2004 ASM International. All Rights Reserved. www.asminternational.

org
Principles of Soldering (#06244G)

Principles of
Soldering

Giles Humpston

David M. Jacobson

Materials Park, Ohio 44073-0002

www.asminternational.org
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

Copyright 2004
by
ASM International
All rights reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the copyright
owner.

First printing, April 2004

Great care is taken in the compilation and production of this book, but it should be made clear that NO WAR-
RANTIES, EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, WARRANTIES OF MER-
CHANTABILITYOR FITNESS FORAPARTICULAR PURPOSE,ARE GIVEN IN CONNECTIONWITHTHIS
PUBLICATION.Although this information is believed to be accurate byASM,ASM cannot guarantee that favorable
results will be obtained from the use of this publication alone. This publication is intended for use by persons having
technical skill, at their sole discretion and risk. Since the conditions of product or material use are outside of ASMs
control, ASM assumes no liability or obligation in connection with any use of this information. No claim of any
kind, whether as to products or information in this publication, and whether or not based on negligence, shall be
greater in amount than the purchase price of this product or publication in respect of which damages are claimed.
THE REMEDYHEREBYPROVIDED SHALLBE THE EXCLUSIVEAND SOLE REMEDYOF BUYER,AND
IN NO EVENT SHALL EITHER PARTY BE LIABLE FOR SPECIAL, INDIRECT OR CONSEQUENTIAL
DAMAGES WHETHER OR NOT CAUSED BY OR RESULTING FROM THE NEGLIGENCE OF SUCH
PARTY.As with any material, evaluation of the material under end-use conditions prior to specification is essential.
Therefore, specific testing under actual conditions is recommended.

Nothing contained in this book shall be construed as a grant of any right of manufacture, sale, use, or reproduction,
in connection with any method, process, apparatus, product, composition, or system, whether or not covered by
letters patent, copyright, or trademark, and nothing contained in this book shall be construed as a defense against
any alleged infringement of letters patent, copyright, or trademark, or as a defense against liability for such in-
fringement.

Comments, criticisms, and suggestions are invited, and should be forwarded to ASM International.

Prepared under the direction of the ASM International Technical Book Committee (20032004), Charles A. Parker,
Chair.

ASM International staff who worked on this project include Charles Moosbrugger, Acquisitions Editor; Bonnie
Sanders, Manager of Production; Kathy Dragolich and Nancy Hrivnak, Production Editors; Kathryn Muldoon,
Production Assistant; and Scott Henry, Assistant Director of Reference Publications.

Library of Congress Cataloging-in-Publication Data

Humpston, Giles.
Principles of soldering / Giles Humpston, David M. Jacobson.
p. cm.
Includes bibliographical references and index.
ISBN 0-87170-792-6
1. Solder and soldering. 2. Brazing. I. Jacobson, David M. II. Title.

TS610.H84 2004
671.56dc22 2003058379
SAN: 204-7586

ASM International
Materials Park, OH 44073-0002
www.asminternational.org

Printed in the United States of America

 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

Contents

Preface .............................................................................................................................................. vii

About the Authors............................................................................................................................ ix
History ................................................................................................................................................ x
Chapter 1: Introduction................................................................................................................... 1
1.1 Joining Methods .................................................................................................................... 1
1.1.1 Mechanical Fastening ................................................................................................. 1
1.1.2 Adhesive Bonding ...................................................................................................... 2
1.1.3 Soldering and Brazing ................................................................................................ 3
1.1.4 Welding ....................................................................................................................... 4
1.1.5 Solid-State Joining ..................................................................................................... 4
1.1.6 Comparison between Solders and Brazes ................................................................. 5
1.1.7 Pressure Welding and Diffusion Bonding ................................................................. 8
1.1.7.1 Pressure Welding .................................................................................................. 9
1.1.7.2 Diffusion Bonding ................................................................................................ 9
1.2 Key Parameters of Soldering ............................................................................................. 12
1.2.1 Surface Energy and Surface Tension ....................................................................... 12
1.2.2 Wetting and Contact Angle ...................................................................................... 13
1.2.3 Fluid Flow ................................................................................................................ 18
1.2.4 Filler Spreading Characteristics ............................................................................... 19
1.2.5 Surface Roughness of Components ......................................................................... 22
1.2.6 Dissolution of Parent Materials and Intermetallic Growth ..................................... 24
1.2.7 Significance of the Joint Gap ................................................................................... 25
1.2.8 The Strength of Metals ............................................................................................ 27
1.3 The Design and Application of Soldering Processes ........................................................ 28
1.3.1 Functional Requirements and Design Criteria ........................................................ 28
1.3.1.1 Metallurgical Stability ........................................................................................ 29
1.3.1.2 Mechanical Integrity ........................................................................................... 29
1.3.1.3 Environmental Durability ................................................................................... 29
1.3.1.4 Electrical and Thermal Conductivity ................................................................. 30
1.3.2 Processing Aspects ................................................................................................... 30
1.3.2.1 Jigging of the Components ................................................................................ 30
1.3.2.2 Form of the Filler Metal .................................................................................... 31
1.3.2.3 Heating Methods ................................................................................................ 33
1.3.2.4 Temperature Measurement ................................................................................. 34
1.3.2.5 Joining Atmosphere ............................................................................................ 35
1.3.2.6 Coatings Applied to Surfaces of Components .................................................. 37
1.3.2.7 Cleaning Treatments ........................................................................................... 37
1.3.2.8 Heat Treatments Prior to Joining ....................................................................... 37
1.3.2.9 Heating Cycle of the Joining Operation ............................................................ 38
1.3.2.10 Postjoining Treatments ....................................................................................... 39
1.3.2.11 Postjoining Cleaning .......................................................................................... 39
1.3.2.12 Statistical Process Control ................................................................................. 42
1.3.3 Health, Safety, and Environmental Aspects of Soldering ....................................... 42
Chapter 1: Appendices ..................................................................................................................43
A1.1 Solid-State Joining with Gold, Indium, and Solder Constituents ........................... 43
A1.2 Relationship among Spread Ratio, Spread Factor, and Contact Angle of
Droplets ................................................................................................................. 44

iii
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

Chapter 2: Solders and Their Metallurgy................................................................................... 49

2.1 Survey of Solder Alloy Systems ........................................................................................ 51
2.1.1 Lead-Tin Solders ...................................................................................................... 56
2.1.2 Other Tin-Base Solders ............................................................................................ 58
2.1.3 Zinc-Bearing Solders ................................................................................................ 60
2.1.4 Gold-Bearing Solders ............................................................................................... 64
2.1.5 High-Lead Solders .................................................................................................... 72
2.1.6 Indium Solders ......................................................................................................... 73
2.2 Effect of Metallic Impurities .............................................................................................. 75
2.3 Application of Phase Diagrams to Soldering .................................................................... 77
2.3.1 Examples Drawn from Binary Alloy Systems ........................................................ 79
2.3.2 Examples Drawn from Ternary Alloy Systems ....................................................... 83
2.3.3 Complexities Presented by Higher-Order and Nonmetallic
Systems ................................................................................................................. 92
2.4 Depressing the Melting Point of Solders by Eutectic Alloying ........................................ 93
2.4.1 Liquid Alloys Based on Gallium ............................................................................. 93
2.4.2 Cadmium-Base Solders ............................................................................................ 93
2.4.3 General Features ....................................................................................................... 93
2.4.4 Implications for Lead-Free Solders ......................................................................... 95
Chapter 2: Appendices ..................................................................................................................96
A2.1 Conversion between Weight and Atomic Fraction of
Constituents of Alloys .......................................................................................... 96
A2.2 Theoretical Modeling of Eutectic Alloying ............................................................. 97

Chapter 3: The Joining Environment........................................................................................ 103

3.1 Joining Atmospheres ......................................................................................................... 103
3.1.1 Atmospheres and Reduction of Oxide Films ........................................................ 105
3.1.2 Thermodynamic Aspects of Oxide Reduction ....................................................... 106
3.1.3 Practical Application of the Ellingham Diagram .................................................. 107
3.1.3.1 Soldering in Inert Atmospheres and Vacuum .................................................. 107
3.1.3.2 Soldering in Reducing Atmospheres ............................................................... 109
3.1.3.3 Alternative Atmospheres for Oxide Reduction ................................................ 111
3.1.4 Forming Gas as an Atmosphere for Soldering ...................................................... 111
3.2 Chemical Fluxes for Soldering ......................................................................................... 111
3.2.1 Fluxes for Tin-Base Solders ................................................................................... 116
3.2.1.1 Soldering Fluxes That Require Cleaning ......................................................... 116
3.2.1.2 No-Clean Soldering Fluxes .............................................................................. 118
3.2.1.3 Measure of Cleaning Effectiveness: The Surface Insulation
Resistance (SIR) Test ................................................................................... 119
3.2.2 Fluxes for Unsolderable Metals ......................................................................... 120
3.2.2.1 Aluminum Soldering Fluxes ............................................................................ 121
3.2.2.2 Stainless Steel Soldering Fluxes ...................................................................... 122
3.2.2.3 Magnesium Soldering Flux .............................................................................. 122
3.2.3 High-Temperature Fluxes ....................................................................................... 122
3.3 Fluxless Soldering ............................................................................................................ 123
3.3.1 Oxide Formation and Removal .............................................................................. 124
3.3.2 Self-Dissolution of Solder Oxides ......................................................................... 125
3.3.3 Reduction of Solder Oxides by Hydrogen ............................................................ 126
3.3.4 Reduction of Solder Oxides by Atomic Hydrogen ............................................... 127
3.3.5 Mechanical Removal of Oxides (Ultrasonic Soldering) ....................................... 128
3.3.6 Reactive Gas Atmospheres for Reduction of Oxides ........................................... 130
3.3.7 Surface Conditioning Processes ............................................................................. 131
3.3.8 Fluxless Soldering Processes Considerations ........................................................ 132
3.3.8.1 Solderable Component Surfaces ...................................................................... 133
3.3.8.2 Preform Geometry ............................................................................................ 133

iv
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

3.3.8.3 Mechanically Enhanced Solder Flow .............................................................. 134

3.3.8.4 Metallurgically Enhanced Solder Flow ........................................................... 134
3.3.9 Example of a Fluxless Soldering Process Using In-48Sn
Solder .................................................................................................................. 135
3.3.10 Fluxless Soldering of Aluminum ........................................................................... 136
Chapter 3: Appendix ...................................................................................................................137
A3.1 Thermodynamic Equilibrium and the Boundary Conditions for
Spontaneous Chemical Reaction ........................................................................ 137

Chapter 4: The Role of Materials in Defining Process Constraints ...................................... 145

4.1 Metallurgical Constraints and Solutions .......................................................................... 147
4.1.1 Wetting of Metals by Solders ................................................................................ 147
4.1.2 Wetting of Nonmetals by Solders .......................................................................... 149
4.1.2.1 Solderable Coatings on Nonmetals .................................................................. 149
4.1.2.2 Active Solders .................................................................................................. 152
4.1.3 Erosion of Parent Materials ................................................................................... 153
4.1.4 Phase Formation ..................................................................................................... 154
4.1.5 Filler-Metal Partitioning ......................................................................................... 155
4.2 Mechanical Constraints and Solutions ............................................................................. 157
4.2.1 Controlled Expansion Materials ............................................................................. 159
4.2.1.1 Iron-Nickel Alloys ............................................................................................ 160
4.2.1.2 Copper-Molybdenum and Copper-Tungsten Alloys ........................................ 161
4.2.1.3 Copper-Surface Laminates ............................................................................... 162
4.2.1.4 Composite Materials ......................................................................................... 163
4.2.2 Interlayers ............................................................................................................... 164
4.2.3 Compliant Structures .............................................................................................. 165
4.2.4 The Role of Fillets ................................................................................................. 167
4.3 Constraints Imposed by the Components and Solutions ................................................. 168
4.3.1 Joint Area ................................................................................................................ 169
4.3.1.1 Trapped Gas ...................................................................................................... 169
4.3.1.2 Solidification Shrinkage ................................................................................... 173
4.3.2 Void-Free Soldering ............................................................................................... 173
4.3.3 Joints to Strong Materials ...................................................................................... 175
4.3.3.1 Joint Design to Minimize Concentration of Stresses ...................................... 175
4.3.3.2 Strengthened Solders to Enhance Joint Strength ............................................ 178
4.3.4 Thick- and Thin-Joint Gap Soldering .................................................................... 178
Chapter 4: Appendices ................................................................................................................180
A4.1 A Brief Survey of the Main Metallization Techniques ......................................... 180
A4.2 Critique of Void-Free Soldering Standards ........................................................... 183
A4.3 Dryness and Hermeticity of Sealed Enclosures .................................................... 184

Chapter 5: Advances in Soldering Technology ......................................................................... 189

5.1 Lead-Free Solders ............................................................................................................. 189
5.1.1 The Drive for Lead-Free Soldering ....................................................................... 190
5.1.2 Compatibility with Lead-Tin Solder ...................................................................... 191
5.1.3 Alternatives to Lead-Tin Solder ............................................................................ 191
5.1.4 Silver-Copper-Tin Ternary Phase Equilibria ......................................................... 193
5.1.5 Metallurgical, Physical, and Chemical Properties of
Lead-Free Solders ............................................................................................... 193
5.1.5.1 Surface Tension ................................................................................................ 193
5.1.5.2 Other Physical Properties ................................................................................. 194
5.1.5.3 Mechanical Properties ...................................................................................... 194
5.1.5.4 Corrosion Resistance ........................................................................................ 195
5.1.5.5 Susceptibility to Tin Pest and Tin Whiskers ................................................... 195

v
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

5.1.6 Process Window for Lead-Free Solders ................................................................ 196

5.1.7 Wetting and Spreading Characteristics of Lead-Free
Solders ................................................................................................................ 197
5.1.8 High-Melting-Point Lead-Free Solders ................................................................. 197
5.2 Flip-Chip Interconnection ................................................................................................. 199
5.2.1 The Flip-Chip Process ............................................................................................ 199
5.2.2 Characteristics of Flip-Chip Technology ............................................................... 202
5.2.3 Underfill .................................................................................................................. 203
5.2.4 Inspection ................................................................................................................ 203
5.2.5 Rework .................................................................................................................... 204
5.2.6 Self-Alignment of Flip-Chip Structures ................................................................ 204
5.2.7 Surface Topography ................................................................................................ 206
5.2.8 Step-Soldered Flip-Chip Interconnects .................................................................. 206
5.3 Solderability Test Methods and Calibration Standards ................................................... 207
5.3.1 Assessment of Wetting ........................................................................................... 207
5.3.2 Assessment of Spreading ....................................................................................... 210
5.3.3 Solderability Calibration Standards ....................................................................... 212
5.4 Amalgams as Solders ....................................................................................................... 214
5.4.1 Amalgams Based on Mercury ............................................................................... 215
5.4.2 Amalgams Based on Gallium ................................................................................ 216
5.4.3 Amalgams Based on Indium .................................................................................. 217
5.5 Strengthening of Solders .................................................................................................. 217
5.5.1 Grain Refinement ................................................................................................... 218
5.5.2 Oxide-Dispersion-Strengthened Solders ................................................................ 218
5.5.3 Composite Solders .................................................................................................. 219
5.6 Reinforced Solders (Solder Composites) ......................................................................... 222
5.7 Mechanical Properties and Numerical Modeling of Joints ............................................. 223
5.7.1 Measurement of Mechanical Properties ................................................................ 223
5.7.2 Numerical Modeling of Joints ............................................................................... 224
5.7.2.1 Dimensional Stability of Soldered Joints ........................................................ 224
5.7.2.2 Prediction of Joint Lifetime ............................................................................. 226
5.8 Solders Doped with Rare Earth Elements ....................................................................... 227
5.8.1 Effect of Rare Earth Additions on Solder Properties ............................................ 227
5.8.2 Implications for Soldering Technology ................................................................. 229
5.9 Diffusion Soldering ........................................................................................................... 230
5.9.1 Process Principles ................................................................................................... 230
5.9.2 Diffusion Soldering of Silver ................................................................................. 231
5.9.3 Diffusion Soldering of Gold .................................................................................. 233
5.9.4 Diffusion Soldering of Copper ............................................................................... 234
5.9.5 Practical Aspects ..................................................................................................... 234
5.9.6 Modeling of Diffusion-Soldering Processes .......................................................... 235
5.10 Advances in Joint Characterization Techniques .............................................................. 235
5.10.1 Ultrasonic Inspection (Scanning Acoustic Microscopy) ....................................... 235
5.10.2 X-Radiography ....................................................................................................... 236
5.10.3 Optical Inspection ................................................................................................... 237

Abbreviations and Symbols.......................................................................................................... 243

Index................................................................................................................................................ 245

vi
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

Preface

Since the first edition of Principles of Soldering and Brazing, published in 1993, the authors have
received valuable feedback from readers representing a wide range of technical interests. This has
prompted the decision to expand the text and organize it into two companion books, one covering
soldering and the other brazing. This first book primarily aims at providing information about
soldering in a form that is hopefully readily accessible and as easy to assimilate as possible. Priority
is given to the fundamental principles that underlie this field of technology rather than recipes for
making joints. The largely artificial distinctions between soldering and brazing are preserved
because, despite their many commonalities, it has been found that practicing engineers are either
concerned with soldering or brazing and seldom are involved with both simultaneously. The planned
companion book, Principles of Brazing, addresses this complementary need. A large proportion of
the literature on soldering and brazing may be charged with being heavy on description and light on
critical analysis. We have endeavored to redress the balance, while striving to avoid being unduly
simplistic or overly mathematical in our approach. Admittedly we may not always have succeeded in
this aim.
As in Principles of Soldering and Brazing, we have striven to maintain the focus on the fundamental
aspects of soldering and have deliberately avoided entering into specific joining technologies in detail.
At the same time, we recognize that the range and extent of the knowledge base of metal joining is
not immediately obvious, and it requires a fairly deep understanding of materials. To cite a single
example, nichrome (an alloy of nickel and chromium), which is a perfectly satisfactory and widely
used metallization for soldering, is rendered useless if the solder contains bismuth. If there is an evident
bias towards electronic and photonics applications, this reflects the recent professional orientation of
the authors. Some topics are inevitably not accorded due consideration, although it is hoped that
sufficient references are provided to enable the reader to pursue these further.
No attempt has been made to gather a comprehensive list of published papers. Those that are
included have been selected because they are useful basic texts, cover important subject matter, or
relate to exemplary pieces of work, whether in respect of methodology, technique, or other noteworthy
features. It was felt that if the value of the book depended on its bibliography, it would rapidly become
dated. The advent of computer search facilities and databases of scientific journal and conference
abstracts should enable the reader who wishes to find references on a specific topic to obtain further
information without too much difficulty. The search term lead-free solder will yield an astounding
25,000 publications in the public domain, virtually none of which are more than 10 years old.
The reader should note that all compositions given in this book are expressed in weight percentage
in accordance with the standard industrial practice. These have, for the most part, been rounded to
the nearest integer. The ratio of elements in intermetallic compounds, again by convention, refers to
the atomic weight of the respective constituents. The general convention used for specifying alloy
compositions is that adopted by the alloy phase diagram community, namely in the alphabetical order
of the elements, by chemical symbol. We have not been entirely rigorous in this regard as it is
sometimes helpful to group alloys by the dominant constituents. Minor additions to bulk compositions
are given in order of concentration; for example, Pb-62Sn-0.5Lu-0.02Ce.
Specific references are given with each chapter. For those wishing to read more generally on
particular topics, the authors would recommend the texts listed as Selected References at the end of
this preface.
Many phase diagrams are subject to ongoing research, resulting in continued improvement in the
accuracy and detail of the information. The most recent version of a diagram may be identified by
consulting the latest cumulative index of phase diagrams, published in the Cumulative Index of the
periodical Journal of Phase Equilibria (ASM International). This will refer to the source of the
thermodynamically assessed diagram of interest. The reader is advised that the four compendia of
binary phase diagrams published in the 1960s, 70s and 80s (colloquially referred to as Hansen,
Elliott, and Shunk) are now known to contain many errors and omissions.

vii
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

Information on new developments in soldering and brazing is scattered throughout a wide range
of periodicals, as reflected in the sources cited in the references appended to the individual chapters.
To keep abreast of the literature, the authors have found especially useful the following abstract
publications: Metals Abstracts and Science Abstracts. Technical libraries can provide automated
searches against specified key words as a monthly service.
We wish to thank our many colleagues and ex-colleagues for their helpful advice and encourage-
ment, particularly James Vincent, for insights into lead free soldering.

Giles Humpston
David M. Jacobson

SELECTED REFERENCES
Soldering
Brandon, D. G., and Kaplan, W. D., 1997. Joining Processes: An Introduction, John Wiley & Sons
Frear, D.R., Jones, W.B., and Kinsman, K.R., 1990. Solder Mechanics: A State of the Art As-
sessment, TMS
Hwang, J.S., 1996. Modern Solder Technology for Competitive Electronics Manufacturing,
McGraw-Hill
Hwang, J.S., 2001. Environment Friendly Electronics: Lead-Free Technology, Electrochemical
Publications
International Organization for Standardization (IOS), 1990. Welding, Brazing, and Soldering
Processes: Vocabulary, (ISO/DIS 857-2), ISO (currently under revision)
Klein Wassink, R.J., 1989. Soldering in Electronics, 2nd ed., Electrochemical Publications
Liebermann, E,, 1988. Modern Soldering and Brazing Techniques, Business News
Manko, H.H., 2002. Solders and Soldering, 4th ed., McGraw-Hill
Nicholas, M.G., 1998. Joining Processes: Introduction to Brazing and Diffusion Bonding, Kluwer
Academic
Strauss, R., 1998. SMT Soldering Handbook, 2nd ed., Butterworth-Heinemann
Thwaites, C.J., 1983. Capillary Joining: Brazing and SoftSoldering, Books Demand UMI
Woodgate, R.W., 1996. The Handbook of Machine Soldering: SMT and TH, John Wiley & Sons
Alloy Constitution
John, V.B., 1974. Understanding Phase Diagrams, Macmillan
Prince, A., 1966. Alloy Phase Equilibria, Elsevier
West, D.R.F., 1982. Ternary Equilibrium Diagrams, Chapman and Hall

viii
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

About the Authors

Giles Humpston took a first degree in metallurgy at Brunel Uni-

versity in 1982, followed by a Ph.D. on the constitution of solder al-
loys in 1985. He has since been employed by several leading indus-
trial companies, where he has been involved with determining alloy
phase diagrams and developing processes and procedures for pro-
ducing precise and high-integrity soldered, brazed, and diffusion-
bonded joints to a wide variety of metallic and nonmetallic materials.
His expertise extends to fine-pitch flip-chip, new materials develop-
ment, and packaging and interconnection for electronics, radio fre-
quency, and optical products. He is the cited inventor on more than
75 patents, the author of more than 60 papers, and recipient of six
international awards for his work on soldering and brazing.
Dr. Humpston is a licensed amateur radio enthusiast and has
published several articles and reviews on electronics, radio, and
computing. His other interests include exploring vertical-axis wind
turbines, building power inverters, flying radio controlled gliders, wine making, and growing bonsai.
He lives with his wife, Jacqueline, and their three children in a small village in Buckinghamshire,
England and San Jos (Silicon Valley), California.
David M. Jacobson graduated in physics from the University
of Sussex in 1967 and obtained his doctorate in materials sci-
ence there in 1972. Between 1972 and 1975 he lectured in
materials engineering at the Ben Gurion University, Beer-
Sheva, Israel, returning as Visiting Senior Lecturer in 1979-
1980. Having gained experience in brazing development with
Johnson Matthey Ltd., he extended his range of expertise to
soldering at the Hirst Research Centre, GEC-Marconi Ltd.,
which he joined in 1980. Currently, he holds the position of
senior research associate at the Centre for Rapid Design and
Manufacture, Buckinghamshire Chilterns University College in
High Wycombe. He is the author of more than 80 scientific and
technical publications in materials science and technology and
more than a dozen patents. He has been awarded three presti-
gious awards for his work on brazing.
Dr. Jacobsons principal outside interests are archaeology and architectural history, focusing on the
Near East in the Graeco-Roman period. He has published extensively in these fields on subjects that
extend to the numismatics and early metallurgy of that region. He recently completed a Ph.D. thesis
on Herodian architecture at Kings College, London, and teaches part-time in this subject area at
University College, London. Dr. Jacobson is married with two grown-up children and lives in
Wembley, England, close to the internationally famous football stadium.

Giles Humpston and David Jacobson are the coauthors of the book Principles of Soldering and
Brazing, which was published by ASM International in 1993, with more than 4000 copies sold.

ix
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

History

Origins of Solders and Soldering

The word solder derives from the Old French, soudure, which in turn stems from the Latin solidare,
which means to fasten together. Its earliest use in a completely English context as a noun meaning
a fusible metallic alloy used for uniting less fusible metal surfaces or parts dates to about 1350.
It is interesting to note that in 19th century English, just as in modern French, the l would have
been omitted and the word pronounced sod-der, a form that still persists in the United States of
America today.
Although the origin of solders and soldering is lost to antiquity, it is possible to speculate on how
the invention arose. Lead was first obtained as a by-product of silver production. Silver extraction
from ores involved cupellation of lead, and the base metal was then recovered from the litharge
[Tylecote 1976]. The softness and malleability of lead were clearly recognized, and there exist
examples of lead being used as a setting agent to fix posts in the ground and lock morticed stones.
It was observed that in this instance the lead filler could give a stronger joint than a simple friction
grip. Lead was used by the Mesopotamians (3000 B.C.) to join pieces of copper together, although
perhaps more by luck than design since pure lead does not wet copper at all readily. The Romans are
known to have produced lead separately from silver, taking advantage of the fact that this metal can
be easily extracted from its sulfide ore, galena, simply by roasting the mineral in air [Tylecote 1976].
The earliest examples of tin are Egyptian and date from 2000 B.C. What might be construed as a
manufactured solder alloy has been found in King Tutankhamuns tomb (1350 B.C.), although there
is some debate among scholars about the deliberateness of the metallurgy of this joint.
Solders comprising alloys of lead and tin were almost certainly used during the Iron Age [Tylecote
1962]. By the Roman Imperial period there is evidence, both from literary sources and from surviving
artifacts, that lead-tin solders were in regular use. Pliny the Elder (1st century A.D.) speaks of tertiarum,
an alloy of two parts of (black) lead and one part of white lead (tin) being used for joining metal pipes
[Pliny, Natural History xxxiv 161 (Rackham 1952)]. Pliny also remarks that the price of this alloy
is 20 denarii per pound. With 25 denarii (silver pieces weighing approximately 4 gm, or 0.14 oz, each)
to 1 gold aureaus of close to 8 gm (0.28 oz), the price of Roman solder works out at $70 per kilogram,
assuming that gold has maintained its purchasing power since Plinys day. The current price for the
same alloy (Pb-33Sn) is lower by an order of magnitude, which indicates how much more precious
solder was in antiquity.
An analysis of soldered joints in Roman artifacts has shown that both tin-rich and lead-rich alloys
were used. The solder in a force-pump from Roman Silchester contains lead to tin in a weight ratio
of close to 3 to 1, which is similar to the composition of plumbers solder [Tylecote 1962]. Elsewhere,
solders containing mainly tin (80 to 100% Sn), have been encountered in finds from 4th and 5th century
sites in Britain [Lang and Hughes 1991].
Soldering, unlike many Roman crafts, either did not die out during the Dark Ages or enjoyed an
early revival. The soldering iron, not mentioned at all in Classical times, was well known and in
widespread use by the early Middle Ages. Soldering was used for joining the lead strips in stained
glass windows, with the oldest complete examples being the Five Prophets windows in Augsburg
Cathedral that date from the late 11th century. From 1700 onwards it is clear that soldering was well
established with the appearance of tinsmiths and white-iron men as trades. Newcomens discovery
of the effectiveness of the internally condensing steam engine in 1708 is attributed to the faulty repair,
by soldering, of a blowhole in the cast bronze cylinder. This permitted a spray of external condenser
water into the cylinder and the development of the internal condenser; a design that was not superseded
until Watt developed the separate condenser nearly 70 years later.
Modern soldering practice dates to the early 20th century when improved extraction techniques,
which enabled exotic metals to be available at affordable cost, coupled with the appearance of alloy
phase diagrams, gave rise to the diversity of alloys now available.

x
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

REFERENCES

Tylecote, R.F., 1976. A History of Metallurgy, The Metals Society

Tylecote, R.F., 1962. Metallurgy in Archaeology: A Prehistory of Metallurgy in the
British Isles, Edward Arnold
Rackham, H., 1952. Natural History, Vol 10, Cambridge, MA, Translation of Pliny
1. Historia Naturalis, Vol 34 (No. 161)
Lang, J. and Hughes, M.J., 1991. Joining Techniques in Aspects of Early Metal-
lurgy, British Museum Occasional Papers, No. 17, British Museum, p 169177

xi
 ASM International is the society for materials
engineers and scientists, a worldwide network
dedicated to advancing industry, technology, and
applications of metals and materials.

ASM International, Materials Park, Ohio, USA

www.asminternational.org
This publication is copyright ASM International. All rights reserved.
Publication title Product code
Principles of Soldering #06244G

To order products from ASM International:

Online Visit www.asminternational.org/bookstore

Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Fax 1-440-338-4634
Customer Service, ASM International
Mail
9639 Kinsman Rd, Materials Park, Ohio 44073-0002, USA
Email CustomerService@asminternational.org
American Technical Publishers Ltd.
27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX,
In Europe United Kingdom
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)
www.ameritech.co.uk
Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550

Terms of Use. This publication is being made available in PDF format as a benefit to members and
customers of ASM International. You may download and print a copy of this publication for your
personal use only. Other use and distribution is prohibited without the express written permission of
ASM International.
No warranties, express or implied, including, without limitation, warranties of merchantability or
fitness for a particular purpose, are given in connection with this publication. Although this
information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be
obtained from the use of this publication alone. This publication is intended for use by persons having
technical skill, at their sole discretion and risk. Since the conditions of product or material use are
outside of ASM's control, ASM assumes no liability or obligation in connection with any use of this
information. As with any material, evaluation of the material under end-use conditions prior to
specification is essential. Therefore, specific testing under actual conditions is recommended.
Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale,
use, or reproduction, in connection with any method, process, apparatus, product, composition, or
system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this
publication shall be construed as a defense against any alleged infringement of letters patent,
copyright, or trademark, or as a defense against liability for such infringement.
Principles of Soldering
Giles Humpston, David M. Jacobson, p1-47
DOI:10.1361/prso2004p001
CHAPTER 1
Introduction
1.1 Joining Methods
SOLDERING AND BRAZING represent one
of several types of methods for joining solid
materials. These methods may be classied as:
Mechanical fastening
Adhesive bonding
Soldering and brazing
Welding
Solid-state joining
Other methods, such as glass/metal sealing, elec-
trostatic welding, and so forth, are dealt with
elsewhere [Bever 1986].
Schematics of these joining methods are given
in Fig. 1.1. These different methods have a num-
ber of features in common but also certain sig-
nicant differences. For example, soldering and
brazing are the only joining methods that can
produce smooth and rounded llets at the pe-
riphery of the joints. The joining methods are
listed in the rst paragraph of this chapter in the
order in which they lead to fusion of the joint
surfaces and tend toward a seamless joint.
Because soldering and brazing lie in the middle
of this sequence, they share several features with
the other methods. For example, soldered and
brazed joints can be endowed with the advan-
tageous mechanical properties of welded and dif-
fusion-bonded joints; at the same time they can
be readily disassembled, without detriment to the
components, like mechanically fastened joints.
These features make soldering and brazing highly
versatile.
The principal characteristics of the various
joining methods are summarized in the para-
graphs that follow.
1.1.1 Mechanical Fastening
Mechanical fastening involves the clamping
together of components without fusing the joint
Copyright 2004 ASM International
All rights reserved.
www.asminternational.org
surfaces. This method often, but not always, re-
lies on the use of clamping members such as
screws and rivets. In crimping, the components
are keyed together by mechanical deformation.
Characteristic features of mechanical fasten-
ing include:
A heating cycle is generally not applied to the
components being joined. A notable excep-
tion is riveting, where the rivets used for
clamping are heated immediately prior to the
fastening operation. On subsequent cooling
the rivets shrink, causing the components to
be clamped tightly together.
The reliance on local stressing to effect join-
ing requires thickening or some other means
of reinforcement of the components in the
joint region. This places a severe restriction
on the joint geometries that may be used and
imposes a weight penalty on the assembly.
Another constraint on permissible joint con-
gurations is the need for access to insert the
clamping member.
The method usually requires special mechani-
cal preparation, such as drilling holes, ma-
chining screw threads, or perhaps chamfer-
ing of abutting surfaces, in the case of
components to be crimped.
The choice of suitable joint congurations is
highly dependent on service conditionsfor
example, whether or not leak tightness is re-
quired. Joints may be designed to accommo-
date thermal expansion mismatch between
the components in the assembly. In the ex-
treme case, joints can be made to permit com-
plete freedom of movement in the plane per-
pendicular to the clamping member, as
applied to the shplates used to couple train
rails.
The electrical and thermal conductance across
the joint is a function of the effective area that
is in contact. This depends on many other
2 / Principles of Soldering
parameters, such as the clamping force and
the materials used, and in service the con-
ductance is unlikely to be constant.
1.1.2 Adhesive Bonding
Adhesive bonding involves the use of a poly-
meric material, often containing various addi-
tives, to stick the components together. The
process involves a chemical reaction, which may
simply comprise exposure of the adhesive to air,
leading to the formation of a hydrogen-type bond
Fig. 1.1 Principal methods for joining engineering materials
between the cured adhesive and the respective
components. The original interfaces of the joint
are preserved in this type of bonding process.
Characteristic features of adhesive bonding in-
clude:
It is inherently a low-stress joining method
because it is carried out at relatively low tem-
peratures and most adhesives have high com-
pliance.
A diverse range of methods are available for
curing adhesives.
 Chapter 1: Introduction / 3

The geometry of the components tends not to
be critical.
Constraints apply to the geometry of the ac-
tual joint; in particular large areas and very
narrow gaps are necessary to ensure me-
chanical integrity.
Joints tend to be weak when subject to forces
that cause peeling. For this reason, adhesive
joints are frequently used in combination with
mechanical fasteningfor example, in air-
frame assembly.
Joint integrity tends to be sensitive to the at-
mosphere of the service environment and to
the state of cleanliness of the mating surface.
The service temperature range of adhesively
bonded joints is usually limited, as is their
compatibility with solvents.
Joints usually possess poor electrical and ther-
mal conductivity, although by loading the or-
ganic adhesive with metal particles moderate
conductance can be achieved that approaches
that of some solder alloys.
Polymer chemistry is a rapidly evolving sci-
ence. As a result, some very advanced adhesives
have appeared on the market in the last few years
with properties highly tailored for particular func-
tion in the electronics industry. These include
thermally conductive adhesives, electrically con-
ductive adhesives, and anisotropically conduc-
tive adhesives. Table 1.1 provides an indication
of the product range of thermally conductive ad-
hesives available from one manufacturer. These
advanced materials are being augmented by the
development of polymers with high imperme-
ability to moisture and low thermal expansion
coefficients for use as electronic packaging ma-
terials. Other polymers have been developed that
function as both ux and underll material for
ip-chip applications. More exciting advances
will no doubt continue to become available.
1.1.3 Soldering and Brazing
Soldering and brazing involve using a molten
ller metal to wet the mating surfaces of a joint,
with or without the aid of a uxing agent, leading
to the formation of metallurgical bonds between
the ller and the respective components. In these
processes, the original surfaces of the compo-
nents are eroded by virtue of the reaction oc-
curring between the molten ller metal and the
solid components, but the extent of this ero-
sion is usually at the microscopic level (<100
m, or 4000 in.). Joining processes of this type,
by convention, are dened as soldering if the

Table 1.1 Selection of commercially available conductive adhesives, used in place of solder for
some applications
4030SD 4030LD 4030Hk 4030SR 4130HT 5030P 6030Hk(a)

Paste Properties
Viscosity at 10 rpm, 3045 2540 2545 3550 3040 3040 30
kcP(b) (typical)
25 C (77 F) shelf life, 6 6 6 6 6 6 6
months
Paste density, g/cm3 3.17 3.42 4.35 3.07 3.33 3.7 3.84.5
Processed Properties
Die shear(c)(d), 56 (800) 56 (800) 42 (600) 98 (1400) 105 (1500) 105 (1500) 105 (1500)
kgf/cm2 (psi)
Bulk resistivity(d), 40 40 13 160 40 25 610
 cm
Thermal conductivity, 15 15 3540 5 20 2025 3060
W/m K
Youngs modulus, 0.9 (125) 0.9 (125) 0.9 (125) ... 1.8 (250) 2.5 (350) ...
GPa (ksi)
Thermal expansion, 15 (28) 15 (28) ... ... 17 (30) 13 (23) ...
106/C (106/F)
Rework temperature(e), 100 (212) 100 (212) 100 (212) 150 (300) 150 (300) 200250 (f)
C (F) (390480)
Designed product use Highly Large area Very high Solder Higher- Withstands Extremely
conductive device thermal and replacement temperature wire bonding high
adhesive attach electrical for SMT applications at 250 C thermal/
conductivity (480 F) electrical
adhesive(g) conductivity

Note in particular the cited thermal conductivity values are comparable to those of many solders. (a) A novel thermoset development material capable of room-temperature
storage. (b) kcP is 1000 centipoise (1 Pa s). (c) 6.48 mm (0.255 in.) die, ceramic. (d) 175 C/15 min prole. (e) 70 kPa (10 psi) force. (f) Rework as for typical epoxies.
(g) Maximum die size 7  7 mm (0.28  0.28 in.). Source: Multicore Solders Ltd.
4 / Principles of Soldering
ller melts below 450 C (840 F) and as brazing
if it melts above this temperature.
Characteristic features of soldering and braz-
ing include:
All brazing operations and most soldering
operations involve heating the ller and joint
surfaces above ambient temperature.
In most cases, the service temperature of the
assembly must be lower than the melting tem-
perature of the ller metal.
It is not always necessary to clean the sur-
faces of components prior to the joining op-
eration because uxes are available that are
capable of removing most oxides and organic
lms. However, there are penalties associ-
ated with the use of uxesfor example, the
residues that they leave behind, which are
often corrosive and can be difficult to re-
move.
The appropriate joint and component geom-
etries are governed by the ller/component
material combination and by service require-
ments (need for hermeticity, stress loading,
positional tolerances, etc.). Complex geom-
etries and combinations of thick and thin sec-
tions can usually be soldered or brazed to-
gether.
Intricate assemblies can be produced with
low distortion, high fatigue resistance, and
good resistance to thermal shock.
Joints tend to be strong if well lled, unless
embrittling phases are produced by reaction
between the ller metal and the components.
Soldered and brazed joints can be endowed
with physical and chemical properties that
approximately match and, in some cases,
even exceed those of the components, but
usually have limited elevated-temperature
service and stability.
Fillets are formed under favorable condi-
tions. These can act as stress reducers at the
edges of joints that benet the overall me-
chanical properties of the joined assembly.
Soldering and brazing can be applied to a wide
variety of materials, including metals, ceramics,
plastics, and composite materials. For many ma-
terials, and plastics in particular, it is necessary
to apply a surface metallization prior to joining.
1.1.4 Welding
Welding involves the fusion of the joint sur-
faces by controlled melting through heat being
specically directed toward the joint. Commonly
used heating sources are plasma arcs, electron
beams, lasers, and electrical current through the
components and across the joints (electrical re-
sistance). Filler metals may be used to supple-
ment the fusion process for components of simi-
lar composition, as for example when the joint
gap is wide and of variable width. In that situ-
ation, the ller is often chosen to have a mar-
ginally lower melting point than the components
in order to help ensure that it completely melts.
Characteristic features of welding include:
Welding invariably involves a heating cycle,
which tends to be rapid, and a very wide
variety of welding processes are available.
Welding cannot be used to join metals to
nonmetals or materials of greatly differing
melting points. There are exceptions to this,
but these are generally limited to precise com-
binations of materials and highly specic
welding methods.
Joint geometries are limited by the require-
ment that all joint surfaces are accessible to
the concentrated heat source.
Welded joints may approach the physical in-
tegrity of the components, but are often in-
ferior in their mechanical properties, particu-
larly fatigue resistance. This is due to stress
concentrations produced by the high thermal
gradients developed during joining and the
relatively rough surface texture of welds.
The heating cycle usually affects the micro-
structure and hence the properties of the com-
ponents over a macroscopic region around
the joint, called the heat-affected zone (HAZ).
The HAZ is often inuential in determining
the properties of welded joints.
Welding tends to distort the components in
the region of the HAZ. This is associated
with the thermal gradients developed through
the use of a concentrated heat source to fuse
the joint surfaces.
1.1.5 Solid-State Joining
The term solid-state joining covers a very wide
range of joining processes. The two extremes are
pressure welding and diffusion bonding. Pres-
sure welding, at its simplest, involves physically
deforming two abutting, faying surfaces to dis-
rupt any intervening surface lms and enable
direct metal-to-metal contact. Diffusion bonding
in its purest form merely requires placing two
faying surfaces in contact and heating the as-
sembly until the voids at the interface have been

removed by diffusion. Further details of process
parameters for diffusion bonding of gold and
indium are given in Appendix A1.1. Pressure
welding generally works better if the compo-
nents are heated (e.g., friction welding), and dif-
fusion bonding is usually greatly accelerated by
the application of pressure or mechanical agita-
tion (e.g., thermosonic ball and wedge bonding,
see Fig. 1.2) to force a greater area of the faying
surfaces into contact.
Solid-state joining constitutes a subject in its
own right, quite separate from soldering and braz-
ing, which rely on liquid-state metal joining.
However with the development of the diffusion-
soldering and diffusion-brazing processes, which
are a hybrid of the two, some consideration of
solid-state joining, in particular diffusion bond-
ing, is merited. Pressure welding is sometimes
used to prepare ller metals in various geom-
etries and to tack preforms in position. Occa-
sionally, conventional ller metals are used to
produce a pressure weld or diffusion bond be-
tween dissimilar metals in a solid-state joining
process. For this reason, further information on
both pressure welding and diffusion bonding is
included in section 1.1.7 in this chapter.
Characteristic features of solid-state joining:
This method generally involves heating the
joint to a temperature below the melting point
of the components.
Pressure welding is often a much faster pro-
cess than soldering or brazing (<1 s), the
Fig. 1.2 An electronic module in which the semiconductor
dies have been interconnected using ne wire at-
tached by thermocompression bonding
Chapter 1: Introduction / 5
extreme being explosive welding, while dif-
fusion bonding is much slower (>10 min).
The joints have no llets.
The service temperature of joined assemblies
can be higher than the joining temperature
and tend toward the melting point of the com-
ponents.
Solid-state joining is limited in application to
specic combinations of materials that pro-
vide specic combinations of mechanical or
diffusion characteristics.
Of all the joining methods, they are the least
tolerant of poor mating of the joint surfaces.
Joint surfaces need to be scrupulously clean
because solid-state joining is a uxless pro-
cess.
The properties of solid-state joints can ap-
proach those of the parent materials.
Further details on pressure welding are given in
section 1.1.7.1 and on diffusion bonding in sec-
tion 1.1.7.2 in this chapter.
1.1.6 Comparison between
Solders and Brazes
In many respects it is fruitful to consider sol-
ders together with brazes. This integrated treat-
ment can be justied on metallurgical grounds.
These two classes of ller cannot be demarcated
by a temperature boundary as is habitually done:
conventionally, solders are dened as ller met-
als with melting points below 450 C (840 F)
and brazes as having melting points above this
value. This distinction has a historical origin: the
earliest solders were based on alloys of tin, while
brazes were based on copper-zinc alloys (see
History of Soldering in this volume and His-
tory of Brazing in the planned companion vol-
ume Principles of Brazing for a brief historical
background of solders and brazes, respectively).
The type of metallurgical reaction between a
ller and parent metal is sometimes used to dif-
ferentiate soldering from brazing. Solders usu-
ally react to form intermetallic phases, that is,
compounds of the constituent elements that have
different atomic arrangements from the elements
in solid form. By contrast, most brazes form
solid solutions, which are mixtures of the con-
stituents on an atomic scale. However, this dis-
tinction does not have universal validity. For ex-
ample, Ag-Cu-P brazes react with steels to form
the interfacial phase of Fe3P in a similar manner
to the reaction of tin-base solders with iron and
steels to form FeSn2. On the other hand, solid
6 / Principles of Soldering
solutions form between silver-lead solders and
copper just as they do between the common sil-
ver-base brazes and copper. Also, there exist
brazes for aluminum that melt below 450 C
(840 F).
Soldering and brazing involve essentially the
same bonding mechanism: that is, reaction with
the parent material, usually alloying, to form me-
tallic bonds at the interface. In both situations,
good wetting promotes the formation of llets
that serve to enhance the strength of the joints.
Similar processing conditions are required, and
the physical properties are comparable, provided
the same homologous temperature (the tempera-
ture at which the properties are measured as a
fraction of the melting temperature expressed in
degrees Kelvin) is used for the comparison.
The perpetuation of the distinction of solders
from brazes on the basis of the 450 C (840 F)
boundary has arisen from the signicant gap that
exists between the melting points of available
solder alloys, the highest being Au-3Si, which
melts at 363 C (685 F), and the lowest tem-
perature standard braze, the Al-4Cu-10Si alloy,
which melts at 524 C (975 F) but, being a
noneutectic alloy, is fully liquid only above 585
C (1085 F). Eutectic alloys are dened in Chap-
ter 2, section 2.3; for the present, it shall suffice
to state that eutectic alloys are akin to pure met-
als in melting and freezing at the same tempera-
Fig. 1.3 Principal solder alloy families and their melting ranges
ture. The temperature ranges of the principal sol-
der and braze alloy families are shown in Fig. 1.3
and 1.4.
For most purposes, the temperature gap be-
tween solders and brazes is substantially wider
than 160 C (290 F). This is because the gold-
base solders are very expensive and are largely
limited in use to the high added-value manufac-
turing of the electronics industry. Removing the
high-gold-content alloys from consideration, the
highest-melting-point solders are the lead-rich
alloys, which melt at about 300 C (570 F). The
lowest-melting-point brazes that are used com-
mercially in signicant quantities are the rea-
sonably ductile aluminum-silicon alloys, which
melt at 577 C (1070 F). A selection of eutectic
alloys with melting points in the temperature
interval 300 to 550 C (570 to 1020 F) that at
some time or other have been promoted as sol-
ders and brazes are listed in Table 1.2. They are,
without exception, brittle and often contain one
or more volatile constituents, notably magne-
sium, cadmium, or zinc. Some multicomponent
alloys that have been developed and are designed
to ll the temperature gap are described in Chap-
ter 2. However, none of them are readily avail-
able from commercial sources.
The dearth of ller metals with melting points
in the range 300 to 550 C (570 to 1020 F) is
not necessarily a handicap; techniques are avail-
 Chapter 1: Introduction / 7

able for making joints using molten ller metal
with effective melting points in this temperature
interval. Transient-liquid-phase diffusion bond-
ing is one such example and is discussed in Chap-
ter 5, section 5.9.
From the maps of solders and brazes in Fig.
1.3 and 1.4, it might appear that there are many
more solders than brazes. In fact, the contrary is
true. The alloys that are specically indicated in
these gures are the mostly eutectic composi-
tions or those characterized by minimum melting
ranges. Most commercially used solders are in-
cluded because these are almost all of eutectic
composition. However, whole families of brazes
have been omitted because there is no eutectic in
the alloy system; instead they exhibit complete
intersolubility. Examples are the copper-nickel,
silver-gold, silver-palladium, and silver-gold-

Fig. 1.4 Principal braze alloy families and their melting ranges

Table 1.2 Selected eutectic alloys that are offered as high-melting point solders and low-melting
point brazes
Solder composition(a), Melting point
wt% C F Problems

5Ag-95Cd 340 644 Toxic fumes, volatile

75Au-25Sb 356 673 High cost, brittle
88Au-12Ge 361 682 High cost, brittle
97Au-3Si 363 685 High cost, brittle
6Al-94Zn 381 718 Volatile, brittle
48Al-52Ge 424 795 High dross, brittle
36Al-37Mg 450 842 Volatile, brittle
75Pb-25Pd 454 849 Poor fatigue resistance, brittle
56Ag-44Sb 485 905 Volatile, brittle
58Au-42In 495 923 High cost, brittle
68Al-27Cu-5Si 524 975 Difficult to clean, brittle
23Ag-53Cd-24Cu 525 977 Toxic fumes, volatile, brittle
24Cu-76Sb 526 979 Volatile, brittle
62Cd-38Cu 549 1020 Toxic fumes, volatile, brittle

(a) All compositions given are in weight percent.

8 / Principles of Soldering
palladium alloys. Alloys in such systems melt
over a temperature range that varies with the
composition.
The higher process temperatures needed to
make a brazed joint have important conse-
quences because more thermal activation energy
is present. These are:
More extensive metallurgical reaction be-
tween the ller metal and the substrate. Sol-
ders typically do not dissolve more than a few
microns of the component surfaces, whereas
brazes often dissolve tens of microns. Larger
changes in the composition of the ller metal
therefore occur during brazing, which in turn
signicantly affects the uidity of and wet-
ting by the molten ller as well as the prop-
erties of the joint.
Greater reactivity with the atmosphere sur-
rounding the workpiece. All other factors be-
ing equal, brazes are less tolerant of oxidiz-
ing atmospheres than solders, but, for the
same reasons, are also better suited to clean-
ing by reducing atmospheres. When joints
are made in air with the aid of a ux, the
greater reactivity of brazes means that a
higher proportion of ux to ller metal is
generally required. In consequence, ux-
cored solders are adequate for use in air, while
brazing rods intended for use in ambient at-
mosphere must be provided with a thick ex-
ternal coating of ux. Fluxes are discussed in
Chapter 3, section 3.2.
Most, but not all, soldering and brazing pro-
cesses are performed at small excess tempera-
tures above the melting point of the ller metal,
commonly referred to as the superheat. Much
higher process temperatures are occasionally
used where it is desirable to exploit thermal ac-
tivation. For example, tin-containing solders can
wet and join nonmetallized ceramics provided
the solder incorporates an active ingredient, such
as titanium, and the alloy is heated above about
900 C (1650 F) [Kapoor and Eagar 1989]. Al-
though the freezing point of the solder is un-
changed at about 250 C (480 F), such acti-
vated solders have several of the characteristics
of brazes at the process temperature. Further in-
formation on these alloys can be found in Chap-
ter 4, section 4.1.2.2. On the other hand, the
aluminum-germanium eutectic alloy melting at
424 C (795 F) behaves like a typical braze on
aluminum and copper surfaces, although by con-
ventional denition, it is classed as a solder.
Several general features distinguish the ma-
jority of solders from common brazes, namely:
Most commercial solders are of eutectic com-
position because there is usually a need to
minimize the processing temperature while
maintaining reasonable uidity of the molten
ller. Also, solders are intrinsically soft and
must be conferred with optimal mechanical
properties; generally these are achieved by
having a ne-grained microstructure, which
is a characteristic feature of a true eutectic
alloy.
Most brazes, by comparison, possess mutual
solid solubility between their constituents and
are therefore offered with a wide range of
compositions and melting ranges. The low
degree of intersolubility and the propensity to
form intermetallic compounds possessed by
solder alloys are related to their constituent
elements having a noncubic crystal symme-
try.
Solders nd application at temperatures at a
fraction between 50 and 90% of their melting
point in degrees Kelvin, under strain levels
that often exceed 10%. At these relatively
high temperatures, the alloys are not metal-
lurgically stable and the joint microstructure
tends to change with time. Brazes tend to be
used at temperatures that are relatively much
lower and usually below half their melting
point in degrees Kelvin.
These points are discussed in further detail in
Chapters 2 and 3, and reference should also be
made to the planned companion volume Prin-
ciples of Brazing. Notwithstanding the differ-
ences, solders and brazes operate on similar prin-
ciples, and hence the frequent use of the
collective term ller throughout this book has
some justication.
1.1.7 Pressure Welding and
Diffusion Bonding
Solid-state joining methods are not new, and
examples of gold-base artifacts fabricated using
pressure welds have been dated to 1000 B.C.
[Tylecote 1968], while a cup and chalice deco-
rated by diffusion bonding have been dated to
3200 B.C. [Tylecote 1967]. Although more re-
cent interest in welding has been almost totally
dominated by fusion welding processes, both
pressure welding and diffusion bonding continue
to satisfy niche applications because of the
unique combination of process and joint param-

eters they offer. Some solid-state joining proce-
dures are a combination of pressure welding and
diffusion bonding, as is evident from the funda-
mental characteristics of each.
1.1.7.1 Pressure Welding
Pressure welding utilizes pressure to rupture
surface lms at the joint interface and also to
extrude virgin parent metal between islands of
surface contamination so that metallic bonding
can take place. Thus, the process is character-
ized by high pressures, applied for short peri-
ods of time, on metals that may be either cold
or hot. By necessity, bulk plastic deformation
of the metals will occur. Possibly the most
common examples of pressure welding that are
pertinent to solders and brazes are butt weld-
ing to join lengths of wires, roll-bonding, and
indentation welding.
In pressure welding, it is generally accepted
that bond formation is controlled by the extent of
deformation of the faying surfaces. The term
threshold deformation is used extensively in
the literature on this subject and is dened as the
minimum deformation needed to achieve any
bonding, although the strength of a bond at this
level of deformation is generally much less than
that of the parent metal (see Fig. 1.5).
The bonding process can be described as four
consecutive stages:
1. Removal of surface contamination and
breakup of brittle surface layers, in particu-
lar oxides. This is frequently accomplished
Fig. 1.5 The strength of pressure-welded joints as a function
of the deformation induced during the bonding pro-
cess. Below the threshold deformation level, no joining occurs.
With increasing deformation the joint strength also increases even-
tually up to that of the parent materials. Note that the joining
process modies the properties of the parent material as it will
work-harden when mechanically deformed.
Chapter 1: Introduction / 9
by mechanically abrading the surface im-
mediately prior to bonding. Adsorbed water
is believed to be the main surface contami-
nant and responsible for preventing bonding
if the deformation is less than 8%. Typically,
40% deformation is required to affect a
sound joint when bonding metals in atmo-
spheres other than vacuum.
2. Establishment of physical contact between
regions of uncontaminated metal as virgin
metal extrudes between gaps in the ruptured
surface lms
3. Activation of contacting atoms to form a
metallic bond. The contact area determines
the extent of bonding.
4. Subsequent atom rearrangement as a con-
sequence of postweld heat treatment and/or
stress relaxation
Pressure welding is particularly effective when
joining dissimilar metals. For good weldability,
the softer metal should have the more brittle and
stronger oxide lm and vice versa. The hard oxide
layer can then promote and assist in the breakup
of the surface layers on the harder metal, but is
itself easily ruptured by the yielding metal sup-
porting it. For example, the oxide on aluminum
fullls the requirements of strength and brittle-
ness compared to the oxides of most other metals,
while the metal is relatively soft, and therefore
pressure welding of aluminum to other metals oc-
cursatlowerdeformationsthanwhenautogenous-
ly welded. Also, the different deformation char-
acteristics of dissimilar metals may result in
interfacial movement that will enhance bonding
compared to autogenous welding.The use of pres-
sure welding to fabricate ductile preforms of
brittle alloys using partitioned constituents is dis-
cussed further in Chapter 4, section 4.1.5.
1.1.7.2 Diffusion Bonding
Diffusion bonding relies on a combination of
temperature and time to remove voids from the
free interfaces between two abutting metal parts.
Fundamentally, the process is dened as one in
which no plastic deformation of the components
being joined takes place, although it is usual to
apply some pressure to ensure that the nominally
at faying surfaces are indeed in intimate con-
tact. Typical process conditions are durations of
up to several hours at temperatures that may be
as high as two-thirds of the melting point of the
least thermally stable metal in the bonded couple.
The use of long times at relatively high tem-
10 / Principles of Soldering
peratures necessitates some form of atmosphere
control to preserve surface cleanliness. Soft
(roughing) vacuum and controlled atmospheres
are equally suitable.
Since diffusion processes are the main mecha-
nisms for bonding, with no means for the physi-
cal displacement of any intervening nonmetallic
surface lms, there are two major considerations
in diffusion bonding. The rst is the necessity to
ensure that these lms do not constitute a barrier
to atom migration. Secondly, in bimetallic sys-
tems the formation of intermetallic compounds
and porosity arising from inequality of diffusion
rate by different species (Kirkendall porosity)
must be controlled. Table 1.3 presents some of
the better-known direct diffusion-bonding com-
binations of metals and metalloids.
Diffusion bonding does not take place by one
dominant mechanism, but is a consequence of
one or more possible mechanisms that often op-
erate in parallel. Each mechanism results in ma-
terial (or void) transport so that the surface en-
ergy associated with the interface is progressively
reduced as joining proceeds. Some possible
mechanisms include:
Plastic yielding of surface asperities
Creep of the surface asperities
Surface and volume diffusion altering the
shape of the voids
Grain boundary and volume diffusion from
the bond interface to reduce the void vol-
ume
A detailed theoretical treatment of solid-state
diffusion bonding is provided by Hill and Wallach
[1989]. In practice, the extent of bonding and the
rate at which it is achieved is governed both by
material properties (such as surface, grain bound-
ary, and volume diffusion coefficients, creep and
yield strength, etc.) and process parameters, of
which the four main variables are:
Pressure. Adequate pressure is required to
achieve contact on an atomic scale by local-
ized deformation of asperities on the nomi-
nally at surfaces being joined and also to
allow creep mechanisms to contribute to
bonding.
Temperature. Thermal energy promotes faster
bonding since plastic deformation, creep, and
all diffusion mechanisms are temperature de-
pendent. Typically, temperatures around 0.7
Tm are used, where Tm is the absolute melting
temperature of the lowest melting point com-
ponent. Heating rates are not critical.
Time. Creep and diffusion mechanisms are
also strongly time dependent, and there must
be a sufficient interval afforded to allow for
void closure by material transfer. As the tem-
perature increases, the time required for bond-
ing decreases.
Surface condition. The height and fre-
quency of surface asperities dening the joint
will control the extent of initial surface con-
tact and thus inuence the bonding rate. Gen-
erally, atter and more highly polished sur-
faces are easiest to bond. The removal of
surface contamination and thick oxides is es-
sential prior to bonding since these will either
persist at the joint line or must be removed by
solution in the parent material as bonding
proceeds. It therefore takes higher relative
temperatures and pressures to bond alumi-
num-base alloys than copper-base alloys.
Paradoxically, tungsten, titanium, and tanta-
lum do not exhibit diffusion-barrier prob-
lems, despite the fact that their oxides or car-
bides are very stable. Certain titanium alloys
are of particular industrial interest since they
can be diffusion bonded and superplastically
shaped in one processing operation, made
possible because above about 900 C (1650
F), titanium can dissolve oxygen into its vol-
ume as fast as a surface scale can form.
Diffusion-bonded joints normally exhibit 80
to 100% of parent material strength. One per-
ceived problem with diffusion bonding is the
thermal cycle time, particularly compared with
fusion welding. However, a complex welding
job might take several hours to prepare and jig
the components, in which case diffusion bonding
might offer advantages. Unlike most welding pro-
cesses, the process-time curve for diffusion bond-
ing is almost at in relation to the job size, be-
cause the process time is essentially independent
of joint area provided adequate compressive
stress is applied.
Interlayers are often used to diffusion bond dis-
similar metals. For example, silver foil is used in
bonding steel to titanium and nickel foil is often
used to bond high-carbon steel to itself and other
materials.Agold ash applied to a precleaned sur-
face permits diffusion bonding of nickel and cop-
per components. An obvious extension to this ap-
proach is making use of interlayers that melt,
thereby increasing diffusion rates, helping to ll
the joint gap and disrupt surface lms; this is the
characteristic feature of diffusion-soldering and
diffusion-brazing processes. Diffusion soldering
is discussed in Chapter 5, section 5.9 and diffusion
Table 1.3 Matrix of metals and metalloids that can be diffusion bonded
X indicates potentially workable combinations
Stainless Cast
Ag Al Au Be Cr Co Cu Fe Nb Mg Mo Ni Pd Pt Ta Ti U V W steel iron Carbides Graphite

Ag X X ... X ... ... X ... ... ... ... ... ... ... ... X ... ... ... X ... ... ...
Al X X ... X ... ... X X ... X ... X ... ... ... X ... X ... X ... ... ...
Au ... ... X X ... ... X ... ... ... ... X ... ... ... ... ... ... ... X ... ... ...
Be X ... X X ... ... X ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
Cr ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... X ... ... ... ...
Co ... ... ... ... ... X ... ... ... ... ... X X ... ... ... ... ... ... ... ... ... ...
Cu X X X X ... ... X X X ... X X ... ... ... X ... X ... X ... ... ...
Fe X ... ... ... ... ... X ... ... ... X X X X ... X ... ... X ... ... ... ...
Nb ... ... ... ... ... ... X ... X ... X X ... ... X X ... X X X ... X ...
Mg ... X ... ... ... ... ... ... ... X ... X ... ... ... ... ... ... ... ... ... ... ...
Mo ... ... ... ... ... ... X ... X ... X X X X X X ... ... X ... ... ... ...
Ni ... X X ... ... X X X X X X X X X ... X X ... X X X X ...
Pd ... ... ... ... ... ... X X ... ... X X X ... ... ... ... ... X X ... ... ...
Pt ... ... ... ... ... ... ... ... ... ... X X ... X ... ... ... ... X ... ... ... ...
Ta ... ... ... ... ... ... ... ... X ... X ... ... ... X X ... X ... ... ... ... ...
Ti X X ... ... ... ... X X X ... X X ... ... X X X ... ... X ... ... X
U ... ... ... ... ... ... ... ... ... ... ... X ... ... ... X X ... ... ... X ... ...
V ... X ... ... ... ... X ... X ... ... ... ... ... X ... ... X ... ... ... ... ...
W ... ... ... ... X ... ... X X ... X X X X ... ... ... ... X ... X ... ...
Stainless X X ... X ... ... ... X ... ... X X X ... ... X ... ... ... X ... ... X
steel
Cast iron ... ... ... ... ... ... ... X ... ... ... X ... ... ... ... ... ... X ... X ... ...
Carbides ... ... ... ... ... ... ... ... X ... ... X ... ... ... ... ... ... ... ... ... X ...
Graphite ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... X ... ... ... X ... ... ...

Adapted from [Feature 1976]

Chapter 1: Introduction / 11
12 / Principles of Soldering
brazing in the planned companion volume Prin-
ciples of Brazing.
1.2 Key Parameters of Soldering
The quality of soldered joints depends strongly
on the combination of ller and component ma-
terials, including surface coatings that may be
applied to the components, and also on the pro-
cessing conditions that are used. It is precisely
for this reason that a sound understanding of the
metallurgical changes accompanying the se-
quence of events that occur in making soldered
joints is so vital for developing reliable joining
processes.
Soldering technology has generally evolved
in an empirical manner, largely by trial and er-
ror. Theoretical principles have helped to fur-
nish insights, guidelines, and qualitative expla-
nations for this technology, but have rarely
provided reliable data for use in the design of
joining processes. The basic difficulty is that
the real situation is highly complex, as it brings
into play a large number of variables, some of
which may not be easy to recognize. Among
the relevant factors are the condition of the
solid surfaces (i.e., the nature of any oxides or
other coatings, surface roughness, etc.) and the
temperature gradients that develop during the
joining operation, as well as the metallurgical
reactions involving the ller and parent mate-
rials and also the chemical reactions with
uxes, where these are used.
Another key aspect of joining with llers is the
manner and extent of ow of the molten ller
into the joint. These are inuenced by:
Dimensions of the joint
Spread characteristics of the ller metal
Surface condition of the components
The limitation of theory in accounting for ob-
served behavior is well illustrated by the clas-
sical model of wetting and spreading. This model
nevertheless does provide useful concepts and
insights. It is given a detailed treatment by Har-
kins [1952] and is not repeated here. For the
purposes of the present discussion, it suffices to
outline the main features of this model.
In the classical model of wetting, the surface
of the solid is taken to be invariant as a liquid
droplet spreads over it. That is to say, the reaction
between the liquid and the solid components
across their common interface is considered to
be negligible. It is also assumed that the com-
position and other characteristics of the solid and
liquid components, likewise, do not change with
time. This assumption is not generally valid, as
is shown in this section.
Modeling of self-alignment and other wetting
and spreading processes can be done with a pro-
gram entitled Surface Evolver. A search of the
World Wide Web using this as the keyword
should identify a site from which the software
can be obtained.
1.2.1 Surface Energy and
Surface Tension
The concepts of surface energy and surface
tension are briey reviewed in this section. Fig-
ure 1.6 provides a simplied representation of
the atomic structure of a solid close to one of its
free surfaces. The atom at position A, in the bulk
of the solid, has a balanced array of neighboring
atoms, whereas atom B at the surface of the solid
is lacking in neighbors above it, apart from the
occasional vapor molecule and, therefore, it has
some unsaturated bonds.
The potential energy of atoms at the free sur-
face, such as B, is higher than the energy of
atoms within the bulk of the solid, such as A, by
the energy of the unsaturated bonds. The aggre-
gate of this excess energy that is possessed by
atoms in the vicinity of the free surface consti-
tutes the surface energy of the solid. In a similar
manner, a liquid also possesses a surface energy,
which is directly manifested in the tendency to
draw up into drops. If small, the droplets are
Fig. 1.6 Simplied diagram of surface energies. Atom B, at
the surface, has unsaturated bonds and thus a higher
energy than atom A. This difference in energy is the origin of
surface energy SV.

perfect spheres. Because a sphere has the small-
est surface-to-volume ratio of any shape, it is
clear that the surface energy of a liquid is greater
than its volume energy. In the classical model,
when a liquid spreads over a surface, the volume
remains constant, because evaporation and re-
action with the substrate are excluded. There-
fore, only surface-energy changes must be con-
sidered.
A surface of a liquid acts like an elastic skin
covering the volume; in other words, the sur-
face is in a state of tension. The tensile force
(F), known as surface tension (), is dened as
the force acting at right angles to a line of unit
length (L) drawn in the surface. The relation-
ship between surface tension and surface en-
ergy under specic conditions can be seen as
follows.
Consider a liquid lm of length L and width
W. Apply a force F at a barrier AB, as shown in
Fig. 1.7, parallel to one surface of the lm, so as
to extend the liquid lm a distance x.
The increase in area of the lm is x L.
The work done in obtaining this increase is the
mathematical product of the force applied times
the distance moved, or F x.
The work done by the liquid lm in opposing
the increase in area, under isothermal conditions
(i.e., constant temperature), is 2  x L, where
 is the surface tension force acting on each
surface at the prescribed temperature.
At a xed temperature (under isothermal con-
ditions):
Fx  2xL
Rearranging, F/L  2 or F/L   for each
surface.
Fig. 1.7 Diagram used to explain the relationship between
surface energy and surface tension
Chapter 1: Introduction / 13
Thus, surface energy is equivalent to surface
tension under isothermal conditions. In the mod-
ern metric or International System of Units (SI),
the unit of surface energy is joule per square
meter (J/m2) and that of surface tension is new-
ton per meter (N/m). Because these parameters
are properties of an interface (e.g., between liq-
uid and air), surface energy and tension must be
dened with reference to the appropriate pair of
materials that meet at the interface, and the test
conditions, such as temperature and atmosphere,
also must be specied.
1.2.2 Wetting and Contact Angle
According to the classical model of wetting,
the liquid will spread over a solid surface until
the three surface tensionsbetween the liquid
droplet and the solid substrate, the liquid droplet
and the atmosphere, and the substrate and the
atmosphereare in balance as shown in Fig. 1.8.
According to the balance of forces:
SL  SV  LV cos  (Eq 1.1)
where SL is the surface tension between the
solid and liquid, LV is the surface tension be-
tween the liquid and vapor, SV is the surface
tension between solid and vapor, and  is the
contact angle of the liquid droplet on the solid
surface.
Equation 1.1, known as the wetting or Youngs
equation, shows that  < 90 corresponds to the
condition SV > SL.This imbalance in surface ten-
sion (i.e., surface energy) provides the driving
force for the spreading of liquid over the solid sur-
face and diminution of the unwetted surface area.
The contact angle  provides a measure of the
quality of wetting. Thus, if 90 <  < 180, some
wetting is said to occur, but a liquid droplet will
not spread on the surface with which it is in
contact. On the other hand, if  < 90, a liquid
droplet will wet the substrate and also spread
Fig. 1.8 Surface tension forces acting when a liquid droplet
wets a solid surface, according to the classical model
14 / Principles of Soldering
over an area dened by the contact angle .
Clearly, the area of spreading will increase with
decreasing contact angle. For further details of
the interrelationship between these two param-
eters, refer to the Appendix A1.2.
Rewriting Eq 1.1 in terms of cos :
SV  SL
cos 
LV
Thus, wetting is improved by decreasing  as
cos increases; that is, as  approaches 0. Cos 
may be maximized by:
Increasing SV
Decreasing SL
Decreasing LV
The term SV can be maximized for a given
solid by cleaning the surfaces. The presence of
adsorbed material, such as water vapor, dust, and
other nonmetallic surface lms on a metal sur-
face, markedly reduces SV and correspondingly
raises the contact angle . Therefore, it is im-
portant in soldering operations that joint surfaces
are clean and metallichence the need for uxes
or protective atmospheres to achieve and then
sustain this condition.
The term SL is a constant at a xed tempera-
ture for a particular solid-liquid combination, ac-
cording to the classical model of wetting. This
parameter can be reduced by changing the com-
position of the materials system, as can be seen
from Fig. 1.9.
However, changing the composition is not usu-
ally easy to achieve in practice because compo-
nent materials are specied to fulll certain other
functional requirements. Fortunately, SL is
Fig. 1.9 Wetting angle of lead-tin solder on copper at 10 C
above the melting point, 1 min after reow using
rosin mildly activated (RMA) ux, as a function of lead concen-
tration. Adapted from Liu and Tu [1998]
highly temperature dependent and usually de-
creases rapidly with increasing temperature
[Schwartz 1987, Table 1.4], thereby providing a
ready means of controlling spreading.
The term LV is a constant at a xed tempera-
ture and pressure for a particular liquid-vapor
combination, but can be varied by altering the
composition and pressure of the atmosphere. Al-
though the composition of the atmosphere used
for the joining operation is known to affect the
contact angle, in practice it is often easier to
promote spreading by reducing the pressure of
the atmosphere. This is one of the reasons for the
popularity of vacuum-based joining processes.
In general, the relative magnitudes of the sur-
face energies are SV > SL > LV. For water
wetting on mica, subjected to an atmosphere of
water vapor, the following values have been mea-
sured [Tabor 1969]:
SV  0.183 N/m
SL  0.107 N/m
LV  0.073 N/m
Thus, cos   (0.183 0.107)/0.073  1 (within
the limits of experimental error) and the contact
angle,   0.
The surface energies of pure metals correlate
quite well with their melting points. This is to be
expected because the refractoriness of metals re-
ects the strength of the bonds between adjacent
atoms in the lattice, and the asymmetry between
this and a free atom is responsible for surface
energy and tension.
Table 1.4 Surface roughness (Ra) of cold-rolled
copper after sanding with wet silicon carbide
paper or polishing with a colloidal suspension
of alumina in water
Nominal particle Ra obtained on
Abrasive size, m cold-rolled copper, m
80 grit 200 2.2
240 grit 63 0.95
400 grit 23 0.51
1200 grit 5 0.23
Polishing alumina 0.05 0.012
For comparison, copper surfaces on electronic component leads usually have an
oriented Ra of approximately 0.1 m (4 in.).
 Chapter 1: Introduction / 15

It is possible to calculate the surface tension
of solder alloys from thermodynamic prin-
ciples using data for the pure metals. It mostly
varies as an essentially linear relationship be-
tween the values for the two pure metals. Sol-
ders that exhibit poor spreading even at large
superheats above the liquidus temperature, such
as Sn-40In and Sn-65Bi, have similar surface
tension to other solders, but the wetting is con-
trolled by the reaction kinetics at the solder/
substrate interface, which are less favorable
[Park et al. 2000].
The wetting equation (Eq 1.1) applies when
the liquid is practically insoluble in the solid
over which it spreads (i.e., the solubility is less
than 0.1%). For binary metal systems where
this condition is satised (e.g., tin-chromium),
it has been shown that the wetting equation can
be reduced to:
to an equilibrium height h where the capillary
force balances the hydrostatic force (as shown in
Fig. 1.10), such that:
2LV cos 
h (Eq 1.2)
gD
where is the density of the liquid and g is the
acceleration due to gravity.
As might be expected, experimental assess-
ment of capillary rise of solders reveals that cap-
illary rise is less than predicted by theory, al-
though the general principles of Eq 1.2 are
substantiated. Meniscus rise is usually greatest
for solders that exhibited the lowest contact angle
and surface tension and in the narrowest gaps.
However, for many solders the correlation with
gap width is weak. This has been studied and
ultimately was attributed to be due to a rapid

cos   1  k
TmS
TmL
1
increase in voids in the joint as the gap width was
decreased progressively (see Chapter 4, section
4.3.1). Indeed, in practical terms capillary ow
in narrow gaps is largely dictated by the efficacy
of the ux/ller combination used [Vianco and
where k is a constant equal to approximately 0.3, Rejent 1997].
T Sm is the melting point of the solid metal, and T Lm The actual situation in soldering is much more
is the melting point of the liquid metal. This ex- complex than that represented by Eq 1.2 and the
pression has been veried experimentally [Eusta- classical wetting model. The irreversible nature
thopoulos and Coudurier 1979]. Higher-order of spreading and the time dependence of contact
metal systems (ternary, quaternary, etc.) are con- angle that is commonly observed are at variance
siderably more complex, and the wetting equation with this simple model. These and other depar-
cannot be truncated to such a simple form.Amore tures from the classical model occur because the
sophisticated analysis of wetting that takes into
account the inuence of certain microscopic fea-
tures, including the inuence of local defects and
van der Waals forces, is provided by de Gennes
[1985]. However, this is still a continuum analysis
and does not consider the local atomic environ-
ment. Indeed, it has further been suggested that
Youngs equation is only valid under certain spe-
cial cases and there are some difficulties with the
theoretical denition of solid-surface tension
[Xian 2000]. Further academic endeavor will
hopefully resolve these issues.
So far, this chapter has idealized ller spread
over a single surface. In a joint there are always
two facing surfaces. If both contact angles are
less than 90, the surface energies will give rise
to a positive capillary force that will act to ll the
joint. For a pair of vertical parallel plates D mm
apart and partly immersed in a liquid, the cap-
illary force per mm length of joint is equal to Fig. 1.10 Rise of a liquid between two parallel plates by
2LV cos . Under this force, the liquid will rise capillary force
16 / Principles of Soldering
joining process almost invariably involves a de-
gree of chemical reaction between the ller metal
and the solid surface, which is neglected in the
conventional model. This is clearly demon-
strated in a study [Schwartz 1987] that showed
that the contact angle for various liquid metals on
freshly cleaned beryllium generally decreased
with time, over a timescale of several minutes,
at a xed temperature. Predictably, perhaps, it
was also found that the contact angle decreased
with increasing temperature and the atmosphere
in which the test was conducted also made a
difference.
Reactions between a ller metal and the sub-
strate often result in dissolution of the surface of
the substrate; this process usually leads to the
growth of new phases. Frequently, these phases
are intermetallic compounds that either appear
distributed throughout the joint or form as layers
adjacent to the surface of the solid substrate.
The energy of formation of an intermetallic
layer by reaction between a molten ller and a
solid substrate has been calculated by Yost and
Gr
Romig [1988] and Wang and Conrad [1995].
The energy of formation considered is the ther-
modynamic function known as the Gibbs free
energy. This function and its properties are
briey explained in the Appendix A3.1. In or-
der to simplify the analysis, Yost and Romig
Gr is the Gibbs free energy of the reaction.
limited their consideration to the clean sur-
faces of pure metals, wetted by liquids of el-
emental metals, in the absence of uxes, to
form binary interfacial phases. It was demon-
strated that the free energy of formation of in-
termetallic phases by reaction of liquid anti-
mony, cadmium, and tin with solid copper was
approximately two orders of magnitude larger
than the energy release created by the surface-
energy imbalance during the advance of a
spreading solder droplet, which is exclusively
considered in the classical model. Therefore, in
these cases, and probably more generally in
soldering processes, the Gibbs free energy
change that occurs on reaction by a ller with
the substrate is demonstrably the dominant
driving force for wetting. Empirical evidence
for this is provided, for example, by the fact
that the measured contact angle of molten ger-
manium on silicon carbide at 1430 C (2600
F) is approximately 120, whereas that of
molten silicon on this ceramic at the same tem-
perature is 38 [Li and Hausner 1991]. The
substantial difference in the two contact angles
cannot be accounted for by the difference in
LV in the wetting equation (Eq 1.1). It can
only be due to the greater intersolubility of sili-
con with silicon carbide. This example clearly
demonstrates that the simple classical wetting
equation cannot be relied on for a quantitative
description of wetting, contact angle, or spread-
ing. A more direct example is provided by Fig.
1.11. In this simple system of copper-silicon
braze wetted onto graphite, the nal contact
angle is insensitive to alloy composition, but
the rate of attainment of equilibrium wetting is
directly related to the concentration of silicon,
which is the active ingredient in the braze.
Modications have been proposed to incor-
porate the Gibbs free energy change accompa-
nying metallurgical reaction into the classical
wetting equation by adding additional terms. In
particular, the following equation has been de-
veloped for the contact angle in reactive wetting
[Kritsalis, Coudurier, and Eustathopoulos 1991;
Laurent, Chatain, and Eustathopoulos 1991]:
SL  SL
0
cos   cos 0   (Eq 1.3)
LV LV
where SL is the solid-liquid interfacial energy
after reaction, SL
0
is the interfacial energy before
reaction, 0 is the contact angle before reaction,
and
Equation 1.3 is probably more of theoretical in-
terest than practical value because its use pre-
supposes knowledge not only of the Gibbs free
energy of reaction, but also values of the before
and after contact angle or interfacial energy.
The effect of metallurgical interaction be-
tween ller and the component material in pro-
Fig. 1.11 Contact angle of copper-silicon brazes of different
composition on vitreous carbon substrates dem-
onstrating the effect of driving force of alloying on wetting rate
and the dependence of the equilibrium wetting angle on the
reaction product. Adapted from Landry, Rado, and Eustathopou-
los [1996]

moting wetting is exploited in active ller met-
als: the addition of a small fraction of a reactive
metal such as titanium, hafnium, or zirconium to
llers enables them to wet and spread over ce-
ramic materials. In this instance, wetting of and
reaction with the ceramic are inextricably linked.
Activated ller alloys are discussed further in
Chapter 4, section 4.1.2.2 and the planned com-
panion volume Principles of Brazing.
Although a low contact angle is used as an
index for judging the quality of wetting, there are
situations where higher contact angles are pre-
ferred. This is illustrated in Fig. 1.12, which
shows two joints, one between two component
surfaces of unequal area and the other between
component surfaces that entirely correspond. In
the rst case, a low contact angle serves to form
a gentle concave llet, which enhances the me-
chanical properties of the joint. In the other con-
guration, a low contact angle encourages the
formation of a neck in the joint, which can be a
source of weakness. A contact angle close to 90
will eliminate this problem.
Fig. 1.12 Effect of contact angle on llet formation and joint lling. Low contact angles tend to be preferred when external llets
can form. In other geometries, higher contact angles result in lower stress concentrations.
Chapter 1: Introduction / 17
It is usually presumed that a substrate surface
is perfectly wettable, or at least will be when the
ux has had sufficient time to perform its clean-
ing action. Sometimes, however, the majority
of the surface area will be wettable and the re-
mainder covered with an array of nonwettable
patches. This can be due to inadequate surface
preparation or an incorrect choice of ux or pro-
cess conditions, or it is simply an inherent feature
of the substrate material. An example of the latter
is some of the new generation of metal-matrix
composites that comprise a nely dispersed mix-
ture of a metal and powder of refractory com-
pounds such as Be/BeO and Au/TiN. The net
effect of these nonwettable patches is to cause
local impediments to wetting and spreading and
an increase in the effective contact angle, as
shown in Fig. 1.13.
A further point to be aware of in connection
with wetting is that a situation can arise where
the molten ller is physically prevented from
achieving its equilibrium contact angle, as, for
example, when a solder droplet is conned to a
18 / Principles of Soldering

Fig. 1.13 Effect of nonwettable surface features on the contact angle of solder on copper. Data of Yost, Hosking, and Frear [1993]
augmented by the authors. Lead-tin solder wetted onto a copper surface containing embedded nonwettable particles
1020 m (400800 in.) in diameter (RMA ux, 180 C, or 356 F).

small metallized area. This is commonly en- liquid ow, dV/dt, between a pair of horizontal
countered on electronic circuit boards, where sol- parallel plates, length l, separated a distance D,
der droplets are constrained to individual metal under a pressure P per unit area transverse to the
pads. In this situation, the pad is often too narrow plates. The viscosity of the liquid is .
to accommodate the spherical metal cap that
would form if this restriction did not apply. The
dV PD3
enforced wetting angle imposes a pressure on the 
droplet that is often adequate to cause the solder dt 12l
to ow along the length of the conductor that
leads away from the pad and, more seriously, to It is assumed that the liquid front will advance
lift and ow under the solder resist that sur- at a rate (dl/dt) equal to the mean velocity of ow,
rounds the pad. An analysis of the pressure aris- that is:
ing from a nonequilibrium contact angle, using
the classical wetting model, is given by Klein
Wassink [1989].
dl
dt

D
1 dV
dt

PD2
12l

1.2.3 Fluid Flow From the wetting equation (Eq 1.1), under iso-
thermal conditions the change in surface energy
The wetting equation determines the degree of as a unit area of a surface becomes wetted by the
wetting for a given liquid-solid combination, but liquid is:
will not provide information on the rate of wet-
ting. Knowledge of the contact angle(s) enables
SV  SL  LV cos 
the surface energy to be determined and hence
the force that acts to ll the joint gap with liquid.
The liquid will ow into the joint under this force Therefore, the change in surface energy when the
at a rate that is governed by its viscosity. Simple pair of parallel plates becomes wetted is:
uid-ow theory assumes that:
There is no interaction between the liquid and 2l (SV  SL )  2l LV cos 
the solid surfaces with which it is in contact.
All surfaces are smooth and perfectly clean.
Flow is laminar, not turbulent. It follows that the force acting on the liquid to
cause it to wet the plates is:
For a detailed treatment of this subject, the reader
is referred to a paper by Milner [1958]. This
2l LV cos 
chapter merely quotes the expression (given as F
Eq 8 in Milners paper) for the volume rate of l

so that the pressure is:
2LV cos 
P
D
and the velocity of ow of the liquid into the
space between two parallel surfaces, of separa-
tion D, according to this simple model is given
by:
dl LVD cos 
 (Eq 1.4)
dt 6l
Equation 1.4 shows that the rate of liquid ow
increases when:
The liquid-vapor surface tension, LV, in-
creases.
The joint gap, D, increases.
The contact angle  decreases.
Filler metal viscosity is low.
Andrade [1952] derived an empirical formula
relating viscosity, when molten, to the molecular
weight of metals (in SI units):
m  1.65  107 T 0.5
m A
0.5
V 2/3
where m is the viscosity at the melting point of
the metal, Tm is the absolute melting point, V is
the molar volume, and A is the atomic weight of
the metal. By assuming limited solubility be-
tween the constituents in an alloy and applying
the rule of mixtures, it is thereby possible to
provide an estimate of the theoretical viscosity of
solder.
Rates of ow calculated from Eq 1.4 for mol-
ten solders in joints 50 m (2000 in.) wide are
typically 0.3 to 0.7 m/s (1 to 2.3 ft/s). In other
words, a joint 5 mm (0.2 in.) long will be lled
in a time of the order of 0.01 s. This implies that
joint lling by the molten solder occurs virtually
instantaneously and that transient effects asso-
ciated with uid ow can generally be neglected
in joining processes. De Gennes [1985] offers a
more developed model of the dynamics of liquid
spreading, in which the surface-energy driving
force is opposed by viscous drag and surface
irregularities. Joint lling times of the order of
0.1 s are routinely measured on instruments used
Chapter 1: Introduction / 19
for determining solderability. It should be noted
that, although the rate of lling is proportional to
the joint gap D, the driving force for lling, ac-
cording to the classical model, is inversely pro-
portional to D; that is, these two aspects of lling
act in opposition.
This simple model needs to be modied in
situations where interfacial reaction occurs while
liquid spreading is proceeding. Models that have
been tentatively proposed for this situation have
been reviewed by Meier, Javernick, and Edwards
[1999]. Currently, the lack of relevant data on
reaction-rate kinetics, interfacial energy before
and after reaction, and diffusion hampers a more
complete understanding of spreading, and also
wetting, of molten llers, especially where in-
terfacial reaction with solid components is sig-
nicant. However, much can be learned from
empirical observations, as shown in the follow-
ing section.
1.2.4 Filler Spreading Characteristics
Molten ller metals do not all have the same
spreading characteristics, although, with few ex-
ceptions, the degree of spread over an ideal
substrate increases as the temperature is raised
and the environment is made more reducing. In
this context, an ideal substrate, suitable for
reference purposes, needs to be dened. This is
understood to possess a perfectly clean metal
surface that is highly wettable by the ller metal
under consideration, but with which it does not
signicantly alloy. Any alloying reactions will be
highly specic to the combination of materials in
question, so that the substrate will lose its ideal
characteristics.
An example of a substrate that approximates
the ideal, and that has been used by the authors
in comparative soldering assessments, com-
prises a at glass plate sputter coated with 0.1 m
(4 in.) of chromium and overlaid with 0.1 m
(4 in.) of gold. The chromium represents a metal
that is essentially insoluble in most solders, and
the gold layer provides this reactive metal with
protection against oxidation. The gold layer is
sufficiently thin to not signicantly alter the com-
position of a solder pellet as it spreads over the
substrate [Humpston and Jacobson 1990].
Eutectic composition alloys are often re-
garded as having the best spreading character-
istics, and this is frequently one of the reasons
cited for their selection in preference to hypo-
eutectic and hypereutectic compositions. The su-
perior spreading of alloys of eutectic composi-
20 / Principles of Soldering
tion in comparison with off-eutectic alloys of the
same system, which is often observed, can be
explained by the different melting characteristics
in the two cases. An alloy of eutectic composi-
tion melts instantly. Spreading of the molten al-
loy is then driven by interaction with the sub-
strate [Ambrose, Nicholas, and Stoneham 1992].
In the case of a noneutectic ller metal, melting,
wetting, and spreading commences before the
alloy is entirely molten and it will tend to be
somewhat viscous. Under these conditions,
movement of the ller will be relatively sluggish.
By the time the alloy is completely molten, the
ller will have partly alloyed with the substrate,
and the driving force for spreading will have
been diminished. Eutectic composition alloys
also have lower viscosity than adjacent compo-
sitions when completely molten; further details
are given in Chapter 2, section 2.3.1.
Whether or not the ller alloy is of eutectic
composition is of much less importance to the
phenomenon of spreading than the composition
per se. The spreading of a ller metal depends
greatly on the elemental constituents present and
their relative proportions. The authors have com-
pared the spreading characteristics, as a function
of excess temperature above the melting point
(superheat), of all combinations of the elements
bismuth, indium, lead, silver, and tin when used as
eutectic solders on ideal substrates [Humpston
and Jacobson 1990]. The results presented in Fig.
1.14 show that the area of spreading increases at
an accelerating rate as a function of the excess
temperature above the melting point of the solder.
Furthermore, this study has demonstrated that
there is a consistent ranking order for these ele-
ments in their ability to promote spreading
namely, tin > lead > silver > indium > bismuth.
This ranking order is maintained even for ternary
and quaternary solders and when applied to a
range of substrates, in air, using mild uxes.
Although high uidity of a ller metal is a de-
sirable property when it is required to ow into the
joint gap of a heated assembly by capillary action,
it is not quite so important when the preferred
method of applying the ller is to sandwich a thin
foil preform between the components, which are
then joined together in an appropriate heating
cycle. For this type of conguration, a high degree
of spreading is detrimental to joint lling, as the
ller tends to ow out of the joint. Placement of
the ller metal and its inuence on joint lling is
discussed in Chapter 4, section 4.3.1.1.
In a vacuum or neutral protective atmosphere,
the spreading of a ller metal will tend to be in-
ferior to that obtained in air in the presence of a
chemical ux. This is to be expected in view of the
limited effectiveness of these environments: nei-
ther a vacuum nor a protective atmosphere is usu-
ally capable of removing oxides that form on the
surface of components or the ller while exposed
to air before the joining operation. In both cases,
the spreading is inferior to that achieved in the
presence of an active ux that can remove the sur-
face oxide [Humpston and Jacobson 1991].
Detailed investigation reveals that even an os-
tensibly simple parameter such as contact angle
exhibits somewhat complex behavior. Figure
1.15 shows the contact angle of Pb-60Sn solder
on copper at 194 C (381 F), protected by a
chemically inert ux, as a function of wetting
time. The results indicate that there are at least
four distinct stages of wetting and spreading.
During the rst 10 s of melting, the solder forms
a spherical cap and there is subsequent rapid
spreading with a corresponding decrease in the
contact angle. The contact angle then tempo-
rarily stabilizes, as a dynamic balance is struck
among the growth rate of interfacial intermetal-
lic compounds, the diffusion rate in the molten
pool of liquid, and the efficacy of the ux in
cleaning the substrate surface. This situation per-
sists for about 500 s. Thereafter, a further re-
duction in contact angle occurs. This is thought
to be associated with a progressive change in the
composition of the solder resulting in a sudden
change in the intermetallic compounds formed in
the halo at the edge of the solder pool and hence
a change, in this case a further decrease, in the
wetting angle. Finally, after many minutes, the
contact angle reaches a settled value as the solder
pool becomes saturated with the substrate. The
molten liquid then commences isothermal freez-
ing as the solidus temperature progressively in-
creases, owing to the alloying with the substrate
[Wang and Conrad 1995].
Some attempt has been made to undertake a
theoretical analysis of the kinetics of spreading of
a molten metal over a wettable solid surface. The
current theoretical approach considers the spread-
ing of an inert sessile drop on a smooth and per-
fectly wetted substrate as a balance between sur-
face-energy drive and viscosity impedance [de
Gennes 1985]. However, comparison of this theo-
retical model with practical experience reveals a
number of aws with the model, not the least is
that measured ow rates are in the region of four
orders of magnitude slower than predicted by
theory (see Fig. 1.16). These discrepancies are
mostly due to the added metallurgical and
Fig. 1.14 Spread characteristics of binary solder alloys on an ideal substrate as a function of excess temperature above the melting
point. The substrate is a at, microscope slide, sputter metallized with 0.1 m (4 in.) of chromium overlaid with 0.1 m
(4 in.) of gold. Spread ratio is dened in Appendix A1.2.
22 / Principles of Soldering
Fig. 1.15 Contact angle of lead-tin solder on copper as a
function of wetting time, using an inert ux and
low superheat. There are four distinct stages of wetting, the last
being the equilibrium contact angle that is obtained using more
typical process parameters.
physical complexity in ller metal wetting and
spreading as discussed earlier and are among the
simplifying assumptions of this model.
Nevertheless, the de Gennes model does pre-
dict some interesting dependencies of spreading.
First, the initial spreading of molten ller metal is
described by the imbalance between Youngs
forces and viscous damping. This model also pre-
dicts a relative insensitivity of spreading to excess
temperature in ller/substrate combinations that
wet well. Continued research in this area may
achieve a complete mathematical description of
wetting and spreading by ller metals that takes
into account isothermal solidication and the
physical and chemical state of the surface [Am-
brose, Nicholas, and Stoneham 1993].
1.2.5 Surface
Roughness of Components
The roughness of joint surfaces can have a
signicant effect on both the wetting and spread-
Fig. 1.16 Comparison between measured and predicted rates of spreading by molten solder. The large discrepancy arises because
the models are based on uid ow and do not take into account the metallurgical driving force for spreading.
ing behavior of a ller. Surface roughness re-
duces the effective contact angle *, where * is
related to , the contact angle for a perfectly at
surface through the relation:
cos *  r cos 
where:
Actual area of rough surface
r
Plan area
At the same time, by producing a network
of ne channels, the texturing may increase the
capillary force acting between the ller and the
component surfaces. Both phenomena will tend
to aid spreading. A directionally oriented sur-
face texture promotes preferential ow parallel
to the channeling [Nicholas and Crispin 1986].
From surface-energy calculations it is possible
to show that if the instantaneous contact angle of
the molten ller is less than the surface angle
(i.e., the root angle of V-shaped valleys), then
profuse wetting tends to occur along the valleys.
This is a frequent observation and, indeed, rep-
resents a problem when soldering to rough ma-
chined surfaces in that the ller does not spread
uniformly in all directions.
Exactly the same situation pertains to mi-
croscopically rough surfaces. An example of a
microscopically rough surface is thick electro-
plated copper that has been deposited without
brighteners (surface leveling agents) or a steel
surface that has been etched so as to furrow
out the grain boundaries. The resulting surface
comprises roughly spherical nodules, which
therefore have a continuous network of valleys

between the nodule peaks. Prolometer mea-
surements indicate that the many valleys have
contact angles in the region of 15 and are
therefore capable of enhancing spreading. The
difference here is that rather than a contact line
that undulates over a rough surface, the wet-
ting front has a fractal, or lacelike character.
The solder wets the channels between the
peaks and gives the wetted solder an extensive
halo ahead of the main molten pool [Yost,
Michael, and Eisenmann 1995].
Where capillary enhancement of spreading is
required, the surface texture should be as jagged
as possible. A surface prepared by grit blasting
or abrading with silicon carbide impregnated pa-
per is therefore preferable to a shot-peened sur-
face. The reason for this is as follows. Sharp
reentrant angles that exist on jagged surfaces co-
incide with sudden changes in the crystallo-
graphic orientation of the exposed parent metal.
The adhesion of native oxides at these microstruc-
tural discontinuities will tend to be relatively
weak and provide sites at which the oxide layer
can be more readily undermined or penetrated.
There is a limit to the roughness of surface that
can be used to promote spreading by a molten
ller. If the texturing is too deep, then capillary
dams can be formed and these will impede the
spreading of the ller metal [Funk and Udin
1952].Another factor that should be considered in
connection with texturing is the extent of alloying
between the ller and the parent material. For ex-
Fig. 1.17 Radiograph of a 50 mm (2 in.) diam component soldered using two 50 m (2000 in.) thick foils of Ag-96Sn solder, in
high vacuum and incorporating a No. 400 gauze. (b) High-resolution radiograph that reveals the true nature of the joint
lling in (a), with a void present at the center of each aperture in the mesh. Magnication: 640
Chapter 1: Introduction / 23
ample, when lead-tin eutectic solder is used to join
gold-coated components, there is a limit of about
4% to the concentration of gold that can be accom-
modated before the solder is embrittled (see Chap-
ter 2, section 2.3.2). If the solder spreads over a
gold-coated surface, the critical thickness of the
gold coating will be reduced by a factor related to
the surface roughness, r, and which must be con-
sidered when calculating the total volume of gold
to be applied corresponding to a given plan area
of spread. Table 1.4 indicates values of surface
roughness that can be obtained by abrasion of cop-
per by various means.
Attempts have been made to improve joint
lling by introducing capillarity enhancers to the
joint gap. Such enhancers include nely divided
powders and ne meshes that are wetted by the
ller but that are effectively inert. This type of
approach has been explored by the authors and
has not been found to radically improve joint
lling. A volume fraction of powder that was
calculated to signicantly increase capillary
forces had an adverse effect on the uidity of the
ller metal. On the other hand, meshes provided
stable traps for air and evolved gases in the joint.
This led to the formation of an array of voids
corresponding to each aperture in the mesh, as
revealed by a high-resolution radiograph of a
joint with a No. 400 gauze that was soldered
using a foil of Ag-96Sn alloy (Fig. 1.17).
The importance of surface roughness is ex-
emplied by a recent study of the fracture tough-
24 / Principles of Soldering
ness of joints made to copper components using
Ag-96Sn solder. It was found that the only sig-
nicant variable was surface roughness. Joint
toughness was found to be largely independent
of joint thickness and soldering time for limited
ranges of these variables (60200 s and 150400
m, or 616 mils, respectively). The response to
variation in surface roughness is reproduced in
Fig. 1.18. The low fracture toughness at inter-
mediate roughness occurs when adjacent sites of
intermetallic compound growth meet at unfa-
vorable crystallographic orientations [Stroms-
wold, Pratt, and Quesnel 1994].
1.2.6 Dissolution of
Parent Materials and
Intermetallic Growth
It is frequently observed that a ller metal will
continue to spread beyond an initially wetted
surface area over an extended period of time
(>10 s), which would not be expected from clas-
sical uid-ow theory. Clearly, classical expres-
sions for uid ow, exemplied by Eq 1.4, do not
strictly apply in such cases. Indeed, this type of
ow can usually be associated with solid-liquid
interfacial reactions, which are neglected in the
model described in Milners paper [1958]. Where
joint lling is sluggish because of reactions oc-
curring between the ller and the solid surface,
increasing the temperature to reduce the viscos-
ity of the molten ller is unlikely to enhance
Fig. 1.18 Effect of the surface roughness of copper substrates
on the fracture toughness of joints made with silver-
tin eutectic solder. It is worth noting that the joints under which
the test joints were made are relatively extreme in terms of joint
thickness (i.e., quantity of solder present) and soldering time, so
that the thickness of the copper-tin intermetallic layers formed is
likely to be substantially thicker than encountered in normal sol-
dering practice.
lling, because the reactions that are occurring
transverse to the ow directions will accelerate
[Tunca, Delamore, and Smith 1990].
Dissolution of the substrate and resulting
growth of intermetallic compounds both follow
Arrhenius-type rate relationships, represented by:
Rate  exp
Q
kT
where Q is an activation energy that character-
izes the reaction taking place at temperature T (in
degrees Kelvin) and k is the Boltzmann constant.
The alternative of increasing the joint gap is not
usually an option because this is likely to lead to
a reduction in joint lling and/or joint strength,
as discussed in Chapter 4, section 4.3. The so-
lution then is to change the materials system;
several means by which this can be achieved
without changing the parent materials are de-
scribed in Chapter 4, section 4.1.
Interfacial reactions are important, not only in
determining the ow characteristics of the ller
and its wetting behavior, but also the properties
of the resulting joints. When a molten ller wets
the parent materials, there is normally some inter-
solubility between them. It is usually manifested
as dissolution of the surfaces of the parent ma-
terials in the joint region and the formation of
new phases at either interface between the parent
materials and the molten ller or within the ller
itself when it solidies. The effects of dissolution
of the parent materials and compound formation
on joints are discussed in detail in Chapter 2,
section 2.3.
The rate of dissolution of a solid metal in a
molten metal is described by Weeks and Gurin-
sky [1958, p 106161] and Tunca, Delamore,
and Smith [1990]:
dC K A (Cs  C )
 (Eq 1.5)
dt V
where C is the instantaneous concentration of the
dissolved metal in the melt, Cs is the concen-
tration limit of the dissolved metal in the melt at
the temperature of interest, t is the time, K is the
dissolution rate constant, A is the wetted surface
area, and V is the volume of the melt. This equa-
tion is known as both the Nernst-Shchukarev and

the Berthoud equation. In the integral form, Eq
1.5 becomes:
KAt
C  Cs 1  exp (Eq 1.6)
V
assuming initial conditions of C  0, t  0.
Equation 1.6 reects the fact that, in general,
the concentration of dissolved metal in the mol-
ten ller increases in an inverse exponential man-
ner with respect to time. That is, the dissolution
rate is initially very fast but then slows as the
concentration of the dissolved parent material
tends toward its saturation limit (i.e., equilib-
rium), as shown in Fig. 1.19. Substituting mea-
sured values into Eq 1.6 shows that, for a sol-
dered joint of typical geometry, the equilibrium
condition is reached within seconds at the pro-
cess temperature. Thus, it is possible to use an
equilibrium phase diagram to predict the change
in the composition of the ller metal that will
occur in typical joining operations and the as-
sociated depth of erosion of the joint surfaces.
Equilibrium phase diagrams and their use in sol-
dering and brazing are considered more fully in
Chapter 2, section 2.3.
In some materials systems, the product of re-
action between molten ller metal and the parent
materials is a continuous layer of an intermetallic
compound over the joint interface. Once formed,
Fig. 1.19 The concentration of a solid metal in a liquid metal
wetted by it changes in an inverse exponential
manner with respect to time and is limited by the saturation
concentration of the solid constituent in the liquid at that tem-
perature.
Chapter 1: Introduction / 25
the rate of erosion greatly decreases because it is
then governed by the rate at which atoms of the
parent material can diffuse through the solid in-
termetallic compound. As a rough guide, solid-
state diffusion processes are two orders of mag-
nitude slower than solid-liquid reactions, and thus
continued dissolution of the parent materials ef-
fectively ceases, within the timescales of typical
joining processes. Intermetallic growth will,
however, continue throughout the life of the prod-
uct, the practical implications of which are dis-
cussed in Chapter 4, section 4.1.4.
1.2.7 Signicance of the Joint Gap
The joint gap at the process temperature in-
uences both the joint lling and the mechanical
properties of the resulting joint. The relationship
between joint dimensions and mechanical prop-
erties is discussed in Chapter 4, section 4.3 and
in the planned companion volume Principles of
Brazing. In summary, the thinner a joint is, the
greater its load-bearing capability tends to be,
until a limiting condition is reached.
Contact angle, surface tension, and viscosity
all reduce with increasing temperature, making
good joint lling in narrow joints more readily
achievable as the joining temperature is raised
above the melting point of the ller metal. A
lower practical limit to the joint gap is imposed
by three factors:
The need to provide a path for vapors to
escape. Flux vapors evolved within the joint
and pockets of air must be allowed to escape,
if the formation of voids through gas entrap-
ment is to be prevented (see Chapter 4, sec-
tion 4.3.1.1). At the same time, any reducing
agent needs to gain access to all joint surfaces
and be present in sufficient concentration to
work effectively.
Reaction with the components. The metal-
lurgical reaction that occurs between a mol-
ten ller metal and the surfaces of the com-
ponents can take one of two forms.
a. The surface region of the work piece has
limited solubility in the molten ller. This
is the preferred situation. The dissolution
of metal from the surface of the compo-
nents can result in either compound for-
mation at the interface, which may prevent
26 / Principles of Soldering

further dissolution, or alloying with the An upper practical limit to the joint gap is
ller, which will change its composition determined by:
and hence its melting point.
On the whole, solders tend to form in-
Mechanical properties of the joint. As the
gap is increased, the mechanical properties of
terfacial compounds with parent materials,
the joint declines progressively to those of
while brazes usually exhibit more exten-
the bulk ller metal, which in the case of
sive alloying between the materials. This
solders are particularly poor in relation to
can be partly explained by the fact that
most structural materials. This aspect is dis-
most solder alloys are based on elements
cussed further in Chapter 4, section 4.3.3.
with crystal structures that differ from those
of most common parent metals. Conse-
Joint lling requirements. As noted in sec-
tion 1.2.2 in this chapter, the capillary force
quently, intermetallic compounds tend to
decreases as the joint gap increases, and this
form in preference to solid solutions.
will place a practical upper limit on the joint
A reaction that depresses the melting
gap. At the same time, a sufficient quantity of
point of the ller metal is desirable for
ller must be supplied to the joint to entirely
narrow joints, because its uidity will be
ll it. Hydrostatic forces will promote the
enhanced by such a reaction at a constant
ow of low-viscosity ller metals out of wide
temperature. A reaction that raises the melt-
gap joints.
ing point of the ller metal will tend to
increase its viscosity and can cause the The optimal balance of these factors is
ller to solidify at the process temperature achieved when the joint gap is about 10 to 100
before it has lled the entire joint. Wider m (400 to 4000 in.), depending on the type of
joints mitigate this effect because the al- reaction that occurs between the ller and the
loying will tend to be diluted. component. This is substantiated by theoretical
b. Dissolution of the ller in the parent metal.
In this situation, the volume of ller will
Table 1.5 Typical thermal expansivities of
shrink as the reaction progresses; therefore,
common engineering materials at normal
a larger volume of ller metal accommo- ambient temperature
dated in a wider joint gap is again preferred
Material Linear expansivity, 106/K
and for similar reasons. However, absorp-
tion of the ller is generally undesirable, be- Polymers
cause its constituents will tend to penetrate Polymers, rubbers 150300
into the parent materials preferentially Polymers, semicrystalline 100200
along grain boundaries, generally to the det- Polymers, amorphous 50100
riment of the mechanical properties of the Metals
assembly and sometimes resulting in em- Zinc alloys 2530
brittlement and/or hot shortness. Aluminum alloys 2023
Copper alloys 1619
Stainless steels 1517
Control of the joint gap. The width of the Iron alloys
Nickel alloys
1315
1215
joint gap must be predictable and stable Cast irons 1013
during the bonding cycle. The size of the Titanium alloys 810
Tungsten/molybdenum alloys 47
gap will be inuenced by the coefficients of Low-expansion alloys (Fe-Ni-base) 15
thermal expansion of the respective com- Graphite 79
ponents, and allowances need to be made Ceramics
for different expansivities of the mating Ceramics, glass 610
components. The expansivities of a repre- Ceramics, oxide 48
Ceramics, porcelain/clay 37
sentative range of engineering materials at Ceramics, nitride/carbide 26
room temperature (25 C, or 77 F) are Diamond/silica/carbon ber 1 to 1
listed in Table 1.5. Temperature gradients
The values given are representative of the most widely used materials, rather than
along the joint must be considered from the provide absolute limits for the different classes listed. The thermal expansivity
same viewpoint. Variations of joint gap will depend not only on elemental composition but also on microstructure and
temper. Composite materials can have expansivities that effectively range be-
should be avoided wherever possible, as tween those of the constituents and depend on the relative proportions of the
this can have a serious effect in impeding matrix and reinforcement phases. To convert to customary units of 106/F,
multiply given values by 0.55556.
ow of the ller by capillary action.

calculations of capillary force and viscous drag
of liquid ow; see Fig. 1.20.
Generally, when components rest freely on one
another and the assembly is heated until the ller
is molten, the joint will tend to self-regulate to
widths around 50 m (2 mils). Indeed, it has been
demonstrated that for a xed combination of ller
metal, component materials, and process condi-
tions the joint gap will tend to a xed value specic
to the combination. This value must be deter-
mined by experiment. If there is insufficient ller
metal to ll this gap then the joint will contain
voids, or if too much ller is applied the excess
will spill out [Bakulin, Shorshorov, and Shapiro
1992]. Where thinner or wider joints are required,
it is necessary to insert spacers (such as wires) of
the desired width between the components and,
for thin joints, to apply pressure during the bond-
ing cycle to overcome the hydrostatic forces that
Chapter 1: Introduction / 27
strong is the interface between the parent mate-
rial and the ller metal in an ideal situation?
The cohesive strength of metals results from
attractive forces between the constituent atoms.
Normally, each atom will occupy a physical lo-
cation where the net force on it is zero. When the
solid metal is strained by the application of an
external load, the atoms move from their equi-
librium positions, and an opposing stress is set up
in the metallic crystal. The attractive force be-
tween atoms that share the same electron cloud
increases with the distance between them up to
a maximum and thereafter decreases abruptly,
when failure occurs. A perfect metal lattice will
fail at this point by cleavage across the crystal-
lographic plane because this is the region where
the interatomic forces are weakest.
To a rst approximation, the interatomic force
per unit area varies with interatomic separation,
will act to levitate the upper component. x, according to a sine wave with wavelength ,
as shown in Fig. 1.21. The interatomic force per
unit area may then be represented by a sine wave:
1.2.8 The Strength of Metals
The purpose of making soldered joints is usu- x
ally to form a metallic bond between compo-

o sin 2 (Eq 1.7a)

nents. A fundamental question, therefore, is how

where o is the maximum theoretical strength.

