.
In Saving the Twentieth Century: The Conservation of
Modern Materials. Proceedings of a Conference Sympo-
sium ‘91, Ottawa, Canada, 15–20 September 1991, ed.
D.W. Grattan, 307–22. Ottawa: Canadian Conservation
Institute.
United States Patent and Trademark Office. www.uspto.
gov (accessed October 13, 2018).

Authors
Claudia Chemello is principal and senior conservator
of Terra Mare Conservation, LLC. She is a Fellow of the
American Institute for Conservation and Coordinator
of the ICOM-CC Metals Working Group.

Emilio Cano is tenured scientist at the Centro Nacional

de Investigaciones Metalúrgicas (CENIM)-CSIC in
Madrid, Spain. He is assistant coordinator of the
ICOM-CC Metals Working Group.

Ana Crespo is currently completing her Ph.D. on Arti-

ficial Patinas in Weathering Steel Sculpture at the Centro
Nacional de Investigaciones Metalúrgicas (CENIM)-
CSIC in Madrid.

Paul Mardikian is co-founder and senior conservator

of Terra Mare Conservation, LLC. He is a Fellow of
the American Institute for Conservation and served as
assistant coordinator for the ICOM-CC Metals Working
Group for nine years.

Curtis Patterson is a sculptor based in Atlanta, Georgia.

Patterson has a Master of Visual Arts in Sculpture from
Georgia State University, Atlanta. He has produced
several major works for prominent cities in the United
States.
The Metal Patina and Surface Layer of El Caballito: Calling
Things by Their Name

Jannen Contreras-Vargas* Jaime Cama-Villafranca

National School of Conservation,
Restoration and Museography–INAH
Mexico City, Mexico
Jannen_contreras_v@encrym.edu.mx
*Author for correspondence
National School of Conservation,
Restoration and Museography–INAH
Mexico City, Mexico
jaime_cama_v@encrym.edu.mx

Abstract
This paper deals with the term ‘patina’ and the difficulty in
defining a priori those materials that should be preserved on
a metal surface. Dogmatic and restrictive ideas exist that only
patinas, understood as visually pleasant corrosion, should be
preserved. This has resulted in treatment criteria that have
promoted the loss of important technological, historical and
aesthetic information and materials. These comments arise
from observation of damage to the sculpture of the Spanish
king Carlos IV, known as El Caballito, in Mexico due to a non-
professional treatment that caused losses to its surface. This
paper proposes the term ‘surface layer’ to identify, describe
and assess all materials on a metal surface, especially metal
sculptures, irrespective of their nature, intention, origin,
temporality or function. The term encourages the understanding
that different materials on a metallic surface can be worthy of
preservation, even if they do not correspond to a patina.
Keywords
El Caballito, patina, surface layer, metal finish, critical
judgment

Introduction
‘A patina is the proper way to confer colour on
metals.’
Based on this phrase, one of the authors started years of
lessons on artificial patination during which it was also
pointed out that the patina of metals was constituted
by ‘visually pleasing and protective corrosion products’
(Scott 2003). This author was not aware that this rein-
forced a dogmatic idea that has underpinned arguments
for inadequate treatments and automatic determinations.
According to this statement, only patinas understood as
pleasant and protective corrosion deserve to be preserved,
particularly on metal sculptures.
After the mandatory mea culpa, the aim of this work is
to discuss such dogmatic ways of characterising metal
surfaces, defining the term ‘patina’ on metals and, there-
fore, understanding the objects treated.
These considerations result from observation of damage
suffered by the equestrian sculpture of the Spanish king
Charles IV in Mexico, known as El Caballito, when it
underwent non-professional treatment in 2013, based on
findings made during the diagnosis work and from the
semantic approach to terminology. When the damage was
made public, the term ‘patina’ was used widely in the media
and on social networks to convey a variety of meanings and
even to diminish the magnitude of the perceived damage
by arguing that the surface affected might not have been
the patina or the original one (Contreras and García 2016).
El Caballito
El Caballito is a sculpture by the Spanish artist and archi-
tect Manuel Tolsá in honour of the Spanish king Charles
IV. It was inaugurated in 1803 and is considered one of
the most beautiful equestrian sculptures in the world. Its
production was a great technological achievement since
it was created in a single casting operation in a country
with no tradition of monumental metallic sculpture and
a lack of the necessary material resources.
Being the portrayal of a Spanish king in a country that
soon sought its independence from Spain, the sculpture
had to be saved from the crucible and from those who
wished to produce coins and guns from it. In its more
than 200-year history, the sculpture has been relocated
three times and has undergone several treatments, but in
THE METAL PATINA AND SURFACE LAYER OF EL CABALLITO: CALLING THINGS BY THEIR NAME
September 2013 it was treated with nitric acid solution
and steel wire brushes to remove all materials from its
surface in order to apply a new patina. These actions were
halted the following day but almost half of its surface had
already been removed (Figure 1).
Figure 1. View of El Caballito at the beginning of the conservation
project, June 2016. Photograph by F. Kochen
The surface of El Caballito
When speaking about the damage suffered by El Caballito,
the term ‘patina’ was used widely and carelessly in the
media, on social networks and in reports and research
proposals made by conservators, chemists, sculptors,
bronze artisans, politicians, historians and even decorators.
It has become so complex over time because of its different
definitions and classifications according to the field in
which it is employed. In this case, it was defined as follows:
a) Metallurgists Something defined a priori and objec-
tively: a thin layer of corrosion (for
example, oxide or a sulfur compound)
that can be protective.
b) Chemists The conversion layer of the metal,
formed by corrosion products (Dent
1996 [1977], 395; Doménech-Carbó
et al. 2009, 129).
c) Bronze arti- An intentional finish that arises from
sans the desire to reproduce the layers of
corrosion formed naturally. This finish
is achieved on metallic objects by the
application of chemicals that produce
colourful corrosion and comprises corro-
sion products that may be enhanced
with waxes, resins and oils that may or
may not be coloured (González 2009, 7;
Contreras y Garcia 2016, 50).
323
d) Artists, A universal darkening that time makes
connoisseurs on pictures – or in this case, the sculp-
and amateur ture – that sometimes favours them
art dealers (Baldinucci 1681).
e) Metal conser-Corrosion, natural or artificial, that
vators is continuous, compact and forms
a homogeneous layer that faithfully
follows an object’s design, enhances its
aesthetic quality and comprises stable,
protective corrosion products which
isolate the metal from aggressive agents
of the environment (Cimadevilla and
González 1996, 6–7; Scott 2003, 10;
Doktor and Mach 2000; Mourey 1987).
f) According Everything on the surface of the sculp-
to one of ture forms superimposed patinas. The
the research original patina was the one ‘developed in
projects the foundry according to the taste and
(FCHCM traditions of the moment of produc-
2014, 23–24) tion’. Over this, historical patinas were
caused by ‘retouching or alteration
processes’. Finally, the current patina
was constituted by ‘intentionally and
environmentally caused corrosion prod-
ucts, pigments, carbon particles and
other organic materials produced by
combustion and deposited over time,
necessarily including materials that can
be classified as dirt or undesirable’.
Reaching agreement about the surface of the sculpture, its
finishing and the damage it had suffered via any discourse
using the term ‘patina’ was therefore impossible.
In 2016, the authors formed part of the team in charge
of the conservation of the sculpture and discovered that
the black finish that provided the aspect most people
recognised was a mixture of asphalt, waxes, soot and
other particles produced in the quite polluted atmosphere
of Mexico City (Figure 2). Below these materials, the
existence of a visually pleasing and protective corrosion
was expected. Instead, a green paint layer made with
drying oils and lead, copper and tin pigments was evident
(INAH 2017) with strata indicating that Tolsá and his
team sought to give the sculpture different green tones
and hues (Figure 3).
Although in western cultures artificial patination by
corrosion dates back to at least ancient Rome (Dent
324  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 2. Detail of the aspect and state of conservation of the layer of
asphalt, wax and soot on the horse’s elbow at the beginning of the
conservation project, 2016. Photograph by J. Contreras
2007), it was achieved mainly by cold techniques, with
patination by heating only possible on small sculptures.
Before the development of suitable torches1 it was very
difficult to reach the temperature required to form an
artificial patina rapidly on monumental sculptures.
Considering the 13-ton mass of El Caballito, and the
limited amount of time that Tolsá and his team had for
the finishes, including repairing, chiselling and colouring
(only fourteen months compared to the five years needed
to finish Bouchardon’s Louis XV sculpture), it was very
unlikely that he would have carried out a cold artificial
patination. In addition, the painting layer found on the
sculpture is consistent with a description by Alexander
Von Humboldt, who witnessed the unveiling: ‘The artist
had the good taste not to gild the horse, which is simply
coated with a brownish olive varnish’ (Humboldt 1974
[1814], 25–26).
CONSERVATION TREATMENTS
The painting on El Caballito is not a patina but is part
of the materiality that conveys its image, setting the
dialectical relationship between material and image
(Brandi 1963).
Lack of knowledge of the history of technology coupled
with the dogmatic idea that copper-alloy sculptures can
only be covered with patinas, understood as visually
pleasing corrosion, and that any other material does
not deserve to be preserved, has resulted (at least in
Mexico) in inappropriate decision-making criteria that
have promoted the loss of important technological,
historical and aesthetic information and materials. In
fact, it is presumed that other sculptures had similar paint
finishes but that as they weathered they were removed
during maintenance treatments that sought to ‘clean’ the
metal surface in order to make a new coloured corrosion
finish, destroying all material evidence and information.
Patina?
This paper does not seek to trace the genealogy of the
term ‘patina’ but to underline that it is linked to very
dissimilar effects by the very definition of the term.
During Catholic mass, the sacred host is placed on
a ‘paten’ – a small, brilliant, gilt dish – or covered by
means of a ‘patina’ to make it shiny. On the other hand,
the dictionary of the Royal Academy of the Spanish
Language defines ‘patina’ as ‘A tone, a kind of hard
varnish, olive-coloured and gleaming, formed due to the
action of moisture on the ancient objects of bronze’ and
Figure 3. Cross sections of the
green paint layer found on
the sculpture viewed under
an optical microscope. The
lower part corresponds to the
metal interface, showing a
minute flake of cuprite. The
size and distribution of the
pigment particles points to
antique painting preparation.
 Photograph by J. Vázquez
THE METAL PATINA AND SURFACE LAYER OF EL CABALLITO: CALLING THINGS BY THEIR NAME
a ‘Seated and soft tone that time gives to oil paintings’.
The Merriam Webster defines it as ‘a usually green film
formed naturally on copper and bronze by long exposure
or artificially (as by acids) and often valued aesthetically
for its colour’, or ‘a surface appearance of something
grown beautiful especially with age or use’.
In Italian, Spanish and English, therefore, the term
‘patina’ is linked to cleanliness, brightness and varnishes
but also to darkening and the development of coloured
corrosion on metal objects. In both cases, there is a rela-
tionship to changes experienced by materials over time.
In conservation, the term developed mainly in the field
of paintings. Years after the completion of a painting,
all of its materials will have changed. The most notable
changes are often seen in varnishes, which become yellow
and dark, modifying the appearance of colour and depth,
sometimes with a positive result:
Artists, connoisseurs, art dealers, collectors and
amateurs of art found themselves embroiled in an
artistic ideological debate on the aging of paintings.
Toward the end of the 17th century a theory had
emerged and continued to gain currency through
the 18th, that ‘Time’ improved and mellowed paint-
ings, increasing their beauty, harmony, subtlety and
mystery. (. . .) To quote Redgrave in A Century of
British Painters, 1866: ‘Connoisseurs believed that
pictures, like coins, obtained a patina from age that
mellowed their tone, and made them more valu-
able than in the state they left the painter’s easel.’
Sir George Beaumont, a British artist who died in
1827, concisely states: ‘A good painting, like a good
violin, should be brown’ (Keck 1984).
It is logical that these changes in appearance were assessed
according to the prevailing taste. That time and changes
endured by materials add value to objects was recognised
and desired by dealers, artists, connoisseurs, collectors,
amateurs of art and, later, also by conservators. Due to
this, the appearance of naturally occurring changes was
artificially reproduced either to achieve greater financial
profit or as part of an aesthetic aim.
Making decisions about objects of public interest is
difficult, even risky, and the results of our actions are
frequently controversial, coming with immense respon-
sibility (Contreras 2017). There has therefore been an
aim to make conservation objective, reversible and
‘irrefutable’ (Cercós 2014). In this context, attempts to
325
define metal patinas in an objective way served a useful
purpose and it was logical that definitions associating
them only to specific chemical compounds, such as
copper carbonate and copper oxide, were well received.
There are extraordinary and useful classifications of the
corrosion of copper alloys according to their appearance,
colour, texture, composition, behaviour and the state
of preservation of the original surface (Robbiola and
Fiaud 1992). From these, it would seem that a conser-
vator should certainly know a patina when they see one
but, in fact, a totally different and problematic scenario
exists, which, it is thought, stems from the fact that the
profession is trying to avoid obligatory subjective ques-
tions and decision-making by using only objective data.
We might say that we know exactly what a patina is,
what it is made of and, on this basis, presume that we
would respect it. However, if we do not point out exactly
what we consider a patina to be, it is naive to expect that
everyone else will have the same idea. From our point of
view, it is like commenting on the Emperor’s new suit.
Cartesian sciences are objective because they focus only
on a small part of reality. Conservation itself is a science
– an anthropological one – that uses tools from the
Cartesian sciences to understand materials, processes,
etc., but whose essence is social. The problems conser-
vation solves are wide, complex and inherently critical.
Materials are analysed in order to solve the problems
their changes cause to image and meaning (Dent 2007).
As Brandi (1963) states, we only restore the material in
which the image manifests itself and which promotes its
transmission. The analysis and interpretation it needs
are obviously subjective, in which positivist objectivity
alone becomes reductionist and limited.
Paul Philippot (1966) has provided the best definition
of patina:
. . . patina is precisely the normal effect that time
has on material. This is not a physical or chemical,
but a critical concept. Patina is nothing more than
the combination of these ‘normal’ alterations as they
affect the appearance of the work without disfiguring
it – precisely because it is the result of normal altera-
tions. Resorting to this notion of ‘normality’ does not
negate the concept of patina; it merely reveals that
the concept does not concern material but rises from
the critical domain and always implies an aesthetic
judgment (Philippot [1966] 1996, 373).
326  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Patina, therefore, is the evidence of the passage of time
and, although it is manifested through matter, is not
something physical that can be determined a priori
and objectively. Defining to what extent the modified
materials should be treated to keep the object legible,
without erasing the evidence of the passage of time, is a
difficult duty that motivates questions such as: How do
we recognise patina? When do material transformations
begin and cease to be considered as patina?
That the term ‘patina’ invariably falls within the field
of subjectivity has been widely disqualified but it is
impossible to determine the composition of patina a
priori and, in fact, the apparently unsolvable problem
of patina disappears if it is assumed to be the result of
a necessary critical judgement (Dent 2007). Carolusa
González (2009, 7), a pioneer of metal conservation in
Mexico, made it very clear: ‘The aim was to provide a
certain scientific objectivity while trying to avoid the
subjectivity of personal interpretation. However, time
and experience have shown that critical interpretation
is inevitable’.
Surface layer
The choice of a term and its meaning defines a position
and a scope. In the case of El Caballito, a discussion of
patina was pointless as it obstructed the dialogue. The
ambiguity of the term vs the need to know what consti-
tutes it has prompted different authors to warn about
the inappropriateness of its use (France-Lanord 1996;
Bertholon 2000, 214). William Mourey (1987) advised
against it, since it might comprise potentially harmful
corrosion products with the potential to cause instability
and the protection it provides is doubtful. He was prone to
speak simply of corrosion and alteration and made clear
that the function of patina is aesthetic and therefore it is
a subjective, not an objective, notion.
Establishing what should be preserved on the metallic
surface implies a very complex assessment and deci-
sion-making process which considers objective and
subjective aspects. From our particular point of view,
it is necessary to start solving the semantic problem of
insisting on the use of the term patina as a synonym for
metallic finish, for the surface of metallic objects, for
stable corrosion products, or for whatever we would
like to imply.
Consequently, for this case, ‘surface layer’ was proposed
rather than patina. It allowed for identification, analysis,
CONSERVATION TREATMENTS
description, interpretation and assessment of all the
materials on the metallic surface. This could be achieved
without a priori judgements about composition and
relevance and while considering their diverse origins,
intentions, meaning and characteristics, their contri-
bution to the appearance of the object, the protection
provided and the aesthetic, historical and technological
information they supply (Contreras and García 2016,
50). It was also possible to include the author’s intention
and the oeuvre’s conservation needs. The surface layer
can therefore be formed by:
• Intentional finishes, or artificial patination, which
may include:
–– Corrosion of the constituent metal resulting from
the application of chemical reagents or particular
reaction conditions such as fire. Corrosion prod-
ucts may be enhanced with pigments and organic
materials such as waxes, resins, dyes, paints, etc., and
even with metal powders (Dent 2007, Hughes 1982).
–– Metallic coatings.
–– Waxes, resins, oils (coloured or not) and paint.
The presence of a finishing material does not preclude
others.
• Remains of foundry materials.
• Products of alteration:
–– Corrosion products that are stable or unstable,
compact or dusty, homogeneous or heterogeneous,
and visually pleasing or deforming.
–– Materials resulting from environmental reactions,
maintenance actions or social demonstrations.
–– Pigments and salts.
–– Metallic coatings.
–– Waxes, resins, oils, asphalt and paints.
–– Fibres, dust, dirt, soot, food debris, droppings
and exudates from plants, animals and humans,
dead animals, leaves, seeds, paper, etc.
All of these materials can undergo modifications –
‘normal alterations’ that can be considered a patina and
others that disfigure the objects and constitute deterio-
ration, according to Philippot’s definition of patina, even
during and after conservation treatments.
In this case, if the patina understood only as the appealing
corrosion products was considered worthy of being
preserved, it would have been necessary to destroy all
remains of the significant green painting layer. Instead,
it was decided to remove the black layer of asphalt, wax
and soot to recover the important original green paint
THE METAL PATINA AND SURFACE LAYER OF EL CABALLITO: CALLING THINGS BY THEIR NAME
layer which covered almost 13 m2, about 30% of the total
surface (46.5 m2) (Figures 4 to 6).
Figure 4. Cleaning
process for the
king’s sandal or
campagus. On
the right side, the
original green
paint is shown
after cleaning.
Photograph by
 F. Kochen
Figure 5. Cleaning process for the horse barrel. On the lower side, the
original green paint is shown after cleaning. Photograph by F. Kochen
Current thoughts
Although the term ‘patina’ for metals has a long tradition,
it has been used in a careless and dogmatic way that
has proved to be limiting and inadequate. Attempts to
produce pre-constructed objective definitions around its
composition did not help either; on the contrary, these
have favoured semantic confusion.
Use of the term ‘surface layer’ to encompass all materials
on the surface of El Caballito allowed them to be assessed
free from a priori judgments. This took into considera-
tion their importance, origins, intentions, meaning and
physical characteristics, along with their contribution
to the oeuvre’s aspect – its protection and aesthetic –
327
and the historical and technological information they
provide (Contreras and Garcia 2016, 50–51), alongside
the intention of the author and the conservation needs
of the monument as a whole.
‘Patina’ remains a valid concept in conservation if a
clear conceptual framework accompanies it. In the
conservation profession, we should call every material
by its name and use the term in line with an agreed
definition – ‘the modification of materials by the
passage of time’ – assessed with critical judgement and
in consideration of the fact that all surface materials,
organic or inorganic and whether formed naturally or
intentionally, can be relevant.
References
Baldinucci, F. 1681. Vocabolario toscano dell’arte del
disegno. Florence: Accademia della Crusca.
Bertholon, R. 2000. La limite de la surface d’origine des
objets métalliques archéologiques : Caractérisation,
localisation et approche des mécanismes de conserva-
tion. PhD dissertation, University Panthéon-Sorbonne
Paris 1, France.
Brandi, C. 1993 (1963). Teoría de la restauración, trans.
María de los Ángeles Toajas. Madrid: Alianza.
Cercós, L. 2014. Restauración científica y poscientífica. In
Atelier du Patrimoine et d’Architecture, 09 February 2014,
available at: http://lc-architects.blogspot.mx/2014/02/
restauracion-cientifica-y-pos-cientifica.html (accessed
November 2017).
Cimadevilla, I. and C. González 1996. La teoría de la
restauración aplicada en la intervención de objetos
metálicos. Imprimatura: Revista de Restauración 12:
25–33.
Contreras, J. 2017. Investigar sirve para resolver prob-
lemas. Investigar en conservación-restauración sirve
para resolver problemas de conservación-restauración.
In Archivo Churubusco año 1, número 2. Available at
https://archivochurubusco.encrym.edu.mx/n2letras7.
html (accessed October 2018).
Contreras, J. and Á. García. 2016. La increíble y triste
historia de la cándida escultura metálica y su ‘pátina orig-
inal’ desalmada. In Estudios sobre conservación, restau-
ración y museología, 43–55. Mexico: ENCRyM INAH.
Contreras, J., G. Peñuelas, M. López, and A. García. 2014.
Metal patina, patina loss and repatination criteria: The
328  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
case of ‘El Caballito’. In Open Air Metal: Outdoor Metallic
Sculpture from the 19th to the Early 20th century. Identifi-
cation, Conservation, Restoration, 15th SFIIC Study Days,
4–5 December 2014, Paris. Champs-sur-Marne: SFIIC.
Dent Weill, P. 1996 (1977). A review of the history
and practice of patination. In Historical and philo-
sophical issues in the conservation of cultural heritage,
eds. N. Stanley, M. Kirby, and A. Melucco, 394–414,
first published in National Bureau of Standards Special
Publication 479, Gaithersburg. Los Angeles: The Getty
Conservation Institute.
Dent Weil, P. 2007. Technical art history and archeom-
etry I, patina: Historical scientific and practical consider-
ations. In Revista Brasileira de Arqueometria, Restauração
e Conservação 1(2): 60–66.
Doktor, A. and M. Mach. 2002. What is bronze In Protect
our European outdoor bronze monuments: Good practice
guide, Culture 2000, technical report, ed. C. Degrigny,
chap. 2.3, 23–24.
Dómenech-Carbó, A., M.T. Dómenech-Carbó, and V.
Costa. 2009. Electrochemical methods in archaeometry,
conservation and restoration. Berlin: Springer Verlag.
Fideicomiso del Centro Histórico del Distrito Federal
(FCHDF). 2014. Proyecto de investigación científica para
la conservación y restauración de la escultura ecuestre de
Carlos IV y su pedestal. Unpublished.
France-Lanord, A. 1996. Knowing how to question the
object before restoring it. In Historical and philosophical
issues in the conservation of cultural heritage, ed. N. Stan-
ley-Price, 244–47. Los Angeles: The Getty Conservation
Institute.
González, C. 2009. La pátina y la restauración de metales:
Entre la objetividad científica y la interpretación crítica.
In Memorias del Primer Encuentro entre Restauradores y
Corrosionistas. Campeche: UACAM.
Hughes, R. and M. Rowe. 1982. The colouring, bronzing
and patination of metals. London: Watson-Guptill.
Humboldt, A.V. 1814. Researches, concerning the institu-
tions & monuments of the ancient inhabitants of America:
with descriptions & views of some of the most striking
scenes in the Cordilleras!, trans. H.M. Williams, vol.
1. London: Longman, Hurst, Rees, Orme & Brown, J.
Murray & H. Colburn.
INAH. 2013. Dictamen de daños a la estatua ecuestre
de Carlos IV conocida como ‘El Caballito. Available at
CONSERVATION TREATMENTS
www.inah.gob.mx/es/comunicados/463-dictamendel-in-
ah-determina-dano-en-50-de-la-superficie-de-escultu-
ra-ecuestre-de-carlosiv (accessed 20 June 2016).
‘Lampe à souder’. In Lamptérophile Club France. Available
at www.lampeasouder.fr/ (accessed 27 November 2016).
Mourey, W. 1987. La conservation des antiquités
métalliques : De la fouille au musée. Draguignan: LCRRA
(Laboratoire de conservation, restauration et recherches).
Philippot P. 1996 [1966]. The idea of patina and the
cleaning of paintings. In Historical and philosophical
issues in the conservation of cultural heritage, 372–76, first
published in Bulletin de l’Institute Royal du Patrimoine
Artistique. Los Angeles: The Getty Conservation Institute.
Robbiola, L. and C. Fiaud. 1992. Apport de l’analyse
statistique de la composition des preduits de corrosion
à la compréhension des processus de dégradation des
bronzes archéologiques. Revue d’Archéométrie 16: 109–19.
Scott, D. 2003. Copper and bronze in art. Corrosion,
colorants, conservation. Los Angeles: The Getty Conser-
vation Institute.
Author
Jannen Contreras-Vargas, M.Sc., is a conservator
who specialises in metallic heritage, materials and
manufacturing techniques, methodology and evaluation
of treatments. She has been a lecturer at the National
School of Conservation, Restoration and Museography
(ENCRyM) of the National Institute of Anthropology
and History (INAH) in Mexico City since 2001 and is
part of the Metallic Heritage Conservation Laboratory.
Jaime Cama-Villafranca obtained his certificate as a
conservator for French museums in 1968 and later as a
mural painting specialist (ICCROM). One of the founders
of ENCRyM, he was the school’s head between 1983 and
1992 and is currently its dean, having been responsible for
the creation of its degree program in conservation. He has
directed and coordinated numerous large conservation
projects and taught many generations of conservators
in Latin America.
Study and Conservation of Lead Curse Tablets
Loretta Rossetti Nicolas Stephant
Arc’Antique, Grand Patrimoine de Loire Institut des Matériaux Jean Rouxel de
Atlantique (GPLA) Nantes (IMN)
Nantes, France Nantes, France
Loretta.Rossetti@loire-atlantique.fr Nicolas.Stephant@univ-nantes.fr
Jean Gabriel Aubert Nicolas Mélard
Arc’Antique, Grand Patrimoine de Loire Centre de Recherche et de Restauration
Atlantique (GPLA) des Musées de France (C2RMF)
Nantes, France Palais du Louvre
Jean-Gabriel.Aubert@loire-atlantique.fr Paris, France
Nicolas.Melard@culture.gouv.fr
Abstract
Lead curse tablets (defixio) served as a support for inscriptions and
were commonly used in antiquity. In some cases, the text engraved
in the lead remains visible, but in many others, it has become
obscured by corrosion products. Corrosion typologies have been
studied on eight tablets belonging to the National Library of
France. Certain tablets originating from North Africa date from
the early centuries AD, whilst information on others is incomplete
or inexistent. Corrosion facies were observed in section using
both macroscope and SEM-EDX. Some corrosion products were
analysed by XRD. Cleaning tests were then carried out. The impact
of cleaning on the readability of the engravings was evaluated by
Introduction
The Arc’Antique laboratory in Nantes has been involved
in the restoration and conservation of lead objects since
the 1990s (Degrigny and Le Gall 1999, Guilminot et al.
2012, Fontaine et al. 2016) and has thus had the oppor-
tunity to treat a range of lead objects which includes
sarcophagi, papal bulls, water pipes, bullets and statuettes.
Initially, work on this material focused on the study of
corrosion mechanisms, the development of electrolytic
treatments and the optimisation of health and safety
conditions for the treatment of lead objects, but Arc’An-
tique gained an opportunity to probe more deeply into
themes related to the conservation and restoration of lead
curse tablets (defixio) in 2012 when it was entrusted with
a series of tablets by the National Library of France (BnF).
Prior to this, the laboratory’s restorers had been able
to process only a small number of tablets directly from
excavations or heritage collections, but the frequency of
cases was too sporadic for these to constitute a subject for
Elodie Guilminot*
Arc’Antique, Grand Patrimoine de
Loire Atlantique (GPLA)
Nantes, France
Elodie.Guilminot@loire-atlantique.fr
Author for correspondence
*
imaging techniques. Measurements by different methods yielded
results with a resolution of 1 μm for microtopography, 2 to 5 μm
for the AICON scanner and 20 to 40 μm for the Leica DMS 1000
macroscope. The most precise techniques (microtopography and
3D scanner) are able to evaluate modifications due to cleaning.
Additional techniques (photography or macroscope) provide
efficient tools to observe the engravings.
Keywords
lead tablets, corrosion, conservation-restoration, imaging
techniques
research. Whether from excavations or heritage collec-
tions, one of the major difficulties is to retain or restore
the readability of incisions made on lead sheets. Cleaning
is very delicate as incisions are often preserved inside the
layer of corrosion products which is typically very fragile.
An additional issue is that, generally, the restorer is not
personally able to interpret the signs, hence the need to
use the most ‘objective’ and gradual methods available
so as to observe results step by step and not to eliminate
data or alter the surface.
The corpus belongs to the Department of Coins, Medals
and Antiques of the National Library of France. It is
composed of about 20 items. Most were acquired in the
19th century when interest in epigraphic collections was
at its peak, notably influenced by Ernest Babelon, the
department director. Others were bequeathed in 1925 by
the epigraphist and eminent collector, Wilhelm Froehner.
Over the years this collection was gradually added to by
330  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
donations and purchases, and was partially studied in the
early 20th century by epigraphists such as Augustus Audol-
lent (North African tablets) and Louis Robert (Froehner
Collection). Some tablets come from North Africa and
date from the early centuries AD; while information on
others is incomplete or even totally non-existent.
Despite the lack of information, it can be assumed that
from the moment tablets entered public collections, they
were probably stored in oak coin cabinets, contributing
thus to their deterioration (Degrigny and Le Gall 1999).
Lead is very sensitive to the organic acid vapours given
off by certain species, such as oak. Most of the tablets
had already undergone cleaning and consolidation
procedures, and many had featured in past publications
(Audollent 1904). However, on arrival at the Laboratory
it was observed that their state of preservation appeared
somewhat unsound, with extensive fracturing and the
surfaces generally not very legible.
Clearly, the specificity of these objects constituted
grounds for further studies with respect to their conser-
vation. Research was carried out on various aspects,
particularly in the aim of improving the legibility of the
texts. Cleaning tests were conducted and suitable imaging
techniques were researched both to evaluate these
cleanings and to preserve the data. Other aspects of the
study concentrated on enhancing storage and preventive
conservation systems, together with the improvement of
parameters for the opening of rolled tablets.
This study focuses primarily on the results of the cleaning
tests and their assessment by various imaging techniques
carried out both at Arc’Antique and at the Centre de
Recherche et de Restauration des Musées de France
(C2RMF).
Operating conditions
Corpus
The tablets constituting the corpus and the relevant tests
and analyses are shown in Table 1. Mechanical cleaning
was carried out either by micro-sandblasting with a
plant-based abrasive (olive-stone powder with a particle
size of <100 μm) or with glass microbeads (particle size
of 45 μm–90 μm). Chemical cleaning was carried out
in an acid medium in order to dissolve lead carbonates.
Solutions of citric acid (pH = 3), oxalic acid (pH = 2) or
nitric acid (pH = 3) were applied in the form of gels: a
chemical gel, Nanorestore new formulation, developed
CONSERVATION TREATMENTS
and supplied by the Consorzio Interuniversitario per
lo Sviluppo dei Sistemi a Grande Interfase (CSGI),
University of Florence; xanthan gum at 2% m; or an agar
gel (Nevek®) at 50% m. A complexing solution (EDTA
disodium) applied as a gel was also tested. Finally, ion
exchange resins (Amberlite® IR 120 H strongly acidic
cation exchanger) were also used.
Analysis techniques
Tablet fragments were embedded in a cold-mounting
epoxy resin and polished with a series of silicon carbide
papers down to 4000 grit and with 1 μm diamond paste.
First, they were examined by a Leica optical macroscope
with a 0.75–6× magnification objective lens. Scanning
electron microscopy (SEM) (JEOL JSM 5800LV) was
used at 20 kV to observe the morphology of corrosion
products, and energy-dispersive x-ray spectroscopy
(EDX) (SAMx) to elucidate elemental composition.
Corrosion products, collected and reduced to powder,
were analysed by x-ray diffraction (XRD) with a diffrac-
tometer (X’Pert pro PANalytical) using Cu Kα1 and Kα2
(40 V and 40 mA) in 2Ѳ configuration. Corrosion prod-
ucts were also analysed by Raman spectroscopy. Spectra
were obtained at room temperature under an excitation
wavelength of 514 nm (Ar), using a Renishaw InVia
reflex Raman spectrophotometer, associated with a Leica
microscope. To avoid any transformation of corrosion
products, nominal power was limited to 0.15 MW with
a laser spot size of 3 µm².
Imaging techniques
The techniques chosen for this study were conventional
and enhanced photography (known as ‘open light’), 3D
structured light scanning, macroscopy and microto-
pography.
Photographs were taken in the studio using a flash, with
constant colour temperature. Photos then underwent
post-processing on Capture One and Photoshop with
a Nikon D810 and D850 (47.5 million pixels). ‘Open
light’ photography is a simplified version of reflectance
transformation imaging (RTI) and is performed in total
darkness with a grazing light source rotating around the
tablet at 360° during shooting. A high-quality LED to
focus the beam was used as the light source.
Macroscope shots were made on a Leica DMS 1000 with
a magnification of up to 300. A motorised micrometer
STUDY AND CONSERVATION OF LEAD CURSE TABLETS 331

tray allowed the tablets to be shot in their entirety. The
macroscope was controlled by Leica Application Suite
V4.12 software. Measurements were made by the Eotech
company using a 3D structured light scanner together
with an Aicon 3D Systems sensor in smartSCAN R5
configuration, two B&W 5 MP cameras and a blue LED
structured light projector. The triangulation angle was
27°, with a measurement field of 47 × 38 × 25 mm, a
resolution of 0.02 mm and a precision of 7 μm. Multiple
views were needed to cover the entire surface of a tablet,
with each view overlapping to allow the final image to
be constructed. Measurements were performed on the
reverse of tablet 2012421 before and after cleaning.
Microtopographic measurements were carried out at the
C2RMF imaging department on an Altisurf 50 (2016
model) micro-metrology station. Measurement consisted
of scanning the surface with a beam of white light along
two axes (x and y). During scanning, the height (Z)
between the probe and the surface was recorded together

Table 1. Table summarising the corpus showing the analyses and imaging tests performed

Cleaning
Imaging tests Analyses
tests
Ref. Inventory Dimensions 3D
Arc’Antique no. BnF (mm) Photos before (and after) Open structured Micro- SEM-
RTI Macroscope XRD Raman
cleaning Light light topography EDX
canner


Audollent 
2012410 144 × 88  (during  
284 (1 sector)
cleaning)

Bronze 
2012413 144 × 88   
2296 (before)

 
95 × 47
Audollent (before (9 sectors)
2012421 (main     
278 and after) (before and
fragment)
(reverse) after)


Audollent  (1 
2012422 127 × 93     
sn (before) fragment, (1 sector)
before)


AA. 60 × 35
 (1 
2012433 Froehner (fragment     
(before) fragment, (1 sector)
15 analysed)
before)

Audollent  
2012467 112 × 102    
276 (before) (before)

Audollent 
2012469 121 × 99    
281 (before)


104 × 65
Froehner  (1 
2012470 (main     
or FG (before) fragment, (1 sector)
fragment)
before)
332  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
with the XY position. This position was recorded at a
cadence defined by the measurement step which can be
fixed between 400 nm and 500 μm. In the present study, an
8 mm probe (depth of field) was used. Measurement steps
were set at 10 μm and cleaning windows, with dimensions
varying in length from 8 to 11 mm, were delimited by
Paraloid. Measurements were carried out before and after
cleaning on tablet 2012421 (on the reverse, engraved side)
in the same areas, designated as A to G (Figure 1). The
Paraloid perimeter was used as a reference and allowed
measurements to be readjusted after cleaning. Data were
processed by MountainsMap software and filtered in order
to distinguish the roughness of the surface from its overall
undulation, this warping effect being due to the plasticity
of the material. Standard surface parameters (ISO 25175)
were calculated for each measured area.
Figure 1. Photos of curse tablet 2012421 showing cleaning windows
before (a) and after cleaning (b)
Results
Description of tablet corrosion facies
On a macroscopic scale (with the naked eye and under
binocular viewing), slightly varying corrosion facies
were identified, often simultaneously present on the
same object. Engravings were either completely masked
by hard, thick corrosion products or lightly veiled with
whitish corrosion products. Layers containing the inci-
sions sometimes either lacked adherence and displayed
extensive cracking or, inversely, were compact and solid.
They also displayed a range of intermediate degrees
between these two states. Observations of cross sections
of the corrosion facies on all the tablets showed various
states of degradation (Figure 2). In some places, tablets
were totally mineralised (in the case of tablets 2012422
and 2012470) (Figure 2d). On certain tablets (such as
2012413 and 2012433), a layer of red-brown corrosion
products delimited the original surface (Figures 2a and
2b). This layer had an average thickness of 150 μm.
Above this surface, there were external white-yellow
corrosion products. Other, whiter, corrosion products
had developed inside the metal core. EDX analysis
revealed these white internal corrosion products were
CONSERVATION TREATMENTS
exclusively composed of lead, carbon and oxygen, while
external white-yellow corrosion products also contained
calcium, silicon and aluminium. Raman spectroscopy
analysis yielded few results due to strong fluorescence.
The only exploitable corrosion product spectra obtained
showed characteristic peaks of lead carbonates: cerus-
site and hydrocerussite. XRD analysis confirmed these
results: most compounds were lead carbonates, gener-
ally associated with the lead oxides PbO (litharge or
massicot) and Pb3O4 (minium). Raman analysis located
lead oxides mainly in the red-brown corrosion layer.
According to XRD analysis, the presence of calcium and
silicon in external products could be linked to a calcium
silicate. It is possible that external and internal (whiter
and calcium-free) lead carbonates were not formed at
the same time:
• External white-yellow corrosion products (with
external markers, such as calcium, silicon and
aluminium) may have developed during the period
the tablet was buried.
• Internal white corrosion products may have appeared
during the period the tablets were stored in oak cabinets.
Tablets 2012421 (Figure 2c) and 2012422 (Figure 2d)
present a less well-defined stratigraphy: lead carbonates
still constitute the main corrosion products, but another
brown-orange compound was detected throughout the
sample. This compound was linked to the presence of
chloride (Figure 3), but its nature has not been clearly
identified. According to Raman analysis, certain spectra
obtained on samples 2012421 and 2012422 display charac-
teristic peaks of lead chlorides associated with carbonates
or phosphates: peaks at 127 and 838 cm-1, for phosgenite
(Pb2CO3Cl2), and peaks at 117, 127, 430 and 970 cm-1, for
phosphohedyphane Ca2Pb3(PO4)3Cl (Frost et al. 2014).
Figure 2. Macroscope photos of cross sections of tablets 2012413 (a),
2012433 (b), 2012421 (c) and 2012422 (d)
STUDY AND CONSERVATION OF LEAD CURSE TABLETS 333

trated and obscured incisions, while preserving internal

Figure 3. SEM and EDX analysis of tablet 2012421: photo in secondary
electrons (a), cartographies of Cl (b), O (c) and Pb (d)
Cleaning tests
The purpose of cleaning was to eliminate the external
corrosion products and superficial dirt which had pene-
corrosion products in whose layers the texts had been
preserved. In view of the difficulty of testing various
cleaning methods on all the corrosion facies present,
tablet 2012421 (Inv. # Audollent 278), displaying moder-
ately corroded surfaces, was selected. Mechanical and
chemical cleaning methods previously used in the labo-
ratory on lead objects, and other techniques which could
be effective in theory, were also selected. The aim was to
verify their impact on the surface and more particularly
on the incisions. The tests were performed first on the
non-engraved reverse side of the tablets (Table 2) and
then on the engraved side (Table 3). Mechanical tech-
niques were the fastest and also under the constant visual
control of the restorer. External corrosion products were
almost completely eliminated, even inside micro-cracks,
which were exposed at the same time. Micro-sandblasting

Table 2. Summary of operating conditions and observations on the first series of cleaning tests (reverse side of tablet 2012421)

ZONES Solution pH Method Observations

A Not cleaned
B Oxalic acid 2 Nanorestore gel, new formulation, Surface not cleaned in areas with insufficient
5 g/l 5 hrs of immersion in treatment solution contact with gel
2 applications: 1 hr + 1 hr (15’ rinsing of gel in treatment solution
during application time); total application time: 2 hrs
C1 Citric acid 3 Nanorestore gel, new formulation, Cleaner surface, no residue
5 g/l 4 hrs of immersion in treatment solution A veil of corrosion products remained. Surface
2 applications: 1 hr + 1 hr (15’ rinsing of gel in treatment solution not cleaned in areas with insufficient contact
during application time); total application time: 2 hrs with gel
C2 Citric acid 3 Gel: Xanthan (2wt%) Gel difficult to remove
5 g/l Applied hot (2 applications 1 hr + 1 hr); total application time: 2 hrs
D1 EDTA 2Na+ 5 Gel: Xanthan (2wt%) Gel difficult to remove
12.5 g/l Applied hot (3 applications :1 hr + 1 hr + 2 hrs); total application Surface with white corrosion products still
time: 4 hrs present
D2 EDTA 2Na+ 5 Nanorestore gel, new formulation Surface not cleaned in areas with insufficient
12.5 g/l 4 hrs of immersion in treatment solution contact with gel
2 applications 1 hr + 1 hr (5’ rinsing of gel in treatment solution Thick, white corrosion products remain on
during application time); total application time: 2 hrs surface after removal of gel
E Nitric acid 3 Nanorestore gel, new formulation Surface not cleaned in areas with insufficient
Mix 50/50% 10-2 and 10-4M 5 hrs of immersion in treatment solution contact with gel
2 applications 1 hr + 1 hr (15’ rinsing of gel in treatment solution
during application time); total application time: 2 hrs
F Ion exchange resins Applied as a paste. Mixed with osmose water Effervescence
(strong cation exchanger) Surface stirring with wooden stick Non-homogeneous surface cleaning
Amberlite® IR 120H Removal by stick and osmose water using cotton bud Localised polishing (bare metal) due to rubbing
Treatment time: approx. 10’ action with stick
G1 Micro-sandblasting with Pressure: 0.3 atm approx.; granulometry: 45 µm–90 µm; time: More uniform and grey surface
glass microbeads approx. 2’ Surface polished
Diam. nozzle: approx. 1 mm
G2 Micro-sandblasting with Pressure: 0.9 atm approx.; granulometry: 150 µm; time: approx. 2’ More uniform and grey surface
plant-based abrasive Diam. nozzle: approx. 0.75 mm Dull surface, cracks tending to crumble around
edges
H1 Not cleaned Reference zone
H2 Citric acid 3 Gel: Nevek® (50 wt%) Corrosion products remain
5 g/l Applied hot: 2 applications (0.5 hr + 4 hrs);
applied as a film: 1 application (1hr); total application time: 5.5 hrs
334  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CONSERVATION TREATMENTS

Table 3. Summary of operating conditions and observations on the second series of cleaning tests (front, engraved surface of tablet 2012421)

ZONES Solution Method Observations

A Micro-sandblasting with Microbeads (45 µm–90 µm); pressure: 0.5 bar; nozzle Superficially shiny, smoother surface, cracking more visible
glass microbeads diameter: 0.6 mm

B Micro-sandblasting with Microbeads (45 µm–90 µm); pressure: 0.5 bar; nozzle
glass microbeads diameter: 0.6 mm
C Ion exchange resins Application of Amberlite IR120H resins (strong cation
exchanger), stirred with wooden stick, then rinsed with
osmose water using cotton bud
D Micro-sandblasting with Lignoblast (CN100), impalpable; pressure: 1.5 bar; nozzle
plant-based abrasive diameter: 0.6 mm
E Micro-sandblasting with Lignoblast (CN100), impalpable; pressure: 1,5 bar; nozzle
plant-based abrasive diameter: 0.6 mm
F Ion exchange resins Application of Amberlite IR120H resins (strong cation
exchanger), stirred with wooden stick, then rinsed with
osmose water using cotton bud
G Oxalic acid 3 successive applications of Nevek gel (50%m in osmose
water) in slab form and soaked in oxalic acid (5 g/l) for 24 hrs
Applied cold to surface with a weight placed on it
(1st application: 2 hrs; 2nd: 2 hrs; 3rd: 4.5 hrs)
Rinsing after each application with a cotton bud soaked in
osmose water
H Oxalic acid 3 successive applications of Nevek gel (50%m in osmose
water) in slab form and soaked in oxalic acid (5 g/l) for 24 hrs
Applied cold to surface with a weight placed on it
(1st application: 2 hrs; 2nd: 2 hrs; 3rd: 4.5 hrs)
Rinsing after each application with a cotton bud soaked in
osmose water
I Ion exchange resins Application of Amberlite IR120H resins (strong cation
+ exchanger), stirred with wooden stick, then rinsed with
Micro-sandblasting with osmose water using cotton bud
plant-based abrasive + Lignoblast (CN100), impalpable; pressure: 1,5 bar; nozzle
diameter: 0.6 mm
Superficially shiny, smoother surface, cracking more visible
Reduction in the superficial layer of corrosion products Whitish
residue in micro-cracks. Some bare metal exposed
Elimination of external corrosion products and dirt
Micro-cracking exposed
Elimination of external corrosion products and dirt
Micro-cracking exposed
Reduction in superficial layer of corrosion products.
Whitish residue in micro-cracks. Some bare metal exposed
Reduction in superficial layer of corrosion products, but only in
areas in direct contact. Formation of a generalised whitish veil
which increased with number of applications
Reduction in superficial layer of corrosion products, but only in
areas in direct contact. Formation of a generalised whitish veil
which increased with number of applications.
2nd and 3rd applications were restricted to the most corroded
areas: superficial whitening is mainly visible on these areas
The most concreted lead carbonate clusters were not attacked.
Micro-cracking was exposed

with glass microbeads polished the surface and, although
performed at very low pressure, was found to be too
aggressive.
The effectiveness of acid solutions depended on how
they were applied. In any case, it was necessary to
proceed cautiously, because the supports used masked
surfaces and results could only be checked after rinsing.
Furthermore, the use of gels required a certain amount
of preparation. On a macroscopic scale, no obvious
differences were visible among acid-cleaned surfaces,
except for those engraved surfaces that were cleaned
with oxalic acid. On these surfaces, a film of white
matter appeared which intensified when the application
time was lengthened. IR analysis indicated that this
white film consisted of lead oxalates. Some differences
were noted in the various gel treatments. Xanthan,
used hot, adhered well to surfaces, but proved difficult
to remove because of its high viscosity. The chemical
gel Nanorestore left no residue, but unfortunately the
film did not adhere perfectly to uneven surfaces, and
its action remained superficial. Nevek gel presented
interesting characteristics, used either hot or cold: in
the first case it moulded to the surface perfectly but
the acidity of the solution modified its film-forming
properties, while using it cold (in a slab) reduced the
effectiveness of contact with the surface. For tests
carried out with ion exchange resins, the action was
visible immediately after application in the form of
effervescence. It was possible to monitor the removal
of corrosion products under binocular microscope and
halt cleaning with their elimination. However, some
dust resin residue remained in the micro-cracks. On
the engraved surfaces, the occasional appearance of the
underlying metal corrosion products was observed – a
sign that the action had been too incisive.
In reality, experience shows that a combination of
different cleaning techniques is generally the best solu-
tion, although in this study the simultaneous use of
more than one technique was avoided so as to enable
differences to be evaluated.
STUDY AND CONSERVATION OF LEAD CURSE TABLETS
Only one test was conducted that combined cleaning with
ion exchange resins and micro-blasting with plant-based
abrasives, because this method had been frequently used
by the laboratory in the past.
Imaging before and after cleaning
‘Open Light’ photos require only basic technical knowl-
edge in photography and give fast results with good
readability of the details of engravings (Figure 4b). On
the other hand, this technique allows only for subjective
comparison of tablet surfaces before and after cleaning
because no measurement parameters can be determined.
This is also the case for macroscope images which yield
very good image quality with a precision of 20 to 40 μm
(Figure 4d). The 3D images generated by 3D structured
light scanning are very high resolution (7 μm) (Figures 4c
and 5) and scanning a whole tablet takes only a few
minutes. Processing allowed images to be compared
before and after treatment and revealed a deformation
of approximately 1.5 mm caused by cleaning (Figure 6).
Figure 4. Photograph of a fragment of tablet 2012421 in compression z
(a), in open light (b), by 3D scanner (c) and by macroscope (d)
Figure 5. Image of tablet 2012422 by 3D structured light scanning (a)
and topography with false colour code to show relief (b)
Deformations were due both to the extremely malleable
nature of lead and the mechanical action of cleaning.
To minimise the impacts of mechanical cleaning, close-
ly-fitting supports adapted to tablet morphology were
designed for the cleaning of further tablets, notably
335
those presenting a high degree of mineralisation. 3D
image processing eliminates differences caused by the
deformation of the tablet to highlight the finer variations
due to cleaning. Results obtained with this technique on
the non-engraved reverse did not show any detectable
changes. For the study of the engraved side, a more
precise technique, microtopography, was used to evaluate
the impact of cleaning on inscriptions. Comparison of
microtopographic images highlighted changes in the
appearance of surfaces in a qualitative manner. For three
of the windows (A: micro-sandblasting with glass beads;
D: micro-sandblasting with a plant-based abrasive; and
I: ion exchange resin and abrasive sandblasting), the
treatments had a homogenising effect on the surface.
This would seem to be induced by the removal of major
corrosion clusters (greater than 200 µm). It should be
noted that window B (micro-sandblasting with glass
beads), which was treated in an identical manner to A,
displayed a surface which appeared more cracked. This
was also the case for zones F (ion exchange resins) and G
(gel impregnated with oxalic acid). Post-cleaning cracks
often appear to be extensions of pre-existing ones. This
raises the question as to whether they could be pre-ex-
isting cracks that are being emptied of corrosion prod-
ucts by the cleaning process. It is necessary, however, to
add that an attenuation of the edges of engravings was
observed with a consequent diminution in readability.
Only window E which underwent micro-sandblasting
with a plant-based abrasive seemed to have retained
the initial properties of the micromorphology of the
profiles, in particular the depth of the groove and the
bulge formed by the engraving along one of the edges
(Figure 6). Additionally, the surface displayed a more
Figure 6. Highlighting the variation in shape before and after cleaning
of the entire reverse side of tablet 2012421 (a) or of a detail (b) by
structured light 3D scanning
336  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
granular appearance after cleaning. In comparison with
window E, it can be observed in zone D (also cleaned
by micro-sandblasting with a plant-based abrasive) that
the engraved line is shallower. This is probably related to
an attenuation of edges. The effect of edge attenuation
together with the appearance of cracks would seem to
indicate that treatments removed clusters and corrosion
products both on the surface and within the engraved
grooves. As a result, both the outlines of engravings and
cracks linked to the deterioration of tablets were revealed.
In other cases, however, it was noted that the edges of
engraved lines had been blunted by cleaning, resulting
in certain features being less marked.
Figure 7. Microtopographic image of zone E. Above: block 3D image (a)
and grey-level image (b); below: evolution of profile in zone E (cleaning
with plant-based abrasive) before and after cleaning
Conclusion
Studies conducted on the curse tablets displayed consid-
erable disparities in their state of conservation, with some
presenting a high degree of mineralisation. Analysis of
corrosion facies showed that lead carbonates had prob-
ably been formed post-excavation due to contact with
wooden storage containers. Various cleaning techniques
were tested on tablet 2012421: mechanical cleaning
(micro-sandblasting with glass or plant-based abrasive
microbeads) and chemical cleaning (ion exchange resin
or acid solutions applied in gel form). The impact of
cleaning techniques was evaluated by different imaging
techniques: open light photography, macroscopy, 3D
structured light scanning and microtopography. Only
the highest-precision techniques (microtopography)
CONSERVATION TREATMENTS
allowed for the impact of cleaning to be measured. 3D
scanner imaging highlighted structural deformation in
tablets due to cleaning-related handling, while microto-
pography enabled the impact of cleaning on inscriptions
to be precisely determined. The cleaning process revealed
both inscriptions and cracks due to deterioration. Finally,
irrespective of the cleaning method employed, engrav-
ings were slightly attenuated. Micro-sandblasting with
plant-based abrasive appeared to be the least invasive
technique, although care should be taken about general-
ising these results and the suitability of a given cleaning
technique should be evaluated on a case-by-case basis
according to the state of surface conservation. All the
techniques selected for this study provided yielded or
3D images of particular interest for the deciphering of
signs by epigraphers.
Acknowledgements
This study was carried out with the financial support
of the French Ministry of Culture. It would not have
been possible without the support of Mathilde Avis-
seau-Broustet, Chief Curator at the Department of Coins,
Medals and Antiques of the National Library of France,
who made the curse tablets available to us. We would
also like to thank Charlene Pelé-Meziani (IR), Jean Yves
Mevellec (Raman spectroscopy) and Pierre Emmanuel
Petit (XRD) for their analyses, not to mention Carl
Colonnier for his precious help in imaging and Gabriel
Koëller from Eotech, who produced the 3D structured
light scanning images.
References
Audollent, A. 1904. Defixionum tabellae, 360–99. Paris:
in aedibus A. Fontemoing.
Degrigny, C. and R. Le Gall. 1999. Conservation of
ancient lead artifacts corroded in organic acid environ-
ments: Electrolytic stabilization/consolidation. Studies
in Conservation 44: 157–69.
Fontaine, C., E. Guilminot, R. Jeanneret, and L. Rossetti.
2016. Determination of parameters for local electrolytic
treatment of corroded lead and lead-tin alloys. Journal
of Cultural Heritage 20: 607–14.
Frost, R., R. Scholz, A. López, B. Firmino, C. Lana, and
Y. Xi. 2014. A Raman and infrared spectroscopic charac-
terisation of the phosphate mineral phosphohedyphane
Ca2Pb3(PO4)3Cl from the Roote mine, Nevada, USA.
STUDY AND CONSERVATION OF LEAD CURSE TABLETS 337

Spectrochimica Acta Part A: Molecular and Biomolecular

Spectroscopy 127: 237–42.
Guilminot, E., L. Rossetti, D. Teigne, N. Faisant, J. Pichon,
A. Pineau, Y. François, A.-L. Carré, S. Maillard, and H.
Chew. 2012. Prévention des risques associés au plomb
en conservation-restauration : de l’atelier de restauration
au musée. In Actes du colloque Sciences des matériaux du
patrimoine culturel, Paris, 20–21 November 2012: 87–90.
Paris: Ministère de la Culture et de la Communication.

Authors
Loretta Rossetti is a metals conservator at Arc’Antique
where she has worked since 1996. She is a graduate of
the School of Restoration, Villa Manin de Passariano,
Udine, Italy (1994).

Jean Gabriel Aubert is a photographer and radiologist

at Arc’Antique.

Nicolas Stephant is a scanning microscope operator at

the Institut des Matériaux Jean Rouxel de Nantes (IMN).

Nicolas Mélard is an archaeologist and curator at the

Centre de Recherche et de Restauration des Musées de
France (C2RMF) and an imaging expert.

Elodie Guilminot completed her engineering diploma

at Polytech Nantes (France) in 1996 and obtained a PhD
in electrochemistry at the Institut National Polytechnique
de Grenoble (France) in 2000. Her PhD thesis focused on
the conservation of waterlogged wood/metal composites
in collaboration with the French conservation-restoration
laboratories Arc’Antique and ARC-Nucléart. She has
worked at the French Institute of the Sea (IFREMER)
and the University of Grenoble on the corrosion and
characterisation of materials. She joined the Arc’Antique
laboratory (Nantes, France) as conservation scientist in
2006 to conduct and coordinate research projects. Her
research interests include the corrosion of metals and
the development of restoration treatments.
Revealing the H.L. Hunley: The Strategic Planning and
Deconcretion Process of an American Civil War Submarine
Johanna Rivera* Michael Scafuri
Warren Lasch Conservation Center Warren Lasch Conservation Center
– Clemson University – Clemson University
Charleston SC, USA Charleston SC, USA
jrd123@clemson.edu scafuri@clemson.edu
Paul Mardikian Liisa Näsänen
Terra Mare Conservation, LLC Warren Lasch Conservation Center
Charleston SC, USA – Clemson University
paul@terramareconservation.com Charleston SC, USA
lnasanen79@gmail.com
Abstract
In August 2014, after years of planning and great anticipation, the
Hunley team was finally ready to reveal the hull of the submarine
for the first time in 150 years. The main conservation objective
was to remove the concretion covering the surface of the hull
and to document and preserve as much of the original metal as
possible without compromising the structural integrity of the
submarine. In parallel to the conservation work, the objective of
the archaeological team was to document and record the hull
Introduction
The H.L Hunley submarine disappeared off the coast
of Charleston, South Carolina, on February 17, 1864,
following an attack on USS Housatonic, becoming the first
submarine in history to sink an enemy ship in battle. On
August 8, 2000, the submarine was recovered from the
ocean and transported to the Warren Lasch Conserva-
tion Center (WLCC) in North Charleston for excavation
and conservation. After an extensive technical analysis
of the submarine, a conservation plan was submitted to
the U.S. Navy for review and formal approval, as well as
peer-reviewed (Mardikian et al. 2006). This document
served as a general roadmap for conservation for the
duration of the project and was centered around three
primary objectives. These were to:
1. preserve H.L Hunley’s structural integrity;
2. identify techniques that would stabilize the submarine
with minimal risk; and
3. develop and define display conditions that would
minimize the need for future conservation work.
Although challenging, the multifaceted objectives of
the conservation plan presented a unique opportunity
Nestor González-Pereyra
Warren Lasch Conservation Center
– Clemson University
Charleston SC, USA
nestg@g.clemson.edu
*Author for correspondence
and its diagnostic features. The challenge of achieving these
objectives had to be combined with the need to ensure safe
working conditions for personnel and colleagues throughout
the duration of the project.
Keywords
H.L. Hunley, underwater archaeology, conservation,
deconcretion, large-scale marine objects
to perform an in-depth analysis of the construction of
the Hunley submarine, revisit the current state of our
knowledge regarding the conservation of large marine
objects, and compare traditional and experimental tech-
niques for the stabilization of archaeological iron objects
(Mardikian et al. 2006, 2009a, 2009b).
Prior to the deconcretion of the submarine, several steps
were taken to improve the access to the hull as well as
its stabilization. These steps involved the removal of
high-risk materials from the submarine (such as organic
elements), the removal of glass viewports (unstable in
strong alkaline solutions), a forensic and structural study
of the hull, the disassembly and removal of mechanical
components (i.e., cast-iron pumps), paint analysis,
metal and biofouling analysis, a feasibility study and
rotation of the submarine to an upright position, and
the modification of the fluid management system to
be able to safely immerse the submarine in sodium
hydroxide (Mardikian et al. 2006). Another part of the
project’s objectives was to allow for public viewing of
the submarine during treatment. All of these critical
REVEALING THE H.L. HUNLEY: THE STRATEGIC PLANNING AND DECONCRETION PROCESS OF AN AMERICAN CIVIL WAR SUBMARINE
steps had to be addressed individually, then rolled into
a working plan so work could start. This paper discusses
the planning and execution of the deconcretion and the
practical challenges faced when working with large-scale
marine iron artifacts.
The Hunley rotation
One of the most important steps was to rotate the
submarine from its original 45-degree angle, remove the
lifting truss and supporting slings, and rest it on its keel
blocks. The purpose was to facilitate access to the surface
of the submarine during conservation work and permit
documentation and deconcretion of inaccessible areas of
the hull. A numerical simulation was developed for the
rotation. This procedure involved gradually adjusting the
support slings to be longer on the port side and shorter on
the starboard side. This “pull-and-release” system would
rotate the submarine in a slow and controlled manner to
an upright position (Blouin et al. 2011). Following this
plan, the team successfully rotated the vessel and placed
it on its keel blocks in 2011 (Figure 1).
Figure 1. H.L. Hunley submarine during rotation
Tank overhaul project
In preparation for the deconcretion and desalination of
the submarine, the team was posed with the challenge of
preparing 284,000 liters of a solution of sodium hydroxide
(pH above 13) needed to fill the Hunley tank. Keeping
the safety of personnel and visitors as a top priority,
a detailed assessment of the capability of the existing
systems was initiated in 2012. The conclusion was that
the fluid handling system originally designed to transfer
water, and eventually an alkaline solution to and from the
tank, needed to be upgraded to comply with federal, state,
and local codes. Therefore, the overhaul of the system
and general tank area was initiated in November 2013.
339
It started with the construction of secondary contain-
ment concrete walls around the tank and around the six
56,000-liter outdoor mixing tanks. The fluid handling
system was reconfigured and upgraded to allow for the
preparation and adjustment of the sodium hydroxide
solution, its transfer to and from the Hunley tank, and
its chemical neutralization before disposal. This work
involved new piping, pumps, and valves, as well as new
electrical wiring for the power and control of the new
automatic fluid control system. The final system enabled
the team to perform most of the operations related to the
handling of the fluid remotely or manually as needed. A
filtering system was added to reduce the amount of solids
in suspension in the Hunley tank in order to improve
viewing during public tours.
Other upgrades included new stairs and scaffolding
to permit better access to the tank, chemical cabinets
for large amounts of sulfuric acid and concentrated
sodium hydroxide, and a harness fall protection system
around the Hunley tank working area. A motor-operated
retractable cover was installed over the tank to minimize
the formation of carbonates in the solution (due to the
chemical reaction between the carbon dioxide in the
air and the sodium hydroxide) and pH drop. For safety
purposes, a folding glass wall was installed to protect
visitors from the tank without obstructing the view of
the submarine.
Exterior deconcretion
The marine concretion covering the submarine was
the result of years of macrofaunal growth and debris
accumulation on the surface of the submarine while
it rested on the seafloor. This growth, combined with
iron corrosion products, formed a very hard protective
marine concretion. The concretion recorded vital site
information regarding the physical and environmental
conditions around the submarine. The environmental
factors that led to the formation of this concretion have
been outlined in a number of studies of the site (Murphy
et al. 1998, Conlin 2005, Jacobsen 2012, Jacobsen et al.
2012, Harris and Neyland 2016).
Following the modification to the fluid management
system, preparations were made to begin the deconcre-
tion of the submarine’s exterior and the active treatment
of the iron hull. A detailed deconcretion plan, written as
an addendum to the Hunley Conservation Plan (2006),
was drafted by former senior conservator Paul Mardikian
340  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
in 2014 and submitted to the U.S. Navy for review and
approval. The deconcretion plan was reviewed and
discussed with the conservation and archaeological teams
to produce a step-by-step guideline with clear objectives
and goals for the team to follow during the process.
The first objective of the deconcretion process was
to “remove the concretion covering the surface of
the hull and document and preserve as much of the
original surface as possible without compromising
the structural integrity and stability of the submarine”
(Mardikian 2014). The archaeological objectives were to
document the concretion as it was being removed and
to record the hull and surface features revealed during
the process. This was critical to the interpretation of
how the submarine functioned and what possibly led
to the sinking 150 years previously (Mardikian 2014,
Rivera et al. 2017, Scafuri 2017).
Prior to the deconcretion, both teams conducted a
preliminary survey of the vessel to identify and mark
sensitive areas of the hull to be initially avoided. These
areas were mostly spots of hull damage including holes
in the forward ballast tank, the aft ballast tank, and the
port viewport on the forward conning tower (Mardikian
2014, Scafuri 2017). Additionally, certain parts of the
submarine deemed to be of forensic importance were
also marked.
In May 2014 the submarine was placed in a recirculated
solution of sodium hydroxide solution designed to let the
hull passivate for several months and mitigate corrosion
on the reactive metal once exposed to the atmosphere
(Mardikian 2014). The anticipation was that subjecting
the hull to a high pH would not only passivate the metal
but would also soften and facilitate the removal of the
concretion.
The deconcretion of the exterior of the H.L. Hunley
submarine started in August 2014. A number of tech-
niques and tools were considered to aid in the removal of
concretion, the most effective being mechanical cleaning
via the use of hand tools such as pneumatic chisels,
surgical chisels, and hammers. The process began with
the removal of the exterior concretion along the riveted
seam lines of the submarine and then gradually transi-
tioned to interior surfaces and operational components.
An initial probe line was made along the length of the
submarine on both port and starboard sides. The purpose
of the probe line was to assess the condition of the subma-
CONSERVATION TREATMENTS
rine’s hull as well as to identify and reveal individual hull
plates, their seams, and structural features (Figure 2).
Figure 2. Probe lines on expansion strake and hull plates seam lines
This method would also help the documentation of the
submarine and measure the thickness of the concretion
per hull plate throughout the process. The thickness of
the concretion was found to vary over the length of the
submarine depending on the location and condition of
the hull surface. In general, the concretion on the port
side was thicker due to its longer exposure to concre-
tion-forming conditions, whereas that on the starboard
side had both a short exposure duration and significant
scouring to thin the layers (Scafuri 2017, 6). The decon-
cretion also revealed evidence of erosion, fractures,
impacts, and inherent material defects. This damage was
more significant on the starboard side of the submarine
where marine currents scoured and eroded the subma-
rine, creating hull breaches related to the thinning of the
metal. This effect was clearly visible at the forward edge
of the bow cap, where it had been eroded and scalloped
reducing the original vertical bow to a convex curved
edge (Figure 3) (Jacobsen et al. 2012, Scafuri 2017).
It took over a year to complete the deconcretion of the
exterior hull. Approximately 1,000 kg of concretion was
removed from the hull by a team of four conservators.
Work was limited to three days a week to reduce expo-
sure of the submarine to air. An automated system was
used to drain the tank Tuesday through Thursday, from
8 a.m. to 1:30 p.m., with the solution otherwise covered
with a retractable pool cover to minimize evaporation
and formation of carbonates. Before any work could
begin on the submarine, conservators would install a
removable railing for the stairs going into the Hunley
tank, followed by portable eyewash stations, a safety
REVEALING THE H.L. HUNLEY: THE STRATEGIC PLANNING AND DECONCRETION PROCESS OF AN AMERICAN CIVIL WAR SUBMARINE 341

Interior deconcretion
The deconcretion of the interior presented a number
of additional technical challenges due to the confined
space and limited accessibility of portions of the vessel
particularly in areas such as the narrow ballast tanks and
around the interior propulsion system.

