ASTM STP 1530 - Selected Technical Papers - Narayan Prabhu - ASTM Committee D-2 On Petroleum Products and Lubricants - Lead-Free Solders (2011)

Journal of ASTM International
Selected Technical Papers STP 1530

Lead-free Solders
JAI Guest Editor:

Narayan Prabu
ASTM International
100 Barr Harbor Drive
PO Box C700
West Conshohocken, PA 19428-2959
Printed in the U.S.A.
ASTM Stock #: STP1530

Library of Congress Cataloging-in-Publication Data
Lead-free solders / JAI guest editor, Narayan Prabu.
p. cm. -- (Journal of ASTM International selected technical papers; STP1530)
Includes bibliographical references and index.
ISBN: 978-0-8031-7516-7 (alk. paper)
1. Lead-free electronics manufacturing processes. 2. Solder and soldering--Materials. I.
Prabu, Narayan.
TK7836.L424 2011
621.9’77--dc22 2010053870
Copyright © 2011 ASTM INTERNATIONAL, West Conshohocken, PA. All rights

reserved. This material may not be reproduced or copied, in whole or in part, in any printed,
mechanical, electronic, film, or other distribution and storage media, without the
written consent of the publisher.
Journal of ASTM International „JAI… Scope
The JAI is a multi-disciplinary forum to serve the international scientific and engineering
community through the timely publication of the results of original research and
critical review articles in the physical and life sciences and engineering technologies.
These peer-reviewed papers cover diverse topics relevant to the science and research that
establish the foundation for standards development within ASTM International.
Photocopy Rights
Authorization to photocopy items for internal, personal, or educational classroom use, or
the internal, personal, or educational classroom use of specific clients, is granted by
ASTM International provided that the appropriate fee is paid to ASTM International, 100
Barr Harbor Drive, P.O. Box C700, West Conshohocken, PA 19428-2959, Tel:
610-832-9634; online: http://www.astm.org/copyright. The Society is not responsible, as
a body, for the statements and opinions expressed in this publication. ASTM
International does not endorse any products represented in this publication.
Peer Review Policy
Each paper published in this volume was evaluated by two peer reviewers and at least
one editor. The authors addressed all of the reviewers’ comments to the satisfaction of both
the technical editor(s) and the ASTM International Committee on Publications. The
quality of the papers in this publication reflects not only the obvious efforts of the authors
and the technical editor(s), but also the work of the peer reviewers. In keeping with
long-standing publication practices, ASTM International maintains the anonymity of the
peer reviewers. The ASTM International Committee on Publications acknowledges
with appreciation their dedication and contribution of time and effort on behalf of ASTM
International.
Citation of Papers
When citing papers from this publication, the appropriate citation includes the paper
authors, ⬘⬘paper title’’, J. ASTM Intl., volume and number, Paper doi, ASTM International,
West Conshohocken, PA, Paper, year listed in the footnote of the paper. A citation is
provided as a footnote on page one of each paper.
Printed in Bridgeport, NJ
February, 2011
Foreword
THIS COMPILATION OF THE JOURNAL OF ASTM INTERNATIONAL
(JAI), STP1530, on Lead-free Solders, contains papers published in
JAI encompassing the environmental and health concerns of the exposure
to lead during soldering and the success and failures of lead-free solders.
This STP is sponsored by ASTM Committee D02 on Petroleum Products and
Lubricants.
The JAI Guest Editor is Professor K. Narayan Prabhu, Department of
Metallurgical & Materials Engineering, National Institute of Technology
Karnataka, Surathkal, Mangalore, India.
Contents
Overview .......................................................................... vii
Wetting Behavior of Solders
G. Kumar and K. N. Prabhu .............................................. 1
A Review of Pb-Free High-Temperature Solders for Power-Semiconductor Devices:
Bi-Base Composite Solder and Zn–Al Base Solder
Y. Takaku, I. Ohnuma, Y. Yamada, Y. Yagi, I. Nakagawa, T. Atsumi, M. Shirai, and
K. Ishida ............................................................. 27
Wetting Behaviour and Evolution of Microstructure of Sn–Ag–Zn Solders on Copper
Substrates with Different Surface Textures
Satyanarayan and K. N. Prabhu ........................................... 50
Solder Joint Reliability of SnBi Finished TSOPs with Alloy 42 Lead-Frame under
Temperature Cycling
W. Wang, M. Osterman, D. Das, and M. Pecht ................................ 74
The Microstructural Aspect of the Ductile-to-Brittle Transition of Tin-Based Lead-Free
Solders
K. Lambrinou and W. Engelmaier .......................................... 89
Loading Mixity on the Interfacial Failure Mode in Lead-Free Solder Joint
F. Gao, J. Jing, F. Z. Liang, R. L. Williams, and J. Qu ........................... 121
Ball Grid Array Lead-Free Solder Joint Strength under Monotonic Flexural Load
P. Geng. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Tensile Properties of Sn-10Sb-5Cu High Temperature Lead Free Solder
Q. Zeng, J. Guo, X. Gu, Q. Zhu, and X. Liu. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
Empirical Modeling of the Time-Dependent Structural Build-up of Lead-Free Solder
Pastes Used in the Electronics Assembly Applications
S. Mallik, N. N. Ekere, and R. Bhatti ........................................ 168
Rheological Characterisation and Empirical Modelling of Lead-Free Solder Pastes and
Isotropic Conductive Adhesive Pastes
R. Durairaj, L. W. Man, and S. Ramesh ...................................... 186
Reprinted from JAI, Vol. 7, No. 5
doi:10.1520/JAI103055
Available online at www.astm.org/JAI
Girish Kumar1 and K. Narayan Prabhu2
Wetting Behavior of Solders
ABSTRACT: Lead bearing solders have been used extensively in the assem-
bly of modern electronic circuits. However, increasing environmental and
health concerns about the toxicity of lead has led to the development of
lead-free solders. Wetting of solders on surfaces is a complex and important
phenomenon that affects the interfacial microstructure and hence the reliabil-
ity of a solder joint. The solder material reacts with a small amount of the
base metal and wets the metal by intermetallic compound 共IMC兲 formation.
The degree and rate of wetting are the two important parameters that char-
acterize the wetting phenomenon. Contact angle is a measure of the degree
of wetting or wettability of a surface by a liquid. Spreading kinetics in a given
system is strongly affected by the experimental conditions. In reactive sys-
tems like soldering, wetting and chemical interfacial reactions are interre-
lated, and hence for successful modeling, it is essential to assess the effect
of interfacial reactions on kinetics of wetting. Solder wetting necessarily in-
volves the metallurgical reactions between the filler metal and the base
metal. This interaction at the solder/base metal interface results in the for-
mation of IMCs. During soldering an additional driving force besides the
imbalance in interfacial energies originates from the interfacial reactions. The
formation of IMC has significant influence on contact angle. The presence of
IMCs 共thin, continuous, and uniform layer兲 between solders and substrate
metals is an essential requirement for good bonding. Optimum thickness of
an IMC layer offers better wettability and an excellent solder joint reliability.
However, due to their inherent brittle nature and tendency to generate struc-
tural defects, a too thick IMC layer at the interface may degrade the joint. In
Manuscript received February 25, 2010; accepted for publication April 15, 2010; pub-
lished online June 2010.
1
Dept. of Mechanical Engineering, St. Joseph Engineering College, Mangalore 575028,
India, e-mail: srigk71@gmail.com
2
Dept. of Metallurgical and Materials Engineering, National Institute of Technology
Karnataka, Surathkal, Mangalore 575025, India, e-mail: prabhukn_2002@yahoo.co.in
Cite as: Kumar, G. and Prabhu, K. N., ‘‘Wetting Behavior of Solders,’’ J. ASTM Intl., Vol.
7, No. 5. doi:10.1520/JAI103055.
Copyright © 2010 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West
Conshohocken, PA 19428-2959.
1
2 JAI • STP 1530 ON LEAD-FREE SOLDERS
this paper, the factors affecting the wetting behavior of solders and evolution
of interfacial microstructure are reviewed and discussed.
KEYWORDS: wetting, lead-free solders, IMC, microstructure
Introduction
Soldering is a low temperature process of joining metallic materials by using a
filler material, known as solder, that melts below 450° C. It is a milder form of
achieving metallurgical continuity. It involves several fields of science such as
mechanics, chemistry, metallurgy, etc. The process consists of placement of
parts to be joined, wetting the surfaces with molten solder, and allowing it to
cool until solidification. The solder reacts with a small amount of the base
metal and wets the metal by intermetallic compound 共IMC兲 formation. Thus,
the important feature of the soldered joint is that a metallurgical bond is pro-
duced at the filler metal/base metal interface. There are several advantages of
soldering compared to other joining methods. Some of them are the simplicity
of the process, economy, easy repair works, etc. In the electronics field, solder
plays a crucial role by providing electrical, thermal, and mechanical continu-
ities in electronic assemblies 关1–3兴.
Conventionally, solders are the alloys of tin 共Sn兲 and lead 共Pb兲. The use of
lead in solders dates back more than 5000 years 关2兴. Lead bearing solders 共eu-
tectic and near eutectic兲 have been used extensively in the assembly of modern
electronic circuits. Pb reduces the surface tension of pure tin, and lower surface
tension of Sn–Pb alloy facilitates wetting. As an impurity, even at very low
concentrations lead prevents the transformation of white tin and thereby main-
tains the structural integrity of Sn based alloys. Also Pb acts as a solvent facili-
tating intermetallic formation between Sn and Cu rapidly 关1–4兴.
Flux is a combination of organic and inorganic chemicals formulated to
react with metal oxides at soldering temperatures and eliminate them. The
reason for the use of flux is that an oxide interface between solder and substrate
acts as a barrier for the good bond. Hence, the major function of the flux is to
chemically clean the surface and keep it in the proper state till the completion
of soldering process. Although the flux should be sufficiently active to remove
any strong oxide, it should not leave any residue at the end of soldering, or the
residue should be cleanable without much difficulty. There are two basic cat-
egories of fluxes, namely, rosin based flux and water soluble flux. Again on the
basis of activeness, there are mild fluxes, medium active fluxes, and fully acti-
vated fluxes.
Lead-Free Solders
Lead contributes many of the desired properties of solders such as cost, avail-
ability, and performance. However, its toxicity towards humans and wildlife has
warranted the elimination of lead from solders. There are some characteristics
that play a major role in the consideration of substituting tin-lead solders in
electronic soldering—lower melting temperature, adequate strength, and envi-
ronmental issues related to the toxicity, good electrical/thermal conductivity,
KUMAR AND PRABHU, doi:10.1520/JAI103055 3
TABLE 1—Potential lead-free candidate alloys 关4,6兴.
Alloy Melting Point 共°C兲 Relative Price Index Remarks