The work done per unit area in completely sepa-
rated neighboring planes of atoms, which are an
equilibrium distance apart (i.e., a  /4), is then:

/2 /2
 o
 dx   o sin 2
x

dx 

0 0

Fig. 1.20 Calculated time for molten tin and copper to ow

up a perfectly wetted capillary [Nicholas 1989]
This work corresponds to the total surface energy
of the two new surfaces created in the fracture,
that is, 2SV, where SV is the surface energy per
unit area of the solid.
Accordingly:

2SV
o  (Eq 1.7b)


Within the elastic range of strain, Hookes law

applies, that is:
x
Fig. 1.21 Variation of interatomic force, per unit area, with   E
distance a
28 / Principles of Soldering

Differentiating Eq 1.7(a) gives: its theoretical strength. Only in special materials

such as carbon ber are the two values remotely

comparable.
d 2 x In ductile metals, application of stress results
 o cos 2
dx   in the movement of dislocations and other de-
fects through the lattice of individual grains. The
interfaces between grains are another region
At zero strain, that is, x  0: where physical material transport and plastic ow
takes place. Failure occurs when the rate of in-

crease in strength of the material due to work
d 2
 o hardening falls below the rate of decrease in the
dx x0 a load-bearing cross section resulting from the
plastic ow.
The preceding discussion pertained to bulk
Hence:
materials, that is, the components and the ller
metal, when considered in isolation. In reality,
E the joint interfaces will often be a source of voids,
o  (Eq 1.8) microcracks, local interfacial mismatch stresses,
2a
and brittle intermetallic layers. These features
tend to be a common source of joint weakness,
From Eq 1.7(b) and 1.8: and they should be minimized through judicious
choice of ller/parent material and joining con-
2SV ditions.
2a
  o
o E

1.3 The Design and Application of

so that:
Soldering Processes

o  ESV
a
1/2 A soldered joint is usually required to satisfy
a specic set of requirements. Most frequently,
it must achieve a certain mechanical strength,
which it must retain to the highest service tem-
perature in the intended application. The joint
The theoretical fracture stress is about o /10
must also endure a particular service environ-
for metals, although in practice strengths of met-
ment, which may be corrosive, and it may have
als tend to be only one-tenth of this value (i.e.,
to provide good electrical and thermal conduc-
o /100), owing to the presence of lattice defects
tance. In addition, the joint must be capable of
and other discontinuities.
being formed in a cost-effective manner without
Possibly somewhat surprisingly, soldered
detriment to other parts of the assembly.
joints subject to simple mechanical stress will
The principal aspects that need to be ad-
often fail in a brittle manner. The reasons for this
dressed can be divided:
are elaborated in Chapter 4, section 4.3.3. In
brittle materials, failure takes place by the ex- The functional requirements of the applica-
tension of cracks that either preexist in the struc- tion and the means of satisfying these through
ture or nucleate at lattice imperfections. The appropriate structural design
stress to cause fracture can be deduced from Eq The achievement of the specied assembly
1.9 by replacing the denominator with c, where through successful processing
c is the crack length, thus:
Each of these stages is examined in this section.

b  (ESV  c)1/2 1.3.1 Functional Requirements and

Design Criteria
Since c is very much larger than a, the mechani- All soldered joints used in manufactured prod-
cal strength of a brittle material is low relative to ucts must remain solid in service and retain the

associated components in xed positions when
subjected to stress. These requirements are usu-
ally satised by suitable design of the geometry
and the metallurgy of the joint, but there are also
other aspects to consider. Not the least among
these is the fact that solders, in particular, are
often operating under conditions that are rela-
tively at least as severe as those encountered in
jet engines [Plumbridge 1995]. Factors that have
a bearing on the functional integrity of soldered
joints are discussed below.
1.3.1.1 Metallurgical Stability
For a joint to remain solid, in most cases the
melting point (solidus temperature) of the ller
metal needs to exceed the peak temperature that
the component is ever likely to experience. There
are exceptions to this rule, which are discussed
in Chapter 5, section 5.9. Because the strength of
all metals decreases rapidly as the melting point
is approached, the peak operating temperature
should not exceed about 70% of the melting point
of the ller, in degrees Kelvin, if the joint is
required to sustain a load.
The phases that form on solidication in a
soldered joint are frequently unstable at elevated
service temperatures. Instability of phases
present in the joint at the service temperature
may be undesirable, because it can affect its in-
tegrity. The effects of continued reactions be-
tween the ller and the components also must be
considered, as explained in Chapters 2 and 3.
Because most solders are softer than the mate-
rials that are commonly joined, the mechanical
properties of a joint are generally limited by those
of the ller metal. An exception arises when the
joints are very thin and constrained between par-
ent materials of high modulus, as described in
Chapter 4, section 4.3.3.
1.3.1.2 Mechanical Integrity
The durability of engineering and consumer
products often depends on joints maintaining
their mechanical integrity for the duration of their
expected service life. The mechanical integrity
of a soldered joint depends on a number of fac-
tors, including:
The mechanical properties of the bulk ller
metal (Chapter 5, sections 5.5 and 5.6)
The joint geometrynamely area, width, and
shape (Chapter 4, section 4.3)
Chapter 1: Introduction / 29
The mechanical properties of any new phases
formed in the joint by reaction between the
ller and the components, either during the
joining operation or subsequently in service
(i.e., there is an interdependence with the mi-
crostructure) (Chapter 4, section 4.1).
The number, size, shape, and distribution of
voids within the joint (Chapter 4, section 4.3)
The quality of llets formed between the ller
and the surface of the components at the edge
of the joint (i.e., their radius of curvature and
extent of continuity) (Chapter 4, section 4.2.4)
The mechanical properties of joints, taking into
account the inuence of joint geometry, are ex-
tensively reviewed elsewhere. The reader is re-
ferred to Schwartz [1987], Frear, Jones, and Kins-
man [1990], Brandon and Kaplan [1997],
Nicholas [1998], and Manko [2002].
1.3.1.3 Environmental Durability
Joints are normally expected to be robust in
relation to the service environment. This most
commonly involves exposure to corrosive gases,
including sulfur dioxide and other constituents
of a polluted atmosphere; to moisture, perhaps
laden with salt; and to variable temperature. The
corrosion and stress-corrosion characteristics of
the joint are then relevant. Corrosion mecha-
nisms are generally very complex and specic to
a given combination of materials, chemical en-
vironment, and joint geometry. Therefore, each
situation should be determined empirically.
The temperature of a joint can be shifted well
beyond a normal ambient range, especially in
aerospace applications and in situations where
heat is generated within the assembly itself. Then,
thermal fatigue and other changes to the metal-
lurgical condition of the joint, such as the growth
of phases, can occur, and these invariably affect
the properties of the joint. In other words, there
is interdependence between environmental sta-
bility and microstructural stability. An appropri-
ate choice of the materials combination used
should enable these changes to be constrained
within predictable and acceptable limits. In this
regard, solders tend to be used at service tem-
peratures that are proportionately very close to
their melting points, with respect to the thermo-
dynamic reference point of absolute zero tem-
perature (273 C, or 459 F). Hence, they are
metallurgically unstable and microstructural
changes take place readily.
30 / Principles of Soldering
1.3.1.4 Electrical and
Thermal Conductivity
In certain applications, soldered joints are re-
quired to provide electrical and/or thermal con-
tact between components. Generally, thin, well-
lled soldered joints amply satisfy this
requirement. Only in a few extreme situations
are the thermal and electrical properties of such
joints close to the allowed limits. A case in point
is high-power silicon device assemblies, where
the joint between the silicon device and the metal
backing plate is required to conduct away >1
W/mm2 of thermal power. Here, it is crucial to
ensure that the joints are kept thin (<30 m, or
1200 in.) and essentially void-free (<5% by
volume) [Humpston et al. 1992].
1.3.2 Processing Aspects
An important aspect that must be considered
when designing a joint is the practicality of the
process involved. Among the relevant issues are:
Jigging of the components
The form of the ller metal
Heating method
Temperature measurement
Joining atmosphere
Coatings applied to surfaces of components
(as necessary)
Cleaning treatments
Heat treatments prior to joining
Heating cycle of the joining operation
Postjoining treatments
1.3.2.1 Jigging of the Components
The components being joined normally must
be held in the required conguration until the
ller metal has solidied. Even if the compo-
nents can be preplaced without a xture, the
use of some form of jig is still frequently ben-
ecial to ensure that the components are not
disturbed by capillary forces originating from
the molten ller metal. On the other hand, it is
also possible to exploit the capillary forces to
make the assembly self-align during the join-
ing process; this is widely practiced in the fab-
rication of electronic circuits using lightweight
surface-mounted components, for example,
style 0201 and ip-chip assembly (see Chapter
5, section 5.2).
The jig can be used to fulll more than a hold-
ing function. For example, it can serve as a heat
spreader, as a heat sink, or as a heat source. A jig
should be constructed from a nonporous material
to prevent contamination of the atmosphere sur-
rounding the workpiece. Moreover, the jig ma-
terial should not be wettable by the molten ller
metal, in case they come into contact through
accidental spillage. Materials such as brass
should be avoided as zinc readily volatilizes.
Care should be taken in designing jigs so as not
to stress the constrained components through
thermal expansion mismatch.
A note of caution needs to be sounded when
joining to stressed components, as this can lead to
brittle failure of the components through a mecha-
nism known as liquid metal embrittlement. The
degradation in the mechanical properties of an en-
gineering steel, when stressed in tension and wet-
ted by Pb-4Sn solder, has been studied by Wat-
kins, Johnson, and Breyer [1975]. In practice, this
failure mechanism is seldom encountered.
Graphite is often favored as a jig material. It
is inexpensive, easy to machine, a good thermal
conductor (making it an effective heat spreader),
is an absorber of radiant heat, and is not wetted
by the majority of molten ller metals. Graphite
also has the merit of mopping-up oxygen in an
oxidizing atmosphere to form CO and CO2. How-
ever, if the oxygen partial pressure is already
low, then its role as a reducing agent is negli-
gible. Care should be taken to ensure that the
graphite used for jigs is of a dense grade so that
it will be mechanically robust and have a low
porosity to minimize outgassing. The desorption
of water vapor, in particular, frequently deter-
mines the quality of a gas atmosphere in the
vicinity of the workpiece.
Jigs are sometimes used to apply a controlled
pressure to a joint in an assembly. One compo-
nent can then be deliberately and elastically dis-
torted to bring it into close and uniform contact
with its mating part. This is an advantage when
very narrow joints are required and when solid-
state diffusion constitutes an important part of
the joining process. Compressive loading on the
joint also aids expulsion of air and vapor from
the joints, which are otherwise trapped in pock-
ets and produce voids. Another application is
outlined in Fig. 1.22. Because this is a uxless
soldering process, the applied load serves a dual
function. First, it helps puncture the oxide lms
on the surfaces of the ller. Second, when the-
ller metal melts, the applied force acts against
any dewetting capillary force of the liquid to
ensure ow. The combination of these two fac-
tors leads to improved joint lling. Additional
 Chapter 1: Introduction / 31

considerations about uxless joining processes
can be found in Chapter 3, section 3.3.
1.3.2.2 Form of the Filler Metal
Filler metals are available in many different
forms. These include congurations that nor-
mally can be produced from an ingot by me-
chanical workingfor example, wire, rings, and
foil. Such geometries are not restricted to ductile
alloys. If the constituents are individually duc-
tile, then the preform can be partitioned. This
method is discussed further in Chapter 4, section
4.1.5. The development of rapid-solidication
processes has led to the availability of foils and
wire of joining alloys that are inherently brittle.
These foils are produced directly from the melt:
the process involves forcing molten metal
through a hole or slot onto a rapidly spinning,
water-cooled, metal wheel. Figure 1.23 shows
such a strip casting process in operation, and Fig.
1.24 gives some typical foils produced by this
route. The high rate of heat extraction that occurs
in this process causes the molten metal to so-
lidify almost immediately on striking the wheel,
resulting in the formation of a strip of the alloy
with a ne crystalline, or occasionally amor-
phous, microstructure. The dimensions of the cast
material can be controlled by varying the nozzle
dimensions, the ejection pressure, the speed of
rotation of the wheel, and other parameters of the
casting process [Jones 1982]. The rened mi-
crostructure of the rapidly solidied alloys gen-
erally improves strength and ductility compared
with the same alloys produced by conventional
casting and mechanical working (see Fig. 1.25).
A good example is the Bi-43Sn eutectic solder,
which is normally brittle. However, when pre-
pared as a foil by rapid solidication, the duc-
tility of this alloy is comparable to that of other
solders.
Solders are also available as nely divided
powders that can be mixed with a binder to form
a paste capable of being screen printed onto a
substrate or applied to the workpiece, via a dis-
penser, in an automated production line. How-
ever, powders and pastes containing powders
have an extremely high ratio of surface area to
volume of ller, which generally produces high
oxide fractions and in the absence of suitable
precautions would be detrimental to the quality
of the resulting joints. Paste manufacturers go to
great lengths to produce solder spheres with pol-
ished surfaces and thin oxide skins specically
for this very reason.
Some of the more common ller metals are
available as wire or rod that incorporates a ux.
Most readers are probably familiar with ux-
cored solder wire. In this form, the ller metals
can be readily used in air without any additional
precautions.
In more specialized joining processes, the sol-
der can be deposited as a coating on the com-
ponents by screen printing, electroplating, and
by vapor deposition techniques such as evapo-
ration. Where it is not possible to deposit the
actual alloy, sequential layers of the constituent
elements can be applied. The former is generally
preferred because the melting point of an alloy

Fig. 1.22 (a) Sensor comprising three piezoelectric ceramic elements. These are metallized and soldered onto a metallized substrate.
The component is then encapsulated in a polymer to provide protection against the environment. (b) To ensure the
designed degree of acoustic coupling between the piezoelectric elements and the substrate, the soldered joints must be of a specied
thickness. This was achieved using tungsten spacer wires in each joint and a spring-loaded jig to apply a compressive stress to each
element during the process cycle. Also shown is the mask used to apply the metallization pattern to the substrate.
32 / Principles of Soldering
is well dened, whereas there is no guarantee
that melting will take place at the desired tem-
perature in the case of the composite layers, un-
less signicant solid-state diffusion has occurred
rst to form the appropriate low-melting-point
phases.
The use of some form of preplaced ller metal
has a number of advantages. Most particularly,
because the thickness and area of ller metal are
predetermined, the volume of molten ller may
be carefully controlled. Also, the number of free
surfaces is reduced from four (corresponding to
a foil preform sandwiched in the joint) to just
two, thereby considerably reducing the propor-
tion of oxides and other impurities deriving from
exposed surfaces.
Tin-lead solder is the only solder that can be
readily applied to faying surfaces as an alloy
directly by electroplating. More recently, and
Fig. 1.23 Production of foil directly from a molten charge by
strip casting. Source: Vacuumschelze GmbH, Ger-
many
Fig. 1.24 Examples of foil strip produced by rapid-solidi-
cation casting technology. Source: Fleetwood et
al. [1988]
aided by the development of advanced pulse-
plating methods, coelectroplating of gold-tin sol-
der has been developed. Prior to this, the solder
had to be realized as a deposit of gold, overlaid
with tin (the respective electropotential of these
elements makes it difficult to reverse this order).
This order is undesirable because the tin is able
to oxidize during storage and on heating to the
process temperature. Also, it is difficult to realize
thick solder deposits using this strategy because
of the need to heat above 420 C (788 F) to
destabilize the layer of AuSn intermetallic that
will otherwise form as a barrier layer at the in-
terface between the two metals. The new method
uses pulse plating to extract a controlled ratio of
gold to tin atoms from a cyanide-free, weakly
acidic solution and offers the prospect of highly
reproducible deposits over large areas to any de-
sired thickness [Sun and Ivey 2001].
Some ingenuity has been applied to other, less-
direct methods of selective application of ap-
plying ller metals. One of these exploits a
chemical reaction to selectively apply tin-lead
solder to ne-pitch pads on printed circuit boards
without the normal complexity of precision sten-
cil printing. The so-called Super Solder system
uses tin powder combined with a lead salt of an
organic acid to make a paste. On heating, the
organic acid decomposes, thereby providing a
partial uxing action, and the lead so liberated
combines with the tin to form the solder alloy in
situ. The paste is applied over the entire surface
of the circuit board. On heating above 183 C
(360 F), any tin-lead formed where there is an
exposed copper land will wet to it. The excess
paste can be simply washed off as there are no
Fig. 1.25 Dendrite arm spacing decreases with increasing
cooling rate and hence ne-grained microstruc-
tures have improved mechanical properties. The data pertain to
hypereutectic cast iron. Adapted from Seah, Hmanth, and Sharma
[1998]

ux residues to promote adhesion. By introduc-
ing the lead content of the solder in essentially
liquid form, the diameter of the tin particles can
also be decreased. By this means, it is claimed
that this process is suitable for component pitches
as ne as 80 m (3 mils). The solder deposit is
typically 100 m (4 mils) thick [Fuse, Obara, and
Irie 1992].
In the electronics industry there exists a very
different method of applying the solder to the
workpiece. It is wave soldering, which is a pro-
cess used to manufacture many millions of
printed circuit boards (PCBs) each year. It is
fundamentally different in that molten solder is
applied to the faying surfaces. There are many
variations of the process, but, in essence, molten
solder is pumped continuously over a weir so
that there exists a stationary wave whose height
is precisely controlled. The circuit board, already
populated with electronic components, is uxed,
preheated, and then passed through the very top
of the solder wave. The transit time is typically
1 to 2 s, making it a very rapid assembly method.
Because the entire area (volume) of the joint is
immersed in molten solder during transit through
the wave, this removes from the process the need
to achieve any spreading, so the solder need only
wet the PCB lands and component terminations.
1.3.2.3 Heating Methods
Heat must be supplied to the joint to raise the
temperature of the ller metal and joint surfaces
above the melting point of the ller. The joint sur-
faces need to be heated; otherwise the ller metal
will be incapable of wetting them and therefore
will ball up. To prevent this situation, it is good
practice always to heat the ller metal via the com-
ponents to be joined and never vice versa.
The available methods of heating are: local
heating, in which only that part of the compo-
nents in the immediate vicinity of the joint is
heated to the desired temperature, and diffuse
heating, where the temperature of the entire as-
sembly is raised.
Common local heat sources include soldering
irons, gas torches, and resistance heating using
the assembly as the resistive element. More so-
phisticated heating techniques, such as induction
heating and laser heating, also fall within this
category. Although some methods of local heat-
ing are applicable to joining in a controlled at-
mosphere, this is not usually the case with a
soldering iron or torch, and a ux must then be
used.
Chapter 1: Introduction / 33
In local heating, the rate of heat energy input
must be high to overcome the heat conducted
away by the components and jigging. A high rate
of heat input can achieve the desirable charac-
teristic of fast heating and cooling of the joint.
Fast heating coupled with short heating cycles
minimizes erosion of substrate surfaces and
therefore restricts the formation of undesirable
phases, while rapid cooling ensures a ne grain
size to the solidied ller and thereby superior
mechanical properties. However, these potential
benets can be offset by the thermal distortion
that might be produced in the components being
joined. Local heating can be used to create speci-
ed temperature gradients that will restrict the
ow of the molten ller metal to the immediate
vicinity of the joint.
Diffuse heating sources include furnaces (both
resistance and optical), hot plates, and induction
coils. The features of diffuse heating methods are
the opposite of those of the local heating meth-
ods. For example, the total energy requirement is
higher, as the temperature of the entire assembly
has to be raised, which also signicantly in-
creases the process cycle time. On the other hand,
there is less risk of thermal distortion and accu-
rate control of temperature is easier to achieve.
Diffuse heating methods tend to impose fewer
constraints on the atmosphere surrounding the
workpiece, because the source of heat is rela-
tively remote from the components. For ex-
ample, a torch is not generally compatible with
a special atmosphere.
If diffuse heating is to be used in the fabrication
of complex assemblies, the designer must ensure
that all of the component parts are able to with-
stand the peak process temperature. With local
heating, heat sinks can be used to protect sensitive
areas from excessive thermal excursions. A re-
lated consideration when using diffuse heating in
a situation where several joints must be made is
that the melting point of the ller metal used for
the preceding joining operations must be higher
than the peak process temperature used in the suc-
ceeding cycle. Several different ller metals will
therefore be required to fabricate a multijointed
product in a step-joining process.
A diffuse heating method often used in the
electronics industry is reow soldering. A printed
circuit board is prepared by populating with com-
ponents and at each joint location a controlled
volume of solder and ux is applied, often in the
form of a paste. Heating is carried out in the
saturated vapor of a very precisely formulated
organic uid that has a condensation temperature
34 / Principles of Soldering
some tens of degrees above the liquidus tem-
perature of the solder. Heaters are used to va-
porize the uid, and the circuit board is placed
in the vapor stream produced. When this inter-
sects with this cold body of the printed circuit
board, the vapor condenses and liberates its heat
of vaporization, thereby heating the board and
components. The process is particularly effective
when the thermal mass of adjacent parts differs
greatly as the coldest components condense more
vapor and therefore receive the greatest heat in-
put. Temperature gradients are therefore auto-
matically minimized. Through the correct choice
of uid, it is also possible to design the working
uid so that ux residues are removed and the
nished assembly emerges from the process
chamber dry and clean. The principal drawbacks
are the cost of the working uids and their at-
tendant health and safety issues.
The oldest method of heating joints is by na-
ked ame. The gases predominantly used now
are acetylene and propane, burnt in oxygen.
These gases are inexpensive, widely available,
easy to use, and can be made oxidizing, reducing,
or neutral by adjustment of the oxygen-to-gas
ratio. These three combustion conditions are also
readily discernible by eye, allowing a skilled
operator to adjust the torch to satisfy the require-
ments of the job at hand. The thermal charac-
teristics of some common fuel gases burnt in
oxygen are given in Table 1.6.
A relatively new development in ame tech-
nology is the microame (Fig. 1.26). This is es-
sentially a conventional gas torch that uses a
hypodermic needle as the gas tip. By this means
the ame diameter can be reduced to submilli-
meter dimensions. This tool is particularly useful
for precision hand (or robotic) soldering, braz-
ing, and indeed welding of three-dimensional
components, such as jewelry items, musical in-
struments, and heavy electronic parts, ex-
amples of which include coils and connector pins.
The ame has an equivalent heat output of some-
where in the region of 500 to 2000 W (370 to
Table 1.6 Thermal characteristics of common
fuel gases burnt in oxygen. In each case, the
ame temperature is in the region of 3000 C
(5430 F)
Fuel gas Thermal output, kJ/cm3/s (kW/cm3)
Acetylene 15
Methane 7
Propane 6 Fig. 1.26
Hydrogen 9 a small, intense heat source. Source: Dipl. Ing.
1475 ft lbf/s), so it is a considerably more
intense heat source than a soldering iron. A fur-
ther convenience of these miniature gas torches
is their compatibility with a gas source derived
from the electrolysis of water, for example, using
an iron cell containing a concentrated sodium
hydroxide solution. Both the energy source (elec-
tricity) and fuel (deionized water) are readily
available; the gas supply is effectively instanta-
neous and does not require pressurized cylinders.
This type of equipment is available commer-
cially from a number of European manufacturers
and possibly several others, worldwide.
1.3.2.4 Temperature Measurement
The liquid-solid metallurgical reactions that
occur during soldering operations are highly tem-
perature dependent. Therefore, reliable measure-
ment of temperature is essential. Thermocouples
and pyroelectric elements are the most common
types of temperature sensor.
A number of precautions should be taken when
employing thermocouples. Regular calibration
checks should be made to determine if the ther-
moelectric characteristics of the thermocouple
materials have altered and to test for electrical
interference affecting the display system. Cor-
rect temperature measurement requires good
thermal contact between the thermocouple and
the object being monitored. This tends to present
a problem in vacuum joining processes where
thermal contact by mechanical meansnamely,
Microame torch to solder small objects requiring
Ernest Spirig

resting the thermocouple against a surface
tends to be inadequate. The thermal mass of the
thermocouple and its protective sheath impedes
the thermocouple junction from sensing the true
temperature of the component surface. These ef-
fects can be minimized by embedding the ther-
mocouple within the workpiece to improve ther-
mal transfer.
Even when thermocouples are used for tem-
perature measurement in gas atmospheres, where
the thermal coupling is better than it is in a
vacuum, a change in the measured temperature
will lag behind that actually occurring. This de-
lay, which can be of the order of seconds, is
difficult to measure accurately, but it must be
taken into account if a thermocouple is being
used to monitor the temperature of assemblies
exposed to high heating and cooling rates.
Pyrometers have one important advantage
over thermocouples: they are noncontacting sen-
sors of temperature. Measurements may be made
remotely from the workpiece, and the response
time of the instrument can be accurately deter-
mined. Traditional pyrometers are primarily de-
signed for operation above about 750 C (1380
F). However, recent advances in pyroelectric
technology have led to the commercial develop-
ment of thermal imaging bolometers that are
capable of measuring radiated energy down to
and even below ambient temperature with a fast
response time. Bolometers are now worth con-
sidering not only for high-temperature brazing,
but also for general application to soldering and
brazing processes. Apart from their exibility,
bolometers are capable of rapidly measuring
temperature over a large surface area. By con-
trast, a thermocouple provides only a highly
localized measurement of temperature at its tip;
this might not be representative of the entire
joint region.
1.3.2.5 Joining Atmosphere
For a molten ller metal to wet and bond to a
metal surface, the latter must be free from non-
metallic surface lms. Although it is possible to
ensure that this condition is met at the beginning
of the heating cycle, by prescribed cleaning treat-
ments, signicant oxidation will generally occur
if the components are heated in air. Steps must
therefore be taken to either prevent oxidation or
remove the oxide lm as fast as it forms.
The approach adopted will depend largely on
the atmosphere surrounding the workpiece. Sol-
dering processes are conducted in one of three
Chapter 1: Introduction / 35
types of atmosphere, dened according to the
reaction that occurs between the atmosphere and
the constituent materials:
Oxidizing (e.g., air, usually in the presence of
liquid ux)
Essentially inert (e.g., nitrogen, vacuum)
Reducing (e.g., carbon monoxide, halogen
containing)
The implications associated with using each of
these atmospheres are considered below.
Oxidizing Atmospheres. Air is the most
common oxidizing atmosphere. The principal ad-
vantages of joining in air are that no special
gas-handling measures are required and that there
are no difficulties associated with access to the
workpiece during the joining process. However,
because most component surfaces and those of
the ller metal are likely to form oxide scale
when heated in air, normally uxes must be ap-
plied to the joint region.
An active ux is capable of chemically and/or
physically removing an oxide lm. The ux may
be applied either as a separate agent or as an
integral constituent of the joining alloy. The sub-
ject of uxes is discussed in detail in Chapter 3,
section 3.2.
Gold and some of the platinum-group metals
do not oxidize when heated in air. Although sil-
ver will oxidize at air ambient temperature, the
oxide dissociates on heating to about 190 C
(375 F). These precious metals are therefore
sometimes applied as metallizations to the sur-
faces of the components being joined in uxless
processes. The use of wettable metallizations is
discussed in Chapter 4, section 4.1.2.1. Solders
that contain signicant proportions of the pre-
cious metals are generally less susceptible to oxi-
dation, enabling mild uxes to be used.
Inert Atmospheres. From a practical view
point, an atmosphere is either oxidizing or re-
ducing. This is because it is not possible to re-
move and then totally exclude oxygen from the
workpiece, except perhaps under rigorous labo-
ratory conditions. Thus, when dening an atmo-
sphere as inert it must be taken as meaning that
the residual level of oxygen present is not suf-
cient to adversely affect the joining process un-
der consideration. An atmosphere that might be
suitable for soldering to silver is likely to be
inadequate for nickel.
Because the inertness of an atmosphere is
judged relative to the specic application, it is
necessary to dene a quantitative measure of the
oxygen present. This parameter is the oxygen
36 / Principles of Soldering
partial pressure. Partial pressure provides a mea-
sure of the concentration of one gas in an at-
mosphere containing several gases. The partial
pressure of a gas in a mixture of gases is dened
as the pressure it would exert if it alone occupied
the available volume. Thus, dry air at atmo-
spheric pressure (0.1 MPa, or 14.5 psi) contains
approximately 20 vol% O2, so that the oxygen
partial pressure in air is 0.02 MPa (2.9 psi).
Typical inert atmospheres among the com-
mon gases include nitrogen, argon, and hydro-
gen. Hydrogen is included here because it is not
capable of reducing the oxides present on the
majority of metals at normal soldering and braz-
ing temperatures. The oxygen partial pressure in
standard commercial-grade bottled gases is of
the order of 10 mPa (1.5  106 psi). Higher-
quality grades are available, but their cost is usu-
ally too prohibitive to permit their use in most
industrial applications.
Vacuum is frequently used as a protective
environment for ller metal joining processes.
Vacuum offers several advantages compared with
a gas atmosphere, particularly the ability to mea-
sure and control the oxygen partial pressure more
readily. In a substantially leak-free system, the
oxygen partial pressure is one-fth of the vacuum
pressure, which is relatively easy to determine,
as compared with direct measurement of oxygen
partial pressure. Although a roughing vacuum of
100 mPa (15  106 psi) will provide an atmo-
sphere with the same oxygen partial pressure as
a standard inert gas, it is possible to improve on
this value, by several orders of magnitude, using
a high-vacuum pumping system. Alternatively, a
low oxygen partial pressure may be achieved by
obtaining a roughing vacuum, backlling with
an inert gas and then roughing out again. The
effect of the second pumping cycle will be to
reduce the oxygen partial pressure to less than
typically one-thousandth of that in the inert gas,
that is, approximately 10 Pa (1.5  109 psi).
This estimate assumes that the furnace chamber
is completely leak tight and does not outgas from
interior surfaces, nor does any oxygen or water
vapor backstream through the pump.
The disadvantages of using a vacuum system
for carrying out a joining process are, principally,
restricted access to the workpiece and the inad-
visability of using either uxes or ller metals
with volatile constituents, such as cadmium, as the
vapors can corrode the vacuum chamber, degrade
its seals, and contaminate the pumping oils.
A frequently overlooked consideration in re-
duced-pressure atmospheres is adsorbed water
that naturally exists on surfaces that are exposed
to ambient atmospheres. The continuous stream-
ing of water vapor that desorbs from surfaces and
ows past the workpiece as the pressure in a
vacuum chamber is reduced is a source of oxi-
dation; this is discussed in quantitative detail in
Chapter 3, section 3.1. In a vacuum system op-
erating at 10 mPa (1.5  106 psi), the desorbing
water vapor constitutes the major proportion of
the residual atmosphere. An adsorbed monolayer
of water vapor of just 100 mm2 (0.16 in.2) in area
desorbs to a gas pressure of 4 mPa (6  107 psi)
per liter of chamber volume. The surfaces of the
chamber should therefore be smooth to minimize
the surface area and also dry. In order to reduce
this problem further, the walls of the vacuum
chamber should be heated and the system should
be vented to a dry atmosphere. To effectively
desorb water vapor, the bakeout temperature
should be at least 250 C (480 F), which may
be difficult to achieve in practice owing to design
constraints and the employment of rubber and
other organic seals.
Another source of oxidizing contamination in
a vacuum system is oil vapor mixed with air and
water vapor, backstreaming from a rotary pump.
This can occur whenever the pressure inside the
vacuum chamber drops below 1 Pa (1.5  104
psi), but can be largely eliminated by employing
a foreline trap or by isolating the pump from the
chamber once the required pressure reduction
has been obtained.
The widespread practice of relying on an open
gas shroud to provide an inert atmosphere is
often unsatisfactory because it is extremely dif-
cult to control such an atmosphere reliably. For
example, turbulence in the inert gas shroud can
result in a supply of air actually being directed
at the workpiece. Recent advances in furnace
technology now permit open furnaces, often belt
furnaces intended for reow soldering of PCBs,
to achieve very high specication atmospheres
in the working zone through careful design of the
gas ow at the open portals.
A reducing atmosphere is one that is ca-
pable of chemically removing surface contami-
nation from metals. Gases that provide reducing
conditions are generally proprietary mixtures that
liberate halogen radicals. Specic gas-handling
systems are usually needed for these in order to
satisfy health and safety legislation.
For a few metals, hydrogen is satisfactory as
a reducing atmosphere. No less important for
meeting its functional requirement than the oxy-
gen partial pressure of the gas is its water con-

tent. Hydrogen is a relatively difficult gas to dry,
and the water vapor present can present a serious
problem. A frost point of 70 C (95 F) is cally. However, mechanical cleaning is generally
equivalent to a water content of 0.0002% by
volumethat is an oxygen partial pressure of
about 10 mPa (1.5  106 psi). There is also the
risk of explosion when dealing with hydrogen at
high temperatures, and hydrogen can embrittle
some materials. A more detailed treatment of
reducing atmospheres and their use is given in
Chapter 3, sections 3.1 and 3.3.
1.3.2.6 Coatings Applied to
Surfaces of Components
Only occasionally is the desired joining al-
loy (chosen on the basis of melting tempera-
ture and physical properties) metallurgically
compatible with the substrate in the sense that
the ller will wet the substrate uniformly, with-
out the consequential formation of embrittling
phases by reaction. A solution is to apply a sur-
face coating that will promote wetting by the
ller and react with it in a benign manner.
Coatings can be applied by a variety of tech-
niques and to thicknesses that suit the particu-
lar application.
On metals, coatings are usually applied by wet
plating methods, which are quick, economical,
and exible with regard to the coating thickness.
Electroplating cannot often be used directly for
most nonmetals, and it is more common instead
to rely on vapor-deposited coatings. If the sub-
strate is refractory in character, adhesion of metal
coatings tends to be poor unless the metal is itself
sufficiently refractory so that it will form a strong
reactive bond to the substrate. Widely used met-
allizations are chromium or titanium as the re-
active layer, overlaid by gold or a platinum group
metal to provide protection from the atmosphere.
These and other metallizations and the principles
on which they are designed are described in
Chapter 4, section 4.1.2.1.
1.3.2.7 Cleaning Treatments
The surfaces of the components to be joined
and the ller metal preforms must be free from
any nonmetallic lms, such as organic residues
and metal oxides, to enable the molten ller metal
to wet and alloy with the underlying metal. Fluxes
are often capable of removing surface oxides,
provided they are reasonably thin.
Organic lms can be removed with solvents,
which obviously should not react with the un-
Chapter 1: Introduction / 37
derlying materials. Thick oxides and other non-
metallic surface layers can be removed chemi-
preferable because chemicals tend to leave resi-
dues, which then also have to be removed. A
procedure that has proved effective with solder
wire is to wipe the latter with a brous tissue
soaked in solvent. The rst few wipes leave a
black mark on the tissue, which is a combination
of native oxides and manufacturing residues (in-
cluding lubricating oils). As shown by the data
presented in Fig. 1.27, the solder wire should be
wiped multiple times and, ideally, used within
1 h (see Chapter 3, section 3.3).
Mechanical abrasion exposes a fresh metal sur-
face. The roughness of the abraded surface can be
readily controlled, and this can be used to advan-
tage in promoting the spreading of the molten
ller metal. The rougher the surface, the better the
wetting and spreading of the molten ller tend to
be, for the reasons given in Chapter 1, section
1.3.5. However, rough surfaces create problems
when it is required to cover them with thin metal
coatings. For example, thickness uniformity of
thin metallic lms is difficult if not impossible to
achieve on a rough-textured substrate.
Soft components, such as solder foils, can be
difficult to mechanically clean using abrasive par-
ticles because these tend to get embedded. These
and thin vapor-deposited and electroplated met-
allizations can be protected against atmospheric
corrosion by the application of a noble metal over-
coat. If correctly stored and handled, such com-
ponents will not require cleaning prior to bonding
and, for obvious reasons, must never be abraded.
1.3.2.8 Heat Treatments Prior to Joining
Prejoining heat treatments are occasionally
useful in providing stress relief and thereby pre-
Fig. 1.27 Improvement in shear strength of a uxless joint
made with In-48Sn wire as a function of the num-
ber of wipes made with a paper tissue soaked in ethanol, im-
mediately prior to melting of the solder
38 / Principles of Soldering
venting unpredictable distortion during heating
of the components to the bonding temperature.
Other situations where prejoining heat treat-
ments can be benecial include those involving
components with nonmetallic surface lms that
are thermally unstable. In the case of silver, for
example, the oxide will readily dissociate when
heated above 190 C (375 F) in an ambient
atmosphere. Likewise, silver sulde dissociates
on heating above 842 C (1548 F).
Heating cycles may be used to produce solder
alloys from layers of the constituent elements
applied to surfaces by screen printing, electro-
plating, or vapor deposition. By heating the sub-
strate above the melting point of the constituent
with the lowest melting point, alloying will occur
by solid-liquid interaction. The joint can then be
formed in a subsequent heating cycle that is usu-
ally referred to as the reow stage.
1.3.2.9 Heating Cycle of the
Joining Operation
The prepared components and ller metal, pos-
sibly mounted in jigs, are joined by applying
heat. The heating cycle involves four important
processing parameters: the heating rate, the peak
bonding temperature, the holding time above the
melting point of the ller, and the cooling rate.
In general, it is desirable to use a fast heating
rate to limit reactions that can occur below the
prescribed bonding temperature. However, the
maximum heating rate is normally constrained
by adverse temperature gradients developing in
Fig. 1.28 Proles of typical temperature cycles. (a) Heating cycle with a controlled prole incorporating dwell stages to reduce
thermal gradients. (b) Heating cycle dened solely by attainment of a peak temperature
the assembly. These can produce distortions in
the components and give rise to nonuniform re-
actions between the ller and the two joint sur-
faces. Also, temperatures are difficult to measure
reliably during fast heating schedules. A better
practice, when joining in a vacuum or special
atmosphere furnace, is to heat the assembly rap-
idly to a preset temperature that is just below the
melting point of the ller metal and then hold at
this temperature for sufficient time (which can
range from a few seconds to over one hour, de-
pending on the size of the assembly and the heat-
ing method) to allow the assembly to thermally
equilibrate and for water vapor to ush out of the
joint. Following this dwell, the assembly may
then be rapidly heated to the bonding tempera-
ture. This method is used to make joints to PCBs
using lead-free solders where only a few degrees
of superheat are permitted. Proles of typical
temperature cycles are shown in Fig. 1.28.
The bonding temperature should be such that
the ller is guaranteed to melt, but at the same
time should not be so high that the ller degrades
through the loss of constituents or by reaction
with the furnace atmosphere. The optimal tem-
perature is normally determined by metallurgical
criteria, most importantly the nature and extent
of the ller-substrate interaction. The peak pro-
cess temperature is frequently set at about 50 and
100 C (90 and 180 F) above the melting point,
because accurate temperature measurement and
control is not always readily achievable, espe-
cially in reduced-pressure atmospheres, where
bulky jigging is used or where conduction be-

tween the heat source and the workpiece is poor.
Moreover, the reported melting temperatures of
some llers are not based on accurate measure-
ments, and it is prudent to make some allowance
for this uncertainty. The minimum time that the
assembly is held above the melting point must be
sufficient to ensure that the ller has melted over
the entire area of the joint and the maximum time
is usually a compromise based on practical and
metallurgical considerations. Extended dwell
times tend to result in excessive spreading by the
molten ller, possible oxidation gradually taking
place, and deterioration of the properties of the
parent materials.
The cooling stage of the cycle is seldom con-
trolled by the operator, but tends to be governed
by the thermal mass of the assembly and jig.
Forced cooling can lead to problems, such as
exacerbating mismatch stresses. Occasionally
one or more dwell stages are required, either to
provide stress relief to the bonded assembly, or
to induce some requisite microstructural change.
An example of the latter would be the aging
treatment of a precipitation-hardening alloy. In
this case, the solution treatment and quenching
stages are carried out in tandem with the actual
reow operation.
A heat treatment temperature of about 75% of
the freezing point of the ller metal in Kelvin
usually provides the optimal relief of residual
stresses. An example of an assembly that re-
quired a stress-relief treatment in order to avoid
catastrophic cracking is shown in Fig. 1.29.
1.3.2.10 Postjoining Treatments
Various types of postjoining treatments can be
applied. A cleaning schedule is generally used to
Fig. 1.29 Large-area silicon chip soldered into a metallized
ceramic package using an alloy based on the Au-
3Si composition. The expansion mismatch between silicon and
the ceramic makes it necessary to cool the assembly at a con-
trolled rate in order to prevent fracture of the components.
Chapter 1: Introduction / 39
remove ux residues and the tarnishing that are
produced when joining in air. Flux residues must
be removed, as they are usually corrosive, es-
pecially when moist and can affect the long-term
reliability of the component in service. Both
chemical and mechanical means of ux removal
are employed. Tarnishing tends to be removed
chemically, often with acids, followed by thor-
ough rinsing. If a heat treatment is not integrated
into the cooling stage of the bonding cycle, a
separate heat treatment may be carried out sub-
sequently.
1.3.2.11 Postjoining Cleaning
For many years it has been standard practice to
clean assemblies after soldering. This need arose
from the corrosive nature of traditional rosin-
based uxes and the subsequent reliability and
aesthetic degradation that can occur if the ux
residues are not removed. Particularly in the elec-
tronics industry, the preferred cleaning agent has
been a blend based on chlorouorocarbons
(CFCs). These liquids were used in vapor clean-
ing equipment of low complexity and therefore
low cost. In addition to being relatively efficient
solvents, CFCs had good compatibility with most
engineering materials including solders, no am-
mability risk, and low toxicity. The process had
the additional benet that the product emerged dry
from the cleaning operation. The ease of cleaning
in this manner was largely responsible for it to be
the norm to clean all parts and assembliesit was
less expensive and easier to clean than to risk sub-
sequent problems in service.
The discovery in the mid-1970s that CFCs
were contributing to stratospheric ozone deple-
tion led to elimination of CFCs from solder clean-
ing and ultimately to a ban on their use and
manufacture. The response by industry and par-
ticularly those engaged in electronics manufac-
ture was threefold.
First, alternative cleaning chemicals and meth-
ods were devised. Many of these are now com-
mercially available systems. As a result of this
substitution endeavor it is perhaps ironic that it
has been proven that cleaning with CFC-based
solvents is actually fairly ineffective, especially
with modern surface-mount and chip-on-board
technologies.
Second, new and improved ux chemistries
were formulated, including water soluble and
no-clean uxes (see Chapter 3, section 3.2).
Finally, many manufacturers undertook a thor-
ough reappraisal of the need for cleaning, par-
40 / Principles of Soldering
ticularly given the short life of some products
owing to technical obsolescence or the tran-
sience of fashion; examples include wrist-
watches, portable computer games consoles, and
mobile telephones.
Each of these options is considered further in
the paragraphs that follow.
Cleaning of Electronic Assemblies. The cur-
rent scene of cleaning methods is very analogous
to the situation regarding lead-free solders as a
replacement for lead-tin eutectic; namely, an al-
most universal cleaning route has now been re-
placed with a plethora of alternatives. This is
partly due to commercial pressures as the various
manufacturers compete for market share. The
available cleaning methods can be broadly
grouped into four categories:
Solvents
Semiaqueous formulations
Aqueous formulations
All of the above with ultrasonic or high-
pressure jet assistance
Comparative studies have been made of the rela-
tive effectiveness of each of these approaches
[Richards et al. 1993], from which the following
conclusions can be made:
There are many CFC-free cleaning options
that are technically and commercially viable.
The choice of which to use requires careful
consideration of the type of assembly for
which cleaning is required, the level of clean-
liness it is necessary to obtain, and the nature
of the residues it is permitted to leave.
Cleaning process times are signicantly
longer (1020 min) compared with that for
CFCs (3 min). Although the intrinsic clean-
ing step is not signicantly longer, the overall
process time is substantially increased owing
to the need to employ multitank processing
and a drying stage.
Mechanical assistance, most practicably in
the form of ultrasonic agitation or high-
pressure jet sprays, always results in a cleaner
product for a given cleaning time. This op-
tion is not always possible as some parts can
be degraded or damaged by aggressive wash-
ing [Richards, Burton, and Footner 1993].
The ability to clean ux from narrow gaps,
such as from underneath components, is
roughly related to h3, where h is the width of
the gap. Consequently, components such as
leadless ceramic chip carriers (LCCCs) with
stand-off heights of 50 m (0.002 in.) are much
more difficult to clean under than plastic
leaded chip carriers (PLCCs), which have
standoff heights around 100 m (0.005 in.).
Different substrates and components have dif-
ferent levels of inherent cleanliness. This is
due to the relative chemical affinity and de-
gree of mechanical adhesion between the
ux/ux residues and the substrate. Thus, for
example, ceramic circuit boards have only
half the contamination of ber-reinforced
laminate (FR4) boards as reowed, specied
in terms of g NaCl equivalence. After clean-
ing, the results are not greatly different.
Best cleaning results are obtained by opti-
mizing the thermal prole of the soldering
cycle for the cleaning system used. Some
cleaners are far more effective at removing
ux residues than unspent ux, and vice versa
for others. Usually, manufacturers sell a pack-
age comprising a ux and recommended
cleaning system and will be able to advise on
the most appropriate thermal cycle as well.
Comparative studies reveal that the solvent, ux,
and equipment manufacturers have a clear un-
derstanding of the capability of their products to
clean (see Fig. 1.30). There is negligible tech-
nical merit in deviating from their recipes.
Water-Soluble and No-Clean Fluxes. Wa-
ter-soluble uxes are designed such that residues
and unspent ux are miscible with water. Thus,
the assembly can be cleaned in water. To avoid
water stains on the product, it is usual to com-
plete the cleaning process with an organic rinse,
usually some form of alcohol.
Another family of uxes has been developed
that carries the designation no-clean. The chem-
istry of these materials has been carefully for-
mulated to ensure that unspent ux and ux resi-
dues are chemically bound or otherwise buffered
and remain relatively inert in the presence of
moisture. Modern versions of these uxes have
been further tailored so that the residue can pro-
vide additional functionality. A prime example is
provided by no-clean uxes developed for ip-
chip underll applications, where the ux resi-
due is designed to have a particular modulus and
expansion coefficient that helps to extend the
fatigue life of the solder joints.
As a general observation, the soldering pro-
cess window for water-soluble and no-clean
uxes tends to be somewhat narrower than for a
process utilizing a more conventional ux for-
mulation.

Alternatives to Cleaning. There are three
alternatives to cleaningsimply do not clean at
all, use uxless processes, or switch to conduc-
tive adhesives [Lea 1998a; Lea 1998b].
Electrically conductive adhesives have made
considerable technological progress and are
widely used to interconnect to LCD displays.
However, they have not yet reached the point of
maturity where they can be used reliably for in-
terconnects that carry more than a few micro-
amperes of current or high frequencies. There are
also issues concerning their stability in service:
deterioration of their adhesion and the slow evo-
lution of volatile species, which is of particular
concern when conductive adhesives are used to
interconnect bare die or are in the vicinity of
naked optical facets. No doubt, improved sta-
bility will be achieved as better-formulated prod-
ucts become available.
At face value, uxless processes are highly at-
tractive because there is no ux and therefore no
cleaning required. However, uxless processes
are extremely difficult to implement and are in-
compatible with low-cost volume manufacture.
To obtain satisfactory wetting and spreading, the
process atmosphere needs to be denuded of oxy-
gen and water vapor to levels that can only be ob-
tained in closed vessels. More details on uxless
soldering are given in Chapter 3, section 3.3.
Some care needs to taken to ensure that a ux-
less process does leave a clean product after sol-
dering. The authors encountered one process in
Fig. 1.30 The relative effectiveness of different cleaning agents on glass-reinforced epoxy laminate FR4 printed circuit boards (PCBs)
as measured by residual sodium chloride contamination. The reowed bar references as-made and uncleaned PCBs.
Adapted from Richards et al. [1993]
Chapter 1: Introduction / 41
which the solder perform was etched in 10% hy-
drochloric acid and then rinsed in deionized water
prior to use. The process was claimed to be ux-
less because no ux was used, and all the acid was
supposed to be rinsed off the solder. However, as
the solubility of hydrochloric acid in water de-
creases faster than does the dilution ratio, in prac-
tice the dried solder preform and nal product
were both proved to be severely contaminated
with chlorides. It transpired that better-quality
joints with lower levels of ionic contamination of
the product were obtained by using a commercial
no-clean ux without cleaning!
The ultimate alternative to cleaning is simply
not to do it. Cleaning adds to the capital and
consumable process cost and results in an as-
sembly yield loss. Provided the ux residues do
not adversely impact the reliability of the prod-
uct, then the merits of cleaning are truly ques-
tionable. This is particularly true for short-life
consumer products, for example, musical greet-
ing cards. Because the principal risk to leaving
ux residues on the product is corrosion arising
after the residues react with moisture, another
frequent approach is to coat the entire product in
lacquer. This provides a sufficient barrier to mois-
ture ingression over the recommended life of the
product. Obviously for life- or mission-critical
products, cleaning will probably always be un-
dertaken, but in many instances acting in rec-
ognition of a nite life is a technically and eco-
nomically sound approach.
42 / Principles of Soldering
1.3.2.12 Statistical Process Control
All processes are subject to variation, and
achieving stability of processes is an important
step in any quality-improvement program. The
application of statistics to monitor and control
the variability is termed statistical process con-
trol (SPC) [Ledolter and Burrill 1999]. Many
industrial soldering processes are subject to SPC.
A modern PCB assembly line can achieve joint
defect rates of a few ppm. This means, quite
literally, that only one or two soldered joints in
every million made would fail a quality inspec-
tion. This degree of manufacturing quality is only
achieved through SPC.
A fundamental tool in SPC is a graphical dis-
play, known as a control chart. This chart provides
the basis for deciding whether the variation in the
output of a process is due to common, randomly
occurring variations or to unusual causes, which
would require investigation and action. The con-
trol chart is a chronological plot of particular char-
acteristics, such as joint strength or peak reow
temperature, sampled at periodic intervals. This
information furnishes data on the process stability
and provides an understanding of improvements,
where made. Whenever a signicant deviation
from the norm is detected, a decision can be made
to adjust a process variable in order to bring the
output back to the required quality level. Obvi-
ously, to accomplish this there must be a proper
understanding of the relationship between the
process variables and the output.
There are different types of control charts, de-
signed for different situations, which are classi-
ed by the type of data they contain. Control
Fig. 1.31 Statistical process control chart for peak reow temperature (indicated by solid squares) measured at a test point on a
printed circuit board, showing the upper and lower control (i.e., intervention) points for this process
charts designed to monitor the proportion of de-
fective items are referred to as p-charts, while
charts that track the number of defects on the
product are known as c-charts. Both are used to
describe attribute data, that is, a record of the
presence and absence of certain characteristics.
Quantitative data are monitored using a mean or
x-bar chart, while process variability is measured
using range charts (r-charts) and standard devia-
tion charts (s-charts).
The basis of SPC is, for each parameter
tracked, to select upper and lower control limits
(see Fig. 1.31). These are set at a multiple num-
ber of standard deviations from the mean such
that there will be a high probability that the data
will fall between these limits when the process
is working as desired. Only when the process
metrics drift further from the mean is interven-
tion required.
It is possible to apply SPC to virtually any
process or machine output. In many instances, it
is a highly effective basis for controlling manu-
facturing processes. As with any tool it is nec-
essary to use some discretion and critical thought
to ensure that SPC is appropriate and the cost of
implementing and sustaining it is justied.
1.3.3 Health, Safety, and
Environmental Aspects of
Soldering
Soldering encompasses the use of a large num-
ber of different materials, covering metallic and
nonmetallic elements for the llers and the par-
ent materials, and organic and inorganic chemi-

cals used in uxes, controlled atmospheres, and
for removing ux residues. Several of these ma-
terials are hazardous in varying degrees to the
operators or to the environment [Sax and Lewis
1989]. Accordingly, they must be handled, used,
and disposed of according to national codes of
practices or regulations governing hazardous
substances. Official listings produced by na-
tional health and safety authorities classify ma-
terials according to their toxicity level, for ex-
ample, the exposure limits for hazardous
vapors and dusts.
The main problem with solders and soldering
uxes arises when they are heated to make a joint.
The fume contains a cocktail of gases that can
cause eye and nose irritation, dermatitis, asthma,
and respiratory problems. The fume contains ne
particles, in the range 0.1 to 1 m (4 to 40 in.),
which is the most dangerous size distribution for
causing long-term lung damage. The recom-
mended solution is to ensure that the workplace or
work chamber is ventilated using an appropriately
designed extraction system that is able to exhaust
the gases and trap the particulates [Jakeway
1994]. Preventing exposure to the hazard by ap-
propriate measures should always be given higher
priority than protective measures.
All materials that are likely to be encountered
in a joining context will have an assigned value
of maximum exposure limit, usually in weight
per unit volume (normally in mg/m3). The form
of the material is also relevant. Powders and
dusts are more hazardous than nonvolatile liq-
uids and monolithic solids: they are ranked ac-
cording to the maximum inhalable quantity in
mg/m3, time weighted over a period of time,
either short term, meaning minutes, or over a
longer period of many hours. For correct inter-
pretation of the rules, regulations, and audits,
reference should always be made to a qualied
safety practitioner, as there are often also legal
aspects to consider.
Care must also be taken in the storage of ma-
terials both prior to use and in the procedures for
the subsequent disposal of residues, exhaust
emissions, and other associated effluent, such as
solutions containing rinsed uxes. These are usu-
ally subject to statutory controls. For many or-
ganic chemicals and gases, in this context solder
uxes, binders used in pastes, and halogenated
gases, there may also be re risks to consider.
The ammability is rated according to ash-
point, which is the lowest temperature at which
the substance can be spontaneously ignited when
it is in a saturated condition.
Chapter 1: Introduction / 43
Appendix A1.1:
Solid-State Joining with
Gold, Indium, and
Solder Constituents
Gold has three desirable characteristics that
render it a most suitable metal for making dif-
fusion-bonded joints. These are its low modulus,
rapid self-diffusion, and absence of an oxide skin
when heated in air. As a result, diffusion bonding
of gold can be achieved at room temperature
with plastic deformation of as little as 20%. The
temperature/pressure curve for a process time of
1 h is given in Fig. 1.32 [Humpston and Baker
1999]. Extrapolation of the graph predicts that a
successful gold-gold bond can be achieved with-
out pressure above approximately 450 C (840
F). Certainly it is the authors experience that
gold rods merely placed in contact bond readily
at 500 C (932 F).
Indium can also be used to make pressure-
welded joints. Because indium tends to be cov-
ered by a relatively thick layer of oxide, the pro-
cess usually relies on extensive physical
deformation to ensure adequate virgin metal con-
tact. Heating is not normally necessary because
of the extensive deformation and, at room tem-
perature, the metal is already very close to its
melting point. A typical temperature/pressure
process curve for indium is given in Fig. 1.33
[Plotner et al. 1991].
Diffusion bonding and pressure welding can,
perhaps surprisingly, be achieved using standard
solders. The authors have witnessed one high-
volume manufacturing line where a hermetic seal
was made between two gold metallized surfaces
using a wire of ordinary (uxless) solder. Two
Fig. 1.32 Temperature/pressure curve for diffusion bonding
of gold, for a process time of 1 h. The line on the
graph differentiates between joints of acceptable (above) and
unsatisfactory (below) tensile joint strength after fabrication.
44 / Principles of Soldering

The droplet resolidies after spreading on the

substrate as a spherical cap of radius R and
height h, its interface with the substrate hav-
ing a diameter 2A, as shown in Fig. 1.34.
The volume of the original pellet is equal to
the volume of the resolidied droplet. This
means that any volatilization of the molten
droplet and reaction with the substrate do not
measurably affect its volume.
The volume of the spherical cap is:
Fig. 1.33 Temperature/pressure curve for diffusion bonding
of indium, for a process time of 1 h. Good-quality
joints are obtained from the conditions above the boundary line 1
and lesser-quality or no joint from those below.
V  h (h2  3A2)
6
solders used in this manner include Sn-40Pb and
In-48Sn. In both instances, the trick to obtaining
a satisfactory joint lies in ensuring that the faying Spread Ratio and Contact Angle
surfaces of both the components and the solder
are as free from nonmetallic contamination as The spread ratio, Sr is dened as:
possible. The joint was made by forming a length
of freshly cleaned solder wire into a ring, butt Plan area of spread on the substrate surface
welding the ends, and simply pressing it between Sr 
Plan area of the original spherical pellet
the two component surfaces, at room tempera-
ture. Sufficient load was applied to cause ow of
the solder out to the joint edge and a substantial A and a are related by the conservation of the
narrowing of the joint gap; so the process was volume of the droplet, that is:
clearly pressure welding. The resulting joint pro-
vided a key seal on a hermetic cavity, and the 4 1
product carried a 20-year guarantee against her- V  a3  h (h2  3A2)
3 6
meticity failure by the manufacturer. The merit
of using solder preforms for this application is
that they are readily available at low cost com- Therefore:
pared with specially deposited coatings of gold
or indium.
a
1
h(h2  3A2)1/3
2

Appendix A1.2:
Relationship among
Spread Ratio, Spread
Factor, and Contact
Angle of Droplets
Expressions describing the spread of a molten
metal droplet are derived under the following set
of idealized conditions:
The original metal pellet is in the form of a
spherical bead of radius a (and diameter
D  2a). Fig. 1.34 Spherical cap geometry
 Chapter 1: Introduction / 45

and A2  h2
R
2h
2
4A
Sr  Therefore:
h(h2  3A2) 2/3
2
From the geometry (Fig. 1.34), A  R sin  sin  
(A/h)  (h/A)
and h  R(1 cos ):
REFERENCES

4A2/h2 Ambrose, J.C., Nicholas, M.G., and Stone-

Sr 
ham, A.M., 1992. Kinetics of Braze Spread-
(1  3A /h )
2 2 2/3
ing, Proc. Conf. British Association for Braz-
ing and Soldering, 1992 Autumn Conference,
4 cot2 /2
 Coventry, U.K.
(1  3 cot2 /2) 2/3 Ambrose, J.C., Nicholas, M.G., and Stone-
ham, A.M., 1993. Dynamics of Liquid Drop
Spreading in Metal-Metal Systems, Acta Met-
for 0 <  < 180 all. Mater., Vol 41 (No. 8), p 23952401
Andrade, E.N.C, 1952. The Viscosity of Liq-
uids, Proc. Royal Soc., A, Vol 215, 36/356
Spread Factor and Contact Angle Bakulin, S.S., Shorshorov, M.Kh., and Sha-
piro, A.E., 1992. A Thermodynamic Ap-
proach to Optimising the Width of the Braz-
The spread factor is dened by:
ing Gap and the Amount of Brazing Alloy,
Weld. Int., Vol 6 (No. 6), p 473475
(h3  3A2h)1/3 h Bever, M.B., Ed., 1986. Encyclopedia of Ma-
Dh terials Science and Engineering, Pergamon
Sf  
D (h3  3A2h)1/3 Press, p 24582463
Brandon, D.G. and Kaplan, W.D., 1997. Join-
1 ing Processes: An Introduction, John Wiley
1 & Sons
(1  3A2/ h2) 1/3 de Gennes, P.G., 1985. Wetting: Statistics and
Dynamics, Rev. Mod. Phys., Vol 57 (No. 3),
1 p 827863
1
(1  3 cot2 /2) 1/3 Eustathopoulos, N. and Coudurier, L., 1979.
Wettability and Thermodynamic Properties
of Interfaces in Metallic Systems, Proc. Conf.
for 0 <  < 180 British Association for Brazing and Solder-
ing, Third International Conference, London,
paper 5
Feature, 1976. Diffusion BondingTomor-
Contact Angle and the rows Low-Cost Fabrication Tool, Met.
Dimensions of the Solidied Mater., Feb, p 3739
Fleetwood, M.J., et al., 1988. Control of Thin
Pool of Filler Strip Casting, Proc. Second International
Conf. Rapidly Solidied Materials, 79
From Fig. 1.34, it can be seen that: March (San Diego, CA)
Frear, D.R., Jones, W.B., and Kinsman, K.R.,
1990. Solder MechanicsA State of the Art
A  (R  h  R
2 )2 2
Assessment, TMS
Funk, E.R. and Udin, H., 1952. Brazing Hy-
according to the Pythagorean theorem. Rear- dromechanics, Weld. Res. Suppl., Vol 6, p
ranging this equation: 310s316s
46 / Principles of Soldering
Fuse, K., Obara, Y., and Irie, H., 1992. Super
Solder for Fine Pitch (0.15 mm) Applica-
tions, Proc. International Electronics Pack-
aging Conference, 2730 Sept (Austin, TX),
p 795811
Harkins, W.D., 1952. Physical Chemistry of
Surface Films, Van Nostrand Reinhold
Hill, A. and Wallach, E.R., 1989. Modeling
Solid State Diffusion Bonding, Acta Mater.
Metall., Vol 37 (No. 9), p 24252437
Humpston, G. and Baker, S.J., 1999. Diffu-
sion Bonding of Gold, Gold Bull., Vol 31
(No. 4), p 131132
Humpston, G. et al., 1992. Recent Develop-
ments in Silicon/Heat Sink Assemblies for
High Power Device Applications, GEC Rev.,
Vol 7 (No. 2), p 6778
Humpston, G. and Jacobson, D.M., 1990. Sol-
der Spread: A Criterion for Evaluation of Sol-
dering, Gold Bull., Vol 23 (No. 3), p 8395
Humpston, G. and Jacobson, D.M., 1991. The
Relationship between Solder Spread and Joint
Filling, GEC J. Res., Vol 8 (No. 3), p 145150
Jakeway, P., 1994. The Healthy Solderer, The
Health and Safety Practitioner,April, p 2023
Jones, H., 1982. Rapid Solidication of Met-
als and Alloys, Monograph No 8, The Insti-
tution of Metallurgists
Kapoor, R.R. and Eagar, T.W., 1989. Tin-
Based Reactive Solders for Ceramic/Metal
Joints, Metall. Trans. B, Vol 20B (No. 6), p
919924
Klein Wassink, R.J., 1989. Soldering in Elec-
tronics, 2nd ed., Electrochemical Publica-
tions
Kritsalis, P., Coudurier, L., and Eustathopou-
los, N., 1991. Contribution to the Study of
Reactive Wetting in the CuTi/Al2O3 System,
J. Mater. Sci., Vol 26, p 34003408
Landry, K., Rado, C., and Eustathopoulos,
N., 1996. Inuence of Interfacial Reaction
Rates on the Wetting and Driving Force in
Metal/Ceramic Systems, Metall. Mater.
Trans., A, Vol 27A, p 31813186
Laurent, V., Chatain, D., and Eustathopoulos,
N., 1991. Wettability of SiO2 and Oxidized
SiC by Aluminum, Mater. Sci. Eng. A, Vol
A135, p 8994
Lea, C., 1998a. Cleaning Electronic Assem-
blies, National Physical Laboratory
Lea, C., 1998b. Metal and Precision Clean-
ing, National Physical Laboratory
Ledolter, J. and Burrill, C.W., 1999. Statis-
tical Quality ControlStrategies and Tools
for Continual Improvement, John Wiley &
Sons
Li, J.G. and Hausner, H., 1991. Wettability of
Silicon Carbide by Gold, Germanium and
Silicon, J. Mater. Sci. Lett., Vol 10, p 1275
1276
Liu, C.Y. and Tu, K.N., 1998. Morphology of
Wetting Reactions of SnPb Alloys on Cu as
a Function of Alloy Composition, J. Mater.
Res., Vol 13 (No. 1), p 3744
Manko, H.H., 2002. Solders and Soldering,
4th ed., McGraw Hill
Meier, A., Javernick, D.A., and Edwards,
G.R., 1999. Ceramic-Metal Interfaces and the
Spreading of Reactive Liquids, J. Met., (No.
1), p 4447
Milner, R.D., 1958. A Survey of the Scientic
Principles Related to Wetting and Spreading,
Brit. Weld. J., Vol 5, p 90105
Nicholas, M.G., 1998. Joining Processes
Introduction to Brazing and Diffusion Bond-
ing, Kluwer Academic Publishers
Nicholas, M.G., 1989. Metal Surface Ener-
gies and Capillarity, Pub 466, Physical and
Electrical Characterisation of Metals, M.
McLean, Ed., The Institute of Metals, p 177
227
Nicholas, M.G. and Crispin, R.M., 1986.
Some Effects of Anisotropic Roughening on
the Wetting of Metal Surfaces, J. Mater. Sci.,
Vol 21, p 522528
Park, J.Y., et al., 2000. Study on the Solder-
ing in Partial Melting State: Analysis of Sur-
face Tension and Wettability, J. Electron.
Mater., Vol 29 (No. 10), p 11451152
Plotner, M., et al., 1991. Aspects of Indium
Solder Bumping and Indium Bump Bonding
for Assembling Cooled Mosaic Sensors, Hy-
brid Circuits, Vol 25, p 2730
Plumbridge, B., 1995. The Gas Turbine Lap-
top: A New Era for Solders, Mater. World,
Sept, p 422424
Richards, B.P., Burton, P., and Footner, P.K.,
1993. Damage-Free Ultrasonically Assisted
Cleaning of Printed Circuit Assemblies, GEC
J. Res., Vol 10 (No. 3), p 140157
Richards, B.P., et al., 1993. The Technical Op-
tions for Replacing CFCs for Cleaning Elec-
tronic Assemblies, GEC Rev., Vol 9 (No. 1),
p 320
Sax, N.I. and Lewis Sr., R.J., 1989. Danger-
ous Properties of Industrial Materials, 7th
ed., Van Nostrand Reinhold
Schwartz, M., 1987. Brazing, ASM Interna-
tional.
Seah, K.H.W., Hmanth, J., and Sharma, S.C.,
1998. Effect of the Cooling Rate on the Den-
drite Arm Spacing and the Ultimate Tensile

Strength of Cast Iron, J. Mater. Sci., Vol 33,
p 2328
Stromswold, E.I., Pratt, R.E., and Quesnel,
D.J., 1994. The Effect of Substrate Rough-
ness on the Fracture Toughness of Cu/96.5Sn-
3.5Ag Solder Joints, J. Electron. Mater., Vol
23 (No. 10), p 10471053
Sun, W. and Ivey, D.G., 2001. Microstruc-
tural Study of Co-Electroplated Au/Sn Al-
loys, J. Mater. Sci., Vol 36, p 757766
Tabor, D., 1969. Gases, Liquids and Solids,
Penguin Books
Tunca, N., Delamore, G.W., and Smith, R.W.,
1990. Corrosion of Mo, Nb, Cr, and Y in
Molten Aluminium, Metall. Trans. A, Vol 21A
(No. 11), p 29192928
Tylecote, R.F., 1967. Diffusion Bonding, Part
l, Weld. Met. Fabr., Vol 35 (No. 12), p 483
489
Tylecote, R.F., 1968. The Solid Phase Weld-
ing of Metals, Edward Arnold
Vianco, P.T. and Rejent, J.A., 1997. Capillary
Flow Solder Wettability Test, Solder. Surf.
Mt. Technol., Vol 25 (No. 2), p 47
Wang, X.H. and Conrad, H., 1995. Kinetics
of Wetting Ag and Cu Substrates by Molten
Chapter 1: Introduction / 47
60Sn40Pb, Metall. Mater. Trans., Vol 26A, p
459469
Watkins, M., Johnson, K.L., and Breyer,
N.N., 1975. Effect of Cold Work on Liquid-
Metal Embrittlement by Pb Alloys on 4145
Steel, Proc. Conf. Fourth Inter-American
Conference on Materials Technologies (Ca-
racas), p 3138
Weeks, J.R. and Gurinsky, D.H., 1958. Liq-
uid Metals and Solidication, American So-
ciety for Metals
Xian, A., 2000. Thermodynamic Discussion
on Youngs Equation in Wetting, Z. Met-
allkde., Vol 91 (No. 4), p 316322
Yost, F.G., Hosking, F.M., and Frear, D.R.,
1993. The Mechanics of Solder Alloy Wet-
ting and Spreading, Van Nostrand Rein-
hold
Yost, F.G., Michael, J.R., and Eisenmann,
E.T., 1995. Extensive Wetting due to Rough-
ness, Acta Metall. Mater., Vol 43 (No. 3), p
299305
Yost, F.G. and Romig, A.D., 1988. Thermo-
dynamics of Wetting by Liquid Metals, Ma-
terials Research Society Symposium Proceed-
ings, Vol 108, p 385390
Principles of Soldering
Giles Humpston, David M. Jacobson, p49-102
DOI:10.1361/prso2004p049
CHAPTER 2
Solders and Their Metallurgy
This chapter presents an overview of solder
alloy systems that one is likely to encounter.
Necessarily, the survey must include consider-
ation of the parent material with which the solder
is used because the suitability of a solder for a
particular joining process will depend largely on
its compatibility with the base materials.
Extensive reference is made to phase dia-
grams in order to highlight particular points. An
introduction to alloy constitution and phase dia-
grams designed for those with no background in
the subject is presented in section 2.3, which
covers interpretation of phase diagrams and the
associated terminology. Methods for their deter-
mination are summarized in the literature [Hump-
ston and Jacobson 1993, Chapter 3].
For a solder to be compatible with a particular
parent material, it must exhibit the following
characteristics:
A liquidus temperature below the melting
point (solidus temperature) of the parent ma-
terials and any surface metallizations. Usu-
ally a margin is required between these two
temperatures in order to achieve adequate u-
idity of the molten ller. Strictly speaking,
the uidity of solders is not a strong function
of temperature, but the overall ow behavior,
which is of immediate practical interest, does
often depend substantially on the application
temperature.
Capability of producing joints at tempera-
tures at which the properties of the base ma-
terials are not degraded. For example, many
work-hardened and precipitation-hardened
alloys cannot withstand elevated tempera-
tures without loss of their benecial mechani-
cal properties. Work hardening involves sub-
jecting the alloy to mechanical deformation
such as rolling or hammering, when reason-
ably cold. As the temperature is raised, the
Copyright 2004 ASM International
All rights reserved.
www.asminternational.org
deformation damage is removed by atomic
rearrangement in the metal. Precipitation
hardening is accomplished by creating a
nely divided phase within the material,
which can be thought of as akin to a com-
posite material. The dispersed phase is pre-
cipitated by means of an appropriate heating
schedule and its strengthening effect is like-
wise degraded by high temperatures or pro-
longed exposure at low temperatures, both of
which tend to coarsen this phase.
The ability of the parent materials, or a met-
allization applied to the parent materials, to
be wetted in order to ensure good adhesion
through the formation of metallic bonds. As
explained in the preceding chapter, it is not
possible to get spreading without wetting, but
an absence of spreading does not automati-
cally imply a lack of wetting. For example,
all solders will wet platinum, but only gold-
base solders will spread on this metal.
Limited erosion of the parent metals at the
joint interface. The associated alloying, which
must occur to form a metallic bond, should
not result in the formation of either a large
proportion of brittle phases within the joint or
of signicant concentrations of brittle phases
along interfaces or other critical regions of
the joint. Even ductile phases can have weak
interfaces with solidied ller alloys. Simi-
larly, the products of interalloying must not
generate other forms of weakness such as
corrosion or voids in the joint.
Elimination of constituents or impurities that
might embrittle or otherwise weaken the re-
sulting joint. Likewise, the parent material
must not contribute constituents or impurities
to the solder that will have a similar effect.
Besides being compatible with the parent ma-
terial, the solder and the joining process used
50 / Principles of Soldering

must be mutually suited. For example, solders The properties of the solder and the resulting
containing zinc, lead, or other volatile constitu- joint must also be compatible with the service
ents are not usually appropriate for furnace join- requirements. These are likely to involve a com-
ing at elevated temperature, especially when bination of at least some of the following con-
these involve reduced pressures. siderations:
The degree of temperature uniformity that can
be achieved over the joint area will have an in-
The strength and ductility of the joint should
meet certain minimum requirements over the
uence in determining the minimum tempera-
range of service temperatures. Soldered joints
ture difference that can be tolerated between the
are primarily used in the assembly of elec-
melting temperature of the ller and the melting
tronic and optical products where dynamic
or degradation temperature of the parent mate-
characteristics, such as resistance to thermal
rial. This consideration is particularly relevant to
fatigue and creep, are usually of greater im-
the joining of aluminum and stainless steel com-
portance than quasi-static mechanical
ponents, for contrary reasons. Stainless steel pos-
strength.
sesses relatively poor thermal conductivity,
whereas aluminum has a low thermal heat ca-
The design of the joint should not introduce
stress concentrations in the assembly that
pacity, both of which make it difficult to obtain
might arise through solidication shrinkage
uniform heating throughout a bulk assembly us-
or formation of intermetallic phases. Like-
ing local heat sources that do not encompass the
wise, the design should not cause undue dis-
workpiece.
tortion of the assembly from bimetallic ef-
The condition of the surface of the parent ma-
fects, excessive thermal expansion mismatch
terial may affect its compatibility with the solder,
strain, or the ller metal being too unyield-
especially when uxes are not used. As an ob-
ing.
vious example, ller metals will less readily wet
an oxidized surface than a freshly cleaned metal
The joint is normally required to be resilient
to the service environment, in terms of cor-
surface. This consideration often determines the
rosion and oxidation resistance and compat-
acceptable shelf life of components prior to join-
ibility with vacuum, in accordance with func-
ing. The term atomically clean should not gen-
tional requirements. Zinc-base solders are
erally be used to describe cleaned surfaces. It is
particularly limited in this regard.
an abstract ideal, not normally relevant in prac-
tical situations!
The ller must be compliant with statutory
needs. These include hallmarking regula-
In order to establish whether a particular par-
tions for precious metals and health restric-
ent metal (or nonmetal with a surface metallic
tions on lead and cadmium for certain culi-
coating) is compatible with a given solder, it is
nary and medical applications and nickel in
necessary for the appraisal to be carried out un-
applications that involve prolonged skin con-
der conditions that are likely to be representative
tact, such as spectacle frames.
of those used in any practical implementation of
the joining process. Parameters such as process
Aesthetic requirements are usually impor-
tant, for example, color, color matching in
time and temperature can be critical in this re-
jewelery and utensils, the ability of joints to
gard. Storage shelf life of the components (and
accept surface nishes such as paints, elec-
the ller when in the form of paste) is another
troplatings, and so on. Good llet formation
relevant factor that needs to be taken into ac-
is often demanded for aesthetic reasons and
count but is often neglected during transfer of a
also as a criterion of acceptable joint quality.
process from the laboratory to the factory. This
The reader is cautioned that the latter infer-
is particularly true where the parent metal has a
ence can be misleading, particularly for large
multilayer metallic coating, a situation that is
area planar joints (see Appendix A4.2).
commonly met with in soldering. Although the
outermost layer, often gold, will have indenite
Joints are normally required to possess cer-
tain thermal and electrical properties. These
shelf life, if it is not fully dense or sufficiently
often represent essential constraints in elec-
thick, oxygen will be able to percolate through
trical and optical assembly, which directly
it and oxidize less noble metals beneath the gold
determine tness for purpose.
layer. The problem then only becomes apparent
when the gold dissolves in the solder, exposing The simultaneous attainment of several of the
the underlying metal that has been rendered non- desired characteristics is frequently achievable
wettable during storage. with common ller metals, provided the basic

design guidelines and process conditions are sat-
ised. Lead-tin eutectic solder, which is taken to
include the Pb-60Sn alloy, accounts for the larg-
est proportion of joints involving ller metals
that are encountered. It is the more exceptional
and demanding service requirements that have
given rise to the development of the hundreds of
additional ller metal compositions. However, in
general, soldering is not a difficult process to
master. The commercially available llers and
matching uxes have been designed so that, when
used in conjunction with the common engineer-
ing parent materials, they meet many if not all of
the requirements listed previously.
2.1 Survey of
Solder Alloy Systems
The survey begins with consideration of lead-
tin solder on account of its preeminent position
among the lower-melting-point ller metals. The
review then extends to consider the less common
solders, arranged in rough order of usage and
decreasing melting point. To readers familiar
with brazing technology, one important differ-
ence of solders will soon be apparent. That is,
solders are largely based on unique eutectic com-
positions. The addition of other alloying ele-
ments usually destroys the eutectiferous charac-
ter of these alloys and the desirable soldering
properties that stem from it. Thus, the alloy com-
positions are normally exact rather than encom-
passing a broad range, an important exception
being the indium-lead series of alloys, which are
not eutectiferous. In any case, there is little in-
centive to introduce such modications because
there is a ready choice of alternative eutectic
solders having similar melting points, as illus-
trated by Fig. 1.3. For detailed coverage of the
available solder alloys, the reader should consult
reference publications [e.g., Klein Wassink 1989,
Manko 2002, Schwartz 2003] as well as suppli-
ers data sheets and manuals.
All solder alloys are based on combinations of
two or more constituents, chosen from nine com-
mon elements, namely antimony, bismuth, cad-
mium, gold, indium, lead, silver, tin, and zinc. Of
these, the cadmium-containing alloys have
largely been removed from manufacturers cata-
logs because they offer no clear advantages over
other solders, and their use is subject to restric-
tions arising from the toxicity of cadmium fumes.
As with beryllium, it transpires that cadmium
metal is relatively safe, and it is the oxide that is
Chapter 2: Solders and Their Metallurgy / 51
hazardous to health. In recent years, the use of
lead-containing solders for joining domestic wa-
ter pipes has been largely banned. This has led
to the development of lead-free solders for
plumbing applications [Irving 1992]. These sol-
ders comprise typically 95% tin with the balance
being one or more of silver, antimony, bismuth
copper, nickel, and zinc. Solders that do not con-
tain lead are also actively being used in elec-
tronics manufacture as part of a portfolio of mea-
sures by companies toward environmental
responsibility. Further information on this ini-
tiative is given in Chapter 5, section 5.1.
High-melting-point solders contain either gold
or lead as the principal constituents. The major-
ity of lower-melting-point solders, that is, those
with melting points below 150 C (300 F), fall
into two groups depending on whether the active
constituent of the alloy is tin or indium. This
classication is determined on the basis of the
likely reaction of the solder with the substrate
material, which, in the temperature range cov-
ered by solders, is necessarily a metal because
there is insufficient thermal activation available
for wetting nonmetals.
In most cases, wetting of a component by a
solder results in the formation of intermetallic
compounds, either within the ller or at the in-
terface between the solder and the parent mate-
rial. These intermetallic phases have a pro-
nounced effect on the mechanical properties of
the joint. As constituents of solders, indium and
tin have a dominant role in determining the in-
termetallic compounds that form by reaction with
the components: the compounds invariably con-
tain either indium or tin. When both indium and
tin are present, the composition of the parent
materials (i.e., the material on the surface of the
components) determines which of these ele-
ments is the predominant constituent of the re-
sulting intermetallic compounds. For solder al-
loys that contain neither element, intermetallic
compounds are not always formed and those that
do necessarily depend on the composition of both
the solder and of the parent materials. The mecha-
nism by which these intermetallics form and
some of the implications of their presence in
joints are discussed in section 2.3 in this chapter.
Each of the major elements in the lowest melt-
ing point soldersbismuth, indium, lead, and
tinconfers different properties to the ller met-
als, often in a manner that is not obvious. For
example, the fact that solders tend to be based on
eutectic composition alloys means that their melt-
ing points are determined by the eutectic reaction
52 / Principles of Soldering
rather than by the melting points of the indi-
vidual elements present. There is no simple re-
lationship between the melting temperatures of
the constituent elements and the eutectic tem-
perature. Therefore, while the elements fall into
the following sequence, lead > bismuth > tin >
indium, when ranked in descending order of
melting point, the solder alloys do not fall into
this pattern; those containing bismuth have the
lowest melting points of all. Despite such com-
plexities, it is possible to make some generali-
zations about the role of each element in a solder.
To this list is also appended antimony, not that
there are common solders based on this element,
but because it is sometimes a minor constituent
in other solder alloy families.
Tin is a preferred constituent of many solder
alloys because it confers uidity, benets wetting,
enhances mechanical and physical properties, and
possesses exceptionally low vapor pressure.After
lead, cadmium, and zinc, tin is the least expensive
ingredient of solder alloys. Nevertheless, the
price of tin is still some 30 times that of lead, which
accounts for the popularity of lead-rich lead-tin
solders. By way of comparison, bismuth and an-
timonyareaboutthesamepriceastin,withindium
and silver being over an order of magnitude more
expensive still. However, tin-bearing solders tend
Fig. 2.1 Gold-indium phase diagram
to form brittle compounds on reaction with many
parent materials and metallizations used in engi-
neering, particularly copper and gold. Silver is
one of the few exceptions, with silver-tin interme-
tallic phases being comparatively ductile. There-
fore, considerable care must be taken in designing
the joining process and specifying the service en-
vironment so as to restrict the formation of inter-
metallic compounds to concentrations below
those that would otherwise weaken and embrittle
the joints. This point is discussed in connection
with the phase diagrams of the relevant alloy sys-
tems in section 2.3 in this chapter.
Indium and lead are the two softest and most
ductile constituents of solder alloys. Despite their
inferior mechanical properties, solders with high
lead concentrations nd wide application be-
cause they are the least expensive and the easiest
to use of the high-melting-point solder alloys
available. The indium-bearing solders are par-
ticularly attractive for use with gold metalliza-
tions because these are not readily dissolved and
the interfacial phases that form are compara-
tively ductile, so that joints are not embrittled by
their presence. The low level of gold erosion
stems from a combination of the steep slope of
the liquidus line on the phase diagram between
indium and gold (see Fig. 2.1) and the formation

of a thin, continuous intermetallic compound
AuIn2 between the molten solder and the gold
metallization. This layer of compound then acts
as a barrier against signicant further gold dis-
solution from taking place and results in the pro-
le of the erosion curves shown in Fig. 2.2. Thus,
indium-containing solders can be reliably used
in conjunction with very thin gold metalliza-
tions.
Silver is a constituent of several solder alloys,
but only as a minor proportion, not simply on
account of its price premium, but because higher
concentrations (more than about 5%) result in a
sudden increase in the liquidus temperature to-
ward those of the silver-bearing brazes. Small
additions of silver are used primarily to enhance
mechanical properties of solders and joints and
to promote uidity by destabilizing native sur-
face oxides on the molten solder. Silver oxide is
not stable in air above 190 C (374 F). Unfor-
tunately, however, owing to industrial pollution
silver tarnish often contains sulde, which is far
more stable against thermal degradation. The Ag-
96Sn solder has among the best mechanical and
physical properties of any low-melting-point sol-
der alloy [Harada and Satoh 1990]. Silver-
containing solders tend to be preferred when join-
ing silver-coated components because the
presence of silver in the solder reduces the rate
and extent of scavenging from the metallization,
as shown by the data in Fig. 2.3.
Bismuth is the most brittle constituent of the
common solders and, for this reason, few solders
contain more than 50% of this element. Bismuth-
bearing alloys comprise the majority of the low-
est melting point solders, as can be seen from
Fig. 1.3. Bismuth exhibits the unusual charac-
teristic of expanding on freezing, enabling sol-
ders to be tailored to have essentially zero liquid-
Fig. 2.2 Erosion of a gold metallization by molten indium as
a function of reaction time and temperature. Similar
results are obtained for indium-base solders, including gold-
indium, silver-indium, indium-lead, and indium-tin.
Chapter 2: Solders and Their Metallurgy / 53
to-solid volume contraction by appropriately
adjusting the bismuth concentration. It has been
claimed that this property can confer benets in
making hermetic soldered joints [Dogra 1985].
Although alloys such as the bismuth-tin eutectic
solder can be prepared in a manner that renders
it soft and ductile, by rapid solidication, it will
subsequently embrittle, even at room tempera-
ture, owing to changes in the atomic lattice spac-
ing of the bismuth phase that is responsible for
the solid-state expansion [Hare, Corwin, and Re-
imer 1985]. The mechanical properties of bis-
muth-tin solders may be improved considerably
by the addition of 0.5% Ag. This low concen-
tration of silver does not appreciably affect the
melting range of the solder, but acts as a very
efficient grain rener that improves the tensile
ductility of the solder by a factor of three and
reduces the susceptibility of the solder to strain-
rate-dependent deformation behavior [McCor-
mack et al. 1997]. The low melting point and
inferior uidity of the bismuthbearing solders
impose constraints on the joint design and pro-
cessing conditions. For example, their relatively
low melting temperature means that aggressive
inorganic uxes are needed to chemically clean
the surfaces of the parent materials.
Antimony is often found as a deliberate minor
addition (<10%) in many solders, particularly
lead-tin alloys. This is because the addition im-
proves some key mechanical properties. In lead-
tin solder, antimony causes solid-solution
Fig. 2.3 Substantial reduction of the dissolution rate of silver
in lead-tin eutectic composition solder obtained by
small additions of silver to the alloy. Adapted from Bulwith and
Mackay [1985]
54 / Principles of Soldering
strengthening, up to a concentration of about 3%
(see Fig. 2.4). As the proportion of antimony
increases, SbSn cuboids form in the solder and,
while this produces further improvements in
properties such as creep, these benets are offset
by a tendency toward brittleness that is respon-
sible for high and undesirable scatter on reli-
ability data [Tomlinson and Bryan 1986].
Gold is the most expensive major constituent
of solders and, for that reason, the applications
of gold solders tend to be limited to high-value
electronics, photonics, and jewelry manufacture.
One of the chief attractions of these solders is
their melting point, which falls within the 300 to
500 C (570 to 930 F) gap that separates the
upper limit of the lead-base solders from the
lower limit of the available aluminum-bearing
brazes.
Zinc forms a eutectic alloy with aluminum at
the composition, Al-94Zn (melting point 381 C,
or 718 F). It is the only low-melting-point metal
that does not grossly enhance the corrosion of alu-
minum alloys because its electrode potential is
close to that of aluminum, as indicated by the data
in Table 2.1. Accordingly, zinc-aluminum solders
nd use in the joining of aluminum alloys. How-
ever, the refractory nature of aluminum oxide re-
quires the use of aggressive uxes to promote wet-
ting by zinc-base solders, and this creates other
problems, namely highly corrosive ux residues.
This outline of solder families is brief and
fairly generalized. Many solder alloys have
unique properties, conrmed in numerous stud-
ies, that show that there is often no regular pat-
tern among solder composition, melting point,
and joint properties. Indeed, the ranking order
for even simple mechanical tests can be radically
changed merely by altering the test conditions
used [Tomlinson and Fullylove 1992].
Fig. 2.4 Shearstrength of -brass joints made with lead-tinsol-
der containing varying concentrations of antimony
A list of some of the more common binary,
ternary, and quaternary composition solder al-
loys is given in Table 2.2, and the associated
phase diagrams of those in widest use are shown
in Fig. 2.5 to 2.15. Most of these alloys are
ductile and can be mechanically worked to pro-
duce preforms of virtually any desired geom-
etry. The majority of solders are based on eutec-
tic compositions, for reasons that are explained
in section 2.3 in this chapter. Many solder manu-
facturers also offer off-eutectic compositions in
these and other alloy systems. They are in-
cluded in the product range because they are
either less expensive, easier to fabricate as wire
and foil, or have a melting range, which is
sometimes a technical advantage. It is usually
the case that uidity and mechanical properties
suffer on moving to an off-eutectic composition,
but sometimes these characteristics are desir-
able. For example, the high-lead solders are
expressly used below their liquidus temperature
for the purpose of bridging wide gaps. On the
other hand, the indium-lead alloys offer a con-
tinuum of solder compositions with intermedi-
ate melting temperatures.
Although not included in Table 2.2, reference
has been noted in the literature to one other high-
melting-point solder. Details are sparse, but it
appears to be a quaternary alloy based on the
copper-silver eutectic braze with substantial ad-
ditions of indium and tin to depress the liquidus
and solidus temperatures. The melting point of
the solder is claimed to be 450 C (842 F), and,
therefore, it technically qualies as a solder
[Hirakawa, Tanahashi, and Terasawa 1995]. Fur-
ther details on conventional brazes based on cop-
per-silver and other alloys can be found in the
planned companion publication Principles of
Brazing.
Table 2.1 Electrode potential of selected
elements at 25 C (77 F)
Element Electrode Potential, V
Gold 1.50
Silver 0.80
Copper 0.34
Hydrogen 0.00
Lead 0.13
Tin 0.14
Nickel 0.25
Cadmium 0.40
Iron 0.44
Zinc 0.74
Silicon 1.30
Aluminum 1.66
Magnesium 2.37
 Chapter 2: Solders and Their Metallurgy / 55

Table 2.2 Low-melting-point eutectic composition alloys used as solders

Composition, wt% Eutectic temperature
Ag Bi In Pb Sn Other C F

... 49.0 21.0 18.0 12.0 ... 57 135

... 33.7 66.3 ... ... ... 72 162
... 52.0 ... 32.0 16.0 ... 95 203
... 67.0 33.0 ... ... ... 109 228
... ... 51.0 ... 49.0 ... 120 248
... 55.5 ... 44.5 ... ... 123 253
... 57.0 ... ... 43.0 ... 139 282
3.0 ... 97.0 ... ... ... 144 291
... ... 96.0 ... ... 4.0Zn 144 291
5.0 ... 80.0 15.0 ... ... 142149(a) 288300(a)
... ... 99.5 ... ... 0.5Au 156 313
... ... 75.0 25.0 ... ... 165170(a) 330340(a)
1.5 ... ... 36.0 62.5 ... 179 354
... ... ... 38.0 62.0 ... 183 361
... ... 50.0 50.0 ... ... 178210(a) 352410(a)
... ... ... ... 91.0 9.0Zn 198 388
... ... ... 85.0 ... 15.0Au 215 419
3.5 ... ... ... 96.5 ... 221 430
... ... ... ... 99.3 0.7Cu 227 441
... ... ... ... 95.0 5.0Sb 235245(a) 455475(a)
... ... ... 85.0 3.5 11.5Sb 240 464
25.0 ... ... ... 65.0 10.0Sb 232234(a) 450453(a)
... ... ... 88.9 ... 11.1Sb 251 484
... ... 25.0 75.0 ... ... 255265(a) 490510(a)
2.5 ... 5.0 92.5 ... ... 300310(a) 572590(a)
2.5 ... ... 97.5 ... ... 304 579
1.5 ... ... 97.5 1.0 ... 309310(a) 588590(a)
... ... ... 99.5 ... 0.5Zn 318 604

(a) These alloys are not eutectic compositions, but have been included on account of their narrow melting range and their industrial exploitation.

Fig. 2.5 Indium-tin phase diagram

56 / Principles of Soldering

2.1.1 Lead-Tin Solders Relatively inexpensive to produce and use.

Lead, in particular, is inexpensive compared
This survey necessarily starts with lead-tin sol-
with other solder metals, notably bismuth, in-
ders because they account for about 95% of all
dium, and silver. Moreover, the fabrication
soldered joints. Soldered joints are made on a
costs are low in comparison with those of most
vast, global scale. It is estimated that more than
solders as lead-tin alloys are readily cold
60,000 tonnes per annum (6  107 kg/yr or
worked.
54,000 tons/yr) of lead-tin solder are used in the
electronics assembly industry alone. The ben-
Greater versatility. Lead-tin solders readily
wet a wide range of metals to produce sound
ecial characteristics of lead-tin alloys have been
joints with minimal substrate erosion. Sol-
appreciated since at least Roman times, and the
dering to thick gold metallizations represents
elder Pliny, writing in the rst century A.D., in
one of the few situations in which restrictions
his Historia Naturalis [Rackham 1952] speci-
need to be observed in order to prevent em-
cally mentions an alloy containing two parts of
brittlement through the formation of unfa-
black lead (modern lead) and one part of white
vorable intermetallic phases. The reaction be-
lead (i.e., tin) used for soldering pipes. He also
tween lead-tin solders and gold metallizations
remarks that the price of this alloy is 20 denarii
is discussed in detail in section 2.3 in this
per pound. It is interesting to note that this works
chapter.
out at roughly $100/kg, assuming that gold has
maintained its purchasing power since Plinys
Ready application as coatings on compo-
nents by electroplating. The lead-tin alloy
day. Todays price for the same solder is lower
system is the only one for which low-cost
by an order of magnitude.
electroplating technology has been devel-
Lead-tin alloys offer the following advantages
oped successfully.
compared with other solders:
Satisfactory mechanical properties for many
Superior wetting and spreading character- applications. In terms of mechanical strength,
istics, compared with most other solders, stiffness, and fatigue resistance, only silver-
especially those containing bismuth, anti- tin solders are superior and even then by much
mony, and indium. The difference is quan- less than an order of magnitude and in par-
tied in Fig. 1.14. ticular circumstance.

Fig. 2.6 Bismuth-tin phase diagram


In most other respects, such as corrosion re-
sistance and electrical and thermal conductivity,
there is little to distinguish lead-tin alloys from
the majority of other low-melting-point solders
(see Table 5.19 in Chapter 5, section 5.7).
Lead-tin solders are seldom pure, but often
contain other elements, either as incidental im-
purities or as deliberate additions designed to
modify specic properties. Three metal addi-
tions are routinely encountered; antimony, bis-
muth, and silver.
Antimony is frequently present to a maximum
concentration of about 1%. It has a benecial role
in improving the mechanical properties of joints
though a solid-solution-strengthening mecha-
nism.Antimony can substitute for double the pro-
portion of tin, without greatly widening the melt-
ing range of the solder and, therefore, is often
favored by solder manufacturers as it decreases
materials costs. At higher concentrations of anti-
mony, there are solder alloys based on the ternary
system Pb-Sb-Sn. The liquidus projection of this
alloy system is given in Fig. 2.13, from which it
can be seen that there is a ternary eutectic at 240
C containing 11.5% Sb.
Bismuth is added for similar functional rea-
sons as antimony. The maximum concentration
is usually limited to about 3%, as higher levels
Fig. 2.7 Silver-indium phase diagram
Chapter 2: Solders and Their Metallurgy / 57
result in a signicant widening of the melting
range and a noticeable impairment to wetting
and spreading behavior. In addition, there is a
range of low-melting-point ternary solders based
on Bi-Pb-Sn alloys in which bismuth is present
in high concentrations. The liquidus surface of
this alloy system is given in Fig. 2.16. Small
quantities of bismuth and also antimony are
added to lead-tin solders to prevent degradation
at low temperatures through a mechanism known
as tin pest. Further details of this failure mecha-
nism are given in section 2.2 in this chapter.
Silver. Lead-silver-tin solders nd application
for soldering silver-coated surfaces. The addi-
tion of 2% Ag to the lead-tin eutectic alloy has
a marked effect in reducing the erosion of silver
coatings, as shown by the data in Fig. 2.3. The
presence of silver in lead-tin solders results in the
formation of a ne dispersion of the relatively
ductile intermetallic compound Ag3Sn, on so-
lidication, which boosts the mechanical prop-
erties of joints. Small quantities of copper and
gold have a similar benecial effect, as described
in section 2.2 in this chapter. However, these
elements are not usually present as deliberate
alloying additions since they are incorporated
automatically when soldering copper- and gold-
coated components, respectively.
58 / Principles of Soldering

All of the aforementioned additions to lead-tin tially with some of the tin and thereby serve to
solder, in small quantities, enhance the strength reduce the thickness of the copper-tin interme-
of joints to copper because they alloy preferen- tallic phases. These form at copper-solder inter-

Fig. 2.8 Lead-tin phase diagram

Fig. 2.9 Silver-tin phase diagram

 Chapter 2: Solders and Their Metallurgy / 59

faces and can be a source of mechanical weak- Attempts have been made to improve the me-
ness (see section 2.3 in this chapter) [Quan et al. chanical properties and resistance to failure of
1987]. lead-tin solders through creep and fatigue by a

Fig. 2.10 Antimony-tin phase diagram

Fig. 2.11 Silver-lead phase diagram

60 / Principles of Soldering
number of strengthening mechanisms. These
novel alloys are discussed in Chapter 5, sections
5.5, 5.6, and 5.8.
2.1.2 Other Tin-Base Solders
Two tin-base solders nd wide industrial ap-
plication. These are silver-tin eutectic and a Ag-
Sb-Sn solder often referred to as Alloy J.
Silver-Tin. Silver-tin is a eutectic alloy that
has a xed point generally accepted to be 221 C.
The composition of the eutectic is close to Ag-
96.5Sn. Alloys offered by manufacturers will de-
viate from this by up to 1.5% on account of cost,
but this results in a widening of the melting range.
The silver-tin eutectic forms the basis of virtually
all of the lead-free alloys offered as replace-
ments for lead-tin solder (see Chapter 5, section
5.1). In many respects, this solder is the closest
binary alloy to lead-tin solder. The melting points
are not greatly different, and their wetting and
spreading characteristics are comparable to those
of lead-tin. So are their mechanical properties,
although silver-tin alloys are generally slightly
superior in this respect for service temperatures
below 150 C (300 F). Both solders form closely
related intermetallic compounds on wetting of
the common engineering parent materials and
Fig. 2.12 Indium-lead phase diagram
metallizations. The hue and luster of solder llets
are visually indistinguishable, but the surfaces of
lead-tin solder llets are often smoother as this
solder contains a more even ratio of the two
constituents and hence a ner grain size.
Alloy J. The majority of solders are referred
to in manufacturers catalogs by a name or ter-
minology that gives a clue to the alloy compo-
sition, although part of the product range of one
well-known solder manufacturer is denoted by a
numbering system that can only be decoded with
aid of a reference card. More recently, it has been
noted that, with the advent of lead-free solders,
marketing departments have had some role in the
naming of new products. These names are de-
signed to convey an impression of green and
environmentally friendly products rather than
technical information about the product. One of
the cryptic solder names in the literature is Al-
loy J, and particular reference is made to this
relatively prominent alloy here to elucidate its
key features.
Alloy J has the composition 25Ag-65Sn-
10Sb. It has a narrow melting range, usually
cited simply as 233 C (451 F) [Olsen and Span-
jer 1981, Mackay and Levine 1986]. Containing
65% Sn, Alloy J qualies as a tin-base solder.
The alloy was originally developed as an alter-

native solder for silicon semiconductor die-
attach with ostensibly more favorable mechani-
cal properties, melting range, and price than the
gold-tin and gold-silicon solders. The name de-
rives from the systematic search of the Ag-Sb-Sn
system that was undertaken for suitable candi-
datesAlloy J was the tenth alloy composition
evaluated and the rst one found largely to meet
the target requirements, that of substantially su-
perior performance compared to lead-tin eutectic
under conditions of power cycling of semicon-
ductor die, as shown in Fig. 2.17 [Mackay and
Levine 1986].
Reference to the ternary phase diagram for
this alloy system (Fig. 2.15) shows that the 25Ag-
65Sn-10Sb solder is a three-phase mixture of tin,
Ag3Sn, and SbSn. The two intermetallic com-
pound constituents are relatively hard and brittle.
When produced and used conventionally, the
presence of these intermetallics makes the fab-
rication into convenient forms such as wire and
preforms extremely difficult and is detrimental to
the joint properties. This problem has been re-
solved through the production of the alloy using
rapid-solidication technology (see Chapter 1,
section 1.3.2.2). The rapid-solidication process
yields an alloy with a substantially ner micro-
Fig. 2.13 Liquidus surface of the Pb-Sb-Sn system
Chapter 2: Solders and Their Metallurgy / 61
structure with a grain size below 0.3 m (12 in.)
compared with 10 m (400 in.) when produced
by conventional ingot casting. The rapid-
solidication process also extends the solid solu-
bilities of both silver and antimony in tin so that
the volumetric proportion of intermetallic phase
in the alloy is correspondingly decreased. These
two features result in improved ductility and a
reduction in hardness.
Provided the soldering cycle is short and the
subsequent service temperature is not excessive,
the benets of a rened grain size persists in
joints made using solder prepared by rapid-
solidication processing. The reasons for this are
not fully understood, but the net effect is that
silicon die soldered to ceramic packages using
rapidly solidied Alloy J have improved resis-
tance to fatigue failure on power cycling over
solder produced by conventional casting, as in-
dicated by the data presented in Fig. 2.18 [Pi-
namaneni and Solomon 1986].
2.1.3 Zinc-Bearing Solders
One of the major applications of solders con-
taining zinc is for joining aluminum and its al-
loys [Finch 1985]. This metal plays an additional
62 / Principles of Soldering
role in helping to disrupt alumina surface layers
and so helps to promote wetting by the solder. A
ux-cored, zinc-base solder that is intended for
joining aluminum is available commercially [Ru-
bin 1982]. Such joints are obviously susceptible
to corrosion and tend to be used for making elec-
trical connections to functional aluminum parts
rather than structural assemblies. The key to mak-
ing a successful soldered joint to aluminum is the
formulation of the ux used. This topic is dis-
cussed in Chapter 3, section 3.2.2.1. When sol-
dering to aluminum parts, a frequently over-
looked consideration is the combination of high
thermal conductivity and expansivity of this
metal, coupled with low heat capacity. There-
fore, particular attention needs to be paid to the
heating method to minimize thermal gradients
and thermally induced distortion.
One of the zinc-bearing solders that is widely
used for joining aluminum is the eutectic com-
position alloy Al-94Zn (melting point is 381 C,
Fig. 2.14 Liquidus surface of the Ag-Cu-Sn system. Adapted from Petzow and Effenberg [1988]
or 718 F). The aluminum-zinc phase diagram
is given in Fig. 2.19. Often a few ppm of
gallium is added to the solder to aid wetting,
molten gallium being one of the few metals that
will not ball up on alumina, which is prevalent
on the surface of aluminum alloys. Spreading of
the solder is promoted by a reaction between
gallium oxide and alumina, rather than between
the respective metals. The relatively small de-
pression in the melting temperature of alumi-
num achieved by alloying with a large propor-
tion of zinc has one negative consequence,
namely the occurrence of signicant erosion of
the aluminum components in the region of the
joint. This can be a problem in situations where
it is required that component geometry is not
perceptibly changed by the joining operation.
Zinc-bearing alloys are not suitable for joining
processes that entail reduced pressure atmo-
spheres, owing to the inherent volatility of this
element.