Conservators implemented a number of safety proto-

cols in compliance with federal and state regulations
for working in a confined space and in contact with
chemicals. These included the installation of eyewash
Figure 3. Cast-iron stations inside the submarine itself and the use of Tyvek
bow cap with suits as protection from the hydroxide solution as well as
a convex edge full-face respirators, gloves, sleeves, and ear protection.
shape due to
Lastly, a harness connected to an overhead crane was
 currents scouring
used for fall protection when getting in or out of the
submarine (Figure 4).
horn, compressed air hoses, and any other tool required
to do the work in the tank.

One of the many challenges of this phase of the project

was to monitor the corrosion of the submarine and to
adjust the concentration of sodium hydroxide to the
Hunley tank to maintain the highest possible pH. This
proved to be a very difficult and expensive task due to
the amount of solution in the tank, the large volume
and surface area of the tank, and the increasing need
for protection as more of the surface of the submarine
became exposed. Despite the use of a retractable cover
above the submarine as often as possible, the transfer of
solution from the Hunley tank to and from the storage
tanks contributed to the formation of carbonates via Figure 4.
stirring and exposure to air, gradually reducing the Conservator
pH from 13.2 to 12.5 within the first year. Despite the getting inside the
submarine using
team’s best efforts to spray the submarine with dilute
fall protection
sodium hydroxide and cover previously deconcreted system required by
areas with plastic sheets, while adjusting the pH to state and federal
maintain values above 13, it became clear that this was  regulations
not enough to prevent additional corrosion. With a low
Due to the limitations of the space, in most cases only
native potential around -0.470 volt vs. Ag/AgCl, it was
two conservators would work on the interior of the
decided to re-start the cathodic protection system that
submarine at the same time. Since the concretion layer
was originally used during the excavation – and kept
from the exterior had been removed, the team was also
in case of emergency – when the submarine was stored
careful not to put extra weight or constraints on the
in city water. After adjusting the current as more metal
already fragile hull plates.
was gradually being exposed, the team successfully
regained control over the corrosion. The potential of The strategy used to deconcrete the interior was to begin
the submarine gradually reached 1.0 volt vs. Ag/AgCl, with two conservators, one conservator working on the
allowing deconcretion work to resume again after propulsion system in the aft portion of the crew compart-
several weeks. ment, and one underneath the forward conning tower
342  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND CONSERVATION TREATMENTS

in the captain’s station of the submarine. Once these two

critical areas were finished, they would move towards
the center of the submarine and begin deconcreting the
remaining top and bottom hull plates.

Features and metal interpretation

Understanding the nature and behavior of cast and
wrought iron is essential to differentiate surface damage
caused by erosion, deformation, or fractures from metal
defects inherent to the metallurgical process indicative of
an incomplete or faulty casting (blowholes or misruns).
Surface damage caused by erosion has equally affected
cast and wrought iron on the Hunley. However, plastic Figure 6. Lower section of the cast-iron stern showing what appears to
be a case of a misrun in which the liquid solidified before being able
deformation that has caused a permanent change in
to fill the mold
shape was only observed on the wrought-iron plates in
the area of the portside dive plane. In addition to evidence of metallurgical defects, the
deconcretion also revealed many archaeological features.
Examples of fractures on the submarine are common
These features provided a number of interesting clues to
and include those found on the bow and on the propeller
the design, operation, and modification of the submarine
shroud. Metal defects such as blowholes were identified
as well as some insights into damage that occurred to the
on the cast-iron elements. A blowhole is a metallurgic
vessel. Because of the controlled nature of the deconcre-
term used to describe a cavity left in a casting near or
tion process, the archaeological team was able to collect
at the surface of the metal after solidification due to
very comprehensive information. Additionally, all of the
entrapped air or gas. Numerous cases of blowholes were
concretion removed from both the exterior and interior
noted on the Hunley’s cast-iron elements during the
was sampled according to the hull plate or the feature it
deconcretion, particularly on the dive planes, snorkel box,
came from. This material was then screened and x-rayed
deadlight covers, and bow and stern castings (Figure 5). A
to ensure that any small artifacts in the concretion would
possible case of misrun was identified on the aft cast-iron
be located and any anomalous material sub-sampled for
cap (Figure 6). A misrun is a term used when the molten
additional analysis (Scafuri 2017).
metal has failed to completely fill the mold due to a lack
of fluidity of the alloy, inadequate pouring conditions, While much of the damage revealed by the exterior
or venting of the mold. These metallurgical defects are deconcretion could be explained by corrosion and
quite interesting as they suggest limited manufacturing scour, some damage on the bow may be the result of
capabilities and resources during the construction of the Hunley’s torpedo attack on USS Housatonic (Scafuri
the submarine. 2017). Other features of the bow showed clear signs of
being later modifications to the hull, reflective of changes
to the submarine’s weapon system. Damage revealed in
the stern area was less conclusive. Both the rudder and
the cast-iron starboard propeller shroud were broken off
at some point. While indicative of physical trauma and
not corrosion, it is not clear how or why this damage
occurred (Scafuri 2017).

The deconcretion of the interior revealed even more


Figure 5. X-ray of the portside cast-iron dive plane showing subsurface
blowholes entrapped in the solidified metal
features of significance to the archaeological interpre-
tation and analysis of the submarine. Many of these
interior features represented components and equip-
ment related to the submarine’s operation, such as those
revealed in the captain’s station at the forward end of
REVEALING THE H.L. HUNLEY: THE STRATEGIC PLANNING AND DECONCRETION PROCESS OF AN AMERICAN CIVIL WAR SUBMARINE
the crew compartment (Figure 7) where he controlled a
number of navigational components, namely the ballast
system, the dive plane counterweight, the steering rod,
the depth gauge system, and the propulsion system.
The hand crank was also deconcreted, revealing metal
sheeting that encased each crank handle as well as textile
wrapped around the metal. The metal and textile were
most likely placed there to protect the hands of the crew
while cranking (Figure 8).
Figure 7. Lt.
George Dixon
 station or Grid 1
Figure 8. Wrought-
iron hand crank
 after deconcretion
Deconcretion of the aft end of the crew compartment
revealed other interesting features, most specifically
related to the propulsion of the submarine. In particular,
the crank terminated in a gear system that would have
added mechanical advantage to the turning of the
propeller. The propulsion shaft then passed through a
343
well-preserved cast-iron flywheel. The flywheel would
have also aided in the rotation of the propeller, giving
the crew additional support in propelling the subma-
rine. Because of the data collected from these areas,
the archaeological team has been able to begin a more
detailed analysis of the speed and propulsive capabilities
of the submarine and its crew (Figure 9).
Figure 9. Grid 8 or propulsion system before and after deconcretion.
Features revealed are the cast-iron flywheel, copper intake ballast pipe
and handle, crank system with discs and chain to help move the flywheel,
and the steering system behind flywheel
Conclusion
As a result of careful planning and preparation, the
deconcretion of the H.L. Hunley submarine thus far has
been largely successful (Figure 10). The scientific team
has had to overcome a number of technical challenges
and adapt to many unforeseen circumstances throughout
the long process. There were very few direct parallels to
draw from, and much of the conservation plan had to
be developed to address the uniqueness of this project.
This required extensive collaboration between the
conservation and archaeological teams and ultimately
that resulted in the successful completion of this critical
phase of the project.
Figure 10. Aft side view of the submarine after deconcretion
Acknowledgments
The authors would like to thank all the people involved
during the deconcretion project that have helped to make
344  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
this happen: Stephanie Crette, Virginie Ternisien, Emily
Schwalbe, Craig Meier, Eddie Piedmond, and Paul Ogle.
References
Blouin, V.Y., P. Mardikian, and C. Watters. 2011. Finite
element analysis of the H.L. Hunley submarine: A
turning point in the project’s history. In Metal 2010:
Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, Charleston, South Carolina, USA,
11–15 October 2010, eds. P. Mardikian, C. Chemello, C.
Watters, and P. Hull, 393–99. Clemson, SC: Clemson
University.
Conlin, D., ed. 2005. USS Housatonic site assessment.
Washington, DC: National Park Service, Naval Historical
Center, South Carolina Institute of Archaeology and
Anthropology.
Harris, S. and R. Neyland. 2016. Environmental context
and site formation. In H. L. Hunley: Recovery operations,
eds. R.S. Neyland and H.G. Brown, 51–65. Washington,
DC: Naval History and Heritage Command.
Jacobsen, M., V. Blouin, and W. Shirley. 2012. Does
erosion corrosion account for intriguing damage to the
Civil War submarine H.L. Hunley? Marine Technology
Society Journal 46(6): 38–48.
Mardikian, P. 2004. Conservation and management strat-
egies applied to post-recovery analysis of the American
Civil War submarine H.L. Hunley (1864). The Interna-
tional Journal of Nautical Archaeology 33(1): 137–48.
Mardikian, P., M. Drews, N. González, and P. de Viviés.
2006. H.L. Hunley Conservation Plan, unpublished report
submitted to the Naval History and Heritage Command,
Department of Navy, Washington DC.
Mardikian, P., M.J. Drews, N.G. González, and P. de
Viviés. 2009a. New perspectives regarding the stabili-
zation of terrestrial and marine archaeological iron. In
Iron, Steel & Steam, Fremantle, Australia, 26–27 June
2006, ed. M. McCarthy, 113–18. Special publication,
Australian National Centre of Excellence for Maritime
Archaeology No. 13.
Mardikian, P., M.J. Drews, N.G. González, P. de Viviés,
and J.W. Hunter. 2009b. Searching for a diamond in
the rust: A review of large-scale archaeological iron
conservation projects, 1980–2006. In Iron, Steel & Steam,
Fremantle, Australia, 26–27 June 2006, ed. M. McCarthy,
82–88. Special publication, Australian National Centre
of Excellence for Maritime Archaeology No. 13.
CONSERVATION TREATMENTS
Mardikian, P. 2014. Deconcretion plan for the H.L.
Hunley submarine, addendum to H.L Hunley Conser-
vation Plan, unpublished report.
Murphy, L.E., ed. 1998. H.L Hunley site assessment. Santa
Fe, New Mexico: National Park Service, Submerged
Cultural Resources Unit.
Rivera, J. and M. Scafuri. 2017. The sea, the sub and mari-
time collaboration: How conservators and archaeologists
worked together to recover and conserve the H.L. Hunley
submarine. In Engaging conservation collaboration across
disciplines, eds. N. Owczarek, M. Gleeson, and L. Grant,
181–89. Philadelphia, PA: University of Pennsylvania
Museum of Archaeology and Anthropology.
Scafuri, M. 2017. The H.L. Hunley revealed: Docu-
mentation, deconcretion, and recent developments in
the investigation of an American Civil War submarine
from 1864. International Journal of Nautical Archaeology
46(2): 303–16.
Authors
Johanna Rivera obtained her B.A. in Fine Arts from
the Universidad de Chile in 2001, and her postgraduate
degree in Conservation of Cultural Heritage in 2004. In
2011, she completed a graduate certificate in Museum
Collections Management and Care from George Wash-
ington University. She joined the H.L. Hunley submarine
project in 2005 and her work has included the conser-
vation and treatment of marine archaeological artifacts
and managing the collection and storage rooms. She is
currently leading the Hunley deconcretion process as
well as the conservation of artifacts from several other
projects.
COMPOSITE OBJECTS
Challenges and Solutions for Conserving Non-Separable
Marine Archaeological Composite Objects

Paul Mardikian
Terra Mare Conservation LLC
Charleston SC, USA
paul@terramareconservation.com

Abstract
This paper discusses the conservation of three non-separable
marine composite artifacts treated over the last 25 years by
the author: a flushable toilet made of ceramic, lead, brass, cast
iron and wood recovered from the Confederate raider  CSS
Alabama (1864); a steel and tin canteen with a cork stopper and
shoulder strap made of hemp rope from the H.L. Hunley Civil
War submarine (1864); and a fuel injector made of copper,
chrome and 3xx series stainless steel recovered from one of
the Apollo 11 Saturn V rocket engines (1969). Treatment of
the objects comprised chemical or electrochemical cleaning
and stabilization techniques in various solutions, including tri-
sodium citrate, alkaline Rochelle salts, ammonium hydroxide
and non-vacuum freeze-drying. This paper demonstrates that
conserving complex multicomponent objects is possible by
fine-tuning traditional techniques and adapting new ones. In
particular, the use of ammonium hydroxide for the conservation
of certain metal/organic materials should offer a new and
valuable alterative to conservators.
Keywords
marine archaeology, composite objects, chloride removal,
ammonium hydroxide, tri-sodium citrate, alkaline Rochelle
salts, non-vacuum freeze-drying

Introduction
Non-separable marine archaeological composite objects
are among the most difficult objects to conserve. These
objects may comprise any combination of materials,
including multiple metals and/or multiple organic
materials. Their component materials may have been
originally fitted together using either reversible or
non-reversible assembly methods, or a combination
thereof. Separating one component from another is risky
but sometimes necessary and is done on a case-by-case
basis depending on the significance of the object and
the state of degradation of the most fragile component
(often an organic component). More often than not,
separation is not a viable option and holistic and custom
treatment solutions must be researched and implemented
(Mardikian 1997a).
Research into the treatment of non-separable marine
composite artifacts emerged in the early 1980s, particu-
larly in Australia and Canada (MacLeod and North
1982, Strang 1983, Cook et al. 1984, MacLeod 1989). The
following decades saw a marked interest in the research
and development of custom treatment protocols using
corrosion inhibitors, neutral salts as electrolytes for the
stabilization of metals in the presence of pH-sensitive
materials, as well as techniques to physically protect
one of the components during stabilization. Examples
of contributions include MacLeod et al. (1993), Selwyn
et al. (1993), Degrigny (1995), Argyropoulos et al. (2000),
Degrigny et al. (2002), Memet and Tran (2005) and
Bruhin et al. (2017).
The three case studies presented in this article comprise
a multicomponent flushable toilet that could only be
partially separated before treatment, a canteen made
from two different metals and two different organic
materials which could not be separated due to their state
of degradation, and a large modern space artifact made
from three different metals that could not be separated
due to the method of fabrication.
Case study: The CSS Alabama flush toilet
Description and condition
The Confederate raider CSS Alabama was equipped
with two pairs of floor-mounted flushable toilets. Toilet
ALS-65 was recovered from the wreck in 1993 in two
CHALLENGES AND SOLUTIONS FOR CONSERVING NON-SEPARABLE MARINE ARCHAEOLOGICAL COMPOSITE OBJECTS
parts, the toilet itself and a fragment of the cast-iron
base that includes the cast-iron counterweight and
flush handle that separated during recovery (Mardikian
1997b). As shown in Figure 1, the object was covered with
voluminous concretions, seaweed and shells, particu-
larly on the cast-iron elements. The lead surface was
covered with a typically thin and adherent concretion.
The transfer-printed ceramic bowl was intact but heavily
soiled and stained with iron corrosion products. A small
fragment of pine planking was still attached to a brass
screw originally securing the toilet to the floor.
Figure 1. Alabama flushable toilet photographed after recovery in 1993.
Note the heavy concretion covering the cast-iron pedestal extending
to the square lead base situated immediately under the ceramic bowl
wrapped with lead
Treatment
After examination of the original brass bolts used to
assemble the different elements together, the object was
partially separated into two major composite components
consisting of the ceramic bowl/lead/brass insert and the
heavily de-alloyed and fragile cast-iron/lead pedestal.
The ability to partially separate the object into two parts
provided clear access to its interior for better cleaning
and stabilization.
The treatment objectives for the ceramic/lead/brass
component were to safely clean and desalinate the
ceramic bowl as well as clean and stabilize the encrusted
lead and brass components. The ceramic bowl was
first cleaned with an ultrasonic dental scaler to remove
the thickest deposit. The next phase of the treatment
addressed the chemical cleaning of the deeply seated
iron stains on the ceramic at the same time as the elec-
trochemical cleaning of the lead/brass component. This
347
objective was accomplished using a 2% (w/v) tri-sodium
citrate (Na3C6H5O) solution at a pH of 6 as an electrolyte
for the cleaning of the metals and gentle chelating agent
for the ceramic. Each liter of solution was prepared by
adding 17.47 grams of Na3C6H5O and 2.53 grams of
C6H8O7 to 1 liter of deionized water. The potential of
the object was maintained under -0.50 volts vs. SHE
(standard hydrogen electrode) to avoid lead dissolu-
tion and generate enough hydrogen to clean the metal
surfaces. Stainless steel anodes and recirculation of the
solution were used at all times. During treatment, large
patches of concretion gradually separated from the lead
and brass base revealing a well-preserved metal surface.
Finally, the object was cleaned with a soft brush and
rinsed in warm tap water for a month under polarization
at about -0.40 volts vs. SHE to eliminate any residual
salts or chemicals. After cleaning and stabilization, the
ceramic regained its original color and luster without
any damage to its glazing (Figure 2).
Figure 2. Partial view of the china bowl after mechanical and chemical
treatment
Treatment of the second non-separable component
of this artifact, the cast-iron/lead pedestal assembly,
began by removing the concretion by hand as much as
possible before stabilization in an alkaline solution to
facilitate chloride diffusion into the solution and reduce
the immersion time to a minimum. A 1% (w/v) recircu-
lated sodium hydroxide (NaOH) solution (pH 13) was
chosen to stabilize the cast iron while under cathodic
protection around -0.75 volts vs. SHE, to avoid corro-
sion of the lead in the high pH solution and excessive
hydrogen evolution on the cast iron. The solution was
renewed every month for three months until the chloride
level dropped under 2 ppm. The object was then rinsed
under cathodic protection for a month in warm tap water
regularly renewed. After drying, the cast-iron parts of
348  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
the pedestal were consolidated in a 20% (w/v) solution
of Paraloid B-48N in toluene for several days and let to
air-dry under a fume hood for two weeks. Finally, the
small piece of pine planking that was originally attached
to the base was easily removed, cleaned, soaked in a
solution of 25% (v/v) polyethylene glycol (PEG) 400 and
then vacuum freeze-dried for a week. After treatment,
the different components of the object were reassembled
using the original brass bolts as shown in Figure 3.
Figure 3. Toilet after treatment and re-assembly of main components
This case study demonstrates that a minimal approach
to disassembly coupled with a custom treatment worked
well for such a complex object. Recent communication
with the Naval History and Heritage Command in
Washington D.C. indicates that the object is still stable
25 years after treatment (Morrand 2018).
Case study: Canteen from the H.L. Hunley
submarine (1864)
Description and condition
A fragile fragmentary and concreted drum canteen,
HL-2159, made of tinned sheet iron was excavated from
the submarine in 2005. The canteen was found under the
crewmembers’ bench fused against the submarine’s hull
(Figure 4). The cork stopper was still in place along with
fragments of hemp rope around the base of the object.
After examination and x-radiography, it was found that
in addition to the corroded sheet iron, hemp rope and
cork, the canteen had a tin spout soldered to its body,
making the conservation and long-term stabilization of
this object even more challenging. Separation was briefly
considered for the cork, but it was so tightly held into the
tin spout that removal would have resulted in damage
to the cork and the spout. The remains of hemp rope
COMPOSITE OBJECTS
were also so firmly embedded in the hard sediment layer
that removal was simply impossible. In some areas, the
sediment had captured the imprint of the decayed rope,
which triggered the decision to try to preserve the object
in its entirety with its concretion, rope and cork in place.
Figure 4. Left: Author deconcreting canteen inside the Hunley.
Right: Canteen after removal. Note hemp rope on bottom left side of
photograph. Courtesy of Friends of the Hunley, Inc.
Treatment
The main conservation issue was to find a way to
stabilize the corroded tinned sheet of steel, and more
particularly to chemically remove the chloride ions from
the metal and surrounding concretion without affecting
the tin and/or the organic materials present. The poten-
tial-pH (Pourbaix) diagram for tin indicates that an
upper limit of pH 12 may be acceptable (Murphy 2006,
Selwyn 2018); however, given the presence of organic
materials, it was decided to keep the pH under 11 to
minimize the risk (Florian 1987). Other factors may
play a role in the corrosion rate of tin, particularly the
presence of other ions that may affect the solubility of
SnO2 (Tin (IV) Oxide), such as tin sulfides, and increase
the region of passivity of tin (Leidheiser 1971, Selwyn
2018). Therefore, a weak base of ammonium hydroxide
(NH4OH) was used with the objective of keeping the pH
under 11 while stabilizing the sheet metal. This decision
was based on similar work successfully carried out by
the author as early as 2005 on composite test samples
and artifacts from the Hunley (Mardikian 2005). Liquid
ammonia (NH3) was identified by Gilberg and Seeley
(1982) to stabilize archaeological iron, ammonia/ammo-
nium sulfate washing procedures were used by Macleod
(1983) for the treatment of corroded aluminum-copper
alloys, and ammonium hydroxide is mentioned by
Scott (2002) as a cleaning agent for bronze; however,
it appears that the use of ammonium hydroxide has
never been published to this day for the conservation
of marine composite objects. Another application of
ammonium hydroxide for the treatment of complex
CHALLENGES AND SOLUTIONS FOR CONSERVING NON-SEPARABLE MARINE ARCHAEOLOGICAL COMPOSITE OBJECTS 349

marine composite artifacts is presented by Argyropoulos

and the author in this volume.

After mechanical cleaning, the object was stabilized in

a 1% (w/w) solution of ammonium hydroxide at room
temperature under a fume hood. Note that a rise in
temperature of the solution from 20°C to 40°C may cause
tin to go from passivity to corrosion (Britton and Michael Figure 5. Left: Concreted side of the canteen after conservation. Right:
1955). Therefore, keeping the temperature of the treat- Opposite side of the canteen originally against the hull after treatment.
Courtesy of Friends of the Hunley, Inc.
ment bath under 20°C is recommended. The 1% (w/w)
solution was prepared by adding 36.1 grams of NH4OH
(29%, ACS Reagent grade) to 1 kilogram of deionized
water, giving a pH of 10.98. The 10-liter tank was stirred
with an aquarium pump for the duration of the treatment
and remained covered at all times with polyethylene film
on the surface and a lid to reduce evaporation, NH3 vola-
tile loss and prevent a drop in pH. Chloride monitoring
was performed using ion chromatography.

None of the exposed cork or rope fragments were coated

or protected during treatment. However, the outlet of
the recirculation pump was placed away from them to
avoid damage. The treatment solution plateaued around
233 ppm of chloride ions after a year. After a quick rinse
in deionized water, the object was placed in a new solution
of ammonium hydroxide. No chloride ions were detected
in the solution for four months, so the object was briefly
rinsed in deionized water and air-dried while keeping the
rope and the cork covered with a thin polyethylene film to
avoid damage. A rope consolidant consisting of 1% (w/v)
Bermocoll®, a non-ionic cellulose ether, 2% (w/v) glyc-
erol and 2% (w/v) PEG 400 (Hawley 1989) was used to
impregnate the rope with a pipette before freezing the
object in a domestic freezer at -23°C. The intent of the
freezing was to achieve slow non-vacuum freeze-drying
of the cork and rope. The cork and rope remained stable
over the three-year drying period. After verifying that
the weight of the object was stable, the canteen was
consolidated and backed with Stabiltex – an open weave
polyester fabric – applied with a 10–30% (w/v) solution
of Paraloid B-48N in acetone (Figures 5 and 6).
Figure 6. Left: View of canteen after conservation treatment showing
remnants of hemp rope. Right: Spout and cork after stabilization.
Courtesy of Friends of the Hunley, Inc.
advantages include excellent wettability and chelating
properties of the ammonium hydroxide solution. Based
on a recent assessment, the canteen is still stable after
eight years with no sign of active corrosion or alteration
of the organic materials. Important limitations do exist,
however, when using ammonium hydroxide on certain
materials. Wood may darken, and copper/copper alloys
or copper corrosion products will be attacked. The exact
impact of ammonium hydroxide on waterlogged wood
and other organic materials needs to be fully investigated
and should be part of a future study.
Case study: Apollo 11 fuel injector
Description and condition

The use of ammonium hydroxide has worked well for this
object and while the chloride extraction may be slower
at pH 11 than it would be at a higher pH, it represents
a good compromise when working on pH-sensitive
materials. Since ammonium hydroxide is a volatile
compound, it does not need to be rinsed out after treat-
ment like other chemicals. This is a real advantage when
working on fragile and composite porous materials. Other
In March 2013, 12.5 tons of Apollo-era (1969–72) Saturn V
rocket engine parts from several Apollo missions were
recovered off the coast of Florida at a depth of approxi-
mately 4,300 meters (Mardikian et al. 2015). Among the
recovered parts, there was a fuel injector belonging to
engine number 5 from Apollo 11 (F1-2013-0005-2) made
from pure copper on one side (Figure 7) and chrome-
plated 3xx series stainless steel on the other (Figure 8);
350  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
both materials were pressed together during fabrication
and could not be separated for treatment. Also, a large
detachable fluoropolymer elastomeric Viton O-ring gasket,
not shown on this image, is part of the object.
Figure 7. Before-conservation view of outer side of fuel injector
belonging to engine number 5 from Apollo 11. Note deformed copper
rings and baffles and extent of the copper corrosion products. The scale
on top of the fuel injector is one meter long
The critical role of the fuel injector needs to be explained
in order to understand the conservation challenge posed
by this object. The fuel injector’s main function was to
direct the propellants (liquid oxygen and refined kero-
sene) into the thrust chamber at the proper mixture
ratio, pressure and spray pattern via 2,832 orifices to
ensure combustion stability and start the engine ignition
sequence. The implication for conservation is that the
internal spaces forming the rings and copper baffles
represent a significant volume of the object and treat-
ment solutions have to reach these restricted areas to
ensure proper stabilization. Figure 7 shows the severely
Figure 8. Before-conservation view of inner side of fuel injector showing
chrome-plated layer present on top of the surface of the stainless steel
COMPOSITE OBJECTS
deformed copper baffles after engine number 5 hit the
surface of the ocean, and gives a good idea of the size of
the object, which is over a meter in diameter and weighs
approximately 544 kilograms. Voluminous green and
brown corrosion products and clay-like sediment cover
the surface of the copper.
Treatment
Work started with the cleaning of the fuel lines and
liquid oxygen orifices to remove the hard sediment using
a pressure sewer drain cleaning hose. This technique
worked well to remove the sediment and soft corrosion
products, but was unable to eliminate the hard copper
corrosion products plugging the orifices on the rings
and baffles and thus preventing access to the treatment
solutions. Analysis of the copper corrosion products
using Raman spectroscopy and x-ray diffraction by
McCrone Associates, Inc. confirmed the presence of
atacamite (Cu2Cl(OH)3 and paratacamite Cu3(OH)6Cl2.
This finding prompted a more interventive treatment to
chemically dissolve the copper corrosion products and
arrest corrosion.
To selectively remove the copper corrosion products
without affecting the other materials present, a 0.1 M
solution of potassium sodium tartrate (also known as
alkaline Rochelle salts, KNaC4H4O64H2O) was prepared
by adding 3.9 g/L of sodium hydroxide and 28.2 g/L of
potassium sodium tartrate. It was anticipated that a short
24-hour treatment at a pH of 12.81 would also boost the
chloride extraction from the stainless steel. The injector
was placed with the copper side facing up and resting
on nylon shims to elevate it from the bottom of the tank
and facilitate the circulation of the solution. The injector
was pulled out of treatment several times using an over-
head crane and examined. The solution quickly turned
blue from the copper tartrate complex and the majority
of the heavy green corrosion was removed. Vigorous
brushing and repeated pressure washing with deionized
water improved the cleaning and helped dislodge the
solubilized corrosion products trapped inside the copper
baffles. After rinsing in warm deionized water for several
days, the injector was placed in a 1% (w/v) solution of
sodium sesquicarbonate (Na3H(Co3)2 for approximately
two months while the chloride level was monitored. The
solution showed no increase of chloride ions into the
solution, indicating that the alkaline Rochelle salts had
performed exactly as planned. The injector was then
CHALLENGES AND SOLUTIONS FOR CONSERVING NON-SEPARABLE MARINE ARCHAEOLOGICAL COMPOSITE OBJECTS
rinsed in recirculated deionized water at approximately
40°C for two months to remove all traces of the chemicals.
The pH and conductivity of the water were monitored
until it reached that of the deionized water. After rinsing,
it was decided to boost the protection of the interior of
the baffles and rings by immersing the fuel injector in
a 5% (v/v) solution of VpCI-377 (a water-based, multi-
metal corrosion inhibitor made by Cortec Corporation)
and leaving it to drain and dry thoroughly at room
temperature. Finally, both sides of the exterior surfaces
of the object were coated with a 5% (w/v) solution of
Paraloid B-48N in toluene applied by spray and the Viton
O-ring gasket (not visible on images) was repositioned on
the object after mounting on a stand (Figures 9 and 10).
Figure 9. After-conservation view of outer side of fuel injector and
mounting on display stand
Figure 10. After-conservation view of inner side of fuel injector
351
While it is too early to say if this artifact will remain
stable in the long term since treatment was completed
in 2015, it is believed that this treatment has dramati-
cally enhanced its stability and overall appearance. The
Rochelle salts solution not only provided a way to clean
the object successfully, it also greatly assisted in removing
the entrapped chloride ions from the copper and stain-
less-steel components. This artifact is now part of the
permanent collection of the Smithsonian’s National Air
and Space Museum.
Conclusion
The three case studies presented here demonstrate that
the conservation of non-separable composites often
requires going off the beaten path to research the most
inclusive and less risky conservation strategies. Since
there is often no textbook to follow, the conservation of
these objects is usually a one-off and a bit experimental.
Defining the right treatment parameters – particularly
the permissible pH window and corrosion potential when
using electrochemical techniques – as well as the correct
treatment sequence is paramount.
While there has been much progress and more published
work in the conservation literature over the last 20 years,
conservators should go back to their treatment records to
evaluate their performance and share their observations
with the conservation community.
Acknowledgements
The author would like to thank Mary Ballard, Claudia
Chemello, Kellen Correia, Christian Degrigny, Ian
MacLeod, John Moore, Kate Morrand, Johanna Rivera,
Elsa Sangouard, Lyndsie Selwyn and Emily Williams.
References
Argyropoulos, V., C. Degrigny, and E. Guilminot. 2000.
Monitoring treatments of waterlogged iron-wood
composite artifacts using Hostacor IT – PEG 400. Studies
in Conservation 45: 253–64.
Britton, S.C. and D.G. Michael. 1955. The corrosion of
tin by aqueous solutions of ammonia. Journal of Applied
Chemistry 5(1): 1–9.
Bruhin, S., E. Hildbrand, E. Sangouard, and J. Schramm.
2017. Protection of organic remains in alkaline iron desal-
ination. In ICOM-CC 18th Triennial Conference Preprints,
352  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Copenhagen, 4–8 September 2017, ed. J. Bridgland, art.
0801. Paris: International Council of Museums.
Cook, C., A. Dietrich, D. Grattan, and N. Adair. 1984.
Experiments with aqueous treatments for waterlogged
wood-metal objects. In Proceedings of the 2nd ICOM
Waterlogged Wood Working Group Conference, 28–31
August 1984, 147–59. Grenoble: Centre d’Étude et de
Traitement des Bois Gorgés d’Eau.
Degrigny, C. 1995. Stabilisation de moteurs d’avions
immerges. Studies in Conservation 40: 10–18.
Degrigny, C., G. Baron, P. Christodoulou, K. Tran, and
X. Hiron. 2002. Conservation of a collection of water-
logged composite rifles dating from the 17th century
recovered from the Brescou II marine site. In Proceedings
of the 8th ICOM-CC Wet Organic Archaeological Mate-
rials Working Group Conference, Stockholm, 11–15 June
2001, ed. P. Hoffmann et al., 399–411. Bremerhaven,
Germany: Deutsches Schiffahrtsmuseum.
Florian, M.-L. 1987. Deterioration of organic materials
other than wood. In Conservation of marine archaeo-
logical objects, ed. C. Pearson, 21–54. Waltham, MA:
Butterworth-Heinemann.
Gilberg, M.R and N.J. Seeley. 1982. Liquid ammonia as
a solvent and reagent in conservation. Studies in Conser-
vation 27(1): 38–44.
Hawley, J.K. 1989. The conservation of waterlogged
rope from a sixteenth century Basque whaling ship.
In Conservation of Wet Wood and Metal: Proceedings
of the ICOM-CC Wet Organic Archaeological Materials
Working Group Conference, Fremantle, Australia, 1987,
ed. I.D. MacLeod, 19–37. Perth, WA: Western Australian
Museum.
Leidheiser, H. 1971. Aqueous corrosion. In The corrosion
of copper, tin, and their alloys, 282–302. New York: John
Wiley & Sons, Inc.
MacLeod, I.D. and N. North. 1982. Conservation of a
composite cannon Batavia (1629). The International
Journal of Nautical Archaeology 11(3): 213–19.
MacLeod, I.D. 1983. Stabilization of corroded aluminium.
Studies in Conservation 28(1): 1–7.
MacLeod, I.D., ed. 1989. Conservation of Wet Wood
and Metal: Proceedings of the ICOM-CC Wet Organic
Archaeological Materials Working Group Confer-
ence, Fremantle, Australia, 1987. Perth, WA: Western
Australian Museum.
COMPOSITE OBJECTS
MacLeod, I.D., P. Mardikian, and V. Richards. 1993.
Observations of the extraction of iron and chloride
ions from composite materials. In Proceedings of the 5th
Interim Meeting of the ICOM-CC Wet Organic Archaeo-
logical Materials Working Group, Portland, Maine, 16-20
August 1993, 199–211.
Mardikian, P. 1997a. Composite gorgés d’eau: technique
et éthique. Conservation-Restauration des Biens Culturels
9: 31–8.
Mardikian, P. 1997b. The conservation of two composite
objects from the Confederate raider Alabama (1864). In
Underwater Archaeology: Proceedings from the Society for
Historical Archaeology Conference, 8–12 January 1997,
ed. D.C. Lakey, 128–34. Corpus Christi. Texas, USA: The
Society for Historical Archaeology.
Mardikian, P. 2005. Unpublished treatment records and
lab notes for canteen HL-2159 and lantern HL-1957. H.L
Hunley database, Warren Lasch Conservation Center,
North Charleston, SC, USA.
Mardikian, P., C. Chemello, and J. Alexander. 2015. One
small step for man, one giant leap for conservation. The
American Institute for Conservation of Historic & Artistic
Works, Objects Specialty Group Postprints, Volume 22, eds.
E. Hamilton and K. Dodson, 123–50. Washington, D.C.:
The American Institute for Conservation of Historic &
Artistic Works.
Memet, J.-B. and K. Tran. 2005. Development of a conser-
vation treatment process adapted to archaeological iron/
waterlogged wood composites. In Proceedings of the 9th
ICOM Group on Wet Organic Archaeological Materials
Conference, Copenhagen, 2004, eds. P. Hoffmann et al.,
437–60. Bremerhaven: Verlag H.M. Hauschild for the
International Council of Museums.
Morrand, K. 2018. Lead Conservator, Naval History &
Heritage Command, Underwater Archaeology Branch,
personal communication, July 2018.
Murphy, T.P. 2006. Tin and tinplate. In Uhlig’s corrosion
handbook, ed. R. Winston Revie, 853–62. Hoboken, NJ:
John Wiley & Sons, Inc.
Scott, D.A. 2002. Copper and bronze in art: Corrosion,
colorants, conservation. Los Angeles, CA: Getty Conser-
vation Institute.
Selwyn, L., D.A. Rennie-Bisaillion, and N.E Binnie. 1993.
Metal corrosion rates in aqueous treatment for waterlogged
wood-metal composites. Studies in Conservation 38: 180–97.
CHALLENGES AND SOLUTIONS FOR CONSERVING NON-SEPARABLE MARINE ARCHAEOLOGICAL COMPOSITE OBJECTS 353

Selwyn, L. 2018. Senior Conservation Scientist, Cana-

dian Conservation Institute, personal communication,
October 2018.
Strang, T. 1983. Those bothersome composites: Wood
and iron. Ph.D. dissertation, M.A.C Seminar, Queen’s
University, Ottawa, Ontario, Canada.

Author
Paul Mardikian is co-founder and senior conservator
of Terra Mare Conservation, LLC. He has graduate
degrees in art history and archaeology from the School
of the Louvre and a graduate degree in conservation
from the Sorbonne University, Paris. Paul is a fellow of
the American Institute for Conservation.
Conservation Treatment and Stabilization of Marine
Composite Medieval Swords from Rhodes, Greece, Using
Ammonium Hydroxide Solutions
Vasilike Argyropoulos* Alexandra-Astrid Papanikou
University of West Attica University of West Attica
Department of Conservation of Department of Conservation of
Antiquities and Works of Art Antiquities and Works of Art
Egaleo, Greece Egaleo, Greece
bessie@uniwa.gr astrid.papanikou@yahoo.com
Abstract
This paper describes the successful application of a 1 M
ammonium hydroxide solution (pH 11.6–11.7) to remove
chlorides from iron without damaging the associated
mineralized organic materials. This treatment was carried out on
four medieval marine iron composite swords with their wooden
scabbards as well as on a dagger, all containing iron, tin and
copper inlays, wood, and textile and leather pseudomorph
imprints. A detailed scientific examination was carried out to
understand the construction technology of the artifacts as well
as their condition. X-rays revealed that a significant iron core was
present in all the swords, including the dagger. Chemical analysis
determined chlorine (Cl) and sulfur (S) on the surfaces of the
concretions in contact with the original surface of the objects,
with Cl and traces of S present on the iron parts and S present in
Introduction
This paper presents the results of chloride extraction
treatments for marine composite artifacts (iron-organics)
using a 1 M ammonium hydroxide1 (NH4OH) solution
in deionized water. The solution pH of 11.6–11.7 is
believed to be less damaging on the mineralized layers of
iron-wood artifacts, and may also be more beneficial in
treating mineralized organics such as wood. To the best
knowledge of the authors, there has been no published
work describing the stabilization of iron-wood composite
artifacts in ammonium hydroxide solutions. Research by
industry has shown that the ammonia/water system acts
as a corrosion inhibitor for mild steel using weight-loss
measurements at a pH of 10.5–10.9 (Shuaib and Kareem
1997), and was patented for such use between a pH range
of 10.5–11.8 (Erickson 1999). Water bound in wood can
be displaced by ammonia through ammonia sorption
(Bariska and Schuerch 1977). In conservation, there has
been research on the application of liquid ammonia in
treating iron artifacts with chlorides (Gilberg and Seeley
1982) and of ammonia vapor as a neutralizing treatment
Paul Mardikian
Terra Mare Conservation LLC
Charleston SC, USA
paul@terramareconservation.com
*Author for correspondence
the mineralized organic remains. The treatment was first tested
on one of the swords found in parts, and chloride removal was
successfully completed after one to two years of immersion in
this solution, with little or no damage to the mineralized layers
of the iron or wood. Currently, the remaining swords and dagger
have undergone a year-long treatment in this solution. This
paper provides details on the monitoring of these treatments,
which have proved to be very effective at removing chlorides
with no dissolution of the mineralized layers on the organic
pseudomorphs.
Keywords
medieval swords, iron-wood, marine composite artifacts,
chloride removal, ammonium hydroxide solution
for polyethylene glycol (PEG)-treated marine wood
(MacLeod and Kenna 1991).
This method of chloride extraction was carried out on
four medieval marine iron-wood composite swords and
a dagger discovered in the port of Rhodes in Greece. The
treatment dilemmas and diagnosis describing the tech-
nological features and condition of each sword and the
dagger are detailed elsewhere (Argyropoulos et al. 2018).
Separation of the different components of the swords
and the dagger for individual treatment was not possible
due to the risk of damage to the objects. Stabilization of
the marine iron artifacts was tested using electrolysis in
a neutral conductive solution (potassium nitrate), but
was found to be very slow in removing chlorides from
the iron and was not considered an option for these
artifacts, especially when the iron blade is encased in
the wooden scabbard. Thus, immersion in a chemical
solution that would passivate the iron, be effective at
removing chlorides, and not alter and/or destroy the
CONSERVATION TREATMENT AND STABILIZATION OF MARINE COMPOSITE MEDIEVAL SWORDS FROM RHODES, GREECE, USING AMMONIUM HYDROXIDE SOLUTIONS
associated organic materials was researched. This paper
describes the effective iron stabilization of one of the
swords treated in fragments after a two-year immersion
in NH4OH, and the ongoing treatment for the remaining
swords and dagger using this method.
Background and condition of the swords
The swords (objects no. 16, 17, 18, 1506, and the dagger
part of no. 17) are approximately 1 m long overall and
were found either near a 13th-century shipwreck or as
scattered remains in the port of Rhodes, buried at a depth
of 8–15 m underneath sea sediment in anoxic conditions
between 2007 and 2013 (Figure 1). In 2014, the swords,
which had been stored in tap water, were brought in a
wet state to the University of West Attica for analysis
and conservation. Although their archaeological study
is not complete, it is believed that they may belong to the
Knights of Rhodes (the Hospitallers) under the Order of
St. John.2 There has been extensive study on the typology
Figure 1. Dr. G. Koutsouflakis from the Hellenic Ephorate of Underwater
Antiquities holding sword no. 17 with the dagger at its underwater
location in Rhodes, Greece
(Oakeshott 1991 and 1997) and archaeometallurgy
(Williams 2012) of medieval swords, and based on the
scientific examination of the technological features of the
four swords in question (Argyropoulos et al. 2018), the
following can be concluded: each sword has a one-hand
grip, given the grip length, with a double-edged blade
covered by a wooden scabbard, at the end of which is
attached a chappe (of leather) to prevent it from wearing
away during use. Three of the blades (swords no. 16, 17,
and 1506) have a long tang extension to which the hilt
(grip, pommel, and guard) is fitted, and all have wood to
help with the grip, with evidence of textile or mineralized
355
textile. The other sword (no. 18) has a tongue instead of a
tang (a narrow piece at the top) with no organic materials
remaining on its grip. Each sword has a pommel with
a bore hole through it attached to the tang or tongue,
and the pommel of sword no. 1506 is decorated with
an eight-pointed cross also known as the Maltese cross.
Also, the blade of sword no. 1506 had evidence of pattern
welding in the corrosion products prior to treatment.
Finally, sword no. 16 has evidence of leather on top of the
wooden scabbard, and all the swords have metal lockets,
rings, etc., made of either copper or tin associated with
the scabbard. The dagger originally attached to sword no.
17 has a blade length of approximately 30 cm; the blade
is tapered and double-edged, with a crescent-shaped
pommel and a short straight cross. The metal decoration
on the sides of the dagger’s grip was found to contain Au,
Ag, and Hg, with the base metal being Cu.
The condition of each sword and dagger was examined
non-invasively, before and during cleaning, using radiog-
raphy and tomography. Also, samples recovered during
the process of removing concretions were analyzed using
scanning electron microscopy with energy-dispersive
x-ray spectroscopy (SEM-EDX) followed by either x-ray
diffraction (XRD) and/or Fourier-transform infrared
(FTIR) spectroscopy to determine the chemical and
mineralogical compositions as well as the presence of
organic remains. Argyropoulos et al. (2018) detail the
location of the samples taken from each sword and the
dagger, and present the results of these analyses and the
condition of the objects as summarized below:
1. The metal part of the swords contains iron-metal in the
blade, tang/tongue, and pommel, but there are parts in
the blades that are completely or partially mineralized.
The dagger only contains some iron-metal in the blade.
2. SEM-EDX analysis determined Cl and S on the
surfaces of the concretions in contact with the original
surface of the objects, with Cl and traces of S present
on the iron parts and S present in the mineralized
organic remains.
3. The main corrosion products of the iron analyzed
(after drying the samples) by XRD and FTIR were
magnetite, siderite, goethite, lepidocrocite, akaganeite,
and aragonite. However, it was difficult to identify the
types of iron sulfides and other iron/sulfur-containing
compounds using these scientific techniques.
4. For corrosion products in contact with organic mate-
rials, FTIR indicated the presence of polysaccharide
material associated with wood, and collagenous
356  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
material associated with the leather. The mineralized
leather was found to contain siderite and aragonite,
whereas the mineralized wood contained magnetite
and lepidocrocite. Their degree of mineralization was
determined by testing the hardness of organic surfaces
using a fruit penetrometer (Petrou and Pournou 2018).
The diagnosis concluded that all the organic materials in
the swords (wooden scabbards and grips, leather chappes
and leather detail on scabbard no. 16, and the textile
associated with the grip) were mineralized as opposed
to waterlogged due to their heavy contamination with
iron corrosion products and/or calcium carbonate. There
were traces of waterlogged or softer wood in some areas
of the scabbards and in textile remains on the grip. Also,
it is assumed that the degree of mineralization is the same
both on the surface and within the artifact. The dagger’s
scabbard was also completely mineralized and found to
contain polysaccharide material after analysis by FTIR.
Treatment approach for the composite swords
During the documentation/examination phase of the
project, the concretions from the swords and dagger
were removed with careful cleaning using a pneumatic
air scribe (Figure 2) with the aid of radiographs. After
concretion removal, the artifacts were maintained
using various mediums, such as Hostacor IT and poly-
acrylamide superabsorbent polymer gel (Aquaperla)
(Argyropoulos et al. 2017); however, a 1% (w/v) sodium
phosphate solution in deionized water was found to be the
most effective at passivating the iron prior to treatment.
Figure 2. The removal of concretion from the pommel of sword no. 17
using a pneumatic air scribe. The concretions are made of aragonite and
iron corrosion products
It was decided that PEG consolidation treatments are
not necessary since the organic parts of the objects are
fully impregnated with iron corrosion products and/or
COMPOSITE OBJECTS
calcium carbonate, and thus not preserved as waterlogged
tissue. As mentioned before, separation of the different
material components of the swords (no. 16, 17, and 18,
and dagger) for individual treatment was not possible due
to the risk of damage to the objects. However, sword no.
1506 was found in fragments, and its blade and wooden
scabbard, as well as other parts, were easily separated.
It was thus decided to first test chloride removal on this
sword using a 1 M ammonium hydroxide (NH4OH)
solution in deionized water before starting treatment on
other pieces of the remaining swords (Figure 3).
Figure 3. (Top) X-ray image of sword no. 1506 before concretion removal;
(middle) sword no. 1506 with concretions; (bottom) sword no. 1506
after concretion removal indicating the parts treated in ammonium
hydroxide solution: (1) iron pommel; (2) wooden grip containing iron
tang; (3) extensively mineralized iron blade; (4) tip of iron blade with
wooden scabbard in place and copper inlay in the wood
Prior to placing the piece with the copper inlay on the
tip of the scabbard in the NH4OH solution, the copper
part was covered with three layers of masking agents for
protection and to prevent it from corroding. The first
layer consisted of a Mowital B 30 H solution (polyvinyl
butyral, PVB) applied with a brush, followed immedi-
ately with a layer of liquid cyclododecane (with mineral
spirit) applied using an eyedropper, and then a final layer
of a two-part Xantopren VL Plus dental wax applied
using a spatula. Drying of the last two layers was almost
immediate.
Chloride results for sword no. 1506
During treatment of the four fragments from sword no.
1506 in a 1 M NH4OH solution, the pH dropped from
11.6–11.7 to 9.5–10.5 in about two to three weeks, after
which the solution was changed. This solution was quan-
titatively analyzed for chlorides using the Mohr titration
CONSERVATION TREATMENT AND STABILIZATION OF MARINE COMPOSITE MEDIEVAL SWORDS FROM RHODES, GREECE, USING AMMONIUM HYDROXIDE SOLUTIONS 357

method (North 1987). Figure 4 provides a graph showing

the total amount of Cl- ions found in the solution plotted
against the square root of time (t1/2) for the iron blade
over a one-year period. The S-shaped behavior indi-
cates the time delay needed for the metal surface to be
passivated by OH- ions and for the Cl- to diffuse through
the corrosion layer and enter the treatment solution as
previously described in the literature (Selwyn et al. 2001).
It is interesting to note that this time delay was observed
for marine iron removed from the wooden scabbard
and concretions encasing it as opposed to terrestrial
Figure 5. (Top) Blade of sword no. 1506 after 60 days of immersion in
iron with concretions, as previously published (Selwyn
ammonium hydroxide solution (June 2016); (middle) after treatment
and Argyropoulos 2005). The results also show that it (May 2017); (bottom) after restoration and storage (October 2018)
took around one year before the Cl- levels fell to a level
at which treatment could be stopped. Figure 5 shows
the before and after stabilization treatment of the blade,
including its restoration. Figure 6. (Top and
middle) Wooden
12000 grip of sword
no. 1506 after
10000
Chloride ions removed in ppm

30 and 400 days

8000 of immersion
in ammonium
6000
hydroxide solution,
4000 respectively;
(bottom) after
2000
drying out the
0 object without
0 5 10 15
Square root of time (days1/2)
20 25 rinsing the
ammonium
Figure 4. Graph for the 1506 iron blade in ammonium hydroxide solution
 hydroxide solution
over a one-year period showing the total amount of Cl- ions found in the
solution plotted against the square root of time (t1/2)

For the pommel, tip of the blade with wood and copper
inlays, and grip (wood and iron), the removal of chlo-
rides took around 1½ to 2 years of treatment in NH4OH,
after which period the treatment was stopped. Again, the
Cl- plots versus t1/2 showed an S-shaped behavior. The
longer treatment time was due to the substantial metal
core found in the pommel and in the iron between the
wooden grip and scabbard, as opposed to the extensively
mineralized blade of the same object. Figures 5–7 show
the before and after stabilization treatment for these
objects.

Rinsing and drying of sword no. 1506

After stabilization treatment, the final NH4OH solution
was not rinsed from any of the pieces, since it was believed
that it would evaporate during the drying procedure.
Figures 5 and 7 show the condition of the blade and
iron pommel after air-drying the artifacts without any
Figure 7. (Top) The pommel and grip of sword no. 1506 during storage in
sodium phosphate solution (June 2015); (bottom left) the same pommel
after immersion for 390 days in ammonium hydroxide solution (May
2017); (bottom right) pommel after treatment, dried without rinsing
the ammonium hydroxide solution (July 2018)
358  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

rinsing. This is an advantage of the NH4OH treatment

compared to other hydroxide ion treatments, where
rinsing is mandatory to avoid contamination of the
residual chemical in the artifact. Figure 8. Application of
masking agents to the
In the case of the drying wooden grip and iron tip with copper rings of sword
scabbard (including copper inlay), it was decided that, no. 16 for protection
despite being classified as “mineralized” wood, it would during the ammonium
hydroxide treatment: a
be safer to carry out a solvent drying technique using
first layer of Mowital B 30
acetone as a dewatering solution. Solvent exchange H solution, a second layer
started progressively with 30% (v/v) and 60% (v/v) of cyclododecane, and a
acetone in deionized water, and gradually reached 100% final layer of a two-part
Xantopren VL Plus dental
acetone for a period of 24 hours at each stage. During this
wax applied using a
process, the pieces were wrapped in Flexifilm X-25346 spatula

and fastened with a nylon string to ensure that if a piece
became detached, its exact location could be determined. was no significant change in the amount of Cl- being
The mineralized wood did not suffer from cracking or released, or the solution pH dropped below 10.
warping after the drying process.
The surfaces of each object were monitored visually and
Finally, the stabilization treatment using the NH4OH with a penetrometer to check for hardness to determine
solution for the iron remaining on each of the fragments if there were any changes to the organics covered and
of sword no. 1506 was effective in removing chlorides uncovered with the masking agents. To date, no visual
since no active corrosion is visible today after one and changes or hardness of the mineralized surface have been
a half years of treatment and storage in an uncontrolled noticed. Figures 9 and 10 show the before and after results
environment. for sword no. 16 after around 300 days in NH4OH solu-
tions, which appear to be similar: the mineralized textile
Treatment of the swords and dagger pseudomorph on the iron tongue or handle (without
masks during immersion) and the mineralized leather
After the successful removal of chlorides using ammo-
surface and chappe between the copper rings (with masks
nium hydroxide for the different parts of sword no.
during immersion).
1506 containing metallic iron, it was decided to use this
method to stabilize the rest of the swords and the dagger.
Special containers of approximately 1.5 m in length were
constructed using PVC piping with lids for their treat-
ment. Two of the swords (no. 17 and 18) were placed
in one of these containers and sword no. 16 in another
immersed in 35 L and 10 L of the 1 M NH4OH solution,
respectively. The dagger was placed in a 5 L solution in a
polystyrene container with lid. The copper and tin inlays
or rings on each sword and copper base grip with gilding Figure 9. Grip of sword
no. 16 showing the detail
on the dagger were protected using the masking agents
of textile pseudomorphs:
previously described (Figure 8). (top) immediately after
concretion removal (June
To date, swords no. 16, 17, 18, and the dagger are contin- 2015); (middle) during
uing treatment in NH4OH solutions. Up to the date of immersion in sodium
this publication, the following total amounts of chlorides phosphate solution (April
2016); (bottom) after
had been removed from each object: swords no. 17 and
over a year in ammonium
18, after 462 days: 22,295 mg of Cl-; sword no. 16, after hydroxide solution without
297 days: 7090 mg of Cl-; and the dagger, after 87 days: masking agents (October
1925 mg of Cl-. The baths were changed either once there  2018)
CONSERVATION TREATMENT AND STABILIZATION OF MARINE COMPOSITE MEDIEVAL SWORDS FROM RHODES, GREECE, USING AMMONIUM HYDROXIDE SOLUTIONS
Figure 10. Details of the
chappe, copper rings, and
leather pseudomorph
of sword no. 16: (top)
immediately after the
concretion removal (June
2015); (middle) during
immersion in sodium
phosphate solution
(April 2016); (bottom)
after almost one year in
ammonium hydroxide
solution protected using
masking agents (October
2018)

The textile and leather pseudomorphs remain preserved
on the surfaces after over one year of immersion in
the solution. Also, there seems to be no change in the
appearance of the wood. The pH of the NH4OH solution,
ranging between 10 and 11.7, may play a role in mini-
mizing the dissolution of the iron and calcium contained
in the corrosion products that make up the mineralized
layers of these objects. Studies have shown that the
aragonite crystal, given its needle-like structure, is much
more difficult to dissolve than calcite crystals (Ryu et al.
2010). Furthermore, Pourbaix diagrams show that the
FeS phase ceases to be stable if more than ~ 0.05 mol/kg
of NaOH is added (pH above 12.5). The stability field of
the sulfide is followed by that of an aqueous species (i.e.,
Fe[OH]3–), which corresponds to corrosion in alkaline
environments (Anderko et al. 1997).
After completion of the chloride removal of the swords
and dagger, the drying protocols will need to be examined
more carefully using either non-vacuum freeze-drying or
air-drying methods. Given the size of the artifacts, dehydra-
tion with acetone will not be carried out for safety reasons.
Conclusion
A 1 M ammonium hydroxide (NH4OH) solution in
deionized water was shown to be effective at removing
359
chlorides from marine iron-wood composite artifacts
without damaging mineralized organic materials. For
the mineralized organics (wood, leather, and textile)
there appeared to be no visible damage or losses due to
immersion in this solution, even after long periods of
time. The treatment benefits of ammonium hydroxide
for the conservation of certain marine composite arti-
facts was demonstrated, but more research is needed
to assess its effect on various organic materials and
other metals.
Notes
1 When handling ammonium hydroxide solution,
protective clothing and masks should be worn since
contact of the skin with the liquid may cause burns and
its vapors can cause irritation to the respiratory tract
when inhaled.
2 The archaeologist responsible for the swords and
dagger is Dr. G. Koutsouflakis from the Hellenic Ephorate
of Underwater Antiquities in Athens.
References
Anderko, A., S.J. Sanders, and R.D. Young. 1997.
Real-solution stability diagrams: A thermodynamic
tool for modeling corrosion in wide temperature and
concentration ranges. Corrosion 53(1): 43–53.
Bariska, M. and C. Schuerch. 1977. Wood softening and
forming. In Wood technology: Chemical aspects, ed. I.
Goldstein, 327–46. Washington, DC: ACS Symposium
Series, American Chemical Society.
Argyropoulos, V., N. Koumparou, A. Bei, M. Giannou-
laki, A. Malea, S. Rapti, A. Pournou, S.C. Boyatzis, and
A. Tsompanidis. 2018. Tackling the difficulties in the
conservation treatment and stabilization of marine
composite medieval swords. In Proceedings of the 13th
ICOM-CC Group on Wet Organic Archaeological Materials
Conference, Florence, May 2016, eds. E. Williams and E.
Hocker, 270–81. Paris: ICOM.
Argyropoulos, V., A. Bei, and S.C. Boyatzis. 2017. The
testing of a potassium polyacrylate (water) gel as pretreat-
ment and storage of wet marine iron-organic composite
artefacts. In Gels in the conservation of art, eds. L.V.
Angelova, B. Ormsby, J.H. Townsend, and R. Wolbers,
267–69. London: Archetype Publications Ltd.
Erickson, D.C. 2001. Aqueous ammonia corrosion
inhibitor. US Patent No. 6,203,718 B1, 20 March 2001.
360  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Gilberg, M.R. and N.J. Seeley. 1982. Liquid ammonia as
a solvent and reagent in conservation. Studies in Conser-
vation 27(1): 38–44.
MacLeod, I.D. and C. Kenna. 1991. Degradation of
archaeological timbers by pyrite: Oxidation of iron and
sulphur species. In Proceedings of the 4th ICOM-CC Group
on Wet Organic Archaeological Materials Conference,
Bremerhaven, 1990, ed. P. Hoffman, 133–42. Paris: ICOM.
North, N.A. 1987. Chloride analysis methods. In Conser-
vation of marine archaeological methods, chap. 10,
Conservation of metals, ed. C. Pearson, 248–52. Toronto:
Buttersworth.
Oakeshott, E. 1991. Records of the medieval sword. Wood-
bridge: The Boydell Press.
Oakeshott, E. 1997. A knight and his weapons, 2nd edition.
Chester Springs: Dufour Editions Inc.
Petrou, M. and A. Pournou. 2018. Testing the efficiency
of a fruit penetrometer to assess the condition of small
waterlogged wooden artifacts. In Proceedings of the 13th
ICOM-CC Group on Wet Organic Archaeological Materials
Conference, Florence, May 2016, eds. E. Williams and E.
Hocker, 47–53. Paris: ICOM.
Ryu, M., H. Kim, M. Lim, K. You, and J. Ahn. 2010.
Comparison of dissolution and surface reactions between
calcite and aragonite in L-glutamic and L-aspartic acid
solutions. Molecules 15: 258–69.
Selwyn, L.S., W.R. McKinnon, and V. Argyropoulos. 2001.
Models for chloride ion diffusion in archaeological iron.
Studies in Conservation 46(2): 109–20.
Selwyn, L.S. and V. Argyropoulos. 2005. Removal of
chloride and iron ions from archaeological wrought iron
with sodium hydroxide and ethylenediamine solutions.
Studies in Conservation 50(2): 81–100.
Shuaib, Y.L.B. and Y.A. Kareem. February 1997. Corro-
sion inhibition of mild steel in ammonia/water solution,
pH 10.5–10.8. Project report, Mechanical Engineering
Department, University of Ilorin, Nigeria.
Williams, A. 2012. The sword and the crucible. Boston:
Brill.
Authors
Vasilike Argyropoulos is a professor of Metals Conser-
vation at the Department of Conservation of Antiq-
uities & Works of Art, University of West Attica. She
has conducted post-doctoral research in conservation
COMPOSITE OBJECTS
science related to treatments for iron artifacts, including
waterlogged iron-wood composite artifacts. She has
coordinated many European and international projects
related to conservation of antiquities in Europe and the
Arab region, namely in Egypt, Jordan, Tunisia, and Syria.
Her most recent national funded research project dealt
with research on the corrosion and in-situ monitoring
of iron shipwrecks in the Aegean Sea.
Alexandra-Astrid Papanikou is a final year student at
the Department of Conservation of Antiquities & Works
of Art, University of West Attica. Her main interest is in
conservation of marine artifacts. She has participated
in fieldwork projects in Crete and Turkey. Her under-
graduate dissertation focuses on the topic of this paper.
Paul Mardikian is co-founder and senior conservator
of Terra Mare Conservation, LLC, a firm specializing in
the conservation of cultural heritage artifacts. He has
graduate degrees in art history and archaeology from the
School of the Louvre and a graduate degree in conser-
vation from the Sorbonne University, Paris. Since 1988
he has led numerous conservation projects including the
RMS Titanic (1912), CSS Alabama (1864), H.L. Hunley
(1863) and Apollo F1 engine conservation project. He
has over 30 years of international experience in cultural
heritage preservation, and extensive experience with
complex conservation programs requiring a multidisci-
plinary approach and international collaboration. He is
a fellow of the American Institute for Conservation of
Historic and Artistic Works.