Sn-37Pb 183 1 Binary eutectic
Sn-58Bi 138 1.32 Binary eutectic
Sn-9Zn 198 1.39 Binary eutectic
Sn-3.5Ag 221 2.70 Binary eutectic
Sn-0.7Cu 227 1.50 Binary eutectic
Sn-3.8Ag-0.7Cu 217 2.80 Ternary eutectic
Sn-3.5Ag-1.5In 218 3.29 Ternary eutectic
Sn-3.5Ag-4.8Bi 205–210 2.68 Non-eutectic
Sn-2.5Ag-0.8Cu-0.5Sb 213–218 2.35 Non-eutectic
low cost, ease of handling, good ductility and workability, and excellent wetting
properties 关3,5兴. Majority of the alternatives has Sn as the major component
because tin has relatively low melting point and has a long history as a solder
constituent. The candidate lead-free solders include Sn–Ag, Sn–Bi, Sn–Zn, and
Sn–Cu with other minor additions 共eutectics and near eutectics兲关3兴. Silver is
added in small quantity with tin to form the eutectic, and the alloy offers higher
strength and fatigue resistance than conventional solder. Copper is abundantly
available, relatively cheap, and soluble in tin. Zn, although has problems of
oxidation, with tin forms a eutectic solder that has melting point very close to
conventional solder. Due to the scarcity of In and Bi, the maximum limit of
these elements in lead-free solders should be within 0.5 % and 15 %, respec-
tively 关1–3兴.
Since the properties of binary lead-free solder alloys cannot fully meet the
requirements for applications in electronic packaging, additional alloying ele-
ments are generally added to improve the performance of these alloys. Thus,
ternary and quartenary lead-free solders have also been developed. Some of
these alloys offer advantages such as higher joint strength, better fatigue resis-
tance, etc. over conventional solders 关1,3,4,6兴.
Some of the important lead-free solders are listed in Table 1.
A drop-in lead-free solder alloy must exhibit various desirable material
characteristics in terms of melting temperature, wettability, electrical and ther-
mal conductivity, thermal expansion coefficient, mechanical strength, ductility,
creep resistance, thermal fatigue resistance, manufacturability, and cost. How-
ever, wettability is the basis of all soldering processes 关7兴. One of the most
sensitive parameters for the quality of soldered joint is soldering temperature
as the current processing equipments are optimized for Sn–Pb solders. The
temperature margin available with conventional solder is about 50° C for re-
flow soldering, whereas it is only 30° C with lead-free solders since the melting
point of lead-free solders is significantly higher as seen from Table 1. Thus the
process window is narrower for lead-free solders. Another problem with lead-
free solders is their inferior wetting behavior 关8兴.
Thus, the important differences between lead-tin and lead-free solders are
summarized as follows.
• Most of the lead-free solders have liquidus temperature 20– 40° C higher
than that of Sn–Pb eutectic. As a result, the soldering temperature is

also higher for lead-free soldering. However, Sn–Bi and Sn–In are the
exceptions.
• Conventional Sn–Pb solder alloy is known for its excellent wetting per-
formance with different substrate/plating materials. But, lead-free alloys
show poor wetting, which requires the use of more active fluxes.
• The constituent materials of lead-free solders are costlier than conven-
tional Sn–Pb solders. Hence, the cost of the lead-free alloys will be at
least by two to three times higher than the tin-lead eutectic alloys.
• The microstructure of conventional solder is lamellar consisting of
phases—lead rich and tin rich phases. On the other hand, majority of
lead-free solders possess composite-like microstructure having a matrix
phase 共generally Sn兲 and a secondary phase, an IMC of Sn with other
elements 关1兴.
Common Lead-Free Solder Systems
Sn–Ag system is considered as first choice for lead-free solder, and the eutectic
temperature is found to be 221° C with composition Sn-3.5Ag. The processing
temperatures for Sn–Ag solder alloys are 235– 250° C for reflow soldering and
250– 260° C for wave soldering. They possess excellent mechanical properties.
Soldering temperatures for these alloys can be lowered by the addition of Bi.
However, lift-off is a common problem with these solders due to partial melting
reaction of Pb and Bi at 139° C. Further, soldering temperature is significantly
higher than conventional solders, and they are poorly compatible with general
printed wiring board finishes. The microstructure of Sn–Ag alloys contains fine
dispersion of Ag3Sn needles in the primary ␤-Sn grains. Ag3Sn precipitates
form a unique crystallographic relationship with ␤-Sn matrix, which produces
good interface bonding and accounts for excellent mechanical properties 关1兴.
Figure 1 shows the photomicrograph of Sn-3.5Ag solder. To improve the prop-
erties of Sn-3.5Ag solder, additional elements such as Bi, Cu, In, and Zn are
added. These additional elements lower the liquidus temperature and improve
tensile strength. For example, the addition of about 9 % In lowered the liquidus
temperature below 210° C and also resulted in a change of microstructures 关9兴.
Sn-58%Bi eutectic solder is known as useful low temperature solder 共eutec-
tic temperature: 139° C兲. However, a low temperature ternary eutectic of Bi, Pb,
and Sn is known to form at 96° C, and hence, the alloy may fail at general
service temperatures even if a small quantity of lead is present in solder or
plated base metal. Sn–Bi solders have reflow processing temperature of
180– 200° C, which is lower than conventional Sn–Pb solders. Addition of Ag
improves the brittle microstructure of these solders. However, the lead con-
tamination from conventional Sn–Pb plating, both on components and on cir-
cuit boards causes serious problems due to the formation of low temperature
phases 关1兴.
Among the various low temperature solders, Sn–Zn solders appear as
drop-in substitutes due to their identical processing temperature as that for
conventional tin-lead solders and good availability 关5,10兴. Sn-9Zn alloy is the
FIG. 1—Photomicrograph of the Sn-3.5Ag eutectic solder 关3兴.
typical eutectic alloy with eutectic composition of Sn-8.8 wt % Zn and tempera-

ture of 198° C. This looks appropriate from the melting point of view, but the
severe oxidation and poor heat resistance limit its use 关1,3,4,11兴. These solders
have significant benefit on cost as well as excellent mechanical properties.
However, poor oxidation resistance in humid/high temperature condition and
poor compatibility with Cu substrate at elevated temperatures are the major
drawbacks. Bi is usually added to Sn–Zn binary alloy in order to improve wet-
ting and to lower melting temperature. But the addition of Bi in excess of 6 %
enhances the brittleness of the alloy. Tensile, creep, and fatigue properties of
these alloys are excellent 关12兴. Al with eutectic Sn–Zn solder has a eutectic
temperature of 199.7° C and offers improved oxidation resistance 关13兴. The
microstructure of Sn–Zn eutectic has two phases—a body centered tetragonal
Sn-matrix phase and a secondary phase of hexagonal Zn containing less than 1
% tin in solid solution. Both elements hardly dissolve in each other 关12兴. The
solidified microstructure exhibits a fine uniform two-phase eutectic morphol-
ogy as seen from Fig. 2 关3,14兴.
Sn–Cu solders are only about 30% costlier than Sn–Pb solders. But their
soldering temperature is high 共about 260° C兲 because of higher eutectic tem-
perature 共227° C兲关1兴. This would cause the deterioration of the board material
in wave soldering. Further, Sn-0.7Cu eutectic solder suffers from the problem
of tin whisker growth.
Sn–Ag–Cu 共SAC兲 is one of the most frequently mentioned potential candi-
date systems for Pb-free solder with a melting point lower than Sn–Ag and
Sn–Cu systems due to its excellent mechanical properties and good wettability
FIG. 2—Photomicrograph of the Sn-9Zn eutectic solder 关14兴.
on Cu substrate 关10,15,16兴. These ternary alloys are known to possess good