All of the zinc-bearing solders contain high
percentages of zinc, generally in the range 75 to
95%, and their solidus temperatures cover the
temperature band 197 to 419 C (387 to 786 F).
The alloying elements present in signicant per-
centages are aluminum, cadmium, copper, and
tin. A representative list of these solders is given
in Table 2.3. Pure zinc is not used as a ller metal
because it tends to ball up on heating when used
in normal atmospheres, whereas when small
amounts of other elements are added, wetting of
surfaces is more readily achieved. The cadmium-
containing alloy listed in Table 2.3 is now gen-
erally avoided because cadmium fume is classed
as a health hazard.
The choice of alloying additions is made on
grounds of reducing the erosion of the substrate
metals and, as far as possible, of improving the
spreading and owing characteristics. In gen-
eral, the additions cannot be entertained as melt-
ing-point depressants because they mostly do not
reduce the liquidus temperature, only the solidus
temperature.
Over the years, considerable research has been
devoted to developing improved new solders
Fig. 2.15 Liquidus surface of the Ag-Sb-Sn system. Adapted from Petzow and Effenberg [1988]
Chapter 2: Solders and Their Metallurgy / 63
based on zinc [Harrison and Knights 1984] be-
cause these alloys have the following potential
attractions:
They can be used at sufficiently low tem-
peratures so as not to destroy the work-
hardened strength of the copper alloys. Yet,
at the same time, the joints that are produced
are typically two to four times stronger than
those obtained with the common solders.
The zinc solders are compatible with galva-
nized steel components.
They are inexpensive in relation to most sol-
ders, being approximately one-quarter the
cost of the lead-tin alloys, for a given volume
of ller metal.
They are lightweight, having approximately
half the density of lead-tin eutectic solder.
They possess high thermal conductivity, ex-
ceeding 100 W/m K (60 Btu/h ft F).
They are not hazardous: unlike cadmium and
lead, the effects of small amounts of zinc
absorbed in the human body are noncumu-
lative and temporary. Zinc is one of the vital
trace elements in the human diet (as are many
64 / Principles of Soldering

elements that are toxic in larger concentra-
tions).
The factors that have greatly limited the adop-
tion of the zinc solders include:
The potential strength of joints to copper al-
loys are compromised by the presence of em-
brittling copper-zinc compounds, which form
as interfacial phases. For this reason, the zinc-
base solders that have been formulated for
use with copper alloys often contain numer-
ous minor alloying additions to modify the
growth of these phases.
Zinc alloys in joints to heavy base metals
(e.g., those based on copper or iron) are sus-
ceptible to galvanic corrosion, a problem that
they share with aluminum ller alloys.
The low intrinsic material costs are largely
offset by high fabrication costs arising from
the low ductility of most zinc-base alloys.
The relatively high volume contraction on
solidication, which is typically 1 to 1.5
vol%, is detrimental to joint lling and can
cause stress concentration in joints.
The zinc alloys generally exhibit poor ow-
ing characteristics. The high oxidation rate of

Fig. 2.16 Liquidus surface of the Bi-Pb-Sn system

Fig. 2.17 Degradation of joint quality, as measured by through-thickness electrical resistivity for silicon semiconductor die attached
joints using three different solders and subject to power cycling
 Chapter 2: Solders and Their Metallurgy / 65

zinc, which is of the order of 2 to 3 m/min
(80 to 120 in./min) at 400 C (750 F), in air,
coupled with high surface tension and vis-
cosity (both approximately double those of
lead-tin alloys), are the main factors.
The high vapor pressure of zinc means that
these solders cannot be used in reduced pres-
sure atmospheres because the volatility of
this element exacerbates the formation of
voids in joints and contaminates furnace
equipment.
The high affinity of zinc for oxygen requires
the use of aggressive uxes when soldering
in ambient atmospheres which, in turn, leads
to cleaning and corrosion problems. The ux
efficacy is highly specic to the particular

Fig. 2.18 Cumulative failure data for TO-220 silicon semiconductor die, subject to power cycling. Joints were fabricated using
preforms prepared using conventional casting (dark bars in chart) and mechanical working or rapid solidication tech-
nology (white bars on chart). Adapted from Pinamaneni and Solomon [1986]

Fig. 2.19 Aluminum-zinc phase diagram

66 / Principles of Soldering
parent metal/solder combination on which it
is used. This makes it extremely difficult to
achieve tolerant soldering processes using
zinc-base ller metals, although consistent
results do appear to have been achieved with
nickel and copper substrates with uxes based
on L-glutamic acid and dimethylammonium
chloride using process temperatures as low as
250 C (480 F).
For many situations, the disadvantages of zinc-
bearing solders outweigh their prospective ben-
ets, which accounts for the fact that they have
not found much favor in industry. Nevertheless,
further development continues. One example is
the development of uxes based on organic tin
compounds. These chemicals are designed to de-
compose at zinc soldering process temperatures
and coat the substrate and ller metal with a thin
lm of tin. By this means, contact angles as low
as 20 can be obtained on copper substrates when
soldered in air using tin-zinc eutectic solder
[Vaynman and Fine 2000]. Another example is
the development of zinc-base solders as lead-
free substitutes for the high-lead family of ller
metals. In particular, an alloy of composition
Al-3Ga-3Mg-90Zn, which has a melting range
of 309 to 347 C (588 to 657 F), has been
proposed as a hard solder for die-attach appli-
cations in the electronics industry [Shimizu et al.
1999]. Flux pastes of zinc-base solders have also
been developed [Noguchi 1999].
2.1.4 Gold-Bearing Solders
Gold is unusual in that it is the only element
on which both brazes and solders are based; that
is, this element is the major constituent of both
solders and brazes. Further details about gold-
base brazes are given in the planned companion
volume Principles of Brazing; also discussed is
the use of gold solders for jewelry applications.
The gold-bearing solders are all gold-rich alloys
of eutectic composition and have melting points
between 278 and 363 C (532 and 685 F). They
are listed in Table 2.4 and their associated phase
diagrams are given in Fig. 2.20 to 2.23. In view
of their high cost, the applications of these alloys
tend to be limited and specialized. One of the
chief attractions of these solders is their melting
point, which falls within the 300 to 500 C (570
to 930 F) gap that separates the upper limit of
the lead-base solders from the lower limit of the
available aluminum-bearing brazes.
Gold-bearing solders have the advantage of
being suitable for joining to gold-metallized com-
ponents. Their compatibility with gold metalli-
zations is explained more fully later in this sec-
tion. Three of these solders (the gold-antimony
alloys being the exception) are used widely in the
electronics industry as high-temperature solders
for attaching semiconductor devices into pack-
ages and building hermetic enclosures for sen-
sitive compound semiconductors and optical de-
vices. Gold-antimony alloys are brittle, even
when prepared using rapid-solidication tech-
nology and, as far as the authors are aware, are
not much used as ller metals.
The principal gold-base solders are consid-
ered briey in the paragraphs that follow.
Gold-Silicon and Gold-Germanium. On ac-
count of their closely similar characteristics as
solders, these alloys may be considered together.
Gold-silicon alloys are primarily used in the form
of a foil preform for bonding silicon semicon-
ductor chips to gold-metallized pads in ceramic
packages. The alloy compositions used as sol-
ders are slightly gold-rich with respect to the
eutectic composition, generally at or close to
Au-2wt%Si. This is deliberate because the eu-
tectic alloy is too hard and brittle even to hot roll
to foil. By making the alloy gold-rich, the pro-
portion of the ductile gold phase in the micro-
structure is sufficient to improve the mechanical
properties to tolerable limits. Rapid solidica-
tion is unable to produce ductile foil because
rapid cooling of molten gold-silicon alloys re-
sults in the formation of a large volume fraction
of (metastable) gold-silicon intermetallic com-
pounds, principally Au3Si, which render the foil
too brittle to handle [Johnson and Johnson 1983].
An alternative method of applying the solder to
silicon components is simply to coat the back
surface of the silicon die with a thin layer of gold,
applied by a vapor-phase technique. On heating
the gold-metallized silicon to above 363 C (685
F), the resulting interdiffusion between the gold
and silicon generates liquid ller metal in situ.
The Au-2Si solder is an exception to the rule
of eutectic composition alloys, possessing fa-
vorable characteristics as ller metals. In par-
ticular, it suffers from high viscosity when mol-
ten. This characteristic is a direct consequence of
the low temperature of the eutectic transforma-
tion, relative to the high melting points of both
constituent phases, together with the silaceous
dross that tends to form on the surface of the
alloy and the absence of suitable uxes that might
enhance wetting. This can be seen by comparing
 Chapter 2: Solders and Their Metallurgy / 67

the spreading behavior of the Au-2Si alloy, as a common solder alloys given in Fig. 1.14. Poor
function of temperature above its melting point, uidity of the solder increases the risk of inad-
shown in Fig. 2.24 with similar data for other equately lled joints, which mar the performance
and reliability of the product.
It is generally recommended that gold-silicon
Table 2.3 Zinc-bearing alloys used as solders foil is given a light etch in hydrouoric acid and
Melting range
then used within 30 minutes. This process step
removes the surface silica and silicon and greatly
Solder composition C F
improves the wetting and spreading behavior.
Zn-6Al 381 718 While technically successful, this approach does
Zn-5Al-2Ag-1Ni 381387 718729
Zn-7Al-4Cu 379390 714735 require handling of this rather aggressive acid
Zn-10Cd 266399 511750 and special operator training is usually required
Zn-25Cd 266370 511700
Zn-1Cu 418424 784795 to satisfy health and safety regulations. Further-
Zn-30Cu-3Sb-1Ag 416424 781795 more, when using gold-silicon solders, precau-
Zn-10Sn-1Pb 197385 387725
Zn-70Sn 199311 390592
tions must be taken to ensure that the initial cool-
Zn-2Ni 419560 7861040 ing rate of solidied joints does not exceed 5 C/s
(9 F/s). If this condition is not satised, Au3Si
The cadmium-containing alloys are no longer used for health reasons.
is formed, and its subsequent decomposition with
time to gold and silicon can produce cracks
within the joint, due to the associated volume
contraction [Johnson and Johnson 1984].
Table 2.4 Gold-bearing solders A number of alloying additions are known to
Eutectic temperature be capable of promoting solder spreading
Composition, wt% C F [Humpston and Jacobson 1990]. For the Au-2Si
Au-20Sn 278 532 solder, one of the most effective promoters of
Au-25Sb 356 673 spreading is tin. Results of spreading tests con-
Au-12Ge 361 682
Au-3Si 363 685
ducted under identical conditions, but with dif-
ferent concentrations of tin in the solder showed

Fig. 2.20 Gold-silicon phase diagram

68 / Principles of Soldering

that increasing the level of tin gave a progressive this was accompanied by a softening of the alloy,
improvement in solder spread, as illustrated in with its hardness decreasing by more than 150
Fig. 2.25. Hardness measurements revealed that HV, a welcome feature because it makes the al-

Fig. 2.21 Gold-tin phase diagram

Fig. 2.22 Gold-germanium phase diagram


loy more amenable to mechanical working into
foil and wire for solder preforms.
The simplied phase diagram of the Au-Si-Sn
alloy system, reproduced in Fig. 2.26, indicates
that, by restricting the concentration of tin to
below 8%, the melting temperature of the alloy
is determined by the reaction among gold, sili-
con, and the Au5Sn intermetallic compound,
which occurs over a narrow melting range of
356.5 to 358.0 C (673.7 to 676.4 F). This is
closely similar to the solidus temperature of the
Au-2Si alloy (363 C, or 685 F), and switching
of this latter alloy for one additionally containing
tin will therefore not upset subsequent manu-
facturing steps.
Semiconductor die attach represents one of
the main areas of application of gold-silicon
and gold-germanium solders. Neither will di-
rectly wet bare silicon. Silicon is a relatively
refractory element that, when exposed to air,
becomes covered with a stable, continuous, and
extremely adherent layer of the native oxide
(silica). For this reason, all conventional sol-
der alloys are unable to wet silicon even with
uxes. In order to achieve wetting, it is nec-
essary to either coat the back of precleaned die
with a solderable metallization, such as gold,
or to manually scrub the molten ller between
the die and the package and thereby physically
Fig. 2.23 Gold-antimony phase diagram
Chapter 2: Solders and Their Metallurgy / 69
displace the silica layer. The rst option is not
always compatible with other stages of semi-
conductor fabrication and so scrubbing is re-
sorted to, using a protective shroud of inert
gas. With the advent of automatic machines to
perform this operation, it has evolved to be a
rapid joining method that delivers consistent
quality joints.
A successful method of promoting wetting of
oxidized metal surfaces is to incorporate ele-
ments into the ller that are capable of reacting
with the oxide to form an adherent bond [Crispin
and Nicholas 1984, Xian and Si 1991]. Research
has shown that addition of just 1% Ti to the
Au-2Si and Au-20Sn solders is highly effective
in promoting wetting of bare silicon at soldering
temperatures of around 400 C. Concentrations
of titanium of up to 2% were found to have little
effect on the melting ranges of the alloys and are
actually benecial in reducing their hardness.
Figure 1.29 shows a silicon die soldered into a
gold-metallized package using a gold solder with
the composition Au-2Si-8Sn-1Ti; the joint was
made by heating in a nitrogen atmosphere with-
out scrubbing. Wetting of the ceramic package
and the silicon die by the solder was found to be
highly uniform.
The gold-germanium eutectic composition al-
loy is sometimes recommended in place of the
70 / Principles of Soldering

gold-silicon solder where cost margins are par-

Fig. 2.24 Spreading behavior of Au-2Si solder as a function
of the excess temperature above its melting point
Fig. 2.25 Spreading behavior of Au-2Si solder containing
tin. Left to right: 0%, 4%, 8% Sn
ticularly tight. As can be seen from the phase
diagram given in Fig. 2.22, it offers the benet
of lower gold concentration with little change in
melting point (note in some older texts the melt-
ing point of the gold-germanium eutectic is er-
roneously given as 256 C). The gold-germa-
nium binary eutectic contains 12.5 wt% Ge
compared with 3.2 wt% Si in the case of the
gold-silicon eutectic alloy. However, the price of
germanium is much higher than that of silicon
(because this element has to be produced as a
minor by-product of zinc), but is considerably
below that of gold. Historically, the price of ger-
manium is about 5% that of gold, although in
2002, it stood at 10% of the gold price, so that
the saving in materials cost with respect to gold-
silicon is fairly signicant. As a solder, gold-
germanium is no easier to use than its silicon
equivalent. The Au-12Ge solder reportedly ex-
hibits excellent wetting characteristics on cop-
per, silver, and nickel surfaces [Hosking,
Stephens, and Rejent 1999]. Where this solder is
to be used in conjunction with steel parts, the
parts should be plated with nickel prior to sol-
dering to avoid the formation of iron germides.
Hosking, Stephens, and Rejent [1999] have also
shown that although Au-12Ge solder is brittle at
temperatures up to 170 C (340 F), above about
220 C (430 F), this alloy is reasonably ductile

Fig. 2.26 Gold-rich portion of the Au-Si-Sn phase diagram. The system can be divided into three regions by melting point. The circle
indicates the alloy composition selected as the solder. Adapted from Prince, Raynor, and Evans [1990]

and it is possible to substantially remove expan-
sion mismatch stresses by a postsoldering heat
treatment at this temperature. The same will pre-
sumably be true of the Au-2Si solder, but the
softening temperature is not known for this alloy.
Gold-Tin. The Au-20Sn eutectic solder is hard
and moderately brittle. These properties arise
from the fact that its constituent phases are two
gold-tin intermetallic compounds, namely, AuSn
() and Au5Sn (). The Au5Sn phase is stable over
a range of compositions and, consequently, has
some limited ductility that is imparted to the
solder alloy (approximately 2% at room tem-
perature). Although difficult, this solder can be
hot rolled to foil and preforms stamped from it.
By using rapid-solidication casting technology,
it is possible to produce thin ductile foil, up to
about 75 m (3 mil) thick and having an amor-
phous microstructure. However, this state is
somewhat unstable, and within about 30 min at
room temperature the rapidly solidied strip is
indistinguishable in its mechanical properties
from foil prepared by conventional fabrication
methods [Mattern 1989]. Nevertheless, this crys-
tallization can be suppressed for about a year if
the quench-cooled material is stored under liquid
nitrogen (196 C, or 320 F), and for close
to one month at 20 C (4 F), so that it is
possible to manufacture shaped foil preforms
while the strip is still ductile, which can then be
either immediately placed in a jig or returned to
cold storage. The alloy can also be readily gas
atomized to form spherical powder and, because
it is relatively inert, will survive for long periods
in an organic binder medium (optionally con-
taining a ux) without degradation. For this rea-
son, the solder alloy is often used in the form of
paste, provided the components being joined are
compatible with the chemicals involved.
An alternative method of introducing the Au-
20Sn solder into an assembly is to selectively
coat the joint area with a thick layer of gold,
overlaid with a thinner layer of tin in the thick-
ness ratio of 3Au to 2Sn. If electroplated, the
layers must be deposited in this order owing to
their electrochemical potentials. By subjecting
the tin to a light acid etch and then immediately
applying an outer layer of gold by evaporation or
immersion plating, the predeposited solder is
suitable for use without ux and can be endowed
with a shelf life of several months. Other meth-
ods of producing ductile preforms of gold-tin
solders are described in Chapter 4, section 4.1.5.
The principal applications of the Au-20Sn
solder are die attach of gold-metallized chips
Chapter 2: Solders and Their Metallurgy / 71
and the hermetic sealing of ceramic semicon-
ductor packages. The die-attach process that is
used when the sealing operation is performed
with the Au-20Sn solder is usually a higher-
melting-point gold-silicon solder. Thus, the
gold-silicon and gold-tin solders are used in the
initial joining operations of a step-soldering se-
quence, which ends with the soldering of the
packages containing the chips onto printed cir-
cuit boards. This nal soldering operation is
most frequently performed using a lead-tin or
lead-free solder at a process temperature in the
region of 240 C (464 F).
Of the gold-bearing solders, only the Au-20Sn
alloy has a signicant degree of uidity when
molten. In their major eld of application, that of
semiconductor manufacture, the use of uxes to
promote spreading is not usually permitted. In-
stead, the techniques and methods normally ap-
plied to uxless joining are invoked to effect
reliable joints. The options available and under-
lying principles of uxless joining are described
in detail in Chapter 3, section 3.3.
The gold-tin eutectic solder tends mostly to be
used for soldering to thick gold metallizations.
There are some difficulties with this, notably the
dissolution of gold increases the freezing point
and prevents good spreading. Identication of
wettable, but insoluble barrier metallizations for
this solder is therefore desirable. Copper, nickel,
chromium, and nichrome are wholly unsuitable
as they all dissolve readily in the Au-20Sn alloy.
Palladium is a good wettable metallization for
the molten solder, but, on extended aging in the
solid state, Kirkendall voids form at the interface
between the residual palladium and the Pd3Sn2
intermetallic compound [Anhock et al. 1998].
The solubility of platinum in Au-20Sn solder is
strongly temperature dependent. At normal sol-
dering temperatures with short process cycles, a
thin layer of platinum (200 nm, or 8 in.) pro-
vides a readily wettable and stable barrier layer
[Wada and Kumai 1991]. However, if large su-
perheats or prolonged thermal excursions are
necessary, then published work suggests that co-
balt is a better choice. The dissolution rate of
platinum in gold-tin eutectic at 330 C (626 F)
is 10 nm/s (0.4 in./s), but nearly an order of
magnitude slower for cobalt. Both gold and tin
have very low solubility for cobalt. The me-
chanical properties of cobalt-tin intermetallics
are not known, but, as the layers formed during
normal soldering cycles will be extremely thin,
they are unlikely to inuence joint properties
signicantly [Park et al. 2002].
72 / Principles of Soldering
A high-melting-point quaternary ller metal
based on the gold-tin eutectic became available
relatively recently, intended for joining to thick-
lm metallizations [DuPont Electronics 1998].
The design strategy appears to have been to sub-
stitute some of the gold by silver and copper,
both of which form solid solutions with gold.
The exact composition and melting range have
not been disclosed, but the recommended pro-
cess temperature is 400 C (750 F), compared
with 350 C (660 F) for the Au-20Sn binary
composition from the same supplier. This im-
plies the melting range is probably somewhere
around 300 to 350 C (570 to 660 F). Among
the claimed benets of the substitution are higher
melting point, greater ductility, and improved
wettability.
A noneutectic gold-indium alloy of composi-
tion 82Au-18In is offered commercially by a
number of manufacturers. The alloy is gold-rich
to the nearest eutectic and deliberately so be-
cause of the poor mechanical properties of the
eutectic composition alloy. The demand for this
solder and its inclusion in catalogs stems from its
melting range of 451 to 485 C (844 to 905 F),
which makes it the highest melting point solder
available. However this gold-indium alloy is a
difficult solder to work with, as it is inferior to
gold-silicon and gold-tin alloys with regard to its
wetting and ow characteristics, as indicated in
Table 2.5. Also, there is no developed ux for
this solder and it must be used uxless, with all
the attendant problems and limitations, although
the process temperature is sufficiently high that
this is one of the rare examples where a hydrogen-
containing atmosphere is chemically active to-
ward surface oxides on the solder (see Chapter
3, section 3.1.3).
2.1.5 High-Lead Solders
High-lead solders can be loosely dened as
alloys containing 95% or more Pb and having
Table 2.5 Contact angle of Au-12Ge and
Au-18In solders on common metal substrates
Contact angle, degrees
Substrate Au-12Ge(a) Au-18In(a)
Copper 4 35
Silver 4 58
Nickel 5 43
(a) The process temperature for gold-germanium was 430 C (806 F) and
gold-indium 550 C (1022 F), with 5 min hold at peak temperature in a vacuum
of 7 mPa ( 5  105 torr). Adapted from Hosking, Stephens, and Rejent [1999].
solidus temperatures above 300 C (570 F). A
representative list of these solders is given in
Table 2.6. They possess small melting ranges,
usually, about 5 to 10 C (9 to 18 F), and their
usage arises largely on account of their melting
point. They are, in effect, inexpensive alterna-
tives to the Au-20Sn eutectic alloy or, with care,
can be used in a step-soldering sequence in which
the next ller metal is the gold-rich solder. These
high-lead solders differ in one important respect
to the Au-20Sn alloy in that they are extremely
soft and creep readily. This can be either an ad-
vantage or drawback, depending on the appli-
cation and the function that the solder joint is
required to provide.
The high-lead alloys possess other advantages
as solders. They are suitable for use with gold-
plated components without the risk of cata-
strophic embrittlement through the formation of
continuous AuSn4 intermetallic phase at the joint
interface. Although there is a small percentage of
tin in some of the high-lead solders, the alloy con-
stitution allows for appreciable gold dissolution
before the formation of undesirable gold-tin
phases because the small amount of tin present is
mostly incorporated in the lead solid solution.
With small adjustments of composition, some
high lead-base solders have wide melting range
and consequently have excellent gap-bridging
characteristics. In the development of electronic
packaging fabrication technology, the Ag-
92.5Pb-5Sn alloy applied in the form of a propri-
etary solder paste was found to produce consistent
hermetic seals for feed-throughs into hybrid mi-
crowave packages despite joint clearances vary-
ing from below 20 m (0.8 mil) to more than 150
m (60 mil), as can be seen in Fig. 2.27 [Jacobson
and Sangha 1997].
An obvious omission from any list of high-
melting-point lead-rich solders is pure lead. The
reason for this is pure lead has atrocious wetting
Table 2.6 Selected high-melting-point
lead-rich solders
Melting range
Solder composition C F
98.5Pb-1.5Sb 310322 590612
95Pb-5Sn 308312 586594
92.5Pb-5In-2.5Ag 300310 572590
97.5Pb-1.5Ag-1Sn 309310 588590
93Pb-3Sn-2In-2Ag 304305 579581
97.5Pb-2.5Ag 304 581
95Pb-5In 300313 572595
93.5Pb-5Sn-1.5Ag 296301 565574
90Pb-5In-5Ag 290310 554590
92.5Pb-5Sn-2.5Ag 287296 549565
 Chapter 2: Solders and Their Metallurgy / 73

and spreading behavior. This is a consequence of 2.1.6 Indium Solders

the fact that lead does not alloy readily with most
Indium-base solders share the common char-
engineering metals and metallizations, so that
acteristics of low melting point and being ex-
there is only marginal thermodynamic driving
tremely soft and ductile. The mechanical prop-
force for wetting and spreading. Indeed, this is
erties are mostly a reection of the fact that at
the sole reason for lead being unsuitable on its
room-temperature indium solders are operating
own as a solder and requiring alloying additions.
at a very high homologous temperature. That is,
The list of high-lead alloys given in Table 2.6
25 C (77 F) is close to their melting point when
is by no means exhaustive. Many other variations
expressed in Kelvin. Indium remains ductile even
exist, but the basis of their formulationnamely,
at cryogenic temperatures. At high homologous
small additions of silver, tin, indium, antimony,
temperatures, the rate of solid-state diffusion in
and other elementsis similar. Data sources cite
metals with simple crystal structures is suffi-
slightly different solidus and liquidus tempera-
ciently fast that microstructural changes can oc-
tures for many of these alloys, but the discrepan-
cur in timescales that are comparable to changes
cies are mostly less than 5 C (9 F).
in the service environment of joints in compo-
Although the mechanical properties of lead-
nents. This is exemplied by the stress-strain
rich solders are often maligned, they will actu-
curve of a thick indium soldered joint and the
ally deliver a joint with respectable mechanical
continuum between stress-strain and creep data
properties. Chadwick peel-test results with a
given in Fig. 2.28 and 2.29, respectively [Freer
98Pb-2Sn solder can have peel-fracture initia-
Goldstein and Morris Jr. 1994; Darveaux and
tion strengths of more than 35 kN/m (200 lb/in.)
Linga Murty 1993]. This means that recovery
when optimally made [Beal 1984]. Lead-rich sol-
and recrystallization occur as fast as work hard-
ders containing 98% Pb appear to be a good
ening is induced, and mechanical failure of joints
initial choice for applications where mechanical
made using indium-base solders tends to be stress
properties are important, as indicated in Table
overload or unidirectional creep.
2.7. At lower lead contents, the deformation char-
The ready creep of indium solders implies that
acteristics are slip bands that are balanced in
joints are unlikely to fail on thermal cycling un-
distribution between the grains and the grain
less the load-displacement curve is asymmetric.
boundaries. At 98% Pb, the grain boundaries ap-
It follows that indium solders are well suited for
pear to be more robust, leading to more heavily
making joints between dissimilar materials that
localized strain effects and superior creep resis-
will be subject to thermal cycling. In fact, creep
tance. Above this concentration of lead, the sol-
in joints made with indium solders can usually
der progressively behaves more like a homoge-
occur sufficiently fast to ensure that the stress is
neous single-phase material.
always close to zero, with roughly 80% stress
relaxation occurring within seconds of a step
change in strain. Then, the solder will not work
harden. If joints are suitably designed to take
advantage of these benecial thermomechanical
characteristics, indium solders can provide su-
perior life (see Fig. 2.30, also Humpston et al.
2001). When indium soldered joints do fail, the
mechanism is predominantly by classical creep
rupture, having its origins in the nucleation

Table 2.7 Mechanical properties and corrosion

resistance of high-lead solders containing tin
Corrosion
Stress-rupture weight loss
Lead Peel fracture initiation time, 142 N (32 ASTM D 1384
content, % kN/m lbf/in. lbf) load, days 2 weeks, mg
Fig. 2.27 Microsection through a joint of varying width made
using the 1.5Ag-92.5Pb-5Sn solder in the form of 100 17.5 100 122 10.1
a paste to two electronic packaging components in lightweight 99 19.25 110 8.3 16.1
Al-70Si (Osprey CE7 alloy), plated with nickel and gold. The 98 36.75 210 28.6 9.2
blackish region constituting the joint is the solder, which com- 97 20.12 115 0.55 16.2
pletely lls the gap, which varies in width from less than 20 to 96 21.88 125 0.79 10.4
more than 150 m (0.79 to 5.9 mil).
74 / Principles of Soldering
and coalescence of cavities, arising from the ma-
terial redistribution associated with stress relax-
ation. In contrast, tin-base solders, when subject
to conditions of low-cycle (low-strain rate) fa-
tigue, fail because the plastic deformation in the
solder leads to microstructural changes in the
region of maximum strain. Internal cavities then
develop, which coalesce to form voids and then
grow into cracks.
Fig. 2.28 Shear stress strain curve for a 500 m thick indium-
tin (In-48Sn) joint held at 40 C (104 F) ambient
and strained at a rate of 5  104/s. Adapted from Freer Goldstein
and Morris Jr. [1994]
Fig. 2.29 Continuum between stress-strain and creep data
for indium-tin eutectic solder (In-48Sn) at room
and elevated temperature. Adapted from Darveaux and Murty
[1993]
Fig. 2.30 Resistance of a ip-chip daisy chain between sili-
con and alumina, versus the number of cycles (N)
of a thermal shock test 25 to 196 C, 30 s dwell for the solders
indicated. Adapted from Shimizu et al. [1995]
Indium-base solders used in optoelectronic and
photonics applications can also fail by a process
known as phase segregation. This occurs when
there is an extreme and unidirectional electrical
and/or thermal gradient sustained across a joint.
The result is migration of indium toward one
joint interface and the accumulation of voids at
the other until the joint fails. This mechanism is
generally only observed in applications such as
die attach of microwave power ampliers and
laser diodes where, although the absolute power
levels are modest, the small physical size of the
parts results in very high energy ux.
Pure indium is not often used as a solder be-
cause the wetting and spreading characteristics
are mediocre as are the mechanical properties of
joints made using it. One exception stems from
exploitation of the complex oxide that forms on
indium. Very-high-purity indium is readily avail-
able because this metal is chemically extracted
from zinc residues as a minor by-product. Pro-
vided the indium is of purity better than
99.99995%, it will wet and spread over unmet-
allized oxide ceramics and glass, in air, without
ux. The resulting joints do not have the same
strength (5 to 10 MPa, or 725 to 1500 psi) and
fracture toughness as conventional soldered
joints, but are nevertheless hermetic and usable
in a limited range of applications [Indium Cor-
poration of America 1994].
The low melting point of indium solders stems
from the low melting point of indium itself, which
is 157 C (315 F). It is interesting to note that
the cited melting point of indium has increased
by nearly 5 C (9 F) over the last 40 years with
the development of improved rening methods;
the melting point of the metal being very sen-
sitive to low levels of metallic impurities. Albeit
largely as a point of metallurgical curiosity, the
melting point of indium is also unusually sen-
sitive to pressure; the application of 4000 MPa
(580 ksi) will cause the melting point to roughly
double to 300 C (570 F; see Fig. 2.31).
Alloying additions made to indium to form
lower-melting-point eutectics confer a number
of practical benets. Table 2.8 lists some of the
more common indium solders and their melting
points. The merits of alloying are principally the
generation of multiphase microstructures, which
improve the mechanical properties of the solder,
and the formation of mixed-composition oxides,
which are generally easier to chemically remove
than pure indium oxide. Indium oxide forms
readily as a sticky and tenacious lm that re-
quires specially formulated uxes to effect its

Fig. 2.31 Melting point of indium as a function of pressure. Adapted from Kennedy and Newton [1963]
removal. Even so, as can be seen from Fig. 1.14,
indium solders are appreciably less uid than
their tin-base counterparts and require greater
superheats to achieve comparable ow.
Indium-lead alloys are a useful class of solders
with readily adjustable melting ranges. Consid-
eration of the phase diagram for this alloy sys-
tem, which is given in Fig. 2.12, reveals that
indium and lead form an essentially miscible
mixture that extends between the melting points
of the two parent materials, that is, 157 to 328 C
(315 to 622 F). Thus, by appropriate selection
of the composition it is possible to obtain a sol-
der, albeit one with a melting range and mediocre
mechanical properties, that has any desired soli-
dus temperature between the two limits. Al-
though indium-lead alloys solidify by peritectic
reaction, the solidication rate of most soldered
joints is usually sufficiently slow to prevent prob-
lems associated with microstructural coring, but
this must be considered if a fast heating method,
such as laser soldering, is being employed.
Indium-containing solders are often recom-
mended for joining to gold-coated components
because gold is less soluble in these alloys than
Table 2.8 Composition and melting ranges of
some common indium-base solders
Melting range
Composition, % C F
In-50Pb 178210
In-30Pb 165172
In-3Ag 141
In-5Ag-15Pb 142149
In-40Sn-20Pb 121130
In-49Sn 120
In-67Bi 110
In-34Bi 72
In-32.5Bi-16.5Sn 60
Chapter 2: Solders and Their Metallurgy / 75
in lead-tin solders. The restricted solubility of
gold in a joint containing indium is largely as-
sociated with the formation of a continuous layer
of the intermetallic compound AuIn2 at the sol-
der/gold interface, which effectively suppresses
further reaction between these metals (see Fig.
2.2). In the presence of lead, the interfacial layer
takes the form of separate grains of the AuIn2
compound embedded entirely in primary lead.
The lead provides an easy diffusion path between
the solder and the gold coating and so permits the
reaction to continue, albeit at a modest rate over
extended timescales [Yost, Ganyard, and Kar-
nowsky 1976; Jacobson and Humpston 1989].
Intermediate melting temperature indium-
base solders, such as the In-15Pb-5Ag compo-
sition, are signicantly more ductile than lead-tin
eutectic and can yield joints with correspond-
ingly superior thermal fatigue performance [Ed-
wards, Nixon, and Lakes 2000]. Although the
melting range of this alloy is close to the eutectic
temperature of lead-tin solder, its price differ-
ential is a factor of about 80, which restricts its
use to high-added-value and more specialist ap-
plications.
2.2 Effect of Metallic Impurities
As might be expected, metallic impurities
352410 present in ller metals can be either highly ben-
329342
286
ecial, benign, or totally undesirable, depending
288300 on the solder, the impurity, and the application.
250266 The effects of impurities have been most widely
248
230 studied for lead-tin solder.
162 Lead-tin solder is generally prepared to high-
140
purity specications because the presence of
76 / Principles of Soldering
many other elements can have a deleterious ef-
fect on joint formation, even in concentrations as
low as 5 ppm [Schmitt-Thomas and Becker 1988;
Becker 1991]. With some elements, joint for-
mation may be inhibited or even prevented. The
detrimental characteristics of the elemental ad-
ditions and the critical concentrations that pro-
duce these adverse effects are summarized in
Table 2.9. There was no minimum level of alu-
minum that did not adversely affect solderability,
except on brass substrates (see below): the low-
est concentration investigated was 0.0005%.
A detailed study of the effect of different el-
ements on the wetting properties of eutectic lead-
tin solder was made by Ackroyd, Mackay, and
Thwaites [1975]. They assessed the solderability
on copper, brass, and steel coupons using area-
of-spread tests, rotary dip tests, and wetting bal-
ance measurements. High-purity Pb-60Sn solder
was deliberately adulterated with low concen-
trations (<1%) of aluminum, arsenic, bismuth,
cadmium, copper, phosphorus, sulfur, antimony,
and zinc and used under an ambient atmosphere
with selected active and nonactivated uxes. Of
the substrate materials, brass exhibited a con-
sistent enhancement in spread area with increas-
ing impurity addition, irrespective of the element
in the above selection. By contrast, the solder-
ability of copper and steel almost invariably de-
teriorated in the presence of subpercentage lev-
els of the impurity and then continued to decline
or tended to a steady value at higher concentra-
tions.
The effect of major ternary additions on the
wetting angle of Pb-80Sn solder on copper, brass,
and mild steel substrates is given in Table 2.10
[Raman 1977]. Bismuth was found to be gen-
erally benecial; other additions such as anti-
mony and indium exhibited a dependence on the
substrate. The ternary additions consistently im-
Table 2.9 Lowest impurity concentrations
producing detrimental effects, in terms of
wetting, in Pb-60Sn solder
Impurity element Impurity concentration, %
Aluminium <0.0005
Antimony 1.0
Arsenic 0.2
Bismuth 0.5
Cadmium 0.15
Copper 0.29
Phosphorus 0.01
Sulfur 0.0015
Zinc 0.003
Adapted from Ackroyd Mackay, and Thwaites [1975]
proved the strength of butt joints to copper and
brass substrates, but markedly decreased the joint
strength for the mild steel substrates.
One important area where impurity elements
have a positive role in solders is in suppressing
tin-pest in lead-tin solders. Pure tin can exist
in two allotropic forms: white tin, which is the
common metallic form and gray tin, to which
white tin can transform below 13 C (55 F).
Owing to the 26% increase in volume that ac-
companies the phase change from white to gray
tin, the solid metal disintegrates into a crumbly
mass having essentially no strength. The trans-
formation from white to gray tin is clearly di-
sastrous for the mechanical properties of sol-
dered joints. Tin-pest can affect electrical
systems used in subzero environments, for ex-
ample, in high-ying aircraft and also soldered
containers and conduits of refrigerants.
The transformation does not occur spontane-
ously, but is always preceded by an incubation
period, which may be as much as several years
if the tin is exceptionally pure. The maximum
rate of conversion occurs at about 40 C (40
F). Figure 2.32 shows the rate of transformation
as a function of time at various temperatures. The
incubation period is much shorter if white tin is
mechanically worked at low temperatures or is
inoculated with either gray tin or other elements
and compounds having similar diamondlike crys-
tal structures, such as silicon and ZnSb [Macin-
tosh 1968]. Copper, too, accelerates the process
[Rogers and Fydell 1953], whereas lead at higher
concentrations than the eutectic retards it slightly
[Williams 1956].
Relatively small additions of certain elements
have been shown to suppress the allotropic trans-
formation of tin. The addition of 0.15% anti-
mony will prevent it in tin-base solders. Bismuth
is even more effective and a 0.05% addition is all
that is necessary [Bornemann 1956]. Existing
solder specications accommodate sufficient
quantities of impurity elements to prevent tin-
pest. However, in recent years there has been a
Detrimental effect
Table 2.10 The effect of major alloying
Oxidation additions on the wetting angle of Pb-80Sn
Reduced spreading solder
Reduced spreading
Oxidation Wetting angle, degrees
Reduced spreading Solder/addition Copper Brass 0.15% C steel
High melting range
Dewetting 80Sn 20Pb 26 50 95
High melting range 4Sb 38 43 82
Oxidation 5Bi 22 36 70
5Cd 30 38 102
5In 34 37 98
 Chapter 2: Solders and Their Metallurgy / 77

trend toward high-purity solders in order to im- Tin also improves the spreading character-
prove certain characteristics in wave soldering istics and reduces the hardness. However, to
machines. In some solder specications, the achieve these benets, the concentration of
maximum level of antimony permitted has been tin ideally needs to be several percent (see
reduced from 0.5 to 0.12%, so that tin-pest section 2.1.4 in this chapter).
could occur in soldered assemblies unless other The addition of germanium to gold-silicon
impurities, notably bismuth, remain at appropri- solder also reduces the alloy hardness with
ate levels [EP&P 1992]. consequential benets for joint ductility. The
Metallurgical reactions of solders with the par- softening of the multicomponent alloy arises
ent materials can also have an effect on the prop- as a result of grain rening of the silicon
erties and performance of joints made with ller phase, the grain size being decreased by
metals. The most common examples are modi- roughly half for comparable cooling condi-
cation of melting range and physical properties. tions.
This subject is dealt with further in section 2.3 Similarly, a 0.5% silver addition provides
in this chapter. grain renement of bismuth-tin solder.
Other examples of favorable effects of impu- Silver, gold, and antimony, at moderate con-
rities on different solders include: centrations (<4%), act as strengthening ele-
ments in lead-tin solder.
Small quantities of the transition metals co- The addition of 0.5 to 0.6% Bi to the Au-20Sn
balt and iron (<0.5%) increase the mechani-
solder is said to improve the wettability when
cal strength of Ag-1.7Cu-93.6Sn solder by
the alloy is used for joining gold-plated prod-
about 25% at room temperature. Cobalt re-
ucts [Rapson and Groenewald 1978].
nes the Ag3Sn phase, while iron promotes
renement of primary tin dendrites. The ben-
Although the presence of aluminum is un-
desirable in conventional solders, as it greatly
ets are lost at elevated temperature (>150
exacerbates oxide formation, the same is not
C, or 300 F) owing to rapid solid-state dif-
true of all solders. A case in point is the lead-
fusion [Anderson et al. 2002]. Rare-earth el-
free alloy Sn-8.5Zn-1Ag. Minor additions of
ements have a similar effect, as discussed in
aluminum to this solder, up to a maximum of
Chapter 5, section 5.8.
1 wt%, actually improves wetting of copper.
Zinc improves the spread and ow of Au-2Si In this instance the increased oxidation is
solder, but only at concentrations above 0.5%.
outweighed by modication of the metal-
Zinc has the additional benet of softening
lurgy such that the formation of copper-zinc
the alloy, reducing the hardness from 320 to
intermetallic compounds is suppressed in fa-
128 HV for a 0.75 wt% addition.
vor of copper-tin phases because of the ready
association between aluminum and zinc in
the molten state [Cheng and Lin 2002].
This list is by no means exhaustive, and some
additional examples are given in Chapter 3, sec-
tion 3.3.8.4. In general, most of these effects
have been discovered by accident rather than by
design. Manufacturers supply solders of suitable
purity for most applications, but the process de-
signer must always be aware of the possibility of
unfavorable composition changes to the solder
as a result of alloying with the components being
joined. This topic is examined in the next section.

2.3 Application of
Phase Diagrams to Soldering
Fig. 2.32 Allotropic transformation of white tin into gray tin
The selection of a solder for a particular ap-
as a function of time and temperature. Adapted
from Bornemann [1956] plication is often based exclusively on the melt-
78 / Principles of Soldering
ing point and mechanical properties of the solder
and its ability to wet the parent materials. The
solder is regarded as a uniform layer of metal that
simply bridges the gap between the components
and binds them together. If only life were that
simple! In reality, the formation of the desired
metallic bond between the solder and a compo-
nent requires a degree of alloying. The ensuing
metallurgical reactions usually lead to a hetero-
geneity of phases in the joint. To further com-
plicate matters, kinetic factors tend to accentuate
the development of this nonuniformity. Such in-
homogeneities often determine the quality and
overall characteristics of joints, such as their me-
chanical properties, the ease and extent of solder
spreading, the nature of any llets formed, and
so on.
Metallurgical reactions do not cease once the
joint has been made, but continue to proceed, to
a greater or lesser extent, during the service life
of the assembly. The rate-controlling step for
reaction between two solid metals is the diffu-
sion of atoms between the reacting phases. The
relative position of the product of the reaction
and the reacting phases will be governed largely
by the diffusion coefficients of the participating
metals. For individual metals, it has been estab-
lished empirically that the rate of diffusion R,
increases rapidly with absolute temperature, T,
following an exponential relationship:
R  A expQ kT
where k and A are constants; k is the Boltzmann
constant, and A is an experimentally determined
factor for each combination of reacting phases
that may vary with concentration. Q is the ac-
tivation energy for diffusion which, to a rst
approximation, is proportional to the melting
point, Tm, of the particular metal [Birchenall
1959].
The rate of reaction will therefore be depen-
dent on the homologous temperature dened as
the ratio of T/Tm and will be more pronounced for
low-melting-point solders that see service at or
close to normal ambient temperatures. The re-
actions produce perceptible metallurgical
changes in the constitution and microstructure of
soldered joints [Frear, Jones, and Kinsman 1990,
Chapter 2].
For a proper understanding of metallurgical
reactions between solders and parent materials,
it is essential to have some grasp of the subject
of alloy constitution. The constitution of an
alloy refers to features such as its composition,
melting range, range of phase stability, solubility
limits, and related parameters that can be de-
duced from the phase diagram of the system in
which the alloy appears. In the following sec-
tions, some attention is given to highlighting the
value of phase diagrams and suggesting how this
valuable source of information might be tapped.
Frequently, the appropriate phase diagram for
elucidating specic solder/substrate reactions and
joint microstructures is not available in the lit-
erature. Experimental techniques and a method-
ology for elucidating gaps in phase diagrams are
described in the literature [Humpston and Ja-
cobson 1993, Chapter 3]. Note that all of the
phase diagrams in this book are dened in weight
percentages of the constituent elements as this is
more appropriate to soldering technology than
the atomic percentage scale. Also, the relative
proportion of the elements in intermetallic com-
pounds, such as Cu3Sn, refer to atomic weights.
General equations for converting atomic to
weight percent of constituents in alloys, and vice
versa, are given in Appendix A3.1.
The fundamentals of alloy phase diagrams are
covered in many metallurgical textbooks and will
not be repeated here. Readers without a back-
ground in this eld are referred to the publica-
tions listed in the Preface under the heading Al-
loy Constitution, which provide an excellent
introduction to the subject. Here it will suffice to
state that a phase diagram is a representation of
the thermodynamic stability of phases as a func-
tion of composition with respect to particular
thermodynamic variables such as temperature or,
less commonly, pressure. What is important to
remember is that the information given by the
diagrams relates to essentially equilibrium con-
ditions. The phase diagram tells us about the
ultimate balance of phases within the joint and
those that are likely to be encountered during the
progression toward equilibrium. A joined assem-
bly in which the solder and abutting components
are different materials are never in true compo-
sitional equilibrium, as long as the joint remains
distinct. In most practical contexts, the compo-
sition of a joint will be tending toward equilib-
rium over most of its width, and therefore phase
diagrams are applicable to an assessment of its
constitution. However, at the edges of the joint,
marked compositional gradients will exist, caus-
ing a signicant deviation from equilibrium.
These will be exacerbated by any temperature
gradients that develop during the process cycle
and are manifested as the appearance of different

phases in those regions. Even here, phase dia-
grams can assist in the elucidation of the met-
allurgical reactions and the resulting phases, as
shown in the following section.
Phase diagrams can provide the following
practical information:
The melting temperatures of the virgin sol- the Bulletin of Alloy Phase Diagrams and Jour-
der and of the abutting components
The probable freezing range of the solder
following alloying with the components and
hence the remelt temperature of the joint
Whether the solder remains homogeneous in
the joint after reaction with the components
and, if not homogeneous, the phases that are
likely to be present, or that may form sub-
sequently, with their elemental compositions
and melting temperatures
What a phase diagram is unable to reveal is:
The rate of reactions that might occur be-
tween the solder and the components and
their variation with time and temperature.
This applies both when the solder is molten
and when it is solid during service.
The spatial distribution and morphology of
phases in the joint, although frequently it is
possible to deduce whether intermetallic
phases are likely to form as interfacial layers
or will be dispersed throughout the solidied
solder. This is explained in section 2.3.1 in
this chapter.
The wetting characteristics of a particular sol-
der/parent materials combination, of even
perfectly clean surfaces. In practice, wetting
is likely to be heavily inuenced by the ox-
ides, impurities, and residues that are inevi-
tably present on component and solder sur-
faces, but that are extraneous to the alloy
phase diagram.
Physical properties of joints, in particular the
mechanical and corrosion characteristics.
However, it is often possible to predict the
likely range of certain physical properties by
comparison with other known alloy systems.
The simplest diagrams that are encountered in
a joining context are those relating to binary
alloys where, for example, the solder is a single
metal being used to join components of another
metal. This situation is represented by the use of
pure tin to solder copper pipes.
Although the available literature on phase dia-
grams may appear to be reasonably comprehen-
sive, it is worth bearing in mind that reliable
diagrams exist for, roughly, only 50% of binary
Chapter 2: Solders and Their Metallurgy / 79
combinations, 5% of ternary systems, and 0.5%
of quaternary mixtures. A compendium of au-
thoritative alloy phase diagrams is being pre-
pared under the auspices of the International Pro-
gramme for Alloy Phase Diagram Data (IPAPD)
and the rst volumes have already appeared. This
work is ongoing, and updates are to be found in
nal of Phase Equilibria. On a periodic basis,
these publications include a cumulative index
that lists all phase diagram evaluations published
by members of IPAPD. It is worthwhile con-
sulting the most recent phase diagram available,
because older diagrams can contain signicant
errors or omissions.
High-order alloy systems are naturally more
complex and are less well documented, as noted
earlier. However, for a given joining process only
a very limited portion of the phase diagram is
required, and if one is unavailable it is often
possible to experimentally determine the neces-
sary data (for example, [Humpston and Jacobson
1993, Chapter 3]).
Recently, an exciting breakthrough appears to
have been achieved in a method that is able to
predict, with claimed accuracy exceeding 99%,
whether compound formation will occur for any
binary, ternary, or quaternary system. This will
be of great assistance in reducing the time and
effort to establish the phase relationships in alloy
systems, particularly those of higher order that
are often necessary to encompass soldering and
brazing processes. This work also proved that
materials properties are quantitatively contained
in the elemental property parameters of the con-
stituent elements, so that once additional infor-
mation retrieval methods are automated, the se-
lection of materials for specic applications will
be greatly facilitated, and it is hoped many new
materials with exciting property combinations
will be discovered [Villars et al. 2001].
The value of alloy phase diagrams for under-
standing and optimizing soldering processes can
best be appreciated by describing a few specic
examples in the following sections.
2.3.1 Examples Drawn from
Binary Alloy Systems
Example 1: A Binary Eutectic Composition
Solder Used with Components of One of the
Constituent Metals, with No Intermetallic
Compound Formation. A representative ex-
ample of this type of reaction is a silver-lead
solder used to join components of pure silver.
80 / Principles of Soldering
The silver-lead phase diagram represented in Fig.
2.11 shows that in this binary alloy system there
is minimum melting point, that of the eutectic
composition (Ag-97.6Pb). Alloys with this
unique composition transform between a liquid
(L) and two solid phases (S1 and S2) at a xed
temperature, 304 C (579 F), according to the
reaction:
L S1  S2
For all other compositions, except those of
the pure metals, there is a separation between
the liquidus and solidus temperatures. On either
side of the eutectic the liquidus and solidus
separate, with the alloys melting over a range of
temperatures. At temperatures within the melt-
ing range, the alloy is partly liquid and partly
solid. On cooling below the solidus tempera-
ture, all alloys in this system exist as duplex of
S1 and S2, in direct proportion to the alloy
composition.
It can be seen from the silver-lead phase dia-
gram (Fig. 2.11) that silver is soluble in the Ag-
97Pb solder when molten. Thus, at the joining
process temperature (say, 400 C, or 750 F), the
solder will dissolve silver from the components
until the equilibrium concentration of silver is
attained, determined by the intersection of a line
drawn on the phase diagram at the process tem-
perature with the liquidus curve. Thus, at 400 C
(750 F) the dissolution of silver by the Ag-97Pb
solder changes its composition to approximately
Ag-93Pb. That is, the dissolution of silver in-
creases the liquidus temperature of the solder in
the joint, but not its solidus temperature, because
eutectic transformations are isothermal.
Because soldering cycles are short, the solder
will not normally dissolve more silver. (In theory,
if left at temperature for a sufficiently long time,
the lead would diffuse through the entire volume
of the silver components so that the assembly
had a uniform composition determined by the
total quantities of silver and lead in contact. This
is actually the basis of diffusion-soldering pro-
cesses (referred to in Chapter 5, section 5.9)). At
the commencement of the cooling stage of the
process cycle, the molten solder no longer cor-
responds to the eutectic composition, but is rich
in silver and, in consequence, now possesses a
freezing (i.e., melting) range. On cooling below
the liquidus temperature, the excess silver will
solidify rst, as indicated by the phase diagram.
This precipitation tends to occur preferentially at
the interface between the components and the
solder because this interface is usually slightly
cooler than the volume of the molten solder. Pre-
cipitation continues until the temperature and
composition of the remaining liquid reach the
eutectic point so that nal solidication by the
molten solder results in the formation of a small
volume fraction of nely divided eutectic. The
alloy microstructure will therefore comprise pri-
mary dendrites of silver with the interdendritic
spaces lled with the duplex eutectic mixture.
The primary silver phase will contain a local
concentration gradient as the amount of lead it
incorporates varies with temperature. The silver
phase is then said to be cored.
It is possible to quantitatively follow the
change in composition of the solder as it cools
by application of the lever rule. Referring to the
enlarged and schematic portion of the phase dia-
gram given in Fig. 2.33, at 350 C (662 F) the
weight fraction of the solder that is solid, under
equilibrium conditions, is:
XO
% solid  100
XY
The remainder of the solder will be molten,
that is:
OY
% liquid  100
XY
where X is the composition of the liquid phase,
Y is the composition of the solid phase, and O
is the composition of the alloy.
Fig. 2.33 Application of the lever rule to an enlarged portion
of the silver-lead system diagram (not drawn to
scale)

Alloys of eutectic composition are preferred
as solders on account of the following charac-
teristics:
Superior spreading behavior when molten.
This feature is an immediate consequence of
there being no temperature range over which
the alloys coexist as solid and liquid. Where
a pasty mixture can occur, alloying with the
materials of the components will diminish
the available driving force for spreading while
the partly molten alloy is unable to ow due
to its high viscosity.
Superior mechanical properties, arising
from the interspersed or duplex character of
the eutectic microstructure and the ne
grain size. Grain renement is the only
metallurgical process that enhances both the
strength and ductility of a metal. The su-
perior mechanical properties of the lead-tin
eutectic solder, with its duplex microstruc-
ture, over the adjacent alloy compositions
is shown in Fig. 2.34.
) as a layer that separates the molten
Joining process temperatures can be chosen
to be only slightly above the melting point of
the alloy, precisely because eutectic compo-
sition alloys melt completely at a single tem-
perature.
A reduced risk of disturbing located compo-
nents, which can easily occur when the solder
appears to be solid but is actually in a pasty
state. A rapid liquid-to-solid transformation
on cooling, without an intervening pasty
Fig. 2.34 Tensile strength of cast bars of lead-tin alloys. Op-
timal mechanical properties are coincident with
the eutectic composition (Pb-62Sn). Adapted from Inoue, Kuri-
hara, and Hachino [1986]
Chapter 2: Solders and Their Metallurgy / 81
stage, minimizes the chance of such an in-
terruption. However, this assumes that alloy-
ing of the solder with the component mate-
rials does not greatly shift the composition of
the solder from its eutectic point. A disturbed
joint generally has inferior mechanical prop-
erties, and the llets will acquire a rough
surface with a frosty appearance.
For these reasons, most solders are of eutectic
composition.
Example 2: A Binary Eutectic Composition
Solder Used with Components of One of the
Constituent Metals, with Intermetallic Com-
pound Formation. The silver-tin phase diagram
is given in Fig. 2.9. The eutectic composition
solder (Ag-96.5Sn) is widely used to join silver-
coated components. One of the main reasons for
this choice is that the erosion of the metallization
on such components is low and highly predict-
able. The restricted erosion is a consequence of
the formation of the Ag3Sn intermetallic com-
pound (
solder from the remaining silver, because it has
a melting point above normal soldering tem-
peratures. This interfacial layer restricts further
reaction to a degree determined by temperature
and the duration of the soldering process, as in-
dicated by the data given in Fig. 2.35. Using such
data, it is possible to determine the minimum
thickness of silver that is required for a given
volume of solder. This feature can be used to
advantage in soldering to nonmetals by ensuring
that the application of a silver metallization of
dened thickness will prevent erosion through to
the nonmetallic base material in a prescribed sol-
dering cycle and thereby avoid catastrophic de-
wetting.
While the phase diagram can provide guid-
ance about whether a new phase will form, it
cannot be used to determine the ultimate distri-
bution within the joint as this is greatly inu-
enced by a combination of factors. Another piece
of important information that cannot be ascer-
tained from equilibrium phase diagrams is the
rate of growth of phases. The different rates at
which solders can react with substrate metals is
vividly illustrated by comparing the rate of ero-
sion of gold by molten indium in Fig. 2.2 with
that by molten tin in Fig. 2.36. Supercially, the
phase diagrams of these binary systems are iden-
tical.
Although it would appear from Fig. 2.2 that
the indium-gold reaction is self-limiting, this is
only true in relation to the short timescales of
82 / Principles of Soldering

typical soldering operations (seconds). Over
longer periods of time, the reaction will proceed
to a signicant extent. Even when the solder is
solid, the extent of the reaction can be appre-
ciable as shown in Fig. 2.37, despite the fact that
reaction rate may be one or two orders of mag-
nitude slower than when the solder is molten, the
exact ratio depending on the respective tempera-
tures. This point must be borne in mind when
considering the stability of a joint over the ser-
vice life of the associated assembly. The design
life may be as long as 25 years, which is longer

Fig. 2.35 Erosion of silver by molten tin as a function of reaction time and temperature. Adapted from Evans and Denner [1978],
augmented by authors own data

Fig. 2.36 Erosion of gold by molten tin as a function of reaction time and temperature

than the duration of the joining cycle by a factor
of six orders of magnitude.
Example 3: A Binary Peritectic Solder, Il-
lustrating Problems Associated with Using a
Solder in This Category. The second common
type of phase transformation is the peritectic re-
action where a liquid, L, on cooling, partly so-
lidies to form a solid phase S1 and at the peri-
tectic temperature the remaining liquid reacts
with S1 to form a new solid phase, S2. This may
be written as:
L  S1 S2
Indium-lead solders are frequently employed
in situations where the standard tin-base solders
cannot be used, for example, for soldering at
intermediate temperatures between the available
eutectic composition solders. The indium-lead
phase diagram, which is given in Fig. 2.12, con-
tains two peritectic reactions. These are:
L  Pb   at 171.6 C (340.9 F)
L    In at 158.9 C (318.0 F)
Alloys exhibiting this type of transformation are
generally undesirable as solders because, during
Fig. 2.37 Continued growth of gold-indium intermetallic
phases at the interface between a gold metalliza-
tion and indium-lead solder at elevated temperature but below
the solidus temperature. Adapted from Frear, Jones, and Kinsman
[1990]
Chapter 2: Solders and Their Metallurgy / 83
a peritectic reaction, it is not possible to maintain
equilibrium conditions. This is due to the fact
that diffusion rates in solids are about two orders
of magnitude slower than in liquids, so that a
nonequilibrium microstructure develops consist-
ing of islands of the primary solid phase, S1,
completely surrounded by a rim of the second
solid phase, S2.
A quaternary aluminum alloy microstructure
exhibiting a peritectic transformation is shown in
Fig. 2.38. In such an alloy, liquid that is rich in
the lower-melting-point elements will be re-
tained below the peritectic transformation tem-
perature. Thus, the melting and freezing range of
the alloy is widened and the remelt temperature
cannot be reliably predicted. Furthermore, its mi-
crostructure will be grossly inhomogeneous and
relatively coarse, to the detriment of the me-
chanical properties of joints made with this alloy.
In higher-order alloys, a number of other types
of phase transformation can occur and these are
generally referred to as transition reactions. The
majority of these possess features akin to a peri-
tectic transformation, and they tend to be avoided
when selecting alloy compositions as solders on
the same grounds.
2.3.2 Examples Drawn from
Ternary Alloy Systems
It is rare for a soldered joint to be limited to a
combination of just two elements, forming a bi-
nary alloy system. Usually the solder is an alloy of
at least two metals, while engineering alloy sub-
strates are frequently multicomponent. Com-
Fig. 2.38 An alloy microstructure characteristic of a peri-
tectic transformation. The alloy contains four con-
stituents: aluminum, copper, nickel, and silicon. The primary
phase is totally surrounded by a rim of a second phase as a result
of the peritectic reaction failing to maintain equilibrium condi-
tions during solidication.
84 / Principles of Soldering
monly, intermetallic compounds form between
the constituents. The volume, distribution, and
morphology of these intermetallic phases in a
joint can have a pronounced effect on mechanical
properties, in particular. From the relevant phase
diagram, it is possible to predict whether the in-
termetallic compound will tend to form as a con-
tinuous interfacial layer against the parent mate-
rials or is more likely to be dispersed throughout
the joint. If the intermetallic compound has poor
mechanical properties, then a dispersion of it is
preferred because this not only avoids the source
of weakness represented by the interfacial inter-
metallic phase, but actually works to advantage by
strengthening the solder. Examples of these dif-
ferent situations are described below.
Some care needs to be taken when referring to
the intermetallic phases that form in a ternary
system. An example is provided by binary cop-
per-tin solder wetted on nickel substrates to form
a mixture of copper-tin and nickel-tin interme-
tallic phases at the interface. The apparent con-
fusion in the literature over the composition of
these phases can be explained by reference to the
Cu-Ni-Sn phase diagram that reveals that the
binary compounds Cu6Sn5, Ni3Sn2, and Ni3Sn4
not only have very extensive ternary solubility
(up to 30%), but there exists a continuous solid
solution between Cu3Sn and Ni3Sn. Hence, there
is considerable scope for composition variation
within the stable phase elds, and the convenient
labels adopted from binary alloy practice do not
adequately describe the real situation [Lin, Chen,
and Wang 2002].
A ternary system is most usually represented
by an equilateral triangle, with each of the ver-
tices corresponding to the three elements. A grid
is normally drawn on the triangle to provide a
linear scale of composition. Temperatures are
then represented by a series of isotherms, so that
the position of the liquidus on the diagram is
mapped as a topographical surface viewed in
plan. Phase stability as a function of temperature
is commonly represented by a diagram resem-
bling a binary alloy phase diagram, where either
one of the constituents or the ratio of two con-
stituents is held xed. A single diagram cannot
be used to track the solidication sequence be-
cause the ensuing composition changes can ex-
tend outside the plane of the diagram. For a simi-
lar reason, the lever rule cannot be applied to this
representation in order to calculate the propor-
tions of phases that exist in equilibrium. How-
ever, the lever rule can be used in conjunction
with a series of isothermal sections.
Example 4: Interfacial Compound Forma-
tion between a Eutectic Solder and the Com-
ponent Metals. As an example of interfacial
compound formation, it is well known that lead-
tin solders, when used with copper components,
form several copper-tin intermetallic com-
pounds, predominantly at the copper/solder in-
terface. Despite the many billions of soldered
joints that have been made to copper using lead-
tin solder since antiquity, a complete and cred-
ible ternary phase diagram for this alloy system
only become available in 1994. The binary cop-
per-tin phase diagram, a liquidus projection, and
the key isothermal sections and isopleths rel-
evant to soldering to copper are reproduced in
Fig. 2.39 to 2.42 [Yost, Hosking, and Frear 1993].
The features to note are:
Copper is soluble in molten lead-tin solders.
The dissolution of copper by the solder mar-
ginally depresses the melting point of the
lead-tin eutectic alloy.
Liquid lead-tin eutectic solder that has dis-
solved a small quantity of copper will so-
lidify via a ternary eutectic at 182 C (360
F). A reduction in melting point of a solder
on dissolution of substrate constituents is a
key driving force for spreading. The ternary
eutectic point is actually at 61.75Sn-38.05Pb-
0.2Cu (wt%).
The third constituent in the majority of re-
actions between lead-tin solder and copper,
in addition to the tin and lead phases, is the
intermetallic compound Cu6Sn5. It is only for
very lead-rich solder compositions or when
other molten solder compositions have had
the opportunity to dissolve appreciable con-
centrations of copper that the intermetallic
compound Cu3Sn is present as a primary
phase.
The rate of dissolution of copper in molten
Pb-60Sn solder is initially rapid, as is the erosion
of most other engineering metals and metalliza-
tions by this alloy, as can be seen from Fig. 2.43.
Consequently, despite the short process cycle
times normally associated with soldering prac-
tice, the molten solder will dissolve sufficient
copper to reach the saturation concentration, dic-
tated by the process temperature and the com-
position of the liquidus surface in the Cu-Pb-Sn
phase diagram.
On cooling this ternary solder, it would be
expected from consideration of the phase dia-
gram that copper-tin intermetallic phases will
precipitate and become distributed within the
 Chapter 2: Solders and Their Metallurgy / 85

bulk of joint. If the peak process temperature is cooling will be Cu6Sn5. This has been conrmed
kept below 375 C (705 F), the phase diagram by experiment. At normal solder process cooling
shows that the primary phase that precipitates on rates, the precipitates are typically of the order of

Fig. 2.39 Copper-tin phase diagram.   Cu6Sn5;

 Cu3Sn

Fig. 2.40 Liquidus surface of the Cu-Pb-Sn system. Adapted from Yost, Hosking, and Frear [1993]
86 / Principles of Soldering

Fig. 2.41 Expansion of the solder-rich, copper-poor, region of the Cu-Pb-Sn system. Regions are identied by the primary or rst
phase that solidies on cooling in that region. Adapted from Yost, Hosking, and Frear [1993]

Fig. 2.42 Isothermal section of Cu-Pb-Sn system at 150 C. Adapted from Yost, Hosking, and Frear [1993]

5 to 20 nm (0.2 to 0.8 in.) in size and therefore
can only be resolved using high-resolution im-
aging techniques such as transmission electron
microscopy [Felton et al. 1991]. However, if the
peak process temperature exceeds 375 C (705
F) and remains above this level long enough for
more than 1.6% of copper to dissolve, then the
rst phase to precipitate will be Cu3Sn.
It is also apparent from the ternary phase dia-
gram that copper-tin intermetallic compounds
will form at the interface between the solid cop-
per and molten solder. Owing also to the pre-
vailing concentration gradient, the copper-rich
Cu3Sn phase forms adjacent to the copper com-
ponent with the more tinrich Cu6Sn5 phase be-
tween it and the solder, as shown in Fig. 2.44.
The rate of formation of interfacial copper-tin
intermetallic phases is governed by the rate of
interdiffusion of copper and tin through them and
Fig. 2.43 Rate of dissolution of a range of engineering parent
metals and metallizations in lead-tin solder as a
function of temperature. Adapted from Klein Wassink [1989]
cm 17.5 8.9 28.5
Fig. 2.44 Phases formed by reaction between a lead-tin sol-
der and a copper substrate, following extended
heat treatment
Chapter 2: Solders and Their Metallurgy / 87
is well characterized. A distillation of the data
that are available in the literature is presented in
Fig. 2.45. The Cu3Sn phase layer does not grow
to any signicant thickness during typical in-
dustrial soldering processes because the inward
diffusion of tin through the Cu6Sn5 phase does
not normally reach sufficient concentration
[Dirnfeld and Ramon 1990]. The formation of
continuous layers of intermetallic compounds be-
tween the copper and the molten solder greatly
restricts the further dissolution of copper in the
solder [Felton et al. 1991].
Copper-tin intermetallic compounds do not
possess particularly desirable mechanical or
physical properties in bulk form (see Table
2.11). These phases will continue to increase in
thickness during elevated-temperature service
of the component, albeit much more slowly
than when the solder is molten as can be seen
Fig. 2.45 Growth of copper-tin intermetallic compounds on
a copper substrate wetted by lead-tin eutectic sol-
der as a function of reaction time and temperature
Table 2.11 Mechanical and physical properties
measured for three common intermetallic
compounds, on bulk specimens
The phases are generally hard with low fracture toughness.
However, their thermal expansivities lie between the substrate
and solder, which probably plays a role in decreasing the stress
concentration in joints.
Property Cu6Sn5 Cu3Sn Ni3Sn4
Hardness, HV 378 343 365
Toughness, MPa m1/2 1.4 1.7 1.2
Electrical resistivity,
Thermal conductivity, W/m K 34 70 20
Thermal expansivity, 106/K 16.3 19 13.7
Adapted from Fields, Low, and Lucey [1991]
88 / Principles of Soldering
from Fig. 2.46. The presence of thick layers of
coppertin intermetallic phases is widely re-
ported as being detrimental to the mechanical
properties of the joints, and this has prompted
many studies of the effect of their presence, for
example Dirnfeld and Ramon [1990]. Figure
2.47 shows the relationship between the ten-
sile strength of joints and the thickness of the
copper-tin intermetallic layer as determined in
these studies, and Fig. 2.48 shows the relation-
ship between intermetallic thickness and creep
rupture life [Hongyuan, Ying, and Yiyu 1994].
Notwithstanding this general concern, there are
Fig. 2.46 Growth of copper-tin intermetallic compounds on
a copper substrate in contact with lead-tin solder
for 100 s at different process temperatures
Fig. 2.47 Effect of thickness of copper-tin intermetallic com-
pounds in soldered joints on tensile strength at
room temperature
very few reports of joint failures that can de-
nitely be pinpointed to the copper-tin interme-
tallic phases. This unclear picture is under-
standable because tin forms hard intermetallic
phases, of comparable thickness, by solid-state
diffusion at elevated temperature with virtually
all the metals commonly used in engineering
[Kay and Mackay 1976]. Some relevant data
are reproduced in Fig. 2.49. Not surprisingly
then, joints made to copper testpieces using
lead-tin solder are found not to be signi-
cantly weaker than joints to other common
substrate materialsin terms of their strength,
ductility, and fatigue resistancewhen formed
using conventional process cycles such that the
intermetallic layers remain relatively thin.
Not all the binary lead-tin solders form two
intermetallic phases on reaction with copper. If
Fig. 2.48 Relationship between thickness of copper-tin in-
termetallic and creep rupture life of joints made to
copper components using lead-tin eutectic solder
Fig. 2.49 Growth of tin-base intermetallic phases by solid-
state diffusion at 170 C (340 F) on various sub-
strate materials. Adapted from Kay and Mackay [1976]

the tin level is less than 25%, the phase diagram
in Fig. 2.41 indicates that only the Cu3Sn inter-
metallic will precipitate at the solder/copper in-
terface under near-equilibrium conditions [Gri-
vas et al. 1986].
Example 5: Distributed Compound Forma-
tion between a Eutectic Solder and the Com-
ponent Metals. Another industrially important
alloy system comprises gold, lead, and tin, on
account of the widespread use of gold as a sol-
derable metallization for uxless joining in the
electronics industry.
The liquidus surface of Au-Pb-Sn ternary
phase diagram is given in Fig. 2.50, and a section
from the eutectic Pb-62Sn composition toward
gold is shown in Fig. 2.51. The principal features
of these diagrams are closely akin to those of the
Cu-Pb-Sn system discussed previously, albeit
with the difference that in the gold-base system
at least three embrittling gold-tin intermetallic
compounds are able to form directly from the
liquid. Therefore, it might be expected that, when
lead-tin eutectic solder is used to join solid gold
components, the resulting microstructure of
joints will be similar. However, there are two
important differences.
Fig. 2.50 Liquidus projection of the Au-Pb-Sn ternary system. The rst phase to form on solidication is labeled for each phase eld.
The 4% Au isoconcentration line is marked on the gure as is the tie line between lead-tin eutectic solder and pure gold.
A concentration of 4% gold in the eutectic is taken as a safe limit because AuSn4 cannot form as a primary phase. Adapted from Humpston
and Davies [1984, 1985]; Humpston and Evans [1987]
Chapter 2: Solders and Their Metallurgy / 89
Gold metallizations are seldom more than a
few microns (sub-mils) thick, largely on account
of the cost of this metal, so that there is a limit
to the concentration of gold that will accumulate
in the molten solder. The precise value is deter-
mined by both the volume of solder alloy and the
thickness of the metallization, but is usually well
below the saturation concentration. Also, due to
the high rate of dissolution of gold in molten
lead-tin solders, even thick metallizations will be
completely dissolved during the heating cycle.
On cooling a Au-Pb-Sn ternary alloy formed
by reaction of lead-tin eutectic solder with a gold
metallization, the phases that precipitate are in-
dicated by the phase diagram. Because of their
unfavorable mechanical properties, it is gold-tin
phases that are of interest. If the gold concen-
tration is below 1%, then the gold remains dis-
solved as a solid solution in the lead and tin
phases. Between 1 and 5% Au, solidication ter-
minates by the ternary eutectic reaction so that
the AuSn4 phase present will be well-dispersed
throughout the volume of the solder and nely
divided. Between 5 and 8% Au, AuSn4 precipi-
tates as a secondary phase and is consequently
present as large conglomerates in addition to be-
90 / Principles of Soldering
ing a constituent of the ternary eutectic. Above
8% Au, AuSn4 is the primary phase formed on
solidication and becomes the dominant con-
stituent of the solder microstructure. At even
higher concentrations of gold (above about 13%
Au), the solder constitution changes abruptly
with gold-tin phases being the major compo-
nents of primary, secondary, and tertiary solidi-
cation.
When gold-tin phases form as the primary or
major phase, they embrittle the soldered joints on
account of their intrinsically low fracture tough-
ness and the weak interface between them and
the lead-phase in lead-tin eutectic [Harding and
Pressly 1963]. A maximum gold concentration of
4% is usually taken as a safe working limit in
industry. As can be seen from Fig. 2.50, high-
lead and high-tin content in lead-tin solders can
accommodate slightly higher levels of gold be-
fore gold-tin intermetallic compounds become
dominant (i.e., constitute the primary phase) and
cause joint embrittlement.
Likewise, from the appropriate sections of the
phase diagrams of other tin-base solder alloys in
combination with gold, it is possible to derive
safe working limits for them in a similar manner.
Fig. 2.51 Vertical section through the Au-Pb-Sn ternary system between eutectic lead-tin (Pb-62Sn) composition and gold, with the
8% Au concentration marked by a dashed line. The plan view of this section is marked by a dashed line on the liquidus
projection of the Au-Pb-Sn ternary system shown in Fig. 2.50.
Low concentrations of the AuSn4 phase actually
enhance the mechanical properties of many tin-
containing solders, including lead-tin [Wild
1968]. This strengthening is illustrated in Fig.
2.52 for the Ag-96Sn solder. From a concentra-
tion of about 3.5 to 8% Au, the solder micro-
structure is characterized by a ne dispersion of
the AuSn4 phase, present as a secondary phase.
However, when the level of gold in the solder
rises toward 10%, there is a sudden change in
properties and ingots of the solder become com-
pletely unworkable. This change corresponds to
the appearance of AuSn4 as the primary phase;
that is, AuSn4 is the rst solid phase to form on
solidication and consequently adopts a massive
form. Above 19% Au the primary phase is AuSn2,
and at even higher gold concentrations (greater
than 38%) it is AuSn. The safe working limit for
this alloy system is usually taken to be 8% Au.
The ternary Ag-Au-Sn phase diagram and rel-
evant pseudobinary section between eutectic sil-
ver-tin solder (Ag-96.5Sn) and gold are shown in
Fig. 2.53 and 2.54. It is noteworthy that although
the Ag-96Sn alloy contains more tin than lead-tin
eutectic solder, it is tolerant to approximately
twice the volume fraction of gold before the al-
 Chapter 2: Solders and Their Metallurgy / 91

loy is embrittled by AuSn4. This is thought to be Compounds that are stable over a range of com-
due to a better match between the atomic lattices positions tend to be ductile and have crystal struc-
of Ag3Sn and AuSn4 than between lead and tures exhibiting high symmetry, as do most el-
AuSn4. The critical levels of gold that give rise emental metals. Therefore, they tend to have a
to AuSn4 as the primary phase are listed in Table benign effect on joint properties. An example of
4.3 for a selection of solders. Soldering pro- such a compound is Ag3Sn, which is stable over
cesses that involve gold metallizations are dis-
cussed in Chapter 4, section 4.1.3 and Chapter 3,
section 3.3.8.1.
The mechanical properties of joints containing
intermetallic phases can be inferred from a phase
diagram according to whether they are com-
pounds of exact stoichiometric composition (i.e.,
they are in integral atomic ratios of their constitu-
ents) or exist over a range of compositions. Exact
stoichiometric compounds tend to form when one
of the two elements is strongly metallic in char-
acter and the other signicantly less so, in terms
of the density of free electrons that bind the atoms
of the metal together. AuSn4, FeSn2, and Cu6Sn5
are typical examples of this type of compound.
These compounds tend to be hard and brittle.
Moreover, because their crystal structures are fre-
quently of low symmetrythat is, they deviate
from simple cubic or hexagonal structuresthe
interfaces of these compounds with other phases Fig. 2.53 Liquidus surface of the Ag-Au-Sn system. The rst
phase to form on solidication is labeled for each
tend to be weak. These characteristics are trans- phase eld.
ferred to the joint unless the compounds form as
a ne dispersion within the solder.Therefore, their
occurrence should be minimized or, even better,
avoided wherever possible.

Fig. 2.54 Vertical section through the Ag-Au-Sn ternary sys-

tem between eutectic silver-tin solder (Ag-96.5Sn)
Fig. 2.52 A selection of mechanical properties of Ag-96Sn as and gold. The plan view of this section is marked by the dashed
a function of gold addition line in Fig. 2.53.
92 / Principles of Soldering
the composition range from 13 to 20%Ag at room
temperature and has a hexagonal close-packed
crystal structure. This compound forms as an in-
terfacial layer when silver-tin solder is used to join
silver-coated components in a manner analogous
to the reaction between leadtin eutectic and cop-
per. An example is shown in Fig. 2.55, and the as-
sociated binary phase diagram in Fig. 2.9.The rate
of growth of the Ag3Sn layer decreases exponen-
tially with time at a xed temperature as shown in
Fig. 2.56. The growth of this intermetallic layer
correlates with the measured rate of erosion of sil-
ver by tin (Fig. 2.35). The growth of the Ag3Sn
layer progressively reduces the dissolution of sil-
ver. For this reason, the silver-tin eutectic solder,
used in conjunction with silver-coated substrates,
Fig. 2.55 Micrograph revealing a continuous interfacial layer
of the intermetallic compound Ag3Sn formed on
reaction between tin-silver solder and silver. Magnication: 20
Fig. 2.56 Thickness of the Ag3Sn intermetallic layer formed by reaction between Ag-96Sn solder and silver as a function of reaction
time and temperature. After Evans and Denner [1978], with authors own data
forms the basis of a soldering process that is highly
tolerant to the processing time and temperature,
inasmuch as the risk of totally dissolving a fairly
thin (10 m, or 0.4 mil) silver metallization can
be minimized.
2.3.3 Complexities Presented by
Higher-Order and Nonmetallic
Systems
More often than not, higher-order alloy sys-
tems are encountered in joining, because both the
solder and the parent materials usually each have
a minimum of two constituents. Combinations
involving ve or even larger numbers of ele-
ments are not uncommon.
The denition of the plethora of phases that
can exist in these higher-order systems repre-
sents a daunting task. In order to make the prob-
lem more tractable, a reductive approach is often
employed. This method usually involves parti-
tioning the multicomponent system into a series
of quasi-binary or quasi-ternary alloy systems,
each containing different but xed proportions of
the other components and ascertaining these sec-
tions of the relevant phase diagrams, in turn.
Much care should be exercised in extracting
quantitative information from these partial phase
diagrams because the tie lines, triangles, quad-
rilaterals, and so forth that are used with the lever
rule to determine the relative proportions of

phases present often do not lie in the plane of the
selected sections.
2.4 Depressing the Melting
Point of Solders by Eutectic
Alloying
The number of commercially available sol-
ders is nite, and it is not uncommon to have an
application where it would be desirable either to
extend the lower temperature limit on the use of
a particular family of solders or to identify ller
compositions that melt within a specied range.
An obvious case in point is the search for re-
placements for lead-tin solder that do not contain
lead, but that can be used at comparable pro-
cessing temperatures.
For the reasons elaborated in section 2.3 in this
chapter, eutectic alloys possess several key char-
acteristics that make them a natural choice for
llers, namely superior spreading behavior when
molten, with complete melting occurring at a
single temperature that is lower than, and usually
well below, that of either constituent metal. This
property of instantaneous melting enables join-
ing operations to be carried out at temperatures
only slightly in excess of the solidication tem-
perature, and molten eutectic alloys generally
possess a high degree of uidity. Eutectic alloys
also possess favorable mechanical properties
arising from their well-distributed duplex or
higher-order microstructure.
As shown for the particular example of the
Bi-Pb-Sn solder in Fig. 2.16, a eutectic alloy
formed of three constituent metals (that is, a ter-
nary eutectic) always has a melting point that is
lower than the three constituent binary eutectics.
What then happens to the melting point if further
constituents are added? In particular, can it be
lowered at will and, if not, how might one de-
termine the limits?
Answers to these questions have been pro-
vided by analyzing the pattern of behavior ob-
served in two simple examples, namely the de-
velopment of cadmium solders and gallium alloy
systems intended as substitutes for mercury.
From their general features it is possible to de-
vise certain empirical ground rules of general
applicability that can be applied to other cases,
such as the quest for lead-free solders with a
similar melting point to lead-tin eutectic (183
C, or 361 F). Other examples, pertaining to
Chapter 2: Solders and Their Metallurgy / 93
low-melting-point aluminum brazing alloys, are
given in the planned companion volume Prin-
ciples of Brazing.
2.4.1 Liquid Alloys Based on Gallium
There are requirements for metals that are liq-
uid at normal ground temperatures. Mercury is
the only metallic element that is molten down to
subzero temperatures (freezing at 39 C, or
38 F). It has therefore enjoyed a near mo-
nopoly in thermometry for more than two cen-
turies as well as in other applications employing
liquid metals such as mirrors, high-current
switches, and slip rings. With the growing aware-
ness of health and safety issues in recent years,
there has been a move away from the use of
mercury in these applications, mostly to func-
tional substitutes, rather than toward nding di-
rect replacements. Some direct mercury-free al-
loy substitutes have been developed, based on
gallium eutectics, gallium being the second-
lowest-melting-point metal next to mercury.
Thus, for example, Ga-In-Sn, melting at 12 C
(54 F), and Ga-In-Sn-Zn, melting at 9 C (48
F), have been used in high-current switches
[Walkden, Kowalczyk, and Cooke 1978]. The
progressive reduction of the melting point of gal-
lium alloys, as the number of constituents is in-
creased, is shown in Tables 2.12 and 2.13.
2.4.2 Cadmium-Base Solders
Cadmium forms a series of binary eutectic al-
loys with other common solder ingredients that
melt at temperatures down to 123 C (253 F), as
shown in Table 2.14. The alloying elements se-
lected in this example are indium, tin, and zinc,
which all melt at relatively low temperatures and
form binary eutectics with one another, as well as
with cadmium. In Table 2.14, it can be seen that
ternary alloys of cadmium with these elements
span a melting range from 163 to 93 C (325 to 199
F) and on down to 90 C (194 F) for the qua-
ternary eutectic in the Cd-In-Sn-Zn system.
2.4.3 General Features
The melting point behavior follows the fol-
lowing pattern for the three alloy systems con-
sidered here:
The melting point drops monotonically with
the addition of each successive constituent.
94 / Principles of Soldering

The size of the melting-point depression is
dependent on the specic alloying addi-
tions. Some elements are more effective
than others in lowering the melting point of
the alloy.
The incremental melting-point depression ac-
companying the addition of each new alloy-
ing element that enters into the eutectic re-
action becomes progressively smaller, so that
the melting point tends to an asymptotic mini-
mum.
These features are consistent with elementary
thermodynamic and statistical models, as ex-

Table 2.12 Temperature of eutectiferous phase transformations in which one of the participating
phases is gallium
Melting point, Temperature depression,
Order Alloy system C T

1 Ga 30 0
2 Ga-Ag 25 5
Ga-In 16 14
Ga-Sn 21 9
Ga-Zn 25 4
3 Ga-Ag-In 14 16
Ga-Ag-Sn 21(a) 9
Ga-Ag-Zn 20(a) 10
Ga-In-Sn 12 20
Ga-In-Zn 13(a) 17
Ga-Sn-Zn 17 13
4 Ga-Ag-In-Sn 9 21
Ga-Ag-In-Zn 9 21
Ga-In-Sn-Zn 9 21
5 Ga-Ag-In-Sn-Zn 6(a) 24

(a) Authors own measurements

Table 2.13 Effect of addition of a fourth element to the temperature of the Ga-In-Sn eutectic
Melting point, Temperature depression,
Order Alloy system C T

3 Ga-In-Sn 12 0
4 Ga-In-SnAg 9 3
Ga-In-SnAl 9(a) 3
Ga-In-SnBi 10(a) 2
Ga-In-SnCd 10(a) 2
Ga-In-SnCu 10(a) 2
Ga-In-SnPb 10(a) 2
Ga-In-SnZn 9 3

(a) Authors own measurements

Table 2.14 Temperature of eutectiferous phase transformations in the Cd-In-Sn-Zn quaternary

system
Melting point, Temperature depression,
Order Alloy system C T

1 Cd 321 0
2 Cd-In 123 198
Cd-Sn 177 144
Cd-Zn 266 95
In-Sn 117 204
In-Zn 142 179
Sn-Zn 199 122
3 Cd-In-Sn 93 228
Cd-In-Zn 116 205
Cd-Sn-Zn 163 158
In-Sn-Zn 108 213
4 Cd-In-Sn-Zn 90 231

plained in Appendix A2.2 to this chapter, and are
generic to eutectic alloying.
From a practical point of view, the implica-
tions are clear. When seeking a reduction in the
melting point of a pure metal or of an existing
eutectic alloy, for use in soldering or other ap-
plications, there is a trade-off between keeping
the number of constituents low and judiciously
chosen to optimize the melting-point depression
and increasing the number of components, which
is likely to produce only a relatively small further
reduction in the melting point. Multicomponent
solder are beset by disadvantages, chief among
which is a considerably more complex phase
diagram, if it is indeed known, and often, a harder
and less workable alloy.
On account of the cumulative complications
associated with increasing the number of con-
stituents in the ller, it is advantageous, where
possible, to limit the choice to alloying additions
that are most effective at depressing the melting
point. Generally, these are elements that have
low melting points and very limited solid solu-
bility in the host metal.
The implications from the foregoing discus-
sion are not encouraging with regard to the search
for drop-in replacements for eutectic lead-tin
solders, as explained below. Further details on
lead-free solders are given in Chapter 5, section
5.1.
2.4.4 Implications for
Lead-Free Solders
The wealth of published studies of alternative
solders has generally arrived at a consensus view
that alloys that provide the closest alternatives to
lead-tin eutectic solder, which can meet the ac-
ceptance criteria for printed circuit board (PCB)
manufacture are tin-rich with the minor constitu-
ents being one or more of the following: anti-
mony, bismuth, copper, silver, and zinc [Hamp-
shire 1993]. Ideally, the replacement solder
should have a melting point close to that of lead-
tin, lying in the range between 150 and 200 C
(302 and 392 F), and be of eutectic composition.
Here, the objective is not to minimize the melting
temperature, but to restrict it to within a pre-
scribed temperature window.
Because none of the known binary eutectic
alloys of tin satisfy the above criteria, a study
was made of higher-order systems to establish
whether these might contain suitable eutectic
compositions. The results are given in Table 2.15.
From the experimental data, conclusions can be
Chapter 2: Solders and Their Metallurgy / 95
drawn about the solidication of tin-rich alloys
and are described below.
Those tin-rich multicomponent alloys that so-
lidify eutectiferously do so at a temperature that
is either above 200 C (392 F) or signicantly
below 150 C (302 F). The available range of
suitable alloying additions will either only lower
Table 2.15 Alloying sequences that show that
drop-in replacements for lead-tin solders, based
on tin, are unobtainable
A suitable replacement for lead-tin eutectic solder would need to
meet the following criteria:
Contain only inexpensive and nonvolatile and nonhazardous
constituents
Be a eutectic with a melting point between 150 and 200 C
(302 and 392 F)
1. Candidate constituent binary systems and their eutectic
temperatures are:
Ag-Sn (221 C)
Bi-Sn (139 C)
Cu-Sn (227 C)
In-Sn (120 C)
Sn-Zn (199 C)
Another binary alloy that might provide the basis for a
multicomponent solder is the noneutectic, namely: Sn-6Sb,
melting range: 232250 C (450482 F).
2. Candidate ternary systems for replacing eutectic lead-tin
solder:
Ag-Cu-Sn, ternary eutectic at 217 C [Miller, Anderson,
and Smith, 1994]
Ag-Sb-Sn(a), ternary transition reaction terminating on
the Ag-Sn binary eutectic (221 C)
Ag-Sn-Zn(a), ternary transition reaction terminating on
the Sn-Zn binary eutectic (199 C)
Cu-Sb-Sn, ternary transition reaction terminating on the
Cu-Sn binary eutectic (227 C)
Cu-Sn-Zn(a), ternary transition reaction terminating on
the Sn-Zn binary eutectic (199 C)
All other combinations result in ternary eutectic alloys melting
below 150 C, thus:
Ag-In-Sn(a) (113 C)
Bi-Sn-Zn(a) (130 C)
In-Sn-Zn(a) (108 C)
or, in the case of the Ag-Bi-Sn and Bi-Sb-Sn alloys, terminate
on the Bi-Sn binary eutectic composition (139 C) [see
Kattner and Boettinger, 1994 for the Ag-Bi-Sn system and
Ohtani and Ishida 1994 and Ghosh, Loomans, and Fine, 1994
for the Bi-Sb-Sn system]
3. Quaternary alloys based on the ternary alloys that melt above
150 C
Freezing of these alloys concludes at the binary eutectic
melting point. Thus, for example: Ag-Cu-Sn-Zn(a) undergoes
a ternary transition reaction terminating on the Sn-Zn binary
eutectic.
(a) Authors own measurements
96 / Principles of Soldering
the melting point slightly below that of tin, melt-
ing point 232 C (450 F), as in the case of the
3.5Ag-0.9Cu-95.6Sn ternary eutectic, melting
point 217 C (423 F) [Loomans and Fine 1999],
or will effect a radical reduction in melting point,
for example, alloying with indium and/or bis-
muth [Yoon et al. 1999; Hassam, Dichi, and Leg-
endre 1998]. A eutectic alloy of silver-tin with
gold, of composition 3.6Ag-3.6Au-92.8Sn, has a
melting point of 206 C (403 F), but the cost
premium associated with the gold content would
be unacceptable for the manufacturing industry.
The Sn-9Zn eutectic alloy, melting at 199 C
(390 F), may be judged unsuitable for most
soldering applications because of its propensity
to corrode with the generation of a conductive
white zinc chloride surface lm when subjected
to a chlorine-containing atmosphere.
Solidication of all other tin-rich multicom-
ponent alloys occurs via one or more transition
reactions in a sequence that terminates on one of
the tin binary eutectics, that is, at a temperature
close to that of tin and above 200 C (392 F).
An example is provided by the Bi-Sb-Sn system
[Ohtani and Ishida 1994; Ghosh, Loomans, and
Fine 1994].
It has been shown that some of the multicom-
ponent alloys that are being put forward as drop-
in replacements for lead-tin are handicapped by
the presence of a low-melting-point eutectic frac-
tion that can severely compromise the reliability
of soldered joints if segregation occurs toward
the low-melting-point alloy composition, either
continuously at joint interfaces or along grain
boundaries [Vincent and Humpston 1994]. One
of these alloys is the Ag-20In-77Sn alloy being
offered as a lead-free solder, which substantially
melts at 177 C (351 F). However, a signicant
fraction of this alloy melts at a much lower tem-
perature of 113 C (239 F) [Korhonen and
Kivilahti 1998], associated with a low-melting-
point Ag-In-Sn composition, close to the In-
48Sn binary eutectic [Artaki, Jackson, and Vi-
anco 1994]. Likewise, the In-86Sn-9Zn solder
largely melts between 188 and 198 C (370 and
388 F), but it contains a proportion of the In-
46Sn-2Zn ternary eutectic melting at 108 C (226
F) [McCormack, Jin, and Chen 1994].
It may reasonably be concluded that there are
no higher-order tin-base eutectics, without gold,
that solidify in the target temperature range.
Hence, the multicomponent alloys offer no major
advantages over the binary eutectic alloys of tin,
with regard to melting point. It follows that lead-
free solders that are developed in the future are
likely to be based on the silver-tin binary alloy
system, which has a eutectic temperature of 221
C (430 F). Economic and functional benets
are obtained by substituting some of the silver
for copper and exploiting the Ag-Cu-Sn ternary
alloy system, which has a eutectic temperature at
217 C (423 F), as mentioned above [Miller,
Anderson, and Smith 1994; Vincent and Hump-
ston 1994]. More information on the rationale for
this choice is given in Chapter 5, section 5.1.
Appendix A2.1:
Conversion between
Weight and Atomic
Fraction of Constituents of
Alloys
In an alloy containing N constituents, con-
version from weight to atomic fraction of con-
stituent, n, may be made using:
Pn /An
at.% n   100
N
 / AiP'i
i1
where P is the weight percentage of the con-
stituent denoted by the subscript, A is the atomic
weight of the constituent denoted by the sub-
script, subscript n refers to constituent n, and
subscript i refers to each constituent in turn.
Similarly, in an alloy containing N constitu-
ents, conversion from atomic to weight fraction
of constituent, n, may be made using:
P'n An
wt% n   100
N
 Pi Ai
i1
where P' is the atomic percentage of the con-
stituent denoted by the subscript, A is the atomic
weight of the constituent denoted by the sub-
script, subscript n refers to constituent n, and
subscript i refers to each constituent in turn.
 Chapter 2: Solders and Their Metallurgy / 97

tion, Eq A2.1 can be extended stepwise to mul-

ticomponent alloys, where, for example, the bi-
Appendix A2.2: nary alloy A  B may be considered as the
Theoretical Modeling of solution matrix AB of a pseudobinary alloy
AB  C. Then Eq A2.1 may be rewritten in the
Eutectic Alloying form:

The laws of thermodynamics account for a

lowering of the melting point, when a substance,
B, is added to a pure solvent, A, by an amount
given by Raoults law in the form of the Clausius-
H(2)
m

R ( 1

T (3)
m

1

T (2)
m
) ( )
 ln
x2

x *2
Clapeyron equation. At the liquidus line repre-
senting the equilibrium between a solid solution
of B in A and a liquid solution of B in A, the with the following conditions satised:
Clausius-Clapeyron equation takes the form:

x2 x1
H (2) (1) (3) (2) (1)
m  H m ; T m  T m  T m ; 
H(1)
R
m

( 1

T (2)
m

1

T (1)
m
) ( )
 ln
x1

x *1
x *2 x *1

The relationship x2x *2  x1x *1 implies that

for the third component, C, the attainable liqui-
where x1 is the mole fraction of component A of dus depression T (3)
m
 T (2)
m
is in most cases sig-
the liquidus composition at temperature T (2) , and nicantly smaller than for the second component
m
*
xi is this mole fraction at the limit of solid solu- B, equal to T (2)
m
 T (1)
m
. If this sequential pro-
bility at the same temperature. T (1) is the melting cedure is applied to additional elements, the gen-
m
eral formula for the ith constituent and the cor-
temperature of pure A, and  H m is its latent
(1)
responding liquidus temperature, T im is
heat of fusion. R is the universal gas constant.
obtained:
The latent heat Hm (or enthalpy of fusion) of
a metal is proportional to its melting point. This
is because entropies of fusion (Sm) have similar
values for all metals (at roughly 10 J/K mol),
and to a rst approximation, in the absence of
any phase changes:
H(i1)
m

R
.
( 1

T (i)
m

1

T (i1)
m
) ()
 ln
xil
*
xi1
(Eq A2.2)

Hm  Tm Sm  0 (Eq A2.1)

that is:
Hm  Tm
Equation A2.1 therefore implies that the attain-
able melting-point depression is determined by
the melting point of the addition; the lower the
melting point of the addition, the lower will be
the melting point of the resulting eutectic (as-
suming that one exists).
The effect on melting point of multiple alloy-
ing additions, all with similar melting points, can
be deduced as follows. As a simple approxima-
This expression is a member of a series, of de-
creasing size for increasing value of i. The melt-
ing-point depression is maximized for a large
difference between the concentration of the ma-
trix xi1 and its solid solubility limit x*i1.
Despite the fact that this model considerably
oversimplies reality, it accounts for the two prin-
cipal features in common with the experimental
results on eutectic alloys, namely:
The progressive reduction in melting tem-
perature as the number of alloying constitu-
ents is increased
The asymptotic narrowing of the melting-
point depression of the alloy with the intro-
duction of each additional constituent
98 / Principles of Soldering
Composition-specic thermochemical data on
multicomponent alloys (Hm, xi) are needed in
order to apply Eq A2.2 in calculating the liquidus
temperature depression through progressive al-
loying. This information is mostly unavailable
for ternary and higher-order systems. However,
from knowledge of thermochemical data for pure
metals and binary alloys, it can be inferred that
the respective values will differ widely from one
metal to another, and therefore large variations in
temperature drop are to be expected between
different alloying additions.
The physical picture of this behavior may be
more clearly understood in terms of the entropy
changes accompanying progressive alloying. At
the microscopic level, a material system may be
viewed as an ensemble of atoms or molecules
and, on this basis, entropy provides a measure of
the degree of atomic or molecular disorder in the
system, according to the relationship:
S  k ln
S3,4  R [ln 4  ln 3]
where S is entropy, k is the Boltzmann constant,
and
is the degree of disorder, as measured by
the number of different distributions available to
the atoms or molecules in the system.
In the simplest case, where the volume of the
material is shared by i different species of atom,
representing different constituents of an alloy,
each present in amounts N1, N2, N3, , N i, such
that:
N1  N2  N3  N i  N
then assuming that the different species of atom
are equally interchangeable, the number of ways,
, in which all the atoms may be arranged among
the N available sites is:
N!
 the system, as one might intuitively expect.
N1! N2! N3! Ni!
and
S  k ln
N!
N1! N2! N3! Ni!
As the number of constituents increases, while
keeping the total number of atoms, N, constant,
there is a tendency for the entropy to increase
as the number of constituents in the material is in-
creased, as follows: If Ni 10 for all values of i,
then Stirlings Formula can be applied, whereby:
ln Ni!  Ni ln Ni  Ni
and the change in entropy in increasing the num-
ber of constituents from two (binary system) to
three (ternary system) is:
S2,3  kN [ln 3  ln 2]  R [ln 3  ln 2]
in the particular case where N1  N2  12 N for
the binary system and N1  N2  N3  13 N for
the ternary system.
Then, on increasing the number of constitu-
ents from three to four (quaternary system), the
entropy rises further by an increment:
S3,4  S2,3
where N1  N2  N3  N4  14 N for the qua-
ternary system.
In general:
Si1,i  R [ln i  ln (i1)]
where N1  N2  N3Ni1  Ni  (1/i)N
and i 2.
The pattern is established where each succes-
sive addition increases the entropy of the system
overall, but by progressively smaller amounts, as
shown by the data in Tables 2.12 to 2.14. In other
words, each additional constituent has a rela-
tively smaller effect on the degree of disorder of
Entropy, S, is related to the Gibbs free energy,
G, by the relationship:
()
dG
dT P
 S
Therefore, as the entropy increases, the depres-
sion of the Gibbs free energy of a system as a

function of temperature increases. This in turn
will tend to depress its melting point, although
the actual relationship will be governed by the
specic free energies of the constituents in the
molten and solid states and of the solution that
they form. In general terms, the picture provided
by this elementary expression is consistent with
that furnished by Raoults law and the Clausius-
Clapeyron equation.
At rst sight, this model may appear inappro-
priate for a eutectic alloy. However, the special
case of a eutectic alloy with a low degree of
intersolubility of the pure metal constituents in
the solid approximates reasonably well, insofar
as each phase is tantamount to a pure constituent
and is well dispersed throughout the alloy. This
model therefore serves as a crude, but neverthe-
less graphical illustration of the physical effect of
progressive eutectic alloying.
REFERENCES
Ackroyd, M.L., Mackay, C.A., and Thwaites,
C.J., 1975. Effect of Certain Impurity Ele-
ments on the Wetting Properties of 60% Tin-
40% Lead Solders, Met. Technol., (No. 2), p
7385
Anderson, I.E., et al., 2002. Microstructural
Modications and Properties of Sn-Ag-Cu
Solder Joints Induced by Alloying, J. Elec-
tron. Mater., Vol 31 (No. 11), p 11661174
Anhock, S., et al., 1998. Investigations of
Au-Sn Alloys on Different End Metallisa-
tions, Proc. Conf. 22nd IEEE/CPMT Inter-
national Electronics Manufacturing Technol-
ogy Symposium, 2729 April (Berlin), p 156
165
Artaki, I., Jackson, A.M., and Vianco, P.T.,
1994. Evaluation of Lead-Free Solder Joints
in Electronic Assemblies, J. Electron. Mater.,
Vol 23, p 757764
Beal, R.E., 1984. Minimum Tin Content Sol-
ders for Automotive Radiators, J. Met., (No.
11), p 6368
Becker, G., 1991. Impurities in Solder
Their Impact on Solderability and Corrosion,
Solder. Surf. Mt. Technol., Vol 7 (No. 2), p
2433
Birchenall, C.E., 1959. Physical Metallurgy,
Metallurgy and Metallurgical Engineering
Series, p 200
Bornemann, A., 1956. Tin Disease in Solder
Type Alloys, STP 189, Proc. Symposium on
Chapter 2: Solders and Their Metallurgy / 99
Solder, 1920 June (Atlantic City, NJ),
American Society for Testing Materials, p
129149
Bulwith, R.A. and Mackay, C.A., 1985. Sil-
ver Scavenging Inhibition of Some Silver
Loaded Solders, Weld. J., Vol 64 (No. 3), p
86s90s
Cheng, S. and Lin, K., 2002. The Thermal
Property of Lead-Free Sn-8.05Zn-1Ag-xAl
Solder Alloys and Their Wetting Interaction
with Copper, J. Electron. Mater., Vol 31 (No.
9), p 940945
Crispin, R.M. and Nicholas, M.G., 1984. Re-
active Brazing of Alumina to Metals, Brazing
Soldering, Vol 6 (No. 1), p 3739
Darveaux, R. and Murty, K.L., 1992. Effect
of Deformation Behaviour on Solder Joint
Reliability Prediction, Proc. Conf. Micro-
structures and Mechanical Properties of Ag-
ing Materials, 25 Nov (Chicago), p 399
408
Dirnfeld, S.F. and Ramon, J.J., 1990. Micro-
structural Investigations of Copper-Tin In-
termetallics and the Inuence of Layer Thick-
ness on Shear Strength, Weld. J., Vol 69 (No.
10), p 373s377s
Dogra, K.S., 1985. A Bismuth Tin Alloy for
Hermetic Seals, Brazing Soldering, Vol 9
(No. 3), p 2830
DuPont Electronics, 1998. Low Tempera-
ture Braze System 5087D Braze Alloy Com-
position, Technical Information Sheet, Sept
Edwards, L.K., Nixon, W.A., and Lakes, R.S.,
2000. Development of a Low-Cycle Fatigue
Life Curve for 80In15Pb5Ag, J. Electron.
Mater., Vol 29 (No. 9), p 10841089
EP&P 1992. Reduction in Antimony Puts Sol-
der Joints at Risk, Electron Packag. Prod.,
Vol 32 (No. 2), p 20
Evans, D.S. and Denner, S.G., 1978. An Ap-
paratus for the Determination of Solid/Liquid
Metal Interactions under Controlled Condi-
tions, Pract. Metallogr., Vol 15, p 486493
Felton, L.E., et al., 1991. n-Cu6Sn5 Precipi-
tates in Cu/Pb-Sn Solder Joints, Scr. Metall.
Mater., Vol 25 (No. 10), p 23292333
Fields, R.J., Low III, S.R., and Lucey Jr.,
G.K., 1991. Physical and Mechanical Prop-
erties of Intermetallic Compounds Com-
monly Found in Solder Joints, Proc. TMS
Fall Meeting, 2024 Oct (Cincinnati, OH), p
165173
Finch, N.J., 1985. Soft Soldering Aluminium
and Its Alloys, Brazing Soldering, Vol 9 (No.
59), p 4649
100 / Principles of Soldering
Frear, D.R., Jones, W.B., and Kinsman, K.R.,
1990. Solder MechanicsA State of the Art
Assessment, TMS
Freer Goldstein, J.L. and Morris Jr., J.W.,
1994. Microstructural Development of Eu-
tectic Bi-Sn and Eutectic In-Sn During High
Temperature Deformation, J. Electron.
Mater., Vol 23 (No. 5), p 477485
Ghosh, G., Loomans, M., and Fine, M.E.,
1994. An Investigation of Phase Equilibria of
the Bi-Sb-Sn System, J. Electron. Mater., Vol
23, p 619623
Grivas D., et al., 1986. The Formation of
Cu(3)Sn Intermetallic on the Reaction of Cu
with 95Pb-5Sn Solder, J. Electron. Mater.,
Vol 15 (No. 6), p 355359
Hampshire, W.B., 1993. The Search for Lead-
Free Solders, Solder. Surf. Mt. Technol., Vol
14, p 4952
Harada, M. and Satoh, R., 1990. Mechanical
Characteristics of 96.5Sn/3.5Ag Solder in
Microbonding, IEEE Trans. Components,
Hybrids Manuf. Technol., Vol 13 (No. 4), p
736742
Harding, W.B. and Pressly, H.B., 1963. Sol-
dering to Gold Plating, 50th Annual Conf. of
the American Electroplaters Society, p 90
106
Hare, E.W., Corwin, R., and Reimer, E.K.,
1985. Structure and Property Changes dur-
ing Room Temperature Aging of Bismuth
Solders, ASM Second Electronic Packaging:
Materials and Processes Conf., 2931 Oct
(St. Paul, MN), p 109115
Harrison, K.T. and Knights, C.F., 1984. De-
velopment of Zinc Based Solders, Brazing
Soldering, Vol 6 (No. 1), p 59
Hassam, S., Dichi, E., and Legendre, B.,
1998. Experimental Equilibrium Phase Dia-
gram of the Ag-Bi-Sn System, J. Alloy.
Compd., Vol 268, p 199206
Hirakawa, T., Tanahashi, S., and Terasawa,
M., 1995. Preattached Chip Capacitors for
Ceramic Packages, Proc. Conf., International
Symposium on Microelectronics, 2426 Oct
(Los Angeles, CA), p 517
Hongyuan, F., Ying, Z., and Yiyu, Q., 1994.
An Experimental Study on Creep Properties
of Solder Joints, Pre-assembly Symposium
47th Annual Assembly of IIW, 12 Sept
(Dalian, Peoples Republic of China), p 342
347
Hosking, F.M., Stephens, J.J., and Rejent,
J.A., 1999. Intermediate Temperature Join-
ing of Dissimilar Metals, Weld. J., April, p
127s136s
Humpston, G. and Davies, B.L., 1984. Ther-
mal Analysis of the AuSn-Pb Quasibinary
Section, Met. Sci., Vol 18 (No. 6), p 329
331
Humpston, G. and Davies, B.L., 1985. Con-
stitution of the AuSnPbSn Partial Ternary
System, Mater. Sci. Technol., Vol 1 (No. 6),
p 433441
Humpston, G. and Evans, D.S., 1987. Con-
stitution of the Au-AuSn-Pb Partial Ternary
System, Mater. Sci. Technol., Vol 3 (No. 8),
p 621627
Humpston, G., et al., 2001. Compliant and
Hermetic Solder Seal, U.S. Patent
10,020,018, Oct
Humpston, G. and Jacobson, D.M., 1993.
Principles of Soldering and Brazing, p 9195
Humpston, G. and Jacobson, D.M., 1995. Do
18 Carat Gold Solders Exist?, Gold Bull., Vol
27, p 110116
Indium Corporation of America, 1994. Re-
search Solder Kits brochure
Inoue, H., Kurihara, Y., and Hachino, H.,
1986. Pb-Sn Solder for Die Bonding of Sili-
con Chips, IEEE Trans. Components, Hy-
brids Manuf. Technol., Vol 9 (No. 2), p 190
194
Irving, B., 1992. Host of New Lead-Free Sol-
ders Introduced, Weld. J., Vol 71 (No. 10), p
4749
Jacobson, D.M. and Humpston, G., 1989.
Gold Coatings for Fluxless Soldering, Gold
Bull., Vol 22 (No. 2), p 918
Jacobson, D.M. and Sangha, S.P.S., 1997.
Novel Low Expansion Packages for Elec-
tronics, GEC J. Technol., Vol 14, p 4855
Johnson, A.A. and Johnson D.N., 1983. The
Room Temperature Dissociation of Au3Si in
Hypoeutectic Au-Si Alloys, Mater. Sci. Eng.,
Vol 61, p 231235
Johnson, D.N. and Johnson, A.A., 1984.
Surface Cracking in Gold-Silicon Alloys,
Solid State Electron., Vol 27 (No. 12), p
11071109
Kattner, U.R. and Boettinger, W.J., 1994. On
the Sn-Bi-Ag Ternary Phase Diagram, J.
Electron. Mater., Vol 23, p 603610
Kay, P.J. and Mackay, C.A., 1976. The
Growth of Intermetallic Compounds on Com-
mon Base Materials Coated with Tin and Tin-
Lead Alloys, Trans. Inst. Met. Finish., Vol 54,
p 6874

Kennedy, G.C. and Newton, R.C., 1963. Sol-
ids under Pressure, McGraw-Hill
Korhonen, T.M. and Kivilahti, J.K., 1998.
Thermodynamics of the Sn-In-Ag System,
J. Electron. Mater., Vol 27 (No. 3), p 149
158
Lin, C., Chen, S., and Wang, C., 2002. Phase
Equilibria and Solidication Properties of Sn-
Cu-Ni Alloys, J. Electron. Mater., Vol 31 (No.
9), p 907915
Loomans, M.E. and Fine, M.E., 1999. Tin-
Silver-Copper Eutectic Temperature and
Composition, Metall. Mater. Trans. A, Vol 31
(No. 4), p 1551162
MacIntosh, R.M., 1968. Tin in Cold Service,
Tin Uses, Vol 72, p 710
Mackay, C.A. and Levine, S.W., 1986. Sol-
der Sealing Semiconductor Packages, IEEE
Trans. Components, Hybrids Manuf. Tech-
nol., Vol 9 (No. 2), p 195201
Mattern, N., 1989. Dynamical X-Ray Dif-
fraction Study on the Phase Transformation
in Rapidly Quenched Au71Sn29 Alloy,
Proc. Conf. Advanced Methods of X-Ray
and Neutron Structural Analysis of Materi-
als, p 7376
McCormack, M., et al. 1997. Signicantly
Improved Mechanical Properties of Bi-Sn
Solder Alloys by Ag-Doping, J. Electron.
Mater., Vol 26 (No. 8), p 954958
McCormack, M., Jin, S., and Chen, H.S.,
1994. New Lead-Free Sn-Zn-In Solder Al-
loys, J. Electron. Mater., Vol 23, p 687690
Miller, C.M., Anderson, I.E., and Smith,
J.F., 1994. A Viable Tin-Lead Solder Sub-
stitute: Sn-Ag-Cu, J. Electron. Mater., Vol
23, p 595601
Noguchi, H., 1999. Solder Paste, Japanese
patent W09901251
Ohtani, H. and Ishida, K., 1994. A Thermo-
dynamic Study of the Phase Equilibria in the
Bi-Sn-Sb System, J. Electron. Mater., Vol 23,
p 747748
Olsen, D.R. and Spanjer, K.G., 1981. Im-
proved Cost Effectiveness and Product Reli-
ability through Solder Alloy Development,
Solid State Technol., (No. 9), p 121126
Park, J., et al., 2002. Reaction Characteristics
of the Au-Sn Solder with Under-Bump Met-
allurgy Layers in Optoelectronic Packages, J.
Electron. Mater., Vol 31 (No. 11), p 1175
1180
Petzow, G. and Effenberg, G., Ed., 1988. Ter-
nary Alloys, Vol 2, VCH
Chapter 2: Solders and Their Metallurgy / 101
Pinamaneni, S. and Solomon, D.A., 1986.
Rapidly Solidied Soft Solder Die-Attach
Technology, IEEE Trans. Components, Hy-
brids Manuf. Technol., Vol 9 (No. 4), p
416422
Prince, A., Raynor, G.V., and Evans, D.S.,
1990. Phase Diagrams of Ternary Gold Al-
loys, Institute of Metals, London
Quan, L., et al., 1987. Tensile Behaviour of
Pb-Sn Solder/Cu Joints, J. Electron. Mater.,
Vol 16 (No. 3), p 203208
Rackham, H., 1952. Natural History, Vol 10,
Cambridge; Translation of Pliny, Historia
Naturalis, Vol 34 (No. 161), p 242
Raman, K.S., 1977. Inuence of Metallic Ad-
ditions on the Spreading Characteristics of
Soft Tin-Lead Solders, Met. Min. Rev., Vol 16
(No. 6), p 120
Rapson, W.S. and Groenewald, T., 1978. Gold
Useage, Academic Press, London, p 80145
Rogers, R.R. and Fydell, J.F., 1953. Factors
Affecting the Transformation to Gray Tin at
Low Temperatures, J. Electrochem. Soc., Vol
100, p 383387
Rubin, W., 1982. Some Recent Advances in
Flux Technology, Brazing Soldering, Vol 2
(No. 1), p 2428
Schmitt-Thomas, Kh.G. and Becker, R.,
1988. Impurities in Solder Baths, Brazing
Soldering, Vol 15 (No. 3), p 4347
Schwartz, M., 2003. Brazing, 2nd ed., ASM
International
Shimizu, K., et al., 1995. Solder Joint Reli-
ability of Indium-Alloy Interconnection, J.
Electron. Mater., Vol 24 (No. 1), p 3945
Shimizu, T., et al., 1999. Zn-Al-Mg-Ga Al-
loys as Pb-Free Solder for Die Attach Use, J.
Electron. Mater., Vol 28 (No. 11), p 1172
1175
Tomlinson, W.J. and Bryan, N.J., 1986. The
Strength of Brass/Sn-Pb-Sb Solder Joints
Containing 010% Antimony, J. Mater. Sci.,
Vol 21, p 103109
Tomlinson, W.J. and Fullylove, A., 1992.
Strength of Tin-Based Solder Joints, J. Mater.
Sci., Vol 27, p 57775782
Vaynman, S. and Fine, M.E., 2000. Flux De-
velopment for Lead-Free Solders Containing
Zinc, J. Electron. Mater., Vol 29 (No. 10), p
11601163
Villars, P., et al., 2001. Binary, Ternary and
Quaternary Compound Former/Nonformer
Prediction via Mendeleev Number, J. Alloy.
Compd., Vol 317318, p 2638
102 / Principles of Soldering
Vincent, J.H. and Humpston, G., 1994. Lead-
Free Solders for Electronic Assembly, GEC
J. Res., Vol 11 (No. 2), p 7689
Wada, O. and Kumai, T., 1991. Preferential
Reaction and Stability of the AuSn/Pt Sys-
tem: Metallisation Structure for Flip-Chip,
Appl. Phys. Lett., Vol 58 (No. 9), p 908910
Walkden, A.J., Kowalczyk, M.R., and Cooke,
R.I., 1978. Liquid-Metal Switching of Ultra-
High Direct Currents, GEC J. Sci. Technol.,
Vol 45 (No. 2), p 6370
Wild, R.N., 1968. Effects of Gold on the
Properties of Solders, Report Number 67-
825-2157, IBM Federal Systems Division,
Owego
Williams, W.L., 1956. Gray Tin Formation in
Soldered Joints Stored at Low Temperatures,
STP 189, Proc. Symp. on Solder, 1920 June
(Atlantic City, NJ), American Society for
Testing Materials, p 149157
Xian, A. and Si, Z., 1991. Wetting of Tin-
Based Active Solder on Sialon Ceramic, J.
Mater. Sci. Lett., Vol 10 (No. 22), p 1315
1317
Yoon, S.W., et al. 1999. Investigation of the
Phase Equilibria in the Sn-Bi-In Alloy Sys-
tem, Metall. Mater. Trans. A, Vol 30A, p
15031515
Yost, F.G., Ganyard, F.P., and Karnowsky,
M.M., 1976. Layer Growth in Au-Pb/In Sol-
der Joints, Metall. Trans., Vol 7A (No. 8), p
11411148
Yost, F.G., Hosking, F.M., and Frear, D.R.,
1993. The Mechanics of Solder Alloy Wetting
and Spreading, Van Nostrand Reinhold, p
130134
Principles of Soldering
Giles Humpston, David M. Jacobson, p103-143
DOI:10.1361/prso2004p103
CHAPTER 3
The Joining Environment
WHEN CONSIDERING the metallurgical as-
pects of soldering in Chapter 2, it is assumed that
components and the ller were perfectly clean
and remained so throughout the process cycle,
enabling the constituents to freely interact so that
the ller metal wets and spreads over the com-
ponent surfaces. However, this situation repre-
sents the ideal case because oxides and other
nonmetallic species are usually present on sur-
faces that have been exposed to ambient atmo-
spheres and these will interfere with or inhibit
wetting and alloying. Any oxygen or moisture
present in the joining environment will further
exacerbate this effect, particularly as the kinetics
of oxidation reactions are highly temperature de-
pendent. Thus, the nature and quality of joints
depend not only on alloying reactions, but also
on the processing environmentin particular on
whether the surroundings are oxidizing, reduc-
ing, or neutral. The term surroundings refers to
both the gas atmosphere itself and any chemi-
cals, such as uxes, that are in the vicinity of the
workpiece. These aspects are considered in sec-
tions 5.2 and 5.3, Chapter 5.
Materials used in joining, whether solders,
uxes, or atmospheres, are becoming increas-
ingly subjected to restrictions on the grounds of
health, safety, and pollution concerns. These
regulations can limit the choice of materials and
processes that are deemed acceptable for indus-
trial use. This issue is a reoccurring theme and
is particularly addressed in Chapter 1, section
1.3.3 and Chapter 3, section 3.2 in the context of
soldering uxes and Chapter 5, section 5.1, which
covers lead-free solders.
Most nonmetallic materials are not wetted by
most conventional solders, even when these have
clean surfaces. Where wetting does occur, the
contact angle between the molten solder and the
parent material is often high, and thus the solder
does not spread over the component surfaces.
Copyright 2004 ASM International
All rights reserved.
www.asminternational.org
This situation cannot be remedied with the help
of chemical uxes because these are unable to
change the physical properties of the intrinsic
materials that govern the wetting characteristics,
as explained in Chapter 1.
Wetting and spreading of a solder on nonmet-
als can be induced by incorporating within the
solder highly active elements, such as titanium,
that react chemically with the base materials to
form interfacial compounds that the solder can
wet. Although the manner in which reactive ll-
ers promote wetting is normally different from
that of chemical uxes, they can also be used to
promote wetting of oxidized metal surfaces and
thereby provide a uxing action. Owing to the
fact that the active constituent of the solder can
reduce the oxides of less refractory metals, it will
remove this surface lm, enabling wetting to
proceed in a conventional manner through al-
loying. Reactive solders are described in Chapter
4, section 4.1.2.2. Active brazes, which are much
more common, are discussed separately in the
planned companion volume Principles of Braz-
ing; these are commercially available and widely
used.
3.1 Joining Atmospheres
There are many types of assembly that de-
mand soldering under a protective atmosphere.
These include assemblies intended for service in
a vacuum environment, which must be free from
volatile contaminants and parent metal compo-
nents that are disgured by oxide scale. The cat-
egories of joining atmospheres that are available
and their interrelationships are shown in Fig. 3.1.
Generally, uxes are needed only when carrying
out the joining operation in air or other oxidizing
environments.
104 / Principles of Soldering
Two distinct types of atmosphere are used for
soldering, namely:
Chemically inert gas atmospheres (e.g., ar-
gon, nitrogen, helium, vacuum). These func-
tion by excluding oxygen and other gaseous
elements that might react with the compo-
nents to form surface lms and inhibit ow-
ing of and wetting by the solder.
Chemically active atmospheres, both gases
and uxes, which are designed to react with
surface lms present on the components and/
or the ller metal during the joining cycle and
remove them in the process. These atmo-
spheres may either decompose surface lms
(as does hydrogen when acting on certain
oxide or sulde layers, for example) or react
with the lms to produce compounds that can
be displaced by the molten ller metal. Tra-
ditional rosin uxes predominantly function
in this manner.
Controlled gas atmospheres require a conn-
ing vessel, and this invariably means a furnace
of some type. Furnace joining also offers other
advantages:
The process may be easily automated for ei-
ther batch or continuous production, because
the heating conditions can be accurately con-
trolled and reproduced without the need for
much operator skill.
Furnace joining offers uniform heating of the
components of almost any geometry and is
Fig. 3.1 Interrelationship of joining atmospheres
suitable for parts that are likely to distort if
heated locally.
The atmospheric protection afforded leads to
economies on the use of ux and on nishing
operations, such as cleaning and the removal
of ux residues.
Against this must be considered the following
potential disadvantages:
Capital costs of the equipment, including the
associated gas atmosphere handling or
vacuum system, may be signicant in rela-
tion to the processing costs.
The entire assembly is heated during the pro-
cess cycle, which can result in a loss of me-
chanical properties, even to components re-
moved from the joint area.
The range of permissible parent materials and
solders tends to be restricted to elements and
chemicals of low volatility to avoid contami-
nation of the furnace. For a similar reason,
most uxes are undesirable.
Certain metals are embrittled on heating in the
presence of standard gas atmospheres (oxygen,
nitrogen, hydrogen, and carbon-containing
gases) and must therefore be joined in a vacuum
furnace. These are principally the refractory met-
als beryllium, molybdenum, niobium (colum-
bium), tantalum, titanium, vanadium, and zirco-
nium, but really only at brazing temperatures
above 750 C (1380 F), which are the same as
those required for using most activated solders.
 Chapter 3: The Joining Environment / 105