Stabilization of Archaeological Composites: A Complex
Challenge Requiring Compromise
Trine Wiinblad
National Museum of Denmark
Lyngby, Denmark
trine.wiinblad@natmus.dk
Abstract
Three archaeological iron-bone composites were treated
electrochemically in sodium sesquicarbonate (pH 10) for
two weeks at a fixed reduction potential. The stability of
the composites was investigated by measuring oxygen
consumption. The effect of the treatment on the organic material
was investigated by Fourier transform infrared spectroscopy.
Introduction
The conservation of archaeological composite objects
composed of both organic and inorganic materials
presents particular challenges since the different prop-
erties of the materials require different preservation
strategies. There is no ideal conservation treatment for
composite objects and even dismantling of an object
followed by separate treatments of each material often
leads to loss of organic material.
Aims
• To investigate whether potential-controlled elec-
trochemical rinsing (PCER) in a 0.03 M solution
of sodium sesquicarbonate (pH 10) can be used to
remove chlorides from archaeological composites
composed of iron and bone/antler.
• To investigate whether the oxygen consumption rate
(OCR) can be used to determine the stability of the
entire object.
Objectives
• Desalinate selected chloride-infested composite
objects at a fixed reduction potential using sodium
sesquicarbonate electrolyte.
• Measure the OCR of untreated objects against controls
comprising fresh organic exemplars and treated objects.
• Use Fourier transform infrared (FTIR) spectroscopy
to assess the structure of the organic component of
selected archaeological samples and fresh controls
before and after treatment.
Young Conservation Professional
Keywords
archaeological iron-bone composites, sodium sesquicarbonate,
potential-controlled electrochemical rinsing, oxygen
consumption, Fourier transform infrared spectroscopy
Materials
This study included ten untreated archaeological iron
knives (COM 1–10) with handles made of bone/antler/
ivory excavated at the Town Hall Square (Rådhuspladsen)
in Copenhagen in 2012–2013 and three archaeological
composite knives (REF 1–3) excavated at an urban site
in Copenhagen (Esplanaden) in 2003–2004 which had
been preserved through desalination by immersion in
0.03 M sodium sesquicarbonate for five years (Table 1).
Both sites were urban dumps from the 17th century
which were the sources of chlorinated metal objects.
Fresh cow bone and recent reindeer antler, along with
non-corroded and pre-corroded iron test coupons (both
Q-panel R46), were used as reference materials.
Methods
Voltammetric studies (VS)
Polarization curves were performed using a potentiostat
(Volta Lab PGZ 301 with Volta Master 4 software) to scan
the response of iron to hydrogen and oxygen evolution.
The working electrodes were new, non-corroded and
pre-corroded iron test coupons, and iron from archaeo-
logical composites (COM 1–2). Titanium mesh was used
as a counter electrode and mercury sulfate as a reference
electrode (SSE, 0.651 V/SHE). The polarization curves
were compared with Pourbaix diagrams calculated using
HSC Chemistry 5.1 software (Outokumpu).
362  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

Potential-controlled electrochemical rinsing Results and discussion

(PCER)
The reduction potential was chosen based on polarization
Three composite knives (COM 3–5) were treated for two curves. Figure 1 shows a detail of a voltammetric curve
weeks at a fixed reduction potential of -1.3 V(SSE) using for iron in the electrolyte.
the potentiostat with 0.03 M sodium sesquicarbonate
(pH 10) as the electrolyte. The amount of chloride in
the electrolyte was measured by ion chromatography
(Metrohm 881 Compact IC Pro with a compact autosam-
pler and conductivity detector, and detection limit of
0.5 ppm chloride and RSD of 0.7%). The initial concen-
tration of chloride in the solutions was under 0.5 ppm.
The electrolytes were not changed during the treatment
period. Samples (5 mL) taken out for analysis every other
day were not replaced with fresh electrolyte.

Oxygen consumption rate (OCR)

Before measuring the OCR, the samples were stabilized
for at least four weeks in a climate chamber conditioned Figure 1. Detail of a voltammetric curve for iron (non-corroded test
coupons) in 0.03 M sodium sesquicarbonate (scan rate = 10 mV/s,
to 50% relative humidity (RH) and 20°C. The composite
number of cycles: 3)
knives (COM 3–8 and REF 1–3), fresh cow bone, and
recent reindeer antler were individually sealed in bags A fixed reduction potential of -1.3 V(SSE), which corre-
made of flexible, transparent barrier film (Escal) with a sponds to -0.65 V(SHE), was selected. At this potential,
sensor spot (PreSens sensor foil) inside each bag along the iron is cathodically protected in an area where the
with a few grams of buffered silica gel to keep the RH evolution of hydrogen is theoretically minimal, the corro-
stable at 50%. A bag filled with air was used as a control. sion products will presumably be reduced, and chloride
The measurements were performed using optical oxygen ions can be leached (Figure 2).
meters (PreSens). All measurements were carried out
in a climate chamber conditioned to 50% RH and 20°C.
The OCR was calculated as described by Matthiesen and
Stemann-Petersen (2013).

Fourier transform infrared (FTIR) spectroscopy

Organic material taken from the three electrochemi-
cally treated objects (COM 3–5) was measured before
and after treatment and compared with the analysis of
organic material from two composites (COM 9–10),
fresh cow bone, and reindeer antler before and after
immersion in 0.03 M sodium sesquicarbonate for two
Figure 2. Pourbaix diagram showing the peaks measured by cyclic
months. voltammetry

The infrared spectra were obtained with a Perkin
Elmer FTIR spectrometer (Spectrum 1000, 4000 cm-1
to 400 cm-1) by taking small samples (bone powder) from
the objects to prepare 5 mm KBr pellets. The splitting
factor (SF) was calculated based on the extent of the
splitting between absorptions at 603 and 565 cm-1, as
described by Weiner and Bar-Yosef (1990).
During the two-week treatments, the amount of chlo-
ride in the electrolyte increased (Table 1), showing that
this method enables the removal of chlorides from the
objects.
However, the appearance of the objects changed.
Hydrogen evolution had a negative impact as the original
iron surface was lifted off in some areas. The organic parts
STABILIZATION OF ARCHAEOLOGICAL COMPOSITES: A COMPLEX CHALLENGE REQUIRING COMPROMISE 363

Table 1. An overview of materials, sample preparation, analytical methods and results

Initial air Oxygen

Analytical Weight SF
Materials Treat volume in Escal consumption SF before
methods (g) after
bag (cm3) (mg/day)
COM 1 UT VS
COM 2 UT VS
COM 3 Pot2 OCR 18.16 18.26 8.6E-03
PCER*
FTIR 2.34 2.43
COM 4 Pot2 OCR 45.25 17.46 8.9E-02
PCER**
FTIR 2.42 2.40
COM 5 Pot2 OCR 17.25 9.34 1.8E-02
PCER***
FTIR 2.31 2.41
COM 6 UT OCR 24.45 58.52 9.7E-02
COM 7 UT OCR 16.00 22.94 1.7E-02
COM 8 UT OCR 26.80 67.66 undetectable
COM 9 NaSes2 FTIR 2.35 2.32
COM 10 NaSes2 FTIR 2.39 2.38
REF 1 NaSes5 OCR 36.94 14.78 1.1E-02
REF 2 NaSes5 OCR 78.58 58.52 1.2E-02
REF 3 NaSes5 OCR 96.64 59.43 4.6E-03
non-corr. iron test Control VS
coupons
pre-corr. iron test Control VS
coupons
cow bone A UT OCR 30.74 24.03 2.9E-03
cow bone B UT OCR 43.03 61.76 8.1E-03
cow bone C NaSes2 FTIR 2.39 2.45
cow bone D NaSes2 FTIR 2.33 2.31
reindeer antler A UT OCR 6.06 13.32 2.5E-05
reindeer antler B UT OCR 10.75 15.62 undetectable
reindeer antler C NaSes2 FTIR 2.28 2.35
reindeer antler D NaSes2 FTIR 2.29 2.32
PCER* = COM 3: Final chloride measurement in the electrolyte after two weeks PCER: 4.71 ppm
PCER** = COM 4: Final chloride measurement in the electrolyte after two weeks PCER: 18.25 ppm
PCER *** = COM 5: Final chloride measurement in the electrolyte after two weeks PCER: 1.19 ppm
UT = Untreated
Pot2 = 2 weeks at a fixed reduction potential of -1.3 V(SSE) in 0.03 M sodium sesquicarbonate (pH 10)
NaSes5 = Treated in 0.03 M sodium sesquicarbonate (pH 10) for 5 years
NaSes2 = Treated in 0.03 M sodium sesquicarbonate (pH 10) for 2 months

of the composites were also affected, leaving the surface
less shiny and more fragile. Analysis of the polarization
curves (Figure 1) showed that the electrochemical treat-
ment could be conducted at a fixed reduction potential
of -1.20 to -1.25 V(SSE) (= -0.55 to -0.60 V(SHE)) to
further minimize hydrogen evolution while still reducing
the corrosion products (Figure 2) and allowing chloride
extraction.
The stability of the composites was investigated by
measuring the OCR (Table 1). The OCR of bone and
antler was negligible in relation to that of the composites
before treatment (COM 3–8). The oxygen consumption
of REF 1–3 was only measured after desalination as a
comparison for the level expected after desalination of
a composite object. To compare the different objects, the
OCR was related to the weight of the individual objects
364  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
(µg O2/day/gram object). Although the results indicate
that the OCRs are on average lower for the desalinated
composites (REF 1–3) than for the composites before
conservation (COM 3–8) (Figure 3), the number of
samples is small. The results show, however, that meas-
uring the OCR is a useful method for determining the
stability of composites which can be used to compare the
stability of the same object before and after conservation.
Figure 3. The OCR for the reference materials and the composites (COM
3–8) before treatment
The effect of the alkaline electrolyte on the organic
material was investigated by analyzing the SF measured
by FTIR (Table 1). The SF is commonly used to deter-
mine the general state of preservation of ancient bone,
antler, and ivory (Weiner 2010). As protein (collagen) is
removed, the remaining inorganic structure (hydroxyapa-
tite) is rearranged, resulting in an increase in the SF
value. The SF indicates the degree of crystallinity of the
bone and thus, indirectly, the amount of organic material
retained (ibid.). Evaluation of the SF before and after
treatment indicated small changes in the chemical struc-
ture when the objects were exposed to 0.03 M sodium
sesquicarbonate. On average, the SF values increased
after exposure to the alkaline solution, which indicates
that protein was removed from the bone and antler. On
average, samples exposed to PCER saw their SF values
increase by 2.4% compared to the same material before
treatment. On average, the SF values for the samples
exposed to sodium sesquicarbonate for two months only
increased by 0.7%.
COMPOSITE OBJECTS
Conclusion
PCER can be used to remove chloride from archaeolog-
ical composites using 0.03 M sodium sesquicarbonate
(pH 10) as the electrolyte. The chosen reduction potential
of -1.3 V(SSE) had a somewhat negative effect on the
integrity of the objects due to hydrogen evolution during
treatment. Future studies should show if the use of PCER
at a slightly higher potential could be beneficial. Based
on experiments with oxygen consumption, it can be
concluded that this method can be used to measure the
OCR for composites, but further experiments should be
conducted before the stability of the objects is assessed
based on the consumption of oxygen alone.
The organic material of the composites was somewhat
affected by 0.03 M sodium sesquicarbonate. The largest
change was detected for the composites that had been
treated with PCER. Treating archaeological composite
objects with 0.03 M sodium sesquicarbonate may cause
damage to their organic components, but stabilization
of this type of object requires compromises.
References
Matthiesen, H. and K. Stemann-Petersen. 2013. A fast
and non-destructive method to document and quantify
the efficiency of metal conservation. In Metal 2013:
Proceedings of the Interim Meeting of the ICOM-CC Metals
Working Group, Edinburgh, 16–20 September 2013, eds. E.
Hyslop, V. Gonzalez, L. Troalen, and L. Wilson, 179–84.
Edinburgh: Historic Scotland and International Council
of Museums.
Weiner, S. 2010. Microarchaeology: Beyond the visible
archaeological record. Cambridge, MA: Cambridge
University Press.
Weiner, S. and O. Bar-Yosef. 1990. States of preservation
of bones from prehistoric sites in the Near East: A survey.
Journal of Archaeological Science 17(2): 187–96.
Author
Trine Wiinblad has a B.Sc. and an M.Sc. in conservation
and restoration from The Royal Danish Academy of Fine
Arts, Schools of Architecture, Design and Conservation
(KADK). She worked on the conservation of different
archaeological objects/materials at an independent
conservation workshop in Denmark and since 2010 has
worked as a conservator of archaeological metal at the
National Museum of Denmark.

Glass-Induced Metal Corrosion: Study and Conservation of
an Enamelled Altarpiece (1954–56) from the Collection of
the University of Applied Arts Vienna
Maria Holzleitner* Meral Hietz
University of Applied Arts Vienna University of Applied Arts Vienna
Institute of Conservation Institute of Conservation
Vienna, Austria Vienna, Austria
maria@holzleitner.com meral.hietz@gmail.com
konservierung-restaurierung.uni-ak.ac.at/
Eva Lenhart
*Author for correspondence University of Applied Arts Vienna
Institute of Conservation
Vienna, Austria
eva.putzgruber@uni-ak.ac.at
Abstract
During the transfer of the collection of the University of
Applied Arts Vienna to the new open storage facility in 2015,
an enamelled altarpiece was noted for its quality as well as for
its state of conservation. The winged altarpiece, composed
of painted wood, gilded copper and enamel, presented
greenish-blue efflorescence mainly at the interface between
the opaque blue enamel and the gilded copper. In addition, the
artwork was affected by different corrosion products. In order
to characterise the materials and identify the degradation
products and their possible causes, scanning electron
microscopy with energy-dispersive x-ray spectrometry, Fourier
Introduction
The investigated object is a winged altarpiece with a
painted wooden frame and enamel decoration on copper
with precious gilding. It was created at the University of
Applied Arts Vienna by Erich Hamann, during a master-
class on enamelwork delivered between 1954 and 1956,
and is therefore a document of the history of the univer-
Figure 1. Enamelled altarpiece, Erich Hamann, 1954–56, collection of the
University of Applied Arts Vienna. Inside of the altar before treatment
Young Conservation Professional
Marta Anghelone Gabriela Krist
University of Applied Arts Vienna University of Applied Arts Vienna
Institute of Conservation Institute of Conservation
Vienna, Austria Vienna, Austria
Marta.anghelone@uni-ak.ac.at gabriela.krist@uni-ak.ac.at
transform infrared spectroscopy and x-ray diffraction analyses
were carried out. The results clarified the enamel composition
and detected the presence of different organic copper salts in
the efflorescence, suggesting the occurrence of glass-induced
metal corrosion. Finally, a conservation strategy was developed
and applied that employed an EDTA rigid gel to remove the
corrosion products.
Keywords
glass-induced metal corrosion, analysis, gilded copper, enamel,
rigid gel, ethylenediaminetetraacetic acid (EDTA)
sity and the craftsmanship of its students. The inside of
the altar is dominated by a large crucifix, surrounded by
32 plates which depict scenes from the Old and the New
Testament (Figure 1). Today, the altarpiece is part of the
University’s collection. Until 2015, the object was stored
in inappropriate climatic conditions, with high humidity
and in direct contact with acidic storage materials which
led to the formation of corrosion products on the metal
surface. In 2017, it was transferred to the Institute of
Conservation for investigation and conservation (Hietz
2018, Holzleitner 2018).
Technology
The artist combined three fundamental techniques of
enamel on copper. The two plates on the external side
of the altar are made of cloisonné enamel. The figure of
Christ is worked in ronde-bosse enamel, while the plates
on the inside are made of champlevé enamel (Brepohl
1992). After the enamelling process, the plates were
366  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
hand-polished and gilded using the galvanic method.
The main enamel used was an opaque blue. Analyses
by scanning electron microscopy with energy-dispersive
x-ray spectrometry (SEM-EDS), applied directly on the
enamel plates and under atmospheric conditions, clarified
the composition of the enamel – it is based on silicon
dioxide and lead oxide with the addition of potassium
and sodium oxides as fluxing agents. In addition, different
colouring agents such as, presumably, cobalt(II) oxide
and chrome(III) oxide were identified.
Condition
During the initial examination, a variety of different
corrosion products were found on the enamelled plates.
Most parts of the gilded copper were extensively covered
with whitish corrosion products. Fourier transform
infrared (FTIR) spectroscopy in reflection mode was used
directly on the object and showed the presence of copper
carboxylates (Figure 2a). In addition, nearly all interfaces
between the opaque blue enamel and the gilded copper
presented distinctive greenish-blue efflorescence which
affected the readability and appearance of the object
(Figure 3). X-ray diffraction (XRD) and attenuated total
reflectance (ATR)-FTIR were used to characterise the
efflorescence. The XRD results revealed the presence of
Figure 2. FTIR spectra of (a) copper carboxylate and (b) copper trihydroxy
formate
Figure 3. Detail of greenish-blue efflorescence at the interface between
the blue enamel and the gilded copper
COMPOSITE OBJECTS
copper(I) oxide, copper(II) carbonate hydroxide hydrate
and sodium copper formate. The FTIR spectra also iden-
tified copper(II) trihydroxy formate (Figure 2b) (Veiga et
al. 2016). Such characteristic corrosion products indicate
the presence of glass-induced metal corrosion (Fischer
2016), which typically occurs at the interface between
leached glass and copper or copper alloys in the presence
of organic acids and moisture from the environment
(Eggert and Fischer 2012, Fischer et al. 2018).
Treatment
The priority of the conservation work was to stabilise the
object’s condition and to achieve an aesthetically pleasing
appearance for future display. The enamelled metal plates
underwent dry cleaning followed by wet cleaning with
deionised water. In order to remove the corrosion prod-
ucts, mechanical methods were tested (Lafitte et al. 2018)
but ruled out to protect the delicate gilding. Removal with
ethylenediaminetetraacetic acid (EDTA) was preferred
instead (ibid.). Several concentrations and methods of
application were studied by analysing the surface before
and after treatment using SEM-EDS, FTIR and conduc-
tivity measurements. By evaluating the applicability and
the effect of the treatments on the metal as well as on
the enamel, the use of a rigid gel was selected in order to
prevent the solution from penetrating the material. Thus,
agar-agar was added to a one-percent solution of EDTA
disodium salt in deionised water and the resulting rigid gel
was cut in shape and applied on the corroded metal areas
(Figure 4). The treatment was followed by a comprehensive
cleaning with deionised water (Figure 5). For long-term
storage, an acid-free cardboard support was prepared.
In the collection’s new open storage facility, the altar will
be presented in a regulated climate of 40%–45% relative
humidity (Barbe and Le Roux 2018).
Figure 4. Removal of corrosion products with rigid gel
GLASS-INDUCED METAL CORROSION: STUDY AND CONSERVATION OF AN ENAMELLED ALTARPIECE (1954–56) FROM THE COLLECTION OF THE UNIVERSITY OF APPLIED ARTS VIENNA
Figure 5. Detail after treatment with rigid gel
Conclusion
Different corrosion products were formed on the
surface of the enamelled altarpiece mainly due to inap-
propriate climatic conditions and storage materials. A
multi-analytical approach identified not only typical
copper corrosion products, but also sodium copper
formate and copper(II) trihydroxy formate, which
are characteristic of a glass-induced metal corrosion
process. To remove the corrosion, an EDTA disodium
solution was applied as an agar-agar rigid gel. The
conservation treatment was successful and enhanced the
condition and the readability of the object (Figure 6).
To prevent further corrosion, preventive conservation
measures and improvement of the storage conditions
were undertaken.
Figure 6. Working photograph of the inside of the altar: (left half) after
treatment, (right half) before treatment
Acknowledgements
The authors would like to thank Silvia Herkt, Kathrin
Schmidt and Hannelore Karl for their support, Rudolf
Erlach for the SEM-EDS analyses and Erich Halwax for
the XRD analyses.
367
References
Barbe, F. and J. Le Roux. 2018. A project to improve an
existing showcase for conservation and presentation
of the Limoges early painted enamels at the Louvre
Museum. In Enamel 2018, Vol. I: Extended Abstracts,
Stuttgart, 7–9 June 2018, ed. G. Eggert, 26–31.
Brepohl, E. 1992. Werkstattbuch Emaillieren. Technik und
künstlerische Gestaltung. Augsburg: Augustus Verlag.
Eggert, G. and A. Fischer. 2012. Gefährliche Nach-
barschaft. Durch Glas induzierte Metallkorrosion an
Museums-Exponaten – Das GIMME-Projekt. Restauro
1: 38–43.
Fischer, A. 2016. Glasinduzierte Metallkorrosion. Ph.D.
dissertation, Academy of Fine Arts Stuttgart, Germany.
Fischer, A., G. Eggert, R. Dinnebier, and T. Runčevski.
2018. When glass and metal corrode together, V: Sodium
copper formate. Studies in Conservation 63(6): 342–55.
Hietz, M. 2018. Ein Emailaltar von Erich Hamann aus
der Wiener Kunstgewerbeschule. Zur Reduzierung
von Kupferkorrosionsprodukten mit EDTA-Agar-Agar
Gelen. Unpublished report, University of Applied Arts
Vienna, Austria.
Holzleitner, M. 2018. Ein Emailaltar aus der Wiener
Kunstgewerbeschule der 1950er Jahre. Forschungen und
Konzeptentwicklung für die Konservierung und Restau-
rierung mit Schwerpunkt auf glasinduzierte Metallko-
rrosion. Pre-diploma thesis, University of Applied Arts
Vienna, Austria.
Lafitte, A., A. Gall-Ortlik, A. Genachte-Le Bail, M.
Loeper-Attia, B. Quette, and M. Saheb-Djahromi. 2018.
Conservation of a Chinese cloisonné enamel from the
Museum des Arts décoratifs of Paris. In Enamel 2018,
Vol. I: Extended Abstracts, Stuttgart, 7–9 June 2018, ed.
G. Eggert, 82–89.
Veiga, A., D.M. Teixeira, A.J. Candeias, J. Mirão, P. Simões
Rodrigues, and J. Ginja Teixeira. 2016. On the chemical
signature and origin of dicoppertrihydroxyformate
(Cu2(OH)3HCOO) formed on copper miniatures of the
17th and 18th centuries. Microscopy and Microanalysis
22(5): 1007–17.
Materials list
Agar-agar (Vollkraft GmbH, Austria)
Titriplex III® (EDTA disodium salt) (Merck KGaA,
Austria)
368  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

Authors
Maria Holzleitner is a student at the Institute of
Conservation, University of Applied Arts Vienna, with
a focus on objects conservation. From March to June
2018 she was an intern at the scientific laboratory of the
Kunsthistorisches Museum Vienna.

Meral Hietz is a student at the Institute of Conservation,

University of Applied Arts Vienna, with a focus on objects
conservation, and also a freelance metal conservator.

Marta Anghelone is senior lecturer at the Institute of

Conservation, University of Applied Arts Vienna. She
completed her PhD in 2017 and was also a researcher
and lecturer at the Academy of Fine Arts Vienna.

Eva Lenhart is assistant professor at the Institute of

Conservation, University of Applied Arts Vienna, with
a focus on objects conservation. She completed her PhD
in 2016 at the University of Applied Arts Vienna.

Gabriela Krist is head of the Institute of Conservation

and a professor at the University of Applied Arts Vienna.
She studied conservation at the Academy of Fine Arts
Vienna as well as art history and archaeology at the
Universities of Vienna and Salzburg.
New Strategies for the Conservation of Paintings
on Metal

Julie Gordon Sigrid Mirabaud Mandana Saheb Ludovic Bellot-Gurlet

Synchrotron SOLEIL and MONARIS
(UMR 8233), Sorbonne Université, CNRS
Paris, France
julie.gordon@synchrotron-soleil.fr
Laura Normand
Synchrotron SOLEIL
Gif-sur-Yvette, France
lauranormand@hotmail.fr
Anne Genachte-Le Bail
Institut national du patrimoine
Paris, France
anne.genachte-lebail@inp.fr
*Author for correspondence
Institut national d’histoire de l’art LISA, Laboratoire Inter-Universitaire MONARIS (UMR 8233), Sorbonne
Paris, France des Systèmes Atmosphériques, Université, CNRS
sigrid.mirabaud@inha.fr Université Paris-Est Créteil et Université Paris, France
Paris Diderot, CNRS, UMR 7583 ludovic.bellot-gurlet@sorbonne-
Marie-Anne Loeper-Attia Créteil, France universite.fr
Institut national du patrimoine mandana.saheb@lisa.u-pec.fr
Paris, France Solenn Reguer*
loeperattia@orange.fr Anne-Marie Geffroy Synchrotron SOLEIL
Independent conservator Gif-sur-Yvette, France
Remi Catillon Paris, France solenn.reguer@synchrotron-soleil.fr
Musée des Arts et Métiers amgeffroy@yahoo.fr
Paris, France
remi.catillon@lecnam.net Celine Paris
MONARIS (UMR 8233), Sorbonne
Anne-Laure Carré Université, CNRS
Musée des Arts et Métiers Paris, France
Paris, France celine.paris@upmc.fr
anne-laure.carre@lecnam.net

Abstract
The conservation and restoration of painted metal objects
presents a challenge due to their composite nature. Both the
metal support and the paint layers can undergo independent
and linked alterations, making it necessary to fully understand
the processes of degradation before attempting to implement a
treatment. This paper presents the first part of the CoPaiM project,
detailing the selection of a corpus from the collections of the
Musée des Arts et Métiers in Paris, France. A condition report on
23 painted metal objects was prepared including a description
of the materials, their visible alterations and a first diagnosis of
whether the degradation originated from deterioration of the
paint or corrosion of the metal support. Samples of paint and
corrosion products were collected from six objects in order to
characterize the system by micro-analysis. A multi-technique
analytical methodology was used to study the morphology, the
elemental composition and distribution and the variety of phases
present in the samples, thus allowing a complete characterization
of the metal, the corrosion products and the components present
in the paint layers. Aesthetic alteration of the paint was observed
in cases where the degradation of the metal support had led
to subsequent migration of the corrosion products into the
paint layers, leaving a brown-orange stain. The paint layers were
determined to have oxidized over time, which could affect their
stability in the composite system. The next steps of this project
are to summarize the analytical data into hypotheses for the
degradation mechanisms of the paint-metal system.
Keywords
paint, metal, alteration, restoration, Raman, X-ray fluorescence
spectroscopy (XRF), X-ray absorption near-edge struc­ture (XANES)

Introduction
The present paper details the CoPaiM project (Conserva-
tion of Painted Metals), which deals with the degradation
and conservation issues related to painted metal objects
from cultural heritage collections. Such composite items
are widely present in museum collections, including
artworks, technical and industrial objects (such as painted
signs, scientific instruments, models, tools), and means
of transport.
Progressive degradation of painted metal objects
is a common issue and its processes and associated
kinetics depend on several factors: storage condi-
tions (controlled or uncontrolled environment with
fluctuations of temperature and relative humidity),
pollution particles, presence of other harmful materials
and previous use or restoration treatments, among
other factors. As a result, metal corrosion products
can form and leave stains on the paint layer affecting
the aesthetics of the object. More importantly, their
continued development can produce an outward pres-
sure, pushing up the paint layer and creating cracks,
blistering and flaking until the paint layer is severely
degraded or completely separated from the metal
(Schröter 2009, Sutter 2014).
370  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
When storage conditions cannot be controlled or are
not suitable for all the materials present (in large and
heterogeneous collections for example), curators and
conservators have looked for surface treatments to slow
down or prevent the processes of degradation. However,
conservation of painted metals is not an easy task as the
treatment must ideally conserve and protect the surface
of the object including its paint layers and corrosion
products that may contain historical information such as
inscriptions or engravings. The preservation of painted
metal heritage therefore requires the development of
conservation strategies while also taking into account the
constraints of conservation ethics. Moreover, non-toxic,
reversible (and if not, re-treatable) and durable treat-
ments with no harmful volatile organic compounds are
nowadays required.
The first phase of the CoPaiM project deals with the
characterization of samples acquired from painted metal
objects with visible surface alteration in the Musée des
Arts et Métiers collection. This first step is essential to
understanding the processes of degradation and interac-
tions taking place in the paint-metal system, and provides
necessary information for a complete assessment of the
conservation state prior to the application of a protective
surface treatment. The second phase of the project, which
is not presented in this paper, concerns the creation of
painted metal plates imitating the degraded surface state
observed on the historic objects. A similar methodology
was proposed in previous studies (Degrigny 2010, Rapp
et al. 2010). These plates will be used as a laboratory test
media for the application and assessment of different
protective surface treatments. The information gained
from this study will then be used to make an informed
decision on the application of surface treatments to real
objects.
Materials: Studied collection and corpus
selection
Studied collection from the Musée des Arts et
Métiers
The painted metallic objects from the large scientific and
technical collections of the Musée des Arts et Métiers
(Paris, France) constitute an ideal corpus for the develop-
ment and assessment of protection protocols on painted
metals. The museum’s indoor storage building contains
a diverse collection of items that vary in size (from a few
centimeters to a few meters), utility (industrial, agricul-
COMPOSITE OBJECTS
tural), conservation state, period and materials (wood,
leather, plastic, rubber, etc.). The present study focused
on objects with metal supports.
Diagnostic report and database elaboration
The first step of the study was to diagnose (i.e., produce
a condition report on) the state of conservation of 23
objects stored in the museum’s reserves and selected for
the painted metals alteration study with help from the
curators and conservators (from the Musée des Arts et
Métiers and Institut National du Patrimoine) involved
in the CoPaiM project. The origin of the degradation
was investigated to determine whether the alteration
was linked to paint layer deterioration, such as lack of
cohesion or binder instability, or if the corrosion of the
metal was responsible for the degradation of the coating.
Following this diagnosis, a database for managing the
condition survey of the 23 objects was designed with
input from the curators and conservators. The objective
was to list the materials and coatings identified on the
object, to describe the current alterations (mechanical
and physico-chemical), to illustrate with photographs
and to identify their state of degradation. The data-
base is divided into four parts: identification of the
object, nature of the different components (including
metal, morphology and type of coating), alteration of
the components (metal and coating) and diagnosis. A
standard datasheet can be seen in Table 1.
Establishment of a corpus with representative
altered surfaces
A first overview of the collection showed that the types
of degradation and their origins were diverse and could
affect the metallic structure, paint layer and upper varnish
layer, or a combination of the coating and the metal. The
database allowed for the selection of six objects (Figure 1)
described below with representative degraded painted
metal system surfaces.
a. The “Quadruplette” (inv. 21811-0001) is a unique
object crafted by its inventor, Boyan Siméonoff,
to accommodate the whole family on road trips: a
four-seater bicycle supplemented by a small auxil-
iary motor. Designed just after World War II, it was
given to the museum in 1965. It is a well-used object
with an undocumented history of interventions. It is
constructed mainly of iron, with some parts in copper
and a white metal, with various forms of generalized
NEW STRATEGIES FOR THE CONSERVATION OF PAINTINGS ON METAL
Table 1. Datasheet established for the diagnosis of the conservation
state of metal painted objects
Identification • Denomination
• Localization
• Function
• Type of storage
• Size
Materials • Metal:
–  nature of the principal metal
– other metals: iron, copper, white metals or
aluminum
• Coating:
– morphology of coating, mono or multi-layer,
over or juxtaposed layer…
– type of coating: transparent or non-transparent,
hot or cold paint, varnish, …
• Other materials: organic, petroleum-based
components
Alterations • Mechanical and chemical degradations of the
metal: uniform or localized corrosion
• Alteration of other materials
• Mechanical and chemical alteration of the coating:
– optical and physical changes of the coating such
as fissures, deformation, …
• Presence of dust
Diagnosis • Evolution
• Causes of the degradation:
– alteration due to the metal part or the coating
– alteration due to the combination of the coating
and metal
– alteration due to the use of the object
and localized corrosion. The observed alterations are
mainly related to coatings with adhesion problems.
b. The reversible Brabant plow (inv. 17001) is an agricul-
tural device dating from the last quarter of the 19th
century that was inventoried by the museum in the
1930s, making it difficult to trace its history in the
museum exhibition display. This iron object is covered
in various areas with decorative multicolored paint.
Part of the iron structure is corroded but alterations
to the painted zones are mainly due to wear.
c. The weighing scale (inv. 5005) is a scientific instrument
used as a 50 kg checkweigher. It is linked to the impor-
tant metrological work performed by the Conservatoire
Figure 1. (a) “Quadruplette” (inv. 21811-0001); (b) reversible Brabant plow (inv. 17001); (c) weighing scale (inv. 5005); (d) three-pane lantern (inv.
16726); (e) model of engine boiler (inv. 36181); (f) set of six measuring containers (inv. 60005)
371
National des Arts et Métiers (CNAM, Paris). It stood
among the collection of weights and measures sent by
the CNAM to The Great Exhibition of 1851 in London.
Originally covered with several layers of paint, this iron
object is completely corroded and only shows a few
painted areas with small flakes of paint.
d. The small three-pane lantern (inv. 16726) is dated
to around 1906 and belongs to the museum’s railway
collection. It is a display piece, never used for its
intended industrial purpose. The corrosion is clearly
visible on the surface of the iron alloy and the handle
rivet and the black paint is showing problems with
adhesion.
e. The model of an engine boiler (inv. 36181) is of
unknown provenance. It is a scale model of a large
marine steam engine of British origin as shown by the
name plaque. The model consists of two parts, one
made of wood and the other of an iron alloy. Physical
and chemical alterations are visible on the metallic
part, as well as on the three types of paint layers.
f. The set of six measuring containers (inv. 60005) is also
of unknown provenance. It is dated to around 1850
by comparison with similar pieces in the collection.
The set is made from an iron alloy which is uniformly
corroded. The coating is the same all over the surface
and consists of a brown-black paint layer over an
orange paint layer. The alteration relates to the corro-
sion of the iron support.
Analytical methodology
The macroscopic approach presented in the database
describing the objects and their visible alterations is
supplemented by microanalyses to identify the materials
for the six selected objects: the nature of the metal, the
phases present in the corrosion layer, the composition
of the painting (pigments and binders) and the possible
stratigraphy. Microsampling allowed for the application
372  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
of a full analytical methodology using a multiscale and
multimodal approach. The development of fast imaging
techniques using synchrotron-based X-ray absorption
and fluorescence spectroscopies but also vibrational
spectroscopies, such as Raman, were used to achieve
local characterization.
Sample preparation
Millimeter-sized samples were collected with a scalpel
from areas of interest in order to include corrosion prod-
ucts, altered paint layers and potential varnish or upper
organic layers. Some samples were embedded in epoxy
resin and cut as cross sections with a diamond wire saw.
The faces were polished with SiC grit papers and with a
diamond paste on a damp polishing plate, rinsed with
ethanol and dried.
Morphological observations
Samples were observed at 5×, 10× and 20× magnification
with an Eclipse LV100 D-Nikon Microscope using visible
light in reflection and dark-field mode. A Zeiss Supra 55
VP scanning electron microscope (SEM-FEG) was used
with an acceleration voltage of 5 to 20 kV to observe the
topography and chemical contrast of the sample layers
thanks to secondary and backscattered electron modes
(SE and BSE).
Elemental and structural analyses
The elemental composition of the paint and corrosion
layers was identified using energy-dispersive X-ray
spectroscopy (EDS) and X-ray fluorescence spec-
troscopy at micrometric scale (µXRF). EDS analyses
were conducted at 20 kV at a working distance of 6.5
mm, using a Quantax EDS nano and microanalysis
system (Bruker) and a silicon drift detector (SDD).
Micro-XRF analyses were performed at 13 keV at the
DiffAbs beamline (Synchrotron SOLEIL, France). XRF
maps were collected using the continuous flyscan mode
(Esteve et al. 2017) thanks to a focused incident beam
of 8 µm × 6 µm, a four-element SDD (Vortex-ME4,
Hitachi) and 50 ms dwell time. All XRF experiments
were performed at room temperature and atmospheric
pressure (Ar K lines from ambient air are visible in XRF
spectra), implying difficulties in detecting the fluores-
cence signal from light chemical elements (such as Si,
P). Data processing was performed using the PyMCA
software (Solé et al. 2007).
COMPOSITE OBJECTS
Raman spectroscopy
Two instruments were employed to identify the struc-
ture of pigments and corrosion products present in the
microsamples. A Raman Senterra spectrometer (Bruker
Optics) was initially used, with an infrared diode laser
(785 nm), 50× IR objective, 3 mW laser power at the
sample surface and acquisition times between 2 and 10 s.
The spectral resolution was about 5 cm-1. Follow-up anal-
yses were performed using a LabRam HR 800 (HORIBA
Jobin Yvon) with the 458 nm line of an Ar+ laser and
a 100× objective. Laser power was adjusted to around
100 µW at the surface of the sample to avoid material
degradation during analysis. Spectra were collected for
3 - 30 s with 3 - 20 accumulations. Spectral resolution
was about 3 cm-1.
Gas chromatography coupled to mass spectrome-
try (GC-MS)
GC-MS was used for the separation and identification
of organic components in the paint layers. A trace GC
Ultra gas chromatograph (Thermo Scientific) and an
ISQ single quadrupole mass spectrometer (Thermo
Scientific) were used with the following specifications:
non-polar fused silica capillary column (Rxi 5Sil MS,
20 m, 0.18 mm di, 0.18 mm df Restek); carrier gas H2 at
1.5 mL/min; a temperature program of 50°C for 1 min,
ramp of 20°C/min until 320°C and 320°C held for 5
minutes; split injector set at 290°C with a total flow of
50 mL/min and a split ratio of 33; automated injection;
an interface temperature set at 320°C; electronic impact
ionization mode of 70 eV and ion source temperature
of 250°C. The samples required a derivatization step
prior to injection, which included mixing fragments of
the samples with 50 µL of Meth-Prep II (Grace Alltech)
and heating at 60°C for 30 min prior to injection. This
transmethylation is needed to cleave the oil-based and
terpenoid components of the binder, and to esterify the
polar groups (acids and alcohols).
X-ray absorption spectroscopy
Following the previously described µXRF maps which
were used to reveal, among other elements, the Pb
distribution, some points of interest were selected and
analyzed by collecting X-ray absorption near-edge struc-
ture (XANES) spectra at the Pb L3-edge to determine the
Pb local environment. The acquisition conditions for
XANES were from 12.6 to 13.3 keV, with steps of 1 eV
NEW STRATEGIES FOR THE CONSERVATION OF PAINTINGS ON METAL
in the pre- and post-edge regions and 0.5 eV in the edge
domain and a counting time set to 1 s.
Results
Metal alteration: Corrosion products
Corrosion products are present at the interface between
the metal part of the object and the paint layers. To
illustrate, a flake from the weighing scale was observed
using SEM and revealed well-crystallized corrosion
products (Figure 2).
Figure 2. SEM image (SE mode, 2 keV) of the corrosion products formed
between the metal and the paint layers of the sample collected from
the weighing scale
Otherwise, the sample (Figure 3) from the measuring
containers (inv. 60005) is an interesting example with a
complex multilayer system. The outer face is very dirty:
a multitude of gray to brown grains covers the painting
layer. The inner face (close to the metal) is brown-orange
due to the presence of corrosion products. This layer is
partly powdery with a shiny appearance.
Figure 3. Paint flake sample from the measuring and cross-section series
observed under optical microscope and XRF maps of major elements
373
The stratigraphy of the cross section is rather complex
with a variable thickness (from 290 to 570 μm). Many
porous layers of variable thickness, morphology and color
(gray, orange, brown) surround a thick orange layer (55
and 200 μm) which contains white particles.
The XRF maps illustrate the distribution of the main
elements: Fe surrounding Pb in the orange layer. The
iron corrosion products were identified thanks to the
analytical methodology previously described. The
Raman spectra of lepidocrocite (Figure 4) but also
goethite, akaganeite and maghemite were revealed
and identified thanks to comparison with references
(Bellot-Gurlet et al. 2009). Further analysis and data
treatment are in progress.
Figure 4. Raman spectra of corrosion products acquired at two points
of interest on the sample from the series of measures
Complexity of painted layers and their alteration
Identification of the pigments was performed for the entire
corpus using the above described methodology through a
combination of Raman, SEM-EDS, XRF and XANES. The
main results are presented here to highlight the variety of
pigments present in the paint layers (Table 2). This diver-
sity and intensity of colors is visible on the cross section of
the micro-samples collected on the following objects (see
Figure 5, from left to right): the quadruplette, the engine
boiler model and the reversible Brabant plow.
Figure 5. Cross section observed under optical microscope for (from left
to right): the quadruplette, the engine boiler model and the reversible
Brabant plow
374  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

Table 2. List of the main pigments identified one corresponding to the orange pigment minium,
Object Color Pigment/paint layer Pb 3O 4, and the other corresponding to lead white,
Quadruplette blue and lazurite Na8[Al6Si6O24]Sn 2PbCO3·Pb(OH)2. In addition, a spectrum similar to
green prussian blue Fe4[Fe(CN)6]3 lead soaps can be found on some samples (not shown
anatase TiO2
here), and can be linked to the degradation of Pb
lead white 2PbCO3.Pb(OH)2
lithopone ZnS and BaSO4 compounds in interaction with the binder of the paint
Rreversible yellow, minium/lead red, Pb3O4 layer. Such amorphous compounds were not revealed
Brabant plow red and chrome yellow PbCrO4 using other analytical techniques.
blue lazurite Na8Al6Si6O24
lithopone zinc sulfide ZnS
barium sulfate BaSO4
Weighing scale blue litharge/massicot PbO
lead white 2PbCO3.Pb(OH)2
lithopone ZnS and BaSO4
lazurite Na8[Al6Si6O24]Sn
anatase TiO2
Small three-pane black
lantern
Engine boiler red and lead white 2Pb(CO3)2,Pb(OH)2
white minium/lead red Pb3O4
vermillion HgS
non-identified organic pigment
Set of six brown minium/lead red Pb3O4
measuring and red lead white 2PbCO3.Pb(OH)2 Figure 7. Cross section of a sample from the series of measures
containers non-identified carbonaceous pigment observed using an optical microscope and XRF maps showing Fe and
Pb distribution. XANES spectra at Pb L edge were acquired at nine
Identified pigments were confirmed to have been in use different points of interest located on the Pb XRF map and compared
with references
during the time periods to which the objects are dated.
As an example, a large variety of elements were detected
on the weighing scale micro-sample using XRF and EDS The identification of the organic content was completed
(Figure 6). The results were correlated to Raman analysis using GC-MS as shown for six samples from five different
in order to identify the pigments. EDS maps revealed objects in Table 3, illustrating evidence of aged oil (oxida-
the main elements of Pb, Na and Ba, and indicate the tion products of unsaturated fatty acids) and an organic
presence of lead white, lazurite and BaSO4, a compound red pigment in the precise case of the engine boiler. For
found in lithopone. example, the main peaks identified during the analysis of
a red and yellow paint sample from the reversible Brabant
plow correspond to methylated suberic, azelaic, palmitic
and stearic acids and also to the methylated oxidation

Figure 6. Cross section observed under optical microscope for the

weighing scale and EDS maps of the main elements

The XANES spectra permitted the identification of

different Pb compounds constituting the paint layers
and determination of the Pb alteration products. To
illustrate, two types of spectra were identified on Figure 8. GC-MS measure obtained from the analysis of the sample from
the paint layer of the series of measures (Figure 7): the reversible Brabant plow
NEW STRATEGIES FOR THE CONSERVATION OF PAINTINGS ON METAL 375

Table 3. List of the organic compounds identified products of abietic acid (Figure 8). The presence of
Analyzed objects palmitic and stearic acids along with azelaic and suberic
diacids indicates that the binder is a drying oil and the

Reversible Brabant plow

large amount of azelaic acid is a marker of oxidized drying

Measuring containers
(samples P08, P09)
oil (Piccirillo et al. 2005). The abietic acid degradation
Organic compounds

Weighing scale
products reveal the presence of an oxidized (aged) resin,
Quadruplette

Engine boiler
(sample P06)

(sample P01)

(sample P02)

(sample P01)
probably from Pinus species (Van den Berg et al. 2000).

Interaction of metal/paint layers

Heptanedioic acid, dimethyl ester
(pimelic acid, dimethyl ester)
Octanedioic acid, dimethyl ester
(suberic acid, dimethyl ester)
Nonanedioic acid, dimethyl ester
(azelaic acid, dimethyl ester)
Decanedioic acid, dimethyl ester
(sebacic acid, dimethyl ester)
Hexadecanoic acid, methyl ester
(palmitic acid, methyl ester)
Octadecanoic acid, methyl ester
(stearic acid, methyl ester)
11-octadecenoic acid, methyl
ester, (Z)-
Nonanoic acid, 9-oxo, -methyl
ester
Nonanoic acid, methyl ester
(pelargonic acid methyl ester)
Glycerol 1,3-TFMP diether,
2-methyl ether
Glycerol 1,2-methyl diether,
3-TFMP ether
Glycerol 1-methyl ether, 2-TFMP
ether
Glycerol TFMP triether
Dimethyl phthalate (alkyd resin
fragment)
Naphthalene, 2-methoxy
Naphthalene, 1,7-dimethoxy-
Methyl dehydroabietate (methyl
ester of Abietic Acid - resin)
Tetradehydroabietic acid, 7
methoxy, methyl ester
7-oxodehydroabietic acid, methyl
ester
15-hydroxy-7-oxo-dehydroabietic
acid, dimethyl derivative
Jalaric acid, trimethyl ester
Laccijalaric dimethyl ester methyl
ether - gomme lacque
Cyclohexasiloxane,
dodecamethyl-
Benzenamine, N,N,3-trimethyl-
Benzenamine, N,N,4-trimethyl-
Synthetic organic red
pigment: naphthol or pyrrole
x x
x x
x x
x x
x x
x collected samples.
x
x x
x x
x
x
x x
x
x
x layer), cross section observed using optical microscope and XRF maps
x x
The condition report was prepared to determine the
x x x origin of the observed alterations: either physical
causing constraint on the paint layers; or of chemical
x x x origin due to diffusion of chemical species throughout
the layers. Chemical interactions between the two layers
x x x
can lead to significant visual changes. The most obvious
x x is due to the migration and diffusion of metallic ions
into the paint layers. Iron corrosion often leads to an
x x x orange-brown color change in the paint layer, especially
when the colors are pale and thus easily overpowered.
x
This phenomenon could be observed on part of the
To illustrate, the sample from the model of an engine
boiler shows white granular paint layers which are tinted
x x due to corrosion of the iron support (orange, brown).
The paint layers are mainly constituted of Pb (inside
x
layer) and Zn (top layer) suggesting the use of white
x
pigments such as lead white and zinc white. The orange
color visible in both layers is clearly due to the presence
x of Fe, as revealed by the XRF map (Figure 9). Fe is not
x present in the original pigment. After oxidation of the
underlying metal, the metallic ions (mainly Fe3+) most
x
likely diffused from the metal into the paint layers.
x
x
x
x
Figure 9. Paint flake sampling from the model of an engine boiler (top
of major elements
376  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conclusion and perspectives
This study has highlighted a rich and promising corpus
both in terms of conservation-restoration and the estab-
lishment of protection protocols for painted metals.
Following the first step of the CoPaiM project, some
requirements for protective surface treatments will be
determined.
Previous studies (Schröter 2009, Sutter 2014) note
the use of coatings to protect painted metal against
possible external aggressions or inhibitors to stabilize
the metal and prevent damage due to corrosion (Rocca
and Mirambet 2007). In fact, the complexity and plurality
of alterations to historical painted metal objects requires
this two-stage restoration: metal stabilization and paint
protection.
The second part of the CoPaiM project will focus on
laboratory tests on painted metal plates in order to assess
and validate a protective surface treatment while fully
respecting the ethical principles of conservation and
restoration. The effectiveness of corrosion inhibitors and
coatings will be evaluated by applying accelerated aging
techniques and monitoring the samples in order to test
the stability of the protected system.
Considering the environment in the museum and storage
room, some of the painted metal plates will be exposed
under light and to regulated climatic cycles corresponding
to atmospheric corrosion conditions, while others will
undergo a more complex environment in which the
insertion of pollution particles, such as concrete dust,
will simulate a specific indoor environment.
Some of the other painted metal plates will experience
natural aging at selected exposure sites in the museum
reserves to assess the treatments over a long period of
time. The degradation of metal and paint layers exposed
to an uncontrolled and polluted environment presents new
challenges for the characterization of their reactivity and
the establishment of efficient protective surface treatments.
The final objective of the project is to develop a formu-
lation and a treatment protocol whose composition
and application have been validated on real museum
collections.
Acknowledgments
The authors would like to thank Julie Sutter for the
initiative and Julie Schröter for her recent and future
COMPOSITE OBJECTS
involvement in the CoPaiM project. They would also
like to thank Emmanuel Rocca and François Mirambet
for their precious scientific advice, Synchrotron SOLEIL
for the provision of synchrotron radiation facilities and
Dominique Thiaudière, Cristian Mocuta and Philippe
Joly for the DiffAbs beamline. They are also grateful
to Ipanema and LPS (UMR 8502, Univ. Paris Sud) for
laboratory access and Cecile Fosse and Dominique
Bazin, respectively, for conducting SEM-EDS analyses.
CoPaiM is partly funded by DIM MAP (Domaine d’In-
térêt Majeur, Matériaux Anciens et Patrimoniaux, Ile
de France).
References
Bellot-Gurlet, L., D. Neff, S. Réguer, J. Monnier, M. Saheb,
and P. Dillmann. 2009. Raman studies of corrosion layers
formed on archaeological irons in various media. Journal
of Nano Research 8: 147–56.
Degrigny, C. 2010. Use of artificial metal coupons to
test new protection systems on cultural heritage objects:
Manufacturing and validation. Corrosion Engineering,
Science and Technology 45(5): 367.
Esteve, E., S. Réguer, C. Boissiere, C. Chanéac, G. Lugo,
C. Jouanneau, C. Mocuta, D. Thiaudière, N. Leclercq, B.
Leyh, J.-F. Greisch, J. Berthault, M. Daudon, P. Ronco,
and D. Bazin. 2017. Flyscan opportunities in medicine:
The case of quantum rattle based on gold quantum dots.
Journal of Synchrotron Radiation 24: 991–99.
Piccirillo, A., D. Scalarone, and O. Chiantore. 2005.
Comparison between off-line and on-line derivatisation
methods in the characterisation of siccative oils in paint
media. Journal of Analytical and Applied Pyrolysis 74:
33–38.
Rapp, G., C. Degrigny, F. Mirambet, S. Ramseyer, and
A. Tarchini. 2010. The application of non-toxic corro-
sion inhibitors for the temporary protection of iron and
copper alloy in uncontrolled environments. In Metal
2010: Proceedings of the Interim Meeting of the ICOM-CC
Metals Working Group, 11–15 October 2010, Charleston,
South Carolina, eds. P. Mardikian, C. Chemello, C.
Watters, and P. Hull, 185–92. Clemson, SC: Clemson
University.
Rocca, E. and F. Mirambet. 2007. Corrosion inhibitors for
metallic artefacts: Temporary protection, eds. P. Dillmann,
G. Beranger, P. Piccardo, and H. Matthiesen, book no.
48. European Federation of Corrosion.
NEW STRATEGIES FOR THE CONSERVATION OF PAINTINGS ON METAL 377

Schröter, J. 2009. L’étude et la conservation-restauration

du fer peint: le cas d’une croix tombale en fer forge
peint au Musée des Arts Appliqués de Vienne, Autriche.
Master’s thesis in Conservation and Restauration of
Cultural Heritage, Université Paris 1 Panthéon-Sorbonne,
France.
Sutter, J. 2014. Aux Trois Rois: Etude et conservation-res-
tauration d’une enseigne en alliage ferreux, peinte sur
ses deux faces (musée Unterlinden de Colmar). Master’s
thesis in Conservation of Cultural Heritage, Institut
national du patrimoine, France.
Solé, V.A., E. Papillon, M. Cotte, Ph. Walter and J. Susini.
2007. A multiplatform code for the analysis of energy-dis-
persive X-ray fluorescence spectra. Spectrochimica Acta
Part B: Atomic Spectroscopy 62: 63–68.
Van den Berg, K.J., J.J. Boon, I. Pastorova, and L.F.M.
Spetter. 2000. Mass spectrometric methodology for
the analysis of highly oxidised diterpenoid acids in
Old Master paintings. Journal of Mass Spectrometry 35:
512–33.

Authors
Julie Gordon is a recent graduate from Sorbonne
Université, having completed a Master of Science in
Analytical Chemistry. Her Ph.D. research focused on the
CoPaiM project at Synchrotron SOLEIL and MONARIS
laboratory.

Solenn Reguer has been beamline scientist on the

DiffAbs X-ray beamline at Synchrotron SOLEIL since
2008.
Preserving Fort Erie’s CN 6218. A Case Study
Susan L. Maltby* Jane Davies
Maltby & Associates Inc. Fort Erie Museum and Cultural Services
Toronto ON, Canada Fort Erie ON, Canada
susan.maltby@utoronto.ca JDavies@forterie.ca
*Author for correspondence
Abstract
The Canadian National Railway (CN) 6218 steam locomotive was
in active service throughout Canada and the USA from 1942 to
1959. In 1973, local rail enthusiasts were successful in lobbying
the Fort Erie (Canada) Town Council to acquire the 6218 from
CN. Its acquisition led to the formation of the Fort Erie Railway
Museum. In 1989, museum volunteers raised considerable
funds to refurbish the locomotive. One of the primary tasks
was the removal of the asbestos insulation around the boiler.
This abatement left the locomotive with an unfortunate and
deleterious legacy. The asbestos was removed and the existing
jacketing discarded. The gap left by the removal of the asbestos
was filled with pink Fiberglas insulation and an inaccurate
jacketing was installed on top. The jacketing was detailed so that
water ran onto the boiler rather than off of it and the insulation
held the moisture against the boiler, guaranteeing corrosion.
In 1999, the conservator was hired to assess the condition
of the 6218 and to make recommendations for its long-term
preservation. From the beginning, the 6218 was treated as a
museum object. The approach taken was preservation, not
Introduction
Figure 1. CN 6218 (spring 2017)
The Canadian National Railway (CN) 6218 steam loco-
motive (Figure 1) has been on permanent display at the
restoration. In 2017, the Town commissioned the conservator to
identify, investigate, analyze and propose options for the future
of the 6218 to help Council decide upon a course of action.
Working with the museum’s curator, the conservator’s study
sought community input, which yielded ten distinct visions. The
options ranged from full restoration (making the locomotive
operational) to disposal. Seven risk criteria were identified and
used to assess each of the ten options. The risk analysis clearly
identified the lowest risk option: repairing and protecting the
locomotive, in stages, at the museum – the recommendation
outlined in the 1999 report. This recommendation was put
forward to Council and accepted. The work began in the spring of
2018. This paper demonstrates how risk analysis can effectively
support and promote the conservation of a significant museum
object.
Keywords
steam locomotive, risk management, consultation,
collaboration
Fort Erie Railway Museum since 1974. In 2017, the Town
commissioned the conservator to identify, investigate,
analyze and propose options for the future of the 6218
in order to assist Council in deciding upon a course of
action. The study was a completely collaborative process.
Throughout, the conservator worked closely with the
museum’s curator.
The study sought community input. All involved in the
consultation were asked to respond to the same six ques-
tions regarding their vision for the future of the 6218.
The consultation resulted in ten distinct vision options.
The options ranged from full restoration (making the
locomotive operational) to disposal. The study took a
risk-management approach in which seven risk criteria
were identified and used to assess each option. The risks
PRESERVING FORT ERIE’S CN 6218. A CASE STUDY
were defined as to what constituted a low, medium or
high risk. These risks were then applied to each of the
ten options. This paper discusses the planning process
and risk analysis that led to the Town’s commitment to
the preservation of the 6218.
Consultation
Throughout the project consultation was carried out with
conservators, curators, historians, archivists and railway
heritage enthusiasts, all of whom have a long-standing
interest and expertise in preserving rail heritage. These
individuals provided invaluable insight into facilitating
and managing the preservation of historic rolling stock
within a museum setting. They generously shared the
lessons learned from their various restoration/conser-
vation projects. All had worked with steam locomotives
comparable to the 6218.
Background/History
Figure 2. CN 6218 in service (Fort Erie Historical Museum)
The 6218 and tender were fabricated at the Montreal
Locomotive works in 1942. The locomotive was in active
service throughout Canada and the USA until 1959
(Figure 2). In 1963, it was restored to working order and
used for steam excursion trips. It finally retired from
service in 1971.
In 1973, local rail enthusiasts were successful in lobbying
the Fort Erie Town Council to acquire the 6218 from CN.
The Council Motion to accept the donation stated that part
of the agreement between the Town and CN was “that the
locomotive would be kept in a clean, attractive condition
at all times, that adequate protection would be provided
to minimize vandalism and that a suitable roof would
379
be erected to provide overhead protection.” Although
the work done over the years on the engine attempted to
uphold this motion, there were some shortcomings.
The Fort Erie Railway Museum was formed in 1973 by a
group of local volunteers. One year later, the 6218 and the
caboose were moved to their current location. Volunteers
installed the track upon which they currently sit. Steps
were taken at the time to try and prepare the 6218 for a
life outdoors. These measures included: cleaning out the
smokebox; sealing the smoke stack (with a steal plate set
into the stack) to prevent water from running into the
smokebox; and welding a steel plate over the coal hopper
to keep rain and snow out.
In 1989, the Railway Museum volunteers successfully
raised $58,000.00 to fund the refurbishment of the loco-
motive. One of the primary tasks was the removal of the
asbestos insulation around the boiler. This abatement
left the locomotive with an unfortunate and deleterious
legacy. The asbestos was removed and the existing
jacketing discarded. The gap left by the removal of the
asbestos was filled with pink Fiberglas insulation and a
less-than-accurate jacketing was installed on top. The
jacketing was detailed so that water ran onto the boiler
rather than off of it and the insulation held the moisture
against the boiler guaranteeing corrosion (Figure 3). Since
1989, additional funds have been allocated to cover the
costs of painting and additional repairs to the locomotive.
Figure 3. Corroded jacketing and boiler below
The ownership and management of the Fort Erie Railway
Museum has evolved over the years from a volunteer-based,
not-for-profit organization to a municipal government
department. This change was precipitated by financial
pressures and a declining volunteer base. The former
management model began in 1970; the Town of Fort Erie
assumed ownership in 1987 when the museum amalga-
380  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

mated with two other museums within the municipality. Condition update
At that time, the first full-time professional staff was
The condition of the 6218 had deteriorated since the
hired to coordinate activities and volunteers for each of
the sites. In 2005, the Town assumed all management initial assessment. This was confirmed by comparing
responsibilities. This change had implications on how the images from 1999 and 2008 with the current condition
CN 6218 was maintained. Emerging from the not-for- (Figures 4–7). The boiler jacketing and the Fiberglas
profit model was a scenario where volunteers made the insulation continued to be a major concern and cause of
conservation decisions and performed the repairs. The degradation. The insulation acted like a sponge holding
government model resulted in conservation decisions water next to the boiler. This, in turn, caused the jack-
being made by professionals, guided by government eting, and the boiler below, to continue to corrode. The
regulations, policies and industry standards. Subsequent true extent of the corrosion was not known.
repairs were then completed by contractors who followed
specifications drawn up by professional conservators. The
decision to limit the use of volunteers was a recognition by
the Town of Fort Erie of the risks with volunteer labour,
including: potential harm to individuals, Town staff and
the public; potential damage to Town’s assets, equipment
and reputation; cost and staff time associated with using
nonprofessional labour; potential violation of collective
agreements; and potential legal liability for the Town. The
Town’s Volunteer Policy requires any volunteer group be
responsible for the health and safety of all their members
while on a specific project.
Figure 4. CN 6218 (1999)
In 1999, the conservator was hired to assess the condition
of the 6218 and to make recommendations for its long-
term preservation. From the beginning, the 6218 was
treated as a museum object (i.e., minimal intervention).
The approach taken was preservation, not restoration.
This approach was shared by both the curator and the
conservator. The condition of the locomotive was such
that it was now a static object. Restoring it to running
condition was not considered a viable option. As such,
the recommended, staged treatment was designed to
stabilize the locomotive in its current setting. Recom-
Figure 5. CN 6218 (2017)
mended work included: removal of the jacketing and
insulation; stabilization of the boiler; fabrication and
installation of an accurate boiler jacketing; protection
from the elements; and painting. Follow-up assessments
were made in 2008 and 2012.

