Tomography

David J.Larson Ty J.
Prosa
Robert M.Ulfig Brian P.Geiser
Thomas F.Kelly
Local Electrode
Atom Probe
Tomography
A Users Guide
Foreword by
Professor Sir Colin J. Humphreys
Local Electrode Atom Probe Tomography
David J. Larson Ty J. Prosa Robert M. Ulfig
Brian P. Geiser Thomas F. Kelly
Local Electrode Atom

Probe Tomography
A Users Guide
Foreword by
Professor Sir Colin J. Humphreys, CBE, FRS, FREng.
David J. Larson Ty J. Prosa
CAMECA Instruments, Inc. CAMECA Instruments, Inc.
Madison, WI, USA Madison, WI, USA
Robert M. Ulfig Brian P. Geiser

CAMECA Instruments, Inc. CAMECA Instruments, Inc.
Madison, WI, USA Madison, WI, USA
Thomas F. Kelly
CAMECA Instruments, Inc.
Madison, WI, USA
ISBN 978-1-4614-8720-3 ISBN 978-1-4614-8721-0 (eBook)

DOI 10.1007/978-1-4614-8721-0
Springer New York Heidelberg Dordrecht London
Library of Congress Control Number: 2013953879
Springer Science+Business Media New York 2013

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief excerpts
in connection with reviews or scholarly analysis or material supplied specifically for the purpose of being
entered and executed on a computer system, for exclusive use by the purchaser of the work. Duplication
of this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publishers location, in its current version, and permission for use must always be obtained from
Springer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center.
Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Preface
Over the years a number of atom probe books have been written by various authors
[110], so why write another one you might ask. We believe that this book is unique
in specifically targeting atom probe adopters who are new to the technique. The
book introduces new users to the process of performing all of the aspects of a Local
Electrode Atom Probe experiment. It includes the fundamentals of preparing
specimens for the microscope from a variety of materials, details of the instrumen-
tation used in data collection, parameters under which optimal data are collected,
current methods of data reconstruction, and selected methods of data analysis. In
addition, certain topics are explained specifically from a user perspective and
include details that are often learned only through trial and error, allowing users
to succeed more quickly in the challenging areas of specimen preparation and data
collection.
This book is meant to be a useful reference for the conventional wisdom type
of information that is not always found in academic books and is usually gained
only through experience. It is not meant to be a comprehensive treatment of atom
probe tomography but rather an everyday reference for data collection on the local
electrode atom probe and for the specimen preparation and data analysis that go
along with such experiments. For the most part, we have dealt with more advanced
topics, such as the details of the spatial reconstruction equations, by including the
information in appendices or by simply referring the reader to other textbooks or
journal articles. In this way we hope to have produced a very usable reference for
both novice users and experienced scientists. The future of atom probe tomography
is bright, and we hope that the path to adoption will be clearer with the availability
of this book.
Madison, WI, USA David J. Larson

Ty J. Prosa
Robert M. Ulfig
Brian P. Geiser
Thomas F. Kelly
v
vi Preface
References
1. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principles and Applications. Elsevier,
New York (1969)
2. Hren, J.J., Ranganathan, S. (eds.): Field-Ion Microscopy. Plenum Press, New York (1968)
3. Bowkett, K.M., Smith, D.A.: Field-Ion Microscopy. North-Holland, Amsterdam (1970)
4. Muller, E.W., Tsong, T.T.: Field Ion Microscopy, Field Ionization and Field Evaporation, vol.
4. Progress in Surface Science (1973)
5. Miller, M.K., Smith, G.D.W.: Atom Probe Microanalysis: Principles and Applications to
Materials Problems. Materials Research Society, Pittsburgh (1989)
6. Sakurai, T., Sakai, A., Pickering, H.W.: Atom probe field ion microscopy and its applications.
Adv. Electron. Electron. Phys. 20, 1299 (1989)
7. Tsong, T.T.: Atom-Probe Field Ion Microscopy: Field Ion Emission and Surfaces and
Interfaces at Atomic Resolution. Cambridge University Press, Cambridge, Great Britain
(1990)
8. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion Micros-
copy. Oxford University Press, Oxford (1996)
9. Miller, M.K.: Atom Probe Tomography: Analysis at the Atomic Level. Kluwer Academic/
Plenum Publishers, New York (2000)
10. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer Series
in Materials Science, vol. 160. Springer (2012)
Acknowledgment
Numerous people have contributed to the development and success of the LEAP1
over the years. We would certainly like to include everyone who participated in this
venture, but we are certain to miss a few names, and so we apologize in advance for
any omissions.
We would like to start with a special acknowledgement to the family of Tom
Kellywithout their encouragement and support over the years, the current LEAP
would not exist. We would like to acknowledge all of the past employees of
CAMECA1 (formerly Imago Scientific Instruments) who played critical roles
in the development of the LEAP; most notably we would like to thank Dr. Tye T.
Gribb for all of his contributions toward making the LEAP what it is today. We
would also like to thank all of the current employees of CAMECA with a special
acknowledgment to the engineering team for creating a marvelous instrument: Joe
Bunton, Dan Lenz, Jesse Olson, Ed Oltman, Tim Payne, Dan Rauls, Jeff Shepard,
Eric Strennen, Mark Lamers, and Dave Sund.
We would especially like to thank our colleagues who have assisted in the
preparation of this book by reviewing individual chapters. Professional colleagues
include Frederic Danoix, Baptiste Gault, Stephan Gerstl, Michael Miller, Sergej
Mutas, John Panitz, Danny Perea, Amanda Petford-Long, Krishna Rajan, C.J.
Humphreys, and Dan Schreiber. We will particularly like to thank Richard Forbes
for assistance with one of the appendices. Likewise, our CAMECA colleagues who
have reviewed chapters include Peter Clifton, Devin Giddings, Daniel Lawrence,
Dan Lenz, Mark Levesque, David Olson, Jesse Olson, Ed Oltman, David Reinhard,
Hugues Francois Saint Cyr, Michel Schumacher, Eric Strennen, Savanna Strennen,
Ludovic Renaud, and John Watson.
We would also like to thank the scientists and publishers who have granted us
permission to include portions of their work in this book: D. Abou-Ras,
H.O. Andren, O.H. Auciello, P.A.J. Bagot, J. Ballard, R. Banerjee, G. Brennecka,
R. Caballero, J. Cairney, A.J. Cavosie, A. Cerezo, V. Chitry, P.P. Choi,
O. Cojocaru-Miredin, X.Y. Cui, T.L. Daulton, A.M. Davis, F. DeGeuser,
S.P. DenBaars, A. Deschamps, D.R. Diercks, D.M. Dimiduk, J.W. Elam,
P. Felfer, C. Floss, H.L. Fraser, B. Gault, M. Gilbert, H. Gnaser, L.M. Gordon,
vii
viii Acknowledgment
B.P. Gorman, T.T. Gribb, G.H. Gu, E.E. Haller, M. Hasegawa, P.R. Heck,
M.A. Hekmaty, A. Henjered, K. Henry, J. Hiller, A.C. Hillier, C. Hombourger,
Y.L. Hu, J.Y. Hwang, K. Inoue, D. Isheim, K.M. Itoh, D. Jager, M.J. Jercinovic,
D. Joester, A.K. Kambham, Y. Kawamura, K.F. Kelton, M. Kimoto, R. Kirchhofer,
S. Koelling, N. Kruse, L.J. Lauhon, W. Lefebvre, J.L. Lensch, J. Lewis, T. Li,
P. Liddicoat, L. Longstreth-Spoor, A. Mane, E.A. Marquis, I. Martin, S.J. May,
B. Mazumder, A.J. Melmed, T.P. Melo, M.K. Miller, J. Mody, M.P. Moody,
D. Moser, C.K.S. Moy, E.W. Muller, Y. Nagai, S. Nakamura, A. Nishida,
O. Nishikawa, H. Norden, J.A. Panitz, C.G. Park, M.J. Pellin, D.E. Perea,
T.C. Petersen, D. Raabe, J.N. Randall, G. Ranzi, R.F. Reidy, L. Renaud,
S.P. Ringer, M. Romanes, M.R. Savina, D.W. Saxey, D. Schreiber,
D.N. Seidman, Y. Shimizu, R. Shivaraman, C. Sigli, V.S. Smentkowski,
G.D.W. Smith, D. Snoeyenbos, J.S. Speck, R. Srinivasan, F.J. Stadermann,
E. Steel, T. Stephan, A. Strickland, G. Sundell, H. Takamizawa, M.C. Teague,
M. Thuvander, J. Tiley, T. Toyama, S.C.E. Tsang, T. Tsunomura, A. Tyagi,
M. Uematsu, T. Ushikubo, J.W. Valley, W. Vandervorst, F. Vurpillot,
G.B. Viswanathan, B.W. Wessels, S. Wilde, C.A. Williams, M.H. Wong,
R. Wuerz, Y. Xiang, F. Yano, Y. Zhang, and H. Zhong. We also thank the
companies who granted us permission to display images, graphs, or schematics or
who aided in some implementation of LEAP analysis, including Evans Analytical
Group1, Dune Sciences, Gatan, Fischione, Omniprobe, Hummingbird Scientific,
and XEI Scientific Inc.
Foreword
Atom probe tomography has had explosive growth in recent years. The technique
was born way back in 1935 when Edwin Muller invented the field emission electron
microscope. This original atom probe instrument was difficult to use and could only
be applied to metals. Modern-day atom probe tomography is straightforward to use
and can be applied not only to metals but also to semiconductors, ceramics,
geological materials, organic materials, biological materials, and composites.
This is a wonderful book which should be in every science library and on the
bookshelves of scientists and engineers interested in the characterization of mate-
rials at the atomic level. The book requires no prior knowledge of atom probe
tomography and takes readers through from the history of the local electrode atom
probe to the latest developments in the field. Although atom probe tomography is
now an easy technique to use, the interpretation of the results requires considerable
care and understanding. This book takes readers step by step through data collec-
tion, data reconstruction, and data analysis. As with all microscopies, good speci-
men preparation is the key to obtaining world-class results. Various methods of
specimen preparation are described in detail, together with clear instructions on
how to optimize specimen preparation using, for example, focused ion beam
methods.
It is a great pleasure for me to recommend this comprehensive yet very readable
book on the local electrode atom probe, which complements other methods, such as
atomic resolution electron microscopy. I believe that atom probe tomography will
increasingly be used by scientists and engineers involved with the characterization
of materials and devices at the atomic level. I have no doubt that this excellent book
will be widely used by the atom probe community.
Professor Sir Colin J. Humphreys

Professor of Materials Science and Director of Research
University of Cambridge
ix
Contents
1 History of APT and LEAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Ancestry of the Local Electrode Atom Probe . . . . . . . . . . . . . . 2
1.2.1 Early History and the Field Electron Emission
Microscope (~1935) . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Field Ion Microscope: The First Images
of Atoms (1955) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.3 Atom Probe Field Ion Microscope (1967) . . . . . . . . . . . 6
1.2.4 The Advent of Atom Probe Tomography . . . . . . . . . . . . 7
1.2.5 The Position-Sensitive Atom Probe (1988) . . . . . . . . . . 8
1.2.6 Electron Beam Pulsed Atom Probe . . . . . . . . . . . . . . . . 10
1.2.7 The Scanning Atom Probe . . . . . . . . . . . . . . . . . . . . . . 10
1.2.8 The Local Electrode Atom Probe (2001) . . . . . . . . . . . . 10
1.3 The State of Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3.1 The Growth of the Local Electrode Atom Probe . . . . . . 14
1.3.2 Laser Pulsing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3.3 Fundamental Considerations for Design
of Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3.4 Reflectron-Based Instruments . . . . . . . . . . . . . . . . . . . . 18
1.4 FIB-Based Specimen Preparation . . . . . . . . . . . . . . . . . . . . . . . 19
1.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2 Specimen Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Electropolishing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3 Needles Versus Microtips . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.4 Electrostatic Discharge Considerations . . . . . . . . . . . . . . . . . . . 30
2.5 Focused Ion Beam Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.5.1 Capping Considerations and Damage . . . . . . . . . . . . . . 33
2.5.2 Standard Lift-Out Process . . . . . . . . . . . . . . . . . . . . . . 36
xi
xii Contents
2.5.3 Sharpening Process . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

2.5.4 FIB Deprocessing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.5.5 Cross-Section Preparation . . . . . . . . . . . . . . . . . . . . . . 40
2.5.6 Backside Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 43
2.6 Hybrid Transmission Electron Microscopy/Atom
Probe Tomography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.6.1 Preparation and Holders . . . . . . . . . . . . . . . . . . . . . . . . 47
2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3 Design and Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.2 How Atom Probes Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.3 LEAP Performance Parameters . . . . . . . . . . . . . . . . . . . . . . . . 58
3.3.1 Field of View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.3.2 Mass Resolving Power . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.3.3 Data Collection Rate . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.3.4 Model Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4 Instrumentation of the LEAP . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4.1 Local Electrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4.2 Detection and Imaging . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.4.3 Transfer and Storage of Consumables . . . . . . . . . . . . . . 69
3.4.4 Field Evaporation Systems . . . . . . . . . . . . . . . . . . . . . . 72
3.4.5 Ancillary Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.2 Data Quality Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.3 Analysis Yield Considerations . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4 Experimental Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.4.1 Pulse Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.4.2 Base Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.4.3 Detection Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
4.4.4 Pulse Fraction (Voltage Mode) . . . . . . . . . . . . . . . . . . . 89
4.4.5 Laser Pulse Energy (Laser Mode) . . . . . . . . . . . . . . . . . 90
4.4.6 LEAP Parameter Ranges . . . . . . . . . . . . . . . . . . . . . . . 90
4.5 How to Start Your Investigation of Any New Material . . . . . . . 91
4.6 Brief Overview of LEAP Operation: Data Collection . . . . . . . . 92
4.6.1 Voltage Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
4.6.2 Laser Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.6.3 Now You Are Atom Probing . . . . . . . . . . . . . . . . . . . . 105
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5 Data Processing and Reconstruction . . . . . . . . . . . . . . . . . . . . . . . 109
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.2 A Word on Data Files and Work Flow . . . . . . . . . . . . . . . . . . . 110
Contents xiii
5.3 Conversion from Detector Space to Specimen

Space Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
5.3.1 Selection of Depth and Areal Regions . . . . . . . . . . . . . . 111
5.3.2 Spectral Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.3.3 Chemical Identification & Ranging . . . . . . . . . . . . . . . . 117
5.3.4 Spatial Reconstruction: Projection and Depth Scaling . . 120
5.3.5 Wide-Angle Reconstruction Protocols . . . . . . . . . . . . . . 121
5.3.6 Tangential Discontinuity . . . . . . . . . . . . . . . . . . . . . . . 127
5.3.7 Reconstruction Explorer . . . . . . . . . . . . . . . . . . . . . . . . 128
5.3.8 Creation of ROOT and POS Files . . . . . . . . . . . . . . . . . 130
5.4 Discussion of Spatial Resolution and Spatial Positioning . . . . . . 131
5.4.1 Spatial Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
5.4.2 Spatial Positioning (Non-specimen Dependent) . . . . . . . 131
5.4.3 Spatial Positioning (Specimen Dependent) . . . . . . . . . . 134
5.5 A Word on Density Relaxation . . . . . . . . . . . . . . . . . . . . . . . . 142
5.6 Reconstruction Case Study: NIST Standard Reference
Material 2134 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
5.6.1 Reconstruction Parameter Discussion . . . . . . . . . . . . . . 145
5.6.2 Experiment and Analysis Details . . . . . . . . . . . . . . . . . 149
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6 Selected Analysis Topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
6.2 Spectral Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
6.2.1 Ranging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
6.2.2 Practical Considerations for Detection Levels . . . . . . . . 166
6.2.3 When Is the Signal Level Statistically Significant
(Critical Level) for a Peak? . . . . . . . . . . . . . . . . . . . . . 167
6.3 Concentration Space Analyses . . . . . . . . . . . . . . . . . . . . . . . . . 173
6.3.1 Gridding, Voxels, and Delocalization . . . . . . . . . . . . . . 173
6.3.2 Interface Creation and Interfacial Roughness . . . . . . . . . 175
6.3.3 Effects of Delocalization on Planar Surfaces . . . . . . . . . 176
6.3.4 The Proximity Histogram . . . . . . . . . . . . . . . . . . . . . . . 178
6.4 Solute Analysis: Cluster Detection Method . . . . . . . . . . . . . . . 180
6.4.1 Description of the Technique . . . . . . . . . . . . . . . . . . . . 180
6.4.2 Example of Cluster Detection . . . . . . . . . . . . . . . . . . . . 182
6.5 Spatial Distribution Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
6.5.1 The SDM Defined and Important Properties . . . . . . . . . 186
6.5.2 Methods Similar to the SDM . . . . . . . . . . . . . . . . . . . . 186
6.5.3 Understanding Basic SDMs . . . . . . . . . . . . . . . . . . . . . 187
6.5.4 Calculating SDMs in IVAS . . . . . . . . . . . . . . . . . . . . . 188
6.5.5 Visualizing Tungsten SDMs with IVAS . . . . . . . . . . . . 189
6.6 Application of Spatial Distribution Maps . . . . . . . . . . . . . . . . . 191
6.6.1 Finding the Crystal Lattice . . . . . . . . . . . . . . . . . . . . . . 191
6.6.2 Using SDMs to Calculate Efficiency . . . . . . . . . . . . . . . 192
6.6.3 Ordered Structures and Site Occupancy . . . . . . . . . . . . . 195
6.6.4 Ordering in Al3Sc Precipitates . . . . . . . . . . . . . . . . . . . 195
xiv Contents
6.6.5 FeCr Precipitates in a NiAlFeCr Alloy . . . . . . . . . . . . . 195

6.6.6 Site Occupancy of Nb in TiAl . . . . . . . . . . . . . . . . . . . 196
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
7 Applications of the Local Electrode Atom Probe . . . . . . . . . . . . . 201
7.1 Metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.1.1 Ordered Alloy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.1.2 Site Occupancy in Precipitates in Aluminum Alloys . . . 203
7.1.3 Imaging Nanovoids . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
7.1.4 Intergranular Attack in Ni-Base Superalloy . . . . . . . . . . 204
7.2 Catalytic Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
7.2.1 Ex Situ Analysis of CoCuMn Nanoparticles . . . . . . . . . 207
7.2.2 In Situ Analysis of PdRd and PtRhRu Catalysts . . . . 209
7.3 Ceramic and Geological Materials . . . . . . . . . . . . . . . . . . . . . . 210
7.3.1 CeO2 as a Model for Nuclear Fuel . . . . . . . . . . . . . . . . 210
7.3.2 Zircons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
7.3.3 Extrasolar Nanodiamonds . . . . . . . . . . . . . . . . . . . . . . . 214
7.3.4 Ferroelectrics/Piezoelectrics . . . . . . . . . . . . . . . . . . . . . 216
7.4 Semiconductor Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
7.4.1 Group IV Semiconductors (Silicon and Germanium) . . . 218
7.4.2 Compound Semiconductors . . . . . . . . . . . . . . . . . . . . . 222
7.5 Organics and Biological Materials . . . . . . . . . . . . . . . . . . . . . . 227
7.5.1 Synthetic Polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
7.5.2 Chiton Teeth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
7.5.3 Ferritin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.6 Composite Structures/Devices . . . . . . . . . . . . . . . . . . . . . . . . . 233
7.6.1 MetalOxide Interfaces . . . . . . . . . . . . . . . . . . . . . . . . 234
7.6.2 MOSFET Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
7.6.3 FinFET Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
7.6.4 Commercial Devices Analysis; Intel i5 . . . . . . . . . . . . . 237
7.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Appendix A: Data File Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
Appendix B: Field Evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Appendix C: Reconstruction Geometry . . . . . . . . . . . . . . . . . . . . . . . 267
Appendix D: Mass Spectral Performance . . . . . . . . . . . . . . . . . . . . . . 281
Appendix E: Additional Considerations for LEAP Operation . . . . . . 289
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Abbreviations: Variable and Function
Descriptions
a Acceleration
A Detected surface area
a 0, a 1, a 2 Voltage fitting constants
Ai Actual count of type i ion
B Background, broadening due to the adsorption process
Bgd Background level in ppm/ns
c Bowl fitting constant
C( ) Gaussian response function
Ci Concentration of ith range
d Separation distance
Dmax Maximum cluster ion separation
DR Detection rate
dR/dZ Radial change as a function of depth change
e Fundamental unit of charge
E Electric field, cluster detection erosion distance
F Force, evaporation field
f Frequency
F0 Field necessary for onset of field evaporation
FB ( ) Bowl correction function
fij Isotopic abundance of isotope type i in range j
fS/C Specimen sphere-to-cone ratio
FV ( ) Voltage correction function
FEV Field evaporation
h Specimen height
In nth ionization energy
k Electric field factor, confidence level factor
Ke Evaporation rate
xv
xvi Abbreviations: Variable and Function Descriptions
L, LDet Flight path length, cluster detection envelope parameter

LPE Laser pulse energy
M Magnification
m Mass
m/n Mass-to-charge-state ratio
m/m Mass resolving power (MRP)
Mj Measured counts in each range j
N Ion sequence number, number of chemical species, number of relative
displacement vectors
n Charge state, number of atoms in the analyzed dataset
ncell Number of atoms in a cell
Nmin Minimum cluster size
small-angle magnification
p Concentration level
PF Pulse fraction
PFI Post-field ionization
Q0 Zero-field activation energy
Qn Activation energy barrier for field evaporation of an n-fold charged ion
r Radius of what or radial distance
R, RDet Detector radius, specimen radius
R0 Specimen radius, initial tip radius
RC Distance from specimen apex to the counter electrode
RCone Cone radius
RF, RFinal Final tip radius
RS Spherical counter electrode radius
RSphere Sphere radius
S Signal
T Temperature
t Time or time of flight (TOF), feature thickness
T, Ti Ranged counts, counts in ith range
t0 Detector system time-of-flight offset constant
tcor Corrected time of flight
traw Uncorrected or raw time of flight
TTotal Total counts in all ranges
v Velocity
V Voltage
VDC Direct current voltage (time independent)
Vext Extraction voltage
Vp Pulse voltage
Vpa Post-acceleration voltage
Vtot Total accelerating voltage
Abbreviations: Variable and Function Descriptions xvii
w Specimen base width

X, Y, W Detector anode references
X, Y, Z Detector coordinates in mm
x, y, z Reconstructed specimen coordinates in nm
xint Distance from the sample surface at which the potential energy curves
intersect
xscr Screening distance
Shank half-angle
a, i, m Polarizabilities of a neutral atom, and ion, and a substrate metal atom
ij Fractions of the ions in range j that are type i
( ) Dirac delta function
(t) Piecewise linear interpolation function
i,j Kronecker delta function
L/L Flight path spread
m Mass spread
m Width of mass peak
m/m Mass resolution
t Total timing imprecision or time-of-flight spread
tD Time-of-departure spread
tP, t Physical timing uncertainty or timing uncertainty ignoring hardware
limitations
tT Timing electronics uncertainty
V/V Voltage spread
z z reconstruction increment, spatial increment for building 1D
histograms
/E Energy spread
Detector efficiency
0 Effective detection efficiency
0 Frequency
Sublimation energy
Mean value of a Gaussian function
a Zero-field dipole moment of the adsorbate
Surface atom vibration frequency
Image compression factor
Number of atoms per unit volume
( ) Function describing the reconstructed ion positions
Width of a Gaussian function, delocalization
i Expected standard deviation of ith range
Angular field of view (FOV), work function
Ratio of the ionization energy (from a solid) to the sublimation energy
Atomic volume
Chapter 1
History of APT and LEAP
1.1 Introduction
Atom probe tomography (APT) is one of the most spectacular microscopies that
exist. A three-dimensional (3D) image at near atomic scale is produced with single-
atom sensitivity where each atom (actually each isotope) in the image is identified.
Because the fundamental data format is the 3D position and identity of individual
atoms in a volume containing potentially hundreds of millions of atoms, many types
of information may be gleaned. Elemental concentration may be determined in any
subvolume size or shape simply by counting atoms. Concentration profiles may be
calculated in any direction, even radially through a spherical feature or normal to
any defined surface. Isoconcentration surfaces can be set to delineate and measure
interfaces. Interatomic distribution functions can be determined for studying order-
ing, dopant interactions, cluster formation, crystal structure, diffusion, and early
stages of precipitation. Once the dataset is obtained, the quality and quantity of
results derived are limited principally by the intrinsic quality of the data and the
microscopists ability to direct a computer to extract the information.
APT is used to study an increasingly wide variety of inorganic materials and is
gradually demonstrating capabilities for analysis of organic materials as well.
The unique power of APT lies in its ability to tie compositional information
to structure. Other characterization techniques can determine physical structure at
sub-nanometer levels, but only APT has the capacity for broad compositional
identification (and quantification) at these length scales. Furthermore, APT accom-
plishes this with high analytical sensitivity which can reach one atomic part per
million (appm).
The atom probe was invented in 1967 at the Pennsylvania State University by
Muller, Panitz, and McLane [1] as a 1D analytical instrument that collected
hundreds of atoms/day with approximately a one-nanometer field of view (FOV).
By 1973, the origins of tomographic analysis were present in Panitzs 10 cm Atom
Probe [2]. A revolution in atom probe technology came when an operational, fully
3D imaging instrument was developed by Cerezo, Godfrey, and Smith [3] that
D.J. Larson et al., Local Electrode Atom Probe Tomography: A Users Guide, 1
DOI 10.1007/978-1-4614-8721-0_1, Springer Science+Business Media New York 2013
2 1 History of APT and LEAP
Fig. 1.1 Ancestry of the LEAP
could record hundreds of atoms/min in a ~15 nm wide FOV. In the past decade or
so, the local electrode atom probe (LEAP) [4, 5] and related contributions like the
scanning atom probe [6] and focused ion beam (FIB)-based specimen preparation
[7] have brought about another revolution in the development of APT [8]. With a
LEAP, millions of atoms/min can be collected with up to ~200 nm wide FOV with
high mass resolution (<1/1,000) on many types of materials.
It is useful to review the APT history so that users can understand the LEAP and
its evolution from its predecessors. This chapter is intended to be historical in nature
and focuses on the people and machines that have led to the current generation of
atom probe instruments. In Fig. 1.1, a flow chart of the predecessor instruments in
the lineage of the modern atom probe is presented. There have been dramatic
improvements in atom probe instrumentation throughout its history. Recent
advances have been driven by the availability of vastly superior commercial
instruments, most notably the LEAP.
1.2 Ancestry of the Local Electrode Atom Probe
First, to clarify some nomenclature: APT is the name of the technique. There are
many types of instruments which may be 1D atom probes (1DAP) or 3D atom
probes (3DAP). Some instruments that are 3DAPs were given names such as the
1.2 Ancestry of the Local Electrode Atom Probe 3
position-sensitive atom probe (PoSAP) [3] or the tomographic atom probe (TAP)
[9]. Instruments that feature a local electrode are LEAPs, which is a name used
by CAMECA Instruments, Inc.
1.2.1 Early History and the Field Electron

Emission Microscope (~1935)
The history of APT begins in the early nineteenth century. The electron was
discovered by Thomson [10] in 1897, and the modern view of the structure of the
atom was developed by Rutherford [11] and Bohr [12] in 1913. Experimental
confirmation of the atomic nature of materials and packing of atoms in crystals
was deduced by Bragg and Bragg [13] in 1913. Charged-particle quantum optics
theory was being developed in the late 1920s, and Fowler and Nordheim [14]
predicted the magnitude of electric field that would be required to strip an electron
from an atom, and physicists of the day could readily show that fields of about
109 V/m (1 V/nm) should suffice.
Physicists next considered how to create an electric field to form a beam of
emitted electrons from a material. Since the electric field in the region between
parallel plates is simply the voltage between them divided by their separation,
109 V/m is achieved by applying 10,000 V1 across a micrometer separation. This
was not, however, a practical scenario, and physicists of the day certainly under-
stood that. It was also known that with a concentric spherical geometry for the
electrodes, the field on the inner spheres surface would vary as the voltage between
the surfaces divided by the radius of the inner sphere. So for 10,000 V it is necessary
only to make the inner radius small enough. If the sphere (or hemisphere) is on
the end of a needle and is opposed by a flat plate normal to the axis of the needle,
one can achieve almost the same result. Thus for 10,000 V, a tip apex radius on a
needle of about 1 m will create these fields of 109 V/m and remove electrons from
the surface. This is a field emission electron source and brings us to about the end of
the 1920s when Professor Gustav Hertz of the Technical University of Berlin hired
Erwin W. Muller as a student to build a field electron emission microscope (FEEM)
which they used to demonstrate this effect and study the phenomenon. Erwin
Muller, Fig. 1.2, eventually became the father of this scientific discipline known
as high-field nanoscience. Incidentally, at the same institute at the same time,
Professor Max Knoll had a student, Ernst Ruska, working on a new instrument
called an electron microscope.
1
10,000 V is a convenient high voltage to produce experimentally in the laboratory without
needing any extreme measures.
Fig. 1.2 Photograph of

Professor Erwin W. Muller
(19111977): Father of high
field nanoscience (figure
courtesy of J. A. Panitz, The
University of New Mexico,
Department of Physics and
Astronomy)
Muller developed the FEEM [15] around 1935. By creating a field electron
emission source and projecting the emission pattern of electrons onto a fluorescent
screen, an image like that in Fig. 1.3b was formed. This was very exciting in the
1940s, when this work was starting to mature, and researchers started to ponder
the ultimate spatial resolution of this technique. Effectively, they were asking the
following question: Could we see atoms? By the mid-1940s, it was clear that the
answer was no. The FEEM achieved about two-nanometer resolution at best and
this has not improved significantly [16]. However, this work did spur some thinking
that led to the next development, which is the field ion microscope (FIM).
1.2.2 Field Ion Microscope: The First Images

of Atoms (1955)
It became common practice to reverse the polarity on an FEEM tip to get rid of
adsorbed hydrogen on the specimen. Instead of simply cleaning the tip, they found a
steadily emitted stream of positive ions which formed a field ion emission pattern
on a fluorescent screen [17]. Higher fields of ~1010 V/m were required, so the
Fig. 1.3 (a) Schematic of the field electron emission microscope. (b) FEEM image of <110>
orientation of tungsten. The twofold symmetry of this pattern is evident. (c) Schematic of the field
ion microscope. (d) Field ion image of <110> tungsten. The first images of atoms were obtained
on October 11, 1955, by Bahadur and Muller. ((b) Reprinted with the permission from Springer-
Verlag GmbH York [17]. (d) Reprinted with permission from Elsevier [20])
needles required a ~0.1 m diameter apex in order to achieve ionization. FIM

produced images of higher spatial resolution than FEEM. The first FIM images
were produced in 1951 by Muller [18]. It took several more years of trying, but in
1955 Kanwar Bahadur and Muller [17, 19] resolved individual atoms with the field
ion microscope, Fig. 1.3d. Recollections of this period in the laboratory may be
found in [20]. This was the first time humans saw atoms some 2,400 years after
Leucippus and Democritus postulated the existence of the undividable particle.
Fig. 1.4 The original atom probe field ion microscope circa 1967. (a) Schematic drawing and (b)
photograph of the apparatus with John Panitz at the helm (figures courtesy of J. A. Panitz, The
University of New Mexico, Department of Physics and Astronomy)
1.2.3 Atom Probe Field Ion Microscope (1967)
After looking at images of atoms for half a dozen years one begins to ask, Cant we
figure out the identity of any or all of those atoms? This was first attempted using a
magnetic-sector mass spectrometer instrument by Barofsky and Muller [21], but
eventually Barofsky became convinced that time-of-flight mass spectroscopy was
the better approach [22]. Muller, Panitz, and McLane [1] thus settled on measuring
time of flight as the way to identify the ions visible in the FIM. This required a
pulsed field evaporation event in order to know when the atoms left the tip. John
Panitz (a Ph.D. student working with Muller), S. Brooks McLane, and Gerald
L. Fowler (technicians in Mullers lab) were responsible for building this instru-
ment under Mullers watchful eye. It was called the atom probe field ion micro-
scope (APFIM), Fig. 1.4, and was a 1D instrument. The operator would press a
button that sent a voltage pulse to the specimen, and the oscilloscope screen would
show a trace when the detector was hit by an ion, Fig. 1.5.
The technique progressed quickly as the groups instrument could collect tens of
atoms/h and had about a 1 nm FOV. The mass resolving power (MRP) was just
sufficient to resolve 56Fe+ from 57Fe+. Often working into the night, Panitz
would record the oscilloscope trace on Polaroid Type 57 film, typically exposing
Fig. 1.5 Data file from the

APFIM. Photograph of an
oscilloscope screenshot
with two traces. The lower
trace shows two atom hits.
The sine wave at the top was
used as a known time base
(figure courtesy of J. A.
Panitz, The University of
New Mexico, Department
of Physics and Astronomy)
a half-dozen traces per sheet to save film. A sine wave from a crystal oscillator was
recorded on the same sheet of film as a time reference independent of the oscillo-
scopes time base generator. He then projected each film positive 10 using an Aero
Ektar lens and, for each ion, measured the distance from a reference object with a
meter stick and converted that distance on the sine wave to a flight time. By morning,
he would have a list of about 100 atoms that had been recorded overnight. Groups at
the University of Pittsburgh under Sid Brenner and at the University of Oxford under
George Smith began developing atom probes. In addition to the group at Oxford,
groups at Chalmers (Hans-Olof Andren) and Cornell (David Seidman) started
computerizing control of the instruments. Soon commercial instruments by Vacuum
Generators (the VG APFIM 100 in 1980) and Applied Microscopy (the APFIM
200 in 1986) appeared and offered data collection rates of 100,000 atoms/day and
MRP (FWHM) of 1 part in 2,000 with voltage pulsing using a Poschenrieder lens
[23] and reflectron optic [24, 25], respectively. These were 1D instruments with a
probe hole for the atoms to pass through and a ~1 nm FOV.
1.2.4 The Advent of Atom Probe Tomography
The 3D atom probe seems a natural progression from the 1D atom probe, yet it was
a revolutionary departure from the APFIM when it was first pursued. John Panitz
developed what he called the imaging atom probe (IAP) [2] in 1972, Fig. 1.6. It was
a short flight path instrument with a large FOV, a design that we shall see below still
exists because of its fundamental merit. Time of flight could be measured anywhere
on the full detector because it used spherical microchannel plate amplifiers centered
on the specimen apex, keeping the distance to the channel plate amplifier the same
in all directions. The IAP would record the hit positions of a single atom type
(or isotope) at a time from any given voltage pulse by time gating the channel
plates. With this approach, an iron map in steel could be recorded, then a carbon
Fig. 1.6 Photograph of the

imaging atom probe circa
1972 (figure courtesy of
J. A. Panitz, The University
of New Mexico.
Department of Physics and
Astronomy)
map could be recorded from the next part of the specimen, and so on. The IAP
collected 3D data by manually examining a series of 2D maps of a single
preselected species as a function of depth. Since the field desorption process is
destructive only a single species could be examined in each section, unlike a
modern 3DAP where a 2D map of all species can be obtained as a function of
depth. Nonetheless, the IAP produced tomographic data from an atom probe, and
Panitz should be credited with the initial inventive step in the development of APT.
1.2.5 The Position-Sensitive Atom Probe (1988)
The advent of a true 3DAP that recorded all atom species had to wait until the 1980s
when position-sensitive detector concepts were adapted to what essentially
amounted to IAP-like instruments. Miller conceived of such an instrument and
presented his concepts in the mid-1980s [26, 27]. The first fully operational 3D
atom probe was developed in 1988 at the University of Oxford by Cerezo, Godfrey,
and Smith and was dubbed the PoSAP [3], Fig. 1.7. These authors adapted a
wedge-and-strip detector, which was prevalent in astronomy to a commercial VG
APFIM 100 which was in their lab. By adding a 2D detector to a 1D atom probe,
Fig. 1.7 Photograph of the position-sensitive atom probe (PoSAP) developed by Cerezo,
Godfrey, and Smith at the University of Oxford in 1988 (figure courtesy A. Cerezo)
they created a working 3D atom probe. The images were spectacular for the time
period. They contained about a million atoms and showed precipitates and inter-
faces in materials. Analytical electron microscopists at the time appreciated how
challenging it would be to do similar work with an electron microscope.
1.2.5.1 The Tomographic Atom Probe
After a postdoctoral experience at Cambridge in the early 1970s, Jean Gallot started
an activity based on field ion microscopy on his return to Rouen. The group
progressively grew from a handful of individuals to more than 150 today. This
growth is partly due to the excellence of their instrumental development of the atom
probe under the successive leadership of Alain Menand and Didier Blavette in what
became the Groupe de Physique des Materiaux (GPM), UMR CNRS 6634. Building
on a unique local expertise, Blavette played a key role in the evolution of the atom
probe from a one-dimensional technique into the tomography technique that we
know today. He supervised the development of the multi-anode position-sensitive
detector in the early 1990s and the methodology that enabled three-dimensional
images to be built, atom-by-atom. The instrumentation development is now in the
hands of a highly specialized team at the GPM, who actively works with CAMECA,
a company that has been working alongside the GPM for 20 years on the commer-
cialization and further advances of the technique.
1.2.6 Electron Beam Pulsed Atom Probe
The PoSAP was an extremely successful instrument, as it ushered in a new era in

microscopy, but it had its limitations. It could take days to collect useful amounts of
data, and it was limited to a ~15 nm FOV. The data collection rate of these
instruments was limited by the high-voltage pulsers available at that time, which
could not go to high repetition rates at high voltages. These limitations are what
drove Kelly et al. [28, 29] to try to go to much higher repetition rates by heating
very small volumes of the specimen apex quickly and achieving both rapid heating
and cooling. At the time it was not feasible to focus lasers to submicron spots in an
atom probe, and this led Kelly to the concept of an electron beam pulsed atom probe
[28, 29]. Focusing electron beams to small diameters (~200 nm) was possible, so if
it were possible to deliver enough energy (~100 nJ) in a short enough time (~100 ps)
in a small enough volume (~106 nm3), high repetition rates with high mass
resolution might be achieved. Larson pursued a Ph.D. thesis on electron beam
pulsing as a way to achieve these goals [30]. This group built an atom probe inside
a scanning electron microscope (SEM). This instrument was never successful at
demonstrating electron beam pulsing due to limitations of the hardware at the time,
but it did demonstrate that there was value in having an electron image of an atom
probe specimen, and we see this theme being pursued again today.
1.2.7 The Scanning Atom Probe
The next step in the development of the atom probe was the SAP [6] invented by
Osamu Nishikawa (who also came from Mullers lab) in 1993, Fig. 1.8. The idea
was to use a funnel-shaped electrode positioned close to a surface with a high
applied potential. Any sharp points present in the region the aperture passed over
would cause ions to start evaporating, a fraction of which would pass through the
aperture. The geometry theoretically enabled a radically different approach to
making specimens. Specimens had always been needle shaped, but now Nishikawa
was suggesting microtip arrays. This concept had a major impact on how one could
think about specimens and instruments.
1.2.8 The Local Electrode Atom Probe (2001)
At this time, Kelly was still trying to address the basic conundrum of the 3DAP; it
was a spectacularly successful analytical tool, but it was too slow and had too
narrow of a FOV to be practical for many applications. Also about this time, field
electron emitter arrays were being developed that had field enhancement factors of
up to 10 or more due to proximal electrodes [31]. This field enhancement
Fig. 1.8 Schematic of the scanning atom probe (reprinted with permission from Elsevier [6])
(producing the same field using a lower applied voltage) provided a means to
achieve higher data collection rates. This all came together in two key thoughts:
(1) a local electrode could be used to lower voltages and attain high data collection
rates without an electron beam and (2) a relative reduction in the energy spread
(due to the voltage reduction) made it possible to achieve high mass resolution over
the entire FOV [4]. So, a local-electrode geometry, like that in Fig. 1.9, actually
mitigated or solved the three biggest issues with the atom probe of the day:
specimen geometry, data collection rate, and mass resolution over a large FOV.
Based on these ideas, in 1994 Thomas Kelly, David Larson, Patrick Camus, and
Louis Holzman incorporated Millennium Instruments. Kelly and Larson carried this
forward, but it was not until 1998 that the first full-time employee, Tye Gribb, was
hired. From the outset, Millennium intended to develop laser pulsing to provide an
instrument that could be used for the analysis of semiconducting materials.
The group raised capital and hired additional people. It took about 2 years to build
the voltage pulsed alpha prototype, Fig. 1.10a. The alpha prototype had two
ultra-high vacuum SEMs mounted on the side for positioning the tip relative to the
local electrode. After the turn of the millennium, the company became Imago
Scientific Instruments. In 2002, Michael Miller at Oak Ridge National Laboratory
purchased the voltage pulsed beta prototype LEAP, which was delivered in 2003.
Imago started working on laser pulsed LEAP in earnest in 2002 and received a grant
in 2003 from the US National Science Foundation [32] to develop laser pulsing.
Eventually Imago shipped the first commercial laser pulsed atom probe in 2006.
Fig. 1.9 Schematic of the

LEAP Si specimen, local
electrode, and detector
geometry
Also in 2006, Imago acquired Oxford Nanoscience which had developed a novel
curved reflectron [33] for large-FOV imaging in voltage pulsing and laser pulsing
modes. In 2010, Imago was acquired by AMETEK1 and was made part of
CAMECA SAS. A photograph of todays LEAP 4000X is shown in Fig. 1.10b for
comparison with the original alpha prototype.
In order for all of these developments to be realized, many things needed to
happen at the right time. If any one of them did not happen, the APT reality today
might be very different. The Imago team was contemplating generating terabytes of
data in 1998, which was a daunting thought at the time. The CPUs and the graphical
processing units back then were not capable of processing, visualizing, and ana-
lyzing a hundred-million-atom dataset. The only known way to cool a specimen
to 20 K without vibration was using liquid helium, which was a less than desirable
option. Fortunately, commercial vibration-isolated mechanical refrigeration sys-
tems appeared in 2000 just in time for the needs of this instrument. The lasers in the
early 2000s were not stable enough for a commercial atom probe offering. The first
suitable laser for small-spot laser pulsing became commercially available about
1 year prior to adoption and commercial FIB tools did not exist 20 years ago. Given
all of this, we now identify three crucial technologies that contributed to the current
status of APT: the local-electrode geometry, FIB-based specimen preparation, and
thermal pulsing with lasers.
Fig. 1.10 (a) Photograph of the alpha prototype of the commercial LEAP at Imago Scientific
Instruments (2001). (b) Photograph of a current LEAP 4000X system for comparison with the
alpha prototype (reprinted with the permission of Cambridge University Press [8])
The local-electrode geometry allowed us to go from a 15 nm FOV to greater than

150 nm FOV. The speed of the instrument increased from ~104 to >107 atoms/h.
The analytical sensitivity improved from 100 to ~1 appm. The analyzed volumes
increased by three orders of magnitude. The MRP at wide FOV has also improved
by a factor of 5 to 1 part in greater than 1,000 (due also to other effects such as laser
pulsing and reflectron compensators). Instead of working with 510 mm long
needles, 100 m tall microtips are available, and it is all compatible with laser
pulsing. A summary of the current performance, along with some predicted
performance from 1990 [28], was recently presented in [8].
1.3 The State of Instrumentation
1.3.1 The Growth of the Local Electrode Atom Probe
As described above, significant changes in atom probe instrumentation came about

in the early part of the 2000s. Prior to this, the technique was constrained by a low
data collection rate (up to ~10 atoms/s or 106 atoms/day) and a small FOV
(~1520 nm diameter if an MRP of 200 was to be realized). With these limitations,
the data collection rate translated to an analyzed depth of about 1525 nm/day.
Larger FOVs were possible but only by sacrificing MRP. The challenge was to find
a way to obtain high data collection rates and deliver high MRP in a large-FOV
configuration. This challenge has been met by at least two different approaches to
date, and these are illustrated in Fig. 1.11: the straight-flight-path LEAP in
Fig. 1.11a and the reflectron-compensated LEAP in Fig. 1.11b.
Fig. 1.11 The two basic

configurations of current
LEAP instruments: (a)
LEAP 4000X Si, which
is a straight-flight-path
configuration, and (b)
LEAP 4000X HR, which
is a curved-reflectron
configuration
1.3 The State of Instrumentation 15
In a remote-electrode geometry, blunter tips require higher voltages to create the

requisite electric field at the apex while projecting a larger area onto the same
detector area. A local-electrode geometry [4, 5, 3436] has the same trend, but for a
given applied voltage the field is higher at the specimen. Consequently, the FOV is
increased at a given voltage relative to the remote-electrode geometry. Practically
speaking, the local electrode doubles the FOV at a given voltage for commercial
LEAP systems. As a result of Faraday shielding of the flying ions from the time-
varying voltage pulse, the potential energy gained by evaporated ions is also more
uniform in LEAP which leads to improved MRP. The lower voltages also make it
much easier to build high-repetition-rate voltage pulsers and achieve up to two
orders of magnitude greater pulse repetition frequency than previous atom probes
(120 kHz). In 2006, the sum of all these advantages provided an instrument
capable of MRP of 500, a 200 nm FOV, and a pulse repetition rate of >200 kHz
yielding >106 atoms/min. This performance was achieved in a straight-flight-path
geometry (Fig. 1.11a) with a large detector that also delivers >50 % detection
efficiency for all ion types. The performance and capability of the LEAP have
continued to improve and by 2008 included laser pulsing and large-angle reflectron
technology.
1.3.2 Laser Pulsing
Historically, voltage pulses have been used to generate pulsed field evaporation
[1]. This requires that the specimen has high electrical conductivity (>102 S/cm),
which limits the technique to metals and heavily doped semiconductors [37].
Alternatively, short laser pulses (<1 ns) directed at the specimen apex can be
used to induce a pulse in the evaporation rate for any material regardless of
electrical conductivity [38]. For proper quantitative operation, the DC voltage on
the specimen should be low enough to prevent evaporation between laser pulses.
Unlike the case of laser ionization mass analysis (LIMA), the temperature rise
required for laser pulsed APT can be small: typically 100300 K [39, 40]. Data
from Kellogg [41] show that for a given evaporation rate, the evaporation field
decreases as temperature is increased. There have also been suggestions that laser
pulsing may invoke athermal field evaporation pulsing mechanisms [42], but
experimental evidence has shown that thermal pulsing dominates the process
[4345].
Experiments performed in the 1980s by Kellogg and Tsong demonstrated that
semiconductor materials and metals could be analyzed with laser pulsing [38, 39,
4648]. The technique was not widely applied because lasers in 1980 were slow,
expensive, and unstable. The first 30 years of atom probe applications were
dominated by metals research using voltage pulsing. Eventually, given the
compelling reasons for using laser pulsing to analyze insulating materials coupled
with advances in laser technology, laser thermal pulsing caught on. The first
commercial laser atom probes were offered in 2005 by all of the commercial
vendors at the time. Now nearly every commercial atom probe being sold
incorporates a laser. The advent of faster laser systems and well-engineered com-
mercial atom probe instruments has made laser pulsed operation similar to that of
voltage pulsing from a users point of view. Laser pulsing opens a broad universe of
applications available for study by atom probe; examples include silicon-based
semiconductors and compound semiconductors [7, 49], multilayer thin films [50],
ceramics [5154], and organics [55, 56] (note that the authors have not attempted to
be complete here in referencing the various applications; a more complete review of
recent applications of APT is provided in Chap. 7). In addition, laser pulsing
produces a very low energy spread from the ions. The result is very high MRP
for long flight times. Indeed, Liu and Tsong [57] built an eight meter pulsed laser
atom probe that achieved an MRP of 4,000.
A main challenge in developing a laser pulsed atom probe was to create thermal
pulses of sub-nanosecond duration at greater than 100 kHz repetition rate that
would work with poor thermal conductors. By focusing the laser energy into a
diffraction-limited spot of less than 3 m diameter (4), a very small volume of the
specimen is heated. This makes it possible for the specimen to cool quickly and
results in an evaporation window in time (known as time-of-departure spread) of
less than 200 ps. It also makes it possible to analyze materials with poor thermal
diffusivity with relatively good MRP.
Another recent development is the migration to shorter wavelengths in commer-
cial atom probe systems. Early systems used a laser fundamental wavelength of
either 800 nm (titanium sapphire or Ti:sapphire) or 1,064 nm (neodymium-doped
yttrium aluminum garnet or Nd:YAG). Systems progressed to a frequency-doubled
green wavelength of 532 nm (Nd:YAG), and in 2009 frequency-tripled ultraviolet
wavelengths of 355 nm [58] or 343 nm [44] (Nd:YAG) were introduced commer-
cially. One reason for utilizing the shorter wavelengths is that the absorption depth
(anomalous skin depth) decreases as the cube of the wavelength, so in addition to a
smaller diffraction-limited spot, a shallower heated depth is also produced. These
advances result in smaller heated volumes and more rapid cooling in the apex of the
specimen. In all these cases, rapid cooling is the key to achieving the small time-of-
departure spread, which translates into high MRP [43].
Because it uses a relatively small, focused laser beam and a local electrode,
CAMECAs LEAP 4000X and its predecessor, the LEAP 3000X, are currently
the only laser pulsed atom probes with the capability to analyze microtip arrays.
The importance of this ability lies in the way specimens may be configured. As
shown below in Fig. 1.12a, arrays of microtips may be fabricated, for example, by
micro-electro-mechanical system (MEMS) processing methods for silicon. Such
microtip arrays can serve many purposes: (1) they can be analyzed directly after
some form of processing (thermal, implantation, etc.) (Fig. 1.12b); (2) they can be
coated with thin films of inorganic or organic materials (Fig. 1.12b); and (3) they
can serve as holders for extractions from wafers or other bulk materials (Fig. 1.12c).
Material extraction methods (lift-out) make it possible to prepare specimens from a
very small volume (<1 m3) and from a wide variety of sample types with excellent
reproducibility. As discussed in Chap. 2, small material wedge extractions can be
1.3 The State of Instrumentation 17
Fig. 1.12 SEM images of a microtip array manufactured by MEMS processing methods. (a)
Microtip array coupon, (b) a single microtip, (c) a flat-top microtip with lift-out specimen in
position for bonding, and (d) the apex of a pre-sharpened microtip
mounted on a microtip and an entire array of microtips populated for analysis. Most
of the applications described in this book were prepared via lift-out of material and
subsequent transfer to microtip carrier arrays.
1.3.3 Fundamental Considerations for Design

of Instrumentation
Atom probe instrumentation experienced a phase of rapid development in the first

decade of the 2000s (see [8] for a brief history). These tremendous gains in
performance of APT instruments [5861] give cause to consider the extent to
which better performance can be achieved. Greater MRP is desirable but not at
the cost of FOV; when MRP is sufficient for a task, then FOV is more important.
The converse is also typically true. The challenge for instrument design is to find
the optimal trade-off. A straight-flight-path atom probe geometry is illustrated in

Chap. 3, Fig. 3.1. When the flight path length, L, is small, the detector subtends a
large solid angle in the data path, which produces a large FOV, . The flight times
are small when the flight path length is small, so high repetition rates may be used.
However, the timing uncertainty will be a finite number, and for small flight times,
it can lead to unacceptable uncertainties in the timing resolution or the resultant
MRP for ions. If the energy spread for similar ions is negligible, the timing
uncertainty does not change during flight, so if the flight path length is increased,
the relative timing resolution will improve, and greater flight path lengths can
produce higher MRP. Thus, there is generally an inverse relationship between
FOV and MRP. The ongoing challenge is to push this envelope of MRP and FOV
to greater values.
In laser pulsing, the energy spread for many materials is almost negligible
(3 eV) which makes it possible to achieve high MRP with long flight times.
However, some materials may have a large intrinsic energy spread (~10 eV) in laser
pulsing that is large enough to defeat the goal of achieving high MRP with long
flight paths. The source of this large energy spread has not been identified, but
fortunately it appears that few materials have a large energy spread in laser pulsing.
1.3.4 Reflectron-Based Instruments
Given that energy spread may be greater than a few electron volts, even in laser
pulsing, the options for improving peak discrimination are energy compensation
and energy determination. Energy compensation seeks to equalize ion flight time by
causing ions of higher energy to take longer pathways to the detector. When done
correctly, all ions with the same mass-to-charge-state ratio have the same flight
time. This approach makes it possible to achieve high MRP in voltage pulsing
mode as well as in laser pulsing mode even when large energy spreads exist.
Mamyrin et al. [25] designed the first reflectrons for general mass spectrometry
following a concept proposed by Alikhanov [24]. Waugh et al. [62] first used
reflectrons in APT for 1D atom probe instruments. Cerezo et al. [63, 64] adapted
reflectrons to an imaging application in 3D atom probes within a few years. All of
these reflectrons used planar electric fields, which became a problem in imaging
instruments because, although they create an output surface with reduced timing
spread, this comes at a cost of introducing spatial aberrations for ions of different
energychromatic aberration. The FOV was therefore restricted in reflectron-
based instruments to small values (~25 nm diameter) to limit this effect. Panayi
[33] solved this problem by designing a curved reflectron in which the output
surfaces for time-focused ions and spatial-focused ions are made to nearly coincide.
As a result, the FOV of curved reflectrons can be made approximately an order of
magnitude larger (>150 nm). These curved reflectrons were adapted to a commer-
cial remote-electrode atom probe [65] and are currently used in the LEAP 4000 HR
instrument [66].
1.5 Concluding Remarks 19
1.4 FIB-Based Specimen Preparation
The second notable component of the significant recent advances in the field of APT
is fabricating appropriate samples from disparate materials (which is discussed in
detail in Chap. 2). Historically, researchers have always desired site-specific sample
preparation so that they could know, in advance, what was contained within the
specimen apex. About 20 years ago Alexander and colleagues at Oak Ridge National
Laboratory used a Gatan precision ion milling system to modify specimens
for APT [67]. The early instruments did not have the performance to meet
APT requirements, but they were the precursor to the FIB work that came in the
late 1990s.
Two FIB-based methods were proposed by Larson et al. [68, 69] in 1998 for
APT specimen preparation. In the first method, ions are scanned in an annular
pattern and the pattern size is subsequently made smaller and smaller until a sharp
needle is formed. A second method involves a sort of whittling away at the sample
until a needle is formed. The first method has become the standard method and is
used by most of the groups around the world; however, neither of these methods
allowed for site-specific sample preparation. About a half-dozen years later, in the
mid-2000s, there began a capitalization on the transmission electron microscopy
(TEM) preparation techniques that were being developed in the FIB [70]. These
became known as lift-out methods. Lift-out methods required a micromanipulator,
either external or internal to the FIB, which was used to pick up and attach extracted
specimen pieces onto sample receptacles (carriers).
There are early patent applications from Imago in 2001 which propose removing
a piece from an arbitrary bulk material and then using the FIB for shaping purposes
[71]. In 2005, Miller and colleagues [72, 73] at Oak Ridge National Laboratory
showed that this was feasible, and then in 2006 work by Thompson and colleagues
[74] laid the basic groundwork for the techniques being used today. It was found
early on in the efforts that finishing specimens with a low-ion-energy milling step
was very critical to preparing quality samples [75]. Next came the dual-beam FIB,
which is the current state of the art and enables very-high-quality imaging of the
specimen with electrons (no ion damage), while the FIB is used to shape the sample.
The results of successful application of FIB-based specimen preparation methods
may be seen in the majority of the application examples presented in Chap. 7.
1.5 Concluding Remarks
The intent of this brief overview and history of APT is to prepare the reader for
exploration of more intimate details of the technique, especially as they pertain to
operation of the LEAP. The remainder of the book is dedicated to that task. Chapter 2
covers specimen preparation. Electropolishing as well as FIB-based methods
are discussed in sufficient detail to aid novice and advanced practitioners alike.
A section discussing considerations for hybrid TEM/APT is also included since this
is an area of active interest. Chapter 3 covers instrumentation specifically as it
pertains to the LEAP platform. First, the basic physics that governs the atom probe
technique is described followed by discussions related to performance and instru-
ment design philosophy. Then, specific components and systems on the LEAP are
described in detail. Chapter 4 deals with data collection. Topics and specific
checklists have been included to enhance APT experimental design and guide
LEAP users through the steps necessary to operate the LEAP. Chapter 5 details
the current reconstruction algorithm and its options (conversion of raw data into 3D
position and chemical information), provides a discussion to aid users in parameter
selection, and includes a case study to bring some substance to the reconstruction
discussion. Chapter 6 delves into the details of the analysis software Integrated
Visualization and Analysis Software (IVAS). Various features are discussed, and
some of the more complicated functions are described in detail. Finally, Chapter 7
includes summaries of published APT applications highlighting the strength of this
technique to provide nanoscale composition for a variety of material systems.
Our hope is that by gaining an appreciation for the evolution of APT and the
many contributions made by the forbearers of the technique, the present-day
practitioners will be better prepared to apply APT to new material systems and
continue the successful advance of the technique in the decades to come.
References
1. Muller, E.W., Panitz, J.A., McLane, S.B.: The atom-probe field ion microscope. Rev. Sci.
Instrum. 39, 8386 (1968)
2. Panitz, J.A.: The 10 cm atom probe. Rev. Sci. Instrum. 44(8), 10341038 (1973)
3. Cerezo, A., Godfrey, T.J., Smith, G.D.W.: Application of a position-sensitive detector to atom
probe microanalysis. Rev. Sci. Instrum. 59(6), 862866 (1988)
4. Kelly, T.F., Camus, P.P., Larson, D.J., Holzman, L.M., Bajikar, S.S.: On the many advantages
of local-electrode atom probes. Ultramicroscopy 62, 2942 (1996)
5. Kelly, T.F., Larson, D.J.: Local electrode atom probes. Mater. Char. 44(12), 5985 (2000)
6. Nishikawa, O., Kimoto, M.: Toward a scanning atom probe - computer simulation of electric
field. Appl. Surf. Sci. 76/77, 424430 (1994)
7. Larson, D.J., Prosa, T.J., Lawrence, D., Geiser, B.P., Jones, C.M., Kelly, T.F.: Atom probe
tomography for microelectronics. In: Haight, R., Ross, F., Hannon, J. (eds.) Handbook of
Instrumentation and Techniques for Semiconductor Nanostructure Characterization, vol. 2, pp.
407477. World Scientific Publishing, London (2011)
8. Kelly, T.F., Larson, D.J.: The second revolution in atom probe tomography. MRS Bull. 37,
150158 (2012)
9. Bostel, A., Blavette, D., Menand, A., Sarrau, J.M.: Toward a tomographic atom-probe. Colloq.
Phys. 50(C8), 501506 (1989)
10. Thomson, J.J.: Phil. Mag. 44, 293 (1897)
11. Rutherford, E.: The scattering of and particles by matter and the structure of the atom. Phil.
Mag. 21(6), 669678 (1911)
12. Bohr, N.: Phil. Mag. 26, 125 (1913)
13. Bragg, W.L.: The diffraction of short electromagnetic waves by a crystal. Proc. Camb. Phil.
Soc. 17, 4357 (1913)
References 21
14. Fowler, R.H., Nordheim, L.: Electron emission in intense electric fields. Proc. Roy. Soc. Lond.
A 119, 173181 (1928)
15. Muller, E.W.: Z. Tech. Phys. 17, 412 (1936)
16. Gomer, R.: Field Emission and Field Ionization. Harvard University Press, Cambridge, MA
(1961)
17. Muller, E.W.: Z. Phys. 120, 270 (1943)
18. Muller, E.W.: Z. Phys. 131(1), 136142 (1951)
19. Muller, E.W.: Resolution of the atomic structure of a metal surface by the field ion microscope.
J. Appl. Phys. 27(5), 474476 (1956)
20. Melmed, A.J.: Recollections of Erwin Mullers laboratory: the development of FIM
(19511956). Appl. Surf. Sci. 94/95, 1725 (1996)
21. Barofsky, D.F., Muller, E.W.: Mass spectrometric analysis of low temperature field evapora-
tion. Surf. Sci. 10(2), 177196 (1968). doi:10.1016/0039-6028(68)90018-6
22. Panitz, J.A.: Anecdotes from an atom-probe original. Microsc. Microanal. 4(S2), 7475 (1998)
23. Poschenrieder, W.P.: Multiple-focusing time-of-flight mass spectrometers. Part II. TOFMS
with equal energy acceleration. Int. J. Mass. Spectrom. Ion Phys. 9(4), 357373 (1972)
24. Falcone, R., Tirrell, M., Adesida, I., Bonnell, D., Broholm, C., Crabtree, G., Gibson, M.,
Hawker, C., Idzerda, Y., Jeelani, S., Murray, C., Phillips, J., Sinclair, R., Spence, J.: Devel-
oping a vision for the infrastructure and facility needs of the materials community: report of
NSF materials 2022(A Subcommittee of the Mathematical and Physical Sciences Advisory
Committee). In. US National Science Foundation (2012)
25. Mamyrin, B.A., Karataev, V.I., Shmikk, D.V., Zagulin, V.A.: The mass-reflectron, a new
nonmagnetic time-of-flight mass spectrometer with high resolution. Sov. Phys. JETP 37,
45 (1973)
26. Miller, M.K.: Presentation at the Microbeam Analysis Society Annual Meeting, Albuquerque,
NM (1986)
Plenum Publishers, New York, NY (2000)
28. Kelly, T.F., Mancini, D.C., McCarthy, J.J., Zreiba, N.A.: Proposed configurations for a high-
repetition-rate position-sensitive atom probe. Surf. Sci. 246(13), 396407 (1991)
29. Kelly, T.F., Zreiba, N.A., Howell, B.D., Bradley, F.G.: Energy deposition and heat transfer in a
pulse-heated field emission tip at high repetition rates. Surf. Sci. 246(13), 377385 (1991)
30. Larson, D.J., Camus, P.P., Kelly, T.F.: Simulated electron beam trajectories toward a field ion
microscopy specimen. Appl. Surf. Sci. 67(14), 473480 (1993)
31. Spindt, C., Brody, I., Humphrey, L., Westerberg, E.R.: J. Appl. Phys. 47, 5248 (1976)
32. Bunton, J.H., Olson, J.D., Kelly, T.F.: Phase I technical report for grant 0340351, US National
Science Foundation (2005)
33. Panayi, P.: Curved reflectron. Patent UK 0509638.3 and US 60/682,863
34. Nishikawa, O., Akimoto, T., Tsuchiya, T., Yoshimura, T., Ishikawa, Y.: Field ion microscope
with photostimulable phosphor screennew approach for quantitative analysis of FIM
images. Appl. Surf. Sci. 7677, 359366 (1994)
35. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.C.,
Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data from a
commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383 (2004)
36. Bajikar, S.S., Larson, D.J., Kelly, T.F., Camus, P.P.: Magnification and mass resolution in
local-electrode atom probes. Ultramicroscopy 65, 119129 (1996)
37. Sijbrandij, S., Larson, D.J., Miller, M.: Atom probe field ion microscopy of high resistivity
materials. Microsc. Microanal. 4(2), 9091 (1998)
38. Kellogg, G.L., Tsong, T.T.: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys. 51
(2), 11841194 (1980)
39. Kellogg, G.L.: Determining the field emitter temperature during laser irradiation in the pulsed
laser atom probe. J. Appl. Phys. 52(6), 53205326 (1981)
40. Cerezo, A., Grovenor, C.R.M., Smith, G.D.W.: Pulsed laser atom probe analysis of semicon-
ductor materials. J. Microsc. 141(2), 155170 (1986)
41. Kellogg, G.L.: Temperature dependence on the silicon field evaporation voltage. Surf. Sci.
124, L55L59 (1982)
42. Gault, B., Vurpillot, F., Vella, A., Gilbert, M., Menand, A., Blavette, D., Deconihout, B.:
Design of a femtosecond laser assisted tomographic atom probe. Rev. Sci. Instrum. 77, 043705
(2006)
43. Bunton, J.H., Olson, J.D., Lenz, D.R., Kelly, T.F.: Advances in pulsed-laser atom probe:
instrument and specimen design for optimum performance. Microsc. Microanal. 13, 418427
(2007)
44. Vurpillot, F., Houard, J., Vella, A., Deconihout, B.: Thermal response of a field emitter
subjected to ultra-fast laser illumination. J. Phys. D Appl. Phys. 42, 125502 (2009)
45. Vella, A., Vurpillot, F., Gault, B., Menand, A., Deconihout, B.: Evidence of field evaporation
assisted by nonlinear optical rectification induced by ultrafast laser. Phys. Rev. B 73, 165416
(2006)
46. Liddle, J.A., Norman, A., Cerezo, A., Grovenor, C.R.M.: Application of position sensitive
atom probe to the study of the microchemistry and morphology of quantum well interfaces.
Appl. Phys. Lett. 54, 1555 (1989)
47. Grovenor, C.R.M., Cerezo, A.: Atom probe studies of the composition of low-temperature
oxides on (100) silicon and gallium arsenide surfaces. J. Appl. Phys. 65, 5089 (1989)
48. Mackenzie, R.A.D., Liddle, J.A., Grovenor, C.R.M.: Compositional homogeneity of
metalorganic chemical vapor deposition grown III-V compound semiconductor epilayers.
J. Appl. Phys. 69, 250 (1991)
49. Lauhon, L.J., Adusumilli, P., Ronsheim, P., Flaitz, P.L., Lawrence, D.: Atom-probe tomogra-
phy of semiconductor materials and device structures. MRS Bull. 34(10), 738743 (2009)
50. Larson, D.J., Cerezo, A., Juraszek, J., Hono, K., Schmitz, G.: Atom-probe tomographic studies
of thin films and multilayers. MRS Bull. 34(10), 732737 (2009)
51. Larson, D.J., Alvis, R.A., Lawrence, D.F., Prosa, T.J., Ulfig, R.M., Reinhard, D.A.,
Clifton, P.H., Gerstl, S.S.A., Bunton, J.H., Lenz, D.R., Kelly, T.F., Stiller, K.: Analysis of
bulk dielectrics with atom probe tomography. Microsc. Microanal. 14(S2), 12541255 (2008)
52. Chen, Y.M., Ohkubo, T., Kodzuka, M., Morita, K., Hono, K.: Laser-assisted atom probe
analysis of zirconia/spinel nanocomposite ceramics. Scr. Mater. 61, 693696 (2009)
53. Ohkubo, T., Chen, Y.M., Nogiwa, K., Nishimura, A., Hono, K.: Quantitative femtosecond
laser atom probe analysis of rare earth magnets. Microsc. Microanal. 15(S2), 310311 (2009)
54. Marquis, E.A., Yahya, N.A., Larson, D.J., Miller, M.K., Todd, R.I.: Probing the improbable:
imaging carbon atoms in alumina. Mater. Today 13(10), 4244 (2010)
55. Kelly, T.F., Nishikawa, O., Panitz, J.A., Prosa, T.J.: Prospects for nanobiology with atom-
probe tomography. MRS Bull. 34(10), 744749 (2009)
56. Prosa, T.J., Greene, M., Kelly, T.F., Fu, J., Narayan, K., Subramaniam, S.: Atom probe
tomography of mammalian cells: advances in specimen preparation. Microsc. Microanal. 16,
482483 (2010)
57. Liu, J., Wu, C.-w., Tsong, T.T.: Measurement of the atomic site specific binding energy of
surface atoms of metals and alloys. Surf. Sci. 246(13), 157162 (1991)
58. Bunton, J.H., Olson, J.D., Lenz, D.R., Larson, D.J., Kelly, T.F.: Optimized laser thermal
pulsing of atom probe tomography: LEAP 4000X. Microsc. Microanal. 16(S2), 1011 (2010)
59. Blavette, D., Al Kassab, T., Cadel, E., Mackel, A., Vurpillot, F., Gilbert, M., Cojocaru, O.,
Deconihout, B.: Laser-assisted atom probe tomography and nanosciences. Int. J. Mater. Res.
99(5), 454460 (2008). doi:10.3139/146.101672
60. Bostel, A., Renaud, L., Yavor, M., Deconihout, B.: Sonde tomographique grand angle a haute
resolution. Cameca Soc Par Actions Simplif April 2009: FR2922350
References 23
61. Yavor, M., Bostel, A.: Sonde atomique a haute acceptance. Cameca May 2009: WO 2009/
065938
62. Waugh, A.R., Richardson, C.H., Jenkins, R.: APFIM 200 - a reflectron-based atom probe. Surf.
Sci. 266, 501505 (1992)
63. Cerezo, A., Godfrey, T.J., Sijbrandij, S., Smith, G.D.W., Warren, P.J.: Performance of an
energy compensated three-dimensional atom probe. Rev. Sci. Instrum. 1(69), 4958 (1998)
64. Sijbrandij, S., Cerezo, A., Godfrey, T.J., Smith, G.D.W.: Improvements in the mass resolution
of the three-dimensional atom probe. Appl. Surf. Sci. 94/95, 428433 (1996)
65. Cerezo, A., Clifton, P.H., Lozano-Perez, S., Panayi, P., Sha, G.: Overview: recent progress in
three-dimensional atom probe instruments and applications. Microsc. Microanal. 13, 408417
(2007)
66. Clifton, P.H., Gribb, T.J., Gerstl, S.S.A., Ulfig, R.M., Larson, D.J.: Performance advantages of
a modern, ultra-high mass resolution atom probe. Microsc. Microanal. 14(S2), 454455 (2008)
67. Alexander, K.B., Angelini, P., Miller, M.K.: Precision Ion Milling of Field-Ion Specimens.
J. Phys. Colloq. 50(C8), 549554 (1989)
68. Larson, D.J., Foord, D.T., Petford-Long, A.K., Anthony, T.C., Rozdilsky, I.M., Cerezo, A.,
Smith, G.D.W.: Focused ion-beam milling for field-ion specimen preparation: preliminary
investigations. Ultramicroscopy 75, 147159 (1998)
69. Larson, D.J., Foord, D.T., Petford-Long, A.K., Liew, H., Blamire, M.G., Cerezo, A., Smith,
G.D.W.: Field-ion specimen preparation using focused ion-beam milling. Ultramicroscopy 79,
287293 (1999)
70. Giannuzzi, L.A., Drown, J.I., Brown, S.R., Irwin, R.B., Stevie, F.A.: Focused ion beam milling
and micromanipulation lift-out for site specific cross-section TEM specimen preparation.
Mater. Res. Soc. Symp. Proc. 480, 1927 (1997)
71. Kelly, T.F., Martens, R.L., Goodman, S.L.: Methods of sampling specimens for microanalysis.
United States Patent 6,576,900, June 10, 2003
72. Miller, M.K., Russell, K.F., Thompson, G.B.: Strategies for fabricating atom probe specimens
with a dual beam FIB. Ultramicroscopy 102, 287298 (2005)
73. Miller, M.K.: Sculpting needle-shaped atom probe specimens with a dual beam FIB. Microsc.
Microanal. 11(S2), 808 (2005)
74. Thompson, K., Lawrence, D.J., Larson, D.J., Olson, J.D., Kelly, T.F., Gorman, B.: In-situ site-
specific specimen preparation for atom probe tomography. Ultramicroscopy 107, 131139
(2007)
75. Thompson, K., Gorman, B.P., Larson, D.J., van Leer, B., Hong, L.: Minimization of Ga
induced FIB damage using low energy clean-up. Microsc. Microanal. 12(S2), 1736CD (2006)
Chapter 2
Specimen Preparation
2.1 Introduction
In 1965 Muller, the father of the atom probe, once observed, . . . the specimen
itself is also the image-forming lens and the imaging ion beams originate at the
specimen surface [1]. In other words, the specimen is the primary optic of the atom
probe microscope, and as such, specimen preparation forms a critical step in a
successful atom probe tomography (APT) analysis. Controlling the tip size and its
shape is essential for understanding and manipulating the ionic trajectories from
specimen surface to detector.
The motivation for nearly every specimen prepared for APT analysis is funda-
mentally the same: (1) the specimen must be sharp enough to allow field evapora-
tion, (2) it must be robust enough to allow for significant evaporation events to
occur, and (3) user-defined features of interest must be present in the near-apex
region to ensure that such features are included in the collected data. Two main
methods of specimen preparation are prevalent for APT: electropolishing and
focused ion beam (FIB) milling, which are the focus of this chapter.
2.2 Electropolishing
There may be several reasons for electropolishing specimens. The method is

relatively easy, often fast, and does not require expensive and complicated instru-
mentation. Often the user may not be interested in any specific feature but rather a
more general survey of what is present in the microstructure (bulk analysis).
Alternatively, the material may not be of interest at all but may serve as a generic
test specimen that is necessary to evaluate some aspect of the LEAP instrument.
Until relatively recently (late 1990s), nearly all specimens prepared for APT
were formed by electropolishing. Standard methods were transferred from electron
microscopy [2] and applied to metals (though not exclusively) beginning with the
26 2 Specimen Preparation
Fig. 2.1 Example of metal