solderability and mechanical properties and have been widely used as reliable
lead-free solders. However, the melting temperature for the alloys of this system
is about 30° C greater than that of eutectic Sn–Pb solder. Accordingly reflow
temperature must be increased which may rise manufacturing concerns.
Hence, high melting temperatures of these alloys limit the adoption of these
alloys to certain applications such as temperature sensitive components, op-
toelctronics modules, etc. This is due to the fact that excess heating during
reflow treatment induces some damage to electronic devices and thus affects
the reliability of the component. In addition Cu content of the solder can ac-
celerate the IMC formation between Ni and Sn and reduces fatigue life 关12,13兴.
Sn–Ag–Cu ternary solder alloy is found to have a eutectic composition of Sn-
3.5Ag-0.7/0.9Cu and eutectic temperature of 217° C. This alloy can be used for
wave, reflow, and rework/repair soldering processes and hence provides a best
all-around solution for lead-free soldering. On solidification under conditions
similar to those found with real electronic assembly joints, the SAC alloys ex-
hibit eutectic-like microstructures with a tin matrix containing distributed dis-
persed particles of Ag3Sn and/or Cu6Sn5 depending on the alloy and the sub-
strate material in contact with it. In conventional lead-tin solders, the lead rich
phase coarsens on ageing, whereas in SAC alloys the dispersed phase tended to
refine and spheroidize on ageing. The alloys were found to soften by about 20 %
on ageing at 125° C for 1000 h. The binary and ternary alloys of Sn with Ag and
Cu have similar microstructures. Coarse ␤-Sn phase is found to form at cooling
rates of 15– 20 K · s−1 关3兴. Sn–Ag–Cu ternary solders are alloyed with a trace
amount of Ni in order to further increase its wettability and mechanical
strength. Microalloying with Ge is also found beneficial in improving mechani-
cal properties and lowering dross formation 关15兴.
Soldering Methods
The simplest method of making a solder joint is to do it manually using solder-
ing iron. It is also known as hand soldering and used in a number of applica-
tions including modification/repair works and touch up/finishing jobs. The en-
tire soldering process can be automated to meet the demands of mass
production. Drag soldering, wave soldering, and reflow soldering are some of
automated soldering techniques. In drag soldering, the boards stuffed with
through-hole components are dragged on the surface of a reservoir of flux,
followed by a reservoir of molten solder bath. The most important trouble of
drag soldering is the formation of dross on the surface of the molten solder
bath, which is a barrier for obtaining a good solder joints. However, this prob-
lem can be overcome by using a standing wave of molten solder bath, and the
technique is known as wave soldering. Here, the stuffed board gently rides on
the crest of the standing wave of molten solder during which the leads pick up
adequate quantities of solder by wetting and wicking action. Although wave
soldering can be used satisfactorily for many components including surface
mount components, such components should be mechanically anchored using
adhesive before wave soldering to negate the turbulence of molten solder wave.
Further, components close to each other and tall components pick up the prob-
lem of solder shorts. The better solution to the soldering of surface mount
components is reflow soldering. As the name implies, the solder made to fuse
and reflow at desired locations. For this, the solder is to be formulated in the
form of a paste, applied at desired locations. The heat is applied so that solder
paste fuses and reflows to make a joint. Depending on the method of applica-
tion of heat, the reflow soldering can be classified as infrared reflow, laser re-
flow, vapor phase soldering, etc. 关17–19兴.
Wetting and Intermetallics

To form a proper metallurgical bond between two metals, wetting must take
place. This means that a specific interaction must take place between the liquid
solder and the solid surface of the parts to be soldered. The ability of the molten
solder to flow or spread during the soldering process is of prime importance for
the formation of a proper metallic bond. Thus, the basic soldering process
depends on wetting for the formation of solder-to-base metal contact. The so-
lidification of molten solder after wetting results in permanent bond. Therefore,
the solderable surfaces must allow the molten solder to wet and spread within
the available time 关18,19兴.
Wetting of a solid by a liquid is a surface phenomenon in which the surface
of the solid is covered by the liquid on placing it over the surface. By definition,
wetting is a measure of the ability of a material, generally a liquid, to spread
over another material, usually a solid. Spreading is a physical process through
which liquid wets the surface. It can be defined as the increase in the area of
coverage by the liquid with respect to time on placing a drop of liquid on the
surface.
The extent of wetting is measured by the contact angle that is formed at the
juncture of a solid and liquid in a particular environment, as shown in Fig. 3. In
FIG. 3—Sessile drop on a solid substrate.
general, if the wetting or contact angle lies between 0° and 90°, the system is
said to wet, and if the wetting angle is between 90° and 180°, the system is
considered to be non-wetting. The contact angle 共␪兲 is determined from the
balance of surface tensions at the juncture, according to the Young–Dupre
equation
␥sv = ␥sl + ␥lv cos ␪ 共1兲
where:
␥sv = surface tension of the solid in the particular environment,
␥sl = interfacial energy 共surface tension兲 between the solid and the liquid,
and
␥lv = surface tension of the liquid in the same environment.
Figure 3 shows the situation, while Fig. 4 is a schematic representation of
wetting and non-wetting with contact angles greater than and lower than 90°,
respectively.
There are two important parameters to characterize the wettability of a
liquid on a solid. They are the degree or extent of wetting and the rate of
wetting. The degree of wetting is generally indicated by the contact angle
formed at the interface between solid and liquid. In the equilibrium case, it is
governed by the laws of thermodynamics. It is dependent on the surface and
interfacial energies involved at the solid/liquid interface. The rate of wetting
indicates how fast the liquid wets the surface and spreads over the same. It is
guided by a number of factors such as the thermal conditions of the system,
capillary forces, viscosity of the liquid, the chemical reactions occurring at the
interface, etc. 关17,18兴.
The wettability measurement has an important role in wetting studies. Re-
liable and reproducible contact angle value should be available from experi-
ments in order to analyze the behavior. Various methods have been developed
over the years to evaluate wettability of a solid by a liquid. Among these, sessile
drop and wetting balance techniques are versatile and popular and provide
reliable data.
FIG. 4—Wetting and non-wetting conditions.
Intermetallic Compound Formation

Solder wetting involves the metallurgical reactions between the filler metal and
the base metal. This interaction at the solder/base metal 共substrate兲 interface
results in the formation of IMCs such as Cu6Sn5, Cu3Sn, AuSn4, Ni3Sn4, etc.
IMCs mainly form due to the interfacial reactions between Sn from the solder
and plating material 共Cu, Ni, Au, etc.兲. The formation of this layer occurs due to
the wetting reaction between the solder and the substrate above the melting
point, whereas its growth occurs by a solid state reaction process below the
melting point. The growth of intermetallic layer can be expressed by simple
Arrhenius type of equations because diffusion of Sn plays a key role in the
formation of reaction layers 关1兴. Figure 5 shows the solder/substrate joint where
the formation of intermetallic is shown schematically. Figure 6 is the actual
photomicrograph of the solder/substrate interface.
During soldering an additional driving force besides the imbalance in in-
terfacial energies originates from the interfacial reactions. The formation of
FIG. 5—Schematic sketch of a solder/substrate interface 关20兴.
IMC has significant influence on contact angle 关22兴. It is found that the wetting
of solid metal by molten solder/metal will be better if intermetallic formation
exists because the formation of IMCs can affect wetting since it alters the value
of ␥sl. Thus the presence of IMCs 共thin, continuous, and uniform layer兲 between
solders and substrate metals is an essential requirement for good bonding 关23兴.
An excellent wettability makes the Cu substrate react with the solder alloy com-
pletely, and IMCs form at the interface between them. Optimum thickness of an
IMC layer offers a better wettability and an excellent solder joint reliability.
However, due to their inherent brittle nature and tendency to generate struc-
tural defects, too thick IMC layer at the interface may degrade the joint. The
optimum thickness of Cu6Sn5 between Cu and 63Sn-37Pb solder alloy was
proposed to be 5 – 7 ␮m 关16,23兴.
The intermetallic reaction layer formation takes place during soldering in
three consecutive stages of dissolution of the base metal, chemical reaction,
and solidification. However, for this to occur, flux has to facilitate wetting by
removing oxide layers 关23兴. For most of the electronic applications, Sn based
FIG. 6—Photomicrograph of a solder/substrate interface 关21兴.

solder is used on Cu substrate 关24兴. When Sn based solder is in contact with Cu

substrate, interfacial reactions occur and the main IMCs formed at the inter-
faces between the Cu substrate and various solder alloys are scallop or penin-
sular shaped ␩-Cu6Sn5 and a planar or layer type ␧-Cu3Sn. Generally, the
former is thicker than the latter.
The type of surface finishes 共platings兲 strongly influence interface reaction
and wetting. Au/Ni, Pd/Ni, and Au/Pd/Ni systems are three of the important
plating materials used as surface finishes. As a result reaction systems such as
Sn–Ni, Sn–Pd, and Sn–Au appear at the surface finish/solder interface 关12兴.
Factors Affecting Solder Wetting and Evolution of Microstructure

The wetting of solid by a liquid is a complex phenomenon sensitive to a large
number of factors. Reactive wetting process such as soldering is affected by
many factors such as material composition, temperature, flux usage, etc. 关25兴.
The important factors that affect the wetting and solidification behaviors of
solders on a substrate are briefly discussed below.
Solder/Substrate Types
The type of the solder as well as the type of substrate are the most important
factors in affecting the wettability and interfacial microstructure. It is well
known that the conventional Sn–Pb solders possess excellent wetting on gen-
eral base materials like Cu and its alloys. Pb reduces the surface tension of pure
tin and lower surface tension Sn–Pb alloy facilitates wetting. As an impurity,
even at very low concentrations lead prevents the transformation of white tin
and thereby maintains the structural integrity of Sn based alloys. Also Pb acts
as a solvent facilitating intermetallic formation between Sn and Cu rapidly
关17,19兴.
Since there is no drop-in substitute for lead-tin solder, different lead-free
solders are used in different situations depending on the method used and type
of component. Hence, the wettability of these solders will not be identical on
the given substrate. Further, the practical soldering of electronic products like
lead frames, PCBs, and semiconductor chips requires suitable metallization to
provide desired functions. The metallizations may include wetting layer and
diffusion barrier. For example, Cu is a common lead frame material, which is
electroplated with Ni or coated with electroless Ni to prevent solder interaction
with Cu. The Ni plating surface may be further modified with a wetting layer to
enhance solderability 关26兴. Thus substrate surface composition is also a vari-
able that affects the wetting and microstructure.
The lead-free solders exhibit inferior wetting properties compared to lead
based solders on most of the substrate surfaces. For example, a contact angle of
17° was reported for Sn-37Pb solder on a copper substrate, whereas the corre-
sponding contact angles under the similar conditions on copper substrate for
Sn-3.5Ag are 36° and 43° for Sn-58Bi 关2兴. Table 2 given here lists contact angle
data reported in the literature for various solder and substrate compositions.
Alloying additions to binary and ternary lead-free solders are generally
made in order to improve wetting and mechanical properties. In fact, it is a
TABLE 2—Contact angle data for various solder/substrate compositions.
Solder/Substrate Condition Contact Angle Reference