Competing municipal priorities in the context of limited

financial resources prevented the implementation of
the 1999 recommendations. In 2017, a number of town
councillors made the 6218 a priority. It was their wish
that the Town Council engage in a formal debate so
that a decision could be made about the steam engine’s
future. Their commitment to the 6218 led to the Town
commissioning the study. Figure 6. Engine letters (1999)
PRESERVING FORT ERIE’S CN 6218. A CASE STUDY
Figure 7. Engine letters (2017)
The stability of the track bed was also a concern as the
engine was listing slightly to one side. The engine did
not appear to have moved since the 2008 assessment.
Public consultation
Figure 8. Open
 house notice
Community input was sought through holding meetings
with identified stakeholder groups who had expressed
interest in the future of the CN 6218; by soliciting public
input through the Town of Fort Erie’s website and social
media platforms; through questionnaires distributed at
the Fort Erie Historical Museum and Fort Erie Railway
Museum; and by hosting a public open house at the
Town Hall (Figure 8). The identified stakeholders were:
the Fort Erie Museum Services Standing Committee;
the Fort Erie Historical Railway Association (FEHRA);
the Niagara Railway Museum (NRA); and the Alberta
Railway Museum (ARM).
381
Some of the stakeholder groups had developed their own
plans for the future of the 6218 and were lobbying Town
Council to support their involvement. Council’s goal
in commissioning this study was to have a transparent
consultation process and objective analysis of all potential
options on which to base its decision.
All involved in the consultation – stakeholders and the
general public – were asked to respond in writing to the
following questions:
1) What is your vision for the future of the 6218 and
caboose1?
2) How much do you think this vision will cost?
3) Do you have a budget for this vision?
4) Can you suggest any potential funding – public and/
or private – sources to support this vision?
5) How would you implement this vision? Please be
very specific outlining all steps involved.
6) Do you have any comparable experience? Please
elaborate.
Each stakeholder group was sent the questions in advance
and asked to bring written responses to the meeting. For
the general public, a notice was posted on the Town’s
website with the date and time of the open house and
the list of questions. Those unable to attend the public
meeting were invited to submit their answers via email
or by dropping them off at either the Fort Erie Historical
or Railway museums.
The visions
Twenty submissions were received during the public
consultation. There were clear overlaps and shared
visions amongst them. As a result, the 20 submissions
were distilled down to 10 distinct options.
Note that the difference between “restore” and “repair”
in the following visions reflects the level of intervention.
Restore refers to restoring missing components (e.g.,
dials, switches, damaged hoses) and/or restoring to
working order. Repair refers to preserving the engine in
its current state.
Option #1 – Restore using volunteers and operate
at the Fort Erie Railway Museum
Restore the CN 6218 to operational condition (using
volunteers), laying track and running the train throughout
Fort Erie or to other towns. The caboose could either be
operated as part of the train or converted into a restaurant.
382  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Option #2 – Restore and fence using volunteers at
the Fort Erie Railway Museum
Restore the engine and caboose and install a fence
around them for protection. This vision included using
volunteer labour.
Option #3 – Repair and protect using volunteers
at the Fort Erie Railway Museum
Repair and protect the engine according to the 1973
agreement between CN and the municipality using
contractors and/or volunteers.
Option #4 – Repair and protect at the Niagara
Railway Museum
Move the CN 6218 to the Niagara Railway Museum
(NRM) at the municipality’s expense where the repairs
would be done by paid contractors at the municipality’s
expense with the assistance of NRM volunteers.
Option #5 – Repair and protect at the Alberta
Railway Museum
Move the engine and caboose to the Alberta Railway
Museum in Edmonton (3500 km away) and either
repair or restore them to operational condition. This
vision included developing strategies for fundraising,
partnerships, etc. Their preliminary implementation
plan suggested moving the equipment within six months,
repairing it within one year and, if feasible, restoring it
to operating condition within two to five years.
Option #6 – Fundraise, repair and protect at the
Fort Erie Railway Museum
Repair and protect the engine and caboose on site. This
vision suggested the municipality hire a professional
fundraiser to raise much of the needed funds for this
project.
Option #7 – Repair and protect at the Fort Erie
Railway Museum
Uphold the 1973 agreement made between CN and the
Town of Fort Erie. That agreement included: keeping
the engine in an attractive condition, adequate protec-
tion from vandalism and covering the engine with a
roof. This vision included repairing and protecting the
engine on site.
COMPOSITE OBJECTS
Option #8 – Repair and protect (staged plan) at
the Fort Erie Railway Museum
Repair and protect the engine according to the 1973
agreement between CN and the municipality in three or
more stages, based upon available funding. This vision
divided the project into three major tasks: mandatory
remediation; protection from the elements; and, finally,
reconstruction of the jacketing.
Option #9 – Defer
This vision suggested deferring the decision as to which
vision to pursue in the hope that grant money might
become available in the future.
Option #10 – Dispose
This vision proposed the disposal of CN 6218 and the
caboose and removal of these assets from the Fort Erie
Railway Museum site. This process would follow the
guidelines outlined in the municipality’s by-laws and
associated policies.
Risk assessment criteria and definitions
Seven risk criteria were identified and used to assess
each option. These were: cost estimate; human resources;
municipal policies and practices; industry standards/
best practice; cost savings; impact on the Fort Erie
Railway Museum; and impact on the 6218 and caboose.
The following defines these criteria and how risk was
determined.
a) Cost estimate
The costs outlined by the person or group providing
the vision were evaluated to determine the risk to the
municipality. The cost estimate was compared with the
municipal financial resources currently allocated towards
this project.
• Low risk: cost estimate affordable based on current
municipal resources allocated towards this project;
• Medium risk: cost estimate close to the amount of
current municipal resources allocated towards this
project or possibilities created to defer expenses;
• High risk: cost estimate unaffordable based on
current municipal resources allocated towards this
project.
PRESERVING FORT ERIE’S CN 6218. A CASE STUDY
b) Human resources
The human resources listed to deliver the vision were
assessed to determine if they were considered viable to
fulfill the vision. The feasibility of the implementation
plan was also assessed.
• Low risk: sufficient and viable human resources and
a feasible implementation plan;
• Medium risk: insufficient human resources and a
feasible implementation plan – or sufficient human
resources and a lack of feasible implementation plan;
• High risk: insufficient human resources and lack of
feasible implementation plan.
c) Municipal policies, practices and regulations
Canadian municipalities must adhere to a number of poli-
cies, practices and regulations. In Fort Erie this included:
the Health & Safety Policy, the Collective Agreement, the
Tendering & Purchasing Policy and the Occupational
Health & Safety Act. How the vision complied and/or
conflicted with municipal policies, practices and regu-
lations was assessed.
• Low risk: complies with relevant municipal policies,
practices and regulations;
• Medium risk: complies with some municipal policies,
practices and regulations;
• High risk: does not comply with relevant municipal
policies, practices and regulations.
d) Industry standards (e.g., provincial museum
standards, Fort Erie Museum policies, best prac-
tice (e.g., Canadian Association for Conservation
of Cultural Property Code of Ethics)
This addresses how the vision complied with established
industry conservation standards/best practice.
• Low risk: complies with industry conservation standards;
• Medium risk: complies with some industry conser-
vation standards;
• High risk: does not comply with industry conservation
standards.
e) Cost savings
Consideration was given to groups willing to assume
some or all of the cost of implementing the vision. Costs
could be reduced with in-kind donations (i.e., labour or
materials) or monies.
383
• Low risk: municipality assumes little to no costs;
• Medium risk: municipality assumes some of the costs
or possibilities created for cost savings;
• High risk: municipality assumes all costs.
f) Impact on the Fort Erie Railway Museum
Removing a large steam locomotive and a caboose from a
railway museum would have an impact on the museum.
The positive and negative effects that the vision would
have on the operation and future of the Fort Erie Railway
Museum were assessed.
• Low risk: positive effect;
• Medium risk: status quo and/or the combination of
positive and negative effects;
• High risk: negative effect.
g) Impact on the CN 6218 and caboose
This addresses how the implementation of the vision
affects the integrity of the 6218 and caboose.
• Low risk: integrity not affected;
• Medium risk: integrity could be compromised;
• High: integrity compromised.
Risk assessment summary
The results of the risk assessment, as applied to each of
the ten options, are summarized in Figure 9.
The risk analysis clearly identified Option #8 as the
lowest-risk option. This option involved repairing and
protecting the locomotive, in stages, at the museum,
using qualified professionals to repair and protect the
6218. This option fulfills the 1973 agreement between
CN and the Town of Fort Erie and mirrors the recom-
mendations made in the conservator’s 1999 report. The
work is divided into three stages: 1) removing the boiler
jacket and Fiberglas insulation; opening the smoke box;
making minor repairs (e.g., repairing the hole in the
smoke box); and applying a protective coat of paint; 2)
protection from the elements by erecting a protective shed
over the locomotive; and 3) fabricating and installing an
accurate boiler jacketing that will be painted to match
the locomotive.
The final stage is cosmetic and could be put off until
sufficient funds are available. Although, from a curato-
rial standpoint, it would be ideal to have the jacketing
on the engine, there are precedents of similar locomo-
384  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 9. Risk assessment summary
tives without a boiler jacket. The Canadian Museum of
Science and Technology has a sister engine, CN 6200,
in its collection. The 6200 is displayed outdoors and has
no jacketing. The museum’s approach to preservation is
periodically repainting the engine with a two-part epoxy
paint system.
The recommendation, with full budget costing for each
of the three stages, was put forward to Town Council and
accepted. The stage 1 work began in the spring of 2018
Figure 10. After treatment (2018)
COMPOSITE OBJECTS
and was completed by October (Figure 10). A full report,
outlining the study and its results, was also submitted to
the Town.
Track stability
The stability of the track bed and the 6218 needed to be
determined. A monitoring programme was established
in stage 1 whereby a professional land surveyor took a
number of elevations from the track bed. The location
of each elevation was recorded and the process will be
repeated annually. If monitoring shows a significant
change in elevation over time, an engineering study will
be commissioned to determine if the track bed needs to
be replaced.
Testing
It was highly likely, given the age of the 6218, that some of
the extant paint contained lead and/or other heavy metals.
The ash and debris in the smokebox was also considered
to be potentially toxic. Prior to commencing work, an
environmental consultant was retained to sample and
analyze potentially toxic materials on the locomotive.
The consultant also tested loose fibrous material for the
presence of asbestos. The testing showed that the paint
did include lead and the ash in the smokebox included
a myriad of elements including antimony, arsenic,
cadmium, cobalt, lead, vanadium and mercury. No
asbestos was found. As a result, a qualified abatement
contractor was hired to remove all nonadherent paint and
to clean out the smokebox. They were also responsible
for the safe disposal of the material.
Conclusion
The preservation of CN 6218 required sufficient financial
resources and the political will to advance this project.
Competing visions from community organizations for
the future of this asset presented an additional challenge
for local politicians.
This challenge was addressed with a thorough and trans-
parent community consultation, designing a compre-
hensive risk analysis and objectively applying the risk
assessment to each vision.
Fort Erie Town Council was then able to debate the merits
of each vision based upon relevant and easily comparable
facts. The ensuing motion to allocate tax dollars towards
PRESERVING FORT ERIE’S CN 6218. A CASE STUDY 385

one of these options had community support because

the community was engaged throughout the process.

Acknowledgements
The authors would like to thank Richard Stromberg for
reviewing this publication. The input of the following
conservators, curators, historians, archivists and railway
heritage enthusiasts was invaluable to this project:
Richard Fuller, Sue Warren, Jean-Paul Viaud, Paul
Robertson, George Prytulak, Katherine McCracken,
Grant Kingsland and R. Patrick McKnight. And, finally,
thanks are due to Nicole Komar for technical support.

Notes
1 Although included in the public consultation, the
caboose – and its long-term preservation – was outside
of the conservator’s scope of work for this project.

Authors
Susan L. Maltby has a master's in Art Conservation
from Queen’s University, Kingston, Ontario, Canada.
After graduation, she worked in the Ethnology laboratory
at the Canadian Conservation Institute for four years. In
1988, she established her own conservation consulting
firm in Toronto. In addition to her consulting work, Ms.
Maltby is adjunct faculty in the Master of Museum Studies
Programme/Faculty of Information at the University
of Toronto and teaches in the University of Victoria’s
Cultural Resource Management Programme. She also
writes a numismatic conservation column – “Preserving
Collectibles” – for Coin World.

Jane Davies graduated from McMaster University in

Hamilton, Ontario, Canada, with a Bachelor of Arts in
History. She has been the Curator of the Town of Fort
Erie’s Museums since 1987. As the first professional at
the Town’s three museums, she was responsible for devel-
oping these three separate volunteer organizations into an
amalgamated Eco-Museum model. She now manages a
team of paid staff at the Fort Erie Railway Museum, Fort
Erie Historical Museum, Ridgeway Battlefield Interpre-
tive Park, Mewinzha site and a 5,000-square-foot artifact
storage facility. She co-authored the books Many Voices
– A Collective History of Greater Fort Erie; Many Voices
II; and First-Hand Accounts of the 1866 Fenian Raid and
Battle of Ridgeway.
Metal Textile Composites and Improved Treatment
Outcomes through Application of In-Situ Corrosion Data
Ian D. MacLeod* Julia Brennan
Western Australian Maritime Museum Caring for Textiles
Fremantle WA, Australia Washington, DC, USA
ian.macleod@museum.wa.gov.au julia@caringfortextiles.com
*Author for correspondence
Abstract
Before embarking on the conservation of the royal collection of
the Thai court, housed in the Queen Sirikit Museum of Textiles,
it was necessary to establish new methods of examination and
assessment using replicas of the traditional court garments. The
silver and copper alloys used in the construction of these highly
decorated objects and the impact of accumulated sweat residues
determined both the extent of corrosion and the nature of the
decay products. Use was made of late 20th-century Khon dance
costumes belonging to private Bangkok collectors in training
exercises in which the critical evaluation of corrosion potential
data, pH and chloride ion measurements was made. Specific
forms of degradation varied and this, in turn, determined what
treatment paths could be pursued with safety to the textile and
Introduction
The royal collection at the Queen Sirikit Museum of
Textiles (QSMT) in Bangkok consists of approximately
5,000 textiles associated with the Queen and includes her
personal clothing, textile collections and textiles arising
from her 50-year SUPPORT Foundation involving rural
crafts development. A program of rotating exhibitions
about Her Majesty, the King and their textiles involves the
full range of conservation activities – from the analysis
of materials to the treatment of objects – on a continuing
cycle. Many textiles from the 18th and 19th centuries are
woven and embellished with metallic threads and adorn-
ments, which makes them metal-textile composites for
which there are no standard operating procedures. The
first training program in Thailand was held in 2016 at
the QSMT to assess the treatment issues associated with
metallic fabrics, since this type of object is common to
the Thai court. Included here are the complex costumes
of the traditional Khon dance, a narrative dance form
in which dancers are literally sewn into the costumes
for periods of up to 8 hours. Such composite textiles
need great flexibility owing to the extreme movements
Rinske Car
Denmark River Textile Conservation Studio
Denmark WA, Australia
rinskecar@gmail.com
comfort to the conservator. The sensitivity of the surface pH
of textiles to the chloride concentration from sweat was very
high. Comparison with data from less humid environments has
quantified the impact of high temperature and relative humidity
on the decay mechanism. The training program resulted in
empowered conservators conducting in-situ measurements on
composite textiles that placed the artefacts onto the decision-
making tree.
Keywords
Queen Sirikit, textiles, metallic threads, sweat, corrosion,
biodegradation, dithionite reduction, Khon dance
performed by the dancers and show copper corrosion
products associated with reactions with human sweat.
The workshop was the first exposure of textile conserva-
tors in Thailand to decision-making based on a knowl-
edge of the pH, chloride and metallic compositions of the
objects. Data gained from in-situ measurements provided
a matrix for determining which treatment methods were
the best to pursue. Planning of the workshop took two
years, during which the team leader (Brennan) built trust
in the processes with the stakeholders.
There are specific challenges associated with the treat-
ment of royal Thai textiles. The development of the
conservation workshop had to consider that many of the
young conservators had learned their skills on the job
through traditional apprenticeships in the conservation
department. Others had had some engagement with a
course from the CollAsia program (Brennan et al. 2014),
but perhaps one of the most significant outcomes of this
work is that the results and the pedagogical approach
are directly transferable to textile collections throughout
Southeast Asia and the Pacific Islands. While signif-
METAL TEXTILE COMPOSITES AND IMPROVED TREATMENT OUTCOMES THROUGH APPLICATION OF IN-SITU CORROSION DATA 387

YES YES Questions NO NO

1 Are there bright green
corrosion products?

Do a solubility
test with cotton 2 Is mineralisation Measure the Ecorr of
bud & ethanol present with blue green oxidized metal
colour? thread
If bud turns Place object in
green, then Wash with diammonium
wash in 50% diammonium 3 Is the metal generally citrate and
ethanol citrate tarnished & dull? monitor Ecorr

Wash with Treat metal with

diammonium Treat textile with buffered pH 6.5
citrate buffered pH 6.5 and S2O42- @5 g/l
and S2O42- @5 g/l
-
Figure 1. Decision-making
tree for treatment of
Rinse 4 times in fresh H2O
 composite textiles

icant inroads into the challenges of stabilising textile
collections in Singapore have been made (Janaway
and Cunningham 1995), this has been a long-term
process involving many millions of dollars. The raison
d’être of the workshop was to use the experience at the
QSMT as the launchpad for an Asia-wide approach to
understanding the decay and conservation needs of
composite metallic textiles. The workshop empowered
the conservators to become more able to achieve good
collection outcomes when negotiating with curatorial
managers who understand the practices. By approaching
collection items in terms of their significance, their
conservation needs and the specific nature of the decay
of the metallic elements, it was possible to demonstrate
that even the most difficult and problematic items could
be managed through a decision-making tree (Figure 1).
This approach removed a lot of stress from the indi-
vidual conservator by developing appropriate treatment
options. In addition, the project sought to build the
confidence and leadership skills of the local team which
should open pathways to additional training, including
their own implementation of workshops in regional and
rural collection centres in Thailand.
Thus, textiles from private collections were sourced
for the workshop since they exhibited a wide range of
deterioration issues. The fabrics were mainly cotton, silk
and some woollen materials. The owners were happy to
have their items tested, assessed and treated as part of
the workshop outcomes. These textiles included typical
netted robes known as ‘sua khrui’, with heavy metallic
thread brocades typical of the late 1800s and 1900s royal
costumes, and long lengths of cloth worn as hip wrappers
by the court and in dance traditions called ‘pha nung’,
as well as many dance costume components such as
collars, arm bands, belts, leggings and jewellery (Salim
2015) (Figures 2 and 3). New approaches were needed to
develop treatment protocols for royal metallic textiles –
which are rarely worn and often highly corroded – and
dance costumes – which are heavily used, poorly stored
and heavily corroded.

Collection samples
Given the experimental nature of the workshop, the
historic textiles belonging to the royal court could not
be directly used due to their high cultural value and
rarity. Protocol demands that only proven approaches to Figure 2. Stained Thai 19th-century textile embellished with gilded
textile conservation are acceptable for any royal items. Indian silver-alloy thread
388  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 3. Owner of a Burmese jacket (centre) watching Rinske Car and
Julia Brennan treating the object
Study methods
The workshop was structured into theoretical and prac-
tical lectures on the nature of metal alloys used for deco-
rative purposes and how they can be identified without
access to modern expensive instrumentation such as
x-ray fluorescence (XRF) and Fourier transform infrared
spectroscopy (FTIR). The program covered the sensitivity
of alloys to corrosive elements found in tropical climates,
which included issues of the high relative humidity, high
temperatures and persistent low levels of hydrogen sulfide
emanating from nearby drains and canals in Bangkok.
All these factors accelerate corrosion of metals. The
loaned textiles provided excellent samples of how metal
decay processes can also impact on the rates of deterio-
ration of the associated textiles. Training presentations
were interspersed with hands-on testing and cleaning
procedures that illustrated the technical sessions, thus
consolidating learning. Treatments included the use of
chelating agents to remove the corrosion products from
the metal surfaces which obscured the structural details of
the textiles. Optical examination under 20× hand lenses
and a binocular microscope proved to be sufficient to
characterise the degradation of the silver-copper alloys.
In the characterisation steps it was important to use only
methods that are inherently available in low-funded
regional conservation centres.
The training workshops involved combinations of
discussions about the nature of metals and what can be
gleaned from the nature of the different corrosion prod-
ucts. Simple spot tests for solubility of blue-green copper
corrosion spews, using ethanol-soaked swabs, clearly
demonstrated the difference between organic copper
complexes and inorganic corrosion products which are
COMPOSITE OBJECTS
only soluble in acidic or chelating solutions. The textile
conservators integrated their topics of the traditional
weaves of village-based looms versus machine-assisted
woven fabrics and how natural silk fibres have a different
response to relative humidity than those of cotton, flax
and hemp. The textile discussions included how the
metallic threads were made and then woven into the
fabrics to create the lustrous and highly valued garments.
Gradually, there was a significant change in the approach
by the local conservators who, emboldened by their own
practical measurements and trained in the interpretation
of the analysis of the results, began to see the strategic
advantages of knowing much more about the nature of
the collections on which they were working. Prior to the
workshop, traditional responses to the artefacts took the
least interventive approach possible, since this appeared
to pose the least risk. Another important aspect of the
training was to develop cheap methods of analysis, as
expensive analytical equipment is currently not feasible
for the QSMT conservation lab.
Corrosion potential measurements
The first essential step for the local conservators was
gaining confidence in taking direct measurement of the
corrosion voltages of the textile alloys. The reference
electrode was a silver chloride (Ag/AgCl, 3 M KCl) and
a platinum working electrode connected to a digital
multimeter, with the reference plugged into the common
terminal. Corrosion potential measurements have previ-
ously been used to determine alloy contents on a wide
range of artefacts (Degrigny et al. 2011) and were made
by inserting the platinum wire electrode (2 mm o.d.)
through a section of a domestic sponge rubber cloth
to touch the metallic elements and establish electrical
contact. The sponge was wetted with carbonated mineral
water (Perrier, France), which acts as a mildly alkaline
electrolyte. The ‘sponge method’ is remarkably effective
at enabling corrosion measurements to be done on
embroidered threads and attached bells, sequins, etc.,
without wetting the textile or leaving any tell-tale tide line,
which could occur from partial immersion. Calibration
voltages were developed using assayed jewellery items
and other known alloys belonging to the training team
and from personal jewellery. Thus, the characteristic
corrosion potentials (Ecorr) of white gold, 18 ct. and 9 ct.
gold alloys, sterling silver, sterling silver with 2% zinc,
metallic copper and brass voltages were all recorded
(Table 1). These measurements were reproducible within
METAL TEXTILE COMPOSITES AND IMPROVED TREATMENT OUTCOMES THROUGH APPLICATION OF IN-SITU CORROSION DATA
± 2 mV by touching the platinum electrode at new contact
points on the same object. Measurements of the surface
and solution pH were made using a Thermo Instruments
flat surface electrode connected to a Thermo Star A121
portable pH meter. Measurements of chloride ion activity
(Merck QuanTab test strips) established that many objects
would need desalination to remove the sweat and to rid
the textiles of acidity associated with the biodeterioration
of the fabrics.
Table 1. Corrosion potentials of metal objects for reference and with
composite Thai textiles
Object
E corr. volt vs.
Object
E corr. volt vs.
Ag/AgCl
white gold earring 0.262 silver Roman coin ring
white gold ring 0.210 rose gold ring
cleaned gold ring 0.182 Khon plated brass sequins
sulfided Ag sheet 0.161 Thai brass
18 ct. gold ring 0.154 Thai soutache threads
Sterling Ag 2% Zn 0.145 Thai Indian Ag
Results
Acidity and sweat
The decision-making tree (Figure 1) was used in group
discussions with the team at QSMT and this approach
enabled the treatment of both acidity and sweat as one
issue. There are specific deterioration problems asso-
ciated with residual sweat which adversely affect both
the textiles and the metals. Confidence-building exer-
cises involved pairs of QSMT staff working since their
natural shyness inhibited individual responses. During
the training workshop it was demonstrated that without
the prior removal of copper corrosion products before
reductive consolidation unsightly bright pink metallic
copper can be deposited on the surfaces. An example
of the ‘invisible copper’ problem was demonstrated by
following the gradual increase by 60 mV in the positive
direction in the Ecorr values of 18-carat gold rings in
diammonium citrate. Although there was no apparent
corrosion problem, the corrosion potential increased
from 0.044 to 0.104 volts vs. Ag/AgCl over 15 minutes as
the diammonium citrate complexed the ‘invisible’ layer of
cuprite. The conservators were amazed at the power that
had been given to them with just a couple of electrodes.
By the end of the program, the conservators at the QSMT
had their tables of Ecorr values and the correlation of the
values with known compositions as discussed above
389
(Table 1). They rapidly developed the confidence to
identify methods to remove copper corrosion products
and to bring about safe and cost-efficient treatments
which also removed chloride ions and the acidic bacterial
metabolites. During the last two days of the workshop,
the conservation staff reported the results of their tests
on the amount of salt (as measured by the chloride test
strips) and the surface pH of the textiles. When all the
results were tabulated, it became apparent that there
was a direct causal link between the amount of sweat (as
indicated by the chloride strips) and the observed acidity
of the textiles, with most of the data coming from the
Khon costumes (Figure 5).
Ag/AgCl
0.119 Bacteria degrade fabrics in the ambient conditions of high
0.095 relative humidity and temperature in Bangkok, with a
0.050 mean annual relative humidity (RH) of 76.0 ± 2.8% and
0.034 temperatures of 28.4 ± 1.9°C. The apparent sensitivity of
-0.040 the rates of deterioration on water activity was indicated
-0.089 by comparison with data from the much drier climate
in Western Australia, which had corresponding values
of RH of 51.8 ± 10.0% and temperature of 20.7 ± 5.7°C
(Figure 4). The correlations between the pH and chloride
were tested using linear regression analyses in Excel. The
square of the correlation coefficient, R2 values, provides
a guide as to the statistical validity of the fit to the equa-
tion. For the Thai Khon costumes, the R2 value for the
correlation between pH and chloride ion (ppm) was
high at 0.98 and Equation 1 describes this relationship.
pH Khon = 5.73 – 0.0007 [Cl] (1)
The slope of the pH vs. ppm [Cl] means that for every
1,000 ppm or gram of chloride in the textile, the pH will
fall by 0.7 units, or there will be a five-fold increase in
the concentration of hydrogen ions. The high degree of
correlation reported in Equation 1 is seen through the
small error of ± 0.05 in the pH intercept value (less than
1%) and an error of ± 0.00004 or 6% in the slope. This
is one of the first reported direct relationships between
the amount of sweat in textiles and the surface pH for
tropical textiles. Similar data had been collected from
a mid-19th-century pretiosa (precious) mitre from
the Benedictine community in New Norcia, Western
Australia (MacLeod and Car 2016). The regression anal-
ysis for the mitre (R2 0.99) showed that the pH decreased
by 0.3 pH per gram of chloride ions, i.e. the acidity of
the textiles increased by a factor of two times per gram
of chloride, as shown in Equation 2. The R2 value of 0.99
for the regression for the mitre resulted in only very
390  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
small errors in the intercept (5.69 ± 0.04) and the slope
(0.00030 ± 0.00003) for Equation 2.
pH front of mitre = 5.69 – 0.0003 [Cl]
The pH of the Khon and of the exterior of the mitre
at zero chloride were the same (i.e. the intercepts of
Equations 1 and 2 are within experimental errors),
which indicates that similar chemical processes had
been used in the manufacture of the silk into the textile
yarn. It is likely that the underlying reason why the
slope of the Thai Khon (Equation 1) was 2.3 times that
of the New Norcia silk mitre is the greater availability
of moisture and the warmer temperatures of Bangkok,
which make for easier microbiological reproduction
rates. An opportunity to test this hypothesis occurred
during a visit from New Zealand to Perth in the form
of an unwashed red cotton dress shirt which had been
3,0
Mitre NN
3,5
BKK Khon
pH Khon = 5.73 - 0.0007 [Cl]
R² = 0.98
4,0
surface pH
4,5
5,0
pH mitre = 5.67 - 0.0003[Cl]
R² = 0.99
5,5
6,0
0 1.000 2.000 3.000
ppm chloride
Figure 4. Plot of the relationship between the chloride concentration
and the pH of textiles in Bangkok and rural Western Australia
COMPOSITE OBJECTS
carried in luggage for four days. The slope of the pH vs.
chloride of the shirt after four days was the same slope
as the Khon textiles (Equation 1), but the pH of the
(2)
intercept was essentially neutral, at pH 7.70. After two
months in a sealed polyethylene container, containing
a beaker of saturated NaCl, the areas previously meas-
ured were reassessed. The intercept value of pH had
fallen from 7.7 to 6.1, indicating increased biodegra-
dation, but the sensitivity to chloride had decreased to
-0.0004 pH/ppm. During the two months in the plastic
container, the mean RH of 70.3 ± 2.3% and temperature
of 25.3 ± 2.4oC equates to a mean vapour pressure of
17.0 ± 0.6 mm of Hg (mercury). Using mean tempera-
ture and relative humidity data for Bangkok, the mean
vapour pressure of water was 16.8 ± 2.7, which is the
same as the red shirt data, while the New Norcia mean
vapour pressure was 9.8 ± 1.8 mm Hg. Work on rock art
sites in the Napier Ranges, in the Kimberley region of
Western Australia, has shown that in the wet season, the
mean pH fell from 7.8 ± 0.8 to 6.4 ± 0.9 due to increased
bacterial counts (MacLeod and Haydock 2008).
Removal of copper corrosion from corroded silver
alloys
Many of the silver embellishments showed a proliferation
of the characteristic blue-green hue of copper(II) corro-
sion products. Organic copper salts from sweat corrosion
4.000 5.000 6.000
are readily removed with methylated spirits, followed
by cleaning using neutral pH (6.5 ± 0.1) solutions of
diammonium citrate to complex and thereby remove any
‘invisible’ copper materials prior to the final dithionite
treatment. The effectiveness of such washing procedures
was demonstrated by placing aluminium foil (commer-
cial cooking grade) into the rinsing solution overnight.
Copper which had been removed from corroded silver
decorations cemented out on the aluminium foil to give
a dull grey-pink deposit and regenerated the complexing
solution in the process so that it could be used again,
while the contaminated aluminium foil could be simply
discarded.
Removal of silver tarnish using neutral dithionite
solutions
The efficacy of using weak solutions (0.1 wt%) of sodium
Figure 5. Detail
of Burmese dithionite in the neutral pH buffered solutions to remove
jacket before silver sulfide corrosion products has recently been
 conservation reported (MacLeod and Car 2014). The treatment for a
METAL TEXTILE COMPOSITES AND IMPROVED TREATMENT OUTCOMES THROUGH APPLICATION OF IN-SITU CORROSION DATA
richly decorated silver and silk jacquard-woven Burmese
frock coat was meant to replicate the successful work
done on the cope. However, it was determined that the
dithionite had partially oxidised to sodium sulfite in
the damp storage conditions of the supplying chemical
merchant. Although sulfite is a reducing agent, it is not
nearly as effective on silver sulfide as the dithionite. This
problem presented real challenges as fresh sodium dith-
ionite would take four weeks for delivery. This issue was
managed by a ten-fold increase in the concentration of
the reducing chemical (Figure 6). This approach demon-
strated that if the underlying chemistry of the treatments
is understood, it is possible to make adjustments on the
spot. An important issue is that when purchasing reactive
chemicals, such as reducing or oxidizing agents for a
workshop, it might be prudent to pre-ship materials by
sea from a known supplier six to ten months in advance,
rather than relying on local suppliers. Having been alerted
to such issues, trainees can do necessary tests on local
chemicals to ascertain their purity and freshness.
Figure 6. Upper
section of Burmese
jacket following
neutral dithionite
 treatment
Conclusion
The success of the workshop was measured by the engage-
ment of the local conservators with the practical exercises
and through their participation in all the learning experi-
ences. One of the keys to the success of this training was
the critical ratio of three students per trainer, especially
important with a non-scientifically trained group. At the
beginning of the program, the standard Thai conservation
approach was only to use known methods; if a new object
and set of conditions presented itself, then an alternative
item would be selected for the exhibition, despite it having
less engagement potential with a visitor. Through the
process of using pH, chloride and Ecorr measurements,
391
which were combined with detailed textile condition
analyses, the conservation team learned how to develop
a decision-making tree. This process increased the
confidence of the participants and resulted in greatly
improved outcomes. Through a combination of skills of
presenters all of the complex processes were able to be
explained in a culturally sensitive context. The cultural
and operational liaison, including program adjustments
as needed, solved many issues. Staff at the Queen Sirikit
Museum of Textiles became empowered and they were
much more confident in asking questions and defending
their proposed treatment programs in front of their
colleagues and teachers.
This workshop on developing skills in managing
composite metal and textile objects was designed for
non-scientifically trained conservators. Amongst the
challenges were the language differences and the strict
guidelines around royal textiles. Future workshops will be
able to develop from the established foundations. It was
good that the textile conservator from the USA had many
years of experience at the museum and had seen it grow
from conceptual plans to the realities of a functioning
organisation. Analytical skills developed daily in rapid
response, once study materials were fully comprehended.
The presentations, which were provided in advance to
the attendees, were a very important preparation tool.
Although the partial oxidation of dithionite to sulfite
resulted in much slower kinetics, it demonstrated ways
in which conservation problems can be managed, and
so the desired outcome of treating the Burmese dress
coat was achieved.
An unexpected bonus of the documentation on the pH
and salt levels in the composite metal-textile objects
was finding the direct link between increasing chloride
levels and the falling pH, which quantified tropical decay
mechanisms. The pH on the Khon costumes was found
to be much more sensitive than on sweat-stained textiles
from temperate environments in Western Australia. It is
likely that a combination of Bangkok’s higher temper-
atures and relative humidity leads to a much higher
absolute humidity that encourages bacterial reproduc-
tion and greater rates of chemical reactions involving
the depolymerisation of the natural polymers found in
the textiles. This data on the rates of biodeterioration
in different climates has provided an insight into the
specific effects of storage in tropical climates compared
with temperate areas. However, it should be noted that the
392  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

Thai textiles were cotton with some woollen elements, the
New Zealand shirt was cotton, whereas the New Norcia
mitre was silk. Plans for a sequential workshop devoted
to analysing storage conditions and providing sustainable
and tailored solutions for the storage of metallic textiles
in tropical climates are already in discussion.
Consolidation of the learning outcomes may also be
achieved through engagement with academic and tech-
nical staff from a local university who have the necessary
expertise to advise the conservators on practical scientific
problems. The installation of temperature and relative
humidity dataloggers in a number of locations would
provide much needed information to assist the trainers
in their tasks. Assessment of the microenvironment of
the storage conditions for private collections as well as
the QSMT royal collection would provide the necessary
data for checking on the universality of the impact of
perspiration leading to biodeterioration of the textiles
and corrosion of the metallic embellishments in tropical
climates.
Bridgland, art. 1306, 7 pp. Paris: International Council
of Museums.
MacLeod, I.D. and R.J. Car. 2016. Conservation of a
mid-19th-century pretiosa mitre from New Norcia,
Western Australia. AICCM Bulletin 37(2): 96–106.
Salim, S.S. 2015. What lies beneath – The safe display of
textiles and garments. On Conservation 3: 47–52.
Authors
Ian MacLeod pioneered in-situ corrosion measurements
on historic shipwrecks.
Rinske Car and Julia Brennan are highly experienced
textile conservators in private practice with more than
90 years’ experience in treating decaying fabrics.

References
Brennan, J.M., P. Saengsirikulchai, and P. Kingpra-
toommas. 2014. Thai textile conservation: Building
bridges regionally and culturally in a twenty-first-century
context. Studies in Conservation 59(1): S5–S8.
Degrigny, C., G. Guibert, S. Ramseyer, G. Rapp, and A.
Tarchini. 2011. Qualitative analysis of historic copper
alloy objects by measuring corrosion potential versus
time. In Metal 2010: Proceedings of the Interim Meeting
of the ICOM-CC Metals Working Group, Charleston,
South Carolina, 11–15 October 2010, eds. P. Mardikian,
C. Chemello, C. Watters, and P. Hull, 450–57. Clemson,
SC: Clemson University.
Janaway, R.C. and R.A.E. Coningham. 1995. A review of
archaeological textile evidence from South Asia. South
Asian Studies 11(1): 157–74.
MacLeod, I.D. and P. Haydock. 2008. Effects of water
vapour and rock substrates on the microclimates of
painted rock art surfaces and their impact on the pres-
ervation of the images. AICCM Bulletin 31: 66–86.
MacLeod, I.D. and R.J. Car. 2014. Conservation of a
mid-18th-century Italian embroidered ecclesiastical cope
and treatment of corroded metal threads using neutral
dithionite solutions. In ICOM-CC 17th Triennial Confer-
ence Preprints, Melbourne, 15–19 September 2014, ed. J.
Cloth of Metal (Indian Shamiana): Scientific
Characterisation of Metal Thread Corrosion
Wafaa Anwar Mohamed Mai Rifai
Conservation Department Conservation Department
Faculty of Archaeology, Cairo University Faculty of Archaeology, Cairo University
Giza, Egypt Giza, Egypt
wafaaanw@cu.edu.eg mairifai@hotmail.com
*Author for correspondence
Abstract
This paper reports the investigation of a 17th/18th-century Indian
shamiana, a cloth canopy decorated with metal threads. The
shamiana was stored at the museum of the Faculty of Applied
Arts, Egypt, and showed signs of metal tarnishing. Environmental
scanning electron microscopy equipped with energy-dispersive
x-ray spectrometry (ESEM-EDS) was used to investigate sectioned
samples of the tarnished spun metal threads, essentially a metal
strip upon a core yarn, to determine the metal composition and
identify the yarn and ground fabric. Polarised light microscopy
(PLM) revealed the morphology of the metal threads and aspects
of their corrosion, while atomic absorption spectrometry (AAS)
confirmed the bulk metal composition. Non-destructive x-ray
diffraction (XRD) was used to analyse the corrosion products on
the tarnished spun metal strips. ESEM-EDS results of the sectioned
samples identified the metal as a gilded copper-based alloy with
minor ratios of silver, gold and other traces, and AAS confirmed
this composition. Tiny ratios of chloride and sulfur contributed
Introduction
The origins of textiles decorated with metal threads are
attributed to royal families due to the use of gold, silver
and copper in the weaving process. Embroidery varied
in its forms over time, whether woven as basic units in
the textile, used for embroidery purposes after weaving,
or both. Metal embroidery also varied, with metal
strips/wires being embroidered directly into the textile
or wrapped around a core yarn (cellulosic or proteina-
ceous) which might or might not be dyed. Metal strips/
wires spun around the core yarn could also be attached
to other organic substrates (e.g. leather, parchment and
gut membrane) and then embroidered into the textile,
a decorative variety known as membrane metal threads
(Harris 1993).
The earliest surviving examples of embroidery are from
ancient Egypt and were found in the tombs of Thutmose
IV (1364 BC) and Tutankhamun (1334–25 BC), though
Fatmaa El-Zahraa Sadat*
Conservation Department
Faculty of Archaeology, Cairo University
Giza, Egypt
fatmasadat@cu.edu.eg
to the corrosion. Micrographs of the spun strip revealed the core
yarn to be cellulosic cotton fibre. ESEM-EDS micrographs of the
etched samples showed remnants of the double-sided gilding of
the strip. Micrographs record evidence of the tool marks from the
rollers and drawing processes used to fabricate the metal strip. This
type of copper-based strip shares similarities with the Nuremberg
thread known in Germany in the 18th century. EDS line analysis of
gold, silver and copper distribution through the sectioned metal
strip indicates variation due to surface enrichment. XRD confirmed
the presence of covellite (cupric sulfide) as black spots and of
paratacamite as slightly greenish spots. ESEM aided identification
of the ground fabric, which was made of cotton and silk, as an
Indian mashru fabric.
Keywords
Indian, mashru, double-sided gilding, enrichment, tarnish,
metal thread
it is presumed that their origins are even earlier. Embroi-
dery also has a long history in China, where embroi-
dered silk and gauze dating to the 4th century BC have
been found (Harris 1993). Metals have long been used
as raw materials in textiles, especially for magnificent
objects with symbolic meaning. In the Middle East, the
fabrication of metal filaments dates to the 3rd century
BC (Lee 1992). Gold, silver and combinations of both
have been used most frequently as textile adornments,
although examples of alloys of baser metals have also been
observed (Hoke and Petrascheck-Heim 1977, Darrah
1989, Indictor et al. 1989, Járó, Gál and Tóth 2000).
As important as preventive and remedial conservation,
investigative conservation is concerned with the mecha-
nisms of deterioration, characterisation of raw materials
and methods of technology and manufacture. These
may be studied using various analytical and micros-
394  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
copy techniques to provide information that will assist
in preventive and remedial conservation. This paper
reports the analytical investigation of a 17th/18th-century
metal-decorated shamiana (an awning or canopy) which
was probably used to cover royal thrones in India. The
study aimed to identify the composition, manufacture
and corrosion of the gilded metal thread and the nature
of the shamiana’s ground fabric using a range of comple-
mentary analytical techniques.
Materials and methods
Sample description
The shamiana textile was stored at the museum of the
Faculty of Applied Arts, Helwan University, in Cairo,
Egypt. It was described in the museum’s records as part
of a pillow cover, thought to have been produced in
Bokhara (present Bukhara, Uzbekistan). Unfortunately,
the textile has never been artistically described or scien-
tifically dated. Provenance and dating studies had been
performed based upon the ornamental patterns and simi-
larity (Figure 1). The shamiana was sampled for analysis.
Figure 1. General view of the studied decorated textile (top left); detail
of the decorative tarnished metal threads (top right; middle left and
right); USB digital imaging of the completely (dull black) and partially
tarnished metal threads (bottom)
Examination and analysis
Metallurgical microscopy, polarised light microscopy
(PLM) and scanning electron microscopy (SEM) were
COMPOSITE OBJECTS
used as investigative tools in combination with analysis
by energy-dispersive x-ray spectroscopy (EDS), x-ray
diffraction (XRD) and atomic absorption spectrometry
(AAS) to produce detailed data about the morphology,
fabrication technique, core yarn, metal composition,
corrosion products and surface alterations of the metal
decoration on the shamiana.
Samples were taken from the brittle fragments of the
decorated textile, secured in fixation clips, mounted,
ground and polished. The polished samples were etched
using alcoholic ferric chloride (70 ml of ethanol, 10 ml of
HCl, 2 g of FeCl2) from 2 seconds to 1 minute, then rinsed
with distilled water and dried with ethanol. They were
then studied using PLM with a Zeiss Axio A1 portable
digital microscope at 200× magnification.
Environmental SEM (ESEM) used a Philips XL30 to
examine samples at low vacuum (0.8 torr). Backscattered
electron (BSE) images of the cross sections were produced
under high vacuum with a 25 kV acceleration voltage,
with the backscatter detector at 5 mm eucentric working
distance and a spot size of 0.7 mm using a JEOL JSM
S400LV linked to an Oxford ISIS EDX system.
In preparation for AAS analysis, a sample of the core
yarn was separated gently from the spun metal strip,
whose corrosion was then reduced in volume. The metal
strips were then digested in concentrated nitric acid
with deionised water (1:1) and analysed using an atomic
absorption spectrometer (EC Model AAS 4129) to give
the bulk composition of the metal/alloy.
Non-destructive XRD using a PANalytical Empyrean
CuKα diffractometer with a wavelength of 1.54 µm,
3D pixel detector, fixed divergence module, an angle
range of 5°–90°, a step size of 0.0131 and mirror optics
was carried out at the National Institute for Standards,
Egypt. Corroded metal strips, separated from the core
yarn inside, were analysed. The x-ray beam was directed
at the thin corrosion spots on the metal strip surface to
aid identification of the existing corrosion products, as
separating the corrosion products from the metal was
challenging.
Results
Morphology and composition of the metal strip
The morphology of the wrapping direction of the metal
strip (cabled yarn) upon the core yarn can be observed
CLOTH OF METAL (INDIAN SHAMIANA): SCIENTIFIC CHARACTERISATION OF METAL THREAD CORROSION 395

on the PLM images. The remaining gilding layer was Core yarn
shown on the etched sample of the sectioned metal
Examination of the PLM and ESEM micrographs of a
strip (Figure 2). AAS helped to identify the bulk metal
sectioned sample of the spun metal strip upon a core
composition of the metal strip as a copper (Cu) alloy
yarn allowed identification of the core yarn as cotton
with silver (Ag), gold (Au), zinc (Zn) and lead (Pb)
(Figure  4). The characteristic kidney or lima bean
(Table 1).
shape with the collapsed inner hollow core is shown
clearly.

Figure 4. PLM imaging of the etched spun metal strip showing the core
yarn inside (left); ESEM micrographs of the kidney- or lima bean-shaped
Figure 2. Light microscopy image showing the spun metal thread as
core yarn characteristic of cotton (right)
cabled yarn (2 S and Z) and the golden layer (left: 120× magnification);
PLM image of the etched section of the metal strip showing remaining
gilding on both sides (right: 340× magnification) Corrosion aspects and products
PLM images show green and black corrosion products
Table 1. AAS analysis results of the metal strip
covering the metal strip (Figure 5). ESEM micrographs
Analysed elements Composition (wt%) reveal pitting corrosion on the surface (Figure 6). In
Cu 87.50% addition, EDS line scans were performed on three points
Ag 1.8% (surface, middle and core) of the sectioned metal strip
Au 0.063% to show variation in ratios of Cu, Ag and Au (Figures 6
Pb 0.023 and 7, Table 2). Non-destructive XRD was used to analyse
Zn 0.016 the corroded metal strips and to identify the corrosion
products present (Figure 8).

Manufacturing method of the metal strip

The manufacturing techniques of the metal strip are
identified as casting, drawing and rolling methods. The
micrographs confirmed the presence of smooth rounded
edges and long striation lines similar to the cast, drawn
and rolled (CDR) type (Figure 3).

Figure 5. Micrographs of the spun metal strip showing the broken core
yarn (top); USB image (200× magnification) showing black corrosion
Figure 3. Long striation lines due to the manufacturing of the metal strips covering the metal threads (bottom left); PLM image of green corrosion
(left); the width of the metal strip was measured to be 306 µm (right) covering the metal strip (bottom right)
396  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

Weaving structure of the ground fabric

The ground was examined by USB digital microscopy
(200×) to show the structure of different yarns of length-
wise and crosswise directions. ESEM micrographs helped
to identify these yarns in their longitudinal sections,
showing cotton fibres for lengthwise yarns as convolu-
tions/twists and silk fibres for crosswise yarns as smooth
tubes (Figure 9).

Figure 6. Line ESEM micrographs of the sectioned spun metal strip and
the corresponding EDS spectra of the surface, middle and core of the
strip. Thickness of the metal strip was measured to be 6.08 µm

Figure 7. Plot
chart showing
variation in
gold, silver
and copper Figure 9. USB image of the ground textile (top: 200× magnification);
ratios on the ESEM of the longitudinal section of the silk yarn as smooth tubes (middle:
 metal strip 1000× magnification) and of the longitudinal section of the cotton as
twists/convolutions (bottom: 800× magnification)

Table 2. EDS comparative analysis results of three spots on the surface,

Discussion
middle and core of the sectioned spun strip sample

Element The studied object was attributed to the Indian Mughal

Analysed spot era dating to about the 17th/18th century AD (11th–12th
Au Ag Cu Cl S
Spot A 7.6 37.9 52.1 1.1 1.0 century AH) based on its ornamental patterns. It was
Spot B 0.7 12.3 84.0 1.6 1.2 suggested that the embroidered textile was probably used
Spot C 0.6 2.1 94.2 0.5 2.3 as a table cloth, a lining or an awning to cover thrones
and cushions.

The EDS and AAS analysis results showed that the

Figure 8. XRD pattern of the corrosion products on the spun metal strip
metal strip is a copper-based alloy with Ag, Au, Zn and
Pb as trace elements. It is thought that the remaining
11% ratio in the AAS results can be attributed to the
corrosion products of chloride (Cl) and sulfur (S),
which could not be removed during sample preparation.
The minute ratios of minor and trace elements such as
iron (Fe), calcium (Ca), aluminium (Al), silicon (Si),
etc., were interpreted as additions by manufacturers to
improve the properties of the alloy. For example, lead
was deliberately added to increase malleability of the
alloy. Some of these elements, such as Ca and Si, may
derive from contaminated ores. Tiny ratios of Cl and
CLOTH OF METAL (INDIAN SHAMIANA): SCIENTIFIC CHARACTERISATION OF METAL THREAD CORROSION
S were also detected and are likely to be present due to
the corrosion.
The results of the study prove the double-sided gilding
of the strips. It is likely that the gilding was applied
by hammering as no evidence for mercury or other
welding materials was found. Due to the corrosion and
the hammered gilding technique, most of the gilding
layers were missing (Peranteau, Carlson and Mass 2005).
Surface enrichment (mise-en-couleur) was detected on
the metal strips, probably because of the preferential
leaching effect of salts removing baser metals from the
surface. This caused an increase in Au content from the
core to the surface and a corresponding decrease in Cu
from the core to the surface.
Covellite, a cupric sulfide (CuS), was responsible for the
black spots covering most of the decorative metalwork.
Paratacamite, a basic cupric chloride (Cu2(OH3)Cl),
represents the most stable phase of copper chlorides and
occurs as chalky, light greenish spots. The presence of
paratacamite can probably be attributed to inappropriate
storage conditions and materials in the museum environ-
ment. Sulfides on copper alloys cause black spots and may
result from the presence of airborne hydrogen sulfide,
either from industrial pollutants or display materials
(Ballard et al. 1989, Indictor and Ballard 1989).
The metal strips were spun upon a cotton core, iden-
tified by its kidney/lima bean structure, in both S and
Z directions to form the final Z twist. This is known as
cabled yarn. The technology used to fabricate the metal
strip (flattened wire) was identified as the CDR tech-
nique which was used in the early days of silver gilding
(Járó, Gál and Tóth 2000). Drawn wire was flattened
into a strip by hammering or passing it between rollers.
The latter technique was proposed to have followed
the introduction of drawplates dated to the 18th/19th
century (Balázsy and Eastop 1998, 129). The resulting
double-sided metal strip is similar to a metal thread that
was known as Nuremberg type.
The Nuremberg thread was a new, cheaper metal thread
made in Germany from copper, thinly silvered or gilt. It
was cast, drawn and rolled in the same manner as silver
and silver-gilt thread (Darrah 1989, Hacke, Carr and
Brown 2004). It is likely that it was transferred over time
from Germany to India, where this type of decoration
was adopted by the Mughal during their reign. This
type of thread was fabricated to economise on the cost
397
of using bulk, pure noble metals in textile decoration by
using thin gilding on less noble metals/alloys, such as Cu.
The etched sample showed remains of a gilding layer on
both sides. A fine grain structure derives from the quick
cooling process, which did not allow sufficient time for
the grains to grow fully during casting.
It was found that both the lengthwise yarn and the core
yarn are cellulosic fibres, identified as cotton by ESEM
examination, while the crosswise yarn of the ground
fabric was silk. This cotton-silk ground was well known
to Indian embroiderers and was called mashru, meaning
‘permitted’ in Persian and ‘mixed’ in Arabic. It was
woven between the 18th and 19th centuries and was an
old traditional craft in India frequently traded to Gulf
countries, Turkey and the Ottoman Empire (Gillow and
Barnard 1991, Khatwani and Khawani 2017).
Conclusion
In this study, a shamiana – a historic textile with gilded
metal threads most probably used to cover thrones
and tents in India – was investigated. A combination
of analytical techniques were employed to identify and
characterise the gilded metal threads, including their
metal composition, core yarn and corrosion products.
The metal strip forming the thread is a copper-based
alloy with minor elements being gold, silver, zinc
and lead. It was gilded on both sides, most probably
by mechanical methods, with most gilding now lost.
The copper alloy was cast, drawn and rolled as flat-
tened wire, and gilded to give a double-sided gilt. The
resulting thread is similar to the Nuremberg type made
in Germany in the 18th century using a technique that
was transferred to India over time. The metal strip is
shown to be wrapped around the cotton core yarn. The
metal thread exhibits black spots due to the reaction
of sulfur pollutants with the copper and silver. Active
corrosion has been identified as paratacamite, a basic
cupric chloride, which appears as light greenish spots.
Results of the study have also revealed that metal
decoration was applied to a mixed ground woven from
cotton and silk.
References
Balázsy, Á.T. 1989. Deterioration aspects of fibres in
different chemical actions. In Conservation of Metals:
Problems in the Treatment of Metal-Organic and Metal-In-
organic Composite Objects. International Restorer Seminar,
398  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Veszprém, Hungary, 1–10 July 1989, ed. M. Járó, 82.
Budapest: Központi Muzemi Igazgatóság.
Balázsy, Á.T. and D. Eastop. 1998. Chemical principles of
textile conservation, 128. Oxford: Butterworth-Heine-
mann.
Darrah, J.A. 1989. The microscopical and analytical
examination of three types of metal thread. In Conserva-
tion of Metals: Problems in the Treatment of Metal-Organic
and Metal-Inorganic Composite Objects. International
Restorer Seminar, Veszprém, Hungary, 1–10 July 1989, ed.
M. Járó, 53–63. Budapest: Központi Muzemi Igazgatóság.
Gillow, J. and N. Barnard. 1991. Traditional Indian textiles.
London: Thames and Hudson.
Hacke, A.-M, C.M. Carr, and A. Brown. 2004. Charac-
terization of metal threads in Renaissance tapestries. In
Metal 04: Proceedings of the International Conference on
Metals Conservation, Canberra, Australia, 4–8 October
2004, eds. J. Ashton and D. Hallam 415–25. Canberra:
National Museum of Australia.
Harris, J., ed. 1993. 5000 years of textiles, 9, 5, 31. London:
British Museum Press.
Hoke, E. and I. Petrascheck-Heim. 1977. Microprobe
analysis of gilded silver threads from mediaeval textiles.
Studies in Conservation 22(2): 49–62.
Indictor, N. and M.W. Ballard. 1989. The effects of aging
on textiles that contain metal: Implications for analysis.
In Conservation of Metals: Problems in the Treatment of
Metal-Organic and Metal-Inorganic Composite Objects.
International Restorer Seminar, Veszprém, Hungary,
1–10 July 1989, ed. M. Járó, 76–82. Budapest: Központi
Muzemi Igazgatóság.
Indictor, N., R.J. Koestler, M. Wypyski, and A.E. Wardwell.
1989. Metal threads made of proteinaceous substrates
examined by scanning electron microscopy-energy
dispersive X-ray spectrometry. Studies in Conservation
34(4): 171–82.
Járó, M., T. Gál, and A. Tóth. 2000. The characterization
and deterioration of modern metallic threads. Studies in
Conservation 45(2): 95–105.
Khatwani, P. and P. Khawani. 2017. Mashru fabrics: Its
production technology and scope for further develop-
ment. International Journal of Science Technology and
Management 6(5): 413–16.
Peranteau, A., J. Carlson, and J. Mass. 2005. The tech-
nical analysis of two pieces of 17th-century English
COMPOSITE OBJECTS
embroidery. In Scientific analysis of ancient and historic
textiles: Informing preservation, display, and interpreta-
tion. Postprints of the AHRC Research Centre for Textile
Conservation and Textile Studies 1st Annual Conference,
13–15 July 2004, eds. P. Wyeth and R.C. Janaway, 24–30.
London: Archetype Publications.
Authors
Wafaa Anwar Mohamed, PhD, is professor of metal
conservation at the Conservation Department of the
Faculty of Archaeology at Cairo University. Her principal
interests are the conservation of metal objects, examina-
tion of metals, composite artefacts and metal technologies
in ancient Egypt. She has taught conservation of metals
to undergraduate and graduate conservation students
and has supervised master and doctoral conservation
research since 2000.
Mai Mohamed Rifai, PhD, is assistant professor of
metal conservation at the Conservation Department
of the Faculty of Archaeology at Cairo University. Her
research interests lie in the analysis and conservation
of metal artefacts with an emphasis on copper-based
alloys. Her current research focuses on green corrosion
inhibitors, protective coatings for outdoor metal and
daguerreotypes.
Fatmaa El Zahraa Sadat, PhD, is lecturer of metal
conservation and composite metal-textile objects at the
Conservation Department of the Faculty of Archaeology
at Cairo University. Her master’s thesis focused on the
analysis and conservation of metal-textile composite arte-
facts and testing of non-conventional cleaning methods
such as laser and ultrasonic dental scalers. Her PhD
dissertation addresses the optimisation of the cleaning
parameters of ultrasonic scaler and laser ablation to be
applied on multiform metal embroidery composed of
multi-organic materials with corroded metal.
Physical and Chemical Interactions between Metallic
Instruments and Plastic Threads
Marie-Anne Loeper-Attia
Cité de la musique – Philharmonie de Paris
Centre de recherche sur la conservation (CRC)
Muséum national d’histoire naturelle (MNHN – CNRS)
Ministère de la Culture
Paris, France
maloeperattia@cite-musique.fr
Abstract
The collections at the Musée de la musique in Paris include several
hundred brass, silver and silver-plated wind instruments that are
often exhibited, loaned or played. For the museum’s opening in
1997, special mountings were proposed and designed by Franck
Hammoutène and various types of thread were used to hold
the instruments by their cylindrical parts. These threads were
regularly untied and re-tied, for instance, for loans to temporary
exhibitions, studies and when the galleries were refurbished
in 2009. They were also used for temporary reinforcement
(instead of adhesives), mounting (sometimes in combination
with foams) and for attaching labels. Various types of thread
were used – rope, fishing line and sewing thread – composed of
nylon, polyester, polyamide or polyethylene. The choice between
these four materials depended on criteria such as the weight and
nature of the instrument, colour of the surface it is attached to
and the friction or tractive forces involved in the instrument’s
use. Over the years, various types of surface damage to the
Introduction
In a museum setting, objects are subjected to a number
of factors, such as climate, lighting and display method.
The items in question, instruments, can be placed
directly on a surface with a screen material that avoids
any contamination with the support. The presentation
can be more constrained if the instrument is mounted.
The display method must be adapted to the materials
the item is made of, taking into account any observed
degradation or fragility. A material’s corrosion resistance,
especially in the case of a metallic material, is not merely
an intrinsic property. The amount of corrosion depends
greatly on the environment. It is therefore essential to set
up sound storage and display protocols that correspond to
an item’s specific needs. This paper discusses the support
and display of musical instruments in a museum setting,
specifically at the Musée de la musique – Philharmonie
de Paris at its opening in 1997 and during the various
metallic instruments have been noticed, including mechanical
damage, such as scratches, and chemical damage, such as
localised corrosion. After a short description of the collection of
instruments and the degradation observed, this paper discusses
the solutions now being applied at the Musée de la musique. The
research addresses four points: 1) identification of the mechanical
and chemical damage observed using visual and microscopic
observations and XRF in situ; 2) the required specifications for
choosing a new type of thread; 3) the types of plastic thread
chosen and their various chemical and physical characteristics;
and 4) corrosion tests in a humid chamber and Oddy tests with
metallic sheets in contact with various types of plastic thread. The
new solutions found will be introduced into the framework of the
museum’s global preventive conservation protocol.
Keywords
brass, exhibition, plastic thread, polypropylene
stages of work in the exhibition areas, taking into account
the constraints inherent to the materials, the location and
the museographical approach.
Description of the museum and the particulari-
ty of the collections
The Institut national de musique was created by decree on
8 November 1793 to ‘[train] the artists needed to perform
on national holidays’ and included an ‘instrument room’
containing pieces confiscated from ‘emigrants’. Under the
terms of the Convention of 17 August 1795, the Institut
became the Conservatoire de musique de Paris and
included a collection of ‘antique or foreign instruments,
and those of our own customs, that can in their perfec-
tion serve as models’. Of the 316 items transferred to the
Conservatoire in 1795, only a dozen remain today. In 1861,
400  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
the French government purchased for the Conservatoire
the instrument collection of Louis Clapisson, a composer
and member of the Institut. Two rooms were officially
opened and, for the first time, a collection of instruments
was made accessible to the public in Paris, marking the
birth of the instrument museum. Clapisson was the first
curator, followed by the composer Hector Berlioz and
then Gustave Chouquet, who significantly developed the
collections. The third major milestone occurred around
1960, thanks to the Countess Geneviève de Chambure
(Getreau 1996). During this period, the role and purpose
of a future Musée de la musique were defined in terms of
conservation, restoration, inventory and documentation
(Figure 1).
Figure 1. Musée
instrumental in 1884.
Engraving by Eugène
Tradoné for the
second edition of
Chouquet’s catalogue

Figure 2. Display case
for 19th-century
instruments
 Figures 3 and 4. Instrument mounting in a display case
In 1978, with the founding of the Cité de la musique, the
decision was made to hand the Conservatoire’s national
collections over to the state, leading to the creation of
the Musée de la musique in 1995. An acquisition and
COMPOSITE OBJECTS
restoration campaign was launched and plans were made
for a new museum, which officially opened at this new
site in 1997. A redesign in 2009 introduced the latest
museographical changes.
The major themes covered are the 17th century and
opera, the music salon of the 18th century, the great
inventions of the 19th century, the new sounds of the
20th century and, finally, the musical traditions of other
continents (Figure 2).
Instrument mounting and display at the Musée
de la musique – Philharmonie de Paris
The mounts created for the museum when it opened in
1997 were designed by architect Frank Hammoutène.
Today, the museum is still required and obliged to use
the same type of mount, based on the principle that the
instruments are (generally) presented vertically or in
playing position. A metal armature holds them up with
half-rings covered in high-density polyethylene foam
and threads encircling the instrument. These museo-
graphical constraints were included in the convention
signed between Frank Hammoutène and the Musée de
la musique. It explains why this display method is imper-
ative and why it must continue to be used even today.
Securing the instrument in a vertical position requires
adjusted grips and a mounting perfectly adapted to the
instrument. This often requires fastidious adjustments
(Figures 3 and 4). There were, however, several excep-
tions to this rule when an instrument’s configuration
was particularly unusual and could not be made to fit
the mounts designed by Frank Hammoutène.

Materials used and types of damage observed
To secure each instrument to its attachment ring, special
threads were chosen, i.e. woven fishing line in two colours
PHYSICAL AND CHEMICAL INTERACTIONS BETWEEN METALLIC INSTRUMENTS AND PLASTIC THREADS 401

Table 1. Composition of different fibres and threads

Woven fibres and threads Non-woven fibres and threads

Composition Commercial name Composition commercial name
cotton aromatic polyamide or aramid (PAA) Kevlar, Technora, Twaron
non-aromatic polyamide (PA) Rilsan, Technyl Nylon aromatic polyester Vectran
non-aromatic saturated polyester (PET) Dacron, Tergal, Tetoron carbon
non-aromatic saturated polyester (PEN) Pentex
polypropylene Prolene
high-density polyethylene (HDPE) Spectra, Dyneema, Nacrylan

(grey and brown). The colour was selected to match the
object. The brands used were either Caperlan thread (by
Decathlon) or Nacrylan thread (by Tortue). Both types
of fishing line are woven and made of Dyneema, the
main component of which is high-density polyethylene
(HDPE). The other types of fishing line are polyam-
ide-based (nylon) or fluorocarbon. In addition to the
main component, the threads also contain micro-pig-
ments or abrasives that make the thread visible in water
and ensure good grip in the reel during fishing. Unfor-
tunately, the exact nature and concentration of these
components were not provided by the manufacturers.
Non-woven threads such as nylon (polyamide) or hemp
(for the labels) were also sometimes used on the instru-
ments. Table 1 indicates the type of fibres and their most
common commercial names.
of use (producers of sound) and art objects (Barclay
2004). During remodelling work at the museum, all of
the brass instruments (which are in fact instruments
made of brass or silver- or gold-plated brass such as
trumpets, saxophones and trombones) were removed
from their mounts to be cleaned (Loeper-Attia 2007). It
was then realised that a large number of the instruments
were scratched and/or corroded where they touched the
supporting threads. This damage was only visible on the
metal objects; other instruments made from organic
materials showed no sign of oxidation. Instruments
with labels attached by old thread, as could be seen, also
presented surface corrosion (Figures 5–7).