Compound semiconductors (e.g., gallium ars- dG  VdP  SdT

enide) can have their electrical functionality poi-
soned by hydrogen, while some brasses are in- V is the volume and S is the entropy, which is
tolerant of ammonia. Thus, the requirements of explained further in Appendix A3.1 Chemical
each component in an assembly must be indi- reactions, such as oxidation and reduction, which
vidually assessed and the atmosphere chosen to are reversible, can take place at constant pressure
suit. and temperature, so that the Gibbs free energy of
the material system does not then change in the
course of the reaction. Table 3.1 shows the Gibbs
3.1.1 Atmospheres and free energy of formation of oxides for a selection
Reduction of Oxide Films of metals at room temperature. This formation
energy is sometimes referred to reciprocally as
A principal process requirement for successful the dissociation potential of the oxide. The least
soldering is to ensure that the joint surfaces are stable metal oxides are those of the noble metals,
free from oxides and other lms that can inhibit gold, silver, and members of the platinum group.
wetting by the molten solder and the formation These metals are therefore the most readily sol-
of strong metallic bonds. The ability to remove dered, while the refractory metals and the light
a layer of oxide from a given metal depends on metalsnotably aluminum, beryllium, and mag-
the ease of either physically detaching the lm nesiumhave particularly stable oxides so that
from the underlying metal or of chemically sepa- these metals are the most difficult to join. It is
rating the oxygen ions from the metallic ions precisely because gold will not form a stable
present in the oxide, that is, the strength of the oxide in air that it is widely used as a surface
relevant molecular bonds. Chemical reduction of coating for uxless joining processes (see Chap-
metal oxide by atmospheres is considered rst. ter 3, section 3.3.8.1).
Chemical thermodynamics can be used to de- Other factors need to be considered in con-
termine the propensity for a metal to spontane- nection with oxide reduction. In particular, many
ously oxidize or, conversely for an oxide to dis- metals form different oxides of varying stabil-
associate. An understanding of oxide reduction ityfor example, cuprous oxide (Cu2O) and cu-
by gas atmospheres is very relevant to brazing, pric oxide (CuO). Furthermore, oxides formed
and for that reason an introduction to the subject on alloy surfaces are not generally pure metal
is provided in the planned companion volume
Principles of Brazing. At typical soldering tem-
peratures there is insufficient driving force avail- Table 3.1 Comparative values for free energies
able to reduce or prevent oxidation of almost all of formation of metal oxides of common solder
constituents and selected metals at room
metals, save those classied as noble and under
temperature (25 C or 77 F)
limited circumstances, possibly also extending
to silver and copper. For this reason, only a very The more negative the value, the more stable the oxide.
supercial treatment of chemical thermodynam- Free energy of
ics is provided here, sufficient only to grasp the formation at
25 C (77 F),
most important fact that gaseous atmospheres,
Element Common oxide kJ/mol (rounded values)
including pure hydrogen, are generally inert to-
ward oxides of solders and common substrate Gold Au2O3 50
Silver Ag2O 10
materials at normal soldering temperatures. Copper CuO 130
There are, of course, exceptions, and some of Cu2O 150
Bismuth BiO 170
these are elaborated on later in this chapter. Bi2O3 460
A measure of the strength of a metal-to- Lead PbO 190
PbO2 210
oxygen chemical bond is given by the change in Pb3O4 570
the Gibbs free energy that occurs when that metal Tin SnO 260
reacts to form the oxide, as detailed in Appendix SnO2 490
Zinc ZnO 300
A3.1. Here, it is noted that the Gibbs free energy, Antimony Sb2O3 580
G, is an important thermodynamic function in Sb2O4 740
Sb2O5 780
chemistry because incremental changes in its Indium In2O3 620
value only involve incremental changes in pres- Chromium Cr2O3 700
sure, P, and temperature, T, for reversible reac- Titanium TiO2 840
Aluminum Al2O3 1580
tions:
106 / Principles of Soldering

oxides but rather compound or other forms of
mixed oxide. Often these are of nonuniform com-
position and structure, adding further complexity
to the subject. This is particularly true of solders
that are almost inevitably multicomponent al-
loys. In its present state of development, chemi-
cal thermodynamics is not able to predict accu-
rately conditions under which dissociation of
oxides will occur, but can only provide a semi-
quantitative indication, particularly when the ki-
netics of reaction are taken into account. Owing
to such complexities, the thermodynamic prin-
ciples for analyzing oxide reduction are consid-
ered only for pure metals.
3.1.2 Thermodynamic Aspects of
Oxide Reduction
All chemical reactions are reversible, includ-
ing oxidation reactions. In general, the oxidation
of any metal can be described by an equation of
the form:
m librium partial pressure of oxygen in the metal
nM  O2 MnOm
2
The reaction will proceed spontaneously in ei-
ther directionnamely, oxidation of the metal,
or, conversely, reduction of the oxide, if it is
energetically favorable to do so. A condensed
treatment of relevant thermodynamic functions
and their relationships, which has by necessity
required a degree of oversimplication, is given
in Appendix A3.1 at the end of this chapter. A
more rigorous treatment is given in standard text-
books on thermodynamics, such as those listed
in the Selected References given in the Preface.
The free-energy change (G) for oxidation re-
actions involving a series of metals can be charted
on a diagram as a function of temperature, as
shown in Fig. 3.2. This representation is known as
an Ellingham diagram or a Richardson-Jeffes dia-
gram. The diagram can be used to determine
whether, in principle, an atmosphere is capable of
reducing surface oxides, although it does not pro-
vide any indication of the kinetics of the reactions.
The use of the Ellingham diagram in soldering
practice is described in the next section.
At any given temperature, the smaller the equi-
oxide, the stronger the bond between the oxide

Fig. 3.2 Simplied Ellingham diagram showing the free-energy change for oxidation of several metals. Oxide stability is reduced
by elevated temperature and decreased oxygen partial pressure. Each dashed line corresponds to the Gibbs free-energy
change as a function of temperature, relating to a particular oxygen partial pressure. mpt, melting point

and the parent metal, that is, the greater is the
stability of the oxide. The partial pressure of a
gas in an atmosphere is dened in Chapter 1,
section 1.3.2.5. Thus, the tendency for the oxide
to decompose will be greater the lower the oxy-
gen content of the atmosphere and the higher the
temperature. The further down the diagram a
particular metal-oxygen reaction curve lies, the
more inherently stable is the oxide, and it is cor-
respondingly more difficult to reduce; in other
words, higher temperatures and atmospheres of
lower oxygen content are required to effect re-
duction.
3.1.3 Practical Application of the
Ellingham Diagram
3.1.3.1 Soldering in Inert
Atmospheres and Vacuum
For many metals, heating alone in air is not ad-
equate to reduce the oxide, because the compo-
nents are degraded or even melt before the critical
temperature, Tc, is reached at which the oxide will
spontaneously decompose. Moreover, the rate of
oxidation roughly doubles with each 25 C (45 F)
rise in temperature. Thus, stable oxides become
progressively thicker and more tenacious, and
consequently more difficult to remove, over the
time interval that the component is being heated
to the critical temperature. Excessive oxidation
can damage component surfaces, particularly if
the lm spalls off locally, because the rate of oxi-
dation will be nonuniform over the surface, pro-
ducing an unsightly nish. For these reasons,
where uxes are not employed, it is usual practice
to heat the components in a protective atmosphere
or vacuum, which will both protect the surfaces
from further oxidation and reduce the partial oxy-
gen pressure, and hence the critical temperature.
The conditions of temperature and oxygen par-
tial pressure required to spontaneously reduce a
metal oxide can be deduced from the Ellingham
diagram. Reduction will occur when the free-en-
ergy curve for metal-oxide formation lies above
the oxygen partial pressure curve at the tempera-
ture of interest; that is, the oxygen pressure in the
atmosphere is less than that which will cause the
metal under consideration to oxidize. Thus, the
critical temperature for the reduction of PdO de-
creases from 920 C (1688 F) in pure oxygen at
atmospheric pressure to 380 C (715 F) if the
oxygen partial pressure is decreased to 1010 atm
(102 mPa). It can be seen from the more detailed
Chapter 3: The Joining Environment / 107
Ellingham diagram reproduced in Fig. 3.3, that
oxide reduction in vacuum is practicable only for
palladium and silver at normal soldering tempera-
tures (<450 C, or 840 F). The minimum partial
oxygen pressure that can be achieved using high-
quality industrial equipment is of the order of
1010 atm (102 mPa). Note that it is convenient
to use the atmosphere as the unit of pressure in
thermodynamic calculations, and this convention
is also applied to Ellingham diagrams.
For metals having oxidizing reaction curves
that are located below the 1010 atm (102 mPa)
oxygen partial pressure curve (that is, a line join-
ing the point O on the T  273 C axis on the left,
to the 1010 atm value on the partial oxygen pres-
sure (PO2 ) scale on the right of the Ellingham dia-
gram, as shown in Fig. 3.2), it will be energetically
favorable for the metal to oxidize by reaction with
the residual oxygen and any water vapor present
in the furnace atmosphere. This includes most
common solder constituents, including tin and in-
dium. Thus, industrial-quality vacuum and inert
atmospheres are incapable of preventing degra-
dation of solders during normal heating cycles.
Obviously, an atmosphere that is largely free of
oxygen and water vapor will greatly slow further
oxidation, but cannot prevent or reverse it.
As mentioned previously, care must be taken to
select an atmosphere that is inert toward all of the
metals in the assembly being joined. Vacuum can
degrade certain materials, notably brass, even at
soldering temperatures, due to the loss of zinc
through volatilizationa consequence of the
high vapor pressure of this element. Likewise,
lead-containing solders are unstable in high
vacuum at temperatures much above 300 C (570
F) and are not recommended for use under these
conditions. Table 3.2 lists the boiling/sublimation
temperatures of selected elements at 1010 atm
(102 mPa). For metals to be joined under reduced
pressure, the process temperature must be consid-
erably less than the boiling/sublimation tempera-
ture (by a factor of < in K/K), if volatilization is
not to be signicant.
The oxygen partial pressure in a vacuum fur-
nace can be reduced substantially below the gas
pressure in the vacuum by repeatedly pumping
out and backlling the chamber with a dry, oxy-
gen-free gas (see Chapter 1, section 1.3.2.5). Care
must be taken to ensure that the inlet system is
completely leak-tight. Otherwise, some oxygen
will be bled into the furnace, and this will impair
or even nullify the benet of the inert atmo-
sphere. A periodic ushing of the chamber with
an inert gas will also serve to minimize any
108 / Principles of Soldering

buildup of oxygen released in the dissociation of
oxides during the heating cycle.
The effectiveness of using the process tem-
perature and oxygen partial pressure to control
oxide reduction, or at least prevent oxidation, is
limited by the presence of adsorbed water vapor
on the walls of the vacuum chamber and on other
free surfaces. The desorption of water vapor ef-
fectively increases the oxygen partial pressure in
the chamber, and this has a deleterious effect on

Fig. 3.3 Ellingham diagram for selected oxides. M, melting point of metal; B, boiling point of metal; M', melting point of oxide
 Chapter 3: The Joining Environment / 109

the oxide-removal process. Therefore, it is good
practice to heat the walls of the chamber to pro-
mote desorption, while simultaneously remov-
ing the vapor from the chamber by alternately
pumping out and/or ushing with dry, inert gas
before commencing the heating cycle. Addi-
tional background information on moisture in
closed spaces in given in Appendix A4.2. Nev-
ertheless, as can be seen from the nomogram in
Fig. 3.2, the quality of vacuum required to pre-
vent adsorption on to the faying surfaces needs
to exceed that which can usually be economi-
cally achieved in an industrial process.
Large-scale industrial processes often rely on
liquid nitrogen for several reasons. Not least of
these is the ease of convenience of delivery and
storage. Furthermore, nitrogen boiled off from a
cryogenic tank containing the liqueed gas pos-
sesses lower levels of oxygen and water vapor
(typically <2 ppm combined) than all but the
purest grades of bottled nitrogen. It is also rela-
tively inexpensive, being comparable in price
per liter to bottled mineral water. Owing to in-
creasingly stringent environmental legislation,
joining in inert atmospheres is gaining in popu-
larity. This is particularly true in the electronics
industry, where the trend is toward nitrogen at-
mosphere furnaces for both wave and reow sol-
dering processes. In commercial systems, the ni-
trogen ambient contains less typically than 10
ppm of other species. The running costs associ-
Table 3.2 Boiling/sublimation temperature of
selected elements at a pressure of 10!10 atm
(10!2 mPa)
Boiling/sublimation temperature
Element C F
ated with the large volumes of nitrogen that are
required to achieve this quality of atmosphere
are offset by the ability to dispense with post-
joining treatments because reduced quantities of
uxes and cleaning uids are required, with their
associated health and environmental problems.
For certain applications, inert gases other than
nitrogen may be more appropriate. Of these less
common inert gases, argon and carbon dioxide
are probably the most widely used. Both can be
purchased in high-purity form. Carbon dioxide is
often recommended in applications where the
atmosphere is conned, but open to air at various
portals, because the greater molecular weight of
carbon dioxide enables it to displace air more
effectively than does nitrogen [Esquivel and
Chavez 1992]. Argon is more expensive than the
other two gases, and its use is therefore largely
conned to joining in closed volumes.
3.1.3.2 Soldering in
Reducing Atmospheres
If the partial oxygen pressure surrounding the
workpiece cannot be sufficiently lowered to ef-
fect oxide removal by introducing a vacuum or
inert gas environment, then a reducing atmo-
sphere might be able to remove the oxide. The
three most widely used reducing gases are hy-
drogen, carbon monoxide, and cracked am-
monia (that is, ammonia dissociated into nitro-
gen and hydrogen).
The basic chemical reduction processes for
hydrogen and carbon monoxide are:
yH2  MxOy xM  yH2O

Cd 100 212
Zn 150 302 and
Mg 210 410
Sb 300 572
Bi 350 662
In 525 977 yCO  MxOy xM  yCO2
Mn 550 1022
Ag 630 1166
Al 725 1337
Sn 730 1346 For convenience to the user, the Ellingham dia-
Cu 780 1436 gram is provided with a series of side scales
Cr 800 1472 giving the partial oxygen pressure corresponding
Au 880 1616
Pd 905 1661 to ratios of H2/H2O and CO/CO2, as shown in
Fe 950 1742 Fig. 3.3. On the left-hand side is shown an axis
Co 1020 1868
Ni 1025 1877 for T  273 C (0 K) with points marked at
Ti 1130 2066 values of the free energies, G, at this tempera-
Mo 1680 3056
W 2230 4046 ture for the hydrogen/oxygen (point H), the car-
bon monoxide/oxygen (point C) and other reac-
Values are rounded. Note the high position of tin and the low position of man- tions. Each of these points is associated with one
ganese and zinc in the table in relation to their melting points.
of the side scales shown in Fig. 3.3.
110 / Principles of Soldering

As an example of the use of the Ellingham
diagram, consider the conditions for reduction of
chromium oxide. The free energy of formation of
Cr2O3 as a function of temperature is represented
by curve AB in Fig. 3.4. Values of the free energy
of formation of water vapor from the reaction of
hydrogen with oxygen are represented by a fam-
ily of curves diverging from the point H, each
curve corresponding to a different molar ratio
H2/H2O in the atmosphere. When curve AB
crosses a particular curve belonging to the family
of water vapor curves representing different wa-
ter vapor/hydrogen partial pressure ratios, the
chromium/oxygen and hydrogen/oxygen reac-
tions are in equilibrium because their respective
free energies are the same. This means that the
oxygen potentials for the two reactions are iden-
tical. When curve AB lies above the water vapor
curve at a particular temperature, chromium ox-
ide will be spontaneously reduced to form chro-
mium and water vapor by the hydrogen because
the latter combination is more stable than Cr2O3.
The reverse is true when curve AB lies below the
water vapor curve. For the equilibrium between
chromium/oxygen and water vapor/hydrogen re-
actions to be achieved at a soldering temperature
of 400 C (750 F), the H2O/H2 ratio must be
lower than 1010indicated by lines AB and
HQ. This condition cannot be achieved in prac-
tice, so that hydrogen is ineffective in removing
chromium oxide in soldering operations.
It is evident from the Ellingham diagram that
the stability of metal oxides decreases as the tem-
perature is increased and the oxygen partial pres-
sure is reduced. Commercial supplies of hydro-
gen, nitrogen, and other gases will inevitably
contain some oxygen and water vapor, which are
oxidizing agents, and component and furnace sur-
faces will usually contribute desorbed water to the
process atmosphere. In addition, atmospheres
with a high concentration of hydrogen present an
explosion risk, which usually precludes their use.
Although not given in Fig. 3.3, the reaction
curve for tin and its common oxide lies in close
proximity to that for carbon monoxide/dioxide.
Thus, at 400 C, it is theoretically possible to re-

Fig. 3.4 Simplied Ellingham diagram illustrating the graphical method for determining the temperature and H2O/H2 ratio that will
spontaneously reduce a metal oxide to metal (here, Cr2O3 to Cr). The set of dashed lines corresponds to the Gibbs free-energy
change as a function of temperature for the reaction of hydrogen with oxygen to produce water vapor for different H2O/H2 ratios.

duce tin oxide to tin with a H2O/H2 ratio as low as
102. However, the reaction rate is so slow that, in
practice, the gases hydrogen and carbon monox-
ide are not effective for reducing the oxides of
most common industrial metals, including tin, be-
low about 500 C (1380 F) at readily obtainable
oxygen partial pressures. The issue of reaction
rate is discussed further in section 3.3.3 in this
chapter.
3.1.3.3 Alternative Atmospheres for
Oxide Reduction
Other gases such as chlorine and uorine are
more effective than hydrogen and carbon mon-
oxide at removing surface oxides of particular
metals, as is clearly indicated on the relevant
Ellingham diagrams [Wicks and Block 1963].
The power of chlorides and uorides as cleaning
agents accounts for the effectiveness of chloro-
uorocarbons (CFCs) and the difficulty experi-
enced in replacing them by more environmen-
tally friendly chemicals, following the
implementation of the Montreal Protocol on Sub-
stances That Deplete the Ozone Layer: 1991
[McLaughlin et al. 1998; Lea 1991].
Such gases partly operate by converting the
oxide to a halide that is volatile at the joining
temperature and that vaporizes during the heat-
ing cycle. These halide atmospheres also chemi-
cally attack the underlying metal and physically
undermine the oxide, as occurs in the uxing of
aluminum. This point is discussed in section
3.2.2.1 in this chapter.
A different option, particularly for tin-base sol-
ders, is to use a process atmosphere that contains
a chemical ux in its vapor state. Formic acid is
one such example. Owing to the absence of a
liquid ux, joining under gaseous uxes is gen-
erally misdescribed as a uxless process. Fur-
ther information on chemically active atmo-
spheres is to be found in section 3.3.6 in this
chapter.
3.1.4 Forming Gas as an
Atmosphere for Soldering
Despite the scientic knowledge embodied in
the preceding discussion, it is a common mis-
conception that forming gas (nitrogen/hydro-
gen mix) is a better choice than nitrogen as a
process gas for soldering because it will reduce
tin and indium oxide. Figure 3.5 shows the result
of melting controlled amounts of four common
solders in atmospheres of pure nitrogen, nitrogen-
Chapter 3: The Joining Environment / 111
3% hydrogen, and nitrogen-40% hydrogen. From
this trial it can be concluded that there is no
particular advantage in using forming gas at small
excess temperatures above the solder melting
point. In fact, the European specication for com-
mercial grade forming gas allows it to contain a
higher level of oxygen and water vapor than is
found in bottled nitrogen. In the United States,
dry nitrogen and forming gas have roughly simi-
lar specications, but forming gas carries a price
premium as a specialty gas mixture. Experimen-
tal investigators are advised that 40% hydrogen
is well above the explosive limit for this gas in
air, so its use requires many safety features to
comply with health and safety legislation or
makes for something of a white knuckle ride
for the process operator! However, as explained
in section 3.3, if a process using higher than nor-
mal soldering temperatures can be adopted, then
a dry hydrogen atmosphere may facilitate ux-
less soldering, particularly for the higher-
melting-point gold-base solders, on account of
the greater thermal activation available and the
higher intrinsic nobility of the alloys.
It is sometimes claimed that better results in
terms of wetting and spreading are obtained when
using forming gas compared with pure nitrogen.
Frequently, this has more to do with the fact that
the quality (i.e., leak tightness) of the furnace and
gas control and conveyance system used with
forming gas are of a higher standard than for a
nitrogen furnace, as required by safety protocols
and hydrogen has a thermal conductivity seven
times that of nitrogen (see Table 3.3). Thus, for
the same process temperature setting, the parts
are likely experiencing additional superheat
when reowed in forming gas.
3.2 Chemical Fluxes for Soldering
Successful soldering is largely dependent on
the ability of the solder to wet and spread on
component surfaces. A major barrier to wetting
is presented by stable nonmetallic lms and coat-
ings on the surfaces, in particular oxides and
carbonaceous residues. Oxide lms on the faying
surfaces present more than a physical barrier to
wetting and spreading. Oxides are typically poor
thermal conductors, compared to metals, and act
as barriers to heat transfer, thereby exacerbating
temperature gradients present and delaying fu-
sion of the solder with the parent metal. The
thermal contact resistance of electroplated and
112 / Principles of Soldering

then reowed lead-tin eutectic solder on copper oxide thickness [Di Giacomo 1986]. Surfaces
is known to increase with the logarithm of the with a coarse-grained microstructure lose sol-

Fig. 3.5 Spread tests of four common solders, melted on NiCr/Au substrates at 10 C (18 F) superheat, in controlled atmospheres.
There is negligible benet from a hydrogen-rich atmosphere. Although somewhat subjective, the solders melted in the 40%
hydrogen atmosphere do appear to be cleaner with a brighter metallic luster. The holes in the substrates are of identical size and spacing
and act as scale markers. The solder disks measured 6 mm (0.24 in.) diameter by 25 m (1 mil) thick and were only supercially cleaned
before use. (a) Ag-80In-15Pb solder melted in nitrogen. (b) Same alloy melted in nitrogen-3% hydrogen. (c) Same alloy melted in
nitrogen-40% hydrogen. (d) Ag-96.5Sn solder melted in nitrogen. (e) Same alloy melted in nitrogen-3% hydrogen. (f) Same alloy melted
in nitrogen-40% hydrogen. (g) Au-20Sn solder melted in nitrogen. (h) Same alloy melted in nitrogen-3% hydrogen. (i) Same alloy melted
in nitrogen-40% hydrogen. (j) Au-12Ge solder melted in nitrogen. (k) Same alloy melted in nitrogen-3% hydrogen. (l) Same alloy melted
in nitrogen-40% hydrogen. Source: BAE Systems
 Chapter 3: The Joining Environment / 113

derability much more readily than those with
ne-grained microstructures because the oxide
lms grown on the former are more continuous,
with fewer defects. Fluxes are chemical agents
that are used to remove these layers and thereby
promote wetting by the molten ller.
In order to be effective in exposing a bare
metal surfaces, a ux must be capable of ful-
lling the following functions:
Removal of oxides and other lms that exist
on surfaces to be joined by either chemical or

Fig. 3.5 (continued) (g) Au-20Sn solder melted in nitrogen. (h) Same alloy melted in nitrogen-3% hydrogen. (i) Same alloy
melted in nitrogen-40% hydrogen. (j) Au-12Ge solder melted in nitrogen. (k) Same alloy melted in
nitrogen-3% hydrogen. (l) Same alloy melted in nitrogen-40% hydrogen. Source: BAE Systems
114 / Principles of Soldering

physical means, often involving reaction of Another method involves introducing the ux
the ux with surface oxides to form metal together with the ller into a joint already held
salts, which are then dissolved by the ux at the bonding temperature, in the form of ux-
Protection of the cleaned joint from oxidation cored solder wire. Although this technique is
during the joining cycle widely practiced because it is fast and conve-
Wetting the joint surfaces, but being dis- nient, it is not recommended because the heated
placed by the molten solder as the latter component surfaces are unprotected until the
spreads ller is applied. More aggressive uxes are then
required, which in turn tends to accentuate cor-
While molten, uxes form a thermal blanket
rosion and cleaning problems.
around the joint that helps to spread the heat
Alternatively uxes can be applied together
evenly during the heating cycle. The ux also
with the ller, prior to the heating cycle, in the
tends to reduce the surface tension between the
form of pastes and creams, which are normally
solder and the joint surfaces, thereby enhancing
proprietary formulations. They comprise mix-
spreading.
tures of the ller metal, which is present as a pow-
Ideally, the ux should leave no residues or
der of a prescribed shape and size distribution to-
produce residues that are easily removed by, for
gether with a ux and an organic binder that is
example, being soluble in water. It should also be
selected to produce the desired viscosity and to
compatible with the ller and substrate materi-
burn off without leaving contaminating residues.
als. For example, ammonia-containing uxes are
These pastes and creams are particularly useful in
not suitable for brass components, because in-
automated reow soldering operations because
tergranular corrosion can result through chemi-
they can be screen printed or dispensed using sy-
cal reaction. Chemical uxes always function
ringes. The large surface area of the powdered
while in a gaseous or liquid form, although the
ller metal in contact with the ux means that cor-
active constituents are frequently solid at room
rosion is inevitable during storage; therefore,
temperature.
these products have a nite shelf life and storage
Fluxes can be introduced to the joint in a num-
requirements, which are best strictly observed.
ber of ways, the most common of which are
The mechanisms of ux action are almost as
discussed here. A ux can be applied in the form
diverse as are the ux formulations that are com-
of a powder, paste, or liquid immediately prior
mercially available. In many cases, the mecha-
to the heating cycle. The joint is then heated to
nisms have not been fully elucidated, which is in
the required bonding temperature, by which point
no short measure due to the commercial secrecy
solid uxes have become molten, ideally just
surrounding ux compositions. While solder
before the ller metal melts. In wave soldering,
compositions are mature, ux chemistry is still
a liquid ux is applied to a circuit board using a
radically evolving. Todays soldering uxes are
foam or spray dispenser only seconds, or less,
very sophisticated chemicals compared with even
before the part enters the solder wave (see Chap-
a few years ago, and new products continue to be
ter 1, section 1.3.2.2).
launched almost monthly. Several different ux-
A ux can also be placed within or adjacent to
ing mechanisms cover the majority of soldering
the joint together with the ller metal as a pre-
operations that are encountered. Even these are
form and the assembly heated to the bonding
sufficiently complex not to be understood in de-
temperature. As a properly chosen ux will melt
tail at the present time. However, uxing mecha-
at a temperature below the melting point of the
nisms can be classied according to whether they
ller, the molten ux is able to spread over the
remove the nonmetallic surface coating by physi-
joint surfaces and clean them before the ller
cal or chemical means.
metal melts and displaces the ux.
A ux can chemically remove a surface oxide
coating by:
Table 3.3 Thermal conductivities of soldering
atmospheres, relative to air Dissolving the coating
Soldering atmosphere Relative thermal conductivity Reacting with the coating to form a product
Carbon dioxide 0.62
that is unstable at the bonding temperature
Argon 0.68 Reducing the oxide to metal in an exchange
Nitrogen 0.99 reaction
Air 1
Helium 5.8 A surface coating can also be physically re-
Hydrogen 6.9
moved. This usually occurs through erosion of
 Chapter 3: The Joining Environment / 115

the underlying metal. In this mechanism, the ux
does not react with the surface coating itself, but
is able to percolate through it and react with the
underlying metal, thereby causing detachment of
the coating. In addition, physical uxing action
can be achieved by applying mechanical agita-
tion, without the need for a material uxing agent,
as discussed in section 3.3.5. Many uxes func-
tion by a combination of mechanisms, and for
this reason uxing action is best illustrated with
reference to specic examples.
Environmental considerations have impinged
heavily on the use of certain uxes, particularly
those incorporating organic materials (volatile
organic components, or VOCs), because these
have tended to rely on CFCs for the removal of
their residues, which attack the earths ozone
layer when they are discharged to the atmo-
sphere. These older VOC formulations have been
largely replaced in manufacturing industry by
uxes whose residues are soluble in water or
cleaning agents that do not contain substances
that are environmentally harmful [Lea 1991; El-
lis 1991] (see Chapter 1, section 1.3.2.11).
Even when soldering using ux, the employ-
ment of an inert atmosphere, such as nitrogen, is
benecial. This is particularly true for the Ag-
Cu-Sn lead-free solders, which are used at much
lower superheats than lead-tin solder. The nitro-
gen atmosphere helps prevent oxidation of the
solder and substrate before the ux becomes ac-
tive and reduces the work that the ux has to do
in maintaining oxide-free interfaces ahead of the
advancing solder front. This means that less ux
is required, making it environmentally and eco-
nomically advantageous. If large superheats can
be applied, the benets of a protective atmo-
sphere are reduced because the solder spreads
and time at temperature all need to be carefully
optimized to achieve low-defect soldering. At
elevated oxygen levels, the rate of dross forma-
tion on the solder becomes signicant and will
interfere with the solder wetting and spreading,
as can be seen in Fig. 3.7 [Nowotarski and De
Wilde 1996].
Soldering in very low oxygen level atmo-
spheres does not necessarily translate to signi-
cantly improved production costs through maxi-
mizing yield. Indeed, just from the standpoint of
the cost of the increased gas consumption nec-
essary to reduce the oxygen level below 20 ppm,
at least one study showed that it was economi-
cally unfavorable (see Table 3.4) [Verite, Ver-
bockhaven, and Alleaume 1997]. Obviously this
analysis will vary with the value of the product,
in this instance the PCB for a mobile telephone,
which is produced in high volumes in a highly
competitive market.
Fig. 3.6 Wetting of copper by Pb-63Sn solder using rosin
ux. Soldering with ux generally benets from a
protective atmosphere (unless the atmosphere detrimentally af-
fects the chemistry of the uxing action), because the ux has to
work less to protect the substrate and ller from oxidation.

much faster, as can be seen from Fig. 3.6 [Buck-

ley 2000]. Further information on lead-free sol-
ders is given in Chapter 2, section 2.4.4 and
Chapter 5, section 5.1.
The effect of different concentrations of oxy-
gen on the results obtained when fabricating
uxed printed circuit boards (PCBs) using lead-
tin solder has been studied. In general, if the
oxygen level can be decreased to below 10 ppm
then the soldering process will generally be in-
sensitive to many process variables, because the
molten solder has a low surface tension and
shorter wetting time. Soldering processes can be
realized in an inert atmosphere where the oxygen
concentration is up to 1% (10,000 ppm). How-
ever, to be successful, most process parameters Fig. 3.7 Effect of oxygen concentration in the atmosphere on
the rate of dross (scum) formation on an exposed
such as ux type, heating rate, peak temperature, solder reservoir
116 / Principles of Soldering
3.2.1 Fluxes for Tin-Base Solders
The overwhelming majority of soldered joints
made are to interconnect electronic components.
The principal material that is joined in these ap-
plications is copper, due to its high electrical
conductivity; the solder is almost invariably a
lead-tin (or lead-free tin-base) alloy. Because
electronic components tend to be manufactured
and stored under reasonably cool, clean, and dry
conditions, they are likely to have only a thin
layer of copper oxide as a barrier to wetting.
This, coupled with the fact that soldering is usu-
ally performed rapidly and mostly below 300 C
(570 F), means that the uxes required tend not
to be highly aggressive chemicals. Major con-
siderations pertaining to uxes intended for elec-
trical/electronic applications, other than their
cleaning ability, are the nature of the residues and
the ease of their removal. These factors are of
concern owing to the need to avoid subsequent
deterioration of the joints and ultimately failure
of the circuitry through corrosion [Turbini et al.
1991].
Indium and zinc solders require uxes with a
chemistry different from tin-base solders, be-
cause of the composition of the oxides and the
different process temperatures involved. As such,
they have an even more specialized formulation
compared with uxes for tin-solders and are best
procured from specialist suppliers and used in
accordance with manufacturers guidelines.
3.2.1.1 Soldering Fluxes That Require
Cleaning
Conventional soldering uxes contain at least
four basic ingredients, each of which has an iden-
tied role [Klein Wassink 1989; Manko 2002]:
Acids or halides to provide the cleaning ac-
tion (the active constituents)
An ingredient that is liquid at the soldering
temperature that seals and protects the
cleaned surfaces against reoxidation
Table 3.4 Normalized cost analysis of
manufacturing cell phone PCBs, illustrating the
effect that the quality of the soldering
atmosphere has on manufacturing economics
Nitrogen  Nitrogen <20
Air 1% oxygen ppm oxygen
Cost of rework 100 37 30
Cost of nitrogen 0 22 32.7
Total cost per PCB 100 59 62.7
A surfactant that promotes wetting of the joint
surfaces by the active and sealing constitu-
ents
A rheological additive to suit the application
method
In practice, commercial uxes often contain
more than these four ingredients in order to meet
the requirements of the soldering process. Fluxes
have been formulated that are satisfactory for
most pure metals and alloys, including stainless
steel. Noted exceptions are beryllium, chro-
mium, magnesium, titanium, and some alumi-
num alloys, which are classied as unsolder-
able in air, unless coated with a different metal
(see section 3.2.2 and Chapter 4, section 4.1.2)
or a very special ux composition is employed
(see section 3.2.2). When selecting a ux for a
particular application, it is usually good practice
to follow the manufacturers guidelines, because
the effectiveness of a particular formulation tends
to be highly sensitive to the combination of met-
als and process conditions with which it is used.
The higher the activity of a ux, the greater is
its ability to remove surface oxides from metal
components, but so is the corrosiveness of the
ux and its residues. The active ingredient of a
solder ux can be either an inorganic acid (hy-
drochloric acid is commonly used) or an organic
acid (e.g., carboxylic acids). A common car-
boxylic acid used in uxes is abietic acid
(R3COOH, where R is an organic radical), which
is a major constituent of rosin uxes. In both
cases, the acid reacts with the surface cupric
oxide and converts it to compounds that are
readily removed, either chemically and/or physi-
cally, from the joint surfaces. Even the basic
form of the chemical reactions between these
uxes and copper oxide is complex, but can be
simplistically described by:
CuO  2HCl  CuCl2  H2O
CuO  2R3COOH  Cu(R3COO) 2  H2O
where R is an organic radical.
Copper chloride is soluble in water, while cop-
per abiet [Cu(R3COO)2] is miscible with rosin.
Hence, the compounds that now contain the
copper oxide dissolve in the excess liquid ux to
leave a clean metal surface when the ux is dis-
placed by the molten ller metal.
Like most active ux constituents, these acids
become progressively more active with increas-

ing temperature until a point is reached when
their effectiveness ceases, because they either
thermally decompose or boil off. The corrosive
properties of acids at room temperature can
present handling problems and application dif-
culties especially when the ux must either be
introduced directly to the components or mixed
with solder to form pastes sometime before the
joining operation. However, there are salts that
liberate acid only when heated, thus avoiding
this type of problem. One of these is zinc chlo-
ride, which produces hydrochloric acid by reac-
tion with moisture at elevated temperatures, ac-
cording to the reaction:
ZnCl2  H2O  Zn(OH)Cl  HCl
By using a mixture of similar salts, the activa-
tion temperature of the ux and its corrosiveness
can be adjusted over a wide range. In all cases, the
residues are highly corrosive. Other halide uxes
operate in a very similar manner to zinc chloride.
For example, the amine hydrohalides, such as hy-
drazine hydrochloride (R2NH2Cl), thermally de-
compose with the liberation of hydrochloric acid:
R2NH2Cl  R2NH  HCl
The ux constituent that protects the clean
metal surface from reoxidation usually also
serves as the carrier for the other ingredients. It
need only be effective for a few seconds in many
soldering processes. Alcohols, oils, esters, gly-
col, and even water are capable of fullling this
function at the relatively low temperatures used
for most soldering operations (<250 C, or 480
F). As the ux carrier constitutes by far the
largest volume fraction of the formulation, it has
a strong inuence on the aggregate properties
even though it is not intended to be chemically
active. The ux carrier is usually a mixture of
chemicals formulated partly to prevent the ux
from boiling violently when the workpiece
reaches a specic temperature. As mentioned pre-
viously, it may also play a role in mopping up the
products of reaction.
Surfactants are added to lower the surface ten-
sion of the liquid ux. The effect of minute ad-
ditions of detergents, soaps, and soluble oils to
water is well known, and these are often added
to water-based uxes to ensure satisfactory wet-
ting when either oil or grease are likely to be
present on the component surfaces. The surfac-
Chapter 3: The Joining Environment / 117
tant needs to be inert toward the other constitu-
ents of the ux and also to the clean metal sur-
face. For this reason, ion-free organic complexes,
similar to those widely used in the electroplating
industry, tend to be favored.
The rheological agent is provided to impart
the correct degree of body to the ux to suit the
application method, whether syringe dispensing,
screen printing, or stenciling.
A common commercial designation of uxes
used for soldering is:
R: rosin
RMA: rosin mildly activated
RA: rosin activated
OA: organic acid
IA: inorganic acid
WS: water soluble
SA: synthetically activated
This classication is somewhat loose and can be
more misleading than helpful. It does not strictly
indicate the chemical characteristics of the ux,
particularly its aggressiveness to specic oxides.
However, because this classication is widely
used, it warrants consideration here. An alterna-
tive classication of soldering ux types, de-
vised by the International Organization for Stan-
dardization, is given in Table 3.5. Although this
classication is scientically more precise, it is
less descriptive; therefore, the traditional terms
tend to be preferred. The different categories can
be grouped as shown below.
R, RMA, RA. Fluxes with these designations
contain rosin as the principal active chemical
ingredient, the difference between them being a
progressive increase in level of chemical activity
from R through to RA. In modern uxes, the
rosin is a synthesized chemical and no longer the
natural product obtained from the sap of pine
trees.
OA. OA stands for organic acid, which pro-
vides the cleaning action in this type of ux. The
organic acid used will tend to be one of those
described previously, such as a carboxylic acid.
These uxes are generally more aggressive than
RA uxes.
IA. IA uxes are based on inorganic acids,
usually hydrochloric acid and are among the most
aggressive of the available uxes.
SA. Synthetically activated uxes are formu-
lated to have residues that are soluble in chlo-
rouorocarbon solvents. With concern growing
about atmospheric pollution, and CFCs in par-
ticular, the SA uxes have been superceded by
118 / Principles of Soldering
newer formulations that have water-soluble (WS)
residues.
WS. Water-soluble uxes containing soluble
halides have been available for several decades.
These are among the most commercially devel-
oped uxes available with chemical activity tai-
lored over a wide range, roughly in proportion to
the total halide content. Particular care needs to
be taken to clean off ionic residues, which can
cause corrosion and generate electrical malfunc-
tion. Water-soluble uxes tend to have shorter
printing life than rosin uxes, have less tackiness
and are therefore more difficult to use for reow
soldering.
3.2.1.2 No-Clean Soldering Fluxes
In response to manufacturers concern over
the cost of cleaning, a new generation of no-
clean uxes has come into use that are said to
leave no deleterious residues (in PCB assembly
applications). These uxes have been formulated
with a wide range of chemical ingredients, but all
share the characteristics of leaving residues that
are judged to be benign and so can be left on the
assemblies.
No-clean uxes mostly contain an alcohol sol-
vent carrier and a small percentage of active in-
gredients, which may comprise resins or simply
organic acids. Some of the newer no-clean uxes
are water-based and are totally free of VOCs,
including alcohols. Fortuitously, using water as
the solvent in place of a VOC, such as an alcohol,
has been found to enhance no-clean uxing. In
particular, water-based no-clean uxes appear to
cope better with circuit boards that enter the sol-
dering operation with surface oxide. Their en-
hanced activity with respect to metal oxides has
Table 3.5 Classication of soldering uxes using the method adopted by the International
Organization for Standardization
Flux type Flux basis
1 Resin 1 Rosin
2 Resin
2 Organic 1 Water soluble
2 Not water soluble
3 Inorganic 1 Salts
2 Acids
3 Alkalis
Thus, a resin-based paste ux with a halogen activator is classed as type 122C.
been ascribed to a more complete ionization of
the activators in a water base than in alcohol
[Hyland and Rao 1996].
The need to clean is obviated by diluting the
activators to low levels. Therefore, these uxes
contain a much lower content of rosin, or ha-
lidesreferred to simply as solidsthan do
traditional uxes. The solids content of no-clean
uxes is normally in the range 2 to 3 wt%, as
compared with 25 to 35 wt% for normal rosin-
based uxes. These low-solid uxes are more
difficult to work with than the rosin uxes pre-
cisely because they are less active.
An alternative approach used for formulating
no-clean uxes is to use activators of normal
strength or thereabouts in conjunction with other
constituents that undergo polymerization when
subjected to soldering temperatures. Gelatin is
one such compound. The resulting polymer frac-
tion encapsulates residual active species, thereby
incapacitating it. However, this mechanism can
be impaired if not all the ux is fully heated,
leaving areas of the joint exposed to active resi-
dues. It is also important that no-clean ux resi-
dues are not cleaned because, unless the cleaning
is scrupulous it leaves behind destabilized en-
capsulant from which corrosive agents can sub-
sequently escape.
From the foregoing outline, it is clear that
no-clean uxes are less forgiving, being more
intolerant to variability in the soldering process
conditions than conventional alternatives. Suf-
cient ux must be provided to achieve the re-
quired solderability and, as noted previously, the
ux must be heated uniformly and in the correct
manner to react fully and leave innocuous by-
products. These considerations are consistent
with the experience of industrial users, who nd
Flux activation Flux form
1 Not activated A Liquid
2 Halogen activated
3 Not halogen activated
B Solid
1 With NH4Cl
2 Without NH4C1
1 Phosphoric acid
2 Other acids
1 Ammonia and/or amines C Paste

that no-clean uxes work best in assembly-line
soldering operations when the process condi-
tions are optimally tuned and strictly maintained
from one assembly operation to another. While
this situation is readily achieved in a factory
using modern equipment, it does mean that no-
clean uxes are not generally suitable for hand-
soldering processes. When correctly instigated,
the residues left on a PCB are not detrimental in
most applications and accelerated life-test envi-
ronments [Anson et al. 1996].
3.2.1.3 Measure of Cleaning
Effectiveness: The Surface
Insulation Resistance (SIR) Test
Because ux residues, like other contaminants,
can easily affect the performance and service life
of electronic circuits, effective cleaning proce-
dures assume a critical importance. A semiquan-
titative test has been devised to assess the effec-
tiveness of a cleaning process for removing ux
residues from circuit boards and to provide assur-
ance that boards soldered using no-clean ux will
not fail in service due to unbound residues. This
test measures the surface insulation resistance
Fig. 3.8 International Electrotechnical Commission (IEC) test coupon for the evaluation of board cleanliness using SIR. Schematic
of typical detail of an interdigitated comb is illustrated. Courtesy of Concoat Systems
Chapter 3: The Joining Environment / 119
(SIR) to current ow between pairs of conducting
tracks in interdigitized comb patterns on boards as
a function of time at a prescribed temperature and
level of humidity. It is capable of catching elec-
trochemical failure mechanisms associated with
vestigial ionic residues, namely unacceptable cur-
rent leakage under humid conditions, corrosion,
metal migration, and dendritic growth.
An electric eld is applied between the two in-
termeshed sets of combs, as shown in Fig. 3.8. The
widely accepted SIR test standard (ISO 9455, part
17) calls for a test pattern with 400 m (16 mil)
wide tracks, separated by 500 m (20 mil) wide
gaps, and a test eld of 100V/mm (2.5V/mil) (i.e.,
a test voltage of 50 V).
The test environment is a constant tempera-
ture of 85 C (185 F) and a relative humidity of
85%. Surface insulation resistance measure-
ments are made twice a day over a test duration
of 168 h. If the surface resistance is maintained
above 108  throughout the test, the surface is
judged to be clean.
The presence of ionic contamination on the
surface of the test board can result in the forma-
tion of conducting dendrites, as shown in Fig.
3.9. These dendrites can grow rapidly (in min-
utes), starting from residues deposited on a cath-
120 / Principles of Soldering
ode and progressing toward a neighboring an-
ode. When a dendrite has bridged the gap, the
SIR value will rapidly decrease. As a dendrite
grows, it will progressively carry a larger pro-
portion of the available current until at some
point it burns out and the SIR will return to a
high value. This sequence of events repeats
until carbon deposits build up sufficiently to
permanently lower the resistance below the ac-
ceptance threshold. Examples of SIR test plots
(log SIR versus time) are shown in Fig. 3.10.
A recent assessment of the SIR test has high-
lighted the fact that the test procedure, as cur-
rently followed, can miss detection of failure
events [Hunt 2000]. It is possible that resistance
drops, due to dendrite formation, might escape
observation if the electrical measurements are
widely spaced in time: half a day is clearly too
long. Moreover, the bridging of a dendrite across
neighboring conductors, which lasts for even
only a few seconds before burnout, can repre-
sent a large number of operations in equipment
operating at a clock frequency of MHz and
above. For such equipment, intermittent den-
drite bridging would cause a corresponding in-
cidence of failures, which the test might not
pick up.
This limitation can be overcome by using mod-
ern high-speed multichannel measuring equip-
ment in the SIR test. Other recent recommen-
dations to improve the test include using a lower
voltage stress (5 V/mm) and ner test structures
(100 m/100 m, or 4 mil/4 mil track/gap), which
would be more appropriate of current circuitry
and operating conditions. It is also proposed to
change the test environment from 85 C (185
F)/85% RH to a more demanding 40 C (104
F)/93% RH.
Fig. 3.9 Example of conducting dendrites growing across the
gaps of a test pattern on a circuit board during a SIR
test. Courtesy of Concoat Systems
As a result of such reviews of the SIR test
procedure, the following compromise set of con-
ditions is being proposed for a new Standard
specication (draft IEC 61189-5):
The test pattern should comprise combs with
a 400 m (16 mil) track width and 200 m (8
mil) gap width.
The test environment should be 40 C (104
F) and 93% RH.
Surface insulation resistance (SIR) measure-
ments should be made at 20 min intervals.
The test voltage should be 5 V, creating an
electric eld of 25 V/mm (0.63 V/mil).
The test patterns should be overmounted
with dummy components to better replicate
the situation in populated circuit boards, when
using the SIR test for cleaning and other pro-
cess characterization purposes.
3.2.2 Fluxes for Unsolderable Metals
Aluminum, chromium, and some other metals
are often classied as unsolderable because
they are not wetted by lead-tin solders using
common uxes. Table 3.6 provides an indication
of relative solderability of some engineering met-
als, alloys, and metallizations. The difficulty of
soldering to many metals is an exaggerated per-
ception because most uxed solders found in
laboratories, factories, and workshops are de-
signed for soldering of electronic components
and in chemical terms are relatively mild. They
are certainly not capable of dealing with the na-
tive oxide on some of the more refractory metals.
Aluminum and stainless steel are two engi-
neering materials for which there is often a re-
quirement to make a solder joint or connection
and where ux is permitted. Commercially avail-
Fig. 3.10 Example of an SIR test plot (log SIR vs. time), show-
ing features usually associated with dendritic
growth. Courtesy of Concoat Systems
 Chapter 3: The Joining Environment / 121

able uxes exist for both of these unsolderable
metals.
3.2.2.1 Aluminum Soldering Fluxes
Aluminum forms a natural refractory oxide
that is remarkably stable and tenacious. It is me-
chanically durable, with a hardness that is only
inferior to that of diamond and its high melting
point (2050 C, or 3722 F) reects its high de-
gree of physical stability. Alumina is also chemi-
cally stable to the extent that it cannot be directly
reduced to the metal by aqueous reagents. On
exposure to air, a layer of alumina will form
almost instantaneously on the surface of alumi-
num, and this will grow to an equilibrium thick-
ness of between 2 and 5 nm (0.08 and 0.2 in.)
at ambient temperature. On heating to 500 to 600
C (930 to 1110 F), the thickness of this surface
coating will increase to about 1 m (40 in.).
Therefore, special uxes have been formulated
for use with aluminum alloys. These have to be
particularly effective in protecting the metal sur-
face from oxidation before the solder melts and
spreads.
Aluminum uxes all contain halide com-
pounds. These are highly corrosive especially in
the presence of moisture, including humid at-
mospheres. Therefore, all ux residues must be
removed as completely as possible, as residues
left behind after the cleaning procedures may
cause corrosion in the vicinity of the joint.
The uxes used for soldering of aluminum and
its alloys are of two types [The Aluminium As-
sociation 1990]: organic uxes and chloride-
based uxes.
Organic uxes contain amines, uoborates,
and a heavy metal compound in an organic car-
rier. They come in the form of viscous liquids or
powders. A typical example of this type of ux
has the composition: 83% triethanolamine, 10%
uoboric acid, and 7% cadmium uoroborate (a
viscous liquid). Its operating range is 180 to 280
C (355 to 535 F).
The uxing action relies on disrupting the ox-
ide, which cracks and crazes during the heating
operation due to the differential thermal expan-
sion between the metal and oxide. This enables
the ux to come into direct contact with the alu-
minum and deposit a lm of the metal ion in the
ux (in this instance, cadmium) onto the alumi-
num surface via an exchange reaction. Organic
uxes must not be exposed to a torch or ame;
otherwise they will char, and this will impede
solder ow. Aluminum alloys containing more
than about 1% Mg cannot be satisfactorily sol-
dered using these uxes, because magnesia is
more refractory than alumina and the ux is cor-
respondingly less effective.
Chloride-based uxes contain zinc or tin
chlorides with ammonium chloride and uoride
and are generally applied as a water-based slurry
or paste to precleaned component surfaces. An
example of such a ux has the formulation: 88%
tin chloride, 10% ammonium chloride, 2% so-
dium uoride (powder) with a working range of
300 to 400 C (570 to 750 F). By substituting
the tin chloride with zinc chloride, the tempera-

Table 3.6 Relative solderability of selected metals and alloys

Parent material Easy Intermediate Difficult Unsolderable

Aluminum ... ... X ...

Aluminum alloys ... ... ... X
Beryllium ... X ... ...
Brass X ... ... ...
Chromium ... ... ... X
Copper X ... ... ...
Copper-nickel ... X ... ...
Gold X ... ... ...
Invar/Kovar ... ... X ...
Lead X ... ... ...
Magnesium ... ... ... X
Nichrome ... ... X ...
Nickel ... X ... ...
Palladium X ... ... ...
Platinum X ... ... ...
Silver X ... ... ...
Stainless steel ... ... X ...
Steel ... X ... ...
Tin X ... ... ...
Titanium ... ... ... X
Zinc ... X ... ...
122 / Principles of Soldering
ture of operation can be raised to 380 to 450 C
(715 to 840 F).
The uxing mechanism is essentially the same
as that for the organic uxesnamely, one in-
volving an exchange reaction whereby alumi-
num on the surface is replaced by zinc or tin. The
effectiveness of these uxes is reduced by the
presence of silicon in the parent material, be-
cause silicon is not as amenable to the exchange
reaction as is aluminum.
Because both of these two uxes operate by
substituting for aluminum a metal that has rea-
sonable oxidation resistance and that is more
readily wetted by the ller metal, they are fun-
damentally different from conventional solder-
ing uxes. The latter act simply by cleaning and
protecting the original surfaces of the compo-
nents. Because the replacement metal has a
higher density than aluminum, this type of ux-
ing process is commonly referred to as heavy
metal deposition.
The quantity of ux that needs to be applied
is a function of the humidity of the ambient at-
mosphere. In moist atmospheres, a proportion of
the ux is rendered ineffective through hydroliza-
tion by reaction with water vapor. It has been
found that the quantities of ux that need to be
applied can be reduced considerably by carrying
out the soldering operation in a completely dry
environment.
3.2.2.2 Stainless Steel Soldering Fluxes
Stainless steel is not an easy material to solder
for two reasons. Firstly, it is covered with an
extremely stable (self-repairing) oxide layer,
which indeed gives it its stainless characteristic.
Secondly, for a metal it has an unusually low
thermal conductivity, one-thirtieth of that of cop-
per, which causes complications when endeav-
oring to heat joints quickly and to a uniform
temperature. Aggressive halogen-based uxes
can attack the oxide layer on stainless steel but,
unless the residues are fully removed, they are
prone to cause pitting corrosion. A solution of
zinc chloride in hydrochloric acid constitutes the
chemically active ingredients of uxes for stain-
less steel [Mei and Morris 1992].
Phosphoric acid is also extremely effective as
a ux on stainless steel, but unfortunately it pol-
ymerizes at temperatures above 200 C (392 F),
which precludes its use with the common tin-
base solders. Multicomponent, lead-free, solders
based on bismuth-tin have been devised for sol-
dering to stainless steel using phosphoric acid
ux [Wakelin 1993]. The solder possesses suf-
cient ductility to survive bending several times
by hand and can be cold worked by rolling. De-
tails of the solder composition have not been
disclosed, and it has not yet been developed as
a commercial product.
3.2.2.3 Magnesium Soldering Flux
Magnesium is also deemed to be unsolder-
able by conventional means. However, an in-
teresting development is the discovery that mol-
ten acetamid will directly dissolve the surface
oxides on both aluminum and magnesium [Gol-
ubtchik 1984]. Because this organic compound
melts at 83 C (181 F), it offers a route whereby
indium-tin solder (melting point 120 C, or 248
F) could be used to wet the exposed metal on
the substrate. An experimental trial conducted by
one of the authors under very rudimentary con-
ditions conrmed that the approach does have
some potential and possibly merits further in-
vestigation, particularly as a method for tinning
magnesium prior to more conventional solder-
ing.
3.2.3 High-Temperature Fluxes
Fluxes for soldering at high temperatures (in
the range 250 to 450 C, or 480 to 840 F) are
formulated differently to uxes designed to work
with lead-tin eutectic solder and alloys of similar
melting point. There are three reasons for this.
First, the higher process temperature means
that there is more thermal activation available so
that the active constituents need not be as chemi-
cally aggressive. Second, these uxes are usually
used with high-lead solders or gold-tin eutectic
solder. Thus, the oxide that impedes wetting and
spreading on the surface of the solder is either
lead oxide or tin oxide diluted by the noble metal
so, again, decreased chemical activity is suffi-
cient. The nal consideration is that the ux needs
to provide signicantly more protection to the
ller metal and component surfaces against oxi-
dation, because higher process temperatures usu-
ally mean that the duration of the process cycle
is longer.
For these reasons, high-temperature uxes
tend only to be mildly active and are based pre-
dominately on high-molecular-weight hydrocar-
bons. As might be expected, the long-chain mol-
ecules are very viscous at room temperature, and
these uxes are very difficult to dispense, espe-
cially on a cold morning. To overcome this prob-

lem, the ux is usually thinned with a lower-
molecular-weight species. On heating, the lighter
fraction of the mixture progressively volatilizes
as the viscosity of the high-molecular-weight hy-
drocarbon declines, so ensuring that the uidity
of the ux remains essentially constant through-
out the working temperature range. A point to be
aware of when using these uxes, especially in
conned spaces such as ip-chip assembly op-
erations, is that if the heating rate is too fast, then
on reaching about 150 C (300 F), the ux can
boil and physically displace components. This
is simply rapid volatilization of the lighter hy-
drocarbon fraction. Incorporating a dwell in the
heating cycle or a slow ramp through the critical
temperature range are appropriate remedies.
In air, high-temperature uxes produce a
plume of smoke when heated to the soldering
temperature. This fume is ammable, so that sol-
dering cannot be undertaken with a naked ame.
In enclosed spaces the fume tends to recondense
on cooler surfaces, which requires that cleaning
is undertaken on a planned basis. Higher-
temperature uxes are available in all of the mod-
ern designations including water soluble and
no-clean.
3.3 Fluxless Soldering
The vast majority of commercial soldering op-
erations involves the use of chemical uxes. The
use of ux makes the soldering operation toler-
ant to an ambient air environment and maintains
the solderability of surfaces through the heating
cycle, as explained in section 3.2. However, there
is a penalty to be paid when uxes are used in
the residues that they leave behind. Flux residues
are never completely removed in normal clean-
ing procedures, and the contamination can im-
pair the product function, performance, and life.
High-performance electronics systems, par-
ticularly those enclosed in hermetic packages,
usually proscribe the use of all organic materials
in their assembly, including uxes, as a means of
guaranteeing service life. A particularly critical
application is hybrid assembly for space elec-
tronics, where even small traces of organic sub-
stances present an unacceptable risk, owing to
the tendency of hydrocarbons to outgas. Because
aerospace electronics are often contained within
hermetic packages, the presence of moisture in
the package environment can activate a litany of
semiconductor failure mechanisms and is there-
fore undesirable.
Chapter 3: The Joining Environment / 123
The problem is no less acute in relation to
optoelectronic and photonic modules. A typical
example is a laser diode soldered to a heat sink
(for example, a thermoelectric cooler) in order to
stabilize the junction temperature. The joining
operation must leave the adjacent light-emitting
facet in a pristine state. Contamination of this
surface with virtually any species will either de-
grade the service life of the device or even have
catastrophic consequences for the operational
characteristics. Similar considerations pertain to
the assembly of tiny microelectromechanical sys-
tems (MEMS) [Kuhmann et al. 1998].
For these and other comparable applications,
there is the need to use uxless soldering pro-
cesses, which will also furnish thin and well-
lled joints to satisfy the thermal and mechanical
requirements of the assembly. Fluxless soldering
offers other attractive features: it eliminates a
processing material and more signicantly in
terms of cost saving, processing steps including
cleaning after the soldering operation. Fluxless
soldering also obviates a major source of voids
in joints, namely, that arising from trapped ux
products.
The crux of a successful uxless soldering
process is to eliminate all surface contamination
from the faying surfaces and to provide sufficient
protection of exposed surfaces from oxidation
through the heating cycle. At the same time, del-
eterious contaminants must be excluded from the
joining environment. A ux is designed to
achieve these conditions, and if it is dispensed
with then other means have to be found to satisfy
them.
As might be expected, there is direct corre-
lation between oxide thickness and solderability.
Similarly, there is a clear relationship between
the solderability of electronic component termi-
nations and defect levels in volume manufac-
turing of electronic circuit boards. Thus, it is
possible to establish a relationship between de-
fect levels and oxide thickness on a given surface
nish. Figure 3.11 presents results from board
trials that were conducted in air using a nonac-
tivated ux, involving copper lands. Clearly, the
rst 5 nm (0.2 in) of oxide layer thickness has
a very signicant impact on solderability and the
number of defective joints that result.
While it is fairly straightforward to achieve a
nonoxidizing atmosphere of sufficient quality to
largely maintain the solderability of joint sur-
faces during a rapid heating cycle, initial re-
moval of surface contamination, and particu-
larly the native oxides, is more complex. Tackling
124 / Principles of Soldering
this problem rst requires knowledge of the
nature and thickness of the oxides involved.
3.3.1 Oxide Formation and Removal
During uxless soldering the prime impedi-
ment to wetting and spreading is the presence of
oxide lms on the surfaces of the parent and
ller metals. All base metals are covered with a
thin lm of oxide through contact with air. In
this respect, the noble metals, gold and plati-
num, are exceptional. Surface coatings of these
metals are exploited for their perpetual solder-
ability, stemming from their inertness to oxy-
gen.
The thickness of the native oxide on a base
metal surface depends very much on the history
of the particular component. The principal fac-
tors governing the thickness of the oxide layer
are time of exposure to air and temperature. The
oxidation of some common metals and metalli-
zations at room temperature, after mechanical
cleaning of the surface, are indicated in Fig. 3.12.
Fig. 3.11 Effect of oxide thickness on copper lands on the
defect level of joints incurred during PCB assembly
Fig. 3.12 Oxide growth on four base metals at room tem-
perature, as a function of time
This graph shows that the surface of these base
metals will be covered with more than 5 nm (0.2
in.) of oxide within 5 min of the cleaning op-
eration, although the oxide layer could be sub-
stantially thicker if the same metal were exposed
to certain chemicals or heated in air. Obviously,
this surface contamination must be removed or
displaced by some means before wetting by the
solder can proceed.
The curves represented in Fig. 3.12 should be
taken as being indicative only, because in prac-
tice the growth rate is affected by a multitude of
factors, including surface roughness, residual
stresses, and humidity. For elevated tempera-
ture, an approximate rule of thumb is that the
native oxide thickness on base metals eventu-
ally doubles with every 200 C (360 F) in-
crease in temperature. For this reason, it is
desirable to heat the joint rapidly and as soon as
possible after cleaning of the surfaces, even
when carrying out a soldering operation with
the benet of ux and especially so in its ab-
sence. A detailed theoretical treatment of native
oxide growth on base metals is given by Martin
and Fromm [1977].
Because oxide lms grow so rapidly on most
base metals, as shown in Fig. 3.12 and 3.13, in
practice clean and solderable component sur-
faces can only be achieved if these are of gold
or platinum. Accordingly, for uxless soldering,
a gold coating is generally applied to the com-
ponent surfaces, which is of sufficient thickness
and adequately pore-free to ensure good solder-
ability over its specied storage life. Further
information on this subject is given in Chapter
4, section 4.1.2.1.
An oxide lm will likewise grow rapidly on
any solder preform exposed to air. To a rst
approximation, metallic oxide growth has a para-
bolic relationship with time, following a stan-
dard diffusion-type equation:
X2  Dt  Dot e(Q/RT)
where X is the layer thickness, D is the diffu-
sivity, t is the time in seconds, Do is the diffusion
coefficient (m2/s), T is the temperature in Kelvin,
Q is the activation energy (J/mol), and R is the
universal gas constant (8.314 J/mol K).
For pure tin, Q  33 kJ/mol and Do3.7 
1018 m2/s, while for eutectic lead-tin solder the
respective values are 40 kJ/mol and 2.5  1017
m2/s. Thus, the native oxide on solid solder pre-

forms will be typically a few nm thick shortly
after cleaning, but will rapidly increase in thick-
ness, as can be seen in Fig. 3.14, and especially
as the heating cycle progresses, unless the at-
mosphere is either inert or actively prevents re-
oxidation.
The dominant species of oxide that forms on
solder is usually consistent with that predicted
based on classical thermodynamics, namely the
stable oxide of the constituent element that has the
lowest free energy of oxide formation. This me-
tallic constituent will tend to oxidize preferen-
tially at the surface of the alloy preform. Thus,
even on the high lead 95Pb-5Sn solder, the oxide
skin contains somewhere in the region of 20 times
more tin than lead. The free energies of oxide for-
mation of typical elements used in solders are
listed inTable 3.1. Metal hydroxides are thermally
unstable with respect to the oxides, so native sur-
face lms are usually not hydrated to any appre-
ciable extent. From Table 3.1 it can be seen that
only the stability of indium and zinc oxides ex-
Fig. 3.13 Oxide growth on three base metals as a function
of temperature. The time at temperature is of the
order of a few minutes and has some correspondence with typical
soldering cycle times once component mass and heating rates are
taken into consideration
Fig. 3.14 Oxide thickness versus oxidation time for a range of solders held in air 140 C (284 F) above their melting point. Adapted
from Dong, Schwarz, and Roth [1977]
Chapter 3: The Joining Environment / 125
ceeds that of tin. Hence the majority of solder al-
loys are predominantly covered with a layer of tin
oxide and only indium- and zinc-containing al-
loys have different surface oxides. This is also the
reason why indium- and zinc-base solders each
have their own unique ux formulations.
The tin oxide growth on molten lead-tin solder
follows a square-root time dependence, whereas
the combination of oxides that form on In-48Sn
solder follows a parabolic growth law, as shown
in Fig. 3.15. Indium oxide initially thickens much
more rapidly than tin oxide, particularly when
the solder is molten and accounts for the relative
difficulty of obtaining satisfactory wetting and
spreading using indium solders without ux. In-
dium-tin solders oxidize to form a mixed oxide
of In2O3 and SnO and so are somewhat easier to
use than other indium solders, but still less so
than tin-base solders. Likewise, on gold-tin al-
loys, the tin oxide is diluted through the incor-
poration of gold. As gold does not oxidize, ex-
cept in very unusual environments, for gold-tin
solder the rate of oxide growth is slightly slower
than on pure tin.
3.3.2 Self-Dissolution of Solder
Oxides
Published work indicates that if a molten metal
or alloy is heated above some critical tempera-
ture then dissolution of thin lms of surface ox-
ide into the bulk of the melt can occur, the extent
depending on the solubility of the oxide. If the
heating cycle is conducted in a controlled atmo-
sphere that prevents reoxidation of the free sur-
face, then a uxless soldering process may be
possible under these conditions.
126 / Principles of Soldering
The superheats (temperatures above the melt-
ing point) required to dissolve the surface oxides
on seven common solder alloys, in nitrogen, have
been determined experimentally and are given in
Table 3.7 [Dong, Schwarz, and Roth 1977]. Inter-
estingly, the high-tin-containing alloys perform
well in this regard, compared with lead-tin eutec-
tic solder, and this phenomenon might partly ex-
plain the success being achieved in running lead-
free soldering process at very low superheats,
using Ag-Cu-Sn alloys with only mildly active
uxes. Solder-spreading tests suggest that the
spreading of high-tin-containing solders has far
higher tolerance to the oxygen level in the process
atmosphere (nitrogen) than other solders, as
shown in Table 3.8 [Dong, Schwarz, and Roth
1977], for the same reason.
The combined effect of the presence of surface
oxide on both the substrate and solder can be seen
in Fig. 3.16, which shows the contact angle for In-
48Sn solder on various substrates, after heating
for 5 min at 250 C (480 C), as a function of oxy-
gen partial pressure. Because the quality of the at-
mosphere used is nowhere near adequate to affect
oxide reduction on the solder, the conclusion is
that spreading is being achieved through a com-
bination of oxide dissolution and other physical
mechanisms that locally rupture the oxide lms
and facilitate sites of direct metal-to-metal con-
tact from which spreading can proceed.
The ability of solders to dissolve their own
oxides has only a limited benet for most ux-
less soldering applications. While wetting and
spreading characteristics may supercially ap-
pear quite satisfactory, the degradation of the
solder resulting from oxide dissolution is mani-
fested in increased viscosity and impaired me-
chanical properties. As a process option it is
therefore only really applicable to circumstances
Fig. 3.15 Oxide growth on molten solders at an oxygen par-
tial pressure of 1 Pa (1.5  104 psi). Adapted from
Kuhmann et al. [1998]
where the joint contains a large volume of solder,
a high ratio of the volume-to-surface area of the
ller metal, and large superheats can be used.
3.3.3 Reduction of Solder
Oxides by Hydrogen
For the reasons that are explained in section
3.1.3.2, whether or not an oxide can theoretically
be reduced by hydrogen depends on the stability
of the oxide at the process temperature and the hy-
drogen/water partial pressure ratio in the solder-
ing atmosphere. Once these fundamental condi-
tions are satised, one then needs to consider the
oxide reduction rate. Even when an oxide is ther-
modynamically unstable, the rate of transforma-
tion of the surface oxides might be so slow as to
render the process practically ineffective.
In a normal soldering process, one might expect
the initial layers of oxide on the surface of the ller
metal to be of the order of, for example, 2 nm thick
(0.8 in.) and the process cycle time available no
more than 1 min. In this case, for reduction by hy-
drogen to be effective, the dynamics of oxide re-
moval has to exceed 2 nm/min (0.8 in./min).This
means that either the quality of the atmosphere
must exceed the thermodynamic minimum value
and/or the temperature must be increased above
the base point. From measurement of the reduc-
tion rate of solder oxides as a function of tempera-
ture in hydrogen gas containing about 10 ppm
combined oxygen and water vapor, it is possible
to dene a threshold temperature above which the
reduction rate exceeds 2 nm/min (0.8 in./min).
As can be seen from Fig. 3.17, for solders covered
with tin oxide the threshold temperature is ap-
proximately 430 C (800 F) and for solders cov-
ered with indium oxide it is slightly higher at 470
C (880 F). The initiation temperature for the re-
duction of zinc oxides exceeds 500 C (930 F),
at which point the volatility of zinc will start to
cause other problems. Hence, hydrogen gas is
Table 3.7 Superheats required to dissolve
native oxides on selected solders
Superheat for dissolving oxides(a)
Solder C F
Sb-95Sn 10 18
Cu-99Sn 18 32
Ag-96Sn 19 34
Pb-63Sn 77 139
In-48Sn 93 167
Sn-9Zn >300 >540
(a) Excess temperature above the melting point

Table 3.8 Tolerance of molten solders to the oxygen level in the atmosphere, as indicated by the
excess temperature above the melting point necessary to achieve solder spreading
Superheat, C above melting point
O2 level O2 level
Solder 10 ppm 100 ppm
Sb-95Sn 6 15
Cu-99Sn 3 7
Ag-96Sn 9 17
Pb-63Sn 22 24
In-48Sn 83 No spread
<300 C (<570 F)
Sn-9Zn No spread ...
<300 C (<570 F)
not an effective substitute for ux unless the sol-
dering process can be undertaken at very high ex-
cess temperatures.
In an investigation of the oxidation and re-
duction kinetics of selected lead-tin, gold-tin,
and indium-tin eutectic solders, it was possible
to use hydrogen to achieve self-alignment of
solder bumps during ip-chip interconnection
[Kuhmann et al. 1998] (further information on
ip-chip interconnection can be found in Chap-
ter 5, section 5.2). However, in order to reach
this satisfactory result it was necessary to achieve
hydrogen/water ratios in excess of the limiting
values derived from the Ellingham diagram.
The investigators satised the requisite condi-
tions by pumping down the soldering oven,
which was an ultrahigh vacuum (UHV) system,
to a pressure of 109 Pa (1.5  1013 psi). This
pressure will have dened the residual partial
pressure of water vapor present. Then, the intro-
duction of sufficiently pure hydrogen at a pres-
sure 1 Pa (1.5  104 psi) was able to satisfy the
thermodynamic requirements for reduction of
oxides, including that of indium, at a suffi-
ciently fast rate that the oxides were actually
removed within a few minutes at normal pro-
cess temperatures. While the ability to achieve
the stringent conditions required for the reduc-
tion of solder oxides by hydrogen has been
demonstrated in a laboratory context, it is doubt-
ful whether this would be realistic or economic
in a manufacturing environment.
3.3.4 Reduction of Solder
Oxides by Atomic Hydrogen
Although molecular hydrogen is not practica-
bly capable of reducing solder oxides at normal
Chapter 3: The Joining Environment / 127
O2 level O2 level
1,000 ppm 10,000 ppm
18 No spread
<300 C (<570 F)
18 No spread
<300 C (<570 F)
19 No spread
<300 C (<570 F)
87 No spread
<300 C (<570 F)
... ...
... ...
soldering temperatures, the same does not apply
to atomic hydrogen. Atomic hydrogen is far
more reactive than the molecular species. In
essence, the energy provided externally to split
the diatomic hydrogen molecule into single at-
oms then becomes available for accomplishing
chemical work. Atomic hydrogen reacts with
surface oxides on solders and joint surfaces to
form hydroxides, hydrogenated complexes, or
water, all of which are volatile at typical solder-
ing temperatures and can therefore easily be
removed from the area of the joint. Molecular
hydrogen can be dissociated into atomic hydro-
gen using a variety of methods including a
heated lament, photo dissociation, electrical
discharge, microwave radiation, and, probably
the most practical, low-frequency alternating
current ionization. Commercial atomic hydro-
gen soldering systems are available from at
least one vendor [ATV 2003].
The concentration of any monatomic hydro-
gen produced decays rapidly with distance from
the source so that it must be generated near the
workpiece. However, because atomic hydrogen
is such a reactive species toward solder oxide, a
commercial argon/hydrogen gas mixture can be
used to feed the plasma generator and at a hy-
drogen concentration that is sufficiently low to
obviate the need for hydrogen-monitoring equip-
ment and the associated safety features.
Because the species in a plasma are not in
thermodynamic equilibrium, application of
chemical thermodynamics to the analysis of such
a system requires modications. A quantitative
process description has been formulated that has
allowed a line corresponding to the following
chemical reaction to be added to the Ellingham
diagram [Jacob, Chandran, and Mallya 2000]:
128 / Principles of Soldering
4H  O2  2H2O
Under standard conditions, monatomic hydro-
gen can reduce almost all metal oxides at tem-
peratures below 500 C (930 F). Currently, it is
not possible to produce monatomic hydrogen at
atmospheric pressure (Po  105 Pa, or 15 psi),
and the dashed line included in Fig. 3.18 repre-
sents the reduction potential of atomic hydrogen
at a base pressure of 1 kPa (0.15 psi). To exploit
fully the reduction potential of monatomic hy-
drogen, it is necessary to remove the products of
reaction and prevent back-reaction resulting from
the formation of molecular hydrogen and oxide.
This is achieved using a ow system where the
hydrogen gas is passed through a dissociator and
then over the workpiece before being transported
away.
Fig. 3.16 Contact angle for In-48Sn solder on four metal
substrates, after heating for 5 min at 250 C (482
F), as a function of oxygen partial pressure. Adapted from Preuss,
Adolphi, and Werner [1994]
Fig. 3.17 Reduction rate of solder oxides in hydrogen, containing 10 ppm total oxygen and water vapor, for a range of solders as
a function of process temperature. Adapted from Dong, Schwarz, and Roth [1977]
Using monatomic hydrogen, ller alloy and
component surfaces can be cleaned in situ and
excellent quality uxless joints obtained at nor-
mal soldering temperatures. The process cycle
time can be less than 1 min, the time being lim-
ited more by the mechanics associated with pass-
ing the part through a heated zone containing the
monatomic hydrogen rather than by the kinetics
of the reduction process itself. Figure 3.19 is an
example of application of hydrogen plasma to
lead-tin solder using a commercial solder reow
oven equipped with a low-frequency plasma gen-
erator. Similar ndings are reported in the lit-
erature [Hong and Kang 2001].
Reduction of metal oxide by hydrogen plasma
has been followed by direct measurement of the
oxygen/silver ratio as a function of process time
(see Fig. 3.20). The substrate was 40 nm (1.6
in.) thick silver, the base pressure 1 Pa (1.5 
104 psi), and the plasma intensity 1 eV. The
kinetic energy of the hydrogen ions was below
the threshold for sputtering processes, verifying
the chemical nature of the reductive mechanism
[Bielmann et al. 2002].
3.3.5 Mechanical Removal of
Oxides (Ultrasonic Soldering)
The native oxides on the surfaces of the solder
and components can, of course, be removed
physically. Although an obvious form of me-
chanical abrasion is to use abrasive paper or a
scalpel, more rened and controlled methods ex-
ist. For example, fast atom bombardment using
a neutral beam can mill surfaces at rates of tens

of nm per minute. The effectiveness of the pro-
cess is demonstrated in Fig. 3.21, which refers to
ip-chip joints made with high-lead Pb-3Sn sol-
der. In this graph, joint strength is held to be an
indicator of surface oxide thickness. The simi-
Fig. 3.18 Simplied Ellingham diagram for oxides with a
line (dashed) on it representing the reduction po-
tential of monotomic hydrogen at a partial pressure of 1 kPa (PO
is dened as this pressure as it is not possible to generate mona-
tomic hydrogen at atmospheric pressure by any method currently
available.)
Fig. 3.19 Balls of lead-tin solder, approximately 100 m di-
ameter, reowed using an atomic hydrogen
plasma. The solidied surface is exceptionally smooth and shiny
and, at high magnication the duplex phase microstructure solder
can be readily seen, attesting to the surface cleanliness. Source:
Agilent Technologies
Chapter 3: The Joining Environment / 129
larity of form between Fig. 3.21 and Fig. 1.27
(showing the effect of mechanical wiping) is
noteworthy as they are essentially the same
physical process. It is reported that an addition
of 10% hydrogen into the sputtering plasma con-
siderably boosts the oxide-removal rate. It is
likely that the hydrogen ions contribute to the
oxide removal through a chemical mechanism,
as discussed in the preceding section [Hong,
Kang, and Jung 2002].
The main drawback of all mechanical meth-
ods of cleaning surfaces, which is illustrated in
Fig. 3.22, is that the oxide lms promptly regrow
if the parts are exposed to air or, indeed, left in
a nitrogen atmosphere of even the highest quality
for more than a few minutes. Therefore, a ux-
less soldering process, not involving an atmo-
sphere containing atomic hydrogen, must be able
to cope with a thin layer of native oxide on, at
least, the solder. Strategies for making a robust
uxless soldering process are given in the fol-
lowing sections.
An automated version of physical abrasion to
remove surface contamination is ultrasonic agi-
tation, which is more correctly called ultrasonic
tinning. This technique is highly effective in re-
moving even the most tenacious oxides and is
therefore applicable to virtually all metals, but is
most widely used with aluminum [Jones and Tho-
mas 1956]. It is not a new technology, and there
exists a patent reference to an ultrasonic solder-
ing iron that predates World War II [Antonevich
1976]. Ultrasonic uxing normally involves ap-
plying an ultrasonically activated soldering iron
to tin the workpiece [Schaffer et al. 1962]. In a
variant arrangement, a solder bath is ultrasoni-
cally excited and the workpiece is dip coated
with the solder. These ultrasonic systems typi-
cally operate at 20 to 80 kHz and 0 to 300 W of
electrical power. Most systems operate at a single
frequency, but some provide for application of
high- and low-frequency acoustic energy simul-
taneously. A more recent development is to in-
corporate an ultrasonic transducer into wave-
soldering equipment. The ultrasonic sonodes are
an integral part of the solder wave and impart
sonic action while the board travels through the
wave [Swanson 1995]. In operation the sonodes
remove the oxide from the component leads and
PCB lands, thereby enhancing wetting of the
solder. The oxide removal is accomplished by
the cavitation action of the ultrasonic waves on
the oxidized surfaces. The system operates with
a frequency in the 20 kHz range and an ampli-
tude in the micron (sub-mil) range. These pa-
130 / Principles of Soldering
rameters are selected in order not to adversely
affect the integrity of the components and circuit
board material.
Ultrasonic uxing is a uxing and tinning op-
eration for applying a metal coating to a clean
surface of the workpiece. Once the metal coating
has been applied, it protects the workpiece sur-
face against reoxidation, and for that reason is
akin to chemical uxing, as described previ-
ously. The coated part can subsequently be sol-
dered (or brazed) using a conventional process.
At present, the method is limited to tinning of
passive electronic and optical devices. Active
devices generally do not respond well to expo-
sure to high-intensity acoustic waves.
Vianco and Hoskin [1992] have studied the
wetting of copper by molten tin when assisted by
a point source of ultrasonic energy (20 kHz, 20
70 W). Some of their results are reproduced in
Fig. 3.23. Visual inspection of ultrasonically
tinned substrates revealed tin lms on both the
front and back surfaces of the copper coupons.
This implies wetting is not solely due to direct
mechanical erosion by the incident pressure
wave, but also secondary propagation within the
substrate and molten metal.
Ultrasonic uxing is attractive as a combined
method of uxing and tinning since it is uxless
and residue-free. With the progressive improve-
ments being made to the understanding of the
process and the push toward greener and
cleaner manufacturing, ultrasonic soldering is
reentering more widespread use, including for
Fig. 3.20 Reduction of silver oxide by atomic hydrogen generated in proximity to the compound using a plasma. The energy of
the plasma was deliberately chosen to be below the threshold at which oxide removal could be attributed to sputtering
processes.
applying metallizations to refractory materials
such as alumina [Naka and Okamoto 1989].
3.3.6 Reactive Gas Atmospheres for
Reduction of Oxides
Working within the denition of a uxless
process being one that does not use a solid or
liquid ux, it is possible to include the option
of using reactive gases to chemically remove
surface oxide lms. A number of gases are suf-
ciently active to be capable of reducing even
indium oxide. Reactive atmospheres that have
been examined and reported in the literature in-
clude formic acid and acetic acid vapor, carbon
monoxide, and halogen gases (CF2Cl2, CF4, SF6,
etc.) [Moskowitz, Yeh, and Ray 1986]. Formic
and acetic acid are probably the most commer-
cialized reactive gas for uxless soldering. Car-
bon monoxide and halogen gases require rela-
tively high temperatures for the reduction of tin
oxides in any reasonable timescale, for example,
350 C (660 F) maintained for 3 to 5 min, which
precludes their use in many soldering processes.
In this regard, these gases are similar to hydrogen
gas.
Formic acid vapor is active at typical solder-
ing temperatures, and the process implementa-
tion is relatively simple. Its effectiveness can be
judged from Fig. 3.24, which illustrates the im-
proved self-alignment of ip-chip assemblies as
formic acid is introduced at progressively higher
concentrations, to the point of reducing surface

oxide on the solder spheres [Lin and Lee 1999;
Deshmukh et al. 1993].
The reaction between gaseous formic acid and
metal oxide can be described by:
MO  2HCOOH  M(COOH)2  H2O
(where M is the metal)
This reaction occurs at a signicant rate at tem-
peratures above about 150 C (300 F) and is
therefore effective at stripping surface oxide lms
prior to the solder melting. When the tempera-
ture is higher than about 200 C (400 F), the
metal formate decomposes to liberate carbon di-
oxide and hydrogen:
M(COOH)2  M  CO2  H2
The concentration of formic acid vapor in the
atmosphere needs to exceed 0.5% in order to
have a useful practical effect. Appropriate pre-
cautions need to be exercised with this process,
as many organic metal compounds, especially
Fig. 3.21 Effect of fast atom cleaning on the strength of high
melting point Pb-3Sn solder joints made without
breaking vacuum between cleaning and soldering. Adapted from:
Kohono et al. [1996].
Fig. 3.22 Effect on joint strength of deliberate delay, in a
high-vacuum environment, between mechanical
cleaning and assembly for high-lead solder (Pb-3Sn) ip-chip
bonds
Chapter 3: The Joining Environment / 131
those of lead, are toxic. Appropriately specied
materials are required for furnacing equipment
capable of utilizing formic acid vapor, owing to
its corrosiveness, but the price premium appears
to be relatively small and off-the-shelf equip-
ment designed to work with it can be readily
purchased.
Acetic acid vapor is effective at reducing tin
oxide on solder. This is evident by the difference
in the time to wetting for Pb-62Sn solder on
copper and gold-coated nickel (see Fig. 3.25).
Short wetting times are always obtained on the
clean gold/nickel metallization, but longer pro-
cess times or higher concentrations of acid are
necessary to clean bare copper and obtain similar
rapidity of wetting [Frear and Keicher 1992].
3.3.7 Surface Conditioning Processes
As an alternative to cleaning the faying sur-
faces on the components and the ller metal in
situ, processes have been developed that convert
the surface layers of oxide to stable compounds
that protect against reoxidation and that are
readily removed during the soldering cycle.
One such process exploits uorine chemistry.
This is the PADS (plasma-assisted dry soldering)
process developed at The University of North
Carolina. It uses a plasma-assisted reaction to
dissociate CF4 or SF6 within an appropriate car-
rier gas to produce monatomic uorine, which,
it is claimed, converts tin oxide to tin oxyuor-
ide:
SnOx  yF  SnOxFy
The reaction time is a few minutes in a re-
duced-pressure atmosphere, and the component
Fig. 3.23 Wetted area of copper coupons by molten tin at
245 C (473 F) as a function of ultrasonic power
at 20 kHz and a process time of 30 s. The degree of wetting was
also found to be directly proportional to the process time for the
process temperature of 245 C (473 F), but the relationship was
less clear at other test temperatures. Adapted from Vianco and
Hoskin [1992]
132 / Principles of Soldering

temperature need not exceed 50 C (122 F).
Treated parts can be stored in air and on subse-
quent heating in a nonoxidizing atmosphere the
tin uoride effectively volatilizes, leaving clean
metal surfaces that are readily soldered. The sol-
derability shelf life for parts treated in this manner
is up to 1 week when stored in normal laboratory
air or 2 weeks in a nitrogen-purged desiccator
[Marczi, Bandyopadhay, and Adams 1990].
Although pH-buffered chemical treatments are
available commercially that are capable of modi-
fying oxide lms on copper, few are available
that work well with solders or other base metals.
One innovation that appears able to overcome
this limitation is the ROSA (reduced-oxide sol-
dering activation) process (Rockwell Scientic).
In this process, an acidic solution containing va-
nadium ions and associated hydrogen is used to
reduce oxides. The vanadium ions, which also
supply hydrogen ions by virtue of their multiple
valency states, are regenerated electrochemi-
cally in a closed loop, so the only effluent is
oxygen gas. This process only removes surface
oxides, and therefore the treated components
must be used before the oxide substantially re-
grows [Tench et al. 1995].
3.3.8 Fluxless Soldering
Processes Considerations
While the methods described in the preceding
sections may be used to minimize the inhibiting
effect of surface oxides on soldering, even when
they are applied in combination it is normally not
possible to make a uxless soldered joint of the
same quality as a uxed joint. The reason is
simply that these agents are not as pervasive at
reaching the critical interfaces in the assembly
and are not as active as a liquid chemical in
rendering all metal surfaces clean at the solder-
ing temperature and simultaneously protecting
the cleaned surfaces from reoxidation. Addi-
tional means are therefore necessary to encour-
age wetting and spreading of molten ller metals
during uxless processing. Generally, such meth-

Fig. 3.24 Effect of formic acid vapor concentration (0.35 to 1.7%) on the pull-in alignment of ip-chip components deliberately
misaligned by 30 m (1.2 mil). Adapted from Deshmukh et al. [1993]

Fig. 3.25 Fluxless wetting of Pb-62Sn solder on copper and gold-on-nickel metallizations, using acetic acid vapor of varying
concentration as an active reductant in the atmosphere

ods involve utilizing nonoxidizable metalliza-
tions on the parent materials, minimizing the
surface-area-to-volume ratio of the ller metal,
applying mechanical means of enhancing solder
ow and, sometimes, metallurgical modication
of the contact angle.
3.3.8.1 Solderable Component Surfaces
Because oxide lms grow so rapidly on most
base metals, as shown in Fig. 3.13, in practice
clean and solderable component surfaces can
only be achieved if they are of gold. Accord-
ingly, for uxless soldering, a gold coating is
generally applied that will be of adequate thick-
ness and sufficiently pore-free to ensure good
solderability over its specied storage life. It is
important to bear in mind the fact that the
solderable shelf life afforded by a gold coating
will be a function of the roughness of the under-
lying surface, the method of application of the
coating and its thickness (see Chapter 4, section
4.1.2). Thus, a 0.5 m (20 in.) thick gold layer
deposited by sputtering can be relied on to
maintain excellent solderability for several
months, even where the coated surface is rela-
tively rough. On the other hand, the solderabil-
ity of a gold layer of the same thickness, but
applied by electroplating to a rough surface (Ra
> 3 m, or 120 in.) may not offer adequate
protection to an underlying base metal from
atmospheric oxidation for more than a few days.
The use of gold-coated surfaces imposes con-
straints on the solders that can be used. In
particular, most tin-base solders, including lead-
tin eutectic, are largely incompatible with thick
gold metallizations, on account of the high solu-
bility of tin in gold, which results in the forma-
tion of AuSn4 and consequential embrittlement
of the joints if this phase becomes dominant.
This topic is discussed in further detail in Chap-
ter 2, section 2.3.2. This restriction can only be
overcome by applying high-quality gold coat-
ings of less than a certain thickness to the joint
surfaces, so as to prevent the formation of
AuSn4 as the primary phase. Where thicker gold
coatings are used, uxless soldering tends to be
conned to high-gold and indium-base solders,
which do not form catastrophic embrittling
phases with gold.
The necessity for a noble metal surface on
the parent materials can be eliminated if the
solder itself is applied as the barrier coating to
Chapter 3: The Joining Environment / 133
previously cleaned component surfaces. This
approach is considered in the next section.
3.3.8.2 Preform Geometry
The form in which solder is admitted into a
joint gap can make a profound difference to the
success of uxed and especially uxless solder-
ing processes. Notwithstanding the condition of
the faying surfaces, it is a general rule that the
greater the solder volume is in relation to its
exposed surface area, then the more readily the
process will work. This is simply because there
is proportionally less oxide to impede wetting
and spreading. It is therefore perhaps not sur-
prising that manufacturers of solder paste go to
quite considerable lengths to ensure that the
solder balls used in its preparation are perfectly
spherical, have exceptionally high surface
smoothness (low Ra), and that the most com-
mon form of solder is round wire.
The most appropriate geometry of preform
for admitting solder into a joint gap depends on
the shape of the components being joined. Ide-
ally, the solder preforms should be designed to
have not only a high volume-to-surface-area
ratio but also to be orientated so that the ad-
vancing front of molten solder will sweep trapped
gas out of the joint gap (see Chapter 4, section
4.3.1.1).
In situations where it is desired to make
joints with very low aspect ratiothat is, thin
in relation to the plan areait is often not
possible to achieve sufficient solder spread to
reliably obtain complete joint lling. This sce-
nario is encountered commonly in the micro-
electronics and photonics industries where there
is a need to join planar components but with
extremely narrow joint gaps on account of the
relatively poor thermal conductivity of most
solder alloys compared with many other metals.
The thinnest solder preforms that can be eco-
nomically purchased are 15 m (0.6 mil) thick.
Handling these foils is almost an art form, and
mechanically cleaning them is virtually impos-
sible. The high surface-area-to-volume ratio of
such preforms also runs counter to the need to
minimize native oxides.
A growing number of companies are now
offering a solution to this problem in the form
of substrates that are sold with the solder com-
position of choice preapplied to surfaces. The
solder is usually applied by either electroplating
134 / Principles of Soldering

or a vapor-phase technique, through a mask, so cased in a skin of oxide. If it is possible to extrude

that only the required areas of the substrate are virgin metal through ssures or other defects in
coated. The solder thicknesses available range the solder, then there is an improved prospect of
from 2 to 50 m (0.08 to 2 mil), and almost achieving a sound joint. One method of doing
every common composition is available. this is to apply a compressive force to the joint
These solder coated substrates offer a number gap.
of distinct advantages compared with a solder The effectiveness of this approach is illus-
foil or wire: trated in Fig. 3.26. Clearly, the higher the com-
pressive pressure, the more effective it is, with
Piece-part inventory and number of suppliers the optimal load in the region of 10 g/mm2 (14
are reduced by dispensing with preforms.
psi), or more. A loading such as this is relatively
Jigging is likely also to be simpler. easy to achieve with weights or spring-loaded
The thickness of the solder joint is decreased jigs for all but the largest components. Precau-
because the solder layer can be substantially
tions need to be taken to ensure that the load is
thinner than the minimum practicable thick-
applied uniformly and parallel to the joint gap,
ness of approximately 25 m (1 mil) required
and that the method of application does not pose
for a solder preform.
a thermal sink on the assembly that would give
Soldering behavior is improved by eliminat- rise to adverse temperature gradients.
ing two joint surfaces with all the attendant
Figure 3.27 shows application of this ap-
problems, including oxide layers, from the
proach to uxless soldering of a GaAs semicon-
joint gap.
ductor die to a gold thick-lm metallized alu-
Solder spread is automatically conned to a mina substrate. The back of the semiconductor is
precise area, and the responsibility of sub-
metallized with gold, and the ller metal is a
strate wettability is also passed on to the sub-
preform of Ag-96Sn solder, 15 m (0.6 mil) thick.
strate producer.
The process superheat was 30 C (86 F). X-
The quality of substrates prepared in this man- radiography reveals the joint to be free of voids
ner has improved substantially, and, although (the black circles are blind vias in the GaAs,
there is a price premium compared to traditional which are included for functional reasons).
foil, they are now available to aerospace and
telecommunications qualied standards.
3.3.8.4 Metallugically Enhanced Solder
Because the success of a uxless soldering
process is critically reliant on the absence of Flow
surface oxide on the molten solder, it is good Occasional reports appear in the published lit-
practice to test whether this condition is being erature of the signicant difference that low con-
achieved. This may be readily accomplished if centrations of other metals can make in promot-
the furnace has a viewing port. On heating to ing wetting and spreading of molten solders.
the process temperature, a source of clean sol-
der will melt and its surface acquires a shiny,
mirrorlike nish. This is often referred to as the
liquid lake condition. Any defects in the liq-
uid solder lm, such as texture, a gray bloom,
or brown spots indicate that some element of
the process, usually either the solder itself or
the reow atmosphere, is in some way decient
and must be corrected.

3.3.8.3 Mechanically Enhanced Solder

Flow
No matter what precautions are taken, in an
industrial uxless soldering process it will al-
ways be the case that by the time the components
and preform have been set in jigs, loaded into the
protective atmosphere, and heated to the process
temperature, then the solder will be totally en-
Fig. 3.26 Void level (as a percentage of the plan area of the
joint) versus the load applied to preforms of three
solders during uxless processing. The soldering conditions used
for the three solders were a superheat of 25 C (or 77 F) and dry
nitrogen as the joining atmosphere. The joints measured approxi-
mately 10 mm  10 mm  25 m (0.4 in.  0.4 in.  1 mil).

Some cases are cited in Chapter 2, section 2.2
and Chapter 5, section 5.8. Virtually all uxless
aluminum brazes make use of the fact that ppm
levels of bismuth and beryllium have a marked
effect on the wetting and spreading characteris-
tics of the Al-12Si eutectic alloy. Further infor-
mation on uxless aluminum brazing is to be
found in the planned companion volume Prin-
ciples of Brazing.
The authors have investigated improving the
uidity of indium by minor additions of other
elements with a view to enhancing the use of
indium ller metals in uxless processes. Figure
3.28 shows the wetting angle as a function of
heating time for pellets of indium solders of con-
trolled weight and geometry wetted on to silver
substrates at 200 C (392 F) in a vacuum of 1
mPa (1.5  107 psi). The measurements were
taken from video stills of the substrate viewed
edge-on, with the timeline beginning at the onset
of observed melting. Of the additions, zinc has
an adverse effect. Antimony is initially neutral,
but appears to impede any additional wetting.
Bismuth and gold are benecial additions that
improve wettability, which requires an extended
dwell at the soldering temperature to take effect,
while the rare earth metal cerium produces a
marked and immediate reduction in contact
angle. This could be a fruitful area for further
research in developing new solder alloys (see
Chapter 5, section 5.8).
3.3.9 Example of a Fluxless Soldering
Process Using In-48Sn Solder
The requirement was to improve the quality a
soldered joint that was being made using a 25 m
(1 mil) thick preform of the In-48Sn solder, with-
Fig. 3.27 Fluxless soldering of a GaAs monolithic micro-
wave integrated circuit, approximately 3  5 mm
(0.12  0.12 in.) achieved by application of a compressive load
of 100 g/mm2 (140 psi) during the heating cycle. Source: BAE
Systems
Chapter 3: The Joining Environment / 135
out resorting to the use of chemical uxes. The
joint measured approximately 10 by 5 mm (0.4
by 0.2 in.), and the components were both met-
allized with 3 to 5 m (120 to 200 in.) of nickel
overlaid with 3 m (120 in.) of gold. In the
industrial environment where this work was car-
ried out, it was necessary to demonstrate the
benecial effect of each change in the process
before other alterations could be investigated.
Shear strength was used to assess the quality of
the soldered joints.
The rst change to be considered was to apply
a load to the joint gap during the heating cycle.
Application of a compressive stress helps ensure
that all free surfaces within the joint are in in-
timate contact. This has the effect of minimizing
temperature gradients across the joints and, as
explained previously, promotes rupturing of the
oxide skin on the solder when it melts.
The data shown in Fig. 3.29 (air) indicates the
relationship between the applied load and result-
ing joint strength, with the joint shear strength in-
creasing by roughly 1 kg (2.2 lb) for every 100 g
(0.2 lb) of applied weight, for the range of applied
pressure shown (0 to 6 g/mm2, or 0 to 75 psi).
A more signicant improvement to the pro-
cess was made by carrying out the joining op-
eration in a furnace maintained under dry nitro-
gen, giving a combined oxygen and water vapor
content of less than 10 ppm in the working at-
mosphere. This measure provided some protec-
tion against oxidation of the parts and solder and
delivered a stepwise improvement in joint
strength; see Fig. 3.29 (nitrogen). It was ob-
Fig. 3.28 The effect of different doping additions on the ux-
less wetting angle of indium on silver substrates, as
a function of time following the commencement of melting of the
solder
136 / Principles of Soldering
served that as a result of changing to the nitrogen
atmosphere, considerably less solder was exuded
from the joint gap, implying that the increase in
joint strength is largely due to improved wetting
of the component surfaces.
Hitherto, the foil preforms of solder had been
used as received from the supplier. These are
covered with a signicant layer of native oxide.
Mechanical abrasion of the preform surfaces with
a glass-ber brush immediately prior to use led
to a further improvement in joint strength, as can
be seen in Fig. 3.29 (cleaned).
A thin foil solder preform is not a particularly
attractive method of admitting solder into a joint
because:
The oxide surface area-to-solder volume ra-
tio is unfavorable.
The foil is difficult to handle and clean if thin.
There is a high risk of trapping pockets of
vapor when the parts and foil are jigged to-
gether (see Chapter 4, section 4.3.1.1).
Thin foils also command a cost premium, espe-
cially when they have to be purchased to custom
dimensions.
Solder in the form of round wire avoids these
drawbacks. The preform was prepared from two
equal lengths of wire, cold compression welded
so as to form a symmetrical cross, as shown in
Fig. 3.30. The surface area-to-volume ratio of
this cruciform conguration is much lower than
for a foil, and wiping the wire several times with
a paper tissue soaked in solvent is an easy method
of striping the solder oxide (see Fig. 1.27). The
benecial effect of substantially removing the
oxide lm from the solder by these means is
attested by the results presented in Fig. 3.29
Fig. 3.29 Shear strength of joints approximately 10  5 mm
(0.4  0.2 in.) in area made uxless using In-48Sn
solder at process temperature of 150 C (302 F) to gold-
metallized components, as a function of the applied compressive
load, showing also the effect of atmosphere quality and condition
of the solder (see text for details)
(wire). The cruciform conguration also pro-
vides for mechanical stability during jigging, and
the direction of solder ow after melting helps
sweep out any trapped gas from the joint gap (see
Chapter 4, section 4.3.1.1).
Examination of the joint surfaces of disas-
sembled parts revealed good wetting of the com-
ponents and a continuous bead of shiny solder
formed at the edge of the joint.
3.3.10 Fluxless Soldering of
Aluminum
From the preceding discussion it might be as-
sumed that to attempt uxless soldering of alu-
minum would be ineffectual. However, uxless
soldering of aluminum is practiced as a repair
technique. It was developed during World War II
as a method of patching small bullet holes in
airplane skins. The original process simply used
zinc as the joining material, but the joints were
weak and susceptible to corrosion. The modern
variation employs a ller alloy containing about
90% Zn, 7% Al, with the balance being magne-
sium, manganese, and other elements. The alloy
melts at approximately 380 C (716 F), and
therefore it qualies as a solder. The joining
method involves heating the aluminum part, of-
ten with a standard blowtorch and puddling (rub-
bing) the solder until a wetted surface lm is
developed over the area to be joined. No ux is
necessary because the alumina is physically re-
moved and the exposed aluminum overcoated
with a more oxidation-resistant alloy. Both joint
surfaces are prepared in this manner, placed in
Fig. 3.30 Cross-shaped preform of In-48Sn solder prepared
by cold welding of two 300 m (12 mil) diam wires
at the common intersection. Source: BAE Systems

contact and the joint made by reheating without
further addition of the ller metal [Phillips 1994].
Additional mechanical agitation helps meld the
two molten liquid skins together. The aluminum
in the solder helps to prevent corrosion of the
joint, the zinc addition lowers the melting point,
while the minor elements assist wetting and
spreading. It is reported that joints made in this
manner can have shear strengths around 90% of
that of the parent metal.
If the layer of applied zinc-base solder can be
made extremely thin, it is possible to diffuse it
completely into the base material if the process
cycle is suitably extended. This technique is prop-
erly known as diffusion soldering, but it does en-
able uxless, effectively corrosion-resistant
joints to be made to aluminum, which are not a
source of mechanical weakness [Ricks et al 1989].
The applicability of the approach is limited by the
long heating cycle times and the need to either use
precisely machined parts or apply pressure to
force the abutting surfaces into intimate contact.
This latter requirement arises from the very lim-
ited volume of zinc in the joint and hence the small
quantity of liquid ller metal available to ll the
joint gap.Adiscussion of diffusion soldering, is to
be found in Chapter 5, section 5.9 and diffusion
brazing in the planned companion volume Prin-
ciples of Brazing.
Appendix A3.1:
Thermodynamic
Equilibrium and the
Boundary Conditions for
Spontaneous Chemical
Reaction
The thermodynamic function that provides a
measure of the driving force of a chemical (in-
cluding metallurgical) reaction is the Gibbs free
energy, which is dened as:
G  E  PV  TS (Eq A3.1)
where E is the internal energy, S the entropy, T
the absolute temperature, P the pressure, and V
the volume of the materials system. The deni-
Chapter 3: The Joining Environment / 137
tion and physical meaning of internal energy and
entropy are explained below. As is shown, an
important property of the Gibbs free-energy func-
tion is that it is always a minimum at equilibrium,
and the extent of its departure from the minimum
value provides a measure of the tendency of a
reaction to proceed spontaneouslythat is, of
the driving force for the reaction.
The First Law of
Thermodynamics and
Internal Energy
The subject of thermodynamics addresses en-
ergy changes in systems. In thermodynamics, the
term system is used to describe a set of ma-
terials that are capable of undergoing a change
as, for example, through a chemical reaction.
The First Law of Thermodynamics is a state-
ment of the Principle of Conservation of Energy.
The various statements of this law are bound up
with the differentiation of various types of en-
ergy and, in particular, with the concept of in-
ternal energy. The internal energy of a system
may be considered as the aggregate of the kinetic
energies and energies of interaction (i.e., poten-
tial energies) of the atoms and molecules of which
the constituent materials are composed. When
the system is isolated from its surroundings, so
that no exchange of energy can take place, then
its internal energy remains xed. However, if
mechanical work can be done on the system, but
no heat is exchanged with its surroundings (i.e.,
the system is adiabatically isolated)for ex-
ample, by an impellor stirring a liquid or gas in
an insulated containerits internal energy, E,
will change by an incremental amount equal to
the work, W, performed:
dE  dW (adiabatic)
The minus sign denotes that work is done on the
system to raise its internal energy. This expres-
sion provides a thermodynamic denition of in-
ternal energy.
The internal energy of a system depends only
on the state of the system (dened in terms of
macroscopic or thermodynamic properties, such
as the pressure and temperature of the system).
For this reason, internal energy is termed a func-
tion of state.
138 / Principles of Soldering
Where work is done in changing the volume
of a chemical system by an increment dV through
the application of external pressure P, then:
dW  PdV
dE  PdV
In practice, most systems are not totally insulated
from their surroundings so that thermal energy
may be exchanged between them.
If an increment of work dW is done on the
system and an increment of heat dQ is exchanged
with the surroundings, then the internal energy
dE will change by the amount:
dE  dQ  dW
This equation is a mathematical expression of
the First Law of Thermodynamics which, for a
chemical system, may be written:
dE  dQ  PdV (Eq A3.2)
Entropy and the Second
Law of Thermodynamics
Internal energy alone cannot determine the
equilibrium state of a system. Although when a
system reaches a state of equilibrium the internal
energy achieves a xed value, this may not be a
minimum. For example, the internal energy will
increase when a solid melts at constant tempera-
ture and pressure through the absorption of latent
heat. For this reason, in addition to the internal
energy, it is necessary to stipulate the value of
another state function of the systemnamely
entropywhich, together with the internal en-
ergy, measures the extent to which the system is
removed from equilibrium.
The concept of entropy arises in connection
with the conversion of heat into mechanical work
and vice versa. The Second Law of Thermody-
namics denes the conditions under which this
conversion from one form of energy to another
can occur. The Kelvin-Planck statement of this
law relates to a device that can perform work by
extracting heat from a particular source and per-
forming an equivalent amount of work, without
any other energy exchange with the surround-
ings. It follows that a reciprocating engine that
operates by extracting heat from one source must
reject some of this heat to a sink at a lower
temperature. If the operating cycle of the engine
is reversible, such that work can be performed to
pump heat from the sink back to the source, it is
possible to show that in accordance with the
Second Law, the integrated ratio dQ/T over one
complete cycle is zero:
dQT  0
R
The circle through the integral sign denotes that
the integration is to be carried out over the com-
plete cycle, and the letter R is a reminder that the
equation applies only if the cycle is reversible.
This result is known as Clausius theorem.
If the integration is carried out over only part
of the cycle, say between two states 1 and 2, then
the integrated ratio dQ/T is not zero, but equals
the difference between the values of a thermo-
dynamic function at the two states:
 dQ
2
 S1  S2
R 1 T
This thermodynamic function of state is called
entropy. If the two states are innitesimally close,
then the relationship can be written:
( )
dQR
T R
 dS (Eq A3.3)
Subscript R indicates that this equation only
holds if the heat increment dQ is transferred re-
versibly. This equation provides a mathematical
expression of the Second Law of Thermody-
namics.
A consequence of the fact that entropy is a
function only of state, a system that has changed
from state 1 to state 2, always has entropy S2,
which differs from that of the initial state S1 by
S1,2  S2 S1, irrespective of the means used to
drive the system. Thus, for example, the system
may have been set in motion, and some of the
kinetic energy converted into heat in overcoming
frictional forces, thereby raising its temperature
to a value that takes the system from state 1 to
 Chapter 3: The Joining Environment / 139

state 2. In this irreversible process, the energy Gibbs free-energy function, G, to determine the
was not supplied to the system as heat so that: temperatures and pressures under which chemi-
cal reactions are thermodynamically favorable,
as well as the direction of the reactions.
 dQ
2
0 In incremental form, Eq A3.1 can be written:
I 1 T

dG  dE  PdV  VdP  TdS  SdT

The letter I denotes that the process is irrevers-
ible.
However, the entropy change S1,2 is still the Substituting for dE and TdS from Eq A3.2 and
same as that obtained by a reversible change A3.3 (all chemical/metallurgical processes being
between states 1 and 2, because it depends only reversible) gives:
on these states and not on the process connecting
them; that is, here too S1,2  S2 S1. Thus, in all
irreversible processes, the entropy change is dG  dQ  PdV  PdV  VdP  dQ  SdT
greater than (dQ/T), where dQ is the heat ab-
sorbed at each incremental step in the irrevers-  VdP  SdT
ible change. This result can be generalized to the
statement that in a spontaneous irreversible For a reversible process at constant tempera-
change, the entropy of an isolated system will ture (isothermal) and constant pressure (iso-
increase, and when in equilibrium, it will remain baric), that is, when the system is in equilibrium:
constant.
Considering the system and its surroundings
together (i.e., the universe), any kind of process dG  0
can be represented in entropy terms by:
and G is constant and has a minimum value. This
is an important result for metallurgical reactions,
dS (universe) 0 because these can be considered as taking place
usually at constant temperature and pressure.
Therefore, from a thermodynamic viewpoint, More generally, at constant pressure, dP  0,
which is macroscopic, entropy can be under- and then:
stood as the propensity of a system to undergo
a change, such as a chemical reaction. A
clearer physical picture of entropy can be ob- dG  S dT
tained at the microscopic level, where a sys-
tem may be regarded as an ensemble of atoms and at constant temperature, dT  0, so that:
or molecules. On this basis, it can be shown
that entropy provides a measure of the degree
of atomic or molecular disorder that exists in dG  V dP
the system, and this will always tend to in-
crease. This concept is consistent with the ob- If the system is an ideal gas, the Gas Law:
servations that all metals are intersoluble, al-
beit in some cases only to a small extent, and
that all liquid metals will wet the clean sur- PV  nRT
faces of solid metals.
applies, where n is the number of moles of gas
and R is the gas constant. Then:
The Dependence of
Gibbs Free Energy on Pressure dG  nRTdP/P at constant temperature
Having dened the thermodynamic func-
tions internal energy, E, and entropy, S, and ex- so that the Gibbs free-energy change resulting
plained their physical signicance, it is pos- from a change from state 1 to state 2 at constant
sible to demonstrate the signicance of the temperature is:
140 / Principles of Soldering

G2  G1  nRT ln P2/P1 Free energy of y moles of solid Y  yG(Y)

The Gibbs free energy, like any other measure  yGo(Y)

of energy, must have some reference point. By
convention, a zero value of G is assigned to the
The Gibbs free energies of the gaseous constitu-
stable form of elements at 25 C (77 K) and 1
ents are:
atm of pressure. Then the Gibbs free-energy
change of a gas at constant temperature from its
value Go at atmospheric pressure, which is de- For a moles of gas A:
ned as its standard state value, is given by: aG(A)  aGo(A)  aRT ln P(A)

For b moles of gas B:

G  G o  nRT ln P (Eq A3.4)
bG(B)  bGo(B)  bRT ln P(B)

where P is the pressure corresponding to the For c moles of gas C:

free-energy state G, expressed in atmospheres. cG(C)  cGo(C)  cRT ln P(C)
Although Eq A3.4 is strictly valid for ideal
gases, it is also approximately applicable to real
gases and can be used for them at pressures close For d moles of gas D:
to normal atmospheric pressure (1 atm). dG(D)  dGo(D)  dRT ln P(D)
In the case of solids, the molar volumes are
small compared with those of gases, so that the where G(A), G(B), etc. are the Gibbs free en-
change in the Gibbs free energy of solids result- ergies of 1 mole of the reagents A, B, etc. at
ing from small pressure excursions, P, such pressures P(A), P(B), etc., and Go(A), Go(B),
that P 1 atm (
100 kPa) at constant tem- etc. are the corresponding values at 1 atm.
perature is small and, to a rst approximation,
may be neglected in reactions involving solids
and gases. It is also assumed that the solubility The free-energy change for the reaction is,
of the gaseous species in the solid phases is neg- from Eq A3.4:
ligible at the temperatures of interest, as is largely
the case in practice.
It is now possible to determine the pressure G  G(products)  G(reactants)
dependence of the Gibbs free energy of the re-
agents that participate in a chemical reaction.
Consider a reaction involving four gases, A, B,  cGo(C)  dGo(D)  aGo(A)  bGo(B)
C, and D and two solids, X and Y, all at constant
temperature T, as follows: P(C)c P(D)d
 RT ln
P(A)a P(B)b
xX  aA  bB yY  cC  dD

P(C)c P(D)d
where a, b, c, d, x, y are the number of moles of  Go  RT ln
each of the reagents. The gaseous reagents are
assumed to behave as though they are ideal gases.
The Gibbs free energies G(X) and G(Y) of the
solid constituents at moderate pressures are ap-
proximately equal to their values at atmospheric
pressure, as explained previously. Therefore:
Free energy of x moles of solid X  xG(X)
 xGo(X)
P(A)a P(B)b
The Gibbs free-energy changes of the solid re-
agents can be neglected, for the reasons given
previously.
Under equilibrium conditions, temperature and
the respective pressures P(A), P(B), and so forth,
are constant, and:
G  0

Hence, the Gibbs free-energy change when the
gaseous reactants A and B in their standard states
are transformed to the products C and D in their
standard states may be expressed in terms of the
partial pressures of the respective reactants in
equilibrium, thus:
Chapter 3: The Joining Environment / 141
per mole of oxygen participating in the reaction.
That is, the driving force needed to oxidize a
metal, as expressed by the Gibbs free-energy
change, is directly related to the oxygen partial
pressure of the atmosphere according to Eq A3.6.