blanks with typical
dimensions used for
electropolishing
inception of high field work in the 1930s. General information on the techniques
can be found in any reference book on the atom probe technique [39], and a large
number of options for electropolishing and chemical polishing solutions may be
found [2, 10].
If bulk samples do not already exist as wires, then blanks must first be cut from
the bulk into appropriately sized pieces (~0.5 0.5 1020 mm3 is common).
This can be accomplished by using a diamond or a rotating wire saw or by electro-
discharge machining. An example of two finished electropolishing blanks is shown
in Fig. 2.1. It is important that the cross section is square or circular in nature as
asymmetric cross sections will lead to blade-shaped specimens that distort the ion
trajectories toward the detector.
The first electropolishing step involves removing material from the middle of a
blank in order to thin that region for separation and to form two roughly needle-
shaped specimens. This is done using a simple setup as shown schematically in
Fig. 2.2a. The blank is connected to a positive DC voltage (although AC is also
often employed), and a metal electrode is connected to the negative polarity and
placed inside a container with a dense inert liquid (such as Galden PFPE heat
transfer fluid) and a layer of electrolyte as shown in the schematic. The electrolyte
layer should be just thick enough to be continuous. The inert liquid is optional,
but dramatically reduces the volume of electrolyte required, and thus lowers storage
and disposal costs. For many ferrous or nickel-based metals and alloys, the standard
electropolishing solution used in this step consists of 10 % perchloric acid in
acetic acid by volume (the user is referred to standard reference books for other
electropolishing and chemical polishing solutions for various materials) [3].
The voltage is then switched on, and the specimen is manually moved up and
down as indicated by the dashed arrow in the figure. During this stage, the sample
may be easily held in locking tweezers or held by other mechanical means, while
the sample blank thins down from its original diameter (the largest amount of
thinning occurs in the center of the blank) and eventually separates into two pieces
2.2 Electropolishing 27
Fig. 2.2 Schematic of two types of electropolishing setups. (a) The suspended electrolyte setup
consists of a relatively narrow film of active electrolyte floating on an inert liquid. A counter
electrode is connected to the specimen blank via a potential difference. The specimen can be
manipulated to control the electropolishing action (dashed vertical arrow). The process ends when
the center of the blank necks until breaking (two-piece image immediately right of panel). This
setup is typically used for rough polishing. (b) An alternative setup that provides the user with finer
control includes a polishing loop that holds the electrolyte and is connected to the specimen via a
potential difference which can be manipulated by motion of the specimen (dashed horizontal
arrow) and gating of the electrical potential (switch)
as shown in Fig. 2.2a, right. At the completion of this process, both samples should
be rinsed in distilled water and ethanol to remove any residual electrolyte and
prevent deposition of any residue that might be dissolved in water remaining on the
tip surface. The blank material can then be mounted into a small copper (or other
metal) tube by crimping to enable easier handling and transfer for APT analysis.
It should be noted however that after completion of this coarse polishing process,
the specimens may not be sufficiently sharp for APT. This is accomplished in a
second stage (often referred to as micropolishing), which is shown schematically in
Fig. 2.2b.
The micropolishing stage forms the specimen into its final shape and may be
accomplished in a variety of ways. The application of this step requires a dynamic
application of voltage between electrolyte and specimen based on observed pro-
gression of the tip shape by the user [10]. The necking method, already described in
the preceding paragraph, can also be used during the micropolishing stage but at a
finer scale and with a smaller electrical potential. This procedure is used with the
aid of an optical microscope, as shown in Fig. 2.3. The specimen, which has a very
small shank angle, is positioned within a metal loop (negative polarity) filled with
electrolyte as shown schematically in Fig. 2.4. A very fine neck is created in the
specimen by moving the metal loop or the specimen back and forth (or up and
down, depending on the orientation of the apparatus) in the loop and leaving a small
region of the specimen outside of the loop (arrowed), Fig. 2.4b. For the final step of
the procedure, the specimen is positioned as shown in Fig. 2.4c (with the loop well
away from the necked region), the potential is switched on, and the specimen is
Fig. 2.3 Photograph of an actual electropolishing setup (also see Fig. 2.2b). The principal
components include microscope, power supply, and switch (specimen and electropolishing loop
are located beneath the microscope optic)
Fig. 2.4 Schematic illustrating evolution in the micropolishing procedure. (a) Initial placement of
the electropolishing loop on a rough-polished specimen. (b) Typical motion of the loop (dashed
arrows) which produces necking near the end of the tip (arrow). (c) Continued motion of the loop
during application of the electric potential (dashed arrow) causing further necking near the end of
the tip. (d) Final tip shape after the process has caused breaking at the neck
2.3 Needles Versus Microtips 29
quickly withdrawn through the loop (dashed arrow in Fig. 2.4c). If the neck is not
removed, the potential is switched off, the specimen is returned to the position
shown in Fig. 2.4c, and the procedure is repeated until the neck is removed, leaving
a completed specimen shape, Fig. 2.4d. Generally, the specimen is sharpest just
after the neck has been removed, but this can be sharper than ideal for some
low-evaporation-field materials. Additional iterations, similar to those described
in Fig. 2.4c, can result in controlled blunting of the apex until the desired curvature
is attained. Again, the specimen should be rinsed in distilled water and ethanol after
the final shape is achieved. Following the end of APT data collection (either by
blunting or fracture), needle specimens can often be repolished many times using
the steps outlined in this micropolishing procedure.
In addition to the standard manual techniques, automated commercial instru-
ments have emerged [11]. Simplex Scientific (Middleton, WI, USA) offers an
instrument called the Electropointer that automatically prepares needle-shaped
specimens with minimal user input. Once the starting wire or blank is inserted
into the apparatus, the instrument automatically controls the voltage and current
is applied to a small wire loop surrounding the specimen normal to its long axis.
The loop is moved longitudinally by a stepper motor. All motions and electrical
signals are controlled and monitored by computer. There is also an integrated light
optical microscope that displays the magnified specimen shape during operation.
The intent of this instrument is to take as input a coarse specimen blank and output a
specimen that is ready for APT analysis. Feasibility for this approach has been
shown for simple metals [11].
2.3 Needles Versus Microtips
The implementation of a local counter electrode in APT enables the use of microtip
arrays or coupons as a means to quickly manufacture extracted material
into APT-compatible specimens, an example of which is shown in Fig. 2.5.
Microtip coupons offer numerous advantages over traditional needle geometry
specimens [12]. The presence of many microtips per coupon makes it easier to
collect a statistically significant number of datasets in a short amount of time, under
very similar processing conditions, and without the need to cycle new specimens
into the vacuum chamber. This improvement of speed and precision not only
improves time to knowledge for typical applications, but it also allows implemen-
tation of a design-of experiment methodology.
The microtip coupon shown in Fig. 2.5 contains an array of microtip specimen
carriers oriented normal to the wafer surface. This provides a common height for all
the tips and offers two additional advantages: (1) APT analysis through multilayer
films is normal to the wafer surface (each carrier tip has exactly the same orienta-
tion, whereas a group of needles are not guaranteed to be identically straight), and
(2) it minimizes ion beam realignment when moving among specimens when FIB
Fig. 2.5 SEM image of a microtip array (coupon). The coupons are 3 mm 7 mm with Eleven
carrier microtips positioned along two of the edges (22 carrier tips total). Three additional large
markers are visible more toward the middle of the coupon. These indicate sides 1 and 2 for
reference purpose
sharpening multiple specimens (required when carrier tips are not identical in
orientation and height).
Pre-sharpened microtips [13], with a tip radius of curvature 50 nm, offer an
additional advantage in that they do not require FIB sharpening. This array type is
designed to be run as a calibration standard for evaluating instrument performance
or to serve as a substrate for deposition of films for APT analysis. Elimination of the
need for any FIB processing also alleviates the detriments associated with gallium
contamination and knock-on or ion-mixing damage of the stack during FIB sharp-
ening [14] but with the caveat that the films are deposited on a highly curved
surface. Furthermore, the samples are ready for analysis immediately upon com-
pletion of film deposition.
2.4 Electrostatic Discharge Considerations
The sharp tip geometry that makes it possible to produce electric fields sufficient to
field evaporate material from APT specimen surfaces also makes the tips suscep-
tible to undesirable electric field-induced issues during handling and transfer.
Electrostatic discharge (ESD) is a single-event, rapid transfer of electrostatic charge
between two objects, usually resulting when two objects at different potentials
come into direct contact with each other. ESD can also occur when a high
electrostatic field develops between two objects in close proximity. ESD is a
common problem in the microelectronics industry, and well-known steps can be
taken to minimize the potential for transfer of static charges to an APT specimen
and to protect insulating specimens from natural fields and voltage differences that
2.4 Electrostatic Discharge Considerations 31
Fig. 2.6 TEM images of a tungsten tip that has been modified due to electrostatic discharge.
(a) STEM-HAADF image showing amorphous WOx layer surrounding W core material deposited
through an ESD event (reprinted with permission from Cambridge University Press [16]).
(b) Stainless-steel specimen tip as polished (upper) and after ESD (lower). The outermost part
of the tip has disappeared and an amorphous outgrowth formed on the truncated tip (reprinted with
permission from IOP [17])
may cause damage [15]. Besides destruction of the APT specimen from a discharge
event between a specimen tip and its surroundings, ESD has also been observed to
initiate blunting, sometimes in conjunction with the growth of amorphous material
[1618], Fig. 2.6.
Figure 2.7 shows TEM images of specimens that have not been carefully
handled to minimize the potential for ESD. In Fig. 2.7a, a thick (>50 nm) WOx
layer is seen to have formed on the specimen surface, which is inconsistent with
normal growth expectations based on reactions of clean tungsten with air. In
Fig. 2.7b, we see an example where a substantial region of oxide has formed at
the end of the tip, consuming the previously FIB-targeted region of interest (ROI) in
the specimen. Since most ESD events that result in tip morphology changes are not
observable visually, good ESD tip handling protocols must be used. Continuous
grounding of specimen holders and tweezers during transfer is common sense to
help prevent ESD. When implemented, good practices can have a profound effect
on the ability to manufacture and image APT specimens.
Some consideration must also be taken when shipping and transferring speci-
mens. Non-ESD-compatible plastics can collect large amounts of static charge on
the surface, which is difficult to dissipate and should be avoided. These charges
may promote ESD by being transferred by non-grounded human touch to the
specimen. Conducting specimen carrier covers and tweezers that are grounded
during transfer provide the best protection. Conductive plastic carriers with anti-
static plastic covers have also proven adequate to protect STM tips and APT
microtip arrays during transport.
Fig. 2.7 Examples of ESD damage. (a) TEM image of a W tip that has been modified due to
electrostatic discharge that results in the growth of WOx. Regions of pure W (dark) and WOx
(light) are observed (reprinted with permission from Cambridge University Press [16]).
(b) Substantial oxide grown on a ferritic alloy after site-specific specimen preparation had been
performed (courtesy D. W. Saxey, University of Western Australia)
2.5 Focused Ion Beam Methods
Using ions to assist in the preparation of specimens for APT goes back nearly
40 years [19]. Broad ion beams have been used by a variety of researchers [2025]
to assist in specimen preparation, and early efforts were also made using FIB
[26, 27], although the capability of these early instruments was not good enough
for them to be broadly applied. The current generation of FIB instruments devel-
oped in the 1990s, however, has brought an entirely new capability to APT
specimen preparation methodology. The combination of the capability to image
specimens (using ions as the probe and secondary or backscattered electrons as the
detected signal) with the capability to remove substantial amounts of material using
ions provided something never before available: the ability to observe the features
of interest in specimens at high magnification while they are being sharpened. Over
the last decade or so, this capability has drastically modified the volume of
applications to which APT may be successfully applied.
The earliest FIB-based methods used to sharpen specimens for APT relied on
attaching a volume of material to the end of a needle [24, 28, 29] (using non-FIB
methods) and shaping the end form for APT analysis with a FIB [14, 3032]. Opti-
mal annular milling methods were gradually developed that provided for improved
tip shapes [12, 3336]. These methods not only allow the user to see the majority of
the specimen apex volume during preparation but are also generally applicable to
almost all materials. If care is taken during the final steps of preparation, minimal
2.5 Focused Ion Beam Methods 33
ion-induced damage will be present in the specimen ROI [37]. APT lift-out
methods [36, 3841] were adapted from transmission electron microscopy tech-
niques [42]. These methods enabled removal of a small region of material, while
being viewed in the FIB, from nearly any starting structure. This small region of
material eventually forms the apex region of the specimen.
Today, many applications and variations of the standard lift-out and sharpening
methods have been reported [36, 4348]. These include variations enabling analysis
parallel to the original specimen surface (cross-section orientation) [49] or inverted
relative to the original specimen surface (backside orientation) [50, 51]. Details of
these methods are described below.
2.5.1 Capping Considerations and Damage
In the context of APT, capping refers to the application of a sacrificial layer of

material to the original surface of a sample. The primary purpose of this procedure
is to protect the specimen from the gallium ions used during the lift-out process and
subsequent sharpening, but there are a number of secondary considerations as well.
These include (1) adequate adhesion of the capping layer to the specimen surface;
(2) material properties which enhance control of the specimen manufacturing
process (i.e., grain size, relative sputtering rate, imaging contrast); (3) potential
mass spectrum peak interferences; and (4) and evaporation field requirements. Each
of these is discussed below.
An illustration of the gallium ion implantation due to FIB milling is shown in
Fig. 2.8. Here, a doped silicon wafer has been prepared using standard lift-out
methods and annular milling techniques (described below). The final milling stage
was performed on separate specimens using 30 and 5 keV gallium ions to investi-
gate gallium ion implantation effects. Figure 2.8a shows an atom map from a
specimen shaped with 5 keV gallium ions. As the analysis proceeds, the gallium
implantation is greatly reduced as the analysis becomes farther removed from the
original specimen surface. Figure 2.8b compares the penetration depth of 30 and
5 keV ions (as analyzed in the LEAP). The implantation region is reduced to
~510 nm for the 5 keV case. When using 2 keV ions for the final preparation
stage, almost no residual gallium ions remain [37, 39]. Figure 2.8c shows a thermal
scale representation (20 % gallium is shown in red) of the gallium implantation into
SiN following a 5 keV final milling step. Note that this specimen was capped a
second time immediately following the FIB sharpening step in order to bury the
final milled surface for APT analysis. For comparison, simulations of gallium ion
implantation into silicon [52] using either 30 or 5 keV ions are shown in Fig. 2.9a, b,
respectively. Figure 2.9c quantifies the simulated implantation and suggests that, at
least for FIB milling of silicon, one should always plan to remove ~20 nm of
material prior to analyzing an undamaged ROI.
APT is a destructive technique (material is removed as the analysis proceeds),
and so damaged regions of the specimen can be removed prior to data collection
Fig. 2.8 Typical gallium implantation caused by FIB milling at 5 and 30 keV. (a) Atom map
obtained from APT of a 5 keV-shaped silicon specimen. The dark dots represent individual
gallium atoms. The dashed lines represent the analysis volume used to produce the atomic gallium
composition as a function of depth shown in (b) (reprinted with permission from Elsevier [39]). (b)
A comparison of the gallium composition profile with depth for a specimen shaped with 5 and
30 keV gallium ions (reprinted with permission from Elsevier [39]). (c) Gallium implantation into
SiN following a 5 keV final milling step
and only undamaged volumes are typically retained for analysis. It is important to
minimize gallium implantation as much as possible as damage may lead to
intermixing of phases and regions of different compositions or even turn crystalline
regions amorphous [30, 31, 36]. The damage may also promote the creation of
regions with decreased structural integrity, increasing the chances that a specimen
will fail prematurely under application of high electric field. In either case, mini-
mizing gallium damage in a finished APT specimen is desirable.
For specimens where the original surface material is part of the ROI, it is
desirable to retain some of the capping material in the final specimen shape for
analysis to ensure damage-free analysis volumes and a buried interface. This
requires a cap with strong interfacial adhesion to the native surface [53]. For
microelectronics applications where silicon is the native surface material, common
seed layer materials (which naturally have good adhesion) are good sources as a
protective cap and include nickel, chromium, titanium, and silicon. Other criteria
should be considered as well and are discussed below, but if the specimen
cap/surface interface does not survive application of high electric field, then other
considerations are moot.
Controlled ion milling properties are important for controlling the shape of the
specimen/optic. Both the grain size and relative ion sputtering rates affect the tip
shaping process. Ideally, a capping material possesses a very large grain size (or is
amorphous) so that differential ion milling rates at grain boundaries do not induce
undesired topography into the forming specimen. Extremely fine grains (much
smaller than the tip dimension) produce a similar effect. Capping materials with
sputtering rates similar to or lower than those of the specimen are also desirable.
Evaporation field [54] is another consideration. Evaporation field differences
between layers are known to cause spatial reconstruction aberrations at interfacial
regions [55, 56]. Considering a microelectronics application again, sputtered nickel
Fig. 2.9 SRIM [52] simulation of gallium ion implantation into silicon using (a) 30 keV and
(b) 5 keV ions. (c) Concentration profile normal to the original silicon surface showing the
implanted gallium ion depth
and polysilicon both have good adhesion and similar evaporation fields (less than
~10 % different) when compared to crystalline silicon and thus make reasonable
choices as capping materials. However, the gallium ion stopping power of
polysilicon is significantly less than that of nickel, and so a thicker cap of
polysilicon would be necessary to provide equivalent protection.
Ion mass peak overlap is another capping consideration. In order to easily
separate out the capping layer from the ROI, overlap of ion peaks from materials
in adjacent layers should be avoided. Materials with a small number of isotopes that
occupy unique regions of the mass spectrum are often considered for this reason
alone. Unfortunately, they commonly suffer serious limitations when considered
against the other capping considerations mentioned above. Gold has a single
isotope far separated from nearly any expected ROI mass peak but has poor
adhesion and high sputtering yield compared to most materials. Cobalt likewise
has a single isotope and is uniquely placed, especially with regard to other transition
metals (no isotopic overlaps with iron or nickel), but it does not adhere particularly
well to other materials. FIB-deposited materials such as tungsten and platinum (and
even carbon) may adhere relatively well and have mass peaks that are often beyond
the mass-to-charge range of most commonly analyzed materials, but they often
contain high levels of carbon and gallium as a consequence of the carrier gas/ion
deposition process [57] and have an extreme evaporation field, resulting in prema-
ture specimen failure.
Finally, very low deposition rates for a capping material can lead to undesirably
long deposition times. For sputter-based deposition, it is the high sputtering yield
that enables fast deposition, but during gallium milling, the same property limits tip
shaping control as discussed earlier.
Multiple capping materials may also be used to manufacture a multilayer cap
with a combination of desired properties [58, 59]. Again using the microelectronics
example, a thin nickel or chromium layer may be selected for good adhesion to a
crystal silicon surface, a layer of gold or platinum might be included for a highly
visible end-point layer (with secondary electron imaging), gold or silver may be
used because of the high deposition rate, and, finally, platinum has durability under
ion milling (sputtering rate and gallium protection).
2.5.2 Standard Lift-Out Process
For material structures where (1) the ROI can be deduced from surface features,
(2) the ROI resides near the surface, and (3) the analysis direction is intended to
proceed from the top surface down into the material (typically referred to as
top-down or normal orientation), the standard lift-out preparation is appropriate
[36, 39, 40]. Although a number of variations of the general method exist [4346,
48, 6062], the specific steps illustrated below describe the method most commonly
used by the current authors.
Figure 2.10 illustrates the general steps that the user should follow during a
standard lift-out procedure. An optional protective capping layer may be applied
over the entire specimen surface. No additional (i.e., non-FIB deposited) capping
layer is used in the current example because the ROI is sufficiently far below the
original surface of the sample. A FIB-deposited platinum strip is added to protect
the surface and to mark the region to be extracted (arrowed in Fig. 2.10a). The
platinum layer is typically 23 m wide and ~100 nm thick with a length that
depends on the geometry of the ROI and the number of desired specimens to be
made from the extracted sample region.
Material is then removed around the specified ROI to create a wedge-shaped
mass of material to be removed from the surface (Fig. 2.10b, arrows). In this case,
trench 1 is created by tilting the stage to 22 (with respect to the electron beam) and
Fig. 2.10 The steps involved in a standard FIB lift-out procedure. (a) A FIB-deposited protective
strip is placed over the region of interest. This protective material is often platinum or tungsten.
(b) The material is then removed by ion milling around three sides of the region (arrows) as well as
underneath to produce a long cantilevered wedge of material. (c) The wedge is removed by using
an in situ micromanipulator (attached to the left end of the wedge) and then cutting the wedge free
from the substrate (dashed line). (d) The micromanipulator is used to position the wedge above
the carrier microtip (plan-view). (e) The wedge is attached to the carrier tip surface with
FIB-deposited platinum (arrow) and then cut free from the carrier tip (dashed line) for transfer
to additional microtips. Once propagation of the wedge is complete, the FIB stage is rotated 180
so that a second platinum deposition can be applied to the opposite wedgepost interface of each
mounted post. (f) The final mounted wedge section is then ready for sharpening: The line of the
targeted FinFET is clearly visible along the line of the arrows (reprinted with permission from
Annual Reviews [77])
milling with a ~6 nA ion current in a ~2 m-by-10 m rectangular pattern. The ion

beam is scanned parallel to the long axis of the wedge starting far from the platinum
strip and proceeding to the near edge. Trench 2 is created using the same procedure
after rotating the stage through 180 . Milling is completed when trenches 1 and
2 meet beneath the ROI. Trench 3 is then cut using a ~12 m long rectangular
pattern of sufficient width to cut across the entire wedge leaving behind a
cantilevered wedge. The 22 tilt coupled with these wedge and trench dimensions
allows the user to observe the bottom of the trench in the SEM and better estimate
the milling time required for complete undercutting of the wedge.
After returning the FIB stage to 0 tilt, a needle on a micromanipulator is
lowered into contact with the free end of the cantilevered wedge (left edge of
Fig. 2.10c). Sufficient FIB-deposited platinum is used to secure the micromanipu-
lator to the top of the wedge. The cantilevered wedge is then cut free by milling a
~12 m long rectangular pattern again (dashed line) with width sufficient to cut
across the entire wedge.
Once the wedge and manipulator are lifted free of the bulk, the wedge can be
repositioned above a carrier tip (Fig. 2.10d). In this image, the wedge is placed
above a ~2 m diameter, flat-topped, silicon microtip post (the post is centered
within the dashed circle). The microtip posts are manufactured as a 450 m array
with each post having an overall height of ~100 m above a planar surface [13]. This
height and spacing combination is sufficient to allow for rapid transfer of wedge
material to multiple posts and to allow independent LEAP analysis of each post
without field evaporating any neighboring specimens. The wedge is carefully
lowered until it comes into direct contact with the flat surface on the top of the
post. An ~1 m-by-1 m-by 0.5 m platinum patch is deposited at the wedgepost
interface to secure the wedge to the post (Fig. 2.10e, arrow). The region of the wedge
secured to the post is then cut free, as indicated by the dashed line in Fig. 2.10e,
and the wedge is moved to the next post and the procedure is repeated until the
wedge is gone.
Once propagation of the wedge to the microposts is finished, the stage is rotated
180 so that a second platinum patch can be applied to the opposite side of the
wedgepost interface for each mounted post. The final mounted wedge is shown in
Fig. 2.10f, with the ROI located along the line indicated by the arrows. Subsequent
cleaning of the micromanipulator to remove any remaining platinum or wedge
material is recommended before reuse.
2.5.3 Sharpening Process
The required specimen dimensions for APT are material and instrument dependent,
and the user must consider this fact, as well as the voltage-range limitation at which
an atom probe can operate, when preparing specimens. If the user is planning to
analyze materials that have particularly high evaporation fields (40 V/nm), this
should be taken into account when considering the final specimen radius. In
addition, features of interest which are relatively large need to be positioned
below the initial apex of a specimen. A general rule of thumb is to prepare a
specimen which has a lateral diameter at the ROI depth position of about two times
the lateral dimension of the ROI. Clearly this limits the size of features which may
be totally contained in the field of view of a LEAP analysis.
The process of converting a lift-out wedge into a sub-200 nm diameter sharp
needle is accomplished through a series of annular milling steps [31, 33] followed
by a low-energy FIB cleanup step [37, 39], as shown in Fig. 2.11. Illustrated in
Fig. 2.11a is a linear fin-shaped field-effect transistor (FinFET) structure centered
within the wedge, shown by the arrow. Tip shaping is accomplished by applying an
annular milling pattern (shown schematically at the top of Fig. 2.11b) with constant
outer diameter (~4 m) and a decreasing inner diameter (Fig. 2.11bd) with a beam
current of ~0.28 nA. The milling proceeds from the outer diameter of the pattern to
the inner diameter of the pattern to best control any potential for redeposition of
removed material. The first pattern (Fig. 2.11b) has an inner diameter of ~1.6 m
Fig. 2.11 The steps involved in FIB sharpening process converting the wedge section in Fig. 2.10
into a specimen for APT analysis. (a) The region of interest is shown by the arrow. (b) The first
annular milling pattern shapes the tip into a cylinder, with the FinFET region clearly visible near
the (c) upper-center portion of the cylinder. The second and third milling patterns produce (c) a
tapered end and (d) a narrowed end. The final sharpening step is performed at low ion energy
(either 2 or 5 keV) and is simply a circular pattern that images the specimen end-on. During this
stage, the specimen is carefully monitored to ensure that the region of interest is positioned near
the apex of the final specimen. The (e) low- and (f) high-magnification images of the specimen that
is then ready for APT (reprinted with permission from Annual Reviews [77])
and produces a long cylindrical shape that proceeds well beyond the platinum-weld
region of the wedge (not shown). The second and third patterns (Fig. 2.11c, d) have
inner diameters of ~0.6 and ~0.3 m, respectively. Each pattern is applied for a
length of time sufficient to achieve the desired tip diameter to a location just beyond
the length of the ROI. The final tip shaping is accomplished with the low-energy
(5 keV) FIB cleanup step described below.
As mentioned previously, the implant and damage region created in silicon by a
30 keV gallium ion beam has been shown to extend significantly into the surface of
the sample, while 25 keV ions limit damage to less than ~5 nm [37]. The goal of
the low-energy (25 keV) milling step is to remove the 30 keV damaged region,
position the apex of the specimen at or slightly above the ROI, and narrow the
bottom section of the ROI to 200 nm or less in width. An annular milling pattern of
>4 m diameter is centered over the tip, and milling proceeds at a reduced beam
current (48 pA in the current example). The diameter of the pattern and beam
current can be adjusted to slow the rate of milling so that the user is able to carefully
control the stopping point. Most current FIB instruments provide live viewing of
ion milling which assists in real-time end-point control. After the final step, the
specimen shape is shown in Fig. 2.11e, f. The total milling time for the final low ion
energy step is of the order of 60 s. Small variations in total milling time can quickly
change the dimensions of the final tip shape. For example, an additional 510 s of
milling in this case would move the tip apex position below the ROI.
2.5.4 FIB Deprocessing
In the preceding sharpening example, the necessary removal of a large amount of

material (~500 nm) to reach the ROI may inhibit the capability to make a repeatable
tip shape because of the influence of differential sputtering due to interaction with
grain boundaries and/or multiple material regions. It is therefore desirable to
control the amount of material above the ROI before initiating the tip-shaping
procedure. Minimizing the capping material thickness is helpful in this regard,
but full control of the distance between the ROI and the original surface is desirable.
Likewise, a problematic region (e.g., an interface between two phases or a region of
highly different evaporation field) may exist near the ROI such that removing that
region from the tip apex with the low-energy cleanup step does not result in an
appropriate final shape. In this case, complete or partial removal of the problematic
region prior to FIB processing and the addition of a new sacrificial capping layer
can enable proper shaping of the ROI volume.
Both of the above objectives can be accomplished via FIB deprocessing
[61]. This process involves removing layers of material parallel to the original
sample surface using FIB ion beam milling in a fashion very similar to TEM
lamella production techniques [63]. An extracted wedge of material, like that in
Fig. 2.10, is first rotated by 90 along its long axis enabling milling parallel to the
original wedge surface. This rotation is usually accomplished by transferring the
wedge to a manipulator that has an axial rotation capability [49]. Similar to final tip
shaping, the SEM is used to control the end-point for the milling while performing
FIB deprocessing.
Figure 2.12a shows a highly magnified image of a transistor exposed at the edge
of an extracted wedge that has been rotated by 90 , with the exposed edge of the
wedge shown at lower magnification in Fig. 2.12b. Approximately 500 nm of
material must be removed to locate the top surface of the wedge near the gate
oxide in this particular example. The user can carefully trim any amount of material
from the surface. Figure 2.12c shows the stopping point one would choose if the top
of the silicide region was the goal of the deprocessing, while Fig. 2.12d shows the
stopping point just above the gate oxide in the transistors.
2.5.5 Cross-Section Preparation
Performing analysis along directions other than from the top surface into the bulk of
the sample can have advantages. For example, analyzing thin films parallel to the
film interface orientation (termed cross-section orientation) serves to both increase
Fig. 2.12 An example of FIB deprocessing. (a) A rotated image of a transistor region within the
wedge. (b) The original extracted wedge after it has been transferred to a manipulator that has been
rotated by 90 . (c) Specimen wedge after it has been deprocessed to a point near the silicide region
above the transistor. (d) Specimen wedge after it has been further deprocessed to a point just above
the gate oxide region (reprinted with permission from World Scientific Publishing [60])
the volume of the ROI (the film interfaces) and improve the analysis yield by
changing the orientation of the applied stress relative to the interface(s) (see Chap. 4
for a description of the stresses applied to an APT specimen). Figure 2.13 illustrates
a cross-section specimen preparation procedure where the ROI consists of layers of
material positioned at the original sample surface.
In such cases, as mentioned previously, a thick sacrificial cap is first added to the
surface of the sample to protect the ROI and add additional material to the sample
so that the surface or the near-surface ROI can be centered in the final tip [49]. In
this case, ~500 nm of nickel has been deposited on the top surface of the sample
before a standard lift-out procedure is used to create a large lift-out wedge
(~5 m wide).
After attaching the manipulator needle and extracting the wedge from the bulk
sample, the wedge is transferred to a horizontal manipulator with a rotational
degree of freedom (Fig. 2.13a). The wedge is transferred by first attaching it to
the second manipulator with FIB-deposited platinum and then cutting the original
manipulator from the wedge. While attached to the horizontal manipulator, one side
Fig. 2.13 The steps for performing site-specific lift-out and manipulation for cross-section APT
specimens. (a) The original capped specimen wedge is transferred to a manipulation probe which
allows rotation of the wedge. (b) Original wedge still in normal orientation is FIB cut to expose
cross-section ROI (see arrow) (c) Wedge is rotated by 90 , and the cross-section ROI surface is
capped ready for final preparation (ROI indicated by arrow). (d) A wedge section is mounted to a
post. (e) The mount is trimmed prior to annular milling. (f) The final tip with cross-section ROI
captured at the tip (reprinted with permission from World Scientific Publishing [60])
of the wedge is milled flat, creating a new surface that is at 90 to the original
surface, as shown in Fig. 2.13b (this will become the top surface of the final tip).
The arrow highlights a dark layer below the 500 nm nickel cap which is the location
of the ROI for this example. For site-specific cross-section preparation, this new
surface also needs to be positioned directly adjacent to (above in the final tip) the
feature of interest.
After this new surface is created, the wedge is rotated back by 90 and an
additional capping layer (~50 nm nickel) is added to the new surface, Fig. 2.13c.
Sufficient capping material on the new surface ensures that the specimen can be
shaped with 30 kV ions and finished with 5 kV ions without exposing any of the
original sample material to gallium ions. Should this second cap be removed before
final specimen preparation is complete, differential milling of the different layers
can cause the tip to form into multiple protrusions.
Once the second capping material has been deposited, the wedge is transferred
back to the original micromanipulator and propagated to microtips in a manner
similar to the standard lift-out method described above (Fig. 2.10). Figure 2.13d
shows a portion of the wedge after it has been successfully propagated to a carrier
tip. Before tip shaping can begin, an effort is made to make the tip more cylindrical.
The portion of the wedge that extended beyond the edge of the microtip is removed
with a rectangular milling pattern before sharpening (Fig. 2.13e). The final speci-
men shape after the standard annular milling and low-energy cleanup steps
described above is shown in Fig. 2.13f.
The standard annular milling process is not quite as straightforward for
cross-section preparation because the ion milling rate of the specimen is likely to
have non-cylindrical symmetry. Any differential ion sputtering between layers will
cause asymmetry in the final shape. Often the center of the milling pattern will need
to be positioned away from the center of the region in anticipation of one side of
the specimen milling at a higher rate. The proper amount of this pattern shift is
usually determined by trial and error. A capping material used on the original
surface chosen to have similar milling properties of the ROI will assist in the
mitigation of this phenomenon.
2.5.6 Backside Preparation
Analyzing materials starting from within the bulk and proceeding toward the
original sample surface (termed backside) [50, 51] can serve multiple purposes.
Should weak or otherwise problematic materials or interfaces exist between the
sample surface and the ROI, analysis from the backside may allow for these regions
to be avoided. If there is a substantial thickness of insulating material (thermal or
electrical) present beneath the ROI, backside preparation may also prove useful, as
this material can be removed from the final specimen. In addition, specimen shape
distortions, and the resultant effects on data reconstruction, can be affected by the
order of evaporation of regions with different evaporation fields. Performing
analysis both in the top-down and backside orientations can provide meaningful
information to help separate real compositional trends from artifacts [64, 65].
Depending on the location of the ROI, backside specimen preparation may
require application of significant material to the top surface of the sample so that
the overall depth of the wedge remains ~23 m after it has been rotated 180 and
formed into a sharp specimen. As in previous examples, adhesion of the added
material to the original sample surface is very important because that interface will
Fig. 2.14 An example of backside specimen preparation. The target ROI is the near-surface
region of a primarily silicon material. An ~2 m layer of sacrificial material (silver) has been added
to bury the original surface. (a) The FIB is used to cut a trench into this composite material to
expose the original surface within the depth of the sample. The interface is observed ~2 m below
the surface. (b, c) A larger-than-normal wedge is extracted from the host material with a
micromanipulator. (d) After transfer to a second micromanipulator with a rotational degree of
freedom, the wedge is rotated by 90 and the material is removed to expose the new uppermost
surface (arrow). (e) The wedge is rotated another 90 so that the uppermost surface is properly
oriented. (f) The wedge is reshaped so that the former top part of the original wedge has a
triangular shape to enable attachment via FIB-deposited platinum. (g) An image of the upside-
down, reshaped wedge after attachment to a microtip post using FIB-deposited platinum. (h) Final
tip shape after all annular milling steps have been completed
reside near the apex of the final specimen. One successful recipe used a palladium
seed layer followed by a silver layer ~3 m thick to provide the necessary thickness
for the backside procedure [51, 65]. The palladium provides good adhesion to the
top metallic surface and to the silver [66], while the use of silver as a capping layer
allows for very fast deposition of the 3 m of material and good electrical and
thermal conduction. The actual steps of the backside specimen preparation proce-
dure are similar to those for a cross-section specimen.
Figure 2.14 illustrates the backside specimen preparation process for a ROI that
resides near the sample surface. First the top surface is sufficiently buried (~2 m
silver with a palladium seed). Then, a trench is cut into the surface of the sample to
expose the depth of the material for SEM imaging. The buried interface is clearly
visible in Fig. 2.14a, confirming that the target ROI resides some 2 m below the
surface. Similar to cross-section specimen preparation, a larger-than-normal wedge
(~5 m wide) of material is extracted and transferred to a micromanipulator with
a rotational degree of freedom (Fig. 2.14b, c). The goal of the next set of steps is
to place the ROI near the top of a new wedge with triangular cross section.
To accomplish this, the wedge (Fig. 2.14c) is rotated 90 and material is removed
(arrow in Fig. 2.14d) with the FIB so that the ROI is near the new surface. Then,
the wedge is rotated another 90 so that the original wedge is upside down
(Fig. 2.14e). At this point an optional second capping layer may be added
to the newly created top surface to protect it during the annular milling process.
2.6 Hybrid Transmission Electron Microscopy/Atom Probe Tomography 45
Fig. 2.15 An example of the mortis and tenon method for wedge attachment to a carrier post.
(a) First a keyhole is milled through the object that is being attached to the carrier post (arrow).
This is commonly performed while the wedge is still attached to the micromanipulator (not
illustrated here). (b) Next the keyhole is filled with FIB-deposited platinum using a pattern of
the same size used in the milling step. (c) Because this wedge surface is oriented 90 to the carrier
surface, it is possible to make an additional FIB-deposited platinum weld to confidently secure the
lift-out to the carrier. Other wedge surface/carrier orientations may only allow for attachment
through the keyhole
To complete the wedge-shaping process, the stage is tilted to 22 with respect to the
electron column and each side of the wedge is trimmed until the triangular cross
section is formed (Fig. 2.14f). This shaping is necessary so that the FIB-deposited
platinum can access the intersection of the wedge and carrier tip with zero stage tilt.
Alternative adhesion strategies, such as the notch or mortise and tenon weld (see
Fig. 2.15) [40], can be used which do not require a triangular wedge cross section to
enable FIB adhesion. Finally, the wedge is transferred back to the mounting
micromanipulator, and the wedge material is transferred to carrier posts
(Fig. 2.14g) and annularly milled to its final tip shape (Fig. 2.14h).
2.6 Hybrid Transmission Electron Microscopy/Atom

Probe Tomography
Transmission electron microscopy (TEM) and APT each has different advantages
and limitations, but together they provide complementary information enabling
more complete characterization of the microstructure and chemistry of a materials
system. Although FIB/SEM images provide some information about the structure
of an APT specimen after the specimen fabrication process, higher resolution
Fig. 2.16 TEM and APT

analysis performed on the
same specimen volume.
(a) Bright field TEM image
of an APT specimen prior
to analysis showing a grain
boundary within the
analysis volume. (b) 3D
reconstruction of the APT
data obtained from the same
tip with oxide particles
present both along the
grain boundary and in the
bulk material (reprinted
with permission from
Elsevier [73])
characterization of specimens using TEM is useful to further increase APT recon-

struction accuracy. TEM imaging can provide specimen radius and shank angle
with high accuracy and also provide some information on the internal structure of
interfaces and precipitates.
Analytical techniques such as energy-dispersive X-ray spectroscopy and elec-
tron energy loss spectroscopy, as well as different imaging modes in scanning
TEM, can provide preliminary information about the composition of precipitates
and interfaces. In addition, TEM can be used to obtain atomic resolution structural
information that can be correlated with the compositional information obtained
from APT.
A number of groups have reported progress in developing general hardware and
methods that allow straightforward TEM and APT analysis of the same specimen
[50, 51, 64, 6774]. Although some physical limitations exist for specimen geom-
etries and holders that are compatible with commercially available TEM and APT
instruments, general wire-based and half-grid-based solutions do exist.
An example of results from TEM and APT of the same specimen is illustrated in
Fig. 2.16 [73]. In this case, an oxide dispersion-strengthened (ODS) variation of the
Eurofer 97 steel was fabricated and analyzed with high-resolution TEM followed
by APT. Oxide particles measuring 510 nm in diameter were observed with TEM,
but only APT could adequately characterize the local composition. As seen in the
figure, particles appear aligned along planar features that could be lath or twin
boundaries. Although large precipitates are obvious in the atom map, the smaller
particles (~2 nm in diameter) are more uniformly distributed with a particle density
of 3.9 0.15 1023m 3 [73].
2.6 Hybrid Transmission Electron Microscopy/Atom Probe Tomography 47
Fig. 2.17 Fixtures available to handle transfer of wire-shaped specimens. (a) Wire secured in a
1.8 mm copper sleeve. (b) SEM/FIB-compatible holder that can accommodate multiple 1.8 mm
objects. (c) 1.8 mm copper sleeve mounted within a Fischione TEM holder. (d) APT holder that
can accommodate multiple 1.8 mm objects
2.6.1 Preparation and Holders
Wire-shaped specimen geometries are common in APT and are very useful for TEM
because the approximately cylindrical specimen cross section presents a constant
sample thickness for all angles in a TEM tomography acquisition [75]. A wire itself
may be difficult to handle and subsequently mount in a holder, so the wire can
be secured in a thicker fixture such as the crimpable 1.8 mm copper sleeve
shown in Fig. 2.17a. Wires are typically electropolished to a sharp tip so that
they can be analyzed as-polished or taken into the FIB for further processing
either as a specimen or a specimen carrier. Multiple wire specimens can be
loaded into the FIB using a carrier similar to that shown in Fig. 2.17b. A set
screw is used to secure Cu stubs into each hole of the holder. After FIB
processing is complete, wires can be placed into appropriate holders for TEM
analysis. For example, the Fischione (Export, PA, USA) 2050 TEM specimen
holder and LEAP wire specimen puck are shown in Fig. 2.17c, d, respectively.
These examples demonstrate how single-wire specimens can be successfully
shared between FIB, TEM, and APT instruments with minimal difficulty.
Half-grid-shaped specimen geometries have become quite commonplace in
TEM and can be accommodated in the LEAP as well. Advantages include a
geometry where the specimen is protected by the superstructure of the grid, the
fact that multiple tips may be carried on a single grid, and the common availability
of TEM holders compatible with grids in non-APT laboratories. In Fig. 2.18, three
different half-grid-format specimen carriers are shown: (a) a tungsten needle
embedded via the Short-Cut into an Omniprobe (Dallas, TX, USA) copper
half-grid [51]; (b) a wire grid that has been cut in half and electropolished suffi-
ciently sharp for analysis or mounting of lift-out specimens; and (c) a Dune
Sciences (Eugene, OR, USA) silicon half-grid with multiple specimen carrier
regions that are compatible for mounting lift-out specimens. Examples of grid-
compatible TEM holders and their LEAP-compatible specimen pucks are shown in
Fig. 2.19. Figure 2.19a, top, shows a generic TEM grid holder and LEAP puck with
Fig. 2.18 Three half-grid-format specimen carrier options: (a) needle embedded via the Short-
Cut into an Omniprobe copper half-grid, (b) a wire grid that has been cut in half and
electropolished so that the ends are sharp, and (c) a Dune Sciences silicon half-grid with multiple
carriers (image courtesy of Dune Sciences, Eugene, OR, USA)
holder designed to hold a half-grid. The grid holder shown in the LEAP puck
(Fig. 2.19b, top) serves double duty as a grid carrier for the LEAP, but it can also be
carried into the FIB. With this approach, the user must handle the grid during some
of the transfers between FIB, TEM, and LEAP instruments. The Hummingbird
Scientific grid holder [50], Fig. 2.19a, bottom, consists of a removable end effector
that can be placed in a TEM holder or a LEAP puck. The advantage is that once a
half-grid is mounted in the end effector, it can be moved from FIB to transmission
electron microscope to LEAP without the need for handling the grid directly with
tweezers. Variations on the half-grid theme exist as well. Gorman et al. have used
the Short-Cut to host a micromanipulator needle, which serves as a specimen carrier
with the sharpened lift-out material at its apex [50]. Similar advances have been
made recently by Felfer et al. [76].
2.7 Summary 49
Fig. 2.19 TEM/APT specimen holder options. (a) TEM grid holders (top reprinted with permis-
sion from www.gatan.com; bottom reprinted with permission from Cambridge University Press
[50]) and (b) their equivalent LEAP-compatible specimen carriers. A generic half-grid handling
strategy is shown at the top of the figure, while the Hummingbird removable end effector strategy
is shown at the bottom of the figure
2.7 Summary
Electropolishing and FIB-based specimen preparation methods give APT practi-

tioners a number of options for manufacture of specimens from bulk materials. The
technologically simple and inexpensive electropolishing method is still highly
useful for metallurgical materials with a uniform distribution of features and an
established polishing recipe. FIB-based methods have recently experienced rapid
development and adoption because FIB instruments are now much more widely
available and because of the compatibility of the lift-out technique with microtips.
The capability of the local-electrode geometry to analyze microtips allows rapid,
consistent, and controlled manufacture of multiple specimens in a short period of
time. Further advantages include the ability to perform site-specific specimen
preparation and FIB deprocessing, which are crucial capabilities for expanding
the application space and success rate of APT experiments in general.
References
1. Muller, E.W.: Field Ion microscopy. Science 149(3684), 591601 (1965)

2. Thompson-Russell, K.C., Edington, J.W.: Electron Microscope Specimen Preparation Tech-
niques in Materials Science. Monographs on Practical Electron Microscopy in Materials
Science, vol. 5. Philips Technical Library, Eindhoven (1977)
Materials Problems. Materials Research Society, Pittsburgh, NJ (1989)
6. Tsong, T.T.: Atom-Probe Field Ion Microscopy: Field Ion Emission and Surfaces and Inter-
faces at Atomic Resolution. Cambridge University Press, Cambridge (1990)
8. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principles and Applications. Elsevier,
New York, NY (1969)
9. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer series
in materials science, vol. 160. Springer, New York, NY (2012)
10. Melmed, A.J.: The art and science and other aspects of making sharp tips. J. Vac. Sci. Tech. B9
(2), 601609 (1991)
11. Kostrna, S.L.P., Peterman, J.W., Prosa, T.J., Wiener, S.A., Larson, D.J., Kelly, T.F.:
An automatic electropolishing system for needle-shaped specimens. Microsc. Microanal. 12
(S2), 17501751CD (2006)
12. Larson, D., Wissman, B.D., Martens, R., Viellieux, R.J., Kelly, T.F., Gribb, T.T., Erskine, H.
F., Tabat, N.: Advances in atom probe specimen fabrication from planar multilayer thin film
structures. Microsc. Microanal. 7, 2431 (2001)
13. Thompson, K., Larson, D.J., Ulfig, R.: Pre-sharpened and flat-top microtip coupons: a quan-
titative comparison for atom-probe analysis studies. Microsc. Microanal. 11(S2), 882 (2005)
14. Larson, D.J., Petford-Long, A.K., Cerezo, A., Smith, G.D.W., Foord, D.T., Anthony, T.C.:
Three-dimensional atom probe field-ion microscopy observation of Cu/Co multilayer film
structures. Appl. Phys. Lett. 73(8), 11251127 (1998)
15. Kaiser, K.L.: Electrostatic Discharge. CRC Press, New York, NY (2005)
16. Gorman, B.P., Ballard, J., Romanes, M., Jager, D., Reidy, R.F., Randall, J.N.: Mediation of
electrostatic discharge induced morphological damage in atomically precise tips. Microsc.
Microanal. 16(S2), 480481 (2010)
17. Andren, H.-O., Henjered, A., Norden, H.: Field corrosion of field-ion microscope specimens
caused by polymer gaskets. J. Phys. E Sci. Instrum. 13, 392395 (1980)
18. Saxey, D.W.: Personal Communication (2011)
19. Walls, J.M., Southworth, H.N., Rushton, G.J.: The preparation of field electron/field ion
emitters by ion etching. Vacuum 24(10), 475479 (1974). doi:10.1016/0042-207x(74)90009-8
20. Henjered, A., Norden, H.: A controlled specimen preparation technique for interface studies
with atom-probe field-ion microscopy. J. Phys. E Sci. Instrum. 16(7), 617619 (1983). doi:10.
1088/0022-3735/16/7/014
21. Hellsing, M.: High-resolution microanalysis of binder phase in as sintered WC-Co cemented
carbides. Mater. Sci. Technol. 4(9), 824829 (1988)
22. Kvist, A., Andren, H.-O., Lundin, L.: A specimen preparation technique for atom probe
analysis of the near-surface region of cemented carbides. Appl. Surf. Sci. 94/95(14),
356361 (1996)
23. Larson, D.J., Camus, P.P., Vargas, J.L., Kelly, T.F., Miller, M.K.: Specimen preparation and
atom probe field ion microscopy of BSCCO-2212 superconductors. J. Phys. 6(C5), 271276
(1996)
References 51
24. Larson, D.J., Petford-Long, A.K., Cerezo, A.: Field ion microscopy of multilayer film devices.
Microsc. Microanal. 4(S2), 112113 (1998)
25. Larson, D.J., Russel, K., Cerezo, A.: Sharpening of field-ion specimens and positioning of
features of interest by ion-beam milling. J. Vac. Sci. Tech. 18(1), 328333 (2000)
26. Waugh, A.R., Payne, S., Worrall, G.M., Smith, G.D.W.: In situ ion milling of field ion
specimens using a liquid metal ion source. J. Phys. 45(C9), 207209 (1984)
27. Alexander, K.B., Angelini, P., Miller, M.K.: Precision ion milling of field-ion specimens.
J. Phys. Colloq. 50(C8), 549554 (1989)
28. Hasegawa, N., Hono, K., Okano, R., Fujimori, H., Sakurai, T.: A method for preparing atom
probe specimens for nanoscale compositional analysis of metallic thin films. Appl. Surf. Sci.
67(14), 407412 (1993)
29. Martens, R.L., Larson, D.J., Kelly, T.F., Cerezo, A., Clifton, P.H., Tabat, N.: Preparation of
three-dimensional atom probe samples containing multilayer thin film structures using a
focused ion-beam. Microsc. Microanal. 6, 522523 (2000)
30. Larson, D.J., Foord, D.T., Petford-Long, A.K., Cerezo, A., Smith, G.D.W.: Focused ion-beam
specimen preparation for atom probe field-ion microscopy characterization of multilayer film
structures. Nanotechnology 10, 4550 (1999)
31. Larson, D.J., Foord, D.T., Petford-Long, A.K., Anthony, T.C., Rozdilsky, I.M., Cerezo, A.,
Smith, G.D.W.: Focused ion-beam milling for field-ion specimen preparation: preliminary
investigations. Ultramicroscopy 75, 147159 (1998)
32. Larson, D.J., Petford-Long, A.K., Cerezo, A., Smith, G.D.W.: Three-dimensional atom probe
studies of metallic multilayers. Acta Mater. 47(15), 40194024 (1999)
33. Larson, D.J., Foord, D.T., Petford-Long, A.K., Liew, H., Blamire, M.G., Cerezo, A., Smith,
G.D.W.: Field-ion specimen preparation using focused ion-beam milling. Ultramicroscopy 79,
287293 (1999)
34. Larson, D.J., Petford-Long, A.K., Ma, Y.Q., Cerezo, A.: Information storage materials:
nanoscale characterization by three-dimensional atom probe analysis. Acta Mater. 52(10),
28472862 (2004)
35. Miller, M.K.: Sculpting needle-shaped atom probe specimens with a dual beam FIB. Microsc.
Microanal. 11(S2), 808 (2005)
36. Miller, M.K., Russell, K.F., Thompson, G.B.: Strategies for fabricating atom probe specimens
with a dual beam FIB. Ultramicroscopy 102, 287298 (2005)
37. Thompson, K., Gorman, B.P., Larson, D.J., van Leer, B., Hong, L.: Minimization of Ga
induced FIB damage using low energy clean-up. Microsc. Microanal. 12(S2), 1736CD (2006)
38. Kelly, T.F., Martens, R.L., Goodman, S.L.: Methods of sampling specimens for microanalysis.
United States Patent 6,576,900, June 10, 2003
39. Thompson, K., Lawrence, D.J., Larson, D.J., Olson, J.D., Kelly, T.F., Gorman, B.: In-situ site-
(2007)
40. Miller, M.K., Russell, K.F., Thompson, K., Alvis, R., Larson, D.J.: Review of atom probe
fib-based specimen preparation methods. Microsc. Microanal. 13, 428436 (2007)
41. Miller, M.K., Russell, K.F.: Atom probe specimen preparation with a dual beam SEM/FIB
miller. Ultramicroscopy 107(9), 761766 (2007)
42. Giannuzzi, L.A., Drown, J.I., Brown, S.R., Irwin, R.B., Stevie, F.A.: Focused ion beam milling
and micromanipulation lift-out for site specific cross-section TEM specimen preparation.
Mater. Res. Soc. Symp. Proc. 480, 1927 (1997)
43. Cairney, J.M., Saxey, D.W., McGrouther, D., Ringer, S.P.: Site-specific specimen preparation
for atom probe tomography of grain boundaries. Phys. B Condens. Mater 394, 267269 (2007)
44. Takahashi, J., Kawakami, K., Yamaguchi, Y., Sugiyama, M.: Development of atom probe
specimen preparation techniques for specific regions in steel materials. Ultramicroscopy 107
(9), 744749 (2007)
45. Saxey, D.W., Cairney, J.M., McGrouther, D., Honma, T., Ringer, S.P.: Atom probe specimen
fabrication methods using a dual FIB/SEM. Ultramicroscopy 107(9), 756760 (2007)
46. Perez-Willard, F., Wolde-Giorgis, D., Al Kassab, T., Lopez, G.A., Mittemeijer, E.J.,
Kirchheim, R., Gerthsen, D.: Focused ion beam preparation of atom probe specimens
containing a single crystallographically well-defined grain boundary. Micron 39, 4552 (2008)
47. Rachbauer, R., Massl, S., Stergar, E., Felfer, P., Mayrhofer, P.H.: Atom probe specimen
preparation and 3D interfacial study of Ti-Al-N thin films. Surf. Coat. Tech. 204(11),
18111816 (2010). doi:10.1016/j.surfcoat.2009.11.020
48. Felfer, P., Ringer, S.P., Cairney, J.M.: Shaping the lens of the atom probe: fabrication of
site specific, oriented specimens and application to grain boundary analysis. Ultramicroscopy
111(6), 435439 (2011)
49. Lawrence, D., Alvis, R., Olson, D.: Specimen preparation for cross section atom probe
analysis. Microsc. Microanal. 14(S2), 10041005 (2008)
50. Gorman, B.P., Diercks, D., Salmon, N., Stach, E., Amador, G., Hartfield, C.: Hardware and
techniques for cross-correlative TEM and atom probe analysis. Microsc. Today 16(4), 4247
(2008)
51. Prosa, T.J., Lawrence, D., Olson, D., Larson, D.J., Marquis, E.A.: Backside lift-out specimen
preparation: reversing the analysis direction in atom probe tomography. Microsc. Microanal.
15(S2), 298299 (2009)
52. Ziegler, J.: The stopping and range of ions in matter. http://www.srim.org/ (2011).
53. Ohring, M.: The Materials Science of Thin Films. Academic, New York, NY (1992)
54. Tsong, T.T.: Field ion image formation. Surf. Sci. 70, 211233 (1978)
55. Vurpillot, F., Cerezo, A., Blavette, D., Larson, D.J.: Modeling image distortions in 3DAP.
Microsc. Microanal. 10, 384390 (2004)
56. Larson, D.J., Geiser, B.P., Prosa, T.J., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F.: Improve-
ments in planar feature reconstructions in atom probe tomography. J. Microsc. 243, 15 (2011)
57. Gerstl, S.S.A., Morrone, A., Kvitek, R.: Three-dimensional nanoscale characterization of pt
deposition from an organometallic precursor induced by a focused ion beam. Microsc.
Microanal. 12(S2), 12521253 (2006)
58. Mutas, S., Klein, C., Gerstl, S.S.A.: Investigation of the analysis parameters and background
subtraction for high-k materials with atom probe tomography. Ultramicroscopy 111, 546551
(2011)
59. Francois-Saint-Cyr, H.G., Mutas, S., Prosa, T.J., Ulfig, R.M., Klein, C., Lawrence, D., Olson,
D., Larson, D.J.: Assessment of protective coatings used in focused-ion-beam specimen
preparation for atom probe tomography. Abstracts of 53rd International Field Emission
Symposium, 216 (2012)
Instrumentation and Techniques for Semiconductor Nanostructure Characterization, vol. 2, pp.
407477. World Scientific Publishing, London (2011)
61. Larson, D.J., Lawrence, D., Olson, D., Prosa, T.J., Ulfig, R.M., Reinhard, D.A., Clifton, P.C.,
Kelly, T.F., Lefebvre, W.: From the store shelf to device-level atom probe analysis: an exercise
in feasibility. In: 36th International Symposium for Testing and Failuer Analysis, San Jose,
CA, 2011, pp. 189197. ASM International
62. Marquis, E.A., Hu, R., Rousseau, T.: A systematic approach for the study of radiation-induced
segregation/depletion at grain boundaries in steels. J. Nucl. Mater. 413(1), 14 (2011).
doi:10.1016/j.jnucmat.2011.03.023
63. Giannuzzi, L.A., Stevie, F.: Introduction to Focused Ion Beams, Instrumentation, Theory,
Techniques and Practice. Springer, New York, NY (2005)
64. Marquis, E.A., Geiser, B.P., Prosa, T.J., Larson, D.J.: Evolution of tip shape during field
evaporation of complex multilayer structures. J. Microsc. 241(3), 225233 (2011)
65. Larson, D.J., Prosa, T.J., Geiser, B.P., Egelhoff Jr., W.L.: Effect of analysis direction on the
measurement of interfacial mixing in thin metal layers with atom probe tomography.
Ultramicroscopy 111, 506 (2011)
References 53
66. Smith, G.D.W.: Pd as an adhesion layer for Ni and Ag. In. Personal Communication, Oxford,
UK (2009)
67. Gorman, B.P.: Atom probe reconstruction refinements by pre- and post- analysis TEM
structure quantification. Microsc. Microanal. 13(S2), 16161617 (2007)
68. Arslan, I., Marquis, E.A., Homer, M., Hekmaty, M.A., Bartelt, N.C.: Towards better 3-D
reconstructions by combining electron tomography and atom-probe tomography.
69. Gorman, B.P., Diercks, D., Kaufman, M.J., Ulfig, R.M., Lawrence, D., Thompson, K., Larson,
D.J.: Atomic scale compositional and structural characterization of nanostructured materials
using combined FIB, STEM, and LEAP. Microsc. Microanal. 12(S2), 1720CD (2006)
70. Gorman, B.P., Diercks, D.: Cylindrical specimen geometries for sub-nm 3-D characterization
of semiconductor devices. Microsc. Microanal. 13(S2), 822 (2007)
71. Petersen, T.C., Ringer, S.P.: Electron tomography using a geometric surface-tangent algo-
rithm: application to atom probe specimen morphology. J. Appl. Phys. 105, 103518 (2009)
72. Antcheva, I., Ballintijn, M., Bellenot, B., Biskup, M., Brun, R., Buncic, N., Canal, P., Casadei,
D., Couet, O., Fine, V., Franco, L., Ganis, G., Gheata, A., Gonzalez Maline, D., Goto, M.:
ROOTA C++ framework for petabyte data storage, statistical analysis and visualization.
Comput. Phys. Commun. 180, 24992512 (2009)
73. Williams, C.A., Marquis, E.A., Cerezo, A., Smith, G.D.W.: Nanoscale characterisation of
ODSEurofer 97 steel: an atom-probe tomography study. J. Nucl. Mater. 400, 3745 (2010)
74. Petersen, T.C., Ringer, S.P.: An electron tomography algorithm for reconstructing 3D mor-
phology using surface tangents of projected scattering interfaces. Comput. Phys. Commun.
181, 676 (2010)
75. Giannuzzi, L.A., Schwarz, S.M.: FIB specimen preparation for STEM and EFTEM tomogra-
phy. Microsc. Microanal. 10(S2), 142143 (2004)
76. Felfer, P., Alam, T., Ringer, S.P., Cairney, J.M.: A reproducible method for damage-free site-
specific preparation of atom probe tips from interfaces. Microsc. Res. Tech. 75, 484491
(2012)
77. Kelly, T.F., Larson, D.J.: Atom probe tomography 2012. Annu. Rev. Mater. Res. 42, 131
(2012)
Chapter 3
Design and Instrumentation
3.1 Introduction
This chapter describes the LEAP instrument and its hardware components with
various levels of detail. Following an introductory section, items are grouped as
they are related to the local electrode, elements critical to detection (detection and
imaging), transfer of specimens and materials though the various vacuum chambers
(transfer and storage of consumables), voltage supplies and laser systems (field
evaporation systems), as well as ancillary systems.
The goal of this chapter is for users to (1) realize that there exist fundamental
limitations inherent in the design of an atom probe and (2) become familiar with
major hardware systems that make up the LEAP.
3.2 How Atom Probes Work
Atom probes work by a process known as field evaporation [1, 2] (see also
Appendix B) where a high electric field at the surface of a material induces
desorption and ionization of atoms even at cryogenic temperatures. At electric
field strengths of 1010 V/m or greater, electrons can be stripped from atoms on
the surface by a tunneling process. The necessary field is created by applying a
modest voltage (510 kV) to a very sharp specimen. The resultant positive ions are
accelerated away by the electric field toward the counter electrode. As discussed in
Chap. 1 and in this chapter, the electric field is proportional to the applied voltage
and inversely proportional to the radius of curvature of the tip of the specimen.
Thus, a specimen with a radius of curvature around 100 nm or less and a voltage
around 10 kV successfully accomplishes this task.
Minimally, an atom probe includes a specimen mount, a counter electrode with
an aperture, and a detector, Fig. 3.1. During analysis, a positive electrical potential
is applied to the specimen, while the counter electrode, located between the
56 3 Design and Instrumentation
Fig. 3.1 Schematic

geometry of a typical atom
probe. L is the flight path
length to the center of the
detector, and R is the radius
of the detector. Vtot is the
total accelerating voltage of
the system, Vext is an
extraction voltage, Vpa is a
post-acceleration voltage
that can range from zero to
greater than Vext, and Vp is a
voltage pulse that is the
time-varying component of
dynamic fields. In the
LEAP, the counter electrode
is a small aperture of about
40 m diameter
specimen and the detector, is either grounded or negatively biased. A pulse, such as
a positive electrical pulse on the specimen (or negative electrical pulse on the
counter electrode) or a thermal pulse from laser irradiation, is periodically applied
to the specimen to initiate ionization of surface atoms at a known time.
A position-sensitive detector records the hit position of individual ions, the
accumulation of which forms an image of the tip surface and ultimately enables a
direct mapping of the atoms position on the original curved surface of the speci-
men apex. This image is a projection, and because the tip apex is about 100 nm
across while the detector is about 100 mm across, the projected image of atoms on
the specimen surface has a projection magnification of about one million times. At
this magnification, interatomic distances of 0.2 nm become 0.2 mm on the detector,
which is easy to resolve with modern detector technology. Because only surface
atoms are removed, the sequence of evaporation events is used to construct the third
dimension of the dataset. Thus, as the specimen is evaporated, the position1 of the
atom in xy is determined from its hit position on the detector and its z-position
(depth into the specimen along its long axis) is determined from the sequence of
1
This book uses the convention that x, y, and z are the coordinates in the real space structure of a
material, X and Y are coordinates of a position on a detector, and N is the sequence number of an
ion in the evaporated dataset.
3.2 How Atom Probes Work 57
evaporation events, N. This creates a map or an image of the specimens constituent

atoms in 3D.
Atoms may be field evaporated in a controlled fashion with simple control of
applied voltage. During data acquisition, the probability of evaporation events per
pulse is kept purposely small (e.g., 1 ion for every 100 pulses). This is both to
suppress the formation of multiple simultaneous hits, which are difficult for single-
particle detectors to encode, and to reduce the field-induced stress on the specimen.
Generally, atom probe tomography (APT) has the advantage that all species of
atoms (including individual isotopes) are detected with uniformly high efficiency.
The efficiency is limited by a number of factors but is dominated by the open-area
fraction of the microchannel plate (MCP) amplifiers, which is typically about 60 %.
The identity of an atom is determined by measuring the time it takes the ion to
travel the distance from the specimen to the detector. This technique is referred to as
time-of-flight (TOF) mass spectrometry (see Appendix D). Since ions experience
an acceleration, a, due to electric fields, E, and there is a uniform electric field at the
specimen apex, all ions of a given charge state, n, experience the same force.
Therefore acceleration depends on mass, m, because force F ma neE for
ions of the same charge state, where e is the elementary unit of charge. Higher
charge states lead to greater acceleration, but these higher charge states are readily
identified and the corresponding mass-to-charge-state ratio, m/n (where n may be
+1, +2, +3, etc.), of the ion can be determined with high precision. The peak width
in a mass spectrum, m, is one of the most important metrics by which atom probe
performance is measured. This width is usually expressed by normalizing it to the
mass of the peak, m/m, which is called the mass resolution and is a number less
than unity. Alternatively, the inverse of mass resolution is called the mass resolving
power (MRP) and is a number greater than unity. Instruments with higher MRP can
better discern or separate individual mass peaks in the mass spectrum.
Peak widths are usually measured and expressed as their full width at half of the
peak maximum (FWHM). Since peak shape is also a very important consideration
in APT, it is desirable at times to know peak width at multiple locations in the peak.
Measurements of peak width at locations below half maximum are also common
since they give some sense of the important property of being able to fully separate
closely spaced peaks. We recognize the following conventions for common peak
heights: full width at one-tenth maximum (FW0.1 M), full width at one-hundredth
maximum (FW0.01 M), full width at one-thousandth maximum (FW0.001 M), and
so on. If the location of the peak width measurement is not stated explicitly, it is
safe to assume that the value is FWHM.
In order to measure TOF, the evaporation process must be limited to a short time
window so that a well-defined departure time is realized. Since the field evaporation
rate increases with electric field and temperature, ionization can be accomplished by
pulsing either the applied voltage or the specimen apex temperature (via a laser
pulse), each on a nanosecond time scale. The time that the specimen apex spends in a
state of high ionization probability is very brief (~100400 ps) so that the time of
departure of the ion is known precisely, and a measurement of the resulting TOF
is obtained. Voltage pulsing requires that the specimen has good electrical
conductivity (>102 S/cm) and work only with metals or other highly conductive
materials. Thermal pulsing has been accomplished by delivering energy with a
pulsed laser [3] and works with a broad range of materials independent of their
electrical conductivity. Semiconductors and insulators are the obvious materials that
benefit from laser pulsing, but metals can also benefit from this form of pulsing.
As a method of mass identification, TOF mass spectroscopy has many advan-
tages including the following:
All atoms and isotopes are detected.
It is not necessary to anticipate which elements to study in advance.
All species are detected with uniform efficiency.
It is possible to achieve high MRP.
3.3 LEAP Performance Parameters
The instrumental design parameters that are of highest interest are field of view
(FOV), MRP, and data collection rate. The adoption of a local-electrode geometry
and the availability of reflectron energy compensation provide certain important
advantages and limitations, and these influence other primary design parameters
(such as flight path length) which directly relate back to FOV, MRP, and data
collection rate. It is very difficult to avoid direct trade-offs inherent in choice of
flight path length (MRP versus data collection rate), although systems with and
without reflectron energy compensation provide some positive options. Ultimately,
the straight-flight-path LEAP 3000/4000X Si system is designed to promote the
highest laser mode data collection rates and the highest detection efficiency, while
the energy-compensated LEAP 3000/4000X HR (high resolution) system provides
superior spectral quality (MRP and background) especially under field-pulsing
acquisition conditions.
3.3.1 Field of View
A large FOV makes it possible to capture larger material volumes with, ideally, no
loss in spatial fidelity. Larger volumes provide an advantage from a statistical
perspective when determining concentrations. In addition, they allow characteriza-
tion of larger scale features and enhance the ease of locating small features (e.g., in
discrete device structures). The first 3D atom probes had limited FOV. Though their
maximum FOV of about 30 nm was a huge improvement over 1D atom probes with
about a 1 nm field, it was insufficient for many material analysis applications. When
the goal is to access as large a FOV as possible, Fig. 3.1 suggests making the
detector size large and the flight path length small. However, the spatial focusing
provided by the reflectron enables longer flight paths without losses in FOV.
3.3 LEAP Performance Parameters 59
Fig. 3.2 Schematic

illustration of a flat
reflectron geometry for an
atom probe. Rays diverging
from the specimen are
reflected by the reflectron.
The more energetic ions
(solid lines) penetrate
deeper into the reflecting
field and begin chasing the
lower energy ions (dashed
lines). The detector is
placed at the location where
the faster ions just catch the
slower ions
In a LEAP 3000/4000 Si, the detector subtends greater than 0.84 steradians
(>60 full angle). This corresponds to a FOV for an aluminum specimen at 10 kV
of greater than 200 nm diameter.
3.3.2 Mass Resolving Power
An atom probe is capable of accurate measurements of concentrations over a wide

dynamic range. This requires that mass peaks from individual species in a spectrum
can be accurately quantified. Ideally, all peaks would be fully separated from each
other, and high values of MRP as expressed by FWHM are clearly needed. In
addition, however, peaks should ideally be Gaussian-like (as opposed to peaks with
large tails) and be narrow not just at mid-height but also at their lower portions.
Thus high values of MRP expressed for FW0.1 M and FW0.01 M are also desirable.
In modern LEAP systems, MRP greater than 1,000 is routine at FWHM. In some
modes of the LEAP, excellent performance is also maintained in lower height
regions of the peak.
The local-electrode geometry produces improvements in MRP in field pulsing
mode due to minimization of field evaporated ion energy spread. Reflectron energy
compensators [4, 5] are needed in field pulsing however to achieve the high MRPs
needed for quality work. Reflectrons have been used in TOF atom probe systems for
both 1D instruments [6, 7] and 3D instruments [8, 9]. Figure 3.2 shows a standard
flat reflectron geometry where the ray paths of the ions diverge continuously. Such a
design is limited because it effectively corrects energy spread in time but not in
space (chromatic aberrations). Panayi [10] developed the concept of a curved
reflectron to overcome this limitation. In the curved reflectron, the curvature of
the electric fields converts the divergent ion ray paths to convergent ray paths. The
flight path length of such instruments is relatively long which has two main
ramifications: (1) the MRP in pulsed-laser mode can be very high due to the long
flight times and (2) the maximum pulse repetition rate may have to be limited
(<500 kHz) due to the long flight time.
With laser pulsing, the energy spread will not be dominated by the thermal
energy of the ions (0.020 eV at 250 K) but rather the energy spread inherent in field
evaporation processes. For most materials, this latter term has been found to be 3 eV
or less. Assuming that the energy spread is low (e.g., E/E V/V 3 eV/
10,000 eV 3 104), then tP for laser pulsing (see Appendix D) will be
small. Longer flight times in laser pulsing can therefore lead to greater MRP
[11]. Whether it is a long straight flight path in a LEAP XSi or a long curved flight
path in a LEAP XHR, high MRP may be achieved with laser pulsing.
3.3.3 Data Collection Rate
The data collection rate of LEAPs is high compared with historical norms for atom
probes. Since the data collection rate is the product of the detection rate and the
pulse repetition rate we can look at these two parameters to understand how a high
rate is achieved. The detection rate is typically about 0.01 ions per pulse. This
seemingly low rate is necessary to keep the number of multiple evaporation events
low since most position-sensitive detectors have difficulties encoding multiple hits.
The ability to properly detect multiple hits occurring during a single pulse enables
acquisition at higher average hits per pulse.
Unless new detector technology comes along, the real opportunity to achieve
high data collection rates is to maximize the pulse repetition rate. The maximum
pulse repetition rate is dictated by the longest ion flight times expected in an
experiment, i.e., for the lowest operating voltage and the heaviest ions for a given
flight path. Short flight paths such as in the LEAP Si thus enable high data collection
rates. The presence of a reflectron constrains flight path lengths to larger values with
correspondingly smaller pulse repetition rates.
Laser pulse repetition rates are not a limitation since they can be much higher
than the physical flight times require. However, voltage pulse repetition rates are
limited by the technology of the pulser. The field enhancement made possible by
the local-electrode geometry means that lower voltages are needed for a given field.
Since much higher pulse repetition rates may be achieved at lower voltages, high
data collection rates under field pulsing conditions are achieved in a LEAP.
3.4 Instrumentation of the LEAP 61
Table 3.1 Summary of trade-offs between LEAP models

LEAP 3000/4000 Si LEAP 3000/4000 HR
Straight flight path Reflectron
Pulsing mode Voltage Laser Voltage Laser
FOV Large Large
MRP Modest Good High High
Data collection rate Modest High Modest
3.3.4 Model Comparison
The trade-offs inherent in the straight flight path LEAP 3000/4000X Si and curved
reflectron LEAP 3000/4000X HR are briefly summarized in Table 3.1. When high
MRP in field pulsing or for general applications is important (e.g., for detection of
dilute species and identification of chemical species), the reflectron instrument is
the best choice. Note that the detection efficiency of the HR model is about 37 %,
whereas the Si model is about 58 %. In some circumstances, this difference in
detection efficiency can be significant. The most common circumstance is analysis
of small clusters of atoms where higher detection efficiency makes it possible to
detect clusters at smaller size. When high data collection rate is premium, the
straight-flight-path instrument is the best choice. For general-purpose laboratories,
either configuration may prove suitable depending on the specific application
needs.
3.4 Instrumentation of the LEAP
In the LEAP, an ultra-high vacuum (UHV) analysis chamber houses the detector,
specimen stage, ion pump, laser optics (optional), and reflectron (optional) as well
as the cryogenic and electronic pathways necessary to enable tool operation.
Ancillary chambers and hardware systems enable safe and clean operation. The
following sections provide descriptions of important hardware and systems present
in the LEAP.
3.4.1 Local Electrode
The local electrode is the distinguishing component of the LEAP and the source of
its name. The funnel-shaped counter electrode [1216] is used to create the electric
field as illustrated in Fig. 3.1. With a 40 10 m diameter aperture at the apex of
the local electrode, the electric field can be confined or localized to very small
Fig. 3.3 Schematic illustration of (a) the electric field, E, on a metal sphere of radius, Ro, opposed
by an outer sphere of radius, RS, with a voltage, V, between them and (b) the electric field of a
metal needle with an apex of radius, Ro, opposed by a counter electrode some distance away, Rc,
with a voltage, V, between them. For both cases, the distance, r, is measured from the center of the
sphere to the bounding counter electrode
features. Though in principle the cathode in an atom probe could be the detector, in
practice a separate counter electrode is always used.
The electric field on a surface depends upon whether the field can penetrate the
surface. In a metal, field penetration may be neglected. In dielectrics, field pene-
tration may largely depend on charge redistribution and available electron energy
states. For the present purposes, we will consider a metal surface with no field
penetration. The easiest geometry to consider is a sphere inside a sphere as shown in
Fig. 3.3a. The magnitude of the electric field vector, E, on the surface of the inner
sphere is simply
V
E 3:1
r
and the orientation is normal to the surface at all points. For a needle with a
spherical apex opposed by a plate, Fig. 3.3b, the electric field magnitude can be
computed from [1] when the distance to the counter electrode RC >> r:
V
E : 3:2
kr
Here k is a constant (greater than unity) which accounts for the fact that the electric
field is reduced in magnitude by the presence of the shank of the needle and the
needle/plane geometry. This latter expression is meant to give the electric field
magnitude normal to the surface at the central axis of the apex. For this example,
k has a magnitude of about 5 [1] and has been found to increase with shank angle,
[1719].
Fig. 3.4 Photograph of a

local electrode with a
properly aligned needle-
shaped specimen in the
LEAP. The local electrode
has a 40 m diameter
aperture at the apex
Fig. 3.5 (a) Schematic of aperture diameter, and hence tip-to-aperture distance, on the field
enhancement at a tip relative to a remote electrode. (b) Parameters for the configuration modeled
are as follows: 50 nm tip radius, 10 shank angle, 50 m diameter wire, 1 mm long, specimen-to-
local-electrode distance is that required to permit a 1.5 steradian (70 ) geometric FOV. Calcula-
tions courtesy T. T. Gribb
A photograph of a metal needle positioned near a local electrode is shown

in Fig. 3.4. When the specimen is just over one aperture diameter (~40 m)
away from the local electrode, the APT image captures up to 1.5 steradian (70 )
of the specimen surface (this assumes an image compression factor of 1.4). Field
enhancement as a function of local-electrode aperture size is illustrated in Fig. 3.5.
The field at the apex increases with decreasing aperture size (or equivalently,
Fig. 3.6 Model parameters for the configuration shown in (a) are 50 nm tip radius, 10 shank
angle, 40 m diameter aperture, and a specimen-to-local-electrode distance of 48 m. The base
width, w, relative to the microtip height, h, is such that h/w << 1. (b) Variation of the field
enhancement at a tip with increasing height of the needle. Calculations courtesy T. T. Gribb
apex-to-aperture separation). For a 40 m diameter aperture, the field factor is

almost two times smaller than the remote-electrode case (two times field
enhancement).
Figure 3.5 begs the following question: why not get more field enhancement by
using smaller apertures? There are at least two reasons. First, the local electrode is a
cathode, and the field on the rim of the electrode can be sufficient to produce
significant electron emission. Whereas field evaporation from the anode (specimen)
requires fields greater than 1010 V/m, field electron emission from the cathode
requires fields about an order of magnitude lower [1]. As the aperture diameter
decreases, the fields on the rim of the local electrode increase [12], and it becomes
more and more difficult to avoid field electron emission for small apertures. When
significant electron emission does occur, the noise level in the data can rise to
extreme values. Second, when specimens fracture the debris can be deposited on the
local electrode. The debris can stick to the local electrode and sometimes obstruct
portions of the aperture or asperities of small radii form and become additional
sources of field-emitted electrons.
As envisioned by Nishikawa [12, 20], a radical change in APT specimen design
is enabled when a local electrode is employed. Specimens need to be less sharp
compared with remote electrode geometries and may be much shorter as well.
Figure 3.6 shows the modeled field for a microtip geometry compared to that of a
needle with a 40 m diameter aperture. Microtips on the order of 100 m tall
experience ~92 % of the field relative to the needle geometry. The microtip base
plane is effectively at infinity since the relative field changes very slowly for longer
microtip heights. Localization of the field allows for microtips (or other protru-
sions) to be only about one to two times (or less) as tall as the aperture diameter in
order to experience significant fields for APT analysis.
Fig. 3.7 (a) SEM image of a microtip array used in the LEAP 3000/4000. There are 22 microtips
in the array located close to the top and bottom edges. (b) A single microtip. (c) The apex of a
pre-sharpened microtip which has a radius at the apex of about 75 nm. (d) A flat-top microtip
which has a 2 m diameter flat at the apex with a lift-out specimen positioned for attachment with
FIB-deposited platinum
3.4.1.1 Microtip Array Specimens
The geometry of the microtip array was developed over many years, driven by need
and practical constraints to improve the speed and precision of specimen prepara-
tion. Some of the first methods to fabricate a large number of specimens took
advantage of existing photolithography and deep reactive ion etching to make
various geometries of specimens that could be mounted to a pin and sharpened
using a focused ion beam (FIB) [21, 22]. As the LEAP and FIB instruments
evolved, so did the design of the specimen carrier resulting in the current generation
of microtip arrays. Ease of use, cost, LEAP design, and performance considerations
all influenced the latest versions which consist of <100>-oriented antimony-doped
silicon coupons with posts that are approximately 150 m tall as shown in Fig. 3.7.
The placement of the tips and marker posts on the 3 mm by 7 mm coupon allows for
straightforward handling and easy alignment in the LEAP while minimizing the
electric field on adjacent tips during application of high field. Microtips come in
both pre-sharpened and flat-topped varieties [23], and both are doped to a conduc-
tivity >100 S/cm (antimony composition on the order of 100 ppm).
Microtip analysis made possible by the local electrode provides some unique
advantages, but there are also potential pitfalls. One notable pitfall is that the
microtip geometry provides a higher relative field at the local electrode compared
to the wire geometry which encourages electron emission and can causes reduced
local-electrode lifetime.
3.4.2 Detection and Imaging
3.4.2.1 Delay Line Detector
The delay line detector (DLD) assembly consists of the MCP assembly and a delay
line anode (DLA). When an ion impacts the MCP assembly and creates a cascade of
secondary electrons, the resulting electrons (on the order of 106 per ion) spread over
a few millimeters (FWHM) and impact several of the serpentine wire traces of the
DLA. Because the electron cascade is larger in width (or equivalently in time) than
the transition time across adjacent delay lines, a single broad electrical pulse is
created. Fortunately, the weighted center of the pulse results in a spatial resolution
finer than the spacing of the individual delay lines.
MCP detection efficiency is affected primarily by the open-pore area and the
incident angle of ion impacts. Because amplification of the single ion into an
electron cloud is largely insensitive to ion mass (for ion energies greater than
about 2 keV), no calibration is necessary for measuring different ion species.
There is however the possibility of variation of amplification across the FOV due
to ion impact angle. At short flight lengths in a straight-flight-path instrument, ion
trajectories may be parallel to the angle of the openings in the MCP causing the
secondary electron cascade to start at a significantly greater depth and resulting in
less amplification. This can be noticed at the very edge of a LEAP dataset at short
flight path lengths at a particular angular location (note that this effect does not
occur in the reflectron-based system because the ion trajectories are more closely
normal to the MCP surface). There are also detection losses due to ions that impact
the MCP between channels as shown in Fig. 3.8 (filled white circle) because they
cannot produce secondary electron cascades.
The electrical pulse from the electron cloud on the delay line is detected at both
ends of the wire (see Fig. 3.9). The time of the leading edge of the pulse is recorded
by the amplifier time-to-digital conversion (ATDC) system as discrete timing events.
A 2D detector is fabricated by wrapping the wire in a meander pattern as shown in
Fig. 3.10 for X and Y. The Y electrode is nominally identical to X but is rotated 90 .
The X anode allows about half of the electron cloud to pass through to be detected by
the Y anode. The resulting timing precision provides approximately 800 800
pixels across the detector face resulting in sub-nanometer spatial resolution in the
final image. This type of detector is typically referred to as a crossed DLA.
Fig. 3.8 SEM image of the

face of a standard
microchannel plate. Ions
impinging on the MCP in a
channel (dark regions) will
likely be amplified whereas
those impinging on the
intrachannel web (white
dot) will likely not be
amplified
Fig. 3.9 Schematic illustration of how a 1D DLA operates. Position (X) is determined from the
difference of the arrival times at the ends of the wire. Time of flight is determined from the average
of the two pulse arrival times. X 0 is the center of the wire. t0 is a constant of the system
Fig. 3.10 Schematic of a