Sn-37Pb/Cu 260° C 17 2
Sn-3.5Ag/Cu 260° C 36
Sn-5Sb/Cu 280° C 43
Sn-58Bi/Cu 195° C 43
215, 230, and 245° C 共RMAa flux is used in all
Sn-50In/Cu cases兲 63, 41, and 33
Sn-3.5Ag/Cu 250 and 270° C 70 and 50 27
Sn-3Ag-0.5Cu/Cu 55, ¯
Sn-3.6Ag-0.7Cu/Cu 55, 40
Sn-4Ag-0.5Cu/Cu 55, 40
Sn-2.5Ag-1Bi-0.5Cu/Cu 55, 52
Sn-2.5Ag-0.7Cu/Cu x = 0 %, 0.1 %, 0.25 % 53 28
Sn-3.5Ag-0.7Cu-xRE/ Cu 48, 41, and 46
Cu= 0 %, 0.5 %, 1 %, 2 %, 4 %, 6 %, 8 %,
Sn-9Zn-xCu/ Cu and 10 % 120, 124, 105, 75, 72, 63, 59, and 54 29
Sn-37Pb/Cu ¯ 10 30
Sn-4Ag-0.5Cu /Cu 30
Sn-37Pb/Cu-Ni-Au 7
Sn-4Ag-0.5Cu/Cu-Ni-Au 27
Sn-37Pb/UBM-1,2,3,4b ¯ 61, 10, 64, and 62 31
Sn-3.5Ag/UBM-1,2,3,4b 64, 27, 67, and 60
Sn-3Ag-xBi/ Fe-42Ni 250° C 70–85 32
450° C 共x = 0 %, 3 %, and 6 %兲 50–65
Sn-37Pb/Cu 250° C 13–32 13
250– 280° C 共x = 0 %, 0.5 %, and 0.75 %;
Sn-3.5Ag-xCu/ Cu different types flux and surface roughness兲 28–55
TABLE 2— 共Continued.兲
Solder/Substrate Condition Contact Angle Reference

60Sn-In-xBi/ Cu 272° C 共x = 5 %, 10 %, 20 %, and 40 %兲 25, 21, 19, and 15 33
Sn-37Pb/Cu 270° C 23 34
Sn-3.5Ag/Cu 43
Sn-3.5Ag/Cu Halide-free flux 43
Sn-3.5Ag/Cu Halide flux 38
Sn-3.5Ag/Cu 1 % Cu 42
Sn-3.5Ag/Cu 5 % In 41
Sn-3.5Ag/Cu 5 % Bi 38
Sn-3.5Ag/Cu 1 % Zn 48
Sn-3.5Ag/Cu Au plated 29
¯

Sn-37Pb/Cu 11.1 35
Sn-0.5Cu/Cu 33.9
Sn-3.5Ag/Cu 34.2
Sn-0.7Cu-xZn/ Cu x = 0 %, 0.2 %, 0.6 %, and 1 % 42, 46, 52, and 50 36
a
RMA: Rosin mildly activated.
b
UBM 1: Au共500 Å兲/Cu共1000 Å兲/Cr共700 Å兲; UBM 2: Cu共5000 Å兲/Cr共700 Å兲; UBM 3: Au共500 Å兲/Cu共5000 Å兲/Cr共700 Å兲; and UBM 4: Au共500
Å兲/Cr共1000 Å兲/Ti共700 Å兲.
method to improve the wetting of a given substrate by a given metal not only in
soldering but also in brazing. The spreading metal is alloyed with a chemical
species that reacts with the substrate to form a dense layer of solid reaction
product. This reaction product is better wetted by the metal than the original
substrate 关37兴.
Sn–Zn solder suffers from easy oxidation and relatively poor wettability.
Generally, there are two methods to improve the wettability of Sn–Zn solder:
One is to develop a new kind of flux that suits Sn-Zn solder, and another is to
improve oxidation resistance of the solder by alloying. Evidence was reported
on the improvement of oxidation resistance, and hence wettability of this sol-
der on Cu substrate by adding Cr, Al, Cu, In, Ag, P, and La. Addition of Ga was
found to be useful 关38兴.
The investigations on the effect of alloying additions to Sn–Zn lead-free
solder revealed that the addition of 1 wt % Ag did not significantly affect wet-
ting property of the Sn-9Zn solder on Cu substrate. On the other hand, the
addition of Al to the above alloy significantly improved wetting. It was found
that the surface tension of Sn-9Zn alloy increased with the addition of Ag as
well as Al. But wettability improvement on Cu was noticed only for Al addition.
It was proposed that Al might enhance the reaction between Cu and solder,
thereby reducing the contact angle. Al tends to diffuse to the interface and
enhancee the wetting between solder and Cu 关7兴.
Chen et al. investigated the influence of gallium addition on wettability of
Sn–Zn–Ag and Sn–Zn–Ag–Al lead-free solder alloys 关39兴. The selection of Sn–Zn
alloys was made since the melting point/liquidus of Sn–Zn alloys is very close to
traditional Sn–Pb eutectic solders. It has been shown in their experiments that
the increase in gallium addition resulted in significant reduction in wetting
time and increase in wetting force. Wetting time reduced from 2.5 s at 0 wt %
Ga to 0.5 s at 3 wt % Ga in the alloy. Similarly, the spreading area of the
Sn-Zn-xGa solder significantly increased with increased Ga addition. Maxi-
mum improvement of nearly 20 % was observed at 0.5 % addition, while the
further increase in Ga addition resulted in marginal improvement. However,
Ga is reported to decrease the microhardness and increase the pasty range at
concentrations⬎ 2 wt %. Thermogravimetric analysis and Auger Electron
Spectroscopy analysis showed that the oxidation resistance of the alloy has
been improved greatly as a result of compact Ga-rich protective film. This re-
sulted in the reduction in ␥LF value and improved wettability 关38兴.
Wang et al. investigated the effect of Zn addition to lead-free Sn-0.7Cu
solder alloy on the wetting behavior of the alloy with Cu substrate 关36兴. The Zn
is varied in the range of 0–1 wt % in the alloy during wetting balance experi-
ments. A deterioration of wetting is observed in their experiments as contact
angle increased from 42° for zero zinc addition to 50° at 1 % Zn addition to the
alloy. But the addition of copper to Sn-9Zn solder alloy was found to be very
much beneficial 关28兴. It was reported that 10 % addition of copper would bring
down the contact angle of the solder on Cu substrate from non-wetting situa-
tion of 120± 8° to 54± 6° 共see Fig. 7兲 in which 0.25–1 % addition brought a
largest drop in the contact angle. It was also reported that the addition of Bi
关40兴 or Al 关41兴 improved the wettability of Sn–Zn solders. The addition of Bi to
FIG. 7—Effect of copper addition on contact angle of the Sn–Zn solder 关28兴.
the conventional Sn-60Pb is found to decrease the surface tension of solder

alloy as well as help in the prevention of dewetting by lowering the surface
concentration of Pb atoms 关42兴.
The investigations on the addition of In to Sn–Ag alloy showed that wetting
angle between copper substrate and solder decreases with increasing amounts
of In 关9兴. Investigations on rare earth 共RE兲 addition to lead-free solders indi-
cated that Ce and La additions improved wetting in most of the lead-free alloys
关3,28,43兴. For example, contact angle of Sn–Ag eutectic alloy reduced from 47°
to 41° on addition of 0.25 % RE. Similar results were found during the RE
addition to Sn–Ag–Cu and Sn–Zn alloys. However, higher RE addition deterio-
rated wetting behavior by lowering the wetting force and by increasing contact
angle. The addition of RE is found to reduce the surface tension between the
solder and the flux since it accumulates at the solder/flux interface, thereby
reducing the contact angle 关3兴. But, at higher amounts they increase the viscos-
ity of the molten solder, tend to oxidize easily, and hence adversely affect the
wetting. Figure 8 shows the effect of RE addition on contact angle of Sn–Ag
solders.
Wettability and mechanical properties improved significantly for the Sn-
3.8Ag-0.7Cu solder with the addition of Pr. Optimal quantity of Pr was found to
be 0.05 wt %, while the properties start deteriorating when Pr exceeds 0.25
wt % 关44兴. Figure 9 shows the effect of Pr addition on the spread area of the
Sn–Ag–Cu solder.
FIG. 8—Effect of RE addition on contact angle of the Sn–Ag solder 关43兴.
FIG. 9—Effect of Pr addition on spread area of the Sn–Ag–Cu solder 关44兴.