P(C)c P(D)d REFERENCES

G  RT ln
o
(Eq A3.5)
P(A)a P(B)b
Anson, S., et al., 1996. Qualifying a Hi-Rel
Since the Gibbs free-energy change Go, for a
particular reaction at a xed temperature and at
atmospheric pressure has a xed value, so too
does the argument of the logarithm. This con-
stant is called the equilibrium constant KP, be-
cause it can be used to determine the equilibrium
state that a reacting system will attain.
No-Clean Process, Surf. Mount Technol., Vol
10 (No. 8), p 116121
Antonevich, J.N., 1976. Fundamentals of Ul-
trasonic Soldering, Weld. J. Res. Suppl., Vol
55 (No. 7), p 200s207s
ATV 2003. ATV Technologie GmbH, Mu-
nich
Bielmann, M., et al., 2002. The H2 Plasma
Treatment of Silver Contacts: Impact on Wire
Bonding Performance, J. Electron. Mater.,
Vol 31 (No. 12), p 13161320
P(C)c P(D)d Buckley, D., 2000. Nitrogen is the Key to
KP  Lead-Free Wetting, Electron. Prod., Vol 29
a b
P(A) P(B)
The subscript P denotes that the equilibrium con-
stant is specied in terms of pressure.
Equation A3.5 becomes:
Go  RT ln KP
For an oxidizing reaction described by:
(No. 7), p 29
Deshmukh, R.D., et al., 1993. Active Atmo-
sphere Solder Self-Alignment and Bonding
of Optical Components, Int. J. Microcircuits
Electron. Packag., Vol 16 (No. 2), p 97107
Di Giacomo, G., 1986. Oxidation of Pb-Sn
Eutectic Solder and Degradation of Thermal
Contact Resistance, Proc. Symp., Interna-
tional Symposium on Microelectronics, 6 Oct
(Atlanta, GA), p 322328
Dong, C.C., Schwarz, A., and Roth, D.V.,
1977. Effects of Atmosphere Composition on
Soldering Performance of Lead-Free Alter-
natives, Proc Conf., NEPCON West 97, 23
xM  y/2O2 MxOy 27 Feb (Anaheim, CA), p 211221
Ellis, B.N., 1991. Water Soluble Fluxes, Their
Reliability and Their Usefulness as a Means
there is one gaseous constituent and two solids, of Eliminating CFC113 Usage, Solder. Cir-
so that the equilibrium constant is simply: cuit Mt. Technol., Vol 8 (No. 6), p 1623
Esquivel, A.E. and Chavez, E., 1992. Ben-
1 ets of Using Carbonic Gas in the Soldering
KP 

Process and Curing Oven for Electronic As-
M y/2 semblies, Proc. Conf., NEPCON West 92,
PO2
2327 Feb (Anaheim, CA), p 219227
Frear, D.R. and Keicher, D.M., 1992. Flux-
less Soldering Using Activated Acid Va-
where POM2 is the partial pressure of oxygen re- pours, Proc. Conf., NEPCON West 92, 23
quired to effect the oxidation reaction, or the 27 Feb (Anaheim, CA), p 17041715
dissociation pressure of the oxide, and Golubtchik, E.M., 1984. Coating of Magne-
sium, Aluminium and Their Alloys with In-
M
Sn Alloy, Zashch. Met., Vol 20 (No. 2), p
GoRT ln PO2 (Eq A3.6) 286289
142 / Principles of Soldering
Hong, S. and Kang, C., 2001. Fluxless Bump-
ing in Flip-Chip Package by Plasma Reow,
Proc. Conf. International Symposium on
Electronic Materials and Packaging, 1922
Nov (Jeju, Korea), p 139144
Hong, S., Kang, C., and Jung, J., 2002. Flux-
Free Direct Chip Attachment of Solder-
Bump Flip-Chip by Ar  H2 Plasma Treat-
ment, J. Electron. Mater., Vol 31 (No. 10), p
11041111
Hunt, C., 2000. Development of Surface In-
sulation Resistance Measurements for Elec-
tronic Assemblies, Report MATC(A) 070,
National Physical Laboratory
Hyland, K. and Rao, S., 1996. An Environ-
mentally Friendly Manufacturing Solution,
Circuits Assem., Vol 7 (No. 7), p 3840
Jacob, K.T., Chandran, A., and Mallya, R.M.,
2000. An Assessment of the Reduction Po-
tential of Hydrogen Plasma, Z. Metallkde.,
Vol 91 (No. 5), p 401405
Jones, J.B. and Thomas, J.G., 1956. Ultra-
sonic Soldering of Aluminium, Proc. Symp.
Symposium on Solder, 1920 June (Atlantic
City, NJ), American Society for Testing and
Materials, p 1529
Klein Wassink, R.J., 1989. Soldering in Elec-
tronics, 2nd ed., Electrochemical Publications
Kohono, A., et al., 1996. Si Circuit Chip Join-
ing Technology Using Ar Atom Bombard-
ment, Proc. Conf. EuPac 96, 31 Jan2 Feb
(Essen, Germany), p 4147
Kuhmann, J.F., et al., 1998. Oxidation and
Reduction Kinetics of Eutectic SnPb, InSn
and AuSn: A Knowledge Base for Fluxless
Solder Bonding Applications, IEEE Trans.
Comp., Pack. Manuf. Technol., Part C, Vol
21 (No. 2), p 134141
Lea, C., 1991. After CFCsMaking the Eco-
nomic and Technical Choice, Circuit World,
Vol 18 (No. 1), p 2833
Lin, W. and Lee, Y.C., 1999. Study of Flux-
less Soldering Using Formic Acid Vapour,
IEEE Trans. Adv. Pack., Vol 22 (No. 4), p
592601
Manko, H.H., 2002, Solders and Soldering,
4th ed., McGraw-Hill
Marczi, M., Bandyopadhay, N., and Adams,
S., 1990. No-Clean, No Residue Soldering
Process, Circuit. Manuf., Feb, p 4246
Martin, M. and Fromm, E., 1977. Low-
Temperature Oxidation of Metal Surfaces, J.
Alloy. Compd., Vol 258, p 716
McLaughlin, M.C., et al., 1998. The Aque-
ous Cleaning Handbook, Morris-Lee
Publishing Group
Mei, Z. and Morris, J.W., 1992. Characteri-
sation of Sn-Bi Solder Joints, J. Electron.
Mater., Vol 21 (No. 6), p 599607
Moskowitz, P.A., Yeh, H.L., and Ray, S.K.,
1986. Thermal Dry Process Soldering, J. Vac.
Sci. Technol. A, Vol 4 (No. 3), p 838840
Naka, M. and Okamoto, I., 1989. Ultrasonic
Soldering and Brazing of Ceramics to Metal
in Metal/Ceramic Joints, Proc. Conf. Mate-
rials Research Society International Meeting
on Advanced Materials, 23 June (Tokyo),
Vol 8, p 7984
Neale, R., Ed., 2002. Reliability Prediction
and PCB Test Coupons, Electron. Eng. Des.,
April
Nowotarski, M. and De Wilde, R., 1996. The
Effect of Oxygen on the Surface of Solder, Sol-
der. Surf. Mt. Technol., Vol 8 (No. 1), p 2226
Phillips, M., 1994. Cool Progress for Alu-
minium Joins, New Sci., 1948, p 24
Preuss, A., Adolphi, B., and Werner, W., 1994.
Benetzungsverhalten von eutektischem InSn-
lot, Z. Metallke., Vol 85 (No. 11), p 796800
(in German)
Ricks, R.A., et al., 1989. Transient Liquid
Phase Bonding of Aluminium-Lithium Base
Alloy AA8090 Using Roll-Clad Zinc-Based
Interlayers, Proc. Conf. Fifth Int. Aluminium
and Lithium Conf., 2731 March (Williams-
burg, VA), p 441449
Schaffer, H., et al., 1962. How to Ultrasoni-
cally Seal Hermetic Ceramic Transistor Pack-
ages, Ceram. Ind., Vol 79 (No. 6), p 5064
Stubbington, C.A., 1988. Materials Trends in
Military Airframes, Met. Mater., Vol 4 (No.
7), p 424431
Swanson, D., 1995. Ultrasonic Replaces Flux
in Wave Soldering, Electron. Pack. Prod.,
Vol 35 (No. 4), p 26
Tench, D.M., et al., 1995. A New Reduced-
Oxide Soldering Activation Method, J. Met.,
Vol 47 (No. 6), p 3641
The Aluminium Association, 1991. Solder-
ing and Brazing of Aluminium, The Alu-
minium Association
Turbini, L.J., et al., 1991. Characterising the
Corrosion Properties of Flux Residues, Part
1: Test Method Development and Failure
Mode Identication, Sold. Surf. Mt. Technol.,
Vol 8 (No. 6), p 2428
Verite, C., Verbockhaven, D., and Alleaume,
J.F., 1997. Reow Soldering under Nitrogen
Based Atmospheres: Industrial Tests at Mit-
subishi Etrelles, Proc Conf., NEPCON West

97, 2327 Feb (Anaheim, CA), p 13451348
Vianco, P.T. and Hoskin, F.M., 1992. Analy-
sis of Ultrasonic Tinning, Proc Conf.,
NEPCON West 97, 2327 Feb (Anaheim,
CA), p 17181729
Wakelin, R.J., 1993. SMARTA Novel
Lead-Free Soft Solder for Stainless Steel,
Chapter 3: The Joining Environment / 143
Proc. Conf. Innovation Stainless Steel, 1114
Oct (Florence, Italy), p 3.3273.332
Wicks, C.E. and Block, F.E., 1963. Ther-
modynamic Properties of 65 Elements
Their Oxides, Halides, Carbides, and Ni-
trides, Bulletin 605, U.S. Bureau of Mines,
U.S. Government Printing Office
Principles of Soldering
Giles Humpston, David M. Jacobson, p145-187
DOI:10.1361/prso2004p145
CHAPTER 4
The Role of Materials in
Dening Process Constraints
THIS CHAPTER CONSIDERS the materials
and processing aspects of soldering and the man-
ner in which these interrelate in the development
of joining processes.
The starting point of any practical joining
process development is a need to fabricate a
unitary assembly or product from a set of com-
ponents. Often, the components are of different
materials in order to maximize the performance
of the product for a given cost. The product
itself will have been designed to satisfy certain
functional requirements, and, for some items,
these can be very diverse. The joint properties
must also be consistent with the specied pur-
pose and application of the product or system.
To identify and develop an appropriate join-
ing process, it is essential to rst give consider-
ation to all possible aspects associated with the
product. This will reveal an array of constraints,
some of which are not immediately obvious,
that govern the feasibility of the joining route.
A ow chart outlining the decision-making steps
and the constraints at each stage of designing
and manufacturing assemblies containing sol-
dered joints is given in Table 4.1. Those con-
straints that are directly linked to the product
itself include the cost tolerance of the product to
the joining process, the scale and throughput of
production that will have to be satised, and the
statutory regulations that apply. The operating
environment must then be considered. Here, the
peak temperature, stress condition, and the cor-
rosion environment tend to be the critical pa-
rameters. Finally, the materials used in the com-
ponents, when taken individually, will impose
an upper limit on the maximum joining tem-
perature that can be used, on the grounds of
thermal degradation, and may restrict the atmo-
Copyright 2004 ASM International
All rights reserved.
www.asminternational.org
sphere in which the joining process can be
carried out. When the overall assembly is con-
sidered, any mismatch in thermal expansivity of
the abutting components can force compro-
mises with regard to the choice of materials and
processes. All of these considerations will inu-
ence the design of the assembly to a greater or
lesser extent.
Having taken account of the more obvious
constraints in the design of an assembly, a se-
lection of ller alloys can be made, each of
which will impose its own set of limiting con-
ditions. Among the most important of these are
the minimum practicable joining temperature
(i.e., liquidus temperature of the ller alloy,
with the addition of a margin to allow for pos-
sible temperature gradients across the joint), the
geometries that the joint can assume and into
which the ller can be fabricated, and the per-
missible joining atmosphere. From the short-
list of ller alloys, it is usually possible to
select at least one whose window of usable
conditions is compatible with the other steps of
manufacture. Up to this point, the selection
procedure is largely a paper exercise, but the
viability of the proposed joining solution needs
to be established by practical trials. This is
because a multiplicity of other features also
enters the equation, such as wetting, erosion of
the parent materials, and intermetallic phase
formation within the joint. While any of these
phenomena can radically affect the integrity of
the product in service, much of this type of
information is unavailable from the published
literature and cannot be correctly surmised.
If problems are identied at this stage of the
joining process development, it may be possible
to obtain remedies by a number of avenues.
146 / Principles of Soldering

Thus, in certain situations, catastrophic mis-
match stresses may be overcome by modifying
the stress distribution in the vicinity of the joint.
Problems associated with the formation of del-
eterious intermetallic phases in the joint, lack of
wetting, and, at the other extreme, excessive
erosion of one or more of the parent materials
can usually be circumvented by interposing a
layer of a different metal between the ller and
the parent material and thereby altering the
metallurgical constitution of the joints. Fluxes
and active ller alloys can be used to improve
wetting and reduce void levels. Some of the
possible remedies that may be brought to bear
on these and other problems are detailed in the
following sections.
If, on the other hand, no joining solution
proves technically tractable, or if the solutions

Table 4.1 Materials systems approach to joining process development

Phase of development Decision making Boundary conditions

Nature of the product and Scale of production

functional requirements Cost constraints
Size and weight limits
Statutory requirements

Denition of the product

Service conditions Thermal environment
Stress environment
Chemical environment

Parent materials Joining temperature
Joining atmosphere
Mismatch stress
Materials and process
selection
Filler alloys Joining temperature
Joining atmosphere
Joint geometry
Filler geometry and
form

Process assessment: Determination of Metallurgical
identication of critical process viability constraints
materials problems

Mismatch Alloying

Wetting Erosion Phases

(Mechanical solutions) (Metallurgical solutions)
Identication and
achievement of solutions
Interlayers Surface processing
Multilayers Wettable coatings
Graded structures Barrier coatings
Compliant structures Active ller alloys
Diffusion soldering/brazing Fluxes
Diffusion soldering/brazing

Prototyping and production Process specied Tolerance of process
established

Adapted from Principles of Soldering and Brazing, 1993, p 112

 Chapter 4: The Role of Materials in Dening Process Constraints / 147
are not economically justiable, then changes
earlier in the decision-making chain are re-
quired. In extreme situations, this may require
drastic revision, perhaps even to the extent that
other means of assembly have to be considered
or that the functional requirements of the prod-
uct must be relaxed.
4.1 Metallurgical
Constraints and Solutions
In principle, most metals can be joined using
ller alloys. However, when there is a require-
ment to join two different parent materials to-
gether, the available choice of llers that are
compatible with both is narrowed somewhat,
especially when the constraints on the solder,
processing conditions, and properties of the joints
mentioned previously are imposed. The prob-
lem is often made more acute by the fact that
the joining processes tend to be left to a later
stage of product design, which further reduces
the available options. The joints are always
considered as an integral part of the overall
design of an assembly if manufacturing is to be
facilitated.
Metallurgical incompatibility of materials and
processing conditions will manifest itself through
poor wetting, excessive erosion of the parent ma-
terials, and/or the formation of undesirable
phases. Means for eliminating or suppressing
these deleterious characteristics are described as
follows.
4.1.1 Wetting of Metals by Solders
Restrictions applied to the choice of joining
temperature and atmosphere, including the use
of uxes, can result in poor wetting of the com-
ponent surfaces by the molten ller if the per-
missible process conditions are inadequate to en-
sure that the joint surfaces are sufficiently clean.
Active solders (described in section 4.1.2.2 of
this chapter) can often help to overcome this
problem, but it is seldom possible to use solders
containing active wetting agents. Alternative
remedies are then necessary.
Poor wetting is a particularly serious problem
when the parent materials are refractory metals.
Their reactivity with oxygen, and, in some cases,
with other elements in the atmosphere, and the
stability of the reaction products on the surfaces
of these materials cause poor wetting. Nonme-
tallic phases present at the surface of materials,
such as graphite inclusions in cast iron and non-
metallic components in metal-matrix compos-
ites, can also inhibit wetting.
These problems are not restricted to the parent
metals but can also encompass the ller metal. A
prime example is provided by zinc-base solders
when these need to be used without an appro-
priate uxing agent. Foils and other preforms of
these alloys tend to produce poor wetting and
spreading over the joint surfaces. However, by
making small additions of elements that lower
the surface tension of the molten ller and that
also help to destabilize the native surface oxide
layer, wetting can be considerably improved.
This is illustrated for indium in Chapter 3, Fig.
3.28, and other examples are outlined in Chapter
2, section 2.2. The concentration of the indi-
vidual additions must be restricted to a maxi-
mum of approximately 1% in order to avoid per-
ceptibly altering the bulk metallurgical
characteristics of the ller alloy.
If the joining environment, which includes
uxes when these are approved, cannot ad-
equately clean and protect the components from
oxidation during the process cycle, a favored
solution is to apply a coating of a more noble
metal to the precleaned joint surfaces. This coat-
ing may be sacricial in that it is subsequently
dissolved by the ller, which then wets the clean
surfaces of the parent material. At the same time,
the composition and thickness of the coating
should be such that when dissolved, it does not
give rise to brittle phases by reaction with the
ller and parent materials.
Copper, silver, and gold are the principal el-
ements used as wettable metallizations because
of their nobility, metallurgical compatibility with
most solder alloys, and ease of deposition. Tin is
also widely used in conjunction with soldering
processes, because tin oxide is readily displaced
by an advancing wave of the molten solder, which
itself often contains tin. The benets of tin are
that it is relatively inexpensive and that it pre-
sents minimal risk of unfavorably altering the
constitution of the solder.
In the semiconductor and optoelectronic in-
dustries, platinum overlaid with a gold ash is a
frequent choice of wettable metallization. Thin-
lm deposition equipment with a platinum sput-
tering source is often readily available, and the
materials cost is insignicant compared to the
process cost. Platinum has the benet of excep-
tionally low solubility in tin- and indium-base
solders and will not oxidize when exposed to air
148 / Principles of Soldering
(Table 4.2). The gold ash ensures good spread-
ing and aids visual inspection of the faying sur-
faces prior to use. The reaction product with
tin-base solder has been identied as an excep-
tionally thin interfacial layer of PtSn4 [Kuhmann
1977]. This metallization scheme confers excel-
lent wetting and spreading with precisely con-
trolled reproducibility, which are useful at-
tributes when joining high-value components.
Solderable metallizations are usually applied
by either wet plating or vapor deposition tech-
niques. Tin coatings can be applied by dipping
methods. The choice of deposition method is de-
termined by various factors, such as the size of the
component, its geometry, the required scale of
production, the capital and running costs of the
deposition equipment, and the thickness of the
metallization.Additional details of the merits and
limitations of each of these coating techniques are
given in Appendix A4.1 in this chapter.
Any metallization applied needs to be suffi-
ciently thick for it to protect the underlying ma-
terial from corrosion and to maintain wettability
of the component over a reasonable storage pe-
riod. The uniformity and density of sputter-
deposited coatings of gold afford such protection
at a thickness of typically 0.3 m (12 in.),
whereas electroplated coatings need to be more
than an order of magnitude thicker to provide the
same shelf life [Humpston and Jacobson 1990].
This point is illustrated in Fig. 4.1, which shows
the wettability, in terms of the wetting force mea-
sured on a wetting balance, of chromium-
metallized coupons coated with different thick-
nesses of gold by sputtering and electroplating,
as a function of the storage time in an open at-
mosphere. The porosity of thin electroplated or
evaporated coatings is frequently overlooked, es-
pecially when a process is transferred from de-
velopment to a less rigorous manufacturing en-
vironment. Further proof of the ineffectiveness
of noble metal coatings in preventing oxidation
of an underlying base metal comes from the ben-
ecial exploitation of their use to grow oxide
lms of controlled thickness in the manufacture
Table 4.2 Dissolution rate of platinum in
tin-base solder at selected temperatures
Temperature
C F Dissolution rate
80 180 (solid-state aging) 1.0 nm/day
200 390 0.3 nm/s
270 520 1.2 nm/s
320 610 2.2 nm/s
of solar cells. In particular, 5 m (200 in.) of
silver, applied by thermal evaporation, to a clean
copper surface permits the growth of Cu2O at the
common interface at a rate of roughly 0.1 m/h
(4 in./h) at 500 C (930 F) [Rosenstock and
Riess 2000].
The alloying behavior of the solder with the
coating metal may establish a maximum limit to
the thickness of the metallization beyond which
joint embrittlement ensues, as explained in sec-
tions 4.1.3 and 4.1.4 of this chapter. Where eco-
nomic and size factors dictate that electroplating
is the preferred method for applying gold coat-
ings, the components should be provided with
typically 3 to 4 m (120 to 160 in.) thick plat-
ings of gold to confer an adequate solderable
shelf life. Then, immediately prior to the sol-
dering operation, the level of gold can be con-
siderably reduced to suit, in particular, high-tin
solders by wicking this metal off the joint sur-
faces. The removed gold can be reclaimed from
the discarded solder.
Metallizations that promote wetting can also
be used to advantage to conne the molten solder
to specic areas on the surfaces of components.
By selectively applying the coating to prescribed
areas, the solder spread can be restricted accord-
ingly.
If the wettable coating is not adherent as ap-
plied, it often can be secured on the surface of a
metal component by a subsequent heat treatment
in a nonoxidizing atmosphere. This process can
promote interdiffusion between the metal and the
Fig. 4.1 Solderability shelf life of gold-coated components.
Thicker and denser coatings are more impervious to
oxygen and water vapor and therefore confer greater protection
to the underlying metal.
 Chapter 4: The Role of Materials in Dening Process Constraints / 149
coating, resulting in a graded interface akin to
that obtained using a carburizing (carbon-
enriched) or sheradizing (zinc-enriched) surface
treatment. For example, titanium is rendered sol-
derable by coating with gold (as applied, the gold
layer will brush off if touched) and then heat
treating at 750 C (1380 F) for 30 min in a
nitrogen atmosphere.
Copper lands on printed circuit boards and
electronic component leads are often coated so
as to maintain solderability in storage and to
permit extremely rapid wetting and spreading by
the molten solder at the appropriate time. Speed
is important, because printed circuit boards pass
through the solder wave of a wave-soldering ma-
chine in 2 s or less. The coatings must also en-
train low materials and application costs. There
are basically three main types: fusible coatings
(tin, lead-tin, and other tin alloys); soluble coat-
ings, that is, noble metals that are readily soluble
in molten solder (silver, gold); and organic coat-
ings (1,2,3-benzotriazole, 1,3-benzodiazole, imi-
dazole, and similar compounds). Often, these
coating strategies are combined. For example,
one commercial product system designed to
maximize the preservation of the solderability of
copper involves application of a thin (0.07 to 0.1
m, or 3 to 4 in.) layer of silver, which is itself
protected from damaging environmental effects
by an organic layer, typically also 0.1 m (4 in.)
thick. Silver is a good choice of overmetal be-
cause it is more noble than copper, and unlike
lead/tin coatings, it cannot form an intermetallic
structure with copper, which generally has poor
solderability. The organic coatings are designed
to form a chemical bond to a clean substrate
surface, and hence protect it during storage, and
then either dissolve in ux or decompose cleanly
just below the soldering temperature (typically
120 to 150 C, or 250 to 300 F). Nickel/gold and
nickel/palladium nishes are more correctly
termed barrier coatings, because the nickel and
palladium protect against reaction between the
solder and the substrate. The gold overcoat pre-
serves the solderability of the nickel or palla-
dium. A comprehensive review of surface n-
ishes relevant to electronic circuit assembly on
printed circuit boards is provided by Vianco
[1998].
4.1.2 Wetting of
Nonmetals by Solders
Nonmetalsnamely, ceramics, glasses, and
plasticsare not wetted by most solders, even
when their surfaces are scrupulously clean. This
is because they are chemically very stable, with
their atoms strongly bound to one another. There-
fore, these materials will not react with and be
wetted by molten solder unless the latter contains
an active element, which can attach itself to the
anionic species of the nonmetallic material. In a
compound or complex, this is normally oxygen,
carbon, nitrogen, or a halide element.
There are two solutions to soldering to ceram-
ics. One is to use an active ller metal, as discussed
in section 4.1.2.2 of this chapter. Active metal
joining is only effective if sufficiently high tem-
peratures, typically above 800 C (1470 F), can
be used for the joining operation, so that the active
ingredient is able to react with the nonmetal.
Where activated llers produce successful joints,
it is observed that the active element concentrates
at the interface with the nonmetallic base material
and imparts metallic characteristics to the surface
of the ceramic.
Where this approach is incompatible with the
joining process, a very similar result can be
achieved via a different route. This involves coat-
ing the joint surfaces with a metal that will bond
strongly to the underlying nonmetal and that is at
once wetted by the ller while not entirely dissolv-
ing in the process. Any erosion through to the
original component surface would result in de-
wetting. This is exemplied by the dissolution of
chromium metallizations by bismuth-containing
solders observed at temperatures above 261 C
(502 F). Bismuth and chromium react to form a
low-melting-point eutectic alloy at this tempera-
ture [Humpston and Jacobson 1990]. Further de-
tails of solderable coatings for nonmetals are
given in the following section.
4.1.2.1 Solderable
Coatings on Nonmetals
Wettable coatings can be applied to the non-
metallic components by methods similar to those
used on metalsnamely, physical vapor depo-
sition, chemical vapor deposition, and wet plat-
ing. Also widely used are red-on glass frits
loaded with particles of metal powder or ake,
often referred to in the literature as thick-lm
metallization techniques.
The vapor phase deposition route tends to be
favored wherever thin coatings of high quality
are required. It is versatile and permits a wide
range of metal and alloy coatings to be applied.
When metallizing a nonmetallic material, it is
usual to deposit more than one layer onto the
150 / Principles of Soldering
joint surface. It is essential that the layer in direct
contact with the nonmetal (henceforth referred to
as the foundation layer) is an active metal, which
is usually nickel, chromium, or titanium. The
choice is largely dictated by the solubility of this
metal in the ller alloy, which must be low but
nite, and the ability of these active elements to
bond strongly to nonmetallic materials, provided
that the surfaces are clean. According to a simple
thermodynamic model, it might be expected that
the adhesion of a metal to, say, an oxide is given
by the Gibbs free energy of formation of the
oxide of the metal: The greater the extent to
which the free energy is reduced when the metal
reacts to form an oxide, the more active is the
bonding. That this model is an oversimplication
of reality is clear from the observation that chro-
mium forms an adherent bond to silica (quartz),
even though the Gibbs free energy of formation
of silica is higher in magnitude than that of
chrome oxide. Evidently, the Gibbs free energy
of oxide formation is only one factor. This ex-
ample merely highlights the complexity of the
subject of metal-to-nonmetal bonding and ac-
counts for the considerable literature that it has
produced [Peteves 1988]. Nevertheless, the de-
sired end result is that the component surface is
rendered sufficiently metallic to enable wetting
by the ller metal.
A sputtering process usually incorporates a
capability for cleaning surfaces prior to deposi-
tion, by being operated in a reverse-bias mode,
so that the atoms of the inert carrier gas in the
deposition chamber are made to bombard the
surface and physically remove lms of contami-
nant. A separate ion source is often included in
evaporation systems to perform the same func-
tion. An in situ preclean stage is essential to
remove absorbed species and achieve good ad-
hesion by the foundation metal.
Chromium, or a mixture of nickel with chro-
mium (Nichrome), is frequently recommended
as the reactive metal for the foundation layer on
nonmetals [Holloway 1980]. Nichrome has a
lower intrinsic stress than pure chromium, which
is benecial for some strain-sensitive electronic
and optical components. Chromium reacts with
many nonmetallic compounds to form complex
chromates. These are not only strongly bonded
to the nonmetal but also act as a barrier to further
interaction taking place by diffusion between
chromium and the nonmetal [Mattox 1973].
Hence, it is possible to use chromium as a high-
integrity metallization on glass and intrinsically
stable ceramics such as alumina and quartz. Ten-
sile strengths are reported to be in the region of
70 MPa (10,000 psi) [Vianco, Sifford, and
Romero 1997]. The benet of this type of chemi-
cal bonding is evident when metallizing rela-
tively unstable ceramics, such as zinc oxide. Zinc
oxide is widely used in electrical voltage surge
suppressors, known as varistors [Leite, Varela,
and Longo 1992; Wersing 1992]. Heating a zinc
oxide ceramic metallized with titanium or zir-
conium causes the metallization to blister and
spall off due to the oxidizing reaction:
2ZnO  Ti  TiO2  2Zn
G2ZNOTi (573 K)  272 kJ (65 kcal)
A by-product of this reaction is the volatil-
ization of free zinc from beneath the coating as
soon as the coated part is heated, which results
in the detachment of the metallization. On the
other hand, a chromium metallization is entirely
benign toward zinc oxide, even upon heating to
400 C (750 F) for prolonged periods, due to the
presence of the intervening chromate barrier
layer. Chromium is also marginally easier to ap-
ply than titanium, because it is slightly less re-
active toward oxygen in the air, and hence, the
quality of the vacuum chamber in which the
deposition is conducted can, accordingly, be
slightly relaxed.
The high reactivity of titanium and other ac-
tive metals toward oxygen and nitrogen in air
means that they rapidly lose their wettability even
after a brief exposure to the atmosphere. To over-
come this problem, a more noble metallization
that offers good wettability to ller metals must
be deposited over the active-metal foundation
layer immediately, while the deposition chamber
is maintained under a protective (low-oxygen)
atmosphere. A metallization system that has been
found to be effective for many soldering appli-
cations is one that comprises a 0.1 m (4 in.)
thick layer of a foundation metal and a 0.3 m
(12 in.) overlay of a wettable metallization, such
as gold, both applied by sputtering. The chosen
foundation-metal thickness is specied to pro-
vide good step coverage even after taking into
account parameter variations in an industrial pro-
cess that may result in local thickness variations
up to 25%. The thicker gold layer provides a
reasonable shelf life and usually permits joints to
be made using tin-base solders without risk of
embrittlement (see Chapter 2, section 2.3.2).
 Chapter 4: The Role of Materials in Dening Process Constraints / 151
An alternative method of coating a nonme-
tallic material with an adherent metal coating is
to use a two-stage joining process in which the
rst step is to allow an activated solder or braze,
capable of providing a solderable surface, to wet
and spread over the component surfaces. The
tinned surface is then usually mechanically
dressed to present a at metal surface for the
actual joining step, which is performed at a tem-
perature below the solidus point of the activated
ller metal. Because of the number of elements
that are likely to be present in the joint, the re-
sulting alloy constitution can be somewhat com-
plex unless the process is designed such that the
alloy used for the coating and the solder alloy
have constituents in common.
A somewhat different approach is to re on a
relatively thick (typically, 1 to 10 m, or 40 to
400 in.) metal coating. This type of metalliza-
tion process tends to be used only on glass and
ceramic materials because of the high tempera-
tures involved [Bever 1986]. In very simple
terms, a thick-lm paste can be considered as a
mixture of metal and ceramic or glass particles.
On ring, the glass or ceramic phase wets the
component surface, while the metal particles oat
to the surface. Adhesion between the solidied
glass or ceramic phase and the metal particles is
by a combination of chemical and physical
mechanisms. Further details of the range of com-
mercial processes that are available are given in
Appendix A4.1 in this chapter.
Reference may sometimes be found in older
technical literature to the active hydride process.
It involves applying a metal hydride, usually of
titanium or zirconium, in the form of a paste to
the component surface. On heating, the hydride
thermally decomposes, liberating hydrogen to
leave a metal lm on the component surface
[Pershall 1949]. This method has lost favor in
recent years to alternatives, because the quality
of the resulting metallization tends to be vari-
able, as reected in the mechanical properties of
the joints [Mizuhara and Mally 1985]. The un-
predictable nature of the metallization quality is
due to its high sensitivity to variations in the
atmosphere during the ring stage. Oxygen and
water vapor contents, in particular, affect the ex-
tent to which the highly reactive metallic con-
stituent oxidizes after decomposition of the hy-
dride and hence the ease of wetting by the molten
ller.
A widely used method for metallizing oxide
ceramics, in particular, alumina, is the so-called
moly-manganese process. It has the advantages
of being relatively straightforward to perform,
highly reproducible, not requiring a vacuum en-
vironment, and thus amenable for processing
large parts. In this process, a slurry of powders
of molybdenum, manganese, and various glass-
forming compounds is applied as a paint to alu-
mina components. Then, the coated ceramic is
red in a wet hydrogen atmosphere (with a well-
controlled dewpoint) at a temperature of 1450 to
1500 C (2640 to 2730 F). The ring operation
results in chemical reaction between the glassy
phase in the alumina and the manganese, while
the molybdenum is established as a surface layer
on the ceramic [Mattox and Smith 1985]. Be-
cause molybdenum is not readily solderable, es-
pecially when the process is limited to mild
uxes, it is common to complete this metalliza-
tion scheme with a coating of electroless nickel,
followed by gold.
An important point to be aware of is that met-
allizations are often put down in a stressed con-
dition. This is an inherent feature of most depo-
sition processes and can be a source of critical
weakness. An example of the manifestation of
excessive stress in a metallization layer is shown
in Fig. 4.2. Here, a silver layer 20 m (800 in.)
thick has cracked away from a ceramic material
through such stress. The stress concentration at
the interface between the component and the met-
allization can be controlled by limiting the thick-
ness of the metallization layers, modifying the
deposition parameters, and using conventional,
postdeposition, stress-relief heat treatments.
Particularly in microelectronics and photonics
manufacturing, there is an ongoing trend toward
product miniaturization. This is demanding the
Fig. 4.2 A porous ceramic material metallized with a thick
silver electroplate. The residual stress in the metal-
lization has resulted in a peel failure through the near-surface
layer of the ceramic.
152 / Principles of Soldering
development of ever-smaller joint geometries and
hence the volume of solder in each joint. At the
same time, pricing pressures mean that any ap-
plied metallizations must be as thin as possible,
because their cost is usually a direct function of
machine time. In theory, a metallization need
only be sufficiently thick that it ensures complete
coverage of the faying surface and, if it is a
barrier metal, will not be eroded back to the
underlying material by reaction with the solder
during the joining process or in service. While it
is possible to give general rules of thumb re-
garding metallization thicknesses, attempts are
now being made to derive more rigorous values
by calculation. One such approach uses a com-
bined thermodynamic and diffusion-kinetic
model, and it is sufficiently broad in its scope to
be able to deal with real systems [Ronka, Van
Loo, and Kivilahti 1998]. No doubt, further ad-
vances will be made in this area over the next few
years.
4.1.2.2 Active Solders
Reactive ller metals are mostly brazes, and
further details of their design and function can be
found in the planned companion volume Prin-
ciples of Brazing. The basis of the approach is to
incorporate into the ller metal small quantities
of elements that are highly reactive. Provided
that at least one of the products of reaction with
the base material is metallic and remains as a
layer on the surface of the nonmetal, then the
ller alloy can wet and form sound joints.
When selecting a reactive ller alloy, the
choice of active ingredient, its optimal concen-
tration, and the appropriate processing condi-
tions need to be considered in relation to the
nonmetal of the component. One of the most
commonly used active constituents is titanium.
Less reactive elements, such as chromium, and
more reactive elements, such as hafnium, are
also used.
The addition of metals such as titanium to
solders will enable them to directly wet non-
metals such as ceramics, provided the process
temperature is sufficiently high to provide the
necessary activation for the chemical bond to
form with the nonmetal. Typically, this activa-
tion takes effect above 750 C (1380 F), which
somewhat negates the benet of the low pro-
cessing temperature characteristic of solders
[Xian and Si 1992]. The process temperature is
governed by the reactivity of the active ingre-
dient, which increases with temperature. If the
temperature margin to the liquidus temperature
of the solder is substantial, the heating cycle
should be kept short in order to prevent extensive
erosion of the joint surfaces or the growth of
thick interfacial phases.
It has recently been reported that melding
small quantities (0.5 to 2%) of lutetium into sol-
ders has a similar effect but enables a reduction
in the process temperature to more normal val-
ues. Rare earth elements have poor solubility in
solder and therefore tend to agglomerate in small
islands within the matrix, where they are pro-
tected from oxidation by the surrounding metal.
However, in the molten state, sufficient rare earth
metal is conveyed to the joint surface, where it
forms a chemical bond with the nonmetal. The
interface layer is 1 to 5 nm (0.04 to 0.2 in.) thick
and appears to comprise lutetium oxide (Lu2O3)
and lutetium-rich solder. By this means, tensile
joint strengths in the region of 800 MPa (120,000
psi) and shear strengths of the order of 10 MPa
(1500 psi) can be achieved to silica-based ma-
terials [Ramirez, Mavoori, and Jin 2002]. Fur-
ther information on solders containing rare earth
elements can be found in Chapter 5, section 5.8.
Hitherto, active llers were found only to be
effective above threshold temperatures of ap-
proximately 750 C (1380 F), and therefore, they
have only been used for brazing. Recently, a
route for substantially reducing the temperature
threshold has been identied, making active sol-
ders a reality. This involves the addition of low
concentrations of gallium and lanthanide ele-
ments to a silver-tin eutectic alloy containing 4%
Ti [Smith 1998]. Although the exact role of the
minor additions is unclear, they appear to fulll
two functions. First, they modify the nature of
the surface oxide that limits its growth. Second,
the minor additions appear to increase the ac-
tivity of the titanium, which enables the solder
alloy to wet and bond directly to nonmetals such
as alumina, aluminum nitride, beryllia, silicon
carbide, silicon nitride, boron nitride, carbon/
carbon composites, aluminum/silicon carbide
composites, copper, silicon, diamond, titanium
and stainless steel, to name but a few, using a
peak process temperature of only 250 C (480
F). The titanium and other additions increase
the strength of the bulk solder and, interestingly,
confer it with signicantly enhanced resistance
to corrosion from aqueous media containing
chlorides. The activated solder can be used in air
without ux, but in order to facilitate wetting, it
is necessary to use some form of mechanical
activation, such as scrubbing. This locally rup-
 Chapter 4: The Role of Materials in Dening Process Constraints / 153
tures the oxide skin and enables wetting and
spreading to proceed from points of contact be-
tween the nonmetal and molten solder.
4.1.3 Erosion of Parent Materials
When a molten ller wets the surface of a
parent material, alloying occurs, leading to a de-
gree of dissolution of the parent material (or
metallization), commencing at the joint inter-
face.
Dissolution of the parent materials occurs be-
cause the materials system encompassing the
joint is not in thermodynamic equilibrium. This
provides the driving force for wetting and spread-
ing. The maximum solubility of the parent ma-
terials in the molten ller can be predicted by
reference to the appropriate phase diagram, as
stated in Chapter 2, section 2.3. In summary,
extensive erosion is likely where the liquidus
surface on the phase diagram between the ller-
metal composition and that of the parent material
has a shallow slope and where the alloying de-
presses the melting point of the ller in the joint.
If the phase diagram exhibits either of these fea-
tures, then it is only possible to limit erosion by
lowering the process temperature, shortening the
heating cycle, and/or restricting the volume of
molten ller metal. Such changes must obvi-
ously not compromise the integrity of the joints
by, for example, reducing the effective uidity of
the molten ller alloy, which in turn will impede
wetting, spreading, and joint lling.
Erosion can be reduced somewhat if interme-
tallic phases form along the joint interface so as
to attenuate the rate at which solid material is
transported from the components into the molten
ller, or vice versa. This approach is only suc-
cessful if the intermetallic phases formed are
reasonably ductile, as exemplied by the case of
tin-base solders used in conjunction with silver
metallizations that is described in Chapter 2, sec-
tion 2.1.2; otherwise, joint embrittlement results.
In other cases, it is possible to protect the com-
ponents against erosion by interposing a metal-
Table 4.3 The effect of gold on the solidus temperature of common, binary tin-base eutectic
solders
wt% gold to form
Solder composition AuSn4 as the primary phase
In-48Sn AuIn2 formed
Bi-43Sn >1
Pb-63Sn >8
Sb-95Sn >10
Ag-96Sn >11
lization that will act as a barrier. An example is
the application of a layer of nickel, typically 2 to
5 m (80 to 200 in.) thick, on aluminum com-
ponents to prevent reaction with lead-tin solder.
Although nickel-tin compounds form at the com-
mon interface, in normal soldering heating
cycles, the intermetallic layer is quite thin and
therefore does not adversely affect the mechani-
cal properties of the joints.
The extent to which dissolution occurs in a
given heating cycle will also depend on the ki-
netics of reaction. The rate of dissolution of com-
mon engineering materials and metallizations in
eutectic lead-tin solder is given in Chapter 2, Fig.
2.53. These data show that decreasing the pro-
cess cycle time and temperature can be used to
reduce erosion, although not always to accept-
able levels in all cases.
The propensity for dissolution (erosion) of the
parent metal by a ller alloy can be reduced by
preloading the ller with this metal so that it is
already saturated and, on becoming molten, will
not dissolve any further quantity. This approach
is used for soldering to gallium arsenide devices
using the Au-20Sn solder modied by the addi-
tion of 3.4% Ga. The addition reduces the solu-
bility of gallium in the solder by half, with
consequential benets to the process yield
[Humpston and Jacobson 1989].
Metallurgical reaction between a solder and
the materials of the components being joined (or
their metallizations) can substantially alter the
characteristics of the ller, which may in turn
have unexpected consequences for the resulting
joints. By anticipating the effects of these reac-
tions, they either can be exploited to advantage
or at least forestall embarrassing problems. Two
examples are described as follows.
Changes to the Melting Point of Soft Solders.
Gold is widely used as the surface layer in met-
allization schemes because of its chemical in-
ertness and ease of visual inspection. The change
in the melting point (solidus temperature) of se-
lected binary tin-base solders, when these have
dissolved gold, is given in Table 4.3. The effect
wt% gold that will dissolve Change in melting point of solder
in solder at 50 C superheat by dissolving 1% Au, C (F)
<1 1 (30)
4 2 (28)
13 6 (21)
30 15 (5)
30 15 (5)
154 / Principles of Soldering

is particularly marked in the case of the Ag- properties of joints. This is particularly true
97.5Pb-1Sn solder, where the solidus tempera- where the compound has low fracture toughness
ture is reduced from 309 to 217 C (588 to 423 and forms as a continuous interfacial layer be-
F) when 4% Au is added. For a layer of the tween the component surfaces and the ller al-
solder 50 m (2 mil) thick, this radical depres- loy. It is sometimes possible to restrict the coars-
sion of the melting point corresponds to a mere ening of these phases by carrying out the joining
1 m (40 in.) of gold coating being dissolved operation under conditions that are unfavorable
from each side of the joint. An example of a for their growthnamely, restricting the heating
change in the contrary direction is the dissolution cycle duration and peak temperature. In some
of gold in the In-18Pb-70Sn solder (melting range
136 to 182 C, or 277 to 360 F), where a 5% Au
addition raises the solidus temperature by 14 C
(57 F). Partial isopleths through these quater-
nary systems as a function of gold concentration
are given in Fig. 4.3 and 4.4.
Changes to Rate of Erosion of Parent Mate-
rials. Minor changes to the composition of the
ller alloy are capable of inuencing the rate at
which dissolution occurs. This is exemplied by
the effect of small additions of silver, typically
2%, on the rate of dissolution of silver by lead-tin
solders, as can be seen from Chapter 2, Fig. 2.3.
The extent of the reduction could not be pre-
dicted from the Ag-Pb-Sn phase diagram and
was determined empirically. Changes of this sort
do not signicantly alter other properties of the
ller, or those of the resulting joints, and can be
readily implemented.

Fig. 4.3 Section through the Ag-Au-Pb-Sn quaternary system

4.1.4 Phase Formation showing the effect on the melting range of adding
gold to Ag-97.5Pb-1Sn solder. Adapted from Evans and Prince
Alloying between a molten solder and parent [1982]
materials more often than not results in the for-
mation of intermetallic compounds. This is par-
ticularly more true of solders, because the major
constituents of most solders are elements with low
crystallographic symmetry, and hence, they do
not readily form solid solutions with engineering
materials that tend to be based on simple body-
centered cubic (bcc), face-centered cubic (fcc), or
hexagonal close-packed (hcp) crystal structures.
The distribution, morphology, and proportion of
these phases will depend on several factors, as ex-
plained in Chapter 2, section 2.3.
Because intermetallic compounds generally
possess a higher elastic modulus (i.e., they are
stiffer) than many ller alloys themselves, well-
dispersed intermetallic phases are often bene-
cial to the stress-bearing capability of the ller
metal. This effect is shown for gold additions to
the silver-tin eutectic solder in Chapter 2, Fig.
2.52. Fig. 4.4 Section through the Au-In-Pb-Sn quaternary system
showing the effect on the melting range of adding
By and large, agglomerations of intermetallic gold to In-18Pb-70Sn solder. (This composition is not a ternary
compounds are deleterious to the mechanical eutectic, as is sometimes stated in older literature.)
 Chapter 4: The Role of Materials in Dening Process Constraints / 155
cases, minor additions can be made to the ller
alloy that will break up agglomerations of in-
termetallic phases present in the joint. This is one
of the reasons for some minor additions made to
zinc-base solders formulated for soldering to cop-
per, which is described in Chapter 2, section
2.1.3. Barrier metallizations applied to joint sur-
faces can be used to prevent alloying with the
parent materials and the consequential formation
of undesirable phases.
Some ller alloys are themselves hard because
they contain intermetallic phases, yet form joints
of high strength with certain parent materials,
provided that the intermetallic phases are nely
divided within the joint microstructure. Gold-tin
solders fall into this category, with rapid solidi-
cation of the solder being a prerequisite for
robust joints. Further details are given in Chapter
2, section 2.1.4.
4.1.5 Filler-Metal Partitioning
Partitioning of ller metals is a fabrication
method that is becoming more common. It is an
approach that can be benecial if an alloy com-
position is relatively brittle when prepared via
conventional casting of ingots, but selected com-
binations of the constituents are ductile. Some
examples taken from the published literature are
listed in Table 4.4. The objective is either one of
attempting to improve the mechanical properties
of the ller metal prior to melting or one of cost
reduction through decreasing the number of stock
alloys required [Mackay and Levine 1986]. It is
usual to have the lower-melting-point combina-
tion in the cladding, with a higher-melting-point
mixture of ingredients in the core. The low-
melting-point fraction then melts at the soldering
temperature and wets the joint interfaces.
The Au-20Sn solder, as produced in the form
of wire or foil, is very costly, partly because of
its precious metal content but mostly because of
the limited reduction in thickness that can be
achieved on successive passes during hot rolling.
This means that the fabrication costs are high.
Table 4.4 Examples of partitioned ller metals
Filler metal Cladding
Silver-tin (mpt, 221 C) Sn Ag
Lead-antimony-tin (mr, 245280 C) Sb-Sn Pb-Sb-Sn
Silver-lead-tin (mr, 303310 C) Ag-Sn Pb-Sn
Gold-tin (mpt, 280 C) Au Sn
mpt, melting point; mr, melting range
Because gold and tin are both very ductile met-
als, composite foil can be prepared by cold roll-
ing. The noble nature of gold is exploited, and,
by placing it on the outside of the sandwich, it
is then able to prevent oxidation of the under-
lying tin during storage and heating to the pro-
cess temperature. A postfabrication heat treat-
ment step is claimed to be benecial in
consolidating the foil and initiating interdiffu-
sion between the constituents prior to use
[Tokuriki Hoten 1981].
An alternative approach is to modify the sur-
face of a ductile foil of tin through electrodepo-
sition of a layer of gold. In this case, it is the
higher-melting-temperature constituent that con-
stitutes the surface layer. If the tin foil is 25 m
(1 mil) thick, and 12 m (0.5 mil) of gold plating
is applied on both sides the calculated compo-
sition of the solder will be hypereutectic at 70
wt% Au, and 30 wt% Sn. From the phase dia-
gram for the gold-tin system, it may be deduced
that if this composition were prepared as a ho-
mogeneous alloy, it would have a melting range
of 280 to 370 C (536 to 698 F).
The melting behavior of this alloy has been
characterized using differential scanning calo-
rimetry. The results are given in Fig. 4.5 and 4.6.
On initial heating (Fig. 4.5), there is an endo-
thermic reaction as the solid-state alloy formed
by the interdiffusion of gold and tin melts at the
binary eutectic temperature of 217 C (423 F).
This is followed by a tin-rich transition reaction
at 252 C (486 F) before onset of the gold-rich
eutectic transformation at 280 C (536 F). There-
after, melting continues until all the solid is con-
sumed, which occurs at approximately 380 C
(716 F). Cooling of this sample back to 150 C
(302 F) and repeating the thermal analysis cycle
(Fig. 4.6) shows that the sample is now homo-
geneous, with a solidus temperature of 280 C
(536 F) and a liquidus temperature of approxi-
mately 380 C (698 F). This result accords well
with the predicted melting behavior deduced
from the alloy phase diagram.
Partitioning of solders can sometimes also be
benecial where the maximum permissible pro-
cess temperature is only marginally above the
melting point of the ller metal. Phased reow
Core
soldering, as this approach is described, has been
exploited in some of the lead-free solder paste for-
mulations based on Bi-Cu-Sn alloys. These re-
placement solders for lead-tin eutectic have to
function at very low superheats, because most
printed circuit boards and components have rated
156 / Principles of Soldering

20.00
mcal/s

10.00

0.00
160.00 190.00 220.00 250.00 280.00 310.00 340.00 370.00 400.00
Temperature, C

Fig. 4.5 Melting behavior of gold-plated tin foil having an effective composition of Au-30Sn

Fig. 4.6 Remelting behavior of gold-plated tin foil

 Chapter 4: The Role of Materials in Dening Process Constraints / 157
maximum process temperatures of 235 C (455
F). Copper-tin eutectic melts at 227 C (441 F),
and bismuth-tin at 139 C (282 F). By partition-
ing the ternary composition so that it includes
some low-melting-point tin-bismuth solder, liq-
uid phase sintering will commence on heating
above 139 C (282 F). The presence of the liquid
phase improves heat transfer to the unmelted par-
ticles and helps minimize temperature gradients
through the solder paste, both of which improve
the reliability of ller-metal joining processes
performed at low homologous temperatures. The
progressive melting of the solder is also reported
to be benecial in reducing the incidence of
tombstoning [Warwick 2002]. This is where
one land on a double-ended surface mount com-
ponent wets more effectively than at the other, and
the imbalance of surface tension forces causes the
component to lift up vertically on one land. Sev-
eral solder manufacturers have led patents on
phased reow soldering.
There have also been attempts to effect sol-
dering by liquid phase sintering, that is, to join
metal components without melting the higher-
melting-point constituent of a partitioned ller-
metal paste. By this means, it has been found
possible to produce joints with reasonable me-
chanical integrity but nowhere near as sound as
that achieved by melting the solder [Palmer, Al-
exander, and Nguyen 1999].
4.2 Mechanical
Constraints and Solutions
In an assembly composed of heterogeneous
materials, there is usually a thermal expansion
mismatch between the abutting components. This
manifests itself as stress on cooling from the
solidus temperature of the ller metal and is a
maximum at the lowest temperature that the as-
sembly experiences. Materials with a relatively
low elastic modulus can accommodate strain and
will tend to deform under the inuence of this
stress, while brittle materials, notably glasses and
ceramics, have a tendency to fracture, particu-
larly if the stress distribution places the compo-
nent in tension. Even if a heterogeneous assem-
bly survives the joining operation, the stresses
arising from the thermal expansion mismatch
can cause it to fail by fatigue during subsequent
thermal cycling in service.
The stress in the region of the joint between
two isotropic materials, designated 1 and 2, with
differing thermal expansivities that develops on
cooling from the freezing temperature of the ller
metal can be approximated by the following
equation [Timoshenko 1925]:
E 1 E2
Stress  (X1  X2)(Tf  Ts )
E1  E2
where E is the modulus of elasticity of materials
1 and 2, X is the coefficient of thermal expansion
of 1 and 2, Tf is the freezing point (solidus tem-
perature) of the ller alloy, and Ts is the tem-
perature of the assembly corresponding to the
stress. In the derivation of this equation, it is
assumed that the materials are only deformed
within their elastic limits and that the joint is
innitely thin. Despite these simplifying assump-
tions and the inaccuracies that they introduce,
this expression is useful in providing an indica-
tion of whether the stress due to thermal expan-
sion mismatch is close to or exceeds the failure
stress of either of the abutting materials, that is,
whether failure of the assembly is likely to occur.
Some worked examples are described by Haug,
Schaefer, and Schamm [1989].
In an assembly with a planar joint between
two elastic but different materials 1 and 2, the
magnitude of the bow distortion in one dimen-
sion can be estimated from the physical prop-
erties of the materials using a simplied model
[Timoshenko 1925]. With reference to Fig. 4.7.
Bow distortion, B  (L2/ 8) / R
Radius of curvature 
(A1  A2)[3(1  M)2  (1  MN)(M 2  1/MN)]
6(X1  X2)(Tf  Ts)(1  M)2
(Eq 4.1)
where
M  A1/A2 and N  E1/E2, B is bow distortion
L is the length of the joint
R is the radius of curvature
A1 and A2 are the thicknesses of materials 1 and 2
X1 and X2 are the coefficients of thermal expan-
sion of materials 1 and 2
E1 and E2 are the moduli of elasticity of materials
1 and 2
Tf is the freezing point (solidus temperature) of
the ller alloy and
Ts is the temperature of the assembly correspond-
ing to the bow distortion
158 / Principles of Soldering
Equation 4.1 assumes that the joint in the hetero-
geneous assembly is innitely thin and totally in-
elastic.
From Eq 4.1, it can be seen that it is possible
to effect some reduction in expansion mismatch
stress, that is, decreasing R, by applying one or
more of the following measures:
Decrease the solidus temperature of the ller
alloy, Tf
Increase the minimum service temperature of
the assembly, Ts
Reduce the dimensions of the joint area, L
Change one or both materials to minimize the
mismatch in thermal expansivity, E1  E2
Occasionally, it may be possible to implement
one or more of these changes, but they are likely
to conict with other processing constraints if not
with the intended functional requirements of the
assembly.Alternative solutions must therefore be
sought.
In practice, it is usually possible to obtain a
small reduction in the distortion of a bowed het-
erogeneous assembly by heat treating it at a tem-
perature below the solidus temperature of the
ller alloy to enable stress relaxation and creep
to occur in the ller metal. However, there is a
limit to the reduction in distortion that can be
obtained by this means, typically 10% or less.
This stems from the fact that stress-reducing
mechanisms are diffusion related and become
more effective as the temperature is raised to-
ward the melting point of the joint, while mis-
match stress increases as the temperature of the
assembly is reduced below that at which stress
relief is effective. Thus, these two tendencies act
Fig. 4.7 Bow distortion of a bimetallic strip
in opposition, and the optimal condition for re-
ducing the distortion of bonded components is a
compromise between the two. In the absence of
other indications, a good starting point is to use
a temperature that is approximately 75% of the
melting point of the ller metal, expressed in
degrees Kelvin.
Some further improvement in the residual-
stress level can be obtained by using a more
compliant ller, especially when the joint is rea-
sonably wide (>25 m, or 1 mil). Solders, almost
without exception, have low elastic moduli,
which is related to their low melting points, but
there are only minor differences between indi-
vidual alloys. Hence, the only possibility for ob-
taining a joint with improved compliance, short
of redesigning the assembly, is to replace a hard
solder with one that will creep. The benet of
such a change might be offset by inferior di-
mensional tolerancing, and the joint design and
operating environment should take this trade-off
into consideration.
The thermomechanical properties of most of
the family of indium solder alloys (pure indium,
silver-indium, indium-lead, indium-tin) are
dominated by creep behavior at all normally ex-
perienced temperatures and strain rates. Thus, a
soldered joint made with one of these alloys will
always creep when subject to stress. The fact that
they creep so readily, even at cryogenic tempera-
tures, means that joints made with indium solders
are substantially more resistant to failure by fa-
tigue than might otherwise be expected (see Chap-
ter 2, section 2.1.6).
Wide joint gaps (>500 m, or 20 mil) can some-
times be used to minimize the effects of expansion
mismatch between two components. The solder
must have high viscosity in order to ll such wide
joints. This is achieved by either using a ller
metal with a wide melting range and performing
the joining process below the liquidus tempera-
ture, so that the alloy is not fully molten, or by mix-
ing in metal powder with a higher melting point.
Spacers are required to control the joint gap. Wide
joints can also be achieved by inserting porous
shims, as described in section 4.3.4 of this chapter.
One particular merit of wide joints to ceramic
components is that they are tolerant to variations
in the width of the joint gap, thereby obviating the
need to closely machine the mating surfaces of the
components, which tends to be costly and can
weaken the material by creating subsurface
cracks. Where the joint is wide, its mechanical
properties are essentially those of the bulk solder
(see section 4.3.3 of this chapter).
 Chapter 4: The Role of Materials in Dening Process Constraints / 159
A variety of mechanical schemes are available
to assist in overcoming the problem of thermal
expansion mismatch. Several approaches that
have proved successful are described follows.
4.2.1 Controlled Expansion Materials
Fabrication of most products requires the use
of many different materials in their assembly,
each selected for a particular property or com-
bination of properties it offers. Ashby and co-
workers have devised a scheme of materials se-
lection charts that pictorially represent materials
according to their properties, which is intended
to facilitate selection [Ashby 1994]. Because sili-
con and most other semiconductor and optical
materials, as well as engineering ceramics, have
Fig. 4.8 An Ashby materials selection chart. The linear expansion coefcient, , plotted against the thermal conductivity, . The
contours show the thermal distortion parameter /.
low thermal expansion coefficients, compared to
metals, there is always interest in controlled ex-
pansion alloys that can be used to bridge dimen-
sional changes between these materials and metal
components in the same assembly (Fig. 4.8).
Single-phase materials, which include many
engineering ceramics, such as alumina and alu-
minum nitride, change dimension in a fairly lin-
ear manner with temperature. This is certainly
true over the temperature range of interest for
consumer products (50 to 150 C). For com-
mon metals, their coefficients of thermal expan-
sivity (CTE) at room temperature are directly
proportional to their melting points. The expan-
sion effect is attributable to the atomic vibrations
of the crystal lattice. Raising the temperature
increases the vibration, which means that each
160 / Principles of Soldering
atom occupies a greater space, and hence, the
material grows in size. Because the maximum
vibration is restricted by a material melting tem-
perature, a low-melting-point material has a
smaller expansion temperature range and will
exhibit a higher CTE value than a metal with a
relatively higher melting point, as demonstrated
in Fig. 4.9, using the data in Table 4.5.
The principal low-expansion metals are tung-
sten and molybdenum, which have CTE values
at 20 C of 4.5 and 5.1 106/K, respectively.
These metals, especially tungsten, are hard and
stiff, offering little compliance, and they are also
relatively difficult to machine. Both are dense
materials, so components fabricated from them
will be heavy. Their high melting point means
that their atoms are strongly bonded, and, con-
sequently, their machining requires considerable
energy. Consequently, the net cost of parts fab-
ricated from these metals can be signicant. Ti-
tanium has a low CTE (5.6 106/K) for a
metal, low density, but a poor thermal conduc-
tivity of only 15 W/m K. This rules it out for
any application where substantial heat transfer
by conduction is an additional requirement. Alu-
mina and aluminum nitride also suffer from en-
ergy-intensive production methods and difficulty
of machining. These ceramics have to be fabri-
cated in near-net shape forms, so the tooling cost
of new components is quite high. Furthermore,
their thermal conductivity is inferior to that of
most metals.
The lack of single-phase materials that offer the
combination of low thermal expansivity, high
thermal conductivity, and, preferably, low density
has led to the development of several families of
multiphase materials with tailored properties.
These include iron-nickel alloys, copper-tungsten
and copper-molybdenum alloys, metal-metal
laminates, metal-ceramic laminates, metal-ma-
trix composites, and metal-metalloid alloys (in
particular, a family of aluminum-silicon alloys).
Fig. 4.9 General relationship between coefcient of thermal
expansion, or CTE (between 273 and 373 K), and
melting point for metals, Tm. Adapted from Li and Krsulich [1996]
These materials and their key physical properties
are listed in Table 4.6. Further information is
given in the following section.
4.2.1.1 Iron-Nickel Alloys
Most readers will probably know iron-nickel
alloys by trade names that include Invar, Kovar,
Nilo-K, Nilo Alloy 42, or by the UNS number
K94610 or DIN WNr 1.3981. Kovar and Nilo-K,
for example, are essentially the same Co-Fe-Ni
alloy, of approximate composition 17Co-54Fe-
29Ni, that has a CTE at 20 C (68 F) of 5.8
106/K. These alloys are readily available from
a number of manufacturers in many shapes and
forms and are competitively priced. Iron-nickel
alloys are widely used in electronic packaging.
Being fairly soft and ductile in the annealed con-
dition, they are also used in shims for joining
low-expansion ceramics to higher-CTE metals,
where they are capable of distributing and ab-
sorbing expansion mismatch stresses.
Iron-nickel alloys offer abnormally low ther-
mal expansion compared to their constituents.
Indeed, over a limited range of temperature, it is
possible to design material that has zero expan-
sivity. The unusual expansion characteristics of
these alloys can be ascribed to the fact that they
are ferromagnetic.
At temperatures above the Curie point (ap-
proximately 450 C, or 840 F), these alloys are
ferromagnetic and exhibit normal thermal ex-
pansion characteristics. Below the Curie tem-
perature, when they are in their ferromagnetic
domain, the actual expansion is the sum of the
normal positive expansion due to lattice vibra-
tion, counteracted by the negative expansion due
to the ferromagnetism. The latter is termed mag-
netostriction. The result is reduced CTE below
the Curie temperature that can be adjusted by
controlling the state of cold work in the alloy. At
very low temperatures, below 100 C (148
F), the expansion coefficient of iron-nickel al-
loys reverts to more normal values (15 106/K).
Expansion curves for some iron-nickel alloys are
given in Fig. 4.10 and 4.11.
It is important to recognize that iron-nickel
alloys only possess low and controlled expan-
sion coefficients over a limited range of tem-
perature. This is due to a combination of the
limited temperature interval over which mag-
netostriction compensates for normal thermal
expansion, plus the necessity for the alloy to
be in a very particular state of cold work.
Thus, further processing that involves either
 Chapter 4: The Role of Materials in Dening Process Constraints / 161

mechanical working or a temperature excur-
sion will change the internal stress in the ma-
terial and thereby its expansion coefficient. Fig-
ure 4.12 illustrates the expansion coefficients of
Fe-36Ni and Fe-42Ni alloys as a function of
temperature and state of anneal. As can be
seen, the lower-nickel-content alloys can be
fabricated to have a smaller overall thermal ex-
pansion, but it is stable over a more limited
range of temperature.
These low-expansion alloys are notoriously
difficult to machine, particularly in thin sec-
tions to a high surface nish. Specialist metal
working companies have developed the neces-
sary skills to deliver products of consistent and
high quality. It is not possible to solder di-
rectly to iron-nickel alloys using electronic-
grade uxes, and parts are normally plated with
nickel and gold. Brittle intermetallic com-
pounds form at the interface when cobalt-
containing mixtures are soldered directly with
tin-base solders.
4.2.1.2 Copper-Molybdenum and
Copper-Tungsten Alloys
Low-expansivity molybdenum (CTE  5.1
106/K at 20 C, or 68 F) or tungsten (CTE 
4.5 106/K at 20 C, or 68 F) is added to
copper (CTE  17.6 106/K at 20 C, or 68
F) to produce controlled-expansion alloys. Mol-
ten copper is virtually insoluble in both molyb-
denum and tungsten, and various techniques have
been devised to enable the manufacture of 100%
dense alloys of these materials. These techniques
include powder metallurgy as well as liquid in-
ltration casting. Controlled-expansion alloys
can be produced with a continuum of properties
that range from essentially pure copper to pure
refractory metal. By this means, families of ma-

Table 4.5 Metals and their properties used to prepare Fig. 4.9
Melting point
Metal C K CTE at 300 K, 106/K

Tungsten 3422 3660 4.5

Molybdenum 2623 2888 5.1
Palladium 1555 1817 11.0
Gold 1063 1336 14.1
Aluminum 660 933 23.5
Cadmium 321 594 31.0
Lithium 181 454 56.0
Mercury 39 235 60.0

CTE, coefficient of thermal expansion

Table 4.6 Indicative physical properties for selected semiconductor and low-expansion materials at
20 C (68 F). Exact values depend on the composition of the material, method of manufacture, test
method, and test conditions. Reference should be made to suppliers data sheets for precise values.
Through-thickness thermal In-plane thermal expansion
Material conductivity, W/m K coefficient, 106/K Density, g/cm3

Gallium arsenide 42 6.5 5.3

Silicon 84 2.5 2.3
Alumina 20 6.7 3.9
Aluminum nitride 165 4.5 3.3
Beryllia 260 7.2 2.9
Molybdenum 140 5.1 10.2
Titanium 15 5.6 4.5
Tungsten 174 4.5 19.3
Copper-alumina-copper 26 7.3 4.1
Copper-molybdenum-copper 166 5.5 10.0
Copper-85% tungsten alloy 180 7.2 16.1
Copper-85% molybdenum 160 6.7 10.0
Invar (Fe-36Ni alloy) 14 2.2 8.1
Kovar (Fe-29Ni-17Co alloy) 17 5.8 8.4
Aluminum-50% silicon alloy(a) 150 11.0 2.5
Aluminum-70% silicon alloy(a) 120 7.4 2.4
Aluminum-68% silicon carbide composite 150 7.2 3.0
Beryllium-30% beryllia composite (E20)(b) 210 8.7 2.1
Beryllium-51% beryllia composite (E40)(b) 220 7.5 2.3

(a) As supplied by Osprey Metals Ltd. (b) As supplied by Brush Wellman Inc.
162 / Principles of Soldering

terials have been developed having a range of Cu, which gives a reasonable boost to the ther-
controlled CTE values. A common ratio is 15% mal conductivity without greatly impairing the
thermal expansivity of the base metal. More im-
portantly, perhaps, is that the addition of a soft
18 copper phase in the otherwise refractory metal
16
matrix greatly improves the machinability. These
alloys still require considerable care to machine,
Coefficient of expansion, 106/K

14 because the copper is much softer than the re-

12 fractory metal constituent, and it is easy to cause
damage in the surface region. The principal draw-
10 back of these alloys is their high density, al-
8 though this is partly offset by their high modulus,
1.8 which means that thinner sections can some-
6 1.6 times be used, depending on the functional re-
1.4
4 1.2
quirements and design of the product. When used
1.0 in precision assemblies, these alloys need to be
2 34 35 36 37
in the annealed condition to ensure dimensional
stability on thermal cycling.
0 10 20 30 40 50 60 70 80 90 100
Nickel, % 4.2.1.3 Copper-Surface Laminates
Fig. 4.10 Expansion coefcient of iron-nickel alloys, at 20 Copper can be attached directly to alumina via
C, as a function of composition in the an- the copper/copper oxide eutectic reaction. These
nealed state
products are often marketed as direct-bonded
copper. The copper can be patterned so that it can
also fulll the function of, admittedly, a very
low-density printed circuit board, and this has
made it a very popular substrate for electronic
power modules.
A more recent development of this material
has been its commercialization in the production
of arrays of through-thickness vias, which are
lled with copper. This arrangement does not
change the in-plane thermal expansivity of the
alumina but signicantly improves the average
through-thickness thermal conductivity. Obvi-
ously, care must be taken when using this ap-
proach with high-intensity heat sources, typied
by optoelectronic and microwave devices, to en-
Fig. 4.11 Total expansion of an Fe-36Ni alloy between 220 sure that the pattern of heat studs matches the
and 250 C
point sources in the semiconductors.

Fig. 4.12 Thermal expansion characteristics of Fe-36Ni and Fe-42Ni alloys as a function of temperature and state of anneal
 Chapter 4: The Role of Materials in Dening Process Constraints / 163
Copper-molybdenum-copper and copper-
Invar-copper laminates are also available but pro-
vide a subtly different balance of properties. They
do not offer the same stiffness and patterning
abilities as the ceramic-cored alternatives and
are less widely used.
4.2.1.4 Composite Materials
Some examples of composite material are in-
cluded in Table 4.6. Each is representative of a dif-
ferent family: metal-metalloid, metal-ceramic,
and metalloid-metalloid. These are all relatively
new materials, and many variations exist on the
market. Metalloid-metalloid composites are very
immature products, and citing reliable property
values from an assessment of published values has
proved problematic. Controlled-expansion mate-
rials comprise two or more components: one, a
metal or metalloid, and the other a lower-expan-
sion metal, metalloid, or nonmetal. By varying the
relative proportions of these constituents, the
CTE values can be adjusted over quite a wide
range.
Alloys of copper-tungsten and copper-molyb-
denum, made by solid-state sintering together of
powders of the separate constituents and referred
to previously, represent controlled-expansion
composites of metal components. Metal-ceramic
composites are typied by aluminum/silicon car-
bide and beryllium-beryllia. These are of lower
density, but their principal limitation stems from
an inability for machining at high speed and ob-
taining an acceptable surface nish, because of
the substantially different hardness of the con-
stituents. Diamond tools need to be used on alu-
minum/silicon carbide because of the presence
of extremely hard silicon carbide particles. After
all, silicon carbide is itself a traditional cutting
tool material, and silicon carbide in a matrix of
aluminum constitutes a cemented carbide. In the
case of the Be-BeO materials, the particles of
BeO are poorly bonded to the beryllium metal
matrix, and machining tends to result in surface-
opening cavities. There is also a potential health
hazard should beryllia dust be generated.
For these reasons, parts fabricated in these
metal-ceramic composites tend to be produced
using near-net shaping methods. Care is taken
during the production to ensure that the part has
an outer skin of metal and that the nishing ma-
chining does not expose ceramic particles. If it
does, then there is a high chance that the particles
will either crack or pull out, and if neither of
these occur, then the presence of an exposed
ceramic phase tends to cause local adhesion prob-
lems if the part needs to be metallized.
Metalloid-metalloid composites are repre-
sented by carbon-carbon ber composites, which
were originally developed for exceptionally high-
temperature applications such as aircraft brakes
and rocket motor nozzles. They can be metal-
lized and also impregnated with copper. This is
desirable because it facilitates soldering and al-
lows the in-plane thermal expansivity to be in-
creased to more normal values. Care needs to be
taken when selecting the carbon bers and the
weave in the component. Carbon ber is a highly
isotropic material and hence, the properties of
parts fabricated by using it can vary accordingly.
Likewise, there are many grades of carbon ber,
and, as one might expect, it is the more expensive
ones that possess the most desirable and stable
properties. Carbon-ber-reinforced aluminum
and copper composites are also available, but
these are highly anisotropic and tend to be di-
mensionally unstable.
Metal-metalloid composites are perhaps best
represented by the series of silicon-aluminum
alloys, marketed as controlled-expansion (CE)
alloys and containing between 27 and 70 wt% Si,
to achieve a range of CTEs, as can be seen in Fig.
4.13 [Jacobson 2000]. These are strictly alloys,
not composites, produced from the melt by the
Osprey spray-forming process. A relatively ne,
two-phase, or duplex, microstructure of continu-
ous, interwoven matrices of silicon and alumi-
num forms naturally by eutectic solidication
during the spray-forming process (Fig. 4.14). The
silicon phase constrains the expansion of the alu-
minum, while the latter fraction is largely re-
sponsible for the transport of heat. Both silicon
and aluminum are elements of low density, with
silicon being lighter than aluminum. The CE7
alloy (Al-70Si) is closely expansion matched to
Fig. 4.13 Coefcient of thermal expansion (CTE) of Osprey
controlled-expansion alloys (based on aluminum-
silicon) as a function of the proportion of silicon, in weight percent
164 / Principles of Soldering
alumina over a wide range of temperatures, mak-
ing it suitable for packaging microwave/radio
frequency circuitry.
4.2.2 Interlayers
One route toward reducing the mismatch stress
concentration that develops in soldered and, in
particular, in brazed assemblies involves a re-
design of the joint to accommodate one or more
interlayers. There are two basic congurations
that are described in the literature.
In the rst approach, a compliant interlayer is
inserted that will yield when the joint is placed
under stress and thereby reduce the forces acting
on the components. This approach is not par-
ticularly effective with soldered joints for the
following reasons. First, the moduli of compliant
metals that are most effective in accommodating
stress (in particular, silver and copper) are too
close to those of many solders to provide much
stress relief, so the solder will tend to yield in
preference to the interlayer. Second, solders tend
to form hard, interfacial phases with most engi-
neering metals and alloys, which will confer a
high modulus to the adjacent interlayer and ac-
tually exacerbate the situation. The approach is
therefore usually only relevant where hard
gold-tin solder is mandated. Filler metals with
intrinsically low and controlled thermal expan-
Fig. 4.14 Photographs of the microstructure (micrographs) of two controlled-expansion (CE) alloys produced by spray forming,
showing their uniform phase distribution. The lighter and darker phases are primary aluminum and silicon, respectively.
(a) CE17 (Al-27wt%Si. (b) CE7 (Al-70wt%Si). Courtesy of Osprey Metals
sion coefficients are discussed in section 4.3.3.2
of this chapter.
An alternative approach for reducing mis-
match stress concentrations is to redistribute the
stresses across a much wider zone, so that the
stress is within tolerable levels everywhere in the
assembly. A graduated redistribution of stress
may be accomplished by inserting into the joint
one or more thick shims or plates that have CTEs
that are intermediate between those of the abut-
ting components. The plates must be sufficiently
thick so that they are not signicantly distorted
by the imposed stresses and therefore are not
usually less than a few hundred microns (several
mils) thick. An assembly containing a single plate
with an intermediate thermal expansivity is
shown in Fig. 4.15.
This approach is particularly suitable where
there is a need to join metals to ceramics and other
ceramic-like nonmetals. If the intermediate plate
is selected to have a thermal expansion coefficient
that is close to that of the nonmetal, then it is pos-
sible to transfer the major proportion of the stress
to the more robust metallic part of the assembly.
Where the two components have greatly different
thermal expansivities, it may be necessary to use
a graduated series of plates to reduce the mismatch
stresses in each joint to an acceptable level.
A monolithic plate of graded composition
and thermal expansivity can be used in place of
a series of discrete homogeneous plates. Typi-
 Chapter 4: The Role of Materials in Dening Process Constraints / 165
cally, these may be prepared from powder com-
pacts. Copper/tungsten components of this type
are made by inltrating a loose compact of
tungsten powder with molten copper. By adjust-
ing the packing density of the powder, the rela-
tive proportions of copper and tungsten will
vary, and the properties of the component can
vary from those of essentially pure copper to
approximately 95% W. This enables one side of
the component to be made tungsten-rich, with a
low expansion coefficient, and the other side
copper-rich, with a much higher expansion co-
efficient. Because there are no abrupt interfaces
in such a component, it can survive thermal
cycling over wide ranges of temperature almost
indenitely without suffering distortion through
creep or fatigue fracture. A graded copper-tung-
sten plate is shown in Fig. 4.16.
Fig. 4.15 Use of a plate of intermediate thermal expansivity
to reduce the stress due to thermal expansion mis-
match in an assembly between an aluminum alloy mount and the
body of a solid-state laser
Fig. 4.16 Monolithic plates of graded composition, varying
from essentially pure copper to approximately 95%
W. The thermal expansivities of the two surfaces differ by ap-
proximately 16 106/C (29 106/F).
If one or both of the components is highly
brittle and vulnerable to fracture under a tensile
or shear stress, it is often the practice to provide
reinforcement by attaching it to a more mechani-
cally robust metal plate of similar expansivity.
This subassembly can then be joined via the re-
inforcing plate to further components of different
expansivities, using approaches detailed previ-
ously. Such a conguration involving a rein-
forced subassembly is used for mounting of semi-
conductor edge-emitting lasers, because the
wavelength of the light produced is a strong func-
tion of the strain in the semiconductor material.
The following disadvantages are associated
with the use of graduated joint structures based
on the use of intermediate plates to accommo-
date mismatch stresses:
An increase in the thickness and often in the
weight of the assembly, which may be signi-
cant. This modication will also introduce ad-
ditional materials and fabrication costs.
At least two soldered joints are used in place
of a single joint. Because further materials
are introduced to the assembly, alternative
ller alloys and joining processes may need
to be developed and qualied.
The thermal and electrical conductance be-
tween the joined components is likely to be
degraded. This is a consequence of the in-
crease in the overall thickness of and number
of interfaces in the assembly. Furthermore,
materials with low expansion coefficients
tend to be poor conductors. An exception is
diamond, and a thin shim (300 m, or 12
mils) will function both as a buffer against
strain and as a heat spreader for optical and
electronic components.
The method is difficult to apply to joints that
do not have simple planar geometries.
4.2.3 Compliant Structures
Equation 4.1, given previously for calculating
the bow distortion of a bimetallic assembly, im-
plies that the mismatch stress is a sensitive func-
tion of joint dimension or, more precisely, joint
area. Although the overall size of the assembly
is likely to be xed by the functional require-
ments of the product, it may be possible to re-
place one of the monolithic components with a
lamentary, brushlike structure. Then, the di-
mensions of each individual bimetallic joint can
be made as small as necessary, thereby effec-
tively eliminating the mismatch stress from this
166 / Principles of Soldering

source, while the high aspect ratio of the la-
ments confers a degree of lateral compliance that
can accommodate the mismatch strain. Ex-
amples of these highly compliant structures are
illustrated in Fig. 4.17 and 4.18, and others are
described in the scientic and technical literature
[Huchisuka 1986].
The so-called ip-chip bonding process used
in semiconductor assembly also results in com-
pliant contacts between the components, al-
though in this case these are provided by the
actual solder joints [Yung and Turlik 1991]. Flip-
chip bonding involves electroplating or vapor
depositing solder bumps on the contact pads of
an electronic component. These bumps are heated
to reform the solder as hemispherical balls, which
are typically 0.1 mm (0.004 in.) high. In a sub-
sequent stage, the components are joined to cir-
cuit boards by reowing the solder bumps. The
entire sequence is shown schematically in Fig.
4.19. The substrate is prepared in such a manner
that solder wetting is laterally conned, in order
to maximize the height of the solder pillars that
constitute the joints. The Pb-5Sn solder is widely

Fig. 4.17 Examples of compliant structures for mitigating mismatch expansivity () of the abutting components

Fig. 4.18 (a) Longitudinal and (b) transverse sections through a compliant structure that is capable of accommodating a thermal
expansivity difference between joined components. (a) 99 . (b) 450
 Chapter 4: The Role of Materials in Dening Process Constraints / 167

used for ip-chip bonding because it has a high strengths by as much as an order of magnitude,
compliance, although its fatigue resistance is the value depending on the geometry of the joint
relatively low for a metal. The ip-chip process and the mode of stressing. This is illustrated for
is described in detail in Chapter 5, section 5.2. resistance to peel initiation in Fig. 4.20 to 4.22.
The use of compliant structures of the forms The improvement in mechanical properties may
shown obviously incurs a cost penalty, due to the be attributed to the gradual transition in geom-
greater complexity of manufacture. The conduc- etry that the llet provides in minimizing stress
tance between the components via the lamen- concentration at the joint periphery and thereby
tary member will also be impaired. Even with joint failure through crack and peel initiation at
laments having a hexagonal cross section to the surface. The stress concentration can be cal-
produce a close-packed structure, it is difficult to culated and is presented as a function of contact
obtain a compliant structure that will work ef-
fectively with a packing density of greater than
approximately 85% [Glascock and Webster
1983]. Furthermore, the ability to simulta-
neously make large numbers of small-area joints
is by no means a trivial exercise but one that
demands stringent control of tolerances and
highly specied joining processes.

4.2.4 The Role of Fillets

Wherever practicable, it is good practice to
design a joint so as to encourage the formation
of a llet. Then, even if the joint contains voids,
for whatever reason, the llet will serve to seal
the joint because there is a higher probability that
where llet formation is promoted, these tend to
be continuous and void-free. Fillets also have a
benecial effect on the mechanical properties of
joints. Well-formed llets of ller metal can en-
hance the measured tensile, shear, and peel Fig. 4.20 Typical peel force (P) prole of original geometry
joints of a at-pack module on a circuit board. The
peaks in peel strength are associated with the llets at each end
of the joint.

Fig. 4.21 Experimentally derived relationship between the

Fig. 4.19 Schematic illustration of the ip-chip joining stress required to initiate peel fracture and the
process height of the solder llet
168 / Principles of Soldering
Fig. 4.22 Typical peel force (P) proles of joints modied by increasing the solder volume and providing for a llet at both ends.
The joint strength is unchanged, but the resistance to peel initiation is greatly improved.
angle in Fig. 4.23 [Eley 1961]. Provided the sol-
der wets to form a llet with a contact angle
below 30, there is no appreciable stress con-
centration at a change in geometric prole. The
subject of stress concentration is discussed in
section 4.3.3 of this chapter.
Because it is difficult to form llets of exactly
reproducible geometry, testpieces for mechani-
cal testing are often designed to exclude llets,
either by preventing their formation through the
use of nonwettable surfaces outside the edge of
the joint or by removing any that happen to form.
Although this practice makes the measurements
more readily reproducible, it modies joint
strengths to an extent whereby they may not be
representative of most practical situations.
Examination of edge llets can provide an
indication of the ller/substrate contact angle at
the onset of solidication. This is usually taken
Fig. 4.23 Role of llets in reducing stress concentration at the
changes in section between abutting components
to be a good indication of the overall quality of
the joint. While this is generally true, visual in-
spection of edge llets can be misleading as to
the quality of the interior of the joint, as dis-
cussed in Appendix A4.2 in this chapter. Infor-
mation on internal integrity can only be obtained
by radiography, scanning acoustic microscopy,
transient thermography, or destructive methods.
4.3 Constraints Imposed by the
Components and Solutions
A large-area soldered joint may be dened as
one where the total joint area exceeds approxi-
mately 20 mm2 (0.3 in.2) and the length in any
direction is greater than approximately 5 mm
(0.2 in.). This denition is based on the following
practical criteria:
The signicant distortion of assemblies that
have joints between materials of CTE dif-
fering by as little as 5 106/C (9 106/
F). Distortion is related to the solidus tem-
perature of the ller alloy, and this problem
is therefore most acute for high-melting-
point ller alloys.
The incorporation of signicant void levels
(above 10%) in joints. This problem tends to
be more pronounced in soldered than in
brazed joints and is associated with the lower
joining temperatures that are used for these.
 Chapter 4: The Role of Materials in Dening Process Constraints / 169
Distortion of assemblies can arise from a num-
ber of causes, some of which were discussed in
Chapter 1, section 1.3.2. Apart from CTE mis-
match, the most common sources of warping or
bowing are uneven heating, which leads to tem-
perature gradients in the components, and re-
sidual stress from earlier stages of fabrication,
which is relieved in the heating cycle. Distortion
from these causes can be avoided by performing
the joining operations under carefully controlled
conditions. Notably, heating should only be car-
ried out in furnaces that provide highly uniform
temperature zones, with the rate of heating tai-
lored to the thermal mass of the components, and
after appropriate stress-relief routines have been
performed.
4.3.1 Joint Area
The strength per unit area of a joint between
components of the same material tends to reduce
in proportion to the area above some lower
threshold, often approximately 20 mm2 (0.3 in.2)
for a soldered joint. This effect can be seen in Fig.
4.24, which shows the percentage of the joint by
plan area that comprises voids as a function of
the (square) component size. There is clearly an
increasing tendency for voids to accumulate in
the joint as its area is increased, with a threshold
joint length of approximately 1 mm (40 mil).
The voids have two causes:
Gas trapped or generated within the joint
Solidication shrinkage of the molten ller
metal
Each of these is considered in further detail in the
following sections.
Fig. 4.24 Void content versus joint length for a range of rep-
resentative solders. The substrates were square
coupons of polished alumina metallized with thin-lm titanium/
gold, applied by sputtering. The joints were made at a superheat
of 25 C, and the void level was assessed by quantitative x-
radiography.
4.3.1.1 Trapped Gas
By far, the largest source of voids in soldered
joints is trapped gas, even for joints made in high
vacuum using both components and ller alloys
that have been given a vacuum bakeout prior to
the joining cycle. Void levels in soldered joints
greater than approximately 100 mm2 (1.5 in.2) in
area and that are no narrower than 5 mm (0.2 in.)
can typically reach 50% of the joint volume,
which is consistent with the entrapment of air or
an evolved gas. Figure 4.25 shows a radiograph
of a joint between a silicon chip and a metallized
ceramic package, made in high vacuum using a
solder preform, that graphically demonstrates the
problem. Figure 4.26 is a scanning acoustic im-
age of the same component and conveniently
illustrates the correspondence that can be ob-
tained between the two analytical techniques.
Air becomes trapped when the components
are assembled, with the mating surfaces forming
an effective seal. As the temperature of the as-
sembly is raised to the peak process temperature,
the trapped air is augmented by gas (usually
moisture) evolved from the joint surfaces. The
total gas volume will increase as the temperature
is raised and the ambient pressure is reduced, in
accordance with the gas law (pressure volume
 constant absolute temperature).
If the path length between a gas bubble and the
joint periphery is small, the gas pressure can
normally exceed the hydrostatic force exerted by
the molten ller metal, allowing the gas to escape
Fig. 4.25 Radiograph of a silicon chip (10 10 mm) sol-
dered into a metallized ceramic package. Voids in
the joint gap are evident as the light areas.
170 / Principles of Soldering
to the surrounding atmosphere. However, the
limit to the path length for this to occur is of the
order of 1 mm (40 mil) for soldered joints (Fig.
4.24), but it is signicantly longer for brazed
joints, because the process temperature, and
hence the pressure developed by trapped gas or
vapor, is higher. Adsorbed water is particularly
detrimental in this regard, because it expands
rapidly as it vaporizes. One method of allowing
the trapped gas to escape is to momentarily split
the joint apart during the reow process. While
this has been demonstrated as being highly ef-
fective in laboratory trials, it is not readily ap-
plicable to volume manufacture [Xie, Chan, and
Lai 1996].
Fig. 4.26 Scanning acoustic microscope image of the sol-
dered joint shown in Fig. 4.25. Voids in the joint
gap correspond to the light areas.
Fig. 4.27 Two congurations showing ow by a molten ller designed to sweep trapped gas out of a joint
When undertaking soldering using preforms,
a common mistake made when admitting the
solder into the joint gap is to use a foil preform
of similar dimensions to the plan area of the joint.
This approach typically results in a high level of
voids because of the large surface area exposed
to the atmosphere.
An effective method of removing trapped air
is to design the joint in such a manner that the
molten solder is made to ow from the center of
the joint out toward the periphery or through the
joint from one edge, as tends to occur when feed-
ing the ller into the joint from a rod or a wire
preform. Suitable arrangements for achieving this
type of ow are illustrated schematically in Fig.
4.27. The advancing front of molten solder is
then able to displace the vapor and air ahead of
it as it ows into the joint gap. However, neither
approach is entirely satisfactory. A preform of
increased thickness and reduced area should not
exceed 2 mm (80 mil) in plan, which can make
jigging of the components difficult. Moreover,
the solution of introducing the solder from one
side of the joint is only effective with ller alloys
that do not react strongly with the substrate ma-
terials to stie ow of the molten alloy in its path
through the joint (see section 4.1 of this chapter).
There are a number of solutions to this prob-
lem. A rst modication is to introduce the solder
from more than one side. The preferred congu-
ration is to place two preform discs at either side
of the joint, as illustrated in Fig. 4.28. Disc pre-
forms are readily available in a variety of sizes,
and, by using a pair, the necessity to have a
highly uid solder is diminished. The preforms
can be placed so that the voids that form will be
 Chapter 4: The Role of Materials in Dening Process Constraints / 171
in prescribed locations and their presence can be
allowed for in the design of the assembly [Weston
1974].
An improvement on this technique uses a
single foil preform in the shape of a cross, with
the arms orientated along the longest diago-
nals of the component. By doing so, the jig-
ging requirements are greatly simplied [So-
colowski 1987]. However, this method requires
a custom-made preform for each application,
and foil preforms are difficult to clean me-
chanically (to remove the native oxide) by me-
chanical abrasion prior to use (see Chapter 3,
section 3.3.5). Admitting the solder in the form
of round wire preform overcomes three of the
fundamental deciencies associated with using
foil preforms:
The surface-area-to-volume ratio is consid-
erably reduced. Correspondingly, the detri-
mental effects of surface oxide on the sur-
face of the solder are greatly diminished.
Furthermore, the solder oxide can be easily
removed by wiping the wire several times
with a paper tissue soaked in solvent (see
Chapter 1, Fig. 1.27 and Chapter 3, section
3.3.5). If this operation is conducted imme-
diately prior to the heating cycle, then voids
stemming from solder oxide can be effec-
tively eliminated.
A round wire has only one small area of con-
tact with each of the faying surfaces. This
precludes gas pockets being trapped during
jigging, and the advancing solder front is al-
ways in the optimal location to sweep out air
from the joint gap. As a side benet, solder
wire is readily available at a low price pre-
mium over the metal content, compared to
foil preforms, and a few stock sizes can be
used for a wide variety of joining applica-
tions.
Fig. 4.28 Dual discs of preform used to reduce the incidence
of voiding in large-area joints
Ductile solder wires can be cold compression
welded together and shaped as a cross (Fig.
3.30). This conguration provides a mechani-
cally stable platform for jigging as well as the
optimal uid ow pattern for the molten sol-
der to sweep out air and ll the joint gap, as
illustrated schematically in Fig. 4.29. Crosses
with short arms and thick solder wire are
preferable, from the considerations of opti-
mal uid ow and minimizing the surface-
area-to-volume ratio of the ller metal.
The solder ll of 100 mm2 (0.155 in.2) joints
could be consistently maintained at over 95%
using this approach, compared with only 45% for
the joints made using a single at preform the
same size as the joint area [Lodge, Humpston,
and Vincent 2001]. A test structure made using
this method, with the difficult process constraints
of an indium solder, uxless, and at only 10 C
(18 F) superheat, is shown in Fig. 4.30.
An aid to reducing the volume of trapped
gas is to reduce the number of surfaces in the
joint. Solders can be applied as vapor depos-
ited, electroplated, or tinned coatings to the
components, thereby eliminating two free sur-
faces. However, considerable care must be
taken to ensure that the coated layers do not
themselves contain signicant volumes of gases
or other volatile constituents. Particularly in the
electronics and photonics industries where the
piece-part cost is relatively high, the selective
application of solders by vapor-phase tech-
niques, which result in predeposited coatings of
high purity and density, can often be justied
on economic grounds if it results in improved
yields.
Original preform Solder flow
Fig. 4.29 Schematic illustration of the outward ow by a cen-
tral cross of ller metal, a conguration that helps to
prevent entrapment of vapor in pockets in a large-area joint
172 / Principles of Soldering
Substrates with solder already applied in de-
ned areas are now available commercially from
a number of suppliers. These have the advan-
tages of:
Piece-part inventory and number of suppliers
are reduced.
Jigging is likely to be simpler.
Thickness of the solder joint is decreased,
because the solder layer can be substantially
thinner than the minimum practicable thick-
ness of circa 25 m (1 mil) required for a
solder preform.
Soldering behavior is improved by eliminat-
ing two joint surfaces, with all the attendant
problems from the joint gap.
Solder spread is controlled.
Soldering operation is reproducible.
For some large components, it may be possible
to incorporate vents through the components to
provide a passage for trapped gas to escape from
within the joint. By careful design, this vent can
be further exploited to increase the effectiveness
of the pressure variation process described sub-
sequently in section 4.3.2 of this chapter [Hump-
ston et al. 2001].
Evolved vapor or gas can originate from sev-
eral sources, in particular:
Organic residues and adsorbed water vapor
on the surfaces to be joined. These species
volatilize as the temperature of the compo-
nents is raised. Residues can be minimized
by carefully precleaning the surfaces. A bake-
out in vacuum immediately prior to joining is
usually effective in removing water vapor and
organic residues, provided that the tempera-
ture used exceeds approximately 150 C (300
F). Reactive ion etching, using a hydrogen
or halogen plasma, and oxygen plasma ash-
ing are component cleaning methods that
have recently gained in popularity because of
their effectiveness at dealing with organic
contamination on surfaces and the greater
availability of off-the-shelf equipment. These
processes have the further benet of usually
involving a combination of elevated tem-
perature and reduced pressure, coupled with
a chemically active ingredient that helps re-
move volatile species.
Volatile materials within the bulk of the com-
ponents. Problems with volatile materials are
most pronounced when the components are
porous or polymeric materials and when the
components (including any metallizations
present), the solder, or the ux contain con-
stituents that volatilize during the heating
cycle. A vacuum bakeout immediately prior
to the joining cycle can help prevent subse-
quent outgassing from porous materials.
Gas evolution from polymeric materials is usu-
ally caused by thermal decomposition. The only
practical remedy in this instance is to use either
a lower-temperature process or to change the
polymeric material to one that has superior ther-
mal stability. Some materials used for printed
circuit boards are less stable than others, often in
relation to their cost. Similar considerations ap-
ply to metallic components and ller metals,
where these contain volatile elements such as
zinc, magnesium, cadmium, and, to a lesser ex-
tent, manganese, and also to uxes and pastes,
which generally contain volatile constituents.
One common method of introducing a solder
into a joint gap is in the form of a uxed paste.
If not correctly formulated and strict process con-
trols observed, joints made with uxed paste tend
to contain a higher proportion of voids than joints
made with a uxed preform. This relationship
has been studied. The voids are primarily caused
by the volatilization of entrapped ux. The pro-
pensity of voids to form increases with decreas-
ing wettability of the faying surfaces, decreasing
Fig. 4.30 Example of a large area, 10 10 mm (0.4. 0.4
in.), made uxless, at 10 C (18 F) superheat,
using In-15Pb-5Ag solder introduced in the form of a wire cross,
shown in Fig. 3.30. The joint ll is revealed by x-radiography. A
line of residual voids marks the location of the original wire cross
that the surface oxide regrew in the interval between cleaning
and melting. Courtesy of BAE SYSTEMS
 Chapter 4: The Role of Materials in Dening Process Constraints / 173

ux activity, increasing ux volume, and in- Table 4.7 Solidication shrinkage of selected
creasing plan area of the joint. The extent of elements common to widely used solders
voiding is also inuenced by the paste design, in Volume Solid
that the sooner coalescence of the metal occurs, contraction expansivity, Liquid
on solidication, linear(a) expansivity,
relative to the ability of the ux to deal with the Element % of solid 106/K cubic, 106/K
oxide present on the faying surfaces, then the
Zinc 6.9 31 167
higher will be the resulting void content. Like- Gold 5.2 14 86
wise, the use of ux media with high boiling Silver 5.0 19 97
Copper 4.8 17 100
points relative to the coalescence temperature of Lead 3.6 29 123
the solder paste tends to exacerbate the forma- Tin 2.6 23 87
tion of voids [Hance and Lee 1992]. Indium 2.5 25 96
Antimony 0.9 10 87
Bismuth 3.3 13 132

4.3.1.2 Solidication Shrinkage
In any soldered joint, a fraction of the residual
voiding does not derive from trapped air, mois-
ture, or gas. These residual voids are extremely
difficult, if not impossible, to remove because
they are intrinsic to the ller metal, being caused
by the shrinkage when it solidies and further
cools. Table 4.7 lists values for the shrinkage
volume contraction of elements common to many
solders. The reservoir of solder represented by
the edge spillage fraction is seldom able to feed
large-area joints and compensate in part for the
contraction, because the outer extremities of
joints usually solidify rst through radiative heat
losses to the surroundings.
The magnitude of solidication shrinkage, as
given in Table 4.7, accounts for the fact that it is
difficult to make joints of large area that contain
less than approximately 3 to 5% voids. Shrink-
age voids tend not to occur in small or narrow
joints (<2 mm, or 0.08 in., in one of the joint area
dimensions). This is because the thermal gradi-
ents that usually develop along a joint when the
assembly is cooled from the joining temperature
are large in relation to the joint dimensions, and
this causes the ller to directionally solidify from
one edge to the other, thereby preventing voids
from forming within the joint.
It is possible to achieve the same effect in
large-area and wide joints by imposing a tem-
perature gradient on the assembly, from either
center-to-edge or edge-to-edge, such that some
periphery of the joint is always the last portion
to solidify. However, this is not always easy to
achieve, especially when large numbers of com-
ponents are involved. Furthermore, the imposi-
tion of a temperature gradient on a large assem-
bly may produce stress gradients and thereby
dimensional distortion of the components, which
becomes xed when the solder solidies.
Where the parent materials and solder are of
similar composition, maintaining the assembly
(a) CTEs of the solids are the average values over the range 0100 C (0212
F), while the liquid CTEs are just above their melting points.
at elevated temperature but below the solidus
temperature of the ller can result in a gradual
reduction in void levels arising from solidica-
tion shrinkage by vacancy diffusion. This is the
mechanism by which dry joint interfaces are re-
moved in diffusion bonding (see Chapter 1, sec-
tion 1.1.7.2). The process times then become
relatively long and not less than 1 h.
Bismuth and, to a lesser extent, antimony are
exceptional among metals in having a volume ex-
pansion rather than a volume contraction on freez-
ing, as shown in Table 4.7. Therefore, by combin-
ing bismuth and/or antimony with other elements,
it is possible to produce solders with essentially
zero volume change on solidication. The Bi-
43Sn solder (melting point of 139 C, or 282 F)
is an example of one such alloy that has been rec-
ommended for applications in the electronics in-
dustry, where joints of guaranteed hermeticity are
required [Dogra 1985]. However, the volume
change that occurs is not instantaneous but often
takes place over several hours after the solder has
solidied [Manko 2002]. The forces accompany-
ing the volume expansion can be signicant and
must be allowed for in the joint design.
4.3.2 Void-Free Soldering
Regardless of the origin of gaseous voids, a
very successful method has been devised to com-
press the trapped gas so that it occupies a smaller
volume fraction of the joint. Because this pro-
cedure works while the solder is molten, it also
helps reduce the volume of voids arising from
the liquid-to-solid phase change through solidi-
cation shrinkage. The pressure variation method
was developed specically to reduce void levels
arising from trapped gas in adhesively bonded
joints [Bascom and Bitner 1975]. It can be used
to make large-area joints using solder preforms
174 / Principles of Soldering
that have void levels consistently below 5% [Mi-
zuishi, Tokuda, and Fujita 1988]. Most of the
residual porosity is due to solidication shrink-
age, as discussed in the preceding section. The
principle of the pressure variation method is to
use the pressure of an external atmosphere to
compress the gas trapped in the joint. The pro-
cedure, as applied to vacuum joining, is as fol-
lows (Fig. 4.31):
The components to be joined are located in
a jig and placed in the bonding enclosure at
temperature T1, which is then pumped to a
reduced pressure, P1.
The temperature of the assembly is raised to
T2 to melt the ller metal, while keeping pres-
sure P1 constant.
The pressure in the enclosure is raised from
P1 to a value of P2, which is several orders of
magnitude higher.
The assembly is allowed to cool to T1 so that
the ller solidies while the pressure is main-
tained at P2.
The voids corresponding to trapped gas are
reduced in volume roughly according to the gas
law (with suitable corrections applied to take
account of the nonideality of the particular gas
used). In the simplest case, corresponding to ide-
Fig. 4.31 Pressure variation method for reducing void levels
due to trapped gas. T, temperature; P, pressure
ality, if the initial void volume at pressure P1 is
V1, then, at constant temperature, the volume at
pressure P2 is:
V1 P1
V2 
P2
Hence, the greater is P2 in relation to P1, the more
effective is the method. This condition is also
found to apply qualitatively to practical situa-
tions.
The experimentally derived relationship be-
tween pressure variation (P2/P1) and the volume
of voids, V2, obtained using this process is shown
in Fig. 4.32. The nonlinearity of the relationship
at large P2/P1 ratios is due to departure from ide-
ality of the gas and the hydrostatic friction at the
interface between the component surface and the
liquid solder. Solder reow ovens are available
commercially that automatically perform the nec-
essary pump/pressure cycles to achieve joints
with few and controlled voids.The pressure varia-
tion method for minimizing voids is obviously not
suitable for situations where vapor is continually
being evolved from volatile constituents.
A general approach that has been found to be
effective in producing well-lled and hermetic
joints is one in which strong metallurgical re-
actions occur across the joint during the heating
cycle while a compressive force is applied (see
Chapter 3, section 3.3, for a discussion on uxless
Fig. 4.32 Experimentally derived relationship between pres-
sure variation and void level obtained in large-area
soldered joints using the pressure variation method. Adapted from
Mizuishi, Tokuda, and Fujita [1988]
 Chapter 4: The Role of Materials in Dening Process Constraints / 175
soldering).The void-free joints obtained using the
diffusion soldering and diffusion brazing pro-
cesses are associated with such reactions. The dif-
fusion soldering process is described in Chapter 5,
section 5.9, and diffusion brazing in the planned
companion volume Principles of Brazing.
4.3.3 Joints to Strong Materials
New materials that have enhanced strengths
are continually coming onto the market. Recent
examples are composite materials such as metal-
matrix composites (MMCs) and precipitation-
strengthened and dispersion-stabilized alloys.
Both of the latter types of strengthening have
been exploited in high-carat gold, suitable for
use in jewelry and bond wire for electronics
[Humpston and Jacobson 1992; Jacobson, Har-
rison, and Sangha 1996; du Toit et al. 2002].
While there is a desire to exploit these materials
in a range of applications, widespread adoption
is contingent on being able to use the favorable
bulk strength levels in joined assemblies. In gen-
eral, the strength of a joint, even when prepared
by welding, is inferior to that of the materials in
monolithic form. Moreover, the heating cycle
used in the joining process can itself degrade the
properties of these materials. For example, alu-
minum/SiC MMCs are susceptible to degrada-
tion when heated above approximately 500 C
(930 F) due to reaction between the constitu-
ents, which results in the formation of a brittle
interfacial layer of Al3C4 [Iseki, Kameda, and
Maruyama 1984]. Although solders are mechani-
cally inferior to welds and brazes, they are nev-
ertheless sometimes used to make joints that are
required to sustain moderate forces. The primary
example is joints in copper water pipes, where
the pressure can easily be 0.7 MPa (100 psi) with
a very modest pump.
In most circumstances, application of stress to
a joint does not result in all regions of the joint
sharing an equal proportion of the load. The un-
evenness of the stress distribution is referred to
as the stress concentration, K. Mathematically,
this is a dimensionless number that simply de-
scribes the magnication factor of the actual
stress at one location compared to the uniform
stress that would prevail in the absence of any
stress concentrations. Expressed as an equation:
K (stress concentration) 
Local stress at a specified location
Applied force/Joint area
The key to making high-strength joints is to pre-
vent the development of stress concentrations
and, at the same time, the strength of the ller
alloy must be maximized. These aspects are con-
sidered in turn.
4.3.3.1 Joint Design to Minimize
Concentration of Stresses
Stress concentrations can be reduced by using
joint congurations that distribute the load away
from the joint. Further details of this subject are
given in the planned companion volume Prin-
ciples of Brazing, because brazed joints tend to
be used more often than ones made with solder
for load-bearing applications.
In order to understand the origin and magni-
tude of stress concentrations that can arise, ref-
erence is made to a single-lap joint loaded in
tension. Stress concentrations arise from two
sources: namely, the differential straining of the
components and ller, and the eccentricity of the
loading path. In lap joints, the shear strength of
the joint per unit area (length) actually decreases
with increase in the joint length. This apparent
anomaly can be explained by the fact that the
shear stress is highest toward the ends of the
joint, so that if the length of the joint is increased
beyond a certain limiting value, the ller in the
central portion of the joint will carry little or no
stress, with the applied stress concentrating at
both ends, as depicted in Fig. 4.33. This explains
why simply increasing the length of the overlap
does not improve the strength of this type of joint
beyond a certain level. The stress concentration
in the joint is proportional to the length of over-
lap, up to a limiting value, the thickness of the
members, and to the thickness of the joint. There-
fore, the stress concentration is least in thin joints
of short overlap.
Far more relevant to the strength of single-
lap joints are the tensile or peeling stresses
that act normal to the ends of the joint and
originate from the eccentricity of the loading
of the assembly. The elastic analysis is rela-
tively complex, but the result obtained is that
longitudinal loading of a single-lap joint effec-
tively applies a perpendicular tensile stress of
approximately four times that amount to the
ends of the overlaps [Harris and Adams 1984].
These perpendicular tensile forces initiate fail-
ure of the joint by peel. With the continued ap-
plication of stress, the sample rotates in an at-
tempt to correct for the axial misalignment, and
176 / Principles of Soldering
Fig. 4.33 Schematic illustration of the stress distribution in the ller metal of lap joints of short and long overlap. When stressed
in shear, the central portion of a long lap joint carries little or no load.
the fracture continues to propagate due to peel-
type debonding. The stress concentrations in a
simple lap joint and their inuence on its re-
sulting failure mode are illustrated in Fig. 4.34.
Fillets at the edges of a joint act to reduce the
stress concentration in that region, as indicated
in Fig. 4.23. They do this by coupling some of
the applied stress into the ends of the laps, thereby
reducing the differential straining between the
Fig. 4.34 Failure in a simple lap joint loaded in tension. (a)
Stress concentrations. (b) Initiation of failure. Edge-
opening crack (free arrow) formed and propagated by the high
normal stress concentration. (c) Progression of joint rotation to
fracture. Plastic bending of the joint region results in the majority
of the failure being due to peel-type debonding. Adapted from
Dunford and Partridge [1990]
components and the ller in the joint and also
shifting the position of the maximum perpen-
dicular tensile stress (originating from the ec-
centricity of loading) to outside the joint. The
magnitude of these effects depends on the radius
of the llets, R, the step height, H, and the elastic
properties of the ller. To be effective, the radius
of the llets must exceed the step height, that is,
R > H, and hence, it is desirable for soldered
joints to have large and well-rounded llets at
their periphery (Fig. 4.35).
The ideal joint is one in which, under all prac-
tical loading conditions, the ller metal is stressed
in the orientation in which it best resists failure.
The complexity of the joint should also take into
account the load intensity to be sustained and any
aesthetic considerations. In general, simple, low-
cost joint designs work well with unobtrusive
joints and low-level loads, while conspicuous
joints with higher and more complex loading situ-
ations demand more elaborate and expensive con-
gurations. Strategies for some of the more com-
mon joint geometries are presented as follows.
Lap joints are probably the most common con-
guration because of their use in electronics (sur-
face mount) and plumbing (pipe joints). Increas-
ing the length of overlap will improve the ability
of the joint to resist load along its length, but,
following the law of diminishing return, this is
Fig. 4.35 A lap joint showing step height, H, llet radius, R,
and contact angle,

 Chapter 4: The Role of Materials in Dening Process Constraints / 177
because the center of the joint carries no effective
load. Further improvement can be made by ta-
pering the ends of the overlap, which is easily
achieved if the components are thin and a llet
is encouraged to form in this region. Lap-joint
styles for different stress regimes are illustrated
in Fig. 4.36(a).
Butt Joints. For a butt joint between two cir-
cular rods subject to tension, there is no stress
concentration. The strength of such joints is
therefore proportional to area. However, the plain
butt joint is only suitable for the least demanding
of applications. The main reason for this is that
the joint has very low resistance to bending
forces. The scarf butt joint has the merit of only
requiring simple machining to prepare the faying
surfaces yet is highly efficient at resisting de-
formation under load. Scarng results in differ-
ential strains and hence the stress concentrations
at the ends of the joint being considerably re-
duced, while the landed or step joint relies on the
step sizes being small to achieve the same effect.
Both congurations are symmetrical, and there-
fore, axial stresses will be balanced over the joint.
They are illustrated in Fig. 4.36(b).
By making the scarf angle sufficiently small,
the joint strength can be made to approach that
of the parent materials; that is, when the scarf
angle is 90, the joint is a butt joint, whereas if
the scarf angle is 0, there is no joint, just two
Fig. 4.36 Recommended designs of (a) lap, (b) butt, and (c) strap joints for different stress environments
parallel pieces of parent material. From a theo-
retical perspective, a radially symmetric tongue-
and-groove joint should be the best able to resist
loads, but unless the components being joined
are particularly large in diameter, achieving ad-
equate lling of the joint could be problematic,
and the component preparation costs could be
quite high.
Strap joints are often used as cheap alterna-
tives to butt joints, because the component pieces
are generally simpler, and less precision machin-
ing is required, although the thickness and weight
of the assembly are increased, and its aero/uid
dynamic performance is often impaired. Some
recommendations for different stress regimes are
shown in Fig. 4.36(c). The main problem with
this style of joint, as with lap joints, is that any
asymmetry in thickness or material properties
results in stress concentrations that cause the par-
ent material to fail prematurely just outside of the
joint region (this can erroneously be taken as an
indication that the joint is stronger than the par-
ent material). The stress concentration can be
reduced by adding taper to the straps by ma-
chining and, ideally, allowing generous and con-
tinuous llets to form.
In the preceding discussion, the assemblies
were considered to be loaded solely in uniaxial
tension. The location of any stress concentration
and its magnitude will change as the stressing
178 / Principles of Soldering
mode is altered, and hence, the optimal style of
joint varies depending on the stress environment
in which the component is required to operate.
4.3.3.2 Strengthened Solders to Enhance
Joint Strength
One of the limiting parameters of joint
strength, especially of thick joints, is that of the
solder itself. Solders can be strengthened by met-
allurgical mechanisms involving elements placed
in solid solution, microscopic second-phase par-
ticles (of either intermetallic precipitates or a
dispersed refractory phase), and renement of
the grains of the ller. These novel solders offer
signicantly improved mechanical properties,
particularly at room temperature. However,
hardly any have yet been developed to the point
of commercialization. An overview of the re-
search in this area is presented in Chapter 5,
sections 5.5 and 5.8.
Another approach is to load the solder with a
uniform distribution of coarse particles or bers
(typically, 100 m to 1 mm, or 4 to 40 mil, in
size) of a refractory or nonmetallic material. The
dimensions of the reinforcement dictate that the
joint gap must be relatively wide. In laboratory
tests using chopped carbon bers as the rein-
forcement, substantial enhancement of the shear
and tensile strength of the joints with respect to
the unmodied solders has been demonstrated
and, more particularly, a signicant reduction in
the thermal expansivity of the solder [Ho and
Chung 1990; Cao and Chung 1992]. Further de-
tails of these investigations can be found in Chap-
ter 5, section 5.6.
4.3.4 Thick- and
Thin-Joint Gap Soldering
Under normal circumstances, a solder joint
will naturally tend to be a few tens of microns
(approximately 1 mil) thick. Sometimes, it is
necessary to create joints that are either signi-
cantly thinner (<10 m, or 0.4 mil) or thicker
(>50 m, or 2 mil). Solders do not have particu-
larly good thermal conductivity, so that if a joint
is required to transport heat through its thickness,
then thin joints are obviously desirable. Thick
joints tend to be encountered where the mechani-
cal tolerance of the components does not allow
for joints to be consistently made narrower, or
where creep is desirable in order to relieve me-
chanical stress.
It is perfectly practicable to make soldered
joints that are as thin as 2 m (80 in.), even in
volume manufacturing. Joints of this thinness
require that the solder is preapplied to one of the
joint surfaces as a high-quality lm. Ion-assisted
vapor phase deposition and sputtering are the
only reliable methods of achieving this. Elec-
troplated and thermally evaporated lms are not
sufficiently dense, and the residual porosity re-
sults in the proportion of oxide being large
enough to interfere with wetting and spreading.
It is generally not possible to make thin soldered
joints by simply using a narrow joint gap and
hoping that the solder from an adjacent reservoir
area will run in and ll it. This is, again, a func-
tion of the very small volume of ller metal
whose composition will change on alloying with
the faying surfaces. Where the joint is thin, the
composition of the advancing solder front rap-
idly becomes uniformly alloyed with material
from the joint surfaces, and isothermal solidi-
cation ensues. In thin joints, liquid uxes inter-
fere with wetting and spreading, because the
volatile species have trouble escaping from a
narrow gap, as can be seen from Fig. 4.37. Nar-
row joints are therefore best made using a gas-
eous ux or uxless (see Chapter 3, sections
3.3.6 and 3.3.8, respectively). Another method of
making a thin joint is to use a more conventional
quantity of solder, appropriate to a wider joint,
Fig. 4.37 Shear strength of soldered joints in brass testpieces
as a function of joint thickness. Narrow joint gaps
are progressively more difcult to ll, thus decreasing the mea-
sured shear strength of thin joints. A gaseous ux is better able to
penetrate narrower joint gaps than a liquid ux; consequently,
thinner joints can be made before the joint-lling problem ap-
pears. Adapted from Manko [1992]
 Chapter 4: The Role of Materials in Dening Process Constraints / 179

and, once the joint surfaces have been wetted by
the molten ller, apply sufficient compressive
stress to overcome the hydrostatic pressure of the
solder to extrude surplus material from the joint
gap. Physical stops can be used to control the
nal joint gap. If the lower component is larger
than the upper one, lands can be provided to
catch the overspill in a controlled manner. The
stress required to do this reliably is of the order
of 100 g/mm2 (1 Pa, or 0.2 lb/ft2), which may
damage some components, but it does usually
enable the soldering process to be uxless (see
Chapter 3, section 3.3.8.3).
A frequently overlooked consideration when
attempting to make thin joints is the cleanliness
of the components and, particularly, the envi-
ronment in which the assembly joining is con-
ducted. When the desired joint gap is just a few
microns wide (typically, 100 in.), there is no
point jigging the components in a room where
the airborne particles are larger! For this reason,
the soldering process must be undertaken in a
semiconductor-grade clean room, and close at-
tention must be paid to the particulate content of
all process gases, cleaning chemicals, tools, and
so on. Table 4.8 shows the correlation between
the various classes of clean room and their par-
ticle size distributions. Clearly, if the require-
ment for a joint gap is below 5 m (200 in.),
then a class M4 (class 100) or better clean room
is required.
When endeavoring to make particularly thick
solder joints, the problem encountered is how to
retain the solder in the joint gap, particularly if
there is compressive stress acting on the molten
ller. The traditional method of solving this prob-
lem has been practiced for generations in the
plumbing industry. The approach is to select a
solder that has a wide melting range and to con-
duct the joining operation below the liquidus
temperature, when the ller alloy is in a pasty
state (a mixture of solid and liquid). The pres-
ence of the solid phase drastically modies the
viscosity of the alloy and prevents it from ow-
ing out of a wide joint gap. The same result can
be achieved by loading a solder with solid par-
ticles. Inserting thin parallel shims, for example,
of copper, into the joint effectively partitions the
joint gap into a series of much thinner joints and
enables conventional joining methods to be em-
ployed. Brazing of wide joint gaps is regular
practice as a crack repair technique in the aero-
space industry and involves inserting a honey-
comb into the joint gap, again to partition the
joint into a number of cells of more conventional
dimensions. Further details can be found in the
planned companion volume Principles of Braz-
ing. The successful use of this approach with
solders has not been documented.
Flip-chip interconnects sometimes make use
of joints that are thick in relation to their plan
area. For example, the solder interconnects on
ball-grid array integrated circuits (BGAICs) can
easily be 1 mm high. Here, surface tension forces
are exploited so that the solder interconnect
adopts the shape of a truncated sphere and can
therefore be tall in relation to its diameter. This
approach to thick joint gaps only works because
the packaged IC is light in relation to the total
joint area (i.e., the sum of all the individual sol-
der balls). Hence, the total surface tension force
is sufficient to levitate the IC and hence achieve
a thick joint gap. Flip-chip technology is dis-
cussed further in Chapter 5, section 5.2.