2D meander DLA. A charge
cloud impinging on this
structure will deposit signal
on both X and Y electrodes
Fig. 3.11 Schematic

illustration of hits on a 2D
anode where ambiguity
exists regarding whether
(X1, Y1) and (X2, Y2), the
cross-hashed circles, or
(X1, Y2) and (X2, Y1), the
diagonal hashed circles, are
the correct answer. A third
axis of information, W,
resolves this ambiguity
3.4.2.2 Multi-hit Detection
A multi-hit occurs when more than one ion hit is detected during a single pulse.
DLD systems in early LEAPs use a two-anode delay line. The addition of a third
anode enables more accurate positioning of the multiple ion impacts that are close
in time and space (described within Fig. 3.11). Current LEAP systems use a
detector with three-anode delay lines, creating up to six timing events for each
ion impact. Occasionally one or more of the six channels may not record a signal
(typically called a partial hit). In theory, as few as three hits (one from each anode)
may be used to record the location of an event, but the more stringent selection
criteria currently employed in hit-finding algorithms require five hit events. Fewer
than five hit events result in significantly increased uncertainty in time (and
therefore in space), and so these hit types are ignored (called unrecoverable
partials).
A three-anode DLD does not, however, overcome the potential overlap of
multiple impacts that are very close in time and space. Multiple impacts that have
a specific spatial relationship can appear at the ends of the DLD as a single pulse
and are recorded as a single discrete timing event. When two ion impacts result in a
single overlapping pulse event, there are ways to extract both impacts from the
eleven recorded signals (six signals per event minus one overlapping event). When
more than one overlapping timing event exists, there may be multiple solutions for
the locations of the two events, so current hit finding typically only faithfully
identifies and records the first arriving event in these latter cases.
Data acquisition rates are typically restricted to minimize the number of multiple
field evaporation events likely to happen during any given pulse. However, multiple
ionization events resulting from a single pulse are much more frequent than random
probability predicts. This is referred to as correlated evaporation, and for some

materials and analysis conditions multiple ionization events may be present in as
many as 60 % of the pulses with recorded events. Common examples of applications
with high multiplicity are oxides, III/V semiconductors, carbon in steels, and carbon
and boron in silicon. Due to the highly correlated evaporation of these alloys, special
care must be taken, and in some cases, highly accurate compositional measurements
cannot be made without correction factors (see [24] for example).
3.4.2.3 FIM
Field ion microscopy (FIM) is the precursor to APT and until recently was a critical
step in each APT experiment [25]. In previous generations of instruments, each
experiment was preceded by FIM to ensure specimen viability, remove oxidation
and damage from specimen preparation, evaporate the specimen until the region of
interest (ROI) was at the surface, and orient the specimen such that the ROI was
aligned with the detector. Previous generations of APT systems typically had a
separate phosphor screen for FIM from the TOF detector. In the LEAP, FIM is not
often used prior to the APT experiment, and the single-detector system has been
optimized for atom probe data collection.
Although the LEAP system has a wide FOV in both APT and FIM modes, there
are times when FIM can be useful as it provides complementary structural infor-
mation about the specimen. FIM has higher spatial resolution than APT due to the
fact that gas ion trajectories are less influenced by local bonding and local changes
to the radius of curvature, so the relationship and size of small phases or precipitates
may be more precisely imaged. In the LEAP, FIM experiments may be digitally
captured, and although the gain and bias of the MCPs cannot be changed after the
experiment, the brightness, contrast, and detector coefficients can be optimized
after data collection. For references on HD-eFIM experimental conditions, see
[2630].
3.4.3 Transfer and Storage of Consumables
3.4.3.1 Specimen and Local Electrode Handling
In LEAP terminology, specimens are affixed to stubs, stubs and local electrodes are
positioned on pucks, pucks are loaded onto carousels, and carousels go into the
LEAP and reside on the vertical transfer rod. Various types of pucks and stubs are
shown in Fig. 3.12. Carousels have six positions, where each position can hold
either a specimen puck or a local-electrode puck. The pucks have a self-locating
wedge shape and are made with a high thermal and electrical conductivity alloy to
facilitate rapid cooling once placed into the cold stage. The local-electrode pucks
are designed for lubricity under UHV conditions.
Fig. 3.12 Solid models of specimen and local-electrode transport system for LEAP; (a) a carousel
has six positions and can be stacked in the loadlock and storage chamber, (b) a local-electrode
puck, (c) a four-needle specimen puck, (d) a single-needle specimen puck with a 3 mm TEM grid
holder, (e) a specimen puck with a Cambridge-style SEM specimen carrier, and (f) a single-needle
specimen puck with a quick-release holder for microtip arrays
Specimens for the LEAP can have nearly any geometry smaller than about
1 1 1 cm3 (due to stage travel and in-vacuum optics limitations). Specimens
made from a bulk sample in a wire geometry can be affixed directly into a puck or
crimped into a cylindrical stub. Specimen extracts can also be affixed to carriers.
The carriers can be wires, microtip coupons, TEM half grids, etc. Specimens of
unique geometries (not accommodated by standard stubs) can be affixed to stubs
with UHV-compatible silver-embedded epoxy to ensure acceptable electrical and
thermal conductivity.
As in all UHV applications, objects entering the vacuum system should only be
handled with clean tools [31]. Analytical sensitivity in a properly designed exper-
iment can be limited by the background noise level, and a clean vacuum will result
in the lowest possible spectral noise level. Of special consideration should be
handling of the local-electrode pucks. While the carousel never enters the analysis
chamber, and the specimen puck will become part of the cryogenically cooled
specimen stage, the local electrode resides in the analysis chamber of the UHV
system at room temperature, so its cleanliness is critical. Although the pumping
system is robust enough to enable a user to introduce a specimen and begin data
collection in only a few minutes, practically speaking one should only introduce
specimens into the analysis chamber after they have been pumped down to base
pressures in the storage chamber. Local-electrode pucks and other in-vacuum
components should be stored in a clean dry environment. Do not use desiccants
that may contaminate the parts with dust, and do not use any oil-based pumps.
Ideally these parts would be stored in a box with a clean, dry cover gas, or in an
oil-free vacuum environment.
3.4.3.2 Specimen Introduction, Transfer, Storage, and Stage Systems
The LEAP system uses a three-chamber design with a loadlock chamber, an

intermediate storage chamber, and a UHV analysis chamber. Carousel and puck
transfers within the LEAP are accomplished by hand and/or motor-driven systems
using magnetic coupling between the portions inside and outside the vacuum. Care
should be taken during all transfer motions, especially with the vertical transfer rod.
3.4.3.3 The Loadlock Chamber
The purpose of the loadlock is to allow fast specimen loading from atmospheric
pressure to high-vacuum conditions. The loadlock chamber is pumped with a high-
speed turbomolecular pump and is vented to atmospheric pressure using dry
nitrogen. Minimizing the introduction of moisture and other contaminants into
the loadlock will maximize vacuum quality. The loadlock chamber allows the
introduction of three or four carousels (depending on LEAP configuration) and
thus the use of up to 24 specimens or local-electrode pucks at the same time.
Pumping is achieved with a turbomolecular pump coupled to an oil-free backing
pump. The transfer pressure (maximum pressure for opening the valve between the
loadlock and storage chambers) must be better than 103 Pa, which is obtained in
less than 5 min. When the required transfer pressure in the loadlock is attained, the
loadlock chamber valve may be opened and the specimen may be transferred to the
storage chamber in a few seconds using the UHV vertical translation/rotation rods.
The UHV loadlock chamber gate valve is provided with automatic valve control
and interlocks to prevent accidental venting of the storage chamber.
To maximize data quality, many groups choose to quarantine freshly introduced
specimen pucks in the loadlock chamber for a minimum period (i.e., overnight),
even though the pressure interlock will allow specimen transfer prior to that time.
There is a trade-off between introduction time and vacuum quality, so some
planning can help avoid the need to introduce a puck immediately before analysis.
For 4000 series instruments that are heavily used, an additional option can be
selected which increases loadlock capacity to include a fourth carousel which
includes a heating capability. Best practice suggests that local electrodes need to
be stored under UHV conditions for outgassing for 34 days prior to use in order to
ensure low-noise, high-quality data. If heating is used in the loadlock, this time for
low noise data can be reduced to less than 8 hours.
3.4.3.4 The Storage Chamber
The purpose of the storage chamber is to maintain specimens and local electrodes
under UHV conditions before analysis. It allows a fast specimen transfer from the
storage chamber to the specimen stage of the analysis chamber. The transfer of the
specimen or the local electrode is achieved by using a translation/rotation rod and
integrated puck exchange video cameras. This configuration makes it possible to

maintain UHV conditions in the analysis chamber. Pumping is achieved with a
turbomolecular pump, and the chamber is bakeable to 150 C. The carousel can be
observed during loading/unloading through a large viewport mounted on the
chamber.
Typically the bottom two carousels remain in the storage chamber at all times,
and the top one (or two) carousel is used to transfer specimen and local-electrode
pucks in and out of the storage chamber. The bottom carousel is typically used to
store local electrodes, a local-electrode testing specimen, and a pre-sharpened
microtip coupon used to focus the laser and perform other quality checks. Be sure
to maintain at least one open carousel position at all times in the storage chamber to
enable specimen transfer.
3.4.3.5 The Analysis Chamber
The analysis chamber is a UHV stainless steel chamber pumped by an ion pump and
titanium sublimation pump. The cryogenically cooled specimen stage acts as a third
pumping device. The chamber is bakeable to 150 C. Base pressure is better than
108 Pa when the stage is at 50 K. The LEAP instrument can be equipped with a
FIM system whose operation requires that the analysis chamber be filled with
ultra-high purity (99.999 %) helium, neon, or argon up to 103 Pa which is achieved
through a precision leak valve.
3.4.4 Field Evaporation Systems
3.4.4.1 The Voltage Control System
In both laser pulsed and voltage pulsed modes, highly stable DC voltage power
supplies provide voltage to the specimen stage, the reflectron (on a LEAP 3000/
4000 HR system), and the MCP assembly. The gain and bias voltages for the MCPs
are set during the detector calibration process upon installation or major servicing
and they are held constant during all experiments. The reflectron voltage and the
pulse voltage (in voltage pulse mode) scale with the specimen voltage. The pulse
voltage is generated from a high-amplitude solid-state pulse generator and further
shaped with a combination of discrete components to ensure a 1090 rise time of
less than 1 ns and a very narrow pulse width.
3.4.4.2 Laser System
The LEAP laser system is an option that can be installed at the time of manufacture
or as a system upgrade after shipment. As with the main emphasis of the design and
construction of the field-pulsed-only LEAP unit (LEAP 3000/4000 HR), the laser
system has been designed to allow the user to conduct APT experiments without
requiring significant knowledge of the sub-components. There are no user-
serviceable parts in the laser enclosure, which is sealed to protect users and the
Class IV laser from damage. This allows the LEAP to be certified as a Class I
(no user training required) system due to its interlocks and lightproof, tamper-
resistant enclosure.
The laser pulse is created first with pump diodes that illuminate an oscillation
medium with an 80 MHz natural oscillation frequency for 1,064 nm light. This
cavity uses a semiconductor saturable absorber mirror (which is a consumable
medium). For the LEAP 4000X systems, this light is frequency tripled to 355 nm,
conditioned with various optics and power/centering/polarization components until
it is pulse-picked such that pulses with a repetition rate of 11,000 kHz enter the
targeting assembly. The beam is then focused and directed with piezoelectric
mirrors through a fused silica air/vacuum barrier.
3.4.5 Ancillary Systems
3.4.5.1 Cryogenics
The atom probe includes a cryogenic cooling device attached to the specimen to
reduce thermal motion of atoms that can result in positional errors in the data
collected, limit preferential evaporation, and also function as part of the vacuum
pumping system. Specimen base temperatures on the order of 2050 K are typically
used. In laser pulsing, since the probability of ionization at the specimen tip is a
rapidly increasing function of both field and temperature, the temperature at the
peak of a thermal pulse needs to be only ~250 K for a very short time (<1 ns). Thus
with a properly chosen laser pulse energy, the time that the specimen spends at
elevated temperature is small (<1 ns) and the effect of atomic thermal motion on
the image resolution is also small. Nonetheless, when careful comparisons are made
between field pulsing and thermal pulsing, it is possible to both observe changes in
charge state ratios and discern a small degradation in spatial resolution in laser-
pulsed data due, presumably, to thermal motion [32, 33].
The cryogenic system is composed of two main parts: the compressor and the
cold head. The former compresses the helium which is then cooled in a heat
exchanger and expanded in the cold head. The cold head is suspended inside an
inverted stainless steel can and isolated from the LEAP frame and vacuum vessel.
The specimen stage is connected to the base of the inverted can for cooling.
Helium gas fills the interspace between the cold head and the can to allow for
natural-convection heat transfer. A rubber membrane seals the helium gas into the
interspace and is the only mechanical connection between the cold head and the rest
of the instrument. The helium gas is kept at a positive pressure to eliminate
the accumulation of air, which would liquefy and freeze between the cold head and
the LEAP vacuum vessel.
The expander portion of the cold head reaches temperatures as low as 10 K. The
minimum achievable temperature at the specimen puck is ~25 K, which varies
slightly with environment. The temperature of the specimen is controlled by a
resistive heating element that is attached to the vacuum side of the cold head can.
3.4.5.2 Safety and Control
Isolation valves (loadlock/storage and storage/analysis) can be activated only if the

transfer rods, high voltage, and laser shutters are in a safe configuration. A master
emergency-off-switch is included that can be connected to external, customer-
supplied shut-off circuits. The entire system, except the cryogenic pump compres-
sor, is run through an uninterruptable power supply, which allows continuous
operation through minor power source fluctuations, or short power losses, and
allows an orderly shutdown in the event of a power loss to enable quick vacuum
recovery. Nearly every function of the LEAP system (other than specimen transfer)
may be controlled by the LEAP Control Center (LCC) software (see Chap. 4).
3.4.5.3 Laboratory Vibration and Temperature Control
In any laboratory setting one must worry about environmental disturbances that can
affect data quality and yield. In general, the imaging capability of APT is insensi-
tive to vibration because micron-level motions of the tip translate to micron-level
blurring in the image (smaller than the resolution of the APT image); however,
alignment of the laser system, which is focused to a micron-sized spot, is greatly
affected by vibration. A site survey tests temperature stability and vibration over a
large range of frequencies. Typically, a room must meet the criteria similar to those
necessary for scanning electron microscopy systems. For LEAP systems that do not
have a laser, there is very little effect of the environment on the operation of the
instrument.
3.4.5.4 Mechanical Vibration
When operating in laser pulsed mode, even minor vibrations can affect the position
of the focused spot relative to the specimen apex. Any vibration visible in the laser
microscope view or any vibration that can be felt by touching the instrument can
affect the laser spot. A slight twinkle of the reflected laser light in the camera view
is normal, but visible vibration or blinking of the reflected light indicates excessive
vibration. This may result in a drop in detection rate, a subsequent rise in voltage, a
decrease in signal-to-noise ratio, and a higher probability of specimen fracture.
When your system has noticeable vibration and the room has been qualified to be
3.5 Summary 75
within specification, first check for any items leaning against the system, especially
items that can transmit vibration from pumps, etc. The most common vibration
source is from the cryogenic cooling system. The LEAP 4000 frame rests on an
integrated air support system, but it may still be sensitive to serious environmental
vibration. It is important to understand that vibration levels may deteriorate with
time as locations become more congested or experience a change of use.
3.4.5.5 Temperature Variation
When inserting a room-temperature specimen puck onto the cryogenically cooled

stage, a temperature rise, and subsequent mechanical excursions, will occur. These
changes are small and typically decay within ~10 min of placing the specimen puck
on the stage. Although modest swings in room temperature will not affect the
temperature of the specimen stage, there are a number of other critical components
in the laser positioning system that are affected by such changes. Even smooth
temperature changes over long periods of time are detrimental as different parts of
the LEAP mechanical frame will change temperature at different rates and lead to
slight changes in mechanical dimensions that may affect laser position. As in all
high-resolution microscopy, the laboratory needs stable room temperature, no
direct sunlight, and no direct heating or air conditioning sources near the instru-
ment. Additionally, be aware that opening and closing the valve for the main LEAP
analysis flange during an experiment can introduce differential temperature
changes to the vacuum chamber and result in laser position drift with respect to
the specimen. Laser drift can lead to a reduction in data quality and potentially
specimen fracture.
3.5 Summary
The state of the hardware used in APT has continuously evolved since its inception
in the late 1960s. Over the past 1015 years however, this evolution has been more
substantial [34] and has culminated in the current-generation LEAP. The benefits of
the LEAP designlarge FOV, high MRP, high data collection rate, and microtip
analysis capabilityare mutually inclusive. It is difficult to envision any one
without the others because a 50 times improvement in FOV requires 50 times
longer acquisition time without a concomitant increase in data collection rate.
APT applications will always benefit from larger analysis volumes, which have
been found to be critical to resolving questions such as dopant distribution in
semiconductor materials [35].
There are undoubtedly directions to move with future atom probe instrument
research and development, including larger detectors, higher detection efficiencies,
improvement in multi-hit detection, improved laser, other thermal pulsing methods,
and more advanced analysis software. The atom probe community continues to
evaluate and develop new technologies to move the capability of APT forward, so
the future of the technology looks extremely promising. We refer the interested
reader to [36, 37] for further speculation as to what the future of APT may hold.
References
(1961)
2. Miller, M.K., Forbes, R.G.: Tutorial review: atom probe tomography. Mater. Charact. 60,
461469 (2009)
3. Kellogg, G.L., Tsong, T.T.: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys. 51
(2), 11841194 (1980)
4. Alikhanov, S.G.: A new pulse method of measuring the masses of ions. Sov. Phys. JETP 4,
452453 (1956)
4548 (1973)
6. Waugh, A.R., Richardson, C.H., Jenkins, R.: APFIM 200 a reflectron-based atom probe.
Surf. Sci. 266, 501505 (1992)
7. Scheinfein, M.R., Seidman, D.N.: Time aberrations of uniform fields: an improved reflectron
mass spectrometer for an atom-probe field-ion microscope. Rev. Sci. Instrum. 64(11),
31263131 (1993)
11. Liu, J., Wu, C.-W., Tsong, T.T.: Measurement of the atomic site specific binding energy of
12. Nishikawa, O., Kimoto, M.: Toward a scanning atom probe computer simulation of electric
field. Appl. Surf. Sci. 76/77, 424430 (1994)
13. Kelly, T.F., Camus, P.P., Larson, D.J., Holzman, L.M., Bajikar, S.S.: On the many advantages
of local-electrode atom probes. Ultramicroscopy 62, 2942 (1996)
14. Kelly, T.F., Larson, D.J.: Local electrode atom probes. Mat. Char. 44(12), 5985 (2000)
15. Cerezo, A., Godfrey, T.J., Huang, M., Smith, G.D.W.: Design of a scanning atom probe with
improved mass resolution. Rev. Sci. Instrum. 71(8), 30163023 (2000)
16. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.
C., Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data
from a commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383
(2004)
17. Larson, D.J., Russell, K.F., Miller, M.K.: Effect of specimen aspect ratio on the reconstruction
of atom probe tomography data. Microsc. Microanal. 5, 930931 (1999)
18. Humphry-Baker, S.A.: Atom Scale Studies of Fusion Materials: Atom Probe Tomography of
TungstenRhenium Alloys. University of Oxford, Part II Thesis (2008)
19. Loi, S.T., Gault, B., Ringer, S.P., Larson, D.J., Geiser, B.P.: Electrostatic simulations of a local
electrode atom probe: the dependence of tomographic reconstruction parameters on specimen
and microscope geometry. Ultramicroscopy 132, 107113 (2013)
20. Nishikawa, O., Kimoto, M., Iwatsuki, M., Ishikawa, Y.: Development of a scanning atom
probe. J. Vac. Sci. Technol. B 13(2), 599602 (1995)
References 77
21. Larson, D.J., Foord, D.T., Petford-Long, A.K., Liew, H., Blamire, M.G., Cerezo, A., Smith, G.
D.W.: Field-ion specimen preparation using focused ion-beam milling. Ultramicroscopy 79,
287293 (1999)
22. Larson, D., Wissman, B.D., Martens, R., Viellieux, R.J., Kelly, T.F., Gribb, T.T., Erskine, H.
F., Tabat, N.: Advances in atom probe specimen fabrication from planar multilayer thin film
structures. Microsc. Microanal. 7, 2431 (2001)
23. Thompson, K., Larson, D.J., Ulfig, R.: Pre-sharpened and flat-top microtip coupons: a quan-
titative comparison for atom-probe analysis studies. Microsc. Microanal. 11(S2), 882 (2005)
24. Thuvander, M., Weidow, J., Angseryd, J., Falk, L.K.L., Liu, F., Sonestedt, M., Stiller, K.,
Andren, H.O.: Quantitative atom probe analysis of carbides. Ultramicroscopy 111(6), 604608
(2011). doi:10.1016/j.ultramic.2010.12.024
26. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principles and Applications. Elsevier, New
York (1969)
27. Muller, E.W., Tsong, T.T.: Field ion microscopy, field ionization and field evaporation, vol.
4. Progress in Surface Science 1139 (1973)
30. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field ion Micros-
31. Hoffman, D., Singh, B., Thomas, J.H.I.: Handbook of Vacuum Technology, 1st edn. Aca-
demic, London (1998)
32. Gault, B., Moody, M.P., De Geuser, F., Haley, D., Stephenson, L.T., Ringer, S.P.: Origin of the
spatial resolution in atom probe microscopy. Appl. Phys. Lett. 95(3), 034103-1 (2009)
33. Gault, B., Moody, M.P., De Geuser, F., La Fontaine, A., Stephenson, L.T., Haley, D., Ringer,
S.P.: Spatial resolution in atom probe tomography. Microsc. Microanal. 16, 99110 (2010)
34. Kelly, T.F., Larson, D.J.: The second revolution in atom probe tomography. MRS Bull. 37,
150158 (2012)
35. Inoue, K., Yano, F., Nishida, A., Tsunomura, T., Toyama, T., Nagai, Y., Hasegawa, M.:
Monolayer segregation of as atoms at the interface between gate-oxide and Si substrate in a
p-MOSFET by 3D atom-probe technique. Appl. Phys. Lett. 92, 103506-1 (2008)
36. Kelly, T.F.: Kinetic-energy discrimination for atom probe tomography. Microsc. Microanal.
17, 114 (2011)
37. Kelly, T.F., Miller, M.K., Rajan, K., Ringer, S.P.: Atomic-scale tomography: a 2020 vision.
Microsc. Microanal. 19, 652664 (2013)
Chapter 4
Data Collection
4.1 Introduction
Data collection for an atom probe tomography (APT) experiment is the process of
(1) introducing an electric field on a specimen in order to initiate ion emission by
field evaporation, (2) using the initial ion events to allow fine positioning of the
specimen, and (3) increasing and adjusting the voltage (and thus the electric field)
to obtain a steady rate of detected ion events. The details of how one can maneuver
through these steps safely and efficiently are the focus of this chapter. Because a
strong electric field is required to initiate field evaporation, the resulting forces
create large stresses that act on the specimen, placing it in continual jeopardy of
mechanical failure. Consequently, a successful atom probe experiment is a
balancing act of minimizing the chance for specimen failure while at the same
time extracting maximum quality from the collected data. After reading this
chapter, it becomes clear that we prioritize yield over data quality (one must have
data before judgments can be made on its quality), but the ultimate decision on this
issue rests with the user. Depending on the details of any particular experiment, the
user may decide that either specimen yield or data quality takes precedence.
The goal of this chapter is to guide the reader in the steps necessary for operation
of the local electrode atom probe (LEAP) in order to collect high-quality data. This
includes a discussion of how the primary experimental parameters (temperature,
ion detection rate, etc.) influence the chances for experimental success (analysis
yield) and affect data quality, especially when approaching a new sample type.
A minimal description of the instrument control software is included as it is this
interface through that the user selects parameters and operates the instrument.
Suggested operating procedures for voltage acquisition and laser acquisition,
including laser pre-alignment, are included to aid novice users through a check
list of activities to begin an analysis. Experienced users may find more value in the
general discussions regarding parameter selection, data quality, and data collection
goals. This chapter is not meant to replace the LEAP User Manual which describes
specimen transfer, alignment, and the acquisition software in detail; instead it is
80 4 Data Collection
meant to serve as an introduction to some important aspects of experimental design,

tool operation, and data collection.
4.2 Data Quality Considerations
Before discussing the various experimental parameters and the effects they have on
data quality, it is important to identify the basic metrics used to assess data quality.
Some of these metrics describe the quality of the mass spectrum, but all impact the
ability to accurately quantify the composition of the specimen. These data quality
metrics include mass resolving power (MRP), background, multi-hit performance,
and complex ion generation.
MRP and background are general metrics used to describe any spectrum. MRP
was defined in Chap. 3 and is a measure of the width of the time-of-flight (TOF)
range (or m/n range) for each ion type. Background is loosely defined as (1) events
detected at arrival times that do not correspond to a specific mass-to-charge (m/n)
peak or (2) ion evaporation events that occur out of synchronization (out-of-synch)
with the voltage or laser pulse. Some of these events originate from the specimen,
while others originate from residual gasses in the analysis chamber that were
adsorbed on the specimen and then field evaporated at a random time. The adsorbed
gasses also field evaporate due to the evaporation pulse, but these events are
observable as families of peaks in the m/n spectrum (commonly H+, H2+, H3+,
OH+, H2O+, H3O+, and others) and can be selectively filtered out during analysis.
The out-of-synch events are evenly distributed in time, which is observed as a
decaying background level in the m/n spectrum.
The presence of a background directly impacts data quality and should be
minimized as much as possible [1, 2]. A background component resides in each
m/n peak and can be estimated and subtracted from each peak when determining
chemical composition; however, correctly accounting for this effect in visualization
(atom maps) is complicated. In principle, each ion in a mass spectrum peak has
some probability that it is a residual gas ion and not an assigned ion type. Due to the
limitation of using a single set of mass ranges, this causes background ions of each
ion type to be distributed throughout the atom map. For example, because of m/n
ranging limitations (see Chap. 6 and Appendix D) a bilayer specimen of pure A on
pure B will display background levels of B ions in A and vice versa in any
atom map.
The out-of-synch component can strongly affect quantification when all ionic
species do not have an equal probability of exhibiting out-of-synch evaporation.
Insufficient pulse fraction (see next section) can cause preferential evaporation
and/or retention of certain ion types resulting in incorrect compositions [3, 4]. Back-
grounds composed primarily of out-of-synch evaporation events of equal probabil-
ity for each ion type do not affect composition accuracy and any potential
reconstruction artifacts can also be corrected (see Chap. 5).
4.3 Analysis Yield Considerations 81
Another limitation for quantification accuracy is multi-hit performance (see

Chap. 3 for a more detailed discussion). Some ion types (e.g., boron, carbon, and
others) have a significant probability of co-evaporation [59]. They tend to evap-
orate from the same region on the specimen surface at the same time as another ion.
When these co-evaporated species have different TOF, they will be detected at a
rate equal to the detection efficiency. When they have the same TOF, then there is a
probability that one of these species will not be detected, leading to an under-
measurement of this ion type and inaccurate compositions. Experimental parameter
selection (primarily detection rate) affects multi-hit performance and should be
appropriately considered.
Lastly, the generation of complex ion species can also affect quantification
accuracy [10, 11]. The potential overlap of spectral peaks (N+ and N2++ for
example) can lead to improper identification or difficulty in accurately separating
a peak into multiple ion types (as well as 3D visualization challenges). Again,
experimental parameter selection (primarily laser pulse energy or (LPE)) can affect
complex ion generation and should be carefully considered.
Ideal data quality would have mass peaks with zero width, zero background,
accurate measurement of all multi-hit events, and simple ion identification with all
evaporated ions arriving at the detector before the next pulse. In practice, this ideal
is never achieved and any given set of acquisition conditions results in a tradeoff
among these data quality metrics and specimen yield, which is discussed in the next
section.
4.3 Analysis Yield Considerations
In an APT experiment, the specimen is subjected to a very high electric field which
produces a very high stress proportional to the electric field squared. Various authors
have calculated the stresses on APT specimens using different methods
[1218]. These stresses can produces elastic strains of 12 % [19] and may be
sufficient to induce motion of unpinned defects such as dislocations [20, 21] and
cause in-situ twinning [19, 22]. Field-induced stress is also a cause of fracture in
specimens, which require a certain amount of elastic strength to produce good yield,
and as such we are interested in practical ways to minimize the effects of stress on
APT specimens. Numerical and analytical estimates of the magnitude of such stress,
similar to calculations from Moy et al. [18], are shown in Fig. 4.1. Note that the
analytical results closely match the previous results from Birdseye and Smith [12].
For specimens with specific regions of interest, a reasonable assumption is that
any specimen will yield relative to the ratio of the adhesion in these regions and the
stress generated by the electric field required to field evaporate the material. It is
often the case that regions of specimen surface which deviate from uniformity can
act as stress concentrators, a notorious example of these being the intersection of an
interface with the specimen surface. In order to improve yield, one can either reduce
the applied stress or increase the effective adhesion.
Fig. 4.1 A plot of the

electric field-induced stress
as a function of distance
from the apex for an atom
probe specimen of typical
dimensions. The stress on
the specimen is proportional
to the voltage (field is
proportional to voltage)
and, as the graph shows,
only the top few tens of
nanometers of the film stack
experience the highest
stress (Reproduced from
[18] with permission from
Ultramicroscopy)
General methods to decrease evaporation field include the following:

1. Use laser pulse mode rather than voltage pulse mode
2. Increase LPE in laser pulse mode
3. Decrease ion detection rate
4. Increase specimen base temperature
5. Increase the background (or other) gas pressure in the vacuum chamber
When laser pulsing rather than voltage pulsing is used to induce evaporation (see
Fig. 4.2) at a constant specimen base temperature (T1), a reduction of the evapo-
ration field from F1 to F2 is achieved (with the arrow LPE1 representing the laser
energy magnitude). In this case the specimen temperature is briefly raised by the
laser pulse, with the evaporation taking place at T2.
Method 2 increases the laser energy from LPE1 to LPE2 so that the evaporation
field is further reduced to F3, with the evaporation taking place at T3, as shown in
Fig. 4.2. It should be noted that caution should be used when using high LPE as
higher temperature evaporation may result in complex molecular ion species
(particularly in semiconductors) and other potential artifacts in the mass spectra,
as well as loss in spatial resolution due to laser overheating [23]. The charge state
ratio of detected ions often may be used as a gauge to determine if laser energies are
too high (see Sect. E.3.4 for discussion).
Method 3 above is the easiest to employ and applies to both voltage pulse and
laser pulse modes of operation. In the case of voltage pulsing, a decrease in the ion
detection rate (from rate 1 to rate 2 in Fig. 4.3) while holding the specimen base
temperature constant (T ) results in a corresponding decrease in the total applied
4.3 Analysis Yield Considerations 83
Fig. 4.2 Constant detection

rate curve showing field as a
function of temperature. For
a specimen analyzed under
constant detection rate
conditions, an increase in
temperature (T1T2T3
brought about by an
increase in laser pulse
energy) causes a reduction
in field (F1F2F3)
Fig. 4.3 Iso-detection rate

curves of field as a function
of temperature. For a fixed
temperature (T ), higher
rates (DR2 to DR1) require
higher fields (F2F1)
specimen field (from F1 to F2). Increasing the base temperature (method 4) also
applies to both voltage pulsed and laser pulsed modes.
Method 5 is seldom employed in the current generation of atom probes, although
hydrogen and some noble gasses have been shown to decrease the evaporation field
of some metals and semiconductors by 10 % or more [24].
Yield may also be increased by improving adhesion within specimens or by
changing the interaction between the applied stress and the inherently weak regions
in the specimen. Specimens with inherently weak interfaces will very seldom run
successfully in APT analysis, as shown by the example in Fig. 4.4a where specimen
failure has resulted in fracture at an interface. In Fig. 4.4b the stresses applied to a
specimen during field evaporation have resulted in shearing along {111} planes.
Fig. 4.4 (a) An example of a failed APT specimen that appears to have delaminated at a film
interface without destroying the rest of the tip and (b) a Ni-4Al alloy specimen which has sheared
on the {111} plane due to field-induced stresses (image in (b) courtesy D.K. Schreiber and
M.J. Olszta)
General methods to increase yield in these cases include:

1. Change the orientation of the features of interest (e.g., interfaces) in the
specimen
2. Remove materials which are not of general interest (and have markedly different
evaporation characteristics)
3. Coat (or etch and coat) the specimen with a layer of material
4. Anneal the specimen
Method 1 uses the cross-section (also referred to as the parallel orientation)
mode of specimen preparation (see Chap. 2) [25], which changes the orientation of
the interfaces in a specimen, as shown Fig. 4.5a. Empirically it is observed that
changing the orientation of interfaces in this way results in improved yield (note
that this also modifies the spatial resolution achieved during data collection).
Another improvement may be made simply by removing regions of the specimen
which are not of interest, method 2 above. Figure 4.5b illustrates this through use of
the concept of focused ion beam (FIB) deprocessing [26, 27]. Using the FIB,
materials which are not of interest are removed and replaced with a suitable capping
material (often nickel or silicon) which then form a portion of the final specimen.
Method 3 involves coating a sharpened, but not field evaporated, specimen with
a thin layer or layers of material. Figure 4.5c shows an example of this concept
using a bilayer of palladium and silver [28]. The palladium provides adhesion
for the silver, which provides good thermal diffusivity properties and
generally improves the mechanical strength of the specimen. By increasing the
cross-sectional area of the specimen with coated material, the overall stress
4.4 Experimental Parameters 85
Fig. 4.5 Reorienting interfaces or layers can lead to improved analysis yield. (a) Multi-layer films
can be prepared in cross-section (film surfaces parallel to the analysis axis) thereby changing the
relative orientation between the applied force and the interface. (b) Samples with differing
material regions can be simplified by removing undesired material and replacing it with more
APT-friendly materials leading to improved analysis yield. (c) An example of an engineered
coating which encapsulates the region of interest and changes the net electrical and thermal
properties of the specimen promoting positive effects on analysis yield
experienced by the specimen may be lowered as well. Additionally, any specimen

may realize an improvement in yield by an annealing process, method 4 above,
however this may cause unwanted diffusion in the specimen [29].
4.4 Experimental Parameters
The philosophy behind the experimental control of an atom probe is to provide the
user with parameters that combine ease of operation with good data quality. Ideally
an APT experiment would extract data from a material with optimum quality and
minimum analysis time. Because each specimen may be unique and all material
types do not perform similarly under APT analysis conditions, the user is often
required to choose control parameters that properly balance data quality, analysis
time, and analysis yield. This section discusses the various acquisition parameters
and algorithms, both basic and advanced, and provides some pointers for the
general LEAP user.
The general experimental parameters are listed in Table 4.1 and include pulse
rate ( f ), base temperature (T ), detection rate (DR in % of pulses that result in a
detected field evaporation event), pulse fraction (PFvoltage APT only
amplitude of the voltage pulse expressed as a percentage of the current DC voltage),
laser pulse energy (LPElaser APT only), and voltage (V ). A number of things
should be considered when selecting specific parameters.
4.4.1 Pulse Rate
The pulse rate is selectable by the user and, for most materials, the best rate is the
maximum rate that enables the m/n range of ions to be captured within a single
pulse window for all acquisition voltages. This keeps acquisition time to a mini-
mum and reduces the opportunity for out-of-synch evaporation to occur within the
pulse window. Because the m/n range is determined by the maximum TOF of
the ions and the TOF for a particular ion species decreases as voltage increases, the
recorded m/n range is both pulse rate and voltage dependent as shown in Tables 4.2,
4.3, and 4.4. Very high pulse rates may result in long TOF ions being detected
during subsequent pulses. A wrap-around effect is then observed with the large m/n
signal from one pulse detected on the next pulse window and observed as a broad
peak in the lower m/n range of the mass spectrum. This signal is generally not
recoverable as identifiable ions for reconstruction and chemical quantification.
Some materials with poor thermal diffusivity analyzed under high-energy laser
pulsing conditions may not cool sufficiently between heating pulses. This causes the
base temperature of the near-apex region of the specimen to remain above the set
base temperature during analysis and can result in long tails for peaks in the mass
spectrum. This base temperature offset effect may also be a function of pulse rate, so
slower pulse rates should be tried for materials exhibiting this behavior.
4.4.2 Base Temperature
APT is performed at cryogenic specimen temperatures to minimize the potential for

surface diffusion [2]. Ideally, the lowest possible base temperature provides the
highest data quality; however, other factors must be considered as well. In voltage
mode, some specimen types have significant yield differences within the range of
2080 K. In laser mode, the base temperature tends to have an effect on MRP and
background. Simple thermodynamics indicate that the lowest base temperatures
Table 4.1 General experimental parameter considerations, data quality trends, and tradeoffs (parameter increases ", trend improves +, or worsens the
metrics)
Acquisition General
parameter Variable trend Metrics Comments Primary tradeoff
Pulse rate " f (Hz) ++ Background Higher is better Almost none except for practical
m/n range Exceptions: high-mass ions considerations for TOF of all
Wrap around with long TOF or extremely potential ion species
Base temperature offset poor thermal diffusivity
4.4 Experimental Parameters
effects (laser) materials in laser mode

Base temperature " T (K) ++ Analysis yield Lower leads to better spatial Data quality vs. analysis yield
MRP resolution, Higher leads to
Background better yield
Heat flow (laser)
Surface diffusion
Detection rate " DR (ions/ ++ Background Higher is better Data quality vs. analysis yield
pulse) Analysis yield Exception: poor yielding mate-
MRP rials and materials prone to
Multi-hit performance multi-hit behavior
Pulse fraction " PF (%) ++ Compositional accuracy Higher is better Data quality vs. analysis yield
(voltage mode) + Background Exception: Extreme PF may
Potential for PF decay limit MRP and pulse ampli-
High cycle fatigue tude is hardware limited
Laser pulse energy " LPE (nJ or pJ) +++ Analysis yield Higher leads to better yield Data quality vs. analysis yield
(laser mode) ++ Background Lower leads to better spatial
+ MRP resolution
+ Effective pulse fraction
Complex ion generation
Surface migration
87
Table 4.2 Approximate LEAP Si platform maximum m/n ranges as a function of

pulse Rate (90 mm flight path)
Table 4.3 Approximate LEAP Si platform maximum m/n ranges as a function of

pulse rate (160 mm flight path)
result in maximum heat flow rates (maximum gradient) out of the apex of the tip,
resulting in higher MRP. Low specimen-stage temperatures can minimize the
mobility of residual gasses present in the analysis chamber, which improves the
vacuum levels and lowers spectral background. Many materials also have thermal
transport properties that improve near 20 K [30].
4.4 Experimental Parameters 89
Table 4.4 Approximate LEAP HR platform maximum m/n ranges as a function of

pulse rate (382 mm flight path)
4.4.3 Detection Rate
Once significant field evaporation is achieved, only small changes in the applied
electric field are necessary to change the DR because of the power-law behavior of
the response [3134]. Large DRs generally promote higher data quality (increased
signal-to-noise [19]). For many materials, this benefit is offset by decreased spec-
imen yield. For some elements (e.g., boron and carbon), increased DR also
increases the potential for lower detection efficiency due to an increase in the
proportion of multi-hit events [7, 3539].
4.4.4 Pulse Fraction (Voltage Mode)
PF is critical for good data quality as measured by compositional accuracy, MRP,

and background including out-of-synch ionization events [3, 4, 19]. To avoid
preferential evaporation in voltage pulse mode, the standing voltage must be low
enough so that there is nearly zero field evaporation, while the DC voltage plus
pulse voltage should provide similar probabilities of field evaporation for all
surface elements. A small pulse fraction will result in evaporation of low evapora-
tion field elements in between pulses. To avoid variation in analysis conditions that
might cause variability in data quality, constant pulse fraction is desirable, but it
should be noted that the LEAP pulser has a maximum pulse voltage, and conse-
quently, for PFs >10 %, a constant PF cannot be achieved over the entire instru-
ment voltage range. Lastly, the repeated application of a pulsed field results in
repeated stresses that may lead to cycle fatigue and specimen failure [29].
4.4.5 Laser Pulse Energy (Laser Mode)
The LPE is a measure of the total photon energy carried by each pulse. This energy
is spread nonuniformly over the laser spot resulting in some unknown energy
density at the specimen tip. Focus quality, spot size, and laser-specimen alignment
will influence the energy density experienced by the specimen. Additional proper-
ties of the specimen (e.g., size, shank angle, composition, reflectivity, absorption
coefficient, thermal diffusivity, and band gap) influence how energy is absorbed,
converted to heat, and removed from the apex region of the tip [2]. Because the
evaporation field required to evaporate ions from the surface of a specimen is
temperature dependent, the thermal pulse created during laser pulsed acquisition
lowers the required applied field, and thus the stress on the specimen, leading to
significant improvement in analysis yield. High LPE generally results in superior
yield, improved MRP, and low background. This provides an incentive to operate at
the highest LPE that does not result in other negative effects (e.g., degraded spatial
resolution).
Insufficient LPE can result in negative effects similar to those resulting from
insufficient pulse fraction in voltage pulse mode. High rates of out-of-synch
evaporation or preferential retention can result when there is too small a field
difference for specimen elements at base temperature and maximum pulse temper-
ature. Also note that peak apex temperature (and resulting required evaporation
field) generally changes as a function of voltage because the absorption properties
and heated volume (which impact apex temperature) evolve with specimen size.
The positive impact on yield and spectral quality suggests higher LPE settings,
but extreme LPE raises concerns regarding surface migration, complex ion gener-
ation, and nonuniform spatial evaporation behavior, negatively impacting the
ability to reconstruct ion positions accurately. In addition, application of laser
pulsed APT to oxides and nitrides suggests that minimal laser energies may be
required in order to obtain correct stoichiometry [40, 41]. One must take this all into
consideration when designing experiments to analyze multiple specimens under
equivalent conditions.
4.4.6 LEAP Parameter Ranges
Each LEAP platform has a unique set of hardware dependent parameter ranges.
Note that some performance metrics evolve as computer power steadily increases.
Hardware parameter ranges are summarized in Table 4.5.
4.5 How to Start Your Investigation of Any New Material 91
Table 4.5 LEAP platform parameter ranges

Specification/model 3000 4000
Voltage pulse rate 1200,000 Hz 1200,000 Hz
Laser pulse rate (HR) 250 kHz Maximum 250 kHz Maximum
Laser pulse rate (Si) 500 kHz Maximum 1,000 kHz Maximum
Minimum base temperature 25 K 25 K
Detection ratea 33 kIons/s live RHIT 33 or 83 kIons/s live RHIT
100 kIons/s post RHIT 100 or 250 kIons/s post RHIT
Pulse fraction (voltage mode) 1.4 kV Si, 1.7 kV HR 1.4 kV Si, 1.7 kV HR
Laser pulse energy (laser mode) 0.012 nJ at 532 nm 0.0011,000 pJ at 355 nm
a
Dependent on manufacture date
4.5 How to Start Your Investigation of Any New Material
The remainder of this chapter operates under the assumption that the user is familiar
with LEAP calibration files, acquisition control algorithms, and local-electrode
maintenance. If this is not the case, it might be a good idea to review the informa-
tion contained in Appendix E.
When attempting APT analysis of a material, the analysis conditions must be
appropriately selected. For materials that are slight variations of previously ana-
lyzed materials analyzed under known conditions, these known conditions make for
a sensible starting point. When attempting analysis of a new material where little
knowledge is available regarding APT analysis, consider the following suggestions:
assume the material is difficult to field evaporate, minimize the necessary evapo-
ration field, vary analysis conditions to obtain data, be prepared to make more
specimens than you think will be adequate, and optimize conditions once minimal
yield has been achieved.
1. Assume that the new material type is difficult to field evaporate. The expectation
is to go through a number of specimens before the analysis objectives are
achieved. It is difficult to gain any insights from specimens that repeatedly fail
before yielding any data, thus conditions that improve yield at the expense of
data quality considerations are often considered as a starting point. Initially, high
data quality (step four) is secondary to analysis yield.
2. Minimize the field necessary to field evaporate the material. Using reasoning that
the lower the evaporation field, the lower the stress on the tip and the greater the
likelihood for successful data collection suggests a certain acquisition parameter
space. Start with a moderate base temperature (50100 K), low DR (0.2 %), and
relatively high LPE. The actual LPE depends on the energy density of that
particular instrument, wavelength, and previous experience with related mate-
rials. Be aware that relatively high LPE for one material is not necessarily the
same as for another (metals versus semiconductors versus oxides for example).
3. Vary conditions to obtain data. Do not get too limited in your parameter
selection. If the first tip fails immediately under a set of conditions, then try a
more extreme parameter selection. A combination of higher base temperature,

higher LPE, and lower DR usually leads to improved yield.
4. Optimize conditions once minimal yield has been achieved. Once data are
acquired, the analysis conditions can be adjusted until the right balance of
yield and data quality is achieved. This is usually a combination of lower base
temperature, lower LPE, and higher DR. It is often advantageous to reconstruct
and quickly investigate data just obtained in order to evaluate it for artifacts,
which then provides input into the choice of parameters for the next specimen.
See [3] for an example on determining acceptable PF and base temperature
conditions in voltage pulse mode.
4.6 Brief Overview of LEAP Operation: Data Collection
Once a specimen has been fabricated, loaded into the LEAP, transferred onto the
cryogenic specimen stage, and coarsely aligned to the local electrode, data collec-
tion can begin. For the uninitiated, the LEAP User Manual provides detailed
descriptions of specimen transfer, coarse alignment, tool operation, and the soft-
ware that is used to accomplish these tasks. Separate software applications are used
to control the instrument and display the critical live acquisition data displays: the
LEAP Control Center (LCC) and the Data Acquisition & Visualization
(DAVis) applications. This section focuses on data collection from the time a
voltage is applied and the detector begins to register hits, to the final storage of data
files (see Appendices A and E). Consequently, only relevant LCC and DAVis
features will be covered. This section is not meant to be comprehensive, but instead
to provide the user with the essentials for getting an experiment under way and
running smoothly. Because the steps and considerations are slightly different for
voltage pulse and laser pulse mode acquisitions, each will be covered in turn.
The general flow of a LEAP acquisition is as follows: (1) the specimen puck is
transferred to the specimen stage and the base temperature is selected; (2) during
the approximately 10 min it takes for the specimen to cool, the specimen is coarsely
aligned to the local electrode using the optical microscope cameras; (3) appropriate
setup parameters are selected;(4) once the specimen temperature has reached
equilibrium, acquisition may be started. During the initial stages of the acquisition,
fine adjustments of the specimen position are made to appropriately center and fill
the detector with ions emitted from the specimen surface. Some users will use the
cool down time to start acquisition and alignment to take advantage of evaporating
through any surface oxides while still at an elevated temperature and thus minimize
early fracture. If the Auto Detection Rate Control has been selected, the instrument
attempts to maintain a voltage such that the actual DR remains near the target rate.
The experiment is terminated based on criteria set by the user (discussed later) or
when the instrument determines that the specimen has failed and self-terminates.
Detailed procedures for voltage pulsed and laser pulsed acquisition follow.
4.6 Brief Overview of LEAP Operation: Data Collection 93
4.6.1 Voltage Acquisition
Check List 1: Voltage Mode Acquisition.

1. Load specimen
2. Visually align specimen to LE
While waiting for base temperature to stabilize . . .
3. Adjust advanced control settings
Acquisition algorithm
Interface Evaporation Control (IEC)
4. Adjust initial setup parameters
DR (typically near the minimum allowed rate)
Initial voltage (500 V unless restarting an acquisition)
PF (user preference)
Auto Detection Rate Control (user preference)
Once base temperature is stable . . .
5. Start acquisition
6. Fine alignment of specimen
7. Adjustments during acquisition
Auto Detection Rate Control
DR changes (increase slowly)
Stopping conditions (note LE turn-on limits)
Notifications
Consider a specimen that is ready for analysis in voltage pulse mode. Before
initiating the acquisition, a number of parameter fields should first be evaluated.
The Voltage Atom Probe tab in LCC contains sections labeled Setup Parameters,
Run Statistics (Fig. 4.6a), and Advanced Controls (Fig. 4.6b). The Setup Parameters
include Detection Rate, Specimen Voltage, Pulse Rate, Pulse Fraction, and Cali-
bration as well as a number of check-boxes which enable Auto Detection Rate
Control, stopping conditions (Stop Elapsed Time, Stop Total Events, Stop Speci-
men Voltage), and Acquisition eMail Notifications. The following set of steps for
selecting important parameters and initiating analysis:
1. Load specimen (see Chap. 3)
3. Make adjustments to the Advanced Control settings.
(a) Algorithm: Choose an algorithm. The most notable impact of this choice will
be minimum DR and initial voltage ramp rate. The minimum DR should be
~0.2 % for fragile specimens. Otherwise select according to personal pref-
erence with the option to change during acquisition.
Fig. 4.6 The LCC voltage atom probe tab: (a) Setup Parameters and Run Statistics and (b)
Advanced Controls
4. Make adjustments to the initial Setup Parameters.

(a) Detection Rate (%): Start out with the lowest rate available. As the specimen
starts to emit ions, this rate can be increased. Analyze at the highest rate
possible that does not negatively impact yield.
(b) Specimen Voltage (V ): The default starting voltage of 500 V is the safest.
Higher voltages can be used in cases where the specimen is known to begin
ion emission at much higher voltages (for example, when a specimen has
already been analyzed to some voltage and is being reanalyzed).
Fig. 4.6 (continued)

(c) Pulse Rate (Hz): Choose the highest pulse rate that results in the heaviest
element arriving within a single pulse window (see Tables 4.2, 4.3, and 4.4).
(d) Pulse Fraction (%): Typical choices are 1030 %. Depending on the mate-
rial, this choice could impact compositional accuracy as insufficient pulse
fraction can lead to low-field species evaporating between pulses. Also,
there is a maximum pulser voltage (instrument dependent). Pulse fractions
greater than ~10 % may not be possible for the top end of the 15 kV range of
the instrument DC voltage. For blunt specimens or specimens with high
evaporation field requirements, the user should consider this limitation. This
parameter cannot be changed during acquisition.
(e) Auto Detection Rate Control check-box: Most users will leave this selected
and the voltage will immediately ramp according to the selected control
algorithm (see Appendix E) until the desired DR is achieved. Advanced
users may choose to leave this unchecked and manually increase the voltage
until some ion emission is observed allowing for fine adjustment of the
specimen position while the voltage remains constant. This control feature
may be toggled on and off during acquisition.
5. Start the acquisition by left-clicking on the Start button.
6. Fine specimen alignment.
(a) Detector Event Histogram: As the voltage ramps (either manually or auto-
matically) pay close attention to the live data plots in DAVis. Of particular
interest is the Detector Event Histogram/Detector Ion Map. Any
misalignment of the specimen to the local electrode becomes clearly evident
as events accumulate on the detector.
(b) Lateral misalignment: This is when the circular shadow of the local elec-
trode is not centered on the detector, Fig. 4.7a. Adjustments are made with
specimen motion control arrows (Fig. 4.7 inset) to move the pattern to the
center of the detector (Fig. 4.7b). Some caution is prudent here in that
motion of the tip while under high field may promote specimen failure.
Adjustments should be completed before the voltage is high enough to
initiate significant field emission (less than about 0.2 % DR).
(c) Specimen-to-local-electrode separation: When the specimen is too close to
the local electrode, the ion pattern overfills the detector so that the circular
shadow of the local electrode does not blank out the outer edge of the data as
shown in Fig. 4.7b. Again, while the voltage is relatively low, the specimen
can be moved away from or toward the local electrode by using the speci-
men motion control + (closer separation or larger ion pattern) and
(farther separation or smaller ion pattern) buttons. It is important to note that
the motions toward the local electrode create rapid increases in the field
experienced by the specimen, possibly promoting tip failure. For especially
delicate specimens (most specimens) it is prudent to manually reduce the
voltage by a few percent, move the specimen closer to the electrode, and
then increase the voltage slowly until minimal ion emission allows you to
reassess alignment using the ion pattern.
Fig. 4.7 Fine alignment of

the specimen to the
LE. (Top) DAVis detector
event histogram showing
the circular emission from
the specimen needs to be
moved down, to the left, and
towards the LE (+). (inset)
Specimen-stage motion
arrows used to guide the
projected image to the
center of the detector
histogram. (Bottom)
Properly aligned specimen
showing an event histogram
that is centered and slightly
under-filling the detector
areathe shadow of the LE
opening masks off the edge
of the detector where only
low-density (nonzero)
background signal can be
observed
7. Adjustments during acquisition. Now that fine alignment of the specimen is

complete, final adjustments can be made by the user and fully automated
acquisition and termination can be implemented.
(a) Auto Detection Rate Control check box: This should now be toggled on. The
detection algorithm can be changed during the run as desired.
(b) Detection Rate (%): The rate can be slowly increased to the desired final rate.
Although the voltage increment is relatively small for increases in target
rate, increasing the detection rate, and thus the electric field, can impact the
longevity of the analysis. Besides having a negative impact on yield, this
parameter positively affects data quality and total time for data collection, so
it is important to experimentally investigate these tradeoffs and optimize this
parameter when frequently analyzing a particular material system.
(c) Stopping conditions and email notification: Before leaving the instrument
unattended, it is advisable to decide on termination conditions. Choosing
reasonable Stop Total Events and Stop Specimen Voltage values is prudent.
Often the voltage stop value may be influenced by the current local-electrode
flat test value (see Appendix E). Likewise, setting up the electronic mail
notifications to make notification when a specimen has completed acquisi-
tion (or failed), or met some other threshold in the Advanced Controls tab
will promote maximum utility from the LEAP.
Some parameter ranges discussed above are hardware dependent and those
generally achievable for the LEAP 3000 and 4000 platforms are listed in Table 4.5.
4.6.2 Laser Acquisition
Consider a specimen that is ready for analysis in laser pulse mode. In voltage pulse
mode, the only alignment concern is the specimen position with respect to the local
electrode. In laser pulse mode specimen alignment is required as well, but now the
laser must also be aligned to the specimen. Any movement of the specimen position
requires realignment of the laser to the new specimen position, so very careful
control and monitoring is required during fine alignment where specimen position
adjustments are followed by laser adjustments. After satisfactory positioning of the
specimen relative to the LE, the laser focus must be adjusted so that optimum laser-
pulse energy density is achieved. Moving the specimen and laser and adjusting the
focus while the specimen is under high field increases the chance for specimen
fracture. Unless there is some confidence that the current position and focus of the
laser is approximately correct (e.g., a specimen has just been analyzed in laser pulse
mode and it was known to be well positioned and focused and the LE has not been
changed), then a high yielding, calibration specimen (pre-sharpened silicon
microtip) should be used to pre-align the laser to the local electrode. Following
this recommendation, there are two procedures for laser pulse mode operation:
(1) laser pre-alignment, and (2) laser acquisition.
Check List 2: Laser Mode Pre-alignment.

1. Acquire data from pre-sharpened silicon microtip to properly align spec-
imen to LE (see Voltage Mode Acquisition)
2. Modify laser camera focus until specimen apex is in best focus
3. Align laser to specimen apex (refocus camera as necessary)
4. Acquire data (see Laser Mode Acquisition steps 38 skipping step 6)
Pre-alignment is complete after successful laser focus scans
As with voltage pulsed mode, a number of parameter fields should first be

considered. The Laser Atom Probe tab in LCC contains sections labeled Setup
Parameters (Fig. 4.8), Run Statistics (Fig. 4.8), and Advanced Controls. The setup
parameters again include Detection Rate, Specimen Voltage, Pulse Rate, and
Calibration as well as a number of check-boxes which enable Auto Detection
Rate Control, stopping conditions (Stop Elapsed Time, Stop Total Events, Stop
Specimen Voltage), and Acquisition eMail Notifications. Instead of Pulse Fraction,
laser pulse mode requires a Pulse Energy setting.
The following is a set of steps for pre-aligning, selecting important parameters,
and getting a laser pulsed analysis under way. First, a procedure is described for
pre-alignment. Then, a procedure is described for starting a specimen after
pre-alignment.
Laser Pre-alignment (alignment of laser to LE)
1. Load and position a pre-sharpened Si microtip specimen for voltage pulsed
acquisition.
2. Align specimen to local electrode in voltage pulse mode by following the
procedure described in the section above for voltage pulse acquisition through
the fine alignment step (step 4). This ensures that the apex of the specimen is
well aligned with the local electrode. It is only necessary to gather enough ions
to ensure that the specimen is properly centered and has the proper spacing to the
local electrode. The specimen is now properly positioned with respect to the LE,
so the specimen should not be moved in subsequent steps.
3. Visually align the laser to the specimen apex.
(a) Choose appropriate laser pulse energy: Because the laser is unlikely to be in
ideal focus, a larger laser spot leads to lower energy density. The way to
compensate for this is by choosing a higher than normal pulse energy. Pulse
energies of ~0.3 nJ and ~40 pJ are appropriate for the LEAP 3000X and
4000X, respectively, when the laser is focused. Starting out with twice those
values is suggested for pre-alignment (this can be adjusted toward the 0.3
and 40 pJ values as the focus improves). In some cases, when the laser is far
from focus, three times these values or more may be necessary. Silicon
microtips are typically used for alignment as they are very tolerant of
extremely high and low laser pulse energy densities compared to almost
any other specimen, but they are not indestructible, so use caution.
Fig. 4.8 The LCC laser atom probe tab showing the Setup Parameters and Run Statistics displays
(b) Manual Alignment: This is accomplished by first moving to the Laser Atom
Probe tab in LCC and left-clicking on the Manual Alignment button. This
will bring up the alignment window. Align the laser to the specimen by
moving the laser until maximum reflected light is observed from the spec-
imen apex.
Camera Focus: The camera laser focus should not be modified unless
alignment is being performed on a newly exchanged LE. For a newly
exchanged LE, the apex of the aligned specimen viewed in the Laser
Targeting camera should be in sharp focus. Manipulate the camera focus
on the left of the Laser Targeting window until the tip apex is in sharp
focus. Remember, the laser passes through this camera, so changing the
camera focus also changes the focal position of the laser spot.
Left-click the Beam On/Off button at the lower right.

Ideally the laser beam will be seen to reflect off some surface within the
camera view. If this is the case, then it is just a matter of using the arrows
to move the laser beam until it is seen to reflect off the specimen apex.
If turning on the laser beam provides no obvious reflection of the laser
spot, manually scan the laser using the Laser Beam Position arrows until
the position of the laser can be determined (a reflection somewhere on the
specimen or local electrode). Note that moving the laser coarse axis
moves the final objective lens so the laser position with respect to the
image center does not change. Proper laser position is always somewhere
near the center of the camera view. When struggling to find the current
laser position, drive the laser position toward the local electrode and then
scan up and down. Some sort of reflection off the local electrode should
be easy to obtain. Then, move the laser to the apex of the specimen. For
final positioning of the laser, the Fine Axis can be enabled and then the
high-precision scanning mirrors are used to control the laser position.
Left-click the OK button to return to the Laser Atom Probe tab.
(c) Fine alignment of laser position and focus. This follows the same procedure
as steps three to seven for Laser Acquisition below with the following
exceptions:
Ignore any suggested motions of the specimen during description of the
fine alignment procedure. The specimen is known to be in the correct
position and should not be moved. Only the laser needs to be scanned to
find the optimal tip position and focus.
Do not use IEC. The laser scanning steps during alignment will often
trigger the IEC and will substantially slow down this process.
Run at higher target DRs. All the scanning algorithms work better at
higher rates. Slowly turn up the rate to ~3 % before doing focus scans.
4.6.2.1 Laser Acquisition
Check List 3: Laser Mode Acquisition.

1. Load specimen
While waiting for base temperature to stabilize. . .
3. Adjust advanced control settings
Acquisition algorithm
IEC (may prefer to disengage during alignment)
(continued)
Check List 3: Laser Mode Acquisition (continued)

4. Adjust initial setup parameters
DR (typically near the minimum)
Initial voltage (500 V unless restarting an acquisition)
LPE (extra energy is necessary if nonoptimal focus)
Auto Detection Rate Control (user preference)
5. Check advanced laser tab
Confirm Scout, Standard, and Drift Comp scan and focus settings
Once base temperature is stable. . .
6. Align the specimen to the laser
7. Start acquisition
8. Fine alignment of specimen (make only small motions before rescanning)
9. Adjustments during acquisition
Auto Detection Rate Control
IEC (confirm desired status)
DR changes (increase slowly)
Consider focus scan frequency
Stopping conditions (note LE turn-on limits)
Notifications
1. Load specimen.
2. Visually align specimen to LE.
3. Make adjustments to the Advanced Laser settings (this is a separate tab in the
Advanced Controls panel).
(a) Algorithm: Choose an algorithm. The most notable impact of this choice will
be minimum DR and initial voltage ramp rate. The minimum DR should
be ~0.2 % for fragile specimens suggesting either algorithm 4 or 5
(see Appendix E). Otherwise select according to personal preference.
4. Make adjustments to the initial Setup Parameters.
(a) Detection Rate (%): See voltage acquisition.
(b) Pulse Rate (Hz): See voltage acquisition.
(c) Specimen Voltage (V ): See voltage acquisition.
(d) Pulse Energy (pJ or nJ): Different material types require different pulse
energies. The heating pulse requirements for field evaporation and the
material properties that effect absorption all play a part here. Here are
some things to consider:
Specimens blunt as they are analyzed thus increasing their heated vol-
umes. Consequently, specimens often require lower pulse energy at the
beginning of an analysis.
The lower the evaporation field needed to initiate field evaporation, the
less likely it is that the specimen will fail. A lower rate of failure may
often be achieved by utilizing higher pulse energies at startup.
Any out-of-focus situation will require extra pulse energy.
Overall it is probably safest to start a specimen with extra laser pulse
energy. Once steady state analysis is achieved, be prepared to make
adjustments.
(e) Auto Detection Rate Control check box: See voltage acquisition.
5. Check the Advanced Laser tab.
(a) Be aware of the Tip Scan Modes: Scout, Standard, and Drift Comp
(b) Be aware of the Focus Scan Modes: Scout, Standard, and Drift Comp
6. Align the specimen apex to the laser:
(a) This is accomplished by first moving to the Laser Atom Probe tab in LCC
and left-clicking on the Manual Alignment button at the lower right of the
window (see Fig. 4.6b). This will bring up the Specimen and Laser Beam
Alignment window.
(b) Left-click the Beam On/off button.
(c) Align the specimen to the laser by moving the specimen using the stage
control arrows at the left of the window until maximum reflected light is
observed from the specimen apex. The laser should be reflected from the
same location relative to the local electrode as is required of all specimens.
Any difficulty in locating the laser suggests something is wrong (specimen is
not centered in both camera views, laser beam is not on, poorly pre-aligned
laser).
(d) Left-click OK to return to Laser Atom Probe tab.
7. Start the acquisition by left-clicking on the start button.
8. Fine specimen alignment:
(a) Detector Event Histogram and Mass Spectrum: As the voltage ramps (either
manually or automatically) pay close attention to the live data plots in
DAVis. Of particular interest is the Detector Event Histogram.
Misalignment of the specimen to the local electrode becomes clearly evident
as events accumulate on the detector. New users should be cautious here
when attempting operation at higher frequencies. Voltages can ramp very
quickly and it is possible to attain voltages in excess of the voltage required
for the desired DR when the laser or specimen is misaligned. This can lead to
immediate specimen failure during tip scanning. In addition, monitoring of
the Mass Spectrum provides information regarding the number of real
ion-based events versus background noise or out-of-synch events.
(b) Laser misalignment: Perform scout scans regularly until the specimen emits
strongly enough so that the laser scan snaps the laser position to the
specimen apex (scan at the beginning and every 5001,000 V to ensure
that the alignment is achieved before the applied field is strong enough to
yield significant field evaporation).
Note that the scan should start beyond the specimen such that you cannot
easily see any reflection as the scan starts. At the end of the scan, there
should be strong reflections as the scan goes across the specimen. If this is
not the case, consider going back to the initial alignment steps.
Note that some emission from the specimen may be observed before the
scout laser scan is completed. If the emission is significant enough such
that you can judge the specimen to be misaligned, then align specimen as
described in the next step and immediately rescan the laser. It is possible
that the tip position is so misaligned that the scout scan only appears to
scan over the tip, but the specimen is actually outside the scan area.
(c) Lateral misalignment: This is when the circular shadow of the local elec-
trode is not centered on the detector, Fig. 4.7a. Make adjustments with
specimen motion control arrows (Fig. 4.7 inset) to move the pattern to the
center of the detector (Fig. 4.7b). Some caution is prudent here in that motion
of the tip while under high field may promote specimen fracture. Adjust-
ments should be completed before the voltage is high enough such that the
specimen has reached a fully developed field evaporated endform. In addi-
tion, only small adjustments in position should be made before realigning the
laser. Large motions may exceed the range of laser scans making laser
realignment difficult without some manual adjustment of the laser position.
(d) Realign laser: After any specimen motion, the laser will need to be realigned
to the specimen apex. Repeat steps 7a, b, and c until the specimen is properly
centered within the local electrode and the laser has properly snapped to the
specimen apex due to a strong, sharp emission spike in the tip scan.
(e) Specimen-to-local-electrode separation: When the specimen is too close to
the detector, the ion pattern overfills the detector so that the circular shadow
of the local electrode does not blank out the outer edge of the detector, as
shown in Fig. 4.7b. While the voltage is relatively low, the specimen can be
moved away from or toward the local electrode by using the specimen
motion control + (smaller separation or larger ion pattern) and (larger
separation or smaller ion pattern) buttons. It is important to note that the
motions toward the local electrode create rapid increases in the field expe-
rienced by the specimen, possibly promoting tip failure. For especially
delicate specimens (most specimens) it is prudent to manually reduce the
voltage by a few percent, move the specimen closer to the electrode, and
then increase the voltage slowly until minimal ion emission allows you to
reassess alignment by way of the ion pattern.
(f) Realign laser: After any specimen motion, the laser will need to be realigned
to the specimen apex. Repeat steps 7ad until the specimen is properly
centered within the local electrode and the laser has properly snapped to
the specimen apex due to a strong, sharp emission spike in the tip scan.
(g) Fine laser alignment: Once the specimen is in the proper position and the
laser has successfully identified the apex via a scout scan, a standard tip scan
at higher scan resolution is necessary to properly place the laser on the
specimen apex. After a successful standard scan, a drift comp scan is used to
make fine adjustments to the laser position. In this mode, the laser never
fully leaves the specimen but only makes small excursions that allow the
control algorithm to make small corrections to the laser position.
(h) Focus Scanning: A good pre-alignment likely ensures that the laser is
focused to within a couple of micrometers of the current specimen position.
For the LEAP 3000X that is likely sufficient since the laser spot size does not
change significantly over a couple micrometers. This is not as true for the
LEAP 4000X. A couple of micrometers misalignment can significantly
reduce the energy density. Therefore, it is recommended that a focus scan
be performed on the 4000X (optional for the 3000X) to place the laser focus
at the specimen apex.
The scout focus scan should be unnecessary if an adequate pre-alignment
has been performed.
Start with a standard scan.
Finish with a drift scan.
9. Adjustments during acquisition: Now that fine alignment of the specimen and
laser is complete, final adjustments can be made by the user and fully automated
acquisition and termination can be implemented.
(a) Auto Detection Rate Control check box: See voltage acquisition.
(b) Auto Tip Scan Settings and Auto Scanning settings in the Advanced Laser tab:
See the Laser Positioning Control section in Appendix D for a discussion
regarding appropriate scan settings. Default scan settings are a good starting
point, but these should be customized depending on sample and laboratory
environment needs. Changes in room temperature should be avoided in
particular as this may cause significant drift of the laser focus position
with respect to the specimen. Monitor auto operation for many scans before
leaving the tool unsupervised in automated mode. Scans that regularly
require laser repositioning to the edge of the scan area suggest that more
frequent and/or wider scans are required.
(c) Detection Rate (%): See voltage acquisition.
(d) Stopping conditions and email notification: See voltage acquisition.
4.6.3 Now You Are Atom Probing
Once automated acquisition is enabled and the specimen is emitting ions in a stable
manner, the LEAP can be left unattended; however, there are a few things one
should keep in mind before proceeding:
1. Select termination conditions. The LEAP will auto-terminate if the analysis

suggests the specimen has failed. Specimen failure should be avoided whenever
possible because each failed specimen may damage the LE. In addition, field
evaporated endforms can provide useful information. Measuring the radius of a
post-analysis specimen provides a constraint on the reconstruction. Direct
observation of the tip shape and knowledge of the deviations from a hemispher-
ical end form may lead to better understanding of reconstruction aberrations and
resulting limitations in reconstruction accuracy. For these reasons a user should
choose termination criteria (number of ions, voltage maximum, etc.) based on
acquisition objectives so that unnecessary specimen failure is avoided.
2. Be wary of local electrode limitations. With each successive specimen failure,
the LE is likely to have reduced performance (i.e. electron emission occurs at
successively lower voltages and can damage the specimen or reduce the data
quality). Ideally, the LE should be tested after each failure to ensure that it will
perform adequately for subsequent analyses. Commonly, a flat test will not have
been performed, so the user should be aware of signs of LE electron emission.
One common sign of this emission is a marked increase in background levels.
After implementing automated acquisition and discerning a normal background
level, it is advisable to select a background threshold and have the LEAP notify
you by email when the threshold has been exceeded, which is an indication of
possible LE turn-on. This allows the user to return to the instrument and
personally assess the current analysis. When degraded LE performance is dis-
covered, it might be possible to replace the LE and restart the analysis.
3. Monitor the instrument during analysis of new materials or materials prone to
failure. The available automated control algorithms are limited and manual
adjustment of analysis parameters during acquisition may lead to improved
specimen survivability. Anything that lowers the field while analyzing through
a failure prone region should increase yield (i.e., higher laser pulse energy and
lower DR). Do not hesitate to initiate this sort of manual control when analysis
yield is an issue.
4. In laser acquisition mode, be wary of laser pulse energy suitability. Insufficient
laser pulse energy requires evaporation at higher DC fields and can lead to higher
backgrounds and shorter specimen lifetimes. Excessive laser pulse energy can
reduce data quality by providing nonuniform areal evaporation and increased
molecular evaporation as well as unstable instrument control (DR fluctuates as a
function of voltage). These effects also change as a function of specimen size or
as the voltage increases. Some experimentation here is usually necessary. Just
because a specimen acts erratically at low voltage for a given LPE does not
imply that this LPE should not ever be exceeded. In fact, it may be an appro-
priate LPE once the specimen size evolves (required voltage increases).
5. Do not be afraid to experimentgood luck!
References 107
References
1. Bevington, P.R., Robinson, K.D.: Data Reduction and Error Analysis for the Physical Sci-
ences, 3rd edn. McGraw-Hill Higher Education, New York, NY (2003)
2. Gault, B., Moody, M.P., Cairney, J.M., Ringer, S.P.: Atom Probe Microscopy. Springer series
in materials science, vol. 160. Springer, New York, NY (2012)
3. Miller, M.K., Smith, G.D.W.: An atom probe study of the anomalous field evaporation of
alloys containing silicon. J. Vac. Sci. Tech. 19(1), 5762 (1981)
Materials Problems. Materials Research Society, Pittsburgh, NJ (1989)
5. Andren, H.O., Henjered, A., Norden, H.: Composition of MC precipitates in a titanium
stabilized austenitic stainless-steel. J. Mater. Sci. 15(9), 23652368 (1980). doi:10.1007/
bf00552329
6. Thuvander, M., Weidow, J., Angseryd, J., Falk, L.K.L., Liu, F., Sonestedt, M., Stiller, K.,
Andren, H.O.: Quantitative atom probe analysis of carbides. Ultramicroscopy 111(6), 604608
(2011). doi:10.1016/j.ultramic.2010.12.024
7. Saxey, D.W.: Correlated ion analysis and the interpretation of atom probe mass spectra.
Ultramicroscopy 111(6), 473479 (2011)
8. Tang, F., Gault, B., Ringer, S.P., Martin, P., Bendavid, A., Cairney, J.M.: Microstructural
investigation of TiSiN hard coatings. Scripta Mater. 63, 192195 (2010)
9. Yao, L., Gault, B., Cairney, J.M., Ringer, S.P.: On the multiplicity of field evaporation events
in atom probe: a new dimension to the analysis of mass spectra. Phil. Mag. Lett. 90(2),
121129 (2010)
10. Tsong, T.T., Liou, Y.: Cluster-ion formation in pulsed-laser-stimulated field desorption of
condensed materials. Phys. Rev. B 32(7), 43404357 (1985)
11. Prosa, T.J., Alvis, R.A., Kelly, T.F.: Observations of cluster ions originating from
non-traditional atom probe materials. Microsc. Microanal. 14(S2), 1236 (2008)
12. Birdseye, P.J., Smith, D.A.: The electric field and the stress on a field-ion specimen. Surf. Sci.
23, 198210 (1970)
13. Birdseye, P.J., Smith, D.A., Smith, G.D.W.: Analogue investigations of electric field distribu-
tion and ion trajectories in the field ion microscope. J. Phys. D 7(12), 16421651 (1974)
14. Eaton, H.C., Bayuzick, R.J.: Field-induced stresses in field emitters. Surf. Sci. 70, 408426
(1978)
15. Smith, P.J., Smith, D.A.: Preliminary calculations of the electric field and the stress on a field-
ion specimen. Phil. Mag. 21, 907912 (1970)
16. Rendulic, K.D., Muller, E.W.: Elastic deformation of field-ion-microscope tips. J. Appl. Phys.
38(5), 2070 (1967)
18. Moy, C.K.S., Ranzi, G., Petersen, T.C., Ringer, S.P.: Macroscopic electrical field distribution
and field-induced surface stresses of needle-shaped field emitters. Ultramicroscopy
111, 397404 (2011)
20. Smith, D.A., Birdseye, P.J., Goringe, M.J.: Forces on dislocations in field-ion specimens;
further analysis of some previous observations. Phil. Mag. 27(5), 11751181 (1972)
21. Loberg, B., Norden, H., Smith, D.A.: Observation of dislocations in a tungsten grain boundary
by combined electron and field-ion microscopy. Arkiv Fysik 40(38), 513519 (1969)
22. Larson, D.J., Miller, M.K.: In-situ micro-twinning of TiAl in the field ion microscope. Mater.
Sci. Eng. A 250(1), 7276 (1998)
23. Koelling, S.: Three dimensional compositional analysis of semiconductors with the atom
probe. Ph.D. Thesis, Katholieke Universiteit Leuven (2011)
24. Nishikawa, O.: Operation of the field ion microscope with neon. J. Appl. Phys. 35(10),
28062812 (1964)
25. Lawrence, D., Alvis, R., Olson, D.: Specimen preparation for cross section atom probe
analysis. Microsc. Microanal. 14(S2), 10041005 (2008)
Instrumentation and Techniques for Semiconductor Nanostructure Characterization,
vol. 2, pp. 407477. World Scientific Publishing, London (2011)
Kelly, T.F., Lefebvre, W.: From the store shelf to device-level atom probe analysis: an exercise
in feasibility. In: 36th International Symposium for Testing and Failuer Analysis, San Jose, CA
2011, pp 189197. ASM International
28. Prosa, T.J., Lawrence, D., Olson, D., Larson, D.J., Marquis, E.A.: Backside lift-out specimen
preparation: reversing the analysis direction in atom probe tomography. Microsc. Microanal.
15(S2), 298299 (2009)
29. Dieter, G.E.: Mechanical Metallurgy, 3rd edn. McGraw-Hill, New York, NY (1986)
30. Powell, R.W., Ho, C.Y., Liley, P.E.: Thermal Conductivity of Selected Materials. In: National
Standard Reference Data Series-National Bureau of Standards. U.S. Government Printing
Office, Washington, DC (1966)
31. Kellogg, G.L.: Determining the field emitter temperature during laser irradiation in the pulsed
laser atom probe. J. Appl. Phys. 52(6), 53205326 (1981)
32. Kellogg, G.L.: Measurement of activation energies for field evaporation of tungsten ions as a
function of electric field. Phys. Rev. B 29(8), 43044312 (1984)
33. Wada, M.: On the thermally activated field evaporation of surface atoms. Surf. Sci.
145, 451465 (1984)
34. Menand, A., Blavette, D.: Temperature dependence of iridium field evaporation rate. J. Phys.
47(C7), 1720 (1986)
35. Thompson, K., Bunton, J.H., Kelly, T.F., Larson, D.J.: Characterization of ultralow-energy
implants and towards the analysis of three-dimensional dopant distributions using three-
dimensional atom-probe tomography. J. Vac. Sci. Technol. B 24(1), 421 (2006). doi:10.
1116/1.2141621
36. Blavette, D., Duval, P., Letellier, L., Guttmann, M.: Atomic-scale APFIM and TEM investi-
gation of grain boundary microchemistry in astrology nickel base superalloys. Acta Mater.
44(12), 49955005 (1996)
37. Tsong, T.T., Ng, Y.S., Krishnaswamy, S.V.: Quantification of atom-probe FIM data and an
application to the investigation of surface segregation of alloys. Appl. Phys. Lett. 32(11),
778780 (1978)
38. Cerezo, A., Smith, G.D.W., Waugh, A.R.: The FIM100 - performance of a commercial atom
probe system. J. Phys. C9(45), 329335 (1984)
39. Menand, A., Al Kassab, T., Chambreland, S., Sarrau, J.M.: Atom-probe study of aluminum-
lithium alloys. J. Phys. C6(49), 353358 (1988)
40. Sha, G., Ringer, S.P.: Effect of laser pulsing on the composition measurement of an
AlMgSiCu alloy using three-dimensional atom probe. Ultramicroscopy 109(5), 580584
(2009)
41. Marquis, E.A., Yahya, N.A., Larson, D.J., Miller, M.K., Todd, R.I.: Probing the improbable:
Chapter 5
Data Processing and Reconstruction
5.1 Introduction
Data processing refers to taking raw atom probe tomography (APT) data and
preparing it to be reconstructed into real-space data, while reconstruction takes
processed experimental data and converts it into 3D spatial coordinates. Once this
has been successfully accomplished, the data can be interrogated for useful nano-
structural information.
The reconstruction algorithm (the act of projecting from detector hit positions
back to a reconstructed specimen surface) is powerful, yet very simple. Accurate
reconstruction of ion position requires an accurate description of the specimen
shape as it evolves during field evaporation; however, the simple assumptions at
the core of the reconstruction algorithms do not always capture all the detector-
to-specimen projection details. Some deviations occur because of the nature of the
bonding in crystallographic specimens, while others originate from evaporation-
field differences between material phases [14].
In this chapter, we will discuss the full data processing cycle for reconstruction.
Details of the voltage, shank, and tip profile reconstruction protocols are included,
as well as common aberrations that result from limitations to these protocols.
Finally, analysis of an arsenic-implanted silicon sample is included, for which the
available data-reconstruction options are discussed and evaluated with respect to
the impact on the expected spatial positioning accuracy.
110 5 Data Processing and Reconstruction
5.2 A Word on Data Files and Work Flow
Before we engage in specific discussions about data processing and reconstruction,

it is worth noting the practical matter of how data are organized and flow through
the various processing steps involved in reconstruction. First, there are several types
of data formats used in reconstruction and data analysis. The file formats discussed
here contain data that have been processed in some way by the user, and include the
ROOT, RNG, RRNG, POS, ePOS, and ATO file formats. For users interested in
more detail about the file formats, or for users who are interested in manipulating
the data outside of the standard software, further information regarding the file
structures is provided in Appendix A.
Creation of a ROOT file is the first step in LEAP data processing. The ROOT
file contains the parameters and information used to convert the data in detector
space (RHIT file) into 3D specimen space coordinates. Many of these parameters
are selected by the user. Many histograms and graphs created as part of the
calibration and reconstruction processing are also stored in the ROOT file for
later reference. Each ROOT file also contains a set of mass range definitions (and
ionic volume definitions) that enable reconstruction of the RHIT file into a POS file,
but mass range files can also exist as separate files. These files use the RNG and
RRNG formats and contain the information necessary to convert a mass-to-charge
(m/n) spectrum into ion identities.
A POS file consists of the reconstructed ion coordinates and the m/n value for
each reconstructed ion. POS files are the natural output of the creation and appli-
cation of a ROOT file, but they can also originate as subvolume extractions from
larger parent POS files.
5.3 Conversion from Detector Space to Specimen