Roughness
Rough surfaces have a significant influence on the wetting behavior of fluids. A
rough surface provides an additional interfacial area for the spreading liquid,
and the true contact angle would be different than the nominal contact angle.
The additional surface area provided by roughening the surface results in the
increase of surface energy. Wenzel studied the effect of surface roughness on
the equilibrium contact angle and proposed an equation that gives a relation
between equilibrium contact angle and the apparent angle formed on a rough
surface 关13,45–49兴
cos ␪w = r cos ␪ 共2兲

where:
␪ = equilibrium contact angle,
␪w = apparent contact angle on a rough surface 共generally known as Wenzel
angle兲, and
r = average roughness ratio, the factor by which roughness increases the
solid-liquid interfacial area.
Hence r is the ratio of actual wetted surface area to projected or geometric
surface area calculated from radius of wetted base. Its value is always greater
than unity except for ideally smooth surfaces for which it becomes equal to
unity. In reactive systems like solder wetting on a substrate, the roughness has
additional effects. The asperities and grooves may act as preferable sites for
reaction, diffusion, adsorption, nucleation, etc. Hence, it would be difficult to
assess the effect of roughness alone on wetting behavior of metallic liquids as
the main effects are masked by other factors.
Yost et al. investigated the wetting behavior of Sn–Pb eutectic solders on
very rough 共⬎50 ␮m兲 copper surfaces 共Ni substrates electroplated with cop-
per兲 and concluded that for extensive wetting ␣ ⬎ ␪. They opined that rough/
grooved surfaces provide an additional driving force for wetting as liquid solder
flows into the valleys by capillary action 关47兴. Lin et al. carried out a number of
experiments on wetting of di-ionized water, Sn–Pb, and Pb-free solders on sur-
faces of different roughnesses, with Ra values varying from 98 to 297 nm 关13兴.
They observed a general trend of decreasing contact angle with increasing sur-
face roughness, as predicted by Wenzel. However, a large scatter was found in
their experiments.
Flux
The flux plays a vital role in wetting/spreading of solders. The reason is the
oxidation of surface of the substrate as well as the liquid solder. The breakdown
of oxide film is vital to achieve true wetting in any system since the film present
on the substrate surface or spreading liquid will alter the interfacial properties.
Thus, to overcome the barrier effects of oxide films, fluxes are generally used
关50兴.
In soldering fluxes keep the solderable surfaces clean and tarnish-free and
influence the surface tension of solder in the direction of solder spreading by
decreasing the contact angle 关2,19,51,52兴. The reactive surfaces present on the
liquid solder and clean metal substrate are highly susceptible to contamination
through adsorption, reaction, and diffusion processes. The flux generally re-
moves oxide layers from substrate and solder surface and improves wetting
关50,53兴.
Fluxes can be broadly classified into two categories, viz., inorganic and
organic. The first category includes inorganic acids, salts, and gases. These
fluxes are not only fast acting but also corrosive in nature. Hence, cleaning is
necessary after their use. Organic fluxes are comparatively milder than inor-
ganic ones. They are either rosin base or resin base fluxes. These fluxes gener-
ally contain a small quantity of activators so as to be used successfully in gen-
eral applications. No-clean fluxes have also been developed, which do not
require post cleaning operations.
Takao et al. performed studies on the action of flux on wettability of lead
based and lead-free solders 关34兴. Contact angles as well as interfacial tensions
were measured in their investigations. The use of a halogen containing acti-
vated flux during wetting test of Sn-3.5Ag on Cu substrate resulted in a decrease
of 5° in the contact angle and 0.064 N/m in the value of liquid solder/flux
interfacial tension.
The experimental evaluation of the effect of low temperature fluxes on wet-
ting by Hubert et al. revealed that the rate of wetting increased with increase in
the acid content and temperature of the solder bath 关54兴. The rate of wetting is
dependent on the degree of oxide removal, which is a function of oxide concen-
tration as well as the temperature.
Wu et al. carried out investigations on the wettability of lead based and
lead-free solders on Cu substrate 关3兴. They observed that for lead based solders,
the use of water soluble flux resulted in best wetting. On the other hand, no-
clean flux gave better results for Sn–Ag solder. Further Sn–Ag and Sn–Ag–Cu
alloys could be soldered by using rosin mildly activated fluxes, whereas for
soldering of Sn–Zn lead-free alloys, the use of rosin activated flux is needed.
However, Kang et al. reported the near identical relaxation behavior of solders
共lead based and lead-free Sn–Ag–Cu兲 on Cu and Cu/Ni/Au substrates irrespec-
tive of the type of flux used, although there was improvement of wetting with
the use of flux 关30兴. Farooq et al. showed that environmental friendly water
soluble fluxes could be used for the soldering of Ni/Au metallized ceramic sub-
strates with lead-tin and lead-free solders without degrading wetting 关55兴.
A detailed investigation of wetting of various binary solders on Cu substrate
stressed the need for more active flux to achieve the same degree of wetting
with Pb-free alloys as obtained with Sn–Pb alloys 关4兴. Thus it is well established
that fluxes improve the wetting force by increasing the solid/vapor interfacial
energy or by lowering the solid/liquid interfacial energy.
Soldering Atmosphere
The soldering atmosphere plays a prominent role in spreading of solders. It has
been established that the reduction in residual oxygen level in the atmosphere
causes the spreading to start at lower temperatures. The oxide surface on the
substrate is detrimental to wetting, and hence the use of flux and/or inert at-
mosphere is inevitable to achieve good wetting. Nitrogen is generally used for
this purpose due to its inertness with most of the metals. Improved wetting is
observed in most of the spreading trials when carried out in N2 atmosphere
than in air particularly in soldering. The presence of inert atmosphere is also
helpful in improving the efficiency/functioning of fluxes in reactive spreading
processes. Further, fluxless soldering can also be carried out in inert atmo-
spheres 关35,51兴.
Temperature
A large number of properties that control wetting are sensitive to temperature
changes. For example, viscosity, surface tension, oxidation behavior, and reac-
tion rate all significantly vary with temperature variations. Viscosity and sur-
face tension of the liquid both decrease with an increase in temperature, result-
ing in the improvement of wetting in any system 关56兴.
In reactive wetting systems, the diffusion rate generally increases with an
increase in temperature. However, the increase in temperature also results in
severe oxidation in most of the metals including solders. The oxide layers
present on the surface of the spreading liquid as well as on the substrate sur-
face alter the interfacial properties and cause inferior wetting. Fluxes are used
in soldering, which are active at temperatures near the melting point of the
solder and remove the oxide layers facilitating efficient wetting. But, the in-
crease in the temperature beyond their activation temperature may cause
evaporation of flux, thereby resulting in a situation where there may not be any
flux available for oxide removal, which in turn results into poor wettability at
higher temperatures.
Bukat et al. carried out investigation on the effect of temperature on wet-
tability of lead-free solders and reported a decrease in interfacial tension and
an increase in wetting force with an increase in solder bath temperature. Both
of these factors contribute to the improvement in wettability 关27兴. Martorano et
al. also made similar observations while investigating the effect of solder bath
temperature on wetting balance curve of tin-zinc-silver solder alloy and attrib-
uted the same to the decrease in liquid viscosity with the increase in solder bath
temperature 关57兴. They also reported that the effect of substrate thickness on
spreading kinetics fades away as the bath temperature is increased. Experimen-
tal observations of Kang et al. on the effect of temperature on lead based and
lead-free solders showed that an increase in temperature resulted in speeding
up of wetting process, which was attributed to the increase in reactivity be-
tween the solder and the substrate material due to the temperature dependence
of diffusion process and drop in viscosity and surface tension of molten solder
关30兴. This is indicated in Fig. 10. A decrease in duration time 共the time required
for the decrease in contact angle from 140° to 50°兲 of about 70 % was observed
in their experiments for the temperature change from 190 to 230° C.
Investigations of Wu et al. on wetting of Sn-9Zn-x solders on Cu substrates
indicated a drop in contact angle of 9°–14° when the temperature was increased
from 245° to 290° 关3兴. Saiz et al. studied the spreading of Sn–Ag–Bi solders on
Fe–Ni alloys at 250 and 450° C and recorded contact angles of 70°–85° and
50°–65°, respectively 关32兴.
Peebles et al. carried out detailed investigations to study the kinetics of the
FIG. 10—Effect of temperature on contact angle relaxation of solders 关30兴.
wetting of tin on copper surface 关53兴. They observed that a slight increase in
substrate temperature results in a large increase in spreading rate. Spreading
rate shows a 1/3 order dependency on time up to 327° C and 1/5 order, there-
after indicating the change of spreading mechanism at that temperature.
Interfacial Reactions
Interfacial reaction rate appears to have no influence on the static contact
angle, while it does control the spreading rate. The reaction product layer ex-
tends on the free surface of the substrate and not only at the solid/liquid inter-
face. As a result, the static contact angle that reached the end of the spreading
is close to the Young contact angle of the liquid on the reaction product 关58兴.
The shorter wetting time indicates that the spontaneous interface reaction pro-
moted the wetting behavior 关24兴.
It is well known that substrate roughness impacts the static contact angle
and the dynamic wetting process as well. For the wetting of Sn based solders on
Cu, the larger substrate roughness will retard the dynamic wetting and increase
the static contact angle owing to the pinning effect, namely, the liquid solder
flowing down to the valley. That means the wetting time will increase, while the
wetting force will decrease during the wetting balance test. Apparently, the
roughness for the Cu6Sn5 / Cu3Sn/ Cu substrate is larger than that of Cu due to
the scallop morphology of Cu6Sn5 IMC. Thus, it can be reasoned that the
FIG. 11—Spreading of the Sn-3.5Ag solder on the IMC 共Cu6Sn5 / Cu3Sn/ Cu兲 surface
共halo effect兲关60兴.
longer wetting times and smaller wetting forces on Cu6Sn5 / Cu3Sn/ Cu sub-
strates than Cu may be attributed to the combined promotion effects of the
inert interfacial reaction and substrate roughness 关24兴.
The recent research on the wetting of lead based and lead-free solders on
virgin Cu and intermetallics of Cu and Sn reveals some interesting results
关8,59,60兴. Both lead based and lead-free solders exhibited extensive wetting on
IMC surfaces 共Cu6Sn5 / Cu3Sn/ Cu兲 compared to that on Cu. This is attributed
to scallop nodular surface morphology of IMCs. More intense spreading has
been observed on fine grained IMC, whereas coarse grained IMC behavior was
similar to virgin Cu. Part of the liquid solder spread on rough IMC surfaces like
a thin film forms a halo around the bulk liquid, as shown in Fig. 11. It was also
observed that there were four stages during the spreading of lead based solder
on bare copper surface where only two stages were observed during the spread-
ing of same solder on intermetallic layer 共Figs. 12 and 13兲
FIG. 12—Stages of spreading of the Sn-37Pb solder on the Cu surface 关8兴.