These threads, especially the woven ones, were chosen

essentially for their strength. All of the data came from
the supplier Kanirope (Table 2).

Table 2. Flexural strength of different threads

Flexural strength
Composition Type of fibre
daN (kg)
nylon (polyamide) 8-strand, 1 mm diameter 30
hemp 8-strand, 1 mm diameter 10 Figure 5. Silver corrosion due to a Nacrylan thread
cotton 8-strand, 1 mm diameter 15
Dyneema (HDPE) 12-strand, 1 mm diameter 195

Even though the values for Dyneema are given for

12-strand weaves rather than the 8-strand weaves of the
others, it is clear that this fibre is much stronger. Given
the low values of hemp and cotton, it is suitable that they
be used only to attach labels.

These threads were regularly cut and removed when

an instrument was taken out of its display to be loaned,
restored, consulted by researchers or played by musi- Figure 6. Brass corrosion
cians. Musical instruments are at once technical objects due to a hemp thread on a
(products of science: acoustics, mechanics, etc.), objects  trumpet (E.734)
402  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 7. Stripe on a brass instrument due to a Caperlan thread
These traces were stubborn and often impossible to
remove completely with chemical processes or mechan-
ical cleaning with abrasive powders. It was decided not
to attempt to remove the scratches observed because it
would have required abrading a significant amount of
material. Moreover, the intervention would have been
clearly visible on these larger, polished areas and it
would have been necessary to clean the entire surface.
The Museum’s team was therefore faced with the need to
find an alternative to these threads to ensure the metal
instruments were properly preserved.
Looking for other materials
Surveying other museums confirmed the regular use of
nylon threads, fishing line and cotton threads to secure
objects to mounts.
In order to choose the threads best suited to the conserva-
tion/restoration of cultural heritage objects, particularly
those made of metal, the team first established appro-
priate specifications and then sought suitable products
on the market and examined their composition. The
threads selected had to meet the following specifications:
• product traceability
• chemical stability
• aesthetic unobtrusiveness
• no chemical or physical interaction with the contact
material
• appropriate tensile strength
• zero hygroscopicity
• zero to low elongation
• no overt sensitivity to UV
• high melting temperature
COMPOSITE OBJECTS
The threads are made from crystalline polymer fibres.
These polymers are elongated macromolecules arranged
in straight, parallel filaments. Their hydrogen bonds hold
the chains together and partly explain their high tensile
strength. The particular positioning of the macromo-
lecular chains occurs after the thread drawing phase
following extrusion (Massardier 2001).
It should be noted that a polymer in its raw form gener-
ally lacks the properties required for the application
in question. Additives need to be added to the plastic
(Cavalier 2013) and these primarily affect the aesthetics,
chemical stability, UV resistance, heat resistance, cost
and plasticity. One function of additives is also to slow
the polymers’ oxidation, which causes yellowing, poten-
tial loss of transparency, the appearance of crazing and
affects the mechanical properties by reducing potential
flexibility and tensile strength. This oxidation is acceler-
ated by temperature and UV rays. Additives trap formed
radicals by reacting with them and/or absorbing UV
energy. Common additives employed are inert compos-
ites, generally mineral-based, such as natural or precip-
itated calcium carbonate, talc and china clay, graphite,
molybdenum disulfide and polytetrafluoroethylene.
Phosphate-, phthalate- or benzoate-based plasticisers
may also be present (Verrier 2013). Table 3 sets out the
data collected on the primary fibres selected during the
first phase of the investigation. All the values given and
protocols of measurement are taken from the Techniques
de l’ingénieur or mentioned reviews (see References).
It was decided to take into account only certain phys-
ical characteristics of these polymers, those that struck
the team as most important to the specifications being
drafted (Dossogne et al. 2010).
• Modulus of elasticity: the tensile stress required to
obtain a given elongation. This is a value intrinsic to
the material, defined as the ratio of stress to the elastic
deformation caused by this stress. It is often called
Young’s modulus and its unit is the Mpa (megapascal).
• Breaking strength: also called stiffness or tensile
strength. It is the resistance to deformation under the
action of continuous stress and therefore corresponds
to the tensile force under which a material reaches its
breaking point. It must not be confused with stretch
resistance: a material or fibre can be highly elastic yet
have very high tensile strength (late rupture).
• Loss of flexion or resistance to flexural stress: loss of
stiffness after 60 cycles of 180° folds.
PHYSICAL AND CHEMICAL INTERACTIONS BETWEEN METALLIC INSTRUMENTS AND PLASTIC THREADS 403

Table 3. Mechanical characteristics of different fibres

Elasticity Flexural strength Flexion resistance Resistance to UV Elongation T fusion

Fibres
modulus (MPa) (g/denier) (60 cycles) (months) % (°C)
nylon 45 9.5 0 3 to 4 13 220
PEN 250 10.2 0 6 6 250
HDPE 1250 33 0 6 to 7 5 150
aramid 550 to 950 24 to 29 7 to 25% 2 to 3 1.5 to 4 NR
Vectran 510 23 15% 1 to 2 2 330
carbon 1350 60 22% without effect 1.5 NR
polypropylene 1200 to 1700 31 to 42 0 6 1 to 6 160 to 170

• UV resistance: this is the resistance of the fibre to were created, i.e. thread and metal in direct contact,
breaking under natural light, measured in months since the damage observed was only the result of direct
(Bohé et al. 2016). contact and not a gaseous effect. The metal support
• Elongation at break or percent elongation: this is a chosen was a form of brass, similar to the alloys found
non-quantified characteristic of materials and defines in brass instruments, i.e. with a mass percentage of
a material’s ability to stretch to its breaking point when around 30% zinc. This content is close to that of alloys
under tensile stress. now referred to as ‘cartridge brass’, characterised by
• Melting temperature. better ductility and higher corrosion resistance. Indeed,
brasses containing up to 35% zinc are single-phase and
The results clearly show the significant differences in
reputed for their ductility and malleability when cold.
these fibres’ behaviour. For example, polyethylene (PEN
Above 35%, hardness increases and ductility when cold
or HDPE) has a very high breaking strength, displays no
is not as good, although malleability when hot is excel-
loss of strength due to bending and has good UV resist-
lent. The maximum threshold for zinc content in brass
ance. Nylon is one of the most elastic fibres (it has high
is 42%, beyond which the alloys become too fragile. Two
elongation at break), which allows it to absorb load peaks.
tubes and a rectangular-section rod were cleaned with
Indeed, the higher a material’s modulus of elasticity, the
abrasive powder (calcium carbonate), then rinsed and
stiffer the material is; in other words, under equal stresses,
degreased with demineralised water, ethanol, acetone
a material with a high modulus of elasticity will stretch
and cyclohexane. Various threads were then knotted to
less than a material with a low modulus of elasticity.
this assembly as shown on one of the tubes (Figure 8).
For instrument-mounting purposes, it is therefore pref-
erable to use a material with a high modulus of elasticity
to ensure a support that is always adapted and adjusted.
The loss of flexion must be low so that it can withstand
bending and folding. The melting temperature must be
high, as must be the breaking strength. It must also have
good UV resistance. In the tests carried out, focus was
therefore placed on polyethylene fibres – both with and
without additives (Nacrylan, Dynema).

Comparative tests
Ageing tests
The goal was to demonstrate proper conservation of
the threads chosen in a harsh environment in contact
with metal. Rather than conducting an Oddy test which
evaluates the gaseous effect of a composite on a metal
plate, conditions closer to those in the museum setting
Figure 8. Brass tube with different threads
Each thread was previously weighed to evaluate whether
there was any weight loss or gain after the experiment, to
help quantify any potential transfer or loss of material.
Each thread was weighed four times using a Mettler
Toledo precision scale (0.0001 g accuracy) before and
404  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

after placing in the chamber. The tubes were placed in a Table 4. Weight variations and effects of different threads on brass
climatic chamber for 30 days. The thermal cycles lasted after oven testing

12 hours with a temperature variation of 18°C–22°C and Weight Effect

relative humidity (RH) of 50%–80%. Commercial name Chemical family variation on
(%) metal
These values corresponded to the possible variations cotton cellulose X
inside the museum (one of the stores had suffered a nylon polyamide 2 X
climatic incident due to a failure in its air conditioning Prolene polypropylene 0.02
system and the RH had risen to 80%). After removal from Spectra high-density polyethylene 0.01 X
the chamber, the tubes were examined and each wire was Spectra 148 high-density polyethylene 0.03 X
weighed. The weight variations were all consistent with Nacrylan high-density polyethylene 0.03 X
the values indicated in the literature and, on average, Dyneema high-density polyethylene 0.03 X
were below 0.05%, except for the Nylon which was Vectran aromatic polyester 0.03 X
slightly more hygroscopic. Observation with the naked polypropylene polypropylene 0.02
eye and a binocular magnifier revealed that every thread polyester 1100/2/3 polyester 0.02 X
tested caused surface corrosion on the metal, except the
polypropylene-based thread.
X-ray fluorescence spectroscopy of corrosion
In phase two, tests were repeated with a thread made of products in ageing tests
polypropylene (the main component of Prolene). This
In an effort to better understand the corrosion these
material can be procured from textile manufacturers who
threads were causing on the alloys, analysis by x-ray fluo-
produce these threads in large quantities. They come
rescence spectroscopy (XRF) was done on three corroded
in various sizes, but those with diameters of less than
areas on one of the tubes after time in the oven and on a
1 mm have ample strength. The thread tested here had
non-corroded area of the same tube. The corroded areas
a diameter of 0.25 mm and can withstand a load of 3 kg.
corresponded exactly to the position of the polyester
The threads are naturally transparent up to a diameter of
1100/2/3 Vectran and Spectra threads. Three shots were
1 mm and slightly opaque in larger diameters. Tests were
taken of the same area in order to obtain the most accurate
also conducted with a monofilament thread compared
possible results. The measurements were taken using an
to a woven or stranded thread to see if the presence
ArtTAX prototype II spectrofluorometer (Intax GmbH).
of multiple strands might retain enough humidity to
The ArtTAX acquisition software connected to the spectro-
cause corrosion in the metal. Two polyethylene threads
fluorometer allowed the spectra obtained to be processed.
were tested: a monofilament thread and a thread woven
ArtTAX has an application that automatically identifies
from two filaments. The same protocol in the climatic
the elements present in each sample. This can be useful
chamber was applied and the same weight measurements
when analysing unknown samples, but it is important to
were taken before and after the test. Other threads were
maintain a critical eye with regard to the elements iden-
knotted onto the brass plate to help characterise the
tified (Gomes 2013). Our analysis of the results did not
corrosion products that formed on the surface of the
provide any information on the nature of the corrosion
metal. Their distribution is shown in Figure 9.
products formed, given that the markings measured by the
spectrometer and the respective spectra of the non-cor-
roded tube and the corrosion products generated by the
three threads are similar, and in fact identical to the naked
eye. There are several possible explanations for this: the
very small quantity of corrosion products formed and the
limited range of use of the spectrofluorometer – it cannot
Figure 9. Brass plate with different threads
determine elements below boron (Z=5) and above uranium
After spending time in the chamber, it was again evident (Z=92). In this case, a Coolidge tube with a molybdenum
that only the propylene threads (monofilament and anode was used, furtherer limiting the range of use (Z=13
woven) did not interact with the metal plate in any way. to Z=92). To determine the degradation product formed
Table 4 sums up the results of these tests. accurately, other analysis methods were needed.
PHYSICAL AND CHEMICAL INTERACTIONS BETWEEN METALLIC INSTRUMENTS AND PLASTIC THREADS
What protocols must be adopted for metal
instruments?
The conducted tests showed that most threads corrode
metal sooner or later. It is highly probable that this is
mainly related to the adjuvants and additives put into
the fibre during manufacturing, given that Dyneema,
Spectran and Nacrylan threads are made primarily
of polyethylene, a material known to have no direct
effect on metal (Lavédrine et al. 2012). The scratches
observed on the metal sheets may be due to the abra-
sive effect of materials added to the threads, such as
calcium carbonate, china clay and talc. It is therefore
recommended that a thread containing no additives be
used. Pure polypropylene thread also meets the other
requirements stipulated in the specifications. Since
2006, we have been using two types of pure polyeth-
ylene thread, with a diameter of 0.25 mm or 0.53 mm.
They are transparent or slightly opalescent (in the case
of the thicker thread) and are used to secure adjusted
mountings without risk of damage to the object. Thus
far, no contact corrosion or scratching as previously
observed has been noticed. Some of the other threads
tested in this study are now being assessed for possible
partial integration with woven threads or cords on
musical instruments from outside Europe.
Conclusion
In developing an adapted protocol for the mounting
and display of musical instruments at the museum, it
was possible to select a fibre – polypropylene without
additives – that is stable over time, has adequate mechan-
ical properties and does not react with the metal in the
instruments. A sound mounting protocol was therefore
successfully established. We have since expanded the use
of these types of thread, whose thinness and transpar-
ency make them ideal for use as supports and for seams
in fabrics, or for reinforcement or partial integration
with weakened woven fibres. Better understanding of
the interactions of the other types of thread with metal
is still needed.
Acknowledgements
The author would like to thank the entire team at the
Cité de la musique – Philharmonie de Paris for making
themselves available and providing assistance during
this research.
405
Materials list
Dyneema:
Gremco Sarl, 1955 chemin de St Bernard,
06220 Vallauris, France
www.gremco-sarl.com
Polyester:
Cousin Trestec, 8 rue Abbé Bonpain, BP39,
Wervicq-Sud, 59558 Comines Cedex, France
www.cousin-trestec.com
Polypropylene:
Société Sagaert (département Zebulon), 58 rue d’Ar-
mentières, 59560 Comines Cedex, France
http://sagaert.com/fr/textile/filature-de-polypropylene.html
Prolene:
Ethicon, Johnson & Johnson Medical SAS,
1 rue Camille Desmoulins,
92 787 Issy-les-Moulineaux Cedex 9, France
https://fr.ethicon.com/produit-catalogue-france/
sutures-non-resorbables/prolene
Spectra:
Honeywell
http://www51.honeywell.com
Vectran:
Laboderm, 12–16 rue Sarah Bernardt, 92600 Asnières-
sur-Seine, France
www.laboderm.fr
References
Barclay, R. 2004. The preservation and use of historic
musical instruments: Display case and concert hall.
London: Earthscan.
Bohé, V., P. Casari, F. Léné, and Peter Davies. 2016.
Comportement des matériaux à voiles de bateau. Revue
des composites et des matériaux avancés 13(3): 251–61.
Cavalier, K. 2013. Adjuvant des plastiques, carbonate de
calcium additif multifonctionnel. Techniques de l’ingénieur.
Plastiques et composites: AM3222/2–10. www.tech-
niques-ingenieur.fr/base-documentaire/materiaux-th11/
adjuvants-des-plastiques-42138210/carbonate-de-calci-
um-additif-multifonctionnel-am3222/ .
Dossogne, R. 2010. Polyéthylène. Techniques de l’ingénieur.
Plastiques et composites: AM3315/1–10. www.tech-
niques-ingenieur.fr/base-documentaire/archives-th12/
406  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND COMPOSITE OBJECTS

archives-plastiques-et-composites-tiaam/archive-3/ University Sorbonne. She is assistant manager of the

polyethylene-haute-densitepe-hd-a3315/ . Metal Workshop in the Department of Ceramics and
Duval, C. 2010. Polypropylène. Techniques de l’ingénieur. Metals at the Institut National du Patrimoine) and head
Plastiques et composites: AM3320/1–16. www.tech- of preservation and restoration at the Laboratory of
niques-ingenieur.fr/base-documentaire/materiaux-th11/ Conservation and Research, Musée de la musique, Cité
matieres-thermoplastiques-monographies-42147210/ de la musique – Philharmonie de Paris.
polypropylenes-pp-am3320/ .
Guérin, B. 2010. Polyamide. Techniques de l’ingénieur.
Plastiques et composites: AM3360/1–23. www.tech-
niques-ingenieur.fr/base-documentaire/materiaux-th11/
matieres-thermoplastiques-monographies-42147210/
polyamides-pa-a3360/ .
Gétreau, F. 1996. Aux origines du musée de la musique.
Les collections instrumentales du Conservatoire de Paris,
1793–1993. Paris: Klincksieck.
Gomes, S. 2013. Vers l’analyse quantitative des alliages de
laiton et bronze de la collection du Musée de la musique
par spectrométrie de fluorescence X. Report completing
an internship at the Laboratoire de recherche du musée
de la musique.
Lavédrine, B., A. Fournier, and G. Martin, eds. 2012.
Preservation of plastic artefacts in museum collections.
Paris: CTHS.
Lechat, C. 2007. Comportement mécanique de fibres et
d’assemblages de fibres en polyester pour câbles d’amar-
rage de plates-formes offshore. PhD dissertation, École
Nationale Supérieure des Mines de Paris, France.
Loeper-Attia, M.-A. 2007. L’impact des restaurations
sur la conservation des instruments de musique de
la famille des cuivres. In Conference proceedings: Un
laboratoire d’idées, facture et répertoire des cuivres entre
1840 et 1930, Cité de la musique, Paris. Historic Brass
Society. www.citedelamusique.fr/pdf/insti/recherche/
colloques_factures/loeper.pdf .
Massardier, V. 2001. Etat de l’art concernant la compat-
ibilité des matières plastiques, Study No. 00-0904/1A.
Laboratoire de Matériaux Macromoléculaires INSA de
Lyon. www.record-net.org .
Verrier, P. 2013. Adjuvant des plastiques, plastifiants.
Techniques de l’ingénieur. Plastiques et composites:
AM3231/2–11.

Author
Marie-Anne Loeper-Attia has a master’s degree in
restoration and preservation of artworks from Paris I
MODERN METALS
Aluminium Alloys in Swiss Public Collections:
Identification and Development of Diagnostic Tools to
Assess Their Condition
Christian Degrigny* Julie Schröter
Haute Ecole Arc Conservation-restauration Haute Ecole Arc Conservation-restauration
HES-SO University of Applied Sciences HES-SO University of Applied Sciences
and Arts Western Switzerland and Arts Western Switzerland
Neuchâtel, Switzerland Neuchâtel, Switzerland
christian.degrigny@he-arc.ch Julie.schroter@he-arc.ch
www.he-arc.ch *Author for correspondence
Abstract
Despite their presence in many museum collections as
testimony to the great technological advances of the late
19th century, aluminium alloy objects are rarely exhibited or
may even be discarded because of their unattractive surface
appearance. Conservation professionals are often helpless
when faced with these materials which they consider to be
unalterable but whose composition, desired appearance
during their manufacture, successive uses and corrosion
forms developed over the long term are not well known. Any
diagnosis is therefore extremely uncertain. The objective of
this project was to cover this gap and provide conservation
professionals with examination tools and diagnostic assistance.
This approach has been applied to more than 400 objects
preserved in four Swiss museum or private collections and
Introduction
Objects made of aluminium alloys are part of our daily
lives. Aluminium was invented in the mid-19th century.
After a short period during which its production cost
made it as precious as gold, it became, because of its
properties (lightness, resistance to corrosion and ability
to be recycled), the main material of emerging industries:
those of transport and packaging. Aluminium adapted
over time to multiple needs and is now part of many
public museum collections.
The durability of aluminium promoted by the indus-
trial world and validated by its behaviour in use makes
it, theoretically, a cultural heritage material without
problems. Yet, aluminium alloys have not received the
attention paid to other metals such as iron, copper, silver
or lead. The materiality of aluminium objects is still little
or not at all studied. The composition of the materials is
rarely known and the original surface appearance and its
evolution over time are poorly documented, as are the
considered sufficiently representative of the problems posed by
aluminium alloys. The materials were systematically analysed,
the corrosion forms developed identified, characterised and
compared from one collection to another. Some alterations
were examined in depth and the observed models were
inserted into the MiCorr participatory database (micorr.org).
The results obtained should make it possible to appreciate
better the nature of the materials studied and understand
their long-term performance in order to propose, if necessary,
appropriate interventions.
Keywords
aluminium, Swiss collections, documentation, analysis,
corrosion forms, diagnosis
risks of alteration incurred and the damage developed
in the long term.
To fill these gaps, more than 400 objects were studied
from four Swiss collections preserved under different
conditions and representative of the issues raised by
aluminium alloys in other collections. Their composition,
surface appearance and corrosion forms were docu-
mented using existing and/or innovative tools considered
accessible to heritage professionals.
A review of aluminium alloys
The discovery of aluminium was only made in 1807
by H. Davy and it was not until 1827 and the work of
F. Wöhler, inspired by that of C. Oersted, that it was
chemically produced. H. Sainte-Claire Deville improved
the process and ensured its industrial production from
1854 to 1886, the year in which the chemical production
method was replaced by the much less expensive elec-
ALUMINIUM ALLOYS IN SWISS PUBLIC COLLECTIONS: IDENTIFICATION AND DEVELOPMENT OF DIAGNOSTIC TOOLS TO ASSESS THEIR CONDITION
trolytic process, developed simultaneously by P. Héroult
in France and C.M. Hall in the USA (Grinberg 2003).
France did not immediately understand the significance
of this discovery, which led Héroult and his partners to
found the very first European electrolytic aluminium
manufacturing plant in Zurich, Switzerland, in 1887.
Chemical aluminium was expensive. Therefore, it
was mainly used to make precious objects (jewellery,
artwork). Following the technological revolution of
1886, the material’s technical performance (lightness
and corrosion resistance) led it to be used to produce
flagship objects for modern societies (Piccadilly Circus
Eros in London, the roof of San Gioacchino church
in Rome). It was only at the beginning of the 20th
century that new emerging industries (transport and
packaging) adopted the material (Plateau and Grinberg
2013). Duralumin, the most important aerospace alloy,
obtained by structural hardening and containing Cu
(3%–5%) and Mg-Mn in lower concentrations, was
invented in 1908 (Polmear 2004) and Alpax, a cast Al-Si
alloy (12%) essential to the automotive and chemical
sectors, in 1920. Gradually, the industry introduced the
seven families of alloys (low-alloy aluminium, Al-Cu,
Al-Mn, Al-Si, Al-Mg, Al-Mg-Si, Al-Zn) to meet the
many needs of modern society, whose designations
at international level were only adopted in 1970 (The
Aluminum Association 2009). The ability of aluminium
alloys to be recycled was also very quickly recognised
and allowed their massive use in Europe in all fields of
activity following the Second World War, as well as their
reuse by African populations during the colonial expe-
ditions of the late 19th century. As a result, 20th-century
recycled objects often have compositions that do not
match usual standards.
Aluminium alloys are known for their high chemical
stability (Vargel 1979). Under normal conditions of use
(neutral pH, ambient temperature), they are naturally
covered with an oxide film that protects them but even-
tually tarnishes over time. However, even under these
conditions, localised corrosions can develop mainly due
to chlorides (pitting corrosion) and evolve into intergran-
ular and filiform corrosions or exfoliation depending on
the environment and/or the composition and micro-
structure of the metal and the presence or absence of a
coating (Vargel 1979, Reboul 2005).
In recent years, and certainly thanks to the growing
interest in industrial heritage, curators have become
409
aware of the cultural values ​​of aluminium alloy objects.
They are often found in museums dedicated to the
historical and applied arts, folk arts and traditions,
military and ethnography. But few examinations have
been carried out on these materials. Their composi-
tion is poorly controlled, although it has been shown
that the distinction between chemical and electrolytic
processes is possible by studying trace elements (iron
[Fe], manganese [Mn], lead [Pb]) (Bourgarit and Plateau
2005). Often the presumed composition of materials is
associated with their field of application or based on the
recommendations of experts and current raw material
suppliers (Euralliage.com, Reboul 2005). However, this
qualitative approach must take into account the evolution
of material designations between the end of the 19th
century and the middle of the 20th century (formerly
and newly used alloys, variations from one country to
another) to avoid any misattribution.
The cultural heritage nature of the objects investigated
means that the invasive analysis of their corrosion forms
through sampling of corrosion products and/or in the
core material is only rarely carried out. Aeronautical
heritage is an exception, certainly because of the mass of
materials considered (Rocca et al. 2008, Guilminot and
Tissier 2015). The corrosion forms usually described in
the literature are pitting corrosion, intergranular corro-
sion and exfoliation. Examples of galvanic corrosion
are also mentioned. Tarnishing, although common, is
poorly documented.
The knowledge of aluminium found in museum collec-
tions is therefore rather approximate and mainly copied
from that of the industrial world. As for the composition
of the materials, it usually fails to take into consideration
the long period of development of alloys prior to the
introduction of international designations (1970s), or
the potential impact of armed conflicts that led to the
use of recycled aluminium. As far as their alteration
is concerned, the possible impact of their use and the
abandonment or reuse of objects are difficult to assess.
Examination conditions: Corpus and tech-
niques used
Corpus
The 405 objects considered come from four collections:
269 from the Swiss National Museum (SNM), 89 from
the Historical Museum of Lausanne (MHL), 31 from the
410  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Ethnographic Museum of Geneva (MEG) and 16 from the
Guex-Joris Foundation (FGJ). The pie chart by category
is shown in Figure 1. The objects studied were chosen
on the basis of their representativeness in the collections
and their conservation condition. The fields of transport
and electricity are excluded.
A few additional objects belonging to Swiss and French
private collections and presenting similar alterations
to those observed on objects in the collections were
considered for invasive examinations, including cross
sections.
Figure 1. Distribution of the objects studied according to their typology
Techniques
In addition to visual observation, the surface appearance
of a few objects was described with colorimetric meas-
urements made with an X-Rite Ci6x spectrophotometer,
the diameter of the measurement being 6 mm. The gloss
of the materials was established for three measuring
angles (30°, 60° and 85°) with a Polygloss GL0030
glossmeter to ISO 2813 standard. Polished black glass
with a defined refraction index was used for calibration
(100 GU [gloss unit]). The values ​​obtained at 60° (the
standard angle adapting to as many possible scenarios
as possible) were first used to obtain an overall reading.
Thus, at a measuring angle of 60°, surfaces with a gloss
greater than 70 GU are considered very bright, those
giving a GU between 10 and 70 are medium bright or
satin, and those with a GU of less than 10 are matte.
Reference materials (industrial samples and repre-
sentative cultural heritage objects) were considered
beforehand to calibrate the visually assessed gloss levels,
and measurements were repeated at least three times
to assess the reproducibility of the results. The same
materials were used for colorimetric measurements.
MODERN METALS
Elemental analyses were carried out with a portable
x-ray fluorescence (XRF) spectrometer (Niton XL3t 950
Air GOLDD +) in parallel with a PIXE particle acceler-
ator. Results on the objects in the collections in question
were very similar (Whitlow et al. 2018), but the use of
portable XRF was deemed more suitable for system-
atic and in-situ analysis campaigns (without surface
preparation). Another qualitative analysis tool called
DiscoveryMat, which monitors the corrosion potential
of materials over time and compares the plots obtained
with those of materials in a database using algorithms,
was tested (Degrigny et al. 2018). As DiscoveryMat
requires working on a clean polished metal surface,
which is not the case for objects from the above-men-
tioned collections, it was applied to aluminium objects
from private collections whose corrosion layers could
be removed mechanically.
Reference materials were used to describe the objects
visually and to characterise the level of alteration of the
objects in the different collections (Figure 2). As this
approach has its limits, non-invasive examinations of the
surface corrosion layers (using an eddy-current sensor
for thickness measurements, x-ray diffraction [XRD]
and Raman spectroscopy) were carried out. Authorised
sampling was sometimes necessary for some invasive
analyses (using powder XRD and Fourier transform
infrared [FTIR] spectroscopy) and for cross-section
examination using a Jeol 6400 SEM with a voltage of
20 kV.
Results
Surface appearance
Beyond the usual characteristics of aluminium alloy
objects (lightness and light grey colour), the material
gloss is often mentioned by professionals. Four levels
of gloss could be observed visually on both reference
coupons and collection objects: glossy, satin, satin
matte and matte. By comparing the visual appearance
(Figure 3a) with the measurements made with a gloss-
meter (Figure 3b) on a few representative objects, a
certain overall consistency was observed, especially at
the extremes (very bright and very matte). On the other
hand, the distinctions between glossy and satin when
observed visually gave inverted gloss measurements in
some cases. Furthermore, it appeared that the gloss-
meter had difficulty differentiating between the matte
and satin matte levels.
ALUMINIUM ALLOYS IN SWISS PUBLIC COLLECTIONS: IDENTIFICATION AND DEVELOPMENT OF DIAGNOSTIC TOOLS TO ASSESS THEIR CONDITION 411

Figure 2. Main forms of

alteration with mention
of certain levels of
development (+: rather
developed, low [T1] to
 high tarnish [T4])

These inconsistencies, which mainly affect the cups of

the SNM flasks (such as KZ 11056 and LM 106770),
could partly be explained by the level of tarnishing of
the objects, measured through colorimeter parameter L*
as indicated in Figure 4a, which shows that the level of
tarnish (as defined on Figure 2) increases as L* decreases.
Indeed, the glossy KZ 11056 and especially satin matte
LM 106770 cups were significantly tarnished (Figure 4b),
which could explain the lower brightness measured. The
matte KZ 11830 cup, on the other hand, had a low level
of tarnishing, therefore its gloss was probably unaffected.

Figure 4. (a) Tarnish levels of reference materials (Figure 2) measured

using a colorimeter parameter L* and (b) comparison with SNM military
flask cups in Figure 3a

Figure 3. (a) Visually observed glossy levels and (b) measurements made Association of composition/material typology
with a glossmeter. Data circled in blue correspond to a slight visual
underestimation of gloss levels and in red, to a visual overestimation Among the 405 objects studied, the main alloys were
of gloss levels those designated as: 1050 – up to Al 99.4%, Si 0.2% and
412  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Fe 0.4% in weight (55 objects); 1110 – up to Al 98.9%,
Si 0.3% and Fe 0.8% in weight (43), corresponding to
low-alloy aluminium; and 4006 – up to Al 98%, Si 1.2%
and Fe 0.8% in weight (133), corresponding to alumini-
um-silicon alloy. These designations are only indicative
since the international system was not yet in use at the
time most of the objects investigated were produced. They
were used in the fields of food (kitchenware – Figure 5a),
outdoor activities, military equipment and commemo-
rative objects, which, as shown in Figure 1, constitute
the vast majority of the objects studied. Technical and
scientific objects require alloys with specific mechanical
and corrosion resistance properties – Al-Cu (22 objects),
Al-Si (22), Al-Mg (19) and Al-Mg-Si (17, Figure 5b).
Some objects, a priori of less beautiful workmanship, are
made of complex alloys combining the elements Al, Cu,
Zn, Fe and Si in high proportions (Figure 5c). As ethno-
graphic objects are often recycled, it is not surprising
that their alloys contain multiple elements but in small
proportion (Figure 5d).
Figure 5. Representative objects studied: (a) coffee maker, Switzerland
(SNM); (b) Nagra III recorder no. B59237, Kudelski Prilly-Lausanne,
Switzerland 1963 (FGJ); (c) crank whisk (SNM); (d) braid needle, North
Africa (MEG)
DiscoveryMat applied to a few objects made it possible
to analyse them qualitatively. The knife knob in Figure
6a studied via the application (Figure 6b) turned out
to be made from a complex alloy containing Cu, Zn,
Fe and Si in addition to aluminium. The proposed
compositions are only indicative due to the low repre-
sentativeness of similar alloys in the database under
construction.
MODERN METALS
Figure 6. (a) Knife with aluminium alloy knob, private collection; (b)
analysis with DiscoveryMat
Corrosion forms
Measurements of the thickness of oxide films (about
10 µm) on the surface of the objects indicated that they
may be as thick as those produced by anodising treat-
ment, although less protective since they did not prevent
the development of the pitting and filiform corrosion
phenomena described below.
Visual observation of the 405 objects in the four collec-
tions, supplemented by the use of a Dinolite digital micro-
scope, showed that, surprisingly, purer aluminium objects
appear to be more susceptible to corrosion than alloyed
ones. This can be explained by the high use of objects
made of purer alloys, which reversed the trends. The
most common forms of corrosion are tarnishing (average
levels T2 to T3 in Figure 2 and more present outside than
inside the objects), pitting corrosion (distributed both
outside and inside) and filiform corrosion (less devel-
oped than pitting but more present outside than inside,
possibly due to the strong tarnishing and therefore the
high thickness of the oxide film under which corrosion
might develop). Filiform corrosion also seems to be an
evolution of pitting corrosion.
Additional studies showed that XRD is not an appro-
priate technique to characterise surface tarnishing (no
additional peaks to those of aluminium). However, in
combination with Raman and FTIR spectroscopy, it may
be used to determine the nature of complex corrosion
products. Observation on cross section and by electron
microscopy of advanced forms of filiform corrosion
(Figure 7) made it possible to validate that intergranular
corrosion extending over the entire thickness of the
material can be generated by surface alteration (Flandro
and Thomas-Haney 2015). Elemental mapping suggests
that the presence of Na and its association with C and O
led to the formation of alkaline sodium carbonate, which
gradually solubilised the metal.
ALUMINIUM ALLOYS IN SWISS PUBLIC COLLECTIONS: IDENTIFICATION AND DEVELOPMENT OF DIAGNOSTIC TOOLS TO ASSESS THEIR CONDITION
Similarly, it was observed that exogenous deposits, which
are theoretically inactive, had built up underneath pitting
corrosion which, as it developed, became intergranular
(Figure 8). The species inducing this active corrosion are
rich in chlorides and sulfides.
Discussion
Recognising an aluminium alloy object in a collection
may be fairly easy: the metal is light, often light grey
in colour, usually has little tarnish and often develops
pitting and filiform corrosion. Other metals may be
mistaken for aluminium, such as tinned iron, zinc alloys
and German silver, yet such assumptions can be easily
eliminated: ferrous alloys are magnetic and zinc alloys
are darker and granular in appearance; as for German
silver, it emerged prior to aluminium, imitates silver and
its production slowed down after 1950.
Gloss measurements are fairly consistent with the gloss
evaluation deduced from visual observation. Therefore,
they might not necessarily be the first to be carried out
by conservators surveying the condition of objects.
Colorimetric measurements of the tarnish level, associ-
ated with those of the thickness of the oxide film formed
on the surface, could be preferred because they allow a
better understanding of the risks of localised corrosion
(pitting and filiform).
413
Figure 7. (a to c) Localised
filiform corrosion extending to
(d and e) intergranular corrosion
over the full thickness (SEM
pictures taken by S. Ramseyer,
HE-Arc I) of an almost pure
aluminium coffee filter, French
 private collection
Figure 8. (a) Exogenous deposits
inducing (b) intergranular
corrosion (SEM picture taken by
S. Ramseyer, HE-Arc I) on this
almost pure aluminium towel
 holder, French private collection
No possible correlation between the composition of the
materials and their corrosion resistance was observed,
but the objects in this study were stored in good envi-
ronmental conditions after a period of use and perhaps
partial abandonment. The effect of the composition of
the alloys on their state of preservation plays a much
more important role for objects exposed to aggressive
conditions, such as aircraft wrecks, ships, etc., found in
a marine environment (Degrigny 2018).
Visual observation of corrosion forms developing on
aluminium alloys does not allow us to understand the
extent of alteration to the core of a material. Corro-
sion models showing corrosion phenomena in their
entirety, such as for a filter (Figure 7) and towel holder
(Figure 8), are essential for the proper diagnosis carried
out by conservation professionals on the collections
studied. Therefore, they were integrated into the MiCorr
online database (micorr.org) to allow the comparison of
schematic representations of corrosion forms observed
visually or under binocular examination of unknown
objects with those of the established models and to refine
the diagnosis (Figure 9).
Conclusion
The composition of objects made of aluminium alloys
does not always match established standards. While a
414  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 9. (a) Observation under binocular of a corrosion form on an unknown object; (b) schematic representation; (c) comparison and identification
of a close stratigraphy in the MiCorr database; and (d) consultation of the corresponding corrosion model to refine the diagnosis
large majority of the alloys studied in this project were
low-alloy aluminium alloys, a significant proportion
consisted of complex alloys containing Cu, Zn, Fe and
Si in addition to aluminium. It has also been shown that
the composition of materials is often chosen according
to their field of application. Thus, low-alloy aluminium
alloys (that would be theoretically more corrosion-re-
sistant) are used in food, sports and outdoor activities,
while alloys with specific mechanical properties are used
in technical and scientific objects. Ethnographic objects
are exempt from this rule because they are often made
of recycled materials.
The characterisation of aluminium alloy composition
is still in its infancy. While it is essential to distinguish
alloys according to their chemical or electrolytic period,
if a more systematic approach were used, it might be
possible to follow the technological evolution of the
material and ensure greater control over time. Portable
and relatively accessible XRF is recommended because it
allows in-situ analysis to be carried out. However, due to
cost and conditions of use, its application in the heritage
field may be limited. The online and easy-to-use Discov-
eryMat application gives encouraging results. However,
the relevance of its use will depend on the enrichment
of its database by conservation professionals.
The tarnishing of aluminium objects can be quantified by
colorimetry, while the thickness of the oxide film formed
can also be measured using an eddy-current sensor. These
data make it possible to understand the risks of local-
ised corrosion that may develop in the long term. Some
corrosion forms are due to an aggressive environment
(pitting, contact corrosion) and are sometimes masked
by exogenous deposits that generate deeper corrosion
(intergranular corrosion). A significant tarnish (thick
oxide layer) favours filiform corrosion which can also
evolve into intergranular corrosion. It was not possible,
MODERN METALS
however, to establish any association between metal
composition and corrosion forms due to the extensive
use of some of the objects studied, which modified
their general behaviour. Nevertheless, some general
conclusions could be drawn. First, tarnishing affects the
vast majority of the materials studied, while pitting and
contact corrosion are rarer. Second, as indicated above,
tarnishing generates a sensitivity to filiform corrosion.
This project tested certain investigation tools that we
believe are appropriate for the diagnosis of museum
aluminium alloys (glossmeter, colorimeter, eddy-cur-
rent sensor, portable XRF analyser and DiscoveryMat
application) and specified the limits of their use. It also
showed the importance of building a database of corro-
sion forms by conservation professionals to allow them
to assess visually the general conservation condition of
the objects studied and to insert the outcomes of their
diagnosis in professional databases such as MiCorr.
Acknowledgements
The authors would like to thank the HES-SO University
of Applied Sciences and Arts Western Switzerland for
the funding of the EtICAL project and those in charge
of the collections (public and private) who made the
objects studied available to us: Markus Leuthard (SNM),
Claude-Alain Künzi (MHL), Kilian Anheuser (MEG),
Pierre-André Perrin (FGJ) and Tobias Schenkel. They
are also grateful to Stephan Ramseyer (Laboratory of
Electronic Microscopy and Microanalysis, Néode, HEI
Arc) for the SEM-EDS analyses.
References
Bourgarit, D. and J. Plateau. 2005. Quand l’aluminium
valait de l’or : peut-on reconnaître un aluminium « chim-
ique » d’un aluminium « électrolytique » ? ArchéoSciences
29: 95–105.
ALUMINIUM ALLOYS IN SWISS PUBLIC COLLECTIONS: IDENTIFICATION AND DEVELOPMENT OF DIAGNOSTIC TOOLS TO ASSESS THEIR CONDITION 415

Degrigny, C. 2018. Conservation of historical objects – Authors

Marine, terrestrial and industrial. In Aluminium 2014,
Christian Degrigny has a PhD in analytical chemistry
Smithsonian Contributions to Museum Conservation,
from the University of Paris IV on the stabilisation of
9, 87–95. Washington.
submerged aluminium remains. He has since specialised
Degrigny, C., E. Menart, and G. Erny. 2018. Easy-to-use, in the application of electrochemistry in conservation
low-cost electrochemical open-source hardware to science and is currently a professor at Haute Ecole Arc
analyse heritage metals: Possibilities and limits. Current Conservation-restauration (HE-Arc CR).
Topics in Electrochemistry 20: 15–23.
Julie Schröter has a master’s in cultural heritage conser-
Euralliage.com. www.euralliage.com/alliage.html
vation from the University of Paris 1 Pantheon-Sorbonne.
(accessed 1 October 2018).
Since 2010, she has specialised in metal conservation as
Flandro, X. and H.M. Thomas-Haney. 2015. A survey of a freelance conservator. In 2014, she joined the research
historic finishes for architectural aluminium, 1920–1960. unit at HE-Arc CR.
Journal of Preservation Technology 46(1): 13–21.
Grinberg, I. 2003. L’aluminium, un si léger métal. Paris:
Découvertes Gallimard.
Guilminot, E. and Y. Tissier. 2015. Étude des alliages
d’aluminium pour la mise au point des traitements de
conservation des avions. Cahiers d’histoire de l’aluminium
54(1): 82–93.
Polmear, I.J. 2004. Aluminium alloys – A century of age
hardening. In Materials forum, eds. J.F. Nie, A.J. Morton,
and B.C. Muddle, vol. 28, 1–14. Melbourne: Institute of
Materials Engineering Australasia Ltd.
Plateau, J. and I. Grinberg. 2013. Passion aluminium.
Trésors de la Collection Jean Plateau-IHA. Paris: Les
éditions du Mécènes.
Reboul, M. 2005. Corrosion des alliages d’aluminium.
In Techniques de l’ingénieur, COR 325.
Rocca, E., C-E. Barchiche, F. Mirambet, M. Dubus, and
C. Tilatti. 2008. Étude de la corrosion des matériaux du
patrimoine aéronautique en aluminium. Techné (Hors-
Série 2008).
The Aluminum Association. 2009. International alloy
designations and chemical composition limits for
wrought aluminum and wrought aluminum alloys.
Registration record series, Teal sheets. Arlington, VA:
The Aluminium Association, Inc.
Vargel, C. 1979. Le comportement de l’aluminium et de
ses alliages. Paris: Dunod Technique.
Whitlow, H.J., L. Wang, E. Guibert, and C. Degrigny. 2019.
Investigations of minor elements in early aluminium
artefacts. Nuclear Instruments and Methods in Physics
Research B 450: 291–93.
Aluminium Surfaces in Heritage Collections: Evaluating
Surface Appearance Modifications

Christine Andraud Clotilde Boust Anne Genachte-Le Bail Marine Page

Centre de Recherche sur la
Conservation, Museum National
d’Histoire Naturelle, MiC, CNRS
Paris, France
christine.andraud@mnhn.fr
Aurélia Azéma
Laboratoire de Recherche des
Monuments Historiques, CRC USR 3224
Champs-sur-Marne, France
aurelia.azema@culture.gouv.fr
Philippe Betinelli
Centre national des arts plastiques
(Cnap)
Paris La Défense, France
philippe.bettinelli@culture.gouv.fr
*Author for correspondence
Centre de Recherche et de
Restauration des Musées de France
Paris, France
clotilde.boust@culture.gouv.fr
Magali Brunet
Université de Toulouse, Cemes-CNRS
UPR 8011
Toulouse, France
magali.brunet@cemes.fr
Michaela Florescu*
Independent metals conservator
Paris, France
michaela.florescu@gmail.com
Institut national du patrimoine
Aubervilliers, France
anne.genachte-lebail@inp.fr
Nicolas Melard
Centre de Recherche et de
Restauration des Musées de France
Paris, France
nicolas.melard@culture.gouv.fr
Anne Michelin
Centre de Recherche sur la
Conservation, Museum National
d’Histoire Naturelle, MiC, CNRS
Paris, France
anne.michelin@mnhn.fr
Centre de Recherche et de
Restauration des Musées de France
Paris, France
marine.page@culture.gouv.fr
Grégory Seine
Université de Toulouse, Cemes-CNRS
UPR 8011
Toulouse, France
gregory.seine@cemes.fr
Annick Texier
Laboratoire de Recherche des
Monuments Historiques, CRC USR
3224
Champs-sur-Marne, France
annick.texier@culture.gouv.fr

Abstract
Surface finishes on aluminium objects play a significant role
in the public perception of an artefact and are therefore a
prominent concern in conservation practice. This paper presents
a project aimed at developing methods that allow stakeholders
to incorporate more conveniently the preservation of surface
finishes into conservation strategies for aluminium objects.
Research is currently underway with support from the Centre
national des arts plastiques (Paris). The methodology uses a
transversal approach to assess variations of surface appearance
based on the evaluation of optical characteristics of aluminium
surfaces. The process combines visual and analytical methods
and has been designed to be suitable for both lab and field-
based projects.
Keywords
aluminium alloys, surface finishes, surface appearance, surface
treatments, methodology, multi-analytical approach, decision-
making tools, conservation strategies

Introduction
Aluminium and its alloys are found in heritage collections
associated with modern artefacts and are typically used
for sculpture, contemporary art, design and technical
and industrial heritage. Where artefacts display bare
aluminium surfaces (without a coating or anodisation),
the conservation of the original surface finish is a promi-
nent concern. The surface appearance can express precise
aesthetic concepts and, as such, plays a significant role
in the public perception of the artefact. As many surface
finishes derive from complex technical processes, they
also form part of the historical record of aluminium
engineering and technology.
The surface appearance can be altered by such factors as
tarnishing by atmospheric pollutants, corrosion and inap-
propriate cleaning processes (Sheasby and Pinner 2001).
Observation or documentation can indicate if the original
surface finishes have been preserved. Conservators are
commonly faced with three types of situations:
• Where they remain intact, it is essential that surface
finishes undergo as little modification as possible
during maintenance interventions or conservation
treatments.
• Where the original surface finishes have been lost but
are documented, conservators and curators may seek
to replicate the original surface appearance through
remedial treatment.
• Where there is no existing reference for the surface
finish intended originally, the aim might just be to
achieve an undefined homogeneous surface aspect.
ALUMINIUM SURFACES IN HERITAGE COLLECTIONS: EVALUATING SURFACE APPEARANCE MODIFICATIONS 417

So far in literature, the focus for aluminium conservation
has mainly been on treatment of corrosion and related
structural issues. In practice, when stakeholders see
preservation of the surface appearance as a priority, they
can find themselves short of solutions, sometimes turning
to treatments derived from industrial processes (which
can be excessively interventive for heritage collections
as they modify the original material) or replacing the
altered elements.
This research aims to establish a diagnostic tool and a
set of decision-making guidelines that will allow stake-
holders to incorporate more conveniently the preserva-
tion of surface finishes into conservation strategies for
aluminium objects. The project focuses on two aspects:
discouraging the application of ‘one-size-fits-all’ treat-
ments and instead encouraging treatment decisions
based on a case-by-case analysis; and developing an
easily implemented methodology for projects that are
sent for competitive tender or conservation projects with
limited resources.
The methodology is based on a transversal approach to
assess variations of surface appearance using evaluation
of the optical characteristics of aluminium surfaces. It
combines the characterisation of the surface aspect with
rugosimetry analysis of the surface. The process has
been designed to be suitable for both lab and field-based
projects. In addition to assessment of aluminium replica
samples, evaluations were carried out on artefacts from
French heritage collections. This paper only describes
the assessments conducted on sample material.
Sample preparation
Smooth (mirror-polished) and rough (mill-finished) plates
(10 × 5 cm) of EN AW-5754 alloy (Al-Mg series) were
used as two distinct sample groups. These samples were
prepared as part of a previous study carried out for the
conservation of a microcar prototype (CNAM – Musée
des Arts et Métiers) (Florescu 2015). Samples were repre-
sentative of the general surface appearance of the artefact.
Based on a literature search and investigation of in-house
aluminium treatment practices across several French and
international museums and conservation studios, a range
of three chemical and three mechanical surface treatments
were applied to the samples, with varying application times
(Florescu 2015, 158–64). For each of the two selected types
of surface, each combination of treatment and application
time was applied to a series of 5 samples, totalling a corpus
of 180 samples. On each sample, the treatment was applied
to only half of the surface, the untreated remaining half
being the reference surface.
The specifications of each selected treatment method
are presented in Table 1. The selected surface treatments
applied to the sample groups should not be considered
exhaustive treatment options. They represent a selective
range of options applied to a single type of alloy with
specific surface finishes.

Table 1. Characteristics of selected surface treatments

CHEMICAL TREATMENTS
NAME SOLUTION COMPOSITION pH APPLICATION POST-APPLICATION
C.I Sodium hydroxide (NaOH) NaOH: 2 g / L demineralized water 13 30 seconds / 2 / 15 / 30 Rinsed with
C.II Phosphoric acid (H3PO4) H3PO4: 74 mL 1 minutes demineralized water
and nitric acid (HNO3) HNO3: 2.5 mL Submersion Dried stamping with
Demineralized water: 23.5 mL a cloth
No further action on
C.III Phosphoric acid (H3PO4) H3PO4: 33% (vol.) / L demineralized water 0.5
the surface

MECHANICAL TREATMENTS
NAME ABRASIVE HARDNESS GRANULOMETRY COMPOSITION APPLICATION POST-APPLICATION
M.I Calcium carbonate Mohs 3 1–36 µm 10 g CaCO3 / 6 mL 1/5 Paste applied on Rinsed with
(CaCO3) demineralized water minutes a cotton cloth, demineralized water
M.II Alumina (Al2O3) Mohs 9 1 µm 2 g Al2O3 / 5 mL Manual wrapped around a Dried stamping with
demineralized water abrasion wooden block a cloth
No further action
M.III Silicon carbide (SiC) Mohs 9.5 unspecified Embedded in Pad attached to a
on the surface
commercially available wooden block
polyester pads (3M™)
418  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Study methods
The optical appearance of bare aluminium surfaces is
largely derived from a ‘reactive layer’, a thin, approximately
1 µm thick surface layer with grains heavily deformed
during the manufacturing processes (AluMATTER,
accessed 28 October 2018) (Figure 1). The main param-
eters influencing the optical appearance include steps in
the surface (dents and scratches at the macrometric scale),
surface roughness (texture at the micrometric scale) and
variable optical constants that relate to different phases
in the reactive layer and to the presence of the naturally
occurring oxide layer on the aluminium surface.
Figure 1. Cross section of an aluminium surface. © AluMATTER
The research focused on determining how conservation
interventions can influence the surface roughness as
they do not affect the composition of the alloy or the
macrometric irregularities.
Roughness determines a surface’s ability to return
received light (reflectivity), influencing the viewer’s
perception of the surface appearance. The rougher the
surface, the more diffuse the reflection it produces. On
the contrary, the smoother the surface, the more specular
the reflection (Seve, Indergand, and Lanthony 2007).
A methodology was developed to correlate the visual
perception of the surface finish with measured changes
in surface roughness and reflectivity.
Characterisation of aluminium surface aspect
using reflective methods
Graphic scale reflection
A scale was created by the main author of this paper and
designed in Adobe Photoshop with sections displaying
graded shades of grey and patterns comprising motifs of
different sizes and shapes. The scale can be reproduced
to match the target surface’s size. The graphic scale
is reflected onto the target surface and the reflection
photographed (Figure 2a). Placing the scale at an angle
MODERN METALS
with the sample produces a clear reflection and facilitates
photographic access. In this case, the scale was adhered
to a wedge inclined at an angle of 7° (Figure 2b). The
photographs are compared to assess modification of the
surface appearance based on the sharpness of reflection
and reproduction of the scale details and surface hue.
Figure 2. Graphic scale reflection method. © M. Florescu/INP
Light reflection
Two light-reflection methods were used during the
research. The first method is suitable for field-based
assessments, providing qualitative results. The second
method provides quantitative data but can only be used
in the lab. Both methods are based on the same principle:
when directed onto a surface, a laser beam reflects a light
field with specular and scattered light (Zhenrong, Jing,
and Peifu 2010).
In-situ laser reflection involves the use of a handheld
laser beam, such as a pointer pen. Pointer pens are
widely available and, as they are low powered (1 mW),
do not fall under use regulations; they run on batteries
and are portable enough to use in all circumstances.
Several colours are available, corresponding to different
wavelengths, and can be used interchangeably.
The set-up requires the laser beam to be directed at the
target surface at an angle of incidence of 45° and reflected
onto a screen positioned at 90° to the surface (Figure 3a
and b). The laser beam is projected on the surface
and the reflection on the screen is photographed. The
photographs are compared to assess any modification
ALUMINIUM SURFACES IN HERITAGE COLLECTIONS: EVALUATING SURFACE APPEARANCE MODIFICATIONS
to the surface appearance based on the shape, size and
brightness of the reflection.
Figure 3. Laser reflection method. © M. Florescu/CNAP
Scatterometry measurements were carried out to measure
the bidirectional reflectance distribution function (BRDF)
of the surfaces. This technique produces a hemispherical
distribution of the light reflected from the surface as a
function of the angle of incidence of the incoming light
ray. The analysis was completed using an Eldim EZ
Contrast device, with a transversal incidence of 45° to
the surface’s direction.
When used on anisotropic surfaces (surfaces having a
preferred orientation due to the manufacturing process),
all three types of projections need to be oriented perpen-
dicular to the direction of lay.
Characterisation of aluminium surface topography
Surface 3D microscopy, using Keyence-VHX 5000 and
Hirox KH-8700 digital microscopes at 800× magnifica-
tion, was used to produce detailed views of the sample
surfaces.
Rugosimetry measurements were taken to quantify the
surface roughness. Analysis was conducted using the
Altimet Altisurf 50, an optical profilometer set at the
419
following working parameters: X, 1 mm; pitch, 0.5 µm;
and Z, 330 µm. Micro-topography profiles were drawn
perpendicular to the surface direction. In general,
micro-topographical irregularities existing on manufac-
tured surfaces are characterised by comparison with a
mean line, defined as the limit where the areas between
the profile and the mean line are equal above and below
the line. Irregularities in the profile are classified in two
categories: asperities or ‘peaks’, and cavities or ‘valleys’.
Each type is further defined by a range of amplitude
and distribution parameters including quantity, altitude,
frequency and distance (Menezes et al. 2013, Interna-
tional Organization for Standardization 2012). Four of
these parameters were included in this study: Rq (root-
mean-square roughness), Rsk (skewness, or symmetry
of the profile about the mean line), Rku (kurtosis, or
sharpness of the profile about the mean line) and Pc
(peak count, or number of local peaks within a selected
band above and below the mean line). These parameters
are detailed and illustrated in Table 2.
Atomic force microscopy (AFM) can measure surface
roughness at the nanometric scale and is considered a
useful complementary analysis method to micro-topog-
raphy. Examination was conducted in tapping mode on
a Bruker Multimode microscope, at a scanning scale of
15 µm.
Results
The results are presented here for the reference surfaces
and for the treatments inducing maximum and minimum
visual variations on the surface appearance of the samples.
They are to be considered for each sample group inde-
pendently.
Mirror-polished sample group
The non-treated mirror-polished reference surface,
when assessed with the graphic scale, showed a very
clear reflection with excellent reproduction of shapes
and details and no change in colour (Figure 4a). The
laser projection showed a concentrated area with a slight
halo (Figure 6a). Scatterometry showed a tall and narrow
peak, which corresponds to specular reflection of the
incident light (Figure 8a). Observation by 3D microscopy
revealed few irregularities on the surface (Figure 9a).
The microtopography profiles confirmed the surface was
smooth (Figure 9a), with a low Rq (average of 0.080). Rsk
averaged 0.39, indicating a balanced distribution of peaks
420  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND MODERN METALS

Table 2. Surface roughness parameters

PARAMETERS DEFINITION ILLUSTRATION

Mean line Defined as the limit where the areas between the profile and the
mean line are equal above and below the line.

Rq Root-mean- Standard deviation of the distribution of surface heights. Very similar

square to the average roughness parameter (Ra), but more sensitive to
roughness peaks and valleys.
Rsk Skewness Measures the symmetry of the profile about the mean line; can be
used to differentiate between surfaces which have the same average
roughness but different shapes.
Positive skewness: profiles with high peaks or more peaks than
valleys; negative skewness: profiles with deep scratches or more
valleys than peaks.

Rku Kurtosis Measures the sharpness of the profile about the mean line.
High kurtosis value (> 3): leptokurtic profile, spiky surfaces; low
kurtosis (< 3): platykurtic profile, bumpy surfaces; kurtosis = 3:
gaussian profile, random surfaces.

RPc Peak count Number of local peaks projected through a selected band located
above and below the mean line by the same distance. The peak
count is determined along the assessment line and given in peaks
per millimetre.
For this project, the bandwidth was 0.5 µm wide, ± 0.25 µm above
and below the mean line.

and valleys and a limited amplitude, Rku averaged 4.5,
indicating a profile with spiky reliefs, and RPc was 0/mm.
Further examination with AFM (Figure 10a) showed
this surface was not entirely smooth and had a periodic
pattern of scratches when observed at the nanometric
scale. The reliefs observed on the sample area cover a
range of 112.54 nm.
Figure 4. Results of graphic scale evaluation, mirror-polished sample
group: (a) non-treated; (b) treated with C.II; (c) treated with M.III; (d)
treated with M.II. © M. Florescu/INP
Surfaces dipped in the phosphoric-nitric acid solution
(C.II) for 30 minutes showed a strongly opacified and
whitish appearance, and graphic reflection scarcely
reproduced the general shapes but none of the details
of the patterns (Figure 4b). The laser projection for this
type of surface produced a large halo with an area of
sharp brightness in the centre (Figure 6b). Scatterom-
etry provided a tall and narrow peak with an enlarged
base, corresponding to a surface with both specular and
diffuse reflection (Figure 8a). Observation by 3D micros-
copy showed extensive and deep pitting on the surface,
confirmed by the microtopography profile (Figure 9a),
which showed a much rougher surface (Rq averaged
0.290). Rsk averaged -2, indicating a profile with low
peaks and deep valleys, Rku averaged 13.7, indicating a
profile with very sharp reliefs, and RPc averaged 31/mm.
AFM (Figure 10b) showed that the initial periodic pattern
of the surface at nanometric scale had been completely
modified and deep pitting was also visible at this scale.
Surface reliefs on the tested area displayed a range of
681.28 nm.
Surfaces abraded with silicon carbide (M.III) for 5 minutes
rendered a very blurry and significantly darkened reflec-
tion, with loss of shapes and detail definition (Figure 4c).
The laser projection had an elongated and narrow form,
without a bright central area (Figure 6c). Scatterometry
showed a very low and large peak, indicating the surface
ALUMINIUM SURFACES IN HERITAGE COLLECTIONS: EVALUATING SURFACE APPEARANCE MODIFICATIONS 421

Figure 5. Results of graphic scale evaluation, mill-finished sample group:

(a) non-treated; (b) treated with M.II; (c) treated with M.III; (d) treated
with M.I. © M. Florescu/INP

Figure 8. Scatterometry results: (a) mirror-polished sample group; (b)

mill-finished sample group. © M. Florescu/CRC/CNAP

Figure 6. Results of
laser evaluation,
mirror-polished
sample group:
(a) non-treated;
(b) treated with
C.II; (c) treated
with M.III;
(d) treated
with M.II. © M.
 Florescu/CNAP

Figure 9. Microtopography and 3D microscopy results: (a) mirror-

polished sample group; (b) mill-finished sample group. © M. Florescu/
C2RMF/LRMH/INP

had a diffuse reflection and very limited specular bright-

ness (Figure 8a). Surface 3D microscopy showed abrasion
had created a net of very fine and regular scratches (Figure
9a). Rugosity parameters showed increased roughness
Figure 7. Results of (Figure 9a), as Rq averaged 0.230. Rsk averaged -0.35,
laser evaluation,
mill-finished
indicating a relatively even distribution between peaks and
sample group: valleys, with the valleys slightly deeper, Rku averaged 3.6,
(a) non-treated; translating into a rather gaussian profile, and RPc averaged
(b) treated with 38/mm. Examination with AFM (Figure 10c) showed abra-
M.II; (c) treated
with M.III; (d)
sion had created a random pattern of scratches, replacing
treated with M.I. the initial periodic pattern. Surface reliefs on the test area
 © M. Florescu/CNAP covered a range of 855.68 nm.
422  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Figure 10. AFM results, mirror-polished sample group: (a) non-treated;
(b) treated with C.II; (c) treated with M.III; (d) treated with M.II. © M.
Florescu/CEMES/CNAP
Abrasion with alumina (M.II) for 5 minutes produced
minor changes. The surface after treatment had very
close characteristics to the non-treated reference surface.
The graphic reflection was very slightly blurred, with the
general shapes and details of the patterns well defined
(Figure 4d). The laser projection had a slightly ovoid
shape, with a small halo around a bright concentrated
central area (Figure 6d). Scatterometry produced a peak
comparable with the reference surface, with a slightly
enlarged base (Figure 8a). Observation by 3D micros-
copy showed abrasion had created few irregularities on
the surface, and surface roughness was low (Figure 9a):
Rq averaged 0.096, Rsk averaged -0.03, indicating the
profile had slightly more valleys than peaks, but of low
altitude, Rku averaged 5.1, indicating a profile with sharp
reliefs, and RPc averaged 2/mm. Examination with AFM
(Figure 10d) showed that the original periodic pattern
had been replaced with a net of random fine and shallow
scratches. Surface reliefs on the test area covered a range
of 477.58 nm.
Mill-finished sample group
The non-treated reference surface provided a blurry
reflection, reproducing distorted shapes and little of
the details of the graphic scale (Figure 5a). The laser
projection had a stretched ovoidal form surrounded
by a halo (Figure 7a). Scatterometry showed a low and
enlarged peak, indicating the surface had mainly a diffuse
reflection and limited specular brightness (Figure 8b).
Observation by 3D microscopy showed an irregular
MODERN METALS
surface condition with scratches of various widths, and
microtopography profiles (Figure 9b) showed an elevated
surface roughness (Rq averaged 0.470 µm), with a large
periodic pattern. Rsk averaged 0.13, indicating there
were slightly more peaks than valleys, Rku averaged
2.73, indicating a rather platykurtic profile, with rounded
reliefs, and RPc averaged 14/mm.
Abrasion with alumina (M.II) for 5 minutes created very
reflective surfaces, with a clear perception of shapes
and details (Figure 5b). Laser projection had a brighter,
more concentrated and rounded form with a small
halo (Figure 7b). Scatterometry provided a higher and
narrower peak than the reference surface, indicating
an increased specular reflexion and decreased diffuse
reflection (Figure 8b). Observation by 3D microscopy
showed an important reduction in surface irregular-
ities (Figure  9b). Microtopography profiles showed
the existing periodic pattern had been significantly
eroded and the surface roughness had decreased, with
Rq averaging 0.220 µm (Figure 9b). Rsk averaged 0.02,
translating into an even distribution of peaks and valleys,
Rku averaged 2.77, indicating a platykurtic profile, and
RPc averaged 4/mm.
The surfaces abraded with silicon carbide (M.III) for
5 minutes showed a very blurry and darkened graphic
reflection, with great loss in quality of reflection (sharp-
ness and details) (Figure 5c). Laser projection had a very
elongated line-shaped form, with no bright central area
(Figure 7c). Scatterometry showed a spread distribution
of the incident light, with a very low and large peak,
indicating a diffuse reflection (Figure 8b). Surface 3D
microscopy showed the abrasion had created a net
of fine and regular scratches, replacing the existing
periodic pattern of the surface, confirmed by micro-
topography profiles (Figure 9b). General roughness
was reduced: Rq averaged 0.365 µm, Rsk averaged
-0.12, indicating a slight predominance of valleys, Rku
averaged 3.02, indicating gaussian type reliefs, and RPc
averaged 14/mm.
Abrasion with calcium carbonate (M.I) for 5 minutes
displayed the least surface modification of this sample
group. Assessment with the graphic scale showed minor
modification in the reproduction of shapes and details
(Figure 5d). The laser projection produced an ovoidal
form with a bright central area surrounded by a halo, a
little wider than the reference (Figure 7d). Scatterometry
provided a peak very similar to the reference (Figure 8b).
ALUMINIUM SURFACES IN HERITAGE COLLECTIONS: EVALUATING SURFACE APPEARANCE MODIFICATIONS
Observation by 3D microscopy showed a decrease in
surface irregularities (Figure 9b). Microtopography profiles
showed the abrasion had reduced the general surface
roughness (average Rq of 0.383 µm) but had preserved
the existing periodic pattern (average Rsk of -0.06; average
Rku of 2.73; average RPc of 14/mm) (Figure 9b).
Examination of mill-finished samples under AFM was
generally inconclusive, as the device had not been able
to acquire clear images of these types of surface.
Discussion
Optical and topographical characterisation of
aluminium surfaces
Depending on the specific characteristics of the arte-
fact considered for intervention, and the goals of the
treatment, each of the treatment methods selected in
this project could prove useful or damaging in equal
measure. However, the results obtained within this
project provide information on the behaviour and appear-
ance of aluminium surfaces after certain treatments,
and therefore provide the conservator with guidelines
in designing their treatment protocol.
The extent to which the treatments modify visual appear-
ance of the surface is dependent on the initial surface
condition. Visual changes derive from variations in
surface roughness at the micrometric and nanometric
scale. Small variations in surface roughness can produce
highly perceptible changes in surface appearance.
Polished surfaces proved to be susceptible to losses in the
quality of the reflection. Mechanical abrasion with Al2O3
produced the least modification in surface appearance
on the mirror-polished surfaces of the sample group.
Rough surfaces proved to be susceptible to losses as well
as gains in reflection. Mechanical abrasion with CaCO3
produced the least modification in surface appearance
on the mill-finished surfaces of the sample corpus.
Application of the assessment methodology
Scientific principles used in lab equipment for optical
and topographical characterisation of surfaces have been
adapted to suit the practical requirements of field-based
projects, therefore offering qualitative visual assessment
methods. For visual assessment, both graphic scale and
laser reflection methods are equally effective on flat,
small-to-medium sized surfaces; the laser-based method
423
is particularly suitable for assessments on more complex
or larger surfaces.
The practical application of the analytical methods in
the field will be limited by the logistical complexity of
manoeuvring cumbersome and sensitive equipment into
contact with more complex or larger target surfaces. For
some of the methods described, such as microtopog-
raphy, current research suggests indirect solutions, such
as making and analysing a high-resolution impression
of the surface.
Conclusion
Understanding the relationship between visual appear-
ance and surface roughness helps to describe the appear-
ance of aluminium surfaces quantitively and define
impacts of surface treatments on aluminium surface
finishes. In that sense, understanding that relationship
is key to defining the strategies for the conservation of
aluminium surface finishes.
The visual and analytical methods tested throughout this
research were successfully combined to assess the appear-
ance of aluminium surfaces. The methodology applied to
the various sample groups can also be applied to objects
and artworks, for both lab and field-based projects. It
provides a convenient decision and communication tool,
allowing all stakeholders in a conservation project to
communicate more objectively the otherwise subjective
characteristics of metallic surfaces on heritage artefacts.
This methodology has shaped the treatment decisions for
the microcar prototype and has since been successfully
used by the main author of this paper in other projects.
The influence of the selected treatments on the appear-
ance of aluminium surfaces depends primarily on the
initial surface aspect. All the treatments tested induced
more or less perceptible optical modifications, which
reflected even small variations in surface roughness.
The results provided by this project are mere stepping
stones. Interactions between surface appearance and
surface roughness require more in-depth study, particu-
larly to understand the role of irregularities at the nano-
metric scale. Conservation practice would also benefit
from research on a wider variety of aluminium alloys and
surface finishes, as well as other treatment options. Future
work can also lead to building a database supplemented
by the community of conservators.
424  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND MODERN METALS

Beyond aluminium alloys, the same methodology is easily

transferrable to other types of opaque materials where
conservation of the surface appearance is a prominent
concern (other metallic or synthetic materials).