Table 4.8 Relationship between clean room class designation and airborne particle size distribution
Federal standard 209F airborne particulate cleanliness classes
Class limits
0.1 m 0.2 m 0.3 m 0.5 m 5 m
Class name Volume units Volume units Volume units Volume units Volume units
Sl English m3 ft3 m3 ft3 m3 ft3 m3 ft3 m3 ft3

M1 ... 350 9.91 75.7 2.14 30.9 0.875 10.0 0.283 ... ...
M1.5 1 1,240 35.0 265 7.50 106 3.00 35.3 1.00 ... ...
M2 ... 3,500 99.1 757 21.4 309 8.75 100 2.83 ... ...
M2.5 10 12,400 350 2,650 75.0 1,060 30.0 353 10.0 ... ...
M3 ... 35,000 991 7,570 214 3,090 87.5 1,000 28.3 ... ...
M3.5 100 ... ... 26,500 750 10,600 300 3,530 100 ... ...
M4 ... ... ... 75,700 2,140 30,900 875 10,000 283 ... ...
M4.5 1000 ... ... ... ... ... ... 35,300 1,000 247 7.00
M5 ... ... ... ... ... ... ... 100,000 2,830 618 17.5
M5.5 10,000 ... ... ... ... ... ... 353,000 10,000 2,470 70.0
M6 ... ... ... ... ... ... ... 1,000,000 28,300 6,180 175
M6.5 100,000 ... ... ... ... ... ... 3,530,000 100,000 24,700 700
M7 ... ... ... ... ... ... ... 10,000,000 283,000 61,800 1750
180 / Principles of Soldering
Appendix A4.1: A Brief
Survey of the Main
Metallization Techniques
The four main techniques that are used for
applying metal coatings to metallic and nonme-
tallic materials are as follows.
Physical vapor deposition (PVD) embraces
all methods where the coating material is physi-
cally converted into a vapor and then made to
condense onto the surface of the substrate with-
out undergoing any fundamental chemical
change in the process. The various methods are
distinguished by the means used to generate and
deposit the vapor of the coating material.
Vacuum evaporation covers those methods
where the source material is thermally vapor-
ized. This is commonly accomplished by resis-
tance heating or by electron beam bombardment.
In sputtering, by contrast, material on the sur-
face of a solid target is vaporized by bombarding
it with inert gas ions, accelerated by a potential of
500 to 5000 V.Aglow discharge in a low-pressure
atmosphere of the inert gaseither self-sus-
tained, as in cathodic sputtering, or supported
thermionically, as in triode sputteringis nor-
mally set up for this purpose.The rate of sputtering
may be increased by magnetically intensifying
the glow discharge, as in magnetron sputtering.
Reverse bias sputtering or fast atom bombard-
ment is normally available as a built-in facility for
cleaning of the substrate surfaces immediately
prior to the sputtering operation. This can consid-
erably enhance the adhesion of coatings.
Where the deposition process takes advantage
of the ionized fraction of the condensing vapor,
the process is described as ion-aided [Martin
1986]. Ion plating and ionized-cluster beam
deposition exemplify two techniques based on
this principle. Instead of generating a discharge
around a target, energetic ion beams may be
aimed directly at the surface of the substrate when
either a surface coating will be obtained, as in ion
beam deposition, or embedded within the sur-
face, as in ion implantation, at higher incident
energies. Because ion plating can develop thick
deposits of high density and purity, it is gaining
preference over evaporation as the vapor phase
process of choice for applying solders to sub-
strates.
Chemical Vapor Deposition. The deposition
of a coating by means of a chemical reaction oc-
curring from a gaseous phase on or immediately
adjacent to the surface of a substrate is known as
chemical vapor deposition (CVD). The substrate
is usually heated to generate the reaction.
Chemical vapor deposition may be classied
according to the type of chemical reaction in-
volved. In a decomposition reaction, a gaseous
compound AB may be decomposed into a solid
condensate A and a gaseous product B when
placed in contact with a colder substrate. If the
compound AB instead dissociates into a solid
phase A and a gas phase AB2, say, then the CVD
process is referred to as one involving a dispro-
portion reaction. Oxidation and reduction of ha-
lides constitute the two other types of reaction
that are widely employed.
Wet plating of metallic layers encompasses
processes where coatings are deposited on a sub-
strate through immersion of the substrate in a
liquid, usually aqueous, containing the appro-
priate metallic ions. The deposition often func-
tions by ionic discharge, with the metal depos-
ited onto an electronegatively charged,
conductive substrate (cathode). The plating pro-
cess can introduce organic compounds into the
metal coatings, although these can often be mini-
mized by judicious choice of the bath formula-
tion.
Thin coatings can be grown autocatalytically
(i.e., without an applied electric eld) through a
reduction of metal ions in the plating bath by the
immersed substrate. This process is known as
chemical displacement and also as immersion
plating. Another autocatalytic method, com-
monly referred to as electroless plating, involves
the deposition of metal from a plating bath con-
taining the metal ions together with a reductant.
This process differs from chemical displacement
in that no signicant reaction occurs within the
volume of the liquid, and the depositing metal
catalyzes further deposition, so that thicker lms
can be grown. It is usually necessary to activate
nonmetallic substrates by chemical treatment for
them to generate the catalytic reaction. Nonme-
tallic elements, principally, phosphorus and bo-
ron, tend to be incorporated into the metallic
coating from the reductant.
Thick-lm formulations usually comprise a
slurry, containing the metals or metal com-
pounds and sometimes a glass in an organic car-
rier, which is intended to be applied by painting
or screen printing onto the desired areas. Sub-
sequent ring drives off the organic fraction and
stabilizes the metallization by producing a dif-
fused interface with the nonmetal substrate. It is
 Chapter 4: The Role of Materials in Dening Process Constraints / 181

usual practice to apply and re each thick-lm
metallization separately, although processes have
been developed whereby at least two thick-lm
layers are red together. Common thick-lm met-
allizations are discussed as follows.
Systems based on reactive metals (zirconium,
tungsten, titanium, manganese, molybdenum).
These formulations are red at approximately
1600 C (2900 F) in a reducing atmosphere. Be-
cause of the relatively refractory nature of the re-
sulting metal surface, either a strongly reducing
environment is required to effect subsequent wet-
ting by solder, or a wettable metallization should
be applied on top. Alternatively, the wettable sur-
face layer may be applied over the reactive metal
layer and the two layers red together. An ex-
ample is a tungsten-loaded frit overcoated with a
nickel paste, which is cored at 1300 C (2370 F).
Even at this temperature, the interdiffusion be-
tween the two metals is slight, so that a discrete
layer of nickel forms on the surface after the heat-
ing cycle [Kon-ya et al. 1990].
Systems based on noble metals (copper, silver,
gold, palladium, platinum). These materials are
red between 850 and 950 C (1560 and 1740
F). The silver, gold, and platinum metallizations
can be red in air, while a reducing atmosphere
is generally required for the less noble metals.
Metal-loaded glass frits are red on the sur-
faces of components above 400 C (750 F) to
form a glaze that is strongly adherent to the non-
metal substrate. The concentration of the glass at
the interface with the component means that the
outer layer of the coating is sufficiently metallic
in character for it to be electroplated or directly
soldered.
Thick-lm metallizations are supplied as com-
plex proprietary formulations and are available
in different physical forms, each tailored for a
limited range of substrate materials. It is advis-
able to consult the supplier on their conditions of
use and likely properties.
As might be expected, there are advantages
and disadvantages associated with the different
metallization techniques. Vapor deposition is
generally superior to wet plating in offering bet-
ter control of impurities and reduced porosity in
thin coatings. Wet plating, by comparison, tends
to be faster and cheaper and can provide thicker
coatings. Thick-lm metallizations may be more
readily applied to selective areas and are more
tolerant of substrate topology. The technique that
will normally be chosen will be the one best
suited to the particular application on the grounds
of its tness for purpose and cost. A brief com-
parison of the characteristics of the principal
methods used for applying metallizations, to-
gether with those of the coatings that they are
capable of producing, is presented in Tables 4.9
to 4.12. It must be pointed out that the entries in
the tables represent the general situation; par-
ticular cases might be out of the ranges indicated.

Table 4.9 Techniques for applying metallizations: characteristic features

Film File
Metallic materials Suitable Throwing thickness thickness Throughput
Process capable of deposition substrates power achievable control of process

Vacuum evaporation Elemental metals and Most Line-of-sight nm-m Good Low, batch
some alloys nonvolatile process
materials
Sputtering Wide range of Most Moderate nm-m Excellent Low, batch
elemental metals nonvolatile (function of
and alloys materials target size,
gas pressure,
and target-
substrate
distance)
Chemical vapor Elemental metals Materials that Good m-mm Good, but High, many
deposition can withstand need to items at a
the high stringently time; batch
temperatures control or
required several continuous
process
variables
simultaneously
Electroplating Elemental metals Electrical Moderate m-mm Generally Very high,
and some binary conductors less precise can be
and ternary alloys than for continuous
vapor
deposition
Electroless plating Elemental metals and Wide range Good m Generally High, can be
a few binary alloys of materials poor continuous
Thick lm Wide range of Materials that Physical access m-mm Moderate High, batch
elemental metals can withstand to surfaces is or conveyor
and alloys the ring required belt
temperatures
182 / Principles of Soldering

Further information on these metallization interface structure and properties of metal/

techniques can be found in the literature, which ceramic interfaces, property measurement, and
is extensive. A comprehensive review that cov- their fracture behavior is given by Howe
ers the chemical bonding, chemical reaction, [1993]. Other useful sources of information are

Table 4.10 Metallization techniques; relative merits

Process Advantages Disadvantages

Vacuum evaporation Relatively simple equipment required for Not suitable for alloys that have constituents with
resistance heating evaporation, which is greatly differing vapor pressures. Meticulous
suitable for coatings of most substrate cleaning prior to deposition is required.
elemental metals
Sputtering Possible to coat a wide range of compositions Requires sophisticated equipment. Low
Dense coatings and good adhesion obtainable throughput. Heating of substrates and low
deposition rates in conventional diode or triode
sputtering
Chemical vapor High-quality coatings are obtainable. Output Equipment is sophisticated and is usually specic
deposition is generally high. to particular coatings.
Electroplating High throughput. Large areas can be coated Chemical handling, vapor, and effluent problems.
with uniform thickness. Limited only Film impurities and imperfections can also present
by the size of the plating bath. Relatively problems. Can only apply coatings to electrically
easy to control conductive materials. Thorough cleaning and
chemical activation of substrates are required
prior to plating.
Electroless plating Large areas can be coated with uniform As above, except that nonconducting materials can
thickness. be plated. Range of available coatings is
Good throwing power. Only very basic equipment restricted.
is required.
Thick lm Requires simple equipment. Lends itself to high Relies on manufacturers proprietary formulations.
volume production using screen printing and Relatively high process temperatures are used.
ring in belt furnaces Only thick lms can be applied by this technique.

Table 4.11 Metallization techniques: important process parameters

Rate of deposition, Pressure in deposition Substrate temperature
Process m/min chamber, mPa during coating process

Vacuum evaporation 0.0015 0.0110 Substrate is often heated to 200 C (390 F)

to promote adhesion
Sputtering 0.0051 10010,000 Mostly below 100 C (212 F)
Chemical vapor deposition 5100 10,000100,000 2002000 C (3903630 F), but usually
400800 C (7501470 F)
Electoplating 0.1100 Ambient 10100 C (50212 F)
Electroless plating 0.11 Ambient 10100 C (50212 F)
Thick lm 1,00010,000 (does not Ambient 4001800 C (7503270 F)
include ring times)

Table 4.12 Metallization techniques: coating quality

Coating thickness Coating Coating Coating adhesion to
Process uniformity continuity purity substrate

Vacuum Variable; determined by Moderate to low Purity limited by source Fair

evaporation source-substrate porosity materials and
geometry deposition atmosphere
Sputtering Higher uniformity Low porosity Purity limited by source Generally excellent
possible than for materials and
vacuum evaporation deposition atmosphere
Chemical vapor Good uniformity Dense and essentially Purity is that of the Variable; dependent
deposition possible; depends on pore-free starting materials or on materials and
design of the even better processing conditions
deposition chamber
Electroplating Good uniformity on Susceptible to porosity May incorporate salts Variable; often excellent
ats, nonuniform at and blistering and gaseous inclusions
edges
Electroless plating Fair uniformity Susceptible to porosity May incorporate salts Variable; often excellent
and blistering and gaseous inclusions
Thick lm Variable Dense coatings are Often contain glass and Variable; dependent on
achievable possibly organic materials and
residues processing conditions
 Chapter 4: The Role of Materials in Dening Process Constraints / 183
given in the selected bibliography appended to
the preface.
Appendix A4.2: Critique of
Void-Free Soldering
Standards
In the attachment of semiconductor die to sub-
strates by soldering, it is frequently a require-
ment that the resulting joint be free of voids. It
is generally true that voids in the joint between
an active device and its heat sink will result in
hot spots and the premature failure of the com-
ponent. Target criteria for the inspection of joints
are dened by standards.
Current inspection standards for void-free
joints are specied in Military Standard (MIL-
STD) 883D, Methods 2010.10 and 2012.7. These
standards are echoed in British Standard (BS)
9450 and European Space Agency (ESA) speci-
cations 20400, 2045000, 4045000, and
2099000. Taken together, they require the fol-
lowing to be met:
>75% of the perimeter to exhibit a llet
>75% of each side to exhibit a llet
Two or more sides to exhibit complete llets
<50% voids in the joint
No single void to traverse either the length or
width of the joint, or to be >10% of the joint
area
Although these standards provided well-
dened criteria for judging an assembly process,
they are not well suited for the purpose of as-
sessing void-free joints. In particular, there is no
external indication of dewetting in the joint gap,
and 49% voids is arguably a somewhat high level
for a joint that is judged to be void-free.
The aforementioned inspection criteria re-
late mostly to visual inspection of the joint and
therefore its external appearance. To properly
assess whether or not a joint is free of voids
requires inspection of the joint interior. This is
often possible using scanning acoustic or x-ray
techniques (see Chapter 5, section 5.10). Stan-
dards exist that dene what constitutes a de-
fect when voids are detected using these in-
spection methods. For example, ESA x-ray
Standard (2099000) species an unacceptable
joint as one containing:
Voids in excess of one-half of the total plan
area
A single void equal to the length of the joint
A single void that traverses the width of the
joint
Again, these criteria permit a joint to contain
large voids that potentially can be responsible for
catastrophic failure yet is able to pass inspection.
For application where void-free joints are es-
sential to the functionality and reliability of the
product, proposed inspection criteria are as fol-
lows:
Visual Inspection. A sensible inspection cri-
terion for void-free joints that relies only on vi-
sual examination is the presence of a complete
and uninterrupted llet around the joint periph-
ery. Although this condition does not guarantee
zero voids within the joint, it does at least pro-
vide assurance that the external surfaces achieved
and maintained wettability during the process
cycle. It also indicates that the joint is leaktight.
According to the same rationale, it is desirable
that the llets are smooth, shiny, and have low
contact angles. A smooth and shiny solder sur-
face is an indicator of the absence of an oxide
skin. Any blemishes, such as nonuniform hue,
discoloration, bumps, cracks, craters, or foreign
material, give cause for concern as to the integ-
rity of the hidden volume of the joint. Low llet
contact angles are desirable as an indicator of
good wetting, but this inspection criterion is not
usually available because solder spread often is
conned by lands, causing the contact angle to
be higher than it would be if the solder were able
to spread freely.
X-Ray or Scanning Acoustic Microscope
(SAM) Inspection. The most demanding target
that can sensibly be set for joints larger than a
few millimeters per side is <5% voids by plan
area. A truly void-free joint is not seen to be an
achievable target in volume production, not least
because even state-of-the-art assessment tech-
niques can lead to ambiguity in identifying a
void, especially when using thin, predeposited
solder and thick components.
For each application, it is should also be pos-
sible to dene a maximum acceptable void size
(either by area or linear dimension) in various
areas of the joint. This gure can be obtained
from thermal and physical modeling. Void frac-
tion, maximum void size, and their impingement
into areas where solder is essential can be gen-
184 / Principles of Soldering
erated automatically by modern x-ray and SAM
equipment in real-time.
The bottom line in devising an inspection strat-
egy is to dene a set of criteria that can offer
assurance of product integrity and thereby add to
its value and reputation for reliability.
Appendix A4.3: Dryness
and Hermeticity of Sealed
Enclosures
A frequent requirement of electronics, optics,
and microelectromechanical systems packaging
is to provide an enclosure that is dry and resistant
to ingress of moisture during the lifetime of the
product. Attainment of such enclosures is gen-
erally referred to as hermetic packaging.
The rst step toward meeting this objective is
to choose a suitable material for the package
construction. No material is truly hermetic to
moisture, although glasses, ceramics, and metals
are obviously superior to any plastic material.
Figure 4.38 provides an indication of the relative
ranking of different classes of materials to
through-thickness moisture penetration, which
accounts for the choice of metal and ceramic
packages for the most demanding applications,
and polymeric packaging solutions for less du-
rable and less cost-tolerant consumer products.
The contents of the package need to be xed
in place, using materials that will not later outgas
Fig. 4.38 Predicted time for moisture to permeate various
packaging materials in one geometry. Adapted
from Traeger [1976]
and contaminate the package atmosphere. Sol-
der, particularly when used uxless, is ideal for
this task. Gold-silicon solder, which is frequently
used for semiconductor die attach applications,
acts as a desiccant after the package is sealed,
because the silicon in the alloy reacts with water
vapor. Tests show that preforms of this material
reduced the moisture content in the package from
15,000 ppm for control samples to 125 ppm [Car-
ley, Nearhoff, and Dennin 1984]. The hydrogen
content in the package was found to increase
from 1800 to 7000 ppm, implying that the likely
reaction taking place is:
Si  2H2O SiO2  2H2
Adhesives have been formulated that do not
greatly outgas. However, care must be taken with
their use in sealed enclosures, because they are
only stable in this regard provided they are not
heated beyond the curing temperature. For ex-
ample, if a semiconductor die is attached using
an epoxy adhesive, but the package is then sealed
using a high-temperature solder, the adhesive
will liberate moisture as the sealed package cools
from the lid-seal process temperature.
Next, the package needs to be dried before it
is sealed. Water adsorbed onto metals needs to be
heated to 115 C (239 F) in order to force it to
desorb. One widespread misconception is the be-
lief that a low moisture reading on a sealing box
hygrometer will ensure a dry ambient inside the
package. The fallacy arises because the major
source of moisture is that adsorbed onto the pack-
age walls, and thus, the dry box reading has no
correlation with the resulting package ambient
atmosphere. Removing the water of hydration
alone is not sufficient. Although this moisture
will desorb at 115 C (239 F), it is not until a
metal surface is heated above approximately 160
C (320 F) that the absorbed hydroxide species
will convert to metal oxide, to the accompani-
ment of further evolution of water [Swartz et al.
1983], according to the reaction:
M(OH)2 MO  H2O
where M represents the metal.
Figure 4.39 shows the moisture content in a
metal package as a function of the bakeout time
and temperature. Low residual moisture content
is obviously favored by high bakeout tempera-
tures sustained for long periods.
 Chapter 4: The Role of Materials in Dening Process Constraints / 185
A problem frequently encountered is that the
package contents or assembly method do not
permit the use of elevated temperature for the
bakeout, while manufacturing economics are not
compatible with extended bakeout times at a re-
duced temperature. One solution to this problem
is to place the package in a sealed chamber and, at
the highest permissible temperature, repeatedly
saturate the atmosphere with dry nitrogen gas, and
then pump it out. Nitrogen gas obtained from a
cryogenic source and conveyed by stainless steel
pipework will have a moisture content of 2 ppm
or less. The moisture on the package walls will
equilibrate with that in the atmosphere, so that
when the nitrogen gas is removed, it will take
moisture with it. Twenty such cycles, conducted
over a period of a few hours, will reduce the mois-
ture content of the package to a few parts per mil-
lion. This approach has the additional advantage
of reducing the time that the package is exposed
to vacuum, which can result in hydrocarbon con-
tamination of surfaces arising from backstream-
Fig. 4.39 Moisture content of metal packages as a function
of bakeout time and temperature. Adapted from
Thomas [1976]
Fig. 4.40 Calculated ingress of water vapor into a hermeti-
cally sealed package having a leak rate of 108 Pa
m3/s (8 1010 ft lb/s). The package volume is 2 mm3 (1.2
104 in.3), and the ambient air is assumed to be at 70% relative
humidity.
ing of vacuum pump oil vapors. Note that the
package must be sealed while still in the bakeout
chamber because, if it is exposed to air, the sur-
faces will instantly become saturated with water
again, even if they are kept hot.
Having hermetically sealed the package, it
might be expected that the contents would now
be permanently protected from the atmosphere
outside. This is not true because, in practice,
hermeticity can only be specied in terms of a
nite leak rate. Small gas molecules will enter
the sealed cavity by diffusion or permeation un-
til, ultimately, equilibrium with the atmosphere
outside is reestablished. Package leak rates are
measured using helium as a tracer gas, and the
acceptable leak rate is dened for the specic
application. For example, a package having an
internal volume of 100 mm3 and a leak rate of 5
107 Pa m3/s (4 108 ft lb/s) will have a
theoretical water buildup rate of roughly 104 ppm
per day, indicating that the package atmosphere
will be in equilibrium with the air in a matter of
days. At a leak rate of 109 Pa m3/s (8 1011
ft lb/s), equilibrium will be achieved within a
year. Figure 4.40 shows predicted moisture pen-
etration rates for a helium leak rate of 108 Pa
m3/s (8 1010 ft lb/s) for ambient air at 70%
relative humidity [Stroehle 1997].
A helium leak rate of 109 Pa m3/s (8 1011
ft lb/s) is generally considered to be a minimum
acceptable level of hermeticity, and 1011 Pa
m3/s (8 1013 ft lb/s) to provide adequate
protection against moisture ingress for all but the
most sensitive components. The latter leak rate
is also the lower detection limit of low-cost he-
lium leak check equipment, while 1014 Pa m3/s
(8 1016 ft lb/s) is about the limit of spe-
cialized laboratory systems.
REFERENCES
Ashby, M.F., 1994. Materials Selection in
Mechanical Design, Pergamon Press
Bascom, W.D. and Bitner, J.L., 1975. Void
Reduction in Large Area Bonding of IC Com-
ponents, Solid State Technol., Vol 9, p 3739
Bever, M.B., 1986. Encyclopedia of Materi-
als Science and Engineering, Pergamon
Press, p 24632475
Cao, J. and Chung, D.D.L., 1992. Carbon
Fiber Silver-Copper Brazing Filler Compos-
ites for Brazing Ceramics, Weld. J., Vol
71 (No. 1), p 21s24s
186 / Principles of Soldering
Carley, D.R., Nearhoff, R.W., and Denning,
R., 1984. Moisture Control in Hermetic Lead-
less Chip Carriers with Silver-Epoxy Die At-
tach Adhesive, RCA Review, Vol 45 (No. 2),
p 278290
Dogra, K.S., 1985. A Bismuth Tin Alloy for
Hermetic Seals, Brazing Soldering, Vol 9
(No. 3), p 2830
Dunford, D.V. and Partridge, P.G., 1990.
Strength and Fracture Behavior of Diffusion-
Bonded Joints in Al-Li (8090) Alloy. Part 1:
Shear Strength, J. Mater. Sci., Vol 25, p 4957
4964; Part 2: Facture Behavior, 1991, Vol 26,
p 26252629
du Toit, M. et al., 2002. The Development of
a Novel Gold Alloy with 995 Fineness and
Increased Hardness, Gold Bull., Vol 35, p
4652
Eley, D.D., 1961. Adhesion, Oxford Univer-
sity Press
Evans, D.S. and Prince, A.P., 1982. The Ef-
fect of Gold on the Pb-1.5%Ag-1%Sn Sol-
der, Mater. Res. Bull., Vol 17, p 681687
Glascock, H.H. and Webster, H.F., 1983.
Structured Copper: A Pliable High Conduc-
tance Material for Bonding Silicon Power
Devices, IEEE Trans. Components Hybrids
Manuf. Technol., Vol 6 (No. 4), p 460466
Hance, W.B. and Lee, N.-C., 1992. Forma-
tion and Control of Voiding in SMT, Proc.
Conf. 1992 International Symposium on Mi-
croelectronics, (San Francisco, CA), 1921
Oct, p 535539
Harris, J.A. and Adams, R.D., 1984. Strength
Prediction of Bonded Single Lap Joints by
Non-Linear Finite Element Methods, Int. J.
Adhesion Adhesives, Vol 4 (No. 2), p 6578
Haug, T., Schaefer, W., and Schamm, R.,
1989. Joining Electrochemical High Tem-
perature Components, Proc. Sixth Interna-
tional Conference on Joining Ceramics,
Glass and Metals (Bad Nauheim, Germany),
p 171178
Ho, C.T. and Chung, D.D.L., 1990. Carbon
Fiber Reinforced Tin-Lead Alloy as a Low
Thermal Expansion Solder Preform, J. Mater.
Res., Vol 5 (No. 6), p 12661270
Holloway, P.H., 1980. Gold-Chromium Met-
allizations for Electronic Devices, Solid State
Technol., Vol 2, p 109115
Howe, J.M., 1993. Bonding, Structure and
Properties of Metal/Ceramic Interfaces (Part
1 and Part 2), Int. Mater. Rev., Vol 38 (No. 5),
p 233271 (entire issue)
Huchisuka, T., 1986. Bonding of Sintered Al-
loys, Met. Technol., Vol 56 (No. 5), p 2127
Humpston, G. et al., 2001. Vented Large
Area Solder Seal, United States invention
disclosure 10020016
Humpston, G. and Jacobson, D.M., 1989.
Gold in Gallium Arsenide Die-Attach Tech-
nology, Gold Bull., Vol 22 (No. 3), p 7991
Humpston, G. and Jacobson, D.M., 1990.
Solder Spread: A Criterion for Evaluation
of Soldering, Gold Bull., Vol 23 (No. 3), p
8395
Humpston, G. and Jacobson, D.M., 1992. A
New High Strength Gold Bond Wire, Gold
Bull., Vol 25, p 132145
Iseki, T., Kameda, T., and Maruyama, T.,
1984. Interfacial Reactions Between SiC and
Aluminum During Joining of MMCs, J.
Mater. Sci., Vol 19, p 16921698
Jacobson, D.M., 2000. Spray-Formed Sili-
con-Aluminum, Adv. Mater. Process., Vol 157
(No. 3), p 3639
Jacobson, D.M., Harrison, M.R., and Sangha,
S.P.S., 1996. Stable Strengthening of Gold,
Gold Bull., Vol 29, p 95100
Kon-ya, S. et al., 1990. New Metallizing Pro-
cess of Alumina Ceramics for Hermetic Seal-
ing, Proc. Third Electronic Materials and
Processing Congress, 2023 Aug (San Fran-
cisco, CA), p 1924
Kuhmann, J.F. et al., 1977. Pt Thin-Film Met-
allization for Fluxless FC-Bonding Using
SnPb60/40 Solder Bump Metallurgy, Proc.
11th European Microelectronics Conference,
1416 May (Venice), p 385391
Leite, E.R., Varela, J.A., and Longo, E., 1992.
Barrier Deformation of ZnO Varistors by Cur-
rent Pulse, J. Appl. Phys., Vol 72 (No. 1), p
147150
Li, J. and Krsulich, V., 1996. Metal Alloy
Applied in Ceramic Package Lids Reduces
Stress, Semicond. Int., Feb, p 105110
Lodge, K.J., Humpston, G., and Vincent, J.H.,
2001. Void-Free, Flux-Free Solder: A Method
of Soldering, U.K. Patent 0104577.2
Mackay, C.A. and Levine, S.W., 1986. Sol-
der Sealing Semiconductor Packages, IEEE
Trans. Components Hybrids Manuf. Tech-
nol., Vol 9 (No. 2), p 195201
Manko, H.H., 2002. Solders and Soldering,
4th ed., McGraw-Hill
Martin, P.J., 1986. Ion-Enhanced Adhesion
of Thin Gold Films, Gold Bull., Vol 19 (No.
4), p 102116
 Chapter 4: The Role of Materials in Dening Process Constraints / 187
Mattox, D.M., 1973. Thin Film Metallization
of Oxides in Microelectronics, Thin Solid
Films, Vol 18, p 173186
Mattox, D.M. and Smith, H.D., 1985. Role of
Manganese in Metallization of High Alumina
Ceramics, Ceram. Bull., Vol 64 (No. 10), p
13631367
Mizuhara, H. and Mally, K., 1985. Ceramic-
to-Metal Joining with Active Brazing
Filler Metal, Weld. J., Vol 64 (No. 10), p
2732
Mizuishi, K., Tokuda, M., and Fujita, Y.,
1988. Fluxless and Virtually Voidless Sol-
dering for Semiconductor Chips, IEEE Trans.
Components Hybrids Manuf. Technol., Vol
11 (No. 4), p 447451
Palmer, A., Alexander, C.N., and Nguyen, B.,
1999. Forming Solder Joints by Sintering Eu-
tectic Tin-Lead Solder Paste, J. Electron.
Mater., Vol 28 (No. 7), p 912915
Pearshall, C.S., 1949. New Brazing Methods
for Joining Nonmetallic Materials to Metals,
Mater. Meth., Vol 30 (No. 6), p 6162
Peteves, S.D., 1988. Designing Interfaces for
Technological Applications: Ceramic-Ce-
ramic, Ceramic-Metal Joining, Elsevier
Ramirez, A.G., Mavoori, H., and Jin, S., 2002.
Bonding Nature of Rare-Earth-Containing
Lead-Free Solders, Appl. Phys. Lett., Vol 80
(No. 3), p 398400
Ronka, K.J., Van Loo, F.J.J., and Kivilahti,
J.K., 1998. A Diffusion-Kinetic Model for
Predicting Solder/Conductor Interactions in
High Density Interconnections, Metall.
Mater. Trans. A, Vol 29, p 29512956
Rosenstock, Z. and Riess, I., 2000. Prepara-
tion of Oxide Thin Films by Controlled Dif-
fusion of Oxygen Atoms, Solid State Ionics,
Vol 136 (No. 7), p 921926
Smith, R.W., 1998. Low Temperature Active
Joining for Electronics and Medical Devices,
Proc. International Conference on Advances
in Welding Technology, 30 Sept to 2 Oct (Co-
lumbus), p 185203
Socolowski, N.J., 1987. Soldering Preform
and Methods of Employing the Same, U.S.
Patent 4,709,849
Stroehle, D., 1997. On the Penetration of Wa-
ter Vapor into Packages with Cavities and on
Maximum Allowable Leak Rates, Proc. 15th
Annual Reliability Physics Symposium, April,
p 101106
Swartz, W.E. et al., 1983. The Adsorption
of Water on Metallic Packages, Proc. 21st
Annual Reliability Physics Symposium, p
5259
Thomas, R.W., 1976. Moisture, Myths and
Microcircuits, IEEE Trans. on Parts, Hybrids,
and Packaging, Vol 12 (No. 3), p 167171
Timoshenko, S., 1925. Analysis of Bi-Metal
Thermostats, J. Opt. Soc. Am. Rev. Sci. In-
strum., Vol 11 (No. 9), p 233255
Tokuriki Hoten, K.K., 1981. Composite Au-
Sn Eutectic Type Solder Alloy, Japanese
Patent JP 81199063
Traeger, R.K., 1976. Hermeticity of Poly-
meric Lid Sealant, Proc. 25th Electronics
Components Conference, April, p 361367
Vianco, P.T., 1998. An Overview of Surface
Finishes and Their Role in Printed Circuit
Board Solderability and Joint Performance,
Circuit World, Vol 25 (No. 1), p 624
Vianco, P.T., Sifford, C.H., and Romero, J.A.,
1997. Resistivity and Adhesive Strength of
Thin Film Metallizations on Single Crystal
Quartz, IEEE Trans. Ultrasonics Ferroelec-
trics Frequency Control, Vol 44 (No. 2), p
237249
Warwick, M., 2002. Tombstone Reduction
via Phased Reow Soldering, Surf. Mount
Technol., Vol 16 (No. 9), p 44-54
Wersing, W., 1992. Improved Ceramics
Through Multilayer Technology, Metals and
Mater., Vol 8 (No. 6), p 326331
Weston, A.D., 1974. Mounting Semiconduc-
tor Bodies, U.S. Patent 3,786,556
Xian, A. and Si, Z., 1992. Interlayer Design
for Joining Pressureless Sintered Sialon Ce-
ramic and 40Cr Steel Brazing with
Ag57Cu38Ti5 Filler Metal, J. Mater. Sci.,
Vol 27 (No. 3), p 15601566
Xie, D.J., Chan, Y.C., and Lai, J.K.L., 1996.
An Experimental Approach to Pore-Free Re-
ow Soldering, IEEE Trans. Components Hy-
brids Manuf. Technol., Vol 19 (No. 1), p 148
153
Yung, E.K. and Turlik, I., 1991. Electroplated
Solder Joints for Flip-Chip Applications,
IEEE Trans. Components Hybrids Manuf.
Technol., Vol 14 (No. 3), p 549559
Principles of Soldering
Giles Humpston, David M. Jacobson, p189-242
DOI:10.1361/prso2004p189
CHAPTER 5
Advances in Soldering Technology
To many readers, this chapter is something of
a disappointment, because it does not contain
any sparklingly new solders, uxes, metalliza-
tions, processes, or diagnostic tools. The reason
for this is quite simply that, as far as the authors
are aware, there have been no signicant com-
mercial developments in the 10 years since the
rst edition of Principles of Soldering and Braz-
ing went to press. Even the hot topic of the
1990slead-free soldersis based on alloys that
were known and largely characterized previ-
ously. Soldering is a relatively mature technol-
ogy, so that while there has been progress and
many undoubted improvements made in recent
years, these have tended to be evolutionary rather
than revolutionary in nature. Composite and
doped solders are, perhaps, some of the few ex-
amples to fall into the latter category, but they are
still a long way from becoming a commercial
reality. Indeed, one might even argue that with
modern health and environmental awareness,
there has actually been a contraction in the num-
ber of soldering materials in general industrial
use in recent years. With the proscription of ma-
terials recognized to pose health hazards, cad-
mium- and lead-containing solders along with
natural rosin-based uxes, volatile organic com-
pounds, and cleaning agents containing chloro-
uorocarbons have either disappeared or are in
the process of being removed from manufactur-
ers catalogs.
This chapter endeavors to present a number of
materials and processes, none of which are en-
tirely new but by virtue of changing industrial
need or technical innovation are likely to be
rather different from the processes the reader
may have been aware of some years ago. The
primary example is, of course, the knowledge
based on lead-free solders. The others are a rep-
resentative selection made by the authors and felt
to be useful or having appeal as laboratory curios
Copyright 2004 ASM International
All rights reserved.
www.asminternational.org
awaiting industrial application but, inevitably,
may not accord with some readers preferences.
To some extent, the new material also is intended
to ll gaps in the previous edition of Principles
of Soldering and Brazing or to provide supple-
mentary detail to the preceding chapters. This is
the reason for the inclusion of sections devoted
to ip-chip processes, diffusion soldering, and
modeling. Scanning acoustic miscroscopy
(SAM) and ne-focus x-ray techniques are in-
cluded because of the considerable technical ad-
vance that has been made with these diagnostic
tools in recent years, coupled with the dramatic
reduction in price of the equipment. It is now by
no means uncommon to see ranks of high-
volume production lines, each with a dedicated
SAM and microfocus x-ray system at the end,
performing 100% product inspection with totally
automated defect-recognition software. Allied
with these facilities are image-recognition sys-
tems that read individual chip, resistor, and ca-
pacitor values and check that printed circuit
boards (PCBs) have been populated with the
specied components in the desired orientation.
The cumulation of these advances is industries
such as the mobile phone market, where total
handset production now exceeds 2 billion units,
achieved in less than 20 years.
5.1 Lead-Free Solders
The literature on lead-free solders is almost
overwhelming. A computer search on the term
lead-free quickly reveals tens of thousands of
technical papers and references to conference
presentations on the subject. This depository of
knowledge is augmented by entire issues of jour-
nals, chapters in several books, and entire books
[e.g., Hwang 2001]. Thus, there is much valuable
190 / Principles of Soldering
technical information on the metallurgy, physi-
cal and chemical characteristics, and reliability
of joints made using lead-free solders. Extensive
as it is, these collective works often do not ad-
equately describe the process window for a par-
ticular joining problem of interest nor predict the
reliability of the resulting joints. This is, perhaps,
not surprising, given such key information is still
not immediately available for lead-tin solder, de-
spite many years of endeavor. Many of the lead-
free alternatives are technically more complex
materials.
The principles of lead-free soldering are not
fundamentally different from those of lead-tin
soldering. Lead-free soldering processes are now
a commercial reality, and within very few years,
it is likely that the vast majority of electronics
and optoelectronics products will be manufac-
tured using this new technology.
5.1.1 The Drive for
Lead-Free Soldering
The drive for lead-free soldering for mass-
market electronics assembly rst came to promi-
nence in the late 1980s with Senator Reads bill
in the United States Senate [U.S. Senate bill S391
1990; U.S. Senate bill S729 1993]. This bill
sought to address the concern for human health
posed by the rapidly increasing quantities of dis-
carded electronics equipment accumulating in
landll waste disposal sites. Although battery
manufacture accounts for the vast majority of the
5.5 million tonnes (6.06 million tons) of lead
consumed each year, battery recycling is almost
100% effective. However, virtually none of the
60,000 tonnes (66,000 tons) of lead used by the
electronics and lighting industries is recycled.
The waste in landll sites is subject to chemical
attack by rainwater, from where leached-out con-
stituents, including lead, eventually nd their
way into drinking water supplies [Smith and
Swanger 1999].
In 1998, the European Union introduced a draft
directive, which is a precursor to law, entitled
Waste Electrical and Electronic Equipment
(WEEE) [The European Commission 2000; The
European Commission 2001]. This followed
Senator Reads initiative and called for a ban on
lead in all electronics, except for automotive use,
by 1 January 2004. The WEEE directive was
intended to ban the selling, importing, and ex-
porting of electrical/electronic equipment con-
taining lead. The extent to which this directive
was politically motivated may be the subject of
informed debate, but it is noteworthy that in the
preceding January, the Japanese Electronic In-
dustry Development Association (JEIDA) and
the Japanese Institute of Electronic Packaging
(JIEP) presented a roadmap to totally lead-free
technology by April 2001.
The combined effects of these proposals was
to put pressure on the rest of the world to follow
suit and generated a large investment in solder-
ing process, equipment, and materials develop-
ment. Many tens of thousands of hours were
committed to research and development of lead-
free solders during the 1990s. Although the ef-
forts were somewhat uncoordinated, the conclu-
sions of these studies conducted across the globe
were remarkably similar: Lead-free soldering
was seen to be a technical possibility, and indeed,
further work has resulted in the commercial avail-
ability of lead-free solders and lead-free elec-
tronics products [IDEALS 1999].
The threat of legislation has now receded.
Senator Reads bill was withdrawn, and the Eu-
ropean Union has greatly expanded the category
of exceptions and pushed back the implementa-
tion date to 2006 or 2008 (the date is not rmly
xed). Nevertheless, most responsible compa-
nies now have accepted a commitment to clean
and green manufacturing, whereby no new
electronic products may contain lead. The au-
thors are of the opinion that there will be a pro-
gressive transition to lead-free manufacturing for
electronics products over the next decade or so,
driven largely by companies wishing to promote
an environmentally aware image.
The majority of the work on lead-fee solders
has concentrated on the need to nd a replace-
ment for lead-tin near-eutectic alloys. While this
endeavor has been successful, no alternatives to
the high-melting-point, lead-rich family of al-
loys with melting points of approximately 300
C (570 F) have been identied. In recognition
of this fact, the pending legislation exempts high-
melting-point solders. Section 5.1.8 of this chap-
ter outlines some of the reported attempts to nd
replacements for these alloys.
One positive benet of the switch to lead-free
solders is elimination of the need to source
low-alpha-emission solder for electronics appli-
cations. Lead, being a heavy metal, contains a
proportion of radioactive isotopes, and the al-
pha particles emitted by spontaneous decompo-
sition can cause electronic circuits to generate
spurious digital signals. Although lead is nomi-
nally cheap, being only an eighth of the cost of
tin, certied low-alpha lead is much more ex-
 Chapter 5: Advances in Soldering Technology / 191
pensive. The price of low-alpha lead roughly
doubles with every tenfold reduction in the
counts per hour of emitted particles, so its re-
moval from premium-quality solders represents
a potential cost saving.
5.1.2 Compatibility with
Lead-Tin Solder
Lead-tin eutectic solder has many desirable
characteristics, namely, excellent wetting and
spreading behavior and a relatively advanta-
geous range of physical properties. Moreover,
because hitherto the entire electronics and light-
ing industries had developed around this alloy
system, all associated materials, components, and
subsequent processes were developed to be com-
patible with its properties and limitations. The
initial goal of the work on lead-free solders was
to nd a drop-in replacement for lead-tin that
could be used at the same process temperature,
on the same equipment, with the same substrate
materials and uxes, and could deliver compa-
rable performance and reliability. For the reasons
that are elucidated, it is now recognized that this
is not technically possible. Instead, there now
exist several families of lead-free solders, each
suited for a different group of applications and
compatible with only a limited subset of metal-
lizations and processes. A further consideration,
of which any practitioner of soldering needs to
be aware, is that many of the lead-free solders are
incompatible with lead-tin solder. This will
present difficulties for many years ahead, be-
cause an attempt to rework or repair a lead-free
joint with lead-tin solder, or vice versa, can result
in a joint with severely compromised function-
ality and reliability. This problem will be further
exacerbated, because it is not easy to distinguish
a lead-tin joint from one made with lead-free
solder without resorting to analytical techniques.
5.1.3 Alternatives to Lead-Tin Solder
In the mid-1990s, it was estimated that the
electronics industry uses approximately 60,000
tonnes (66,000 tons) of lead-tin solder each year
and the lighting industry somewhat more. This is
used to make approximately 1013 soldered joints
per annum [Vincent and Humpston 1994]. For
any alloy to be a worthwhile solder for the elec-
tronics industry, it must meet specic qualities
under the following criteria:
Melting range: An alternative solder must
have an upper process window that is suffi-
ciently low in order that existing components
and boards are not damaged by the soldering
process. In practice, this means a peak process
temperature of 260 C (500 F), so that the liq-
uidus of the alloy must therefore be somewhat
lower. On the other hand, the solder must have
a solidus that is sufficiently high that the joints
are robust in service, and its use must be com-
patible with existing step-soldering opera-
tions. This means that the solidus, after alloy-
ing with any metallizations present on the
boardandcomponentleads,shouldnot be sub-
stantially below 170 C (340 F).
Physical and chemical characteristics: The
new lead-free solder must wet and spread on
the common engineering metals and metal-
lizations, namely, gold, silver, platinum, pal-
ladium, nickel, tin, copper, and iron. Ideally,
the solder should also be compatible with
existing ux technology and certainly not re-
quire a more aggressive or environmentally
adverse material. The solidied solder also
must be sufficiently inert to resist the corro-
sive environments associated with electronic
equipment (for example, the electronic con-
trol boards inside domestic washing ma-
chines and dishwashers are subject to incred-
ibly harsh chemical, physical, and thermal
environments).
Environmental, health, and safety aspects:
The alloy and its components must be non-
toxic. This automatically rules out alloys con-
taining cadmium, thallium, mercury, and also
possibly nickel. The intrinsic toxicity of these
metals means that they cannot be considered
as constituents of replacement solders (Table
5.1).
Similar considerations apply to the ux for-
mulations for use with these new solders and
the chemicals used to remove the ux residues
from the assembly after fabrication. A further
consideration is that the pollution and envi-
ronmental damage arising from the increased
mining and extraction of the alloy ingredients
in response to new demand must be at minimal
cost and environmental damage.
Economics and availability: For any alloy to
be considered as a potential replacement for
lead-tin solder, its constituents must be suf-
ciently abundant so that the needs of the
market for the new solders can be met fairly
readily. This condition on availability is nec-
essary to ensure that the alloy or alloys are
192 / Principles of Soldering

not subject to supply or price constraints that and indium are simply too scarce and too ex-
would prevent their widespread adoption. pensive to use for other than niche applications.
Based on the current world consumption of The situation with regard to bismuth, silver, cop-
lead-tin solder, each 1% substitution for lead- per, antimony, and zinc is less restrictive, and
tin solder by another element represents a new virtually all of the lead-free solders available
annual consumption for that element of the or- commercially are based on tin with selected ad-
der of 500 to 600 tonnes. This means that sup- ditions of these elements.
ply levels should be higher by severalfold. Almost without exception, lead-free solders
There are, for example, some excellent alloys are based on the silver-copper-tin eutectic with
that contain indium, but only 100 tonnes of in- small optional additions of bismuth, antimony, in-
dium are produced each year, and most of that dium, and zinc. Many of the most promising al-
quantity is recovered as a by-product of zinc loys have been assigned patent protection. Some
extraction. World production of elements po- of the key compositions are listed in Table 5.3.
tentially suitable as constituents of lead-free Bismuth is often recommended as an addition
solder is listed in Table 5.2. to lead-free solders because of its benecial ef-
fect on melting point and mechanical properties,
Taking into account availability, cost, melting as shown in Fig. 5.1 and 5.2.
point, and environmental issues leaves tin as the Users of bismuth-containing lead-free solders
only element on which new lead-free solders for need to be aware of a potential incompatibility
the mass market can be developed. Because tin when undertaking repair work or using it to sol-
melts at 232 C (450 F), alloying additions are der to lead-tin component or board nishes. The
necessary to depress the liquidus temperature presence of lead in the alloy mixture results in
and, preferably, to reduce the cost, because tin is nal solidication of the alloy at temperatures
relatively expensive compared to lead. Gallium that can be below 100 C (212 F), depending on
the composition and composition gradient in the
soldered joint. Obviously, this could have a cata-
Table 5.1 Key characteristics of the strophic effect on the reliability of an assembly,
lower-melting-point metals because it would normally be expected that the
Melting joint would remain solid up to at least 150 C
Element point, C Comments
(300 F).
Mercury 38.9 Toxic Lead-free solders are more expensive than
Cesium 28.5 Highly reactive
Gallium 29.8 Extremely rare
lead-tin solder, because lead is substantially
Rubidium 38.9 Highly reactive cheaper than all of the other constituents, and
Potassium 63.7 Highly reactive only zinc comes close to it in price. While this
Sodium 97.8 Highly reactive
Indium 156.9 Low availability and high cost is important for wave-soldering processes where
(more expensive than silver) an industrial machine will require tens of kilo-
Lithium 179.0 Highly reactive
Tin 231.9 Nontoxic. Ample supply grams of solder to ll the solder bath, the price
Bismuth 271.3 Obtained as a minor premium of lead-free solders over lead-tin solder
by-product of lead is minimal, when in the form of dispensable
Thallium 303.5 Toxic
Cadmium 320.8 Toxic vapor
Lead 327.5 Toxic
Table 5.3 Composition ranges (wt%) claimed
for some lead-free solder alloys
Table 5.2 Availability of potential alloying Bismuth
or
elements in lead-free solder Tin Silver Copper antimony Indium Zinc

Metal World production in 2000(a), tonnes 8899.4 0.053.0 0.56.0 0.13.0 ... ...
9093.5 2.05.0 0.32.0 0.57.0 ... ...
Silver 17,700 9299 0.053.0 0.76.0 ... ... ...
Bismuth 5,880 Bal 3.57.7 1.04.0 010 ... 01
Copper 13,200,000 Bal 1.03.0 0.52.0 110 ... ...
Gallium 100 Bal 0.120 ... 0.125 0.120 ...
Indium 335 Bal 0.53.5 0.52.0 ... ... ...
Antimony 118,000 Bal 3.05.0 0.53.0 05 ... ...
Tin 238,000
Zinc 8,730,000
Note: The obvious overlap in composition ranges makes the legal fortitude of
many of the patents questionable. However, cross-licensing of nearly all the
(a) Source: U.S. Geological Survey data. To convert to U.S. (short) ton, multiply silver-copper-tin family of lead-free solder alloys now appears complete, mean-
by 1.10 ing that most formulations are readily available from the majority of suppliers.
 Chapter 5: Advances in Soldering Technology / 193
pastes. The economics of paste production are
such that they are dominated by the costs of
preparing clean, spheroidized solder powder in
precise size distributions and those of the spe-
cialist organic chemicals that comprise the binder
and ux. Thus, while silver-copper-tin eutectic
alloy costs approximately ve times that of lead-
tin eutectic, when based on metal prices, the
premium paid for paste is typically only of the
order of 1.1 to 1.
5.1.4 Silver-Copper-Tin
Ternary Phase Equilibria
Because of the importance of silver-copper-tin
alloys to lead-free soldering technology, it is im-
perative to have a reliable determination of the
constitution of the silver-copper-tin phase dia-
gram. With the aid of modern analytical tech-
niques, the phase relationships in this system
have recently been revisited and precisely de-
termined. An excellent publication on this topic
is one by Loomans and Fine [1999], which is
well worth consulting as a model on how to
determine a phase diagram using thermal analy-
sis and metallography, particularly for a system
where the composition of invariant reactions is
Fig. 5.1 Reduction in the liquidus and solidus temperature of
an off-eutectic silver-tin alloy as a function of the
bismuth addition
Fig. 5.2 Effect of additions of bismuth on the tensile strength
of bulk samples of silver-tin solder
cooling-rate dependent. It is now generally con-
sidered that the composition for the ternary eu-
tectic under equilibrium conditions is 3.5Ag-
0.9Cu-95.6Sn, with a eutectic point at 217.2 C
(423.0 F). A complete liquidus projection of the
ternary system is given in Fig. 2.14. Substitution
of 0.9% Sn in the binary Ag-96.5Sn alloy by
copper reduces the eutectic temperature by al-
most 4 C (7 F). Even this small temperature
reduction can be signicant for electronic as-
semblies with components tailored for use with
the lower-melting-point lead-tin solder, which
can be detrimentally affected by higher soldering
temperatures.
5.1.5 Metallurgical, Physical, and
Chemical Properties of
Lead-Free Solders
The metallurgy, physical, and chemical prop-
erties of lead-free solders are not markedly dif-
ferent from other solders. They wet, spread, and
alloy with substrate metals in a similar fashion,
and all other properties are broadly similar. How-
ever, there are small differences and various nu-
ances that have been identied that may or may
not be critical for certain processes. Some of the
key parameters are presented in the following
section, which draws extensively on the review
by Glazer [1995]. As explained previously, the
term lead-free solders is generally taken to mean
alloys based on the silver-copper-tin ternary eu-
tectic, although the data used for comparison
also include other low-melting-temperature lead-
free alloys.
In attempting any comparison of solder prop-
erties, it must be remembered that for common
solder alloys, room temperature represents close
proximity to their melting points, in terms of
chemical thermodynamics. Thus, in any test, it is
virtually impossible to standardize factors that
are governed by diffusion. This makes it ex-
tremely difficult to compare work by different
investigators. While the data presented here may
appear to be quantitative, it is best interpreted as
semiquantitative and as an indicator of trends.
5.1.5.1 Surface Tension
The surface tension of tin is either raised or
lowered by the addition of other elements in low
proportions. Lead, bismuth, and antimony cause
a reduction, while silver and copper increase sur-
194 / Principles of Soldering

face energy. Values for surface tension are given
in Table 5.4. Surface tension values of solders
exposed to air are typically lower than in an inert
atmosphere, because oxidation lowers the sur-
face energy of the molten solder.
5.1.5.2 Other Physical Properties
Data for the density, electrical resistivity, ther-
mal conductivity, and coefficient of thermal ex-
pansion (CTE) of a few solders are given in Table
5.5. Density and thermal conductivity largely fol-
low the rule of mixtures. On the other hand,
resistivities and CTEs are all closely similar for
the different solders, with the exception of Bi-
43Sn. Bismuth is a semimetal, rather than a nor-
mal metal, that is present in near-equal propor-
tion by volume as the tin phases in the
microstructure of the Bi-43Sn solder, which ac-
counts for its higher resistivity and lower thermal
expansivity.
5.1.5.3 Mechanical Properties
Because solders creep rapidly at room tem-
perature, measurement of mechanical properties
is extremely sensitive to strain rate. An addi-
tional complication arises because mechanical
properties are also very sensitive to grain size,
Table 5.4 Measured values of surface tension
for binary solder alloys in air and nitrogen
Surface tension, mN/m
Alloy Air
Bi-43Sn 319
Sn-9Zn 518
Pb-62Sn 417
Ag-96Sn 431
Cu-99Sn 491
Sb-95Sn 468
Test conditions: test temperature, 250 C (482 F); polytetrauoroethylene sub-
strate; SNMA, synthetic, mildly activated. (RMA-type, rosin, mildly activated)
ux
joint thickness, and alloying with the parent ma-
terials.
For solder alloys that are composed primarily
of a phase mixture of two pure metals, the elastic
modulus can be estimated from the rule of mix-
tures, based on the volume fraction of each phase
present. Of the common lead-free solder alloys,
only the elastic modulus of indium-tin alloys
cannot be deduced from data on the respective
metals but must be measured, because the in-
dium-tin eutectic is formed between two inter-
metallic compounds of indium with tin.
The yield strength of solders is a strong func-
tion of temperature, because thermal activation
assists dislocation movement, while testing at a
high homologous temperature renders the results
particularly sensitive to strain rate. However, be-
cause solder generally has very low yield strength
and is mostly used in a temperature regime where
creep is important, this parameter is often of little
practical value. More relevant is ultimate tensile
strength, although measurement of this property
is equally unreliable for the same reasons. Shear
strength normally correlates closely with tensile
strength, except where plastic ow plays a sig-
nicant role, as it does in most solders at room
temperature. Therefore, for all three of these
propertiesyield strength, tensile strength, and
shear strengthit is probably fair to deduce that
the differences between the solder alloys are
small, with the exception of indium-tin alloys,
which are noticeably weaker.
Elongation is an important property, because
high ductility enhances the low cycle fatigue re-
Nitrogen sistance of a material. Also, reasonable elonga-
349 tion to failure at high strain rates is benecial in
487 preventing catastrophic failures if a joint is in-
464
493 advertently overstressed. The Bi-43Sn and In-
461 49Sn eutectic solders are greatly superior to Pb-
495
62Sn and Ag-96Sn in this respect.
Comparative mechanical properties for se-
lected solders are given in Table 5.6.

Table 5.6 Comparative mechanical properties

Table 5.5 Physical properties of selected for selected solders at room temperature
solders at room temperature
Ultimate
Coefficient Elastic tensile Shear
of Solder modulus, strength(a), strength(a), Elongation(a),
Thermal thermal alloy GPa MPa MPa %
Solder Density, Resistivity, conductivity, expansion,
alloy kg/m3 ohm cm W/m K 106/K
Pb-62Sn 39 50 35 50
Ag-96Sn 50 60 30 70
Pb-62Sn 8400 14 33 25 Bi-43Sn 42 70 30 10
Ag-96Sn 7360 11 50 26 In-49Sn 24 20 10 100
Bi-43Sn 8700 32 21 15
In-49Sn 7300 15 34 20 (a) Measured at a strain rate of 0.001 s1
 Chapter 5: Advances in Soldering Technology / 195

The resistance of metal to creep can be mea- ne two-phase lamella structure of lead-tin sol-
sured in many ways. Creep is a complicated pro- der provides a modicum of strength right up to
cess, and most solders exhibit at least three types the melting point, whereas above 150 C (300
of creep, denoted as primary, secondary, and ter- F), the large crystals of the dominant tin fraction
tiary. To provide a very rough guide to the rela- in silver-tin and copper-tin eutectics are less re-
tive ranking of different solders in their resis- silient to fatigue.
tance to creep, stress-rupture life is a useful
parameter, because there is a linear relationship
between the applied stress and the life of the 5.1.5.4 Corrosion Resistance
testpiece when plotted on logarithmic scales. Rel-
Lead-free solders generally do not suffer from
evant data for some solders at room temperature
corrosion in normal environments. The excep-
are given in Fig. 5.3. Silver-tin solder is the most
tion is alloys that contain zinc, because of the
resistant to failure by creep, whereas indium-tin
large difference in electrochemical potential be-
solder is the least resistant among the four sol-
tween this metal and the other constituents. It is
ders represented.
for this reason that zinc-containing alloys are not
The fatigue resistance of different solders is
considered as suitable replacements for lead-tin
one of the most relevant parameters, from the
solder for most applications. A further disincen-
point of view of their application, but also the
tive to the use of zinc was that, until recently, the
most difficult to determine reliably. The difficulty
wetting and spreading on copper surface was
stems from the fact that the results are extremely
extremely poor, compared with lead-tin. How-
sensitive to the specimen conguration and test
ever, recent work using uxes containing tin
conditions, and these factors can greatly affect
2-ethylhexanoate shows considerable promise.
their relative ranking. However, there is a gen-
This compound decomposes at the soldering tem-
eral consensus that fatigue resistance of the four
perature to leave metallic tin on the faying sur-
solders may be ranked as follows: In-Sn < Bi-Sn
face just ahead of the advancing solder front. By
< Pb-Sn < Ag-Sn. Trials have shown that, for
this means, contact angles as low as 20 can be
surface-mount assembly of components on FR4
achieved, which is low enough to be widely ac-
printed circuit board, both silver-tin and copper-
ceptable [Vaynman and Fine 2000].
tin lead-free solders are superior to lead-tin eu-
tectic under most conditions. However, if the
application involves operation above approxi-
5.1.5.5 Susceptibility to
mately 150 C (300 F), then, surprisingly, the
lower-melting-point lead-tin solder actually ex- Tin Pest and Tin Whiskers
hibits the superior fatigue performance. This is Tin undergoes an allotropic transformation
because the almost equally divided and fairly from its normal metallic white form, which has

Fig. 5.3 Stress-rupture life of joints made with low-melting-point solders, tested at room temperature. Silver-tin solder is more resilient
than lead-tin eutectic, while indium-tin alloys are less able to resist creep.
196 / Principles of Soldering
a body-centered tetragonal crystal structure, to a
powdery gray material, with a diamond cubic
crystal structure, at, nominally, 13 C (55 F)
(see Chapter 2, section 2.2). However, in order
to initiate the transformation, it is generally nec-
essary for white tin to be subject to strain, and the
maximum rate of conversion occurs at approxi-
mately 40 C (40 F). Because the formation Lead-free solders are therefore required to wet
of gray tin from the white allotrope is accom-
panied by a 26% expansion in volume of the
metal, the individual grains separate, and the sol-
der disintegrates into ne powder. The transfor-
mation is termed tin pest. Clearly, with most
lead-free solders having a higher proportion of
tin than the lead-tin eutectic alloy they are re-
placing, there exists the possibility that tin-rich
lead-free solders will be vulnerable to failure by
this mechanism. Lead-tin solders are carefully
formulated with controlled impurity additions to
prevent the transformation from taking place.
This is explained further in Chapter 2, section
2.2.
The susceptibility of three pure tin-base sol-
ders to tin pest has been studied [Kariya, Gagg,
and Plumbridge 2000]. It was found that on stor-
age at 18 C (0.4 F), a proportion of the
tin-rich phase in silver-tin, silver-copper, and tin-
zinc solders transformed to gray tin. The incu-
bation period was shortest for the zinc-contain-
ing alloy and longest for the silver-containing
solder. Further trials are necessary to establish
whether this allotropic transformation can be sup-
pressed by controlled doping, as it can with lead-
tin solder (see Chapter 2, section 2.2).
Pure tin, under certain conditions, has a ten-
dency to grow long crystals, or whiskers, that can
cause electrical shorts between adjacent conduc-
tors on PCBs. The elimination of lead from sol-
ders applies equally to the solderable nishes
applied to component leads and lands on PCBs.
Some of the replacement nishes are tin-base.
Therefore, both lead-free solders and some PCB
nishes are potentially susceptible to tin whisker
growth. Work is believed to be underway to as-
sess this vulnerability, and the results are awaited
with interest.
5.1.6 Process Window for
Lead-Free Solders
Lead-free solders intended to be as close as
possible form-t-function replacements for lead-
tin solder generally have a melting range of ap-
proximately 210 to 230 C (410 to 446 F) [Bra-
dley, Handwerker, and Sohn 2003]. This is
approximately 30 to 50 C (54 to 90 F) higher
than for lead-tin solder. However, the upper tem-
perature for soldering of electronics and opto-
electronics assembly is xed at 260 C (500 F)
by the design of existing parts. The peak tem-
perature of 260 C (500 F), historically, gave an
adequate margin for use with lead-tin solder.
and spread with considerably less superheat than
was previously considered good practice for lead-
tin solder. This has two consequences:
In lead-free soldering processes, much greater
attention must be paid to the thermal cycle.
In particular, the preheat cycle needs to be
carefully optimized to ensure that all parts of
the assembly where joining is to be effected
are ultimately raised above the liquidus tem-
perature of the solder. This can require con-
siderable attention to processing conditions
for such products as double-sided PCBs
populated with tiny surface-mount compo-
nents and massive programmable logic array
chips in ceramic packages.
Care should be taken to ensure that the com-
ponent metallizations and solder are compat-
ible. As discussed in the following section,
different lead-free solders wet standard met-
als and metallizations (copper, tin, nickel,
gold, etc.) at different rates, so what works
with one alloy from one manufacturer might
not be appropriate for an ostensibly similar
alloy composition from another supplier.
The reduced superheat also has consequences
for the choice of ux. At the same time as the
move toward lead-free solders, concern was also
growing over the damage to the ozone layer
above the Earth, with the principal culprit iden-
tied as the family of chlorouorocarbons. The
progressive withdrawal of these standard de-
uxing agents led to the reassessment of clean-
ing procedures and prompted the development of
no-clean uxes (see Chapter 3, section 3.2.1.2).
A further environmental concern arose over the
use of volatile organic compounds (VOCs). Hith-
erto, these were a major ingredient of solder
uxes, and emissions of VOCs from an industrial
wave-soldering machine can be many kilograms
per hour. The VOCs play a major role in the
generation of photochemical smog.
No-clean and low-VOC uxes have inher-
ently low chemical activity that is further di-
minished by the low superheat of lead-free sol-
dering processes. This combination results in
 Chapter 5: Advances in Soldering Technology / 197
processes with unacceptably narrow windows, to
which the response has been the widespread
adoption of nitrogen-inerted soldering. The ni-
trogen atmosphere retards oxidation of the solder
and substrate before the ux becomes active and
thereby reduces the work that the ux has to do
in maintaining oxide-free interfaces ahead of the
advancing solder front. If large superheats can be
applied, the benets of a protective atmosphere
are reduced, because solder is able to spread
further and faster (Fig. 1.14 and 5.4) [Buckley
2000].
In recent years, much effort has gone into the
design of nitrogen-inerted soldering equipment,
and remarkably good-quality atmospheres can
now routinely be achieved on open-access sys-
tems for surprisingly modest rates of gas con-
sumption. Equipment with specications of 10
ppm oxygen, or below, in the working zone is
available commercially.
5.1.7 Wetting and Spreading
Characteristics of Lead-Free
Solders
The wetting and spreading characteristics of
lead-free solders are not straightforward. There
have been many studies on this topic, and much
data exist in the published literature. Under cer-
tain regimes, it is possible for lead-free solders
to wet and spread better than lead-tin solder, but
for many it is not. The root of this complexity is
that the wetting and spreading of lead-free sol-
ders is inuenced by the alloy composition, the
process temperature, the process atmosphere, the
ux, and the substrate metallurgy. This difficulty
is exemplied by the data presented in Fig. 5.5
and 5.6, which show the time required for the
wetting force to reach the acceptance value (23
Fig. 5.4 Wetting speed of lead-tin solder on copper using a
rosin-based ux in air and nitrogen atmospheres.
Nitrogen reduces the propensity for the solder and substrate to
oxidize and thereby decreases the cleaning action demanded of
the ux to effect wetting.
of the theoretical maximum) in wetting balance
tests for a range of process and materials com-
binations [Lau and Ricky Lee 2001]. It is im-
possible to discern any consistent trends in the
data.
Fortunately, the manufacturers of the various
lead-free solder alloys have undertaken sufficient
development to know what materials and pro-
cess conditions are necessary to obtain wetting
and spreading behavior that is at least as good as
for the lead-tin process being substituted. There-
fore, when making the change to lead-free sol-
dering, it is advisable to adopt one alloy, ux,
process window, and recommended component
nish as a complete and integrated package. It is
the authorsexperience that a surface-mount PCB
assembly line can be switched from using lead-
tin to a lead-free solder in less than 20 min. This
line was used for the mass production of circuit
boards, and the manufacturing yield and in-
service reliability obtained from lead-tin and
lead-free solder were indistinguishable. The tran-
sition time was essentially that required to swap
the solder paste print cartridges, allow the reow
ovens to stabilize with a different temperature
prole, and change some of the reels on the com-
ponent placement machines, because some parts
come supplied in a choice of lead-free or lead-tin
plating.
5.1.8 High-Melting-Point
Lead-Free Solders
High-lead solders (>85% Pb) are currently
granted an exception, subject to review, to the
ban on lead in solders [European Union Direc-
tive 2001]. This dispensation has been granted
on the grounds that there are no technically and
economically viable alternatives. Lead-rich sol-
ders are used in large quantities by the lighting
industry, and most discarded light bulbs are dis-
posed of in landll sites, which provides an in-
teresting perspective on the merits of the ban on
lead-tin eutectic solder.
Extensive phase diagram modeling suggests
that the prospects for nding a new multicom-
ponent solder with a melting point in the region
of 300 C (570 F) are low [Lalena, Weiser, and
Dean 2002].
Shimizu et al. [1999] have proposed the qua-
ternary alloy Zn-4Al-3Mg-3.2Ga as one possi-
bility. This alloy exhibits acceptable melting be-
havior (309 to 345 C, or 588 to 653 F), but
there are difficulties associated with its use be-
198 / Principles of Soldering

cause of the unfavorable combination of high
hardness, low ductility, and ease of oxidation.
Nevertheless, this solder has a higher thermal
conductivity (77 W/m K) than high-lead sol-
ders, lower thermal expansivity (20 106/K),
and, moreover, foil can be produced by hot roll-
ing at 250 C (482 F). Fluxless attachment of
gold-metallized silicon die to silver-plated cop-
per leadframes has been demonstrated with this
solder using a die-bonding machine that pro-
vides for a mechanical scrubbing action during
the heating cycle. The resulting joints, which
measured 5 by 5 mm (197 by 197 mil), were well
lled and passed standard thermal cycling and
wet-high temperature storage tests without dif-
culty.
Another candidate for a lead-free high-
melting-point solder is the ternary alloy Bi-11Ag-
0.05Ge, which has a liquidus of 360 C (680 F)
and a solidus of 262 C (504 F). Although the
wide melting range will be unattractive for some
applications, this alloy does have the merit that
ingots can be mechanically processed by hot
working into foil and wire. Molten bismuth oxi-
dizes readily in air to an extent that renders the
binary Bi-11Ag alloy unsuitable as a solder. Ger-
manium has a higher oxygen affinity than bis-
muth and is largely insoluble in the binary base
alloy. Because GeO sublimes only at 480 C (896
F), it is reported that at process temperatures in
the region of 400 C (750 F), wetting by this
solder can be obtained on nickel, silver, and gold
but not copper, in air, using an inorganic aqueous
ux. At room temperature, the alloy is softer than
the Au-20Sn solder and possesses similar, lim-
ited ductility but has a thermal conductivity of
only 9 W/m K, which may be restrictive for
some applications [Lalena, Dean, and Weiser
2002].
Because of their limitation, both of these al-
loys are likely to be of interest only for niche
applications.

Fig. 5.5 Wetting rate, as measured by the time for the wetting force to reach an acceptable value, for a range of solders on different
component and board nishes. There is no conclusive trend of superiority for any solder or solderable coating, emphasizing
the need to tailor the lead-free process to each situation. OSP, organic surface preservation

Fig. 5.6 Wetting rate, as measured by the time for the wetting force to reach an acceptable value, of solders on different substrates
in different atmospheres. The relative ranking varies greatly, depending on the metals and process atmosphere involved.
 Chapter 5: Advances in Soldering Technology / 199
5.2 Flip-Chip Interconnection
Flip-chip interconnection is a method of mak-
ing electrical connection and providing physical
attachment between two components, often a
semiconductor die and substrate. Although the
process was established at the birth of the semi-
conductor revolution, it was largely abandoned in
favor of wire bonding. Flip-chip bonding was rel-
egated to niche applications but nevertheless con-
tinued to evolve and develop to a high degree of
technological sophistication. With the ongoing
drive toward smaller, lighter, and cheaper elec-
tronics systems of greater functionality, ip-chip
bonding experienced a renascence as an essential
enabling technology. Extrapolation of current
trends suggests that within ten years, the majority
of semiconductor die may be ip-chip bonded.
Flip-chip interconnection is being advocated or
used in virtually every advanced packaging con-
ceptball grid arrays (BGAs), chip-scale pack-
ages (CSPs), direct chip attach (DCA), and mul-
tichip modules (MCMs), to name but a few.
The invention, or, more strictly, the commer-
cialization, of ip-chip bonding is credited to
IBM Corporation in the early 1960s [Miller
1969]. The process was developed at the same
time as the rst integrated circuits began to ap-
pear, in response to the need to simultaneously
make a large number of extremely miniaturized
connections. The technique involved soldering a
copper ball at every joint to provide the physical
separation and electrical interconnection be-
tween the transistors on the chip and the sub-
strate. Solder-based ip-chip technology came
about from the realization that the copper ball
was not required and that precise z-axis separa-
tion is maintained by the relatively high surface
tension of solder alloys. This modication,
known as the C4 process, standing for controlled
collapse chip connection, forms the basis of the
two ip-chip technologies practiced today. There
now exist many variants of this process, and only
the key features are elucidated in the following
sections. For a more complete description of ip-
chip technology and methods of its implemen-
tation, the reader is referred to Lau [1996, 2000].
An overview of ip-chip technology, with par-
ticular emphasis on optoelectronic applications,
is provided by Lee and Basavanhally [1994].
5.2.1 The Flip-Chip Process
A schematic illustration of the solder ip-chip
process is given in Fig. 4.19. The two parts to be
joined are rst patterned with mirrored, match-
ing arrays of a solderable metal. One of these
parts is then further processed to deposit inter-
connect metal (solder) on every pad. Either part
is then ipped over (hence, the process name)
and the metal patterns aligned. Finally, they are
brought together to form a joined assembly.
The ip-chip process ow is set out in greater
detail in Table 5.7. It is described by reference to
two basic inputs, an electronics wafer and a sub-
strate, although either or both of these parts could
be substituted by other parts of different func-
tionality, such as a semiconductor component
package and a PCB, as in a BGA mounting
scheme. Processed semiconductor wafers are
supplied coated with a passivation layer, typi-
cally either a polymeric or oxide/nitride lm, to
provide some protection to the devices during
nal assembly and packaging. Apertures are cut
through this passivation layer at discrete points
to allow contact to the top metal of the semi-
conductor fabrication process. This metal tends
to be aluminum or copper on silicon devices and
gold on III-V devices, such as gallium arsenide.
The substrate will usually have a track pattern in
a conductive metal such as copper or gold, with
pads at the connection points.
The rst step of the ip-chip process involves
application of a wettable metal, or underbump
metal, on the areas of the chip and substrate to
be mated (Fig. 5.7). The deposit is often a mul-
tilayer metallization, the functions of which are
to provide ohmic contact to and metallurgical
compatibility between the contacts on the chip
and substrate and the interconnect metal. Typi-
cally, it comprises a sequence of up to three metal
layers. The rst metal is called the foundation
metal, and its role is to provide the initial ohmic
contact and a strong physical bond to the faying
surfaces of the components. The choice of foun-
dation metal varies with the nature of the com-
Table 5.7 Flip-chip process ow
Individual processes differ greatly in their details, but the
general scheme is common to both solder bump bonding and
compression bump bonding. The steps shown in parentheses
are optional.
Dene underbump metal pattern areas
Deposit underbump metals
Dene interconnect pattern areas
Deposit interconnect metal
Clean and inspect
(Fuse interconnect deposit to form alloy)
Align mating parts
Bond
(Underll)
Inspect
200 / Principles of Soldering
ponents being joined and the method of depo-
sition. Common examples are titanium and
chromium, if applied by a vapor-phase tech-
nique, and zinc, if applied by wet plating. The
second metal is the barrier metal and is included
to prevent metallurgical reaction between the ad-
hesion metal and the interconnect metal. The
choice of barrier metal is extremely wide and
includes both pure elements, such as platinum,
nickel, and copper, as well as mixtures of metals,
such as titanium-tungsten (oxy-nitride) and tita-
nium nitride. The barrier metal needs to be of
sufficient thickness to ensure that it remains in-
tact during the making of the joint and the service
life of the assembly, while remaining electrically
conductive. The former requirement means that
this barrier layer must be able to satisfy any
dissolution into the molten solder and interme-
tallic growth in the solid state without being de-
pleted. Finally, it is common practice to com-
plete the underbump metal with a sacricial
metal, which is often gold, so as to maintain the
solderability shelf life and/or aid visual inspec-
tion. The underbump metal typically has a total
thickness of less than 1 m (40 in.).
One of the two parts is removed from the
processing chain at this juncture, because it is
usually neither cost-effective nor often necessary
to have interconnect metal on both parts. The
interconnect metal is mostly applied by screen
printing or electroplating, although for very small
bumps (<10 m, or 400 in., in diameter), vapor-
phase techniques are generally necessary. His-
torically, lead-tin alloys and pure indium were
used for the interconnect metal, but the process
works equally well with lead-free compositions
and other solders [Lau 1996]. Finally, the com-
ponents are cleaned, ready for bonding. A num-
Fig. 5.7 Schematic illustration of a silicon semiconductor de-
vice prepared for ip-chip bonding. A hole has been
cut through the passivation layer, and the three-layer underbump
metal applied. The foundation metal, M1, is followed by a barrier
metal. The interconnect metal has been deposited on top and
fused to form a hemispherical bump.
ber of direct deposition processes have been de-
veloped for applying the interconnect metal. One
of the more interesting of these uses a jetting
system to generate a stream of precisely sized
molten droplets. These are steered by an elec-
trostatic deection system, and an exact number
of droplets can be deposited at any location over
the entire area of a 6 in. diameter substrate (Fig.
5.8) [Priest, et al. 1994; Hayes, Wallace, and
Boldman 1996].
The interconnect metal is frequently an alloy.
Sometimes, for ease of process control, the sol-
der is deposited as sequential layers of the con-
stituent metals. It is then common practice to
perform an additional fusion step immediately
after deposition, whereby the part is heated until
the layers constituting the solder melt and alloy,
forming a hemispherical bump in the process.
This opens the possibility of an interesting pro-
cess variant, called the wet back process,
whereby the diameter of the interconnect metal
areas, as deposited, can be up to twice that of the
underlying wettable metal pad. During fusion,
surface tension forces pull the molten droplet
back to the wettable pad, thereby increasing the
height of the bump while limiting its width. This
trick simplies the creation of particularly tall
bumps. By making any tracks connected to the
bump pad sufficiently narrow, no masking or
other measures are required to prevent lateral
solder ow along the tracks.
The interconnects are not by any means re-
stricted to a spherical geometry. Indeed, ad-
vances in photoresist technology make it pos-
sible to fabricate interconnects with aspect ratios
of 5 to 1, even at tiny feature sizes. One pub-
lished example shows solder columns 5 m (200
in.) in diameter and 20 m (800 in.) high at 10
m (400 in.) pitch [Yamada et al. 1998]. The
column grid arrays so formed have substantially
improved fatigue life compared with ball grid
arrays, because the thermal expansion mismatch
strain between the abutting components is dis-
tributed over a greater height of material. A three-
fold improvement in fatigue life is thereby re-
alized [Sturcken 2000].
During the bonding step, the mating intercon-
nect areas on the two components are aligned in
plane, translation, and rotation and brought into
contact. The connection is made by either cold
compression welding, commonly referred to as
indium bump bonding, or by heating to melt the
interconnect metal, known as solder bump bond-
ing (Fig. 5.9). Although ip-chip compression
bonding and ip-chip solder bonding super-
 Chapter 5: Advances in Soldering Technology / 201

KHz signal
Circuit board
High voltage

Charge electrode
Piezo crystal

Orifice

Charge
driver Catcher
Nitrogen
gas valve
Liquid
metal
supply

CAD/CAM
Temperature input
controller

Fig. 5.8 Schematic layout of an apparatus designed to deposit individual solder balls at any desired location on a 6 in. diameter
substrate. CAD/CAM, computer-aided design/computer-aided manufacturing

cially appear to be similar processes, they are

implemented by exploiting different joining pro-
cesses. Consequently, the resulting interconnects
have signicantly different characteristics. The
key features of each process and the character-
istics of the resulting interconnects are summa-
rized in Tables 5.8 and 5.9.
It should be pointed out that indium and in-
dium alloys are sometimes used as solder inter-
connects, and lead-tin alloys can be used as com-
pression interconnects, so some care needs to be
taken with terminology. Some companies are of-
fering ip-chip processes based on conductive
polymeric media. Generally, these have charac-
teristics similar to compression interconnects, but
they are not able to achieve interconnect diam- Fig. 5.9 Flip-chip interconnect schemes. The prepared com-
eters much below 100 m (4 mil), and of course ponents are aligned and joined by either applying
they lack the self-alignment feature, which is a pressure (solid-state compression bonding) or heat (soldering).
characteristic of molten metal.
The self-aligning feature of solder bump bond- of the interconnect height. A schematic illustra-
ing requires some explanation. The surface en- tion of the self-aligning process is given in Fig.
ergy of a molten solder pillar between two con- 5.10. Typical alignment accuracy is better than
tact pads will always attempt to minimize the 2 m (80 in.) in-plane and 0.5 m (20 in.)
surface area of the liquid. This results in a re- vertical, and these gures can be improved by
storing force that regulates the relative locations nearly an order of magnitude by careful geo-
of each mating pair of pads. The force balance metrical design of the interconnects and atten-
derives from the contact angle of the solder onto tion to process detail. Methods of achieving pre-
the wettable pads, while the weight of the upper cise alignment between ip-chip bonded parts
component controls the planarity and magnitude are discussed in section 5.2.6 of this chapter.
202 / Principles of Soldering

5.2.2 Characteristics of acteristic is particularly important for high-

Flip-Chip Technology
Flip-chip bonding can be dened as a method
of providing simultaneous electrical connection
and physical attachment between two compo-
nents. It is attractive for low-cost, volume manu-
facturing because it is an inherently high-
yielding, wafer-scale process. The key features
and associated benets of this technology are:
Attachment and interconnection are per-
formed simultaneously. This provides a ma-
jor cost and yield advantage over other forms
of interconnection where the component must
rst be rigidly xed to the substrate before
proceeding with wire bonding or lead sol-
dering. Fully automated ip-chip bonding
machines are commercially available that can
ip, align, place, and reow (if required) parts
at throughput rates well in excess of 500/h.
Self-alignment occurs between the compo-
nents and substrates, through surface tension
forces in the molten solder (this characteristic
applies only to solder bonding, not compres-
sion bonding). Thus, optical components can
be assembled in which a semiconductor laser
or detector is reliably and accurately posi-
tioned relative to other parts, such as optical
bers.
Flip-chip interconnects provide a low and re-
peatable inductance, essentially a conse-
quence of the short bond length. This char-
Table 5.8 Process characteristics
Compression bump
Key parameter bonding
Process Solid-state diffusion
Materials In, Au, Pb, Pb/Sn
alloys
Temperature 20200 C (68392 Solder melting
F)
Pressure 10100 MPa (1.515 0 MPa
103 psi)
Max height: 1:5
pitch
frequency circuits. A typical 25 m (1 mil)
diameter bond wire interconnection 0.5 mm
(0.02 in.) long will have an inductance of
0.25 nH, compared with approximately 0.05
pH for a typical ip-chip solder bond.
Parasitic capacitance is small, predictable,
and reproducible. Again, this becomes a most
important issue as the operating frequency
increases. While the interconnection capaci-
tance of a typical bond wire at 150 m (6 mil)
pitch is 25 fF, it is only 2 fF for ip-chip
bumps at 30 m (1.2 mil) pitch.
Interconnections can be accommodated over
the entire area of the component. Wire bond-
ing and tape interconnection can only be
made around the periphery of chips. This re-
stricts the total number of interconnections
that can be made and requires all signal paths
on the chip to be suboptimally routed out to
the chip edge. Flip-chip bonding is, for simi-
lar reasons, the ideal interconnection tech-
nology for pixelated component architec-
tures, one example of which is given in Fig.
5.11.
The interconnection area is smaller than the
component footprint. Close packing of digi-
tal information processing chips is necessary
in order to minimize bus delay times. All
ip-chip connections are located underneath
the chip, whereas all other interconnection
schemes tie up substantial space all around
the chip periphery.
This interconnection method results in low
prole assembly, with ip-chip bumps just a
Solder bump few microns (<1 mil) high (Fig. 5.12). There-
bonding
fore, the chip packages and the whole elec-
Solid-liquid tronics system can be made correspondingly
alloying
Any solder
point
3:1

Table 5.9 Technology characteristics

Compression bump bonding Solder bump bonding

Short (coin) interconnects Tall (pillar) interconnects

Filled joint gap
Fluxless
Low residual stress
Service temperature >
bonding temperature
Alignment as placed
Open joint gap
Flux usually required
Residual stress
Service temperature <
bonding temperature
Self-aligning Fig. 5.10 Schematic illustration of the self-aligning mecha-
nism of ip-chip solder interconnects
 Chapter 5: Advances in Soldering Technology / 203

Fig. 5.11 (a) Semiconductor component of a pixelated imaging device. (b) Close-up view of a single solder ball. The imaging device
utilizes a 16 by 128 array of 33 m (1.3 mil) diameter solder bumps for the interconnects. Each pixel is connected to its
own amplier and processing circuitry electronics. Courtesy of CERN

thinner. This characteristic is particularly

valuable in portable electronics products.
Flip-chip processing is compatible with deli-
cate materials. The upper frequency limit for
silicon-base semiconductors is approxi-
mately 5 GHz. Above this frequency, com-
pound semiconductors such as gallium ars-
enide (GaAs) or gallium nitride (GaN) must
be used. Compared with silicon, these ma-
terials are extremely fragile, so that alterna-
tive interconnection techniques are preferred
to mechanical wire and tape bonding. The
latter can generate surface damage on the Fig. 5.12 An array of exceptionally small ip-chip solder
semiconductors because of the forces in- bumps ready for bonding. Each is mounted on a
pedestal of polyamide for functional reasons. Solder bumps of this
volved in these processes. dimension can only be realized using semiconductor processing
The interconnect is packageless, and the join- techniques. 5000. Courtesy of BAE Systems
ing medium is applied during component pro-
cessing. Therefore, application of the mate-
rials required for ip-chip bonding only Increase the mechanical strength of the as-
requires a few additional processing steps, sembly
and all components may be processed simul- Enhance heat sinking of the die
taneously. Removal or minimization of pack- Extend the fatigue life of the interconnects
aging costs is always attractive.
The process generally involves placing an ac-
curate volume of material along one or more
5.2.3 Underll edges of the die and allowing it to be pulled into
To improve the operational life of ip-chip the gap by capillary action. The adhesive is cured
assemblies, it is common to apply an underll by the action of time, temperature, exposure to
adhesive into the gap between the die and the ambient atmosphere, or a combination of these.
substrate on which it is mounted. The objectives Provided that the modulus and thermal expan-
of using an underll are fourfold: sion coefficient of the underll material are ju-
diciously chosen, then the fatigue life of solder
Improve environmental protection of the bumps between a silicon chip and an FR4 PCB
semiconductor can be increased by as much as two orders of
204 / Principles of Soldering
magnitude on thermal cycling between 65 C
(85 F) and 100 C (212 F) [Doi et al. 1996]. frared radiation.
Flip-chip bonding is a rapidly evolving eld,
and a number of underll materials are being
offered that use different chemistries. Some of
these are applied prior to assembly of the parts
and actually ux the solder during the heating
cycle before providing conventional postbond-
ing functions.
5.2.4 Inspection
One of the challenging features of ip-chip
bonding is that the interconnects are not easy to
inspect. Some process engineers consider this to
be a major benet, because it restricts the scope
for quality inspectors to reject assemblies!Acom-
parative review of ip-chip inspection methods
is given by Burdett, Lodge, and Pedder [1988].
The problem is that only the peripheral bonds
can be observed visually, and then often only
with some considerable difficulty. The other
bonds are hidden from view beneath the com-
ponents. Special microscopes have been devel-
oped that are able to look a considerable distance
into the gap between bonded parts (see section
5.10.3 in this chapter). X-ray inspection is an
obvious solution to this problem, stemming from
the ability of x-rays to penetrate most materials.
Fortuitously, the materials making up most com-
ponents and substrates are lighter and less ab-
sorbent to x-rays than the solder, which therefore
shows up clearly. The main defects that can be
isolated using x-ray equipment include missing
bumps, bridged bumps, excess bump material,
and porosity in the interconnects. However, it
cannot easily detect the most usual cause of yield
loss in ip-chip technology, which is intercon-
nects that have failed to produce joints, that is,
leave gaps. The reason for this is that small air
gaps negligibly affect x-ray absorption, particu-
larly in comparison with solder balls. Further
information on x-ray inspection is given in sec-
tion 5.10.2 in this chapter.
Infrared microscopy can be used to inspect
certain types of ip-chip assemblies. Infrared mi-
croscopy is similar to traditional optical micros-
copy but uses much longer wavelengths. Silicon
and many III-V semiconductor materials are
transparent in this portion of the electromagnetic
spectrum. Infrared microscopy permits the tracks
and the backside of the underbump pads to be
examined but not the detail of the bumps them-
selves, because solder bumps are opaque to in-
Arguably, the most appropriate technique for
inspection of ip-chip bonded components is
acoustic microscopy. An acoustic microscope
generates and focuses an acoustic pressure wave
into the device under examination and uses the
reected echoes to construct an image of the
object at a specic depth. The method is sensitive
to boundaries between regions of different den-
sity and is therefore ideally suited to discrimi-
nating planar joints and transverse defects such
as unwetted joints, missing bumps, cracks, and
voids. Acoustic microscopy is a rapidly evolving
eld, and instruments with truly remarkable reso-
lution are available commercially. Defects in 20
m (800 in.) diameter solder bumps can be re-
solved in real-time. Further information on
acoustic microscopy can be found in section
5.10.1 in this chapter.
5.2.5 Rework
Successful rework of a ip-chip mounted die
is difficult but not impossible. Local hot gas
nozzles can be used to heat an individual chip
until the interconnect metal melts, whereupon
the component can be lifted clear. The difficulty
lies in dressing the used bond pads on the sub-
strate so they are left in an adequate state of
cleanliness and parallelism to accept another
chip. For these reasons, rework tends not to be
practiced for ne-pitch interconnects or with die
that have been underlled, although it is per-
fectly practicable for larger bump bonds such as
those used with BGAs, which have coarser ge-
ometries.
5.2.6 Self-Alignment of
Flip-Chip Structures
One of the remarkable features of solder ip-
chip interconnects is their ability to self-align.
Accuracies at the 100 to 200 nm (4 to 8 in.)
level, with respect to the underbump pattern ac-
curacy, are possible. The factors that need to be
considered to achieve these precisions are now
reviewed.
Self-alignment of ip-chip interconnects was
noted almost from the inception of this soldering
method, and subsequently, there was a lot of
interest in its exploitation for the assembly of
optical components. This interest has been re-
viewed by Tan and Lee [1996]. The best align-
ment from self-alignment is quoted as better than
0.8 m (30 in.) for four 130 m (5 mil) diameter
 Chapter 5: Advances in Soldering Technology / 205
bumps, while Hayashi [1992] and Tsunetsagu et
al. [1997] suggest that better than 0.25 m (10
in.) can be obtained with a larger number (20)
of smaller bonds (25 m, or 1 mil, diameter). The
best claimed alignment accuracy for systems with
deliberately misaligned solder bumps pulling the
device against a mechanical stop was <0.3 m
(12 in.) in the stop axis but greater in other
directions [Lin et al. 1993]. Attempts to use mul-
tiple stops in different orthogonal directions were
unsuccessful. This is because the friction of mov-
ing the device past the rst stop is greater than
the available restoring force.
As stated previously, the driving force for self-
alignment of solder bumps is surface tension,
which seeks to minimize itself by adopting the
shape of a perfect sphere, having the minimum
surface area for a given volume. At that point, the
surface tension is balanced by the internal pres-
sure of the solder. Distorting the sphere, by
squashing, stretching, or shearing sideways, will
be resisted by a restoring force that is approxi-
mately proportional to the additional surface area
created.
When a solder bump is used to bond two com-
ponents together, the situation is complicated
slightly by the presence of at, wettable pads,
plus the weight of the top chip. The solder will
then try to adopt a shape as close to a truncated
sphere as the volume of solder and the size and
shape of the pads allow. The effect of the weight
of the top chip is to slightly increase the internal
pressure in the solder, and this will tend to reduce
the restoring force. Software is available that
allows the restoring force to be calculated
[Brakke 1990], from which the following geo-
metric design guidelines emerge for achieving
good in-plane alignment:
Reduce the compressive loading per bond,
because some of the surface tension has to be
used to support this weight. This can be
achieved by having the smallest chip on top
during reow, thinning the chip or reducing
its size, and increasing the number of solder
bonds.
Use thinner bonds because for a xed pad
size and number, they have a higher restoring
force than ones with thicker solder, because
the shear distortion per unit height is much
greater. However, this implies that the solder
volume should be reduced, but this could have
a detrimental effect on the joint reliability in
fatigue situations and on the ease of making
a joint, because the solder volume-to-surface-
area ratio becomes progressively more un-
favorable with smaller bonds.
Use smaller joints because they have slightly
higher restoring force than a larger joint of the
same aspect ratio. This is because a given mis-
alignment distorts a small sphere much more
than a larger one. On the other hand, because
many more ball bonds can be accommodated
in the same area than can larger bonds, there is
less loading per joint and additional restoring
force from each additional bond.
Design the pads so that at the neutral posi-
tion, all bonds are equally offset. Because the
restoring force diminishes rapidly as the axial
alignment improves, there can be insufficient
net force available to move the upper com-
ponent, particularly if the joint gap is lled
with liquid ux. By offsetting the bonds, the
available force for a given displacement is
larger and is never zero.
The other route of increasing the restoring
force and thus the accuracy of self-alignment is
to increase the surface tension of the molten
metal. In fact, this can be changed reasonably
easily. Lead-tin eutectic alloy has a surface ten-
sion of 0.483 N/m (2.8 103 lbf/in.) at 350 C
(660 F) in inert atmosphere, compared to 0.55
N/m (3.1 103 lbf/in.) for tin at the same
temperature and 0.44 N/m (2.51 103 lbf/in.)
for lead. Replacing the lead with cadmium (0.59
N/m, or 3.37 103 lbf/in.), zinc (0.78 N/m, or
4.45 103 lbf/in.), or silver (0.92 N/m, or 5.25
103 lbf/in.) should increase the surface ten-
sion of the solder alloy and hence its restoring
force against misalignment. The other major fac-
tor altering surface tension is the presence of
ux. The surface tension of eutectic solder at 350
C (660 F) drops from 0.48 N/m (2.74 103
lbf/in.) in inert atmospheres to 0.41 N/m (2.34
103 lbf/in.) in the presence of nonactivated R-
type ux. Additions of chlorine to this ux drop
the surface tension even further, so that using a
1% Cl activated ux lowers the surface tension
to 0.31 N/m (1.77 103 lbf/in.). Liquid uxes
should therefore be avoided and gaseous uxes
used instead (see Chapter 3, section 3.3 on ux-
less soldering).
Z-axis, or height, control is a function of the
volume of the solder sphere, modied as neces-
sary by the weight of the upper component.
Achieving good z-axis control principally re-
quires that accurate solder volumes are placed
on pads whose surface areas are also well de-
ned. These factors are primarily process con-
206 / Principles of Soldering

trol issues, although some design inputs will 5.2.8 Step-Soldered

help. As before, many bonds mean that indi- Flip-Chip Interconnects
vidual bond variations are less signicant [Mc-
Groarty et al. 1993]. For the vacuum deposition Solder ip-chip interconnects can be made
and wet back process, where solder is deposited with essentially every known alloy. Step ip-
over a greater area than the wettable metal pads, chip soldering is therefore both possible and prac-
any variation in the thickness of solder depos- ticed. Figure 5.14 is an example of a substrate
ited is exaggerated in the reowed solder bumps. that contains both wire-bonded and ip-chip-
Similarly, plating a thick solder deposit over a mounted die. By using the high-melting-point
smaller part of the wettable metal will dilute
these thickness errors. Solder deposition sys-
tems that work on mass change rather than
thickness control tend to be superior in this
context. Bond design also helps, with more
spherically shaped bonds being less sensitive to
deposited thickness variations than more co-
lumnar-shaped interconnects.
Predicting the extent of self-alignment that
can be achieved from a ip-chip structure is pos-
sible using the SURFACE EVOLVER computer
code. Good agreement has certainly been dem-
onstrated between the model and lead-tin solder
bumps wetted onto copper pads [Josell et al.
2002]. This suggests that such models can be
used as a tool for predicting the alignment and
stand-off height of specied interconnect geom-
etries.

Fig. 5.13 Flip-chip land designed to cope with surface to-

pology. The solder is initially conned by surface
5.2.7 Surface Topography tension to the central circle but will slowly ow through the
constricting necks into the overspill areas. The ensuing reduction
One of the difficulties with ip-chip processes in solder volume, and hence bond height, enables adjacent in-
terconnects of reduced height a chance to reach and wet their
is that surface topography is difficult to accom- mating pads.
modate in a reliable manner. Quite simply, the
out-of-atness of the mating parts often repre-
sents a signicant proportion of the total height
of the interconnects.
One method of addressing this problem is to
endow each of the mating bond pads with a
solder land that is connected via a narrow path,
as shown in Fig. 5.13. This provides a means
whereby the excess solder on the rst pads to
mate is progressively drawn off, reducing the
total height of the interconnect. Thereby, shorter
interconnects are able to form as the two parts
are drawn together. The narrow land between
the pad and overspill area greatly slows the rate
at which the solder height can decrease and
therefore ensures sufficient material is available
to make all of the interconnects. Obviously, this
system demands additional surface area to ac-
commodate these solder drains, and the addi- Fig. 5.14 Radio frequency die mounted using high-tempera-
ture ip-chip interconnects onto a substrate, which
tional electrical capacitance that arises must be is itself populated with lower-melting solder balls ready for direct
within design limits. attach to a printed circuit board. Courtesy of Intarsia Corporation
 Chapter 5: Advances in Soldering Technology / 207
Au-20Sn solder (melting point 280 C, or 536
F) for this application and making the inter-
connects as short as possible (30 m, or 1.2 mil),
it is possible to use a second ip-chip process to
attach the substrate to a PCB, using, in this case,
solder bumps of lead-tin eutectic solder (melting
point 183 C, or 361 F).
5.3 Solderability Test Methods and
Calibration Standards
The formation of strong joints is contingent on
the ability of the molten ller to wet the joint sur-
faces over their entire surface area. Wetting of
metal surfaces tends to be inhibited by surface ox-
ides and contaminants. Visual inspection is not re-
liable for ascertaining the condition of surfaces
with regard to solder wetting. The surface appear-
ance of metallizations, in particular, can be decep-
tive, because color and reectivity are dependent
on a variety of factors, such as grain size and type,
whether equiaxed or columnar, surface texture,
and any lms present on the surface. Therefore,
wettability has to be determined by a direct mea-
surement involving exposure of a part represen-
tative of a component to be joined to molten sol-
der. These wettability measurement methods can
also be used to assess the effectiveness of a ux,
the sensitivity of surfaces to exposure to various
atmospheres, and also shelf life, in general. The
evaluation of solderability shelf life tends to in-
volve articially accelerated testing to obtain data
within a reasonable timescale.
Because soldering processes are dynamic in
naturethat is, they are sensitive to the time
over which the ller is moltentwo aspects of
solderability need to be considered, namely:
The readiness with which substrates and com-
ponents are wetted by a ller
The extent of spreading that is obtained by
the end of the process cycle, when the ller
solidies
A number of evaluation procedures and tests
have been developed to measure these charac-
teristics and are described in the subsequent sec-
tions. Reference is also made to the production
of calibration standards of solderability.
5.3.1 Assessment of Wetting
The simplest and most popular test used for
assessing wettability is the dip-and-look (DNL)
test. It is a standard test method for solderability
used by the electronics industry for testing the
leads of integrated circuit devices with appro-
priate nishes (e.g., ANSI/J-STD-002 and IPC/
EIA J-STD-003A). In this test, the component
leads are immersed into a suitable rosin-type ux
for 5 to 10 s and then immersed into molten
solder (generally, lead-tin eutectic) for 5 to 10 s.
The component leads are removed from the sol-
der, cleaned, and then visually inspected. The
percentage of the lead surface that has remained
unwetted by the solder is then visually assessed.
The standards specify the acceptance condition
that all terminations shall exhibit a continuous
solder coating free from defects for a minimum
of 95% of the critical surface area of any indi-
vidual termination.
The chief disadvantage of the DNL test is that
the conditions used differ considerably from those
used in a manufacturing environment. In a typical
production line, an infrared or convection furnace
or vapor-phase reow chamber is used to heat the
assembly, with solder paste preapplied to the
joints.The dynamic thermal environment is there-
fore very different and the relative volumes of sol-
der and ux very much smaller than that afforded
by the test.
A variant of the DNL test, called the surface-
mount solderability test (SMT), was devised to
more closely simulate the actual environment that
surface-mount devices encounter during the sol-
der reow process. In this test, solder paste is
screened onto a thin, unmetallized ceramic plate
(0.9 mm, or 35 mil, thick) using a solder stencil.
The paste print used is the pattern of the footprints
of the leads to be assessed. The device to be tested
is then placed onto the solder paste print. Next, the
ceramic substrate is passed through a reow cycle
and allowed to cool. The devices are removed
from the ceramic substrate, and the leads are ex-
amined for solder wetting. The advantage of this
test is that leaded devices are subjected to a similar
thermal environment to that experienced in the ac-
tual assembly. Furthermore, the leads receive the
same volume of solder paste as in the actual as-
sembly operation.The SMTalso neatly avoids the
occurrence of solder bridging of ne pitch leads,
to which the conventional DNL test is prone. This
test method has been adopted as an Electronics In-
dustriesAssociation (EIA) interim standard (EIA/
IS-86).
The new SMT test method has been shown to
be especially effective for providing a reliable
assessment of the solderability of palladium-
plated integrated circuit (IC) leads. Palladium-
plated devices have been marketed since 1989 and
now account for a signicant portion of the IC
208 / Principles of Soldering

market. The palladium is applied as a 0.076 m (3 and provides automatic tareing of specimen
in.) plated nish over a nickel-plated lead. Its weight
function is simply to act as a protective barrier to A temperature-controlled solder bath
oxidation for the nickel. During the soldering A bath lift or specimen fall mechanism with
cycle, the palladium is required to completely dis- speed and positional control
solve into the solder with the joint being formed A computer to display the force/time curve
with the underlying nickel layer [Abbott et al. and derive key metrics from the data
1991]. Trials using the SMT method have dem-
The specimen under test is held in a holder
onstrated that the palladium coating is completely
that is itself suspended from the load cell. The
dissolved, as designed. By comparison, the DNL
bath containing the molten ller metal is raised
test, with its short solder immersion time and ab-
at a preselected speed to immerse the testpiece
sence of preheat, hinders the dissolution of the
to a given depth (on some equipment, the
palladium from the lead surface.
testpiece is lowered into the solder reservoir).
Neither of the two DNL-type tests described
The bath is held in this position for a set dwell
previously provide information about the dynam-
time and is then returned to its rest position.
ics of wetting, and neither is quantitative. A test
The values of these parameters are preselected
that has been devised to quantitatively measure
before commencing the test cycle and are de-
wetting as a function of time is the wetting balance
ned by standards.
solderability test.The method has been adopted as
The resolved vertical forces acting on the
a standard test for measuring the solderability of
specimen are recorded as a function of time
electronic component leads and substrates under
over the whole test cycle. Figure 5.16 shows the
a closely specied set of conditions that include a
typical form of the trace that is recorded, to-
prescribed atmosphere and ux. Solderability
gether with the corresponding position of the
testers, as they are known colloquially, are avail-
specimen relative to the solder bath at each
able from several manufacturers, one example of
stage.
which is shown in Fig. 5.15. Atypical wetting bal-
The wetting balance provides a measurement
ance comprises:
of the vertical component of the force exerted on
A load cell and signal processing system that the testpiece as it is lowered into a reservoir of
furnishes a measurement of load versus time the molten solder or braze, as a function of time.