FIG. 13—Stages of spreading of the Sn-37Pb solder on the IMC layer 关59兴.
The phase transformation from Cu6Sn5 to Cu3Sn is a large energy-

releasing process. If a better wetting property 共shorter time and smaller angle兲
is observed on reactive rough surface, then the interfacial reaction would be a
predominant factor in promoting wetting, compared to the side effect from
substrate roughness. It should be noted that the formation energy of Cu3Sn is
2.5 times greater than Cu6Sn5. Hence, the occurrence of phase transformation
from Cu6Sn5 to Cu3Sn may be easier than that of solder on pure Cu substrate
directly 关24兴.
Summary
Soldering is a milder form of achieving metallurgical continuity that involves
several fields of science. From simple hand soldering to fully automated wave
and reflow soldering, it has a large number of applications. Conventional
Sn–Pb solders are being replaced by lead-free solders due to increasing con-
cerns about the hazards caused by the presence of lead.
The basic soldering process depends on wetting for the formation of solder-
to-base metal contact. The solidification of molten solder after wetting results
in permanent bond. The extent of wetting is measured by the contact angle that
is formed at the interface of a solid and liquid in a particular environment.
Fluxes are generally used in soldering to overcome the barrier effects of oxide
films. They keep the solderable surfaces clean and tarnish-free and influence
the surface tension of solder in the direction of solder spreading by decreasing
the contact angle. However, fluxless soldering can be carried out in certain inert
atmospheres. Solder wetting involves the metallurgical reactions between the

filler metal and the base metal. This interaction at the solder/base metal 共sub-
strate兲 interface results in the formation of IMCs.
The important factors that affect the wetting and solidification behaviors of
solders on a substrate are material composition, temperature, flux usage, etc.
Conventional Sn–Pb solders possess excellent wetting on general base materials
like Cu and its alloys. The lead-free solders exhibit inferior wetting properties
compared to lead based solders on most of the substrate surfaces. Hence, al-
loying additions to binary and ternary lead-free solders are generally made in
order to improve wetting and mechanical properties. It is difficult to assess the
effect of roughness alone on wetting behavior of metallic liquids as the main
effects are masked by other factors. A large number of properties that control
wetting are sensitive to temperature. Knowledge of wetting behavior, evolution
of microstructure, and IMC formation and their interaction is useful in the
design and development of new lead-free solders.
References
关1兴 Suganuma, K., “Advances in Lead-Free Electronics Soldering,” Curr. Opin. Solid
State Mater. Sci., Vol. 5, 2001, pp. 55–64.
关2兴 Vianco, P. T. and Frear, D. R., “Issues in the Replacement of Lead-Bearing Sol-
ders,” JOM, Vol. 45, No. 7, 1993, pp. 14–19.
关3兴 Wu, C. M. L., Yu, D. Q., Law, C. M. T., and Wang, L., “Properties of Lead-Free
Solder Alloys with Rare Earth Element Additions,” Mater. Sci. Eng. R., Vol. 44,
2004, pp. 1–44.
关4兴 Abtew, M. and Selvaduray, G., “Lead-Free Solders in Micro Electronics,” Mater.
Sci. Eng. R., Vol. 27, 2000, pp. 95–141.
关5兴 Kamal, M. and Gouda, E. S., “Enhancement of Solder Properties of Sn-9Zn Lead-
Free Solder Alloy,” Cryst. Res. Technol., Vol. 41, No. 12, 2006, pp. 1210–1213.
关6兴 Suraski, D. and Seelig, K., “The Current Status of Lead-Free Solder Alloys,” IEEE
Trans. Electron. Packag. Manuf., Vol. 24-4, 2001, pp. 244–248.
关7兴 Cheng, S. C. and Lin, K. L., “The Thermal Property of Lead-Free Sn-8.55Zn-1Ag-
XAl Solder Alloys and Their Wetting Interaction with Cu,” J. Electron. Mater., Vol.
31, No. 9, 2002, pp. 940–945.
关8兴 Zhao, H., Nalagatla, D. R., and Sekulic, D. P., “Wetting Kinetics of Eutectic Lead
and Lead-Free Solders: Spreading over the Cu Surface,” J. Electron. Mater., Vol. 38,
No. 2, 2009, pp. 284–291.
关9兴 Sebo, P. and Stefanik, P., “Effect of In Addition on Sn–Ag Solder, Its Wetting and
Shear Strength of Copper Joints,” Kovove Mater., Vol. 43, 2005, pp. 202–209.
关10兴 Ozvold, M., Hodulova, E., Chriastelova, J., Janovec, J., and Turna, M., “Lead-Free
Solders: Comparative Study of Thermal and Wetting Properties,” Metal 2008,
Hradec and Moravici, May 13–15, 2008, Tanger, R.O., pp. 1–8.
关11兴 Wu, C. M. L., Yu, D. Q., Law, C. M. T., and Wang, L., “The Properties of Sn-9Zn
Lead-Free Solder Alloys Doped with Trace Rare Earth Elements,” J. Electron.
Mater., Vol. 31, No. 9, 2002, pp. 921–927.
关12兴 Suganuma, K. and Kim, K. S., “Sn–Zn Low Temperature Solder,” J. Mater. Sci.:
Mater. Electron., Vol. 18, 2007, pp. 121–127.
关13兴 Lin, C. T. and Lin, K. L., “Contact Angle of 63Sn-37Pb and Pb Free Solder on Cu
Plating,” Appl. Surf. Sci., Vol. 214, 2003, pp. 243–258.

关14兴 Islam, R. A., Wu, B. Y., Alam, M. O., Chan, Y. C., and Jillek, W., “Investigations on
Microhardness of Sn–Zn Based Lead-Free Solder Alloys as Replacement of Sn–Pb
Solder,” J. Alloys Compd., Vol. 392, No. 1–2, 2005, pp. 149–158.
关15兴 Chuang, T. H., Yen, S. F., and Cheng, M. D., “Intermetallic Reactions in
Sn3Ag0.5Cu and Sn3Ag0.5Cu0.06Ni0.01Ge Solder BGA Packages with Au/Ni Sur-
face Finishes,” J. Electron. Mater., Vol. 35, 2, 2006, pp. 302–309.
关16兴 Chang, T. C., Wang, M. C., and Hon, M. H., “Thermal Properties and Interfacial
Reaction Between the Sn-9Zn-xAg Lead-Free Solders and Cu Substrate,” Metall.
Mater. Trans. A, Vol. 36A, 2005, pp. 3019–3029.
关17兴 Schwartz, M. M. and Aircraft, S., Metals Handbook, 10th ed., American Society of
Metals 共ASM兲, Metals Park, OH, 1991, Vol. 6, pp. 126–129.
关18兴 Frear, D. R., Jones, W. B., and Kinsman, K. R., Solder Mechanics—A State of the Art
Assessment, TMS Publication, Warrendale, 1991, pp. 1–104.
关19兴 Manko, H. H., Solders and Soldering, 3rd ed., McGraw-Hill, Inc., New York, 1992,
pp. 1–153.
关20兴 Lee, C. C. and Kim, J., ‘‘Fundamentals of Fluxless Soldering Technology,’’ in Pro-
ceedings of the 10th IEEE International Advanced Packaging Materials Symposium,
Irvine, CA, March 16–18, 2005, pp. 33–38.
关21兴 Arenas, M. F. and Acoff, V. L., “Contact Angle Measurements of Sn–Ag and Sn–Cu
Lead-Free Solders on Copper Substrates,” J. Electron. Mater., Vol. 33, No. 12, 2004,
pp. 1452–1458.
关22兴 Ghosh, G., “Interfacial Reaction Between Multicomponent Lead-Free Solders and
Ag, Cu, Ni and Pd Substrates,” J. Electron. Mater., Vol. 33, No. 10, 2004, pp. 1080–
1091.
关23兴 Laurila, T., Vuorinen, V., and Kivilahti, J. K., “Interfacial Reactions Between Lead-
Free Solders and Common Base Materials,” Mater. Sci. Eng. R., Vol. 49, 2005, pp.
1–60.
关24兴 Wang, H., Gao, F., Ma, X., and Qian, Y., “Reactive Wetting of Solders on Cu and
Cu6Sn5/Cu3Sn/Cu Substrates Using Wetting Balance,” Scr. Mater., Vol. 55, No. 9,
2006, pp. 823–826.
关25兴 Shibata, H., Jiang, X., Valdez, M., and Cramb, A. W., “The Contact Angle Between
Liquid Iron and a Single Crystal Magnesium Oxide Substrate at 1873 K,” Metall.
Mater. Trans. B, Vol. 35, 2004, pp. 179–181.
关26兴 Lin, K. L., and Wang, Y. C., “Wetting Interaction of Pb-Free Sn–Zn–Al Solders on
Metal Plated Substrate,” J. Electron. Mater., Vol. 27, No. 11, 1998, pp. 1205–1210.
关27兴 Bukat, K., Sitek, J., Hozu, L., and Bulwith, R., “Solderability Assessment of Pb-
Free Alloys Using VOC-Free Flux,” Global SMT and Packaging Journal, Vol. 2, No.
9, 2002, pp. 1–8.0002-7820
关28兴 Yu, D. Q., Zhao, J., and Wang, L., “Improvement on the Microstructure Stability,
Mechanical and Wetting Properties of Sn–Ag–Cu Lead-Free Solder with the Addi-
tion of Rare Earth Elements,” J. Alloys Compd., Vol. 376, 2004, pp. 170–175.
关29兴 Yu, D. Q., Xie, H. P., and Wang, L., “Investigation of Interfacial Microstructure and
Wetting Property of Newly Developed Sn–Zn–Cu Solders with Cu Substrate,” J.
Alloys Compd., Vol. 385, 2004, pp. 119–125.
关30兴 Kang, S. C., Kim, C., Muncy, J., Schmidt, M., and Baldwin, D., “Experimental
Wetting Dynamics Study of Eutectic and Lead-Free Solders Varying Flux, Tem-
perature and Surface Finish Metallization,” IEEE/SEMI Int’l Electronics Manufac-
turing Technology Symposium, July 14–16, 2004, IEEE/CPMT/SEMI, San Jose, CA,
pp. 1–8.
关31兴 Hong, S. M., Park, J. Y., Kang, C. S., and Jung, J. P., “Fluxless Wetting Properties of
UBM-Coated Si-Wafer to Sn-3.5 wt% Ag Solder,” IEEE Trans. Compon. Packag.