References
AluMATTER. http://core.materials.ac.uk/repository/
alumatter/surface-properties.htm (accessed 28 October
2018).
Florescu, M. 2015. « Voyage en Carrosse ». Etude et
conservation-restauration d’un prototype de micro-car
en aluminium de Paul Arzens (Musée du Conservatoire
des Arts et Métiers, Paris ; en dépôt à la Cité de l’Auto-
mobile, Mulhouse). Evaluation de différents traitements
des surfaces en aluminium, en vue de l’obtention d’un
rendu uniforme des finitions d’origine. Mémoire de fin
d’études, Institut national du patrimoine, Paris, France.
International Organization for Standardization. 2012.
Geometrical product specifications (GPS) – Surface texture:
Areal – Part 2: Terms, definitions and surface texture
parameters (ISO 25178-2:2012, last revised 2017). www.
iso.org/obp/ui/#iso:std:iso:25178:-2:ed-1:v1:en.
Menezes, P.L., S.P. Ingole, M. Nosonovsky, S.V. Kailas, and
M.R. Lovel. 2013. Tribology for scientists and engineers,
1–39. New York: Springer.
Seve, R., M. Indergand, and P. Lanthony. 2007. Réflexion.
Facteur de réflexion. Facteur de réflectance. Réflectivité.
In Dictionnaire des termes de la couleur. Avallon: Editions
Terra Rossa.
Sheasby, P.G. and R. Pinner. 2001. The surface treatments
of aluminum and its alloys. Ohio: American Society for
Metals.
Zhenrong, Z., Z. Jing, and G. Peifu. 2010. Roughness
characterization of well-polished surfaces by measure-
ments of light scattering distribution. Optica Applicata
XL(4): 811–18.

Authors
Michaela Florescu is a metals conservator with an
MA in conservation from the Institut national du patri-
moine (Paris). She currently works as an independent
conservator for French and international museums and
is involved in several professional bodies (ICOM-CC,
ICON, INCCA-f). Her main interests include the conser-
vation of modern metals, industrial heritage, metallic
sculpture and contemporary art.
Preserving Aviation Heritage: The Need for an Approach
Based on Materials Analysis
Magali Brunet* Luc Robbiola
CEMES-CNRS Laboratoire TRACES UT2J
Université de Toulouse Université de Toulouse – CNRS
Toulouse, France Toulouse, France
magali.brunet@cemes.fr robbiola@univ-tlse2.fr
*Author for correspondence
Abstract
Most of the metals used in aircraft construction are aluminium
alloys. These alloys, poorly investigated for conservation
purposes, require specific characterisation due to the presence
of intermetallic precipitates at various length scales influencing
mechanical properties and resistance to corrosion, and
because of different surface treatments. This study undertook
an investigation into the different aluminium alloy parts of a
currently renovated Breguet aircraft and, as a result, revealed
the need to research and characterise these materials. An
adapted methodological approach was used and the results of
the materials characterisation focused on the main materials
Introduction
Preserving the industrial heritage of the 20th century
presents many challenges. The conservation of old aircraft
is one of them. Around the world, there are about 40
air and space museums with dedicated sheds and many
more private collections which are managed by different
associations. In most cases, aircrafts are kept outdoors
which very often leads to the corrosion of metal parts
(Rocca et al. 2010). To ensure the safety of visitors and
to present aircrafts as they would have once appeared,
corroded parts are replaced rather than restored. As a
consequence, the original parts of old aircrafts disappear
little by little without any traceability.
In this context, there is a real need to study these old
materials, mainly for two important reasons. Firstly,
for historical purposes: although industrial archives on
aeronautical materials exist, technical documents were
often not systematically preserved. Secondly, the study
of the metal parts of an old aircraft allows damage to be
understood by identifying the degradation processes.
This helps to ensure long-term conservation. Corrosion is
an important source of damage to these cultural heritage
materials that needs to be clearly understood prior to
deciding conservation protocols.
Philippe Sciau Monique Drieux-Daguerre
CEMES-CNRS Materia Viva
Université de Toulouse Toulouse, France
Toulouse, France materia-viva@orange.fr
philippe.sciau@cemes.fr
and on the corroded surfaces. Alongside performing material
analysis, archival documents were used to provide additional
and complementary historical data. From this, it was concluded
that a specific understanding of the alteration processes in
aluminium alloys is required in order to develop conservation
strategies.
Keywords
industrial heritage, aeronautical materials, aluminium alloys,
multiscale analysis
This study presents the investigation carried out to attain
these objectives by analysing the different metallic parts
from a 1950s’ aircraft undergoing renovation.
Method of study and sampling
For several years, the research team has been following
the renovation of a big carrier airliner – a Breguet 765
Sahara no. 504 64-PH (wing span: 45.5 m; length: 29 m;
height: 10.15 m; dry weight: 15 t) – which is a part of
the aircraft collection preserved by the Ailes Anciennes
Toulouse association. This collection represents aeronau-
tical adventure and local industrial history. The Breguet
765 (Figure 1a), a double-decker aircraft built in 1958 by
the local industrialist Louis Breguet for military usage
(Cochard 2016), is one of the most valuable aircraft
in the collection: only four were built in 1959 due to
drastic budget cuts by the French government. After its
last flight in 1974, this aircraft was stored outside at the
Evreux air base (Normandy, France). It was transported
after partial dismantlement from Evreux to Toulouse by
truck and was added to the collection of the Ailes Anci-
ennes Toulouse association in 1984, since when it has
been stored uncovered outdoors and subject to ongoing
426  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
renovation. The aircraft has been suffering from severe
corrosion for several decades. Figures 1c and 1d show
an example of a corroded stringer made of aluminium
alloy that had to be removed.
Figure 1. Photographs of (a) the Breguet 765; (b) the trailing edge; and
(c) and (d) a corroded stringer on the centre wing section that was
removed by a member of the association
After discussing the issue of corrosion with members
of the association, it was decided to focus the study on
parts made of aluminium alloy in order to understand the
specific corrosion involved and to propose a diagnosis.
Two lines of action were pursued: in-situ observation of
the evolution of the corrosion and the physicochemical
analysis of the collected parts.
Corrosion evolution was observed and documented on
various parts of the aircraft by taking pictures every six
months over a period of two years. For material anal-
ysis, some parts were collected after being removed by
members of the association. These parts suffered from
corrosion and needed to be replaced. The first step
was to document these pieces thoroughly, i.e. their
functional role (structural, cladding, etc.), their exact
location on the aircraft and environmental conditions.
In the second step, a full metallographic study was
carried out on selected representative parts: stringers,
plates, rivets (Cochard et al. 2017, Brunet et al. 2019).
It consisted, for the bulk, of determining the elemental
composition, the microstructure and the precipitation
by optical and electronic microscopy techniques at
different scales. For this, samples were embedded in
epoxy resin and mechanically polished. On the cross
section, the bulk metal, the different protective coatings
employed and the corrosion features could be observed.
Scanning electron microscopy was carried out with a
JEOL JSM 6460 LV operated at 20 kV and equipped
with an Oxford INCA PentaFETx3 energy dispersive
x-ray spectrometer (SEM-EDS). The use of destruc-
tive methods was possible because the removed parts
were numerous and were otherwise disposed of by the
MODERN METALS
association members. The results of the metallographic
study are summarised below. This article focuses on the
alterations and diagnosis.
Results
General observations on the alterations and on
the environment
A first observation of the aircraft revealed that corrosion
had affected the stringers in general, whereas the skin
(the outer surface of an aircraft’s fuselage and wings) was
better conserved (Figure 2a).
Figure 2. Photographs of (a) the interior; (b) exfoliation corrosion on
stringers; and (c) pitting corrosion on fuselage plates
Stringers are part of the primary structure of an aircraft.
Their role is to stiffen the fuselage, wings, empennage
and central section, etc. and they are generally flat or
possess L-shaped profiles riveted to the plates between
the fuselage frames. They are visible in the interior of the
Breguet 765 as the protection plates had been removed
for renovation purposes.
Environmental conditions inside the plane are quite
unstable as it has never been sealed since it was stored in
Evreux or in Toulouse and the humidity rates can easily
be high. On the outside, rain can reach unprotected
parts and water can remain stagnant in gaps. Although
all parts are covered by a protective coating, such as
primer (greenish) and paints (yellow/beige), corrosion
has developed locally.
For stringers, severe exfoliation is sometimes observed.
This corrosion causes the aluminium alloy to delaminate
and eventually turn into powder. For plates, whether
structural (skin of the wings, fuselage) or non-struc-
tural (shields, cladding), corrosion in the form of pits is
generally observed.
Figure 3 shows the evolution of the corrosion on particular
zones of the aircraft interior: a stringer (Figure 2c) and
a plate (Figure 2d).
PRESERVING AVIATION HERITAGE: THE NEED FOR AN APPROACH BASED ON MATERIALS ANALYSIS 427

higher than A-U4G (AIR 3350/C 1957) which justifies

its use for the primary structure of the aircraft.

Table 1. Elemental composition (wt%) of plates and stringers analysed

by SEM-EDS vs reference alloy composition provided in AIR 3350/C
(1957)

Elements
Al Cu Mg Mn Si Fe
in wt%
Structural Base 4.8–4.9 1.3–1.5 0.5–0.8 0.1–0.2 0.2–0.3
plates

Analyses
Non-structural Base 4.3 0.6-0.8 0.6-0.8 0.6-0.7 0.3–0.5
plates
Stringers Base 4.9–5.2 1.5–1.6 0.8–0.9 0.1–0.2 0.2
Figure 3. Photographs of (a) exfoliating corrosion on stringers (interior)
A-U4G Base 4 0.7 0.5 0.5 0.5 max

Ref.
and (b) pitting corrosion on the fuselage (interior) on March 2017 and
A-U4G1 Base 4.25 1.5 0.7 0.4 max 0.5 max
after two years (b and d)

There is a clear sign of the development of corrosion
on the stringers (about 1 cm per year), whereas for the
plates the evolution is not so clear: pit diameters remain
constant but with more corrosion products. According
to these observations, stringers seem to have a lower
resistance to corrosion than plates. To understand and
appreciate the extent of these specific alterations, several
stringers and plates were collected and studied.
Material analysis: Bulk alloy
In the first step, analysis by SEM-EDS on polished
samples provided identification of the bulk alloy of the
parts collected on the Breguet 765. The elemental compo-
sition for each alloy is given in Table 1. All collected
parts are found to be made of duralumin, a family of
alloys containing copper (3.5–5.5 wt%), magnesium
(0.4–1.5 wt%) and manganese (0.5–0.7 wt%). Over time,
these age-hardening alloys, with their excellent properties
(light weight and high strength, in particular), became the
material of choice for aircraft construction (Starke Jr and
Staley 1996). In France, an alphanumeric denomination
(A for aluminium, U for copper and G for magnesium
followed by the nominal content in wt%) was adopted in
the early 1940s (Lebouteux 1956) and this denomination
was followed by the aeronautic standard AIR 3350/C of
1957 (AIR 3350/C 1957). The alloys used in the Breguet
765 were thus identified according to this standard.
The collected stringers are made of A-U4G1. As for the
plates, A-U4G was used for the non-structural parts
whereas A-U4G1 was used for the parts in the primary
structure. Here, the choice of alloy was governed by the
mechanical properties resulting from its composition and
thermomechanical treatments. The ultimate strength,
yield strength and hardness of A-U4G1 is significantly
In the second step, the microstructure of the different
alloys was studied. Figure 4 shows representative SEM
images in backscatter mode for a plate made of A-U4G
(Figure 4a) and a stringer made of A-U4G1 (Figure 4b).
Coarse intermetallic particles analysed by EDS were iden-
tified as S-Al2CuMg, AlCuMgSi, θ-Al2Cu and particles
containing Al, Cu, Fe, Mn and Si with varying stoichi-
ometry. It should be noted that in A-U4G alloys, where
the magnesium content is lower, the S-Al2CuMg phase
is absent. Coarse intermetallic particles are deleterious
to the macroscopic properties, potentially causing cracks
and corrosion (Hughes et al. 2011, DeRose et al. 2013). It
is therefore important to document them in terms of size
and density to understand the alteration in old alloys. In
stringers, particles are smaller and more numerous than
in plates (see Table 2 for statistics).
Figure 4. SEM images (backscatter electron mode) of collected pieces:
(a) plate made of A-U4G; (b) stringer made of A-U4G1
Table 2. Average density, size and area of particles as a fraction of the
total alloy surface for a representative plate and stringer
Density Particle size
Area fraction
(/mm²) (µm²)
Plate (A-U4G) 12000 1.9 ± 4.2 2.3%
Stringer (A-U4G1) 31200 1 ± 1.7 3.1 %
After etching in Keller’s solution (HF 1 mL, HNO3 2.5 mL,
HCl 1.5 mL in 95 mL demineralised water), metallic
grains were revealed.
428  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND MODERN METALS

as well as aluminium sulfates (Graedel 1989). The blue

colour comes from the copper ions released by the
alloy. When observing these corrosion features on cross
sections in larger scale with SEM (Figure 7), it is clear that
intergranular corrosion has developed on both the plates
and stringers. On the plates, the intergranular corrosion is
Figure 5. Optical microscope images after Keller’s reagent etching: (a) coupled with evolving pitting in the crevices (Figure 6b),
plate; (b) stringer at 5× and 50× while on the stringers the intergranular corrosion further
develops during exfoliation (Figure 6d).
The microstructure observed in the different parts can
partially reveal the fabrication process. Plates have equi-
axed grains (Figure 5a) typically obtained after successive
rolling passes and heat treatments leading to recrys-
tallisation, whereas stringers exhibit a highly oriented
fibrous structure in the bulk and large grains on parts
near the surface up to a depth of a couple of hundred
microns (Figure 5b). This peculiar microstructure is
linked to the manufacturing process where extruding
generates severe plastic deformation giving rise to a very
fine and oriented texture – the presseffekt (Herenguel
and Scheidecker 1954, Develay 1997) – to create higher
mechanical resistance. In subsequent heat treatments, the
microstructure undergoes an inhomogeneous recrystal- Figure 6. OM images on plates: (a) top view; (b) cross section showing
lisation with the formation of large grains on the cortex, pitting corrosion; and on stringers: (c) top view, (d) cross section showing
as observed here. These large grains induce a reduction intergranular corrosion leading to exfoliation

in mechanical properties and also influence resistance

to corrosion.

In terms of protective coatings, and depending on

whether the part was for internal or external use, it was
found that anodisation and/or a primer was applied
underneath several layers of paint. The full characterisa-
tion of the parts can be found in a previously published Figure 7. SEM images (backscatter electron mode) of corroded parts: (a)
work (Brunet et al. 2019). plate made of A-U4G; (b) stringer made of A-U4G1

Alterations and diagnosis The observed forms of corrosion (pitting, intergranular,

exfoliation) can be explained by environmental and
As mentioned above, two types of alterations (pitting
intrinsic factors (the composition and microstructure of
and exfoliation) were identified macroscopically on
the alloy). According to the literature, when exposed to
the parts collected from the Breguet aircraft. These
air, an amorphous oxide film (Al2O3) is formed instan-
forms of corrosion are highly representative of outdoor
taneously on the aluminium surface. Upon prolonged
corrosion of aluminium alloys (De Damborenea and
exposure to humidity, the film grows and transforms
Conde 1995, De La Fuente et al. 2007). Figure 6 shows
into hydroxides (boehmite, bayerite, nordstrandite, etc.)
pictures of the corrosion, via a top view and in cross
(Graedel 1989). The resulting film is highly porous. In
section, on plates (Figures 6a and 6b) and on stringers
atmospheres that contain corrosive salts, humidity and
(Figures 6c and 6d).
constituents such as sulfur compounds and chlorides
Inside pits, white and blue corrosion products were are absorbed and cause the film to rupture. By reaching
observed. They are known to be hydrated variants of the bulk metal, these contaminants may generate pitting
aluminium oxides and hydroxides (Degrigny 1991), corrosion (De la Fuente et al. 2007).
PRESERVING AVIATION HERITAGE: THE NEED FOR AN APPROACH BASED ON MATERIALS ANALYSIS
In duralumin alloys, different types of intermetallic
particles are present (as shown in Figure 4) with different
dissolution potentials (Hughes et al. 2011). Copper-rich
particles (θ-Al2Cu) and iron-rich particles (Al-Cu-Fe-
Mn-Si) are cathodic compared to the solid solution,
whereas Mg2Si and S-Al2CuMg particles are anodic. In
the presence of aqueous electrolytes, micro-galvanic
coupling occurs between these particles and the matrix.
This can occur readily after the protective coating is
broken allowing humidity to penetrate. Anodic particles
dissolve preferentially. Cathodic particles, on the other
hand, deplete copper concentration in the surrounding
matrix at microscale. This copper-depleted zone is more
anodic compared to the matrix and oxidises easily, with
the matrix dissolving preferentially. The case of S-Al-
2
CuMg particles is specific: preferential dissolution of
magnesium and aluminium creates a copper-rich zone
around the particle. This zone, more cathodic than the
particle, stimulates secondary pitting (Guillaumin and
Mankowski 1998).
All of these phenomena are localised corrosion processes
originating from surface heterogeneities inducing either
pitting or intergranular corrosion. Intergranular corro-
sion develops specifically when grain boundary regions
are preferentially attacked, due to the presence of inter-
metallic particles at grain boundaries or the presence of
a precipitate-free zone, as well as other factors such as
misorientation (Hughes et al. 2011).
In the case of the parts from the Breguet 765, intermetallic
particles are more numerous and denser on stringers
than on plates (Figure 7), which partly explains why
corrosion is an issue on stringers. To explain why the
intergranular corrosion transforms into exfoliation on
stringers, an important factor is the microstructure linked
to the manufacturing process and heat treatments. On
plates, exfoliation does not occur because of the homo-
geneous microstructure obtained after recrystallisation
annealing. On stringers, due to the particular oriented
and fibrous microstructure mentioned above, the inter-
granular corrosion develops easily into exfoliation due
to the growth of corrosion products which force metal
to move away from the body of the material.
To summarise, intergranular corrosion begins in the
intermetallic precipitates present in the alloys as well as
in the grain microstructure. It is a self-generating process
as long as humidity can penetrate and thus very difficult
to stop. The nature of the alloy is important: A-U4G1
429
containing S-Al2CuMg particles would be expected to
be less resistant to corrosion; however, similar corrosion
features could be observed on different alloys of the
same family, i.e. A-U4G and A-U4G1, which leads to
the conclusion that thermomechanical treatment during
fabrication was, in this case, a determining factor for the
corrosion mechanism.
Discussion
Corrosion affects aluminium alloys differently depending
on many factors: intrinsic factors (such as chemical
composition and thermomechanical treatments during
fabrication) and extrinsic factors (such as protective
coating efficiency, mechanical stress during usage and
environmental conditions). Due to these specificities
of aluminium alloys, it remains difficult, at first sight,
to develop a global conservation strategy. Obtaining
as complete a view as possible of the materiality of the
cultural heritage object, via sampling and/or archival
research, is therefore a particularly important step to
evaluate the potential risks of alteration. In fact, when
considering the long-term preservation of old aircraft
exposed to the elements, decisions on conservation
protocols are potentially easier to make if knowledge on
the resistance to corrosion of specific aluminium alloys
is acquired upstream. In the case of the Breguet 765,
thanks to the study on various collected parts, a better
understanding of factors which caused corrosion was
achieved. Following this, several preventive conservation
measures were transmitted to the association members.
In particular stringers, because of their low resistance
to corrosion, have to be closely controlled. More gener-
ally, for aircraft kept outside, humidity/rain/pollution
penetrates easily through damaged coatings initiating
corrosion. Original protective coatings must therefore
be conserved as far as possible.
When corrosion is detected, the methodology currently
used by the association’s members is the following:
polishing with a non-metallic brush until corrosion prod-
ucts are completely removed, degreasing of the area and
immediate application of an epoxy-base primer (used in
aircraft industry maintenance). This method is not one
hundred percent efficient as corrosion can start again.
When the corrosion is more severe, such as exfoliation,
the association’s members have no choice but to change
the affected parts, as this type of corrosion is difficult to
stop. For this particular aircraft, it means changing a lot
of parts, which is far from ideal.
430  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Right now, adapted methods need to be defined upstream
by the conservation-restoration community, together
with renovation organisations, to conserve aluminium
alloys in cultural heritage artefacts exposed to outdoor
conditions. In particular, a methodical process should
be undertaken to ensure aluminium alloys are exposed
to corrosion conditions as little as possible.
Conclusion
This study highlights the interest in developing a major
strategy when investigating complex industrial and
cultural heritage. The relationship between the role of
the parts (structural or non-structural) and the choice of
materials was revealed through material analysis coupled
with researching the constructor’s technical documents.
It was found that A-U4G and A-U4G1 were employed
depending on the desired mechanical characteristics.
Corrosion on the plates and stringers was then investi-
gated and it was observed that the exfoliation on the plates
is closely related to the microstructure created during the
fabrication process. It was therefore shown that identi-
fying the chemical nature and the microstructure of the
alloys was a necessary step to a better understanding of
the alterations involved.
Conservation and restoration practices on ancient aircraft
remain complex due to the size of the aircraft and the high
number of alloy types. However, similar aluminium alloys
can be found in many aircraft from different periods.
Knowing in advance what the potential risks are for one
part or another due to the chemical nature, fabrication
process or environmental conditions is a useful source
of information to be filed in a database and should help
actors in the conservation field facing similar problems
in the future.
Acknowledgements
The authors would like to thank the volunteers of the
Ailes Anciennes Toulouse association, especially the
Breguet Deux-Ponts team. This study was partially
supported by the NEXT grant (no. ANR-10-LABX-0037)
within the framework of the Programme des Investisse-
ments d’Avenir.
References
Air 3350/C. 1957. Tableaux des alliages d’aluminium et
des alliages de magnésium, édition n.° 4 du 15 juin 1957.
Ministère de la Défense Nationale et des Forces Armées,
MODERN METALS
Secrétariat d’Etat aux Forces Armées (AIR), Direction
Technique et Industrielle.
Brunet, M., A. Cochard, C. Deshayes, C. Brouca-Cabar-
recq, L. Robbiola, J.-M. Olivier, and P. Sciau. 2019. Study
of post-World War 2 French aeronautical aluminium
alloy and coatings: Historical and materials science
approach. Studies in Conservation [in press].
Cochard, A. 2016. Microstructure et propriétés mécan-
iques des alliages de type Duralumin du Breguet 765 n.°
504 64-PH. PhD dissertation, Université de Toulouse,
France.
Cochard, A., K. Zhu, S. Joulie, J. Douin, J. Huez, L.
Robbiola, P. Sciau, and M. Brunet. 2017. Natural aging
on Al-Cu-Mg structural hardening alloys – Investigation
of two historical duralumins for aeronautics. Materials
Science and Engineering: A 690: 259–69.
De La Fuente, D., E. Otero-Huerta, and M. Morcillo.
2007. Studies of long-term weathering of aluminium in
the atmosphere. Corrosion Science 49(7): 3134–48.
De Damborenea, J. and A. Conde. 1995. Comparison of
accelerated and atmospheric exposure tests for corrosion of
aluminium alloys. British Corrosion Journal 30(4): 292–96.
Degrigny, C. 1991. Conservation et stabilisation d’alli-
ages d’aluminium prélevés sur des épaves aéronautiques
immergées en eau douce. Conservation Restauration des
Biens Culturels 311: 27–39.
DeRose, J.A., A. Balkowiec, J. Michalski, T. Suter, K.J.
Kurzydlowski, and P. Schmutz. 2013. Microscopic and
macroscopic characterisation of an aerospace aluminium
alloy (AA2024). In Aluminium alloy corrosion of aircraft
structures. Modelling and simulation, eds. J.A. DeRose,
T. Suter, T. Hack, and R.A. Adey, 23–38. Southampton,
Boston: WIT Press.
Develay, R. 1997. Mise en forme de l’aluminium – Filage
ou extrusion. In Techniques de l’ingénieur. Mise en
forme des aciers, aluminium, autres métaux et alliages,
TIB356DUO, ref. M651. www.techniques-ingenieur.fr/
base-documentaire/materiaux-th11/mise-en-forme-des-
aciers-aluminium-autres-metaux-et-alliages-42356210/
mise-en-forme-de-l-aluminium-m651/ .
Graedel, T.E. 1989. Corrosion mechanisms for aluminum
exposed to the atmosphere. Journal of Electrochemical
Society 136(4): 204C–212C.
Guillaumin, V. and G. Mankowski. 1998. Localized
corrosion of 2024-T351 aluminium alloy in chloride
media. Corrosion Science 41(3): 421–38.
PRESERVING AVIATION HERITAGE: THE NEED FOR AN APPROACH BASED ON MATERIALS ANALYSIS 431

Herenguel, J. and M. Scheidecker. 1954. Influence d’une Philippe Sciau is a research scientist in materials science
addition de zirconium sur la texture cristalline des at CNRS, working at CEMES in the Cultural and Indus-
demi-produits en alliage d’aluminium filés à la presse et trial Heritage Materials team.
traités thermiquement. La Revue de Métallurgie 51(3):
Monique Drieux-Daguerre has been the head of
173–78.
Materia Viva since 2001, a conservation laboratory in
Hughes, A.E., N. Birbilis, J.M.C. Mol, S.J. Garcia, X. Zhou, Toulouse specialising in the preservation of archaeolog-
and G.E. Thompson. 2011. High strength al-alloys: Micro- ical and historical heritage, in particular metals. She has
structure, corrosion and principles of protection. In Recent a master’s in Archaeology and Art History and a master’s
trends in processing and degradation of aluminium alloys, in Conservation of Cultural Heritage from the University
ed. Z. Ahmad, 223–62. IntechOpen. www.intechopen. of Paris 1 Panthéon-Sorbonne.
com/books/recent-trends-in-processing-and-degrada-
tion-of-aluminium-alloys/high-strength-al-alloys-mi-
crostructure-corrosion-and-principles-of-protection .
Lebouteux, H. 1956. La symbolisation de l’aluminium et
de ses alliages en France et dans quelques pays étrangers.
In Encyclopédie du travail de l’aluminium, extract of the
Revue de l’Aluminium no. 227, 1–3. Paris.
Rocca, E., F. Mirambet, and C. Tilatti. 2010. Long-term
corrosion of aluminium materials of heritage: Analysis
and diagnosis of aeronautic collection. Corrosion Engi-
neering Science and Technology 45(5): 345–49.
Starke Jr, E.A. and J.T. Staley. 1996. Application of modern
aluminium alloys to aircraft. Progress in Aerospace
Sciences 32(2–3): 131–72.

Authors
Magali Brunet is a research scientist in materials science
at CNRS (Centre national de la recherche scientifique)
who has worked at CEMES (Centre d’Élaboration de
Matériaux et d’Etudes Structurales) on the Cultural and
Industrial Heritage Materials team since 2015. She studies
alterations to old materials, in particular aluminium
alloys, using physicochemical analysis techniques. She
holds a PhD in Microelectronics Engineering from Cork
University (2003) and a master’s degree in Conservation
and Restoration of Cultural Heritage from the University
of Paris 1 Panthéon-Sorbonne (2015).

Luc Robbiola is a research engineer in cultural mate-

rials science at CNRS. He holds a PhD in Metallurgy
(1990) and a master’s degree in Conservation and
Restoration of Cultural Heritage from the University of
Paris 1 Panthéon-Sorbonne (1985). He presently works
at Toulouse University in the TRACES (Travaux et
Recherches Archéologiques sur les Cultures, les Espaces
et les Sociétés) laboratory for the new ArchéoScience
platform.
POSTER SESSION
434  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conserving Conservation: The Critical Application of
Traditional Metalworking Skills and the Importance of
Knowledge Transfer to New Generations of Conservators
Jill Saunders* Hal Jacob
Plowden & Smith Plowden & Smith
London, UK London, UK
jill.saunders.09@ucl.ac.uk hal.jacobs@plowden-smith.com
*Author for correspondence
Contemporary conservation of metallic heritage has
emerged from an extensive body of highly skilled
construction, repair, and finishing techniques, some
of which have been employed for centuries, if not
millennia. In many contexts, the application of traditional
methods and their modern-day counterparts remains
vital. Conserving working objects, outdoor monuments,
items on open display, as well as handling collections and
historic interior features (e.g. furniture fittings, etc.), may
involve re-patination, tinning, blueing, or traditional
structural repairs. However, a working knowledge of
such processes is hugely beneficial even when this level of
intervention may not be necessary. It equips conservators
with an intimate understanding of the methods by which
most objects, whether ancient artefacts or contemporary
art, have been originally made, or indeed historically
repaired. A conservator experienced in soldering, pati-
nation, or plating may be more likely to know obscure
techniques, recipes, and/or methodological details which
contribute to current condition issues – or indeed be
likely to affect the success of potential treatments or
the impact of different environments. Such skills can
also give conservators the confidence to conduct more
extreme yet ultimately beneficial treatment steps (e.g. the
safe dismantling and reconstruction of a medieval gun
to remove rust and treat active corrosion on otherwise
inaccessible internal fittings). Of course, experience of
POSTER SESSION – TECHNICAL STUDIES
Peter (Mick) Willett †
Plowden & Smith
London, UK
manufacturing techniques and their implications will be
naturally accumulated over the course of a conservator’s
career and a great deal can be learnt from research. Yet,
even the most sophisticated observational or theoretical
knowledge cannot be equated with practical proficiency.
However, often there are insufficient resources – whether
facilities or time – to include these interventive skills in
degrees and educational courses, especially at a point
in time when ethical theory recommends caution with
regard to activities which alter material values. None-
theless, this may result in the undesirable incongruity
of highly educated and experienced metals conservators
without what are in one sense quite basic skills in their
own material specialism. There is also an irony to the
threatened disappearance of such valuable knowledge
within a profession which considers protecting the
longevity of intangible socio-cultural assets a central aim.
Traditional techniques should be a key continuing profes-
sional development topic for metals conservators. This
poster presents captivating case studies of metalworking
and smithing skills in action and explores key ethical
issues and object complexities. Examples incorporate a
range of fascinating methods such as tinning, as well as
various repairs including the handmade manufacture
and replacement of missing object parts.
† Deceased April 2019.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Two Iron Age Barrel Bracelets – Investigation, Technology
and Conservation
Irina Marilyn Huller* Kathrin Schmidt
Institute of Conservation Institute of Conservation
University of Applied Arts Vienna University of Applied Arts Vienna
Vienna, Austria Vienna, Austria
irina.huller@student.uni-ak.ac.at kathrin.schmidt@uni-ak.ac.at
*Author for correspondence
In the autumn of 2012, two Iron Age barrel bracelets
(named after their bulbous shape) were found together
with 130 other prehistoric metal objects alongside a
prehistoric route in the municipal area of Ehrwald, Reutte
district, Tyrol (Austria). The bracelets date back to the
late Hallstatt period, around 600 BC, and are made of an
embossed bronze sheet with a side opening and repoussé,
punched and engraved patterns.
These bracelets were found in an unusual state as they
were intertwined through their side openings. As a result,
the overlapped areas were better preserved, whereas the
remaining exterior surface was heavily corroded, fragile,
slightly deformed and had suffered more material losses.
The aim of the conservation treatment was to prevent
further material loss and to prepare the objects for
display. The decision was made not to separate the two
very fragile bracelets, as they support each other in the
current position.
In order to gain the necessary information to develop a
POSTER SESSION – TECHNICAL STUDIES  435
Marta Anghelone Gabriela Krist
Institute of Conservation Institute of Conservation
University of Applied Arts Vienna University of Applied Arts Vienna
Vienna, Austria Vienna, Austria
marta.anghelone@uni-ak.ac.at gabriela.krist@uni-ak.ac.at
proper conservation strategy, the bracelets were inves-
tigated by x-ray diffraction and scanning electron
microscopy with energy dispersive x-ray spectroscopy.
The corrosion products were identified as mushistonite
(CuSn(OH)6), a copper-tin hydroxide, and traces of
brochantite (Cu4(SO4)OH6).
The surface was cleaned with deionized water and etha-
nol-wetted cotton swabs, as well as mechanically with fine
tools. To secure the corroded surface and fragile edges
of the bronze sheet, an acrylic resin (Paraloid B-44) was
selected and applied in different concentrations. In addi-
tion, silk fabric (crepeline) and glass fibers were used to
support the exposed edges and fragile re-attached pieces.
After the successful conservation treatment, the bracelets
will be presented in a local exhibition in Ehrwald. For
this purpose, an individual display case and a transpor-
tation box were designed to ensure the necessary climatic
conditions are met. The transportation box can also be
used for storage.
436  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – TECHNICAL STUDIES

Gilding Iron in the 16th Century

Dominique Robcis* Emmanuel Plé

Centre for Research and
Restoration of the Museums of
France (C2RMF)
Paris, France
dominique.robcis@culture.gouv.fr
*Author for correspondence
Centre for Research and
Restoration of the Museums of
France (C2RMF)
Paris, France
emmanuel.ple@culture.gouv.fr

On 16th-century gilded iron objects, two techniques
are generally observed: so-called damascening and leaf
gilding. The technical study on François de Montmor-
ency’s armour from Draguignan Museum prior to resto-
ration was an opportunity to examine another gilding
technique still little studied to date: mercury amalgam
on a plated copper substrate.
While fire gilding on copper alloys and silver is well
known, its use on iron raises difficulties. This is due to
the fact that it is technically impossible to gild mercury
amalgam directly onto iron, as mentioned in Roret’s
manual on gilding and silvering techniques. However,
study of the armour in the Draguignan Museum by
x-ray fluorescence spectroscopy, particle-induced x-ray
emission and Rutherford backscattering spectrometry
showed the presence of mercury in the gilding alongside
copper in significant proportions. On the basis of this
observation, the possibility of copper plating prior to
gilding with amalgam was proposed.
Study was divided into two areas. First, an experimental
approach was conducted through the creation of a tasset
and its decoration. Chemical copper plating was carried
out using a solution of copper sulfate followed by gilding.
Scientific analysis showed results close to those for
the armour. Second, a historiographical approach was
used to find a very precise description of the process.
This was discovered in Vannoccio Biringuccio’s De la
pirotechnia, written at around the same time Montmor-
ency’s armour was created (circa 1560–70):
Still you can make your iron gold by virtue of another
water made of grey-green copper, tartar, and common
salt, adding white wine to it: and in this water bath your
iron till it is the colour of copper, but the iron must
first be very polished and clean; then place it back in
this water and dry it well; and after being heated, rub
it with mercury in which gold has been dissolved; and
then heat it until the mercury has evaporated.
Comparative analysis conducted on several suits of
armour in the musée de l’Armée, Invalides, Paris, showed
the same characteristics. The corpus was then expanded
by examining several pieces from the Renaissance
Museum in Ecouen, and the same process was found
on early 16th-century militaria, and also on the famous
jeweller’s bench made by Leonhard Danner for Prince
Augustus of Saxony in 1565.
In conclusion, this quite specific technique of gilding
with amalgam on iron was previously mostly associated
with the 18th century and afterwards. This study shows
that it was in fact used well before this and was probably
much more widespread than previously suspected.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Two Bronze Statuettes from the Estense Gallery:
Materials, Casting Technology and Conservation
Intervention
Cristina Chiavari Carla Martini*
Cultural Heritage Department Industrial Engineering Department
University of Bologna University of Bologna
Ravenna, Italy Bologna, Italy
cristina.chiavari@unibo.it carla.martini@unibo.it
Two small bronze figures (Kouros, Corinth or Magna
Graecia, 5th century BC; Proserpina, South Etruria,
2nd century BC) from Estense Gallery, Modena (Italy),
underwent a technical study as part of an ongoing
conservation project whose results should facilitate
subsequent comparison with similar artifacts from the
same period. Three-dimensional digital microscopy was
used to assess the surface condition and identify features
related to the casting technology and/or the artificial
patination method. Surface layers were identified by
Raman micro-spectroscopy, while the metal microstruc-
ture of both figures was determined by in-situ optical
observation after minimal polishing. Scanning electron
microscopy coupled with energy-dispersive x-ray spec-
troscopy was used on the smallest of the two figures,
i.e., the Kouros, to identify the alloy composition, while
x-ray radiography was carried out on the hollow-cast
Proserpina bronze.
Both statuettes consisted of a ternary Cu-Sn-Pb alloy,
with the typical as-cast cored dendritic microstructure
where the alpha + delta eutectoid and insoluble Pb
globules are preferentially located in inter-dendritic
spaces. The alloy for the Kouros was Cu-15Sn-2Pb.
Typical compounds related to archaeological exposure,
mainly consisting of thin layers of copper oxide (cuprite)
POSTER SESSION – TECHNICAL STUDIES  437
Lorenzo Morigi
Private metals conservator
Bologna, Italy
lorenzo@morigi.it
*Author for correspondence
and carbonates (malachite and cerussite), together
with soil-related species (such as silicates, calcite, and
quartz), were detected on both the bronze surfaces. The
dendritic structure of the Kouros was already detectable
by 10× microscopic observation of the unetched surface
due to mild preferential corrosion of the Cu-rich bronze
matrix. The presence of such an undeformed dendritic
microstructure, with no evidence of cold working by
hammering or chiseling, could be evidence that only
minimal polishing by mild abrasion was performed after
casting in order to obtain a shiny surface. The fine details
in the hair of the statuette are clearly due to the wax
model, with no (or negligible) cold working. Inside the
Proserpina, clear traces of indirect casting were visible,
suggesting small serial production and quite advanced
casting techniques.
The conservation intervention mainly consisted of
surface cleaning under stereomicroscope, followed by
the creation of a tailored support for the Proserpina.
Three-dimensional scanning and CAD/CAM systems
were used to produce a dedicated support system located
inside the lower part of the bronze statuette. This support
system did not affect the original material as it was based
on an expansion mechanism using the internal irregu-
larities of the surface to guarantee adequate grip.
438  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – TECHNICAL STUDIES

Bells in Use: From Living Heritage to Legend

Daniela Lira* Gabriela Peñuelas

National School of Conservation,
Restoration and Museography
Mexico City, Mexico
daniela_lira_p@encrym.edu.mx
*Author for correspondence
National School of Conservation,
Restoration and Museography
Mexico City, Mexico
gabriela_penuelas_g@encrym.
edu.mx

In Mexico, the conservation of bells has recently come
under analysis by several specialists in architecture and
restoration as well as anthropologists. After the 2017
earthquakes, which affected seven of the country’s states,
bells were placed at risk because of damaged belfries.
The authors realized that the decay of this particular
kind of heritage is indissolubly linked to cultural prac-
tices. In many cases, conservators assume that bells are
only designed to be rung at particular hours and for the
local congregation, but the variety of ringing is related
to the religious beliefs and traditions and customs of
a society.
The authors, as conservators, emphasize the need to
discuss three principal factors in order to understand
and preserve bells according to their symbolic function:
their sound, the way they are rung, and the reasons for
doing so.
First of all, the sound is related to the alloy employed in a
bell’s manufacture, as well as a bell’s shape and thickness.
It is also related to the alloy in the clapper and the way
the bell ringer uses it. Because of the way a bell is rung,
with the constant striking of the clapper against the sound
bow, many bells can crack, and all of them are constantly
subject to tensions that might cause damage to other
parts of the bell like the crown or the crown staple. Does
a crack therefore imply that a bell has lost its function?
Should we conserve the material or the significance?
How do we find a balance between the significance and
the material aspects? The study examines cases where
material damage has varying implications in terms of
the use of a bell. Cuentepec, for instance, has a bell with
a hole in it produced by a cannon shot which is used by
people for particular announcements. Nayar Church has
two bells, the most damaged of which is still in use and
has greater sacred significance. Finally, the cracks and
decay of the bells at Escobedo prevent them from being
rung. Traditionally, conservators used to think that if a
bell was cracked, then the sound would also be compro-
mised and the bell would no longer be used. Conservators
need to be aware and open to all possibilities in order
to understand customs and cultural significance and to
act accordingly.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Preliminary Results of Research on 17th–19th-Century
Coffin Portraits from the Collection of the National
Museum in Krakow
Maria Labut* Maria Goryl
National Museum Faculty of Conservation and
Krakow, Poland Restoration of Works of Art,
mlabut@mnk.pl Academy of Fine Art
Krakow, Poland
*Author for correspondence
maria@goryl.eu
Coffin portraits first appeared in Poland at the end of
the 16th century and their popularity in the 17th–18th
centuries owes to an old Polish funeral rite related to the
death of members of the nobility. The occurrence of this
type of portrait was a characteristic feature of Poland
during the Sarmatian era, when the nobility sought to
trace their roots back to ancient families to justify their
social superiority. Depending on the wealth and rank of
the deceased, the funeral ceremony was often prepared a
few weeks or even months in advance. In some cases, the
expense could be equivalent to that required to purchase
several villages. The funeral resembled a show whose
centerpiece was the castrum doloris (castle of grief),
consisting of a decorative catafalque decorated with coats
of arms, inscriptions, and a coffin with portraits attached
to the shorter side. These portraits were very realistic to
evoke the impression of the deceased’s presence during
the ceremony and, at the end, were attached to the church
walls with the coats of arms and epitaphs. The portraits
were usually made on a hexagonal or octagonal metal
substrate with a simple or more decorative frame. The
metal sheets were delivered to the painter by a tinsmith,
POSTER SESSION – TECHNICAL STUDIES  439
Julio del Hoyo-Melendez Aldona Stepien
National Museum National Museum, Krakow, Poland
Krakow, Poland aldona.stepien1@gmail.com
jdelhoyo@mnk.pl
Katarzyna Stefanczyk
Research and Development Centre
for Renovation and Conservation
Nysa, Poland
katarzyna.stefanczyk@pwsz.nysa.pl
while copper or silver sheets were ordered from a gold-
smith. The portraits had a shiny metal background (rarely
painted) with inscriptions and coats of arms. Their artistic
level depended on the painter’s talent, who remained
mostly anonymous.
This ongoing project focuses on research conducted on
ten coffin portraits from the 17th to early 19th century.
Due to the large time frame, differences can be observed
in the way the portraits were made, as well as in their
style. The compositions of the metal alloy and painting
layers were analyzed using scanning electron microscopy
coupled with energy-dispersive x-ray microanalysis,
x-ray fluorescence spectrometry and Fourier transform
infrared spectroscopy. The stratigraphic structure of the
portraits was determined on the basis of cross sections
and analytical photography in the visible, infrared and
ultraviolet areas of the electromagnetic spectrum. These
studies allowed a better understanding of the technique
and technology used to create these objects. Due to
their uniqueness, as well as their cultural and historical
significance, an in-depth study of these objects was
considered necessary.
440  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
All That Glitters: Treatment and Technical Study of an
Ormolu Surtout de Table
Sarah Barack* Jessica Walthew
Cooper Hewitt Smithsonian Design Cooper Hewitt Smithsonian Design
Museum Museum
New York NY, USA New York NY, USA
baracks@si.edu walthewj@si.edu
*Author for correspondence
Cooper Hewitt Smithsonian Design Museum (CHSDM)
recently conserved a spectacular ormolu  surtout de
table (ca. 1805) by Pierre-Philippe Thomire (1751–1843),
chaser-engraver to Napoleon (ciseleur de l’Empereur).
This elaborate dining-table centerpiece consists of an over
ten-foot-long, silver-leaf mirrored tray set in an ormolu
frame and over 50 gilt-brass elements comprising tiered
servers with cut-glass bowls, candelabra, and other deco-
rative elements that sit atop the mirrors. Thomire’s inven-
tive approach to iconography is apparent throughout the
metalwork, as shapes morph from graceful neo-classical
figures to fantastical hybrids of animals and architectural
elements. While stable prior to treatment, thin layers of
soiling, grime, and oxidation products obscured the care-
fully chased details, thereby attenuating the dramatic play
between burnished and matte surface. The Smithsonian
Women’s Committee awarded a grant for the treatment
of the surtout, and a metals conservator was contracted
to lead the ormolu conservation. The team performed
extensive tests to determine effective cleaning methods,
eventually comparing two gel systems and concerns
related to their full clearance.
Though treatment focused on cleaning the metal surfaces,
POSTER SESSION – TECHNICAL STUDIES
Jakki Godfrey
Boro 6 Art Conservation
Jersey City NJ, USA
jakki@boro6artconservation.com
the project provided the opportunity to conduct a tech-
nical study using advanced imaging techniques. Varied
chasing techniques are used with great sophistication
to contrast finishes of satiny flesh and finely deline-
ated feathers with burnished flat surfaces. Toolmarks
were documented with reflectance transformation
imaging, and previous repairs and replacement parts
(likely surmoulages, a copy made from a finished object
rather than from the original mold) were observed.
Three-dimensional scanning, performed by New York
University’s LaGuardia Studio with a FARO Edge
ScanArm, facilitated closer study of the original and
recast components, and allowed exact measurements to
be taken and compared.
The final, ongoing component of the project focuses
on public outreach. CHSDM’s in-house videographer
documented several steps of the project, and a didactic
video will be featured in the exhibition. Technical images
are also being shared through our varying social media
channels in order to illustrate how conservators study and
record artists’ working processes. In this way, cutting-
edge technology will be used to celebrate 19th-century
expert craftsmanship.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – TECHNICAL STUDIES  441

Marine Precious Metals After Excavation

Janneke van der Stok*
University of Amsterdam
Amsterdam, The Netherlands
janneke@metals-inc.nl
*Author for correspondence
Maarten van Bommel Ineke Joosten
University of Amsterdam Cultural Heritage Agency of the
Amsterdam, The Netherlands Netherlands
m.r.vanbommel@uva.nl Amsterdam, The Netherlands
i.joosten@cultureelerfgoed.nl
Tonny Beentjes
University of Amsterdam
Amsterdam, The Netherlands
t.p.c.beentjes@uva.nl

Archaeological precious metal artefacts contain a wealth
of information. The surface, often with its original deco-
ration and plating, is especially important. However, that
same surface is subject to degradation and interventive
conservation treatments like polishing can have a negative
influence on the readability of the metal. It is therefore
important that relevant information about the object
and its context is not lost after excavation and the best
ways found to extract it before post-excavation degrada-
tion and/or conservation treatments. This is the aim of
the AMOR (translated as ‘archaeological metal surface
research’ from the original Dutch) project which started
in September 2018.
As part of the project, 17th-century silverware, retrieved
from a shipwreck recently found in the Netherlands, have
been investigated using multiple analytical techniques.
These archaeological artefacts probably belong to a
well-preserved 17th-century personal ensemble, a unique
set with no known historical parallels. The items are
also interesting from a material-technical point of view
since they are in practically pristine condition. During
the technological investigation, special attention has
therefore been paid to the occurrence and application
of the decorative techniques, like engraving and gilding.
The purpose of this research is to create a better under-
standing of the artefacts and their mutual association and
also their connection to the other finds from the wreck.
Some of the objects have recently been treated while
others have remained untreated. This group of silver
artefacts therefore offers a unique interdisciplinary
opportunity to study historical technology and the
establishment of an optimal conservation strategy for
precious metals from a marine context at the same time.
This poster highlights the silver artefacts, the conser-
vation methods used and the different possibilities for
extracting technological information.
442  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Developing a Sustainable Approach for Screening and
Studying Functional Metal Objects
Silvia Russo* Eleonora Balliana
DAIS, Ca’ Foscari University of Venice DAIS, Ca’ Foscari University of Venice
Venice, Italy Venice, Italy
scomegaia@gmail.com eleonora.balliana@unive.it
*Author for correspondence Federico Thaler
DSU, Ca’ Foscari University of Venice
Venice, Italy
federico.thaler@gmail.com
Functional metal objects (e.g., nails, hinges and other
junction elements) are often found during excavation
campaigns and they can provide fundamental infor-
mation on the purpose of the item to which they are
eventually associated, or on local building technologies.
Unfortunately, such objects too often do not qualify to
be investigated, as their poor state of conservation does
not allow a proper recognition and identification of their
morphology and shape. In addition, since in most cases
they do not have high artistic value and provide only
indirect information, they often are not considered a
priority in terms of time and funds allocation.
The aim of this research was to identify a sustainable and
low-cost approach relying on the use of graphic software
and radiographies to obtain the maximum information
from a set of iron nails coming from the renowned Roman
sites of Aquileia and Torcello (Italy).
The iron finds were poorly preserved and the original
material was highly compromised due to the aggres-
sive contribution of the lagoonal environment. For the
purpose of this study, no cleaning treatment was carried
POSTER SESSION – ANALYTICAL INVESTIGATION
Andrea Cipolato Elisabetta Zendri
DSU, Ca’ Foscari University of Venice DAIS, Ca’ Foscari University of Venice
Venice, Italy Venice, Italy
andrea.cipolato@gmail.com elizen@unive.it
out on the objects. Photogrammetry was applied as a
preliminary tool for inspecting and documenting the state
of conservation and the shape of the objects through the
3D model. Together with the 3D reconstruction, x-ray
radiography was performed and used to create a digital
rendering of the original surface of the objects, which was
crucial for identifying their function and for quantifying
the loss of material due to degradation.
Iron nails from Torcello, dating from different periods
(7th–12th century AD), were selected to find possible
relations between the age of the objects and the metal
loss. Moreover, an attempt to highlight possible correla-
tions between the loss percentages due to corrosion and
the characteristics of the burial soil was performed by
comparing the two sites. For a more comprehensive study,
the composition of the iron nails was also measured by
x-ray fluorescence.
This cost-effective approach can be particularly useful
when a large number of functional objects is found, as it
allows for the building of a statistically significant model
and the selection of samples for further investigation.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – ANALYTICAL INVESTIGATION  443

Microstructural Examination of Kaushambi Copper-Alloy

Coins

Preeti Verma* Devashish Rajpoot Ghanshyam Lal B.V. Kharbade

National Research Laboratory for
Conservation of Cultural Property
Aliganj, Lucknow, India
preeti.rs.met12@itbhu.ac.in
*Author for correspondence
Department of Mechanical
Engineering
Indian Institute of Technology
Bombay
Mumbai, India
devashish1994@gmail.com
National Research Laboratory for National Research Laboratory for
Conservation of Cultural Property Conservation of Cultural Property
Aliganj, Lucknow, India Aliganj, Lucknow, India
ghanshyamlallko@gmail.com directornrlclucknow@gmail.com

Corrosion is the main form of degradation observed on
ancient metal objects. It is, therefore, very important to
understand the behavior of corrosion and to propose a
plausible method for preserving ancient objects. This
investigation consisted in a chemical composition anal-
ysis, microstructural examination, and corrosion product
study of coins excavated at Kaushambi, Uttar Pradesh,
India, dating from 200 BC to AD 100. Kaushambi was an
important and prosperous town in India during the time
of Buddha and retained its importance until at least the
6th century AD, as it was visited by the Chinese pilgrims
Faxian and Xuanzang.
Four representative coins were selected for the present
study. Macroscopic examination indicated that some
inscriptions were present on the obverse side of the
coins, while the reverse side was flat. Critical exami-
nation revealed that these inscriptions were related to
the Lord Shiva, from which it can be inferred that the
rulers of Kaushambi worshipped this deity. Elemental
analysis was carried out using portable micro-x-ray
fluorescence on both sides of the coins and confirmed
the presence of copper, tin, and iron. The percentage
of iron was found to decrease from the surface to the
center, with similar results for all the coins. This may
indicate that an iron die was used in their manufacture.
Different phases, such as Cu4O3, Cu10Sn3, and CuS,
were observed on the coins and the presence of CuS
during analysis by x-ray diffraction indicates that the
copper used to make them may have been extracted
from sulfide ores. A dendritic and columnar grain
structure observed on all the coins confirmed that
they were manufactured by casting. The coins do not
appear to have been subjected to any heat treatment.
One important observation is the occurrence of slip
lines and the intersection of slip lines on two of the
coins. This may be due to cold deformation. There
was a slight variation in hardness and this may be due
to a slight variation in the microstructure and chemical
composition. The corrosion products present on the
coins were analyzed by scanning electron microscopy
and energy-dispersive spectroscopy and a layer of
oxides was found to exist.
444  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Examining Metal Corrosion Using Electrochemistry
Hazel M.A. Hunter* David Thickett
English Heritage English Heritage
London, UK London, UK
hazel.hunter@english-heritage.org.uk david.thickett@english-heritage.org.uk
*Author for correspondence
Within conservation science, accurate atmospheric
pollution monitoring benefits both the care and the
management of collections. However, it can be expen-
sive and instrument-intensive to deliver. As a result,
low-cost, inventive yet appropriate analysis solutions
are sought as these have the potential to give mean-
ingful results in real-world situations. Various reactivity
monitoring approaches have been considered, with
metal coupons being a relatively inexpensive way to
assess actual environments for display and storage of
heritage metals.
Given that a metal object responds to its local envi-
ronment, an understanding of the surface corrosion
products can indicate whether preventative conservation
intervention is necessary in order to minimise or prevent
lasting damage from occurring to our metallic heritage
objects. In particular, analysis and regular monitoring
of the metal artefact offers an opportunity to proactively
manage historic collections.
Linear sweep voltammetry (LSV) as an electrochemical
method is well developed to assess the level of corrosion
to silver and copper coupons. The collections at English
Heritage, however, can contain other metals for which an
POSTER SESSION – ANALYTICAL INVESTIGATION
Jens Najorka Chris G. Jones
Natural History Museum Natural History Museum
London, UK London, UK
j.najorka@nhm.ac.uk c.jones@nhm.ac.uk
equivalent inexpensive, efficient and effective assessment
technique would be valuable. This work aims to extend
the LSV assessment method to real-world artefacts,
particularly those containing lead and iron, where further
understanding for lead and development of a method
for iron is sought.
Initial experiments have focussed upon identifying the
LSV response to standard compounds known to exist
on corroded metal artefact surfaces. These control
experiments ensure an accurate interpretation of the
signal response for each metal compound. In turn, these
standard responses may be used to interpret the results
from sacrificial metal coupons (metal effect sensors) that
have been exposed to the same local climatic conditions
as those experienced by the historic artefacts within the
collection. Various electrolyte concentrations were also
examined, as the solution strength is known to influence
the signal response position and thus can be used to iden-
tify the presence of individual corrosion products. X-ray
diffraction and Fourier transform infrared spectroscopy
were used as supporting materials characterisation tech-
niques in this study.
This poster presents our recent findings.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
A Project with Potential: Evaluating a New Monitoring
System for Artifacts Undergoing Electrochemical
Treatment
William Hoffman* Molly K. McGath
The Mariners’ Museum and Park The Mariners’ Museum and Park
Newport News VA, USA Newport News VA, USA
whoffman@marinersmuseum.org mmcgath@marinersmuseum.org
*Author for correspondence
Conservation personnel at The Mariners’ Museum
and Park in Newport News, Virginia, have undertaken
the treatment of 210 tons of excavated materials from
the wreck site of the American Civil War ironclad
USS Monitor over the last 20 years. Many of the recov-
ered objects are under electrolytic reduction treatment
to facilitate the breakdown of corrosion products and
the extraction of chlorides that are a result of marine
burial. Reference electrodes are employed to identify
object potentials and monitor the electrochemical
process. Early in the project, a centralized data acqui-
sition system was installed which could record multiple
months of information to determine efficiency of the
treatment over time by capturing the potential and pH
data. However, one of the major drawbacks of the system
POSTER SESSION – ANALYTICAL INVESTIGATION  445
was that, once set in record mode, it was not possible
to see what was happening to the objects in real time.
Therefore, in 2015, conservators began researching a
more effective recording system that maintained an
active display and could collect information not only on
potential but also on pH and temperature. The results
of their research identified the Multitrend GR graphic
recording system by Honeywell as meeting the required
criteria. A grant from the Institute of Museum and
Library Services provided funds to acquire the system
and install it in the spring of 2017. This system has
proved valuable in providing real-time readouts as well
as records that enable the conservation staff to review
and interpret how the chemistry of their treatment
systems has evolved over time.
446  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Diagnostics and Monitoring of Historical Vehicle Engines
by Acoustic Emission Testing
Laura Brambilla* Alejandro Roda-Buch
Haute Ecole Arc Ecole Polytechnique Fédérale de
Conservation-restauration Lausanne
Neuchâtel, Switzerland Lausanne, Switzerland
laura.brambilla@he-arc.ch alejandro.rodabuch@epfl.ch
*Author for correspondence Emilie Cornet
Haute Ecole Arc
Conservation-restauration
Neuchâtel, Switzerland
emilie.cornet@he-arc.ch
Among technical and industrial heritage objects with
moving mechanisms, vehicles are particularly complex
because of the number of parts and sub-mechanisms
involved. The functionality of these objects, i.e., the
possibility of making them work, is an integral part of
the objects themselves. Without it, they lose part of their
cultural value. However, the reactivation of these mecha-
nisms is often a challenge for conservators, in particular
if they have ceased to function for a long period. In fact,
the presence of corrosion products, deposits, oxidation,
particles, and the scaling of oils or lubricants can prevent
mechanisms from working properly or even lead to their
breakdown during reactivation.
Vehicles are conserved, exhibited, and sometimes used
in different kinds of museums and collections. These
include dedicated museums (transport museums,
car museums, and railway museums), car company
museums (each major car marque has its own), science
and technology museums, and private collections.
Given the aforementioned difficulties with reacti-
vating engines, there are usually only two choices as
far as exhibition and maintenance of these vehicles is
POSTER SESSION – ANALYTICAL INVESTIGATION
Guillaume Rapp Brice Chalançon
Haute Ecole Arc Association de Gestion du Musée
Conservation-restauration National de l’Automobile
Neuchâtel, Switzerland Mulhouse, France
guillaume.rapp@he-arc.ch brice.chalancon@museedelauto.org
concerned: (a) The static option: the vehicle is exhib-
ited and its engine is never started, or sometimes the
engine is even removed from the vehicle for practical
and safety reasons. (b) The dynamic option: the vehicle
is used, or at least started periodically. In the case of
dynamic maintenance, the originality of the engine and
other parts of the automobile is not always respected, as
many parts of the engine are usually replaced in favor
of ensuring the functioning of the vehicle instead. This
is common, in particular, among collectors, who use
the vehicles for historical races, meetings, and rallies.
In both cases, some of the value and history of the
object is lost.
In order to be able to reactivate the mechanisms without
damage and preserve the cultural value and original mate-
rials of these objects, conservators need diagnostic tools
to detect malfunctions at the very early stages. In this
respect, the use of non-invasive techniques to evaluate
the condition of a mechanism can help to avoid damage
during reactivation. The ACUME_HV project aims to
develop acoustic emission methods as non-invasive
techniques for historical vehicle diagnostics.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – ANALYTICAL INVESTIGATION  447