Space Coordinates
After completion of each LEAP acquisition, STR and RHIT files are created. The
STR file replaces the RRAW file format in LEAP systems after the 4000 base model.
The STR file contains unfiltered instrumental data with an entry for every pulse
event. The STR is then processed to identify ion hits and stored as an RHIT file.
The IVAS Cal/Recon Wizard guides the user through a series of steps which
facilitates the conversion of raw RHIT data into reconstructed positions and
chemical identities. These steps include: Select Ion Sequence Range, Select
Detector ROI, TOF Corrections, Mass Calibration, Ranged-Ion Assignment, and
Reconstruction. The functions of these steps are discussed in the following sections.
First, the user is asked to choose an appropriate subset of the raw data for
processing. This is done by selecting an ion collection range and areal region of
the detector (Sect. 5.3.1). Then, the individual ion flight times are corrected for
variations due to actual flight path and voltage, resulting in the generation of a mass
5.3 Conversion from Detector Space to Specimen Space Coordinates 111
Fig. 5.1 An appropriate

continuous range of
voltages is selected in the
IVAS Cal/Recon Wizard by
adjusting the voltage box
overlay (dashed line) to
capture the appropriate
region of the Voltage vs. Ion
Sequence curve
spectrum (Sect. 5.3.2). This allows the user to associate chemical identities and
volumes to each ion type (Sect. 5.3.3). Finally, each ion is reconstructed by
assigning it with a 3D position. The reconstruction process is covered by sections
that detail the basic reconstruction algorithm (Sect. 5.3.4), the available IVAS
reconstruction protocols (Sect. 5.3.5), details of how the specimen apex relates
to the specimen body (Sect. 5.3.6), and tools available to assess reconstruction
parameter selection (Sect. 5.3.7). Finally, at the end of the process, the ROOT and
POS files are created so that the reconstructed data are available for material
analysis.
5.3.1 Selection of Depth and Areal Regions
When performing a reconstruction it is wise to limit the analysis volume to the

region of the data that is relevant to the questions at hand. Besides contributing
to unnecessary computing requirements, including irrelevant volume contributes to
increased spectral background levels and additional mass peaks that unnecessarily
complicate the analysis and can negatively affect the algorithms used to recon-
struct, calibrate, and optimize the data.
The first step in data processing is to choose a range of ions to include in the
reconstruction. By using the Voltage vs. Ion Sequence Number plot in the Select Ion
Sequence Range step, a range of voltage and ion numbers is selected by adjusting the
selection rectangle (dashed in Fig. 5.1). Only the ions that fall within this selection
box are included in the reconstruction. Information is also available (such as the
shape of the voltage curve, the mass history, changes in the background level, and
changes in analysis conditions) for making intelligent decisions on range selection in
order to avoid inclusion of data containing artifacts that can lead to inaccurate
reconstruction. Examples of regions to avoid include: contamination from sample
Fig. 5.2 An appropriate

elliptical detector area is
selected in the IVAS
Cal/Recon Wizard by
adjusting the hit overlay
to capture the appropriate
region of ion hit-map
on the detector
preparation (gallium implantation or surface oxidation), partial specimen failure,

local-electrode noise, or the beginning of the acquisition when conditions were not
equilibrated.
The next step in data processing, Select Detector ROI, is to choose the spatial
region of the detector from which to accept recorded ions. Data from different
regions of the detector may not possess similar quality and the region of interest
(ROI) may only fill a subset of the field of view (FOV), so again it is wise to limit
the ROI in order to optimize the data quality and enhance the ability to answer the
question at hand. Figure 5.2 illustrates areal selection from the detector hit-map
which avoids a low ion hit density region near the bottom of the detector.
Statistical arguments suggest that inclusion of all detected data enhances the
ability to measure compositional variation, but other considerations may be more
important. Removing rough edge regions in the data (caused by nonuniform edges
on the local electrode) improves not only the esthetic quality of the 3D image but
also the accuracy of the reconstruction since the reconstruction algorithms utilize
the selected ROI to calculate the depth increment. Uniform hit densities that fill the
selected area to the boundaries produce more accurate depth reconstructions. One
may also choose to select a region of highest mass resolving power (MRP) or some
section of the detector to investigate how instrumental or specimen limitations
affect reconstruction.
5.3.2 Spectral Calibration
An atom probe measurement is a combination of time-of-flight (TOF) mass spec-

trometry and projection imaging of a sequence of ions leaving a specimen surface.
The aim of the mass spectrometry component is to arrange the events detected
during an experiment into a single mass (or time) spectrum which is usually
displayed as a histogram plot of ion count versus either m/n or TOF. This allows
for identification of particular ranges of events as belonging to a particular atomic or
molecular species (in the form of singly or multiply charged ions). Due to the physics
of field evaporation and instrument limitations, measured flight times must be
corrected for various effects. The corrections necessary to obtain the most accurate
TOF values vary from specimen to specimen and therefore require optimization
for each individual dataset. Although these specimen-to-specimen differences are
small, they can result in significant variations in mass spectral quality. The following
sections detail the procedures used to calibrate individually measured events into a
coherent and optimized mass spectrum. In IVAS these calculations are performed in
the TOF Corrections step.
5.3.2.1 Voltage and Bowl Corrections
Individual ions are identified by measuring their flight times to reach the detector
(time between voltage or laser pulse and ion arrival at detector). To first order, m/n
of the incoming ion is calculated by equating the kinetic and potential energies:
1 2 d m 2eV
mv neV and v ! 2 t2 5:1
2 t n L
where e is the electron charge, n is the charge state, V is the total voltage applied to the
specimen during evaporation, v is the speed, L is the flight distance, and t is the TOF.
The calculation of m/n is based on two assumptions: first, the potential energy of the
ion is entirely converted to kinetic energy by the end of the flight path, and second, the
acceleration of the ion up to its terminal velocity is instantaneous (see Appendix D for
a more detailed discussion). Correction terms have been suggested that can deal with
the incompleteness of the acceleration and with the amount of applied voltage
actually transmitted to the specimen [5], but the parameterized approach applied to
LEAP data (described below) generally achieves satisfactory results.
Since the data collected is a collection of times of flight, corrections are first
made to the raw TOF data (Fig. 5.3) and then those values are converted to m/n.
Without these corrections, unique m/n ions do not present a narrow TOF range, but
instead a wider distribution of TOFs that depend on ionization voltage and impact
location on the detector. As illustrated in Fig. 5.3a, the raw TOF spectrum here
consists of three broad peaks that originate from the groups of well-defined peaks
shown in the corrected TOF spectrum in Fig. 5.3b. Motivated by (5.1), a corrected
TOF may be expressed as a separable set of functions
tcor FV V DC FB X; Y traw t0 , 5:2
where FV(VDC) accounts for TOF variations due to voltage (i.e., oversimplified
electrostatic assumptions, linear detector response, detector bias, etc.), FB(X,Y )
Fig. 5.3 Example TOF spectrum (a) before and (b) after voltage and bowl corrections. Although
this dataset is composed of a number of well-defined ion species, variations in flight times, and
path length as a function of location on the detector and voltage cause the raw flight data in (a) to
appear broadly distributed. After application of appropriate corrections, the corrected TOF
spectrum in (b) exhibits well-defined flight times for each ion species which can be optimized
for each individual dataset
takes into account flight path length differences as a function of detector hit location
(i.e., an appropriately curved detector would not require this correction; hence this
is often referred to as the bowl correction), and (traw t0) provides a TOF offset
which is required due to various signal cables and electronic processing.
p
According to (5.1), the voltage correction to the TOF should follow a 1= V DC
dependence; however, a three-parameter fit of the following ad hoc form captures
the experimental variations more completely [6]:
a0
FV V DC q : 5:3
a1 V DC a2 V 2DC
Because the detector is flat and the detected ions effectively originate from a
point source, the flight distances (and consequently TOF) for a straight flight path
instrument increase outward from the center of the detector (see Fig. 5.4a). Much
like the voltage correction, a parameterized bowl shaped function, FB(X,Y ) is
defined and its shape is chosen to fit the 2D TOF distribution originating from a
single ion peak. A complex bowl shape is required because this function corrects for
more than simple geometrical variations in flight distances. It also accounts for
things such as detector nonlinearities which affect calculations of detected posi-
tions. Bowl asymmetry provides optimized TOF for each dataset and so the user is
allowed to choose the subdivision level of the bowl for each optimization effort.
An example of a typical bowl shape and the residual TOF values for an example
peak are illustrated in Fig. 5.4. Optimizations of the various corrections are applied
Fig. 5.4 (a) An example bowl correction color coded to show the depth of the bowl (or TOF
in nanoseconds) as a function of location on the 2D detector. (b) The resulting flight time residuals
as a function of radial location on the detector. A properly applied bowl correction will result in
uniform and minimal residuals as a function of detector location
in step four of the IVAS Cal/Recon Wizard. The initial values for the various
parameterizations are based on previously optimized calibration files created
when a LEAP system is set up by a service engineer. The correction parameters
in the calibration files are also applied during live data collection on a LEAP
system.
As part of the TOF Corrections step, optimization of a given dataset proceeds by
first selecting an appropriate mass peak to maximize MRP (if the user does not
manually select a mass peak, the largest peak is selected by default). An appropriate
peak will include ions that are distributed across the entire FOV, else the optimi-
zation algorithm for the flight path (bowl) function may fail. Next, the parameters
defining the voltage correction FV(VDC) are optimized. Then, the parameters defin-
ing the bowl correction FB(x,y) are optimized. These voltage and bowl optimiza-
tions are iterated until MRP stops improving by a predefined threshold. During this
process, t0 remains fixed to the initial calibration value. Values of t0 have not been
found to change significantly from specimen to specimen, so fine adjustments to
this value are accounted for in the conversion from TOF to m/n (see next section).
After successful application of these corrections, the TOF spectrum should consist
of a collection of sharp peaks (Fig. 5.3b).
5.3.2.2 Mass-to-Charge Conversion
Once the voltage and bowl corrections have been satisfactorily completed, conver-
sion from TOF to m/n is performed (Mass Calibration step). Two strategies are used
for this conversion, the parametric fit method and the linearization method. The
parametric fit method is based on the relationship described by (5.1) and is usually
used only to create a calibration file. The linearization method, which assumes that
Fig. 5.5 Example

transformation from TOF to
m/n using the parameterized
function defined in (5.4).
A variable number of
known m/n peaks are
identified by the user
in the IVAS Cal/Recon
Wizard to determine
k and t0
the general relationship defined by the calibration file is reasonably accurate, applies
a piecewise continuous set of linear corrections for each two-point region defined by
the user-selected peaks and is the default method in IVAS.
Parametric Fit Method (Calibration)
As discussed previously, the relationship between m/n and TOF as described by

(5.1) is incomplete, so a parameterized version is applied as a fit to m/n peaks
identified by the user in the IVAS reconstruction wizard. The parameterized
function
m
c t t 0 2 , 5:4
n
is fit to the identified peaks (Fig. 5.5) to determine c and t0. This t0 is the same value
used during the voltage and bowl iterations when generating a calibration file, so it
may be necessary to go back to that step and re-optimize the voltage and bowl fits
given the new t0 value (especially if t0 changes by more than a few percent). Note
that when a single peak is chosen for fitting, t0 remains fixed and only c is adjusted
in (5.4). For two or more peaks, both parameters are adjusted. Because this is a
two-parameter fit, it is advisable to pick more than two peaks to adequately
transform TOF to m/n over the entire range of TOF values. As t0 tends to have
little variation from analysis to analysis, the parametric fit method is generally
suggested for calibration file generation only. However, when data have been
collected with an inappropriate calibration file (e.g. different operational mode,
tip geometry, or outside the calibrated voltage range), then this method is favored.
Linearization Method (Default)
This method uses the c and t0 values from the previously created calibration file.
Single or multiple peaks are selected by the user and corrected with the function
m=n ct t0 2 t, 5:5
where (t) is a piecewise linear interpolation function that ensures that all identified
peaks have the correct m/n and regions between identified peaks are interpolated.
This method, which applies a linearization on top of the parametrically optimized
correction, is usually superior for properly aligning spectra from multiple datasets or
when unknown or poorly resolved peaks dominate the mass spectrum. Choose three
or more high signal-to-noise, well-separated peaks, whenever possible to get the
maximum benefit from the mass spectral calibration. If comparing different datasets,
be sure to use the same calibration peaks in each one.
5.3.3 Chemical Identification & Ranging
Peak identification in a mass spectrum can appear to be the most simple of tasks,
but ranging choices have a significant impact on analytical results. Chemical
identification comes in two steps. First, one must accurately associate a peak with
an ion type, and then an appropriate choice of lower and upper limits is made to
identify where the relevant peak signal is located. For appropriate range selections,
a mass calibration must have been successfully completed.
Peak identification may take only minutes, for a simple or familiar mass
spectrum, or may take hours for unknown, complicated mass spectra. Complications
in mass spectra originate from the variations in elemental species and the number of
isotopes for each element, the distribution of charge states, the frequency of molec-
ular ion production, the frequency of closely timed (or overlapping) events, and peak
shapes. IVAS provides several tools for the identification of peaks including an
elemental overlay tool, an ion overlay tool, and a peak decomposition algorithm
(refer to Chap. 6 and the IVAS User Manual for more information).
Peak identification is approached in terms of the level of complexity. Simple
identifications include those peaks that are isolated and can be identified by simply
looking up masses in a periodic table where relative peak intensities are consistent
with the known isotopic distributions (isotopic fingerprints). It is useful to recog-
nize that peaks appearing near or fractions of a Dalton must represent doubly
or quadruplely charged ions. This allows the user to simply multiply m/n values by
2 or 4 to find the true mass of the ion (e.g., 13.5 Da could be 27Al2+). Peaks found
near the 1/3 or 2/3 fractions of a Dalton represent triply ionized ions. Again,
multiplying by three then provides the mass of the ion (e.g., 10.33 Da could be
31 3+
P ). Reference [7] provides tabulation of the most likely charge states as a
function of field for a variety of elements.
Overlaying plots of relative isotopic abundances onto a mass spectrum is

valuable for recognizing and identifying the isotopic fingerprints such as the
strong-weak-weak pattern exhibited by natural abundances of silicon at 28, 29,
and 30 Da. The same fingerprint idea can be used to identify molecular evaporation.
For example recognizing a pair of peaks whose ratio is similar to 69Ga and 71Ga, but
are located at 41.5 and 42.5 Da might lead one to conclude these peaks originate
from doubly charged Ga-containing ions (peaks appearing at Da positions) and
because two times these peak positions are 7 Da beyond the normal gallium
positions, they likely originate from GaN2+ molecular ions. Users can quickly
become proficient at recognizing offsets of , 1,2, 6, 7, 8, 12, 14, 16 as molecular
species likely containing hydrogen, carbon, nitrogen, and oxygen. Users must also
be aware of ion distributions that are expected to be unnatural. Elements like
gallium and boron may have single isotope distributions if they originate from a
focused ion beam (FIB) or another ion implantation process, others, such as lead,
may deviate from natural abundances due to mineral aging for geologic time scales
or due to neutron irradiation.
Further complicating peak identification and ranging are direct peak overlaps, or
peaks from different ion species that share the same or overlapping ranges in m/n
[8]. Overlaps can occur due to various reasons and have been categorized by the
ability to discriminate them [8]. In some cases poor MRP will cause peaks to
overlap (94Mo3+ and 63Cu2+), while in other cases charge states may cause overlaps
(14N+ and 28Si2+ or 16O22+ and 16O+), and yet in other cases isobaric overlaps may
occur, which are isotopes of different elements overlapping (58Fe+ and 58Ni+) (see
Fig. 5.6 for example). These overlaps add confusion to the isotopic distribution and
can take time to decipher when they are initially encountered. In high signal-to-
noise data with good peak shapes, the peak decomposition algorithm available in
IVAS (see Chap. 6) can identify peaks with unexpected isotopic distributions,
hinting at yet unidentified direct peak overlaps.
In some cases, peak assignments may be ambiguous, or otherwise impossible to
decipher the first time a peak is encountered. An additional bit of information that
can be very useful for peak assignment is spatial association with other ions. One
can assign the peak as an unknown ion and proceed to complete an initial recon-
struction. Observing what ions the unknown peak is associated with in the 3D
reconstruction can lead to a logical determination in many cases.
Once the chemical identities of all the peaks have been satisfactorily identified,
fine details of the peak ranging must be considered. Ranging not only provides
identification of a peak, but also impacts compositional accuracy and determines
the analysis volume, which affects reconstruction accuracy and precision. As
discussed in detail in Chap. 6, the considerations used to set the lower and upper
limits of a range depend on peak shape, MRP, signal-to-noise, and proximity to
closely spaced peaks. There is currently no standard, universally accepted, method
for determining peak ranges, so it is very important to indicate the procedures used
and to be consistent in their application throughout multiple analyses. The risks of
inconsistent range application have been clearly identified in trials where groups
Fig. 5.6 This example mass spectrum from a sample containing Ti, Fe, Co, and Ni ions illustrates
typical mass ranging challenges. The five distinct Ti peaks pose no ranging issues as the ranges
necessary to identify these five isotopes do not conflict with any other elements. This is not the
case for the Fe, Co, and Ni peaks. The region near 58 Da contains both 58Fe and 58Ni isotopes.
Likewise, a significant fraction of the 59Co peak contains the tail from the 58 Da region. Therefore,
careful consideration is necessary for the ranging boundary of this particular element
of people experienced in APT have come to significantly different estimates of

composition, even for simulated data [9].
Caution should be used in applying range files created during live acquisition for
the purpose of real-time estimation of composition. These range files often identify
only a fraction of the observed peaks and may employ inaccurate lower and upper
limits. A range file used for reconstruction should identify every significant peak
and thus maximize the compositional accuracy in order to optimize the accuracy of
the reconstructed volume. While working in the IVAS Cal/Recon Wizard only a
sampling of the ions is displayed to decrease computation time. Unsampling a given
mass spectrum in the Ranged-Ion Assignments step of the IVAS Cal/Recon Wizard
provides additional spectrum counts and adds detail to properly identify fine
features at the expense of computation time but is often worth the sacrifice. As
discussed more fully in Chap. 6, a range file used for accurate image formation may
employ narrow ranges to improve the signal-to-noise in images or charts, while
broader ranges may be necessary for compositional accuracy. In general, a good
starting point is to consider setting the lower and upper bounds for a range at two
times the local background level. It is not unusual for several rounds of ranging to
take place. For example, after examining the fully reconstructed data, a different
region of interest may be identified or a minor unassigned peak may be identified as
a key component of the analysis. Saving range files to a library to be used as a
starting point in subsequent analyses is a smart and efficient protocol.
Fig. 5.7 This schematic shows the geometrical relationship between the tip radius, R, and the
image compression factor, , which is directly related to the angular magnification of the tip image
5.3.4 Spatial Reconstruction: Projection and Depth Scaling
APT data are collected as individual ions. Data reconstruction consists of calculating
3D positions in specimen coordinates (referred to as real space, and denoted as lower
case x, y, and z) from a time-series of 2D positions in detector coordinates (referred
to as detector space, and denoted in upper case X and Y). The standard reconstruc-
tion methodology [1016] involves calculating a magnification in order to convert
from detector hit positions (X and Y) into real-space coordinates (where x and y are
contained in the plane normal to the specimen axis at the current apex position). The
depth of the ion along the specimen axis (z) is based on the assigned ionic volume
and the estimated imaged area of the specimen surface.
In a point projection, the magnification (M ) is simply the distance (L ) between
the sample and the detector divided by the radius of curvature of the specimen (R).
Ion trajectories from a needle (not an isolated point in space) are compressed by the
Image Compression Factor (), Fig. 5.7 and (5.6), which reduces the magnification,
thus the projection magnification is:
L
M 5:6
R
Note that expressions accurate for wide-FOV acquisitions are used in the Recon-
struction step in the IVAS Cal/Recon Wizard and are derived in Appendix C;
however, for the sake of simplicity small FOV expressions are used for illustrating
the main points in this section.
It is generally accepted that the nominal value of is between one and two.
Deviation from unity is attributed to the effects of the shank region of the specimen
compressing the electric field lines inwards toward the detector as shown in Fig. 5.7.
Before the z-coordinate value for each ion is calculated, an estimate of the imaged
specimen surface area is required. This is accomplished by converting the detected
surface area (A) to specimen surface area using a geometrical model of the apex
and assuming the magnification computed above. For each reconstructed ion, the
spherical model z-coordinate is adjusted by a small increment equal to the atomic
volume () divided by the image areathe volume of the reconstructed ion is
spread uniformly over the entire apex. The z increment is also a function of the
Detection Efficiency () [11, 12]:
L2
z : 5:7
A2 R2
Prior to the advent of wide-FOV instruments [17, 18], the geometric assumptions
described by Blavette [10], and later applied by Bas [11], were widely considered as
the standard reconstruction technique. This model assumes that the original shape of
an acquired volume is small enough in lateral extent to be considered cylindrical and
the radius of curvature of the specimen is determined ion-by-ion from the specimen
voltage. In early reconstruction algorithms, the shank angle was assumed to be zero
in the calculation of the volume increment, although Walck [19] published a
description of specimen geometry that could be used to compute depth increments
that are appropriate for a wide FOV and included shank angle effects. These effects
were later taken into account by Geiser et al. [14].
A complete review of the current state of APT reconstruction, which currently is
a topic of much research within the atom probe tomography community, is beyond
the scope of this chapter, but some relevant references are provided for the
interested reader [3, 4, 1214, 16, 2027].
5.3.5 Wide-Angle Reconstruction Protocols
5.3.5.1 Basic Considerations: Voltage and Shank Angle Evolution
Instrument Parameters
Figure 5.8 shows the controls available on the Expert Tab within the Reconstruction
section of the IVAS Cal/Recon Wizard. Many of these variables are functions of
each other. The parameters at the top of the figure are those that might generally be
expected to depend on the instrument configuration.
For example, Detector Efficiency is an estimate of the overall detection
efficiency due to the MCP efficiency (and reflectron mesh when present). The
Fig. 5.8 The IVAS Cal/Recon Wizard Expert Tab contains all the parameter details necessary for
reconstruction
efficiency has an obvious impact on depth scaling as indicated by (5.7). The depth
scaling applied ion-by-ion as reconstruction occurs is driven by the accumulation
of atomic volumes spread out across the apex FOV. The efficiency correction is
applied by scaling each ionic volume by the system efficiency factor to account for
undetected atoms. When the specimen radius evolution is driven by a model like
fixed shank angle or tip profile, an incorrect efficiency can promote errors in tip
radius by evolving the radius too rapidly or too slowly with accumulated volume.
The k-factor (k) [28], is important in the estimation of the Initial Tip Radius, R0,
from the applied voltage difference, V, between the specimen and the local
electrode:
V
R0 5:8
kF
where F is the Evaporation Field of the material being analyzed. The k-factor
depends upon geometric considerations such as shank angle and radius of the
specimen and separation distance between the specimen and local electrode [29].
As explained in the previous section, the Image Compression Factor, , is a

primary contributor to lateral magnification and, because this impacts the FOV, it
also contributes significantly to the depth increment calculated per ion. The phys-
ical origins of image compression depend on shank angle and, to a lesser extent, on
tip radius and tip-aperture separation [29]. In this geometric calculation, all image
compression occurs near the launch point, but physical contributions to the overall
level of image compression, such as the schematic ion trajectories shown in
Fig. 5.7, may also occur near the local electrode and at the detector. The
Specimen-to-Detector Distance, L, is important in calculating the magnification
and estimated FOV. It also impacts the depth increment and is automatically
inserted into the default reconstruction parameters from the RHIT file. This distance
will also typically impact the alignment of the specimen to the local electrode due to
the change in the angular FOV, which may again impact several of the other
variables such as and k.
The Reconstruction Center specifies the detector location that corresponds to the
center of the reconstruction apex. Adjusting this point has the effect of tilting
the reconstruction. Since the spatial distribution map (SDM) tools (see
Chap. 6) are limited to orthogonal projections, setting the Reconstruction Center
to a pole location causes those atomic planes to lie in the z-plane of the reconstruc-
tion allowing examination with those tools. As a convenience an option is available
to set the reconstruction center to the bowl center determined during mass calibra-
tion, or to the center of the selected detector ROI. The default setting is the center of
the detector ROI, which will result in the best choice for the majority of analyses.
Material Parameters
Based on ion count, the highest concentration element becomes the Primary
Element and it will be used to automatically select a default evaporation field as
shown in Fig. 5.8. This value is then used to calculate the Initial Tip Radius from the
initial voltage as determined by the starting point selected in the Select Ion
Sequence Range step. The Primary Element will also be used to suggest an average
Atomic Volume to use for reconstruction if element-specific atomic volumes are not
used. If the material is homogeneous over the length scales of interest and has a
known ionic volume, one should deselect the Element-Specific checkbox and use
the known value. An information field next to the Primary Element indicates the
temperature under which ions were collected while an information field immedi-
ately to the right of the Evaporation Field indicates the temperature at which the
estimate of the Evaporation Field is based. When there is a significant difference
between these two temperatures, or when the material is an alloy, or when data
are collected in laser pulsed mode, adjustment from the default field value is
warranted.
Specimen Parameters
This section of the Reconstruction tab is focused on specifying the tip radius
throughout the reconstruction. There are three different modes of Radius Evolution
that can be chosen: Voltage, Shank, and Tip Profile.
The Voltage Evolution mode uses (5.8) above so that the tip radius used for
reconstruction is strictly based on the total voltage for a given ion. The Evaporation
Field and k Factor directly update the Initial Tip Radius as they are modified, but if
desired, the user can directly modify the Initial Tip Radius by simply entering a
numerical value.
The Shank Evolution mode uses the same Initial Tip Radius as the Voltage
Evolution reconstruction, but the evolution of the tip radius is determined by the
specific choice of geometric parameters, as described in Appendix C. An estimate of
average shank angle is available which is based on the evolution of the voltage curve
and assigned atomic volumes. The correspondence between the shank angle used
in the reconstruction and the physical shank angle (as measured from scanning or
transmission electron microscopy (SEM, TEM)) will depend on how the field
evaporated end form intersects the specimen shank. Historically, the reconstruction
geometry required a tangential continuity between the hemispherical apex and the
shank. This could require entering a shank angle which was significantly different
from the measured shank angle to achieve a reasonable reconstruction because many
specimens exhibit an apex that is significantly blunter than the diameter of the conic
shank on which it resides. Some reconstruction algorithms allow the specification of
this discontinuity in the form of the sphere-to-cone ratio. The ratio is calculated
from the ratio of the radius of the hemispherical end form to the radius of the shank
where the two intersect and results in a specified shank angle more closely
corresponding to the physical shank angle. This enables high magnification SEM
or TEM images to be used successfully to set shank angles and initial tip radii. The
use of the SphereCone Radius Ratio parameter does not change the reconstruction
in any way other than scaling the constant shank angle entered by the user (for more
on the SphereCone Radius Ratio see section (Sect. 5.3.6) below).
Although the voltage and shank angle reconstructions are simple and powerful,
there may be a need to deviate from voltage or single shank angle prescriptions for
radius evolution as a function of depth. For example, multi-layer materials
exhibiting multiple evaporation fields can require significant deviation from a
single evaporation field relationship. Similarly, a lengthy analysis with evolving
shank angle and multiple evaporation fields is frustrated by a single shank angle
reconstruction. By obtaining a high quality electron microscopy image prior to
LEAP analysis, a radius evolution function can be based on the actual tip profile as
shown in Figs. 5.9 and 5.10 [30]. This approach can take into account the evolving
shank and mitigate some of the issues regarding evolving evaporation field. How-
ever, reconstruction challenges caused by non-hemispherical and/or evolving tip
shapes still remain.
The Tip Profile reconstruction is implemented during the final step in the IVAS
Cal/Recon Wizard (Figs. 5.9 and 5.10 show the image and corresponding radial
Fig. 5.9 The IVAS

Cal/Recon Wizard allows
users to define the
reconstruction radius
evolution based on the
actual tip profile. (a) By
measuring various portions
of the tip (colored profile
tracer lines), (b) a radius
evolution function is
defined. The field
evaporated endform of this
tip enables accurate
determination of the initial
radius, R0, in relation to
the other tip measurements
evolution functions defined by the user). First, an image of the pre-LEAP-analysis

tip profile is loaded. Then, the stage tilt used to collect the image is appropriately
set (Typically 52 for an imaged collected on an FEI dual-beam FIB, 90 if
special mounting of the specimen has been performed to allow a true profile
measurement via SEM, or 90 for a TEM). Next, an imaging scale factor is defined
to relate pixel dimensions to nanometers. Finally, a set of profile tracers are used
to define the radial evolution function used in the reconstruction. Of special
importance is the uppermost tracer which defines the Initial Tip Radius of the
reconstruction. Although a reconstruction profile and defined volume are created
as deep into the sample as the traces are drawn, the reconstruction will only fill
this volume up to the point that all the ions collected are used. Figure 5.9 represents
a tip that has been field evaporated to a steady-state end form, imaged in an
Fig. 5.10 Application of

the tip profile method for
un-evaporated tips requires
extra effort from the user
when compared to the
example in Fig. 5.9. (a) The
placement of the R0 tracer
affects the entire
reconstruction because it
impacts the entire radius
evolution function shown in
(b), and is not well defined.
Additional criteria (such as
known RFinal) can constrain
the location of R0, but
ultimately the user must
make some estimate as to
where on the tip the first
reconstructed ions
originate. In this case, the
first ~10 million ions were
ignored and R0 was placed
to result in a known RFinal
SEM, and subsequently analyzed in a LEAP system. This procedure allows for
simple identification of R0 for the subsequent reconstruction. Figure 5.10 is a more
typical example where only a post-analysis image is available and some uncertainty
exists for placement of the R0 contour. For any analysis that does not start with a
field evaporated endform and does not utilize all the ions in the analysis, this
uppermost tracer must be approximately placed to accommodate those realities.
In this case, the sample did not end in failure, so a post-LEAP-analysis image was
taken and RFinal was measured. Variations of the R0 tracer can then be made until
the resulting RFinal matches the measured value in the SEM.
Fig. 5.11 (a) Tangential continuity is shown where the sphere that represents the tip
surface intersects the shank () so that the intersection is smooth fs/c 1. (b) This tangential
discontinuity better reflects physical field evaporated surfaces where the sphere that represents the
tip surfaces intersects the shank nonsmoothly fs/c 6 1
We note that the user is not constrained to the tip profile for defining the radius
evolution. An experienced user might prescribe deviations from the tip profile near
interfaces known to produce non-hemispherical tip shapes [26]. Such deviations
might better approximate the radius evolution as the tip shape becomes highly non-
hemispherical or even non-singly curved during analysis through these regions and
result in more accurate reconstructions. In this sense, the tip profile method might
also be considered a custom radius evolution method.
An important advantage of the tip profile method is that the variables, k-Factor
and Evaporation Field, which can have significant uncertainties, are not used. With
additional images, additional constraints (R0 and/or RFinal) and the extent of tan-
gential discontinuity (see next section) may be measured and utilized. High mag-
nification images enable more accurate rendering of the profile, but some caution
must be used because of possible negative interactions with imaging and APT
analysis yield [31].
5.3.6 Tangential Discontinuity
Historically, a simplifying assumption has been made in reconstruction software that

there is tangential continuity between the evaporating surface and the shank region
of the specimen. The impact of this assumption is that it forces a specific evolution of
the estimated tip radius with analysis depth. Expressions for this evolution in the
case of tangential continuity were written down by Blavette [10], and are shown in
Appendix C (14). The geometry is illustrated in Fig. 5.11a. However, as observed
experimentally [3, 4, 24, 25, 3235], field evaporated specimens do not necessarily
assume a shape with tangential continuity. In recent work we have relaxed this
assumption by allowing the apex radius to maintain an arbitrary but fixed ratio to the
cone radius [25].
Rsphere
f s=c 5:9
Rcone
Using this lessened geometric restriction the evolution of the specimen radius
with analysis depth is described in Appendix C (17). This more general geometry is
illustrated in Fig. 5.11b.
As shown in Fig. 5.8, the option exists for the user to disable the tangential
continuity condition and directly enter the SphereCone Radius Ratio value. The
assumption of tangential continuity biases reconstructions toward parameters with
unphysically large shank angles. The redefinition of the dR/dZ term to allow for a
tangential discontinuity enables APT reconstructions to take advantage of electron
microscopy-based information about the actual geometry of specimen shank angle
and the level of discontinuity between the apex and the shank, both of which may be
used to provide constraints on the selection of other reconstruction parameters.
Reference [25] explains the effect of relaxing the tangential continuity constraint on
the reconstruction evolution in further detail.
5.3.7 Reconstruction Explorer
For the majority of the reconstruction procedures outlined above, processing is

performed on a sampled subset of the data. Unsampled versions of many of the plots
can be made by selecting the Unsample option beneath the plot in question, but the
actual calibration calculations and preview reconstructions are performed on a
random sampling of the data. For visual inspection of reconstruction quality,
however, the sampling can reduce the ability to observe very fine structures like
precipitates or atomic planes. The Reconstruction Explorer, Fig. 5.12, supports
inspection of these details by allowing the user to focus on a reduced region or
species of interest with less data sampling. This allows for live adjustment of
parameters and immediate visual feedback of the reconstructed result. It is impor-
tant to note that even though only a small subset of the atoms in the full fiducial
region may be reconstructed in the Reconstruction Explorer, reconstructed posi-
tions are determined based on a reconstruction of the full fiducial region chosen in
the Select Detector ROI step.
Two primary methods exist for specifying an ROI for use in the Reconstruction
Explorer: (1) a cylindrical section from detector space may be chosen or (2) all
atoms of a specific element can be chosen. If a detector ROI is chosen, the user can
specify the Center X Position and Center Y Position of the cylinder, the Outer
Radius around the chosen center, and if desired, an excluded Inner Radius around
the center. Note that these controls are all in millimeters since they are in detector
space. The exclusion radius is intended to allow the exclusion of low index
crystallographic poles. Although the ROI location can be set to extend outside of
the earlier selected fiducial cut, data will not be present in this region. Ion Sequence
Fig. 5.12 (a) Parameter selection portion of the Reconstruction Explorer, activated from within
IVAS Cal/Recon Wizard Expert Tab, allows the user to quickly manipulate the relevant recon-
struction parameters and assess the consequences by showing a computationally efficient
reconstruction preview of a user selected subvolume. (b) The ion display portion of the Recon-
struction Explorer showing reconstructed atomic planes
Range and Ion Sequence Center allow the user to specify a sequence number range
for reconstruction. As illustrated in Fig. 5.12, an Ion Sequence Range of 50 %
(25 %) around the Ion Sequence Center of 50 % results in a range of 2575 % of
the data to be included.
The same reconstruction parameters available on the Expert tab are available in
the Reconstruction Explorer, as shown in Fig. 5.8. When dragging the slider bar for
any of the parameters, new reconstructions are only triggered upon release of the
slider. Note that at least one parameter must have changed from the original settings
when entering the Reconstruction Explorer in order for any parameters to be
transferred to the Expert tab upon exit.
When generating spatial distribution maps (see Chap. 6 for further details), the
Center Element specifies the species of ions around which the average neighbor-
hood intensities are computed. The Neighbor Element specifies the species of ions
considered as nearest neighbors for the distributions plotted. The Neighborhood
Extent specifies the size of the neighborhood plots to be created. Z Cut in XY Map
describes how close in the Z neighborhood ions must be to the Z Plane Offset in
order to be included in the plot. For example, by default, the Z Plane Offset is 0.0
(neither above or below), so neighbor atoms must have a z coordinate within
0.01 nm of the center atom in order to be included in the TM. Changing the
Z Plane Offset allows the XY distribution of nearby layers of ions to be observed in
well-resolved crystalline materials.
In the visual display region of the subvolume reconstruction, Fig. 5.12b, various
orthogonal projections of the current reconstruction can be chosen. Additionally,
for real-space projections of the reconstruction, the rendered depth can be adjusted
by the application of cut-planes, controlled by the double-ended scroll bar. If an
SDM has been generated, the user can choose between the three 1D projections
and three 2D projections. The 1D X neighborhood, the 1D Y neighborhood, and the
2D XY neighborhood plots are all subject to the Z distance cut from the center
atoms. The Z, ZX, and ZY plots are not subject to any cuts on distance from the
center atoms. By default the reconstruction plot aspect ratio will be physically
correct, but by selecting Fit to Window under Plot Aspect Ratio, the display can be
automatically rescaled to more completely fill the display area.
5.3.8 Creation of ROOT and POS Files
At the Reconstruction step, Fig. 5.8, the user has the option to create multiple
reconstructions using the Create Preview button. No information is saved however,
until the Save Reconstruction button is used to create the ROOT file which will
preserve all of the work done until this point. This file contains all the necessary
information to create a POS file from the RHIT file. By default (Open Analysis
checkbox option), the POS file is also created and opened in IVAS for further
analysis.
5.4 Discussion of Spatial Resolution and Spatial Positioning 131
5.4 Discussion of Spatial Resolution and Spatial Positioning
5.4.1 Spatial Resolution
Ideally, spatial resolution is a local measurement that should express the perfor-
mance of the instrument and not limitations originating from specimen or recon-
struction imperfections. The reality for APT spatial resolution is that it is not really
possible to decouple specimen effects from instrument effects because, as stated in
Chap. 2, the specimen itself is the primary optic of the microscope. Known imper-
fections that degrade resolution include: (1) specimen temperature, (2) detector and
optical limitations, (3) reconstruction errors, (4) trajectory aberrations in monophase
materials, (5) local magnification in polyphase materials, as well as others that will
not be discussed in this section. The spatial resolution achieved in a given dataset
will vary within the image and may never reach the ultimate potential of the
technique. Nonetheless, it is instructive to consider the best demonstrated spatial
resolution as a starting point and then consider each imperfection individually.
Vurpillot et al. [36] defined spatial resolution in APT as the inverse of the width of
a damping function convoluted with the reciprocal-space structure of an atom probe
dataset. This parameter gives information about the uncertainty in interatomic
distances as determined by APT. Geiser et al. [37] introduced the spatial distribution
map as a method of averaging real space information about interatomic distances in a
dataset and measured interatomic separation variations in the data. Kelly et al. [38]
summarized these approaches and proposed a standard measurement of spatial
resolution using tungsten specimens. Gault et al. [39, 40] proposed an expression
for spatial resolution in crystalline systems based on statistical considerations of
planar spacing and temperature.
These definitions all utilize to the same essential information to describe spatial
resolution: real-space atom-location uncertainty or the reciprocal-space counter-
part. Furthermore, there are two physically different mechanisms involved with
resolving information in the longitudinal and lateral directions of APT, and mea-
surements for both are required. Each of the references above concludes that a
lateral resolution of ~0.15 nm has been demonstrated in tungsten at ~50 K. Simi-
larly, the longitudinal (depth) spatial resolution has consistently been found to be
about 0.06 nm.
5.4.2 Spatial Positioning (Non-specimen Dependent)
As mentioned previously, there are many factors that can cause the observed spatial
resolution to be less than optimal. They will be discussed individually in the next
sections.
5.4.2.1 Effect of Temperature
The achievable spatial resolution in APT is a function of specimen temperature

since surface diffusion can contribute to a blurring of determined atom positions
[3943]. In order to obtain a reasonable spatial resolution, specimens need to
remain below a base homologous temperature of approximately 0.1. Even with
tungsten specimens, the image resolution will begin to degrade above 30 K. It is
therefore common practice to operate atom probes with base specimen tempera-
tures of 2080 K depending on the material.
With laser pulsing, the specimen apex temperature is being raised in order to
create a very brief period where the probability to field evaporate is high for a short
time (~100500 ps). Thus the best spatial resolution is generally achieved with
voltage pulsing. It is certainly possible to reach temperatures with laser pulsing
where spatial resolution of the image has been severely compromised [44]; how-
ever, often other quality metrics (spectral signal-to-noise for example) are superior
in laser pulsing mode for many materials (many materials exhibit a narrower
thermal pulse than the state-of-the-art electric field pulse). As will become appar-
ent, this trade-off in quality metric improvement/degradation is a recurring theme in
this discussion. Nevertheless, it is important to understand that no level of improve-
ment in the limitations listed below can compensate for thermally degraded spatial
resolution.
5.4.2.2 Detector and Optical Limitations
The position-sensitive detector used in an atom probe must have sufficient pixel
resolution to realize the ultimate achievable resolution. For example, in a 100 nm
diameter FOV, there will be about 500 atoms on the diameter of the image. The
detector will need at least 1,000 pixels across the diameter to position atoms with
precision. Stated differently, 1,000 pixels in 100 nm means that 0.1 nm spatial
resolution is the approximate geometric limitation of the detector. As specimens
blunt during analyses, magnification decreases correspondingly, and a greater
number of atoms are imaged on the detector diameter. It is therefore possible that
in the early part of an analysis the optimal spatial resolution is achieved but in the
latter stages of an analysis, the spatial resolution is limited by detector pixel size.
Optically, the spatial resolution can be limited by voltage instabilities that cause
variations in magnification. These aberrations affect only the lateral resolution.
In instruments using reflectron energy compensators [4549], ions may pass
through a metallic mesh causing a measureable loss in spatial resolution.
Fig. 5.13 Schematic diagrams of the projection for (a) a hemispherical model and (b) a simulated
field evaporated model
5.4.2.3 Basic Reconstruction Algorithm Limitations
The quality of reconstruction algorithms is a central question in spatial resolution

achieved in APT and there are some key elements that should be discussed. As
described above, the standard reconstruction algorithms today have been adapted to
wide FOV images in IVAS [14]. To the extent that the approximations made are
accurate, there is often uncertainty as to the parameter values to be used. For
example, in the absence of microscopy data from the initial endform, the initial
specimen radius is inferred from the evaporation field of the specimen material
and the specimen voltage required to sustain field evaporation at a specified
rate [28]. The evaporation field is typically a function of temperature [50] and
sensitive to specimen composition [51] and crystalline orientation [52]. The values
of the k-factor will depend on the details of the specimen shape [13, 28, 29, 53] and
spatial position relative to other components of the atom probe flight path [5456].
Procedures for determining global values for these parameters from crystalline
specimen APT data have been suggested by Gault et al. [12].
In addition, current algorithms still assume a hemispherical apex on the specimen
and may assume tangential continuity from the apex to the shank. Many investiga-
tors have observed that the spherical shape is inaccurate in both experimental and
simulated images of field evaporated specimens [24, 21, 24, 26, 3335, 5759].
Evaporated specimens have a blunter apex near the center with greater curvature
near the edges, as shown in Fig. 5.13. The evaporated endform also is not tangential
to the cone of the specimen, as already discussed above. The principle effect of this is
a non-uniform angular magnification [33] (white arrows in Fig. 5.13b show regions
of higher magnification which correspond to low radius regions).
Although the following two aberrations are arguably reconstruction limitations,
they are specimen dependent and have therefore been separated out for clarity.
5.4.3 Spatial Positioning (Specimen Dependent)
5.4.3.1 Monophase Aberrations
The next level of complexity that limits spatial resolution is an aberration found in
crystalline materials. Atoms generally field evaporate from the edges of closely
packed planes and not from the interior of a plane [41]. As a result, in areas of high
atomic packing density, planar facets form on the evaporating specimen. These
planar facets create deviations from a smooth surface and result in local nonuni-
formities in image compression as a function of the azimuthal angle about a central
axis of the specimen [60, 61].
5.4.3.2 Polyphase Aberrations
Specimens with multiple phases present the most challenging reconstructions.

As the specimen is the primary optic, understanding of the shape of the specimen
during the process of field evaporation is an important factor in reconstruction.
Field Evaporation of Precipitates
In principle, not only is it necessary to take all of the above effects into account, but
also it must be done for phases within phases. Using a previously described method
[24, 26, 62, 63] we have simulated the cases of a high-field precipitate in a low-field
matrix (matrix evaporation field is half that of the precipitate evaporation field) as
well as the case of a low-field precipitate in a high-field matrix (matrix field is 15 %
higher than precipitate field) [64]. Consider the former case of a high-field spherical
precipitate as shown in Fig. 5.14. After some evaporation, the specimen will form
an apex structure as shown in the series Fig. 5.14ac. Once the second phase
emerges at the surface, the entire endform is altered (with the high field phase
protruding from the surface) and the trajectories of ions in both the matrix and the
precipitate are different from those in a monophase material. In general, the atoms
of a high-field phase will see greater angular magnification than the low-field phase
[1, 2, 65]. If the evaporation field of the second phase is less than the surrounding
material as shown in the series Fig. 5.14df, the situation is reversed, and the
precipitate region is flattened on the surface. For both cases, once sufficient material
is evaporated following the disappearance of the precipitate, the specimen resumes
its equilibrium electrostatic endform.
The errors that occur when using standard reconstruction methods may be
evaluated by plotting the original atom positions (from the unevaporated simulation
data) vs. the detector impact positions from the simulation [26]. These data are
shown in Fig. 5.15 for the precipitate shapes presented in Fig. 5.14b, e. The slope of
the curves in Fig. 5.15 provides an estimate of the magnification. The data in
Fig. 5.14 Simulated evaporated specimen shapes for (ac) a precipitate with an evaporation field
twice that of the matrix and (df) a precipitate with an evaporation field 15 % lower than the matrix
Fig. 5.15 The errors

introduced when using the
standard reconstruction
model for (a) a precipitate
with an evaporation field
twice that of the matrix and
(b) a precipitate with an
evaporation field 15 %
lower than the matrix
Fig. 5.15a suggest a very high magnification for the region of the precipitate near
the edge of the matrix where there is a narrow range of original atom positions
spread over a wide range of detector impact positions and a narrow region of
potential ion crossing in the matrix near the precipitate (arrowed). The results in
Fig. 5.15b suggest a low magnification in the precipitate region with a narrow
region of higher magnification in the matrix near the precipitate. Note that the solid
lines in the curves in Fig. 5.15 denote data that would be obtained if there were no
precipitates present in the simulations.
In either case, the effects of the resulting shapes on ion trajectories from either
phase raise the possibility of ion trajectory crossing [66]. If ion paths do cross, the
task of correcting this aberration gets more challenging and may be uncorrectable in
a traditional ion-by-ion treatment.
Reconstruction of Precipitates
Let us examine the result of reconstructing a simple dataset containing precipitates

within the context of the Wide-Angle Reconstruction Protocols (Sect. 5.3.5)
discussed above. Note that we are not exploring the variations just discussed
above so much as we are presenting a typical case where we desire to reconstruct
features believed to be spherically shaped.
Figure 5.16 shows various reconstructions of a section of a dataset containing
Mg precipitates in a GaN matrix. All reconstructions have been created using Shank
Evolution (10 ) with a 70 % Efficiency and a Sphere-Cone Radius Ratio of 1.2 but
with a variety of R0 and values.
For the case of R0 20 nm, all of the reconstructions over the range of image
compression values shown (Fig. 5.16a, d, and g) result in an elongation of the
precipitates along the z axis, particularly near the top regions of the data. For all
three sets of reconstructions using a constant image compression (left to right in the
figure), the effect of increasing R0 is observed. For the cases where is 1.65 and 1.8
(Fig. 5.16f, i), increasing R0 to 40 nm has resulted in a flattening of the precipitates
along the z axis. The volume shown in Fig. 5.16e (R0 30 nm and 1.65)
appears to be a reasonable reconstruction, assuming that the precipitates are
approximately spherical. However the similarities between precipitate shapes in
this reconstruction and the one shown in Fig. 5.16c are evident. This illustrates the
effect of R0 and appearing together in the denominators of the reconstruction
equations for magnification and z increment (5.6 and 5.7) and indicates, that
within the current model of reconstruction, the system is under-constrained. This
effect has been discussed by Vurpillot et al. [67], and further by Gault et al. [68],
who suggest the additional reconstruction constraint of a constant ratio between
k-factor (inversely proportional to R0) and .
The LEAP reconstruction for multiple precipitates within a dataset is difficult to
optimize, even if the results of any differences between evaporation field of the
precipitates and evaporation field of the matrix are ignored [1, 2, 23, 65]. The user
must take into account all possible information, ideally including some estimation
Fig. 5.16 An array of reconstructions for a dataset containing Mg precipitates in a GaN matrix
created using shank angle (10 ) reconstruction mode with 70 % efficiency and fs/c of 1.2 with a
variety of initial radii and values. The array is organized with fixed radii columns and fixed
rows (see text for details). Note that all reconstruction volumes have been scaled to ~80 nm in the
z dimension
of evaporated specimen shape (radius and shank angle), in order to minimize the
errors. In the case where no additional information is available for a starting
reconstruction, the user can make multiple reconstructions like those shown in
Fig. 5.16 and visually select the one which results in spherical precipitate shapes
and also maintains the image compression near the nominal mean value of the
instrument.
Fig. 5.17 Simulated

evaporated shape and
corresponding detector
hit-maps for (a) ABA and
(b) BAB configurations
where the evaporation field
for material A is to be 20 %
higher than that of material
B and the initial shape of the
specimen was a hemisphere
for each case
Field Evaporation of Thin Films
In this section we present field evaporation simulation examples for thin films in two
different specimen orientations: plan-view and cross-section (see Chap. 2 for details
about specimen orientations) so that the user may understand how to interpret the
results of reconstructions obtained for such structures. While the cross-section
orientation may seem more simple due to its steady-state nature, for a zero shank
angle case, it has the more difficult condition of not possessing azimuthal symmetry.
Two cross-section examples consisting of ABA and BAB layered structures
(each layer is 5 nm) were simulated using an electrostatic simulation [3, 24, 62, 63]
similar to that developed by Vurpillot et al. [69, 70]. Each specimen was assumed to
be face-centered cubic and oriented in the <001> direction with a lattice parameter
equal to that of aluminum. The evaporation field for material A was chosen to be
20 % higher than that of material B and the initial shape of the specimen was a
hemisphere for each case. Figures 5.17a, b show the evaporated specimen surface
and the simulated detector field evaporation patterns for the ABA and BAB cases,
respectively [64]. In Fig. 5.7a, a flattening of the low field B region is evident in the
specimen center. This causes a lower magnification in this region (compared to
the A regions) and thus, in the hit-map, the B region is compressed compared to the
A regions. For the case of BAB, Fig. 5.17b, a protruding region of high field A
material in the specimen center is observed, with the same result in the hit-map;
that is, the higher density regions correspond to the B material, which has flattened
Fig. 5.18 Simulation of a plan-view multilayer structure containing layers of Si Substrate/Cr(5)/

Co(5)/Cu(5)/Co(5)/Cr(50) on silicon (numbers in parentheses indicate layer thickness in nm)
at the surface in both cases. The important thing for the user to understand in these
cases is that, due to the anisotropic magnification across the surfaces, the B layer
in Fig. 5.17a will be reconstructed as narrower than 5 nm while the A layer in
Fig. 5.17b will be reconstructed as wider than 5 nm.
For the plan-view thin film example, a simple five layer system was simulated [3].
Image hump model evaporation fields [51] were used (30 V/nm for Cu, 37 V/nm for
Co, 27 V/nm for Cr and 33 V/nm for Si) for a simulation of a plan-view multilayer
structure containing layers of Si Substrate/Cr(5)/Co(5)/Cu(5)/Co(5)/Cr(50) (num-
bers in parentheses indicate layer thickness in nm), Fig. 5.18a. The images in
Fig. 5.18a show the shape of the specimen as it is evaporated through the layers.
An estimate of the radius of curvature (measured on the specimen axis) during the
evaporation of this structure is presented in Fig. 5.18b. A highly non-monotonic
result is observed as the specimen radius (on axis) varies from ~20 nm to nearly
140 nm during the course of the shape evolution [3]. Note the differences in the
behavior of the specimen radius when comparing the case of evaporation from
low-field material to high-field material with the opposite case. When a low-field-
on-high-field interface is encountered the specimen radius becomes very high, while
evaporation of a high-field-on-low-field interface produces a very low radius [24]
and a high magnification.
Fig. 5.19 LEAP atom map

showing the repeating
layered structure of Ni
(56 nm) and Cr (57 nm)
layers from the NIST SRM
2135C
Reconstruction of Thin Films
It is often the case when reconstructing complex datasets that the user has some
known information regarding portions of the microstructure. Examples include an
interface that is nominally planar, a measured thickness of a region or phase, the
shape of precipitates, or combinations of any of the above. In this section we will
present and discuss the trends in interface curvature and layer thickness with
variation in the reconstruction variables: Initial Tip Radius, Image Compression
Factor, Shank Angle, and Sphere-Cone Radius Ratio (R, , , and fs/c). Our
reference sample is a repeating layered structure of Ni (57 nm) and Cr (56 nm)
layers from the National Institute of Standards and Technology Standard Reference
Material (NIST SRM 2135C). Figure 5.19 shows the LEAP data on which the
analysis below was performed.
Fig. 5.20 Variable radius analysis of the NIST NiCr thin films. (a) R0=75nm, (b) R0=100nm,
and (c) R0=125nm
Figure 5.20 shows three reconstructions using increasing initial radii from 75 to
125 nm. The reconstructions shown in this figure were created using 15 ,
1.5, and fs/c 1.5. Varying the radius changes both the curvature of the upper
Ni/Cr interface as well as the Cr layer thickness. For the purposes of this discussion
we will define an interface curvature like that in Fig. 5.20a (concave up) as negative
(with a flat interface having a curvature of zero). The trends in the reconstruction
variables on the upper Ni/Cr interface curvature and the Cr layer thickness are
shown in Fig. 5.21 (for 15 ). Note that a smaller shank angle ( 5 ) has
the effect of shifting the reconstructed Cr layer thickness to greater values while
only slightly shifting curvature values to be more negative, Fig. 5.22. Both and
R have a significant effect on the curvature and thickness values in these examples
while the effect of fs/c is less evident. The best reconstruction parameters in this
situation both minimize the curvature and provide the correct Cr thickness (56 nm).
By fitting a polynomial model using all of the input parameters to the reconstruction
and investigating the (R, ) solution space to simultaneously optimize both Cr
thickness and Ni/Cr interface curvature, we can draw several conclusions:
1. Being constrained to a small shank angle of 5 with tangential continuity
( fs/c 1.0) forces the best fit (intersection of the minimum curvature and the
correct Cr thickness) toward low radius values (~75 nm) and high image com-
pression (~2.1).
2. Shifting to greater fs/c (1.5) slightly shifts the intersection (in R, space) of the
minimum curvature and the correct Cr thickness.
3. A greater has more impact than fs/c on shifting the minimum curvature and the
correct Cr thickness toward solutions that are parallel (in R, space).
4. Finally, greater fs/c and greater shank angle values shift the predicted lines of
minimum curvature and correct Cr thickness on top of each other, thus resulting
in a series of (R, ) solutions that both minimize curvature and result in the
correct Cr thickness.
Fig. 5.21 Trends with specimen radius for curvature and Cr layer thickness using 15 for
(a and c) fs/c 1.0 and (b and d) fs/c 1.5. The dashed lines show the nominally correct values
for Cr thickness and curvature
5.5 A Word on Density Relaxation
In APT datasets, the reconstructed density is not uniform due to specimen endform
deviations away from the hemispherical shape that is assumed in the reconstruction
protocol. Improving the aberration by gridding the data and smoothing the density
variations (density relaxation, with the objective of density correction) has been
proposed by various authors as a way to rectify these errors [23, 67, 71, 72]. IVAS
enables the user to perform Density Relaxation on a ROOT or POS file. For the
former, the feature requires an active link to the corresponding RHIT file. The user
can perform the density relaxation only after a reconstruction has been completed
(see the IVAS User Manual for further details). The output from this procedure is a
new POS file. A brief explanation of the variables used in the density relaxation
procedure is included below:
Smooth Iterations specifies how many smoothing iterations the software
performs. In order to attempt to affect different scale features, different XY
gridding is used (both grid-size and grid-orientations are randomly varied) for
5.5 A Word on Density Relaxation 143
Fig. 5.22 Trends with specimen radius for curvature and Cr layer thickness using 5 for (a and
c) fs/c 1.0 and (b and d) fs/c 1.5. The dashed lines show the nominally correct values for Cr
thickness and curvature
each iteration. Note that processing time for the correction increases with
increased iterations. Artifacts can become visible if too many iterations are
performed. This parameter applies only to density relaxation using the XY
Relax option.
Ions Per Shell specifies how many ions to relax per depth shell. Smaller values
allow smoothing of smaller gaps, but can cause low-statistics artifacts. Larger
selected fiducial regions, from the Detector Event Histogram, require a higher
Ions Per Shell value than do smaller regions. Ion values of 5,000 for small fields
of view and 20,000 for larger fields of view detectors are typical. This parameter
applies only to density correction using the XY Relax option.
Z Extent is a slider bar for adjusting the percentage of the reconstruction data or
POS file volume to be density corrected. For ROOT files, the reconstructed apex
of the specimen is 0 %, and the end of the dataset is 100 %. This parameter limits
the density-corrected POS file output to a particular specimen-depth ROI.
Do MCP Correction is a toggle switch that allows the user to prohibit ions from
being relaxed into any low-efficiency microchannel plate region contained in the
data. By default, the density correction procedure relaxes ions into the MCP
Fig. 5.23 Example of a transistor atom map (a) before and after (b) density relaxation
low-efficiency region (if present), which is not an accurate correction. If you

select this option, IVAS prompts you to enter, in the X and Y text boxes, the
nominal position of the low-efficiency region as read from the Detector Event
Histogram in IVAS Cal/Recon Wizard.
XY Relax is a toggle switch that allows the user to disable the density relaxation
occurring in xy.
Z Reorder is a toggle switch that uniformly redistributes the z positions of data
within full z-length volumes. Note that when XY Relax and Z Reorder are both
selected, then XY Relax is performed first.
Clear Voids is a toggle switch that allows XY Relax to fill large internal voids in
the data. In all cases of density relaxation, the filling of bins that should be empty
(edges) is prevented by locking bins with zero data. When the Clear Voids
switch is not toggled on, all grid cells with no data will be locked. When this
switch is toggled on, then cells internal to the data will be allowed to relax. Bins
which are interior to the data are not locked.
Open Analysis will automatically open the density-relaxed POS in the IVAS
image pane after density relaxation is completed.
It should be noted that when density relaxing a ROOT file, the XY relaxation
occurs in groups of atoms collected in a thin shell in detector space. In contrast,
when density relaxing a POS file, atoms of similar real space z position are relaxed
together. These atoms may have been acquired at different times, especially when
the POS file is the result of a coordinate-transformed ROI export. Because different
atom groups are relaxed together, POS-based and ROOT-based density relaxations
will be different. Also, because of the possibility of POS files having their coordi-
nates arbitrarily transformed, the option to avoid MCP efficiency filling is not
supported.
Figure 5.23 is an example of a POS file showing the gate region from test
transistor structure [73] before (Fig. 5.23a) and after (Fig. 5.23b) real-space density
5.6 Reconstruction Case Study: NIST Standard Reference Material 2134 145
lateral relaxation. The light gray points are Si ions and the dark gray points are O
ions. It is believed that the oxide layers are artificially broadened due to local
magnification (as illustrated in Fig. 5.17). Correcting the resulting lower local
density restores the oxide layers to a more physically correct proportion while the
overall size of the structure is approximately preserved. Also note that the edge
atoms are locked in place and thus subject to more distortion as the rest of the ROI is
corrected.
5.6 Reconstruction Case Study: NIST Standard

Reference Material 2134
5.6.1 Reconstruction Parameter Discussion
The 3D reconstruction of ion positions is most often performed using one of the
three radius evolution approaches described previously: Voltage, Shank, or Tip
Profile. The number of user-selected reconstruction parameters varies depending
on the algorithm chosen, as summarized in Fig. 5.24. Extra measurements can be
taken to establish, or narrow the range of, the values of some of these parameters;
however, because the algorithms are approximations and do not capture the exact
field evaporated shapes for real materials, nonphysical parameter selections may
result in more accurate reconstructions. In this section, general considerations for
selection of these parameters are discussed in the context of a relatively simple
material system, the National Institute of Standards and Technology Standard
Reference Material (NIST SRM 2134), which is As-implanted silicon.
Fig. 5.24 Reconstruction process flow chart including the relevant reconstruction parameters
(highlighted in yellow)
There may be as many as five key reconstruction parameters necessary to define

a reconstruction in IVAS (see Fig. 5.24). The first two, Detector Efficiency and
Image Compression Factor are considered hardware dependentto first order
these parameters originate from detector limitations and ion flight path geometries
and are fixed for all analyses. In reality, these values are not always well known and
some variation is expected from analysis-to-analysis [13]. The next parameter is the
product of the k Factor and the material Evaporation Field (kF in Fig. 5.24)
(equivalently equal to the product of the specimen voltage divided by R) and is
the primary free reconstruction variable because it is easy to manipulate and directly
impacts the shapes of geometrical features within the reconstruction. Steps can be
taken to constrain this parameter but, in general, a value is usually chosen to provide
the most accurate reconstruction as interpreted by the user. Finally, the last two
parameters are specimen-based (Shank Angle and SphereCone Radius Ratio) and
describe the specimen shape and response of specimen shape to field evaporation. A
practical discussion of each of these parameters follows.
5.6.1.1 Parameter #1: Efficiency
Detector Efficiency results from a combination of considerations which include the

open area of the microchannel plates (discussed in Chap. 3), the response of the
MCP pores to ion impacts, specific electronic thresholds set for each detector, and
other factors as well. The default efficiency values in IVAS are primarily based on
geometrical considerations for the MCP open area and any meshes in the flight path
(HR model only). Three different detector types have been provided with LEAP
systems to date and these have default efficiencies of 37 % (HR model), 50 %
(DT-200 Si model), and 50 % (DT-100 Si model) as provided by IVAS. The actual
efficiencies may be subject to some variation as a function of specimen and voltage,
as well as some nonuniformity as a function of impact location on the detector.
Attempts to experimentally measure the efficiency with SDM methods (see Chap. 6)
find typical variations of as much as 10 % or more due to a wide variation in
specimen preparation precision and reconstruction parameter uncertainties.
Variations in efficiency as a function of impact location are easy to understand
based on the detector design, as shown in Fig. 5.25. In order to be detected, ions
must strike the secondary-electron-generating coating located on the internal wall
of a tilted MCP pore. The higher the ion strikes in the pore, the greater the level of
electron multiplication and the higher the signal amplification (higher likelihood
of detection). For some ion trajectories, the ions can strike very deep in the MCP
pore of the first plate (or possibly travel through it entirely), as shown schematically
toward the bottom of Fig. 5.25. This leads to reduced amplification and
nondetection of some fraction of these events. This spatial variation in efficiency
is difficult to model or measure, and is thus not taken into account in the general
reconstruction algorithm. Due to the focusing lens of the reflectron, ion arrival
angles vary only slightly across the MCP in LEAP HR systems and these variations
in efficiency are not measurable. Reflectron-based systems have a metallic
Fig. 5.25 Specimen, MCP detector, and resultant detector hit density map showing ion trajecto-
ries and their path into the MCP. The resulting efficiencies are reflected in variations in the hit
density map. A low-efficiency region of the detector is highlighted by the dashed-red circle where
ion trajectories are such that they pass through the top MCP plate without causing a significant
electron cascade
mesh in the flight path which further reduces the efficiency in proportion to the
transmissivity.
Hardware limitations are not the only factors which affect efficiency. For all
available reconstruction algorithms, efficiency modifies the z increment of each
ion (5.7 above) to compensate for the missing volume of undetected ions. For
analyses that have a significant number of events that are not contained in ranged
peaks, this is effectively a loss in detection efficiency. In such cases, the overall
efficiency should be modified to take this into account. The actual percentage of
unranged peaks can be measured, or the constant time-independent background
estimate provided by IVAS can be used to estimate both the unranged ions and
those missed while the detector is off between pulses:
f
0 5:10
f Bgd
where is the effective detection efficiency, is the estimated hardware-limited

Detector Efficiency, f is the pulse frequency in kHz, and Bgd is the background level
in counts per million ions per nanosecond.
5.6.1.2 Parameter #2: Image Compression Factor
Image compression has been defined previously and results from a combination of
flight path, specimen, and electrode geometry effects [74]. Figure 5.26 shows the
modeled variation in image compression for various tip dimensions and shank
angles for the straight flight path configuration [29] (note that the LEAP
reflectron configuration of reflectron and detector leads to differences in these
variations) as well as experimental estimates based on hit-map and pole projection
evolution [31]. IVAS implements a constant for the entire reconstructed volume,
Fig. 5.26 (a) Modeled 4000X Si as a function of tip radius for various shank angles (Reprinted
with permission from Elsevier [29]). (b) Experimental 4000X Si as a function of voltage as
determined by variation in magnification of crystallographic features
which may limit reconstruction accuracy for analyses with large changes in voltage
(specimen radius). A reconstruction approach where a single dataset of large
voltage range is broken into multiple reconstruction regions, each with a limited
range and thus better approximated by a constant , may improve overall accuracy.
This limitation is mitigated in specimens that have narrow analysis voltage range.
Specimens with nearly parallel sides (limited radius evolution during analysis) are
more accurately reconstructed using to single value, although such geometries
may be more prone to premature failure or provide poor thermal/electrical proper-
ties due to their narrow cylindrical shape.
5.6.1.3 Parameter #3: Initial RadiusR0 (or Product of k-Factor

and Electric FieldkF)
The most commonly adjusted parameter is R0, which is determined by using (5.8).
Unfortunately, the product (kF) is prone to large uncertainties. The k-factor may
vary substantially depending on the specific local electrode-specimen geometry.
Likewise, evaporation field tables exist [51], but they are typically model driven
and list values at a single temperature. Even in tables which include experimental
measurements, temperature variation is not normally reported and thus these values
can only be considered as estimates. Experienced users commonly choose values
based on past experience. If there are features in the ROI, such as a planar interface
or spherical precipitates, then R0 is often adjusted until these geometries have the
correct shape.
At this point, it is useful to summarize the effects of and R0 on reconstruction.
detector efficiency affects only the z-scale of the reconstruction. For a voltage
reconstruction, changing the efficiency expands or contracts the z-positions of the
ions and does nothing else. For shank angle and tip profile reconstructions, effi-
ciency changes the depth of the reconstruction and affects regions of the radius
evolution function it is more complicated than just skewing the z-axis, but the net
effect is similar. On the other hand, and R0 both tend to change the shapes of
geometrical features and their effects can overlap [29], so users commonly fix one
parameter () and adjust the other to achieve an accurate result as interpreted by the
user. To some degree, evaluation of validity of the reconstruction by the user is
generally an unavoidable requirement of current reconstruction algorithms.
Much of the anecdotal knowledge for the physical basis of is related to the
electrostatic relationship between the specimen and the aperture. It should be noted
however that other optical elements contribute to as well, in particular detector
bias and reflectron elements [47].
5.6.1.4 Parameters #4 & #5: Shank Angle & SphereCone