Technol., Vol. 26-1, 2003, pp. 255–261.
关32兴 Saiz, E., Hwang, C. W., Suganuma, K., and Tomsia, A. P., “Spreading of Sn–Ag
Solders on Fe–Ni Alloys,” Acta Mater., Vol. 51, 2003, pp. 3185–3197.
关33兴 Yoon, S. W., Choi, W. K., and Lee, H. M., “Calculation of Surface Tension and
Wetting Properties of Sn-Based Solder Alloys,” Scr. Mater., Vol. 40, No. 3, 1999, pp.
297–302.
关34兴 Takao, H., Tsukada, T., Yamada, K., Yamashita, M., and Hasegawa, H., “Develop-
ment of Wettability Evaluation Technique Applying Contact Angle Measurement
During Soldering,” R&D Rev Toyota CRDL, Vol. 392, 2004, pp. 41–48.0002-7820
关35兴 Mackie, A. C., “Reflow Atmospheres in the Lead-Free Era,” Circuits Assem., 2003,
pp. 26–35.
关36兴 Wang, H., Wang, F., Gao, F., Ma, X. and Qian, Y., “Reactive Wetting of Sn-0.7Cu-
xZn Lead-Free Solders on Cu Substrate,” J. Alloys Compd., Vol. 433-1, No. 2, 2007,
pp. 302–305.
关37兴 Mortensen, A., Drevet, B., and Eustathopoulos, N., “Kinetics of Diffusion-Limited
Spreading of Sessile Drops in Reactive Wetting,” Scr. Mater., Vol. 36, No. 6, 1997,
pp. 645–651.
关38兴 Chen, W. X., Xue, S. B., and Wang, H., “Wetting Properties and Interfacial Micro-
structures of Sn-Zn-xGa Solders on Cu Substrate,” Mater. Des., Vol. 31, No. 4,
2010, pp. 2196–2200.
关39兴 Chen, K. and Lin, K. L., “Effect of Gallium on Wettability, Microstructure and
Mechanical Properties of the Sn–Zn–Ag–Ga and Sn–Zn–Ag–Al–Ga Solder Alloys,”
International Symposium on Electronic Materials and Packaging, December 4–6,
2002, I-Shou University and IEEE/CPMT, Kaohsiung, Taiwan, pp. 49–54.
关40兴 Suganuma, K., Muata, T., Noguchi, H., and Toyoda, Y., “Heat Resistance of Sn-9Zn
Solder/Cu Interface with or Without Coating,” J. Mater. Res., Vol. 15, No. 4, 2000,
pp. 884–891.
关41兴 Kitajima, M., and Shono, T., “Reliability Study of New SnZnAl Lead-Free Solders
Used in CSP Packages,” Microelectron. Reliab., Vol. 45, No. 7–8, 2005, pp. 1208–
1214.
关42兴 Prasad, L. C. and Mikula, A., “Role of Surface Properties on the Wettability of
Sn–Pb–Bi Solder Alloys,” J. Alloys Compd., Vol. 282, 1999, pp. 279–285.
关43兴 Wang, L., Yu, D. Q., Zhao, J., and Huang, M. L., “Improvement of Wettability and
Tensile Property in Sn-Ag-RE Lead Free Solder Alloy,” Mater. Lett., Vol. 56, 2002,
pp. 1039–1042.
关44兴 Gao, L., Xue, S., Zhang, L., Xiao, Z., Dai, W., Ji, F., Ye, H., and Zeng, G., “Effect of
Praseodymium on the Microstructure and Properties of Sn3.8Ag0.7Cu Solder,” J.
Mater. Sci.: Mater. Electron., Vol. 2009, in press.
关45兴 Adamson, A. W. and Gast, A. P., Physical Chemistry of Surfaces, 6th ed., Wiley-
Interscience, New York, 1997, pp. 347–379.
关46兴 Morra, M., Occhiello, E., and Garbassi, F., “Knowledge About Polymer Surfaces
from Contact Angle Measurements,” Adv. Colloid Interface Sci., Vol. 32, 1990, pp.
79–116.
关47兴 Yost, F. G., Michael, J. R., and Eisenmann, E. T., “Extensive Wetting Due to Rough-
ness,” Acta Mater., Vol. 43, No. 1, 1995, pp. 299–305.
关48兴 Wolansky, G. and Marmur, A., “Apparent Contact Angles on Rough Surfaces: The
Wenzel Equation Revisited,” Colloids Surf., A, Vol. 156, 1999, pp. 381–388.
关49兴 Nakae, H., Inui, R., Hirata, Y., and Saito, H., “Effects of Surface Roughness on
Wettability,” Acta Mater., Vol. 46, No. 7, 1998, pp. 2313–2318.
关50兴 Lopez, V. H. and Kennedy, A. R., “Flux-Assisted Wetting and Spreading of Al on
TiC,” J. Colloid Interface Sci., Vol. 298, No. 1, 2006, pp. 356–362.
关51兴 Claesson, E., Choquenet, L., and Nilson, M., “Atmosphere Influence on Reflow
Soldering,” Brasage, Vol. 97, 1997, pp. 1–5.
关52兴 Vaynman, S. and Fine, M. E., “Development of Fluxes for Lead-Free Solders Con-
taining Zinc,” Scr. Mater., Vol. 41, No. 12, 1999, pp. 1269–1271.
关53兴 Peebles, D. E., Peebles, H. C., and Ohlhausen, J. A., “Kinetics of the Isothermal
Spreading of Tin on the Air-Passivated Copper Surface in the Absence of a Fluxing
Agent,” Colloids Surf., A, Vol. 144, 1998, pp. 89–114.
关54兴 Plas, H. A. V., Cinque, R. B., Mei, Z., and Holder, H., “Assessment of Low Tem-
perature Fluxes,” Hewlett-Packard J., Vol. 48, No. 4, 1997, pp. 1–7.
关55兴 Farooq, M., Ray, S., Sarkhel, A., and Goldsmith, C., “Evaluation of Lead 共Pb兲-Free
Ceramic Ball Grid Array 共CBGA兲: Wettability, Microstructure and Reliability,”
Electronic Components and Technology Conference, May 29–June 1, 2001, IEEE/
CPMT,ECA, Orlando, FL, pp. 978–986.
关56兴 Bernardin, J. D., Mudawar, I., Walsh, C. B., and Franses, E. I., “Contact Angle
Temperature Dependence for Water Droplets on Practical Aluminum Surfaces,”
Int. J. Heat Mass Transfer, Vol. 40, No. 5, 1997, pp. 1017–1033.
关57兴 Martorano, K. M., Martorano, M. A., and Brandi, S. D., “Effects of Solder Bath
Temperature and Substrate Sheet Thickness on the Wetting Balance Curve,” Tech.
Bull. Polytech. Sch. of USP, Dept. Metall. Eng. Mater., 2004, pp. 1–14.0002-7820
关58兴 Landry, K., Rado, C., and Eustathopoulos, N., “Influence of Interfacial Reaction
Rates on the Wetting Driving Force in Metal/Ceramic Systems,” Metall. Mater.
Trans. A, Vol. 27, 1996, pp. 3181–86.
关59兴 Zhao, H., Wang, H. Q., Sekulic, D. P., and Qian, Y. Y., “Spreading Kinetics of Liquid
Solders over an Intermetallic Solid Surface. Part 1: Eutectic Lead Solder,” J. Elec-
tron. Mater., Vol. 38, No. 9, 2009, pp. 1838–1845.
关60兴 Zhao, H., Wang, H. Q., Sekulic, D. P., and Qian, Y. Y., “Spreading Kinetics of Liquid
Solders over an Intermetallic Solid Surface. Part 2: Lead-Free Solders,” J. Electron.
Mater., Vol. 38, No. 9, 2009, pp. 1846–1854.
doi:10.1520/JAI103042
Y. Takaku,1 I. Ohnuma,2 Y. Yamada,3 Y. Yagi,3 I. Nakagawa,4

T. Atsumi,4 M. Shirai,4 and K. Ishida2
A Review of Pb-Free High-Temperature

Solders for Power-Semiconductor Devices:
Bi-Base Composite Solder and Zn–Al
Base Solder
ABSTRACT: Pb-base high-temperature solders 共mass % Sn⫽5–10, melting

point 共m.p.兲⫽300–310°C兲 are widely applied under severe conditions, al-
though the harmful nature of Pb is recognized. Bi-base alloys 共m.p. of Bi
⫽270°C兲, Zn-base alloys 共m.p. of Zn⫽420°C兲, and several Au-base eutectic
alloys 共m.p. of Au-20Sn and Au-3.6Si⫽280 and 363°C, respectively兲 are pro-
posed as candidates for Pb-free high-temperature solders. This paper re-
views the features of Bi-base composite solders containing reinforcement
particles of a superelastic Cu–Al–Mn alloy in a Bi matrix to relax thermal
stress and to prevent the propagation of cracks, and Zn–Al base solders,
which have high stability and high reliability enough to be utilized in practical
applications under severe thermal cycle tests between ⫺40 and 230°C more
than 2000 cycles.
KEYWORDS: high-temperature solder, intermetallic compound 共IMC兲,
thermal cycle test
Manuscript received March 11, 2010; accepted for publication September 24, 2010; pub-
lished online November 2010.
1
Dept. of Material Science, Graduate School of Engineering, Tohoku Univ., 6-6-02 Aoba-
yama, Sendai 980-8579, Japan, e-mail: yskz-ta@material.tohoku.ac.jp
2
Dept. of Material Science, Graduate School of Engineering, Tohoku Univ., 6-6-02 Aoba-
yama, Sendai 980-8579, Japan.
3
Toyota Central R&D Laboratories, Inc., 41-1 Yokomichi, Nagakute, Nagakute-cho, Ai-
chi 480-1192, Japan.
4
Toyota Motor Corporation, 1 Toyota-cho, Toyota City, Aichi 471-8572, Japan.
Cite as: Takaku, Y., Ohnuma, I., Yamada, Y., Yagi, Y., Nakagawa, I., Atsumi, T., Shirai, M.
and Ishida, K., ‘‘A Review of Pb-Free High-Temperature Solders for Power-
Semiconductor Devices: Bi-Base Composite Solder and Zn–Al Base Solder,’’ J. ASTM
Intl., Vol. 8, No. 1. doi:10.1520/JAI103042.
27
Introduction
Increased awareness of the seriousness of environmental issues mandates a