Use of the MiCorr Application for the Local Diagnosis of

Metal Artefacts: The Conservator’s Point of View

Mathea Hovind* Christian Degrigny

Department of archaeology,
conservation and history
University of Oslo (IAKH-UiO)
Oslo, Norway
mathea91@gmail.com
*Author for correspondence
Haute Ecole Arc
Conservation-restauration (HE-Arc CR)
University of Applied Sciences and Arts
Western Switzerland (HES-SO)
Neuchâtel, Switzerland
christian.degrigny@he-arc.ch

The conservation of archaeological and historical metal
artefacts requires a thorough diagnosis of their condi-
tion based on an understanding of the composition of
the remaining metal components and their associated
corrosion products.
MiCorr is an online decision support system (DSS) for
the local diagnosis of heritage metals developed within
Haute Ecole Arc (HE-Arc), Neuchâtel, Switzerland, as a
partnership between its conservation-restauration (CR)
and management departments. It is based on the compar-
ison of digitally constructed stratigraphies of observed
corrosion structures with similar existing stratigraphies
from a database.
The use of MiCorr (MiCorr.org) still remains limited
within the conservation community. A project aimed at
assessing its user-friendliness and efficiency was carried
out within the framework of the author’s internship at
HE-Arc CR for her master’s degree in object conservation
from the University of Oslo, Norway.
In order to test the application, a series of modern
20th-century metal artefacts were investigated. The
following procedure was adopted: visual macroscopic
observation and documentation using Bertholon’s
method, construction of preliminary stratigraphies
using the MiCorr application and comparison with the
stratigraphies on the MiCorr database. As the results were
rather poor and the artefacts were considered as testing
materials, they were sampled, examined in cross section
and analysed chemically. As a result, the matching with
the MiCorr database produced better results.
After discovering MiCorr, it would seem that this intui-
tive self-learning tool can aid conservators in their work
to understand the corrosion structures of an object.
Like any other tool, it requires the user to invest some
time to learn how to use it effectively. Furthermore,
the success of the tool depends on the enhancement
of the database with models and forms of corrosion by
active contributors.
448  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – PREVENTIVE CONSERVATION

Classification of Indoor Air Quality with Respect to Lead

Corrosion

Milan Kouril* Kristyna Charlotte Strachotova Zdenek Bartak Katerina Kreislova

University of Chemistry University of Chemistry and Technology, Prague SVUOM, Ltd. SVUOM, Ltd.
and Technology, Prague Prague, Czech Republic Prague, Czech Republic Prague, Czech Republic
Prague, Czech Republic strachotova.tyna@gmail.com bartak@svuom.cz kreislova@svuom.cz
kourilm@vscht.cz
Sarka Msallamova
*Author for correspondence University of Chemistry and Technology, Prague
Prague, Czech Republic
msallams@vscht.cz

The ISO 11844-1 standard evaluates the aggressiveness
of indoor atmospheres according to the corrosion of four
reference metals: steel, zinc, copper and silver. Corrosion
of these metals does not sufficiently reflect the typical
pollutants prevalent in the air in archives, libraries and
museums, in particular volatile organic acids. Lead
is sensitive to the vapours of organic acids in indoor
environments and may be used as an indicator of this
pollutant in objects in collections. It should inevitably
be included between the reference metals in order to
evaluate the aggressiveness of indoor atmospheres. This
study provides further research into defining corrosivity
categories for lead and evaluates techniques for meas-
uring the lead corrosion rate. The results of corrosion
rate measurements on standard metals and lead in several
archives and depositories are presented.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
The Patina on Amilcar de Castro’s Weathering
Steel Sculpture: The Influence of the Museological
Environment
Tatiana Russo dos Reis* Teodora Camargo Carneiro
Pinacoteca do Estado de São Paulo Pinacoteca do Estado de São Paulo
São Paulo SP, Brazil São Paulo SP, Brazil
trusso@pinacoteca.org.br tcarneiro@pinacoteca.org.br
*Author for correspondence Márcia de Almeida Rizzutto
Instituto de Física da Universidade de
São Paulo
São Paulo SP, Brazil
rizzutto@if.usp.br
Over the years, artists have distanced themselves from
traditional techniques by incorporating non-traditional
artistic media such as industrial materials in their work.
This, in turn, inspired a number of research studies in
the field of conservation due to the interest generated
by either conceptual issues underlying their work or
the wealth of possible practical investigations into the
behavior and properties of its constitutive materials.
Among these materials are steels that are highly resistant
to corrosion and which have been used in artistic produc-
tion since the 1960s.
These steels have excellent resistance to corrosion due to
the addition of small quantities of copper, phosphorus
or other elements which reduce their degradation rate
when exposed to air. What differentiates this type of steels
from common steels is the development of an adherent
protective layer on the material’s surface, forming a
patina that hinders the action of corrosive agents present
in the atmosphere. The formation of this patina, often
considered visually attractive thanks to its “rusted”
appearance, depends on the chemical composition of
the steel, environmental factors and aspects related to
the geometry of the piece.
When in contact with high levels of chlorine, such as in
maritime environments, akaganeite, a non-protective
layer of iron oxyhydroxide – unlike alpha-goethite and
lepidocrocite, which are oxyhydroxides that possess
POSTER SESSION – PREVENTIVE CONSERVATION  449
Antonio Augusto Couto Pedro H.O.V. Campos
Universidade Presbiteriana Instituto de Física da Universidade de
Mackenzie São Paulo
São Paulo SP, Brazil São Paulo SP, Brazil
antonioaugusto.couto@mackenzie.br pcampos@usp.br
good adherence and form corrosion-resistant protective
layers – can appear on the surface of the weathering steel.
During an investigation into the conservation of steel
sculptures displayed at the Pinacoteca do Estado de São
Paulo, a number of analyses were performed using x-ray
diffraction equipment and x-ray fluorescence. In the
work Cavalo [Horse], by artist Amilcar de Castro, the
elements that compose the oxide layer on the surface of
the sculpture were identified, along with the elements of
the steel alloy and the anomalous presence of akaganeite.
This study was aimed at analyzing the factors that
contributed to the formation of akaganeite on the surface
layer of the artwork. In particular, the use of sodium
hypochlorite to clean the museum’s floor resulted in the
accumulation of chlorine on the work’s surface, which in
turn enabled the formation of the oxide. In view of this
fact, a number of conservation and restoration measures
were discussed that included the possibility of using a
coating method such as wax or varnish in an attempt to
protect the work. However, it was concluded that any
alternative employed could lead to changes in color or
gloss, and, in addition, the current coating itself could
contribute to initiate a corrosion process. Thus, in this
case, the best method of conservation is to clean the
artwork with water frequently to prevent the accumula-
tion of salts like chlorine and sulfur, which may lead to
a corrosion process.
450  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Display Cases for Metal and Composite Objects at the
National Museum in Warsaw, Poland
Anna Mistewicz
National Museum in Warsaw
Warsaw, Poland
amistewicz@mnw.art.pl
This study presents our experiences with new display
cases designed to exhibit decorative art objects in the
Gallery of Old Masters at the National Museum in
Warsaw, Poland, opened in December 2016.
The presence of different types of materials (paper, wood,
stone, plaster, metal, ceramics, textiles, ivory, amber,
glass and also enamel) in one exhibition area required
the introduction of monitored, airtight cases equipped
with silica gel and, in some cases, anti-corrosive materials.
Special cabinets were designed for glass objects and deli-
cate metal and composite artifacts with ivory and enamel.
The display cases were made of inert materials (metal,
glass and Forex) that had positive Oddy test results.
However, two months after the opening of the Gallery,
it became clear that conditions inside the display cases
did not significantly differ from the conditions outside.
The contractor was asked to conduct tests, but over time
it became apparent that the problem was complex, as it
was connected not only to the amount of silica gel but
also to the design of the cases and their airtightness, and
the way the lightening system was installed.
The airtightness of the cabinets was first checked
by two different methods – with an ultrasonic leak
detector and an N2O tracer gas concentration decay
test. The air change per day (ACD) rate was measured
POSTER SESSION – PREVENTIVE CONSERVATION
at around 1, so it was decided to seal the cabinets to
increase airtightness performance and achieve an ACD
level of below 0.4. Another test proved that specially
designed areas for silica gel and anti-corrosion materials
did not create adequate airflow, so the display cases
had to be partially redesigned. Although all heating
elements were located outside the cases, the interior
LED spotlights were causing a rise in the temperature
level whenever the lights were switched on and a drop
in the temperature whenever they were switched off.
This resulted in daily humidity fluctuations of up to
4 percent. In the end, a new method of calculating the
amount of gel was introduced based on airtightness
measurements and volume capacity. In this way, the
amount of gel needed was doubled or tripled in compar-
ison with the distributor’s recommendations.
Luckily, all of the solutions applied until now have
improved the performance of the display cases. The
environment inside the cases is being monitored with
Rotronic LOG-HC2-RC and LOG-HC2-P1-SET remote
wireless dataloggers. The condition of the objects is
examined regularly, mainly visually, and there have been
no signs of deterioration. The improper design of the
display cases and their hasty assembly turned out to be
the main causes of the problem.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – PREVENTIVE CONSERVATION  451

Study of a Preventive Conservation Procedure for

Paintings on Iron Plates

Eva Menart*
National Museum of Slovenia
Ljubljana, Slovenia
eva.menart@nms.si
*Author for correspondence
Nataša Nemeček
National Museum of Slovenia
Ljubljana, Slovenia
natasa.nemecek@nms.si
Irena Kralj Cigić
University of Ljubljana
Faculty of Chemistry and Chemical
Technology
Ljubljana, Slovenia
irena.kralj-cigic@fkkt.uni-lj.si
Helena Pucelj Krajnc
University of Ljubljana
Academy of Fine Arts and Design
Ljubljana, Slovenia
helena6pucelj@gmail.com

The investigated object, a 19th-century painting enti-
tled The fire in Kranj (orig. Požar v Kranju) which – unlike
most paintings in the collection of the National Museum
of Slovenia – was painted on an iron plate, posed an
interesting problem that required the collaboration of
experts from different fields and institutions.
During the challenging conservation-restoration proce-
dure, a decision was made to include the painting in the
museum’s permanent exhibition. However, this involved
creating display conditions that were not straightforward.
The painting is composed of two types of materials
(paint and iron) which are commonly stored/displayed
at different environmental humidity conditions: the
standard recommendation for paintings is around 50%
RH, whereas iron is known to corrode in conditions
above 30% RH.
Consequently, it was decided to investigate appropriate
conditions using accelerated ageing experiments on
model samples. To ensure sample representativeness,
paint layers from the original painting were analysed
using gas chromatography coupled with mass spectrom-
etry. The results showed that the main paint binder, used
throughout the entire painting, is linseed oil. Mock-up
samples were therefore painted with linseed oil-based
paints on iron plates. To monitor potential change/
damage to the samples, ten spots on each sample were
selected and examined under a microscope. The samples
were aged under different environmental conditions that
represented suitable levels of humidity for the two mate-
rials comprising the original painting: 30% RH (suitable
for the iron plate) and 50% RH (suitable for the oil-based
paint), both at 60°C. In parallel, model samples on canvas
were also aged to determine the influence of the iron plate
support on the ageing of the painted layer and to study
the difference between the ageing of the paint layers on a
stiff support (prone to corrosion) and a flexible one (more
prone to embrittlement and degradation). After ageing,
selected spots on the studied samples were investigated
with a microscope (50×, 100× and 200× magnification).
It was then possible to evaluate the influence of different
humidity conditions on composite objects such as the
painting in question and therefore to establish suitable
conditions to display The fire in Kranj in the museum’s
permanent exhibition.
452  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – CORROSION STUDIES

Appearance and Corrosion of Weathering Steel at

an Urban-Coastal Site: Influence of Environmental
Conditions during the First Months of Exposure
Elena Bernardi* Lara Nobili
Dept. of Industrial Chemistry “Toso Dept. of Industrial Chemistry “Toso
Montanari” Montanari”
University of Bologna University of Bologna
Bologna, Italy Bologna, Italy
elena.bernardi@unibo.it lara.nobili3@unibo.it
*Author for correspondence Ivano Vassura
Dept. of Industrial Chemistry “Toso
Montanari”
University of Bologna
Bologna, Italy
ivano.vassura@unibo.it
Cristina Chiavari
Dept. of Cultural Heritage
University of Bologna
Ravenna, Italy
cristina.chiavari@unibo.it
Carla Martini
Dept. of Industrial Engineering
University of Bologna
Bologna, Italy
carla.martini@unibo.it
Daniel de la Fuente
National Centre for Metallurgical
Research (CENIM-CSIC)
Madrid, Spain
delafuente@cenim.csic.es
Manuel Morcillo
National Centre for Metallurgical
Research (CENIM-CSIC)
Madrid, Spain
morcillo@cenim.csic.es

Weathering steel, commonly known by the commercial
name “Cor-Ten,” is widely used in both historic and
contemporary contexts for sculptures, private and public
buildings, structures, and urban furniture. This alloy has
fascinated artists and architects worldwide for its corrosion
resistance, which is related to the formation of a passivating
patina that changes composition and appearance over
time depending on the environment. This work evalu-
ated whether and how the appearance and corrosion of
weathering steel exposed for one year at the same site was
influenced by different initial atmospheric conditions.
Specimens of weathering steel, whose composition
corresponded to that of Cor-Ten A, were exposed both
in sheltered and unsheltered conditions at the urban-
coastal site of Rimini, Italy. Three one-year campaigns
were conducted, starting progressively in different
seasons (summer, winter, and spring). During exposure,
the climatic parameters and main air pollutants were
monitored. Specimens were characterized by surface
and cross-section analysis (color measurements, x-ray
diffraction, and scanning electron microscopy/ener-
gy-dispersive x-ray spectroscopy) and by ion chromatog-
raphy. Patina protective ability, the environmental ions in
patinas, and corrosion rates were determined. Data were
also processed by principal component analysis (PCA).
After one year, differences were observed in patina evolu-
tion (composition, stratification, and color) and corrosion
rate (Vcorr) between samples exposed in different seasons
and in sheltered or unsheltered conditions. Differences
between sheltered and unsheltered samples dimin-
ished when the specific combination of environmental
parameters (rainfall, solar radiation, T, RH%, pollutant
concentrations) became similar. PCA revealed a specific
trend in color change associated with the percentage of
akageneite, lepidocrocite, and goethite, suggesting the
possibility of monitoring patina evolution by processing
spectrophotometric data.
In general, sheltered samples had a higher Vcorr than
unsheltered ones, except during summer exposure,
due to abundant rainfall and higher accumulation of
chloride on the latter. Vcorr values do not always exactly
correspond to the trend expected on the basis of surface
analysis. In fact, less protective external layers may form
above a more stable inner layer, due to initial favorable
conditions, leading to a lower Vcorr.
Results showed that corrosion of weathering steel was
strongly influenced by the initial environmental condi-
tions. Therefore, to reduce material decay, a preliminary
evaluation of the environmental parameters and their
trends at the exposure site is recommended.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Atmospheric Alteration of α/δ Bronze Bells:
Micro-Infiltrating Corrosion
Aline Petitmangin*
Laboratoire Inter-Universitaire des Systèmes Atmosphériques (LISA)
UMR 7583 CNRS/Université Paris-Est-Créteil/Université de Paris/Institut Pierre
Simon Laplace (IPSL)
Créteil, France
aline.petitmangin@lisa.u-pec.fr
Anne Chabas
Laboratoire Inter-Universitaire des Systèmes Atmosphériques (LISA)
UMR 7583 CNRS/ Université Paris-Est-Créteil/Université de Paris/Institut Pierre
Simon Laplace (IPSL)
Créteil, France
Anne.Chabas@lisa.u-pec.fr
Mandana Saheb
Laboratoire Inter-Universitaire des Systèmes Atmosphériques (LISA)
UMR 7583 CNRS/ Université Paris-Est-Créteil/Université de Paris/Institut Pierre
Simon Laplace (IPSL)
Créteil, France
Mandana.Saheb@lisa.u-pec.fr
Bronze bells are part of our intangible and material world
heritage. Their ornamentation makes them unique works
of art and their musicality provides rhythm to time and
civilizations. In France, there are 450,000 actively used
bronze bells. Their composition (22 wt% Sn) consists
of a Cu-Sn alloy with a two-phase α/δ structure. The
δ-phase improves the metal hardness which confers
high resonance for good sound properties. Bells are
subjected to atmospheric, anthropic, and biological corro-
sion representative of all environments. They undergo
complex alterations which are both physico-chemical
(environment-dependent) and mechanical (through the
effect of percussion).
For a better understanding of bell corrosion, samples
were collected from two French bells made in the 1930s
by gravity casting. The first, by the Cornille Havard
Company, was located in the steeple of Trélévern Church
and had been exposed to a marine environment. The
second, by the Blanchet foundry, belonged to Saint-Maur
Church and had been exposed to an urban environ-
ment. Both alloys have inclusions and porosity due to
micro-shrinkages and poor gas evacuation. In the marine
environment, the patina was formed by alternating layers
POSTER SESSION – CORROSION STUDIES  453
Ivan Guillot
Institut de Chimie et des Matériaux Paris-Est (ICMPE)
UMR 7182 CNRS/UPEC
Thiais, France
Ivan.Guillot@glvt-cnrs.fr
Barbara Laik
Institut de Chimie et des Matériaux Paris-Est (ICMPE)
UMR 7182 CNRS/UPEC
Thiais, France
Barbara.Laik@icmpe.cnrs.fr
*Author for correspondence
of atacamite (Cu2Cl(OH)3) and cuprite (Cu2O) on an
inner layer of tin oxide (SnO2). In the urban area, antlerite
(Cu3SO4(OH)4) and tenorite (CuO) were observed with
fly ash from unburnt residues. On the surface, there was
a transformed medium with quartz clays due to the way
these bells were manufactured. A preferential internal
galvanic corrosion of α, Sn poor, thus strongly subjected
to selective dissolution of Cu, was observed associated
with the formation of alternating oxide stripes that are
more or less pronounced according to the environment.
However, the alteration of δ, rich in Sn, dominated at the
outer edge of the alloy pores. Using an electrochemical
zero resistance ammeter (ZRA) method, the cyclical
oxidation of α/δ was evidenced. Isotopic tracing (18O) of
the alteration of old samples examined the micro-infil-
trating networks of environmental fluids and quantified
the reactive zones of the bells. All this work will then
allow an alteration mechanism to be proposed.
This study is part of the BellACorr project supported by
the French National Research Agency. Based on “respect
for original material” (Article 9, The Venice Charter of
1964), it is helping to improve restoration processes for
heritage bronzes.
454  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Whiskers and Strands: Unexpected Corrosion on Early
Medieval Metals on Display at the Museum of London
Helen Ganiaris* Rebecca Lang
Museum of London Museum of London
London, UK London, UK
hganiaris@museumoflondon.org.uk rlang@museumoflondon.org.uk
*Author for correspondence
In October 2017, a black corrosion was noticed on several
early medieval silver coins in the Museum of London’s
Medieval Gallery. Scanning electron microscopy/ener-
gy-dispersive x-ray analysis showed this to be silver and
sulfur in a 2:1 ratio, confirming the initial supposition
of a silver sulfide-related corrosion product. On closer
inspection, it was found that some of the copper-alloy
objects also had fresh black corrosion. Two types of
sulfide corrosion were identified on both the silver
and copper alloys, matching past descriptions of types
known as ‘whisker’ and ‘filiform’. Other similar objects
displayed in an identical adjacent case in the same gallery
were unaffected. All display materials in the showcases
had been tested for emission of corrosive gases by Oddy
testing when the gallery was installed in 2005. A similar
outbreak of copper sulfide corrosion at the British
Museum in the 1990s had been shown to be caused by
the emission of hydrogen sulfide from low-fired ceramic
POSTER SESSION – CORROSION STUDIES
Adriana Paez Cure
Miguel Urrutia Art Museum (Central
Bank of Colombia)
Bogota, Colombia
apaezcur@banrep.gov.co
vessels excavated from waterlogged archaeological sites.
This poster describes the investigations that have been
carried out to identify the source of the sulfides causing
this outbreak of corrosion.
The case in question contains a mixture of organic and
inorganic archaeological objects, including one large
heavily corroded iron cauldron from a pit with organic
residue, excavated in the 1990s. Polished silver coupons
placed in the affected case became tarnished within a
few weeks whereas coupons in the adjacent case were
unaffected. Ongoing investigation includes comparisons
with polished coupons enclosed in sealed containers
with excavated archaeological objects from the same
and similar sites. Diffusion monitoring tubes to detect
hydrogen sulfide were also placed in the affected case
and other gallery locations for comparison.
Conclusions are drawn about these potential sources of
sulfides when planning mixed archaeological displays.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Corroding Lead Cames on Stained Glass Windows: A Case
Study of Two 16th-Century Panels in Their Display Case
Alice Gerber
Haute Ecole Arc Conservation-restauration
Neuchâtel, Switzerland
alice.gerber@he-arc.ch
This study was carried out as part of a thesis for a
bachelor’s degree at the Haute Ecole Arc in Neuchâtel,
Switzerland.
Over the last few years, the lead cames on two 16th-cen-
tury stained glass windows, displayed at the Museum
zu Allerheiligen in Schaffhausen since 2009, have been
subject to corrosion. The corrosion products consist
of various forms of white and gray efflorescence. The
purpose of this study was to gain an insight into the dete-
rioration mechanisms of these visually different corrosion
forms in order to find out whether the source of their
emergence lies in the composition of the stained glass
windows themselves or in their display environment.
The objects and their degradation, namely the corro-
sion products, and their expansion and structure, were
characterized first by documenting and evaluating their
deterioration state and then by analyses with x-ray
fluorescence spectroscopy, Fourier transform infrared
spectroscopy and scanning electron microscopy coupled
with energy-dispersive x-ray spectroscopy. The results
showed that the lead cames had been attacked by acetic
acid, generating lead acetates and carbonates. The loca-
POSTER SESSION – CORROSION STUDIES  455
tion of the corrosion products depended on the elemen-
tary composition of the metal. The percentage of tin in
the alloy in each came was a key reason for the uneven
spread of the corrosion, but it also depended on the
presence of previous restoration interventions or exog-
enous deposits. The way the corrosion had attacked the
metal was potentially influenced by the type of alloy and
the metallographic structure of each of the lead cames.
This study revealed that a Pb-Sb alloy was attacked by a
more aggressive form of corrosion than other lead alloys,
indicating a possible form of galvanic corrosion.
Alongside this, the exhibition environment, i.e., the
display case, was assessed. The presence of airborne
pollutants was determined with the help of a VOC
detector and an Oddy test. This brought to light that the
air in the display case was contaminated by a very high
level of acetic acid. This rate of acetic acid concentra-
tion can probably be explained by the presence of MDF,
the main material used in the case’s construction. The
presence of wooden and paper objects inside the same
display case might also have increased the concentration
of this specific pollutant.
456  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – CORROSION STUDIES

Study of Dark-Red Staining on Lead Rooftops and

Sculptures

Damien Lopez Aurelia Azema* Suzanne Joiret Annick Texier

Laboratoire de Recherche des
Monuments Historiques (LRMH-CRC)
CNRS USR 3224
Paris, France
dlopez3@hotmail.fr
*Author for correspondence
Laboratoire de Recherche des
Monuments Historiques (LRMH-CRC)
CNRS USR 3224
Paris, France
aurelia.azema@culture.gouv.fr
Laboratoire Interfaces et Systèmes Laboratoire de Recherche des
Electrochimiques (LISE) Monuments Historiques (LRMH-CRC)
CNRS UMR 8235 CNRS USR 3224
Paris, France Paris, France
suzanne.joiret@upmc.fr annick.texier@culture.gouv.fr
Solenn Reguer
Synchrotron SOLEIL
Gif-sur-Yvette, France
solenn.reguer@synchrotron-soleil.fr

Due to its interesting physico-chemical properties,
lead has been extensively used in various domains, in
particular as a roof-covering material or in the manu-
facture of outdoor sculptures. Exposure to weather
conditions induces the formation of protective insoluble
salts, such as hydrocerussite (2PbCO3.Pb(OH)2), on its
surface, giving it its usual light grey appearance. Never-
theless, many occurrences of dark-red staining of lead
employed in historical monuments have been reported.
Though not extensively studied, it has been proved that
the compound responsible for this is not organic but a
product of corrosion. A previous study conducted on
lead statues identified the compound as lead(IV) oxide
(plattnerite, ßPbO2).
Our work was aimed firstly at verifying whether the
same product is also present on the stained rooftops of
four monuments – Beauvais Cathedral, Versailles’ Royal
Chapel, Azay-le-Rideau Castle and the Great Stables at
the Chantilly Castle. The second goal was to determine
the factors influencing the formation of the red corrosion.
Each rooftop and/or roof decoration was sampled for
analysis and cross sections prepared. By using optical
microscopy, scanning electron microscopy with ener-
gy-dispersive x-ray spectroscopy, x-ray diffraction,
micro-Raman spectroscopy, and x-ray absorption near-
edge structure spectroscopy, different observations were
made that allowed for the stratigraphic characterization
of the samples.
The red layer was thoroughly analyzed and found to be
essentially composed of ßPbO2. Study of the organization
of the corrosion layers showed that ßPbO2 was almost
always in contact with lead carbonate and sulfate. It
was hypothesized that these lead(II) compounds are
precursors of lead(IV) ßPbO2. From metallic lead and
in classical conditions, this compound is formed under
the action of a strong oxidizing agent which is difficult
to identify in atmospheric conditions. Nevertheless,
the formation of the red lead seems to be influenced by
several factors such as geographic orientation and the
composition of the lead. Indeed, not only is the phenom-
enon more pronounced in south-facing lead roofs, but
analysis of the lead alloy used for brazing (with a high
concentration of tin) showed that the red layer coincides
with the tin-rich zone.
This study provides a first insight into the nature of the
dark-red staining that appears on lead rooftops and
possible approaches to the issue when conservation of
historical monuments is required.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
The Use of Industrial Vapor-Phase Corrosion Inhibitors on
Large and Composite Objects
Jerrad Alexander
Alexander Conservation Services, LLC
Omaha NE, USA
jalexanderconsult@gmail.com
Traditional methods of metals conservation often involve
the use of chemicals and hazardous materials (benzotri-
azole, xylene), coating systems that are ultimately tempo-
rary and can be difficult to apply and remove (waxes,
lacquers), and corrosion or coating removal methods that
require specialized equipment (dry ice, media blasting).
The development of aqueous, vapor-phase corrosion
inhibitors (VpCIs) such as those marketed by Cortec
Corporation represent a possible new means for treating
and preserving metal artifacts. These environmentally
friendly vapor-phase (or volatile) corrosion inhibitors use
water as a carrier solvent and offer a means of reducing
or eliminating many issues often found when treating
large, industrial, mechanical, or technological objects.
These products are effective at low concentrations, offer
protection to multiple materials, and are applicable by
various methods. With a customizable range of film-
forming and UV-resistant additives, conservators are
better able to tailor their treatments to the specific
object, material, alloy, and future environmental needs
of the objects under their care. The low vapor pressure
POSTER SESSION – PROTECTION  457
of the inhibitors offers protection in spaces that may be
inaccessible to conservators. This can equate to consider-
able savings in treatment times, health and safety issues,
and costs for large technology and industrial heritage
objects, particularly those made with modern alloys
and composites, while maintaining a comparable level
of protection with that of traditional treatments. There
is already evidence of the use of VpCIs in museums and
conservation work, but this remains a developing area
of research as the products were originally designed for
manufacturing and industrial applications. Due to their
proprietary nature, their exact formulation is secret; their
aging properties are yet to be determined. However, given
what is known about their functionality, the benefits they
offer, and the anecdotal evidence of their effectiveness,
the continued study and use of these products seems
warranted. This poster recounts the uses of some of
these Cortec VpCI products across a five-year span in
three separate projects, including a mid-19th-century
marine artifact, World War II-era military vehicles, and
Saturn V rocket engines.
458  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Application of Self-Assembled Monolayers of Long-Chain
Organic Acids in the Corrosion Protection of Bronze
Dajana Mikic* Helena Otmacic Curkovic
University of Zagreb University of Zagreb
Faculty of Chemical Engineering Faculty of Chemical Engineering
and Technology and Technology
Zagreb, Croatia Zagreb, Croatia
dmikic@fkit.hr helena.otmacic@fkit.hr
*Author for correspondence
One of the challenges today’s chemical engineers face
is to find environmentally and economically acceptable
solutions to protect against metal corrosion. With regard
to bronze sculptures, this challenge is even greater due to
the special requirements of the conservation and restora-
tion profession and the necessity to preserve a sculpture’s
original appearance. Taking all of these requirements into
account, the great potential for application in this field are
self-assembled monolayers of long-chain organic acids
that are formed by adsorption of acid on the surface of
oxide-covered metal, resulting in an acid-base reaction
between metal oxide and acid. This reaction is followed
by the self-organization of alkyl chains that creates a very
thin and compact protective film on the metal surface,
thus preventing the penetration of water molecules and
aggressive ions.
The aim of this work was to examine the efficiency of
POSTER SESSION – PROTECTION
self-assembled monolayers of 16-phosphonohexadeca-
noic acid (COOH-PA), obtained using various prepara-
tion methods, in protecting bronze surfaces. In addition,
the ability of self-assembled monolayers to improve adhe-
sion between transparent organic coatings (Paraloid B-72)
and bronze surfaces was also investigated. Electrochem-
ical measurements were conducted to test the protective
properties of COOH-PA monolayers on bronze surfaces
using linear polarization and electrochemical impedance
spectroscopy. These measurements were carried out in
simulated acid rain (0.2 g/L NaNO3 + 0.2 g/L Na2SO4 +
0.2 g/L NaHCO3 at pH 5) to replicate the kind of rain
common in urban atmospheres. In addition to electro-
chemical testing, the properties of the protective films
obtained were examined by infrared spectroscopy anal-
ysis with Fourier signal transformation, goniometry, and
scanning electron microscopy.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Assessing the Performance of Protective Treatments on
Outdoor Sculptures: To Be or Not To Be a Representative
Model Sample?
Edith Joseph* Paola Letardi
University of Applied Sciences National Research Council – Institute
Western Switzerland of Marine Sciences
Haute Ecole Arc Genoa, Italy
Conservation-restauration paola.letardi@ias.cnr.it
University of Neuchâtel, Laboratory of
Technologies for Heritage Materials
Neuchâtel, Switzerland
edith.joseph@he-arc.ch
edith.joseph@unine.ch
*Author for correspondence
Protective treatments applied in the field of cultural
heritage need to take into account several constraints that
have gradually been better outlined over the years. Along
with their impact on aesthetic appearance and reversibility,
it is nowadays well recognised that patinas and/or original
historical coatings and finishings can play a relevant role
in the behaviour of protective treatments. Nonetheless,
no standard characterisation method has been drafted to
test new coatings and their application on real artefacts.
In this study, an analytical protocol to assess the perfor-
mance of protective treatments on outdoor sculptures
is discussed, in particular the use of electrochemical
impedance spectroscopy combined with colorimetric
measurements.
To this purpose, coupons as similar as possible to the
metal artefacts for which the treatment is being tested
were selected to evaluate the performance of different
treatments. Naturally and artificially aged coupons
presenting copper sulfate or copper chloride patinas
POSTER SESSION – PROTECTION  459
Monica Albini Mathilde Monachon
National Research Council – Institute University of Neuchâtel,
for the Study of Nanostructured Laboratory of Technologies for
Materials Heritage Materials
Rome, Italy Neuchâtel, Switzerland
monica.albini@cnr.it mathilde.monachon@unine.ch
were prepared or produced. Naturally corroded ternary
(Cu80/Sn8/Pb2) and quaternary (Cu85/Sn5/Zn5/Pb5)
bronze coupons were obtained after an 18-month expo-
sure in an urban-marine environment (Genoa, Italy). In
parallel, sets of 12 coupons (60 × 60 × 4–5 mm) were cut
from cast ingots of ternary and quaternary bronze alloys
and coated with patinas traditionally used in foundry by
the Kunstgiesserei St. Gallen AG. The behaviour of the
different alloys/patinas was characterised before and after
ageing, both alone and treated with a protective layer of
Cosmoloid 80 wax, a largely adopted treatment, or with
biopatina, a green corrosion inhibitor.
The representativeness of the model samples prepared was
validated through measurement campaigns performed
on outdoor bronze sculptures ranging from 19th-century
works to contemporary art. The results obtained offer
valuable information to assist in the choice of model
samples during the decision-making process of artwork
preservation.
460  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Protection and Meaning: A Reflection on the Role of
Painted Surfaces on Outdoor Metal Sculptures and Their
Conservation
Alice Camoli* Alessandra Carrieri
Laboratorio Restauro Metalli e Centro di Restauro Lì Punti-
Ceramiche of Vatican Museums Soprintendenza di Sassari e Nuoro
Fucina Ervas di A. Ervas Accademia di Belle Arti di Verona
Vatican City State, Italy Lì Punti (Sassari), Italy
camolialice@gmail.com ale_carrieri@hotmail.com
*Author for correspondence
Observing the painted surface of an outdoor metal sculp-
ture implies understanding its dual and complementary
function of protection and artistic meaning.
Reproducing a painted surface is not a foregone conclu-
sion. Beyond the aesthetic requirements (such as the
degree of tonality and gloss) and authorization (i.e.,
the permission of the artist or the related estate and
foundations), there are several technical factors that
cannot be avoided, such as the execution of the methods
used to prepare the metal supports and the quality of
the painting products and their application. A prac-
tical experiment was therefore carried out to produce
different specimens in an effort to outline the following:
The technical functionality of the preparation methods
(from manual cleaning to the degree of sandblasting
and the passivation of surfaces), the technological
characteristics of coating products (notably red lead
in linseed oil primer, micaceous iron primer, zinc and
zinc-phosphate base, alkyd-urethane, and two-compo-
nent epoxy primers), and the application methods (by
brush, roller, and spray).
The experiment made it possible to analyze, by means of
POSTER SESSION – PROTECTION
Alessandro Ervas
Fucina Ervas di A. Ervas
Preganziol (Treviso), Italy
fucinaervas@fucinaervas.it
visual and operative study, not only how the execution of
the process and the technological properties of a coating
can guarantee the protection of the metal support and
the aesthetic finish, but also how the modification of a
material has influenced the preparation methods over
time and vice versa. Furthermore, the elimination of some
toxic products from the market (such as lead oxide from
minium or top coats based on chromium or cadmium)
has allowed the development of more environmentally
friendly and safety-conscious materials, methods of
intervention and restoration operations, the patenting
of new water-based coatings, the development of site
logistics and the disposal of sandblasting waste (also using
chemical or solvent-based methods or new methods of
removal such as dry-ice blasting). As a result, to improve
the methodological and technical approach to the conser-
vation of artistic sculptures, it is essential to know and
study the materials used both old and new, comparing
and updating depending on different circumstances and
professionals. It is only in this way that the transmission
of the meanings characteristic of every work of art can
be guaranteed.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
The Mystery of Ancient Metal Amulets
Helena Kupershmidt
Israel Antiquities Authority
Metals Conservation Laboratory
Artifacts Treatment, Conservation and Laboratories Department
Jerusalem, Israel
lenaku@israntique.org.il
Israel is one of those countries where a vast quantity of
masterpieces from ancient civilizations are concentrated.
Every year about 300 archaeological excavations are
carried out all over Israel. Thousands of archaeological
objects representing the rich and varied cultural heritage
of the area are discovered and researched. Inscribed metal
artifacts have a special place among them.
In antiquity, various types of metal sheets (lead, copper,
copper alloy, silver and gold) were used as materials on
which to inscribe texts in Greek, Aramaic, Hebrew and
Arabic. Usually these inscribed objects were magical
amulets containing curses and blessings. These magical
texts and signs are significant documents of the cultural
environment of the time, recreating and reflecting the
intellect, mentality and daily life of real people in the
ancient world.
Metal amulets were usually tightly rolled-up scrolls made
of thin metal. They were not meant to be opened and
read after being written and closed, as it was forbidden
according to beliefs. As the content of the texts is of vast
importance, these scrolls have to be opened and unrolled
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  461
to reveal their secrets and provide the epigrapher with
the possibility to decipher them.
This study examines a variety of amulets from different
periods made from different metals, and the varied
conservation treatments applied to them. A conserva-
tion strategy was designed for each amulet depending
on its condition. In general, the conservation strategy
included:
• unrolling tightly rolled-up objects that were not
intended to be unrolled;
• consolidation and cleaning of the heavily damaged
inscribed surfaces for philological examination,
making the invisible ancient script as visible as
possible; and
• arranging suitable support for storage and handling
for scholarly research.
These objects presented a challenge with regard to finding
appropriate conservation and preservation treatments
and also as far as providing researchers with access to
their invaluable information about life in this region in
the past.
462  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Non-Destructive Investigation of the Conservation
and Restoration of Ductile Metals in Cultural Heritage
Artifacts
Salvador Borrós* Patricia Giménez-Barrera
Instituto Químico de Sarrià Instituto Químico de Sarrià
Barcelona, Spain Barcelona, Spain
salvador.borros@iqs.url.edu patriciagimenezb@iqs.url.edu
*Author for correspondence
Ductile metals permeate a large part of our past and
culture. In ancient times, they were used to create utensils,
coins, medals, and plates in which historically important
inscriptions were engraved using artistic processes such
as embossing, one of the most characteristic metalworking
techniques that takes advantage of their malleability to
obtain a particular relief.
Notwithstanding the studies devoted to understanding
the degradation process in this type of material, there
is still a lack of consolidated protocols for their restora-
tion. It should be pointed out that resistance to corro-
sion depends not only on the chemical environment,
but also on the type of metal, mechanical forces, and
physical parameters involved, which increases the need
to develop alternatives for specific restoration. Thus,
the objective of this study was to offer an approach to
the conservation and restoration of ductile metals that
protects the integrity of cultural heritage artifacts. In this
sense, some of the problems associated with traditional
restoration and conservation techniques are described,
and basic information is provided about the metallurgical
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES
properties of ductile metals with respect to their possible
forms of deterioration.
First, corrosion was induced on lead specimens by means
of different aggressive solutions at different degrees of
concentration. Traditional corrosion treatments such as
chemical cleaning and electrochemical reduction were
then tested against a new methodology based on the
use of cold plasma which has been recently adopted for
the restoration of ductile metals. Since plasma generates
reactive species, its interaction with the surface of the
metal is very efficient and has very fast kinetics, which
allows for optimization of the restoration time. In addi-
tion, it should be noted that it is a controlled process with
regard to parameters such as gas composition and flow,
intensity, and reaction time. In this context, the poster
provides a useful discussion of the process, as well as
the results obtained on true-scale lead samples. Since
this technique reduces the risk of creating by-products
and thus constitutes a harmless treatment, it represents
a substantial improvement over more traditional tech-
niques used in the restoration and conservation field.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Freshly Excavated Coins: Best Practice for Rapidly
Obtaining Numismatic Information
Sabine Brechbühl
Archäologischer Dienst des Kantons Bern
Bern, Switzerland
sabine.brechbuehl@erz.be.ch
This poster focuses on the conservation of archaeological
coins. In the last ten years, the Archaeological Service
of Bern has seen its numismatic collection grow from
50 to 350 freshly excavated coins per year. This increase
has led to a reflection on our methods of conservation,
work practice, and organization of resources in order to
ensure that all incoming finds are stabilized, summarily
cleaned, and identified within a year.
The conservation-restoration unit handles all freshly
excavated finds that come from about 20 excavations per
year. In order to secure the preservation of these objects,
the main emphasis is generally on preventive conserva-
tion. However, the information potential concealed in
each object is often only revealed by curative cleaning
methods. Because archaeological coins are primary
sources that help date remains and structures in the
field, numismatic evidence is often rapidly requested.
To address this urgent need, a minimum standard was
set for succinct cleaning and stabilization.
The airbrush method was found to be very useful for
the rapid cleaning of coins. Known as an instrument for
applying color, it is used for this purpose with a solvent
(e.g. water). Both a cheap and accessible method, it
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  463
removes soil and soft corrosion products quickly and
effectively without direct contact on the fragile surface of
the coins. Serial procedures for stabilization treatments
involving chemicals are applied using ice cube containers.
Minimum standards were also set for documentation
and inventory. Each coin has an individual data entry,
its treatment procedure is assessed using predefined
keywords, and photographs are taken serially before and
after treatment.
Although treatments can be accelerated by adapting
methods and standards, the reach of such measures is
limited. As institutional resources decrease, it is necessary
to consider the scientific value of finds and to restrain
the entry of new artefacts according to archaeological
criteria. Therefore, recent coins that are still currently
in use are registered but not photographed or treated,
and then discarded.
In conclusion, the best practice seems to be the assess-
ment of needs and the definition of standards throughout
the whole archaeological process, with the engagement
of all partners. These standards must be regularly
adjusted in order to best fit the institutional goals and
resources.
464  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Comparing Different Deconcretion Strategies for the
Ballast Tanks of the H.L. Hunley Submarine
Anna Funke* Johanna Rivera-Diaz
Warren Lasch Conservation Center Warren Lasch Conservation Center
– Clemson University – Clemson University
North Charleston SC, USA North Charleston SC, USA
alfunke@clemson.edu JRD123@clemson.edu
The American Civil War submarine H.L. Hunley (1864)
has been going through an extensive deconcretion
process since 2014. A team of conservators removed
all the concretion from the exterior of the hull as well
as the interior of the crew compartment. The process is
almost finished, with only a few very challenging areas
left to deconcrete. The two most complicated sections
of the submarine left are the forward and aft ballast
tanks. Due to the small space and the personal protective
equipment needed by the conservators to work inside the
submarine, it is impossible for any members of the team
to physically enter the tanks and remove the concretion
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES
Gyllian Porteous *Author for correspondence
Warren Lasch Conservation Center
– Clemson University
North Charleston SC, USA
gporteo@clemson.edu
manually. Several different options have been considered
and tested out. Laser cleaning, oxyacetylene torches,
high-pressure water, air abrasive and dry ice are the main
methods that are currently being considered. Flexibility
in terms of accessing restricted and small spaces as well
as limited amounts of residue left behind are two of
the main considerations in choosing the best strategy
for deconcreting these difficult-to-reach areas. This
poster compares and contrasts the different approaches
considered for the deconcretion of the ballast tanks of
the submarine, taking a close look at both the advantages
and disadvantages of each method.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Custom-Made Desalination Structures for Iron and
Copper-Alloy Composite Objects
Alexander Dittus* Marian Schüch
Swiss National Museum Swiss National Museum
Affoltern am Albis, Switzerland Affoltern am Albis, Switzerland
Alexander.Dittus@nationalmuseum.ch Marian.Schuech@web.de
*Author for correspondence
Since the earliest days of metal casting, craftsmen have
combined different types of metals in their products.
Objects containing both iron and copper alloys are of great
beauty and functionality. Nowadays, however, they provide
archaeological conservators with complex conservation
issues. While the iron needs to be stabilized by alkaline
desalination, these baths cause significant surface damage
to the copper alloys. Protective layers have proved insuffi-
cient in providing complete protection for these surfaces
during desalination and therefore this research developed
methods to create “partial baths” for these objects.
Using acrylic glass, a box was created around the iron
blade of a folding knife with a bronze handle. Openings
were created to fit the bronze parts into the box in such
a way that they were not in contact with the desalination
solution. The contact zones of the bronze and acrylic
glass were sealed with Paraloid B-42 and then the box
was filled with cotton wool and placed on a high tray
with an alkaline desalination solution (20 g NaOH and
6.3 g Na2SO3 per liter of H20). Through capillary forces,
the solution came into contact with the folding knife’s
iron blade. At least once a day, some solution from the
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  465
tray was also placed brought into the top of the box with
a pipette, thus increasing the diffusion of fresh desali-
nation solution. The greater amount of liquid solution
(when compared to using pads) was expected to be more
efficient as it did not become as quickly saturated with
salts. Furthermore, it had the advantage of providing
solution samples for titration to record the amount of
salts removed.
For another iron object that was covered with solid bronze
decoration on one side, a stand was built from polysty-
rene so that the artifact could be supported over a bath
in such a way that only the iron part was immersed in
the solution, preventing the need for further protection
of the bronze.
As shown, custom-made desalination structures for
Cu/Fe composite objects provide a practical and creative
solution for preserving these delicate artifacts. Though in
some cases the initial design process may be protracted,
these structures may provide the only solution to preserve
copper and iron composite objects. Each object’s long-
term stability must be monitored on a regular basis since
it can only be partially desalinated.
466  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Use of a Heated Rinse to Facilitate the Thorough Removal
of Sodium Sesquicarbonate from USS Monitor’s Copper
Alloys After Desalination
Elsa Sangouard* Ralph Spohn
The Mariners’ Museum and Park The Mariners’ Museum and Park
Newport News VA, USA Newport News VA, USA
esangouard@marinersmuseum.org rjspohn@cox.net
*Author for correspondence
During a rehousing initiative for a portion of the
USS Monitor collection at the Mariners’ Museum and Park
(Newport News, Virginia), a powdery white bloom was
noticed scattered across the surfaces of several previously
treated copper-alloy artifacts. It was hypothesized that it
was residual sodium sesquicarbonate (Na3H (CO3)2) left
over from its use in storage solutions, as well as an elec-
trolyte during electrolytic reduction treatment which had
not been fully rinsed out. Samples of both the unknown
white powder and sodium sesquicarbonate crystals were
compared using Raman spectroscopy, which verified they
were the same material. Since sodium sesquicarbonate
increases in solubility with increasing temperature, it
was suggested that a potential way to remove the residue
from the artifacts was to re-rinse them using a hot water
bath. To provide proof of concept, heating trials were
undertaken using several blooming copper-alloy artifacts
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES
from USS Monitor. The objects were placed in succession
within a stainless-steel pot containing deionized water
and a thermometer. The pot was set on a hotplate and
brought to 70°C, seeing as it was the maximum temper-
ature reachable by a newly created hot rinse apparatus
in the lab. Within a short period of time, the pH of the
deionized water increased from 6 to 10, indicating that
sodium sesquicarbonate was being released. Using several
drops of phosphoric acid (H3PO4), the pH was adjusted
down to near neutral, with the process being repeated
over the following few days. Ultimately, a consistently
neutral pH was reached, implying that the remaining
sodium sesquicarbonate had been removed. The objects
were dried through solvent replacement and, to date,
blooming has not reoccurred. Currently, a procedure is
being developed that uses the lab’s hot water apparatus
for large copper alloys and for batch rinsing.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  467

Passive Removal of Silver Sulfide Layers from Metal

Threads on Textiles by Storage in an Intercept Bag

Ian D. MacLeod* Rinske Car

Western Australian Maritime Museum
Fremantle WA, Australia
ian.macleod@museum.wa.gov.au
*Author for correspondence
Denmark River Textile Conservation
Studio
Denmark WA, Australia
rinskecar@gmail.com

The results of a 20-month project to treat a woollen
and silver composite textile during its storage in a
controlled relative humidity microenvironment inside
a ziplock garment storage bag from Intercept are
presented. The jacket had been part of the uniform
of a member of the Papal Guard from circa 1830–50.
Large quantities of silver polish (Silvo) residues were
removed by handwashing in methylated spirits. Tests
on the previously covered metal thread showed it was
originally gilded but that the polishing had removed all
the surface gold. Approximately 85% of the silver surface
was significantly tarnished. Optical microscopy and
microhardness testing confirmed that the grey/black
coating was silver sulfide. In order to quantify the reac-
tion microenvironment, a datalogger was placed inside
the bag and confirmed the conditions were 23.6 ± 2.2°C
and 59.7 ± 3.7% RH during a typical 3-month interval.
The RH was controlled with pre-conditioned silica gel
held inside the Intercept bag as part of the standard
operating conditions.
A visual indicator of the rate of change of the sulfide was
taken using a dithionite-treated strip of woven metal
which had undergone standard chemical reduction
treatment at the start of the experiment. After 20 months,
the metal threads had become much lighter in colour,
which indicated the reaction was complete. The nature
of the conversion was established via a scanning electron
microscopy study of the external and interior surfaces of
one of the decorative elements from just above the split
tail at the rear of the jacket. The exterior metal strips
had only traces of Ag2S present in isolated patches and
the metal surfaces were 98 wt% silver. Stitched to the
surface of the wool were areas showing original gilding
with up to 65 wt% gold and the presence of both Ag2S
and AgCl on other sections of the braiding. The AgCl was
associated with a reaction to sweat and the Ag2S with the
high humidity hydrolysis of the sulfur-containing amino
acids in the wool or airborne H2S. Lightly burnishing the
treated silver surface with a soft cotton cloth changed
the metal from a dull grey to a fully lustrous condition.
468  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conservation Strategies for Cans in Collections
Aline Michel* Régis Bertholon
Haute Ecole Arc Haute Ecole Arc
Conservation-restauration Conservation-restauration
Neuchâtel, Switzerland Neuchâtel, Switzerland
aline.michel@he-arc.ch regis.bertholon@he-arc.ch
*Author for correspondence
The CANS project – Conservation of Cans in Collections
(2014–2017) – is the first study aimed at deepening
knowledge about heritage cans conserved with their
original content in museums and collections. The project
enabled conservation solutions to be proposed for these
complex composite objects.
After initially surveying cans in collections and their main
degradation issues (a study presented at Metal 2016, the
interim meeting of the ICOM-CC Metals Working Group
in New Delhi, India), the project focused on developing
conservation strategies centered on two main aspects:
preventive conservation and minimally invasive inter-
vention methods.
Cans are composite objects. The paper of the label, the
organic material of the content, and the corroded metal
of the container need different temperature and relative
humidity conditions. These apparent incompatibilities
in can conservation make it difficult to define compre-
hensive preventive conservation strategies.
Starting with a review of the literature on the average
environmental conditions suggested for each of the mate-
rials and considering the values attributed to the various
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES
Anaïs Vecten Laura Brambilla
Haute Ecole Arc Haute Ecole Arc
Conservation-restauration Conservation-restauration
Neuchâtel, Switzerland Neuchâtel, Switzerland
anais@vecten.net laura.brambilla@he-arc.ch
parts of the cans, different scenarios were simulated. The
study concluded that the metal container had a major
role in the conservation of the whole can and therefore
environmental conditions which safeguard its “survival”
should be prioritized.
Preventive conservation procedures should also comprise
periodic monitoring of the conditions of cans in collections
including control of internal and external corrosion. For
external corrosion, the number and size of corrosion spots
should be controlled every 6 to 12 months. Monitoring of
gas production inside the can (internal corrosion) should
be verified by measuring the height at the center of the can.
In some extreme cases, preventive conservation meas-
ures are insufficient to conserve cans safely. If cans place
the rest of the collection at risk, an intervention may be
necessary. This can be of two kinds: gas release using
adapted medical equipment (needle and injection site);
or opening and emptying the can in an aesthetically
compatible way by abrading the double seam. Special
procedures for a minimally invasive intervention in both
cases were developed by HE-Arc CR during the CANS
project and are described here.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conservation of Damaged Architectural Aluminum
Elements
Martina Raedel* Michael Buecker
Federal Institute for Materials Federal Institute for Materials
Research and Testing (BAM) Research and Testing (BAM)
Berlin, Germany Berlin, Germany
martina.raedel@bam.de michael.buecker@bam.de
*Author for correspondence
Aluminum is an oft-used building material in modern
architecture. In recent years, buildings from the
1950s and 1960s, including their aluminum parts, have
been increasingly subject to conservation. A common
sign of surface damage is corrosion caused by weath-
ering processes or scratches on the anodized layer due to
extensive wear. To repair damaged aluminum surfaces,
there are usually two options: use of a touch-up pen on
smaller areas and complete removal of the anodized
layer on larger ones. Both possibilities are disadvan-
tageous for the objects. The touch-up pen often does
not match the color and provides insufficient corrosion
protection, while newly anodized layers often differ in
color and gloss from the original surface.
For this reason, a research project was initiated to
develop a mobile method to repair anodized aluminum
parts. The first step focused on using an electrolyte with
a gel substrate. Different cathode materials were tested
together with an appropriate cooling material. The
anodized area was isolated to protect undamaged ones.
Examinations of the layers produced were conducted
using a Keyence microscope, eddy current testing and
environmental scanning electron microscopy (ESEM).
Further steps included transforming the coloring and
sealing process for mobile application. Color can be
applied by using a brush or spray, while the sealing
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  469
Ines Feldmann Mandy Reimann
Federal Institute for Materials Haber & Brandner Metal
Research and Testing (BAM) Conservation Studio
Berlin, Germany Berlin, Germany
ines.feldmann@bam.de mandy.reimann@haber-brandner.de
process can be performed with water vapor. A heatable
putty knife can be used when it is necessary to heat the
area. Once the application process has been developed,
anodization can be tested on samples with artificial
damage fixed in horizontal and vertical positions. The
stability of the anodized surface can be examined by
accelerated aging in a climate chamber and outdoor
weathering.
The electrolyte (diluted sulfuric acid) was combined
with a gel binder to enable mobile application. Several
thickening agents were tested for conductivity and
stability in acid systems. Anodization tests with
different cathode materials and shapes were conducted.
The temperature during anodization was controlled
and adjusted where necessary. The anodized area was
restricted by using either a lacquer, an adhesive, or a
removable silicone barrier. The first results showed
the feasibility of the method on enclosed areas. The
achieved thickness was measured by eddy current
testing and the structure was controlled by ESEM.
Examinations showed a connection between the
thickness and porosity of the anodized layers and the
temperature during the application process. The next
steps are to test mobile coloring and sealing methods
followed by mobile anodization on artificially damaged
areas.
470  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES

The INTERFACE Project: Conservation of Painted Metal

Artefacts

Monica Albini* Cristina Riccucci Chiara Giuliani Elena Messina

National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
monica.albini@gmail.com
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
cristina.riccucci@cnr.it
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
chiara.giuliani@ismn.cnr.it
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
elena.messina@cnr.it

Gabriel Maria Ingo
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
gabrielmaria.ingo@cnr.it
*Author for correspondence
Maria Paola Staccioli
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
mariapaola.staccioli@ismn.cnr.it
Gabriella Di Carlo
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
gabriella.dicarlo@cnr.it
Marianna Pascucci
National Research Council of Italy
(CNR)
Institute for the Study of
Nanostructured Materials (ISMN)
Rome, Italy
marianna.pascucci@cnr.it

The use of science in the conservation of cultural
heritage is nowadays widespread. Many studies have
been conducted on artworks made of single materials
(e.g. painting, stone, metal), but a novel research field
has attracted growing interest among European conser-
vation scientists: the characterisation and conservation
of composite artefacts. The INTERFACE (paINTed
mEtal aRteFActs ConsErvation) project, funded by the
European Union’s Horizon 2020 research and innova-
tion programme under the Marie Skłodowska-Curie
Action, is focused on composite artworks made of
painted metal. Indeed, the use of metals as ‘canvas’ has
been poorly investigated, even though many master-
pieces were created using this particular technique. The
degradation mechanisms that occur on metal artefacts
and paintings, as single materials, are well known, but
studies on painted metals and paint-metal interactions
are still rare. There is, in fact, an extensive lack of
knowledge about the degradation processes that occur
on such artefacts and the conservation methodology to
be adopted. The aim of the INTERFACE project is to
fill this gap in scientific information by achieving two
main objectives:
1. the characterisation of the degradation mechanisms,
with particular attention to the processes occurring at
the paint-metal interface (e.g. the formation of metal
soaps due to the interaction between the fatty acids
of the binder and the metal substrate);
2. the development of a conservation methodology to
preserve both the paint film and the metal substrate.
In particular, copper and iron/low carbon steel as
substrates decorated with linseed oil paints are inves-
tigated. This study presents the preliminary results of
the project.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conserving Clocks: Thoughts and Challenges
Ryma Hatahet
Freelance conservator
Besançon and Paris, France
ryma.hatahet@gmail.com
Having learned and practiced clock conservation for four
years after a degree in metal conservation, the author has
been confronted with numerous challenges and thoughts
in the process of offering clock conservation services to
museums.
A ticking and accurate clock mechanism is the successful
outcome of the clockmaker’s craftsmanship and technical
choices. The clockmaker selects appropriate metals
(hard or soft) and prepares the functional surfaces (by
polishing, burnishing, hardening) in order to avoid
wearing. A clock’s construction starts with calculating
the gear arrangement and angles of its parts, which
play a prominent role in defining the clock’s identity. In
addition, the craftsman will pay attention to the object’s
aesthetics. Therefore, an important connection exists
between technique and aesthetics that can be held as a
central feature in clock conservation projects.
Throughout the conservation process, the clock conser-
vator is in a position to collect precious technical infor-
mation about the clockmaker’s choices and skills, as well
as about the possible tools used. This intangible heritage
is crucial and may help curators and conservators deter-
mine or adjust the conservation project.
All objects contain intangible heritage (for example,
bronze sculptures tell us about the foundry where they
were made). All technical objects have something to
teach us about the skills, manufacturing processes and
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  471
industries that produced them. What makes this heritage
particularly interesting and central in clockmaking is that
most of the techniques and tools of production devel-
oped are still the only way to perform a conservation
treatment, especially when the goal is to make the clock
work again. By documenting a clockmaker’s technical
choices and skills, a clock conservator also widens their
own knowledge. They also take part in the survival of
clock craftsmanship by practicing clock making and
repairing techniques. In parallel, as horological heritage
encompasses a wide range of subjects and skills, a
clock conservator works closely with clock restorers.
Clock conservation brings together museums and living
heritage. Sadly, most of the time this collaboration stays in
the shadows: professionals do not seem to communicate
about it between them or to the public.
Often, the main question and project is whether to
put a clock back in working order. Unfortunately, this
question prevents us from addressing other important
issues. What methods apply when conserving intangible
heritage related to clocks? What are the specific ethical
issues? What percentage of modification (old and new)
is tolerable?
The poster focuses on the various issues and challenges
faced recently and discusses the preliminary responses
that the author came up with as a young freelance
conservator.
472  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Exploring Suitable Parameters for Electrochemical
Cleaning/Consolidation of Museum Objects and Practical
Use of the Pleco Electrolytic Pencil
Nataša Nemeček* Eva Menart
National Museum of Slovenia National Museum of Slovenia
Ljubljana, Slovenia Ljubljana, Slovenia
natasa.nemecek@nms.si eva.menart@nms.si
*Author for correspondence Domen Lavriha
University of Ljubljana
Faculty of Chemistry and Chemical
Technology
Ljubljana, Slovenia
domen.lavriha@gmail.com
In the past, many objects in the collection of the
National Museum of Slovenia were severely damaged
due to electrolytic/electrochemical cleaning. In the
early 1990s, the decision was made to abandon this
kind of treatment due to its harshness, although at the
time important developments in electrolytic/electro-
chemical cleaning were underway in Europe that used
stricter control parameters such as cathodic potentials
determined by voltammetry.
Although this cleaning procedure was avoided in the
museum for two decades, an eye was kept on the new,
strictly controlled electrolytic/electrochemical processes,
particularly on those that could be carried out locally.
To prepare for their re-introduction in the museum’s
laboratory, extensive testing of different metal/solution
combinations was carried out through voltammetric
measurements.
The Pleco, an open-source electrolytic pencil developed
by the research unit of the conservation department at the
University of Applied Sciences and Arts Western Swit-
zerland (Neuchâtel), was first tested on metal mock-up
samples and then used on real museum objects. The most
promising and easily achieved results were obtained on
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES
Marko Jeran Christian Degrigny
University of Ljubljana Haute Ecole Arc Conservation-
Faculty of Electrical Engineering restauration
Ljubljana, Slovenia Neuchâtel, Switzerland
marko.jeran@fe.uni-lj.si christian.degrigny@he-arc.ch
Tomaž Lazar
National Museum of Slovenia
Ljubljana, Slovenia
tomaz.lazar@nms.si
tarnished silver, both in artificially induced tarnishing
on a silver sample (using liver of sulfur) and natural
tarnishing accumulated over different periods of time on
different (real) objects. The most extensive work carried
out using the Pleco was successfully completed on a lead
seal from a bull issued by Pope Alexander IV in 1257,
which was attached by a set of strings to a parchment
document with which the Pope confirmed the privileges
granted to one of the oldest monasteries in present-day
Slovenia. Before the consolidative reduction procedure,
the parchment was protected using a polyethylene bag
and the strings were covered with cyclododecane. After
obtaining the necessary parameters based on literature
data, the Pleco was used for local consolidation. The
bull had a very thick layer of white corrosion products
covering its entire surface, which led to the procedure
taking place over several weeks and being stopped and
restarted each day.
Both in the case of silver and lead objects, the length
of the cleaning/consolidation procedure was found to
depend greatly on the thickness of the corrosion layer,
which is presumably influenced by the length of exposure
to harmful atmospheric pollutants.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES  473