Radius Ratio and fs/c
These final two parameters are only relevant for nonvoltage reconstructions and
are determined from the shape of the specimen. The shank angle or evolution of
the shank angle (see Appendix C for details of calculation of shank angle) and the
sphere-to-cone ratio can be used to dictate the radial evolution of the reconstruction.
Because apex curvature can become complicated (nonspherical) by nonuniform
evaporation fields within a single specimen, deviations from the actual tip profile
may be required to provide for more accurate reconstructions than the physical
profile would provide.
5.6.2 Experiment and Analysis Details
Analysis of silicon implanted with arsenic provides a relatively simple example to

aid in our examination of reconstruction parameter selection [31]. The sample is a
bulk, crystalline silicon substrate that has been implanted with arsenic ions. The
arsenic composition, which varies as a function of depth, has been characterized by
multiple methods for use as a standard reference material for secondary ion mass
spectrometry (SIMS) [74, 75]. The dose, or number of arsenic atoms per unit area,
is the certified metric for this material, but the variation of composition as a function
of depth as measured by SIMS has also been reported in the literature [74, 75]. The
purpose of this section is to understand both the experimental precision and
accuracy of APT measurements by comparing multiple measurements from the
same material with established values. Although this is a relatively simple system
with relatively homogeneous material properties, choices pertaining to specimen
preparation, reconstruction method, and reconstruction parameters all affect the
resulting dose measurements.
5.6.2.1 Specimen Preparation and Shape Manipulation
In general, variations in analysis voltage may require alteration of the values

of efficiency and needed to obtain accurate reconstructions. Thus specimens of
similar size and shape analyzed under similar conditions should simplify the
reconstruction process. Specimens with a narrow shank angle over the implant
region of the specimen were manufactured with the aid of a multi-layer cap (50 nm
nickel followed by 120 nm amorphous silicon). Standard annular milling steps were
utilized (see Chap. 2) except that the final annular milling pattern was applied for a
significantly longer time to create a more cylindrically shaped specimen (~24
shank angle). Subsequent application of a low-kV cleanup step resulted in a large
shank angle over the length of the capping material region with little change in
previously formed shank angle over the implant. The 120 nm poly-silicon thickness
is sufficient to allow removal of damaged material and most of the nickel cap during
low-kV milling without affecting the shape of the implant region.
5.6.2.2 Analysis Conditions
Although this material (silicon) has suitable electrical conductivity to attempt

voltage pulsed acquisition, the yield benefits gained by laser pulsed acquisition
are substantial. Low base temperature, high pulse frequency, and medium-to-high
laser pulse energy all typically provide for superior data quality, but also have
effects on analysis yield. Standardized conditions of 30 K, 500 kHz, 30 pJ and 1 %
detection rate were used to ensure good data quality and high analysis yield [31].
5.6.2.3 Reconstruction Methodology
For these particular specimens, smooth variations in analysis voltage suggest that the
reconstruction algorithm plays only a minor role in final reconstruction accuracy.
Since voltage reconstruction assumes a constant relationship between voltage and
radius, let us consider the voltage evolution and silicon charge state ratio (CSR or Si2
+
/Si+), which provide some indication of changes in evaporation field [7]. Figure 5.27
shows the silicon CSR evolution and analysis voltage and as a function of detected
ion number (or depth). Except for the region near the native oxide, both curves
follow a similar trend. Ignoring the region near the native oxide, the CSR evolution
curve is easy to understand. When the overall specimen shape is small, the heated
volume is small. As the analysis proceeds, the specimen radius increases, the heat is
spread throughout larger and larger volumes providing for a lower peak apex
temperature (for fixed laser pulse energy and absorption). Under the constraint of
an approximately constant evaporation rate, a lower evaporation temperature
requires a higher evaporation field, which is exhibited by the increase in CSR.
Effectively, the evaporation field is changing as a function of analysis depth.
Fig. 5.27 (a) Evolution of

silicon CSR as a function
of depth and (b) evolution
of voltage as a function of
ion sequence (or depth).
This figure illustrates some
of the pitfalls of using CSR
as a proxy for apex
temperature. In this case,
changes in apex curvature
(multiple material layers of
differing evaporation field
at the surface
simultaneously) result in
complex changes to the
CSR. This makes it difficult
to decouple effects due to
changes in temperature and
changes in tip shape
(materials with dissimilar
evaporation fields)
This suggests that voltage reconstruction may not be appropriate, and a tip profile or
constant shank angle reconstruction may be more accurate (especially given the tip
shape near the implant).
It should be noted that an opposite CSR trend is observed in voltage pulse mode.
As in the case above, as the specimen blunts during analysis, the FOV increases and
the constant detection rate causes the number of events per unit specimen area per
unit time to decrease. This results in a lowering of the evaporation field over time
and a decrease in the CSR. For the laser pulsed case, this effect is usually
outweighed by the required field increase due to a lower evaporation temperature.
In order to maximize precision (and secondarily accuracy), reconstruction
parameters should be independent of arbitrary judgments by the user. The following
analysis protocol was used:
1. Images were taken of specimens before analysis to make accurate measurements
of the shank angle in the vicinity of the implant region.
2. Only analyses which successfully obtained data from the entire implant and were
stopped in a controlled manner (without specimen failure) were considered.
3. Images were taken of the specimens after analysis to make accurate measure-
ments of the final radius (RF) and sphere-cone radius ratio ( fS/C).
4. Specimens were fabricated to be of similar sizes, and specimen-electrode dis-
tances were chosen so that projected ion patterns filled similar regions of the
detector at similar magnification values (minimizing detector efficiency and
variations from analysis to analysis).
5. The default efficiency value of 0.5 was assumed (LEAP 4000X Si default).
6. Two doses were extracted based on differing constraints:
a. A single, fixed (1.45) was used for all analyses.
b. Each analysis utilized an independent based on a quantitative optimization
of some aspect of the reconstructed geometry (other than the known arsenic
distribution).
In this procedure each free reconstruction parameter is either held constant
(efficiency), determined by microscopy (shank angle, RF (which may be used in
place of R0), and fS/C), or determined by quantitative appraisal of some feature
in the reconstruction (). The ultimate result is a high precision and an acceptable
accuracy.
The reconstruction procedure benefits from selection of some internal feature, or
features, to judge the quality of the reconstruction as a function of . Figure 5.28
shows an atom map from an example implant to illustrate the various features
present in the sample. Potential internal calibration features within the sample
volume include: flatness of the various interfaces (nickel/poly-silicon and native
oxide), thickness of the poly-silicon region, and any atomic plane spacing that
might be observed in crystalline regions of the acquisition when the data quality is
sufficient to resolve such features.
Quantitatively assessing these features is time consuming, but relatively straight
forward. IVAS allows for chemical isoconcentration surfaces to be identified and
measures of roughness are provided for each surface (see Chap. 6). The roughness
values are calculated from a best-fit plane, so minimization of the roughness
parameters quantitatively determines the optimal flatness. Once interface curvature
is minimized, measurements of layer thickness can be performed by interpreting 1D
composition plots and enforcing critical concentration criteria as defining the
location of layer boundaries. Atomic plane spacing information can be extracted
from SDM calculations (see Chap. 6) also available in IVAS. We now consider the
merit of the measurable features available in this specimen type one by one.
Interface Flatness and Silicon Layer Thickness
The nickel/poly-silicon interface and the poly-silicon thickness are possible feature
choices. In principle, the interface should be approximately flat and the poly-silicon
thickness constant and easy to measure to high precision (194 2 nm). However,
Fig. 5.28 An example arsenic implant dataset which illustrates the various features present in the
capped sample. The colored atom map shows a sampled number of ions (Ni green, Si gray, O blue,
As magenta). A 3 % O isoconcentration surface is also shown. The nickel and poly-silicon caps are
above the blue native oxide region that formed on the silicon wafer surface prior to deposition of
the cap (near the center of the image). The changing arsenic composition is highlighted by the
varying density of magenta atoms below the native oxide. Plotting the arsenic composition as a
function of depth results in a 1D composition profile (red graphic to the right of atom map) and
integrating this curve results in the dose
this particular interface is not very well behaved and has a very small area (high
magnification), providing a roughness metric with a poorly defined minimum. Also,
the fact that the top portion of the tip has a larger shank angle, and that it differs
significantly from the shank angle in the implant region, makes it difficult to
reconstruct both the cap and implant regions accurately. In other words, choosing
a to give the correct cap thickness will provide a that produces a distorted
implant region. For Fig. 5.28, the thickness of the poly-silicon is ~170 nm while the
implant depth is reasonably well approximated when compared to SIMS.
By the time the analysis proceeds to the native oxide interfaces, microscopy
images reveal a shank angle that is nearly constant and small, while reconstruction
Fig. 5.29 Optimization

of oxide interface flatness
as indicated by roughness
of the 3 % oxygen
isoconcentration surface.
The top and bottom
interfaces required different
values for optimal flatness
(Reprinted with permission
from Elsevier [31])
trends show the native oxide layer is well defined with significant surface area. As
shown in Fig. 5.28, the native oxide actually consists of two separate surfacesthe
top and bottom chemical interfaces. A series of reconstructions were generated with
a variety of values using the constant shank angle option. Only the region
beginning 1020 nm above the oxide and continuing through the implant was
evaluated. Because RF was known from electron microscopy measurements, R0 was
adjusted until the reconstruction resulted in the correct RF, and fS/C was extracted
from the micrographs. The quality of the oxide interfaces was found to depend on
the definitions of the 3D gridding parameters (see Chap. 6) as well as the specific
isoconcentration surface. In order to produce two distinct surfaces that are not in
contact, smoothing (or delocalization) of the data was necessary to damp statistical
fluctuations while leaving enough detail to define the curvature of the surface.
Oxygen composition values between 0 and 5 at.% identified appropriate top and
bottom oxide surfaces. Smoothing greater than 2 nm often failed to produce
independent surfaces while no smoothing produced highly roughened surfaces
that again made it difficult to produce independent surfaces. Voxels smaller than
1 nm3 provided insufficient statistics to provide smooth composition values, while
larger volumes smeared the data and made it difficult to identify the interfaces.
Ultimately, 3 % oxygen surfaces with voxel sizes of (1 1 1) nm and (x,y,z)
delocalization of (2,2,1) nm provided well defined, independent surfaces without
holes as well as roughness values that followed understandable trends, as illustrated
in Fig. 5.29. Unfortunately, the top and bottom portions of the interface require very
different values to obtain flat interfaces. This is in part due to the different
specimen shapes that occur during field evaporation of the two different types of
interfaces. Using one of these values or some combination (average) may provide a
systematic way to choose an image compression yielding repeatable results, but
with some significant reduction in accuracy. To provide additional insights into this
behavior, a comparable model structure was field evaporated via simulation and
reconstructed.
Fig. 5.30 Atom map of

the reconstructed positions
from a field evaporation
simulation. The field ratio
between the dark blue
atoms and the light blue
atoms is 1.5 and
approximates the field ratio
for Si and SiO2. Regions of
the atom map with distorted
atomic planes suggest
regions likely to be
aberrated in reconstructions
of real experimental data
Figure 5.30 shows the simulated shank angle reconstruction of a 2 shank angle
specimen created with a 2 nm thick region of material (light blue) having 1.5 times
the evaporation field of the surrounding material (dark blue). The reconstruction of
the simulated data results in obvious distortions surrounding the oxide region
(in agreement with previously observed trends [26]). These distortions result
from the non-hemispherical surfaces that occur as the specimen is field evaporated
through the differing evaporation field materials. A positive feature of the
reconstructed simulation is that while the reconstructed atoms above the oxide
are aberrated for 5 nm of depth (note the curved atomic planes and nonuniform
reconstruction atomic densities between the dashed red lines), the atoms immedi-
ately below the oxide appear unaberrated. Thus it should be possible to reconstruct
the entire region below the oxide accurately.
Fig. 5.31 SDM for {100}

planes in arsenic implanted
silicon (Reprinted
with permission
from Elsevier [31])
Silicon Atomic Planer Spacing
The final available feature is the atomic plane spacing exhibited by the crystalline
silicon bulk. Figure 5.31 shows the spatial distribution map calculated from the
{100} planes of an example implant dataset. Creation of relatively good quality
spatial distribution maps allow the user to vary the image compression value until
the atomic plane spacing matches the ~0.543 nm {100} spacing known for this
material. Of course, in order to create an SDM, there must be an initial value used
for the image compression. As long as this initial guess for the image compression
was less than ~20 % of the optimal value determined at the end of the process, good
quality SDMs were produced and this method is relatively straightforward.
The atomic plane spacing does not suffer from the same limitations as features
discussed above. The thickness is known to an extremely high accuracy and
precision from x-ray diffraction methods. The layers do not suffer from evapora-
tion field differences as do almost any other layer-based metric, and thus the
specimen shape evolution is uniform (outside of crystallographic effects) and
resides within the ROI. Therefore, this feature is the most appropriate for
Fig. 5.32 Comparison

of a LEAP arsenic
composition profile
(triangles) and the same
profile as measured by
SIMS (dashed line) [76]
quantifying the accuracy of the reconstruction for choosing . In practice how-

ever, the very regular layer-by-layer evaporation order required to be able to
resolve planes in a reconstructed image may be common in pure materials, but is
rare in materials of interest.
5.6.2.4 Reconstruction Results
With the reconstruction parameters determined, simple 1D dopant profiles may be

calculated. In order to account for background signal in the arsenic peaks, pre-peak
neighboring ranges (see Chap. 6) were identified and subtracted from the arsenic
ranges to provide background corrected compositions. These compositions were
then converted to atomic density and compared to the SIMS profile, as shown in
Fig. 5.32. Integration of the curve results in the implanted dose which was then
compared to the certified values.
As mentioned in describing the reconstruction protocol, dose values were
calculated for a constant of 1.45 as well as a variable used to obtain correct
plane spacing (Table 5.1). Note that of the nine datasets included in this study, only
one SDM was of insufficient quality to determine the image compression. The peak
arsenic composition was extracted from the 1D dopant profiles as this value is
expected to be less sensitive to any errors in reconstruction. Experiments were
performed on two different LEAPs utilizing three different local electrodes, so we
may consider the variability of the results in two ways: pool all the data together to
achieve improved statistics but incorporating more variability, or utilize data from a
single instrument and single local electrode achieving poorer statistics, but also
minimizing potential variability. These comparisons are made in the results table.
For a fixed value of 1.45 and considering all nine datasets, the dose fluctuates
by 5.4 % which is reduced to 4.7 % when a variable approach is used. For a single
tool and local electrode, the deviations are smaller, 4.0 % and 2.2 % respectively.
158
Table 5.1 Arsenic dose values calculated for a constant of 1.45 as well as a variable
Measured dose Peak as
(1e14 cm2) 1.45 Measured dose (1e14 cm2) composition
f R (nm)
Analysis Instrument Lift-out Local electrode S-C F 0.5 Variable 0.5 (1e20 cm2)
1 41 A 1.16 99.5 6.95 NA NA 1.051
2 41 A 1.16 99.5 7.51 7.28 1.470 1.062
3 41 B 1.14 86 8.17 7.06 1.545 1.048
4 41 B 1.14 85.5 7.86 7.26 1.510 1.03
5 41 C 1.15 101.5 7.38 6.69 1.420 1.086
6 44 D 1.15 93 7.65 7.75 1.435 1.088
7 44 E 1.15 92.5 7.07 7.61 1.410 1.111
8 44 E 1.14 90 7.03 7.59 1.405 1.035
9 44 E 1.14 90 7.39 7.35 1.455 1.069
All analyses Average 7.45 7.32 1.46 1.064
Standard deviation 5.4 % 4.7 % 3.4 % 2.5 %
Instrument 44 Average 7.29 7.58 1.43 1.076
Standard deviation 4.0 % 2.2 % 1.6 % 3.0 %
5 Data Processing and Reconstruction
References 159
Noting that the counting statistics limit the ultimate expected fluctuations to ~1 %,
the reconstruction limitations here conspire to provide a similar level of uncertainty
under the best of conditions (limited acquisition and specimen variability).
The process of taking raw LEAP data and transforming it into 3D real space
coordinates appears deceptively simple in commercially available software like
IVAS but, as we have shown, to take full advantage of the impressive capabilities of
the resultant APT data one must be aware of the subtleties of the interrelated
parameters to maximize accuracy and precision of conclusions taken from the data.
References
1. Miller, M.K., Hetherington, M.G.: Local magnification effects in the atom probe. Surf. Sci.
246(13), 442449 (1991)
2. Sha, G., Cerezo, A.: Field ion microscopy and 3-D atom probe analysis of Al3Zr particles in
7050 Al alloy. Ultramicroscopy 102, 151159 (2005)
3. Larson, D.J., Prosa, T.J., Geiser, B.P., Egelhoff Jr., W.L.: Effect of analysis direction on the
measurement of interfacial mixing in thin metal layers with atom probe tomography.
6. Sebastian, J., Hellman, O., Seidman, D.: New method for the calibration of three-dimensional
atom-probe mass spectra. Rev. Sci. Instrum. 72(7), 29842988 (2001)
7. Kingham, D.R.: The post-ionization of field evaporated ions: a theoretical explanation of
multiple charge states. Surf. Sci. 116, 273301 (1982)
17, 114 (2011)
9. Ulfig, R.M., Kelly, T.F., Gault, B.: Promoting standards in quantitative atom probe tomogra-
phy analysis. Microsc. Microanal. 15(S2), 260261 (2009)
10. Blavette, D., Sarrau, J.M., Bostel, A., Gallot, J.: Direction et distance danalyse a la sonde
atomique. Rev. Phys. Appl. 17, 435440 (1982)
11. Bas, P., Bostel, A., Deconihout, B., Blavette, D.: A general protocol for the reconstruction of
3D atom probe data. Appl. Surf. Sci. 87/88, 298304 (1995)
12. Gault, B., de Geuser, F., Stephenson, L.T., Moody, M.P., Muddle, B.C., Ringer, S.P.: Estima-
tion of the reconstruction parameters for atom probe tomography. Microsc. Microanal. 14,
296305 (2008)
13. Gault, B., Moody, M.P., De Geuser, F., Tsafnat, G., La Fontaine, A., Stephenson, L.T., Haley,
D., Ringer, S.P.: Advances in the calibration of atom probe tomographic reconstruction.
J. Appl. Phys. 105, 034913/034911034919 (2009)
14. Geiser, B.P., Larson, D.J., Oltman, E., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F., Prosa, T.J.:
Wide-field-of-view atom probe reconstruction. Microsc. Microanal. 15(S2), 292293 (2009)
16. Gault, B., Haley, D., De Geuser, F., Moody, M.P., Marquis, E.A., Larson, D.J., Geiser, B.P.:
Advances in the reconstruction of atom probe tomography data. Ultramicroscopy 111(6),
448457 (2011)
17. Kelly, T., Larson, D.J.: Local electrode atom probes. Mat. Char. 44(12), 5985 (2000)
18. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.C.,
Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data from a
commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383 (2004)
19. Walck, S.D., Buyuklimanli, T., Hren, J.J.: Extended depth profiling with the IAP. J. Phys.
47-C2(3), 451458 (1986)
20. Gault, B., Moody, M.P., Marquis, E.A., De Geuser, F., Geiser, B.P., Larson, D.J., Kelly, T.F.,
Ringer, S.P., Smith, G.D.W.: Tomographic reconstruction in atom probe microscopy: past,
present. . . future? Microsc. Microanal. 15(S2), 1011 (2009)
21. Petersen, T.C., Ringer, S.P.: Electron tomography using a geometric surface-tangent algo-
rithm: application to atom probe specimen morphology. J. Appl. Phys. 105, 103518 (2009)
22. Gault, B., Haley, D., De Geuser, F., Moody, M.P., Marquis, E.A., Larson, D.J., Geiser, B.P.:
Advances in the reconstruction of atom probe tomography data. Ultramicroscopy 111,
448 (2010)
23. De Geuser, F., Lefebvre, W., Danoix, F., Vurpillot, F., Blavette, D., Forbord, B.: An improved
reconstruction procedure for the correction of local magnification effects in three-dimensional
atom-probe. Surf. Interface Anal. 39(23), 268272 (2007)
25. Larson, D.J., Geiser, B.P., Prosa, T.J., Ulfig, R., Kelly, T.F.: Non-tangential continuity
reconstruction in atom probe tomography data. Microsc. Microanal. 17(S2), 740741 (2011)
26. Larson, D.J., Geiser, B.P., Prosa, T.J., Kelly, T.F.: On the use of simulated field-evaporated
specimen apex shapes in atom probe tomography data reconstruction. Microsc. Microanal. 18
(5), 953963 (2012)
27. Vurpillot, F., Gault, B., Geiser, B.P., Larson, D.J.: Reconstructing atom probe data: a review.
(1961)
29. Loi, S.T., Gault, B., Ringer, S.P., Larson, D.J., Geiser, B.P.: Electrostatic simulations of a local
electrode atom probe: the dependence of tomographic reconstruction parameters on specimen
and microscope geometry. Ultramicroscopy 132, 107113 (2013)
30. Vurpillot, F., Da Costa, G.: unpublished
31. Prosa, T.J., Olson, D., Geiser, B., Larson, D.J., Henry, K., Steel, E.: Analysis of implanted
silicon dopant profiles. Ultramicroscopy 132, 179185 (2012). doi:10.1016/j.ultramic.2012.
10.005
32. Loberg, B., Norden, H.: Observations of the field-evaporation end form of tungsten. Arkiv.
For. Fysik. 39(25), 383395 (1968)
33. Wilkes, T.D., Smith, G.D.W., Smith, D.A.: On the quantitative analysis of field-ion micro-
graphs. Metallography 7, 403430 (1974)
34. Gerstl, S.S.A., Geiser, B.P., Kelly, T.F., Larson, D.J.: Evaluation of local radii of atom-probe-
tomography specimens. Microsc. Microanal. 15(S2), 248249 (2009)
35. Shariq, A., Mutas, S., Wedderhoff, K., Klein, C., Hortenbach, H., Teichert, S., Kucher, P.,
Gerstl, S.S.A.: Investigations of field-evaporated end forms in voltage- and laser-pulsed atom
probe tomography. Ultramicroscopy 109(5), 472479 (2009)
36. Vurpillot, F., Da Costa, G., Menand, A., Blavette, D.: Structural analyses in three-dimensional
atom probe: a Fourier transform approach. J. Microsc. 203(3), 295302 (2001)
37. Geiser, B.P., Kelly, T.F., Larson, D.J., Schneir, J., Roberts, J.P.: Spatial distribution maps for
atom probe tomography. Microsc. Microanal. 13, 437447 (2007)
38. Kelly, T.F., Voelkl, E., Geiser, B.P.: Practical determination of spatial resolution in atom probe
tomography. Microsc. Microanal. 15, 1213 (2009)
39. Gault, B., Moody, M.P., De Geuser, F., Haley, D., Stephenson, L.T., Ringer, S.P.: Origin of the
spatial resolution in atom probe microscopy. Appl. Phys. Lett. 95(3), 034103 (2009)
40. Gault, B., Moody, M.P., De Geuser, F., La Fontaine, A., Stephenson, L.T., Haley, D., Ringer,
S.P.: Spatial resolution in atom probe tomography. Microsc. Microanal. 16, 99110 (2010)
References 161
41. Muller, E.W., Tsong, T.T.: Field Ion Microscopy Principals and Applications. Elsevier, New
York (1969)
43. Gault, B., Muller, M., La Fontaine, A., Moody, M.P., Shariq, A., Cerezo, A., Ringer, S.P.,
Smith, G.D.W.: Influence of surface migration on the spatial resolution of pulsed laser atom
probe tomography. J. Appl. Phys. 108, 044904/044901044906 (2010)
44. Cerezo, A., Clifton, P.H., Gomberg, A., Smith, G.D.W.: Aspects of the performance of a
femtosecond laser-pulsed 3-dimensional atom probe. Ultramicroscopy 107(9), 720725
(2007)
4548 (1973)
46. Scheinfein, M.R., Seidman, D.N.: Time aberrations of uniform fields: an improved reflectron
mass spectrometer for an atom-probe field-ion microscope. Rev. Sci. Instrum. 64(11),
31263131 (1993)
50. Wada, M.: On the thermally activated field evaporation of surface atoms. Surf. Sci. 145,
451465 (1984)
52. Michaelson, H.B.: The work function of the elements and its periodicity. J. Appl. Phys. 48(11),
47294733 (1977)
53. Larson, D.J., Russel, K.F., Miller, M.K.: Effect of specimen aspect ratio on the reconstruction
of atom probe tomography data. Microsc. Microanal. 5, 930931 (1999)
54. Gipson, G.S., Yannitell, D.W., Eaton, H.C.: On the electric field distribution within the field
ion microscope and near the surface of field emitters. J. Phys. D. Appl. Phys. 12, 987996
(1979)
55. Gipson, G.S., Eaton, H.C.: The electric field distribution in the field ion microscope as a
function of specimen shank. J. Appl. Phys. 51(10), 55375539 (1980)
56. Gipson, G.S.: An improved empirical formula for the electric field near the surface of field
emitters. J. Appl. Phys. 51(7), 38843889 (1980)
57. Smith, R., Walls, J.M.: Ion trajectories in the field-ion microscope. J. Phys. D. Appl. Phys. 11,
409419 (1978)
58. Waugh, A.R., Boyes, E.D., Southon, M.J.: Investigations of field evaporation with a field-
desorption microscope. Surf. Sci. 61, 109142 (1976)
59. Oberdorfer, C., Schmitz, G.: On the field evaporation behavior of dielectric materials in three-
dimensional atom probe: a numeric simulation. Microsc. Microanal. 17, 1525 (2011)
60. Moore, A.J.W.: The structure of atomically smooth spherical surfaces. J. Phys. Chem. Solids
23(7), 907912 (1962)
61. Moore, A.J.W., Ranganathan, S.: The interpretation of field ion images. Phil. Mag. 16,
723737 (1967)
62. Geiser, B.P., Larson, D.J., Gerstl, S.S.A., Reinhard, D.A., Kelly, T.F., Prosa, T.J., Olson, J.D.:
A system for simulation of tip evolution under field evaporation. Microsc. Microanal. 15(S2),
302303 (2009)
evaporation of complex multilayer structures. J. Microsc. 241, 225233 (2010)
64. Larson, D.J., Gault, B., Geiser, B.P., Vurpillot, F.: Atom probe tomography spatial reconstruc-
tion: status and directions. Curr. Opin. Solid. State. Mater. Sci. 17(5), 236247 (2013)
65. Marquis, E.A., Vurpillot, F.: Chromatic aberrations in the field evaporation behaviour of small
precipitates. Microsc. Microanal. 14(6), 561570 (2008)
66. Birdseye, P.J., Smith, D.A., Smith, G.D.W.: Analogue investigations of electric field distribu-
tion and ion trajectories in the field ion microscope. J. Phys. D 7(12), 16421651 (1974)
67. Vurpillot, F., Gruber, M., Da Costa, G., Martin, I., Renaud, L., Bostel, A.: Pragmatic recon-
struction methods in atom probe tomography. Ultramicroscopy 111(8), 12861294 (2011)
68. Gault, B., Loi, S.H., Araullo-Peters, V., Stephenson, L.T., Moody, M.P., Shrestha, S.,
Marceau, R.K.W., Yao, L., Cairney, J.M., Ringer, S.P.: Dynamic reconstruction for atom
probe tomography. Ultramicroscopy 111(11), 16191624 (2011)
69. Vurpillot, F., Bostel, A., Blavette, D.: The shape of field emitters and the ion trajectories in
three-dimensional atom probes. J. Microsc. 196(3), 332336 (1999)
70. Vurpillot, F., Bostel, A., Blavette, D.: A new approach to the interpretation of atom probe field-
ion microscopy images. Ultramicroscopy 89, 137144 (2001)
71. Sauvage, X., Renaud, L., Deconihout, B., Blavette, D., Ping, D., Hono, K.: Solid state
amorphization in cold drawn Cu/Nb wires. Acta Mater. 49, 389394 (2001)
72. Vurpillot, F., Larson, D.J., Cerezo, A.: Improvement of multilayer analyses with a three-
dimensional atom probe. Surf. Interface Anal. 36, 552558 (2004)
73. Lauhon, L.J., Praneet, A., Ronsheim, P., Flaitz, P.L., Lawrence, D.: Atom-probe tomography
of semiconductor materials and device structures. MRS Bull. 34, 738743 (2009)
74. Simons, D., Kim, K., Benbalagh, R., Bennett, J., Chew, A., Gehre, D., Hasegawa, T., Hitzman,
C., Ko, J., Lindstrom, R., MacDonald, B., Magee, C., Montgomery, N., Peres, P., Ronsheim,
P., Yoshikawa, S., Schuhmacher, M., Stockwell, W., Sykes, D., Tomita, A., Toujou, F., Won,
J.: Round-robin study of arsenic implant dose measurement in silicon by SIMS. Appl. Surf.
Sci. 252(19), 72327235 (2006). doi:10.1016/j.apsusc.2006.02.152
75. Simons, D.S., Downing, R.G., Lamaze, G.P., Lindstrom, R.M., Greenberg, R.R., Paul, R.L.,
Schiller, S.B., Guthrie, W.F.: Development of certified reference materials of ion-implanted
dopants in silicon for calibration of secondary ion mass spectrometers. J. Vac. Sci. Technol. B
25, 13651375 (2007)
76. Depth profile data provided by Evans Analytical, Santa Clara, CA 95051
Chapter 6
Selected Analysis Topics
6.1 Introduction
LEAP data analysis is performed using the CAMECA Integrated Visualization and
Analysis Software (IVAS) package. IVAS was introduced in 2004 on a Windows
XP software platform based on Java and Netbeans and continues to require a
Windows operating system environment. CAMECA offers several levels of licens-
ing from free and short term use to permanent full versions. Through the years,
IVAS development has closely followed a guiding principle that prioritizes ease of
use and availability of open file formats so that analysis can be performed both with
IVAS as well as user-developed software. Discussing all the features and concepts
pertaining to IVAS would require a separate book (see the IVAS User Guide).
Consequently, this chapter describes only some key IVAS topics and features
useful for atom probe tomography (APT) practitioners. The features and topics
addressed are in no particular order, but each should have some relevance for
experienced and novice analysts alike.
The first topic considers spectral ranging. Because APT is fundamentally a time-
of-flight (TOF) mass spectrometry technique, ranging and identification of chem-
ical species are important for 3D reconstruction of detected ions (see Chap. 5), and
especially important for chemical quantification. Mass spectra from various mate-
rials are considered as well as ranging criteria. The impact on compositional
precision and accuracy is a key component of the discussion. These concepts are
applied to an example mass spectrum and the resulting composition, compositional
uncertainty, detection and quantification limitations of the data are highlighted.
The next topic considers key aspects related to the organization of the 3D data.
The quantity (many millions of ion positions) and 3D nature of the APT informa-
tion often requires some form of gridding or voxelization and delocalization to
enable reasonable calculation times and enable various forms of visualization
beyond simple atom maps. Understanding how these gridding and smoothing
choices affect extracted compositional trends is necessary for understanding the
accuracy of any analysis.
164 6 Selected Analysis Topics
The next topic considers identification of clusters and small precipitates.

Since identification and analysis of clusters is historically a key strength of the
APT technique, the implemented nearest-neighbor algorithm in IVAS is described
and discussed.
Finally, the spatial distribution map (SDM) is defined and the various SDM
visualization tools are discussed. One use of the SDM is to estimate detection
efficiency. An example of this SDM application is included.
6.2 Spectral Analysis
Determining accurate composition is a key goal of any APT experiment, so the

process of converting a mass spectrum into an atomic composition needs to take
into consideration the effects of data processing choices. The basic statistical
concepts of accuracy (or how representative the measured value is of the actual
value being measured) and precision (or what is the spread of multiple
measurementsrepeatability) are critical components of the discussion [1].
6.2.1 Ranging
An APT mass spectrum is composed of a series of sharp features, or signal peaks,

resting on a broad decaying count level (or background). The signal peaks represent
the pulse-coincident (in-time) field evaporated ions from the specimen surface,
while for this discussion, the background results primarily from non-pulse origi-
nated (out-of-time) events. Without going into a detailed discussion about the true
quantitative content within these features (see Chap. 4), we will treat the mass
spectrum as a simple analytical exercise: the total ranged signal in a peak (T )
contains a combination of ion signal (S) from the specimen and background (B),
Fig. 6.1. To calculate compositions, the background must be removed from each
peak to yield the signal (S T B).
In IVAS, a mass spectrum is realized as a histogram with a user-defined,
constant bin width. Typical bin widths range from 0.001 to 0.1 Da. Depending on
whether the user is looking to optimize the histogram for subsequent analytical
accuracy or to more clearly display particular features in the mass spectrum,
different binning choices might be made. The atomic identities of the various
mass peaks are defined by the user via mass ranging (the gray region labeled T in
Fig. 6.1 for example). Based on the particular peak shape, the user is required to
select minimum and maximum m/n values to capture the events, T, within the peak.
For general analysis, all ions falling between these values are considered to have a
single chemical identity (single or polyatomic ions) as defined by the user. For bulk
analysis, options exist for decomposing a range into multiple chemical identities
(see Sect. 6.2.3.4). Considering this, two questions seem immediately apparent:
6.2 Spectral Analysis 165
Fig. 6.1 A generic

description of a peak, T,
containing a combination of
signal, S, and background,
B. Two ranges are defined
by the gray shaded areas.
The central range is used to
determine the total number
of counts in T while the
other is used to estimate
B present in T. Background
estimates are also made by
IVAS (red line)
(1) What is an appropriate histogram bin width? and (2) How should the range
boundaries be chosen if one desires both compositional accuracy and precision?
The width of the histogram binning determines the number of bins over which
T is spread. Coarse binning has the effect of smoothing features and helps smooth
background fluctuations to better define small amplitude signal peak. Overly fine
binning can make it difficult to identify a weak signal on a relatively high back-
ground as a signal peak. In terms of APT quantification, the goal is to choose a
histogram bin width such that measurement of peak location and T are insensitive to
the bin boundaries. A general rule of thumb is for the full width at half maximum
(FWHM) of all peaks to extend over 5 bins [2]. The peak in Fig. 6.1 follows this
rule with bins widths of 0.005 Da yielding 12 bins across the FWHM.
Range boundaries determine the fraction of the signal, S, that is measured. An
accurate measure of S requires sufficiently large ranges to ensure that ~100 % has
been captured within the range; however, large ranges also increase the
B component which, if extreme enough, can lower the overall precision of the
result (increases the spread in multiple measurements). Generally, for high intensity
non-overlapping peaks, it is relatively easy to capture all of the peak intensity with
optimum precision. For low intensity peaks (and overlapping peaks), overly narrow
or wide ranges can result in systematic loss in accuracy or precision respectively.
Ultimately, a good combination of accuracy and precision can be achieved by
capturing uniform signal fraction from each peak and also sufficient signal fraction
such that a small change in range width does not significantly change the measured
fraction.
LEAP mass spectra generally possess asymmetric (non-Gaussian-shaped)
mass spectrum peaks, but some insights can be gained from considering analytic
peak shapes. For a Gaussian peak shape, describes the location of the peak
Table 6.1 Integral of the Gaussian probability density distribution

Confidence level 1-Tailed k-value 2-Tailed k-value
(integral total) ( k to 1) 1-Tailed k 2 ( k to + k)
68.3 % 1
95 % 1.645 2.33 1.96
99 % 3.29 4.65 2.58
99.9 % 3.49 4.94 3.30
center and is related to the FWHM (FWHM 2.354) and describes the peak
width [2]. The integrated peak fractions or confidence levels (CL listed in %)
contained within a region related to the peak center are defined by a numerical
factor k and listed in Table 6.1. Commonly quoted 2-tailed regions of , 2,
3 capture 0.683, 0.954, and 0.997 of the total peak intensity, respectively. By
capturing >99 % of the total signal and requiring that that FWHM spans more than
5 bins, one can generally achieve the goal of capturing a uniform and sufficient
fraction of the signal.
In terms of application of the above discussion to real APT peak shapes, there is
significant added complexity. Generally, APT peaks have steep rise on the low m/n
side and an extended tail on the high m/n side. Real APT spectra may have peaks
with differing peak shapes, shapes that broaden as a function of m/n, significant
signal fraction contained within the peak tail that extends into neighboring peaks, or
other features that make it difficult to assign ranges that achieve the goal of uniform
and sufficient signal fraction. Unfortunately, application of custom range sizes may
be necessary in these cases, and again a good criterion is uniform and sufficient
signal fraction to maintain good accuracy and precision.
6.2.2 Practical Considerations for Detection Levels
There are some basic questions one asks when considering detection of dilute
species and almost all of the answers to these questions depend on the background
level: (1) When is a net signal statistically significant? (2) When a net signal is not
statistically significant, then what is the maximum level that could exist without
detection? and (3) What signal level is necessary to achieve some minimal level of
measurement precision? For a more detailed statistical discussion of these questions
see [3].
Data collected from an APT experiment can be described as a Bernoulli
sequence and so follows a Binomial distribution [4]. For any viable LEAP mea-
surement, the Binomial distribution is in the limit where it can be approximated by a
Poisson distribution for weak peaks or a particular Gaussian distribution for strong
peaks [1, 5, 6]. For either case, this presents at least two options for estimating the
expected standard deviation, i, for many measurements estimated from a single
measurement range containing Ti counts. The first is useful when performing

calculations where background correction is necessary
p
i Ti: (6.1)
because each range i has a i that can be properly propagated when performing
general calculations [1].
The second is useful for calculating concentrations where background contribu-
tions are negligible (Ci Ti/TTotal) [4, 7]
s
Ci 1 Ci
i (6.2)
T Total
where Ti represents the total number of counts for an atomic species i and TTotal is
the total number of counts for all the species. For accurate measurement of dilute
species, it is important to correct the measured signal for background effects (e.g.,
ions from adjacent peaks, out-of-time ion events, etc.). General approaches to these
corrections are illustrated in [8]. For APT data, approximating the background
signal by making use of adjacent range(s) with the same range width is useful,
although there are other approaches as well. With this approach in mind consider
the questions in the following sections.
6.2.3 When Is the Signal Level Statistically

Significant (Critical Level) for a Peak?
This is achieved when the background fluctuations cannot account for the entire
measured signal to some desired CL. For the discussions that follow, the total
number of counts in any given range is composed of the signal plus the background
(T S + B). So consider the number of counts measured for a range defining a
peak as (T ) and the adjacent range(s) as a measurement estimating B. The
corresponding standard deviations for the counts in the peak and the background are
p p
T T and B B: (6.3)
The estimated net sample signal then becomes

p p
S T B and S T B or S 2B: (6.4)
Consider the situation when there is no signal or the limit as S ! 0 so

p
S ! 2B. Any net values greater than k S would be considered as a statistically
significant signal at the CL chosen, where k captures the chosen confidence
p
interval (see Table 6.1). For example, any net signal > 2:33 B would be
statistically significant at a 95 % CL. Such measurements should be reported

with the appropriate significance range described. Any values below that crite-
rion result in no detection and in such cases the critical level should be reported.
In addition, the minimum detection level could be reported as well.
6.2.3.1 What Is the Minimum Detection Level?
This is achieved when the low-side tail of the statistical fluctuations in a normally
detected signal exceeds a critical level based on the background measurement. The
result again depends on the CL chosen (for a complete derivation see [3]). If the CL
for the critical level and the detection level are the same, then the minimum
detection level becomes
p p
k2 2k 2B or 2:71 4:65 B for a 95% CL: (6.5)
In other words, for the case where no signal was detected given the desired CL
(here 95 %), the actual signal level could be as high as this and not have been
detected at least 1-CL (here 5 %) of the time because of statistical fluctuations.
6.2.3.2 What Signal Level Is Necessary to Achieve

a Minimal Level of Measurement Precision?
This is achieved when the estimated error of S, S, meets the criteria specified.
Summarizing from [3] the minimum number of signal counts must exceed the
values below for the relative precision in the net signal, k S/S, to reach 10 %
(kQ 1/0.1) minimum:
q
kkQ 2 kkQ 4 8BkkQ 2
S> (6.6)
2
p p
S > 200 + 40; 000 800B ! S 14.1k B for k S/S < 10 % for a 95 % CL.
6.2.3.3 Example
Consider the arsenic implant case study from Chap. 5. Assessing the background
levels in the tail of the arsenic implant can help guide decisions for generating 1D
composition profiles to meet analysis goals. Fine resolution of the implant profile
(z on the order of 0.1 nm resolution along the depth-dimension of the composition
profile) is possible when arsenic concentration is sufficiently high, but more coarse
resolution (z ~ 110 nm) may be necessary to allow quantification as concentra-
tions become more dilute. Assessing the background level allows one to determine
Fig. 6.2 Relevant Si (light gray) and As (dark gray) mass ranges and their equivalent background
ranges used to illustrate best methods for determining composition and estimates of uncertainty
Table 6.2 Summary of Criteria: 3 FW0.01M (~99 % Peak fraction)

ranging and peak counts
from Fig. 6.2 Identity Peak location Range size (Da) T Ba
28 2+
Si 14 0.369 514,890 147
29
Si2+ 15 0.382 30,167
30 2+
Si 16 0.394 18,065
28 +
Si 28 0.522 12,134 180
29 +
Si 29 0.531 816
30 +
Si 30 0.540 480
75
As2+ 37.5 0.604 285 52
75
As+ 75 0.854 297 46
a
Total for three peaks
the volume (and z) necessary to achieve some quantification level at desired

minimum concentration level.
A mass spectrum has been extracted from the example in Chap. 5 using a
cylindrically shaped subvolume located approximately 135 nm below the oxide
surface (full field of view and 5 nm thick in z). This volume includes 609,762 ions
of which 577,134 are ranged as Si or As. The primary signal peaks include Si2+, Si+,
As2+, and As+ peak shown in Fig. 6.2ad. These peaks have very similar shapes
when viewed as a TOF spectrum, so the mass spectrum peak widths broaden
p
proportional to ~1= m=n. Range widths defined by three times the full width
p
at hundredth maximum (3 FW0.01M) for the 28Si2+ and scaled to 1= m=n for
the remaining peaks satisfy the goal of uniform and sufficient signal fraction. The
range widths and resulting peak totals are listed in Table 6.2 with the resulting
compositions and estimated precisions listed in Table 6.3.
Multiple options for background correction exist. In IVAS, a background is
estimated based on the baseline of the TOF spectrum and is shown as the red line in
Fig. 6.1. An alternative is to manually range a region devoid of signal to approx-
imate the background relevant for nearby peaks. The easiest implementations are to
either make a background range of the same size as the nearby peak T to estimate
the background B, or make two ranges of exactly half the width and placed
Table 6.3 Composition and propagation of errors as compared to simulation

S Composition S Simulated Simulated
(fraction) (fraction) Composition S
Si 576,552 327 576,225 760 0.999161 0.001317 0.999175 0.000046
As 582 98 484 26 0.000839 0.000045 0.000825 0.000046
Total 577,134 Binomial (Eq. 2) 0.000038
Table 6.4 Statistical Counts

threshold levels for As
measurements based Background level 98
on background level Critical level (95 % CL) 23
Detection level (95 % CL) 49
Quantification level (10 % at 95 % CL) 544
symmetrically on either side of peak T. The latter approach has the advantage of
estimating any linear variation in the background level. The former approach is
shown in Fig. 6.1.
For this example, backgrounds for each peak were estimated by use of an
equivalent sized range at the low m/n side of the peak as indicated in Fig. 6.2.
Because the silicon isotope peaks are somewhat closely spaced, the Si2+ and Si+
ranges utilize a single background range of width equal to the sum of the three
individual isotope range widths. The number of counts in each peak (T) and the
number of counts in each background range (B) are listed in Table 6.2.
In order to determine the background corrected composition, the uncertainty in
the number of counts in each range must be estimated using (6.1) and these errors
propagated to yield uncertainties for the reported concentration. Because the
background levels in this example are relatively low, the final results and uncer-
tainties should be similar to those predicted by (6.2) without background subtrac-
tion. It is important to note, however, that as the arsenic signal decreases and the
background level remains relatively constant, the calculated uncertainties with and
without background subtraction will diverge.
The relevant steps for determining the composition and estimate of uncertainty
for this binary system are summarized in Table 6.4. First, the net number of Si ions
is determined by pooling all the counts from the Si ranges together (576,552) and
subtracting from it all the counts from the pooled Si background ranges (327). The
uncertainty in this value becomes the square-root of the sum of the uncertainties
p
of the individual uncertainties added in quadrature ( 576, 552 327 760 ).
The same steps were performed for the As ions. The composition becomes CSi
TSi/TSi+As and the uncertainty is estimated by Si/TSi+As.
There is an apparent contradiction here in terms of the uncertainty estimates.
Because this is a binary composition, the uncertainties of the two compositions are
systematically linked and should be equal because any change in the composition of
one requires an equal change in the other. Using (6.2), which does not allow us to
consider the effects of background subtraction, the resulting uncertainty is close to
that of the As fraction estimate using (6.1). The apparent discrepancy lies in the fact
that there are two ways to estimate the uncertainty. The table illustrates using
the counts of each ion type and the total ions. An alternative would be to use the
As result to determine the Si result (CSi 1 CAs) which results in an equal
uncertainty for both numbers. The best estimate of uncertainty requires the practi-
tioner to use a relationship that results in minimum uncertainty. For a system with
N chemical species, the N 1 species with minimal fractional uncertainty should
the Nth X
be calculated as in the table, while species should
iN1
be determined as the
balance from the N 1 species CN 1 i1
Ci .
The claims of the preceding paragraph can be verified via simulation. 100,000
randomly generated compositions were generated based on a 577,559 ion dataset
and average compositions for SiS+B, SiB, AsS+B, and AsB based on data from
Table 6.2. Simulated background corrected compositions of Si and As were deter-
mined by subtracting the simulated background counts for each ion type for each
simulation. Average Si and As compositions (Simulated Composition) and their
standard deviations (Simulated S) are reported in Table 6.3. There is excellent
agreement with the estimated uncertainty of the As listed in the table consistent
with the preceding discussion.
Estimating the background levels for the As ions allows one to consider how 1D
composition bin sizes impact the critical level, minimum detection level and
quantification levels for an analysis. Based on the background level measured
from the preceding example (98 counts), the three statistical thresholds previously
discussed have been calculated and listed in Table 6.4: (1) The net As counts must
be greater than 23 for an As signal to be detected at a 95 % CL. (2) When a signal is
not considered as detected, then an As level as high as 49 counts might be present in
the sample and not reach the detection threshold at least 5 % of the time. (3) The net
As counts must be greater than 544 to have 95 % CL error bars smaller than 10 % of
the reported value for this sample size.
It is important to emphasize that the preceding discussions on measurement
uncertainties are solely based on the counting statistics contained within a given
mass spectrum. They cannot account for other sources of error (e.g., systematic
measurement errors from multi-hit/pile-up effects, complex ion disassociation,
peak overlap issues, poor detector calibration, etc.).
IVAS has an ROI Simulation Tool that can be used to investigate quantification
outcomes based on variations in ranging, peak shape, and background level.
Consider a high-quality mass spectrum, like the Asimplaned Si example above,
but with some unknown level of As to be analyzed on a LEAP 4000HR. For the
same material system, the HR may provide a typical MRP of 1,100 FWHM and
lower background levels to go with lower detection efficiency. One may be
interested in the As detection limits. To address this issue, mass spectra were
simulated with various levels of background and three levels of As concentration
(atoms/cm3). The As was distributed at a ratio of 1.5:1 in the As2+ and As+ charge
states, respectively, and the background levels for each range were estimated
based on equivalent sized ranges as in the previous example. The results of the
simulations were minimum detection levels as plotted in Fig. 6.3. The y-axis is the
Fig. 6.3 Results for minimum detection level based on simulated LEAP 4000 HR mass spectra.
The estimated analysis volume (plotted as the cube root of the analysis volume) necessary to
achieve detection levels of 1e18cm3, 1e19cm3, and 1e20cm3 are illustrated. These level
depend directly on the average background level in each spectrum
cube-root of the volume necessary to reach a minimum detection level (95 % CL)
given a background level in ppm/ns.
6.2.3.4 Peak Decomposition
There are a number of situations where the determination of composition in APT is

compromised due to the relative position of peaks in TOF. These have been
categorized for clarity with examples below [9]:
Category 0: wellseparated [10B+ and 28Si2+].
Category 1: close, but can be sufficiently separated for quantification in a LEAP
system [94Mo3+ and 63Cu2+].
Category 2: closely overlapping and requires MRP greater than a LEAP 4000X
can provide [14N+ and 28Si2+] at different charge states.
Category 3: overlapped exactly due to multi-charged molecular species [16O22+
and 16O+].
Category 4: closely overlapped and at the same charge state such that they
cannot currently be discriminated in a LEAP 4000X [14N2+ and 28Si+].
Category 5: Overlapped at the same change state beyond any expectation of
resolution currently or in the future [54Cr2+ and 54Fe2+].
In some cases, the relative natural abundances of the isotopic peaks can be used
to decompose closely or exactly overlapping peaks. Decomposition of peaks should
not be confused with decomposing molecular ions for a compositional calculation,
or deconvoluting a peak in the tail of another peak. Peak decomposition is a useful
technique to more accurately describe the chemical composition of a volume where
peaks overlap, but is subject to several constraints. If one or more of the following
constraints are violated, then there is an increased chance that results may be
skewed:
6.3 Concentration Space Analyses 173
1. All peaks in the spectrum should be identified [no unknowns].

2. Fewer than two single isotope elements involved [P, Al, Co, etc.].
3. Elements must be at natural abundance [no ion implanted or enriched isotopes].
4. Well-shaped peaks without overlapping tails.
5. Low rate of correlated evaporation and multiple ion evaporation events (pile-up).
In the case of just two elements and a single overlapping peak, 64Ni2+ and 96Mo3+
for example, one can simply utilize the information from the non-overlapping peaks
to infer the expected number of ions for each element type in the overlapping peak.
When three or more overlapped peaks are involved the calculation quickly becomes
more complicated involving a large number of unknowns and constraints. The
following mathematical concept can be applied to any mass spectrum of any region
of interest in IVAS:
f ij Ai ij Mj (6.7)
Ai: actual count of type i ion (still unknown), Mj: measured counts in each range j,
fij: isotopic abundance that isotope type i has in range j, ij: (unknown) fractions of
the ions in range j that are type i.
Peak decomposition is accomplished by calculating the unknown fractions ij
and the Ai by reformulating (6.7) as an optimization problem with linear constraints
(i.e., sum of the counts from all peaks before and after decomposition must be
conserved). For example, in an ironchromiumnickel system (doubly charged
chromium isotopes at 25, 26, 26.5, and 27 Da) the overlap at 27 Da is constrained
by ACr M25 26.5 + / 27 Cr M27 and / Cr + / Fe 1. These types of constraints
27 27
are continued for all overlaps with the additional constraints 0 / ji 1

(no overlaps can have negative or >100 % of the counts assigned to any single
isotope). Once the operation is complete, IVAS will report a great deal of informa-
tion describing what counts have been identified in each peak. It will also, if
requested, flag values that cannot be solved within statistical limits.
6.3 Concentration Space Analyses
6.3.1 Gridding, Voxels, and Delocalization
In order to investigate the 3D compositional nature of LEAP data, it is useful to

create a 3D grid in concentration space. In IVAS this grid is automatically created
whenever a POS file is opened and an estimate of the concentration at every grid
point is calculated (see Chap. 5 and Appendix A for a description of file types). Once
this grid is created it may be used to display the position of 3D surfaces as defined by
elemental concentration (isoconcentration or iso-density surfaces). Such surfaces
may then be used to make estimates of morphology of internal features, composition
Fig. 6.4 2D representation

of the Gaussian kernel used
to delocalize or distribute
ion content among multiple
voxels
of such features, and chemical intermixing between these features. Single elements
or combinations of elements may be used in the definition of the surface.
First, a concentration space grid is specified by the user by defining voxel (volume
pixel) size and delocalization. The user may individually set the x-, y-, and z-dimen-
sions of the voxels keeping in mind that very small voxel sizes retain a high resolution
in concentration space but suffer from high statistical noise (in the limit of small voxel
sizes, many of the voxels may have only a few or no ions at all). Larger voxel sizes
have the advantage of better counting statistics for the concentration calculation, but
result in degraded spatial resolution. A delocalization function is also applied to the
data following the concentration calculation, which distributes the contribution of
each ion over a range of voxels (an example 2D version is shown in Fig. 6.4).
Several of the spatial analyses performed by IVAS (isosurface, proxigram,
surface roughness, preview compositions) require that the acquired hit-data be
reformatted from an ionic point-list into a set of 3D voxels. The simplest way to
grid the data is to simply assign each ion entirely to the 3D voxel into which it falls.
This has the effect of relocating ions from their various locations within the voxel to
the centers of the voxel. If an ion is detected near the edges of a voxel, then it is
desirable to share it among the neighboring voxels according to a localized transfer
function to smooth or delocalize the count values in these grids. One 3D grid is
created per mass range. At least schematically, IVAS follows the procedure
described by Hellman [10].
Ideally, for any level of delocalization desired by the user, IVAS would perform
a delocalization by convolving the desired 3D kernel with the list of detected points.
It is frequently useful to use rather large kernels, so this procedure can be extremely
slow. However, grid-based convolutions can be much faster because of tools like
the Fast Fourier Transform (FFT).
Based on these two considerations, IVAS performs a two-step delocalization
process. The first step applies a mask-based convolution with an extremely localized
Gaussian kernel to move ions from the point-list into the grid. IVAS uses a kernel in
which only the tails beyond 3 of a Gaussian kernel are outside of a distance of 0.5
voxel lengths. This is only applied once during the initial creation of the grids.
Then a second delocalization is performed directly on the grid, also by convolving

with a Gaussian kernel, but this time using the much faster FFT. Because both
kernels are Gaussian we use the fact that delocalization widths add in quadrature.
The width of the second-pass kernel to be applied is computed by subtracting in
quadrature the width of the first-pass kernel from the users desired width.
6.3.1.1 Sequential Gaussian Delocalization Example
The delocalization is specified at the 3 distance. For a default 1 1 1 nm3 grid,

we specify a uniform first-pass delocalization distance 0.5 voxel ! 0.5 nm is
specified. So points are transferred into the grid using a first-pass kernel with width:
x, y, z (0.5/3, 0.5/3, 0.5/3) 0.1667 nm, 0.1667 nm, 0.1667 nm.
The default user requested total delocalization is a nonuniform 3 3 1.5 nm

at 3. So the desired final width of the effective total deconvolution kernel is
x, y, z 1,1,0.5 nm. Taking the first-pass into account, we apply a second-pass
kernel with
x, y, z (12 0.16672, 12 0.16672, 0.52 0.16672)1/2
0.9860 nm, 0.9860 nm, 0.4714 nm.
6.3.2 Interface Creation and Interfacial Roughness
Isoconcentration surfaces (or isosurfaces) are surfaces drawn within a 3D dataset

representing some threshold value [11]. The threshold can be calculated for single
ions, decomposed ions, or combinations of ions. Isosurfaces can also be created
based on ion or atom concentration, ion or atom counts, or a user-defined ratio of
ions or atoms. Isosurfaces are the first step in creating interfaces and generating
volume rendered displays.
The location of specific concentration values are interpolated from average
voxel values as illustrated in Fig. 6.5. Here the location of 60 % and 60.3 % are
shown as linearly interpolated 1D positions between voxels with average values of
59.9 % and 60.4 % respectively. In the more general 3D volume a marching cubes
algorithm will construct a set of triangles at various orientations to the grid to
approximate the actual isosurfaces.
Once an interface is created and selected, there are several properties or descrip-
tors of the interface that are immediately available to the user. A subset of these
properties is listed in Table 6.5. St, Sa, and Sq are measures of the roughness of the
selected interface, or more precisely, estimates of the difference between the points
describing the surface and a best fit plane. For truly nonplanar features, these three
values lose their practical meaning.
As is the case for measurements of feature size and concentration using
isosurfaces, the voxel and delocalization parameters will affect interfacial
Fig. 6.5 The process of interpolating specific isoconcentration points. Here the location of 60 %
and 60.3 % are shown as linearly interpolated positions between voxels with average values of
59.9 % and 60.4 % respectively
Table 6.5 Interfacial property descriptions

St The area peak-to-valley height; the vertical distance between the maximum height and
the maximum depth of an interface
Sa The average roughness; the average of the absolute values of the measured height
deviations from an ideal flat interface
Sq The root mean square (RMS) roughness; the RMS average of the measured height
deviations from the mean surface taken within the evaluation area
Area The surface area of the interface, in nm2
Volume The enclosed volume of the interface, in nm3. Displays a ? when the interface is not
closed
roughness. See [12] for an example of the effect of voxel size and delocalization on
the resultant roughness measurements of a single interface between two layers.
6.3.3 Effects of Delocalization on Planar Surfaces
The detriment to any smoothing is that some of the spatial information of the
original feature(s) is lost. This section examines the sensitivity of the grid delocal-
ization parameter on the capability to measure the composition and thickness of a
simple planar feature. As a test, consider a synthetic dataset where perfectly sharp
chemical interfaces exist on each side of a single thin film (denoted B in Fig. 6.6),
and all atoms are on a simple cubic lattice. The feature is characterized by a
thickness (t) and a concentration level ( p) and centered at z 0.
In this case of a Gaussian response characterized by delocalization sigma (), the
resulting concentration profile can simply shown to be the following:

p z 2t z 2t
Cz erf p erf p : (6.8)
2 2 2
The resulting profile is linear in generated feature concentration and in terms of

the Gaussian width; it depends only on sigma/feature-size. Note that when com-
paring to measurements made using IVAS, some discrepancy will be observed at
small delocalization values due to the binning of the data. Various levels of a
Fig. 6.6 ABA thin film

schematic example to
evaluation delocalization
effect on data fidelity
Fig. 6.7 B Layer peak

concentration as a function
of distance for various ratios
of delocalization sigma to
feature size. The 100 % pure
B Layer thickness was
originally 1 nm (Note that
IVAS users do not specify a
delocalization sigma, they
set a Delocalization which
is equal to 3)
delocalized 3D grid affect 1D concentration feature size and peak concentration

as shown in Fig. 6.7. Notice that as the delocalization value increases the layer
shape is broadened and the apparent peak concentration is reduced. By the time the
delocalization is the same as the generated feature-size (note for a 1 nm thick
feature this corresponds to an IVAS delocalization value of 3 nm) the apparent
measured composition is ~40 % of the original value. The user needs to keep this in
mind when considering the use of isoconcentration surfaces and large delocaliza-
tion values to visualize small features of interest. IVAS calculates a 1D concentra-
tion with the original discrete data to avoid these effects.
The apparent size of the feature, as represented by the delocalized surface, is a
function of the delocalization sigma and the concentration value chosen to define
the isoconcentration surface (see further description of the creation of
isoconcentration surfaces below). Surface values near the peak concentration, for
any given delocalization level, results in the feature size being underestimated.
Very low concentration surfaces result in the feature width being overestimated.
Figure 6.8 compares measured feature-size as a function of delocalization sigma and
selected isoconcentration surface values and demonstrates that an appropriate
Fig. 6.8 Measured size of

B layer in Fig. 6.6 as a
function of delocalization
sigma and selected
composition isosurface
isosurface value should be near the midpoint of the difference in concentration

for the different phases, in this case (a pure B feature embedded in pure A) at a
50 % concentration value. Increasing the level delocalization can be thought of as
increasing the statistical power of the isoconcentration surface calculation (larger
volumes are used to achieve more precise composition values at each point) at
the expense of absolute accuracy in the localization (any fine scale fluctuations
are averaged out). The curves are values predicted by our Gaussian expression, the
overlaid points are measured results produced by IVAS on the synthetic example
data. The right level of delocalization to use depends on whether your application
calls for lowered noise in the surface shape or in accuracy of the spatial extent [10].
6.3.4 The Proximity Histogram
The proximity histogram (proxigram) [13] bears some similarity to a 1D concen-

tration profile, but rather than calculating a profile along an analysis axis of a region
of interest, the proxigram calculates a profile with respect to distance from an
interface. A proxigram can be computed for one interface or multiple interfaces.
There are several benefits to the calculation of a proxigram. When creating a 1D
concentration profile across an interface from one region of a dataset to another, the
size of the profile should be such that significant interface curvature is not contained
within the profile. If this is the not the case, then a concentration profile which
artificially widens the interfacial region will result. Thus, concentration profiles are
often created from regions of small cross-sectional area (normal to the profile
direction) and statistical variations may result from low ion counts. In the
Fig. 6.9 (a) Data showing two phases in a nickel-based alloy with the interfaces created at the
10 at.% aluminum level, (b) proxigram created using only the upper interface shown in (a), and (c)
proxigram created using all of the interfaces shown in (a)
proxigram calculation, each ion is placed into the distance bin which is the smallest
distance away from any point on the interface, thus effectively taking into account
any curvature in the interface while still maintaining good counting statistics.
An estimate of the standard deviation is calculated on each bin of the histogram,
and as discussed previously, the counting statistics are not the only source of error.
Proximity histograms are a very powerful tool for detecting variations in concen-
trations near buried interfaces, something atom probe tomography is uniquely
suited to do.
We will use the dataset shown in Fig. 6.9 as an example of a simple proxigram
analysis. A dataset containing two phases in a nickel-based alloy is shown in
Fig. 6.9a. A 10 at.% aluminum isoconcentration surface has been used to define
the interfaces shown in the figure. Creation of a proxigram with 0.1 nm bins and a
maximum distance of 5 nm from the interface arrowed in Fig. 6.9a is shown in
Fig. 6.9b. The position of 0 on the x-axis of the proxigram corresponds to the
interface from which the profile was calculated (10 at.% Al), although it may be
slightly offset in some cases depending on the bin size selected for the proxigram
because the surface concentration is calculated from the 3D grid and the proxigram
is based on the discrete data. Negative x-axis values correspond to lower concen-
tration side of the isosurface. From Fig. 6.9b, the user is immediately able
to determine the partitioning of the elements Ni, Al, Mo, Co, and Cr between
the phases and to observe that no obvious segregation is apparent at this interface.
All six of the interfaces visible in Fig. 6.9a may be used to create a single
proxigram, Fig. 6.9c, which shows features very similar to those in Fig. 6.9b, but
with less statistical variation. In the case of a large number of secondary phase
regions of different sizes, the user may wish to consider separate proxigrams based
on the size of the phase (using either volume or surface area) to look for evidence of
size affecting observed composition.
6.4 Solute Analysis: Cluster Detection Method
6.4.1 Description of the Technique
Investigation of the proximity relationship between solute atoms is often used as a

method for defining which atoms belong to a single cluster. This is known as the
friends-of-friends method [1422] and is based on the principle that, within a
cluster, the spacing between solute atoms will be smaller than it is in the matrix.
Any two solute atoms which are nearer than a specified distance, d, are determined
to be in the same cluster (see [23] for a recent review on this topic). Note that some
algorithms [24] use a requirement of more than one solute atom in close proximity
to another as a requirement for defining the cluster. This is often termed the Order
of the method. The variables typically involved in the cluster detection process are
Dmax, Nmin, L, and E. We will define these variables as we proceed through an
example of how they are applied in the cluster detection methodology, Fig. 6.10.
A schematic diagram of solute atoms (red) in a matrix is shown in Fig. 6.10a.
The distribution is clearly nonrandom and we would like to estimate which of the
solute (and matrix) atoms belong to a cluster. Based on a simple visual inspection,
the dashed black line shows our initial guess at which atoms form the cluster.
In Fig. 6.10b Dmax is illustrated as the radius of the green circles. This circle
(actually a sphere in 3D) is positioned onto each solute atom (blue atoms in
Fig. 6.10b) and its interior is examined for any other solute atoms. If any other
solute atom is found (assuming that we are using an Order 1 analysis) within the
circle then the solute atom at the center is determined to be part of the potential
cluster. For the green circle on the left we have found three other solute atoms and
so this atom is determined to be part of the cluster, while for the green circle on the
right no solute atoms are found. Proceeding with this process for every solute atom
leads us to Fig. 6.10c, in which the ten solute atoms determined to be part of the
cluster are shown in blue.
Not all clusters are pure solute however, and we also wish to determine if there
exist some matrix atoms which should be denoted as being contained in the cluster.
This is done in a similar fashion as above using the variable L, which is sometimes
referred to as the Envelope Parameter. It is often set to a value equal to Dmax, which
is shown as the gray circle in Fig. 6.10c. This circle is centered on all ten of the solute
atoms so far determined to be in the cluster in order to determine all potential matrix
6.4 Solute Analysis: Cluster Detection Method 181
Fig. 6.10 Atom diagrams illustrating the cluster-finding algorithm: (a) solute atoms (red) in a
matrix (white) suggest the presence of a cluster (dashed line), (b) the Dmax parameter shown as a
green circle positioned above two solute atoms (black) of interest, (c) results of application of the
Dmax parameter (atoms shown in blue are determined to be part of the cluster) with the size of
the envelope parameter (L ) used to determine potential matrix atoms contained in the cluster
illustrated in as the gray circle, (d) shows all matrix atoms (gray) which may be evaluated to
determine if they are contained in the cluster, (e) the erosion distance parameter (E) (shown as a
purple circle) is applied to all non-cluster atoms to remove atoms around the cluster edges (purple
spheres), and (f) the final estimate of the atoms (6.14) contained in the cluster
atoms which may be in the cluster. For the gray circle shown in Fig. 6.10c, we
identify six potential matrix atoms which may be part of the cluster (gray atoms in
Fig. 6.10c). Figure 6.10d shows the result of using this process to determine all of the
potential matrix atoms (solid gray circles) which may be contained in the cluster. We
now have cluster atoms defined as shown in blue (matrix) and gray (solute).
It is clear however, that while this method allows us to identify some obvious
matrix atoms contained in the cluster (arrow in Fig. 6.10d), it has also added matrix
atoms (usually around the outside of the cluster) which we do not wish to identify as
being part of the cluster. Thus we need to remove these atoms and this is done using
another parameter, which we will call E (also known as the erosion distance). E is
often taken to be slightly smaller than Dmax, and is shown by the purple circles in
Fig. 6.10e. A circle of radius E is positioned over all of the atoms in the dataset
which have been determined to not be part of the potential cluster and are also
within at least a distance E from the cluster. Any cluster atom found within this circle
Fig. 6.11 A
20 20 2 nm3 slice
of data containing
approximately 3 at.%
solute (red atoms) which
appears to be nonrandomly
distributed
is removed from the potential cluster. The purple atoms shown in Fig. 6.10e are an
example of this erosion process. After it is completed, we have the final estimate of
the atoms contained in the cluster, Fig. 6.10f. We can see that our L process added
four matrix atoms (gray) to our initial determination of solute cluster atoms (blue).
The cluster contained 14 atoms and is approximately 71 % solute.
An additional variable often used to refine the cluster detection process is Nmin.
Clusters which contain a number of atoms less than Nmin are ignored because they
are assumed to be due to random fluctuations in the material [18]. Randomized
version of experimental data is often used to estimate random cluster distributions.
6.4.2 Example of Cluster Detection
We will now create a cluster count distribution analysis in order to determine an

appropriate Dmax value for a real dataset. Note that if Dmax is too low then larger
single clusters may be incorrectly identified as several small ones. If Dmax is too high
then separate clusters which are close together may be incorrectly identified as a
single large cluster. A 20 20 2 nm3 slice of data containing approximately
3 at.% solute which appears to be nonrandomly distributed is shown in Fig. 6.11. The
cluster count distribution is shown in Fig. 6.12 and compared to random data
generated using the estimated concentration of the material (97 at.% matrix 3 at.%
solute). In creating this distribution we have estimated Nmin at ten atoms. From
Fig. 6.12 we can choose a Dmax value of 0.5 nm (dashed black line) which will
remove the majority of detected clusters which are due simply to random fluctua-
tions of a 3 at.% solute alloy.
6.4 Solute Analysis: Cluster Detection Method 183
Fig. 6.12 Cluster count distribution for the data shown in Fig. 6.11 using parameters Order 1
and Nmin 10. The count distribution that results from randomizing all of the ions (using the
measured concentration of 3 at.% solute) is shown as the gray line
Fig. 6.13 Cluster size distribution using a Dmax value of 0.5 nm as determined from Fig. 6.12. The
size distribution that results from randomizing all of the ions (using the measured concentration
of 3 at.% solute) is shown as the gray line
We will now create a cluster size distribution analysis in order to determine an

appropriate Nmin value. Note that if Nmin is too low then random solute fluctuations
may be incorrectly identified as clusters (note that this probability increases with
the level of solute). If Nmin is too high then small clusters may be ignored. Using our
Dmax value of 0.5 nm we obtain the size distribution shown in Fig. 6.13. The dashed
black line shows a reasonable estimate for Nmin of eight atoms when compared to
randomized data.
If we now run the cluster detection algorithm with these parameters of Dmax and
Nmin, we obtain a distribution of 35 clusters having an average of 1112 solute
atoms per cluster with a mean volume of ~1 nm3 and a maximum concentration of
~35 at.% solute. For the slice of data shown in Fig. 6.11, the detected clusters are
shown in Fig. 6.14. Summing all of the clusters to create a radial concentration
profile, shown in Fig. 6.15, agrees well with the mean cluster concentration of
approximately 30 at.% solute.
Note that optimal detection parameters depend on the cluster size being
investigated and on the amount of solute in the clusters. Stephenson [24] has
Fig. 6.14 Results of the cluster detection algorithm with parameters Dmax 0.5 and Nmin 8
showing a distribution of 35 clusters having an average of 1112 solute atoms per cluster with a
mean volume of ~1 nm3 and a concentration of ~30 at.% solute. Note that this figure only shows a
slice of the original 20 20 20 nm3 dataset. Matrix atoms are shown in gray and solute atoms
are shown in black
Fig. 6.15 A summed concentration profile for all of 35 clusters showing a maximum average
cluster concentration of ~35 at.% solute. The scaling of the distance axis is performed by scaling
atom 3D displacements from the cluster center of mass to the 3D radius of gyration vector or to the
3D vector of maximum displacements of cluster atoms
6.5 Spatial Distribution Maps 185
demonstrated that a suitable value of Dmax is inversely proportional to the cube root
of the local density of solute atoms in the cluster. Hyde has recently simulated a
number of situations and made some recommendations with respect to the starting
point for the variables. He points out that (1) Dmax is the most important variable,
(2) Nmin is a function of Dmax and ion detection efficiency, and (3) both Dmax and
Nmin need to be increased as the ion detection efficiency decreases [25].
6.5 Spatial Distribution Maps
For crystalline materials, there is often significant order present in the atom
positions in APT reconstructions. This circumstance originates from the naturally
ordered progression over which atoms are removed from the curved atom probe
specimen surface. Evaporation is biased toward evaporation at terrace edges,
especially in pure materials. Consequently, atoms are removed from the terrace
edges where the electric field is highest (ignoring variations in surface evaporation
field from crystallographic orientation). After a single atom is removed from a step
edge, some subsequent step edge experiences the highest electric field and begins to
field evaporate which can be a very ordered process under certain conditions. Some
materials exhibit more ordered evaporation than others. Tungsten and aluminum
are two metals that are known to exhibit strongly ordered evaporation.
Ideally, reconstructions of the detected ions from all crystalline specimens
would result in perfect arrays of ion positions, but in reality, even the best recon-
structions have some inherent level of disorder. Some of the known sources of
imperfection include:
1. Out-of-sequence evaporation
2. Trajectory aberrations due to crystallography and local magnification effects
3. Over-simplified reconstruction algorithm
4. Nonspherical tip shape
5. Nonuniform reconstruction parameters (kF, , etc.)
When evaluating the quality of a reconstruction, it is desirable to validate the
periodicity of the reconstruction against known planar spacings within the speci-
men. Some quantitative measure of both short-range (nearest-neighbor arrange-
ments) and long-range order is necessary to allow assessment of the accuracy of
reconstructed image and any general limitations of the technique. The SDM [26]
was developed to enable validation of reconstructions for highly ordered systems as
well as look for types of chemical ordering with APT tomograms. The result of an
SDM may be applied to a variety of analyses including trajectory aberrations,
atomic ordering and clustering, element- or isotope-specific radial distribution
functions, quantum detection efficiency, and real-space crystal structure. A very
similar type of data analysis is known as the near atom vicinity method [27].
6.5.1 The SDM Defined and Important Properties
The SDM allows for visualization of nearest-neighbor correlations, measurement of

the number of nearest-neighbors, and quantification of longer-range correlations
within a 3D dataset. The SDM is an autocorrelation calculation of the reconstructed
ion positions minus the ion self-correlations. This type of function is closely related
to the Patterson function from x-ray diffraction [28]. A 3D map or histogram of all
vectors between all pairs of ions within the reconstruction is calculated ignoring the
self-correlation vectors of zero length that normally form the origin peak in the
Patterson function. Mathematically, the function takes the form
1

SDMX; Y; Z x; y; zx X, y Y, z Z dxdydz NX, 0 Y , 0 Z, 0
1
N X
X N
xi , xj X yi , yj Y zi , zj Z
j6i i
(6.9)
XN
where x; y; z i
x xi y yi z zi is composed of N Dirac delta
functions that represent the N ion positions. A useful property of the SDM is that the
SDM of a full dataset and the SDM of the same dataset where only a random fraction
of the dataset is used to calculate the SDM are directly related by :
Efficiency ! SDM x; y; z SDM100% x; y; z: (6.10)
6.5.2 Methods Similar to the SDM
Crystallographic structure in APT images has been investigated by a variety of

authors using Fourier transform based [2933] and autocorrelation based [27, 34]
methods. Fast Fourier Transforms (FFTs) require a very fine binning of individual
ion positions onto a regular grid and requires a large amount of computer memory
[30, 31] while direct Fourier calculations treat each ion position as a delta function
requiring extensive CPU use [31, 32]. In addition to these resource issues, analysis
becomes difficult away from positions of finest lateral resolution because the signal
in the intensity maxima becomes too weak. However, it has been reported that useful
information can be discovered by these methods in small, specially placed regions
of interest [31, 32]. In particular, they have shown the ability to detect multiple,
non-collinear reciprocal space vectors and their corresponding atomic planes, mea-
sure spatial resolution in an image, create noise-filtered bright-field and dark-field
images, and determine mean precipitate size, shape, and composition in a specimen.
In addition, other real space methods, as well as variations on the basic SDM
concept, have also been developed [27, 35, 36].
Fig. 6.16 A simple 1D

example of how spatial
distribution maps may be
understood by comparison
to autocorrelation of ions
represented by delta
functions: (a) delta
functions at A, (b) the
autocorrelation of (a),
(c) the SDM of (a)
6.5.3 Understanding Basic SDMs
For simplicity, consider the 1D example shown in Fig. 6.16. A pair of ions is
represented by a pair of delta functions at positions A on the x-axis (Fig. 6.16a).
Consider a copy of this same function (two delta functions at A) translated some
arbitrary distance x along the axis. The autocorrelation for some value x is the
integral of the product of these two functions (the original and the copy translated a
distance x). In this case, the result will be zero for all values of x except when
x 2A, 0, and 2A. When x 2A, the product is a single delta function squared
and the integral over all space is 1. When x 0, the product is a pair of delta
functions squared and the integral over all space is 2. As a result, the autocorrelation
of the two points in Fig. 6.16a consists of the three points shown in Fig. 6.16b
where the middle point has a value of 2 while the adjacent points have values
of 1. Calculation of the SDM follows the same process except that the translation
represented by x 0 (the self-correlation) is excluded, so the result is two points
each with a value of 1 as shown in Fig. 6.16c. Alternatively, one can think about
(or calculate) the SDM from a nearest-neighbor point-of-view. Since the point at
Fig. 6.17 A simple 2D example of how spatial distribution maps may be understood by compar-
ison to autocorrelation of ions represented by delta functions: (a) an arrangement of five delta
functions, (b) the autocorrelation of (a), (c) the SDM of (a)
x A has a single neighbor at +2A and the x A point has a single neighbor at
2A, the SDM has two histogram values of 1 at 2A respectively.
Next consider the 2D example shown in Fig. 6.17. Here we want to consider the
product of the five points as shown in Fig. 6.17a multiplied by all possible 2D
translations of a copy of that same function. For all possible translations of the copy
(except when the translation is zero), only a single point from the original and the
copy overlap resulting in a value of 1 for the autocorrelation for those 20 xy
translations. When the translation is zero, all five points overlap giving an integral
value of 5. So the autocorrelation plotted in Fig. 6.17b has 20 points each with a
value of 1 and a central point with a value of 5. The SDM, shown in Fig. 6.17c, is
the same pattern except that the central point is missing. Alternatively, one could
create the SDM by considering a histogram of all the neighbors for each of the 5 ion
sites and it would result in the same plot for the SDM as shown in Fig. 6.17c.
6.5.4 Calculating SDMs in IVAS
The SDM is a record of all inter-ionic position vectors in a dataset from zero to
some prescribed maximum vector length (Neighborhood Extent). For the ith ion in
an image at (xi,yi,zi), the relative position of the jth ion (xij, yij, zij) is recorded.
In principle, the relative position of every atom pair in the image can be recorded
(the infinite SDM). In practice, we have limited the information retrieved to a cell of
dimensions (xmax, ymax, zmax). Once the positions of all atoms in the cell
relative to the ith atom have been recorded, the process is repeated for the jth atom
and so on for all atoms in the image (or subset of the image). The result is a three-
dimensional histogram of all relative positions out to the dimensions of the cell.
This histogram maps the average near-neighbor environment in the image. Note
that there should be no data at (0,0,0) since there should be no atom pairs where
each atom occupies the same location. If we disregard edge effects for simplicity,
the SDM size grows as
N n ncell (6.11)
where N is the number of relative displacement vectors, n is the number of atoms in

an analyzed dataset and ncell is equal to
ncell V cell 8xmax ymax zmax (6.12)
where Vcell is the cell volume and is the number of atoms per unit volume in the
material. For a cubic cell, this becomes
ncell 8xmax 3 : (6.13)
Thus, the SDM size increase is linear in n and cubic in xmax. This growth rate of
the SDM data size is modest and is much lower than other common techniques that
grow as n2 or n ln n especially as n gets large. For example, the complexity of
Fourier techniques mentioned above grows as n ln n, where n grows as the cube of
the linear spatial extent of the image. This small data file size is one of the
advantages of the SDM over other types of analyses.
6.5.5 Visualizing Tungsten SDMs with IVAS
Once the SDM cell is fully populated, it is informative to set a region of interest
around the principal signal so that periodicities corresponds to the structure may be
more readily visualized. This is illustrated in Fig. 6.18, which shows A 3D plot from
a <110>-oriented tungsten dataset where an iso-intensity surface has been imposed
on the SDM such that 15 % of the signal is contained within the surface. This
process effectively removes the background signal for greater visibility. The
enclosed surfaces in Fig. 6.18 are the atomic separations represented in the cell
that have a high repeat frequency in the data. The structure of these atomic
separations corresponds directly to the real space lattice.
IVAS allows the user to perform SDM calculations both during the reconstruc-
tion phase (within the Cal/Recon Wizard as part of the Reconstruction Explorer)
and after reconstruction. The types of 1D and 2D plots are illustrated in Fig. 6.19.
The range of the SDM is defined over a volume defined by Neighborhood Extent
(in nm) along all three axes. Once the SDM has been calculated, the user then
chooses how to visualize the data. Projections of the entire 3D range are possible as
ZX, ZY, and Z projections. Alternatively, single planes within the SDM range can be
selected by utilizing Z-Cut and Z-Plane-Offset (Reconstruction Explorer only) and
projected onto the XY-plane or along the x- or y-axes. Note that when visualizing the
SDM within the Reconstruction Explorer, the SDM is sampled, while SDMs
generated after reconstruction (on the POS file) are not sampled.
Besides calculating SDMs that utilize all the ion events, it is possible to choose the
chemical identities of the center elements as well as the neighbor elements. This
allows one to investigate neighborhood relationships such as site-occupancy [27, 37].
Fig. 6.18 3D plot of the SDM data from a <110>-oriented tungsten dataset using a 15 %
iso-intensity surface has been imposed. The bounding box of the cell is a cube with xmax
ymax zmax 1 nm. (a) View at an angle to the {110} (x axis tilted up). (b) View along the
<100> which is 45 from the x and z directions. Reprinted from [26] with permission from
Cambridge University Press
Fig. 6.19 (a) A full 3D SDM is represented as an array of high intensity regions (red regions).
IVAS provides multiple ways of viewing this 3D volume. (b) A limited subvolume is first defined
by the user. This is equivalent to considering nearest-neighbors out to a maximum distance DNE.
(c) Users can select to view a 2D projection of this volume in the ZX or ZY plane, or as a 1D
projection along the Z axis. (d) Users can also select an individual Z plane of features using ZCUT
and displaying them as a 2D YX projection or a 1D projection along the X or Y axis
6.6 Application of Spatial Distribution Maps 191
6.6 Application of Spatial Distribution Maps
6.6.1 Finding the Crystal Lattice
Let us consider a straightforward application of the SDM as an example. The

tungsten image in Fig. 6.20 shows atomic planes ({110} in this case) normal to
the analysis direction. We should expect to find a signal from these planes in an
SDM. There are however, very significant aberrations at the zone axes and zone
lines as shown in Fig. 6.20b. These aberrations make it difficult to observe directly
the atomic planes at other orientations in the structure. Despite the aberrations,
Fig. 6.20 (a) Side view of a

thin slice through a 3D atom
map of tungsten where the
atomic planes are apparent.
(b) 2D cumulative field
evaporation histogram of
tungsten revealing the
presence of a <110> zone
axis and zone lines.
Reprinted from [26] with
permission from Cambridge
University Press
Fig. 6.21 Intensity

distribution along the z axis
in the SDM (shown in
Fig. 6.20). The <110>
planes of tungsten are
clearly resolved as the peaks
in this plot. The measured
spatial resolution in the
z direction is r 0.06 nm.
University Press
the interatomic spacing along this direction in the tungsten specimen is clearly
resolved by the SDM in Fig. 6.21. The <110> zone axis of the tungsten specimen is
visible near the center of Fig. 6.20b and is oriented close to the z-axis in
Fig. 6.21. In the data, a {110} plane normal is near parallel to the z-axis, so the
signal in the SDM can be integrated in xy to reveal the planes, Fig. 6.22. Each peak
is the cumulative signal from the <110> planes in Fig. 6.20. Thin slices normal to
z and centered on the peaks in Fig. 6.21 are used to isolate the strongest signal and
obtain the images shown in Fig. 6.22. The z-oriented slices are created with
zslice 0.01 nm. These slices clearly reveal the presence of the periodicity in
the <110> planes which is consistent with the body-centered cubic (BCC) real-
space crystal lattice.
6.6.2 Using SDMs to Calculate Efficiency
SDMs may also be useful in estimating system detection efficiency. We start by

briefly describing how to calibrate a reconstruction using information from SDMs.
For a given set of standard reconstruction parameters (tip radius, efficiency, ,
etc.), we will apply a set of post-reconstruction adjustments that improve the
agreement of the information contained within the tungsten SDM with the known
crystallography. We typically follow a procedure that starts with the XY SDM and
applies a shear and scale factors as required to make the closest 4 XY peaks match the
expected geometry. We then examine the Z SDM and adjust the spacing by applying
Fig. 6.22 2D slices through

data in Fig. 6.20 normal to
<110> showing spatial
periodicities evident:
(a) thin slice through the
origin, (b) thin slice
through the first atomic
layer above (a), and (c)
the second atomic layer
up above (a). Each slice is
(z)slice 0.01 nm thick.
University Press
Fig. 6.23 (a) The portion of the XY SDM to be used for SDM-based efficiency calibration. (b) The
z cut is loosened to include an entire atomic layer. (c) This results in the much lower-resolution XY
SDM that is integrated to estimate the efficiency
a third scale factor. It should be noted that the application of these post-reconstruction
scale factors does not inherently preserve the reconstruction density implied by the
specified atomic volumes. Depending on the quality of the dataset, we will typically
need to smooth and threshold the SDMs in order to process them this way.
Once a dataset has been reconstructed and calibrated we can begin the efficiency
estimation procedure. The idea is that since we have calibrated the reconstruction
we should be able to investigate the occupancy of the average atoms neighbor-
hood. Specifically, for each atom we see and treat as a center, we will look at the
SDM volume that ought to occupied by some number of neighbors and ask how
many were actually seen.
We have chosen to focus on a method that counts the observed fraction of atoms
found in the central 8 in-plane spots closest to the analyzed center atom. In a
correctly calibrated <110> W reconstruction the region occupied by these 8 spots
is not square. So to facilitate our integration of the SDM histogram we apply a 2
stretch to the reconstructed x positions before filling the SDM. Now, the central
eight SDM intensity maxima are easily counted within a square area as shown in
Fig. 6.23a. (This adjustment is not necessary if calibrating on <100> FCC poles for
example.) We then loosen the tight Z cuts generally applied to make the XY SDM so
that they include the full width of the z-plane, Fig. 6.23b. This obviously degrades
the signal-to-noise ratio (SNR) of the resulting SDM calculation but is necessary to
ensure we count all the appropriate atoms, Fig. 6.23c. The actual efficiency
calculation is simply the observed occupancy of the very loosely cut SDM. In
this example we are interested in the spatial region defined by the 8 closest in-plane
spots. So our efficiency estimate is:
N SDMCounts
8 : (6.14)
8N Centers
In principle, efficiency estimates could also be obtained by looking at the closest