Fig. 5.15 Commercially available wetting balance. Courtesy of Concoat

 Chapter 5: Advances in Soldering Technology / 209
This force, FR, is theoretically equal to the sum
of the vertical component of the surface tension
force, F, between the ller and the testpiece and
the buoyancy of the testpiece, FB. Figure 5.17
shows an equilibrium situation appropriate to
partial wetting. The resolved force in the vertical
direction, FR, is the parameter measured in the
test. The variation of this force as a function of
time provides information on the dynamics of
the wetting process.
Typical wetting balance force/time traces are
given in Fig. 5.18. The graphs show the effect of
different cleaning methods on the wetting be-
havior of leadtin eutectic solder, heated to 235
Fig. 5.16 Typical trace of the wetting force during a solder-
ability test cycle, with the corresponding position
of the specimen relative to the solder bath
Fig. 5.17 Forces diagram for a solid plate partially immersed
in a liquid. P, specimen periphery length; , liquid
surface tension; , contact angle, , liquid density; gravitational
constant, g, 9.81 m2/s; V, immersed solder volume
C (455 F), on mild steel uxed with a mildly
activated rosin ux. In Fig. 5.18(a), the steel
coupon is in the as-received condition, and wet-
ting by the solder is consequently very poor.
Abrasive cleaning of the steel improves wetting,
as shown in Fig. 5.18(b), but chemical cleaning
is necessary in order to meet the quality-
acceptance criteria, which are indicated by the
box in the lower left-hand corner of the graph
(Fig. 5.18c). The acceptance criterion for elec-
tronic components specied in national stan-
dards is a wetting force that exceeds two-thirds
of the theoretical maximum force, achieved in a
time representative of the soldering process to be
used. For wave soldering, this is set at 2.1 s.
A wetting balance test can be performed rap-
idly, and the results are quantitative, inasmuch as
reproducible numerical data can be obtained for
a well-dened set of sample and instrumental
parameters and operating procedures, as ex-
plained in Barranger [1989] and Lea [1991]. Fur-
thermore, the change in wetting as a function of
time can be monitored. The surface tension of the
molten ller can be calculated from data ob-
tained on the wetting balance using nonwetted
( 180) substrates, such as polytetrauoro-
ethylene (PTFE) or ceramic coupons, using the
equation:
FR Pcos gV (see Fig. 5.17)
From this value and the measured wetting
force, the angle of contact between the molten
ller and the testpiece can be calculated.
Attempts have been made to correlate wetting
balance data with the results given by other meth-
ods used for assessing wetting [Thwaites 1981;
Wooldridge 1988]. Moreover, adaptations have
been made to the wetting balance for solderabil-
ity testing on specic types of components and,
in particular, for surface-mounted electronic de-
vices and also in controlled atmospheres, includ-
ing vacuum [Gunter and Jacobson 1990].
The only other test in common use that evalu-
ates the dynamics of wetting is the measurement
of contact angle in an enhanced version of the
sessile drop test described by Matienzo and Schaf-
fer [1991]. In that test, a xed volume of solder is
melted on a at coupon of the substrate of interest,
which is held at a xed temperature. A ux is gen-
erally added if the test is performed in air.The con-
tact angle of the molten pool of solder is dynami-
cally observed and measured. The principle of the
method is illustrated in Fig. 5.19.
210 / Principles of Soldering

There is not necessarily a close correlation
between solderability test measurements and the
wetting characteristics of actual joints. One of
the reasons is that the conguration of the test
does not strictly mirror that of an actual joint. In
particular, the volume of the molten ller relative
to the volume of the components, including sur-
face metallizations that react with it, may be
signicantly different in the two cases. For ex-
ample, the extremely fast dissolution rates of
gold coatings in solder, reported for samples
dipped into the reservoirs of solder in a wetting
balance, typically 1 m/s (40 in./s) or more, do
not apply to joints soldered using foil preforms
of much smaller volume. Thus, gold coatings
that completely dissolve in the solder bath can
partly survive in actual joints after the same time
and temperature of the bonding operation.
Another difference is the nonequivalence of
the thermal environment, as in the DNL test. In
many materials systems, the ller metal will re-
act differently with the testpiece as the tempera-
ture changes. This is a critical aspect because, as
explained in Chapter 1, section 1.2.2, metallur-
gical reaction and not surface tension is usually
the dominant driving force for wetting in liquid-
solid metal systems. Because such a metallur-
gical reaction occurs, to an extent that depends
on the materials involved and their temperature,
the wetted interface changes its composition and
geometry with time. This, in turn, means that the
measured wetting force tends to vary in the
course of a test, and only in exceptional cases is
the wetting force equation strictly valid, depend-
ing as it does on the classical model of wetting.
Considerations such as this limit the value of
wetting balance tests for absolute measurements.
However, as a means of obtaining comparative
data, and, in particular, for quality-control as-
surance purposes (pass/fail determination), this
method is most useful [Thwaites 1981].
5.3.2 Assessment of Spreading
One of the simplest and most direct methods
that have been devised for assessing wetting by
liquid metals on solid substrates involves mea-
suring the area of spreading by the molten metal.

Fig. 5.18 Wetting behavior of mild steel by lead-tin eutectic solder, measured on a wetting balance at 235 C (455 F). (a) As-received
condition. Wetting occurs slowly and at an inconsistent rate. (b) Following mechanical abrasion of the coupon surfaces
immediately prior to testing. Wetting occurs more rapidly because of denudation of the oxide scale, but the pass condition, which is
a wetting force of 4.5 mN achieved within 2.1 s of immersion, is not achieved in this case. (c) Following chemical cleaning. The
component solderability now satises the acceptance criterion.
 Chapter 5: Advances in Soldering Technology / 211
This type of basic spreading test measures wet-
ting by a molten solder of a solid over the in-
terface of contact and not the enhancement that
occurs in narrow joints through the action of
capillary forces (see Chapter 1, section 1.2.4).
The procedure that is widely adopted in area-
of-spread measurements is to melt a ller-metal
pellet of known volume in a specied atmosphere,
with or without uxes, and to allow it to spread
over the surface of a testpiece for a xed period of
time under controlled conditions. These condi-
tions, where possible, should be representative of
the intended application, because the spreading of
the ller metal is usually sensitive to component-
specic variables (e.g., surface nish) and to pro-
cess variables (e.g., time at temperature).
The area of spreading of the ller metal is
measured; this provides a relative index of wet-
tability for comparative purposes. This index is
the spread ratio, which is dened as:
Spread ratio (Sr)
Total plan area wetted by the molten metal
Original plan area of a metal pellet
(of a specic geometry)
Fig. 5.19 Principle of the sessile drop test used to assess
wettability. (a) A controlled volume of ller metal
(solder) is melted onto the substrate under controlled conditions.
(b) The contact angle is measured with a calibrated viewnder.
In an enhanced form of this test, the contact angle is recorded
dynamically as a function of time.
Because the geometry of the solidied ller
metal is seldom perfectly circular, image analy-
sis should be used to provide an accurate mea-
sure of the total wetted area. The technique of
image analysis is capable of providing an as-
sessment in real-time, so that with suitable in-
strumentation, the spread area can be monitored
as a function of time on a single testpiece. A
perpendicular view also enables the contact angle
to be similarly monitored. A sequence of spread
tests made to evaluate small changes to the com-
position of a ller alloy is shown in Fig. 2.25.
An alternative means of determining relative
spreading involves quantitatively dening a
spread factor in terms of the volume, V, of ller
metal used and the maximum height, h, of the
solidied pool:
(6V / ) 1/3 h
Spread factor (Sf) 100
(6V / ) 1/3
where (6V/)1/3 D, the diameter of a sphere
corresponding to volume V of the metal pellet
before the spreading test. This term can be cal-
culated from the density of the metal pellet and
its mass.
The normal method used for measuring the
height (h) uses a micrometer. However, this
approach introduces errors because of the ne-
cessity to subtract the thickness of the substrate,
which is likely to be much greater than h, from
the measured height. A more accurate method
of measuring h is to determine the peak height
from metallographic sections or prolometer
traces. Both of these methods also enable si-
multaneous measurements to be made of the
contact angle between the resolidied pellet and
the substrate.
Assuming that the initial pellet of ller can be
approximated to a sphere and the resolidied
ller to a spherical cap of radius R on the surface
of the substrate, the spread ratio (Sr) and spread
factor (Sf) can be expressed in terms of the angle
of contact (), as follows:
4 cot2 /2
Sr (Eq 5.1)
(1 3 cot2 /2) 2/3
1
Sf 1 (Eq 5.2)
(1 3 cot /2)
2 1/3
212 / Principles of Soldering

The contact angle, , can be written in terms
of the radius, A, and height, h, of the resolidied
pool of ller:
2
sin (Eq 5.3)
(A/h) (h/A)
The mathematical derivations of Eq 5.1 to 5.3
are given in Appendix A1.2. Some numerical
values relating spread ratio and spread factor to
the wetting angle (contact angle) are given in
Table 5.10. Obviously, the situation represented
by these expressions is an idealized one, and the
assumptions made to derive them become less
realistic with increasing solder spread. Never-
theless, these relationships do provide a reason-
ably close concordance with measured values.
Some results of spreading tests are given in Fig.
1.14, which provides spread ratio data for a range
of solder alloys melted on thin chromium met-
allizations covered with a ash of gold, as a
function of the process temperature.
The spread ratio becomes progressively more
sensitive as the contact angle declines and wet-
ting improves, whereas the spread factor varies
almost linearly with contact angle, from a value
of 1 at 0 to 0 at 180, as shown in Fig.
5.20. Spread ratio, therefore, provides a better dif-
ferentiation between small differences in mea-
sured contact angle when the values of the latter
are small. However, the converse is true when the
contact angle is greater than approximately 60.
The results of spreading tests over a single
surface must be treated with some caution when
attempting to relate them to the wetting and ll-
ing of joints. Because a joint comprises a pair of
facing solid surfaces, with which the ller metal
can react, the capillary forces can govern the
spreading characteristics; there are hydrostatic
forces to consider as well. The relevance of the
spreading test is also questionable when the joints
are made using foil preforms. For this type of
conguration, it is not necessary for the ller
metal to spread signicantly in order to ll the
joint, so that a low spread in the test does not
necessarily mean that a joint formed under simi-
lar conditions will be poor. Indeed, a high degree
of spreading can be detrimental to joint lling,
because the ller metal can ow out of the joint,
resulting in voids and unwanted coverage of other
parts of the component. Nevertheless, it is often,
but erroneously, assumed that low spreading in
the test necessarily implies weak interaction be-
tween the ller and the substrate and therefore
poor bonding and weak joints.
An additional problem in attempting to ex-
trapolate spreading test results to predict joint
quality is that the ller metal reacts and spreads
over a single surface in a conventional spreading
test to produce a somewhat different microstruc-
ture from that of an actual joint. This is dem-
onstrated in Fig. 5.21. Therefore, the wetting and
spreading characteristics might be different in
the two cases. Large-area testpieces are needed
for measuring spread parameters in situations
where there is good wetting.
5.3.3 Solderability
Calibration Standards
There are occasions where it is necessary to
consider solderability tests from another stand-
point, namely, where the solderability of the
testpiece is dened and some other property is
under investigation, such as occurs during the

Table 5.10 Calculated values of spread ratio

and spread factor corresponding to selected
contact angles. The values are derived from the
expressions given in the text (Eq 5.1 and 5.2).
Spread ratio Spread factor
Contact angle (), degrees (Sr) (Sf)

180 0 0
90 1.6 0.37
40 3.7 0.65
10 9.7 0.86
0 Innite 1.0 Fig. 5.20 Relationships among spread ratio, spread factor,
and contact angle

development of a new ux or when comparing
solders from different manufacturers. In other
words, there is a need for solderability calibra-
tion standards. Perfectly wettable and nonwet-
table testpieces are readily obtainable; gold and
anodized aluminum are examples of the former
and latter, respectively. The problem arises where
there is a need for testpieces having intermediate
and known degrees of solderability.
The published literature cites at least two meth-
ods of preparing reference materials [Norme
Francaise 1987; Lea 1990]. Neither method is
entirely satisfactory. Both depend on chemical
treatments to modify a clean copper surface and
thereby degrade the solderability. The modied
surfaces are not chemically stable, so that the
testpiece must be freshly prepared before each
use. Hence, there is considerable batch-to-batch
variability that becomes amplied when differ-
ent chemists at different sites are involved.
Fig. 5.21 (a) Solder spread sample on a gold-plated sub-
strate. The solder is Pb-60Sn, and the substrate is
copper plated with 5 m (200 in.) of nickel and then 5 m (200
in.) of gold. At least three distinct microstructural bands are
visible. (b) Micrograph of a joint made using the same substrate
and solder described in (a). The joint has a regular microstructure,
and all of the gold coating has dissolved in the solder.
Chapter 5: Advances in Soldering Technology / 213
In recognition of this deciency, several teams
of researchers, funded by the United Kingdom
Department of Trade and Industry, were given
the goal of developing a solderability standard
reference material. The solution nally devised
was a two-layer electrodeposit consisting of 10
m (390 in.) pure nickel, applied from a low-
stress sulfamate solution, overlaid with 5 m
(197 in.) of pure, soft gold [Hunt and Wallis
1995]. The substrate can be any inert material
(from a soldering point of view), and nickel was
selected because it does not contaminate the
nickel sulfamate plating bath. Gold applied di-
rectly to nickel strip was not successful, because
the composition of readily available nickel strip
is not sufficiently tightly controlled. Problems
were encountered with impurities in the base
metal, especially zinc, diffusing through to the
outer gold plating. Plating baths are formulated
to very precise specications from high-purity
chemicals, so the quality of the metals obtained
from this source tends to be highly repeatable.
The as-prepared testpieces have perfect sol-
derability, because the wettable surface is pure
gold, with the plated nickel interlayer providing
an effective barrier to diffusion of impurities from
the nickel substrate to the gold. Because the gold
coating is thick and the diffusion of nickel
through gold at room temperature is negligible,
the reference standards have a long shelf life.
Trials indicated no change in solderability when
stored for a year in an open laboratory environ-
ment. In order to degrade solderability of the
testpieces in a controlled manner, the plated ma-
terial was heat treated in air for 2 h. The long heat
treatment time was chosen to ensure that errors
arising from the heating and cooling stages,
which tend to be difficult to control, are minimal.
It is desirable to load plated sheets directly into
a preheated furnace and, at the end of the allotted
time, remove them to ambient to achieve the
fastest possible temperature change without in-
troducing surface contamination.
The heat treatment temperature needs to be
quite tightly controlled, because the tightly
grouped process temperatures of 295, 310, and
328 C (563, 590, and 622 F) yield three in-
termediate degrees of solderability. Because of
the small temperature interval between these
three temperatures, some care needs to be taken
to ensure that the furnace temperature is cali-
brated and the reference material experiences the
specied set-point temperature.
During the heat treatment, nickel will diffuse
through the gold and, on reaching the free sur-
214 / Principles of Soldering

face, will oxidize. Because nickel oxide is only
removed by the most aggressive of uxes, the
reference material has variable solderability, de-
pending on the proportion of nickel oxide in the
gold. Figure 5.22 shows the concentration of oxy-
gen (atomic percent) in the gold at a depth of 3
nm (0.1 in.). Figure 5.23 represents the wetting
force measured for each solderability reference
standard using three commercially available
uxes having different activities. The results were
found to be consistent, with scatter bars of 2
mN (0.007 ozf) able to fully accommodate all
of the experimental results obtained from an in-
terlaboratory comparison involving ve sources
of solderability reference material, six test labo-
ratories, and three different makes of solderabil-
ity test equipment. The uxes used in all the trials
were of the same type, obtained from one manu-
facturer, but were from different batches and
Fig. 5.22 Oxygen concentration at 3 nm (0.1 in.) depth in
the gold surface of solderability reference standard
material as a function of the heat treatment condition. The sub-
strate material is nickel sheet, electroplated with 10 m (390 in.)
nickel, overlaid with 5m (200 in.) gold.
dates of production. This reference material with
the plated layers therefore fullls the require-
ment of being a readily reproducible calibration
standard for assessment of solderability.
5.4 Amalgams as Solders
An amalgam is a mechanically alloyed mix-
ture of liquid metal and solid powder. They are
usually designed such that metallurgical reaction
at room temperature or when slightly warmed
results in isothermal solidication, and the mix-
ture sets solid. Amalgams therefore merit con-
sideration as solders. The molten constituents are
a low-melting-point metaleither mercury, gal-
lium, or indiumor a mixture of these, while the
solid powder contains metals having a consid-
erably higher melting point. A list of liquid met-
als and solid powders that have been explored as
amalgams is given in Table 5.11.
Because amalgams initially exist as pasty u-
ids, they offer a number of unique advantages
over conventional solders. The bonding process
occurs at low temperature, without ux, so that
it greatly reduces the equipment complexity and
confers much process exibility. In particular,
amalgams can accommodate large out-of-plane
engineering tolerances, because they are semi-
solids, but yet can be dispensed precisely at the
bond location. Furthermore, because amalgams
take a nite time to set, often many tens of min-
utes, and are electrically conductive while mol-
ten, a product can be tested for electrical func-
tionality while the amalgam is still liquid, with
time to replace any faulty components before the
bond is cured. This property is also likely to be
of benet in the assembly of optoelectronic de-
vices, where it is frequently necessary to under-

Table 5.11 Solid and liquid metals that have

been evaluated as amalgams
Melting point
Solid powders Liquid metals C F

Antimony Mercury 39 38
Chromium Gallium 29 84
Cobalt Indium 159 318
Copper Gallium-indium-tin 5 41
Germanium Gallium-indium 15 59
Gold Gallium-tin 16 61
Iron
Nickel
Manganese
Palladium
Fig. 5.23 Wetting force at 2 s, determined using a wetting Platinum
balance, for three commercial uxes, as a function Silver
of the heat treatment condition used to produce the solderability Vanadium
reference standard
 Chapter 5: Advances in Soldering Technology / 215
take active alignment before the parts are xed
in place, yet one does not want the further move-
ment that can occur when a higher-melting-point
solder solidies and cools. Amalgams are also
suited for integration in step-assembly pro-
cesses, because the melting point of a solidied
amalgam is typically 200 to 600 C (390 to 1110
F). Therefore, multiple joints can be made se-
quentially using the same process, which can be
carried out below the upper service temperature
of the product. The liquid component of common
amalgam systems also possesses the character-
istic of being able to wet directly many non-
metals, including ceramics, glass, and, of course,
tooth enamel.
There have been some attempts over the years
to devise amalgam systems for electronic as-
sembly. These have met with only partial suc-
cess. It is not clear from this work whether there
is a fundamental limitation that will preclude the
widespread availability of amalgams with favor-
able functional properties and process charac-
teristics or whether it is simply a lack of research
effort. Certainly, dental amalgams have beneted
from very detailed study and, as a result, are
highly effective for the function for which they
are designed.
5.4.1 Amalgams Based on Mercury
Dental amalgams have been in existence since
before 659 A.D.; the alloy rst appears in a Chi-
nese book with a title translated into Latin as
Materia Medica, written by Su Kung in the fourth
year of the Tang emperor Hsien Ching (659 A.D.)
[Chu 1958], but the formulation currently used
in restorative orthodontology was not devised
until the early 20th century.
The basic method for making amalgam in-
volves a diffusion reaction between silver pow-
der and mercury in a ratio such that all of the
mercury is eventually consumed in the formation
of an intermetallic compound. The reaction may
be simply written as:
2Ag 3Hg Ag 2Hg 3
(melting point 127 C, or 261 F)
The intermetallic compound is often referred
to as the gamma phase in the solidied alloy.
Modern amalgams contain approximately 25%
Sn. The presence of tin accelerates the consump-
tion of the liquid mercury and also results in an
alloy that has a small, controlled expansion on
setting. This is necessary to effect a good seal to
the walls of the tooth cavity but obviously must
not be so great as to cause pain or crack the tooth.
Copper is added to tie up the tin as Cu6Sn5 in-
termetallic compound, because this phase is
proven to have superior mechanical properties
and resistance to corrosion by oral uids than the
HgSn7 phase that would otherwise prevail. Amal-
gam metallurgy is a relatively complex subject,
and a whole family of alloys is available with
varying proportions of minor elements to tailor
the properties of the amalgam for its position in
the mouth and the type of cavity it is being used
to ll.
The mechanical properties of mercury-base
amalgams cannot be expressed in the same lan-
guage as used in mainstream metallurgical tech-
nology without distinct risk of misinterpretation.
For example, the tensile and compressive be-
havior are totally different, and the creep behav-
ior does not correspond with conventional be-
havior. The most accurate description of
mechanical properties is obtained when using
rheological concepts, which are applicable to vis-
coelastic materials. The deformation process in-
volves dislocation climb and grain-boundary
sliding, and, when overstressed, a dental amal-
gam fails by intergranular brittle fracture [Wa-
terstrat 1990]. A stress-strain curve for dental
amalgams at low strains is given in Fig. 5.24.
The mercury in dental amalgams is bound up
in the Ag2Hg3 intermetallic compound, but there
is concern that some mercury might be slowly
released through extended contact with saliva.
Mercury is classed as a hazardous substance.
Efforts have been underway to develop resin ma-
terials for dentistry that are color-matched to
teeth, but despite recent advances in the tech-
nology, these remain inferior to conventional
Fig. 5.24 Stress-strain curve for a dental amalgam (mercury-
silver base) at low strains. Adapted from Dickson,
Oglesby, and Davenport [1968]
216 / Principles of Soldering
amalgams, particularly in their durability. They
mostly contain beryllium, which is also hazard-
ous in some forms, including the oxide.
The requirements of an amalgam intended for
use as a solder are very different to those used for
dentistry. A dental amalgam works by lling a
specially shaped cavity in the tooth that is cre-
ated by the dentist. On curing, the amalgam
swells slightly to lock the cohesive mass of metal
in place. By contrast, a solder must wet and join
two parallel surfaces. Also, a dental amalgam
must set in minutes and attain useable strength
within 2 h. The amalgam used as solder must
afford sufficient persistence of uidity to give
sensible bench life, yet only slightly elevated
temperature is desired to accelerate the harden-
ing process. Nevertheless, the detailed under-
standing of mercury amalgams that now exists
may facilitate the development of amalgams for
manufacturing based on gallium and indium.
5.4.2 Amalgams Based on Gallium
Gallium melts at 29 C (84 F) and is therefore
a potential base for formulating very low-
process-temperature amalgams without the toxic
hazard associated with mercury. Gallium is not
a particularly expensive metal, having a com-
parable price by weight to solder pastes. Most
gallium amalgams exhibit signicant volume
change during curing, but certain compositions
have volume change that is similar to mercury
amalgams. Indeed, because of toxicity concerns
with mercury, gallium-base amalgams have been
developed for dentistry and are commercially
available on the Japanese and Australian markets
[Reusch, Geis-Gerstorfer, and Zeigler 1988].
Over the typical service temperature range of
electronics equipment, the electrical and thermal
conductivities of gallium amalgams are similar
Table 5.12 Properties of some gallium-copper-nickel amalgams
Property 5Ni-20Cu
Set-up time at 35 C (95 F), min 530
Curing time at 100 C (212 F), min 90
Expansion on curing, % 31
Shear strength, MPa (psi) 23.5 (3410)
Elongation, % 8.5
Electrical resistivity, ohm cm 18.2
Thermal expansivity, 106/K 10.4
(a) Balance gallium
to other solder alloys, as can be seen from the list
of properties cited for a range of gallium-nickel-
copper amalgams in Table 5.12 [Mackay 1993].
Gallium forms amalgams with several metal
powders, of which copper and nickel are the
most intensively studied. Unfortunately, the pub-
lished literature does not elucidate on the im-
portant parameters of powder condition, particle
size and shape, nor often the amalgamation
method. However, from the available micro-
graphs, it is possible to deduce that the nickel and
copper powders used with gallium amalgam were
probably spherical and roughly 25 m (1 mil) in
diameter. Gallium is also known to form amal-
gams with silver, and it is regrettable that this
potentially interesting combination has not been
evaluated further.
To form a gallium amalgam, a mechanical
mixture of the constituent metals is warmed until
the temperature reaches 35 C (95 F) and then
thoroughly blended. A convenient method of do-
ing this is in a commercial dental amalgamator,
which contains a pestle to assist in obtaining an
intimate mixture of powder and liquid in a short
period of time. The amalgam will then directly
wet most common metallizations used in the elec-
tronics industry and also many nonmetals, in-
cluding alumina. Due to the high proportion of
solids in the amalgam, some mechanical assis-
tance in the form of scrubbing or wiping is nec-
essary to spread the mixture. Although gallium
alloys will cure at room temperature, to achieve
acceptable process times, the use of elevated tem-
perature is benecial. The graphs in Fig. 5.25
show the cure behavior of Ga-5Ni-30Cu amal-
gams at different temperatures (for comparison,
solidied lead-tin eutectic has an equivalent hard-
ness, on the same Shore Durometer scale, of
approximately 80).
When fully cured, high-strength and well-
lled joints with good fatigue resistance can be
Powder content(a), wt%
5Ni-30Cu 5Ni-40Cu 10Ni-20Cu 10Ni-40Cu
40 15 250 8
10 2 70 1
12 21 15 8
48.5 (7040) ... 29.9 (4340) 39.2 (5690)
7.4 ... 9.3 12.4
11.8 10.8 ... 12.1
4.6 5.3 7.3 0.2
 Chapter 5: Advances in Soldering Technology / 217
obtained, which gives an indication of the po-
tential of this soldering method. Gallium amal-
gams have been demonstrated as ller metals in
a number of electronics-oriented applications,
including semiconductor die attach, ip-chip
interconnection, and via-lling in PCBs. In the
latter example, the amalgam was applied by
screen printing [Bhattachatya and Baldwin
2000]. However, it is generally the case that the
results to date show considerable scatter in terms
of joint quality, and further research is required
to develop processes satisfactory for industrial
use [Baldwin, Deshmukh, and Hau 1996].
5.4.3 Amalgams Based on Indium
Indium is another liquid metal that can be
considered as a base for amalgam systems. Par-
ticularly after alloying with tin and bismuth, the
melting point of the liquid can be depressed to
below 100 C (212 F). Because indium alloys
melt at temperatures that are substantially above
the melting points of mercury and gallium, it is
potentially possible to premix indium amalgams
using liquid indium. Then, the curing reaction
can be suppressed temporarily by quench cool-
ing the liquid-powder mixture until it is needed.
The authors are only aware of attempts to form
indium amalgams using silver powder. As with
the gallium amalgams, the very restricted trial
did not yield a useable mixture, which was at-
tributed to the availability of only relatively
coarse crystalline silver powder. Amalgams ap-
pear to be more successful if the powder used is
ne and spherical. Nevertheless, the mechanical
properties of the silver-indium intermetallic com-
pounds are known to be favorable in indium
Fig. 5.25 Curing curves for Ga-5Ni-30Cu amalgams at a
range of temperatures
joints to silver-metallized surfaces, and there-
fore, further endeavor in this area may prove to
be fruitful.
5.5 Strengthening of Solders
Compared with most metals and alloys in
everyday use, soft solders have intrinsically poor
mechanical properties, that is, low strength and
inferior resistance to creep and fatigue. For the
commonly used lead-tin solders, application tem-
peratures can easily be up to 90% of the melting
point, expressed in Kelvin. This means that the
microstructure of solidied lead-tin joints tends
to be unstable under typical service conditions,
and frequently, grain growth will occur, to the
detriment of the creep and fatigue resistance of
joints. Many solders are also susceptible to stress-
induced microstructural changes. The superpo-
sition of these microstructural changes on a
cyclic strain regime results in concentrations of
shear bands developing in the alloy microstruc-
ture. Such features are prone to initiating fa-
tigue cracks. Hence, solders differ from com-
mon metals in that fatigue failure is initiated
internally, whereas in engineering structures,
fatigue cracks almost always start at an exterior
surface. The poor mechanical properties be-
come most apparent when the joint gap is wide
(>25 m), so that a proportion of the ller metal
through the joint width is free from the con-
straint of the parent materials.
It is precisely because the failure mode of sol-
dered joints in electronics systems is so complex
and depends on many interrelated variables that
the test methods used to assess their integrity and
reliability are specic to that industry, with the
testing usually being carried out on fabricated
assemblies. Further details are given by Coombs
[1988].
So long as the primary function of solders has
been to provide electrical contact, their mechani-
cal weakness could largely be tolerated, although
creep and fatigue failure arising from thermal
expansion mismatch of abutting components
have caused occasional reliability problems. This
was the situation in electronics prior to the adop-
tion of surface-mount technology. This devel-
opment and, in particular, the reliance on support
structures of solder, coupled with the continuing
trend toward miniaturization, have focused at-
tention on the mechanical weakness of soldered
joints.
218 / Principles of Soldering
The key to boosting strength and, in particular,
conferring fatigue resistance to joints made with
solder is the development of a thermally stable
and ne-grained microstructure. Three strategies
have been pursued in an attempt to address this
need:
Grain renement
Dispersion strengthening
Composite solders
All three approaches actually lead to a similar
result, namely, renement of the solder micro-
structure. They are discussed here mostly in re-
lation to the lead-tin solder system, for which
much of the development work has been carried
out because of their ubiquitous use and poor
mechanical properties, although the same ap-
proaches could be adapted to other solder sys-
tems.
5.5.1 Grain Renement
Grain renement is conventionally achieved
by small additions (0.001 to 0.5%) of selected
elements, such as lithium, beryllium, indium, and
gallium, to lead-tin solder [Klein Wassink 1989;
Wade 1999; Tribula and Morris 1990]. This ap-
proach operates by creating a ne dispersion of
either oxide or nitride particles or stable inter-
metallic phases when the solder is molten. These
then act as sites from which the solid phases can
grow on solidication (heterogeneous nucle-
ation). The presence of a large number of such
ne particles when the alloy is molten is reected
in a ne alloy microstructure. The grain size that
can be achieved is typically 150 m (6 mils),
compared with 300 m (12 mils) for pure lead-
tin solder solidied at the same rate. Although
grain renement is benecial to the mechanical
properties of metals, improving both strength and
ductility, the ne microstructure is not thermally
stable and will gradually coarsen by solid-state
diffusion when thermal energy is provided by
exposure to elevated temperature (typically,
above 75 C, or 165 F). As can be seen from the
example given in Fig. 5.26, the creep-rupture
time of a solder doped in this manner is an order
of magnitude longer than that of regular lead-tin
solder. Similar improvements to the creep prop-
erties have been achieved for other binary tin-
base solders by grain renement [McCabe and
Fine 2000]. Similarly, examples of ternary alloy
solders, which have been shown to benet from
grain renement accomplished by minor addi-
tions, include 5In-87Sn-8Zn (melting point, 188
C, or 370 F) and 3.5Ag-95.5Sn-1Zn (melting
point, 217 C, or 423 F) containing additions of
silver or copper, respectively, as the source of the
heterogeneous phase at concentrations in the re-
gion of 0.1 to 0.5% [McCormack and Jin 1994].
This method of improving resistance to failure
by creep or fatigue of soldered joints has the
benet that it is simple to achieve, because it
merely involves controlled doping of the solder
during manufacture.
5.5.2 Oxide-Dispersion-Strengthened
Solders
Dispersion strengthening involves the incor-
poration of ne and intrinsically insoluble par-
ticles into the solder, usually by mechanical
means. Examples are TiO2, SiO2, and Al2O3, all
of which are cheap and inert compounds (not-
withstanding the normal hazards associated with
handling dust) and readily available as ne pow-
der with precisely controlled size distributions.
The particle size is typically 0.1 m (4 in.)
diameter but frequently much less. The loading
of refractory compound in the solder is typically
3% by volume. These dispersoids provide some
grain renement on solidication of the solder
by heterogeneous nucleation, but their principal
role in improving the mechanical properties is by
other means. First, they inhibit coarsening of the
microstructure; because they are insoluble in the
solder, it is thermodynamically favorable for the
particles to reside at regions in the alloy matrix
where there are natural departures from a regular
atomic lattice, such as the boundaries between
the tin- and lead-rich phases in lead-tin solder.
Secondly, provided the particles are sufficiently
ne, they will, for similar reasons, impede the
movement of dislocations through the solder ma-
Fig. 5.26 Creep curve for lead-tin eutectic solder and a dis-
persion-hardened equivalent alloy containing 0.5
wt% Ag, 0.5 wt% Sb, 0.1 wt% Cu, and 0.003 wt% Ga at a constant
stress of 10 MPa (1450 psi) and a test temperature of 60 C (140
F)
 Chapter 5: Advances in Soldering Technology / 219
trix. Dislocations are defects in the atomic lattice
that can be induced to move through it by the
application of mechanical stress; the movement
of many such dislocations results in plastic ow
of the material. Hence, by restricting the move-
ment of dislocations, the mechanical properties
of the solder are enhanced. This type of disper-
sion is less sensitive to thermal degradation, and
therefore, the strengthening effect is more resil-
ient to elevated temperatures than is simple grain
renement from metallic precipitates. Table 5.13
demonstrates the boost to the creep resistance of
lead-tin eutectic solder that can be obtained even
at elevated temperature through dispersion
strengthening.
To be effective obstacles to dislocation move-
ment, the particles must be within a certain size
range, be stable in size and interparticle spacing,
and have a higher ow resistance than the matrix.
Dispersion strengthening is quite well under-
stood, from a theoretical perspective, and the
properties of suitable dispersoids can be calcu-
lated from rst principles. It transpires that ne
oxide particles are a good choice for inclusion in
solders.
Because the dispersoids are insoluble in the
solder, dispersion strengthening remains effec-
tive even when the alloy is heated almost to its
solidus temperature. No change in microstruc-
ture is reported even after heating for 48 h at 120
C (248 F). Furthermore, because the particles
are unreactive toward the constituents of the sol-
der, the dispersion strengthening occurs equally
in both the tin-rich and lead-rich phases. This
results in the modied solder exhibiting im-
proved ductility, tensile strength, and resistance
to creep [Mavoori and Jin 1998]. With a suffi-
cient dispersion of nanosized TiO2 particles, it
has been possible to boost the creep resistance of
soft solder at room temperature to a level com-
parable to that of the Au-20Sn alloy and increase
its tensile strength fourfold, albeit with a corre-
sponding reduction in ductility [Mavoori and Jin
2000]. One useful side benet from stabilization
Table 5.13 Compressive creep rates of normal
and dispersoid-strengthened solders at 100 C
(212 F) and an applied stress of 1.7 MPa (247
psi)
Solder Steady-state creep rate at 100 C, s1
Pb-63Sn 1.26 106
Pb-63Sn 3 vol% Al2O3 4.25 108
Pb-63Sn 3 vol% TiO2 2.61 108
Au-20Sn 8.40 108
of the alloy microstructure is that it helps sig-
nicantly with simplifying reliability prediction
modeling.
The principal impediment to the widespread
adoption of dispersion-strengthened solders is
the difficulty of manufacture. In order to be ef-
fective, the particles need to be present as a ne
and uniform dispersion in the solidied joint.
The problem is that particles tend to agglomer-
ate, trap porosity, and adversely affect the vis-
cosityand hence, spreading abilityof the sol-
der. While solder alloys can be produced
satisfactorily on a small scale in the laboratory,
the specialized equipment and processes re-
quired for their preparation and use currently
preclude them from being drop-in replacements
for conventional solder in most applications.
5.5.3 Composite Solders
Composite solders are not fundamentally dif-
ferent from the other two types of strengthened
solders described previously, in that they are con-
ventional solder alloys with improved mechani-
cal properties that arise from the presence of
small, hard particles. Thus, the strengthening
mechanisms are as mentioned previously,
namely, a combination of grain renement, grain-
boundary pinning, and impediment of the move-
ment of dislocations. What differentiates these
materials is the choice of particle to provide the
reinforcement, being predominantly copper-tin
(Cu3Sn, Cu6Sn), silver-tin (Ag3Sn), nickel-tin
(Ni3Sn4), or copper-nickel-tin (Cu9NiSn3) inter-
metallic compounds [Guo and Subramanian
2002; NEPCON West 1992; Betrabet, McGee,
and McKinlay 1991; Marshall et al. 1991]. These
additions satisfy the following conditions [Guo
and Subramanian 2002]:
The molten solder matrix wets the additives
and bonds strongly to it.
The reinforcements are partially soluble in
the molten solder at normal temperatures used
for reow, so that the reinforcements are es-
sentially stable during reow and aging.
The density of the reinforcements and solder
matrix are sufficiently similar so that gravi-
tational separation does not occur, and the
mixture remains homogeneous when the sol-
der is molten.
These reinforcing phases do not appreciably
coarsen during normal service.
220 / Principles of Soldering

They tend to retard the growth of the inter-
metallic layers at the solder/substrate inter-
face. Although such layers are essential to
successful bonding, their growth in thickness
results in deterioration in the mechanical
properties of joints.
Composite solders come in two avors. The
more common contains ne particles (<5 m, or
0.2 mil, in diameter) at low volume fractions (0.5
to 20 vol%) to produce grain renement. How-
ever, it has been found that the reinforcing par-
ticles are most effective in stabilizing the mi-
crostructure of the solder when they are of the
order of 1 m (40 in.) in size. Figure 5.27 and
Tables 5.14 to 5.17 illustrate some of the re-
ported benets to mechanical properties. In gen-
eral, metallic additions are less effective in boost-
ing creep resistance of soft solders than lower
percentages of dispersed oxides (compare Tables
5.14 and 5.13). Metal additions have a benecial
effect on creep resistance to high cycle fatigue
but are detrimental for low cycle fatigue (Tables
5.15, 5.16). Copper particulate reinforcement is
superior to silver from the point of view of creep
resistance (Table 5.14). However, in situ Cu6Sn5-
reinforced solders are superior to those strength-
ened with copper or silver. Solders containing
distributed Ni3Sn4 particles are better still, in re-
spect of high cycle fatigue and tensile properties
(Tables 5.15, 5.17).
A second and less common type of composite
solder is lled with larger particles (typically 5
to 25 m, or 0.2 to 1 mil, in diameter) at volume
fractions of 10 to 40%. These are akin to metal-
matrix composite materials and benet from en-
hanced tensile strength, resistance to creep, and
improvements to other mechanical properties at
the expense of ductility and ease of application.
The combination of higher volume fraction and
larger particle size is effective in strengthening
Fig. 5.27 Yield strength of composite solders at room tem-
perature plotted as a function of volume fraction of
the added intermetallic powder [Yost, Hosking, and Frear 1993]
the solder matrix by restraining the yield behav-
ior of the soft matrix. The material combinations
that have been evaluated are largely the same as
those described previously for the rst type of
composite.
To date, both types of composite solders have
been difficult to use in joints without a substan-
tial concentration of micropores developing,
which offset much of the gains in properties out-
lined previously. Indeed, while there is a rea-
sonable body of literature largely extolling the
mechanical properties of composite solders, the
data almost exclusively pertain to specially pre-
pared bulk samples of the ller metal and not to
the properties of joints of typical geometry and
aspect ratio.
There are three common methods by which
composite solders are prepared: powder blend-
ing, mechanical mixing, and rapid solidication.
In powder mixing, powder of the reinforcement
of interest is simply stirred into a vat of molten
solder prior to use. The difficulty with this method
is that it is difficult to obtain a uniform disper-
sion, and small particles tend to dissolve and
reprecipitate on the surface of larger particles,
resulting in a relatively coarse dispersion. Me-
chanical mixing entails placing solder powder
and powder of the reinforcing constituent in a
ball mill and working the mixture until the de-
sired distribution is obtained. Obviously, some
of the benets of preparation in this manner will
be lost when the solder is melted to effect joining.
The third method involves gas atomization of
essentially off-eutectic composition alloys. The
reinforcement arises from the ne dispersion of
primary and secondary phases of intermetallic
that precipitate out on cooling. Again, some
coarsening is inevitable when the solder is melted
in the joining operation. Mechanical mixing and
gas atomization yield powder that then needs to
be added to a binder and ux to form paste.
Composite solders generally have inferior wet-
ting and spreading characteristics, compared to
normal solders, and the visual appearance of the
joints is impaired. A frequent and very apt de-
scription is that the solder is gritty. Wetting
balance tests conrm that the wetting time of
composite solders is affected detrimentally and
the spreading is decreased by as much as 25%
[Steen and Becker 1986]. However, the wetting
angles of molten composite solders, in sessile
drop tests using ux, are not appreciably higher
than the unmodied alloys [Subramanian, Bieler,
and Lucas 1999], as shown in Table 5.14. Pro-
ponents of composite solders argue that the poor
 Chapter 5: Advances in Soldering Technology / 221

Table 5.14 Wetting and creep properties of composite solders reinforced with metallic additions
Average wetting angle on copper Creep rate at a steady stress of 17 MPa, s1
Solder (fluxed) substrates, degrees 25 C 65 C 105 C

Sn-3.5Ag 18.8 2.62 105 1.50 104 1.9 103

15 vol% Cu 45.7 (18.0 at 6 vol%) 4.70 106 2.03 105 3.95 104
15 vol% Ag 19.8 1.91 105 8.37 105 1.80 103
20 vol% Cu6Sn5, 17.6 7.6 106 (at 9.8 105 (at 5.8 104 (at
(produced in situ from 25 MPa) 13 MPa and 12 MPa and
added Cu and Sn in the 85 C) 125 C)
solder)

Adapted from Guo and Subramanian [2002]

spreading is not a great impediment for elec-
tronics assembly applications where solder
(paste) is usually preplaced and minimal further
spreading is required to effect joining. The stiffer
rheology of the semimolten composites results in
wider joints than for conventional solders, which
may be advantageous where there is a desire to
bridge gaps, although wider solder joints are in-
trinsically weaker, counterbalancing potential
benets to mechanical properties. Wide joints
are a decided disadvantage for most electronic
and optical assembly because of the relatively
poor electrical and thermal conductivity of sol-
Table 5.15 High cycle fatigue life (expressed
as cycles to failure) of composite solders,
determined at different stress levels, reversed
30 times/min (0.5 Hz)
Solder 28 MPa
Pb-63Sn 42,000
13 vol% Cu6Sn5(a) 116,000
18 vol% Ni3Sn4(b) 285,000
(a) Cu6Sn5 dispersoids, 0.42 m (16.5 in.) in diameter. (b) Ni3Sn4 dispersoids,
0.25 m (10 in.) in diameter
Table 5.16 Low cycle fatigue life (expressed as
cycles to failure) of composite solders,
determined at strain rates of 0.05 and 0.005 Hz
and a total strain of 1%
Solder 0.05 Hz, 1% strain
Pb-63Sn 980
13 vol% Cu6Sn5 750
18 vol% Ni3Sn4 160
ders, compared with metal conductors, notably,
copper, silver, and gold.
The improvements in materials properties re-
sulting from the inclusion of dispersoids in the
ller metal do not greatly expand the possibili-
ties for using lead-tin solders in load-bearing
applications. This is because the reinforcing pro-
cesses are only effective at low strain rates, as
can be seen in Fig. 5.28. At higher strain rates
likely to be experienced in load-bearing struc-
tures (102 s1 and above), properties such as
tensile strength are no different between com-
posite and conventional solder [Mavoori and Jin
1998, 2000]. Furthermore, although the disper-
soids improve the resistance to creep, high cycle
fatigue life, and stress-rupture life of solder, it is
mostly at the expense of ductility, low cycle fa-
tigue life, and fracture toughness, which are prop-
erties likely to be of greater importance in struc-
34 MPa 42 MPa
tural applications.
18,000 4,000 An interesting variation of composite solders
32,000 10,000
167,000 54,000 has been proposed that utilizes insoluble par-
ticles in the form of iron powder. Iron is wetted,
but not consumed, by alloying at a particularly
fast rate by tin-base solders and therefore pro-
vides a similar degree of reinforcement as other
dispersants. However, because iron is a soft mag-
netic material, the natural shape of the solder,
when molten, can be altered by application of an
external magnetic eld, and the modied prole
will be frozen when the solder solidies. Fields
0.005 Hz, 1% strain
in the region of 0.05 to 0.5 T (500 to 5000 G) will
2100 produce signicant height change of molten sol-
460
75
der spheres. It is suggested that this effect might
be exploited to help remove the effects of joint

Table 5.17 Tensile properties of composite solders

0.2% offset yield, Ultimate tensile Elastic modulus, Elongation to Reduction in
Solder MPa strength, MPa GPa failure, % area, %

Pb-63Sn 35 37 12 48 331
13 vol% Cu6Sn5 52 60 21 18 37
18 vol% Ni3Sn4 63 73 24 15 37
222 / Principles of Soldering
Fig. 5.28 Tensile strength of lead-tin eutectic solder with and without 3 vol% of Al2O3 dispersoids as a function of strain rate,
measured at 80 C (176 F)
gap variation when attempting to make multiple
joints in parallel, for example, ip-chip inter-
connection [McCormack, Jin, and Kammlott
1994]. Iron-containing composite solders could
be of interest for microelectromechanical sys-
tems (MEMS) fabrication.
At the time of writing, neither grain-rened,
dispersion-strengthened, nor composite solders
have yet been adopted in commercial use by
industry. Although they are undoubtedly attrac-
tive for some applications, because of the prob-
lems outlined previously, there is clearly a re-
quirement for further research before the
transition can proceed with condence.
5.6 Reinforced Solders
(Solder Composites)
Reinforced solders are ller metals that in-
corporate a reinforcing medium that is physi-
cally large in relation to the joint width. On a
simplistic level, they can be considered as visu-
ally equivalent to the use of steel bars within
reinforced concrete constructions. Because the
strengthening mechanism is bulk physical con-
straint of the solder by the reinforcement, it is
fundamentally different from the mechanisms for
strengthening solders described in section 5.5 of
this chapter.
Reports in the published literature reveal some
of the results of attempts made to load solder
alloys with a uniform distribution of strong (high-
modulus) particles or bers (typically, 100 m to
1 mm, or 4000 in. to 0.04 in., in size) of non-
metallic materials. Refractory metals have been
tried, but these tend to agglomerate when the
ller is molten, because of their higher density
[Ho and Chung 1990]. The most successful av-
enue to date has been to incorporate into the ller
chopped carbon bers, electroplated with nickel
or copper so they are wettable by solder. With
lead-tin eutectic solder, the results show a three-
fold enhancement of the shear and tensile
strength of the joints with respect to the unmodi-
ed llers and, more particularly, a signicant
reduction in the thermal expansivity of the ller.
The published data show that the tensile
strength of joints formed with lead-tin with re-
inforcement can be as high as 65% of the rule-
of-mixtures for the relative proportions by vol-
ume of solder and reinforcement present.
Approximately 15% by volume of bers is the
maximum that can be incorporated into the sol-
der while retaining acceptable workability in the
molten state. With this level of ber loading,
joint strengths in excess of 250 MPa (5.2 lb/ft2)
at room temperature can be achieved [Ho 1996].
An alternative approach to attempting to ll a
joint with a molten alloy loaded with insoluble
bers is to pack the joint with dry bers and use
a conventional solder, with or without ux, as
required, to inltrate and ll the interstices. Ob-
viously, the bers need to be metallized so they
can be wetted by the solder. Much higher load-
ings of reinforcement can then be achievedup
to 55% of the volume of the solder. As alluded
to previously, because carbon bers have a small
but negative coefficient of thermal expansion, at
42% by volume, the thermal expansivity of the
solder composite declines to zero over the tem-
 Chapter 5: Advances in Soldering Technology / 223

perature range of 25 to 100 C (77 to 212 F).
Solders reinforced in this manner might there-
fore offer benets where there is a requirement
to join low-expansivity materials, such as ce-
ramics to metals, where a wide joint gap is man-
datory. The high thermal expansivity of a con-
ventional solder alloy (typically, >20 106/K,
or 20 ppm/K) introduces a shear stress at the
component/solder interface, which is overcome
by using a carbon-ber-loaded solder. The re-
duction in the thermal expansivity brought about
by the ber addition reportedly accounts for a
threefold enhancement in fatigue life on thermal
cycling that is observed in bonded assemblies of
this type.
An additional benet of reinforced solders is
that carbon bers can have very high longitudi-
nal thermal conductivity, so the presence of a mat
of bers in a joint can help to redistribute local
hot spots within the plane of the joint. Clearly,
there is scope for further research in this area.
When contemplating using ber-reinforced
solders, one of the key targets is to obtain a
void-free joint; otherwise, poor joint lling miti-
gates the strengthening effect. This end is greatly
assisted when the inltration of solder into the
ber bundle is promoted not only by metallur-
gical wetting of the solder, but when surface
tension forces are exploited to achieve sponta-
neous inltration into the interstices. This has
been studied from a theoretical standpoint, albeit
simplied, and some of the key results are pre-
sented in Table 5.18. In summary, provided that
the wetting angle of the solder to the metalliza-
tion on the reinforcement material is below 45,
then spontaneous inltration should take place
irrespective of the aspect ratio of the reinforce-
ment.
If the reinforcement medium is not closely
packed, then the critical wetting angle decreases
accordingly. The corollary is that unless the mini-
mum conditions given in Table 5.18 are achieved,
the resulting joint will contain voids, unless ex-
ternal pressure is applied to force the molten
metal into the interstices of the reinforcement
material [Yang and Xi 1995].
5.7 Mechanical Properties and
Numerical Modeling of Joints
This section considers methods for quantify-
ing the mechanical integrity of joints and pre-
dicting their dimensional stability under speci-
ed environmental conditions. These are really
two independent issues that are addressed sepa-
rately. Although there have not been recent sig-
nicant advances in methods of measuring the
mechanical properties of solders and joints, val-
ues for various material parameters, such as
Youngs modulus, are required for numerical
modeling techniques. It is therefore important to
have an appreciation of the limitations inherent
in the derivation of parameters and the scope of
their applicability. Modeling of the lifetime of
joints, when subject to cyclic conditions, has
made considerable progress in recent years and
is also considered in this section.
5.7.1 Measurement of
Mechanical Properties
Measurement of the mechanical properties of
bulk solder specimens and soldered joints should
be an uncomplicated process. Suitable test meth-
ods, specimen designs, and methods of prepa-
ration for metallurgical specimens are all dened
by standards. However, even obtaining what
might appear to be relatively straightforward
property data on bulk solders is fraught with
problems. A few of the factors that inuence the
mechanical strength of solders include:
Test method used
Procedure used to prepare the samples

Table 5.18 Calculated critical angle for a liquid to spontaneously inltrate the interstices in
selected close-packed structures, and the minimum packing density necessary to achieve lling even
with perfect wetting. Above the minimum packing density, spontaneous inltration is relatively easy to
achieve, even when the wetting is relatively poor.
Minimum volume fraction of
Critical wetting angle reinforcement for spontaneous
Reinforcement type for spontaneous inltration, degrees inltration at 0 contact angle, %

Unidirectional bers 45 40
Body-centered cubic packed mono-sized 65 20
spheres
Face-centered cubic/hexagonal close-packed 50 40
mono-sized spheres
224 / Principles of Soldering
Precise sample geometry
Test temperature
Strain rate employed in the test
Microstructure of the ller metal
These sensitivities help to explain why pub-
lished values of the properties of solders can vary
by at least an order of magnitude for what, to the
inexperienced observer, might otherwise appear
to be identical samples [Plumbridge 1996]. The
complexity of solder behavior, especially that of
the softer indium-containing alloys, arises partly
from the fact that at room temperature these al-
loys are working close to their melting point,
expressed in Kelvin. This means that atomic dif-
fusion can occur rapidly, and hence, the pro-
cesses of annealing, alloying, and precipitation
are observed during testing, in addition to con-
ventional bulk metallurgical phenomena such as
necking. In actuality, the response of a solder to
mechanical stress is a complex combination of
elastic, anelastic, and viscoplastic behavior.
Obtaining consistent values for the mechani-
cal properties of soldered joints is even more
difficult. In addition to the variables cited pre-
viously in respect to bulk solders, the strength of
a joint is additionally inuenced by a number of
factors, including:
Dimensions (thickness and area)
Method used for making the joint
Heating excursion (integrated temperature
and time)
Cooling rate
Substrate materials
Impurity content of the ller
Age of the joined assembly
Variables such as strain rate and temperature
alone can be manipulated to give joint strengths
that vary by more than two orders of magnitude
from identical samples prepared under rigorous
laboratory conditions [Jones et al. 1997; ITRI
1987].
Notwithstanding the previously mentioned
considerations, a further issue then arises as to
whether the mechanical properties measured are
satisfactory in the context of the service require-
ments of the assembly. For example, a solder
joint between a silicon die and a ceramic package
may achieve a shear strength of, say, 40 MPa
(5800 psi). Because the die weighs only a few
grams and there are no mechanical contacts made
to it, then clearly, the joint is more than strong
enough to hold the die in place. The majority of
soldered joints in electronic products are not
made to meet load-bearing requirements but to
effect electrical connectivity, thermal conductiv-
ity, or a hermetic seal between components. Cor-
relation between these variables and basic me-
chanical properties (e.g., tensile and shear
strength) is often obtuse. Failure of the joined
assembly to pass a die-shear test is only an in-
dicator that it falls short of a minimum require-
ment. A joint can possess high levels of voids and
therefore possess impaired local thermal con-
ductivity and leakage paths, giving rise to her-
meticity failure long before this is reected in its
shear strength.
In the electronics and photonics industries, the
difficulty of obtaining consistent mechanical
property data and correlating this information
with the parameters of real interest have been
recognized, and, as a result, simple mechanical
testing has largely fallen out of favor. Instead, it
is more common to build either complete prod-
ucts or representative subparts and subject them
to some form of accelerated or extended test that
is considered to encompass the rigors of life in
service. The components are then assessed for
signs of functional degradation. Any deciencies
that are detected and can be attributed to unsat-
isfactory joints are addressed accordingly.
More information about mechanical property
tests can be found in the planned companion
volume Principles of Brazing. Brazed joints are
often expected to carry mechanical loads, and
hence, issues such as the choice of test method
used to measure the mechanical integrity of a
joint become more directly relevant.
5.7.2 Numerical Modeling of Joints
Computational methods are nding growing
use in modeling how components and products
will respond to various stimuli, such as changes
in temperature, mechanical loads, and exposure
to chemicals. These naturally include consider-
ation of soldered joints.
There are basically two types of model: those
that calculate the dimensions of the product, and
others that predict the lifetime of joints. Models
have also been devised that describe the prole
of molten solders wetted onto solid substrates.
An example of this type of model is referred to
in section 5.2.6 in this chapter.
5.7.2.1 Dimensional Stability of
Soldered Joints
One of the better known methods of calcu-
lating the dimensional stability of joints is nite-
 Chapter 5: Advances in Soldering Technology / 225
element analysis (FEA). In FEA, a component or
assembly is modeled in a geometrical manner in
terms of a mesh of smaller units or elements,
with other dimensions of each element scaling to
a set of properties of interest. The visible mesh
usually represents two or three orthogonal di-
mensions of the part, while the other dimensions
are used to set how each element can respond to
stimuli. Common parameters are Youngs modu-
lus for mechanical behavior, and thermal expan-
sivity and thermal conductivity for thermal be-
havior. In a realistic model, these parameters are
not xed but vary with temperature. Constraints
are then applied to the mesh, with matrix algebra
used to obtain a comprehensive and numerically
convergent solution. The large number of cal-
culations that are needed requires a computer for
this task.
There are now a number of commercial FEA
software packages available to suit a range of
applications. A FEA prediction of the deforma-
tion that occurs in a ceramic-metal bond on cool-
ing from the solidus temperature of the ller is
given in Fig. 5.29.
Despite the complexity of the modeling tech-
niques, simplifying assumptions have to be made.
Often, an axis of spatial symmetry is dened,
and only half or one-quarter of the assembly is
modeled. The oversimplication represented by
this assumption can easily be demonstrated for
many situations, especially when transient con-
ditions and thermal gradients are taken into
consideration. Also, when these models are ap-
Fig. 5.29 Finite-element analysis prediction of the geometry of a ceramic-metal brazed joint, at its periphery, at the solidus
temperature of the ller alloy and on cooling to room temperature
plied to soldered assemblies, it is normally as-
sumed that the joints are uniform in their com-
position and physical properties. This is patently
not correct. A further restriction of applicability
arises because most modeling programs assume
that the interface between the components and
the solder is relatively abrupt, whereas, in prac-
tice, it is a highly complex region that has
signicant compositional and microstructural
differences, generated by reaction across inter-
faces, and often these are not stable with time,
temperature, or stress. Fortunately, the greater
sophistication of software and increased com-
puting power are enabling these features to be
built into FEA models.
To successfully apply such models requires
knowledge of the properties of materials in an
assembly and how those properties change in
different environments. For mechanical models,
the environmental variables are usually stress,
strain, and temperature. Thus, there is often a call
on the joining technologist to provide data on the
mechanical and other physical properties of ller
metals. For the reasons outlined in the preceding
section, this is almost an impossible requirement
to fulll reliably. However, the potential savings
afforded by modern computer modeling tech-
niques in their ability to assist in achieving right-
rst-time designs and thereby speed product time
to market means that there is often considerable
pressure to provide appropriate material prop-
erty data! Table 5.19 represents an attempt to
provide indicative data of the bulk properties of
226 / Principles of Soldering

a few common solders and some of their com-
mon constituents at room temperature.
When running a model, the values given in
Table 5.19 can be taken as a starting point and
modied to allow for the joint thickness, inter-
metallic formation, compositional change, and
the properties of the components on either side
of the joint. Temperature-dependent terms can
then be added as appropriate. A literature search
will usually throw out values for temperature-
and strain-rate-dependent terms that can be
adopted if they appear valid for the situation
under consideration. Often, it is necessary to de-
sign and test joints of simplied geometry, in
order to provide condence in the values and
their sensitivity to second-order effects.
While acknowledging the many limitations of
the data, as mentioned previously, Table 5.19
does reveal some interesting trends in comparing
the alloys with the constituent pure elements.
Almost invariably, solders possess greatly infe-
rior thermal and electrical conductivity, com-
pared with pure metals. These characteristics fol-
low from their heterogeneous microstructure and
phase boundaries, which impede the ow of elec-
trons and phonons. The effect is more pro-
nounced on thermal conductivities because of
the dominant contribution of phonons, or lattice
waves, to thermal conductivity at room-ambient
and elevated temperatures. Phonons are more
strongly scattered by the inhomogeneous micro-
structure than are electrons, which dominate elec-
trical conductivity and low-temperature thermal
conductivity. At rst sight, it might be expected
that the gold-rich solders would have properties
not dissimilar to pure gold. However, the large
difference in density between gold, on the one
hand, and the alloying elements, on the other,
means that in terms of volume fraction, the gold-
silicon eutectic alloy, for example, possesses
close to 20 vol% Si. The gold-tin eutectic solder,
Au-20Sn, is not a mixture of the constituents but
an alloy of two intermetallic compounds of gold
and tin (AuSn and Au5Sn), both of which have
very different characteristics to the pure ele-
ments. In general, if compound formation occurs
in the creation of solders, then the properties
cannot be predicted reliably without direct
knowledge of the mechanical and physical prop-
erties of the constituent phases.
5.7.2.2 Prediction of Joint Lifetime
The traditional methods for predicting the life-
time of a fabricated part and the joints contained
therein are accelerated and extended testing. Be-
cause these are time-consuming and expensive
to undertake, much effort has been devoted to
attempting to predict joint life by numerical mod-
eling. The ability to do this is particularly rel-
evant to the needs of the PCB industry, where it
is crucial to know the lifetime of the product and
also for manufacturers to convincingly demon-
strate to customers that the soldered joints will
meet long-term expectations of reliability.
The approach to this problem started with clas-
sical fatigue theory, which was rapidly found to
be wholly inadequate for describing the behavior
of solders. Gradually, more and more rate- and
temperature-dependent parameters were added
to the models. Although the quality of the analy-
sis steadily improved, unexpected outcomes were
as common as accurate predictions. Some test-
ing, albeit less extensive, remained necessary to
validate the model.
Over the last few years, an alternative ap-
proach that uses a concept termed strain energy
density has been gaining favor [Morrow 1964;

Table 5.19 Indicative property values of selected solders and pure metals in bulk form at room
temperature
Tensile Youngs Elongation Electrical Thermal Thermal
Element Hardness, strength, modulus, Poissons to failure, resistivity, conductivity, expansivity,
or solder HV MPa GPa ratio % cm W/m K 106/K

Ag/Au 20 150 80 0.40 40 2 350 20

In 5 5 10 0.45 50 10 80 25
Pb 10 15 15 0.42 50 20 35 30
Sn 15 30 50 0.35 30 15 70 25
Ag-97In 5 5 10 0.45 50 10 50 25
Ag-96Sn 15 60 40 0.35 30 15 50 25
Bi-50Sn 25 60 ... ... 1 35 20 20
Cu-99Sn 10 30 50 0.35 30 15 50 25
In-50Sn 5 20 ... ... 50 30 20 20
Pb-63Sn 15 40 30 0.25 30 15 40 25
Au-3Si 100 300 80 ... 1 20 30 15
Au-20Sn 100 275 60 0.30 5 10 60 10
 Chapter 5: Advances in Soldering Technology / 227
Vaynman and McKeown 1993]. In very simplis-
tic terms, this model attempts to examine whether
the deformation of a joint on single or multiple
stress cycles exceeds the ability of the solder to
absorb and/or dissipate the energy imparted to it,
through creep or microstructural changes. Where
microstructural changes do occur, the model is
able to account for the material change on sub-
sequent cycles. Although its complexities are be-
yond the scope of this book, the important point
to note is that the quality of the joint lifetime
predictions now being made generally accord
with experimental results. For examples of ap-
plication of this method, reference may be made
to Lau [2000], Lau and R. Lee [2001], Lau et al.
[2002], and similar publications.
5.8 Solders Doped with
Rare Earth Elements
A relatively new area of solder research con-
cerns doping of traditional ller metals with rare
earth elements. Some results are reported in the
literature, and patents have been led in respect
of certain composition ranges. Rare earth doping
of solders is of interest because it affords a pos-
sible means of favorably modifying the charac-
teristics of solders in the molten and solid states
and enabling them to approximate more closely
to the materials properties desired by the manu-
facturing industry. A large proportion of the work
done in this area has been in conjunction with
lead-free solders. These solders are discussed in
section 5.1 of this chapter.
The rare earth elements are so described be-
cause they were originally thought to have a low
abundance in the Earths crust and to be difficult
to win from it and to separate from each other.
It is now known that lanthanum, cerium, and
neodymium are actually more abundant than
lead, and vast ore reserves have been found in
China and the United States. There are thirty rare
earth elements, which is really another name for
the elements contained in the lanthanide and ac-
tinide series of group three of the periodic table.
However, one element of the lanthanide series
(promethium) and most of the actinides are trans-
uranium elements, that is, manmade and atomi-
cally unstable. Sometimes, reference will be seen
to a rare earth called mischmetal. This is actually
an alloy containing a high concentration of the
rare earth elements in proportion to their natural
abundance, and therefore, its composition varies
with the ore from which it was obtained: mona-
zite, xenotime, or bastnasite. Mischmetal is
sometimes given the chemical symbol M, and it
contains, typically, 50% Ce and 30% La. The
general symbol used for rare earth is RE. Rare
earth elements have the common attribute that
they are extremely reactive toward other metals
and most atmospheres.
The addition of rare earth elements to solders
has a number of effects on properties that vary
with concentration.
5.8.1 Effect of Rare Earth
Additions on Solder Properties
Recent research has shown that doping of sol-
ders with rare earths appears to enhance their
properties while introducing few drawbacks, and
the melting range of solder alloys is hardly af-
fected (Fig. 5.30) [Ramirez, Mavoori, and Jin
2002; Wang, Yu, and Huang 2002; Chen et al.
2003]. Small additions of rare earths:
Grain rene solders and produce within them
a dispersion of hard, insoluble intermetallic
particles that benets their mechanical prop-
erties. Thus, for example, the addition of
0.4% rare earths to the Ag-96.5Sn solder
halves the average grain size, and this pro-
duces a 12% boost in strength and ductility,
as can be seen in Fig. 5.31 [Wang, Yu, and
Huang 2002]. Likewise, a 0.083% addition
of rare earths to 3.33Ag-4.83Bi-91.84Sn sol-
der has similar effects on grain size, strength,
and ductility [Xia et al. 2002]. However, the
effect of rare earth doping and the associated
grain renement of the microstructure on duc-
tility appears to be solder specicin the
case of tin-zinc eutectic, this property is ac-
tually impaired (Fig. 5.32) [Wu et al. 2002].
Thermodynamic analysis shows that the
dispersed intermetallic phase, in tin-base sol-
ders, is based on the composition Sn3RE [Ma
and Yoshida 2002]. These intermetallic
phases predominantly congregate at grain
boundaries [Ying, Hongyuan, and Yiyu
1994].
The agglomeration of rare earth interme-
tallic compounds at grain boundaries is re-
sponsible for considerably increasing the re-
sistance of solders to creep at room
temperature, as shown in Fig. 5.33 and 5.34,
because creep in solders occurs primarily by
grain-boundary sliding [Chen et al. 2002].
Progressively increasing the concentration of
228 / Principles of Soldering
Fig. 5.30 Solidus and liquidus temperatures of 3.8Ag-0.7Cu-95.5Sn solder as a function of the added rare earth (RE) concentration
the rare earth elements increases the prefer-
ence for discrete compound formation, which
would explain why the creep properties peak
at 0.1% RE addition in 3.8Ag-0.7Cu-95.5Sn
solder. Exposure to elevated temperature re-
duces the benet to be derived from grain-
Fig. 5.31 Tensile strength and elongation to failure of Ag-
96.5Sn solders doped with cerium and lanthanum.
The samples were chill-cast ingots of solder, 20 mm (0.8 in.) long
by 10 mm (0.4 in.) in diameter, tested at room temperature and
a strain rate of 4 103/s. RE, rare earth
Fig. 5.32 Tensile strength and elongation to failure of Sn-9Zn
solders doped with lanthanum. The samples were
chill-cast ingots of solder, 25 mm (1 in.) long by 5 mm (0.2 in.)
in diameter, tested at room temperature and a strain rate of 5
103/s. RE, rare earth
boundary pinning by the rare earth interme-
tallic compounds, because other material
transport mechanisms take effect. Thus, at 65
C (149 F), the creep-rupture life is only
double that of the simple ternary solder, and
any advantage disappears completely above
Fig. 5.33 Stress-rupture life of Sn-3.5Ag and Sn-3.5Ag-
0.25RE solders for an applied stress of 20 MPa
(2900 psi) at 50 C (122 F)
Fig. 5.34 Stress-rupture life of 3.8Ag-0.7Cu-95.5Sn solder as
a function of the rare earth (RE) content for an
applied stress of 16.5 MPa (2400 psi) at room temperature
 Chapter 5: Advances in Soldering Technology / 229

Fig. 5.35 Contact angle and spread area of Ag-96.5Sn solders doped with cerium and lanthanum melted on copper at 300 C (572
F) for 30 s under cover of RMA ux. RE, rare earth

100 C (212 F). This result puts a question

mark on the practical benet of the observed
improvement in creep resistance. Also, the
available data pertain to tens of hours of test
duration, not thousands of hours of service
life, and this issue needs to be investigated.
Decrease the wetting angle of the solder on
metal substrates and generally enhance
spreading. The effect on contact angle, wet-
ting force, and wetting time is shown in Fig.
5.35 for silver-tin eutectic solder on copper.
Rare earth doping of the Sn-9Zn solder im-
proves its wetting on metal substrates in a
similar way (Fig. 5.36) [Wu et al. 2002]. Like-
wise, small additions of rare earths have been
found to reduce the contact angle of indium
solders on silver under inert atmosphere, as
described in Chapter 3, section 3.3.8.4 (Fig.
3.28). The tests described were conducted in
air using ux. It is presumed that the role of
the rare earths is to reduce the interfacial
tension between the solder and ux [Wang,
Yu, and Huang, 2002]. It would seem that the
reactive rare earth elements also help desta-
bilize oxide lms on the surface of the solder
and substrate. Certainly, once the rare earth
concentration exceeds approximately 0.2%,
the modied solders are able to directly wet
and bond to nonmetals, such as silica. This
has been demonstrated for both silver-tin eu-
tectic solder and the high-melting-point Au-
20Sn alloy. The bonding mode is as de-
scribed in Chapter 4, section 4.1.2.2, namely,
migration of the rare earth element to the
solder/substrate interface and chemical bond-
ing by reduction of the nonmetal surface.
Unfortunately, the concentration of rare earth
elements necessary to achieve this behavior
results in a loss of many of the boosted
Fig. 5.36 Wetting force of Sn-9Zn solders doped with lute-
tium on copper using rosin-activated ux at 245 C
(473 F) in air. RE, rare earth
mechanical properties referred to earlier and
impairs the ductility of all solders [Ramirez,
Mavoori, and Jin 2002].
5.8.2 Implications for
Soldering Technology
Most results reported to date relate to bulk
solder specimens. The strength of lap joints in
copper testpieces was found to be indifferent to
rare earth doping. This somewhat disappointing
result was attributed to porosity in the joints,
originating from the use of a simple organic
ux for the joining operation. Whether this po-
rosity is an intrinsic characteristic of rare-earth-
doped solders or indeed merely due to inappro-
priate choice of ux is uncertain, because
currently, there are few reported studies of me-
chanical properties of joints made and assessed
under controlled conditions.
As far as the authors are aware, solders doped
with rare earth elements cannot yet be found in
manufacturers catalogs.
230 / Principles of Soldering
5.9 Diffusion Soldering
In soldering, wetting of the component sur-
faces is not always easy to achieve, and when it
does occur, the resulting alloying between the
ller and components can cause excessive erosion
of the parent materials, embrittlement of joints
due to the formation of phases with inferior me-
chanical properties, and other undesirable effects.
These problems notwithstanding, solders have
the singular merit of being able to ll joints of ir-
regular dimensions and produce well-rounded l-
lets at the edges of the joint.
Diffusion bonding sidesteps the need for wet-
ting and spreading by a ller metal (see Chapter
1, section 1.1.7.2). Once formed, diffusion-
bonded joints are stable to high temperatures,
so that the service temperature of the assembly
can actually exceed the peak temperature of the
joining process without risk of the joint remelt-
ing. While the formation of undesirable inter-
metallic phases can also occur in diffusion bond-
ing, because there are usually fewer constituents
involved, it is easier to select a safe combina-
tion of materials. However, diffusion bonding
tends to be limited as a production process,
because it is not tolerant to joints of variable
width, and moreover, its reliability is highly
sensitive to surface cleanliness. High loads (typi-
cally 10 to 100 MPa, or 1.4 103 to 1.4 104
psi) have to be applied during the bonding cycle
to ensure good metal-to-metal contact across
the joint interface. Also, the duration of the
heating cycle is typically hours, compared with
seconds for soldering, because solid-state diffu-
sion is much slower than wetting of a solid by a
liquid. These factors and the absence of any
signicant llets to minimize stress concentra-
tions at the edges of joints (see Chapter 4,
section 4.2.4) considerably limit the applica-
tions of diffusion bonding.
There exists a hybrid joining process that com-
bines the good joint lling, llet formation, and
tolerance to surface preparation of conventional
soldering, together with the greater exibility
with regard to service temperature and metallur-
gical simplicity that is obtainable from diffusion
bonding. This process is called diffusion solder-
ing, sometimes also referred to as transient liquid-
phase (TLP) joining. Its higher-temperature ana-
logue, diffusion brazing, is an established joining
and repair process that has been used for decades
in the aerospace industry and is described in the
planned companion volume Principles of Braz-
ing.
5.9.1 Process Principles
Diffusion soldering uses a thin layer, typically
5 m (200 in.) or less, of molten ller metal to
initially ll the joint clearance, but during the
heating stage, the ller diffuses into the material
of the components to form solid phases, raising
the remelt temperature of the joint. At this stage,
isothermal solidication occurs, and further re-
action proceeds by solid-state diffusion until the
process cycle is complete. Due to the generation
of liquid in the joint, the necessary applied pres-
sures are much less than those required for nor-
mal diffusion bonding and are typically in the
range of 0.5 to 1 MPa (70 to 140 psi). Diffusion
soldering therefore provides the ready means to
ll joints that are not perfectly smooth or at (a
feature of liquid-phase joining) while offering
great exibility with regard to process tempera-
ture in relation to the service temperature of the
product. Because of these features of the process
and also the precise conditions used in imple-
menting it, especially the controlled thinness of
the layer of low-melting-point ller and speci-
ed loading applied to the joint, the following
additional advantages are obtained:
Good reproducibility of joints
Excellent joint lling that applies to both
small- and large-area joints. Joints free of
voids ensure leaktightness, which is impor-
tant in situations where the joint is made to
provide a seal.
Tight control of edge spillage, which can be
kept to a minimum
Attainment of very narrow joints, typically
less than 10 m (400 in.), which benets the
mechanical properties, as compared with con-
ventional solder
Exploratory work on diffusion soldering was
done in the mid-1960s [Bernstein and Bartho-
lomew 1966; Bernstein 1966]. This activity,
which was described as solid-liquid interdiffu-
sion bonding, was primarily concerned with low-
temperature bonding of power semiconductor die
involving indium and gold. In the processes that
Bernstein investigated, the emphasis was on low
temperatures, typically below 160 C (320 F),
and the endpoint of the process left thick inter-
metallic phases in the joint, which tended to com-
promise the mechanical integrity, although it is
adequate for the intended function.
Alloy systems suitable for diffusion soldering
will possess a phase constitution that includes a
relatively low-melting-point constituent
ideally, a eutectic reactionto initiate the melt-
 Chapter 5: Advances in Soldering Technology / 231

ing process and a higher-melting-point phase or
solid solution on which to terminate solidica-
tion. Wilde and Pchalek [1993] identied the
binary combinations of the precious metals (sil-
ver, gold, palladium) and also copper and nickel,
together with either tin or indium as the low-
melting-point constituent, as among the most
suitable systems for diffusion soldering. Some of
the process details for these combinations, as
reported in the published literature, are listed in
Table 5.20.
In most of the published studies of diffusion
soldering, the emphasis has been on limiting the
process temperature to make the joining opera-
tion suitable for the attachment of dies and other
electronic components, which are temperature-
sensitive. In consequence, a compromise has had
to be struck whereby intermetallic phases are
tolerated at the expense of mechanical strength.
This trade-off is usually acceptable, because the
strength requirement in die attachment is gen-
erally low. For example, the gold-indium joints
obtained by Wang et al. [2000], which achieved
a shear strength of 12 MPa (1740 psi), were
sufficiently strong to satisfy MIL STD 883E,
method 2019.7 acceptance criteria.
At rst sight, the cost of the precious metals,
especially gold, might be considered an impedi-
ment. However, the depth of interaction of these
metals with the ller metal is shallow in diffu-
sion-soldering processes, so that they can be ap-
plied as thin metallizations (<10 m, or 400 in.)
to the component surfaces without the danger of
the component materials entering the reaction.
The contribution to component cost from such
thin layers is relatively small. The use of copper,
silver, or gold as surface coatings confers the
particular advantage that, being relatively noble,
they are readily wetted, even in slightly oxidiz-
ing atmospheres. This makes diffusion-soldering
processes using these metals relatively tolerant
to the condition of the atmosphere in which the
joining operation is carried out. In the published
reports referred to in Table 5.20, the authors em-
phasize their ability to achieve satisfactory joints
without the use of uxes in moderately protect-
ing atmospheres, such as a shroud of nitrogen.
The sequence of steps involved in making a
diffusion-soldered joint is shown schematically
in Fig. 5.37.
5.9.2 Diffusion Soldering of Silver
The authors have found the silver-tin system
to be one of the most versatile for diffusion sol-
dering of components for use in the electronics
industry, because well-lled and ductile joints
can be produced using compressive loads of as
little as 0.5 MPa (70 psi) [Jacobson and Hump-
ston 1992]. Silver-tin solder of eutectic compo-
sition reacts with silver to form the Ag3Sn phase
as a continuous interfacial layer, which is both

Table 5.20 Selected diffusion-soldering systems and process details, as reported in the published
literature
Maximum
ller Process Remelt
Filler thickness, temperature, Applied temperature,
Substrate metal m C Time, min load, MPa C Ref

Copper Indium 1.5 180 4 0.5 >307 Sommadossi et al. [2000]

Copper Tin 1.0 280 4 0.51.0 >415 Bartels et al. [1994]
5.0 300 20 Not >676 Kato, Horikawa, and
specied Kageyama [1999]
3050 300 300 Not >676 Kang et al. 2002
specied
Silver Tin 5.0 250 60 1.0 >600 Jacobson and Humpston
[1992]
2.0 250350 10 0.17 >600 Wilde and Pchalek [1993]
Silver Indium 6.0 175 120 0.5 >880 Jacobson and Humpston
[1992]
Gold Tin 4.0 450 60 1.0 >900 Humpston, Jacobson, and
Sangha [1993]
2.25 310 13 0.28 >278 Matijasevic, Lee, and
Wang [1993]
2.0 260 15 0.3 >278 Lee and Wang [1992]
Gold Indium 5.0 200 0.5 1.25 >495 Wang et al. [2000]
2.0 160240 10 0.05 >495 Wilde and Pchalek [1993]
Nickel Tin 1.8 300 6 0.8 >400 Khanna, Dalke, and Gust
[1999]
1.8 400 2160 0.8 >977 Khanna, Dalke, and Gust
[1999]
232 / Principles of Soldering
tough and strongly adherent to the other phases
in this alloy system. The rate of reaction between
silver and molten tin has been characterized and
is represented graphically in Fig. 2.35. The con-
trollable nature of the alloying reaction in the
conventional soldering system is indicated by
the general prole of the erosion curves, which
show that the reaction is self-limiting in char-
acter, within the context of realistic processing
cycle times and temperatures. If a thin layer of
tin, typically 5 m (200 in.) thick, is sand-
wiched between two components, each covered
with a 10 m (400 in.) thick layer of silver, and
heated to 250 C (480 F), the tin will melt and
react with the silver to form Ag3Sn. On continued
heating, the tin is progressively converted to this
compound until no liquid tin remains. By keep-
ing the tin layer thin, it completely reacts to form
solid phases at the joining temperature in less
than 1 min.
The remelt temperature of a silver-tin diffu-
sion-soldered joint is determined by the phases
that are present. Immediately after the liquid tin
has been consumed by reaction, the remelt tem-
perature is that of the Ag3Sn compound, which is
480 C (895 F). Longer heating times promote
Fig. 5.37 Schematic illustration of the steps involved in mak-
ing a diffusion-soldered joint
continued diffusion of tin from theAg3Sn reaction
zone into the silver. Consequently, the width of
this zone decreases as it is replaced rst by Ag5Sn
() and, ultimately, by a solid solution of tin in sil-
ver, as anticipated from the silver-tin phase dia-
gram given in Fig. 2.9. This progression is illus-
trated by the series of micrographs shown in Fig.
5.38. As the reaction with the silver proceeds, the
remelt temperature rises progressively toward the
melting point of silver (962 C, or 1764 F). Me-
chanical property measurements have shown that
the shear strength of diffusion-soldered joints
containing theAg3Sn phase is close to the 25 MPa
(3600 psi) value for conventional soldered joints
made with theAg-96.5Sn eutectic solder to silver-
coated components. As the joint microstructure
converts to a silver solid solution, the mechanical
properties shift in tandem toward those of pure sil-
ver, with the shear strength increasing toward 75
MPa (11,000 psi). This can be signicant from an
applications point of view, because the strength of
silver is approximately three times that of the Ag-
96.5Sn eutectic alloy, which itself is superior in
this respect to the common Pb-62Sn solder at
room temperature by a factor of two to three
[Harada and Satoh 1990]. Another attractive fea-
ture of the silver-tin alloy system for diffusion sol-
dering is that there is negligible volume contrac-
tion as the reaction proceeds, which is a fortuitous
consequence of the closely similar specic vol-
umes of the various phases. Therefore, the ten-
dency to form voids or cracks as a result of volume
change is minimal.
Diffusion-soldering processes are not routinely
encountered but are used commercially. One ex-
ample is as a method of attachment of silicon
power devices to molybdenum heat sinks [Jacob-
son and Humpston 1992; Humpston et al. 1991].
Replacing brazed joints made using an industry
standard process, involving the Al-12Si alloy,
with silver-tin diffusion soldering provided a
means for reducing the process temperature from
over 600 to 275 C (1112 to 527 F), which de-
creased the bimetallic, center-to-edge bow by
240% for a typical 50 mm (2 in.) component. Be-
sides silver-tin, a silver-indium diffusion-solder-
ing process is an alternative, offering a lower pro-
cess temperature [Humpston and Jacobson 1990].
However, the associated processing involved is
more complex the plating of indium is less stan-
dardthanthatoftin,andthemorerefractorynature
of indium oxides makes it necessary to apply spe-
cial surface treatments to exposed indium sur-
faces prior to the bonding operation [Sommadossi
et al. 2002].
 Chapter 5: Advances in Soldering Technology / 233

5.9.3 Diffusion Soldering of Gold
The gold-tin alloy system has provided the
basis for the diffusion-soldering process for join-
ing items of high-carat gold jewelry below 450
C (842 F). The traditional gold jewelry manu-
facturing route involves the use of the so-called
carat gold solders, which are actually brazing
alloys with working temperatures above 800 C
(1470 F) [Rapson and Groenewald 1978; Nor-
mandeau 1996]. The high temperatures involved
are detrimental to the mechanical strength of
high-carat gold jewelry, because they anneal and
soften rapidly when heated above approximately
450 C (842 F). Further details on carat gold
solders and the metallurgy of gold jewelry can be
found in the planned companion volume Prin-
ciples of Brazing. Diffusion soldering provides a
satisfactory alternative joining process. In trials,
it was found that a tin coating 3 to 4 m (120 to
160 in.) thick was generally sufficient to ensure
complete joint lling and the formation of small
edge llets. Provided that the peak process tem-
perature exceeds 419 C (786 F), the melting
point of the AuSn intermetallic compound, the
tin will transform initially to the high-gold in-
termetallic compound Au5Sn and, on continued
heating, to gold solid solution. The Au5Sn com-
pound contains approximately 90 wt% Au and so
meets the 18 carat requirement of the jewelry
item. Prolonged heating is undesirable, because
it results in softening of the gold assembly, as

Fig. 5.38 Series of micrographs showing the progressive change in joint microstructure that occurs on making a diffusion-soldered
joint using tin in combination with silver metallizations applied to copper substrates. 400
234 / Principles of Soldering
reected by the grain growth, and also in Kirk-
endall voiding in the centerline of the joint. One
hour at 450 C (842 F) under a compressive
loading of 1 MPa (145 psi) was found to be an
acceptable compromise for the processing con-
ditions [Humpston, Jacobson, and Sangha, 1993].
Figure 5.39 shows a bracelet and matching ear-
ring set that was assembled by this method and
exhibited at the World Jewelry Trade Fair held in
Basel in 1992.
5.9.4 Diffusion Soldering of Copper
Copper-tin and copper-indium are also suit-
able systems for diffusion soldering. However,
when the copper-tin joining process is carried out
below 676 C (1249 F) and the copper-indium
process is operated below 631 C (1168 F), they
result in the formation of planar intermetallic
phases that have limited fracture toughness and
are responsible for relatively weak joints. In the
case of copper-tin, these intermetallic phases are
Cu6Sn5 () and Cu3Sn (), while Cu7In4 () and
Cu7In3 () are found in the copper-indium sys-
tem. The Cu3Sn () and Cu7In3 () phases form
adjacent to the surface of the copper layer [Kato,
Horikawa, and Kageyama 1999; Kang et al.
2002].
In the copper-tin process, it has been shown
that the formation of the brittle intermetallics can
be suppressed by raising the joining temperature
above the melting point of the Cu3Sn interme-
tallic (676 C, or 1249 F). The successful dif-
Fig. 5.39 Parts of an 18 carat gold bracelet and matching
earring set assembled using the gold-tin diffusion-
soldering process at 450 C (842 F). The unusually low process
temperature enables the face plates to retain much of their work-
hardened strength and thereby accept a particularly high surface
polish. Each box of the bracelet measures approximately 8 mm
(0.3 in.) wide. Courtesy of the World Gold Council
fusion-brazing process that has been developed
using this approach is described in the planned
companion volume Principles of Brazing (see
also Sangha, Jacobson, and Peacock 1998).
An interesting variant of this diffusion-
soldering process has been investigated whereby
the tin solder in the copper-to-copper joint is
replaced by the In-49Sn eutectic alloy (melting
at 120 C, or 248 F). The joining operations
were carried out at temperatures up to 400 C
(752 F) [Sommadossi, Gust, and Mittemeijer
2002]. In this case, two intermetallic compounds
form by reaction, but these are different from the
and phases produced in the absence of in-
dium. They are both ternary alloys. One of these
is based on the Cu10Sn3 () phase, which, in the
binary alloy system, is only stable at elevated
temperature but is stabilized at room temperature
by the addition of indium. The other phase is
designated (confusingly, here) as but is based
on the Cu2In rather than the Cu6Sn5 intermetallic.
This phase can also dissolve the third element, in
this case, tin. Below 200 C (392 C), only the
phase grows, and above this temperature, both
grow together, with the phase steadily out-
growing its sister phase. The joints made were
relatively thick (typically 50 m, or 2 mils),
which did enable signicant dilution of the in-
dium and tin in copper to occur, so as to dissolve
the intermetallic phases. However, joints made at
350 C (662 F) using 10 m (400 in.) thick
sputtered layers of In-49Sn that were etch cleaned
prior to bonding achieved a shear strength in
excess of 155 MPa (22,500 psi) and a tensile
strength of 160 MPa (23,200 psi) [Sommadossi
et al. 2002]. These relatively strong joints con-
tained only the homogeneous [Cu10(Sn, In)]
phase, which grows entirely by solid-state dif-
fusion and has a relatively small grain size.
5.9.5 Practical Aspects
There are often practical difficulties with ap-
plying the layer of ller as an electroplated or
vapor-deposited coating to the intended joints,
including the need to mask off other areas of the
surfaces of the components. It has often proved
more convenient to use a foil preform of the
relevant precious metal, typically 25 to 100 m
(1 to 4 mils) thick, that is coated with the nec-
essary thickness of the low-melting-point ller
(tin or indium). An appropriate area of the foil is
cut out and sandwiched in the joint gap. The use
of a foil of soft precious metal offers the further
merit of acting as a stress absorber, which is most
 Chapter 5: Advances in Soldering Technology / 235
useful in situations where the parts to be joined
have local topography or signicantly different
coefficients of thermal expansion. This latter as-
pect is treated in greater detail in the planned
accompanying volume devoted to brazing. Be-
cause of the higher temperatures involved in
brazing operations, thermal expansion mismatch
stresses can be a more serious problem in that
context.
5.9.6 Modeling of
Diffusion-Soldering Processes
Some attention has been devoted to the theo-
retical modeling of transient liquid-phase joining
processes, but the published studies to date have
been conned to the higher-temperature diffu-
sion-brazing process [Isaac, Dollar, and Massal-
ski 1988]. The analytical models generally as-
sume that the process kinetics are governed by
diffusion, so that the phases that solidify from the
melt at the joining temperature grow at a rate that
is proportional to the square root of the bonding
time. Clear evidence for the classical diffusion-
controlled relationship has been provided for the
copper-tin system at 300 C (572 F) [Kato,
Horikawa, and Kageyama 1999]. However, there
has been little systematic work in modeling the
kinetics of the various solid phases that grow and
subsequently are replaced by other solid phases,
or of the primary solid solution, as usually occurs
in diffusion soldering. Such an endeavor would
greatly facilitate the design and development of
this interesting joining method.
5.10 Advances in Joint
Characterization Techniques
Of the many techniques available to assess the
integrity of soldered joints, particularly those
used to attach electronic components to PCBs,
three have beneted greatly from the advent of
computer technology. These are ultrasonic in-
spection, x-ray inspection, and optical inspec-
tion.
5.10.1 Ultrasonic Inspection (Scanning
Acoustic Microscopy)
Ultrasound is dened as pressure waves with
frequencies higher than sound waves and that
cannot be heardin practice, in the range of 0.5
MHz to 5 GHz. The particular characteristic of
ultrasound that is exploited in nondestructive
technology is its ability to travel through solid
materials while obeying the same laws of re-
ection and refraction as light. Because ultra-
sound travels with xed velocity through a given
material, echoes produced by reection at dif-
ferent surfaces and interfaces will be temporally
resolved, and the corresponding distances can
then be calculated. This is the basic principle of
pulse-echo ultrasonic inspection.
In commercial instruments, ultrasound is gen-
erated by a piezoelectric transducer mounted as
a probe that is coupled to the surface of the
testpiece via a liquid or pasty coupling agent.
Two or more probes tend to be used: one to
transmit the pulse and the others to detect ech-
oes. The higher the ultrasonic frequency, the
higher is the resolution, but the stronger is the
signal attenuation due to absorption by the ma-
terials through which it travels. Ultrasonic sig-
nals, like other forms of wave energy, can re-
solve features down to approximately the size
of a single wavelength. Accordingly, ultrasound
of 10 MHz frequency should be capable of
detecting cracks down to 0.5 mm (0.02 in.) in
cross section in metal. This is clearly inad-
equate for the inspection of defects in joints that
may themselves be of comparable size or even
smaller, for example, those made to surface-
mount electronic components. To improve on
this level of discrimination, higher frequencies
must be used.
The scanning acoustic microscopy (SAM)
technique has been developed to operate in the
frequency range of 20 MHz to 2 GHz and offers
the nest level of resolution of the ultrasonic test
methods, although the depth of penetration is
limited to below 10 mm (0.4 in.).
The SAM technique is capable of nondestruc-
tively assessing the distribution of voids, cracks,
inclusions, and other hard phases over the area
of essentially parallel-sided joints [Matuasevic,
Wang, and Lee 1990; Kauppinen and Kivilahti
1991]. It involves focusing an acoustic wave,
generated from a piezoelectric transducer, via a
sapphire lens onto the specimen and scanning it
in a raster fashion. Changes in the reected acous-
tic signal from boundaries between features hav-
ing different acoustic properties are recorded and
mapped to produce the image. The correspon-
dence that can be obtained between the images
of a voided joint produced by x-radiography and
by SAM is illustrated in Fig. 4.25 and 4.26.
The interpretation of a scanning acoustic mi-
crograph of a joint can present difficulties, for
236 / Principles of Soldering

example, determining whether a certain feature movement enables defects to be viewed at an

corresponds to a crack, void, or intermetallic par- optimal angle.
ticle. Similarly, the need to mechanically scan An important innovation in x-radiography is
the probe in very close proximity to the surface the development of systems incorporating mi-
of the components restricts the geometries that crofocus sources. Small focal-spot sizes can be
can be examined. Moreover, the limited depth of achieved at the x-ray target by electromagneti-
sample from which clear images can be obtained cally focusing the incident electron beam, ex-
means that one of the components must be thin. ploiting a technique that is widely used in elec-
Modern SAM equipment overcomes all these tron microscope technology. The focal-spot size
difficulties by using a variable or multiple fre- can be as small as 1 m (40 in.) in diameter. The
quency system, often with multiple transducers, benets offered by a microfocus x-ray system
and by automating the probe positioning and im- are:
age processing. A modern SAM system is able to
discriminate between internal and external
Fine rod anode sources can be inserted into
hollow assemblies, such as the cavity of an
boundaries of components and present the user
optoelectronic package, permitting single-
with a three-dimensional image that can be elec-
wall exposures to be obtained in situations
tronically rotated and sliced similar to a com-
where conventional x-ray systems could only
puter-aided design drawing. By this advance,
provide double or multiple-wall radiographs
SAM has changed from being a laboratory di-
because of their bulky tube heads. This sim-
agnostic tool to part of the suite of in-line quality-
plies the projection geometry of the radi-
assurance methods essential to ensure low-
ography and increases the relative sensitivity
defect-rate manufacturing. Some examples of the
of the radiograph to defects with respect to
use of SAM to inspect the interior of electronic
the absorbing material of the assembly.
components in a nondestructive manner are given
by Adams [2001].
There is the possibility of obtaining geo-
metrically magnied images of high deni-
tion by distancing the camera from the
5.10.2 X-Radiography testpiece. The magnication obtainable from
an idealized point source, M, is given by the
The x-ray spectrum comprises electromag-
expression:
netic radiation of short wavelengths in the range
1016 to 1021 Hz. The high frequency and energy
of x-rays enables them to penetrate materials and t Do
reveal internal features, including defects, pro- M
Do
vided they absorb the radiation to a different
extent than does the surrounding material. It is
this characteristic that provides the contrast in This is illustrated in Fig. 5.40.
the x-ray image. Accordingly, voids, inclusions
of heterogeneous material, and cracks parallel to
the x-rays will be more conspicuous to this tech-
nique than cracks and interfaces that are per-
pendicularly oriented. Radiographs revealing
voids in joints are shown in Fig. 1.17 and 4.25.
In a modern industrial x-ray machine, a tele-
vision camera system is employed in place of
traditional lm. This enables the x-ray image of
the object to be viewed in real-time. The com-
ponent undergoing inspection is held by a free-
space manipulator. The combination of move-
ment of the testpiece coupled with real-time
viewing permits a comprehensive examination
to be made rapidly. The nature of visual percep-
tion is such that an area that differs marginally in
contrast from its surroundings is more easily de-
tected when in motion, so that defects are more Fig. 5.40 Representation of geometric magnication in mi-
readily noticed. Additionally, the controlled crofocus x-radiography
 Chapter 5: Advances in Soldering Technology / 237
Steady improvements in microfocus x-ray
tube design have resulted in geometrical mag-
nication of over 2000 being realized in
commercial systems, with the tubes operating
in the low-kilovolt range. With further digital
processing of the optical image, total magni-
cations of 7000 are now possible. This en-
ables features of the order of 250 nm (10 in.)
to be resolved, and the system is then said to
be operating in the nanofocus mode.
A diminution of the focal-spot size means
that the geometric unsharpness is correspond-
ingly reduced; that is, the precision with
which edges can be resolved is improved.
A consequential benet of spatially separat-
ing the image sensor from the testpiece is a
reduction in the scattered ray fraction gen-
erated by the testpiece itself, which contrib-
utes to the image.
The x-ray beam emerging from a microfocus
source can be proled to give a controlled
cone of radiation, again decreasing scatter
and improving the sensitivity to aws.
Finally, by simply moving the testpiece along
the axis of the x-ray beam, it is possible to
continuously zoom in on detail.
Multifocus x-ray tubes have been introduced
that enable the user to choose between micro-
focus, nanofocus, and high-power modes, using
software control and a single mouse click. The
microfocus mode is preferred for applications
where the area being examined is millimeters on
a side, such as inspection of circuit boards and
discrete modules. By contrast, the nanofocus
mode is suitable for examining ip-chip assem-
blies and wire bonds. High-power settings are
appropriate for denser samples, such as die
mounted on heat studs and generally, bulkier
items where the boosted x-ray energy provides
deeper penetration of the radiation.
5.10.3 Optical Inspection
Optical inspection of joints normally de-
scribes examination by eye, often with the aid of
a magnifying device. Usually, this will be an
optical microscope but may be stretched to in-
clude scanning electron microscopy operated in
its usual backscattered mode. Despite being a
very old technology, optical inspection methods
continue to improve at a remarkable rate. The
quality of the image that can be obtained from a
modern microscope in terms of its resolution,
depth of focus, eld area, and working distance
is vastly improved compared to even only a few
years ago. Unfortunately, these improvements
come with a price, and a standard objective lens
can easily carry a four-gure (dollars) price tag.
Notwithstanding the fundamental improvement
in lens design and manufacture, two relatively
recent innovations in optical inspection are hav-
ing an impact on soldering technology. These are
automated optical inspection (AOI) and endo-
scopes.
Automated optical inspection is the product of
a highly successful marriage between optical and
digital electronic technology. In essence, the part
to be inspected is placed under an optical mi-
croscope, and the image is electronically pro-
cessed to identify certain features, usually against
specied pass/fail criteria. The viability of this
inspection method hinges greatly on the camera
system. The camera used in an AOI system is a
digital camera with a sensor having a minimum
resolution of six million pixels. To maximize the
quality of the information acquired by each pixel,
it is exported directly to the computer in digital
format, without conversion to a video signal.
Very sophisticated analysis is now possible on
the acquired digital image. Some of the outputs
that are available from a commercial AOI system
include:
Reading of component labels and markers to
check that correct device types and values
have been placed on a PCB
Reading of serial numbers on larger compo-
nents, for quality and archiving purposes
Verication of component orientation and
measurement of alignment
Checking for solder bridges and lack of sol-
der
Measurement of contact angles of llets and
ranking of surface reectivity
Validation of wire bond patterns and isolation
between adjacent loops
Conrmation of electrical and optical cable
routing and termination
Identication of regions requiring further
manual inspection or rework
Creation of an archive of board integrity
The continued development of improved op-
tics, cameras, and more powerful software,
coupled with improved processing speeds, can
only enhance the adoption of AOI as an integral
part of electronics and photonics assembly lines.
Endoscopes are essentially miniaturized mi-
croscopes that use optical bers or other means
to transmit the image from the location of the
238 / Principles of Soldering
examination to the viewing point. They are
widely used in medicine for conducting internal
examinations in a noninvasive manner. Endo-
scopes have been developed for examining hid-
den soldered joints. Examples include intercon-
nects of ball grid arrays (BGAs), chip-scale
packages (CSPs), and ip-chip components. Cus-
tom optical heads enable side-on viewing of fea-
tures, while their small size and long working
distance permit inspection of otherwise inacces-
sible areas. Inspection of solder llets, solder
paste print proles, via-hole plating integrity, and
conformal coating uniformity are but a few ex-
amples of their application.
REFERENCES
Abbott, D.C. et al., 1991. Palladium as a Lead
Finish for Surface Mount Integrated Circuits,
IEEE Trans. Compon. Hybrids Manuf. Tech-
nol., Vol 14(No. 9), p 567572
Adams, T., 2001. Vox Pop, Inside Electron.,
March, p 1314
Baldwin, D.F., Deshmukh, R.D., and Hau,
C.S., 1996. Preparation and Properties of Gal-
lium Alloys for Use as Microelectronic Inter-
connect Materials, Int. J. Microcircuits Elec-
tron. Packag., Vol 19(No. 1), p 3745
Barranger, J., 1989. Critical Parameters of
Measurement Using the Wetting Balance,
Solder. Surf. Mt. Technol., Vol 1(No. 2), p
1113
Bartels, F. et al., 1994. Intermetallic Phase
Formation in Thin Solid-Liquid Diffusion
Couples, J. Electron. Mater., Vol 23(No. 8),
p 787790
Bernstein, L., 1966. Semiconductor Joining
by the Solid-Liquid-Interdiffusion (SLID)
Process, J. Electrochem. Soc., Vol 83(No. 12),
p 12821288
Bernstein, L. and Bartholomew, H., 1966.
Applications of Solid-Liquid Interdiffusion
(SLID) Bonding in Integrated-Circuit Fabri-
cation, Trans. Met. Miner. Soc. (AIME), Vol
236(No. 2), p 405412
Betrabet, H.S., McGee, S.M., and McKinlay,
J.K., 1991. Processing Dispersion-Strength-
ened Sn-Pb Solders to Achieve Microstruc-
tural Renement and Stability, Scr. Metall.
Mater., Vol 25, p 23232328
Bhattachatya, S.K. and Baldwin, D.F., 2000.
Gallium Alloy Breakthrough for Via-Filling
Application, Adv. Packag., Sept, p 6164
Bradley, E., Handwerker, C., and Sohn, J.E.,
2003. NEMI Report: A Single Lead-Free Al-
loy is Recommended, Surf. Mt. Technol., Vol
17(No. 1), p 2425
Brakke, K.A., 1990. The Surface Evolver,
(Version 2.17) Susquehanna University
Buckley, D., 2000. Nitrogen is the Key to
Lead-Free Wetting, Electron. Prod., Vol
29(No. 7), p 29
Burdett, P.A., Lodge, K.J., and Pedder, D.J.,
1988. Techniques for the Inspection of Flip
Chip Solder Bonded Devices, Proc. Eighth
Annual International Electronics Packaging
Conference, 710 Nov (Dallas, TX), p 4448
Chen, Z.G. et al., 2002. Study of the Micro-
structure of a Novel Lead-Free Solder Alloy
SnAgCu-RE and Its Soldered Joints, J. Elec-
tron. Mater., Vol 31(No. 10), p 11221128
Chen, Z. et al., 2003. Properties of Lead-Free
Solder SnAgCu Containing Minute Amounts
of Rare Earth, J. Electron. Mater., Vol 32(No.
4), p 235243
Chu, H.T., 1958. The Use of Amalgam as a
Filling Material in Dentistry inAncient China,
Chin. Med. J., Vol 76(No. 6), p 533555
Coombs, Jr., C.F., 1988. Printed Circuits
Handbook, 3rd ed., McGraw-Hill
Dickson G., Oglesby, P.L., and Davenport,
R., 1968. The Steady State Creep Behaviour
of Dental Amalgam, J. Res. Natl. Bur. Stand.,
Vol 72C, p 215218
Doi, K. et al., 1996. Prediction of Thermal
Fatigue Life for Encapsulated Flip-Chip In-
terconnection, Int. J. Microcircuits Electron.
Packag., Vol 19(No. 3), p 231237
European Union Directive, 2001.
COM(2001) 316 nal, On the Restriction of
the Use of Certain Hazardous Substances in
Electrical and Electronic Equipment, 2000/
0159 COD
Glazer, J., 1995. Metallurgy of Low Tem-
perature Lead-Free Solders for Electronic As-
sembly, Int. Mater. Rev., Vol 40(No. 2), p
6593
Gunter, I.A. and Jacobson, D.M., 1990. The
GEC Meniscograph Solderability Tester: Ad-
aptation to Vacuum Soldering, GEC Rev., Vol
6(No. 2), p 8689
Guo, F. and Subramanian, K.N., 2002. Sol-
ders Strengthened with Copper and Silver
Particles, Adv. Mater. Process., Vol 160(No.
12), p 4143
Harada, M. and Satoh, R., 1990. Mechanical
Characteristics of 96.5Sn/3.5Ag Solder in Mi-
crobonding, IEEE Trans. Compon. Hy-
 Chapter 5: Advances in Soldering Technology / 239
brids Manuf. Technol., Vol 13(No. 4), p 736
742
Hayashi, T., 1992. An Innovative Bonding
Technique for Optical Chips Using Solder
Bumps that Eliminate Chip Positioning Ad-
justments, IEEE Trans. Compon. Hybrids
Manuf. Technol., Vol 15(No. 2), p 225230
Hayes, D.J., Wallace, D.B., and Boldman,
M.T., 1996. Solder Jet Printing for Low Cost
Wafer Bumping, Proc. Conf. International
Society for Hybrid Microelectronics, 811
Oct (Minneapolis, MN), p 296301
Ho, C.T., 1996. Nickel and Copper Coated
Carbon Fibre Reinforced Tin-Lead Alloy
Composite, J. Mater. Sci., Vol 31, p 5781
5786
Ho, C.T. and Chung, D.D.L., 1990. Carbon
Fibre Reinforced Tin-Lead Alloy as a Low
Thermal Expansion Solder Preform, J. Mater.
Res., Vol 5(No. 6), p 12661270
Humpston, G. and Jacobson, D.M., 1990.
U.K. Patent 2,235,642; U.S. Patent 5,106,009
Humpston, G. et al., 1991. Recent Develop-
ments in Silicon/Heat-Sinks for High-Power
Device Applications, GEC Rev., Vol 7(No.
2), p 6778
Humpston, G., Jacobson, D.M., and Sangha,
S.P.S., 1993. Diffusion Soldering: A New
Low Temperature Process for Joining Carat
Gold Jewelry, Gold Bull., Vol 26(No. 3), p
90104
Hunt, C. and Wallis, D., 1995. Solderability
Standard: A Gold-Plated Nickel Reference
Material, Solder. Surf. Mt. Technol., Vol
21(No. 1), p 69
Hwang, J.S., 1996. Environment Friendly
Electronics: Lead-Free Technology, Electro-
chemical Publications
IDEALS, 1999. Improved Design Life and
Environmentally Aware Manufacturing of
Electronics Assemblies by Lead-Free Sol-
dering, BE95-1994, Commission of the Eu-
ropean Community, 30 June
Isaac, T., Dollar, M., and Massalski, T.B.,
1988. A Study of the Transient Liquid-Phase
Bonding Process Applied to a Ag-Cu-Ag
Sandwich Joint, Metall. Trans. A, Vol 19(No.
3), p 675686
ITRI, 1987. Solder Alloy DataMechani-
cal Properties of Solders and Soldered
Joints, Publication 656, International Tin
Research Institute
Jacobson, D.M. and Humpston, G., 1992. Dif-
fusion Soldering, Solder. Surf. Mt. Technol.,
Vol 10(No. 2), p 2732
Jones, W.K. et al., 1997. The At-Temperature
Mechanical Properties of Lead-Tin Based Al-
loys, Adv. Microelectron., Vol 24(No. 5), p
3034
Josell, D. et al., 2002. Misaligned Flip-Chip
Solder Joints: Prediction and Experimental
Determination of Force-Displacement
Curves, J. Electron. Packag., Vol 124, p 227
233
Kang, J.S. et al., 2002. Isothermal Solidi-
cation of Cu/Sn Diffusion Couples to Form
Thin-Solder Joints, J. Electron. Mater., Vol
31(No. 11), p 12381243
Kariya, Y., Gagg, C., and Plumbridge, W.J.,
2000. Tin-Pest in Lead-Free Solders, Solder.
Surf. Mt. Technol., Vol 13(No. 1), p 3940
Kato, H., Horikawa, S., and Kageyama, K.,
1999. Inuence of Solder Thickness on In-
terfacial Structures and Fatigue Properties of
Cu/Sn/Cu Joints, Mater. Sci. Technol., Vol
15(No. 7), p 851856
Kauppinen, P. and Kivilahti, J., 1991. Evalu-
ation of Structural Defects in Brazed Ce-
ramic-to-Metal Joints with C-Mode Scan-
ning Acoustic Microscopy, Nondestruct. Test.
Eval. Int., Vol 2494, p 187190
Khanna, P.K., Dalke, G., and Gust, W., 1999.
Morphology and Long Term Stability of Ni/
Ni Interconnections Based on Diffusion Sol-
dering, Z. Metallkd., Vol 90(No. 9), p 722
726
Klein Wassink, R.J., 1989. Soldering in Elec-
tronics, 2nd ed., Electrochemical Publica-
tions
Lalena, J.N., Dean, N., and Weiser, M.W.,
2002. Experimental Investigation of Ge-
Doped Bi-1Ag as a New Pb-Free Solder Al-
loy for Power Die Attachment, J. Electron.
Mater., Vol 31(No. 11), p 12441249
Lalena, J.N., Weiser, M.W., and Dean, N.F.,
2002. Die Attach Solder Design, Adv.
Packag., Feb, p 2530
Lau, J.H., 1996. Flip Chip Technologies,
McGraw-Hill
Lau, J.H., 2000. Low-Cost Flip Chip Tech-
nologies for DCA, WLCSP, and PBGA As-
semblies, McGraw-Hill
Lau, J.H. and Ricky Lee, S.W., 2001. Mi-
crovias for Low Cost, High Density Inter-
connects, McGraw-Hill, p 332375
Lau, J. et al., 2002. Nonlinear Analysis of
the Lead-Free Solder Sealing Ring of a Pho-
tonic Switch, Proc. Conf. Pb-Free Solder for
Electronic, Optical, and MEMS Packaging
Manufacturing, 56 Sept
240 / Principles of Soldering
Lea, C., 1990. Surfaces of Standard Solder-
ability, Solder. Surf. Mt. Technol., Vol 5(No.
6), p 4655
Lea, C., 1991. Quantitative Solderability
Measurement of Electronic Components, Part
5: Wetting Balance Instrumental Parameters
and Procedures, Solder. Surf. Mt. Technol.,
Vol 7(No. 1), p 1013
Lee, Y.C. and Basavanhally, N., 1994. Solder
Engineering for Optoelectronic Packaging, J.
Met., Vol 46(No. 6), p 4650
Lee, C.C. and Wang, C.Y., 1992. A Low Tem-
perature Bonding Process Using Deposited
Gold-Tin Composites, Thin Solid Films, Vol
208, p 202209
Lin, W. et al., 1993. Study of Soldering for
VLSI/FLC Spatial Light Modulators, Proc.
43rd Electronic Components and Technology
Conference, 14 June (Orlando, FL), p 491
497
Loomans, M.E. and Fine, M.E., 1999. Tin-
Silver-Copper Eutectic Temperature and
Composition, Metall. Mater. Trans. A, Vol
31(No. 4), p 11551162
Ma, X. and Yoshida, F., 2002. Interaction
Relation in 60Sn-40Pb-0.05La Ternary Sol-
der Alloy, Mater. Lett., Vol 56, p 441445
Mackay, C.A., 1993. Amalgams for Im-
proved Electronics Interconnection, IEEE
Micro., Vol 13(No. 2), p 4658
Marshall, J.L. et al., 1991. Composite Sol-
ders, IEEE Trans. Compon. Hybrids Manuf.
Technol., Vol 14(No. 4), p 698702
Matienzo, L.J. and Schaffer, R.R., 1991. Wet-
ting Behaviour of Eutectic Tin/Lead Solder
and Fluxes on Copper Surfaces, J. Mater.
Sci., Vol 26, p 787791
Matijasevic, G.S., Lee, C.C., and Wang, C.Y.,
1993. Au-Sn Alloy Phase Diagram and Prop-
erties Related to its Use as a Bonding
Medium, Thin Solid Films, Vol 223, p 276
287
Matuasevic, G.S., Wang, C.Y., and Lee, C.C.,
1990. Void Free Bonding of Large Silicon
Die Using Gold Tin Alloys, IEEE Trans.
Compon. Hybrids Manuf. Technol., Vol
13(No. 4), p 11281134
Mavoori, H. and Jin, S., 1998. New, Creep-
Resistant, Low Melting Point Solders with
Ultra-Fine Oxide Dispersions, J. Electron.
Mater., Vol 27(No. 11), p 12161222
Mavoori, H. and Jin, S., 2000. Dispersion
Strengthening for Dimensional Stability in
Low-Melting-Point Solders, J. Met., Vol
52(No. 6), p 3032
McCabe, R.J. and Fine, M.E., 2000. The
Creep Properties of Precipitation-Strength-
ened Tin-Based Alloys, J. Met., Vol 52(No.
6), p 3335
McCormack, M. and Jin, S., 1994. Improved
Mechanical Properties in New Pb-Free Sol-
der Alloys, J. Electron. Mater., Vol 23(No. 8),
p 715720
McCormack, M., Jin, S., and Kammlott,
G.W., 1994. Enhanced Solder Alloy Perfor-
mance by Magnetic Dispersions, IEEE Trans.
Compon. Packag. Manuf. Technol., Part A,
Vol 17(No. 3), p 452457
McGroarty, J. et al., 1993. Statistics of Solder
Joint Alignment for Optoelectronic Compo-
nents, IEEE Trans. Compon. Hybrids Manuf.
Technol., Vol 16 (No. 5), p 527529
Miller, L.F., 1969. Controlled Collapse Re-
ow Chip Joining, IBM J. Res. Dev., Vol
13(No. 3), p 239250
Morrow, J., 1964. Cyclic Plastic Strain En-
ergy and Fatigue of Metals, STP 378, Ameri-
can Society for Testing and Materials, p 45
NEPCON West, 1992. Composite Solders,
Technical Session 34, Proc. National Elec-
tronic Packaging and Production Confer-
ence, 2327 Feb (Anaheim, CA), p 1245
1248
Normandeau, G., 1996. Cadmium-Free Gold
Solder Alloy, Gold Technol., Vol 18(No. 4),
p 2024
Norme Francaise, 1987. Electronic Compo-
nents: General Alloys, Fluxes and Creams
for Soft Soldering. Test Method 1, NF C90-
551, LUnion Technique de lElectricit,
Paris
Plumbridge, W.J., 1996. The Mechanical Be-
haviour of Solders, Solder. Surf. Mt. Tech-
nol., Vol 8(No. 1), p 2731
Priest, J. et al., 1994. Liquid Metal-Jetting
Technology: Application Issues for Hybrid
Technology, Int. J. Microcirc. Electron.
Packag., Vol 17(No. 3), p 10631674
Ramirez, A.G., Mavoori, H., and Jin, S., 2002.
Bonding Nature of Rare-Earth-Containing
Lead-Free Solders, Appl. Phys. Lett., Vol
80(No. 3), p 398400
Rapson, W.S. and Groenewald, T., 1978. Gold
Usage, Academic Press
Reusch, B., Geis-Gerstorfer, J., and Ziegler,
C., 1988. Scanning Probe Microscopy of New
Dental Alloys, Appl. Phys. A, Vol 66(No. 7),
p S805S808
Sangha, S.P.S., Jacobson, D.M., and Pea-
cock, A.T., 1998. Development of the Cop-
 Chapter 5: Advances in Soldering Technology / 241
per-Tin Diffusion Brazing Process, Weld. J.
Res. Suppl., Vol 77(No. 10), p 432s438s
Shimizu, T. et al., 1999. Zn-Al-Mg-Ga Al-
loys as Pb-Free Solder for Die-Attaching Use,
J. Electron. Mater., Vol 28(No. 11), p 1172
1175
Smith, E.B. and Swanger, L.K., 1999. Are
Lead-Free Solders Really Environmentally
Friendly?, Surf. Mt. Technol., Vol 12(No. 3),
p 6466
Sommadossi, S. et al., 2000. Development of
Cu/Cu Interconnections Using an Indium In-
terlayer, Proc. Conf. EuroMat 2000, 2730
Sept (Munich), p 16
Sommadossi, S. et al., 2002. Mechanical
Properties of Cu/In-48Sn/Cu Diffusion-
Soldered Joints, Z. Metallkd., Vol 93, p 496
501
Sommadossi, S., Gust, W., and Mittemeijer,
E.J., 2002. Characterization of the Reaction
Process in Diffusion-Soldered Cu/In-48 at.%
Sn/Cu Joints, Mater. Chem. Phys., Vol 77, p
924929
Steen, H.A.H. and Becker, G., 1986. Effect of
Impurity Elements on the Soldering Proper-
ties of Eutectic and Near-Eutectic Tin-Lead
Solder, Brazing Soldering, Vol 11(No. 4), p
411
Sturcken, K., 2000. Column Grid Array, High
Reliability Option for Packaging, Adv.
Packag., Aug, p 4550
Subramanian, K.N., Bieler, T.R., and Lucas,
J.P., 1999. Microstructural Engineering of
Solders, J. Electron. Mater., Vol 28(No. 11),
p 11761183
Tan, Q. and Lee, Y.C., 1996. Soldering Tech-
nology for Opto-Electronic Packaging, Proc.
46th Electronic Components and Technology
Conference, 2831 May (Orlando, FL), p 26
36
The European Commission, 2000. Proposal
for a Directive of the European Parliament
and of the Council on Waste Electrical
and Electronic Equipment and on the Re-
striction of the Use of Certain Hazardous Sub-
stances in Electrical and Electronic Equip-
ment, COM (2000) 347 nal - 2000/0159/
COD
The European Commission, 2001. Restric-
tions on the Use of Hazardous Substances,
Document 501PC316, 6 June, Legislation in
preparation, Commission proposals
Thwaites, C.J., 1981. Solderability and Some
Factors Affecting It, Brazing Soldering, Vol
1(No. 3), p 1518
Tribula, D. and Morris, Jr., J.W., 1990. Creep
in Shear of Experimental Solder Joints,
J. Electron. Packag., Vol 112(No. 2), p 8793
Tsunetsagu H. et al., 1997. Accurate, Stable,
High Speed Interconnections Using 2030
m Diameter Microsolder Bumps, IEEE
Trans. Compon. Packag. Manuf. Technol.,
Vol 20(No. 1), p 7682
Tuah-Poku, I., Dollar, M., and Massalski,
T.B., 1988. A Study of the Transient Liquid
Phase Bonding Process Applied to a Ag/Cu/
Ag Sandwich Joint, Metall. Trans. A, Vol
19(No. 3), p 675686
U.S. Senate Bill S391, 1990. Lead Exposure
Reduction Act, May
U.S. Senate Bill S729, 1993. Lead Exposure
Reduction Act, April
Vaynman, S. and Fine, M.E., 2000. Flux De-
velopment for Lead-Free Solders Containing
Zinc, J. Electron. Mater., Vol 29(No. 10), p
11601163
Vaynman, S. and McKeown, S., 1993. En-
ergy-Based Methodology for the Fatigue Life
Prediction of Solder Materials, IEEE Trans.
Compon. Hybrids Manuf. Technol., Vol
16(No. 3), p 317322
Vincent, J.H. and Humpston, G., 1994. Lead-
Free Solders for Electronic Assembly, GEC
J. Res., Vol 11(No. 2), p 7689
Wade, N. et al., 1999. Effect of Microalloying
on the Creep Strength and Microstructure of
a Eutectic Sn-Pb Solder Alloy, J. Electron.
Mater., Vol 28(No. 11), p 12861289
Wang, T.B. et al., 2000. Die Bonding with
Au/In Isothermal Solidication Technique, J.
Electron. Mater., Vol 29(No. 40), p 443447
Wang, L., Yu, D.Q., and Huang, M.L., 2002.
Improvement of Wettability and Tensile Prop-
erty in Sn-Ag-RE Lead-Free Solder Alloy,
Mater. Lett., Vol 56, p 10391042
Waterstrat, R.M., 1990. Brushing Up on the
History of Intermetallics in Dentistry, J. Met.,
Vol 42(No. 3), p 814
Wilde, J. and Pchalek, N., 1993. Kontak-
tierung von Solarzellen durch Isotherme Er-
starrung, Verbindungstech Elektron., Vol
5(No. 4), p 172179 (in German)
Wooldridge, J.R., 1988. Lessons Learned dur-
ing a Year of Production Solderability Test-
ing with a Wetting Balance, Brazing Solder-
ing, Vol 15(No. 4), p 2427
Wu, C.M.L. et al., 2002. Improvements of
Microstructure, Wettability, Tensile and
Creep Strength of Eutectic Sn-Ag Alloy by
242 / Principles of Soldering
Doping with Rare-Earth Elements, J. Mater.
Res., Vol 17(No. 12), p 31463154; The Prop-
erties of Sn-9Zn Lead-Free Solder
Alloys Doped with Trace Rare Earth Ele-
ments, J. Electron. Mater., Vol 31(No. 9),
p 921927; Microstructure and Mechanical
Properties of New Lead-Free Sn-Cu-RE Sol-
der Alloys, p 928932; The Wettability and
Microstructure of Sn-Zn-RE Alloys, J. Elec-
tron. Mater., Vol 2(No. 2), p 6369
Xia, Z. et al., 2002. Effect of Rare Earth El-
ement Additions on the Microstructure and
Mechanical Properties of Tin-Silver-Bis-
muth Solder, J. Electron. Mater., Vol 31(No.
6), p 564567
Yamada, H. et al., 1998. Advanced Copper
Column Based Solder Bump for Flip-Chip
Interconnection, Int. J. Microelectron.
Packag. Soc., Vol. 21(No. 1), p 417422
Yang, X.F. and Xi, X.M., 1995. Critical Wet-
ting Angle for Spontaneous Liquid Inltra-
tion into Orderly Packed Fibres and Spheres,
J. Mater. Sci., Vol 30, p 50995102
Yost, F.G., Hosking, F.M., and Frear, D.R.,
1993. Mechanics of Solder Alloy Wetting and
Spreading, Van Nostrand Reinhold
Ying, Z., Hongyuan, F., and Yiyu, Q., 1994. A
Study of Sn-Pb-RE Solder, Proc. Electronic
Packaging Materials Science Conf. VII, 29
Nov to 3 Dec (Boston, MA), p 137143
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)