Use of Gels for the Treatment of Metals

Elodie Guilminot* Alban Gomez Aymeric Raimon Manuel Leroux

Arc’Antique – Grand Patrimoine de
Loire-Atlantique
Nantes, France
Elodie.Guilminot@loire-atlantique.fr
*Author for correspondence
Arc’Antique – Grand Patrimoine de
Loire-Atlantique
Nantes, France
Alban.Gomez@loire-atlantique.fr
Independent conservator
Ormes, France
aymeric.raimon@outlook.fr
Centre de Recherche et de
Restauration des Musées de France
(C2RMF)
Paris, France
manuel.leroux@culture.gouv.fr

Since 2017, the Arc’Antique conservation laboratory has
been researching gel cleaning for historical objects. The
aim is to test certain physical gels (agar, and gellan and
xanthan gum) commonly used in other conservation
fields, such as on paintings and porous materials, on
metal objects. At the same time, the laboratory has tested
new gels such as Nanorestore and PVA.
To further this study, a participatory method was put in
place. In 2018, institutional and private conservators and
scientists met in collaborative workshops to share their
experiences, discuss their needs, and contribute to the
development of research themes. An online collaboration
tool was then created to share and find information on
the advanced research initiated.
Five main points about gels and metal cleaning were
identified in 2018 leading to the following advances:
improvement in gel preparation protocols and applica-
tion; study of the compatibility of gels with active agents
by characterizing their physico-chemical properties
(IR spectroscopy, Young’s modulus, etc.); interaction and
depth of action of a gel treatment; evaluation of the pres-
ence of gel residues on different surfaces; and widening
of the field of application of gels to metal conservation,
such as electrochemical treatment (dechlorination), and
archaeological objects.
The first results of this study showed the importance
of the gel preparation protocol. Agar gels form very
resistant films that peel off when heated twice to 90°C.
Nevek gel (ready-to-use agar gel marketed by CTS) is
particularly suitable for treating metals. PVA-borax films
require rigorous preparation to form a gel (known as
a slim) but their high viscosity makes them unsuitable
for metal treatment. The compatibility of the different
gels with agents generally used in metal treatments
was also studied. Agar gels and gellan gum cannot be
prepared hot with basic or acidic solutions, but once
the films are formed, it is possible to incorporate basic
or acidic solutions by immersion. PVA-borax gels are
incompatible with basic, acidic, or complexing solu-
tions. Xanthan gum is incompatible with basic solu-
tions. Nanorestore chemical gel has the best chemical
compatibility with all tested agents. Initial tests on iron
dechlorination (electrolysis in KNO3 or NaOH) and on
copper alloy chemical cleaning (with complexing agents)
were very promising. Agar films and gellan gum give
very good results when the gel is applied hot. Xanthan
gum, a viscous gel, is difficult to remove. Tests with
Nanorestore films were often inconclusive because of
insufficient contact between the surface of the object
and the gel film.
By leading this project on multiple fronts and involving a
wide range of practitioners and theorists up to an inter-
national level, the aim was to improve knowledge of the
properties of gels in order to facilitate decision-making
based on the product treated and the constraints iden-
tified.
474  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Finding the Way Forward with Atomic Layer Deposited
Coatings on Silver Heritage Objects: Moving from the
Laboratory to Conservation Practice
Eric Breitung* Sara Creange
The Metropolitan Museum of Art Rijksmuseum
New York NY, USA Amsterdam, The Netherlands
breite@metmuseum.org Creange@rijksmuseum.nl
*Author for correspondence
Silver objects require protection from corrosive pollut-
ants such as sulfides to minimize tarnishing. Preventive
tools such as polymeric coatings, inert gas enclosures,
and sulfide scavenging “silver cloth” are used to achieve
a corrosion-free environment. However, these tools can
be labor-intensive to implement, difficult to maintain,
or enclose the object in a visually disruptive container,
making viewing impossible without handling. When
preventive methods are not used or fail, polishing is
commonly implemented to regenerate an aesthetically
inappropriate surface. Repeated polishing unfortunately
removes significant amounts of material and ultimately
results in loss of surface detail.
Atomic layer deposited (ALD) films of aluminum oxide
have been proposed by a handful of laboratories as a
long-term, transparent, and durable coating for silver
heritage objects. Silver coupons have been coated and
tested for resistance to tarnishing, reversibility, and
optical clarity with much success, with ALD-coated
samples being predicted to resist tarnishing for decades
longer than nitrocellulose-coated samples. The coat-
ings proved removable with little impact on the alloy’s
surface composition, and conservators were unable to
distinguish coated and uncoated samples. Much of this
work, however, used pure or sterling silver that was
flat, polished, or otherwise not pickled, treated, cast,
POSTER SESSION – DEVELOPMENT OF CONSERVATION PRACTICES
Tamar Davidowitz
Rijksmuseum
Amsterdam, The Netherlands
T.Davidowitz@rijksmuseum.nl
or worked. This poster summarizes these results and
introduces the current collaborative investigations of the
Metropolitan Museum of Art and the Rijksmuseum into
the coating properties of ALD on samples and objects
that are more representative of actual historic silver in
real-world conditions.
To assess ALD coatings for “real-world” objects and
environments, the effects of various silver alloys, shapes,
and textures were considered, along with the effects of
wear, handling, and aging. The first goal was to deter-
mine a coating thickness that is optically invisible on
actual silver objects. The effects of alloy variations and
impurities such as lead, gold, bismuth, and zinc on the
ALD coatings were examined by fabricating “histor-
ical” silver samples and testing them alongside sterling
silver, commercially available silver standards, and
real-world silver objects with a range of surface patinas
and bulk compositions. Further experiments addressed
the practical concerns of conservators: whether ALD
coatings can be applied to “imperfectly cleaned” silver,
as [invisible] tarnish or polish residues often remain
present on the surface after cleaning. The durability
of ALD coatings was evaluated by subjecting coated
objects to handling and dusting. Finally, removability
of the coating was tested on a range of silver alloys,
even after aging.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – CONSERVATION TREATMENTS  475

Conservation and Stabilization of Iron Objects from the

Archaeological Site of La Tène (Neuchâtel) in View of New
Investigations and a New Exhibition in the Laténium
Emmanuelle Domon Beuret* Christian Cevey Gianna Reginelli Servais
Laténium – Parc et Musée Laténium – Parc et Musée Office du Patrimoine et de l’Archéologie (OPAN) –
d’Archéologie de Neuchâtel d’Archéologie de Neuchâtel Section Archéologie
Hauterive, Switzerland Hauterive, Switzerland Institut d’Archéologie – Université de Neuchâtel
emmanuelle.domonbeuret@ne.ch christian.cevey@ne.ch Hauterive, Switzerland
*Author for correspondence gianna.reginelli@ne.ch

This poster presents the methodology implemented at the
conservation-restoration laboratory of the Laténium in
Neuchâtel, Switzerland, to safeguard the archaeological
iron objects in its La Tène collection. After a condition
report was drawn up, the objects were cleaned, photo-
graphically documented, and the active corrosions
stabilized.
This conservation-restoration campaign was undertaken
as part of the museological renovation of the Iron Age
section of the Laténium’s permanent exhibition, which is
being carried out under the framework of an international
cooperation research project aimed at re-examining La
Tène, the eponymous archaeological site from the Second
European Iron Age (450–15 BC). It includes the inven-
tory, documentation, and valorization of the La Tène
collections housed in many different museums in Europe.
476  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Alkaline Sulfite Desalination of Formerly Restored
Damascened Merovingian Belt Buckles: A Possible
Challenge
Florence Lagger
Service archéologique de l’Etat de Fribourg
Fribourg, Switzerland
florence.lagger@fr.ch
There are several hundred examples of restored iron
Merovingian belt buckles (some of which are decorated)
in the Service archéologique de l’Etat de Fribourg.
Half of the collection is now showing signs of active
corrosion, so it was decided to desalinate all of them
and to perform restoration. This process afforded some
unexpected surprises.
As most restoration work was done in the 1970s and
1980s, it was difficult to find precise documentation
about the treatments, but they can be summarized as
follows: the buckles were soaked in epoxy resin and
then milled; the decorative silver and brass wires were
polished to make them shine; some of the buckles
seem to have been pre-treated with plasma, or tannic
acid; and, in some cases, a protective coating (such as
microcrystalline wax or an acrylic resin) was applied
to the surface.
The desalination treatment in Fribourg consisted in
immersing the objects in a solution of sodium hydroxide
(0.5 mL) and alkali sulfite (0.2 mL) in demineralized
water. When possible, every coating on the belts was
removed before treatment.
POSTER SESSION – CONSERVATION TREATMENTS
Some interesting reactions were observed on the belt
buckles during and after the baths. These included strong
silver and brass tarnishing, re-deposition of restoration
products on the surface, embrittlement, color change
of resins (fillings, adhesives), etc. Sandblasting with
crushed glass was found to be the best way to remove the
tarnishing and/or the re-depositions of former restoration
products on the surface. This technique even helped to
discover new decorations on the belt buckles, such as
engraved lines never seen before on an iron belt with
a very smooth, flat surface due to the milling of epoxy.
The silver and brass wires were then sandblasted with
glass beads to give them a less matte, but not too shiny,
appearance. A protective coat of Paraloid B-44 diluted at
3% in acetone was applied on the surface at the very end.
Desalinating and re-restoring these artifacts is a long-
term and challenging task. One disadvantage was some
weakening of the wires and some possible loss during
sandblasting. But ensuring the long-term stabilization
of the belt buckles, with more visible decoration on the
surface and, in some cases, new information about them,
was well worth the effort.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conservation Treatment of the Six-Inch, Thirty-Caliber
Mark III Gun and Carriage on the USS Maine (ACR-1)
Justin Schwebler
Warren Lasch Conservation Center – Clemson University
North Charleston SC, USA
jschweb@clemson.edu
This poster is a detailed review of a multiyear conser-
vation treatment of a historic naval gun, carriage, blast
shield, and memorial plaque from the United States
battleship USS Maine. The poster covers all aspects of the
project, from initial site visits and transportation from
the US Navy Yard in Virginia to South Carolina for the
conservation work.
In the fall of 2016, the external projects team at Clemson
University’s Warren Lasch Conservation Center began
using a high-pressure, superheated water blasting
system to remove failing coating layers from the severely
corroded gun and carriage. The composite nature of the
gun required careful planning and work by the team
to make sure that all aspects of the gun, made of both
iron and copper alloy, were being properly cared for and
conserved throughout the process. The gun assembly
weighs a staggering 17.5 tons which made the trans-
POSTER SESSION – CONSERVATION TREATMENTS  477
portation and treatment of the gun very challenging.
Following coating removal, corrosion mitigation, and
cleaning the gun was treated with a mild phosphoric
acid solution to stabilize the surface before the appli-
cation of new protective marine coatings for its display
outdoors. The team also carried out a careful repair and
conservation of a large copper-alloy memorial plaque
that had been affixed to the front of the blast shield in
the 20th century.
The poster provides a brief history of the battleship and
the guns context as a part of the ship’s service, sinking,
and recovery. The main body of the poster contains the
conservation work in detailed literature and photography.
The bottom section covers the final application of coat-
ings and the finished gun in photographs. The gun was
completed and delivered back to the United States Navy
in the spring of 2018.
478  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Backyard Treasures: A Case Study from Kulautuva
Cemetery, Lithuania
Asta Vasiliauskaite* Mantas Daubaras
Freelance conservator Freelance archaeologist
Kaunas, Lithuania Vilnius, Lithuania
vasiliauskaite.asta@gmail.com mantas.daubaras@gmail.com
*Author for correspondence
This study discusses the conservation of an extraordinary
archaeological discovery in Lithuania – the grave goods
from burial no. 27 in Kulautuva cemetery. The objects
date from the 12th to the 14th century.
Kulautuva cemetery is listed as a national archaeological
site and is located in Kaunas district in central Lithuania.
Currently, this protected area is occupied by private
landowners and their houses. According to Lithuanian
law, archaeological excavations must be carried out in
the protected areas before any ground disturbance takes
place. One of the landowners was planning to build a
garage and water and electricity lines and so commis-
sioned an archaeological excavation.
During the excavations at the site in 2018, seven burial
pits were found, one of which, no. 27, had a very impres-
sive collection of grave goods. The burial pit was 2.2
m in length, 1.3 m in width and 1.1 m in depth and
was filled with light grey sand. Its contents included a
large number of disarticulated horse bones and richly
decorated riding equipment: iron binary bridle bits,
POSTER SESSION – CONSERVATION TREATMENTS
iron bridle bits with massive profiled crossings, iron
stirrups, iron spurs, copper-alloy spirals, various shapes
and sizes of iron buckles, fragments of riding equipment
and organic material (fragments of leather and textile).
Some of the iron riding equipment was decorated with
silver plates.
After burial pit no. 27 was excavated, five other small pits
were discovered in the natural soil. All five pits contained
a hoard of cut and broken silver jewellery weighing a
total of 273 grammes.
The objects were in poor condition. The composition
of the metal alloys was identified using x-ray fluores-
cence spectroscopy. The iron and copper-alloy objects
were cleaned mechanically, while the silver objects were
cleaned chemically. The iron objects with silver plates
were cleaned using both methods as well as polymer
film peeling. The cleaned objects were consolidated and
a protective wax coating applied.
The conserved objects are stored at Vytautas the
Great War Museum.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Conservation of a Restored Late Iron Age Strainer Bowl
Saray Naidorf* Maickel van Bellegem
British Museum Al Ain Museum
London, UK Al Ain, United Arab Emirates
snaidorf@britishmuseum.org mjtmvb@gmail.com
*Author for correspondence
A late Iron Age (AD 1–100) copper-alloy strainer bowl
with red enamel decoration was found on the grounds
of the Chettle Park Estate, Dorset, in 2003, together with
seven other objects. Discovered by a local metal detec-
torist and excavated by archaeologists from the Wessex
Archaeology Unit, the assemblage consisted of an Iron
Age decorated mirror, an Iron Age strainer bowl and four
other copper-alloy vessels, a large glass cinerary urn and
a small blue glass jar of Roman origin.
Once the assemblage was acquired by the British Museum
in 2009, all the artefacts were re-examined by the Depart-
ment of Conservation and Scientific Research before
any decisions were made on the overall conservation
treatment. As one of the most complex objects in the
group – composed of various elements – the strainer
bowl had been heavily restored in 2006. The preliminary
study using binocular stereo microscopy, x-radiography
and x-ray fluorescence indicated that, in its previous
restoration, large missing areas of the bowl had been
filled and in-painted, some fragments had been placed
incorrectly or floated in the filled areas and some others
possibly belonged to another copper-alloy object from
the assemblage, as identified by the metallurgical anal-
POSTER SESSION – CONSERVATION TREATMENTS  479
Duncan Hook
British Museum
London, UK
dhook@britishmuseum.org
ysis. It was also found that AJK dough had been used
as a gap-filling material, posing major irreversible and
shrinkage issues that could cause further damage to the
object, making it ethically ‘unacceptable’ within current
conservation practices.
The main purpose of the conservation work was,
therefore, to remove the unethical restoration and to
reconstruct the strainer bowl by using materials that are
easily reversible, whilst aiming to give the object a more
authentic appearance, compatible with its fragmentary
condition. This poster focuses on the intricacy of this
conservation treatment, the results of the scientific
analyses carried out and how these shaped the decisions
on the conservation approach. It also discusses the inter-
esting aspects revealed during conservation, including
its ancient repairs and replacement parts, which gave an
insight into the archaeological history of the object as
well as its importance during its original domestic use.
The thought process that led to the production of a
minimalist support mount, which would enable this
strainer bowl to be permanently displayed in the
Britain and Europe gallery at the British Museum, is
also presented.
480  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Restoration of a Decorative Baroque Door Lock from
ca. 1720–40
Jan Růžička
National Technical Museum
Prague, Czech Republic
ruzick.jan@email.cz
The study describes the restoration of a wrought-iron
Baroque warded door lock made probably between
1720 and 1740. The top external right door lock with
a rectangular lockbox and a German-type mechanism
comes from the collections of the Regional Museum
in Kolín, Czech Republic. The chaplet, key guiding
pin, and the wards and their individual ward plates
are doubled, which is typical of Central European
door locks from the mid-18th century. An embossed
volute ‘C’ decoration on the cover panel is in typical
Baroque style—symmetrical along both vertical and
horizontal axes. The basic sheet metal is decorated with
two embossed convex scrolls. The escutcheon, which
was located on the outer side of the door, is attached
to the spindle of the exterior handle. The original key
POSTER SESSION – CONSERVATION TREATMENTS
is missing. Restoration research showed the presence
of non-original oil-resin varnish and corrosion under
the paint. The paint has already lost its protective and
aesthetic function and, due to its acidic nature, has had
a negative influence on the base material. The object’s
original surface finish was a hot dip coating of tin alloy.
This was damaged by the corrosion of the basic iron
surface underneath. Due to the presence of chloride ions
in the object and increased humidity during storage, the
corrosion of the base material was relatively intense.
Restoration intervention involved dismantling, cleaning,
deionizing, removal of iron corrosion products that
were not firmly bonded to the base, surface stabilization
with tannin solution, air-drying, microcrystalline wax
coating, and re-assembly.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – CONSERVATION TREATMENTS  481

Conservation of 8th–9th-Century Curonian Women’s

Jewellery

Sandra Garšvienė* Vytautė Lukšėnienė

Lithuanian Art Museum/Pranas
Gudynas Restoration Centre
Vilnius, Lithuania
sandra.garsviene@gmail.com
*Author for correspondence
Lithuanian Art Museum/Pranas
Gudynas Restoration Centre
Vilnius, Lithuania
vlukseniene@yahoo.com

The Curonians were a Baltic tribe living in the north-
west of Lithuania whose burial sites are distinctive for
including rich grave goods. Stray artefacts from the
8th–11th-century Curonian burial site of Kalnuvenai
(Klaipeda district) first came to light at the end of the
19th century. The burial ground was re-discovered in
2001.
In 2008, the disturbed burial site of a rich Curonian
female dating from the 8th–9th century was discovered.
The grave goods consisted of a cruciform copper-alloy
pin with pendants and a chain ensemble decorated with
chased silver plates and blue-glass inlays, three copper-
alloy bracelets, six copper-alloy coil rings, a round
copper-alloy pendant with a chased silver plate and
blue-glass inlays, an iron knife, an iron awl, and a blue
glass bead. Cruciform pins were common among the
Curonians in the 8th–12th century, while round pendants
were especially found in the graves of rich women dating
from the late 8th–9th century.
The cruciform pin with pendants and chain and the
round pendant were in poor condition. The metal of the
pin and pendant was very corroded, the silver plates were
very fragile, and many of the small chains were broken.
The silver plate on the pin’s head was detached from the
base and had broken into fragments. The archaeologist
retrieved these fragments and glued them onto paper.
The accurate on-site work of the archaeologist facilitated
further laboratory conservation.
The artefacts were x-rayed and the composition of the
metal alloys and solder was identified using x-ray fluores-
cence spectroscopy. The artefacts were cleaned mechani-
cally and chemically. The corrosion products were cleaned
using the method of polymer film peeling. Japanese paper
was used as the backing material for the glued parts.
482  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Just When You Thought You Knew How to Polish Silver:
An Analysis of Silver Polish Applicators
Tina Gessler* Kate McEnroe
Colonial Williamsburg Foundation Colonial Williamsburg Foundation
Williamsburg VA, USA Williamsburg VA, USA
cgessler@cwf.org kmcenroe@cwf.org
*Author for correspondence
Treatment of decorative arts silver objects in museum
and historic house collections often involves some form
of tarnish reduction, whether mechanical, chemical,
or electrochemical. This poster focuses on specific
aspects of tarnish reduction with steam and solvents
as well as mechanical cleaning with a well-tested abra-
sive (precipitated CaCO3). While significant research
has been undertaken by others to determine the most
effective and safe polishing compounds, the variable
introduced by the polish applicator has not been fully
studied. Additionally, an informal survey conducted
by the authors  indicates that many labs are using
solvent cleaning as a form of light tarnish reduction or
preliminary cleaning prior to polishing with an abra-
sive. This practice has the potential to emphasize the
impact of the applicator, despite the fact that solvent
cleaning is likely intended to be less interventive than
abrasive polishing. Through controlled experimentation
accompanied by instrumental analysis, the authors have
analyzed the role of pre-cleaning with various solvents
and steam plus the interplay of polish and its applicator,
be it silk, a cotton swab, Tyvek, cotton flannel, a stencil
brush, or other material. This poster serves to share
our observations and recommendations regarding an
POSTER SESSION – CONSERVATION TREATMENTS
optimal workflow to achieve efficient tarnish removal
with minimal surface abrasion.
The experimental methods chosen were based on our
survey and on the Colonial Williamsburg Foundation’s
silver treatment protocol. Mounted sterling silver coupons
imaged at macro and micro levels were used throughout
the research process. Imaging was done at Colonial
Williamsburg with a Nikon stereomicroscope and Phenom
Pro-X desktop SEM-EDS system. Results indicate that a
qualitative difference between scratch frequency among
applicators all initially assumed to be good choices for
polishing silver can be noted even at the macro level. In
addition, SEM-EDS data collected through elemental
mapping is indicative of the success of tarnish removal
based on the presence or absence of sulfur. Final conclu-
sions are based on analytical data and qualitative obser-
vations made by conservators who regularly care for
historic silver. Pre-cleaning was found to be effective
using steam and certain solvents. Various forms of cotton
were preferred in terms of polish application, while other
applicators proved too abrasive. The results of this study
complement the large body of research on silver polishes by
contributing valuable information regarding non-abrasive
cleaning methods and the effects of polish applicators.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND POSTER SESSION – CONSERVATION TREATMENTS  483

Restoration and Conservation in Unfavorable Working

Conditions: A Case Study

Zoran Kirchhoffer* Sonja Kirchhoffer

Nikola Tesla Technical Museum
Zagreb, Croatia
zoran.kirchhoffer@tmnt.hr
*Author for correspondence
Crafts for the Restoration and
Conservation of Industrial Archaeology
Kumrovec, Croatia
sonjakirchhoffer03@gmail.com

This poster provides an overview of the basic restora-
tion-conservation principles that apply when working
on technical objects and their parts under unfavorable
conditions, i.e., when time is limited and the resources
available are insufficient. At the Nikola Tesla Technical
Museum, there are numerous different objects made of
very different materials and constructed using different
technologies. The aim is to show how an artifact can
be restored when it is not possible to use complex and
expensive technologies, or when the approach is largely
at the experimental level involving “reading” information
about given technologies and materials. Through exam-
ples from the early 21st century, such as a Maserati car
and solar panels and cells, emphasis is placed on the idea
that even if it is not possible to use costly technology, it
is possible to establish a connection between science and
museum practice in the sense of applying experimental
methods.
484  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Treatment of a Composite Sock-Knitting Machine Using
Dry-Ice Cleaning
Paul Mardikian* Claudia Chemello
Terra Mare Conservation, LLC Terra Mare Conservation, LLC
Charleston SC, USA Charleston SC, USA
paul@terramareconservation.com claudia@terramareconservation.com
*Author for correspondence
Terra Mare Conservation, LLC, was contracted by
Bynon Art Services, LLC, to undertake conservation of
a sock-knitting machine for the Tennessee State Museum
in Nashville, Tennessee, in preparation for display in the
museum’s new building.
This machine was made by the Hemphill Manufac-
turing Company around 1913–15 and was used in one
of the hosiery mills in Englewood, McMinn County,
Tennessee, for most of the 20th century. At one time
Englewood had over 30 mills and hosiery mills were
important sources of employment for women in the
southern United States. This intricate and sophisticated
machine is made of hundreds of moving parts and a
variety of materials, including painted cast iron for the
base and housing, steel, brass, wood, rubber, textile (a
partially knitted sock is inside the canister), cotton fiber
wrapped around a belt pulley handle, spools of textile
and several pieces of paper.
The object was completely caked with oil, black grease,
dust and debris, and could not be properly handled,
studied or displayed due to the extent of the grease. The
interior of the machine and the horizontal ledge above
the legs were covered with thick and hardened layers of
built-up oil and fibers from the sock-making process.
POSTER SESSION – COMPOSITE OBJECTS
Mark Bynon
Bynon Art Services, LLC
Taylorsville NC, USA
markbynon@gmail.com
About 75% of the exterior surface of the machine was
painted black, but the paint was barely visible beneath
the thick layers of grease, was heavily abraded, pitted
and damaged overall and, in some areas, completely lost.
For removal of the contaminants, it was decided to use dry
ice, a technique that has become more widely accepted in
conservation as a tool for specific applications. Due to the
complexity of the object, a smaller dry-ice machine and
finer media than those typically used on larger industrial
artifacts or sculptures were chosen. The Cold Jet i3 Micro
Clean dry ice machine was chosen for its great versatility
and precision. Instead of frozen carbon dioxide pellets,
blocks of dry ice are shaved into sugar-sized particles
and propelled through small, pencil-shaped nozzle tips,
offering maximum flexibility. The blast pressure used
was between 1.4–2.0 bar/20–30 psi (with a flow rate of
ice of 0.3 kg/0.6 lbs. per minute).
Cleaning enabled the machine to be further studied and
handled and revealed the extent of the original paint
system. Overall, this technique was very successful in
removing the caked oil and grease, even from areas that
were difficult to reach. The treatment avoided the need
for toxic solvents and caused no damage to the original
paint system, movable parts or substrate.
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Spectacular Spectacles: Corrosion Cases
Judith Huber*
State Academy of Art and Design Stuttgart
Stuttgart, Germany
ma.ju@bluewin.ch
*Author for correspondence
Vision aids such as spectacles have been systematically
collected for over 100 years. They show traces of usage,
care and repair by owners and opticians that have
influenced their ageing. As composite objects made of
different materials, spectacles show a range of degrada-
tion mechanisms. Their preservation is thus as interesting
as it is challenging. Glasses can be ground from different
types of glass. Through the ages, they have been framed
in, for example, horn, baleen and tortoiseshell. After
1600, frames were chiefly made of metal, especially
copper alloys, iron, steel, silver, gold and nickel alloys. In
the late 19th century, plastics like cellulose nitrates and,
later, acetates were introduced.
The condition of glasses varies a great deal – including all
stages of glass corrosion. Specific corrosion phenomena
on metals were observed and investigated using scanning
electron microscopy/energy-dispersive x-ray, µ-Raman
and Fourier transform infrared spectroscopies, and x-ray
diffraction based on a survey of the remarkable spectacle
collection of the Institute for the History of Medicine,
University of Bern, complemented by further examples
with particularly typical damage patterns:
• On frames of a substantial number of spectacles made
of non-ferrous metals, glass-induced metal corrosion
POSTER SESSION – COMPOSITE OBJECTS  485
Andrea Fischer
State Academy of Art and Design Stuttgart
Stuttgart, Germany
andrea.fischer@abk-stuttgart.de
is found in the contact zone between glass and metal.
Glass hydrolysis forms alkaline liquid films which
induce corrosion on neighbouring metals as research
at the Stuttgart State Academy of Art and Design has
shown. Special products are formed, such as sodium
copper formate (Cu4Na4O(HCOO)8(OH)2·4H2O). A
survey at the Swiss National Museum Zurich in 2012
proved that this corrosion is not rare and occurs on
many combined glass/metal objects.
• Non-ferrous frames are often covered with green
metal soaps, such as copper and zinc stearate or oleate,
particularly where they have had contact with the
wearer’s head or have been oiled.
• Varied and typical degradation phenomena of cellu-
lose nitrate and acetate are very obvious with strongly
corroding metal parts, e.g. commonly basic copper
nitrate (Cu2(NO)3(OH)3). Acidic fumes from degra-
dation also damage other nearby stored objects such
as iron frames, generating a rust film which catalyses
the degradation of plastics and other materials.
A well-planned storage concept with preventive measures
is therefore essential in the preservation of spectacle collec-
tions. The main focus lies in reducing the catalysing effect
of various degradation products in material combinations.
486  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Author index
A Bynon, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . 482
Abu-Baker, A.N.  .. . . . . . . . . . . . . . . . . 202
Alexander, J.  . . . . . . . . . . . . . . . . . . . . . . . 455
Andraud, C.  .. . . . . . . . . . . . . . . . . . . . . . . 414
Andrey, L.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Anghelone, M.  . . . . . . . . . . . . . 363, 433
Agnoletti, S.  . . . . . . . . . . . . . . . . . . . . . . . . 271
Ahmadi, A.  .. . . . . . . . . . . . . . . . . . . . . . . . 148
Albini, M.  .. . . . . . . . . . . . . 240, 457, 468
Apchain, E.  . . . . . . . . . . . . . . . . . . . . . . . . . 210
Argyropoulos, V.  . . . . . . . . . . . . . . . . . 352
Azéma, A.  . . . . . . . . . . . . . . . . . . . 414, 454
Aubert, J.G.  . . . . . . . . . . . . . . . . . . . . . . . . 327
Aufray, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . 220
B Chabas, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Balbo, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Balliana, E.  . . . . . . . . . . . . . . . . . . . . . . . . . 440
Barack, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
Baranowska-Fietkiewicz, A.  .. 295
Basile, S.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Battault, J.-C.  . . . . . . . . . . . . . . . . . . . . . . . . 49
Bayle, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Beentjes, T.  . . . . . . . . . . . . . . . . . . 166, 439
Benetti, F.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Bellot-Gurlet, L.  . . . . . . . . . . . . . 58, 367
Bernardi, E.  . . . . . . . . . . . . . . . . . 220, 450
Bertholon, R.  . . . . . . . . . . . . . . . . . . . . . . 466
Betinelli, P.  . . . . . . . . . . . . . . . . . . . . . . . . . 414
Bette, S.  .. . . . . . . . . . . . . . . . . . . . . . 125, 156
Beuret, E.D.  . . . . . . . . . . . . . . . . . . . . . . . . 473
Biemond, D.J.  .. . . . . . . . . . . . . . . . . . . . . . . 33
Bignozzi, M.C.  . . . . . . . . . . . . . . . . . . . . 220
Bloser, J.  . . . . . . . . . . . . . . . . . . . . . . 139, 166
Borrós, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 460
Boust, C.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Brambilla, L.  . . . . . . . . . . . 58, 444, 466
Brechbühl, S.  . . . . . . . . . . . . . . . . . . . . . . 461
Breitung, E.  . . . . . . . . . . . . . . . . . . . . . . . . 472
Brini, A.   . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Brunet, M.  . . . . . . . . . . . . . . . . . . . 414, 423
Bruni, T.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Buecker, M.  . . . . . . . . . . . . . . . . . . . . . . . . 467
C Dillmann, P.  .. . . . . . . . . . . . . . . . . . 49, 210
Cagnini, A.  . . . . . . . . . . . . . . . . . . 259, 271
Cama-Villafranca, J.  . . . . . . . . . . . . . 320
Camoli, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Campos, P.H.O.V.  . . . . . . . . . . . . . . . . 447
Cano, E.  . . . . . . . . . . . . . . . . . . . . . . . . 83, 312
Car, R.  . . . . . . . . . . . . . . . . . . . . . . . . 384, 465
Carneiro, T.C.  . . . . . . . . . . . . . . . . . . . . . 447
Carò, F.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
Carré, A.-L.  . . . . . . . . . . . . . . . . . . . . . . . . 367
Carrieri, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 458
Catillon, R.  . . . . . . . . . . . . . . . . . . . . . . . . . 367
Cevey, C.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Chalançon, B.  . . . . . . . . . . . . . . . . . . . . . 444
Chemello, C.  . . . . . . . . . . . . . . . . 312, 482
Chiavari, C.  . . . . . . . . . . . 220, 435, 450
Cigić, I.K.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
Cipolato, A.  . . . . . . . . . . . . . . . . . . . . . . . . 440
Clarke, C.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Comensoli, L.  . . . . . . . . . . . . . . . . . . . . . 240
Contreras-Vargas, J.  . . . . . . . . . . . . . 320
Cornet, E.  .. . . . . . . . . . . . . . . . . . . . . . . . . . 444
Cotrebil, Y.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Couto, A.A.  . . . . . . . . . . . . . . . . . . . . . . . . 447
Creagh, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Creange, S.  . . . . . . . . . . . . . . . . . . . . . . . . . 472
Crepeau, M.  .. . . . . . . . . . . . . . . . . . . . . . . . . 76
Crespo, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 312
Cure, A.P.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
Curkovic, H.O.  .. . . . . . . . . . . . . . . . . . . 456
D Gessler, T.  .. . . . . . . . . . . . . . . . . . . . . . . . . . 480
Dajnowski, B.  .. . . . . . . . . . . . . . . . . . . . . 295
Daubaras, M.  . . . . . . . . . . . . . . . . . . . . . . 476
Davidowitz, T.  .. . . . . . . 108, 174, 472
Davidson, F.  .. . . . . . . . . . . . . . . . . . . . . . . 132
Davies, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376
de la Fuente, D.  . . . . . . . . . . . . . . . . . . . 450
Degrigny, C.  . . . 259, 406, 445, 470
del Hoyo-Melendez, J.  . . . . . . . . . . 437
de Lapérouse, J.-F.  .. . . . . . . . . . . . . . . 304
de Viviés, P.  . . . . . . . . . . . . . . . . . . . . . . . . 210
Di Carlo, G.  . . . . . . . . . . . . . . . . . . . . . . . . 468
Dinnebier, R.E.  . . . . . . . . . . . . . . . . . . . 156
Dittus, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
Drieux-Daguerre, M.  . . . . . . . . . . . 423
Durier, M.G.  . . . . . . . . . . . . . . . . . . . . . . . . . 49
Duvauchelle, A.  . . . . . . . . . . . . . . . . . . . . 68
E
Eggert, G.  . . . . . . . . . . . . . . . . . . . . 125, 156
Elmarazky, A.  . . . . . . . . . . . . . . . . . . . . . 291
Emmerson, N.J.  . . . . . . . . . . . . . . 76, 279
Erdmann, R.G.  . . . . . . . . . . . . . . . 33, 108
Ervas, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 458
Esvan, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
F
Fabjan, E.Š.  . . . . . . . . . . . . . . . . . . . . . . . . . 220
Faifer, M.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Feldmann, I.  . . . . . . . . . . . . . . . . . . . . . . . 467
Ferrari, E.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Fischer, A.  . . . . . . . . . . . . . . . . . . . 156, 483
Florescu, M.  .. . . . . . . . . . . . . . . . . . . . . . . 414
Fujisawa, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 16
Funke, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 462
G
Galeotti, M.  . . . . . . . . . . . . . . . . . . . . . . . . 271
Ganiaris, H.  . . . . . . . . . . . . . . . . . . . . . . . . 452
Gartner, N.  . . . . . . . . . . . . . . . . . . . . . . . . . 220
Garšvienė, S.  . . . . . . . . . . . . . . . . . . . . . . . 479
Geffroy, A.-M.  . . . . . . . . . . . . . . . . . . . . 367
Genachte-Le Bail, A.  . . . . . 367, 414
Gerber, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 453
Gillioz, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Giménez-Barrera, P.  . . . . . . . . . . . . . 460
Giuliani, C.  . . . . . . . . . . . . . . . . . . . . . . . . . 468
Godfrey, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . 438
Goidanich, S.  . . . . . . . . . . . . . . . . . . . . . . 118
Gomez, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 471
González-Pereyra, N.  . . . . . . . . . . . 336
Gómez Núñez, M.F.  . . . . . . . . . . . . . . . 45
Gordon, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . 367

Goryl, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Grassi, V.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Grayburn, R.  .. . . . . . . . . . . . . . . . . . . . . . 100
Guillot, I.  . . . . . . . . . . . . . . . . . . . . . . 49, 451
Guilminot, E.  . . . . . . . . . . . . . . . 327, 471
Gutknecht, N.  . . . . . . . . . . . . . . . . . . . . . 255
H Krist, G.  . . . . . . . . . . . . . . . . . . . . . . 363, 433
Haddad, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 139
Hallam, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Hatahet, R.  . . . . . . . . . . . . . . . . . . . . . . . . . 469
Heidarpour, B.  . . . . . . . . . . . . . . . . . . . . 148
Hemuki, N.  . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Hietz, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
Hoffman, W.  . . . . . . . . . . . . . . . . 263, 443
Holzleitner, M.  . . . . . . . . . . . . . . . . . . . . 363
Hook, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
Houssay, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 49
Hovind, M.  . . . . . . . . . . . . . . . . . . . . . . . . . 445
Howard, D.L.  . . . . . . . . . . . . . . . . . . . . . . 108
Huber, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483
Hubert, V.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Huller, I.M.  .. . . . . . . . . . . . . . . . . . . . . . . . 433
Hunter, H.M.A.  . . . . . . . . . . . . . . . . . . . 442
Huynh, C.  . . . . . . . . . . . . . . . . . . . . . . . . . . 287
I Lukšėnienė, V.  .. . . . . . . . . . . . . . . . . . . . 479
Ikeda, Y.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Ingo, G.M.  . . . . . . . . . . . . . . . . . . . . . . . . . . 468
J Maltby, S.L.  .. . . . . . . . . . . . . . . . . . . . . . . . 376
Jacob, H.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432
Jeran, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
Jiménez Cosme, I.K.  . . . . . . . . . . . . . . 45
Joiret, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 454
Jones, C.G.  . . . . . . . . . . . . . . . . . . . . . . . . . 442
Joosten, I.  . . . . . . . . . . . . . . . 41, 166, 439
Joseph, E.  . . . . . . . . . . . . . . 240, 255, 457
Josse, C.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Jovanovic, D.  . . . . . . . . . . . . . . . . . . . . . . 304
Junier, P.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
K Meijer, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Kasthuri, K.A.A.  .. . . . . . . . . . . . . . . . . . . 25
Kharbade, B.V.  . . . . . . . . . . . . . . . . . . . . 441
King, L.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Kirchhoffer, S.  . . . . . . . . . . . . . . . . . . . . . 481
Kirchhoffer, Z.  . . . . . . . . . . . . . . . . . . . . 481
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  487
Kohdzuma, Y.  . . . . . . . . . . . . . . . . . . . . . 183
Kooli, W.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Kosec, T.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Kouril, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . 446
Krajnc, H.P.  . . . . . . . . . . . . . . . . . . . . . . . . 449
Kreislova, K.  . . . . . . . . . . . . . . . . . . . . . . . 446
Krieg, M.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Kuiter, R.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Kupershmidt, H.  . . . . . . . . . . . . . . . . . 459
L N
Labut, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Lagger, F.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
Laik, B.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Lal, G.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
Lang, R.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452
Lavriha, D.  . . . . . . . . . . . . . . . . . . . . . . . . . 470
Lazar, T.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
Lehmann, E.H.  .. . . . . . . . . . . . . . . . . . . . . 68
Lenhart, E.  . . . . . . . . . . . . . . . . . . . . . . . . . 363
Leroux, M.  . . . . . . . . . . . . . . . . . . . . . . . . . 471
Letardi, P.  . . . . . . . . . . . . . . . . . 83, 92, 457
Lira, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436
Loeper-Attia, M.-A.  . . . . . . 367, 397
Lopez, D.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . 454
M Paris, C.  . . . . . . . . . . . . . . . . . . . . . . . . 58, 367
MacLeod, I.D.  . . . . . . . . 108, 384, 465
Mannes, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Mardikian, P.  . . . . . . . . 312, 336, 344,
352, 482
Mariani, B.  . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Martini, C.  . . . . . . . . . . . 220, 435, 450
Martins, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 139
Masi, G.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Matsuda, K.  . . . . . . . . . . . . . . . . . . . . . . . . 183
McEnroe, K.  . . . . . . . . . . . . . . . . . . . . . . . 480
McGath, M.K.  . . . . . . . . . . . . . . . . . . . . . 443
McGlinchey, C.  . . . . . . . . . . . . . . . . . . . 139
Mélard, N.  . . . . . . . . . . . . . . . . . . . 327, 414
Memet, J.-B.  .. . . . . . . . . . . . . . . . . . . . . . . 210
Menart, E.  . . . . . . . . . . . . . . . . . . . 449, 470
Messina, E.  . . . . . . . . . . . . . . . . . . . . . . . . . 468
Michel, A.  . . . . . . . . . . . . . . . . . . . . . 58, 466
Michelin, A.  .. . . . . . . . . . . . . . . . . . . . . . . 414
Mikic, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
Mirabaud, S.  . . . . . . . . . . . . . . . . . . . . . . . . . 58
Mistewicz, A.  . . . . . . . . . . . . . . . . . . . . . . 448
Mohamed, W.A.  . . . . . . . . . . . . . . . . . . 391
Monachon, M.  . . . . . . . . . . . . . 240, 457
Monticelli, C.  . . . . . . . . . . . . . . . . . . . . . . 220
Morcillo, M.  . . . . . . . . . . . . . . . . . . . . . . . 450
Morigi, L.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
Msallamova, S.  . . . . . . . . . . . . . . . . . . . . 446
Naidorf, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . 477
Nagar, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
Najorka, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . 442
Näsänen, L.  . . . . . . . . . . . . . . . . . . . . . . . . 336
Naseri, R.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Neff, D.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Nemeček, N.  . . . . . . . . . . . . . . . . 449, 470
Nobili, L.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 450
Normand, L.  . . . . . . . . . . . . . . . . . . . . . . . 367
O
Ortolani, C.  . . . . . . . . . . . . . . . . . . . . . . . . 259
Oudbashi, O.  . . . . . . . . . . . . . . . . . . . . . . 148
P
Page, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Papanikou, A.-A.  . . . . . . . . . . . . . . . . . 352
Pascucci, M.  . . . . . . . . . . . . . . . . . . . . . . . 468
Paterakis, A.B.  .. . . . . . . . . . . . . . . . . . . . 291
Patterson, C.  . . . . . . . . . . . . . . . . . . . . . . . 312
Peñuelas, G.  . . . . . . . . . . . . . . . . . . . . . . . . 436
Petiti, C.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Petitmangin, A.  . . . . . . . . . . . . . . . . . . . 451
Pires, R.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Plé, E.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
Porcinai, S.  . . . . . . . . . . . . . . . . . . . . . . . . . 271
Porteous, G.  .. . . . . . . . . . . . . . . . . . . . . . . 462
R
Raedel, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . 467
Raimon, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 471
Rajpoot, D.  . . . . . . . . . . . . . . . . . . . . . . . . . 441
Ramírez Barat, B.  .. . . . . . . . . . . . . . . . . 83
Rapp, G.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 444
Reimann, M.  .. . . . . . . . . . . . . . . . . . . . . . 467
488  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
Reguer, S.  . . . . . . . . . . . . . . . . . . . . 367, 454
Revay, Z.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Riccucci, C.  . . . . . . . . . . . . . . . . . . . . . . . . 468
Ricotta, N.  . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Rifai, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Rivera-Diaz, J.  . . . . . . . . . . . . . . 336, 462
Rizzo, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
Rizzutto, M. de A.  .. . . . . . . . . . . . . . . 447
Robbiola, L.  . . . . . . . . . . . . . . . . . 220, 423
Robcis, D.  .. . . . . . . . . . . . . . . . . . . 287, 434
Roda-Buch, A.  . . . . . . . . . . . . . . . . . . . . 444
Rossetti, L.  .. . . . . . . . . . . . . . . . . . . . . . . . . 327
Russo, S.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
Russo dos Reis, T.  . . . . . . . . . . . . . . . . 447
Ruvalcaba Sil, J.L.  . . . . . . . . . . . . . . . . . . 45
Růžička, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . 478
S van Bommel, M.  .. . . . . . . . . . . . . . . . . 439
Sadat, F.E.-Z.  .. . . . . . . . . . . . . . . . . . . . . . 391
Saheb, M.  . . . . . . . . . . . . . . . . . . . . 367, 451
Salvadori, B.  . . . . . . . . . . . . . . . . . . . . . . . . 92
Sangouard, E.  .. . . . . . . . . . . . . . . . . . . . . 464
Santagostino Barbone, A.  .. . . . . 271
Saunders, J.  . . . . . . . . . . . . . . . . . . . . . . . . . 432
Savelli, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Scafuri, M.  .. . . . . . . . . . . . . . . . . . . . . . . . . 336
Schaefer, K.  .. . . . . . . . . . . . . . . . . . . . . . . . 295
Schmidt, K.  . . . . . . . . . . . . . . . . . . . . . . . . 433
Schmutzler, B.  . . . . . . . . . . . . . . . . . . . . . 248
Schüch, M.  . . . . . . . . . . . . . . . . . . . . . . . . . 463
Schröter, J.  . . . . . . . . . . . . . . . . . . . . . 58, 406
Schwebler, J.  .. . . . . . . . . . . . . . . . . . . . . . . 475
Sciau, P.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
Seine, G.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Servais, G.R.  . . . . . . . . . . . . . . . . . . . . . . . 473
Singh, M.R.  . . . . . . . . . . . . . . . . . . . . . . . . 196
Škrlep, L.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Southwick, C.  .. . . . . . . . . . . . . . . . . . . . . 174
Sperotto, W.  . . . . . . . . . . . . . . . . . . . . . . . . 220
Spohn, R.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 464
Staccioli, M.P.  . . . . . . . . . . . . . . . . . . . . . 468
Stanley, B.  .. . . . . . . . . . . . . . . . . . . . . . . . . . 230
Stefanczyk, K.  . . . . . . . . . . . . . . . . . . . . . 437
Stelzner, J.  .. . . . . . . . . . . . . . . . . . . 125, 156
Stephant, N.  .. . . . . . . . . . . . . . . . . . . . . . . 327
Stepien, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 437
Strachotova, K.C.  .. . . . . . . . . . . . . . . . 446
Stieghorst, C.  . . . . . . . . . . . . . . . . . . . . . . 248
T
Tauber, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Tedesco, E.  . . . . . . . . . . . . . . . . . . . . . . . . . 220
Texier, A.  .. . . . . . . . . . . . . . . . . . . . 414, 454
Thaler, F.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
Thickett, D.  . . . . . . . . . . . . . . . . . 230, 442
Thunberg, J.C.  . . . . . . . . . . . . . . . . . . . . 279
Thurrowgood, D.  .. . . . . . . . . . . . . . . . 108
Toro, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
U
Urban, F.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
V
Vaiedelich, S.  . . . . . . . . . . . . . . . . . . . . . . . . 49
van Bellegem, M.  . . . . . . . . . . . . . . . . . 477
van Bennekom, J.  .. . . . . . . . . . . . . . . . 174
van Bork, E.  . . . . . . . . . . . . . . 33, 41, 174
van Lookeren Campagne, K.  . 174
van der Stok, J.  . . . . . . . . . . . . . . . . . . . . 439
Vasiliauskaite, A.  . . . . . . . . . . . . . . . . . 476
Vassura, I.  . . . . . . . . . . . . . . . . . . . . . . . . . . 450
Vecten, A.  . . . . . . . . . . . . . . . . . . . . . . . . . . 466
Vega, E.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Verma, P.  .. . . . . . . . . . . . . . . . . . . . . . . . . . . 441
Villiaume, L.  . . . . . . . . . . . . . . . . . . . . . . . . . 49
Vion, M.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
W
Wagner, A.  . . . . . . . . . . . . . . . . . . . . . . . . . 100
Wakiya, S.  . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Walthew, J.  . . . . . . . . . . . . . . . . . . . . . . . . . 438
Watkinson, D.  . . . . . . . . . . . . . . . . 76, 279
Wesolowska, C.V.  . . . . . . . . . . . . . . . . 295
Wiinblad, T.  .. . . . . . . . . . . . . . . . . . . . . . . 359
Willett, P. †  . . . . . . . . . . . . . . . . . . . . . . . . . 432
Y
Yanagida, A.  . . . . . . . . . . . . . . . . . . . . . . . 183
Z
Zanoni, M.  . . . . . . . . . . . . . . . . . . . . . . . . . 118
Zanotto, F.  . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Zdenek Bartak, Z.  . . . . . . . . . . . . . . . . 446
Zendri, E.  . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
Ziegler, J.  . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Zycherman, L.  .. . . . . . . . . . . . . 139, 166

Keyword index
Symbols
19th-century weaponry. . . . . . . . . 130
A
acetic acid. . . . . . . . . . . . . . . . . . . . 196, 291
aeronautical materials. . . . . . . . . . . 423
alkaline Rochelle salts. . . . . . . . . . . 344
alteration. . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
aluminum. . . . . . . . . . . . . . . . . . . . . 137, 406
aluminum alloys. . . . . . . . . . . 414, 423
ammonium hydroxide. . . . . . . . . . 344
ammonium hydroxide solution
.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352
analysis.. . . . . . . . . . . . . . . . . . . . . . . 363, 406
archaeological bronze. . . . . . . . . . . 148
archaeological iron. . . . . . . . . . 74, 248
archaeological iron-bone
composites. . . . . . . . . . . . . . . . . . . . . . . 359
archaeological metal. . . . . . . . . . . . . . . 39
archaeological munitions.. . . . . . 130
archaeology. . . . . . . . . . . . . . . . . . . . . . . . . 304
atmospheric corrosion. . . 210, 220
augmentation. . . . . . . . . . . . . . . . . . . . . . . . 31
automated recognition. . . . . . . . . . . . 31
Aventicum. . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
B De Vlamingh. . . . . . . . . . . . . . . . . . . . . . . 106
basic zinc copper formate. . . . . . 156
benzotriazole. . . . . . . . . . . . . . . . . . . . . . . 196
biocleaning. . . . . . . . . . . . . . . . . . . . . . . . . 240
biodegradation. . . . . . . . . . . . . . . . . . . . 384
biogenic minerals. . . . . . . . . . . . . . . . . 240
bluing.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
brass. . . . . . . . . . . . . . . . . . . . . . 66, 156, 397
bronze. . . . . . . . . . . . . 23, 202, 220, 255
Buddha.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
burial environment.. . . . . . . . . . . . . . 148
Butvar. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
C eddy current. . . . . . . . . . . . . . . . . . . . . . . . . . 56
casting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
cellulose nitrate.. . . . . . . . . . . . . . . . . . . 123
chloride-induced corrosion. . . 255
chloride removal. . . . . . . . . . . 344, 352
chrome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  489
cleaning. . . . . . . . . . . . . . . . . . . . . . . . . 23, 271
coating process. . . . . . . . . . . . . . . . . . . . 220
coatings. . . . . . . . . . . . . . . . . . . . 81, 90, 137
collaboration. . . . . . . . . . . . . . . . . . . . . . . 376
composite objects. . . . . . . . . . . . . . . . . 344
confocal microscopy. . . . . . . . . . . . . . . 56
conservation.. . 202, 304, 312, 336
conservation materials. . . . . . . . . . 123
conservation-restoration. . . . . . . 327
conservation strategies. . . . . . . . . . 414
consultation. . . . . . . . . . . . . . . . . . . . . . . . 376
convolutional neural network. . 31
copper. . . . . . . . . . . . . . . . . . 116, 210, 295
copper alloys. . . . . . 56, 90, 263, 304
corrosion. . . 14, 74, 183, 202, 240,
271, 279, 304, 327, 384, 406
corrosion inhibitor. . . . . . . . . . . . . . . 210
corrosion layers. . . . . . . . . . . . . . . . . . . 116
Corten steel. . . . . . . . . . . . . . . . . . . . . . . . . 312
critical judgment. . . . . . . . . . . . . . . . . . 320
cultural heritage. . . . . . . . . . . . . . . . . . . 240
cyanide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
D Fourier transform infrared
database.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
decision-making tools.. . . . . . . . . . 414
deconcretion. . . . . . . . . . . . . . . . . . . . . . . 336
decorative arts. . . . . . . . . . . . . . . . . . . . . . . 98
desalination.. . . . . . . . . . . . . . . . . . . . . . . . 248
diagnosis. . . . . . . . . . . . . . . . . . . . . . . 14, 406
display recommendations. . . . . . 137
dithionite reduction.. . . . . . . . . . . . . 384
documentation. . . . . . . . . . . . . . . . . . . . 406
double-sided gilding. . . . . . . . . . . . . 391
dry-ice blasting. . . . . . . . . . . . . 263, 271
E glass. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
El Anatsui. . . . . . . . . . . . . . . . . . . . . . . . . . . 137
El Caballito.. . . . . . . . . . . . . . . . . . . . . . . . . 320
electrochemical impedance
spectroscopy (EIS). . . . . . . . 81, 312
electrochemical techniques. . . . 220
electrolytic cleaning. . . . . . . . . . . . . . 259
electroplate. . . . . . . . . . . . . . . . . . . . . . . . . . . 98
enamel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
enrichment. . . . . . . . . . . . . . . . . . . . . . . . . 391
environmental corrosivity. . . . . 116
environmentally friendly
chemicals. . . . . . . . . . . . . . . . . . . . . . . . . 191
epoxy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
ER sensors. . . . . . . . . . . . . . . . . . . . . . . . . . 116
ethylenediaminetetraacetic acid
(EDTA).. . . . . . . . . . . . . . . . . . . . . . . . . . . 363
exhibition. . . . . . . . . . . . . . . . . . . . . . . . . . . 397
experimental setup. . . . . . . . . . . . . . . . . 81
F
ferrous metal corrosion. . . . . . . . . 191
filling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
flame cleaning. . . . . . . . . . . . . . . . . . . . . 279
focused ion beam/field emission
gun-scanning electron
microscopy (FIB/FEG-SEM)
.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
formic acid.. . . . . . . . . . . . . . . . . . . . . . . . . 196
microspectroscopy (FTIR).. 304
Fourier transform infrared
spectroscopy (FTIR). . . . . 56, 359
Frigilene. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
fused plate.. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
G
gap filler. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
gilded bronze.. . . . . . . . . . . . . . . . . . . . . . 271
gilded copper. . . . . . . . . . . . . . . . . . . . . . 363
gilding. . . . . . . . . . . . . . . . . . . . . . . . . 287, 295
gilt silver. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
GIMME.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
glass-induced metal corrosion
.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
gold leaf. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
graphite. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
gray cast iron. . . . . . . . . . . . . . . . . . . . . . . 263
green chemistry. . . . . . . . . . . . . . . . . . . 240
490  ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND
H marine silver. . . . . . . . . . . . . . . . . . . . . . . . . 39
hallmarks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Hartog Plate. . . . . . . . . . . . . . . . . . . . . . . . 106
historically informed music. . . . . 47
H.L. Hunley. . . . . . . . . . . . . . . . . . . . . . . . . 336
HMG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
hollow-casting. . . . . . . . . . . . . . . . . . . . . . . 23
hydrochloric acid. . . . . . . . . . . . . . . . . 196
I metallurgy. . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
imaging techniques. . . . . . . . . . . . . . 327
Indian. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
indoor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
industrial heritage. . . . . . . . . . . . . . . . 423
Industrial Revolution. . . . . . . . . . . . . . 47
inhibitor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
in-situ preservation. . . . . . . . . . . . . . 183
iron. . . . . . . . . . . . . . . . . . . . . . . . 74, 240, 279
iron artifact. . . . . . . . . . . . . . . . . . . . . . . . . 183
iron-wood.. . . . . . . . . . . . . . . . . . . . . . . . . . 352
J multi-analytical approach. . . . . . 414
Japanese Institute of Anatolian
Archaeology. . . . . . . . . . . . . . . . . . . . . 291
K
Kamakura. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Khon dance. . . . . . . . . . . . . . . . . . . . . . . . . 384
L . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
large-scale marine objects.. . . . . 336
laser. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
laser ablation-inductively coupled
plasma-mass spectrometry
(LA-ICP-MS). . . . . . . . . . . . . . . . . . . . . . 66
lead. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166, 196
lead dioxide.. . . . . . . . . . . . . . . . . . . . . . . . 166
lead tablets. . . . . . . . . . . . . . . . . . . . . . . . . . 327
loss compensation. . . . . . . . . . . . . . . . 287
lost-wax. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66, 191
machine learning. . . . . . . . . . . . . . . . . . . 31
Maillol.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
malachite. . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
marine archaeological
conservation. . . . . . . . . . . . . . . . . . . . . 263
marine archaeology. . . . . . . . . . . . . . 344
marine composite artifacts.. . . . 352
mashru. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
medieval swords. . . . . . . . . . . . . . . . . . 352
metal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81, 367
metal artworks.. . . . . . . . . . . . . . . . . . . . 271
metal cleaning. . . . . . . . . . . . . . . . . . . . . . . 98
metal conservation. . . . . . . . . . . . . . . 174
metal corrosion.. . . . . . . . . . . . . . . . . . . 156
metal finish. . . . . . . . . . . . . . . . . . . . . . . . . 320
metal thread. . . . . . . 39, 43, 384, 391
methodology. . . . . . . . . . . . . . . . . . . . . . . 414
micro-abrasive blasting. . . . . . . . . . . 74
microballoon. . . . . . . . . . . . . . . . . . . . . . . 291
microbiologically influenced
corrosion. . . . . . . . . . . . . . . . . . . . . . . . . 183
microcrystalline waxes. . . . . . . . . . . . 90
microorganisms. . . . . . . . . . . . . . . . . . . 240
monitoring.. . . . . . . . . . . . . . . . . . . . . . . . . 116
monumental sculpture. . . . . . . . . . 312
multiscale analysis. . . . . . . . . . 210, 423
N
National Museums Scotland.. 191
natural patina. . . . . . . . . . . . . . . . . . . . . . 210
NdT measurements. . . . . . . . . . . . . . . . 90
neutron and x-ray tomography
nickel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
nitrile gloves.. . . . . . . . . . . . . . . . . . . . . . . 174
non-destructive analysis. . . . . . . . . . 98
non-vacuum freeze-drying. . . . 344
O Santa Vittoria. . . . . . . . . . . . . . . . . . . . . . 259
occupational hazard tests.. . . . . . 220
Oddy test. . . . . . . . . . . . . . . . . . . . . 123, 174
organic coatings. . . . . . . . . . . . . . . . . . . 220
outdoor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
outdoor bronze. . . . . . . . . . . . . . . . . . . . . . 14
outdoor sculpture.. . . . . . . . . . . . . . . . 166
oxime. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
oxygen consumption
.. . . . . . . . . . . . . . . . . . . . . . . . 248, 279, 359
P silver. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31, 123
paint. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
Paraloid B-44.. . . . . . . . . . . . . . . . . . . . . . 291
patina. . . . . . . . . . . . . . . . . . . . . 81, 312, 320
patination. . . . . . . . . . . . . . . . . . . . . . . . . . . 271
piano strings.. . . . . . . . . . . . . . . . . . . . . . . . . 47
plastic thread. . . . . . . . . . . . . . . . . . . . . . . 397
plated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
polarisation resistance.. . . . . . . . . . 196
polypropylene.. . . . . . . . . . . . . . . . . . . . . 397
potential-controlled
electrochemical rinsing. . . . . . 359
practitioner. . . . . . . . . . . . . . . . . . . . . . . . . 279
primers.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
prompt gamma activation
analysis (PGAA).. . . . . . . . . . . . . . . 248
propellants. . . . . . . . . . . . . . . . . . . . . . . . . . 130
Q
Queen Sirikit. . . . . . . . . . . . . . . . . . . . . . . 384
R
radiography.. . . . . . . . . . . . . . . . . . . . . . . . . . 74
Raman spectroscopy. . . . . . . 66, 304,
312, 367
reburial.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
relative humidity. . . . . . . . . . . . . . . . . . 291
reliquary bust. . . . . . . . . . . . . . . . . . . . . . 259
Renaissance Wax. . . . . . . . . . . . . . . . . . 230
residual chloride. . . . . . . . . . . . . . . . . . 248
restoration. . . . . . . . . . . . . . . . . . . . . 23, 367
rigid gel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363
risk management.. . . . . . . . . . . . . . . . . 376
Roman archaeology. . . . . . . . . . . . . . . . 66
S
Saint George.. . . . . . . . . . . . . . . . . . . . . . . 295
scanning electron microscopy
(SEM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
scanning electron microscopy
with energy-dispersive x-ray
spectroscopy (SEM-EDS)
scanning electron microscopy-
backscattered electron imaging
(SEM-BEI). . . . . . . . . . . . . . . . . . . . . . . . . 74
Sicily. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
silver plate.. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
sodium hydroxide solution. . . . 248
 ICOM-CC  |  METAL 2019  |  NEUCHÂTEL, SWITZERLAND  491

sodium sesquicarbonate. . . . . . . . 359 X

soil analysis. . . . . . . . . . . . . . . . . . . . . . . . . 148 x-ray absorption near-edge
solvents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 structure (XANES). . . . . . . . . . . . 367
spectroscopic ellipsometry.. . . . . . 56 x-ray diffraction.. . . . . . . . . . . . . . . . . . . . 14
Sri Lanka. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 x-ray fluorescence (XRF).. 14, 304
stabilization.. . . . . . . . . . . . . . . . . . . 23, 240 x-ray fluorescence (XRF)
stabilization treatment. . . . . . . . . . 255 microscopy . . . . . . . . . . . . . . . . . . . . . . 106
statue. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 x-ray fluorescence (XRF)
steam locomotive. . . . . . . . . . . . . . . . . 376 spectroscopy . . . . . . . . . . . . . . 295, 367
steel. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 x-ray photoelectron (XPS)
surface appearance. . . . . . . . . . . . . . . 414 spectroscopy. . . . . . . . . . . . . . . . . . . . . 220
surface finishes. . . . . . . . . . . . . . . . . . . . 414
surface layer. . . . . . . . . . . . . . . . . . . . . . . . 320 Y
surface treatments. . . . . . . . . . . . . . . . 414 Yemen. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
sustainability. . . . . . . . . . . . . . . . . 174, 255
Z
sweat. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
zinc formate. . . . . . . . . . . . . . . . . . . . . . . . 156
Swiss collections. . . . . . . . . . . . . . . . . . 406
synchrotron.. . . . . . . . . . . . . . . . . . . . . . . . 106

T
tarnished silver. . . . . . . . . . . . . . . . . . . . 259
tarnishing. . . . . . . . . . . . . . . . . . . . . 174, 391
textiles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
Theophilus Presbyter. . . . . . . . . . . . . . 43
the Pleco. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
tinned surface.. . . . . . . . . . . . . . . . . . . . . 255
tinning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
treatment. . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
triple expansion steam engine. 191
tri-sodium citrate. . . . . . . . . . . . . . . . . 344
type I corrosion. . . . . . . . . . . . . . . . . . . 148

U
underwater archaeology. . . . . . . . 336
underwater site. . . . . . . . . . . . . . . . . . . . 183
unexploded ordnance. . . . . . . . . . . 130
UV-induced fluorescence. . . . . . . . 56

V
Vannoccio Biringuccio. . . . . . . . . . . . 43
varnishes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
visual analysis. . . . . . . . . . . . . . . . . . . . . . . . 98
visualization. . . . . . . . . . . . . . . . . . . . . . . . 106

W
weathering steel. . . . . . . . . . . . . . . . . . . 312
wrought iron. . . . . . . . . . . . . . . . 263, 279
The Metals Working Group is part of the
International Council of Museums – Committee
for Conservation (ICOM–CC).
ISBN 978-92-9012-458-0