4 in-plane spots or looking at the occupancy of spots in planes above and below the
center atom being analyzed.
We have found that the efficiency estimate obtained can strongly depend on the
projection settings used to perform the initial reconstruction. Simply scaling an
arbitrary reconstruction to correct the SDM spacings is not sufficient to specify
a unique efficiency estimate. We have found that even after calibration,
efficiency estimates decrease if the reconstruction radius is too large. A 20 % increase
in reconstruction radius can result in a factor of 2 decrease in estimated efficiency.
6.6.3 Ordered Structures and Site Occupancy
SDMs have multiple applications to ordered alloys. They may be used to demon-
strate that long range order is present in a dataset, investigate whether a particle or
phase is coherent with its matrix, and estimate site occupancy of solute atoms on a
sub-lattice. Examples of these three applications will be presented in this section.
6.6.4 Ordering in Al3Sc Precipitates
The high-temperature mechanical properties of dilute Al-base alloys are promising

and atom-probe methods have historically provided high-resolution microstructural
information on materials based on aluminum. In AlSc based alloys, a coarse-
grained matrix forms upon homogenization at high homologous temperatures in the
single-phase -Al region. Upon age-hardening, a high number-density of nano-
scale, coherent Al3Sc precipitates with the ordered L12 structure form and the
resultant alloy displays a high creep resistance [38].
Figure 6.24 shows an example of the SDM applied to the L12 ordered structure of
Al3Sc [39]. A selected region of LEAP data is shown in Fig. 6.24a. where the <100>
direction is nearly parallel to the z direction of the analysis. A 1 at.% Sc
isoconcentration surface has been used in order to extract this region of the data from
the overall dataset. Z-SDMs of the AlAl atoms and the ScSc atoms within the Al3Sc
precipitate are shown in Fig. 6.24b and c, respectively. The dashed lines show that the
spacing of the ScSc atoms is twice the spacing of the AlAl atoms, which matches
the expected crystal structure. In addition, the Al variations along {200} planes are
as expected with every other {200} plane having a higher Al content, Fig. 6.24d.
6.6.5 FeCr Precipitates in a NiAlFeCr Alloy
Many heat resistant ferritic steels have superior thermal expansion and conductivity
properties compared to Ni-based alloys but are somewhat limited in their applica-
tions above ~600 C by their mechanical properties. Efforts are being made to use
precipitation strengthening to improve these properties and thus the usefulness of
such alloys. One such effort uses coherent (Ni,Fe)Al precipitates in a ferritic matrix
and has shown good initial results [40].
Fig. 6.24 (a) LEAP data of two Al3Sc precipitates in an Al matrix with the <100> direction
nearly parallel to the analysis direction. A 1 at.% Sc isoconcentration surface has been used in
order to extract the precipitates only. Analysis of the planar spacings along the <100> direction is
shown by the z-SDMs of (b) the AlAl atoms and the (c) ScSc atoms within the Al3Sc pre-
cipitates. This comparison is quantified in (d)
A subset of LEAP data from a NiAlFeCr alloy [41] is shown in Fig. 6.25a, where
small FeCr precipitates are clearly visible in the microstructure. SDM analysis may
be performed on either single elements or combinations of elements, and in this
case the analyses were performed by grouping together the NiAl atoms and the
FeCr atoms. The results are shown in Fig. 6.25b where the gray line denotes the
NiAl SDM results and the purple line denotes the FeCr results. The curves match
well in the overlaid plot, indicating that the spacing between Fe and Cr atoms (in the
bcc phase) is the same as the spacing between the Ni and Al atoms (B2 phase).
6.6.6 Site Occupancy of Nb in TiAl
Nb is a common alloying element added to -TiAl alloys in order to improve

oxidation resistance. In order to understand this alloying effect, it is important
to know on which sublattice site in the L10 structure, Fig. 6.26a, the Nb atoms
are located.
References 197
Fig. 6.25 (a) LEAP data from a NiAlFeCr alloy where the microstructure consists of small FeCr
precipitates in a NiAl matrix. (b) NiAlNiAl and FeCrFeCr z-SDMs showing very similar planar
spacing between the matrix and the precipitates
Fig. 6.26 Site occupancy of Nb atoms in a 46 at.%Ti52 at.%Al2 at.%Nb alloy which exhibits
the L10 crystal structure (a) as determined by (b) z-SDM analysis of AlAl, AlTi, and AlNb [37]
SDM analyses along the <001> direction in a 46 at.%Ti52 at.%Al2 at.%Nb

alloy shows the separation of Al planes and Ti planes equal to 0.4 nm, as expected
from the L10 structure. A z-SDM analysis of AlAl, AlTi, and AlNb [37] is
shown in Fig. 6.26b. During field evaporation the Nb is preferentially retained
within its plane (due to its high evaporation field) [27, 42] and this must be taken
into account in the interpretation of the results in order to determine site occupancy.
The AlNb SDM (black line Fig. 6.26b), taking into consideration the direction of
field evaporation, indicates that Nb atoms are positioned on Ti planes, in agreement
with previous findings [27, 43, 44].
References
1. Bevington, P.R., Robinson, K.D.: Data Reduction and Error Analysis for the Physical
Sciences, 3rd edn. McGraw-Hill Higher Education, New York, NY (2003)
2. Gedcke, D.A.: How histogramming and counting statistics affect peak position precision.
ORTEC Application Note AN58 (2005)
3. Currie, L.A.: Limits for qualitative detection and quantitative determination: application to
radiochemistry. Anal. Chem. 40, 586592 (1968)
5. Meier, P.C., Zund, R.E.: Statistical Methods in Analytical Chemistry. Wiley, New York, NY
(2005)
6. Boyd, R.K., Basic, C., Bethem, R.A.: Trace Quantitative Analysis by Mass Spectrometry.
Wiley, West Sussex (2008)
in Materials Science, vol. 160. Springer, New York, NY (2012)
8. Gedcke, D.A.: How counting statistics controls detection limits and peak precision. ORTEC
Application Note AN59 (2001).
17, 114 (2011)
10. Hellman, O.C., Rivage, J.B., Seidman, D.N.: Efficient sampling for three-dimensional atom
probe microscopy data. Ultramicroscopy 95, 199205 (2003)
11. Lopes, A., Bordlie, K.: Interactive approaches to contouring and isosurfaces for geovisua-
lization. In: Dykes, J., MacEachren, A.M., Draak, M.J. (eds.) Exploring Geovisualization,
p. 345. Elsevier, London, UK (2005)
12. ONeill, R.W., Larson, D.J., Thompson, K., Kunicki, T.C., Geiser, B.P.: Measuring the
roughness of buried interfaces in nanostructures by local electrode atom probe analysis.
Microsc. Microanal. 12(S2), 1746CD1748CD (2006)
13. Hellman, O.C., Vandenbroucke, J.A., Rusing, J., Isheim, D., Seidman, D.N.: Analysis of three-
dimensional atom-probe data by the proximity histogram. Microsc. Microanal. 6, 437444
(2000)
14. Hyde, J.M., English, C.A.: An analysis of the structure of irradiation induced Cu-enriched
clusters in low and high nickel welds. MRS Symp. Proc. 650(R6.6), R6.6.1R6.6.12 (2000)
15. Vaumousse, D., Cerezo, A., Warren, P.J.: A procedure for quantification of precipitate
microstructures from three-dimensional atom probe data. Ultramicroscopy 95, 215221 (2003)
16. Heinrich, A., Al-Kassab, T., Kircheim, R.: Investigation of the early stages of decomposition
of Cu-0.7%at.% Fe with the tomographic atom probe. Mater. Sci. Eng. A 353(12), 9298
(2003)
17. Stephenson, L.T., Moody, M.P., Ringer, S.P.: Techniques for the analysis of clusters and
aggregations within atom probe tomography data. Microsc. Microanal. 12(S2),
1732CD1733CD (2006)
18. Cerezo, A., Davin, L.: Aspects of the observation of clusters in the 3-dimensional atom probe.
Surf. Interface Anal. 39(23), 184188 (2007)
19. Ceguerra, A.V., Moody, M.P., Stephenson, L.T., Ringer, S.P.: Analysis techniques for nano-
scale solute clustering in atom probe tomography. Microsc. Microanal. 13(S2), 16021603
(2007)
20. Serizawa, A., Miller, M.K.: Influence of analysis parameters on the microstructural character-
ization of nanoscale precipitates. MRS Symp. Proc. 1231, 1925 (2010)
21. Stephenson, L.T., Moody, M.P., Gault, B., Ringer, S.P.: Estimating the physical cluster-size
distribution within materials using atom-probe. Microsc. Res. Tech. 74(9), 799803 (2011).
doi:10.1002/jemt.20958
22. Stephenson, L.T., Moody, M.P., Ringer, S.P.: Theory of solute clustering in materials for atom
probe. Phil. Mag. 91, 2200 (2011). http://dx.doi.org/10.1080/14786435.2011.554909
23. Marquis, E.A., Hyde, J.M.: Applications of atom-probe tomography to the characterisation of
solute behaviours. Mater. Sci. Eng. R Rep. 69(45), 3762 (2010)
24. Stephenson, L.T., Moody, M.P., Liddicoat, P.V., Ringer, S.P.: New techniques for the analysis
of fine-scaled clustering phenomena within atom probe tomography (APT) data. Microsc.
Microanal. 13, 448463 (2007)
References 199
25. Hyde, J.M., Marquis, E.A., Wilford, K.B., Williams, T.J.: A sensitivity analysis of the
maximum separation method for the characterisation of solute clusters. Ultramicroscopy 111
(6), 440447 (2011). doi:10.1016/j.ultramic.2010.12.015
26. Geiser, B.P., Kelly, T.F., Larson, D.J., Schneir, J., Roberts, J.P.: Spatial distribution maps for
atom probe tomography. Microsc. Microanal. 13, 437447 (2007)
27. Boll, T., Al-Kassab, T., Yuan, Y., Liu, Z.G.: Investigation of the site occupation of atoms in
pure and doped TiAl/Ti3Al intermetallic. Ultramicroscopy 107(9), 796801 (2007)
28. Fultz, B., Howe, J.: Transmission Electron Microscopy and Diffractometry of Materials.
Springer, New York, NY (2013)
29. Camus, P.P., Larson, D.J., Kelly, T.F.: A method for reconstructing and locating atoms on the
crystal lattice in three-dimensional atom probe data. Appl. Surf. Sci. 87/88, 305310 (1995)
30. Cerezo, A., Abraham, M., Lane, H., Larson, D.J., Thuvander, M., Seto, K., Warren, P.J.,
Smith, G.D.W.: Three-dimensional atomic scale analysis of interfaces. Paper presented at the
Electron Microscopy and Analysis Group Conf. EMAG99, Sheffield, UK, 1999
31. Vurpillot, F., Da Costa, G., Menand, A., Blavette, D.: Structural analyses in three-dimensional
atom probe: a Fourier transform approach. J. Microsc. 203(3), 295302 (2001)
32. Vurpillot, F., Renaud, L., Blavette, D.: A new step towards the lattice reconstruction in 3DAP.
33. Warren, P.J., Cerezo, A., Smith, G.: Towards 3D lattice reconstruction with the position
sensitive atom probe. Microsc. Microanal. 4(S2), 8687 (1998)
34. Vurpillot, F., Geuser, F.D., Costa, G.D., Blavette, D.: Application of Fourier transform and
autocorrelation to cluster identification in the three-dimensional atom probe. J. Microsc. 216,
234 (2004)
35. Moody, M.P., Gault, B., Stephenson, L.T., Ringer, S.P.: Applications of spatial distribution
maps for advanced atom probe reconstruction and data analysis. Microsc. Microanal. 15(S2),
246247 (2009)
36. Suram, S.K., Rajan, K.: Refining spatial distribution maps for atom probe tomography via data
dimensionality reduction methods. Microsc. Today 18(5), 941952 (2012)
37. Marquis, E.A., Geiser, B.P., Hekmaty, M.A., Larson, D.J.: Application of spatial distribution
mapping to the investigation of alloys. Microsc. Microanal. 13(S2), 196197 (2007)
38. Seidman, D.N., Marquis, E.A., Dunand, D.C.: Precipitation strengthening at ambient and
elevated temperatures of heat-treatable Al(Sc) alloys. Acta Mater. 50, 40214035 (2002)
39. Larson, D.J., Kelly, T.F.: Nanoscale analysis of materials using a local-electrode atom probe.
Microsc. Anal. 78, 5962 (2006)
40. Stallybrass, C., Sauthoff, G.: Ferritic FeAlNiCr alloys with coherent precipitates for high-
temperature applications. Mater. Sci. Eng. 387389, 985990 (2004)
41. Larson, D.J., Prosa, T.J., Kostrna, S.L.P., Ali, M., Kelly, T.F., Stallybrass, S., Schneider, A.,
Sauthoff, G., Degass, J.: Local electrode atom probe study of phase transformations in an
Fe-Al-Ni-Cr alloy. Microsc. Microanal. 12(S2), 968CD970CD (2006)
42. Al-Kassab, T., Yuan, Y., Kluthe, C., Boll, T., Liu, Z.-G.: Investigation of the ordering and
atomic site occupancies of Nb-Doped TiAl/Ti3Al intermetallics. Surf. Interface Anal. 39(23),
257261 (2007)
43. Yang, R., Hao, Y.L., Song, Y., Guo, Z.X.: Site occupancy of alloying additions in titanium
aluminides and its application to phase equilibrium evaluation. Zeitschrift Fur Metallkunde 91
(4), 296301 (2000)
44. Hao, Y.L., Yang, R., Cui, Y.Y., Li, D.: The influence of alloying on the alpha(2)/(alpha
(2) + gamma)/gamma phase boundaries in TiAl based systems. Acta Mater. 48(6),
13131324 (1999)
Chapter 7
Applications of the Local Electrode
Atom Probe
Over the past decade there has been a substantial expansion of the applicability of
atom probe tomography (APT) to materials of all types. This expansion has been
spurred on by major instrumentation developments resulting in the achievement of
high data collection rates, large fields of view, and renewed utilization of laser
pulsing (see Chap. 3). Furthermore, the advent of focused ion beam (FIB) methods
for specimen preparation (see Chap. 2) in APT has had an equally profound impact.
FIB-based methods have not only made it possible to create a LEAP specimen from
nearly any bulk material type, but have also provided the capability to create
specimens from specified regions of a sample and in nearly any orientation. At the
turn of the century, random site specimen creation was the norm and site-specific
specimen creation was a time-consuming process involving many handling steps in
conjunction with electropolishing and electron microscopy. Today, FIB-based site-
specific specimen creation is routine and electropolishing is often only used when it
is more practical. While the use of thermal pulsing is necessary for materials that
cannot be evaporated successfully with field pulsing, thermal pulsing also has been
found to improve yield for a range of metal applications as well. Certain metal
specimens, even high-strength steels, that will not run with adequate yield in voltage
pulsing have been found to run with much higher success rates in laser pulsing. With
these major advances, applications of APT have blossomed.
This final chapter is a survey of some of the most recent APT applications. The
content is organized by material type as opposed to structural type or industry.
Thus, there are sections on metals, ceramics and geological materials, semiconduc-
tors, organic and biological materials, and composite structures and devices for
combinations of the above. Note that there are literally thousands of examples
which could be used in this chapter (see [1] for a recent review of many additional
examples), and no attempt has been made to be comprehensive in any of these
categories. Rather, we have chosen examples that illustrate the different ways
that APT has been used or contributed to material characterization challenges. By
describing a wide array of examples, we hope that the reader may gain a broader
perspective on the many ways that atom probes are being used today and perhaps
this may spur users to develop new applications for the technique.
202 7 Applications of the Local Electrode Atom Probe
7.1 Metals
Because atom probe analysis relied almost exclusively upon voltage pulsing in its
first three decades, virtually all of the work from the 1970s through 2000 was on
metals. As a result, these applications tend to be the most mature. However, new
and more sophisticated applications are constantly being pursued and reported in
the literature, as highlighted in the following examples.
7.1.1 Ordered Alloy
Srinivasan et al. [2] have used high-angle annular dark-field scanning transmission
electron microscopy (HAADF STEM) and APT to study the orderdisorder transi-
tion between the disordered matrix and ordered 0 precipitates in a nickel-base
superalloy, Fig. 7.1. STEM was used to elicit the variation in order across the
coherent <001> hetero-interface through intensity ratios of the aluminum-rich/
aluminum-poor {002} plane signal. APT was used to measure the aluminum
composition profile and estimate the ordering transition width across this same
interface type. The two techniques agree very well. Furthermore, the experimental
results matched with high precision the atomistic modeling results of Mishin [3]
Fig. 7.1 Orderdisorder transition at a hetero-interface. (a) {002} plane intensity in HAADF
STEM at the 0 interface. (b) APT image with phase (Co atoms shown in blue) and 0 phase
(Al atoms shown in red). (c) Close-up of the lower interface in (b) showing the Al atoms in the
{002} where the interface transition based on atomic ordering is about 46 atomic layers
(0.71.0 nm) thick. (d) A proximity histogram across the same interface showing a compositional
interface width of about 2 nm. Reprinted from Srinivasan et al. [2]. Copyright (2009) American
Physical Society
7.1 Metals 203
and Ardell and Ozolins [4]. They found that the orderdisorder transition width
(about 0.71 nm) is smaller than the compositional width of the interface (about
2 nm), which raises fundamental questions concerning the definition of interface
width at orderdisorder interfaces. This work illustrates an important trend in APT
work today: there is a strong convergence between the increasing capabilities of
atomistic modeling and the experimental data available from APT (and TEM). This
synergistic relationship is expected to continue as it is a natural marriage of
predictive and characterization capabilities at the atomic scale.
7.1.2 Site Occupancy in Precipitates in Aluminum Alloys
As the scale and detail available in atom probe data are directly comparable with
that available in atomistic simulations, synergistic integration of the two techniques
is expected to grow rapidly. Gault et al. [5] have used APT to determine the
occupancies of various sites in an ordered phase, L12 (0 , Al3(Li,Mg)), in a model
aluminum alloy. In particular, it was not known whether magnesium would inhabit
the aluminum sites or the lithium sites in an alloy precipitate and APT appeared to
be the only experimental technique that might readily resolve the question. The
results are directly compared with atomistic simulations.
Figure 7.2 shows LEAP data for two heat treatments with precipitates present.
Subvolumes that contained closely spaced atomic planes were extracted from each
dataset. Spatial distribution maps (SDMs, see Chap. 6) were used to analyze the
subvolumes for atomic position correlations. The result is shown in Fig. 7.3. In the
disordered fcc matrix, aluminum, lithium, and magnesium each occupy any lattice
site with equal probability. In the precipitates, however, it is clear that the magne-
sium preferentially occupies the lithium sites in the Al3Li structure. The relative
occupancy fraction between lithium and magnesium on the lithium sites was not
determined precisely but is a subject for further investigation. Atomistic simula-
tions show that the lowest energy configuration for magnesium in the Al3Li pre-
cipitates is for magnesium to sit on the lithium sites; this is directly aligned with the
experimental LEAP findings.
7.1.3 Imaging Nanovoids
Voids in a material generally have a major, usually deleterious, effect on the

mechanical properties. Historically, TEM, positron annihilation spectroscopy, and
even field ion microscopy have been the principal methods of identifying and
cataloging voids. Miller et al. [6] have shown that voids may be visualized by
taking thin slices through a material where the slice thickness must be on the order
of the void diameters, Fig. 7.4a. This approach should work for voids that are on the
order of 20 nm diameter or less. Field ion microscopy was shown to aid in the
identification of the voids.
Fig. 7.2 Typical three-dimensional reconstruction from atom probe analyses of the alloy after (a)
8 h at 150 C and (b) 24 h at 120 C shown at the same scale. (c) A close-up showing the {022}
atomic planes imaged in the dataset displayed in (a). (d) A close-up showing the {002} atomic
planes imaged in the dataset displayed in (b). (e, f) Model view of the fcc Al matrix and of the L12
structure. Reprinted from Gault et al. [5] with permission from Scripta Materialia
Miller et al. [6] have also investigated the use of iso-density surfaces as a way to
reveal the presence of small voids in a structure. They used a combination of
simulations and experimental results to determine that voids as small as 1 nm
diameter should be detectable in APT. Figure 7.4b shows an iso-density surface
that reveals three spheroidal nanovoids in an aluminumyttriumiron metallic
glass. Carbon and oxygen segregation to the nanovoids was also observed by
superimposing atom maps onto the iso-density surfaces. The data show that the
trace concentrations of carbon and oxygen in this material are preferentially
segregated to the voids.
7.1.4 Intergranular Attack in Ni-Base Superalloy
Intergranular attack (IGA) in nickel-base superalloys causes premature failure of

major components in many energy systems including nuclear power reactors.
However, the fundamental mechanisms of IGA have been difficult to ascertain
because of the complex 3D character of the problem and the low concentrations of
7.1 Metals 205
Fig. 7.3 Normalized species-specific SDMs for the two heat treatments in Fig. 7.2 in the matrix
(top row) and the precipitates (bottom row). Note the change in the periodicity between the
precipitates and the matrix in both cases. The corresponding structure is shown in the inset for
each case. Reprinted from Gault et al. [5] with permission from Scripta Materialia
pertinent elements involved. In a very thorough investigation that represents a tour

de force in analytical microscopy, Schreiber et al. [7] investigated this problem with
a combination of SEM, TEM, NanoSIMS, and LEAP. This group found two distinct
regions along grain boundaries near the surface, Fig. 7.5. Up to depths of about
200 nm into the surface, oxidation along the boundary occurred leading to the
formation of predominantly chromium oxide species with the metal oxide structure.
Grain boundary migration and depletion of chromium accompanied the oxide
formation. Below the metal oxide region there is a transition to a sulfurized region
where discrete, separated sulfide precipitates formed on the boundary, Fig. 7.6. The
sulfurized regions also exhibited grain boundary migration and total depletion of Cr
in the migrated volumes. Such complete analytical work on grain boundaries that
had undergone IGA had previously not been performed. These findings help
explain properties measurements and may lead to new understanding of the funda-
mental mechanism in IGA in these important alloys. This work is an excellent
Fig. 7.4 (a) Nanovoids in nanocrystalline copper revealed by a thin section atom map. Oxygen
and sulfur are evidently segregated to the voids. (b) Nanovoids in an aluminumyttriumiron
metallic glass revealed by an isodensity surface. Note that carbon and oxygen are found to
segregate to the nanovoids in this material. Reprinted from Miller et al. [6] with permission
from Ultramicroscopy
Fig. 7.5 STEM X-ray map of the near surface region of alloy 600 after IGA. Chromium oxides
form a continuous film to depths of 200 nm. Sulfides form discontinuous precipitates at greater
depths up to 500 nm. Reprinted from Schreiber et al. [7] with permission from Microscopy and
Microanalysis
7.2 Catalytic Materials 207
Fig. 7.6 (a, b) LEAP reconstruction of the IGA in alloy 600 depicting isoconcentration surfaces
of oxides (blue), sulfides (red), and a large carbide (maroon) in crosssection and plan-view,
respectively. The surface of the specimen prior to FIB lift-out is noted by a yellow dashed line,
above which exists some oxide overgrowth. (cf) Atom maps (10 nm image depth) depicting Ni
(green), Cr (yellow), 16O+ (black), 32S2+ (red), and Cu (blue). Reprinted from Schreiber et al. [7]
with permission Microscopy and Microanalysis
example of the need for, and execution of, site-specific analysis with APT and its
complementary nature with SIMS and TEM.
Metal use in nanoparticles is industrially important, but offers further challenges
in preparation for APT analysis. Some examples of this application are given in the
next section.
7.2 Catalytic Materials
7.2.1 Ex Situ Analysis of CoCuMn Nanoparticles
Large-scale industrial production of long-chain terminal alcohols is economically

important as they are used as feedstock for plasticizers, lubricants, and detergents.
There are several routes to the catalytic preparation of these alcohols based on
homogeneous catalysis using noble metals. Xiang et al. [8] have recently developed
an alternative synthesis approach based on heterogeneous catalysis using transition
metal catalysts. Prior work had shown that cobaltcopper alloys produce good
catalytic activity but insufficient selectivity to the alcohols of interest. Moreover,
according to the cobaltcopper phase diagram, a maximum of 9 at.% copper can be
dissolved in cobalt. Additions of manganese have been shown to increase the
amount of copper that can be dissolved and should provide a route to greater
catalytic selectivity. A ternary metallic alloy, if formed, is expected to have the
highest catalytic selectivity. Another key question is whether the ternary alloy
contains major amounts of oxygen.
Fig. 7.7 (a) Atom probe microscopy results from the CoCuMn catalyst nanoparticles. The images
provided here are sections through 3D tomographic reconstructions so as to reveal the
intragranular structure of the nanoparticles. The data are presented in the form of atom maps
where cobalt atoms are depicted as blue spheres, copper atoms as orange spheres, manganese
atoms as green spheres, and oxygen as white spheres. A faceted coreshell structure that
contained intracore clusters was observed. The data displayed are from a 10 nm thick slice through
one grain. (b) An enlarged view from a 5 nm thick slice of the data at the coreshell interface.
Oxygen (white) was distributed throughout the coreshell interface. (c) The intracore clusters
were observed to contain a high concentration of copper and manganese; 5 nm thick atom map. (d)
A 3D sectional view of the element distribution. Reprinted from Xiang et al. [8] with permission
from Journal of the American Chemical Society
After activating the catalyst precursor material, a catalyst powder was obtained
which had to be passivated prior to APT analysis. An APT specimen was prepared
using FIB-based techniques to encapsulate and then sharpen the specimen. This
preparation made it possible to analyze a single nanoparticle as shown in Fig. 7.7.
The particle appears faceted and a striking coreshell structure is observed,
Fig. 7.7a, b. The shell thickness is about 2 nm with all three elements present in
the shell. There is also a region of the nanoparticle that contains elevated levels of
copper and manganese relative to the cobalt, Fig. 7.7a, c, and which may be
considered a precipitate. Note also that there is very little oxygen in the shell
which indicates that metal oxides are not forming. The key finding in this work is
the metallic shell structure with high levels of intermixed cobalt, copper, and
manganese. The finding is consistent with expectations and explains the high
catalytic selectivity of these processed nanoparticles. This work is an illustration
of how APT may impact the development of real catalyst nanoparticles and can
help in the design of catalyst compositions that lead to large improvements in a
pervasive industrial process.
7.2 Catalytic Materials 209
Fig. 7.8 A serial expose of the evolution of the microstructure in a Pt-24 at.%Rh-10 at.%Ru alloy
after oxidation at 873 K with increasing time. After 10 min (second from left), a
rhodiumruthenium oxide phase forms on a grain boundary. After 30 min (middle), this oxide
volume fraction has increased markedly and small amounts of a ruthenium-rich oxide have
formed. After 90 min, distinct rhodium-rich and ruthenium-rich oxide phases have formed. The
oxide is now the majority of the specimen. Finally, after 300 min, the metallic phase has almost
been fully replaced by oxide and the volume fraction of the ruthenium-rich oxide has increased. At
the bottom, a specimen after 120 min reduction at 673 is shown rhodium-rich and ruthenium-rich
metallic phases have been produced. Reprinted from Li et al. [11] with permission from Journal of
Physical Chemistry C. Copyright (2012) American Chemical Society
7.2.2 In Situ Analysis of PdRd and PtRhRu Catalysts
Heterogeneous catalysis using noble metals is usually performed with nanoscale

particles to increase the surface area and minimize the loading of expensive metals.
These nanoparticles are commonly in the 100 nm diameter range or less. This is
typically the size of the apex of an atom probe specimen. The group at the University
of Oxford has built a catalysis reaction chamber that attaches to an atom probe [9]. They
have used this system to study two noble metal catalyst alloys: palladiumrhodium
[10, 11] and platinumrhodiumruthenium [11].
A series of experiments was performed to monitor the evolution of the
microstructure of the catalyst specimens in time at 873 K, Fig. 7.8. There is a
very clear progression toward increasing oxide content with increasing time of
oxidation. A rhodium-rich oxide phase (of the type MO2) forms first but eventu-
ally a more complex microstructure with two oxides, a rhodium-rich oxide phase
(of the type M2O3) and a rhodium-rich oxide phase (of the type MO2) forms.
Upon reduction at 673 K, a rhodium-rich metallic phase and a rhodium-rich
metallic phase form. The authors suggest that this reduced microstructure, a
novel complex island structure, may be a candidate for an improved catalyst
with extremely small metallic nanophases on the order of 5 nm diameter.
Certainly, the atomic-scale characterization used in these studies is essential to
developing an understanding of the microstructures on a scale that is relevant for
catalysis work.
7.3 Ceramic and Geological Materials
The reemergence of pulsed lasers in APT analysis has opened the way for the
analysis of insulating materials including oxides of various elements like lithium
[12] magnesium [13, 14], aluminum [1517], silicon [15, 1820], iron [21], nickel
[17, 22], zinc [13, 23, 24], zirconium [12, 25], indium [26], and cesium [12,
2729]. In addition, some theoretical studies have also been published [3034]. Sev-
eral examples of the application of APT to ceramic materials are presented in this
section.
7.3.1 CeO2 as a Model for Nuclear Fuel
Cerium oxide is a model surrogate material for the various nuclear fuel materials in
both light water reactors and fast breeder reactors. Kirchhofer et al. [29] have
studied CeO2 to establish appropriate handling and analysis conditions without
the complication of radioactivity. Their work is noteworthy because they made a
concerted effort at the outset to study a host of analysis conditions and in particular,
to establish the conditions under which LEAP tomography will produce the correct
composition in these materials. Figure 7.9 shows a mass spectrum from CeO2 taken
under the optimal conditions: 1 pJ energy per pulse, 20 K base specimen temper-
ature, and a charge state ratio (Cs++/Cs+) of 40. Under these conditions, the
composition is determined to be 67 at.% oxygen as shown in Fig. 7.10. Higher
laser pulse energies were postulated to lead to evaporation of neutral molecular
oxygen in increasing fraction of the total. Neutral atoms are not accelerated
following field evaporation and are not therefore detected. In addition, the authors
found that oxygen ion evaporation is not homogeneous across the apex of the
specimen and that the inhomogeneity gets worse with increasing laser energy,
Fig. 7.11.
In this work, higher laser pulse energy also produced lower mass resolving
power (MRP) in the spectra, which was attributed to greater heating and thus
greater cooling time required for the specimen. In many other materials, higher
laser power leads to greater MRP since the cooling rate from higher temperatures is
greater. However, if the specimen does not cool sufficiently to the base temperature
7.3 Ceramic and Geological Materials 211
Fig. 7.9 Mass spectrum of a cerium oxide used for the experiments obtained at 20 K specimen
base temperature and 1 pJ laser energy. Note that Hf was present in the sample at concentration
below 1 at.%, which is commonly found as an impurity in cerium oxide. Some minor peaks are not
labeled for clarity. Reprinted from Kirchhofer et al. [29] with permission from Journal of Nuclear
Materials
Fig. 7.10 Measured oxygen content for the cerium oxide on the laser side of the specimen, or
opposite from the laser side, shown for different laser energies and for a specimen base temper-
ature of 20 K. The oxygen peak at 16 Da was ranged as both O+ and O2++ for the composition
analysis; the best composition was obtained using the latter ionic ranges. Reprinted from
Kirchhofer et al. [29] with permission from Journal of Nuclear Materials
Fig. 7.11 2D concentration maps showing the relative distribution of ionic clusters (using a
thermal color gradient, where the red color indicates the highest frequency of evaporation events)
across the specimen for the conditions that yielded the best stoichiometric evaporation of ions. The
laser pulse was incident on the specimen at the lower left corner of each plot. Note that evaporation
of O+/O2++ ions was not even across the specimen, while field evaporation of CeO+/CeO++ clusters
was more homogeneous. Reprinted from Kirchhofer et al. [29] with permission from Journal of
Nuclear Materials
between pulses, then there is an effective rise in base temperature, resulting in data
quality reduction. Indeed, increasing thermal tails were observed in the spectra with
increasing laser energy. Another observed benefit of lower laser energy was higher
spatial resolution in the data which is attributed to diminished potential surface
diffusion at the lower temperatures associated with lower pulse energy. One
disadvantage of lower laser pulse energies was a higher fraction of multiple hits
which was explained as due to a lower fraction of atoms evaporating as
molecular ions.
Fig. 7.12 (a) Yttrium L X-ray image from an SXFive electron probe microanalyzer of a 0.5 Ga
zircon. The yttrium signal is elevated at the fossil resorption front. (b) The image from (a)
superimposed on a backscatter image of the specimen after a FIB lift-out from the region of the
resorption front [35]
7.3.2 Zircons
Zircons (ZrSiO4) are valuable minerals scientifically due to their extreme stability
and hence longevity in solar (terrestrial and extraterrestrial) and extrasolar envi-
ronments. Like most materials, their microstructure contains a record of their
history. Two such examples illustrate how LEAP data have contributed to the
understanding of not just Earths history, but possibly the history of Mars as well.
Snoeyenbos et al. [35] were the first to demonstrate LEAP analysis of ancient
zircons. They used an electron probe microanalyzer (EPMA) to map the low
concentration element, yttrium, in a 0.5 Ga zircon (1 Ga109 years old),
Fig. 7.12, which guided the subsequent specimen preparation by FIB to a resorption
front in the zircon. The yttrium concentration at the resorption front is too low to
detect in standard energy-dispersive X-ray spectroscopy (EDX) analysis in an
SEM. The LEAP data show that a band of elevated uranium, thorium, yttrium,
phosphorous, and aluminum at the coreovergrowth boundary is consistent with a
fossil resorption front. Some yttrium, phosphorous, uranium and thorium from the
resorption front was remobilized up to several m into the overgrowth along
subparallel, sub-planar structures, Fig. 7.13. Despite considerable compositional
heterogeneity in the overgrowth, there is no evidence of migration of radiogenic
lead from its uranium parent. This research demonstrates that the decades of work
with SIMS instruments to use lead isotope ratios for dating was valid because lead
did not migrate away from its point of origin.
Valley et al. [36] were able to reaffirm that dating based on 206Pb207Pb isotope
ratios in LEAP data for 4.37 Ga zircons match SIMS-based dating within the
statistical uncertainty of the smaller number of atoms in LEAP datasets. Further-
more, yttrium-rich clusters were found, Fig. 7.14, that were interpreted as having
Fig. 7.13 Atom maps of the key elements of interest for a specimen taken near the resorption front
of Fig. 7.12. The lead does not appear to have migrated away from its parents uranium and thorium
[35]
Fig. 7.14 Atom map containing 260 million atoms (only yttrium shown) of a 4.4 Ga zircon.
Yttrium-rich clusters are clearly visible
formed, not during their original crystallization, but later during the magmatic
upheaval event at 3.4 Ga. Lead ratios inside the yttrium-rich clusters, Fig. 7.15,
match models for diffusive migration of the lead to the clusters during the upheaval
event [36]. Geological applications of LEAP have just begun, but they are already
having impact on the field. There is a bright and exciting future for geological
science that utilizes APT to acquire heretofore unattainable information about the
largest-scale events in our universe from the smallest-scale traces of their existence.
7.3.3 Extrasolar Nanodiamonds
If you think 4.4-billion-year-old zircons are ancient, to some exogeologists they

seem young. There is an ongoing search to qualify the source of meteoritic
nanoscale diamonds that are suspected to be of extrasolar origin; that is, they may
Fig. 7.15 Segment of the mass spectrum from the dataset shown in Fig. 7.14 that contains the
Pb++ isotopes. (a) Lead signal from the full dataset, (b) lead signal from inside the yttrium-rich
clusters, and (c) lead signal from outside the yttrium-rich clusters. It is clear that the lead is
segregated to the yttrium-rich clusters and that 207Pb is even more preferentially found in these
clusters
have been created in another star prior to our solar system. In truth, there are many
events that can lead to the formation of nanoscale diamonds, including: between
presolar supernovae, condensation in the interstellar medium, formation in the
protosolar nebula, and even formation in the early solar system. The relative
amounts of stable isotopes of elements such as carbon are known to be different
for each of these sources and can be used as a signature. However, any given
collection of nanodiamonds could have representatives from any or all of these
sources. Thus, though isotope ratios may be determined from a collective of
particles by any of several techniques, what is needed is analysis of isotope ratios
from many individual nanodiamonds. APT is the only technique that can deliver
such information. Heck et al. [37] have spent several years trying to employ APT to
discern this very information. They faced two primary obstacles: (a) specimen
preparation of sub-10 nm diamonds and (b) qualification of the accuracy of the
isotope ratios determined from APT.
Specimens were prepared from nanodiamonds using an acid-dissolution residue
Allende DM, a well-studied nanodiamond isolate [38]. The isolate was suspended
in alcohol, dispersed on a silicon wafer, and then covered with electron-beam-
deposited platinum in a FIB. A small wedge of material was removed and
microtips were fabricated (see Chap. 2). Data were acquired as shown in
Fig. 7.16. Great care was taken to ensure that the isotope ratios of 12C13C were
not influenced by such experimental parameters as multi-hit detection efficiency,
choice of region of interest, and hydride formation. Their preliminary results are
shown in Fig. 7.17. Note that the accepted terrestrial 12C13C ratio is 89 [39]
(shown by the solid line). Within the scope of this first effort, no statistically
significant difference was found between terrestrial detonation nanodiamonds
(DND), terrestrial ultrananocrystalline diamonds (UNCD), and meteoritic
nanodiamonds (Allende DM). Such careful analyses and attention to detail are
essential to developing a method and establishing the reliability of APT in deliv-
ering accurate isotopic ratios for forensic cosmology.
Fig. 7.16 3D-tomographic reconstruction of meteoritic nanodiamonds from the Allende DM

residue embedded in a platinum matrix. Each dot represents a single atom of Pt (orange) or C
(black). Note that, for clarity, only a fraction of the total platinum atoms are shown. Other atoms
present within the analysis are not shown. Reprinted from Heck et al. [37] with permission from
Meteoritics & Planetary Science
Fig. 7.17 Carbon isotope peak ratios of regions of interest (ROIs) defined from selected synthetic
(DND and UNCD) and meteoritic Allende nanodiamond tips. The horizontal line corresponds to
the terrestrial 12C/13C ratio of 89 [39]. Error bars are 2 and are based on counting statistics.
Reprinted from Heck et al. [37] with permission from Meteoritics & Planetary Science
7.3.4 Ferroelectrics/Piezoelectrics
Ferroelectric oxides are a class of technologically important ceramics that are used
in a wide range of electronic components, from capacitors and transistors to
piezoelectric transducers. Relaxor ferroelectrics, in particular, have extraordinarily
Fig. 7.18 Mass spectrum

for 53/47 PZT material
obtained via APT under
acquisition conditions that
yielded the correct
composition of the oxide.
Reprinted from Kirchhofer
et al. [40] with permission
Microscopy and
Microanalysis
Table 7.1 Composition Element Measured Nominal

(at.%) of PZT specimen as
determined by APT analysis Pb 19.2 20.0
Zr 9.2 10.6
Ti 11.1 9.4
O 60.5 60.0
high piezoelectric and electrostrictive coefficients which makes them valuable for
electromechanical applications [40]. Through substitutions of niobium or lantha-
num for some of the lead in lead zirconium titanate (PZT), PNZT and PLZT relaxor
ferroelectrics are created. It has been proposed that relaxor ferroelectrics achieve
their electrostrictive properties through nanoscale phase separation. APT, therefore,
is a logical tool to investigate these compositional variations. Since these complex
oxide materials have not been significantly studied using APT, Kirchhofer
et al. [40] have performed a series of test experiments to establish the optimal
data collection conditions. In their work, APT and TEM are used to complement
each other and ensure the accuracy of the APT results. APT instrumental parame-
ters such as laser pulse energy, acquisition rate, and specimen base temperature
were optimized to obtain accurate cation and anion stoichiometry from the oxides.
The mass spectrum for PZT in the optimized run condition is shown in Fig. 7.18.
The composition determined from this analysis closely matches the expected
composition of the material, Table 7.1. This research is at an early stage, but it
has already demonstrated that this important class of materials runs well in APT and
that such studies can lead to new understanding of piezoelectric behavior.
Analysis of ceramic-based materials and composites continues to develop,
taking the field of APT into nontraditional applications. Further examples of
novel applications are summarized in the sections Organics and Biological Mate-
rials and Composites Structures/Devices below.
7.4 Semiconductor Materials
Over the past few decades, characterization of semiconductor devices has been
primarily achieved through use of electron microscopy, ion microscopy, various
X-ray technologies, as well as some emission and light optical techniques. As the
industry has moved further into the nanoelectronics realm, the use of these methods
has become more challenging due to the atomistic scale and low dimensionality of
the devices of interest. One significant goal in the application of APT to silicon-
based structures has been the analysis of actual product devices instead of test
structures. Down-scaling of devices and the reduction of sourcedrain junction size
to avoid short-channel effects combined with the increased field of view (FOV)
now available in modern atom probes [41, 42] now make the analysis of entire
device structures a reality. Several researchers recently have used APT in the
analysis of basic configurations used in semiconductor devices or portions of
partially processed device structures and these are presented in this section. Anal-
ysis of more complete or entire devices [4355] is explored in the section below
titled Composite Structures/Devices.
7.4.1 Group IV Semiconductors (Silicon and Germanium)
7.4.1.1 Isotopic Multilayer Structures
Quantitative evaluation of the spatial resolution at an interface is a critical mea-

surement for establishing the performance of any analytical technique. In APT, the
highest spatial resolution is obtained in the direction parallel to the specimen long
axis. In layered materials with a hetero-interface (two different materials on either
side of the interface), this measurement could be severely compromised by trajec-
tory aberrations and reconstruction artifacts associated with field evaporation of
dissimilar material layers (see Chap. 5). However, pure materials with layers that
differ only in isotope concentration do not have these aberrations and therefore can
be used to study the fundamental spatial resolution at an interface.
Shimizu et al. [5658] (and others [59]) have, in several different studies, used
distinct isotopes of silicon and/or germanium to make multilayered structures to
assess the fundamental spatial resolution of APT. In the most recent work [58],
nearly pure 70Ge was used for one layer (actual composition was 4 % 72Ge, 96 %
70
Ge) alternating with a layer of natural abundance germanium (natGe), Fig. 7.19.
The composition profile across the interfaces was compared with similar measure-
ments made using SIMS, Fig. 7.20. In all layers, the APT estimate of the interfacial
width was sharper than the SIMS measurement. The authors used proxigrams to
assess the APT composition profile across the interface and found the transition
width for 1684 % 70Ge to be 0.8 nm (less than two lattice spacings). These
structures in germanium were also compared with earlier work on silicon isotopic
7.4 Semiconductor Materials 219
Fig. 7.19 (a) Atom maps of the five Ge stable isotopes (70Ge, 72Ge, 73Ge, 74Ge, and 76Ge) in the
germanium isotopic multilayers. Doubly charged (Ge++) and singly charged (Ge+) peaks in the
mass spectra in the 1st pair of the (b) 70Ge and (c) natGe layers (selected volume: 20 20 3
nm3) are shown in (b) and (c), respectively. Reprinted from Shimizu et al. [58] with permission
from Journal of Applied Physics. Copyright (2013) American Institute of Physics
Fig. 7.20 (a) Depth profiles by APT (solid curves) of 70Ge and 74Ge in the germanium isotopic
multilayers obtained from an area of 10 10 nm2, overlaid with the SIMS profiles (broken
curves). (b) 70Ge and 74Ge mappings in a volume of 10 10 100 nm3 with the isoconcentration
surfaces of 70Ge at 50 % for highlighting the 70GenatGe interfaces. Reprinted from Shimizu
et al. [58] with permission from Journal of Applied Physics. Copyright (2013) American Institute
of Physics
multilayers and the same result was found for both studies to within experimental
error. Thus, where the field difference between two layers or sections of a structure
is not large, APT has been shown to deliver quite high spatial resolution across
hetero-interfaces.
7.4.1.2 Doping in Silicon and Germanium Nanowires
Nanowires have been around now for well over a decade and they remain an
interesting prospect for nanoelectronic device components. Understanding and
developing doping methods for nanowires during growth are the aims of current
research. However, there has been little information about the effects of growth
conditions on dopant concentrations in nanowires. APT analyses of nanowires
would seem to be a logical undertaking to answer these questions.
Lauhons group at Northwestern University has pioneered many of the tech-
niques needed to perform APT on nanowires. Surprisingly, specimen preparation is
more difficult than one might guess given the near-perfect shape and size of most
Fig. 7.21 (a) End view through the first 20 nm of the dopant decay region in the phosphorus-
doped germanium nanowire. (b) Schematic of part a, illustrating the hexagonal cross section
of the catalyst-defined core and associated radial doping gradient. (c) End view of the first
50 nm of the dopant decay region of a phosphorus-doped germanium nanowire with a
chromium shell and (e) the first 60 nm of the dopant decay region of a boron-doped silicon
nanowire. Scale bars are 10 nm. (d, f) Radial composition profile derived from the region in
the red rectangle in (c, e), respectively. The trijunction location is indicated by a
dasheddotted line and serves as the reference point (r 0) for each graph. Error bars
represent the statistical uncertainty in the number of counts. Reprinted from Perea
et al. [60] with permission from Applied Physics A
nanowires for APT specimens. The biggest challenge has been handling the
nanowires such that a single wire may be positioned and analyzed in an atom
probe. One way that Lauhons group solved this challenge was by growing single
nanowires on top of flat-top microtips [60]. In some recent work, silicon and
germanium nanowires were grown with boron and phosphorus dopants, respec-
tively, and the spatial distribution of the dopants was assessed, Fig. 7.21. A rim of
elevated concentration of the dopants is found as observed in Fig. 7.21a and in the
radial composition profiles of Fig. 7.21d, f. The growth direction of the wires is
<111> and the symmetry of the phosphorus distribution in the cross section of the
wire appears in accordance with this symmetry. The authors posit that there is
faceting at the liquidsolid interface that leads to this distribution. The advances in
understanding of the vaporliquid-solid growth process derived from this study are
an important step toward developing practical processing approaches for fabrica-
tion of doped nanowires. This work is a good example of a characterization task for
which APT may be the only way to obtain the desired information.
Fig. 7.22 Atom map of a multilayer structure containing silicon nanocrystals. The red surface is
an isoconcentration surface at 70 % silicon. The nanocrystals are formed in layers by depositing
alternating layers of silicon dioxide with silicon-rich silicon oxide that is subsequently
decomposed into nanocrystals. Image courtesy H. Gnaser [62]
7.4.1.3 Silicon Nanocrystals
Silicon nanocrystals [61] have a potential role as a nonlinear optic in integrated

photonics structures [62]. These structures consist of amorphous silicon dioxide
with layers in which crystalline silicon precipitates are formed. These equiaxed
precipitates are on the order of a few nm in diameter. Determination of the
concentration of dopants and their disposition within the precipitates has been a
major challenge. APT appears to be highly suited to this task [18, 63].
Gnaser [62] has studied a multilayer silicon nanocrystals structure, Fig. 7.22.
The silicon-rich nanocrystals are evident as precipitate volumes in the atom map
where a silicon isoconcentration surface of 70 % silicon has been used to surround
the nanocrystals. By using a phosphorus (dopant) isoconcentration surface as well,
it was found that phosphorus is segregated to the outer surfaces of the nanocrystals.
Further examination of isolated nanocrystal volumes using a cluster identification
algorithm was performed, as shown in Fig. 7.23. The phosphorus concentration of
the nanocrystal volumes was found to be 0.77 0.4 %. The mean size of the
clusters was 3.7 0.8 nm diameter.
7.4.2 Compound Semiconductors
7.4.2.1 GaN Optoelectronic Structures
Gallium nitride (GaN) has achieved great success as the basis for a wide range of
optoelectronic devices. Updoped GaN emits in the ultra-violet (UV), and alloying it
with indium nitride (InN) or aluminum nitride (AlN) produces emission with
wavelengths ranging from the green to deep UV. GaN-based multiple quantum
well (MQW) structures are used as the active regions of commercial light-emitting
diodes (LEDs) and laser diodes. Such structures often exhibit bright-line emission
Fig. 7.23 A thin section of the atom map in Fig. 7.22 that contains a single layer of nanocrystals.
A cluster finding approach has been used to isolate the nanocrystals and determine their average
phosphorus concentration at 0.77 0.4 %. The nanocrystals are about 4 nm diameter. Image
courtesy H. Gnaser [62]
despite very high dislocation densities. Recently, APT has been used to investigate
the chemical and morphological microstructure of GaN- and GaAs-based structures
[6479].
It has been proposed that carrier localization can originate from structural and
compositional effects like quantum well width fluctuations and indium clustering
[80]. APT is ideally suited to describe compositional fluctuations and look for
nonrandom statistical fluctuations such as those proposed by Miller et al. [81]. An
LED device structure (50 nm p-GaN/30 nm GaN/3.5 nm In0.15Ga0.85N QW/n-
GaN) grown by metalorganic chemical vapor deposition on a semi-polar
10 1 1 GaN substrate has recently been analyzed with APT [71]. Figure 7.24a
shows the APT reconstruction of a section of the quantum well. A proxigram
analysis, Fig. 7.24b, was done relative to the 2 % indium composition surfaces
for the top (blue line in figure) and bottom (magenta line in figure) interfaces. The
bottom interface is abruptno spectral evidence for indium is present for regions
below 0.1 at.% indium, while the top interface has a ~2 nm tail of indium into the
GaN layer above. Frequency distribution analyses were also performed [71], and
for this particular device structure, no evidence was obtained to cause rejection of
the null hypothesis at the 95 % confidence level. In other words, the indium
fluctuations in this device were consistent with random fluctuations and could not
cause carrier localization.
Fig. 7.24 (a) APT reconstruction of a section of a quantum well. (b) Proxigram analysis relative
to the 2 % indium composition surfaces for the top (blue line) and bottom (magenta line)
interfaces. The bottom interface is abruptno spectral evidence for indium is present for regions
below 0.1 at.% indium, whereas the top interface has a 2-nm tail of indium into the GaN layer
above. Atom colors: In, magenta; Ga, orange; N2, green. Reprinted from Prosa et al. [71] with
permission from Applied Physics Letters. Copyright (2011) American Institute of Physics
7.4.2.2 Commercial Device Analysis: OSRAM Golden Dragon LED
APT analyses can be challenging enough when a researcher has thorough knowl-
edge of the global structure of a specimen. However, when the structure of a
macroscopic specimen is unknown, it can be a daunting task to find the region of
interest at the atomic scale. Reverse engineering of devices can pose just such a
challenge (also see section Commercial Devices Analysis; Intel i5-650 below for
another example of competitive analysis on a logic device).
In order to explore the feasibility of competitive analysis on whole devices,
LEAP analyses were obtained from individual commercial LEDs from fully pack-
aged devices. OSRAM 455 nm Golden Dragon Plus LEDs were purchased from a
retail supplier, depackaged, and fabricated into LEAP compatible specimens using
standard FIB methods ([82, 83] and Chap. 2), Fig. 7.25a, b. The LEDs were first
depackaged to expose the active component, Fig. 7.25a. FIB was used to cut into the
structure to expose the electrically active region, Fig. 7.25b. At this point the project
encountered difficulty. The light-emitting components of the device were not
obvious in the SEM image and the depth of this component was almost 10 m.
Without a priori knowledge it could have taken dozens of specimens prepared at
different depths to target the region of interest given that the average depth of these
LEAP analyses are on the order of a few hundred nanometers. A CAMECA
SXFiveFE EPMA was used to generate an X-ray map of indium and aluminum in
this region, Fig. 7.25c. It is clear from this X-ray map that an aluminum blocking
layer was present as evidenced by the bright aluminum band across the region. The
FIB lift-out effort was then able to focus on this region and produce a specimen that
led to the LEAP image in Fig. 7.26.
Fig. 7.25 (a) Image of a commercial white LED with the active region in the center. (b) Top
surface of white LED after FIB processing to expose the active component for lift-out. (c) X-ray
map of the active area cross section in (b) taken with a CAMECA field-emission SXFiveFE EPMA
which shows the location of the respective regions based on composition
Clearly identifiable in the structure are at least four distinct regions [84] (from
top to bottom in the atom map): (1) magnesium doping (to form p-type GaN)
near the base of the structure, (2) an aluminum-rich electron blocking layer,
(3) an indium-rich quantum well region, and (4) an indium-based superlattice
structure near the top of the dataset. A composition profile vertically through this
structure is shown in Fig. 7.27. This work illustrates the feasibility of using
LEAP for both competitive analysis and failure analysis within the commercial
LED market.
7.4.2.3 LEAP Analysis of Solar Cells
Several compound semiconductors are candidate materials for mass-produced solar

cells. Many materials have been used in demonstrations of solar energy conversion
efficiencies in excess of 20 %. However, most of these materials could potentially
reach even greater efficiencies. The following two examples illustrate the role that
APT is playing in this search for improved photovoltaic solar cell materials.
Fig. 7.26 Atom map of the

active component region of
a commercial white LED.
There are four distinct
regions visible; (from the
top) (1) a magnesium-doped
region, (2) an aluminum-
doped electron-blocking
layer, (3) indium quantum
wells, and (4) an indium
superlattice. An
isoconcentration surface of
1 at.% magnesium reveals
clusters of magnesium in
region one while a 4 at.% In
isoconcentration surface
shows the position of
indium quantum wells [84]
Fig. 7.27 Concentration profile from a 20 nm diameter cylindrical volume through the center of
the atom map in Fig. 7.26
7.5 Organics and Biological Materials 227
7.4.2.4 Cadmium Telluride (CdTe)
CdTe is a candidate thin-film solar cell material because it may be readily deposited
on large areas of a wide range of materials including metals, glass, and polymers.
APT data have been collected from a CdTe layer within a CdTe solar cell [85, 86]
that has been prepared with standard CdCl2 and copper treatments [87]. Figure 7.28a
shows the mass spectrum which is one of the most complex APT spectra published
to date. The spectrum contains a number of high mass molecular ions. Figure 7.28b
contains 2D atom maps of CdS, CdCl, and TeS molecular ions, with a color overlay
of CdCl (orange), Cd (black), and Te (dark green). These 2D renderings show
enrichment of sulfur and chlorine at the grain boundary which is suspected to cause
degradation of performance of the solar cells. Quantitative composition profiles
generated from the dashed-box region show peak chlorine concentrations at the
grain boundary to be well over 1 at.% and sulfur concentrations of about 0.5 at.%,
which are very large amounts for impurities. Despite the complexity of the mass
spectrum, the observation of grain boundary segregation was a fairly straightfor-
ward result that provided crucial insight into the processing as it may have affected
the performance of this device.
7.4.2.5 Cu(InGa)Se2 (CIGS)Solar Cells
Thin film solar cells based on CuInSe2 (CIS) and Cu(InGa)Se2 (CIGS) offer
potentially high efficiency and stability combined with a low cost of production.
Improvements in efficiency have been held back by uncertainties in the mechanism
of charge carrier losses at the pn junction created with an overcoat layer of CdS.
Cojocaru-Miredin et al. [88, 89] have analyzed this interface with a CIS layer,
Fig. 7.29. Using a proxigram analysis, they found copper, indium, and selenium in
the CdS layer. More importantly, they found cadmium diffused into the CIS layer
on a 2-nm scale, Fig. 7.29b, in sufficient quantities to create a buried pn
homojunction rather than a heterojunction. This homojunction formation could be
responsible for the high photo-conversion efficiencies of these solar cells.
7.5 Organics and Biological Materials
If APT can be applied to organic materials with similar success to that of its
application to inorganic materials, it would be a major development. To date,
there have been several attempts and progress is being made. However, there has
yet to be an analysis of a monolithic soft biological specimen in an atom probe. This
can be ascribed, in part, to the lack of necessary equipment: a complete cryogenic
preparation, transfer, and analysis capability has yet to be fully realized. Note,
therefore, that each of the results below have a key component in common: they are
Fig. 7.28 (a) APT mass spectrum and (b) 3D atom maps of CdS, CdCl, and TeS molecular ions,
with a color overlay of CdCl (orange), Cd (black), and Te (dark green). These two-dimensional
renderings show enrichment of sulfur and chlorine at the grain boundary. Quantitative composition
profiles that were generated from the dashed-box region show peak chlorine concentrations at the
grain boundary to be well over 1 at.% and sulfur concentrations of approximately 0.5 at.%, which
are very large amounts for impurities. Reprinted from Choi et al. [85] with permission from
Microscopy Today
Fig. 7.29 (a) APT atom maps of copper, indium, selenium, cadmium, and sulfur. The volume size
shown is 18.5 16 45.5 nm3. (b) Proximity histogram with respect to the cadmium 26 at.%
isoconcentration surface and 0.3 nm bin size. Reprinted from Cojocaru-Miredin et al. [88] with
permission from Applied Physics Letters. Copyright (2011) American Institute of Physics
all projects where the specimen preparation was accomplished at room temperature.
Two universities, the Eidgenossische Technische Hochschule Zurich and the Uni-
versity of Michigan each have implemented cryogenic preparation and transfer
capabilities for the LEAP, including a cryostage for FIB. What is still missing is a
cryogenic micromanipulator for a FIB, although there is progress in this area for
electron microscopy [90]. When this technique is applied for atom probe specimen
preparation, we should see an expansion of biological applications for the LEAP.
Kelly et al. recounted the history of attempts to analyze organic materials
[91]. A couple of examples will be given here to illustrate the state of this art.
Fig. 7.30 Typical mass spectrum for C60-doped P3HT polymer, highlighting the rich ion frag-
ment distribution observed for polymer films deposited on sharp carrier tips (inset). Expanded
mass-to-charge-state ratio range showing the large number of C60 ions observed. Reprinted from
Joester et al. [97] with permission from Cambridge University Press
7.5.1 Synthetic Polymers
Synthetic polymers, such as self-assembled monolayer-forming oligomers [9294],

larger-molecular-weight polymers [95], and natural biopolymers [91] made from
assemblies of proteins to make subcellular systems have yielded to APT analysis.
The potential to access nanoscale chemical imaging information on these com-
plexly organized materials promises to open up new understanding of interchain
interactions as well as interfacial interactions which are important in organic device
development [96] and biology.
One concern for APT analysis of polymeric materials is the potential for large
molecular-ion evaporation. The accuracy and precision of the three-dimensional
(3D) reconstruction is partially predicated on ion size. For atomic ions, one
achieves maximum precision, while for large molecular ions, all the atoms com-
posing the molecule are essentially placed at the same position in 3Dno molec-
ular orientation information is collected [92]. As one might expect, the nature of the
bonding in polymers leads to field evaporated mass spectra with a wide variety of
ion sizes. An example of this variety is evident in mass spectra from poly
(3-hexylthiophene) (P3HT) films prepared for APT by different methods [95, 97],
Fig. 7.30. For thin films deposited on carrier tips, large molecular ion fragments are
observed, while for FIB-extracted sections, primarily monatomic and diatomic ion
species are detected. Ion fragments are observed for nearly every integer mass
position except that there are no peaks representing single-carbon ions. For P3AT
films doped with C60 molecules, both fragmented and whole C60 ions were
Fig. 7.31 Mass spectrum for PCBM doped P3HT sample prepared by FIB lift-out extraction from
a spin-coat deposited film. The ion fragment distribution has fewer peaks and chemical species like
individual carbon ions are unambiguously identifiable. The chemical structures for PCBM and
P3HT are shown as well (inset). Reprinted from Joester et al. [97] with permission from Cam-
bridge University Press
observed with the relative abundances varying as a function of analysis conditions

(see Fig. 7.30 inset).
Composite P3HT and [6, 6]-phenyl-C61 butyric acid methyl ester (PCBM) films,
a popular bulk heterojunction material for organic photovoltaic devices such as
solar cells [96], were deposited and prepared via FIB lift-out. Chemically, these are
similar in composition and structure to the C60-doped films described above;
however, the mass spectra collected from these films are starkly different as
shown in Fig. 7.31. Here, carbon is a dominant monatomic ion and even moderate
atomic-mass ions are absent from the mass spectrum. This is consistent with mass
spectra observed from other FIB-prepared polymer specimens including cell biop-
sies where atomic carbon, nitrogen, and oxygen were observed with very few large
mass ion fragments [91].
7.5.2 Chiton Teeth
Among the most advanced analysis of a biological material comes from the recent
work of Gordon and Joester on chiton teeth [98]. Specimens were FIB-extracted
from the cutting edge of the magnetite-based teeth which contain 510 nm diameter
chitin protein fibers. The mass spectra were logical for the materials and showed
clearly distinct peaks from the organic and mineral components. Atom maps of two
analyses are shown in Fig. 7.32 (a and d). The organic-matter-derived ions are
shown to originate from fibers included within the mineral. The total amount of
Fig. 7.32 Atom maps of two analyses using chitin fibers. (a, d) 3D reconstructions and proxigrams.
Two representative samples containing organic fibers that exclusively bind Na+ (ac) or Mg++
(df). For clarity, only approximately 5 % of the Fe/O ions are rendered; the edge of the FOV is
marked (dashed line in panel b). (b, e) Overlay of Na+ (b, red spheres) and Mg++ (e, magenta
spheres) ion positions on carbon concentration maps integrated over the boxed regions indicated in
panels a and d. Some regions of the fibers appear devoid of sodium or magnesium (arrows). (c, f)
Proximity histograms (error bars, 1) of (c Na/C and f) Mg/C across the organicinorganic
interface of fibers indicated by arrows in panels a and d. Interfaces appear graded over 24 nm.
Images courtesy L. Gordon and D. Joester (Northwestern University)
carbon is less than what would be expected from crystalline chitin fibers which may
be a consequence of overlapping peaks in mass spectra, preferential evaporation, or
biological remodeling of fibers during mineralization. Nevertheless, the fiber diam-
eter (510 nm) closely resembles the organic fibers observed in STEM images of
the mineralized tooth [98]. Furthermore, sodium and magnesium clearly co-localize
with organic fibers, Fig. 7.32b, c and e, f, respectively. This work was the first to
show microstructural information from an APT image of a material of biological
origin. It will stand as a portent of great things to come.
7.5.3 Ferritin
The Joester group has also recently analyzed a portion of a single biomolecule:
ferritin. This molecule has been the object of efforts over the past three decades
[99102] to observe its structure in APT precisely because it is a model molecule
that contains about 4,500 iron atoms in an oxide core with a protein shell surround-
ing it. Gordon and Joester [103] were able to prepare isolated molecules by
depositing a sparse layer of ferritin molecules on goldpalladium-coated flat-top
microtip arrays, Fig. 7.33a. After deposition, the specimens were overcoated with
7.6 Composite Structures/Devices 233
Fig. 7.33 (a) Schematic illustration of the preparation of LEAP specimens of ferritin molecules.
A flat-top microtip is coated with goldpalladium (left), a sparse layer of ferritin molecules is
deposited (middle left), the microtip is then encapsulated (middle right), and the FIB is used to
sharpen the resultant microtip (right). (b) Atom map (left) and proxigram (right) for an Fe-rich
core in a ferritin molecule. An isoconcentration surface for 5 iron atoms/nm3 is shown. Note the
elevated iron and oxygen signals inside the isoconcentration surface and the elevated carbon signal
outside the surface. (a) courtesy L. Gordon and D. Joester (Northwestern University). (b) reprinted
from Gordon et al. [103] with permission from Microscopy and Microanalysis
more goldpalladium and the result was sharpened into a LEAP specimen using
FIB techniques. Though the LEAP analysis was not as precise as obtained with
most inorganics, it was consistent with expectations: iron and oxygen in a core with
carbon and nitrogen surrounding it, Fig. 7.33b. A proximity histogram composition
profile about the molecule center shows an expected iron-rich core with carbon
reaching a maximum just outside the core, Fig. 7.33c. It is likely that there is a
significant local magnification effect for the molecule since goldpalladium and the
ferritin molecule likely have different evaporation fields. Nonetheless, work like
this is starting to show that organic materials with observable microstructure may
be analyzed at the atomic scale with APT.
7.6 Composite Structures/Devices
There are many types of complex structures that are candidates for APT analysis.
These structures may contain a variety of material types: metals, semiconductors,
dielectrics, and polymers, in various morphological configurations. Having
reviewed several examples of each type of material analysis above, we will now
look at combinations of these materials as found in such structures as microelec-
tronic devices and oxidized metals.
7.6.1 MetalOxide Interfaces
The simplest composite structure is a bilayer. Most metals are not inert to environ-
mental degradation which is an expensive constraint on their utilization. Oxides
grown on metals form a bilayer where the details of the interface between layers can
be critical to the performance of the part. APT of these interfaces has been fruitfully
applied to learn novel information in many cases.
Zirconium and its alloys have been investigated by a variety of researchers using
APT [104112], often due to the fact that zirconium alloys are used in nuclear
applications for fuel rod cladding. They are exposed to high temperature water or
steam in a radiation environment. The protective oxide coating that forms on
zirconium alloys is crucial to their utility in this application. Transition metal
elements are known to improve the performance of these alloys in the reactor
environment, but the details of the physical mechanism by which this occurs is
not exactly known. Sundell et al. [110] have studied corrosion of Zircaloy-2 in
steam at 400 C which formed a 1.3 m thick oxide in 3 days. APT images of a
metaloxide interface are shown in Fig. 7.34. Two planar features are visible in this
image from the atom distribution pattern of Fe and Ni. Since low-angle grain
boundaries are known to form during oxidation of this alloy, these features are
presumed to be evidence of segregation of iron and nickel to low angle grain
boundaries near this interface. The presence of iron and nickel at boundaries
associated with this interface may have profound effects on the corrosion behavior
of this alloy in practice because oxygen transport is assumed to occur through oxide
grain boundaries.
7.6.2 MOSFET Structures
Inoue and co-authors have recently done substantial work investigating the distri-
bution of dopants in various parts of MOSFETs [4446, 55]. Knowledge of the
detailed dopant distributions in microelectronic devices is becoming more and more
important as the devices get smaller and are using increasingly complex 3D internal
structures. In the n-MOSFET gate region shown at the left in Fig. 7.35, phosphorus
atoms segregate to grain boundaries and polycrystalline siliconoxide interfaces. In
contrast, no segregation of boron atoms is observed in the p-MOSFET gate (right
side of Fig. 7.35) [46]. Arsenic and boron atoms were detected in the source/drain
regions and channel regions, respectively for p-type FET (and vice versa for the
n-type structure). Statistical evaluation of the data suggested that the fabrication
process increases the boron fluctuation but not the arsenic fluctuation, thus produc-
ing more variations from device to device in p-type MOSFETs [46].
Fig. 7.34 LEAP images of the interface between the oxide and metal in a Zircalloy-2 alloy. (a, b)
are the same image rotated by 90 about the long axis. A presumed low-angle grain boundary is
visible in each of the metal and oxide layers. Both nickel and iron are segregated to these
boundaries. Fe is shown as purple spheres, Ni is shown as green spheres, O is shown as blue
dots, and Zr is shown as orange dots. The size of the box is 50 50 140 nm3. Reprinted from
Sundell et al. [110] with permission from Corrosion Science
In current generation devices the entire transistor channel may have on the order
of only a few hundreds of dopant atoms, which leads to potential sources for
variability in performance. To investigate this variability, Takamizawa et al. [55]
have made repeated observations of channel doping in 65 nm node n-MOSFETs by
measuring the dopant concentration ~30 nm below the gate oxide, Fig. 7.36a. Only
transistors having threshold voltage values ranging from 4 to +4 in one million
transistors were selected. This data was compared to the threshold voltage (1 M tran-
sistors) of transistors on with five different boron implant doses as measured with
SIMS, Fig. 7.36b. The APT boron concentrations (vs. measured threshold voltage)
show a correlation to electrical performance when compared to transistors with
known dopant concentration [55].
7.6.3 FinFET Structures
FinFETs [113] are field effect transistors that maximize the area of the gate oxide
by wrapping the oxide and gate structure around a fin-shaped feature, Fig. 7.37a.
Intel started using these structures in its 22 nm node products in 2012 [114]. This
type of structure is actually part of a larger group of devices called multi-gate
transistors. The international technology roadmap for semiconductors (ITRS) refers
to these types of structures as advanced nonclassical CMOS devices [113].
Fig. 7.35 Atom maps of MOSFET gate regions. Left: In the n-MOSFET gate region, phosphorus
atoms segregate to grain boundaries and polycrystalline siliconoxide interfaces. Right: In con-
trast, no segregation of boron atoms is observed in the p-MOSFET gate. Arsenic and boron atoms
were detected in the source/drain regions and channel regions, respectively, for the p-type FET
(and vice versa for the n-type structure). Statistical evaluation of the data suggested that the
fabrication process increases the boron fluctuation but not the arsenic fluctuation, thus producing
more variations from device to device in p-type MOSFETs [46]. Reprinted from Kelly and Larson
[1] with permission from Annual Review of Materials Research
Various authors have recently investigated FinFET-type devices using APT.