Index

A Aluminum fluxes, 121

Aluminum-gallium-magnesium-zinc solders, specific
Abbreviations and symbols, 243 types
Abietic acid, 116 Al-3Ga-3Mg-90Zn, lead-free solders, 66
Acceptance criteria, 231 Al-3Ga-3Mg-90Zn, zinc-base solders as, 66
Acetic acid vapors, 132(F) Aluminum-germanium eutectic alloy, 8
for Pb-62Sn solders, 131 Aluminum nitride, 152
as reactive atmospheres for fluxless soldering, 130 Aluminum quaternary alloy, 83
toxicity of, 131 Aluminum-silicon alloys, 6
Acetylene, 34 Aluminum-silicon alloys, specific type
Acoustic microscopy, 204 Al-12Si, additions for wetting and spreading, 134
Activated filler metals, 151 Al-12Si, diffusion soldering processes with, 232
Activated solders, 8 Al-70Si, alumina matched thermal expansion, 164
Activation energy, 8, 24, 124 Aluminum-silicon carbide composites, 152, 175
Activation temperature of fluxes, 117 Aluminum-zinc phase diagrams, 66(F)
Active brazes, 103 Aluminum-zinc solders, specific types
Active filler metals Al-94Zn, with aluminum, 54
for ceramics, 149 Al-94Zn, zinc alloys, 54
contact angle in, 17 Al-94Zn, zinc-bearing solders, 62
high temperatures of, 149 Amalgams
Active hydride process, 151 advantages of, 214
Active solders, 147, 152153 based on gallium, 216217
Additives, 135 based on indium, 217
Adhesive bonding, 23 based on mercury, 215216
Adhesively bonded joints, 3, 173 defined, 214
Adhesives, 184 dental, stress-strain curve for, 215(F)
Aesthetic requirements, 50 as solders, 214217
Airborne particle size, 179(F) solid and liquid metals evaluated for, 214(T)
Alcohol solvent carrier, 118 Ammonia fluxes and intergranular corrosion with
Allotropic transformation, 77(F), 195196 brasses, 114
Alloy constitution, defined, 78 Anisotropically conductive adhesives, 3
Alloy Constitution (reference text), 78 ANSI/J-STD-002 (test method), 207
Alloy J, 60, 61 Antimony
Alternative atmospheres for oxide reduction, 111 added to lead-tin solders, 57
Alumina, 121 as additive to indium, 135
alloy matching thermal expansion, 164 effects of, on surface tension of tin, 193
and gallium, 62 as impurity, 76
nonmetallic bonding to, 152 Pb-Sb-Sn ternary system, 57
Aluminum and aluminum alloys as solder constituent, 5354
Al-94Zn solders with, 54 and solid-solution strengthening, 5354
brazes for, 6 Antimony-tin phase diagrams, 60(F)
corrosion of, 54 Application methods, 148
diffusion bonding of, 10 Argon, 36, 109
diffusion soldering for, 137 Arrhenius-type rate relationships, 24
fluxes for, 121122 Ashby materials selection chart, 159(F)
heat capacity of, 62 Asthma, 43
as impurity, 76, 77 Atmospheres. See also inert atmospheres
oxides on, 9 categories of, for joining, 103
and temperature uniformity, 50 chemically active, described, 104
thermal conductivity of, 62 chemically inert, described, 104
thermal expansivity of, 62 controlled gas, 104
thermal heat capacity of, 50 effect of, on spreading, 20
zinc alloys with, 54 and fluxes, 8
zinc-bearing solders for, 61 and heating method, 33
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
246 / Principles of Soldering

Atmospheres (continued) Bismuth-lead-tin ternary system, 57

joining, interrelationship of, 104(F) Bismuth-silver-germanium alloys, specific type
nonoxidizing, and heat treatment for wetting, 148 Bi-11Ag-0.05Ge ternary, 197198
and oxide film reduction, 105106 Bismuth-tin lead-free solders, 122
reactive gas, and oxide reduction, 130131 Bismuth-tin phase diagrams, 56(F)
reduced pressure, and zinc bearing-solders, 62, 65 Bismuth-tin solders, specific types
soldering, costs and benefits of, 116(T) Bi-43Sn, 31
soldering, thermal conductivities of, 114(T) Bi-43Sn, elongation, 194
types of, 104 Bi-43Sn, for hermetic joints, 173
Atmospheric corrosion, 37 Bi-43Sn, melting point of, 173
Atmospheric quality, 136(F) Bi-43Sn, physical properties of, 194
Atomic diffusion, 224 Boiling/sublimation temperatures, 107
Atomic fraction of constituents, 96 Boiling, temperatures of, 109(T)
Atomic hydrogen Bolometers (thermal imaging), 35
creation of, 127 Boltzmann constant, 24, 78, 98
reduction of silver oxide by, 130(F) Bond formation in pressure welding, 9
solder oxides reduction by, 127128 Bond quality, 212
Atomically clean, 50 Bond wire, 175
ATV 2003, 127 Bond wire interconnections, 202
Automated optical inspection (AOI), 237 Bonding process
at low temperatures, 214
stages of, 9
B Bonding temperature, peak, 38
Boron nitride, 152
Bakeout temperature, 36 Bow distortion
Ball grid array integrated circuits (BGAICs), 179 of bimetallic assembly, 165
Ball grid arrays (BGAs), 199, 204 of bimetallic strip, 158(F)
Balling up, 6263 equation for, 157
Balls (of lead-tin solders), 129(F) Brasses
Barrier coatings, 133, 149 and ammonia, 105
Barrier metal, 200 intergranular corrosion and ammonia flux, 114
Barrier metallizations, 155 vacuum atmospheres, and zinc metallization, 107
Basic spreading test, 211 Braze alloy families
Berthoud equation, 25 and melting ranges, 7(F)
Beryllia, 152 temperature ranges of, 6
Beryllia dust, 163 Braze alloy, specific type
Beryllium, 104, 216 Al-4Cu-10Si, 6
Bimetallic expansion, 50 Brazes and brazing
Bimetallic strip, 158(F) filler metal temperatures in, 5
Binary alloys and phase diagrams, 79 service temperature of, 8
Binary compounds, 84 British Standards (BS) 9430 (void free joints), 183
Bismuth Brittle failure, 28
added to lead-tin solders, 57 Brittle intermetallic layers, 28
as additive to indium, 135 Brittle joints, 52. See also embrittlement
Bi-Pb-Sn ternary system, 57 Brittle materials, 28
effects of addition to, on liquidus and solidus Brittle phases, 49
temperatures of silver-tin off-eutectic alloys, 193(F) Bulk filter mechanical properties, 29
effects of addition to, on tensile strength of silver-tin Bulk properties of solders, 225226
solders, 193(F) Bulletin of Alloy Phase Diagrams (phase diagrams), 79
effects of, on surface tension of tin, 193 Butt joints, 177, 177(F)
expansion on freezing, 53 Butt welding, 9
in hermetic soldered points, 53
as impurity, 76, 77
as solder constituent, 53 C
Bismuth-antimony-tin system, 96
Bismuth-bearing solders C-charts, 42
dissolution of chromium metallizations to, 149 C4 process, 199
inferior fluidity of, 53 Cadmium and cadmium alloys
inorganic fluxes for, 53 health hazard of, 63
Bismuth-lead-tin ternary eutectic solders, 93 restrictions in use of, 50, 51
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 247

toxicity of, 51, 191 Cleaning

Cadmium-base solders alternatives to, 41
melting point depression, 93 benefits and costs of, 41
melting points for, 93 CFC-free, 40
Cadmium-indium-tin-zinc quaternary eutectic system, chemical, 37
93 costs of, 118
Cadmium-indium-tin-zinc quaternary system, 94(T) electronic assemblies, 40
Cadmium solders as substitutes for mercury, 93 measure of effectiveness of, 119120
Calibration standards, 212214 mechanical, 37
Capillary action, 23 methods of, 40
Capillary dams, 23 postjoining, 3941
Capillary flow by reverse-gas bias mode, 150
in narrow gaps, 15 Cleaning agents
time for molten tin and copper in, 27(F) chlorides as, 111
Capillary forces, 15, 15(F), 212 fluorides as, 111
Carbon-carbon fiber composites, 152, 163 relative effectiveness of, 41(F)
Carbon dioxide, 109 Cleaning treatments, 37
Carbon-fiber-loaded solders, 223 Cleanliness of IEC board test coupons, 119(F)
Carbon fibers, 222 Coatings
Carbon monoxide gold, and embrittlement, 133
as reactive atmospheres for fluxless soldering, 130 gold, shelf life of, 133, 148(T)
uses of as reducing gases, 109 non-metallic removal of, 114
Carboxylic acids, 116 onto component surfaces, 37
Cast iron, 147 solderability shelf life of, 133
Cathodic sputtering, 180 solderable, 149152
CE7 alloy, 163 soluble, 149
Ceramics storage shelf life of, 50
active filler metals for, 149 thickness of, 148
soldering to, 149 thin, by autocatalytic method, 180
wide-gap joints in, 158 types of, 149
zinc oxide, metallization of, 150 vapor deposition, 234
Cerium, 227 Cobalt metallization, 71
as additive to indium, 135 Coefficient of thermal expansion (CTE)
CFC-free cleaning operations, 40 of carbon fibers, 222
of components, 26
CFCs. See chlorofluorocarbons
of copper-molybdenum alloys, 161
Chadwick peel tests, 73
of copper-tungsten alloys, 161
Channeling, 22
of iron-nickel alloys, 160
Chemical cleaning, 37, 209
and melting point for metals, 160(F)
Chemical displacement, 180 of metals, and their melting point, 159
Chemical fluxes. See fluxes of molybdenum, 160
Chemical properties of lead-free solders, 193196 of non-nickel alloys, 162(F)
Chemical reduction of metal oxides, 105 of Osprey controlled expansion alloys, 163(F)
Chemical vapor deposition (CVD) of titanium, 160
metallization techniques, 180 of tungsten, 160
on nonmetallic components, 149150 Coelectroplating, 32
Chemically active atmospheres, 104 Cold compression welding, 200
Chemically inert atmospheres, 104 Compatibility, characteristics of, for solder, 49
Chip-scale packages (CSPs), 199 Compliant structures
Chloride-based fluxes, 121 accommodating thermal expansivity difference,
Chlorides, 111 165167, 166(F)
Chlorofluorocarbons (CFCs), 39, 111, 115, 117, 118, economics of, 167
196 for mitigation of mismatch expansivity, 166(F)
Chopped carbon fibers, 222 Component cleaning methods, 172
Chrome oxide, 110 Component surfaces
Chromium, 150 dissolution of, by brazing, 8
Chromium metallizations, 149 dissolution of, by soldering, 8
Clausius-Clapeyron equation, 97, 98 Component testing, 224
Clausius theorem, 138 Composite materials
Clean room class designation, 179(F) controlled expansion materials, 163164
 2004 ASM International. All Rights Reserved.
Principles of Soldering (#06244G)
248 / Principles of Soldering
Composite materials (continued)
families of, 163
Composite solders
high and low-cycle fatigue life of, 221(T)
intermetallic compounds in, 219
iron containing, 222
with large particles, 220
preparation methods of, 220
of sessile drop tests, 220
tensile strength of, 221(T)
types of, 220
wetting and spreading characteristics of, 220
wetting properties of, 221(T)
wide joints and, 221
yield strength of, 220(F)
Compositional equilibrium, 78
Compound formation predictions, 79
Compound semiconductors, 105
Compression bump bonding, 202(T)
Compressive loading
fluxless soldering by, 135(F)
and joint shear strength, 136(F)
Concentration of solid metal in liquid metal, 25(F)
Conductive adhesives, 3(T)
Conductive polymeric materials, 201
Conductivity, 30
Constraints
imposed on by components and solutions, 168179
joint area, 169173
Contact angle
of Ag-96.5Sn solders, 228(F)
of copper-silicon brazes, 16(F)
effect of on fillet formation joint filling, 17(F)
effective, 22
and fillet formation, 17
in lap joint, 176(F)
on lead-tin solders, 18(F)
of lead-tin solders, 22(F)
measurement of, 211212
metallurgical modification of, 133
of Pb-60Sn solder, 20
and quality of wetting, 17
rare earth doping of indium solders, 228
in reactive wetting, 16
and spread factor, 4445
and spread factor and spread ratio, relationship
between, 212(F)
and spread ratio, 4445
temperature effects of, 25
time dependence of, 15
and wetting area, 1314
Contact angle (of droplets), 45
Control charts, 42
Controlled expansion materials
alloys, 164(F)
components of, 163
composite materials, 163164
copper-molybdenum alloys, 160161
copper-surface laminates, 162163
copper-tungsten alloys, 161162
www.asminternational.org
interlayers, 164165
iron-nickel alloys, 160161
mechanical constraints and solutions, 159164
Cooling rate, 38
Cooling stages, 39, 39(F)
Copper
diffusion soldering of, 234
effects of, on surface tension of tin, 193
lead-free tin-base solders for, 116
lead-tin solders for, 116
nonmetallic bonding to, 152
Pb-63Sn solder wetting using rosin flux, 115(F)
rate of dissolution of, in molten Pb-60Sn solder, 84
as wettable metallizations, 147
wetting of, by Pb-63Sn solder, 115(F)
Copper abiet, 116
Copper-base alloys, 10
Copper-base brazes, 6
Copper coupons, 131(F)
Copper-indium alloy systems, 234
Copper-indium intermetallic compounds, specific types
Cu2In, for diffusion soldering, 234
Copper-Invar-copper laminates, 163
Copper-lead-tin phase diagrams, 8485
Copper-lead-tin system
diagram sector of, 86(F)
isothermal section of, 86(F)
liquidus surface of, 85(F)
Copper metallizations, 132(F)
Copper-molybdenum alloys
coefficient of thermal expansion (CTE) of, 161
as composite materials, 163
controlled expansion materials, 160161
Copper-molybdenum-copper laminates, 163
Copper-nickel-tin phase diagrams, 84
Copper particle reinforcement, 220
Copper powder, 216
Copper-surface laminates, 162163
Copper-tin alloy systems, 234
Copper-tin intermetallic compounds
effect of thickness on, 88(F)
growth of, 87(F), 88(F)
interfacial, rate of formation of, 87
from lead-tin solders, 84
presence of thick layers of, 88
properties of, 87
Copper-tin intermetallic compounds, specific types
Cu10Sn, for diffusion soldering, 234
Cu3Sn, binary compounds, 87
Cu3Sn, precipitates at copper/solder interface, 89
Cu6Sn5, as stoichiometric compounds, 91
Cu6Sn5, binary compounds, 84, 85
Copper-tin phase diagrams, 85(F)
Copper-tin solders, specific types
Cu6Sn reinforced, 220
Copper-to-aluminum direct bonding, 162
Copper-to-copper joint, 234
Copper-tungsten alloys
coefficient of thermal expansion (CTE) of, 161
as composite materials, 163
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 249

controlled expansion materials, 161162 Defect-recognition software, 189

Copper-zinc compounds, 64 Defective items, 42
Cored silver, 80 Defects. See also tests/testing; voids
Corrosion distortion, 38, 50, 168170
atmospheric, 37 dross formation, 115, 115(F)
in joints, 49 inspection of, 235
mechanisms of, 29 joint embrittlement, 52, 90, 153
of Sn-9Zn eutectic alloy, 96 levels of, and oxide thickness effect of, 124(F)
Corrosion resistance of lead-free solders, 195 in liquid solder film, 134
Costs. See also economics rates of, in joints, 42
of 3.6Ag-1.6Au-92.8Sn eutectic solders, 96 and x-ray inspection, 204
of cleaning, 41, 118 Delay effect on joint strength, from cleaning to
of forming gas, 111 assembly, 131(F)
of furnace joining, 104 Dendrites
of gold-germanium vs. gold-silicon solders, 70 arm spacing and tensile strength, 32(F)
of lead-free solders, 192 bridging, 120
of low-alpha lead, 190191 on circuit board, 120(F)
of machining molybdenum, 160 growth of, 119120
of machining tungsten, 160 primary, of silver, 80
of nitrogen, 109 Dendritic growth, 120(F)
of precious metals in metallizations, 231 Dental amalgams, 215, 215(F)
of solder coated substrates, 134 Depression of melting point, 9396
of solder elements, 52 Dermatitis, 43
Costs and benefits of soldering atmospheres, 118(T) Design criteria of soldering processes, 2830
Crack repair, 179 Design guidelines for various metals for in-plane
Cracked ammonia, 109 alignment, 205
Cracks Diamond
from low-cycle-fatigue, 74 as intermediate plate material, 165
orientation of, and x-rays, 236 nonmetallic bonding to, 152
from volume contraction, 67 Die attach of gold-metallized chips, 71
Creep Differential scanning calorimeter, 155
and heat treating, 158 Differential thermal expansion
of indium-base solders, 73 steps to reduce stress from, 158
to relieve mechanical stress, 178 stress from, 157
resistance of metals to, 195 Diffuse heating, 33
and strain rate measurement, 194 Diffusion
thermal cycling without, 165 activation energy for, 78
types of, 195 rate of, 78
Creep behavior of indium solder alloys, 158 surface, 10
Creep curve for lead-tin eutectic solders, 218(F) volume, 10
Creep properties of composite solders, 221(T) Diffusion bonding
Creep rates of various solders, 219(T) of aluminum-base alloys, 10
Creep resistance of copper-base alloys, 10
effect of metal additions on, 220 of gold, 45, 43
of lead-tin eutectic solders, 219 of gold, temperature/pressure curve for, 43(F)
of solder, 217 of indium, 45
Creep rupture of indium, temperature/pressure curve for, 44(F)
of indium-base solders, 73 interlayers for dissimilar metals, 10
and intermetallic thickness, 88(F) limitations of, 230
Creep stress, 10 process, 9
Critical angle, 223(T) with standard solders, 43
Critical temperatures, 107 of tantalum, 10
Curie point, 160 of titanium, 10
Curing curves, 217(F) of titanium alloys, 10
of tungsten, 10
of various metals, 10
D Diffusion brazing, 10, 175
Diffusion rates
De Gennes model, 20, 22 inequality of, 10
Decomposition reaction, 180 in solids, 83
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
250 / Principles of Soldering

Diffusion reaction, 215 of silver with lead-tin solders, 53(F)

Diffusion-soldered joints Dissolution rate constant, 24
micrographs of, 233(F) Distortion
steps involving, 232(F) of assemblies, causes of, 168170
strength of, 232 from bimetallic expansion, 50
Diffusion soldering, 7 during heating, 38
advantages of, 230 from thermal expansion mismatch strain, 50
for aluminum, 137 Doping, 227229
binary combinations for, 231 Doping additions, 135(F)
of copper, 234 Drop-in replacement, 191
copper-indium alloy systems for, 234 Dross formation
copper-tin alloy systems for, 234 interference of, with wetting and spreading, 115
Cu10Sn intermetallic phase for, 234 oxygen concentration and rate of, 115(F)
described, 230 Ductile foils, 155
of gold, 233234 Ductility
interlayers in, 10 improvement of, in filler metals, 155
modeling of, 235 of indium-base solders, 73
practical aspects of, 234235 of indium-bearing solders, 52
process principles of, 230231 of silver-tin intermetallic phases, 52
of silver, 231232 of tin-based solders, 52
Diffusion-soldering processes of zinc-based alloys, 64
with Al-12Si alloy, 232 Duration of diffusion bonding process, 9
basis of, 80 Dwell stages, 38
for gold jewelry, 233
jewelry made by, 234(F)
with silver-indium alloys, 232 E
Diffusion soldering systems, 231(T)
Dimensional stability of soldered joints, 224226 Economics. See also costs
DIN WNr 1.3981 (iron-nickel alloys), 160 of compliant structures, 167
Dip-and-look (DNL) test, 207 of fluxless soldering, 123
Dipping methods, 148 of forming gas, 111
Direct-bonded copper, 162 of joining process, 145
Direct chip attach (DCA), 199 of lead-free solders, 192
Disc preforms, 170 of low oxygen atmospheres, 115
Dislocation climb, 215 of thin foil preforms, 136
Dispersion-stabilized alloys, 175
of vapor-phase techniques, 171
Dispersion-strengthened solders, 219(T)
Economics and availability of lead-free solders,
Dispersion strengthening, 218, 219
191193
Dispersoids, 218, 219, 221
Edge fillets examination, 168
Disproportion reaction, 180
Effective contact angle, 22
Dissimilar materials and thermal cycling, joints with,
73 EIA/IS-86 (test method), 207
Dissimilar-metal joints, 9 Elastic modulus, 194
Dissimilar metals, 10 Electric conductance of mechanical fasteners, 12
Dissolution Electrical conductivity
of chromium metallizations to bismuth-containing of adhesive joints, 3
solders, 149 of composite solders, 221
of component surfaces, 8 joint requirements, 30
of parent materials, 153 Electrical properties of joints, 50
rate of, 2425 Electrically conductive adhesives, 3
rates of, 153 Electrode potential of selected elements, 54(F)
Dissolution of parent materials and intermetallic Electrodeposition, 155
growth, 2425 Electronic assemblies
Dissolution rate cleaning, 40
of copper in molten Pb-60Sn solders, 84, 89 joining methods for, 116
of gold in molten lead-tin solder, 89 Electroplating
of metals and metallizations in lead-tin solders, 87(F) difficulties with, 234
of platinum, in gold-tin eutectic solders, 71 as manner of solder deposition, 31
of platinum, in tin-base solders, 148(T) in metallizations, 37
and saturation limit, 25 tin-lead solder, 32
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 251

Ellingham diagram spreading of, 81

adjustments for plasma state, 127 theoretical modeling of, 9799
application of, 107111 used for solders and brazes, 6
free energy change for oxidation of several metals, Eutectic solders
106(F) composition of, 8
for oxides with hydrogen at a partial pressure, 129(F) melting temperatures of, 52
for selected oxides, 108(F) Eutectiferous character of alloys, 51
temperature and O/H ratio for metal oxide reduction, Evolved vapor, 172
110(F) Exchange reactions, 122
Elongation, 194 Expansion coefficients. See coefficient of thermal
Elongation to failure expansion (CTE)
of Ag-95Sn solders, 228(F) Expansion mismatch
of Sn-9Zn solders, 228(F) compliant structures for mitigation of, 166(F)
Embrittlement and cooling stages, 39(F)
caused by impurities, 49 steps to reduce stress, 158
of copper-zinc compounds, 64 Expansion on freezing, 53
and gold-based coatings, 133 Explosion risk, 110
liquid metal, 30 Explosive limit, 111
of solder, 23 Exposure limits for hazardous dusts and vapors, 43
tin-based solders and AuSn4, 133 External magnetic fields of iron powder solders, 221
Embrittling phases, 4 Extrapolation of results in spreading test, 212
Endoscopes, 237 Eye and nose irritation, 43
Engineering ceramics thermal expansion, 159
Enthalpy of fusion, 97
Entropy F
changes in, 9899
in chemical bond, 105 Fast atom bombardment, 128, 180
defined, 137, 138, 139 Fast atom cleaning, 131(F)
of fusion, 97 Fatigue cracks
Environmental concerns, 191 internal initiation of, 217
Environmental considerations. See also health and wide-gap joints and, 217
safety issues Fatigue failure, 157
chlorofluorocarbons (CFCs), 196 Fatigue fracture, 165
volatile organic compounds (VOCs), 115 Fatigue life, 200
Environmental durability, 29 of carbon-fiber-loaded solders, 223
Environmentally friendly chemicals as cleaning agents, Fatigue resistance
111 of Alloy J, 61
Equilibrium constant and oxidizing reaction, 141 high-cycle, 220
Equilibrium contact angle, 1718 indium solder alloys and, 158
Equilibrium partial pressure of oxygen. See oxygen low-cycle, 220
partial pressure ranks of lead-free solders, 195
Equilibrium phase diagrams, 25 of solder, 217
Erosion for soldered joints, 4
changes to rate of, in parent materials, 154 Fatigue situations, joint reliability in, 205
conditions for, 153 Fatigue theory, 226
of gold, by molten indium, 81 Ferromagnetic alloys, 160
of gold by molten tin, 82(F) Filler metal temperatures in brazing and soldering, 5
of parent materials, 49 Filler metals
of silver by molten tin, 82(F) and dissolution with component pieces, 26
of substrate surfaces, 33 form of, 3133
Erosion of parent materials, 153154 with limited component solubility, 2526
European Space Agency (ESA) Specifications (void molten, flow influences to, 12
free joints), 183 molten, solidification shrinkage of, 169
Eutectic alloys and alloying partitioned, 155(T)
benefits of, 81 partitioning of, 155157
grain refinement of, 81 spread characteristics of, 12
melting point depression by, 9396 spreading of, 20
melting point of, 81 surface area to volume, ratio of, 133
vs. noneutectic alloys, 20 for welding, 4
spreading characteristics of, 19 Filler spreading characteristics, 1922
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
252 / Principles of Soldering

Fillet formation for aesthetic requirements, 50 Flux action mechanisms, 114

Fillet radius, 176(F) Flux activators, 118
Fillets Flux boil, 123
beneficial effects of, 167 Flux carriers, 117
enhancements of, 167 Flux chemistry, 116
with frosty appearance, 81 Flux-cored solder wire, 31, 114
inspection of, 238 Flux-formulations, 114
integrity of, 168 Flux removal, 39
quality of, 29 Flux residues, 123
to reduce stress concentrations, 176 Flux vapors, 2526
role of, 167168 Fluxes
and stress concentration reduction, 168(F) activation temperature of, 117
and stress concentrations, 167 activity of, 116117
as stress reducers, 4 for aluminum, 121122
Finishes, 149 and atmospheres, 8
Finite element analysis (FEA) based on organic compounds, 66
of ceramic-metal brazed joint, 225(F), 226(T) chemical activity of, 118
modeling assumptions in, 225 chemical, described, 111112
Fire risks, 43 chloride-based fluxes for, 121122
First Law of Thermodynamics, 137138 classification of, 118(T)
Flammability, 43 commercial designations of, 117
Flip-chip assembly operation, 123 containing 2-ethylhexanoate, 195
Flip-chip bonding, 200(F) with gold-tin eutectic solders, 122
Flip-chip bonding process with high-lead solders, 122
defined, 201 high-molecular-weight hydrocarbons as, 122
described, 166 high-temperature, 122123
Pb-5Sn solders for, 166 IA type, 117
Flip-chip components pull-in alignment, 132(F) ingredients in, 116
Flip-chip daisy chain, 74(F) for lead-tin eutectic solders, 122
Flip-chip inspection methods, 204 for magnesium, 122
Flip-chip interconnection no-clean, 39, 40, 118119
high-temperature, die mounted using, 206(F) OA type, 117
self-alignment of, 202(F) organic, 121
self-alignment of solder bumps during, 127 with oxide scale, 34
thermal expansion mismatch in, 200 R, RMA, RA type, 117
Flip-chip interconnects, 179 requirements for, 113114
Flip-chip joining process, 167(F) role of, in wetting and spreading, 114
Flip-chip joints with high-lead solders, 129 SA type, 117
Flip-chip lands, 204(F) for stainless steels, 122
Flip-chip process flow for range of solders, in different that require cleaning, 116118
atmospheres, 199(T) for tin-base solders, 116120
Flip-chip processes for unsolderable materials, 120122
Au-20Sn eutectic solders for, 207 water soluble, 39, 40
interconnection schemes for, 201(F) WS type, 118
lead-tin eutectic solders for, 207 for zinc-bearing solders, 6566
step soldered flip-chip interconnects, 206207 Fluxless joining
surface topology of, 206 gold as solderable metallization, 89
Flip-chip solder bumps, 204(F) using Au-20Sn eutectic solders, 71
Flip-chip structures, 204206 Fluxless joining process, 3031
Flip-chip technology Fluxless process, 111
characteristics of, 202203 Fluxless soldering
rework, 204 of aluminum, 136137
underfill, 203 by compressive loading, 135(F)
Flow, molten filler metal influences to, 12 economics of, 123
Fluid flow, 1819 of gallium arsenide (GaAs), 134
Fluidity gold coating for, 124
inferior levels of, with bismuth-bearing solders, 53 process considerations for, 132133
of molten filler metals, 49 reactive atmospheres for fluxless soldering, 130
Fluorides, 111 using 82Au-18In solders, 72
Fluorine chemistry, 131 using In-48Sn solder, 135136
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 253

Fluxless wetting Gaseous reagents, 140

on copper metallizations, 132(F) Geometry of adhesive joints, 3
on gold-on nickel metallizations, 132(F) Germanium, 77
of Pb-62Sn solder, 132(F) Gibbs free energy
Fluxless wetting angle of indium-silver substrates, changes in, 105, 141
135(F) defined, 137
Foil preforms depression of, as a function of temperature increases,
deficiencies of, 171 98
production of by solidification casting, 32(F) pressure dependence of, 139141
production of by strip casting, 32(F) reference point of, 140
and spreading tests, 212 Gibbs free energy function (G), 16, 139
Foils and sheets, 23, 31 Gold
Force diagram for immersed solid plate, 209(F) as additive to indium, 135
Formic acid vapor, 111 characteristics of, 43
concentration, effect of, 132(F) as constituent of high-melting-point solders, 51
as reactive atmospheres for fluxless soldering, 130131 critical levels of, 91
in self-alignment of flip chip assembly, 130 diffusion bonding of, 45, 43
Forming gas, 111 diffusion soldering of, 233234
Foundation layers, 150 effect of, on tin-base eutectic solders, 153(T)
Foundation metal, 199 effects of addition to Ag-97.5Pb-1Sn solder, 154(F)
Free energies effects of addition to In-18Pb-70Sn solder, 154(F)
comparative values of metal oxide formation, 105(T) erosion of, by molten indium, 81
of oxide formation, 125 erosion of, by molten tin, 82(F)
Free-energy change high rate of dissolution of, in molten lead-tin solder, 89
for oxidation of several metals, 106(T) melting point of, 54
for oxidation reactions, 106 in oxidizing atmospheres, 35
Free energy of formation, 110 as solder constituent, 54
Fuel gases, 34(T) as solderable metallization, 89
Furnace joining, 104 temperature/pressure curve for diffusion bonding of,
Fusible coatings, 149 43(F)
Fusion, 97 as wettable metallizations, 147
Gold-antimony alloys, 66
Gold-antimony phase diagrams, 69(F)
G Gold-base brazes, 66
Gold-base solders, 111
Gallium Gold-bearing solders
and alumina, 62 for gold-metallized components, 66
amalgams based on, 215216 melting point of, 66
liquid alloys based on, 93 used as solders, 67(F)
Gallium alloy systems, 93 Gold-coated components
Gallium arsenide (GaAs), 134 electrodeposition of, 155
Gallium-base amalgams, 216 solderability shelf life of, 148(T)
Gallium-copper amalgams, 216(T) Gold coating, 133
Gallium-indium-tin alloys Gold flash, 10, 147148
Ga-In-Sn, melting points of, 93 Gold-germanium, 6671
Gallium-indium-tin solders Gold-germanium eutectic alloy, 70
Ga-In-Sn, effect of additions to, 94(T) Gold-germanium phase diagrams, 68(F)
Gallium-nickel-copper amalgams, specific types Gold-germanium solders, specific types
Ga-5Ni-30Cu, 216 Au-12Ge, characteristics of, 70
Ga-5Ni-30Cu, curing curves for, 217(F) Au-12Ge, contact angle of, 72(T)
Gamma phase intermetallic compound, 215 Au-12Ge, iron germides, 70
Gap size, 178179 Gold-indium intermetallic phases, 83(F)
Gaps and x-ray inspection, 204 Gold-indium joints, 231
Gas atomization, 220 Gold-indium noneutectic alloys
Gas evolution from polymeric materials, 172 82Au-18In solders, 72
Gas Law, 139 Gold-indium phase diagrams, 52(F)
Gas torch, 34 Gold-indium solders, specific types
Gaseous fluxes 82Au-18In, fluxless soldering, 72
as fluxless process, 111 82Au-18In, gold-indium noneutectic alloys, 72
narrow joints with, 178 82Au-18In, highest melting-point solders, 72
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
254 / Principles of Soldering

Gold-indium solders, specific types (continued) Gold-tin solder, 32

Au-18In, contact angle of, 72(T) Gold-tin solders, specific types
Gold jewelry, 233 Au-20Sn, application of, 71
Gold layers in metallization, 150 Au-20Sn, bismuth as impurity of, 77
Gold-lead-tin ternary system Au-20Sn, cobalt metallization for, 71
gold limit in, 89 Au-20Sn, fabrication costs of, 155
liquidus projection of, 89(F) Au-20Sn, fluxless joining of, 71
liquidus surface of, in phase diagrams, 89 Au-20Sn, foils and preforms of, 71
vertical section through, 90(F) Au-20Sn, for flip-chip process, 207
Gold limit, 89 Au-20Sn, for soldering to gallium arsenide, 153
Gold metallizations Au-20Sn, forms of, 71
characteristics of, 89 Au-20Sn, hermetic sealing of ceramic semiconductor
erosion of, by molten indium, 53(F) packages, 71
and indium-bearing solders, 5253 Au-20Sn, high-lead solders as alternate to, 72
Gold-metallized components Au-20Sn, intermetallic compounds of, 226
gold-bearing solders for, 66 Au-20Sn, melted in controlled atmosphere, 113(F)
In-48Sn solder to, 136(F) Au-20Sn, palladium metallization for, 71
Gold-on-nickel metallizations, 132(F) Au-20Sn, rapid solidification casting, 71
Gold-plated tin foil, 156(F) Au-20Sn, titanium as additive to, 69
Gold-silicon, 6671 Au-30Sn, foil melting point, 156(F)
Gold-silicon intermetallics, 66 Graduated joint structures, 165
Gold-silicon phase diagrams, 67(F) Grain-boundary sliding, 215
Gold-silicon solders Grain refinement, 81, 218
Au-2wt%Si, 66 Graphite brazing, 30
cost advantage of gold-germanium eutectic alloy, 6970 Green manufacturing, 190
as a desiccant for hermetic packaging, 184
for silicon semiconductor chips to gold-metallized pads,
66 H
vapor-phase technique, 66
Gold-silicon solders, specific types Halide atmospheres, 111
Au-2Si, 71 Halide fluxes, 117
Au-2Si, spreading behavior of, 69(F) Hallmarking regulations, 50
Au-2Si, titanium as additive to, 69 Halogen-base fluxes, 122
Au-2Si, zinc as impurity of, 77 Halogen gases, 130
Au-2wt%Si, alloy additions for spreading, 67 Health and safety issues
Au-2wt%Si, gold-silicon solders, 66 asthma, 43
Au-2wt%Si, high molten viscosity of, 66 from beryllia dust, 163
Au-2wt%Si, silaceous dross, 66 of beryllium, 216
Au-2wt%Si, tin as alloy element for, 68 of cadmium and cadmium alloys, 50, 51, 63, 191
Gold-silicon-tin alloy system phase diagrams, 69 of electronics equipment disposal, 190
Gold-silicon-tin phase diagram, 70(F) explosion risk of hydrogen atmospheres, 110
Gold-silicon-tin-titanium solders, specific types exposure limits for hazardous dusts and vapors, 43
Au-2Si-8Sn-1Ti, silicon wetting, 69 eye and nose irritation, 43
Gold-tin, 68(F), 7172 flammability, 43
Gold-tin alloy system, 233 of hydrogen, 37
Gold-tin alloys, specific type of lead, 50
Au-20Sn, wetting effect or rare earth doping, 228229 of mercury, 93, 191, 215
Gold-tin eutectic solders of nickel, 50
fluxes with, 122 of organic metal compounds toxicity, 131
high-melting-point solders, 72 soldering fumes, 43
intermetallic compounds of, 226 of thallium, 191
Gold-tin intermetallic compounds, 32 Heat-affected zone (HAZ), 4
AuSn4, and embrittlement, 133 Heat capacity, 62
AuSn4, and tin-base solders, 133 Heat treatment
AuSn4, as stoichiometric compounds, 91 and creep, 158
Gold-tin intermetallic compounds, specific types in non-oxidizing atmosphere, 148
Au-Sn4 phase, and joint embrittlement, 90 vs. oxygen concentration, 214(F)
Au-Sn4 phase, in Au-Pb-Sn ternary system, 89 prior to joining, 3738
Gold-tin partitioned filler metals, 155 stress relaxation, 158
Gold-tin phases and joint embrittlement, 90 temperature of, 39
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 255

Heat treatment temperature, 213 Hydrogen safety, 127

Heating cycles Hydrostatic forces, 15, 27, 212
of joining operations, 3839 Hydrostatic pressure, 178179
parameters of, 38 Hypereutectic alloys, 19
profiles of, 38(F) Hypoeutectic alloys, 19
Heating methods, 3334
Heating rate, 38
Heavy metal deposition, 122 I
Helium, 184
Helium leak rate, 184 IA type fluxes, 117
Hermetic sealing Ideal substrate, 19
Bi-43Sn solders for, 173 Image analysis, 211
of ceramic semiconductor packages, 71 Immersion plating, 180
with compressive force, 174 Impurity
dryness of, 184185 aluminum as, 76, 77
In-48Sn solder, 44 antimony as, 76
Sn-40Pb solder, 44 bismuth as, 76, 77
with standard solders, 43 cobalt and iron as, 77
High creep levels of high-lead solders, 72 germanium as, 77
High-cycle fatigue life of composite solders, 221(T) silver as, 77
High-cycle fatigue resistance, 220 tin as, 77
High-lead solder alternative, 197 zinc as, 77
High-lead solders Indentation welding, 9
98Pb-2Sn solders, 73 Indicative property values of selected solders and pure
as alternate to Au-20Sn solders, 72 metals, 226(T)
fluxes with, 122 Indium
high creep levels of, 72 amalgams based on, 217
mechanical properties and corrosion resistance of, diffusion bonding of, 45
73(T) fluidity of, 135
mechanical properties of, 73 melting point of, 75(F)
step-soldering sequence for, 72 molten, and gold erosion, 81
High-melting-point solders, 54 in pressure-welded joints, 43
constituents of, 51 production levels of, 192
eutectic alloys of, 7(F) pure, as a solder, 74
fast atom cleaning effect of, 131(F) temperature/pressure curve for diffusion
lead-free, 197198 bonding of, 44(F)
lead rich, 72(T) and tin in intermetallic compounds, 51
for thick film metallizations, 72 Indium amalgams, 217
High-molecular-weight hydrocarbons, 122 Indium and indium alloys, 201
High-purity solders, 77 Indium and lead, 5253
High-volume contraction, 64 Indium-base solders
Highest melting-point solders, 72 alloy additions to, 74
Highly active elements, 103 benefits of platinum in, 147148
Holding time, 38 composites and melting points of, 75(T)
Homologous temperatures, 6 creep rupture of, 73
Hookes law, 2728 ductility of, 73
Hot shortness, 26 intermediate melting temperature solders, 75
Hydrochloric acid (HCL), 116 low melting point of, 73, 74
Hydrogen in optoelectronic applications, 74
explosion risks of, 37 phase segregation failure of, 74
in inert atmospheres, 36 in photonic applications, 74
solder oxides reduction by, 126127 rare earth doping and contact angles of, 228
uses of as reducing gases, 109 superheats for, 75
Hydrogen atmospheres Indium-bearing solders
and explosion risk, 110 ductility of, 52
and gold-base solders, 111 and gold-metallizations, 5253
oxides with hydrogen at a partial pressure, 129(F) Indium bump bonding, 200
tin oxide reduction in, 111 Indium-gold reaction, 82
Hydrogen plasma, 128 Indium-lead alloys
Hydrogen poisoning, 105 non-eutectiferous character of, 51
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
256 / Principles of Soldering

Indium-lead alloys (continued) Interfacial reactions, 19, 24

peritectic reaction, 75 Intergranular brittle fracture, 215
Indium-lead phase diagrams, 61(F) Intergranular corrosion, 114
Indium-lead-silver solders, specific types Interlayers
In-15Pb-5Ag, 172(F) controlled expansion materials, 164165
In-15Pb-5Ag, indium-base solders, 75 for diffusion bond dissimilar metals, 10
In-15Pb-5Ag, intermediate melting temperature solders, gold flash, 10
75 nickel foil, 10
In-15Pb-5Ag, thermal fatigue performance of, 75 silver foil, 10
Indium-lead-tin solders, specific types Intermetallic compounds
In-18Pb-70Sn, gold addition, effect of, 154(F) barrier metallizations to avoid, 155
In-18Pb-70Sn, gold addition effects on, 154 in composite solders, 219
Indium oxides
duration of growth, 25
alloying additions to facilitate removal of, 7475
effects of, 153
stability of, 125
elastic modulus of, 154
temperature for reduction of, 126
formation of, 10, 51, 154
Indium-silver substrates, 135(F)
Indium solder alloys gamma phase, 215
creep behavior of, 158 growth of, 16, 25
fatigue resistance and, 158 indium and tin in, 51
Indium solders, 7375, 116 mechanical and physical properties of, 87(T)
Indium-tin phase diagrams, 55(F) in solders, 5
Indium-tin solders in ternary systems, 84
for magnesium, 122 thickness of, and creep rupture, 88, 88(F)
Indium-tin solders, specific types Intermetallic phase layer, 16
In-48Sn binary, melting point of, 96 Internal energy, 137138
In-48Sn, contact angle for, 128(F) International Electrotechnical Commission (IEC),
In-48Sn, continuum between stress strain and creep 119(F)
data for, 74(F) International Organization for Standards
In-48Sn, elongation, 194 flux classifications of, 117
In-48Sn, fluxless soldering, 135136 soldering flux classifications, 118(T)
In-48Sn, hermetically sealed enclosures-hermetically International Programme for Alloy Phase Diagram
sealed enclosures, 44 Data (IPADA), 79
In-48Sn, joining operations with, 135 Invar, 160
In-48Sn, oxide growth on, 125 Invariant reactions, 193
In-48Sn, preform of, 136(F)
Ion-aided deposition process, 180
In-48Sn, stress-strain curve for, 74(F)
Ion-assisted vapor spray deposition, 178
In-48Sn, to gold-metallized components, 136(F)
Ion plating, 180
Indium-tin-zinc solders, specific types
Ionic contamination, 119
5In-87Sn-8Zn, grain refinement in, 218
In-46Sn-2Zn ternary, melting point of, 96 Ionized-cluster beam deposition, 180
In-86Sn-9Zn, melting point of, 96 IPC/EIA J-STD-003A (test method), 207
Inert gas atmospheres Iron and tin dendrites, 77
described, 3536 Iron germides, 70
industrial quality, 107 Iron-nickel alloys
soldering in, 107109 controlled expansion materials, 160161
types of gases in, 36 machinability of, 161
uses of, 109 trade names of, 160
Infrared microscopy, 204 Iron-nickel alloys, specific type
Inorganic acid fluxes, 117 17Co-54Fe-29Ni, coefficient of thermal expansion
Inorganic acids, 116 (CTE), 160
Inorganic fluxes, 53 17Co-54Fe-29Ni, iron-nickel alloys, 160
Inspections of joint interior, 183 17Co-54Fe-29Ni, Kovar, 160
Instantaneous melting properties, 93 Fe-36Ni, coefficient of thermal expansion (CTE), 161
Interatomic force Fe-36Ni, thermal expansion characteristics of, 162(F)
per unit area, 27 Fe-36Ni, total expansion of, 162(F)
variations of, 27(F) Fe-42Ni, coefficient of thermal expansion (CTE), 161
Interdiffusion, 148149 Iron-tin intermetallic compound, specific types
Interfacial compound formation, 84 FeSn2, as stoichiometric compounds, 91
Interfacial compounds, 26, 103 Irreversible processes, 139
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 257

J Joints, 175178
butt joints, 177, 177(F)
Jetting system schematic, 201(F) cleaning for solid-state joining, 5
Jewelry, 175, 233 cleanliness of, 4
Jigging, 3031 corrosion in, 49
Joining atmospheres, 35, 103111 dimensional stability of, 224226
Joining environments, 103141 with dissimilar materials and thermal cycling, 73
Joining methods, 112, 2(F) fitness for purpose tests, 224
Joining operations, heating cycle of, 3839 landed butt, 177
Joining process lap joints, 175, 176(F), 176177, 177(F), 229
cost tolerance of, 145 large area, 168, 171(F), 174(F)
fluxless, 3031 lifetime prediction of, 226227
Joining process development, 146(T) measurement of mechanical properties of, 223224
Joining temperature, minimum practical (liquidus), mechanical properties of, 26, 224
145 modeling lifetime of, 223
Joint defect rates, 42 narrow, 178
Joint design numerical modeling of, 224227
dimensions, 12 peel force profiles of, 168(F)
dimensions and mechanical properties, 25 quality of, 12
issues regarding, 30 recommended designs for, 177(F)
to minimizes concentration of stresses, 175178 scarf butt joints, 177
remedy of problems in, 146 shear strength and compressive loading, 136(F)
strengthened solders, 178 shear strength as a function of thickness, 178(F)
trials of, 145 shear strength of, 136(F), 178(F), 222
Joint embrittlement solidification shrinkage, 173
effects of, 153 step butt joints, 177
effects of intermetallic phases on, 153 strap joints, 177, 177(F)
and gold-tin phases, 90 strength factors influencing, 224
of indium-bearing solders, 52 surface roughness of, 22
of tin-based solders, 52 tensile strength of, 222
Joint filling requirements, 26 tongue and groove joints, 177
Joint gaps trapped gas, 169173
compressive forces applied to, 134 trapped gas sweeping, 170(F)
control of, 26 voids in, 49
limits to, 2526 wide gaps in, 158
optimal balance of, 26 Journal of Phase Equilibria (phase diagrams), 79
self regulation of, 27
significance of, 2527
size of, 26 K
upper practical limit to, 26
voids in, 169(F), 170(F) Kelvin-Planck statement, 138
Joint geometries, 29 Kinetics of reaction, 153
for brazed joints, 4 Kirkendall porosity, 10
length vs. void content, 169(F) Kirkendall voids, 71, 234
of mechanical fasteners, 1 Kovar, 160
for soldered joints, 4
for welding, 4
Joint integrity L
advances in techniques for assessing, 235238
optical inspection, 237238 Landed butt joints, 177
scanning acoustic microscopy, 235236 Lanthanum, 227
ultrasonic inspection, 235236 Lap joints, 176177
x-radiography, 236237 failure in, 176(F)
Joint quality degradation and power cycling, 65(F) geometry of, 176(F)
Joint strength recommended designs for, 177(F)
of Ag-96.5Sn eutectic solders, 232 shear stress on, 175
of Ag3Sn intermetallic phase, 232 strength of, and rare earth doping, 229
effect on, from delay from cleaning to assembly, 131(F) stress distribution in, 176(F)
and wetting of component surface, 136 Large-area joints
Joint weakness, 28 definition of, 168
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
258 / Principles of Soldering

Large-area joints (continued) Lead-rich alloys, 6

due to trapped gas, 174(F) Lead-silver solders, specific types
pressure variation and void levels, 174(F) 1.5Ag-92.5Pb-5Sn, micrograph of, 73(F)
vapor pockets in, 171(F) Lead-tin alloys
voiding in, 171(F) as compression interconnects, 201
Latent heat of fusion, 97 tensile strength of, 81(F)
Lattice waves, 226 Lead-tin-copper solders, specific types
Lead 61.75Sn-38.05Pb-0.2Cu(wt%), lead-tin eutectic solders,
as constituent of high-melting-point solders, 51 84
effects of, on surface tension of tin, 193 Lead-tin intermetallic compounds, specific types
health restrictions on, 50 Pd3Sn2, with palladium metallization, 71
in landfills, 190 Lead-tin phase diagrams, 58(F)
pure, as solder, 7273 Lead-tin solders
Lead-antimony-tin system, 62(F) additions to, 57
Lead-antimony-tin ternary system, 57 advantages of, 56
Lead-free solder paste, 155 antimony in, 57
Lead-free solders application of hydrogen plasma to, 128
Ag-20In-77Sn alloy for, 96 balls of, 129(F)
Al-3Ga-3Mg-90Zn solders, 66 bismuth in, 57
Alloy J, 60 change of dissolution rate of silver, 154
availability of potential alloying elements for, 192(T) characteristics of, 191
composition ranges for, 192(T) contact angle of, 22(F)
corrosion resistance of, 195 for copper, 116
costs of, 192 and copper substrate, reaction phases formed by, 87(F)
drive for, 190191 copper-tin intermetallics from, 84
economics and availability of, 191193 creep curve for, 218(F)
economics of, 192 creep-resistance of, 219
elongation of, 194 dissolution rate of metals and metallizations in, 87(F)
fatigue resistance ranking of, 195 dissolution rate of silver in, 53(F)
and fluxes, 196 drop-in replacement for, 95, 96, 191
health, safety and environmental aspects of, 191 for flip-chip process, 207
high-melting point, 197198 fluxes for, 122
history of, 190 grain refinement, 218
In-48Sn binary eutectic solders, 96 history of, 56
In-86Sn-9Zn solders, 96 lead-free solders, alternatives to, 191193
lead-tin solder, alternatives to, 191193 lead-free solders, compatibility with, 191
lead-tin solder, compatibility with, 191 molten, high rate of dissolution of gold in, 89
literature on, 189190 repairs to, 191
mechanical properties of, 194195 shear strength of joints with, 54(F)
melting point depression, implications for, 9596 silver in, 57
melting ranges of, 191, 196 vs. silver-tin solders, 60
metallurgical, physical, and chemical properties of, solder alloy systems, 5660
193196 tensile strength of, 222(F)
other physical properties of, 194 wetting angle of, 14(F)
physical and chemical characteristics of, 191 wetting behavior of, on mild steel, 210(F)
plastic flow of, 194 wetting speed of, 197(F)
for plumbing applications, 51 Lead-tin solders, specific types
for printed circuit board (PCB) components, 95 98Pb-2Sn, Chadwick peel tests of, 73
process window for, 196197 Pb-3Sn, effect on joint strength on delay from cleaning
reduction in superheat for, 196 to assembly delay, 131(F)
shear strength of, 194 Pb-3Sn, fast atom cleaning effect of, 131(F)
silver-copper-tin ternary phase equilibria, 193 Pb-3Sn, flip-chip joints, 129
strain rate of, 194 Pb-3Sn, high melting point, 131(F)
surface tension of, 193194 Pb-4Sn, 30
tin-based solders as key to replacement for, 192 95Pb-5Sn, oxide growth on, 125
tin pest and tin whiskers, 195196 Pb-5Sn, for flip-chip bonding process, 166
ultimate tensile strength of, 194 Pb-26Sn, eutectic composition, 89
wetting and spreading characteristics of, 197 Pb-60Sn, 51
yield strength of, 194 Pb-60Sn, contact angle of, 20
Lead-free tin-base solders, 116 Pb-60Sn, effect of major ternary additions to, 76
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 259

Pb-60Sn, impurity concentrations producing detrimental Lutetium, 152

effects for, 76(T) Lutetium oxide (Lu2O3), 152
Pb-60Sn, lead-tin eutectic solder, 51
Pb-60Sn, molten, rate of dissolution of copper in, 84
Pb-60Sn, spreading of, on gold-plated sample, 213(F) M
Pb-62Sn, elongation, 194
Pb-62Sn, fluxless wetting of, 132(F) Machinability and machining costs, 160, 161
Pb-62Sn, for copper components, 131 Magnesium, 122
Pb-62Sn, for gold-nickel components, 131 Magnetostriction, 160
Pb-63Sn, copper wetting using rosin flux, 115(F) Maps of brazes and solders, 7
Pb-80Sn, effect of alloying additions on wetting of, Materials
76(T) storage of, 43
Pb-80Sn, effect of major ternary additions to, 76 strength of, 2728
Leadless ceramic chip carriers (LCCs), 40 Materials systems approach, 146(T)
Lever rule, 80, 80(F), 9293 Maximum exposure limits, 43
Lifetime prediction of joints, 226227 Mechanical cleaning, 37, 209
Limitations of solderability test measurements, 210 Mechanical constraints and solutions, 157168
Linear expansion coefficient vs. thermal conductivity, Mechanical fasteners and fastening, 12
159(F) Mechanical integrity, 29
Liquid flow, 19 Mechanical properties. See also tests/testing
Liquid infiltration casting, 161 of intermetallic compounds, 87(T)
Liquid lake condition, 134 of joints, 26
Liquid metal embrittlement, 30 of joints, measurements of, 223224
Liquid nitrogen, 109 of lead-free solders, 194195
Liquid phase sintering, 157 of selected solders, 194(T)
Liquid solder film, 134 of soldered joints, 224
Liquid-solid metallurgical reactions, 34 tests/testing of, 224
Liquid spreading, 19 Mechanical strengths in solders, factors in, 223224
Liquidus projection, 89(F) Melting, instantaneous, 93
Liquidus surface Melting point
of Ag-Cu-Sn system, 63(F) of Ag-20In-77Sn solders, 96
of Ag-Sb-Sn system, 64(F) of Ag-Cu-Sn ternary system, 96
of Cu-Pb-Sn system, 85(F) of aluminum-silicon alloys, 6
of Pb-Sb-Sn system, 62(F) effect of, on multiple alloying additions, 9798
of Si-Pb-Sn system, 65(F) of eutectic alloys, 81
Liquidus temperature, 49 of eutectic solders, 52
of 3.8Ag-0.7Cu-95.5Cu solders, 228(F) gold addition effects on, 153154
calculating depression of, 98 of gold-bearing solders, 66
of eutectic silver solders, 80 of In-46Sn-2Zn ternary eutectic, 96
increases of, from silver additions, 53 of In-48Sn binary eutectic, 96
Load applied to preforms vs. void level, 134(F) of In-86Sn-9Zn solders, 96
Local heating, 33 of indium, 75(F)
Local interfacial mismatch stresses, 28 of liquid metal, 15
Low-alpha-emission solders, 190 of lower-melting point solders, 51
Low-alpha lead, 190191 of metals and their coefficient of thermal expansion
Low-cycle fatigue life, 221(T) (CTE) of, 159, 160(F)
Low-cycle fatigue resistance, 220 and peak operating temperature, 29
Low-expansion materials, 161(T) of silver-tin binary system, 96
Low-expansion metals, 160 of solid metal, 15
Low-expansivity materials, 223 Melting point depression
Low-melting-point eutectic alloys, 55(F), 149 behavior of, 9395
Low-melting-point metals, 192(T) cadmium-base solders, 93
Low-melting-point solders by eutectic alloying, 9396
Ag-96Sn properties of, 53 general features of, 9395
eutectic alloys of, 7(F) lead free solders, implications for, 9596
of indium-base solders, 73 liquid alloys based on gallium, 93
stress-rupture life of, 195(F) Melting ranges
Low-solid fluxes, 118 of Alloy J, 60
Low spreading and bond quality, 212 and braze alloy families, 7(F)
Lower-melting-point solders, 5152 of lead-free solders, 191
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
260 / Principles of Soldering

Melting ranges (continued) Metallurgical reactions, 12

and solder alloy families, 6(F) erosion rate changes to parent materials, 154
Mercury liquid-solid, 34
amalgams based on, 215216 melting point changes, 153154
health hazards, 215 Metallurgical stability of soldering processes, 29
health safety issues of, 93 Metals
melting point of, 93 cohesive strength of, 27
and silver powder, diffusion reaction of, 215 and their properties, 161(T)
substitutes for, 93 wetting of by solders, 147157
toxicity of, 191 Metals and metallizations in diffusion bonding, 11(T)
Mercury-base amalgams, 215 Microcracks, 28
Metal-ceramic composites, 163 Microelectromechanical systems (MEMS), 123, 222
Metal-loaded glass frits, 181 Microflame torch, 34, 34(F)
Metal-matrix composite (MMC), 175 Microfocus x-ray systems, 236(F), 236237
Metal-metalloid composites, 163 Micrographs
Metal oxide reduction, 110(F) of Ag3Sn intermetallic phase, 92(F)
Metal oxides of controlled expansion (CE) alloys, 164(F)
bond strength of, with parent metals, 106107 of diffusion-soldered joint, 233(F)
chemical reduction of, 105 of lead-tin solder on gold-plated copper substrate,
and formic acid, reaction between, 131 213(F)
Metal-to-oxygen chemical bond, 105 of peritectic transformation, 83(F)
Metallic impurities, 7577 Microstructural coring, 75
Metallization layer, 151 Military Standards (MIL-STD)
Metallization techniques 883D, (void free joints), 183
advantages and disadvantages of, 181 883E, method 2019.7 (acceptance criteria), 231
as a barrier, 153 Mischmetals, 227
characteristics of, 181(T) Mismatch stress concentration, 164
chemical vapor deposition, 180 Mismatch stresses, 39
coating quality of, 182(T) elimination of, 165166
of porous ceramic materials, 151(F) Modeling
process parameters of, 182(T) joint lifetime, 223, 224
relative merits of, 182(T) of lifetime of joints, 224
with silver electroplating, 151(F) Modeling assumptions in finite element analysis (FEA),
thick-film formulations, 180181 225
wet plating, 180 Modeling elements, 226
Metallizations
Modulus of elasticity, 157
of alumina, 151
Moisture content
and control of spreading, 148
as a function of bakeout time and temperature, 185(F)
costs of, 152, 231
as a function of bakeout times, 184
firing on glass and ceramics, 151
gold layers in, 150 Moisture in semiconductor failure mechanisms, 123
in layers, 150 Moisture permeation, time predicted for, 184(F)
and metals in diffusion bonding, 11(T) Moisture removal, 184
moly-manganese process of, 151 Molar volumes, 140
noble metals, 181 Molecular hydrogen, 127
of oxide ceramics, 151 Molten acetamid fluxes, 122
of parent materials, 49 Moly-manganese process, 151
for refractory materials, 130 Molybdenum
solderable, gold as, 89 of coefficient of thermal expansion (CTE), 160
stresses in, 151 joining atmospheres for, 104
with titanium, 150 as low-expansion metals, 160
widely used, 37 Monatomic hydrogen, 127128
with zirconium, 150 Monolithic plates, 165(F)
Metalloid-metalloid composites, 163 Montreal Protocol on Substances That Deplete the
Metallurgical considerations for solder selection, 78 Ozone Layer: 1991, 111
Metallurgical constraints and solutions, 147157 Multichip modules (MCMs), 199
Metallurgical incompatibility of materials and Multilayer metallic coatings, 50
processing conditions, 147 Multilayer metallization, 199
Metallurgical modification of contact angle, 133 Multiphase materials, 160
Metallurgical properties of lead-free solders, 193196 Multiple frequency SAM, 236
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 261

N Outgassing, 30
Oxidation
Narrow joints, 178 rate of, 107
Native oxides, 126(T) water vapor as source of, 36
Neodymium, 227 Oxidation reactions, 106
Neutral flame, 34 Oxide-dispersion-strengthened solders, 218219
Nichrome, 150 Oxide films
Nickel destabilization of, and rare earth doping, 228
health restrictions on, 50 effects of, 111
toxicity of, 191 reduction of, 105106
Nickel foil interlayers, 10 regrowth of, 129
Nickel powder, 216 removal of, 35
Nickel-tin intermetallic compounds, specific types Oxide formation and removal, 124125
Ni3Sn2 phase, binary compounds, 84
Oxide growth
Ni3Sn4 phase, binary compounds, 84
on 95Pb-5Sn solders, 125
Nilo-Alloy 42, 160
on base metals, temperature dependence, 125(F)
Nilo-K, 160
on base metals, time dependence, 124(F)
Niobium (columbium), 104
Nitrogen equation for, 124
costs of, 109 on molten solders, 126(F)
dry, for moisture removal, 184 Oxide layers, 122
in inert atmospheres, 36 Oxide reduction
Nitrogen-based soldering, 197 alternative atmospheres for, 111
No-clean fluxes, 39, 40, 118119 dynamics of, 126
Noble-metal metallizations, 181 thermodynamic aspects of, 106107
Noble metals thermodynamic principles for analysis of, 106
embrittlement from, 147 Oxide reduction rate by hydrogen, 126
oxides of, 105
Oxide scale, fluxes with, 34
stability of, 105
Oxide thickness
Non-nickel alloys, 162(F)
Nonmetallic bonding, 152 and defect levels, 124
Nonmetallic components effects of, on defect levels, 124(F)
chemical vapor depositions on, 149150 vs. oxidation time, 125(F)
physical vapor depositions on, 149150 temperature effects on growth of, 124
wet plating on, 149150 Oxides
Nonmetallic materials, wetting and spreading with, on aluminum, 9
103 growth of, 125, 148
Nonmetallic phases and wetting problems, 147 with hydrogen at a partial pressure, 129(F)
Nonmetallic surface coating, 114
mechanical removal of, 128130
Nonmetals
native, superheats to dissolve, 126(T)
foundation layers for, 150
solderable coatings on, 149152 reduction of, by reactive gas atmosphere, 130131
wetting of by solders, 149153 surface, removal of, 37
Nonoxidizable metallizations, 133 Oxidizing atmospheres
Nonwettable patches, 17 gold in, 35
Numerical modeling, 223224 graphite effect in, 30
platinum-group metals in, 35
Oxidizing flame, 34
O Oxygen atmospheres and solder spreading, 127(T)
Oxygen concentration
OA type fluxes, 117
effect of, on rate of dross formation, 115(F)
Off-eutectic compositions, 54
vs. heat treatment, 214(F)
Optoelectronic devices, 214215, 221
Organic acids, 116 Oxygen partial pressure, 3536
Organic acids fluxes, 117 to effect oxidation reaction, 141
Organic coatings, 149 and metal oxide bond strength, 106
Organic films, 37 method for reduction of, 107
Organic fluxes, 121 and water vapor desorption, 108
Organic metal compounds, 131 Oxygen, residual levels of, 35
Osprey spray-forming process, 163 Ozone depletion, 39
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
262 / Principles of Soldering

P of gold-silicon, 67(F)
of gold-silicon-tin, 69
P-charts, 42 of gold-tin, 68(F)
Package leak rates, 184 of indium-lead, 61(F)
Palladium metallization, 71 of indium-tin, 55(F)
Palladium oxide, 107 of lead-tin, 58(F)
Palladium plated devices, 207208 of silver-indium, 57(F)
Parent materials of silver-lead, 61(F)
erosion of, 49, 153154 of silver-tin, 59(F)
metallization of, 49 Phase formation, 154155
wettability of, 49 Phase segregation failure, 74
Partial pressure. See also oxygen partial pressure, 36 Phased reflow soldering, 155, 157
Partitioned filler metals, 155(T), 155157 Phases
Pastes, 31 instability of, 29
Path length, 169170 rate of growth of, 81
PCBs. See printed circuit boards (PCBs) Phonons, 226
Peak bonding temperature, 38 Phosphoric acid fluxes, 122
Peak operating temperature, 29 Physical abrasion for surface cleaning, 129
Peak process temperature, 33 Physical properties
Peel force, 167(F) of Ag-96Sn solders, 53
Peel force profiles, 168(F) of copper-tin intermetallic compounds, 87
Peel fracture and height of the solder film, 167(F) of intermetallic compounds, 87(T)
Peeling stresses, 175 of lead-free solders, 193196
Peritectic reaction for low-expansion materials, 161(T)
of aluminum quaternary alloy, 83 of selected solders, 194(T)
described, 83 for semiconductors, 161(T)
indium-lead alloys, 75 Physical vapor depositions, 149150
Peritectic transformation, 83(F) Piezoelectric ceramic elements, 31(F)
Phase diagrams Planar joints, 50
availability of, in literature, 79 Plasma, 127
binary alloy systems, 7983 Plasma-assisted dry soldering (PADS), 131
and binary alloys, 79 Plastic flow, 194, 219
binary eutectic composition solder with Plastic leaded chip carriers (PLCCs), 40
intermetallics, 8183 Plastic yielding, 10
binary eutectic composition solder with no Platinum, 147148, 148(T)
intermetallics, 7981 Platinum-group metals, 35
binary peritectic solder, 83 Platinum-tin intermetallic compounds, specific types
described, 78 PtSn4 interfacial layer, 148
distributed compound between eutectic solder and Polymeric materials, 172
component metals, 8992 Polymers, 3
higher order systems, 9293 Porosity. See voids
interfacial compound between eutectic solder and Porous ceramic materials, 151(F)
component metals, 8489 Powder metallurgy materials (P/M), 161
limitations of, 79 Powdered solders, 31
non-metallic systems, 9293 Power cycling and joint quality degradation, 65(F)
overview of, 49 Precious metals
soldering applications of, 7779 binary combinations for diffusion soldering, 231
ternary alloy systems, 8392 hallmarking regulations for, 50
of ternary systems, 84 Precipitation hardening, 49
uses of, 79 Precipitation-strengthened alloys, 175
in weight percentages, 78 Preform geometry, 133134
Phase diagrams, specific alloy systems Preforms
of aluminum-zinc, 66(F) Alloy J for, 61
of antimony-tin, 60(F) disc shaped, 170
of bismuth-tin, 56(F) dual disc, 171(F)
of copper-tin, 85(F) foil, deficiencies of, 171
of gold-antimony, 69(F) hermetically sealed enclosures-hermetically sealed
of gold-germanium, 68(F) enclosures, 44
of gold-indium, 52(F) round wire, cross shaped, 171
of gold-lead-tin, 89 thickness of, 133
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 263

Pressure dependence of Gibbs free energy, 140 Range chart (r-chart), 42

Pressure in diffusion brazing, 10 Raoults Law, 97, 98
Pressure variation method Rapid-solidification process, 66
to minimize voids, 174 Rapidly solidified alloys, 31
for reduction of level of voids, 173 Rapidly solidified filler metals, 6061
for void level reduction, 174(F) Rapidly solidified processes, 31
Pressure welding, 45 Rare earth doping
of aluminum, 9 contact angle of indium solders, 228
and dissimilar-metal joints, 9 effects of, on solders, 227
indium in, 43
oxide film destabilization, 228
joint strength of, 9(F)
of Sn-9Zn solders, 228
with standard solders, 43
Rare earth elements
Pressure welding and diffusion bonding, 812
effect of additions on solder properties, 227229
Primary dendrites of silver, 80
Principle of Conservation of Energy, 137 function of, in nonmetallic bonding, 152
Printed circuit boards (PCBs) implications of, for soldering technology, 229
assembly line change to lead-free solders, 197 solders doped with, 227229
lead-free solders for, 95 Rate of dissolution. See dissolution rate
reflow soldering of, 36 Rate of reaction, 78
solders for low process temperature, 155, 157 Reactive filler alloys, 152
tin whiskers on, 196 Reactive filler metals, 152
Process control, 42 Reactive gas atmospheres, 130131
Process control chart of peak reflow temperature, Reactive ion etching, 172
42(F) Reactive-metal metallizations, 181
Process cycle time, 33
Reactive metals, 150
Process variability (r-chart), 42
Reactive solders, 103
Process window for lead-free solders, 196197
Processing aspects, 3042 Reactive wetting, 16
Product miniaturization, 151 Reduced-oxide soldering activation (ROSA) process,
Profilometer traces, 211 132
Progressive alloying, 98 Reducing atmospheres, 8, 3637, 109111
Progressive eutectic alloying, 98 Reducing gases, 109
Progressive melting, 157 Reduction, 130(F)
Propane, 34 Reduction atmospheres, 34
Pulse-echo ultrasonic inspection, 235 Reduction flame, 34
Pure metals, 14 Reference standards, 213214
Pyroelectric elements, 34
Reflow soldering, 3334, 36
Pyrometers, 35
Reflow stage, 38
Pythagorean theorem, 45
Refractory metals
list of, 104
Q oxides of, 105
stability of, 105
Quality acceptance criteria, 209 wetting problems with, 147
Quality-control testing, 210 Reinforced solders (solder composites), 222223
Quality of soldered joints, 12 Reinforcing plates, 165
Quality of wetting and contact angle, 17 Relative spreading, 211
Quasi-binary alloy systems, 92 Resin materials, 215
Quasi-ternary alloy systems, 92
Respiratory problems, 43
Quenching stages, 39
Restoring force, 205
Reverse-gas bias mode, 150
R Reversible chemical reactions, 105
Reversible processes, 138, 139
R-chart (range chart), 42 Rheological concepts, 215
R, RMA, RA type fluxes, 117 Richardson-Jeffes diagram. See Ellingham diagram
Radiographs Roll-bonding, 9
of Ag-96Sn foil solders, 23(F) Rosin, 116
of joint integrity, 236237 Rosin fluxes, 115(F), 117
of silicon chips, 169(F) Rule of mixtures, 194
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
264 / Principles of Soldering

S as solder constituent, 53
sulfides of, 38
S-charts (standard deviation chart), 42 as wettable metallizations, 147
SA type fluxes, 117 Silver additions, 53
Sacrificial metal, 200 Silver-antimony-tin solders, specific types
Safety. See health and safety issues Alloy J, 60
Scanning acoustic microscopy, 170(F) Silver-antimony-tin system, 64(F)
Scanning acoustic microscopy (SAM), 189, 235236 Silver-base brazes, 6
Scanning acoustic techniques, 183 Silver-base filler metals, 57
Scarf angle, 177 Silver-coated components, 81
Scarf butt joints, 177 Silver-copper-phosphorous brazes, 5
Screen printing, 31 Silver-copper-tin alloys
Second Law of Thermodynamics, 138 lead-free solders, 193
Self-alignment superheats of, 126
of flip chip interconnection, 202(F) Silver-copper-tin lead-free solders, 115
predictions of, 206 Silver-copper-tin phase diagrams, 193
of solder bump bonding, 201 Silver-copper-tin solders, specific types
Semiconductor die attach, 69 3.5Ag-0.9Cu-95.6Sn ternary, cooling rate dependency
Semiconductors, physical properties for, 161(T) of, 193
Service requirements of joints, 50 3.5Ag-0.9Cu-95.6Sn ternary, melting point of, 96
Service temperature range, 3 3.8Ag-0.7-95.5Sn, stress-rupture life of, 228(F)
Service temperatures 3.8Ag-0.7Cu-95.5Cu, liquidus temperatures of, 228(F)
of brazed and soldered joints, 4 3.8Ag-0.7Cu-95.5Cu, solidus temperatures of, 228(F)
of brazes and solders, 8 Ag-1.7Cu-93.6Sn, cobalt and iron as impurity of, 77
of solid-state joining, 5 Silver-copper-tin system, 63(F)
Sessile drop tests, 211(F) Silver-copper-tin ternary system, 96
of composite solders, 220 Silver electroplating, 151(F)
tests/testing, 209 Silver foil interlayers, 10
Shear strength Silver-gold-tin solders, specific types
of fluxless joints, 37(F) 3.6Ag-1.6Au-92.8Sn, costs of, 96
as a function of joint thickness, 178(F) Silver-gold-tin ternary systems
of gold-indium joints, 231 liquidus surface of, 91(F)
of joints, 222 vertical section through, 91(F)
of joints with lead-tin eutectic solders, 54(F) Silver-gold-tin ternary systems phase diagrams, 90
of lead-free solders, 194 Silver-indium alloys, 232
of soldered joints, 178(F) Silver-indium intermetallic compounds, specific types
Shear stress, 175 AuIn2, 75
Shelf life and storage requirements of fluxed Silver-indium-lead solders, specific types
components, 114 Ag-80In-15Pb, melted in controlled atmosphere, 112(F)
Shelf life of multilayer metallic coatings, 50 Silver-indium phase diagrams, 57(F)
Silaceous dross, 66
Silver-indium-tin solders, specific types
Silica, 69
Ag-20In-77Sn, melting point of, 96
Silicon, 152
Silver-lead phase diagrams, 61(F), 80
Silicon-aluminum alloys
Silver-lead solders, 6, 7980
coefficient of thermal expansion (CTE), 163
Osprey spray-forming process, 163 Silver-lead solders, specific types
Silicon carbide, 152 Ag-97.6Pb, silver-lead phase diagrams for, 80
Silicon chip radiograph, 169(F) Silver-lead system diagram, 80(F)
Silicon-lead-tin system, 65(F) Silver-lead-tin solders, specific types
Silicon semiconductor chips to gold-metallized pads, Ag-97.5Pb-1Sn, gold addition, effect of, 154(F)
66 Ag-97.5Pb-1Sn, gold addition effects on, 154
Silver Silver-mercury intermetallic compounds, specific types
coring of, 80 Ag2Hg3, 215
diffusion soldering of, 231232 Silver oxide, 130(F)
dissolution rate of, in lead-tin eutectic solders, 53(F) Silver particle reinforcement, 220
effects of, on surface tension of tin, 193 Silver powder
erosion of, by molten tin, 82(F) in gallium-based amalgams, 216
as impurity, 77 indium amalgams with, 217
oxides of, 38 and mercury, diffusion reaction of, 215
primary dendrites of, 80 Silver-tin alloys, 232
 2004 ASM International. All Rights Reserved.
Principles of Soldering (#06244G)
Silver-tin alloys, specific type
Ag-96Sn, joints made by, 24
Silver-tin binary system, 96
Silver-tin eutectic alloy, 152
Silver-tin eutectic solders
3.6Ag-1.6Au-92.8Sn eutectic solders, 96
effect of gold additions to, 154
lead-free solders, 60
reaction of, to silver, 231
remelt temperature of, 232
wetting effect or rare earth doping, 228229
Silver-tin intermetallic compounds, specific types
Ag3Sn, 57
Ag3Sn, as continuous interfacial layer, 231232
Ag3Sn, as stoichiometric compounds, 91
Ag3Sn, characteristics of, 9192
Ag3Sn, joint strength of, 232
Ag3Sn, micrograph of, 92(F)
Ag3Sn, rate of growth of, 92
Ag3Sn, thickness of, 92(F)
Au5Sn, creation of, 233
Silver-tin intermetallic phases, 52
Silver-tin off-eutectic alloys
effect of bismuth additions to, 193(F)
liquidus and solidus temperatures of, 193(F)
Silver-tin phase diagrams, 59(F), 81
Silver-tin solders
eutectic alloys, 60
vs. lead-tin solders, 60
Silver-tin solders, specific types
Ag-95Sn, elongation to failure of, 228(F)
Ag-95Sn, tensile strength of, 228(F)
Ag-96.5Sn, 90, 193
Ag-96.5Sn, contact angle of, 228(F)
Ag-96.5Sn, for silver-coated components, 81
Ag-96.5Sn, joint strength of, 232
Ag-96.5Sn, spread area of, 228(F)
Ag-96Sn, elongation, 194
Ag-96Sn, gallium arsenide (GaAs) with, 134
Ag-96Sn, in foil, radiograph of, 23(F)
Ag-96Sn, mechanical properties of, 91(F)
Ag-96Sn, properties of, 53, 90
Silver-tin systems, 231
Silver-tin-zinc solders, specific types
3.5Ag-95.5Sn-1Zn, grain refinement in, 218
Single phase materials, 159
Solder
compatibility of, 49
creep resistance of, 217
deposition methods, 3132
fatigue resistance of, 217
Solder alloy families
and melting ranges, 6(F)
temperature ranges of, 6
Solder alloy systems
antimony in, 5354
bismuth in, 53
gold in, 54
indium and lead in, 5253
overview of, 49
www.asminternational.org
Index / 265
silver in, 53
survey of, 5175
tin in, 52
zinc in, 54
Solder alloys
constituents of, 51
surface tensions of, 15
volatile constituents of, 6
Solder bridging, 207
Solder bump bonding
process characteristics of, 202(T)
process flow for, 199(T)
self-aligning feature of, 201
semiconductor components in, 203(F)
technology characteristics of, 202(T)
Solder bumps, 205
Solder coated substrates, 134
Solder composites (reinforced solders), 222223
Solder drains, 206
Solder elements, 52
Solder flow
mechanically enhanced, 134
metallurgically enhanced, 134135
Solder foil vs. solder coated substrates, 134
Solder oxides
reduction of, by atomic hydrogen, 127128
reduction of, by hydrogen, 126127
reduction rate of, in hydrogen, 128(F)
self-dissolution of, 125126
Solder pastes, 31
Solder preforms. See preforms
Solder reflow ovens, 174
Solder reinforcement, 178
Solder selection, 78
Solderability
evaluation of, 207
of selected metals and alloys, 121(T)
Solderability calibration standards, 212214
Solderability shelf life, 132
of gold-coated components, 148(T)
of gold coating, 133
Solderability test cycle, 209(F)
Solderability test measurements, 210
Solderability test methods, 207214
Solderability testers, 208
Solderable component surfaces, 133
Soldered joints. See joints
Soldering
and brazing, distinction between, 6
chemical fluxes for, 111123
design criteria, 2830
filler metal temperatures in, 5
flip-chip-interconnections, 199207
fluxless, 123137
functional process of, 2830
health, safety and environmental aspects of, 4243
key parameters of, 1228
processes of, 2845
Soldering and brazing, 38
Soldering fumes, 43
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
266 / Principles of Soldering

Soldering iron, 129 defined, 44, 211

Soldering standards. See tests/testing for a range of alloys, 212
Solders. See also specific solders and spread factor and contact angle, relationship
active, 152153 between, 212(F)
bulk properties of, indicative data on, 225226 values for, 212(T)
for domestic water pipes, 51 Spread tests, 112113(F)
eutectic compositions of, 8 Spreading
flow enhancement, 133 area of, 20
mechanical strengths, factors in, 223224 assessment of, 210212
metallurgy of, 4999 atmosphere effects on, 20
powders, 31 capillary action of, 23
service temperature of, 8 classical model of, 12
strengthening of, 217222 driving force for, 84
theoretical viscosity of, 19 of eutectic alloys, 81
wetting of metals by, 147157 irreversible nature of, 15
wetting of nonmetals by, 149153 metallization control of, 148
Solid-liquid interdiffusional bonding. See diffusion of Pb60Sn solders, on gold-plated sample, 213(F)
soldering rate of, 22(F)
Solid-liquid interfacial reactions, 24 vs. wetting, 49
Solid-liquid reactions, 25 Spreading behavior, 70(F)
Solid-solution strengthening, 5354 Spreading characteristics of lead-free solders, 197
Solid solutions, 5 Spreading test, 212
Solid-state diffusion Sputter-deposited coatings, 148
tin-base intermetallic phases growth of by, 88(F) Sputtering, 178, 180
tin intermetallics by, 88 Sputtering process, 150
Solid-state diffusion processes, 25 Stainless steels
Solid-state joining bismuth-tin lead-free solders for, 122
characteristics of, 5 fluxes for, 122
with gold, 4344 halogen-based fluxes for, 122
with indium, 4344 nonmetallic bonding to, 152
joint cleaning for, 5 oxide layers of, 122
with solder constituents, 4344 phosphoric acid fluxes for, 122
temperature levels of, 5 and temperature uniformity, 50
Solid surfaces, 12 thermal conductivity of, 50, 122
Solidification casting, 32(F) Standard deviation chart (s-charts), 42
Solidification shrinkage Standards
control of voids from, 173 Scanning acoustic microscope, 183184
magnitude of, 173 in soldering, 183184
of molten filler metals, 169 visual inspection, 183
of selected elements used in solders, 173(T) x-ray inspection, 183184
stress concentrations from, 50 Statistical process control (SPC), 42
by vacancy diffusion, 173 Step butt joints, 177
Solids content, 118 Step height, 176, 176(F)
Solidus temperatures, 29, 62, 80, 228(F) Step-joining process, 33
Soluble coatings, 149 Step soldered flip-chip interconnects, 206207
Soluble halides, 118 Stirlings Formula, 98
Solution treatment stages, 39 Stoichiometric compounds, 91
Spherical cap, 44 Storage of multilayer metallic coatings, 50
Spherical cap geometry, 44(F) Strain energy density, 226227
Spontaneous chemical reaction, 137 Strain rate, 194
Spontaneous filtration, 223 Strap joints, 177, 177(F)
Spread characteristics on binary solder alloys, 21(F) Strength
Spread factor of lap joints, and rare earth doping, 229
and contact angle, 4445 mechanical, of brittle materials, 28
defined, 45, 211 of pressure welded joints, 9(F)
and spread ratio and contact angle, relationship Strengthened solders, 178, 217222
between, 212(F) Strengths of metals, practical, 28
values for, 212(T) Stress
Spread ratio from differential thermal expansion, 157
and contact angle, 4445 reduction in, from differential thermal expansion, 158
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 267

reduction of, by creep, 178 Surface roughness

Stress concentrations of cold-rolled copper, 14(T)
defined, 175 of components, 2225
and fillets, 167 effect of, on fracture toughness, 24(F)
fillets to reduce, 176 and fracture toughness, 2324
by high thermal gradients, 4 and spreading, 23
of metallization layer, 151 Surface temperature, 25
origins and magnitude of, 175 Surface tensions
from solidification shrinkage, 50 of binary solders, 194(T)
Stress cycles, 227 changes to various metal alloys, 205
Stress distribution, 176(F) diagram of forces of, 13(F)
Stress relaxation, 158 of lead-free solders, 193194
Stress-rupture life between liquid and vapor, 13, 14
of 3.8Ag-0.7-95.5Sn solders, 228(F) of molten filler metals, 209
of low-melting-point solders, 195(F) of solder alloys, 15
as ranking for creep resistance, 195 solder bump self-alignment, 205
of Sn-3.5Ag-0.25RE, 228(F) between solid and liquid, 13, 14
of Sn-3.5Ag solders, 228(F) between solid and vapor, 13, 14
Stress-strain curves for dental amalgams, 215, 215(F) and surface energy, diagram of, 13(F)
Strip casting, 32(F) Surface topography, 206
Sub-zero service temperature, 76 Surfactants, 117
Sublimation, 109(T) Surroundings, defined, 103
Super Solder system, 3233 Sweeping of trapped gas, 133
Superheat Symbols and abbreviations, 243
defined, 8 Synthetically activated fluxes, 117
to dissolve native oxides, 126(T)
to dissolve surface oxides, 126
for indium-base solders, 75 T
and oxygen levels for melting point, 127(T)
and solder spreading, 115 Tantalum
Surface-area-to-volume ratio, 133 diffusion bonding of, 10
Surface cleaning, 129 joining atmospheres for, 104
Surface condition Tape bonding, 203
of components, 12 Tape interconnections, 202
requirements of, in diffusion brazing, 10 Temperature and O/H ratio, 110(F)
Surface conditioning, 131132 Temperature effects
Surface diffusion, 10 of contact angle, 25
Surface energy of oxide thickness growth, 124
diagram of, 12(F) of surface temperature, 25
of a liquid, 13 of viscosity, 25
of pure metals, 14 Temperature gradients, 12
of a solid, 12 Temperature levels
and surface tension, 13 of diffusion bonding process, 9
and surface tension, diagram of, 13(F) of solid-state joining, 5
Surface energy and surface tension, 1213 Temperature limits for brazing and soldering, 4
Surface erosion, 3 Temperature measurements, 3435, 38
SURFACE EVOLVER (software), 12, 206 Temperature/pressure curve
Surface finishes, 149 for diffusion bonding of gold, 43(F)
Surface insulation resistance test (SIR), 119 for diffusion bonding of indium, 44(F)
International Electrotechnical Commission (IEC) test Temperature uniformity, 50
coupons for, 119(F) Temperatures. See also melting point
proposed changes to, 120 activation, of fluxes, 117
test plot of, showing dendritic growth, 120(F) active filler metals requirements, 149
Surface metallization, 4 of boiling, 109(T)
Surface mount solderability test (SMT), 207 of boiling/sublimation, for selected elements, 107
Surface-mount technology, 217 control of, 38
Surface-opening cavities, 163 in diffusion brazing, 10
Surface oxides of eutectiferous phase transformations, 94(T)
removal of, 37 of heat treatment, 39
superheats to dissolve, 126 of heat treatment for reference standards, 213
 2004 ASM International. All Rights Reserved.
Principles of Soldering (#06244G)
268 / Principles of Soldering
Temperatures (continued)
peak operating, 29
peak process, 33
peak reflow, process control chart of, 42(F)
of reactive filler metal bonding, 152
for reduction of indium oxides, 126
for reduction of tin oxides, 126
for reduction of zinc oxides, 126
of sublimation, 109(T)
Tensile strength
of Ag-95Sn solders, 228(F)
of composite solders, 221(T)
and dendrites arm spacing, 32(F)
of joints, 222
of lead-tin alloys, 81(F)
of lead-tin eutectic solders, 222(F)
of Sn-9Zn solders, 228(F)
Tensile stresses, 175176
Ternary systems
intermetallic phases in, 84
liquidus representation in phase diagrams of, 84
representation of, in phase diagrams, 84
Tests/testing
acceptance criteria, 209, 231
ANSI/J-STD-002 (test method), 207
basic spreading test, 211
bolometers (thermal imaging), 35
Chadwick peel tests, 73
component testing, 224
dip-and-look (DNL) test, 207
edge fillets examination, 168
EIA/IS-86 (test method), 207
fitness for purpose tests, 224
IEC board test coupons, 119(F)
inspection methods for voids detection, 183
inspections of joint interior, 183
International Electrotechnical Commission (IEC),
119(F)
IPC/EIA J-STD-003A (test method), 207
of mechanical properties, 224
Military Standards (MIL-STD), 183, 231
of palladium plated devices, 207208
quality assurance testing, 210
quality-control testing, 210
scanning acoustic microscopy (SAM), 170(F), 189,
235236
scanning acoustic techniques, 183
sessile drop tests, 209, 211(F), 220
solderability calibration standards, 212214
solderability test measurements, 210
solderability test methods, 207214
solderability testers, 208
spread tests, 112113(F)
spreading test, 212
standards, 183184
surface insulation resistance test (SIR), 119, 120
surface insulation test plots for dendritic growth,
120(F)
surface mount solderability test (SMT), 207
test coupons, 119(F)
www.asminternational.org
test plot showing dendritic growth, 120(F)
wetting balance solderability test, 208209
wetting force during solderability test cycle, 209(F)
Thallium, 191
Theoretical viscosity of solders, 19
Thermal activation energy, 8
Thermal characteristics of fuel gases, 34(T)
Thermal conductance of mechanical fasteners, 12
Thermal conductivity
of adhesive joints, 3
of aluminum and aluminum alloys, 62
of carbon fibers, 223
of composite solders, 221
in finite element analysis, 225
of lead free solders, 198
limits to, 30
vs. linear expansion coefficient, 159(F)
and rules of mixtures, 194
of soldering atmospheres, 114(T)
of stainless steels, 50, 122
of titanium, 160
of zinc-bearing solders, 63
Thermal cycling
without creep, 165
without fatigue fracture, 165
Thermal distortion, 33
Thermal distortion parameter, 159(F)
Thermal expansion
of engineering ceramics, 159
mismatch in, 145
of single phase materials, 159
Thermal expansion mismatch
effects of, 157
in flip chip interconnection, 200
stress reduction technique, 165(F)
Thermal expansion mismatch strain, 50
Thermal expansivity
of aluminum and aluminum alloys, 62
of bismuth solders, 225
of engineering materials, 26(T)
reduction of, 222223
Thermal fatigue, 29
Thermal heat capacity, 50
Thermal imaging bolometers, 35
Thermal properties of joints, 50
Thermally conductive adhesives, 3
Thermally induced distortion, 62
Thermocompression bonding, 5(F)
Thermocouples, 3435
Thermodynamic and diffusion-kinetic model of
metallizations, 152
Thermodynamic equilibrium, 137
Thermodynamic principles for analyzing oxide
reduction, 106
Thermodynamics
first law of, 137138
second law of, 138139
Thick-film formulations, 180181
Thick film metallizations, 72
Thick-gap soldering, 178179
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 269

Thick solder joints, 179 Tin-zinc alloys, specific type

Thickness of coatings, 148 Sn-9Zn eutectic, corrosion of, 96
Thin coatings, 180 Tin-zinc-silver solders, specific types
Thin foil preforms Sn-8.5Zn-1Ag, aluminum as impurity in, 77
disadvantages of, 136 Tin-zinc solders
economics of, 136 Sn-9Zn eutectic alloy, 96
shapes of, 136 Tin-zinc solders, specific types
Thin-gap soldering, 178179 Sn-9Zn, elongation to failure of, 228(F)
Thin (narrow) joints Sn-9Zn, rare earth doping of, 228
with hydrostatic pressure, 178179 Sn-9Zn, tensile strength of, 228(F)
in semiconductor grade clean room, 179 Sn-9Zn, wetting force of, 228(F)
Threshold formation, 9 Tinned surfaces, 151
Through-thickness vias, 162 Titanium
Time requirements of diffusion brazing, 10 as additive to Au-20Sn solders, 69
Tin as additive to Au-2Si solders, 69
activation energy of, 124 of coefficient of thermal expansion (CTE), 160
allotropic transformation of, 76, 77(F), 195196 diffusion bonding of, 10
allotropic transformation suppression, 196 joining atmospheres for, 104
as alloy element for Au-2wt%Si solders, 68 as low-expansion metals, 160
as impurity, 77 nonmetallic bonding to, 152
and indium intermetallic compounds, 51 thermal conductivity of, 160
as solder constituent, 52 for zinc oxide ceramic metallization, 150
Tin-base eutectic solders, 153(T) Titanium alloys, 10
Tin-base intermetallic phases, 88(F) TO-220 semiconductor die, 66(F)
Tin-base solders Tombstoning, 157
3.5Ag-0.9Cu-95.6Sn ternary eutectic solders, 96 Tongue and groove joints, 177
Alloy J, 6061 Toxicity, 191
dissolution rate of platinum in, 148(T) Transient liquid phase (TLP) joining. See diffusion
failure modes of, 74 soldering
fluxes for, 116120 Transition reactions, 83
and gold coatings, 133 Transuranium elements, 227
as key to lead-free solders replacement, 192 Trapped gas
silver-tin, 60 in adhesively bonded joints, 173
Tin-bismuth solders, specific types described, 169173
Sn-65Bi, 15 pressure variation method for, 174(F)
Tin dendrites, 77 reduction, by vapor-phase techniques, 171
Tin-indium solders, specific types sweeping of, 133, 170(F)
Sn-40In, 15 volume reduction of, 171
Tin intermetallics, 88 Triode sputtering, 180
Tin-lead solder, 3132 Tungsten
Tin-lead solders, specific types of coefficient of thermal expansion (CTE), 160
61.75Sn-38.05Pb-0.2Cu(wt%), 84 diffusion bonding of, 10
Sn-40Pb, hermetically sealed as low-expansion metals, 160
enclosures, 44
Tin oxides
reduction in hydrogen atmospheres, 111 U
temperature for reduction of, 126
Tin pest Ultimate tensile strength of lead-free solders, 194
described, 7677, 195196 Ultrahigh vacuum system (UHV), 127
lead-free solders, 195196 Ultrasonic fluxing, 129, 130
sub-zero service temperature, 7677 Ultrasonic power and wetted area, 131(F)
susceptibility of puree tin-base solders to, 195196 Ultrasonic soldering, 128130
Tin-silver-rare earth solders Ultrasonic systems, 129130
Sn-3.5Ag-0.25RE, stress-rupture life of, 228(F) Ultrasound, defined, 235
Tin-silver solders, specific types Underbump metal, 200
Sn-3.5Ag, stress-rupture life of, 228(F) Underfill adhesive, 203
Tin whiskers Universal gas constant, 97, 124
lead-free solders, 195196 UNS K94610, 160
on PCBs, 196 Unsolderable materials, 120122
 2004 ASM International. All Rights Reserved.
Principles of Soldering (#06244G)
270 / Principles of Soldering
V Volume diffusion, 10
Vacancy diffusion, 173
Vacuum atmospheres
brasses and zinc metallization, 107
industrial quality, 107
soldering in, 107109
Vacuum bakeout, 172
Vacuum deposition process, 206
Vacuum evaporation, 180
Vacuum furnace, 104
Vacuum joining, 174
Vacuum system, 36
Van der Waals forces, 15
Vanadium, 104
Vanadium ions, 132
Vapor deposition, 31
Vapor deposition coating, 234
Vapor deposition technique, 148
Vapor-phase techniques, 171
Velocity of liquid flow, 19
Ventilation, 43
Viscosity
and fluid flow of solders, 18
and molecular weight of metals, 19
of selected conductive adhesives, 3(T)
temperature effects of, 25
theoretical, of solders, 19
Void content vs. joint length, 169(F)
Void-free joints, 183
Void levels
due to trapped gas, 174(F)
vs. load applied to preforms, 134(F)
and pressure variation, 174(F)
Void size, maximum acceptable, 183184
Voids
and carbon-fiber-loaded solders, 223
causes of, 169
control of, 173
and diffusion bonding, 910
effect of, on joint mechanical integrity, 29
with fluxed paste solder, 172173
formation of, through gas entrapment, 25
as function of component size, 169
and gap width, 15
incorporation of, 168
inspection methods for detection of, 183
in joint gap, 169(F), 170(F)
joint interfaces as source of, 28
in joints, 49
Kirkendall voids, 71, 234
in phase segregation, 74
relationships of, to flux conditions, 172173
Volatile organic compounds (VOCs), 118
environmental considerations from, 196
environmental considerations of, 115
Volatilization, 107
Volume contraction
cracks from, 67
of silver-tin alloys, 232
www.asminternational.org
Volume freezing, 173
Volume of spherical cap, 44
W
Water-soluble fluxes, 39, 40, 118
Water vapor
bakeout temperature of, 36
in hydrogen reducing atmosphere, 37
ingress into hermetically sealed package, 185(F)
as source of oxidation, 36
Water vapor desorption and oxygen partial pressure,
108
Wave soldering, 33, 114
Wave soldering machines, 77
Weight fraction of constituents, 96
Welding, 4
Welding process, 4
Wet back process, 206
Wet plating
for application of solderable metallizations, 148
metallization techniques, 180
on nonmetallic components, 149150
Wettability of parent materials, 49
Wetted area and ultrasonic power, 131(F)
Wetting
along surface valleys, 2223
assessment of, 207210
classical model of, 12, 14
difficulties of silica in, 69
driving force for, 16
effects of, by rare earth doping, 228229
and fillet formation, 6
of metals by solders, 147157
of nonmetals by solders, 149153
problems with, 17, 147
rate of, 18
refractory metals problems with, 147
with refractory parent material, 147
Wetting and contact angle, 1318
Wetting and spreading, 20, 135
Wetting and spreading characteristics, 220
Wetting angle, 14(F), 212
Wetting area, 1314
Wetting balance, 208(F)
Wetting balance solderability test, 208209
Wetting balance tests, 197
Wetting behavior, 210(F)
Wetting characteristics, 197
Wetting equation, 13, 18
for binary metal systems, 15
Wetting force
for commercial fluxes, 214(F)
of Sn-9Zn solders, 228(F)
during solderability test cycle, 209(F)
time of, to reach acceptance value, 197
Wetting front, 23
Wetting properties, 221(T)
 2004 ASM International. All Rights Reserved. www.asminternational.org
Principles of Soldering (#06244G)
Index / 271

Wetting rate, 198(F) galvanic corrosion of, 64

Wide-gap joints Zinc-aluminum alloys, specific type
brazing of, 179 90% Zn, 7% Al filler metal, 136
in ceramic components, 158 Zinc-aluminum-magnesium-gallium alloys, specific
composite solders and, 221 type
Wire bonding, 203 Zn-4Al-3Mg-3.2Ga quaternary, high-melting-point
Wire cross preform, 171(F), 172(F) solders, 197198
Work-hardened alloys, 49 Zn-4Al-3Mg-3.2Ga quaternary, high thermal
Work hardening, 49 conductivity of, 197198
WS type fluxes, 118 Zinc-base alloys, 64
Zinc-base solders
high-volume contraction of, 64
X as lead-free solders, 66
limitations of, 50
X-bar chart, 42 stress concentrations of, 64
X-radiography, 172(F) wetting additives for, 147
X-ray inspection and defects, 204 Zinc-bearing alloys, 67(F)
X-ray systems Zinc-bearing solders, 6164
fine-focus, 189 advantages of, 63
microfocus, 236237 Al-94Zn eutectic solders, 62
X-ray techniques, 183 for aluminum and aluminum alloys, 61
X-rays, cracks and orientation of, 236 cadmium alloy of, 63
common alloys of, 6263
fluxes for, 65
Y limitations of, 64
and reduced pressure atmospheres, 62, 65
Yield strength solidus temperature range of, 62
of composite solders, 220(F) thermal conductivity of, 63
of lead-free solders, 194 Zinc chloride, 117
Youngs equation, 13, 15 Zinc-containing alloys, 195
Youngs modulus, 223, 225
Zinc metallization, 107
Zinc oxide ceramic
metallizations of, 150
Z
metallized with titanium, 150
Z-axis control, 205 metallized with zirconium, 150
Zinc Zinc oxides
as additive to indium, 135 stability of, 125
as impurity of Au-2Si solders, 77 temperature for reduction of, 126
as solder constituent, 54 Zinc solders, 116
Zinc alloys Zirconium
Al-94Zn solders, 54 joining atmospheres for, 104
with aluminum, 54 for zinc oxide ceramic metallization, 150
 ASM International is the society for materials
engineers and scientists, a worldwide network
dedicated to advancing industry, technology, and
applications of metals and materials.

ASM International, Materials Park, Ohio, USA

www.asminternational.org
This publication is copyright ASM International. All rights reserved.
Publication title Product code
Principles of Soldering #06244G

To order products from ASM International:

Online Visit www.asminternational.org/bookstore

Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Fax 1-440-338-4634
Customer Service, ASM International
Mail
9639 Kinsman Rd, Materials Park, Ohio 44073-0002, USA
Email CustomerService@asminternational.org
American Technical Publishers Ltd.
27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX,
In Europe United Kingdom
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)
www.ameritech.co.uk
Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550

Terms of Use. This publication is being made available in PDF format as a benefit to members and
customers of ASM International. You may download and print a copy of this publication for your
personal use only. Other use and distribution is prohibited without the express written permission of
ASM International.
No warranties, express or implied, including, without limitation, warranties of merchantability or
fitness for a particular purpose, are given in connection with this publication. Although this
information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be
obtained from the use of this publication alone. This publication is intended for use by persons having
technical skill, at their sole discretion and risk. Since the conditions of product or material use are
outside of ASM's control, ASM assumes no liability or obligation in connection with any use of this
information. As with any material, evaluation of the material under end-use conditions prior to
specification is essential. Therefore, specific testing under actual conditions is recommended.
Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale,
use, or reproduction, in connection with any method, process, apparatus, product, composition, or
system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this
publication shall be construed as a defense against any alleged infringement of letters patent,
copyright, or trademark, or as a defense against liability for such infringement.