Figure 7.37 shows both a schematic and an APT data reconstruction of an entire
high- metal gate FinFET structure (100 nm in height and 20 nm in width) starting
with NiSi (near the top of the image) and ending near the bottom of the fin (bottom
of the image). This structure was studied by Gilbert et al. [54], who found signif-
icant segregation of boron at the TiNSi interface at the top of the fin that was not
present at the same interface on the device sidewalls. Gilbert et al. also reported
several observations of 3D reconstruction artifacts of such structures, which high-
lights the need for improvements in this area [115118].
In FinFETs, as in MOSFETs, device performance may be reduced by an
inhomogeneous distribution of dopants. Recently, Kambham et al. [48] have
analyzed the conformality of boron implantation as a function of implantation
angle in FinFET-type structures. They found that for the 45 implantation, the
peak concentration vertically down from the top of the fin is approximately two
times the peak concentration laterally across the fin, Fig. 7.38. However, in an
implantation done at 10 (measured between the plane of the wafer surface and
incoming dopant trajectory), the dopant concentrations were found to be over ten
times different between vertical and lateral profiles. Observations of high
non-conformality of the dopants compare favorably with previous modeling results
from Vandervorst et al. [119].
Fig. 7.36 (a) Cross-

sectional TEM image of a
typical 65 nm node
MOSFET. The dashed
curve shows the
approximate region
analyzed by APT. (b)
Correlation between
threshold voltage (V) and
channel boron
concentrations in the region
just below the gate-oxide.
Closed circles represent the
APT data. Open triangle
data points are threshold
voltage measurements from
five different channel doses
as measured by SIMS.
Images courtesy
H. Takamizawa and
Y. Shimizu, (Tohoku
University)
7.6.4 Commercial Devices Analysis; Intel i5
While APT has been applied to transistor and FinFET-type structures as shown
above, these structures often have been stopped at some point in the fabrication
process in order to accommodate the analysis. There are many situations in semi-
conductor processing where APT analysis of a finished product is desirable;
competitive analysis and failure analysis are two good examples. Only recently
have APT results been obtained from fully processed, off-the-shelf transistor
structures that are part of a finished product [49, 51, 53]. This section presents
data showing the feasibility of APT analysis for fully packaged integrated-circuit
microelectronic devices.
Fig. 7.37 (a) A schematic and (b) an APT data reconstruction of an entire high- metal gate
FinFET structure (100 nm in height and 20 nm in width) starting with nickel silicide (near top) and
ending near the bottom of the fin (bottom). Reprinted from Kelly and Larson [1] with permission
from Annual Review of Materials Research
Fig. 7.38 Composition profiles across and along fins in FinFETs ion implanted at two different
impingement angles as shown in the figure. Note the relative homogeneity of the implanted
concentration at 45 compared with the large variations that resulted from implantation at 10 .
Reprinted from Kambham et al. [48] with permission from Ultramicroscopy
Fig. 7.39 Specimens created from a commercially available 32 nm node CMOS logic device
that was depackaged and then prepared by FIB methods. (a) APT analysis shows a Hf-based
high- gate oxide near the top of the dataset. (b) Regions of SiGe are visible on either side of a
predominantly silicon region. (b, c) Spatial distributions of the Ge, As, B, and C atoms.
Whereas the arsenic is distributed relatively uniformly (with a concentration of 0.016 at.%),
the boron is correlated with the regions of high-germanium content [49], and the carbon atoms
appear to be clustered [53]. (d) A high-angle annular dark-field STEM image [51] (courtesy
W. Lefebvre, Universite de Rouen) of a structure similar to the one analyzed in panels ac. An
undercut of the gate oxide into the channel is apparent, matching the germanium distribution
shown in panel b. Reprinted from Kelly and Larson [1] with permission from Annual Review of
Materials Research
Fig. 7.40 Growth of the number of publications per year referencing 3DAP or atom probe
overlaid with the relative number of LEAP instruments in the field. The implication is that the
proliferation of LEAP instruments and their inherently high productivity rate are behind the upturn
in number of publications per year using APT
A commercially available 32 nm node logic device was depackaged and FIB

preparation was used to create specimens. APT analysis shows a Hf-based high-
gate oxide (Fig. 7.39a) near the top of the dataset. Regions of SiGe are visible on
either side of a predominantly Si channel region (Fig. 7.39b). The spatial distribu-
tions of the arsenic, boron, and carbon atoms are shown in Fig. 7.39b, c. While the
arsenic is distributed relatively uniformly (with a concentration of ~0.016 at.%), the
boron atoms are clearly correlated with the regions of high germanium content [49]
and the carbon atoms appear to be clustered [53].
Figure 7.39d shows a high-angle annular dark-field transmission electron
microscopy image [51] (image courtesy W. Lefebvre, Univ. Rouen) of a similar
structure to the one analyzed in parts a through c of Fig. 7.39. There is an apparent
undercut of the gate oxide into the channel which matches the germanium distri-
bution seen in Fig. 7.39b.
The FOV of the atom probe for this device is just adequate to observe portions of
the source and drain on either side of the channel. In future technology nodes, the
transistor characteristic lengths will be smaller and the present FOV achieved will
cover a larger portion of the components outside the channel. Although atom probe
analysis of packaged devices is still developing, we expect that this approach to
specimen preparation and analysis will mature quickly.
References 241
7.7 Conclusions
APT has entered a new stage in its maturation. When some of the authors of this
book were introduced to the technique, there were literally only a handful of people
in the world who could fluently describe the operation of an atom probe. We learned
through years of experience to recognize when an audience needed a primer
in the technique. As years have passed and both LEAP and FIB technology
have coevolved, we have found that fewer and fewer minutes of a presentation
were needed to introduce the technique. The number of users worldwide has now
expanded from dozens to many hundreds and will soon pass thousands. The number
of publications per year referencing APT may be taken as an indication of this
fact as illustrated in Fig. 7.40. Yet we may all consider that APT is still in its
formative years. As a reader of this book, you are part of this growth and the authors
want to thank you and your contributions to this rapidly evolving and very suc-
cessful community. The future will be truly exciting and it will be observed at the
atomic scale.
References
1. Kelly, T.F., Larson, D.J.: Atom probe tomography 2012. Ann. Rev. Mater. Res. 42, 131
(2012)
2. Srinivasan, R., Banerjee, R., Hwang, J.Y., Viswanathan, G.B., Tiley, J., Dimiduk, D.M.,
Fraser, H.L.: Atomic scale structure and chemical composition across orderdisorder inter-
faces. Phys. Rev. Lett. 102, 086101 (2009). doi: http://dx.doi.org/10.1103/PhysRevLett.102.
086101
3. Mishin, Y.: Atomistic modeling of the and 0 -phases of the NiAl system. Acta Mater. 52
(6), 14511467 (2004). doi:10.1016/j.actamat.2003.11.026
4. Ardell, A.J., Ozolins, V.: Trans-interface diffusion-controlled coarsening. Nat. Mater. 4(4),
309316 (2005)
5. Gault, B., Cui, X.Y., Moody, M.P., De Geuser, F., Sigli, C., Ringer, S.P., Deschamps, A.:
Atom probe microscopy investigation of Mg site occupancy within delta precipitates in an
AlMgLi alloy. Scripta Mat. 66, 903906 (2012)
6. Miller, M.K., Longstreth-Spoor, L., Kelton, K.F.: Detecting density variations and
nanovoids. Ultramicroscopy 111(6), 469472 (2011). doi:10.1016/j.ultramic.2011.01.027
7. Schreiber, D.K., Olszta, M.J., Saxey, D.W., Kruska, K., More, K.L., Lozano-Perez, S.,
Bruemmer, S.M.: Examinations of oxidation and sulfidation of grain boundaries in alloy
600 exposed to simulated pressurized water reactor primary water. Microsc. Microanal. 19,
676687 (2013)
8. Xiang, Y., Chitry, V., Liddicoat, P., Felfer, P., Cairney, J., Ringer, S., Kruse, N.: Long-chain
terminal alcohols through catalytic CO hydrogenation. J. Am. Chem. Soc. 135(19),
71147117 (2013). doi:10.1021/ja402512r
9. Bagot, P.A.J., de Bocarme, T.V., Cerezo, A., Smith, G.D.W.: 3D atom probe study of gas
adsorption and reaction on alloy catalyst surfaces I: instrumentation. Surf. Sci. 600,
30283035 (2006)
10. Li, T., Bagot, P.A.J., Marquis, E.A., Tsang, S.C.E., Smith, G.D.W.: Characterization of
oxidation and reduction of a palladiumrhodium alloy by atom-probe tomography. J. Phys.
Chem. C 116(7), 47604766 (2012). doi:10.1021/jp211687m
11. Li, T., Bagot, P.A.J., Marquis, E.A., Tsang, S.C.E., Smith, G.D.W.: Characterization of
oxidation and reduction of PtRu and PtRhRu alloys by atom probe tomography and
comparison with PtRh. J. Phys. Chem. C 116(33), 1763317640 (2012). doi:10.1021/
jp304359m
12. Ohkubo T, Chen YM, Kodzuka M, Li F, Oh-ishi K, Hono K (2009) Laser-assisted atom probe
analysis of bulk insulating ceramics. In: MRS Fall Meeting, Materials Research Society,
Boston, MA
13. Chen, Y.M., Ohkubo, T., Hono, K.: Laser assisted field evaporation of oxides in atom probe
analysis. Ultramicroscopy 111(5), 562 (2011)
14. Kuchibhatla, S.V.N.T., Shutthanandan, V., Arey, B.W., Kovarik, L., Wang, C.M., Prosa, T.J.,
Ulfig, R., Thevuthasan, S., Gorman, B.P.: Embedded nanoparticle analysis using atom probe
tomography and high-resolution electron microscopy. Microsc. Microanal. 17(S2), 760761
(2011)
15. Larson, D.J., Alvis, R.A., Lawrence, D.F., Prosa, T.J., Ulfig, R.M., Reinhard, D.A., Clifton, P.
H., Gerstl, S.S.A., Bunton, J.H., Lenz, D.R., Kelly, T.F., Stiller, K.: Analysis of bulk
dielectrics with atom probe tomography. Microsc. Microanal. 14(S2), 12541255 (2008)
16. Marquis, E.A., Yahya, N.A., Larson, D.J., Miller, M.K., Todd, R.I.: Probing the improble:
17. Stiller, K., Viskari, L., Sundell, G., Liu, F., Thuvander, M., Andren, H.O., Larson, D.J., Prosa,
T.J., Reinhard, D.: Atom probe tomography of oxide scales. Oxid. Met. 79, 227238 (2013)
18. Talbot, E., Larde, R., Gourbilleau, F., Dufour, C., Pareige, P.: Si nanoparticles in SiO2: an
atomic scale observation for optimization of optical devices. Eur. Phys. Lett. 87, 26004-
2600126004-26005 (2009)
19. Prosa, T.J., Lawrence, D., Olson, D., Lenz, D., Bunton, J.H., Larson, D.J.: Atom probe
tomography analysis of thick film SiO2 and oxide interfaces: conditions leading to improved
analysis yield. Microsc. Microanal. 17(S2), 750751 (2011)
20. Roussel, M., Chen, W.H., Talbot, E., Larde, R., Cadel, E., Gourbilleau, F., Grandidier, B.,
Stievenard, D., Pareige, P.: Atomic scale investigation of silicon nanowires and nanoclusters.
Nanoscale Res. Lett. 6, 271 (2011). 27110.1186/1556-276x-6-271
21. Bachhav, M., Danoix, R., Danoix, F., Hannoyer, B., Ogale, S., Vurpillot, F.: Investigation of
wustite (Fe1-xO) by femtosecond laser assisted atom probe tomography. Ultramicroscopy
111, 584588 (2011)
22. Chen, Y., Reed, R.C., Marquis, E.A.: As-coated thermal barrier coating: structure and
chemistry. Scr. Mater. 67(9), 779782 (2012)
23. Larde, R., Talbot, E., Vurpillot, F., Pareige, P., Schmerber, G., Beaurepaire, E., Dinia, A., Pierron-
Bohnes, V.: Investigation at the atomic scale of the Co spatial distribution in Zn(Co)O magnetic
semiconductor oxide. J. Appl. Phys. 105(12), 12610713 (2009) doi: 10.1063/1.3152579
24. Larde, R., Talbot, E., Pareige, P., Bieber, H., Schmerber, G., Colis, S., Pierron-Bohnes, V.,
Dinia, A.: Evidence of superparamagnetic Co clusters in pulsed laser deposition-grown Zn
(0.9)Co(0.1)O thin films using atom probe tomography. J. Am. Chem. Soc. 133(5),
14511458 (2011). doi:10.1021/ja108290u
25. Chen, Y.M., Ohkubo, T., Kodzuka, M., Morita, K., Hono, K.: Laser-assisted atom probe
analysis of zirconia/spinel nanocomposite ceramics. Scr. Mater. 61, 693696 (2009)
26. Payne, D.J., Marquis, E.A.: Three-dimensional spatial distribution of Cr atoms in doped
indium oxide. Chem. Mater. 23(5), 10851087 (2011)
27. Li, F., Ohkubo, T., Chen, Y.M., Kodzuka, M., Ye, F., Ou, D.R., Mori, T., Hono, K.: Laser-
assisted three-dimensional atom probe analysis of dopant distribution in Gd-doped CeO2. Scr.
Mater. 63, 332335 (2010)
28. Li, F., Ohkubo, T., Chen, Y.M., Kodzuka, M., Hono, K.: Quantitative atom probe analyses of
rare-earth-doped ceria by femtosecond pulsed laser. Ultramicroscopy 111(6), 589594
(2011)
References 243
29. Kirchhofer, R., Teague, M.C., Gorman, B.P.: Thermal effects on mass and spatial resolution
during laser pulse atom probe tomography of cerium oxide. J. Nucl. Mater. 436, 2328 (2013)
30. Oberdorfer, C., Schmitz, G.: On the field evaporation behavior of dielectric materials in three-
dimensional atom probe: a numeric simulation. Microsc. Microanal. 17, 1525 (2011)
31. Tsukada, M., Tamura, H., McKenna, K.P., Shluger, A.L., Chen, Y.M., Ohkubo, T., Hono, K.:
Mechanism of laser assisted field evaporation from insulating oxides. Ultramicroscopy 111
(6), 567570 (2011)
32. Mazumder, B., Vella, A., Deconihout, B., Al-Kassab, T.: Evaporation mechanisms of MgO in
laser assisted atom probe tomography. Ultramicroscopy 11(6), 571575 (2011)
33. Karahka, M., Kreuzer, H.J.: Field evaporation of oxides: a theoretical study. Ultramicroscopy
132, 5459 (2013) doi: http://dx.doi.org/10.1016/j.ultramic.2012.10.007
34. Tamura, H., Tsukada, M., McKenna, K.P., Shluger, A.L., Ohkubo, T., Hono, K.: Laser-
assisted field evaporation from insulators triggered by photoinduced hole accumulation.
Phys. Rev. B 86, 195430, 195431195436 (2012)
35. Snoeyenbos, D., Jercinovic, M.J., Reinhard, D.A., Hombourger, C.: Characterization of
minerals of geochronological interest by EPMA and atom probe tomography.
Am. Geophys. Union Abstr. V23C-2830 (2012)
36. Valley, J.W., Cavosie, A.J., Ushikubo, T., Reinhard, D.A., Snoeyenbos, D., Lawrence, D.,
Larson, D.J., Clifton, P.H., Kelly, T.F., Strickland, A., Wilde, S., Moser, D.: Radiogenic
dating at the atomic scale in 4.4Ga Zircons. Nature Geoscience in press (2013)
37. Heck, P.R., Stadermann, F.J., Isheim, D., Auciello, O.H., Daulton, T.L., Davis, A.M., Elam,
J.W., Floss, C., Hiller, J., Larson, D.J., Lewis, J., Mane, A., Pellin, M.J., Savina, M.R.,
Seidman, D.N., Stephan, T.: Atom-probe analysis of nanodiamonds from allende. Meteoritics
Planet. Sci. in press (2013)
38. Lewis, R.S., Anders, E., Draine, B.T.: Properties, detectability and origin of interstellar
diamonds in meteorites. Nature 339, 117121 (1989)
39. Coplen, T.B., Bohlke, J.K., De Bievre, P., Ding, T., Holden, N.E., Hopple, J.A., Krouse, H.R.,
Lamberty, A., Peiser, H.S., Revesz, K., Rieder, S.E., Rosman, K.J.R., Roth, E., Taylor, P.D.
P., Vocke, R.D., Xiao, Y.K.: Isotope-abundance variations of selected elements(IUPAC
Technical Report). Pure Appl. Chem. 74, 19872017 (2002)
40. Kirchhofer, R., Diercks, D.R., Gorman, B.P., Brennecka, G.: Atomic scale composition
profiling of ferroelectrics via laser-pulsed atom probe tomography and cross-correlative
transmission electron microscopy. Microsc. Microanal. 19(s2), 908981 (2013)
41. Kelly, T., Larson, D.J.: Local electrode atom probes. Mat. Char. 44(12), 5985 (2000)
42. Kelly, T.F., Gribb, T.T., Olson, J.D., Martens, R.L., Shepard, J.D., Wiener, S.A., Kunicki, T.
C., Ulfig, R.M., Lenz, D., Strennen, E.M., Oltman, E., Bunton, J.H., Strait, D.R.: First data
from a commercial local electrode atom probe (LEAP). Microsc. Microanal. 10, 373383
(2004)
43. Moore, J.S., Jones, K.S., Kennel, H., Corcoran, S.: 3-D Analysis of semiconductor dopant
distributions in a patterned structure using LEAP. Ultramicroscopy 108, 536539 (2008)
44. Inoue, K., Yano, F., Nishida, A., Tsunomura, T., Toyama, T., Nagai, Y., Hasegawa, M.:
Monolayer segregation of As atoms at the interface between gate-oxide and Si substrate in a
p-MOSFET by 3D atom-probe technique. Appl. Phys. Lett. 92, 103506/103501103503
(2008)
45. Inoue, K., Yano, F., Nishida, A., Takamizawa, H., Tsunomura, T., Nagai, Y., Hasegawa, M.:
Dopant distribution in gate electrode of n- and p-type metal-oxidesemiconductor field effect
transistor by laser-assisted atom probe. Appl. Phys. Lett. 95, 043502 (2009)
46. Inoue, K., Yano, F., Nishida, A., Takamizawa, H., Tsunomura, T., Nagai, Y., Hasegawa, M.:
Dopant distributions in n-MOSFET structure observed by three dimensional atom probe
microscopy. Ultramicroscopy 109(12), 14791484 (2009)
47. Lauhon, L.J., Adusumilli, P., Ronsheim, P., Flaitz, P.L., Lawrence, D.: Atom-probe tomog-
raphy of semiconductor materials and device structures. MRS Bull. 34(10), 738743 (2009)
48. Kambham, A.K., Mody, J., Gilbert, M., Koelling, S., Vandervorst, W.: Atom-probe for
FinFET dopant characterization. Ultramicroscopy 111(6), 535539 (2011)
49. Larson, D.J., Lawrence, D., Olson, D., Prosa, T.J., Reinhard, D.A., Ulfig, R.M., Clifton, P.H.,
Bunton, J.H., Lenz, D., Olson, J.D., Renaud, L., Martin, I., Kelly, T.F.: Prospects for atom
probe tomography of commercial semiconductor devices. Microsc. Microanal. 17(S2),
752753 (2011)
Instrumentation and Techniques for Semiconductor Nanostructure Characterization, vol.
2, pp. 407477. World Scientific Publishing, London (2011)
51. Larson, D.J., Lawrence, D., Lefebvre, W., Olson, D., Prosa, T.J., Reinhard, D.A., Ulfig, R.M.,
Clifton, P.H., Bunton, J.H., Lenz, D., Renaud, L., Martin, I., Kelly, T.F.: Toward atom probe
tomography of microelectronic devices. J. Phys. 326, 012030 (2011)
52. Panciera, F., Hoummada, K., Gregoire, M., Juhel, M., Bicais, N., Mangelinck, D.: Three
dimensional distributions of arsenic and platinum within NiSi contact and gate of an n-type
transistor. Appl. Phys. Lett. 99(5), 0519111 (2011). doi: 05191110.1063/1.3616150
Kelly, T.F., Lefebvre, W.: From the store shelf to device-level atom probe analysis: an
exercise in feasibility. In: 36th International Symposium for Testing and Failuer Analysis,
San Jose, CA, pp. 189197. ASM International (2011)
54. Gilbert, M., Vandervorst, W., Koelling, S., Kambham, A.K.: Atom probe analysis of a 3D
finFET with high-k metal gate. Ultramicroscopy 111, 530534 (2011)
55. Takamizawa, H., Shimizu, Y., Inoue, K., Toyama, T., Okada, N., Kato, M., Uchida, H., Yano,
F., Nishida, A., Mogami, T., Nagai, Y.: Origin of characteristic variability in MOSFET
revealed by three-dimensional atom imaging. Appl. Phys. Lett. 100, 253504/
253501253503 (2012)
56. Shimizu, Y., Kawamura, Y., Uematsu, M., Itoh, K.M., Tomita, M., Sasaki, M., Uchida, H.,
Takahashi, M.: Atom probe microscopy of three-dimensional distribution of silicon isotopes
in 28Si/30Si isotope superlattices with sub-nanometer spatial resolution. J. Appl. Phys. 106,
076102 (2009)
57. Shimizu, Y., Kawamura, Y., Uematsu, M., Tomita, M., Kinno, T., Okada, N., Kato, M.,
Uchida, H., Takahashi, M., Ito, H., Ishikawa, H., Ohji, Y., Takamizawa, H., Nagai, Y., Itoh,
K.M.: Depth and lateral resolution of laser-assisted atom probe microscopy of silicon
revealed by isotopic heterostructures. J. Appl. Phys. 109, 036102/036101036103 (2011)
58. Shimizu, Y., Takamizawa, H., Kawamura, Y., Uematsu, M., Toyama, T., Inoue, K., Haller, E.
E., Itoh, K.M., Nagai, Y.: Atomic-scale characterization of germanium isotopic multilayers
by atom probe tomography. J. Appl. Phys. 113, 3 (2013). doi:10.1063/1.4773675
59. Moutanabbir, O., Isheim, D., Seidman, D.N., Kawamura, Y., Itoh, K.M.: Ultraviolet-laser
atom-probe tomographic three-dimensional atom-by-atom mapping of isotopically modu-
lated Si nanoscopic layers. Appl. Phys. Lett. 98, 013111/013111010113/013113 (2011)
60. Perea, D.E., Lensch, J.L., May, S.J., Wessels, B.W., Lauhon, L.J.: Composition analysis of
single semiconductor nanowires using pulsed-laser atom probe tomography. Appl. Phys. A
85(3), 271275 (2006)
61. Canham, L.: Gaining light from silicon. Nature 408, 411412 (2000)
62. Gnaser, H.: Atom probe tomography of Si nanocrystals embedded in silicon oxide. Paper
presented at the SiSS-15, Seikei University, Tokyo
63. Roussel, M., Talbot, E., Gourbilleau, F., Pareige, P.: Atomic characterization of Si
nanoclusters embedded in SiO2 by atom probe tomography. Nanoscale Res. Lett. 6,
164 (2011)
64. Galtrey, M.J., Oliver, R.A., Kappers, M.J., Humphreys, C.J., Stokes, D.J., Clifton, P.H.,
Cerezo, A.: Three dimensional atom probe studies of an InxGa1xN/GaN multiple quantum
well structure: assessment of possible indium clustering. Appl. Phys. Lett. 90, 0619031
(2007)
References 245
65. Gorman, B.P., Norman, A.G., Yan, Y.: Atom probe analysis of IIIV and Si-based semicon-
ductor photovoltaic structures. Microsc. Microanal. 13, 493502 (2007)
66. Galtrey, M.J., Oliver, R.A., Kappers, M.J., McAleese, C., Zhu, D., Humphreys, C.J., Clifton,
P.H., Larson, D.J., Cerezo, A.: Compositional inhomogeneity of a high-efficiency
InxGa1xN based multiple quantum well ultraviolet emitter studied by three dimensional
atom probe. Appl. Phys. Lett. 92, 041904 (2008)
67. Galtrey, M.J., Oliver, R.A., Kappers, M.J., Humphreys, C.J., Clifton, P.H., Larson, D.J.,
Saxey, D.W., Cerezo, A.: Three-dimensional atom probe analysis of green- and blue-emitting
InxGa1xN/GaN multiple quantum well structures. J. Appl. Phys. 104, 013524 (2008)
68. Galtrey, M.J., Oliver, R.A., Kappers, M.J., McAleese, C., Zhu, D., Humphreys, C.J., Clifton,
P.H., Larson, D.J., Cerezo, A.: Atom probe reveals the structure of InxGa1xN based
quantum wells in three dimensions. Phys. Stat. Sol. B 245, 861867 (2008)
69. Muller, M., Cerezo, A., Smith, G.D.W., Chang, L., Gerstl, S.S.A.: Atomic scale character-
ization of buried InxGa1xAs quantum dots using pulsed laser atom probe tomography.
Appl. Phys. Lett. 92, 233115 (2008)
70. Kodzuka, M., Ohkubo, T., Hono, K., Matsukura, F., Ohno, H.: 3DAP analysis of (Ga, Mn)As
diluted magnetic semiconductor thin film. Ultramicroscopy 109(5), 644648 (2009)
71. Prosa, T.J., Clifton, P.H., Zhong, H., Tyagi, A., Shivaraman, R., DenBaars, S.P., Nakamura,
S., Speck, J.S.: Atom probe analysis of interfacial abruptness and clustering within a single In
(x)Ga(1x)N quantum well device on semipolar (1011) GaN substrate. Appl. Phys. Lett.
98(19), 191903191905 (2011). doi:19190310.1063/1.3589370
72. Tomiya, S., Kanitani, Y., Tanaka, S., Ohkubo, T., Hono, K.: Atomic scale characterization of
GaInN/GaN multiple quantum wells in V-shaped pits. Appl. Phys. Lett. 98(18), 181904
(2011). doi:10.1063/1.3585118
73. Bennett, S.E., Saxey, D.W., Kappers, M.J., Barnard, J.S., Humphreys, C.J., Smith, G.D.W.,
Oliver, R.A.: Atom probe tomography assessment of the impact of electron beam exposure on
In(x)Ga(1x)N/GaN quantum wells. Appl. Phys. Lett. 99(2), 0219061 (2011). doi:
02190610.1063/1.3610468
74. Tomiya, S., Kanitani, Y., Tanaka, S., Ohkubo, T., Hono, K.: Atomic scale characterization of
GaInN/GaN multiple quantum wells in V-shaped pits. Appl. Phys. Lett. 98(18),
1819041 (2011) doi: 18190410.1063/1.3585118
75. Agrawal, R., Bernal, R.A., Isheim, D., Espinosa, H.D.: Characterizing atomic composition
and dopant distribution in wide band gap semiconductor nanowires using laser-assisted atom
probe tomography. J. Phys. Chem. C 115(36), 1768817694 (2011). doi:10.1021/jp2047823
76. Bennett, S.E., Clifton, P.H., Ulfig, R.M., Kappers, M.J., Barnard, J.S., Humphreys, C.J.,
Oliver, R.A.: Mg dopant distribution in an AlGaN/GaN p-type superlattice assessed using
atom probe tomography, TEM and SIMS. In: Walther, T., Nellist, P.D., Hutchison, J.L.,
Cullis, A.G. (eds.) 16th International Conference on Microscopy of Semiconducting Mate-
rials, vol. 209. J. Phys. Conf. Ser. (2010)
77. Holzworth, M.R., Rudawski, N.G., Pearton, S.J., Jones, K.S., Lu, L., Kang, T.S., Ren, F.,
Johnson, J.W.: Characterization of the gate oxide of an AlGaN/GaN high electron mobility
transistor. Appl. Phys. Lett. 98(12), 122103 (2011). doi:12210310.1063/1.3569715
78. Bennett, S.E., Smeeton, T.M., Saxey, D.W., Smith, G.D.W., Hooper, S.E., Heffernan, J.,
Humphreys, C.J., Oliver, R.A.: Atom probe tomography characterisation of a laser diode
structure grown by molecular beam epitaxy. J. Appl. Phys. 111, 053508 (2012)
79. Giddings, A.D., Keizer, J.G., Hara, M., Hamhuis, G.J., Yuasa, H., Fukuzawa, H., Koenraad,
P.M.: Composition profiling of InAs quantum dots and wetting layers by atom probe
tomography and cross-sectional scanning tunneling microscopy. Phys. Rev. B 83, 205308
(2011)
80. Ali, S., Cooley, L.D., McCammon, D., Nelms, K.L., Peck, J., Prober, D., Swetz, D., Timbie,
P.T., Weide, D.V.D.: Planar antenna-coupled transition-edge hot electron microbolometer.
IEEE Trans. Appl. Supercond. 13(2), 184187 (2003)
81. Miller, M.K., Cerezo, A., Hetherington, M.G., Smith, G.D.W.: Atom Probe Field Ion
Microscopy. Oxford University Press, Oxford (1996)
82. Thompson, K., Lawrence, D.J., Larson, D.J., Olson, J.D., Kelly, T.F., Gorman, B.: In situ site-
(2007)
83. Miller, M.K., Russell, K.F., Thompson, K., Alvis, R., Larson, D.J.: Review of atom probe
FIB-based specimen preparation methods. Microsc. Microanal. 13, 428436 (2007)
84. Larson, D.J., Prosa, T., Olson, D., Lefebvre, W., Lawrence, D., Clifton, P.H., Kelly, T.F.:
Atom probe tomography of a commercial light emitting diode. In: 18th International Con-
ference on Microscopy of Semiconducting Materials. J. Phys.: Inst. Phys. Conf. Ser. in press
(2013)
85. Choi, P.-P., Cojocaru-Miredin, O., Abou-Ras, D., Caballero, R., Raabe, D., Smentkowski, V.
S., Park, C.G., Gu, G.H., Mazumder, B., Wong, M.H., Hu, Y.-L., Melo, T.P., Speck, J.S.:
Atom probe tomography of compound semiconductors for photovoltaic and light-emitting
device applications. Microsc. Today 20(3), 1824 (2012). doi:10.1017/S1551929512000235
86. Larson, D.J., Smentkowski, V.S., Reinhard, D.A., Prosa, T.J., Olson, D., Lawrence, D.,
Clifton, P.H., Ulfig, R.M., Kelly, T.F.: Atom probe tomography analysis of grain boundaries
in CdTe. Microsc. Microanal. 18(S2), 928929 (2012)
87. Paudel, N.R., Kwon, D., Young, M., Wieland, K.A., Asher, S., Compaan, A.D.: In: 35th IEEE
PVSC, pp. 10091013. (2010)
88. Cojocaru-Miredin, O., Choi, P., Wuerz, R., Raabe, D.: Atomic-scale characterization of the
CdS/CuInSe2 interface in thin-film solar cells. Appl. Phys. Lett. 98, 103504/103501103503
(2011)
89. Cojocaru-Miredin, O., Choi, P., Wuerz, R., Raabe, D.: Atomic-scale distribution of impuri-
ties in CuInSe2-based thin-film solar cells. Ultramicroscopy 111(6), 552556 (2011). doi:10.
1016/j.ultramic.2010.12.034
90. Ditto, J.J.: Devices and methods for cryo lift-out with in situ probe. U.S. Patent Application
Patent 13/570,127, Accessed 14 Feb 2013
91. Kelly, T.F., Nishikawa, O., Panitz, J.A., Prosa, T.J.: Prospects for nanobiology with atom-
probe tomography. MRS Bull. 34, 744749 (2009)
92. Gault, B., Yang, W., Ratinac, K.R., Zheng, R., Braet, F., Ringer, S.P.: Investigation of self-
assembled monolayer by atom probe microscopy. Microsc. Microanal. 15(S2), 272273
(2009)
93. Prosa, T.J., Kostrna, S.L.P., Kelly, T.F.: Laser atom probe tomography: application to poly-
mers. 50th International Field Emission Symposium, pp. 533534. (2006)
94. Zhang, Y., Hillier, A.C.: Three-dimensional atom probe tomography of oxide, anion, and
alkanethiolate coatings on gold. Anal. Chem. 82(14), 61396147 (2010). doi:10.1021/
ac1009035
95. Prosa, T.J., Kostrna Keeney, S., Kelly, T.F.: Atom probe tomography analysis of poly
(3-alkylthiophene)s. J. Microsc. 237, 155 (2010)
96. Li, H.C., Rao, K.K., Jeng, J.Y., Hsiao, Y.J., Guo, T.F., Jeng, Y.R., Wen, T.C.: Nano-scale
mechanical properties of polymer/fullerene bulk hetero-junction films and their influence on
photovoltaic cells. Sol. Energy Mater. Sol. Cells 95(11), 29762980 (2011). doi:10.1016/j.
solmat.2011.05.039
97. Joester, D., Hillier, A.C., Zhang, Y., Prosa, T.J.: Organic materials and organic/inorganic
heterostructures in atom probe tomography. Microsc. Today 20, 2631 (2012)
98. Gordon, L.M., Joester, D.: Nanoscale chemical tomography of buried organicinorganic
interfaces in the chiton tooth. Nature 469, 194198 (2011)
99. Panitz, J.A., Giaver, I.: Ferritin deposition on field-emitter tips. Ultramicroscopy 6, 36
(1981)
100. Panitz, J.A.: Point-projection imaging of unstained ferritin clusters. Ultramicroscopy 7,
241248 (1982)
References 247
101. Panitz, J.A.: Biomolecular deposition on multiple field-emitter tips. Rev. Sci. Instrum. 56(4),
572574 (1985)
102. Greene, M.E., Kelly, T.F., Larson, D.J., Prosa, T.J.: Focused ion beam fabrication of
solidified ferritin into nanoscale volumes for compositional analysis using time-of-flight
mass spectrometry methods. J. Microsc. 247(3), 288299 (2012). doi:10.1111/j.1365-2818.
2012.03644.x
103. Gordon, L.M., Joester, D.: Towards atom probe tomography of hybrid organicinorganic
nanoparticles. Microsc. Microanal. 19(S2) 952952 (2013)
104. Andren, H.O., Mattsson, L., Rolander, U.: Atom-probe analysis of zircaloy. J. Phys. 47(C-2),
191196 (1986). doi:10.1051/jphyscol:1986228
105. Sano, N., Takeda, K.: Atom probe analysis of Sn in Zr-based alloys. J. Nucl. Mater. 252(12),
6370 (1998). doi:10.1016/s0022-3115(97)00302-4
106. Hudson, D., Cerezo, A., Smith, G.D.W.: Zirconium oxidation on the atomic scale.
Ultramicroscopy 109(5), 667671 (2009). doi:10.1016/j.ultramic.2008.10.020
107. Hudson, D., Smith, G.D.W.: Initial observation of grain boundary solute segregation in a
zirconium alloy (ZIRLO) by three-dimensional atom probe. Scr. Mater. 61(4), 411414
(2009). doi:10.1016/j.scriptamat.2009.04.032
108. Hudson, D., Smith, G.D.W., Gault, B.: Optimisation of mass ranging for atom probe
microanalysis and application to the corrosion processes in Zr alloys. Ultramicroscopy 111
(6), 480486 (2011). doi:10.1016/j.ultramic.2010.11.007
109. Thuvander, M., Andren, H.O.: Methods of quantitative matrix analysis of Zircaloy-2.
Ultramicroscopy 111(6), 711714 (2011). doi:10.1016/j.ultramic.2010.12.008
110. Sundell, G., Thuvander, M., Andren, H.O.: Enrichment of Fe and Ni at metal and oxide grain
boundaries in corroded Zircaloy-2. Corrosion Sci. 65(0), 1012 (2012). doi: http://dx.doi.org/
10.1016/j.corsci.2012.08.061
111. Sundell, G., Thuvander, M., Andren, H.O.: Enrichment of Fe and Ni at metal and oxide grain
boundaries in corroded Zircaloy-2. Corros. Sci. 65, 1012 (2012). doi:10.1016/j.corsci.2012.
08.061
112. Cockeram, B.V., Leonard, K.J., Snead, L.L., Miller, M.K.: The use of a laser-assisted Local
Electrode Atom Probe and TEM to examine the microstructure of Zircaloy and precipitate
structure following low dose neutron irradiation at nominally 358 C. J. Nucl. Mater. 433
(13), 460478 (2013). doi:10.1016/j.jnucmat.2012.10.006
113. Colinge, J.-P. (ed.): FinFET and Other Multi-Gate Transistors. Springer, New York (2008)
114. Markoff, J.: To Enhance Chip Speed, Intel Enters 3rd Dimension. New York Times, New
York (2011)
117. Gault, B., Haley, D., de Gueser, F., Moody, M.P., Marquis, E.A., Larson, D.J., Geiser, B.P.:
Advances in the reconstruction of atom probe tomography data. Ultramicroscopy 111,
448457 (2011)
118. Larson, D.J., Geiser, B.P., Prosa, T.J., Kelly, T.F.: On the use of simulated field-evaporated
specimen apex shapes in atom probe tomography data reconstruction. Microsc. Microanal. 18
(5), 953963 (2012)
119. Vandervorst, W., Jurczak, M., Everaert, J., Pawlak, B.J., Duffy, R., Del-Agua-Bomiquel, J.,
Poon, T.: J. Vac. Sci. Technol. B 26, 396 (2008)
Appendix A: Data File Formats
A number of file formats have been utilized for APT over the years (see [1] for
additional examples). As the field continues to grow, new file and data standards
may be required [2]. This appendix describes the file formats pertinent to LEAP.
The RRAW, STR, RHIT, and ROOT files have formats based on the ROOT
software libraries provided by CERN [3]. These files contain binary objects that
represent either data structures native to the ROOT system (such as histograms and
graphs) or proprietary objects developed by CAMECA. The POS, ePOS, ATO,
and range files are more readily interpretable by users, and their formats are
described in detail below.
RRAW or STR Files (Raw Data)
RRAW files contain the raw detector data and the order of ion collection from the
LEAP as well as the history of various parameters measured or controlled through-
out the acquisition. Raw detector data consists of measurements of the time interval
between the evaporation pulse (voltage or laser) and the detection of event pulses at
either ends of multiple delay lines. The RRAW data is the primary input to the ion
hit finding process and includes an entry for every evaporation pulse and any
associated entries for event pulses. STR files replaced RRAW files in LEAP 4000
Systems starting in 2013 and contain the same information but in a modified format.
General users do not interact with RRAW and STR files, but they can provide useful
diagnostics to service personnel. For additional information on detection and hit
finding see Chap. 3.
DOI 10.1007/978-1-4614-8721-0, Springer Science+Business Media New York 2013
250 Appendix A: Data File Formats
RHIT Files (Hit Data)
RHIT files contain the nonzero entries of the RRAW or the STR file that meet
certain quality criteria. These criteria collectively make up the algorithm known as
hit finding. The RHIT file also contains other pertinent information for each event
such as specimen voltage, pulse number, and temperature. An RHIT file contains a
single interpretation of the hit finding of RRAW or STR data. If the hit finding
criteria change, then alternative RHIT files could be generated.
ROOT Files (Reconstruction Data)
A ROOT file consists of a set of parameters used to convert the RHIT data into a
specimen-space spatial reconstruction. Parameters included in the ROOT file
include evaporation field, k factor, image compression factor, ion detection effi-
ciency, atomic density, specimen shank angle, apex continuity relationships, as well
as calibration factors used for conversion from time-of-flight spectra to mass-to-
charge-ratio (m/n) spectra (see Chap. 5 for a detailed discussion of these parame-
ters). In addition, histograms and graphs created as part of the calibration and
reconstruction processing are stored for later reference. Unlike the RHIT and
RRAW files which are generally created once per acquisition, multiple ROOT
files may be created as per acquisition based on the specific parameters selected
by the user: mass spectrum ranging definitions, the specific reconstruction algo-
rithm, and individual reconstruction parameters. A ROOT file contains all of the
parameters required to create POS or ATO files from an RHIT file.
POS Files (Position Data)
A POS file consists of the reconstructed ion positions and the m/n for each
reconstructed ion as indicated in Table A.1. The POS file is created by applying
the reconstruction algorithms specified in the ROOT files to the RHIT data and is
the minimum set of information (with an appropriate range file described below)
required to perform subsequent data analysis. This file type is typically used for
Table A.1 POS file Data (4 4 bytes)

record format
Ion position (nm) m/n
x y z (Da)
Float Float Float Float
Appendix A: Data File Formats 251
higher level analysis (3D display, compositional analysis, creation of concentration

profiles) and is the most widely used format for APT. The data for each of these
fields are stored in the order of assignment from reconstruction as 4-byte IEEE
Standard 754 single-precision floats in most-significant bit (big-endian) byte order.
It should be noted that many systems read by least-significant bit order (small-
endian) and that byte swapping may be required for successful interpretation
outside of commercial CAMECA software.
RNG Files (Ranging Data)
Range files refer to files that describe a set of mass ranges associated with particular
ion types. A range file applied to the m/n information in a POS file defines the
elemental identity of each of the ions for 3D compositional analysis. Range files are
ASCII (1 byte per character) format files that can be opened, read, and edited with
many simple programs. At a minimum, they describe a lower limit and an upper
limit to a range of bins in the m/n histogram that are identified with a specific ion
type. Range files often include volume and color information as well. A hybrid
version of a range file, the LAWATAP ENV file, includes instrument and recon-
struction information [1].
An example RNG file is shown in Table A.2. The format originated from work at
Oak Ridge National Laboratory [4] and the latest version includes fields defining
ion color, lower/upper m/n limits, and composition (elemental or molecular). IVAS-
generated RNG files are used with software developed for the LEAP instrument and
may contain fields that are incompatible with the original format (ion descriptions
with more than two characters and polyatomic ion for example). These incompat-
ible items are stored after the last carriage return as a comments field that can be
reincorporated into an RRNG file used in IVAS.
Each RNG file (Table A.2) begins with two integers that indicate the number of
atom types and ion ranges (Line 1). Next, each 12-character atom type is followed
by that same 12-character atom type and three values indicating the RGB color of
the atom (Lines 210). Next, a table is included (Lines 1019) which associates
each ion m/n range (minimum and maximum value) with an integer combination of
elements from the atom list in the first section (Lines 19). For example, Line
19 indicates that an ion with m/n between 71.5600 and 72.6640 Da is FeO
(indicated by the entries of 1 under Fe and O and entries of 0 under Ni and
H). After these required fields are additional fields added by IVAS to allow for
proper translation into an RRNG format (see next section). In this example, the FeO
ion is provided with a defined color (Lines 2126).
Table A.2 RNG file example

Line
1 4 9
2 Fe
3 Fe 1.00 0.00 1.00
4 Ni
5 Ni 0.00 0.80 0.00
6 O
7 O 0.00 0.80 1.00
8 H
9 H 0.80 0.80 0.00
10 Fe Ni O H
11 . 53.6940 54.4660 1 0 0 0
12 . 55.6010 58.6890 1 0 0 0
13 . 59.6420 60.4820 0 1 0 0
14 . 60.6860 61.2080 0 1 0 0
15 . 61.6620 62.3430 0 1 0 0
16 . 63.7050 64.2280 0 1 0 0
17 . 15.9730 16.4960 0 0 1 0
18 . 0.9000 1.5590 0 0 0 1
19 . 71.5600 72.6640 1 0 1 0
20
21 polyatomic extension
22 1 1
23 FeO
24 FeO 0.00 1.00 0.00
25 FeO
26 . 71.5600 72.6640 1
RRNG Files (Range Data)
An example RRNG file is shown in Table A.3. First, an ion section is defined (Lines
59) followed by range definitions (Lines 1419). Each range includes the range
minimum and maximum, ion volume, ions atomic composition, and an optional
color. Further descriptions of the required and optional fields are included within
commented lines of the file itself.
ePOS Files (Extended Position Data)
The ePOS (or extended POS) file is the same headerless big-endian binary format as
the POS file, but it includes additional four-byte columns describing detector
information that enable more sophisticated analyses. The format detailed in
Table A.4 includes ion position, m/n, far (uncorrected) time of flight, voltage,
detector hit positions, and information on hit multiplicity. The extended informa-
tion increases the file size but also expands the users ability to analyze correlative
Appendix A: Data File Formats 253
Table A.3 RRNG file example

Line
1 # Comment lines can be included for any complete line entry; empty lines are also allowed.
2 # Ion and range sections must be properly identified with brackets [].
3 # Number identifier is required in each section to indicate the number of ion or range fields.
4 # Ion and range fields must follow Ion_Number and Range_Number formats.
5 [Ions]
6 Number 3
7 Ion1 Al
8 Ion2 H
9 Ion3 O
10
11 # Range definitions require maximum and minimum range values.
12 # Mission vol: fields default to IVAS internal values.
13 # Color: field is optional.
14 Number 7
15 Range1 13.3629 13.8335 vol:0.01661 Al:1 Color:33FFFF
16 Range2 26.7840 27.2700 vol:0.01661 Al:1 Color:33FFFF
17 Range3 28.9310 29.1160 vol:0.01661 Al:1 H:2 Color:FF0000
18 Range4 27.8820 28.1180 vol:0.00000 H:1 Color:CCCC00
19 Range5 0.9590 1.1720 vol:0.00000 H:1 Color:CCCC00
20
21 # The Name: field is optional for giving ranges custom names.
22 # Custom ion names cannot be used to define ion types.
23 Range6 106.1250 213.4110 vol:0.00000 Name: Noise Color:0000FF
24 Range7 42.8160 43.3110 vol:0.04543 Al:1 O:1 Name: AlOLikely Color:00FFFF
evaporation (the spatial or the chemical influences on evaporation sequence) and

perform alternate spatial reconstruction of the data.
Hit multiplicity is recorded in the last two columns of each record. When more
than one ion is recorded for a given pulse, only the first event will have an entry in
the Pulses since last event pulse column. Each subsequent event for that pulse
will have an entry of zero because no additional pulser firings occurred before that
event was recorded. Likewise, the Ions Per Pulse column will contain the total
number of recorded ion events for a given pulse. This is normally one, but for a
sequence of records for a pulse with multiply recorded ions, the first ion record will
have the total number of ions measured in that pulse, while the remaining records
for that pulse will have 0 for the Ions Per Pulse value.
ATO Files (Atom Files)
The ATO file format evolved to support TAP instrumentation and includes four
major versions sharing the same ATO extension. The two contemporary versions
are the version-three (V3) ATO which is associated with the LAWATAP instru-
ments and the FlexTAP ATO format.
254
Table A.4 ePOS file record format

Data (11 4 bytes)
Ion position (nm) m/n TOF DC voltage Pulse voltage Detector position (nm)
x y z (Da) (ns) (kV) (kV) X Y Pulses since last event pulse Ions per pulse
Float Float Float Float Float Float Float Float Float Unsigned Int Unsigned Int
Table A.5 ATO file record format (version 3)

Header (8 1 byte)
NA NA NA NA Version NA NA NA
3
Data (14 4 bytes)
Ion position (angstroms) m/n ClusterID Pulse DC voltage TOF Detector position (cm) Pulse voltage (kV) VVolt FourierR FourierI
x y z (Da) Number (kV) (s) X Y
Float Float Float Float Float Float Float Float Float Float
255
Table A.5 summarizes the LAWATAP V3 ATO file format. The first record is a
header with subsequent records containing information about individual ions. Each
record is stored in big-endian format. The header consists of eight 1-byte entries
where the fifth entry contains the value 0x03. The data section of the file consists of
56-byte fixed-length records recorded in acquisition order. The fourteen 4-byte
floating point columns include ion position, m/n, pulse number, voltage, time of
flight, and detector hit position. Additional fields exist but are not defined for use
with IVAS software. It is worthy to note that the pulse number is a 32-bit floating-
point binary, and unless it is reset to zero every 224 (~16.7 million) pulses (as is
typical for the LAWATAP system) it will lose precision as modern datasets
typically contain additional pulse events.
The FlexTAP version ATO file is similar to a tagged image file format (TIFF)
and contains a 16-byte header followed by a table describing the tags. FlexTAP
ATO files are converted automatically to RHIT files after acquisition of data from a
FlexTAP instrument.
References
2. Miller, M.K.: Proposed XML-based three-dimensional atom probe data standard. Surf. Inter-
face Anal. 36(56), 601605 (2004)
3. Brun, R., Rademakers, F.: ROOT - an object oriented data analysis framework. Nucl Instrum.
Meth. Phys. Res. A 389, 8186 (1997)
4. Miller, M.K.: The ORNL atom probe software package. J. Phys. 47-C2(3), 499504 (1986)
Appendix B: Field Evaporation
Field ionization is the physical mechanism that makes field-ion microscopy

possible. Field evaporation is the physical mechanism that makes atom-probe
analysis possible. These processes are introduced below, but it is important to
note that the theories to describe these processes are still early in their development,
even after 60 years.
Field Ionization and Field-Ion Microscopy
After Muller demonstrated experimental evidence of field ionization using the FIM
in 1951 [1], a theoretical explanation was initiated and developed by Inghram and
Gomer [2] and was continued by Muller and Bahadur [3]. In field-ion microscopy,
imaging gas atoms in the vicinity of the specimen undergo field ionization. Due to
an applied voltage on the specimen, the gas atoms are attracted to the sample
surface by an induced dipole force [4]. The gas atoms impact the specimen and
rebound from the surface, losing a portion of their kinetic energy. This process
occurs repeatedly until the gas atom is ionized by the high field directly above the
most protruding atoms. During ionization, the gas atoms topmost electron tunnels
into the sample surface, and the gas ion is then accelerated away from the specimen
by the electric field (typically of strength 20 V/nm).
Ionization occurs because the potential-energy barrier for the topmost electron is
reduced both by the applied field and by interactions with the specimen surface [5].
The highest probability of tunneling occurs at a distance of ~0.20.5 nm above the
specimen and is greatest in the high-field regions directly above the nuclei of
protruding specimen atoms. Because the gas atoms are preferentially ionized at
these positions, an image is obtained which shows the protruding surface atoms as
bright spots. Thus, an FIM image maps the high-electric-field locations above a
specimen surface. The emission current density is determined both by the tunneling
rate constant and by gas-distribution effects that influence the probability that a gas
atom is in the right place to be ionized [6, 7].
258 Appendix B: Field Evaporation
Introduction to Field Evaporation
Field evaporation (FEV) is the removal of specimen atoms, in the form of ions, from
a surface, as result of applying a very high electric field. This is in contrast to
adsorbate field desorption, which is closely related to field evaporation and refers
to the removal of adsorbed foreign atoms from the specimen surface. The two
processes are basically the same from a theoretical standpoint.
FEV is sometimes a two-stage process, involving first the escape of an atom into
an escape charge state n (i.e., its charge is ne, where e is the elementary positive
charge), and then post-field ionization of the ion, in one or more steps, to a higher
final charge state. The escape step is normally taken to be thermal activation over a
barrier of height Qn, with FEV rate constant kn(F,T ) given by
kn F; T AexpQn F=kB T, B:1
where F is the atom-probe (AP) operating field (as applied to surface atoms that are
at high risk of FEV), T is the thermodynamic temperature, kB is the Boltzmann
constant, and A is the rate constant pre-factor (often taken as equal to the surface-
atom vibration frequency). The rate constant kn is measured in s1. If nhr denotes the
count of specimen surface atoms, within the AP detectors field of view, that are at
high risk of FEV, then the effective operating flux (or rate) R during FEV is given
by the FEV emission equation
R nhr kn F; T fnhr AgexpQn F=kB T: B:2
The value of nhr is not well established, but is sometimes taken as 0.01 layers or
100 atoms. The flux or rate R is measured in layers/s, atoms/s, or ions/s.
In principle, two parameters in basic FEV theory are particularly important for
atom-probe operation: (a) the zero-barrier evaporation field (ZBEF) FE, which is
the operating field at which Qn becomes equal to zero, and (b) the field sensitivity
dQn/dF, for F-values near FE. Practical operating-field values depend both on the
chosen (but often not well known) operating flux and on emitter temperature (which
is usually not well known for laser pulsed AP). Consequently, practical F-values
are not themselves well known. However, in principle they must be slightly or
somewhat less than the ZBEF. Hence, a theoretical estimate of the ZBEF can be
used as a first estimate of the AP operating field. Theoretical ZBEF estimates can
be derived from an atomic-level version of the principle of conservation of energy,
as below.
Knowledge of the field sensitivity dQn/dF, near FE, is needed if it is wished to
predict how, for given flux R, the corresponding operating field FR varies with
temperature T, and consequently how the fractional field reduction (FR%) varies
with temperature. This information is important for correct AP operation. To derive
dQn/dF, it is necessary to know how Qn varies with F, near FE. This, in turn,
requires that a reliable physical model of the FEV escape process be available.
Appendix B: Field Evaporation 259
In the early days of field-ion microscopy, it was assumed that FEV could be
described by the classical image-hump model [8] (for reviews, see [9, 10]), but
tests have repeatedly shown that this model is unable to accurately describe exper-
imental data relating to field sensitivity (e.g., [11]). Hence, it came to be thought that
escape takes place via some form of charge-exchange mechanism [12, 13], in which
escape and ionization take place simultaneously at/near the top of an activation-
energy hump.
It is now believed [14] that FEV escape takes place by a particular form of
charge-exchange, termed the revised charge-draining mechanism. In this, FEV is
described as the evaporation of a partial-ion, in a partially charged state, over a
charge-draining hump. (A partial-ion is an atom that does not have an integral
number of electrons in the region of space associated with the atom. Gauss theorem
implies that atoms at highly charged surfaces must behave like this.)
Charge-draining is an intrinsically quantum-mechanical (QM) phenomenon.
Hence, Qn(F) can be reliably determined only by QM calculations. In the high-
field situation of the AP, it is intensely difficult to make such calculations fully
reliable. Useful FEV calculations have been performed only for tungsten (e.g., [15])
and (perhaps) for aluminum [16]. Consequently, for most materials, data relating to
fractional field reduction needs to be acquired via experiment.
The Prediction of Zero-Barrier Evaporation Field
Simple (approximate) ZBEF prediction can be carried out using the following more
general mathematics. First, obtain an expression for the potential energy (PE) of an
n+ ion, measured relative to the energy level of the base of the partial-ions bonding
well. This PE can be denoted by Wn(x, F), where x is the distance of the ion nucleus
from the specimens electrical surface. (The electrical surface is near the outer edge
of the surface-atom electron charge clouds.) The FEV activation energy is then
given by Qn(F) Wn(xe, F), where the escape distance xe is the distance of the
top of the activation-energy hump from the specimens electrical surface and is a
function of operating field F. The zero-activation-energy condition is then
expressed as

Qn FE Wn xe0 ; FE 0, B:3
where xe0 is the zero-barrier escape distance, i.e., the escape distance
corresponding to the condition Qn 0. The resulting expression, sometimes called
an energy-balance equation, contains three variables {n, xe0, FE}. This equation is
then solved for FE, using auxiliary methods to determine values for xe0 and n, as
described below.
The mathematics just set out is general: the main approximation comes in
finding a suitable simple expression for Wn(x, F). This can be done using three
physical steps. First, calculate the local zero-field activation energy (Q0n) needed
to create an n+ ion at a position x somewhat outside a defined part of the surface,
outside the reach of all short-range ion-surface interactions, in the absence of any
applied field. This is done by the following formal cycle. (1) Remove the bound
surface atom to become a neutral atom in remote field-free space: assume this
takes work 0, where 0 is the zero-field bonding energy of the atom, normally
taken equal to the tabulated sublimation energy. (2) Remove n electrons from the
neutral atom, one by one, leaving them in remote field-free space, well separated
from each other and from the resulting ion: this takes work Hn equal to the sum of
the first n ionization energies of the atom. (3) Take the electrons, one-by-one, to the
emitter Fermi level, via paths that all pass through the position x somewhat
outside the surface, and then move the ion to position x: in this combined process,
an amount of energy n is retrieved, where is the local work function of the part
of the specimen surface that faces the external ion. It follows that
X
n
Q0n 0 Hn n 0 Is n, B:4
s1
where Is is the sth ionization energy. This quantity Q0n is not really an activation
energy as usually defined, hence it is sometimes called the thermodynamic term
and denoted by Kn0.
In the second step, move the ion closer to the surface, to a new position x within
the range of attractive ion-surface forces but outside the range of short-range
repulsive forces, and assume (somewhat arbitrarily and not necessarily correctly)
that the strongest attractive force is due to image forces. This step introduces an
image-PE contribution n2e2/160x, where 0 is the electric constant. In the third
step, apply an electric field of mean strength F: this introduces an electrostatic PE
contribution neFx. The final (approximate) expression for Wn(x, F) is
Wn x; F Q0n neFx n2 e2 =160 x B:5
and the corresponding (approximate) energy-balance equation is
Q0n neFE xe0 n2 e2 =160 xe0 0: B:6
Now ask what is the maximum field FnM that (for a given value of n) can be
predicted from this equation. Applying the condition dFE/dxe0 0 finds the max-
imum field as

FnM 40 =n3 e3 Q0n 2 0:6944616 V nm1 eV2 Q0n 2 =n3 : B:7
This field corresponds to an escape distance xnM given by

1=2
xnM ne=160 FnM 2n2 e2 =160 Q0n : B:8
Equation (B.7) has been called Mullers formula, because a formula of this type
(albeit incorrect in detail) was first given by Muller [8]. (In fact, both Gomer and
Swanson [13] and Brandon [17] contributed to giving the correct Muller-type
formula for an n+ ion.) Correspondingly, FnM can be called the Muller escape
field for an n+ ion, and xnM the Muller escape distance for an n+ ion.
Evaluation of values for xnM suggests that, for many species, the distance xnM is
noticeably different from estimates (made in other ways) of the distance xb of the
base of the bonding well from the specimens electrical surface. (In particular, when
xnM < xb, the calculation cannot be physically realistic, because strong repulsive
forces are being ignored.) Physically, one expects the position of the base of the
bonding well to be the main factor that influences where escape occurs. Hence, it is
better in principle to set xe0 xb, and rearrange (B.6) into the form

FnE xb Q0n n2 e2 =160 xb =nexb : B:9
This formula, introduced by Tsong [18], was used by Muller and Tsong for their
systematic 1969 calculations [19], in which they set xb equal to the Pauling atomic
radius [20] for the element of interest.
Equation (B.9) can also be used to define a correction factor n(xb) by

FnE xb n xb FnM : B:10
It has been shown [21] that n(x) is a slowly varying function of distance; explicitly,
calculations for 55 elements (Forbes, unpublished work) have shown that for many
elements n(xb) is in the range: 0.90 n(xb) 1. Consequently, for a charge-
exchange-type model, the numerical value of FnM can be used as a satisfactory

estimate for FnE xb . Values of FnM were first systematically calculated by Tsong
[22] and are tabulated in many AP textbooks, usually for n 1, 2, 3.

The value of the parameter FnE xb depends on n. Brandons criterion [17] is to
take the lowest of these values as the theoretical estimate of zero-barrier evapora-
tion field FE, and the corresponding value of n as the theoretically predicted escape
charge state. Elementary FEV theory (as found in APT textbooks) normally uses the
Muller-Brandon approach, in which ZBEF estimates are obtained by applying
Brandons criterion to the Muller escape fields as determined above.
Note that the derivation of Mullers formula given above relates primarily to the
energetics of the escape process, does not involve the assumption that escape takes
place via the classical image-hump mechanism, but does show that Mullers
formula is applicable (as an approximation) to charge-exchange-type escape mech-
anisms [14, 21].
Mullers formula was originally derived mathematically in a different way, by
assuming that escape takes place via the classical image-hump mechanism, and was
originally called the image-hump formula. The continued use of this name is
unfortunate, because it can give the spurious (logically false) impression that the
success of Mullers formula justifies the use of the image-hump (IH) model in the
Fig. B.1 The melting temperature of various elements versus their theoretical and experimentally
determined evaporation fields. This chart allows a zeroth-order estimate of the evaporation field
for an element or compound if one is unavailable from the Muller-Brandon approach
context of field sensitivity, whereas the facts of the matter are that the IH-model
field-sensitivity predictions have failed badly (quantitatively) on almost every
occasion in the last 70 years on which they have been seriously tested. There can
also be a spurious (logically false) conclusion that because the image-hump model
fails badly, the so-called image-hump formula is not useful. In order to avoid both
these types of false impression/conclusion, the name Mullers formula is to be
preferred. However, it does need to be emphasized that the Muller-Brandon
approach estimates the ZBEF and that it is an approximation. For many materials,
Mullers formula appears to be satisfactory as a simple first approximation, but we
do not currently know how good it is. Significant work remains to develop a theory
of FEV escape that is predictive and applicable for both metals and nonmetals in
both laser and voltage pulse modes.
If the user desires an estimate of an evaporation field for an element or com-
pound, a zeroth-order estimate may be obtained by plotting the melting temperature
of various elements versus their theoretical and experimentally determined evapo-
ration field as shown in Fig. B.1. Although there is significant scatter in the
phenomenological power law fit (R2 ~ 0.7) shown on the plot, this method may
provide at least a starting point for the evaporation field if such a value is
unavailable. It is clear that for compounds such an estimate will be more accurate
for miscible alloys as opposed to ordered phases.
Field Evaporation Charge States and Post-field-ionization
The Muller-Brandon approach, as described above and as implemented by Tsong

[22], predicts that most materials will escape in a singly charged or doubly charged
state (with a few of the most refractory metals predicted to escape as 3+ ions).
Although there are individual exceptions, this prediction does agree surprisingly
well with many experimental FEV results, for the low-temperature cases where the
activation energy Qn has to be very low [5, 10]. However, in some cases, higher
charge states are found (as high as 6+ for tungsten at very high fields [23]).
Following definitive experimental work by Ernst [24], it is now generally accepted
that these higher charge states are formed by a process called post-field-ionization
(PFI) (or simply post-ionization), in accordance with a theory developed by
Haydock and Kingham [25, 26].
For PFI to occur, it is necessary both for the energetics to be right and for the
probability of PFI to be high enough. Examination of the ionic PE expression, (B.5),
shows that as x gets large, the energy curves for high charge-state values must
eventually cross those for lower charge-state values [9]. This implies that escaping
ions can be post-field-ionized at some distance from the specimen surface.
Kingham has concluded that it is likely that, for most (probably all) metals, PFI
can occur and field evaporation can be a multi-stage process. He has performed
calculations predicting charge state as a function of applied field for a large number
of elements [5]. These curves can also be found in APT textbooks (e.g., [10, 27]).
Comparisons of theory and experiment, for how charge state varies with oper-
ating field, show good agreement for many metals. In suitable cases, where a
species is found to field evaporate in two adjacent charge states, and one can be
reasonably sure that the higher charge state is formed by PFI, measurements of the
counts of ions in the two charge states can, via the Kingham diagrams, be used to
estimate the field over the part of the emitter that generates the measured ion counts.
Field Evaporation Induced by Laser or Electron Pulses
From the above discussion it is clear that the rate of field evaporation can be
enhanced simply by increasing the applied field on the specimen. Field evaporation
may also be significantly enhanced by irradiation of the specimen by a beam of
photons or electrons. It is exactly this effect that provides the operating mechanism
for the pulsed-laser atom probe [28]. Tsong [5] proposed that the enhancement in
field evaporation could be caused by two effects. The first, and simplest, effect is
emitter heating by the particle beam. The inelastic interaction of the particles with
the emitter produces a temperature rise in the specimen, and hence an increase in
the FEV rate constant in accordance with (B.1). With a reliable model for Q(F)
(or reliable empirical data for the FEV operating flux in use), the change in the field
evaporation characteristics could be calculated.
The second possibility is a direct particle-induced electronic excitation. An

incoming particle (photon or electron) will have a probability of exciting an electron
of a specimen surface atom to a higher energy state (thereby allowing field ionization
from this excited state) or of ionizing the atom directly. If this occurs for a metal
atom, the ion would usually be quickly re-neutralized as a result of electron transfer
from an energy level in the metal conduction band [5], because the filled conduction-
band levels in the metal are higher than the energy level of the electron that was
ejected to form the ion.
Tsong has found no effect of direct electronic excitations by photons in pulsed-
laser stimulated field desorption from metal surfaces if the photon intensity is not
excessively high [5]. This result was deduced by comparing the energy distributions
of pulsed-laser field desorbed ions with those found in ordinary field desorption. An
excess energy tail, or a separate peak, is evidence of direct excitation of the surface
atom by incoming particles before evaporation occurs. It should be noted, however,
that if the laser power is very intense, an excess energy can be detected irrespective
of whether the specimen is nonmetallic [5].
Work continues on understanding photon/specimen interactions, but current
understanding is that the dominant effect is as follows. The laser energy creates a
very brief rise in specimen temperature for both metals and nonmetals, thereby
inducing field evaporation [29]. The existence of different specimen cooling times
(as a result of differences in shape and/or thermal conduction properties) seems to
be the main reason why different specimens demonstrate differences in mass
resolving power [30].
We thank Dr. Richard G. Forbes for assistance with the preparation of this
Appendix.
References
1. Muller, E.W.: Z. Phys. 131, 136142 (1951)

2. Inghram, M.G., Gomer, R.: Mass spectrometric analysis of ions from the field microscope.
J. Chem. Phys. 22, 12791280 (1954)
3. Muller, E.W., Bahadur, K.: Field ionization of gases at a metal surface and the resolution of
the field ion microscope. Phys. Rev. 102(3), 624634 (1956)
4. Tsong, T.T., Muller, E.W.: Field adsorption of inert-gas atoms on field ion emitter surfaces.
Phys. Rev. Lett. 25(14), 911913 (1970)
5. Tsong, T.T.: Atom-Probe Field Ion Microscopy: Field Ion Emission and Surfaces and
Interfaces at Atomic Resolution. Cambridge University Press, Cambridge, UK (1990)
6. Forbes, R.G.: Seeing atoms: the origins of local contrast in field ion images. J. Phys. Appl.
Phys 18, 9731018 (1985)
7. Forbes, R.G.: Field ion imaging old and new. Appl. Surf. Sci. 94/95, 116 (1996)
8. Muller, E.W.: Field desorption. Phys. Rev. 102, 618624 (1956)
Materials Problems. Materials Research Society, Pittsburgh, PA (1989)
11. Forbes, R.G., Biswas, R.K., Chibane, K.: Field evaporation theory: a re-analysis of published
field sensitivity data. Surf. Sci. 114, 498514 (1982)
12. Gomer, R.: Field desorption. J. Chem. Phys. 31, 341345 (1959)
13. Gomer, R., Swanson, L.W.: Theory of field desorption. J. Chem. Phys. 38, 16131629 (1963)
14. Forbes, R.G.: Field evaporation theory: a review of basic ideas. Appl. Surf. Sci. 87, 111
(1995)
15. Wang, L.C., Kreuzer, H.J.: Kinetic theory of field evaporation of metals. Surf. Sci. 237,
337346 (1990)
16. Sanchez, C.G., Lozovoi, A. Y., Alavi, A. Field evaporation from first principles. Mol. Phys.
102, 10451055 (2004)
17. Brandon, D.G.: The structure of field-evaporated surfaces. Surf. Sci. 3, 118 (1964)
18. Tsong, T.T.: On the mechanism of field evaporation. Surf. Sci. 10, 102117 (1968)
19. Muller, E. W., Tsong, T.T.: Field Ion Microscopy: Principles and Applications. Elsevier,
New York, NY, 1969
20. Pauling, L.: The Nature of the Chemical Bond. Cornell University Press, Ithaca, NY, 1960
21. Forbes, R.G.: A new formula for predicting low temperature evaporation field. Appl. Phys.
Lett. 40, 277279 (1982)
23. Muller, E.W., Krishnaswamy, S.V.: High ion charges in field-evaporating 5d transition
metals. Phys. Rev. Lett. 37, 10111014 (1976)
24. Ernst, N: Experimental investigation on field evaporation of singly and doubly charged
rhodium. Surf. Sci. 87 469482 (1979)
25. Haydock, R., Kingham, D.R.: Post-ionization of field-evaporated ions. Phys. Rev. Lett. 44,
15201523 (1980)
27. Gault, B, Moody, M.P., Cairney, J.M., Ringer S.P.: Atom Probe Microscopy. Springer,
New York, NY, 2012
28. Kellogg, G.L., Tsong, T.T.: Pulsed-laser atom-probe field-ion microscopy. J. Appl. Phys.
51(2), 11841194 (1980)
29. Bunton, J.H., Olson, J.D., Lenz, D., R., Kelly, T.F.: Advances in pulsed-laser atom probe:
(2007)
30. Vella, A., Mazumder, B., Da Costa, G., Deconihout, B.: Field evaporation mechanism of bulk
oxides under ultra fast laser illumination. J. Appl. Phys. 110, 044321 (2011)
Appendix C: Reconstruction Geometry
Overview
In this appendix several geometric exercises are used to illustrate the definitions
essential for spatial reconstructions by using and building on approaches previously
established in the literature [14]. We start by defining common terms that are
relied upon later, and then calculations are developed for the lateral projection and
the volume increment. We finish by showing a calculation of the shank angle from
the voltage curve. In all of these calculations the apex is assumed to be spherical
and situated on a conical shank. We try to follow the convention that distances on
the scale of the whole instrument will be in upper case (flight path LDet for example)
while local properties of the tip will be in lower case (rTip for example).
Image Compression
The projection model is illustrated in Fig. C.1 and assumes that a single projection
point exists at a distance mrTip beyond the center of the spherical apex, where
1 + m is known as the image compression factor. Assuming that an ion is
launched normal to the sphere surface (with angle Launch), the effect of is to
reduce the projection angle to Det and with it the magnification.
The reconstruction process is the task of transforming a detector position into a
launch position on the tip, rLaunch. By considering the detector/flight path triangle,
we know the compressed angle:
LDet RDet
cos Det q and sin Det q : C:1
L2Det R2Det L2Det R2Det
268 Appendix C: Reconstruction Geometry
Fig. C.1 Tip geometry

relevant for image
compression
We will approach this by trying to solve for

r Launch
tan Launch : C:2
zS
We start by looking at rLaunch, inside the tip:

r Launch
sin Det C:3
s
s can be computed using the law of cosines and writing:
r 2Tip m2 r 2Tip s2 2mr Tip s cos Det :
Gathering s terms on the left we get
s2 2mr Tip s cos Det r 2Tip m2 r 2Tip :
Completing the square gives us
s2 2mr Tip s cos Det m2 r 2Tip cos 2 Det r 2Tip m2 r 2Tip m2 r 2Tip cos 2 Det
Appendix C: Reconstruction Geometry 269
and we find
p
s r Tip m cos Det 1 m2 sin 2 Det :
Because it will be useful to us later we define

p
m cos Det 1 m2 sin 2 Det C:4
and we write
s r Tip C:5
r Launch r Tip sin Det : C:6
Using the Pythagorean theorem we can write

2
zS mrTip r 2Launch s2
Substituting in our expressions for s and rLaunch we find
zS r Tip cos Det m C:7
so substituting (C.6) and (C.7) back into (C.2) we obtain
sin Det
tan Launch : C:8
cos Det m
To evaluate the small-angle limit of this expression we will substitute back in for
the definition of to write
p
m sin Det cos Det sin Det 1 m2 sin 2 Det
tan Launch p :
m sin 2 Det cos Det 1 m2 sin 2 Det
We make the approximations sin (Det) Det and cos (Det) 1. Noting that
we have a term proportional to unity in the denominator but must keep Det to first
order in the numerator we get
m 1Det
tan Launch :
1
With a small-angle limit on the (larger) launch angle this reduces to the useful
approximation
Launch Det : C:9

Lateral Magnification
In developing our expression for the compressed trajectory we found (C.6). This
equation is very important in itself as it relates the position on the detector to the
radial launch position on the tip. So we will go back to (C.1)(C.3) in order to
evaluate the ratio RDet/rLaunch.
sRDet
r Launch q
L2Det R2Det
and then
sRDet cos Det

r Launch :
LDet
So our expression for magnification is
RDet LDet LDet

:
r Launch s cos Det cos Det r Tip
This is the expression that applies to large field of view (FOV) local electrode
atom probes. In order to obtain the classical small FOV expression for lateral
magnification we take the small-angle limits for s and cos(Det) to get
RDet r Tip
r Launch
LDet
so the small-angle magnification is
RDet LDet
: C:10
r Launch r Tip
Variation of Tip Radius with Analysis Depth
The radius of the apex is a key component in understanding all aspects of the spatial
reconstruction. It is important to both the radial magnification as well as the depth
increment per atom. As shown in Fig. C.2, we start with the simple case where the
sphere of the apex and the cone of the shank meet with tangential continuity.
Fig. C.2 Tip geometry for variation of tip radius as a function of analysis depth and tangential
continuity
Tangentially Continuous Case
We begin by writing down the expected variation of tip radius with analysis depth
for a fixed shank half-angle, . Note that the angle describes how the radius of the
spherical cap varies with the depth of the spherecone intersection:
dr Cap
tan : C:11
dzShank
We are trying to evaluate how the apex radius varies with analysis depth, zA:
dr Tip
K : C:12
dzA
Just by looking at our definitions, we can write zA zShank zCap and

zCap rTip zS. By looking at the triangle including the sphere center we can
see zS rTip sin(). Substituting into the definition of zA we get
zA zShank r Tip sin 1: C:13
We want the derivative of zA, so we get ready by finding the derivative of zShank
first. Again from the central triangle we have
r Cap
cos
r Tip
Fig. C.3 Tip geometry for variation of tip radius as a function of analysis depth without tangential
continuity
which we can substitute into our definition of tan () to see
dzShank cos
:
dr Tip tan
So now we just take the derivative of the analysis depth to obtain
dzA cos
sin 1
dr Tip tan
Just simplifying, this gives us [1]
sin
K : C:14
1 sin
Non-tangentially Continuous Case
In this more general specimen geometry shown in Fig. C.3, we relax the require-
ment that the apex meets smoothly with the shank. We hold that the tip radius rTip is
just a fixed multiple of the shank radius rCap:
r Tip
f S= : C:15
C r Cap
We will define the angle so that
r Cap 1
cos : C:16
r Tip f S=
C
Writing zA zShank zCap and zCap rTip zS and considering the central
triangle to see that zS rTip sin () we can write
zA zShank r Tip sin 1
We calculate derivatives as before. This time our expression for the derivative of
zShank is
dzShank cos

dr Tip tan
so our expression for the full derivative is
dzA cos
sin 1:
dr Tip tan
Changing from terms of to f we get
dr Tip f S= tan
K , f C
r C:17
dzA 1 f S= tan tan f 2S 1
C =C
Note that the condition for tangential continuity is f S= sec .

C
Analysis Volume Calculation
In this section we calculate the size of the analysis volume incorporating a fixed
shank angle and image compression. Looking at Fig. C.4, one can see that the
acquired analysis volume has three geometric components:
A spherical cap on top defined by the initial tip radius r0Tip and the point of
0
intersection of the sphere with the analysis cone, Z0
0 0
The main analysis cone section from Z0 to Z1
A spherical cap, not acquired, on the bottom defined by the final tip radius r1Tip
0
and the point of intersection of the sphere with the analysis cone, Z1
Rd
qAn
ZZ0 ZZ1 ZAC
Rd
Ld
Z0
Z '0
r0An
Z1 Z '1 r0Tip
r1An
qF
mr0Tip q 0
z
r1Tip
qF
mr1Tip
q 1z
Fig. C.4 Tip geometry for analysis volume calculations
So our analysis volume equation is
V Spec V Cap0 V Cone V Cap1 C:18
zR 3
z3 2R Z3
V Cap R2 z R2 Z :
3 zZ 3 3
For both the top and bottom cap sections, the relevant radius is the current tip
radius rTip, and the Z of interest is zS as defined by (C.7):
zS r Tip cos Det m
so each of the two cap sections can have their volumes written as
!
2 cos Det m3
V Cap r 3Tip cos Det m : C:19
3 3
For the purposes of presentation convenience we define the constant
f cos Det m
Note that this is different from the apex shape parameter f S= . So

C

2 f3
V Cap r 3Tip f :
3 3
Generically, the volume of a cone section is

3 zZ1
z
V Cone tan
2
3 zZ0
In our case, we need to evaluate

V Cone tan 2 An ZZ 31 ZZ30 :
3
It is important to note that An is not the specimen shank angle but rather
describes how the FOV evolves through the analysis depth. From Fig. C.4, we
see that
r 0An
tan An
ZZ 0
where
0
ZZ0 Z AC Z0 :
Recalling (C.6)
r 0An r 0Tip sin Det

by definition of K,f, we have
r 0Tip
ZAC
K , f
Using (C.7),
0
z0 r 0Tip r 0Tip cos Det m
So
r 0Tip
ZZ 0 r 0Tip 1 cos Det m
K , f
Then
r 0Tip sin Det

tan An r0Tip
:
K , f r 0Tip 1 cos Det m
We simplify this to
sin Det K , f
tan An :
1 cos Det K , f
For the purposes of presentation convenience we define the constants
g sin Det
1 cos Det K , f
With this notation change
gK , f
tan An

r 0Tip
ZZ 0
K , f
r 1Tip
ZZ 1
K , f
So our final VCone is

gK , f 2 3 1 3 0 3
V Cone r Tip r Tip
3 K , f
3 3
g2
V Cone r 1Tip r 0Tip :
3K , f
Putting the volume terms all together we find

3 3
g2 3 2 f3 3
V Spec r 1Tip r 0Tip f r 0Tip r 1Tip
3K , f 3 3
Simplifying

g2 1 3 0 3
V Spec 3f 2 f 3 r Tip r Tip : C:20
3 K , f
Volume Increment
During reconstruction, we process a sequence of atoms that have various assigned

volumes. After reconstructing atom i with assigned volume i, we need to adjust
the height of the model specimen as if this volume had been removed uniformly
from the spherical surface of the FOV. Accounting for an expected detector
efficiency , the z adjustment we make is

dzTip i
zTip :
dV Spec
So we need to take the derivative of volume expression (C.20). This is pretty

straightforward since for our standard reconstruction all of the terms are constants
except rTip:
dr Tip
K , f :
dzTip
So the reconstruction volume change per unit volume is
dV Spec
g2 K , f 3f 2 f 3 r 2Tip
dzTip
and our volume increment becomes

zTip i :
g2 K , f 3f 2 f 3 r 2Tip
If we assume that the shank angle is zero, then K,f is zero. In the limit of
small-angle we obtain a per-atom z increment of
i L2Det
zTip :
r 2Tip R2Det 2
Determining Specimen Shank Angle from the Voltage

History Plot
Starting with our formula for the acquisition volume (C.20) we note that all shank
angle dependence is in the K,f terms, so we consolidate the remaining terms into a
placeholder, , and write

3
g2 3
V Spec r 1Tip r 0Tip
3 K , f
where

g2 cos Det 3f 2 f 3 :
Solving for K,f we find
g2 SVR
K , f C:21
3 SVR
where
3 3
r 1Tip r 0Tip
SVR :
V
The volume VSpec is typically accumulated from the assigned atomic volumes,
and the initial and final tip radii are evaluated with R v/KF. So we will
typically use
3
3
vTip v0Tip
1
SVR P
FK 3
Substituting (C.17) into (C.21) and gathering terms of tan () we find
g2 SVR
tan q C:22
f S=C 3 SVR g2 SVR g2 SVR f 2S=C 1
In the case of tangential continuity, satisfies
g2 SVR
sin : C:23
3 SVR g2 SVR
References
1. Blavette, D., Sarrau, J.M., Bostel, A., Gallot, J.: Direction et distance danalyse a la sonde
atomique. Rev. Phys. Appl. 17, 435440 (1982)
2. Bas, P., Bostel, A., Deconihout, B., Blavette, D.: A general protocol for the reconstruction of
3D atom probe data. Appl. Surf. Sci. 87/88, 298304 (1995)
3. Walck, S.D., Buyuklimanli, T., Hren, J.J.: Extended depth profiling with the IAP. J. Phys. 47-
C2(3), 451458 (1986)
Plenum Publishers (2000)
Appendix D: Mass Spectral Performance
Analytical Basis for TOF Spectrometry
As a time of flight (TOF) mass spectrometer, the mass resolving power (MRP) of an
atom probe is one of the critical parameters that describe its performance. On the
surface, this parameter seems straightforward to define and measure. However, as
we shall see below, there are many subtleties that can influence the determination
and reporting of MRP [1].
Determining the Timing and Energy Components

of Time of Flight
A first-order analysis is commonly used to derive the functional dependence of

timing uncertainties in APT. If the potential energy gain of an ion (neV) is equated
to the final kinetic energy (mv2/2) and the limit is considered where potential energy
is immediately converted to kinetic energy so that ion speed can be considered as
constant over the flight path (v L/t), then
1 1 L2
nev mv2 m 2 D:1
2 2 t
where n is the charge state, e is the elementary unit of charge, V is the potential
difference, m is the ion mass, v is the ion speed, L is the ion flight path length, and
t is the ion TOF; then the resulting equation may be solved for TOF as
m L2
t2 : D:2
n 2ev
282 Appendix D: Mass Spectral Performance
For a given ion type, m, n, and e are constants and differentiating (D.2) gives

t L V
t 2 : D:3
2 L V
If a single ion type is measured multiple times with a perfect timer, then the
timing variation that develops is due to voltage (V/V ) or flight length (L/L)
variations. This component of timing precision will be referred to as the physical
timing uncertainty, tP:

L 1 V
tP t t : D:4
L 2 V
Timer electronics achieve a finite precision which will be denoted as tT. This
imprecision is independent of the physical timing imprecision. Furthermore, there
is timing uncertainty associated with the temporal duration of field evaporation
during a pulse. This term is called time-of-departure spread (TODS) and is denoted
as tD. This timing imprecision is also independent of both the timer precision and
the physical timing precision. These three terms may be added in quadrature as
follows:
t2 tP 2 tT 2 tD 2 : D:5
Putting this all together yields

L 1 V 2
t2 t2 tT 2 tD 2 : D:6
L 2 V
Finally, MRP is related to the time resolving power through (D.2) by noting that
m,
t
2 n m D:7
t
n
or
r
m 1 t 1 m 1
MRP L: D:8
m 2 t 2t n 2eV
p
From (D.2) it is apparent that flight time, and hence MRP, will vary with mn,
p
V 1 , and L. Thus if MRP is to be compared from one experiment to the next, it
must be reported at standard values for m/n, V, and L, or the values must be stated
explicitly.
Appendix D: Mass Spectral Performance 283
Sources of V/V
Normalized energy deficits, E/E, which can come from a variety of sources, are
indistinguishable from voltage fluctuations in the accelerating voltage, V/V, and
so all such terms will be grouped in the V/V term. When doing this, it is essential
to remember that a V operating on an ion of n +2 produces a E that is twice
that of an n +1 ion. When a large energy spread exists, flight time differences for
a given ion type get worse with longer total flight time, and longer flight paths do
not lead to improved timing resolution. In laser pulsing, energy deficits may be
small if they are primarily due to thermal energy spreads (0.025 eV at 300 K). Thus
the TOF spread, t, is unchanged during ion flight, and greater MRP can be
achieved by making the flight time longer. This effect was demonstrated dramat-
ically by Liu et al. [2] where they achieved MRP of 4,000 by making very long
flight times in an 8-m-long laser pulsed atom probe.
In voltage pulsing, energy spreads develop because the accelerating voltage is
pulsed by V/V 20 % to induce synchronized field evaporation events. The field
evaporation event actually occurs over a narrower range of voltage which induces
energy spreads of E/E V/V 12 % for n +1 ions. This is a large value
which limits MRP in voltage pulsing to about 200 regardless of flight path length
unless some additional measure is taken (e.g., local electrode or reflectron) to
counter this effect. In LEAPs, the energy spread in field pulsing is reduced com-
pared to a remote electrode geometry because the distance over which any ion is
accelerated is reduced (distance between tip and aperture of local electrode) and the
ion enters the field-free region of the flight path more quickly.
Sources of L/L
The flight length of an ion, L, is generally stated for the distance to the center of the
detector (or the shortest distance to the detector in the case of a reflectron-based
system). In APT, the detector records hit position coordinates on the detector which
makes it possible to correct for geometric differences in flight length with angle
from the central axis such that all flight times are corrected to correspond to L. In the
LEAP, a flight length correction known as the bowl correction was developed by
Oltman [3] and is discussed in detail in Chap. 4.
Sources of tT/t
The imprecision of the clocks used to time ion flights is a hard limit on the timing
precision. However, there are other factors that may limit precision further. The
imprecision introduced by the electronics that pick an arrival time from pulses that
have traversed the delay line anode (DLA) is likely to be greater than the clock
imprecision and is usually the largest limitation imposed by the hardware. Further-
more, this limitation can vary across the face of any DLA. MCPs generally have a
fast response and low time spread relative to the timer precision, so they should not
degrade the precision overall.
Sources of tD/t
Physically, the temporal spread of the evaporation window is the sum in quadrature
of the timing imprecisions. In field pulsing, voltage pulses of a few nanoseconds are
used, but the evaporation events are restricted to the topmost portion of the pulse
and the TODS is correspondingly small for good electrical conductors. The energy
spread terms dominate in field pulsing.
For thermal pulsing, energy variations can be small, and it is the TODS that
usually dominates. The challenge for making small TODS is cooling the specimen
quickly after heating. Certainly, heating the smallest possible volume helps accom-
plish this. It is for this crucial reason that laser-capable LEAPs have always been
designed to focus to the smallest possible spot size. The shortest TODS reported to
date was 114 ps on an aluminum specimen (good thermal diffusivity) with 355 nm
laser light on a LEAP 4000X (<3 m diameter, 4 spot) [1]. Note that this TODS is
much greater than the 12 ps pulse width of the laser which is a confirmation that the
cooling time dominates the TODS. Other materials, such as steels, can have more
shallow skin depths (depth over which energy is absorbed) than aluminum which
can also aid in heating small volumes and result in a reduction in TODS.
Discrete vs. Collective Discrimination
In assessing the impact of MRP on a set of measurements, it is important to

distinguish between whether the goal is image quality or composition determina-
tion. For image quality, the goal is high confidence in the identity of each atom
shown in the image. Ideally, this requires full discrimination and very high MRP for
all peaks in the mass spectrum. For composition determination, the goal is to
correctly determine the number of atoms of each type in some volume of material.
Methods are available to decouple overlapping features in a mass spectrum to
obtain accurate bulk compositions. Thus a distinction is made here between discrete
and collective discrimination.
Discrete discrimination is defined as the ability to discern the identity of each
individual ion in a dataset. In APT, discrete discrimination is required for imaging
so that each ion in the dataset may be correctly assigned an identity. An example of
the importance of discrete discrimination is shown in Fig. D.1. Here the very low
oxygen signal (obtained from a peak at m/n 16 Da) is obscured by the tail of
Fig. D.1 The role that discrete discrimination plays in image quality in APT. In (a) the apparent
oxygen signal in the silicon substrate (right side of image) is due to overlap with the large tail from
the silicon peaks. In (b) with greater MRP comes improved discrete discrimination, so there is no
misidentified oxygen signal in the silicon substrate to the right. Although the peaks may be
decomposed collectively to obtain accurate bulk composition, (b) shows the benefit of discrete
discrimination for imaging as very few oxygen atoms are actually present in the substrate silicon
strong silicon signals (located at m/n 14, 14.5, and 15 Da) in the mass spectrum.
Even though a clear oxygen peak is observed at 16 Da, a majority of the counts in
this mass range are actually silicon ions. Since there is no way to discriminate
individual ions that have different identities but the same m/n, these signals cannot
be discriminated discretely in an image.
Collective discrimination is defined as the ability to discern, as a fraction of
some group, the identity of each subgroup. Collective discrimination would be
accomplished, for example, by decomposition of a single unresolved peak or
overlapping peaks into their constituents using deconvolution or isotopic distribu-
tion information (see Chap. 6). Collective discrimination is important for determi-
nation of bulk composition within a particular volume.
In Fig. D.2, this distinction is illustrated with regard to resolving two peaks.
For collective discrimination, it should be possible to separate two peaks by
deconvolution when there are enough counts to perform a statistically significant
separation. For peaks of about equal height, as shown in Fig. D.2a, discrimination is
possible when they are separated by their FWHM.
For discrete discrimination, the requirements on MRP are much greater. Ideally,
every atom displayed would be unambiguously identified. For discussion purposes,
the minimum requirement is five times that for collective discrimination (5
FWHM), as shown in Fig. D.2b. When peaks are not of equal magnitude or width,
or when the number of counts is low, then these peak separations may need to be
even greater.
A practical example is shown in Fig. D.3. Here the 28Si+ peak and the 14N2+ peak
are just separated enough for good collective discrimination. A composition deter-
mined from this volume would be expected to have good accuracy; however, if
Fig. D.2 Examples of peak overlaps for peaks of equal magnitude. The MRP in (a) is adequate for
collective peak discrimination through deconvolution. The peaks in (b) are much more separated,
relative to their width making discrete discrimination possible
Fig. D.3 Collective discrimination of the 14N2+ and 28Si+ peaks. With sufficient counts, numerical
decomposition is straightforward. The two shaded boxes are placed such that they meet at the
minimum between the peaks to suggest the usual range delimiters that might be used in plotting
these counts in an image
ranges were set as shown by the two shaded areas, then many of the counts in each
range would be misidentified. Discrete display of the hits in this spectrum using
these ranges would lead to a more ambiguous result. The conclusion of this
discussion is that the required MRP depends strongly on whether composition
determination or image display is the analytical goal.
Standard MRP
MRP values can be experimentally measured for a wide range of operating

conditions. As shown in (D.8), MRP depends on m/n, L, and V. However, there
are situations where a single standard value of MRP is useful. For example,
comparison of general instrument performance demands a metric that enables
comparison of mass spectrum quality. In order to compare MRP from one mea-
surement to another, one must either make measurements at the same m/n, L, and
V or refer measurements to the same m/n, L, and V.
For these reasons, MRP at a standard operating condition on a standard specimen
is highly desirable. Aluminum is a convenient material on which to make MRP
measurements because it is easy to prepare in wire form (which makes it universally
applicable to all atom probes), it has a single isotope, and it has a very high thermal
diffusivity (it can produce high MRP in both laser and voltage pulsing modes).
Therefore, we propose that when making claims about instrument performance, a
standard MRP is reported based on the 27Al+ peak. Ideally, the operating voltage,
field of view, crystallographic orientation, and other factors should be standardized
as well.
References
1. Oltman, E., Kelly, T.F., Prosa, T.J., Lawrence, D., Larson, D.J.: Measuring contributions to
mass resolving power in atom probe tomography. Microsc. Microanal. 17(S2), 754755 (2011)
2. Liu, J., Wu, C.-w., Tsong, T.T.: Measurement of the atomic site specific binding energy of
3. Oberdorfer, C., Eich, S.M., Schmitz, G.: A full-scale simulation approach for atom probe
tomography. Ultramicroscopy 128, 5567 (2013), doi: http://dx.doi.org/10.1016/j.ultramic.
2013.01.005
Appendix E: Additional Considerations
for LEAP Operation
Before routine work on a LEAP can be pursued, there are certain instrument
calibrations and preparations that must be made. This appendix summarizes the
essential considerations that should be made before routine operation of the LEAP.
Calibration Files
The conversion of 2D time-of-flight (TOF) data into a mass spectrum is instrument

dependent and, to a lesser extent, specimen dependent. The first data collected from
a calibration specimen is used to determine characteristic voltage and bowl correc-
tions (see Chap. 5) used for subsequent analyses. In order to optimize each mass
spectrum, a unique voltage and bowl correction must be done for each LEAP
analysis; however, the characteristic corrections from the calibration analysis
serve to enable accurate conversion from TOF to mass-to-charge ratio (m/n) and
enables live display of the mass spectrum during data collection. For details on how
to create a calibration file, see the LEAP User Manual.
A properly calibrated LEAP likely has multiple calibration files available for
selection. Choosing the one that best matches the experimental geometry ensures
that a high-quality live mass spectrum is generated. For example, upon initial
installation, a LEAP might have four calibration files: silicon microtip voltage
pulse and laser pulse analyses as well as aluminum wire voltage pulse and laser
pulse analyses. Ideally, each analysis was from a specimen with data acquired over
a large range of acquisition voltages to ensure a stable calibration over a wide
variety of experimental conditions. The characteristic calibration performs best
when a specimen of similar geometry (microtip or wire) and acquisition voltage
is used. Failure to use an appropriate calibration file may cause the live mass
spectrum to have broader mass peaks that drift in m/n position as the acquisition
290 Appendix E: Additional Considerations for LEAP Operation
voltage changes. Fortunately, this only impacts the live mass spectrum quality.
Each individual analysis is properly corrected as a step in data processing (Chap. 5),
so the quality of the reconstructed data will not be impaired by a choice of a
calibration file that does not match the experimental conditions well.
The Local Electrode
Local electrodes (LEs) are critical for good LEAP performance (for a detailed
description of the LE see Chap. 3). Each LE is factory tested to ensure good
ultrahigh-vacuum (UHV) compatibility, correct shape, and high resistance to elec-
tron emission through the voltage operation range of the LEAP. The nature of APT
is to analyze specimens that fail with some significant probability. With each failure
there is opportunity to damage the LE leading to deceased performance (i.e., partial
occlusion of the open aperture itself or electron emission from the LE that results in
a very poor signal-to-noise ratio). A challenge of tool management is to store, test,
and clean LEs in a way that ensures maximum performance and lifetime. New
electrodes from CAMECA have been qualified for electrical performance to 15 kV
with less than 0.2 % data rate (DR) using the procedure outlined in the Local
Electrode Flat-Test Procedure technical bulletin.
Storage
LEs are designed to operate in both voltage pulsed and laser pulsed modes. The
plastic insulator (polyether ether ketone or PEEK) that separates the voltage pulsed
nickel apex from the grounded puck body absorbs moisture when exposed to a
non-UHV environment. Consequently, LEs should be stored in the LEAP when
possible and for best results should reside in the instrument for 37 days before
being used to analyze specimens. A dust-free, dry-pumped vacuum storage unit is
the best alternative to the LEAP, followed by a dust-free desiccator.
Heating
It is well known that removal of adsorbed material is enhanced by heating that

material while under vacuum (commonly referred to as baking) [1, 2]. LEs stored
in the laboratory can be conditioned to the UHV environment of the LEAP more
rapidly when properly baked. An optional LEAP upgrade is the load-lock heater
which is designed to significantly reduce the conditioning period. Those without
a load-lock heater can consider baking in a vacuum oven and placing a hot
carousel of LE pucks immediately into the LEAP to gain some reduction in the
Appendix E: Additional Considerations for LEAP Operation 291
Fig. E.1 Typical

improvement in mass
spectra for water-based ion
contamination originating
from LEs after 4 h under
vacuum. Spectra from an
LE with no load-lock
heating (top) are compared
to an LE which has been
baked for 4 h at 150
C
(bottom). Note the Y-axis
scale offset
conditioning period. Figure E.1 shows the mass spectra from a silicon microtip
analyzed with an LE that has been put into service 4 h after storage outside of
vacuum (unbaked) and another LE that has been baked for 4 h and put into service.
The microtip is made of antimony-doped silicon and, after analyzing through the
surface, should not contain oxygen or hydrogen. The relative number of detected
events associated with contamination (peaks labeled H2O, SiO, and SiH2O) is
reduced from 3.5 to 0.35 % of the total spectrum for the baked LEs.
Generally, electrodes stored in the LEAP do not experience changes in electrical
performance unless put into service. For example, an LE that tests to 10 kV (shows
electron emission) will continue to test to 10 kV even after a month of storage
within the LEAP buffer chamber. Though time in UHV does not affect electrical
performance it does impact the presence of water-bearing and hydrogen-bearing
contamination emitted during acquisition. For any unbaked LEs that experience
contact with the ambient environment, it is recommended that they be placed in
LEAP UHV for a minimum of 37 days before being put into service. Long-term
trends for a group of baked and unbaked LEs are shown in Fig. E.2. A bake of local
electrodes in the load lock for 4 h at 150
C appears equivalent to conditioning LEs
under vacuum without baking for 4 days (note that the unbaked LE results are
shifted by 96 h to highlight the trend). The additional downward curvature of the
baked LE contamination trends immediately after baking for the H2O and SiO
peaks suggests that time in vacuum alone may not be sufficient to achieve the best
cleanliness for LEs.
Fig. E.2 Typical

improvement in total water-
based ion contamination for
baked and unbaked LEs
over a 4-day period in
vacuum
Testing
Electron emission from the local electrode, referred to as electrode turn-on, is the
primary failure mode that limits LE lifetime. Sharp debris or damage to the LE
surface produces a geometry that enables FowlerNordheim electron emission [3].
It is believed that this emission current terminates somewhere on the specimen,
enabling ion emission that at some critical voltage overwhelms the normal imaging
operation of the atom probe (see Fig. E.3). Depending on the type of debris or
damage, either these emission regions on the LE can be mitigated through plasma
treatments or they may require refurbishment of the LE assembly with a new
aperture cone.
Testing for electron emission is illustrated in Fig. E.4. The LE is positioned over
a flat test sample (aligned in this case near a microtip for reference) in distance
similar to normal microtip analysis (~150 m above the test flat). This distance is
based on the measured separation between the LE and its reflection (see the
CAMECA Local Electrode Flat-Test Procedure technical bulletin for instrument-
specific details). LEs are considered safe to use with a specimen up to the voltage
for which the flat-test begins to exhibit significant emission (0.2 % DR or measured
electron current). Minimum electrical turn-on performance guidelines suggest that
the maximum operational voltage for analysis of a wire specimen is 1.52.0 times
the flat-test value and for a microtip specimen is 1.21.4 times the flat-test value due
to the higher electric field imposed upon the electrode surface when running
microtips due to the microtip geometry.
Fig. E.3 Schematic illustrating possible sources of observed LE electron emission. (1)
FowlerNordheim electron emission from LE. (2) Ionization of gas and/or substrate atoms when
electron beam travels to high-voltage surface and strikes substrate surface. (3) Ionized material
follows electric field to terminate on LE or passes through to the detector. (4) High-ion-emission
region (hot spot) observed at detector (as well as broader emission region) indicating significant
LE emission
Fig. E.4 SEM image of LE

alignment used for the
electron emission flat-test.
The separation between
LE-reflected image (left)
and the direct image (right)
is used to estimate the
distance between flat
substrate and the LE
Plasma Treatment
Plasma treatment has been found to improve the electron emission characteristics of
LEs in general. The precise nanoscale modifications to the LE are not well
understood, but the benefits are clear. During the LE plasma treatment process,
the plasma likely removes carbonaceous contamination leading to a cleaner surface.
For recovering damaged or contaminated LEs, plasma treatment can improve flat-
test turn-on voltages by 510 kV in some cases.
Best Known Practices
Once electrodes leave vacuum storage at the factory, they are prone to pick up
small particle contamination of various sorts. Should these particles be of the right
size, shape, and location, they can degrade the electrical performance of the
electrode. This degradation in flat-test performance can almost always be undone
(~95 % of cases) with proper conditioning of the electrode. In order to remove
problematic contamination, the following procedures should be followed for new
electrodes introduced from an ambient environment into the UHV environment of
the LEAP:
1. Immediately before entry into the load lock, electrodes should be physically
treated with a flow of clean, dry, ~15 psig pressure gas (canned dry air or N2) to
remove any loose particle contamination on the electrode surface. Spray gas
directly at the front opening of the electrode (position a few centimeters away
and spray for 12 second bursts).
2. If available, then plasma treat the electrode for 12 h with the integrated plasma
treatment system (check with CAMECA for guidelines regarding gas type,
power, and pressure of operation for the plasma).
3. Perform a flat-test procedure as described in the technical bulletin by turning
up voltage until a DR of 0.2 % is achieved or upon reaching 11 kV. Sometimes
an electrical discharge may occur, requiring the operator to perform the test
again. These electrical discharges often remove stubborn particle contamina-
tion that has not been removed via the high-pressure gas and plasma treat-
ments and should not necessarily be avoided for new electrodes. These
electrical discharges often happen after introduction of new electrodes into
the LEAP from storage in air and do not negatively affect the performance of
new electrodes.
4. Should electrodes fail to yield satisfactory flat-test results, then repeat the gas
flow and plasma treatment an additional time. If plasma treatment is unavailable,
then remove and repeat physical treatment with gas and re-test.
5. Any new electrode failing to meet minimal electrical performance characteris-
tics upon entry and after testing/conditioning described above should be imme-
diately returned for replacement/requalification (contact your CAMECA service
representative for details).
6. For the purposes of acquiring the highest quality data possible, electrodes should
be placed into the LEAP as soon as received from CAMECA and tested/
conditioned for subsequent use. Should space considerations require that some
electrodes need to be stored external to the LEAP, keep them covered and in a
dry, oil-free, dust-free environment and test/condition them upon entry into
the LEAP.
Acquisition Control
Modern atom probes utilize some form of automated acquisition algorithm to

control data collection [4]. Ideally, automated control would adjust conditions
to optimize analysis yield, data quality, and acquisition time; however, current
algorithms merely utilize some form of voltage manipulation in order to maintain a
desired DR. Alternative control strategies continue to be developed. See your
control software user manual for the latest implementation.
Main Voltage Control Algorithm
LEAP automated acquisition control dictates voltage steps by way of a proportional

control algorithm defined in the EvapControl.INI file using six parameters. Opening
or editing INI files should only be done after consultation with CAMECA to avoid
potentially serious problems. The general control behavior is illustrated in Fig. E.5.
Here the target DR is established, and the voltage increases or decreases dependent
on the average rate measured during some previous acquisition time window. As
shown in the figure, when the target DR is changed from 3 to 14 %, the voltage
quickly increases to raise the actual rate. Note that the actual rate is not immediately
achieved. The rate at which the set point is achieved and the propensity to overshoot
or undershoot the target are standard issues for control algorithms [5].
The parameterization of the LEAP control algorithm allows the user to optimize
the efficiency for reaching and maintaining the DR set point by modifying the
algorithm parameters. Default Algorithm 1 from LCC 4.2 is illustrated as an
example below followed by a description of each of the fields following:
[Algorithm 1]
Algorithm Name Standard
Evaporation rate moving average filter depth (pulses) 10000
Pulses Per Voltage Update 5000
Proportional Coefficient (%/%) -10.0
Max upward voltage step (V) 0.5
Max downward voltage step (V ) 0.5
Algorithm Name is the name of the algorithm, up to 100 characters long. This
name appears in the LEAP Control Center (LCC) Advanced Acquisition Setup
window, Evaporation Control Algorithms pane, which adds the item to the pick
list. If the name is set to UNUSED, the algorithm is disabled and does not appear
in the pick list.
Evaporation rate moving average filter depth (pulses) is the number of pulses
over which the current evaporation rate is calculated. The resultant moving
averaged evaporation rate is then input to the algorithm to determine the size
of the next specimen-voltage change.
Fig. E.5 The proportional control algorithm in action showing (top) target and actual detection
rate and (bottom) voltage response for the same 10 million detected events. The arrow shows
excursions in the detected rate that occur during a typical laser Drift Comp scan. The large +
shows the occurrence of an IEC event and the resulting fast drop in voltage which resulted in an
equally fast reduction in observed detection rate
Pulses Per Voltage Update specifies that after every n pulses the firmware
calculates the current DR (based on the filter depth as determined above),
calculates an error term based on the difference between the current and target
DRs, and calculates a voltage-update step size equal to the product of the error
term and the proportional coefficient setting. This parameter defines n and hence
determines the frequency of voltage adjustments.
Proportional Coefficient (%/%) is a multiplier for converting the detection-rate
error term to a voltage adjustment value.
Max upward voltage step (V ) is the maximum upward-step size per voltage-
update interval. The final voltage step for an increasing voltage step is clipped
against this setting.
Max downward voltage step (V ) is the maximum downward-step size per
voltage-update interval. The final voltage step for a decreasing voltage step is
clipped against this setting.
Proportional Control Algorithm DetailThe voltage step is governed by the

following formula which is limited by maximum positive and negative voltage
steps (examples based on Algorithm 1 for f 200 kHz are illustrated following
the general formula):

Prop:Coeff
V Step V Max Down , V DRCurrent% DRTarget% , V Max Up
100
f 200 kHz 1
! 0:5, V 0:1 DRCurrent DRTarget , 0:5 Volts per s
40
Maximum voltage slew:
f 200 kHz
V Up Volts=s f V Max Up =N Pulses Per Update ! 20 Volts=s
f 200 kHz
V Down Volts=s f V Max Down =N Pulses Per Update ! 20 Volts=s
Advanced users can edit and modify the EvapControl.INI file to include addi-
tional algorithms or modify the default algorithms. In order to reset the
EvapControl.INI file to its default settings, delete or rename the EvapControl.INI
file and then close and reopen the LEAP Control Center software, and a new default
file is generated.
The behavior of each default algorithm is summarized in Table E.1 where
the number of pulses per update is also converted to time per update for a
200 kHz pulse frequency. Typical rate determinations are averaged over a 10,000
or 50,000 pulse window. Voltages are updated every 5,000, 10,000, or 25,000
pulses depending on the specific algorithm. The proportional coefficient ranges
from 0.05 to 10, while the maximum voltage change per step is between 0.5 and 5 V
going upward or 0.5 and 1,000 V going downward.
When operating the LEAP, the user selects target DR and initial voltage
(or engages the algorithm during the experiment), and the algorithm manipulates
the voltage from there. Two of the most obvious effects of the algorithm choice are
the minimum DR allowed by the software and the maximum rate of voltage
increase (which will often affect the time it takes to go from near-zero ion emission
to the target DR). For many materials that require low DRs to survive, only the Low
Detection Rate and Interface Low Rate algorithms are appropriate, although these
algorithms have reasonably large maximum upward voltage steps.
In terms of optimizing analysis yield or data quality, the current algorithms are
blind. They merely try to hold the evaporation process relatively constant.
Multiphase materials or layered structures can present widely varying local evap-
oration effects. Alternative control algorithms, dependent on acquisition informa-
tion other than the DR, might allow for improved yield and/or data quality.
298
Table E.1 Default acquisition parameters for LCC 4.2

Default algorithms 1 2 3 4 5 6
Name Standard Soft turn-on Free running Low detection rate Interface low rate Fast slew
Evaporation rate moving average 10,000 10,000 10,000 50,000 50,000 10,000
filter depth (pulses)
Minimum DR (%) 0.5 0.5 0.5 0.1 0.1 0.5
Pulses per voltage update 5,000 10,000 10,000 25,000 25,000 5,000
At 200 kHz 1 1 1 1 1 1
s s s s s s
40 20 20 9 9 40
Proportional coefficient (%/%) 10.0 0.1 0.05 0.1 0.2 10.0
Max upward voltage step (V) 0.5 1 5 2.5 2.5 1
At 200 kHz 20 V/s 40 V/s 200 V/s 100 V/s 100 V/s 40 V/s
Max downward voltage step (V) 0.5 1,000 1,000 1,000 1,000 1
At 200 kHz 20 V/s 40,000 V/s 40,000 V/s 40,000 V/s 40,000 V/s 40 V/s
Appendix E: Additional Considerations for LEAP Operation
Interface Evaporation Control Algorithm
Over time, the LEAP acquisition control algorithm has evolved to include a fast
response feature termed interface evaporation control (IEC) that has demonstrated
yield benefit for some sample types. The goal is to detect abnormal detection
statistics (a possible signal preceding specimen failure) and minimize the chances
for failure by quickly lowering the evaporation rate. It is difficult to rapidly change
the specimen voltage due to the relatively high capacitance of the specimen stage
and HV power supply, so schemes were developed to rapidly stop the applied
pulse to the tip. The original implementation in software calculated an average DR
over a pulse window (defined by the user). When the average rate exceeded the
target rate by some factor, defined by the user, the software would stop firing of
the voltage pulser (or deflect the laser off the tip) and lower the voltage by some
amount (also defined by the user). Although the response of this algorithm is
relatively quick compared to the control algorithm, a firmware implementation
is significantly faster. Note that not all LEAP tools have the firmware implemen-
tation option.
The modern implementation of the IEC in LCC 4.2 uses the firmware to monitor
ions emitted during consecutive pulses and responds immediately (voltage pulse
generation is interrupted or laser pulse picking terminates, and the standing voltage
is lowered prior to the next pulse). The user selects the sensitivity of the response
(low, medium, or high), so when an extreme number of consecutive instrument
pulses result in a detected ion (36 out of 40 pulses, 18 out of 20 pulses, or 9 out of
10 pulses, respectively) the instrument immediately lowers the voltage by an
amount defined by the user. For typical target DRs (1 % or 1 ion for every
100 pulses) it is statistically unlikely for this sequence of detected events to occur
randomly and indicates initiation of rapid tip change (the odds are less than 1067,
1033, and 1017 for low, medium and high IEC, respectively). Quickly stopping
this behavior should improve the chances for specimen survival for a wide variety
of sample types. The occurrence of an IEC event is illustrated in Fig. E.5 by the
large plus sign positioned immediately following the lowering of voltage which
results in a correlated drop in DR. IEC settings are sample type specific, and a too
sensitive setting will result in a lower average detection rate and a degraded signal-
to-noise ratio (SNR).
Constant Flux Control
Keeping DR constant by manipulating only the voltage may not provide for the best
yield or data quality. For example, background counts resulting from the field
evaporation of adsorbed vacuum contaminants should be a function of DR, apex
surface area, and pulse rate. Consequently, background levels might be improved
by operating under constant evaporation flux (ion detection rate per unit area) rather
than DR (ion detection per time) [6]. For material systems that rarely fail during
analysis, constant ion flux likely provides for a better SNR, more consistent
evaporation physics, and possibly improved spatial reconstructions. In order to
estimate the ion flux from the surface, an estimate of the specimen surface area is
necessary. LCC allows the operator to enter an estimated evaporation field that it
then uses with the specimen voltage to estimate surface area (see Chap. 5).
There is some danger or added complexity in implementing a constant flux
control for inhomogeneous materials, especially for multilayer systems. First,
relative to the constant DR algorithm, the electric field will be higher in order to
increase the DR as emission area increases. This increases the chance of specimen
failure as the analysis proceeds. Second, flux estimates are complicated for mate-
rials with differing evaporation fields present in the field of view simultaneously.
Underestimating flux would again result in higher DRs, higher applied field, and
higher likelihood of specimen failure. Finally, for multilayer systems, proceeding
from a layer of low-field material to high-field material for example, the increasing
specimen voltage necessary for evaporation of the high-field material may be
incorrectly interpreted by a constant flux algorithm as an increase in emission
area. The algorithmic demand for voltage (field) increase will be twofold: Constant
flux requires the DR to be increased because of a perceived increase in emission
area (increase in applied voltage), and the increasing field required by the material
requires higher field (increase in applied voltage) to maintain even a steady
DR. The combination of these two demands for increased field (voltage) increases
likelihood of fracture, so any implementation of this type of algorithm must be
considered carefully.
Constant Charge-State-Ratio Control
For pulsed laser acquisition, constant apex field/temperature is desirable for high
data quality based on reasoning similar to that for constant flux. This type of control
requires an ability to monitor the apex temperature and an algorithm to interpret this
variable in order to modify pulse energy during acquisition. Because the evapora-
tion field required to reach a particular rate is a function of temperature, charge-
state ratios (CSRs) can provide an indirect indication of apex temperature [7]. The
sensitivity of any given CSR to the actual temperature (at the time of evaporation)
depends on the details of that particular ion species (rate of CSR change per rate of
field change) and its response as a function of temperature (rate of field change as a
function of temperature). A variety of theoretical and experimental behaviors have
been observed [8, 9]. Implementation of such an algorithm might be straightfor-
ward because the DR could be controlled by voltage and the laser pulse energy
could be controlled by CSR, so both controls are separate and decoupled; however,
some of the same dangers recognized for constant ion flux exist for constant CSR
mode applied to inhomogeneous systems. The particular ion species used to control
laser pulse energy needs to be present everywhere in the specimen; otherwise,
sufficient conditional alternatives need to be in place to transition from one set of

control criteria to another. In addition, the calibration file must be of sufficient
quality to allow accurate interpretation of the mass spectrum during acquisition as a
function of voltage.
Laser Positioning Control
The LEAP is designed to deliver maximum LPE density to the tip apex by utilizing a
diffraction-limited laser spot (see Chap. 3) [10]. This approach avoids heating
unnecessary regions of the tip and optimizes heat transfer from the apex to the
bulk, thereby improving mass resolving power [10, 11]. This approach however is
not without design challenges. The small spot size requires accurate targeting of the
laser to the apexas the specimen moves, the laser must follow, and when the laser
moves or drifts it must be promptly repositioned back to the specimen apex. This
coordinated alignment is susceptible to various instrumental instabilities which
might include thermal expansion or contraction of various elements comprising
the laser path, positioning optics, specimen, or specimen stage and can cause the
laser position and/or apex position to drift on the order of a micron to a few microns
during a multi-hour acquisition. Sub-micrometer repositioning corrections are
required to keep the laser optimally aligned to the tip. A set of laser scanning
algorithms are available and designed for periodic tracking of the laser spot to the
tip apex as determined by the maximum observed DR.
Three scanning types are defined and can be modified by the user: Scout,
Standard, and Drift Comp scans (see Table E.2). The Scout scan is intended to
sweep large areas around a specimen to allow initial alignment of the laser spot to
the specimen apex. For the LCC 4.2, a 12 12 m2 scanning area with 0.4 m
resolution is a practical limit to the scan size. The number of points in any scan is
limited, and scanning with resolutions coarser than 0.4 m risks missing detection
of the maximum entirely. The Standard scan is intended to more accurately place
the laser at the maximum detection position by utilizing a smaller scanning area
which allows for finer scan resolution. Here the scan area is intended to include the
entire area over which emission from the specimen can be observed and is a
function of the actual tip dimensions coupled with the laser spot size. The overall
dimensions of this scan can vary as a function of specimen type, DR, and LPE, and
Table E.2 Default scans defined for LCC 4.2 (see Appendix H of the LEAP User Manual for
instrument-specific details on peak finding, thresholds, dwell time, and other laser scan
information)
Laser position scans (m) Scout Standard Drift Comp
Tip scan edge size 12 1.2 0.25
Tip scan step size 0.4 0.12 0.025
Focus scan depth 20 10 3
Focus scan step size 4 2 1
so some adjustment to the default settings may be warranted by the user. The Drift
Comp scan is intended for use during stable acquisition. In this mode only small
corrections in laser position are necessary, and it is not desirable to allow the DR to
be significantly reduced (any rate reduction likely causes a relative increase in
signal background); hence, the drift scan covers only a small scan area. Similar to
the Standard scan, the area over which the data rate is observed to change notice-
ably is specimen dependent, and so some tailoring of the scan range to specimen
type may be warranted. A simple guideline is for the Drift Comp scan edges
(corners) to dip to not more than ~50 % of the target DR and for the number of
scan steps to fall between 25 and 100 (5 510 10 array steps). The control
algorithm adjusts the dwell time for each step to attempt to ensure adequate
counting statistics, so at low DRs, scan times can become significant.
Besides coordination of the lateral laser position with the specimen apex, the
small focus depth of the laser requires coordination of the laser focus with
the specimen apex as well. The elongated depth of focus reduces the relative
sensitivity to motion along this direction, and so the correction is less significant.
For LEAP 3000X systems, a focus correction at the beginning of an analysis may
not need significant correction for many hours. LEAP 4000X systems have a lesser
depth of focus requiring more frequent focus corrections (~hourly).
Also shown in Table E.2 are the default settings for the LCC 4.2 Scout, Standard,
and Drift Comp focus scans. The Scout focus scan is intended to cover a wide range
of focus depth using the Scout aerial scan and might only be needed after a local
electrode change when the required position of the laser focus changes signifi-
cantly. The Standard focus scan uses the Standard aerial scan and is intended to
cover a more narrow range of focus depth. The Standard focus scan is suggested as
an initial scan at the beginning of an analysis that has significant and stable ion
emission. The Drift focus scan uses the Drift aerial scan and is intended to be quick
and provide only small adjustments to the laser focus position. This is the scan that
is used during automated acquisition. Note that not all materials run well under
default automated focus scan conditions.
Two notable features exist to aid the performance of the tip tracking algorithm:
Mass Filter Ranges and Extended Dwell Time are both defined on the LCC
Advanced Controls tab within the Laser Atom Probe tab. When scanning over a
specimen, a variety of emitted ion types can be observed. The goal of proper tip
tracking is for the laser to be positioned in the place that results in maximum in-time
specimen ion emission. Depending on the specimen structure, there may be places
on the specimen that will emit ions at greater intensities but are dominated by
out-of-synch or adsorbed ion species (the platinum weld on a lift-out specimen for
example). To avoid this, the user can define up to three ranges to use for calculating
DR during a laser scan. Default ranges exclude hydrogen and water ion peaks. The
Extended Dwell Time check box allows the user to triple the amount of time
devoted to individual laser steps. This may be necessary when scan statistics at
low DRs are insufficient to result in acceptable tip scanning. All of the user-selected
features on the Advanced Control tab can be saved, enabling customized scan sizes
for various specimen classes.
Fig. E.6 IVAS Wizard plot of laser position and ambient temperature for a 24-h microtip run
(top). The laser scanning frequency and size were insufficient to compensate for an external
temperature variation that occurred during the middle of the analysis (indicated by the accompa-
nying rise in voltagebottom plot). Higher scan frequency or larger area may be warranted.
Hourly focus scans reveal changes in focus of up to 1.5 m, and these are correlated to large lateral
laser shifts as well. For the 4000X, 1 m focus changes are significant. More frequent focus scans
may be warranted in this example
In order to assess laser tracking stability, it is suggested to monitor laser position

over the course of a long analysis (silicon microtip analyzed overnight for example)
during which regular lateral and focus scans are performed. Any unexplained
excursions in the voltage plot (not attributable to normal evolution due to changes
in specimen composition and structure) may be a sign of poor laser tracking. Using
the Cal/Recon Wizard step 2 in IVAS (see Chap. 5), a Laser Pos. and Ambient
Temp. Hist. plot is available that shows laser position change vs. ion sequence
number (Fig. E.6). Analysis of this plot should allow for sensible scanning fre-
quency adjustments to be made.
References
1. Lunger, M.J., Vergason, G.E.: Reducing pump-down time with dry air venting. Soc of
Vacuum Coaters 43rd Annual Technical Conference Proceedings, p 432 (2000)
2. Dylla, H.F., Manos, D.M., Lamarche, P.H.: Correlation of outgassing of stainless-steel and
aluminum with various surface treatments. J. Vac. Sci. Tech. 11(5), 26232636 (1993),
DOI:10.1116/1.578617
3. Fowler, R.H., Nordhiem, L.: Electron emission in intense electric fields. Paper presented at
the Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathe-
matical and Physical Character
5. Li, Y., Ang, K.H., Chong, G.C.Y.: Patents, software and hardware for PID control - an
overview and analysis of the current art. IEEE Contr. Syst. Mag. 26(1), 4254 (2006)
9. Southworth, H.N., Ralph, B.: J. Microsc., 90 (1969)
10. Bunton, J.H., Olson, J.D., Lenz, D.R., Larson, D.J., Kelly, T.F.: Optimized laser thermal
pulsing of atom probe tomography: LEAP 4000X. Microsc. Microanal. 16(Suppl 2), 1011
(2010)
11. Bunton, J.H., Olson, J.D., Lenz, D.R., Kelly, T.F.: Advances in pulsed-laser atom probe:
(2007)
Glossary
1D 2D 3D The dimensionality of data. 1D refers to a one-dimensional analysis

such as a mass histogram or a 1D concentration profile. 2D refers to a
two-dimensional dataset where the data lie in a single plane, such as a hitmap
or a picture. 3D refers to a three-dimensional analysis, such as a position file.
1DAP 3DAP 1D or 3D atom probe instruments. 3DAP was also the model name of
an instrument produced by Oxford Nano Science.
AMETEK AMETEK, Inc is a global manufacturer of electronic instruments and
electromechanical devices with headquarters in the United States and over
80 manufacturing sites worldwide. CAMECA SAS is part of AMETEKs Mate-
rials Analysis Division.
Amplifier time-to-digital converter (ATDC) A time-to-digital converter that
records the time of occurrence of independent electrical pulse inputs, each
measured relative to a single, common reference input. It records timing events
from each side of the three DLD anodes.
Analysis chamber The vacuum chamber in which specimens are analyzed in
the LEAP.
Annular milling A type of FIB milling pattern.
Aperture The opening in a local electrode through which ions pass after evapo-
rating from the specimen tip.
Appm Atomic part per million.
APT See atom probe tomography.
ASCII American standard code for information interchange.
ATDC See amplifier time-to-digital converter.
ATO A file format containing the same information contained in a POS file with
additional data such as raw time of flight and other information used in data
reconstruction.
Atom probe tomography (APT) The subject of this handbook. Please read the
entire handbook for more information.
306 Glossary
Background Recorded event in the mass spectrum which do not originate from
the specimen or are not found within a peak (not correlated to a voltage or a laser
pulse).
Backside Specimen preparation orientation such that the analysis proceeds from
below the original surface and toward the exposed surface (opposite of top-side
or top-down analysis).
Bakeout The process of heating the LEAP instrument to approximately 150
C.
A bakeout is performed after the LEAP vacuum chambers have been exposed to
air and assists in achieving high or ultrahigh vacuum.
Base pressure The standard operating pressure within a particular vacuum cham-
ber in the LEAP instrument.
BCC Body-centered cubic.
Bgd See background.
Bias voltage The DC voltage applied to the specimen during data acquisition.
Bowl correction A time-of-flight correction to retain the relationship that all ions
of the same mass and charge record the same time flight for all locations on the
detector.
Buffer chamber A vacuum chamber between the load-lock and analysis cham-
bers. This is the second of two chambers in which a carousel may be located and
the chamber from which specimen and local electrode pucks are stored for use.
Calibration specimen A standard specimen (often aluminum or doped silicon)
that produces consistent data quality when the LEAP system is operating
properly such that it can be used to monitor system performance.
CAMECA ROOT A software application developed within the framework of
ROOT, a Lesser General Public License object-oriented framework for data
analysis developed by the European organization CERN.
CAMECA SAS A company based in Gennevilliers, France, that produces scien-
tific instrumentation including SIMS, EPMA, LEXES, and APT. CAMECA was
originally an acronym for Compagnie des Applications Mecaniques et
Electroniques au Cinema et a lAtomistique but has been out of the film business
since the late 1960s. It has several subsidiaries, including CAMECA Instru-
ments, Inc., Madison, WI-based headquarters for CAMECA in North America.
Capping Deposition of sacrificial material on top of a sample to protect it during
specimen preparation.
Carousel A carousel is a platter onto which is loaded either local electrode pucks
or specimen pucks. The carousel resides in the load-lock or storage chamber.
Carousel heater An optional load-lock upgrade that enables baking of local
electrodes and/or specimens in the load-lock chamber to rapidly degas them.
CERN Conseil Europeen Pour La Recherche Nucleaire, now referred to as the
European Laboratory for Particle Physics.
Client PC A desktop computer loaded with the LCC software through which a
LEAP system operator controls LEAP functions such as data acquisition, spec-
imen-to-local electrode alignment, and ISdb functions.
Glossary 307
Co-evaporation When multiple ions evaporate from the specimen during a single
pulse.
Cold finger The part of the cryostat at which refrigeration is concentrated. The
cold finger is mechanically isolated from the specimen stage to reduce specimen
vibration during data acquisition. It is thermally connected to the specimen stage
by thermally conductive helium gas.
Complex ion An ion that contains more than one atom, also known as a molecular
ion.
Constant CSR A data collection mode where the laser power control algorithm
calculates and maintains a constant CSR for a species in the specimen.
Constant flux A data collection mode where the specimen voltage control algo-
rithm calculates and maintains a constant ion/nm2/s detection rate.
Cross Section Refers to preparing a specimen such that the analysis proceeds
parallel to the original exposed surface (also known as parallel orientation).
Cryogenic temperature controller This device, located in the Instrument rack,
controls the temperature of the cold finger that cools the specimen. The cold
finger has a constant cooling power; the cryogenic temperature controller adds
heat to the cold path with a resistive coil to regulate temperature.
CSR Charge-state ratio.
Da Daltonthe SI atomic mass unit (AMU).
Data acquisition See atom probe analysis.
DAVis Data acquisition and visualization system. The LEAP software component
that provides 2D, digital specimen analysis data to the workstation. This soft-
ware also provides run-time mass spectra and monitors specimen voltage, ion
detection rates, etc.
Delay line detector A PSD designed to provide high-speed resolution of the time
of flight and XY-coordinate encoding of a single or multiple ions. The DLD
consists of serpentine DLAs and timing circuits. A charge cloud produced by the
microchannel plates impinges on the DLD creating an electromagnetic pulse that
travels each direction to the ends of the delay line and out to the ATDC timing
circuits.
Delocalization A smoothing function allowing discrete data produced by the
LEAP to create the 3D concentration grid used to create surfaces in IVAS.
Deprocessing A specimen preparation step where unwanted regions are removed
from near the region of interest. Often used when preparing microelectronic
samples.
Desiccant A drying agent typically used in dry-box applications to prevent mois-
ture from accumulating on local electrodes and specimens in storage. Only dust-
free desiccants are recommended for local electrode storage.
Desorption Field evaporation of gas ions adsorbed onto the specimen surface.
Detector A device within the LEAP instrument that includes the MCP and DLD
and which detects ions released during data acquisition.
Detector space A coordinate system based on the curved specimen surface
projected onto the flat detector and is displayed in live time in DAVis. Detector
308 Glossary
space is visualized in a real space reconstruction as hemispheres centered on the

reconstruction center.
Discrete discrimination The ability to discern the identity of each individual ion
in a dataset, which can be complicated by high background noise or a small peak
in the tail of a large peak.
DLA Delay line anode.
DLD See delay line detector.
DR Detection rate.
Dry box A storage device specifically designed to maintain a moisture-free envi-
ronment for electrodes and specimens.
EDS Energy-dispersive X-ray spectroscopy (also known as EDX).
EDX Energy-dispersive X-ray spectroscopy (also known as EDS).
Electric field The distribution in space of the strength and direction of the force
that would be exerted on an electric charge at any point in that space. Units for
APT are typically V/nm.
Electrode See local electrode.
Electropointer A motorized electropolishing system with automatic and man-
ual modes (produced by Simplex Scientific).
Electropolishing The process of applying a positive charge to a sample while
immersed in an electrolyte in order to remove material and sharpen the sample
into a specimen.
ENV The environment file used in TAP instruments that contains ranging and
reconstruction information.
ePOS The extended POS file containing the information in the POS file and
additional data such as raw time of flight and other information used in data
reconstruction.
ESD Electrostatic discharge.
Evaporation See field evaporation.
Experiment See atom probe analysis.
FCC Face-centered cubic.
FEEM Field electron emission microscopy.
FET Field effect transistor.
FFT Fast Fourier transform.
FIB See focused ion beam.
FIB-LO Focused ion beam lift-out. The process of making APT specimens with
an FIB.
Field enhancement Increase in electric field due to the proximity of the local
electrode to the specimen.
Field evaporation The phenomena that results in specimen atoms being ionized in
the presence of extremely high electric field of the order of a few to tens of
V/nm.
Field ion microscopy The process where gas atoms adsorbed on the tip of a
specimen are ionized by a high electric, becoming positively charged, and are
Glossary 309
repelled from the tip. The curvature of the surface near the tip causes a natural
magnificationions are repelled in a direction roughly perpendicular to the
surface, producing a point projection effect. A DLD or a phosphor screen is
used to visualize the ions.
FIM See field ion microscopy.
Flash See specimen fracture.
Flat-test The procedure to test the quality of a local electrode.
Flat-top microtip array A silicon coupon with an array of blunt-topped specimen
posts. These coupons provide an efficient way to do precise and accurate FIB
lift-out specimen preparation techniques.
Flight path The space inside the atom probe that ions pass through when they
leave the specimen and travel to the detector.
Focused ion beam A specimen preparation and milling technique typically using
gallium as the focused ion medium to micro-machine samples into specimens.
FOV Field of view.
Fracture See specimen fracture.
FW0.01M Full width 1/100th maxa measure of a spectral peak width or mass
resolving power.
FW0.1M Full width 1/10th maxa measure of a spectral peak width or mass
resolving power.
FWHM Full width half maxa measure of a spectral peak width or mass resolv-
ing power.
HAADF High-angle annular dark fieldan imaging technique and detector type
used in TEM.
Hit event A hit event is a per-pulse detection of an ion on any of the detector
channels in the LEAP microscope.
Horizontal transfer rod The manually operated, magnetically coupled, horizon-
tal rod that moves a specimen or a local electrode puck from the carousel in the
buffer chamber to the analysis chamber.
HR The LEAP model including the large-angle reflectron flight path.
HVAC Heating, ventilation, and air conditioning.
IEC See interface evaporation control.
Image correction factor A variable in a large field-of-view reconstruction to
correct for the fact that ions off of the center of the specimen apex do not travel
in straight lines toward the detector, but rather ions leaving from locations closer
to the specimen shank are increasingly compressed toward the center of the
specimen apex.
Imago The proper noun for, and an acceptable truncation of, Imago Scientific
Instruments Corporation. Acquired by AMETEK, Inc. Imago is now part of
CAMECA Instruments, Inc.
Instrument flange The front flange on the LEAP instrument Analysis chamber,
upon which is mounted the specimen stage, detector, pulser electronics, etc.
310 Glossary
Instrument rack A rolling rack containing the LEAP microscope component

computers, controllers, power supplies, a backup power system, and emergency
system-power-cutoff switch.
Integrated scientific database See ISdb.
Integrated visualization and analysis software See IVAS.
Interface evaporation control A LEAP instrument feature that automatically and
rapidly reduces field ionization when detection rate increases beyond a specified
threshold to help reduce the incidence of specimen fracture.
Ion gauge A gauge that measures pressure in a vacuum chamber. The Instrument
rack contains an ion gauge controller that converts the analog pressures into
digital readouts.
ISdb The LEAP software component integrated into the LCC control system
that queries, and updates, the specimen database. ISdb keeps track of locations
and conditions of specimens and local electrodes as well as records experimental
setup information. A trademark of CAMECA Instruments, Inc.
ITRS International technology roadmap for semiconductors.
IVAS Integrated visualization and analysis software. The LEAP software compo-
nent that reconstructs specimen data into three-dimensional images and is used
for data analysis. It is based on Java and Netbeans.
k-factor A geometry-driven electric field enhancement factor used in reconstruc-
tion calculations.
LAR Large-angle reflectron. A specially designed electrostatic mirror for ions.
It uses an electric field to reflect incoming ions from a specimen toward a
detector. In the process, the reflectron improves mass resolving power by
time-focusing ions of different energy onto the detector. With a LAR, it is
possible to record a large field of view while maintaining high mass resolution.
LAS LEAP acquisition server. The LEAP software component through which
acquisition commands are sent and that transfers acquisition data to the client
PC.
Laser head Mode-locked laser cavity which produces pulses at a rate of 80 MHz
with a fixed pulse energy.
Laser scan A process of rastering the laser beam over the specimen apex while
monitoring the DR to identify the optimal laser spot location on the
specimen apex.
LAWATAP Laser-assisted wide-angle tomographic atom probe. A previous gen-
eration of a 3D straight flight path atom probe.
LCC LEAP control center. The primary software user interface that controls the
LEAP microscope.
LEAP Local-electrode atom probe. A proper noun for, and abbreviation of, the
LEAP microscope. See LEAP Microscope.
LEAP acquisition server See LAS.
LEAP control center See LCC.
Glossary 311
LEAP microscope The part of the LEAP system that is mounted on a chassis,
contains vacuum chambers and other equipment and instruments, and from
which specimen data originates. A CAMECA product designed to analyze
specimens at the atomic scale.
LEAP system schematic The LEAP system schematic is a digital, pictorial, user
interface in the LCC software representing the functions and features of the
LEAP microscope, from which parts of the microscope are controlled.
LEAP system server See LSS.
LEAPnet A LEAP ethernet network that allows the workstation to communicate
with the instrument-rack PC, which, in turn, communicates with other instru-
ments in both the rack and on the LEAP instrument.
Load-lock chamber This chamber is the chamber into which you place carousels
and is the first step in loading specimens and local electrodes into the instrument.
Local electrode The cone-shaped piece of metal mounted on a puck and posi-
tioned directly opposite the specimen in the analysis chamber. A brief, high-
voltage pulse, supplied by the high-voltage pulser, which is applied to the local
electrode during a voltage pulsed atom probe experiment. The local electrode
improves the mass resolution, field of view, and acquisition speed of atom probe
analysis and simplifies specimen preparation for atom probe analysis.
Local electrode puck The local electrode puck provides a mounting surface for
the local electrode within the LEAP instrument. See Local Electrode.
LPE Laser pulse energy.
LPS Laser positioning server. The LEAP software component that controls laser
position.
LSS LEAP system server. The LEAP software component that allows all LEAP
software and components to communicate and interact.
m/n Mass-to-charge-state ratio, also expressed as m/c, m/e, m/q, m/z, and u/e.
Mass-to-charge-state ratio The mass-to-charge-state ratio of an ion is the ions
mass (in mass units) divided by its charge state (in charge units). The mass-to-
charge-state ratio is on the x-axis of a mass spectrum and is indicated in SI units
of Daltons (Da).
MCP See microchannel plate.
Metadata Metadata is information a user enters into a database and that can be
used for sorting and searching the database. Adding specimens to the ISdb often
requires adding to the database metadata about the specimen.
Microchannel plate Part of the atom probe detector system. The MCP is a
transducer/amplifier that amplifies a single, charged species by approximately
1 106. Gain is controlled by the voltage difference between the front and back
plate.
Microchannel plate amplifier See microchannel plate.
Micro-positioning stage controller An electronic component in the instrument
rack that controls the micro-positioning motors in the specimen stage. These
motors move the specimen relative to the fixed position of the local electrode.
312 Glossary
Microtip array One of the two primary specimen-mounting geometries in pre-

paring specimens for analysis. Microtip array coupons are ~3 7 mm2 chips of
silicon with an array of approximately 150 m tall microtips etched on the
surface. A trademark of CAMECA Instruments, Inc.
MRP Mass resolving power. A measure of the spectral quality of a peak in the
mass-to-charge-state ratio histogram.
Multi-hit event A multi-hit event is the detection of two or more independent ions
on the detector in the LEAP microscope.
Parallel-orientation analysis Refers to preparing a specimen such that the anal-
ysis proceeds across the surface of the original exposed surface (also known as
cross-section orientation analysis).
Peak overlap The superposition of two peaks in a histogram. Si-28+ and Fe-562+
for example.
PEEK Polyether ether ketone.
PF Pulse fraction.
Pixel The smallest addressable picture element in a display device.
PLAP Pulsed laser atom probe.
Plasma cleaner An optional hardware upgrade, the plasma cleaner is used to treat
local electrodes which have been in storage for an extended period of time and to
recover local electrodes which have failed due to specimen fracture.
POS A data file type consisting of the reconstructed ion position (x, y, z) and the
mass-to-charge ratio for each reconstructed ion.
PoSAP Position-sensitive atom probe. It is also the software that was developed by
Oxford Nano Science for use with the 3DAP instrument.
Puck A small carrier unit (~3 cm long) for specimens and local electrodes. These
are mounted into a carousel to be transferred to the specimen stage in the
analysis chamber.
Pulse generator A device that generates the electrical pulse delivered to the local
electrode to ionize atoms from a specimen apex for voltage pulsed data
acquisition.
Pulser See high-voltage pulser.
Ranging The process of identifying all the peaks in the mass-to-charge-state ratio
histogram and choosing lower and upper bounds.
Reconstruction The process of turning the timing signals from the DLD into real
space positions and identified ions.
Reflectron A specially designed electrostatic mirror for ions. It uses an electric
field to reflect incoming ions from a specimen toward a detector.
Residual gas analyzer An optional LEAP instrument that measures the concen-
trations of gases within the analysis chamber.
RGA See residual gas analyzer.
RHIT A raw data file created by the LEAP that the user analyzes with IVAS.
ROI Region of interest: The region of special interest for analysis.
ROOT See CAMECA ROOT.
Glossary 313
Roughing pump Roughing pumps are mechanical pumps for initial pumping of
the load-lock and buffer chambers. Roughing pumps are required before apply-
ing the turbo pumps, the latter capable of achieving ultrahigh vacuum (UHV).
RRAW The raw data file: RRAW files contain the raw detector data and the order
of ion collection from the LEAP as well as the history of various parameters
measured or controlled throughout the acquisition.
RRNG ASCII format files that describe how mass ranges are associated with
particular ion types.
Sample A sample is a representative part or a single item from a larger whole or
group. For example, a sample is obtained from a wafer, and a specimen is
prepared from a sample.
Sampling Taking a random subset of the collected information to increase calcu-
lation or display speeds. Typically, reconstruction calculations and charts in the
IVAS wizard are limited to 500,000 ions or fewer.
SDM Spatial distribution map.
Shank angle A measure of specimen taper.
SIMS Secondary ion mass spectrometry.
Site specific Specimen preparation that targets a specific ROI.
SNR Signal-to-noise ratio.
Specimen The needle-shaped object of interest from which the LEAP microscope
attains an image.
Specimen analysis See atom probe analysis.
Specimen database A LEAP software component where all specimens, local
electrodes, and run conditions are recorded. See ISdb.
Specimen DB See ISdb.
Specimen failure See specimen fracture.
Specimen fracture A specimen fracture occurs when a specimen mechanically
fails. Many common terms have been invented to describe failure including
flash, rupture, failure, or zot.
Specimen puck See puck.
Specimen stage That part of the LEAP hardware, located inside the analysis
chamber and part of the Instrument flange assembly, to which specimen and
local electrode pucks are mounted.
Specimen stage controller See micro-positioning stage controller.
SRM Standard reference material.
Stage controller See micro-positioning stage controller.
Storage chamber See buffer chamber.
STR STR files contain the raw detector data and the order of ion collection from
the LEAP as well as the history of various parameters measured or controlled
throughout the acquisition. They are used in the latest LEAP systems to create
RHIT files.
Stub A tube of metal, usually copper or nickel, which can accept wire geometry
specimens and can be secured in a specimen puck. Also known as crimps or
tubes.
314 Glossary
TDC See amplifier time-to-digital converter.

Time of flight A mass spectrometry method for determining ion mass by measur-
ing the time it takes to travel a known distance.
Tip tracking A LEAP run mode where the laser position control algorithm
maintains the laser position in the optimal location over the specimen apex.
TMP See turbo pump.
TODS Time of departure spread.
TOF See time of flight.
Top down Refers to specimen preparation where analysis proceeds from the
original exposed surface into the sample (it is also known as normal orientation
and is the opposite of backside sample preparation).
Torr A unit of pressure equal to 1/760 of an atmosphere or about 133.3 Pa or
1.333 103 bar. The primary unit of measuring pressure in the vacuum
chambers of the LEAP instrument.
Trajectory aberrations Any effect that causes ions leaving the surface to travel
toward the detector in unexpected paths.
Turbo pump Turbomolecular pump. One of several vacuum pumps on the LEAP
system that helps the LEAP vacuum chambers achieve ultrahigh vacuum.
Turn-on A phrase referring to the onset of emission (either electron or ion
emission).
UHV Ultrahigh vacuum. Any vacuum lower than 109 Torr.
UHV CLEAN A condition that indicates that the object is ready to be used in a
UHV environment. It typically requires 35 cleaning steps, ending with several
rinses, often in an ultrasonic bath, of high purity, low-moisture alcohol.
UHV foil An aluminum foil, applied as a dust cover to UHV clean parts. It is made
without the use of manufacturing oils. UHV foil is the preferred foil for protec-
tively wrapping UHV clean CAMECA LEAP microscope parts.
Unsampled The unsampled option displays all ions rather than a sampling of the
entire dataset.
UV Ultraviolet
Vertical transfer rod The motor-driven (vertical motion), magnetically coupled
vertical rod that moves a specimen or an local electrode puck between the
carousel in the buffer chamber to the specimen stage in the analysis chamber.
Voltage correction A time-of-flight correction to retain the relationship that all
ions of the same mass and charge record the same time of flight for all analysis
voltages.
Voxel A volume or 3D pixel.
Workstation See client PC.
Yield Analysis yield is fraction of all APT measurements.
Index
A Constant CSR, 300301

Accuracy, 8081, 87, 96, 112, 118119, Constant flux, 299300
148152, 157, 164166, 178, 185, 230 Continuity (tangential, non-tangential), 111,
Activation energy, 258263 124, 127128, 133, 141, 271273
Analysis chamber, 61, 7072 Convolution, 174, 175
Analysis direction, 33, 36, 4044, 84, 85, 138, Cooling, 10, 16, 69, 73, 75, 210
195, 196 Correction, 69, 110, 113117, 122, 142145,
Analysis yield. See Yield 167, 169
Annular milling, 32, 33, 38, 39, 4244 Corrections (voltage and bowlflight path),
Autocorrelation, 186188 113116
Counting statistics, 159, 171, 174, 179
Coupon, 17, 29, 30, 65, 70, 72
B Critical level. See Level
Background, 70, 8082, 8690, 97, 103, 106, Cross-section. See Analysis direction
111, 119, 147, 157, 164172 Cu(InGa)Se2 (CIGS), 227
Backside. See Analysis direction
Baking, 72
Binning, 164, 165, 171, 176, 186 D
Binomial, 116 Damage (ESD, ion), 19, 30, 3236, 106, 150
3DAP. See 3D atom probes (3DAP)
Data collection rate (DR), 7, 10, 11, 14, 58, 60,
C 61, 75, 201
Calibration, 72, 99, 110, 112117, 123, 152, 3D atom probes (3DAP), 2, 79, 18, 58, 240
194, 195 Decomposition, 117, 118, 172173
file, 91, 115117, 289, 290 Delay line, 6668
specimen, 30, 98 Density correction (relaxation), 142144
Capping, 3336, 4044, 84, 85, 150, 153 Depackaging, 224, 239, 240
Carousel, 6972 Deprocessing (FIB, wet, dry), 40, 41, 49, 84
Catalysis, 207210 Desorption, 8, 55
Ceramic, 16, 210217 Detection efficiency, 15, 5758, 61, 66, 81, 89,
Charge-exchange model, 259261, 284 121, 122, 144, 146148, 171, 185, 186,
Charge state ratio (CSR), 18, 57, 82, 150, 151, 192195
210 Detection level. See Level
Cluster, 61, 164, 180185, 208, 212215, 222, Detector space, 110130, 144
223, 226, 239, 240 Diamond, 215
Co-evaporation, 81 Disassociation, 171
Complex ion, 80, 81, 87, 90, 171 Drift (laser, specimen), 75, 105
316 Index
Drift compensation scan, 105 H

Drift scan. See Laser positioning control (drift, Hit events, 68
focus, scan) Holders (specimen, TEM), 16, 4649, 70
E I
Efficiency. See Detection efficiency IEC. See Interface Evaporation Control (IEC)
Electron emission, 3, 5, 64, 66, 106 Image
Electropointer, 29 compression (ICF), 120, 123, 134, 136,
Electropolishing, 19, 2529, 49, 201 137, 140, 141, 146148, 154, 156, 157,
Energy 267269
spread, 11, 16, 18, 59, 60 hump model, 139, 259262
Erosion parameter, 181, 182 Imaging atom probe (IAP), 7, 8
Evaporation Implant, 16, 3335, 39, 118,145, 149154, 156,
field, 15, 3436, 38, 40, 43, 8284, 90, 91, 157, 168, 235
96, 103, 122124, 133136, 138, 139, Initial radius, 125, 148
146, 148151, 155, 197, 233 Interface, 34, 40, 41, 43, 44, 81, 8385, 127,
in-time, 16, 66, 80, 164 139141, 148, 152155, 175, 176, 178,
out-of-time, 80, 86, 89, 90, 103, 164, 167 179, 202, 203, 208, 218, 223, 224, 227,
232, 234236
Interface Evaporation Control (IEC), 93, 101,
F 102, 299
Faceting, 221 Intergranular attack (IGA), 204207
Feature size, 175177 In-time. See Evaporation, in-time
FEEM. See Field electron emission microscope Iso-concentration, 1, 152154, 173, 175179,
(FEEM) 195, 196, 207, 220, 222, 226, 229, 233
Ferritin, 232233 Isosurface, 174, 178, 179
Ferroelectric, 216217
Field
emission, 3, 96, 225 K
enhancement, 10, 60, 63, 64 k-Factor, 122, 124, 127, 133, 136, 146, 148
evaporation, 15, 55, 64, 79, 134136,
258264
ionization, 5, 5557, 73 L
of view (FOV), 1, 58, 112, 218 Laser positioning control (drift, focus, scan),
Field electron emission microscope (FEEM), 75, 98105, 301303
35 Launch angle, 267270
Field ion microscope (FIM), 46, 69, 72, 257 LE. See Local Electrode (LE)
File formats, 110, 249256 LED. See Light Emitting Diode (LED)
FIM. See Field ion microscope (FIM) Level,
Fin-shaped field-effect transistor (FinFET), critical, 167, 170, 171
3739, 235238 detection, 166, 170, 171
Flat test, 98, 290294 quantification, 168, 170, 171
Flight, 6, 7, 16, 60, 67, 113 Lift-out, 16, 17, 19, 32, 33, 3638, 4143,
path, 7, 18, 56, 5861, 66, 88, 110, 45, 4749, 65, 158, 201, 213, 224,
113115, 133, 146, 147 225, 231
Fracture, 64, 75, 81, 83, 92, 98, 104 Light emitting diode (LED), 223226
Friends-of-friends, 180 Loadlock, 7072, 74
Local electrode (LE)
care, 6971, 106, 290294
G general, 2, 3, 1016, 49, 55, 5866, 91,
Gaussian, 59, 165, 166, 174176, 178 92, 96104, 106, 112, 122, 123,
Geologic material, 210217 148, 157
Grid, 142, 144, 154, 163, 173177, 186 Low energy cleanup, 3840, 43, 150
Index 317
M Precipitate, 9, 46, 69, 128, 134137, 140, 148,

Magnification, 56, 120, 121, 123, 131, 139, 164, 186, 195197, 222
145, 148, 152, 153, 270 Precision, 29, 57, 65, 66, 118, 132, 146, 149,
Mass resolution, 2, 11, 57 151, 152, 156, 159, 163166, 168, 169,
Mass resolving power (MRP), 6, 7, 1318, 202, 230
5761, 75, 80, 8690, 112, 115, 118, Preferential evaporation/retention, 73, 80, 89,
171, 172, 210, 281287 90, 232
Maximum m/n, 8690, 164 Probe hole, 7
MCPs. See Microchannel plates (MCPs) Proxigram, 174, 178180, 202, 218, 223, 224,
Mesh, 121, 132, 146, 147 227, 229, 232, 233
Microchannel plates (MCPs), 7, 57, 66, 67, 69, Puck, 47, 48, 6972
72, 121, 143, 144, 146
Micropolishing, 2729
Microtip (pre-sharp, flat-top, coupon), 10, 13, Q
16, 17, 2931, 37, 38, 43, 44, 49, 6466, Quantification level. See Level
70, 98, 99, 221, 232, 233
Mortise and tenon, 45
MRP. See Mass resolving power (MRP) R
Multi-hit, 68, 69, 76, 80, 81, 87, 89, 171, 215 Range file, 110, 119
Ranging, 80, 117120, 163166, 169, 171, 222
Reconstruction
N explorer, 128130, 189
Nanowire, 220221 general, 20, 34, 43, 46, 80, 86, 106159,
Neighborhood. See SDM, neighborhood 163, 185, 186, 189, 192, 194, 195, 204,
NIST SRM (2134, 2135c), 140, 141, 145159 207, 208, 216, 218, 223, 224, 230, 232,
Non-spherical apex, 149 236, 238
method (voltage, shank, tip profile), 109,
124, 137, 148, 150, 151, 153, 155
O Reflectron, 7, 1214, 18, 5861, 66, 72, 121,
Order-disorder, 195197, 202, 203 132, 146, 147, 149
Organic material, 16, 223, 227233 Relaxation. See Density correction
Out-of-sync. See Evaporation, out-of-time Remote electrode, 15, 18, 63, 64
Out-of-time. See Evaporation, out-of-time Reorder. See Density correction
Oxide, 31, 32, 40, 41, 46, 69, 9092, 145, 150, Residual gas, 80, 88
152155, 205211, 216, 217, 222, 232, Reverse engineering, 224
234, 240 Roughness, 152154, 174176
P S
Parallel. See Analysis direction Sampling, 119, 128
Parallel-orientation analysis deprocessing, Scan size. See Laser positioning control (drift,
40, 84 focus, scan)
Passivation, 208 Scanning. See Laser positioning control (drift,
Patterson function, 186 focus, scan)
Peak identification, 117, 118 Scout scan. See Laser positioning control (drift,
Piezoelectric, 73, 216217 focus, scan)
Pile-up, 171, 173 SDM, neighborhood, 130, 188, 189, 194
Plasma, 292294 Segregation, 179, 204, 227, 234, 236
Poisson, 166 Shank angle, 27, 46, 6264, 90, 121124, 128,
Poschenrieder lens, 7 137, 140, 141, 146155, 278279
Position-sensitive atom probe (PoSAP), 3, Shape (apex, hemispherical, non-spherical, non-
810 hemispherical, geometry), 10, 11, 29, 30,
Position sensitive detector, 8, 56, 60, 67, 39, 47, 6266, 70, 106, 124, 127, 133,
68, 132 135, 138, 139, 142, 149, 150, 154, 155
318 Index
Sharpen. See Annular milling or Thermal energy, 60, 283

Electropolishing Thin films, 16, 40, 138141, 230
Signal-to-noise ratio (SNR), 75, 194 Time resolving power, 282
Simulation, 33, 35, 134, 136, 138, 139, 154, Timing precision, 66, 281284
155, 170, 171, 203, 204 Tip continuity. See Sphere-cone radius ratio
Site-specific. See Lift-out Tip tracking. See Laser positioning control
Software (drift, focus, scan)
Davis, 92, 96, 97, 103 Tool (element overlay), 117
IVAS, 20, 110, 126, 129, 130, 133, 142, Top-down. See Analysis direction
144, 146, 147, 152, 159, 163165, 169, Trajectory aberrations, 131, 185
171, 173, 178, 188190 Transfer rod, 69, 71, 74
LCC, 74, 92,-94, 99, 100, 103 Transistor, 38, 40, 41, 144, 216, 235, 237, 240
Root, 110, 111, 130, 142144 Transmission electron microscopy (TEM), 19,
Solar cell, 225, 227, 231 20, 31, 32, 40, 4549, 70, 124, 125, 203,
Solute, 180185, 195 205, 207, 217, 237
Spatial distribution map (SDM), 123, 130, Turn-on (electrode), 106, 292, 293
146, 152, 156, 157, 164, 185192,
194197, 203
Spatial resolution, 4, 5, 66, 69, 73, 82, 84, U
87, 90, 131142, 174, 186, 192, 212, Uncertainty, 18, 68, 126, 131, 133, 159, 163,
218, 220 169171, 213, 221
Spectrum (mass, TOF), 33, 35, 57, 80, 86, Unsampling, 119
103, 113, 115, 117119, 163166,
169, 171, 173, 210, 211, 215, 217,
227, 228, 230, 231 V
Spherecone radius ratio, 124, 128, 136, 271 Vaporliquid-solid, 221
Sphere-to-cone ratio. See Sphere-cone radius Vibration, 12, 74, 75
ratio Void, 203, 204, 206
Standard deviation, 158, 166, 167, 171, 179 Voltage stability, 74, 282283
Standard scan. See Laser positioning control Voltage vs. sequence plot, 111
(drift, focus, scan) Voxel, 154, 174176
Storage chamber, 7072
Straight flight path, 14, 15, 18, 58, 60, 61, 66,
114, 147 W
Stress, 41, 57, 79, 8185, 90, 91 Width, 37, 39, 57, 64, 66, 72, 80, 81, 164167,
Stub, 47, 69, 70 169, 170, 175177, 194, 202, 203, 218,
Sublimation energy, 72, 260 223, 238
Superalloy, 202, 204207
Surface diffusion, 86, 87, 132
Y
Yield, 41, 74, 8187, 8992, 106, 150
T
TEM. See Transmission electron microscopy
(TEM) Z
Temporal spread, 282, 284 Zircon, 213, 214

Tomography

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Tomography

Загружено:

Авторское право:

Доступные форматы

David J.Larson Ty J.

Local Electrode Atom

Robert M. Ulfig Brian P. Geiser

ISBN 978-1-4614-8720-3 ISBN 978-1-4614-8721-0 (eBook)

Springer Science+Business Media New York 2013

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Madison, WI, USA David J. Larson

Professor Sir Colin J. Humphreys

1 History of APT and LEAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2.5.3 Sharpening Process . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5.3 Conversion from Detector Space to Specimen

6.6.5 FeCr Precipitates in a NiAlFeCr Alloy . . . . . . . . . . . . . 195

Appendix A: Data File Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

Appendix B: Field Evaporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257

Appendix C: Reconstruction Geometry . . . . . . . . . . . . . . . . . . . . . . . 267

Appendix D: Mass Spectral Performance . . . . . . . . . . . . . . . . . . . . . . 281

Appendix E: Additional Considerations for LEAP Operation . . . . . . 289

L, LDet Flight path length, cluster detection envelope parameter

w Specimen base width

Fig. 1.1 Ancestry of the LEAP

1.2 Ancestry of the Local Electrode Atom Probe

1.2.1 Early History and the Field Electron

Fig. 1.2 Photograph of

1.2.2 Field Ion Microscope: The First Images

needles required a ~0.1 m diameter apex in order to achieve ionization. FIM

1.2.3 Atom Probe Field Ion Microscope (1967)

Fig. 1.5 Data file from the

1.2.4 The Advent of Atom Probe Tomography

Fig. 1.6 Photograph of the

1.2.5 The Position-Sensitive Atom Probe (1988)

1.2.5.1 The Tomographic Atom Probe

1.2.6 Electron Beam Pulsed Atom Probe

The PoSAP was an extremely successful instrument, as it ushered in a new era in

1.2.7 The Scanning Atom Probe

1.2.8 The Local Electrode Atom Probe (2001)

Fig. 1.9 Schematic of the

The local-electrode geometry allowed us to go from a 15 nm FOV to greater than

1.3 The State of Instrumentation

1.3.1 The Growth of the Local Electrode Atom Probe

As described above, significant changes in atom probe instrumentation came about

Fig. 1.11 The two basic

In a remote-electrode geometry, blunter tips require higher voltages to create the

1.3.2 Laser Pulsing

1.3.3 Fundamental Considerations for Design

Atom probe instrumentation experienced a phase of rapid development in the first

the optimal trade-off. A straight-flight-path atom probe geometry is illustrated in

1.3.4 Reflectron-Based Instruments

1.4 FIB-Based Specimen Preparation

1.5 Concluding Remarks

There may be several reasons for electropolishing specimens. The method is

Fig. 2.1 Example of metal

2.3 Needles Versus Microtips

2.4 Electrostatic Discharge Considerations

2.5 Focused Ion Beam Methods

2.5.1 Capping Considerations and Damage

In the context of APT, capping refers to the application of a sacrificial layer of

2.5.2 Standard Lift-Out Process

milling with a ~6 nA ion current in a ~2 m-by-10 m rectangular pattern. The ion

2.5.3 Sharpening Process

2.5.4 FIB Deprocessing

In the preceding sharpening example, the necessary removal of a large amount of

2.5.5 Cross-Section Preparation

2.5.6 Backside Preparation