further increase of energy consumption efficiency. Next-generation semicon-
ductors, such as SiC and GaN instead of conventional Si, are expected to im-
prove the efficiency of vehicle on-board power devices. For such application,
however, the operating temperature of their assemblies, which consist of semi-
conductor chips, solder layers, substrates, and a heat sink plate, exceeds 200° C
关1–3兴. Therefore, heat resistance and reliability as well as environmentally
friendly characteristics are required for solder materials. To evaluate properties
of heat resistance and reliability, the assemblies must be able to survive severe
thermal cycle tests, for instance, thousands of thermal cycles at temperatures
between −40 and 200° C or higher with various heating and cooling rates,
which simulate severe Winter and on-stream engine room temperature condi-
tions. Recently, Pb–Sn 共mass % Sn= 5 – 10兲 solders are applied under such se-
vere conditions, even though the harmful nature of Pb is recognized. Therefore,
the immediate development of reliable Pb-free high-temperature solders is re-
quired.
The melting temperature of practical Pb-free solders, such as Sn-3.5Ag 关4兴
共melting point 共m.p.兲 = 220° C兲 and Sn-0.7Cu 关5兴共m.p. = 227° C兲 eutectic alloys,
is above 200° C. After high-temperature operations around 200° C, however,
intermetallic compounds 共IMCs兲 formed with substrates during soldering grow
and coarsen rapidly at the interface, which results in fracture origins due to
thermal stress. Thus, the reliability of solder joints is lost. Recently, the use of
Zn-共10–30兲Sn alloys has been proposed due to their economical advantage and
superior thermal and electrical conductivities 关6–8兴. These hypereutectic alloys
have a solidus temperature of 199° C 共eutectic temperature兲 and a liquidus
temperature of 360° C. A key aspect of using these alloys is to avoid liquid
formation at a reflow temperature around 250° C. However, as the surface of
the Sn–Zn alloy may possibly undergo oxidation from ambient gas, careful
attention to the high-temperature operation is required 关9兴. Other high-
temperature solders are Au-20Sn 关10兴共m.p. = 280° C兲 and Au-3.2Si 关11兴共m.p.
= 363° C兲 eutectic alloys, whose expensiveness prevents them from being ap-
plied practically for electronic packaging.
This review focuses on the design of new high-temperature Pb-free solders
using a thermodynamic database, Alloy Database for Micro-Solders 共ADAMIS兲
关12兴, in which 11 elements, Ag, Al, Au, Bi, Cu, In, Ni, Pb, Sb, Sn, and Zn, are
available. Using this database with certain software programs, such as Thermo-
Calc, Pandat, CatCalc, etc., the calculation of multi-component phase dia-
grams, solidification simulations, prediction of physical properties of the liquid
phase, etc., can be carried out 关13兴. In this paper, Bi-base and Zn–Al base alloys
are proposed as candidates for Pb-free high-temperature solders 关14–16兴. In
addition, based on the result of thermal cycle tests, optimized structures of the
next-generation power semiconductor devices using Pb-free high-temperature
solders are proposed.
TAKAKU ET AL., doi:10.1520/JAI103042 29
Bi-Based Solder
Alloy Design and Properties

Figure 1共a兲 and 1共b兲 shows phase diagrams of the Bi–Cu binary system and the
Bi–Cu–Al ternary system at 700° C calculated by Thermo-Calc using the ADA-
MIS database 关12兴. The calculated phase diagram of the Bi–Cu system suggests
that a metastable miscibility gap of liquid shown by a dashed line is hidden by
a stable equilibrium between the liquid and 共Cu兲 phases. The miscibility gap
becomes stable with the ternary addition of Al, as shown by the solid curve in
Fig. 1共b兲. In this ternary system, a Bi-rich liquid and a Cu–Al liquid are immis-
cible, which suggests that Cu–Al base fine particles can disperse in Bi-rich mol-
ten solder without formation of IMCs. A Cu-23Al-2Mn 共at. %兲 alloy was chosen
for use as reinforcement particles because its ␤-phase exhibits martensitic
transformation above room temperature and pseudo-elasticity with low applied
stress can be expected, which could relax the thermal stress loaded on the
Bi-base matrix. In addition, such particles are also expected to terminate the
growth of cracks formed in Bi-base solder. A synthesis procedure of the particle
reinforced Bi-base solder is shown in Fig. 2 关17兴. Powder of the Cu–Al–Mn alloy
was prepared by a gas atomizing method. The surface of the powders with a
diameter of 10– 50 ␮m was plated with electroless Ni 共⬃1 ␮m thickness兲 to
improve its wettability with molten Bi. The plated powders were then mixed
with molten Bi at 500° C, and the mixture was cast into a solder ingot. The
constitution and melting temperature of three kinds of Bi-base composite sol-
ders are listed in Table 1. The melting temperatures were measured by a differ-
ential scanning calorimeter 共DSC兲 and determined to be about 270° C. Back
scattered electron 共BSE兲 images of the microstructure of the solder were exam-
ined by an electron probe micro-analyzer 共EPMA; JEOL: JXA-8100兲. Figure 3
shows an example of a BSE image taken from No.1 solder sample, which con-
sists of 60 Bi, 30 CuAlMn-particles, and 10 Bi3Ni 共vol %兲. The volume fraction
of the Bi matrix 共white兲 was optimized to be 60–70 %. About 30–40 % of Cu–
Al–Mn共black兲- and Bi3Ni 共gray兲-particles from the plated Ni reacting with mol-
ten Bi at 500° C were dispersed in the matrix homogeneously.
The tensile strength of the prepared samples is shown in Fig. 4. A pure Bi
specimen was also prepared for comparison. Figure 4 shows that composite
solder specimens have higher tensile strength than the pure Bi one. Figure 5
shows the temperature dependence of the stress-strain curve of the No.1 solder
sample, which suggests that mechanical properties depend strongly on tem-
perature. The fracture elongation increases and the tensile strength decreases
with increasing test temperature. Even at 195° C, however, the tensile strength
remains at 13 MPa. It is widely known that Bi-base alloys show weak impact
resistance, as shown by a fracture surface of pure Bi after the tensile test in Fig.
6共a兲. On the other hand, the fracture surface of the No.1 composite solder
seems to be improved, as shown in Fig. 6共b兲, due to the presence of reinforce-
ment particles. Consequently, the Cu–Al–Mn reinforcement particles are very
effective for the improvement of the mechanical properties of Bi-base alloys.
FIG. 1—Calculated phase diagrams of 共a兲 the Bi–Cu binary system and 共b兲 the Bi–Cu–Al
ternary system at 700°C using the ADAMIS database.
CuAlM powder (10~50μm）

CuAlMn
Electroless nickel
c plating
by dropping
n method
et (~1μm)
ri powder
Dried Pure Bi
Mixing in evacuated
quartz capsule.
c
Melting at 500ºC for 5min to produce
Bi and CuAlMn particle composites.
Casting
FIG. 2—Synthesis procedure of Bi-base composite solders.
Interfacial Reaction between Bi-Base Solders and Substrates

During soldering and the thermal cycle test, the formation and growth of IMCs
occur at the interface between the solder and substrates, which strongly affects
the reliability of solder joints. Figure 7 shows the microstructure near inter-
faces between a Bi-base solder and substrates of 共a兲 Ni and 共b兲 Cu heat-treated
at 330° C for 60 s. A thick layer of Bi3Ni formed as shown in Fig. 7共a兲, the
TABLE 1—Volume fraction of each phase and m.p. of Bi-base composite solders.
Bi Cu–Al–Mn Bi3Ni Melting Temperature

共vol %兲共vol %兲共vol %兲共°C兲
1 60 30 10 268.5
2 70 20 10 269.5
3 60 20 20 271.0
Bi3Ni
Cu-Al-Mn
20μm
FIG. 3—BSE image of microstructure of a Bi-base composite solder 共No. 1:
60 Bi+ 30 CuAlMn particles+ 10 Bi3Ni in vol %兲.
growth rate constant of which is estimated to be 1.1⫻ 10−6 m · s−1/2. Dybkov

and Duchenko reported that the growth rate of the Bi3Ni layer is too fast to
maintain the reliability of solder joints during aging 关18兴. In contrast to the Ni
substrate, no IMC layer formed at the Cu interface, as shown in Fig. 7共b兲, which
can be predicted from the immiscible tendency of the Bi–Cu system shown in
Tensile Strength / MPa
FIG. 4—Comparison of tensile strength of the Bi-base composite solders and pure Bi.
30 105 o
C
25oC 105oC
25 -40oC
Stress(MPa)
20
195oC
15
10
5
0
0 1 2 3 4 5 6 7 8 9 10
Strain(%)
FIG. 5—Stress-strain curve of No. 1 solders tested at −40, 25, 105, and 195°C.
Fig. 1共a兲. Consequently, the Cu substrate is appropriate for Bi-base solders. In

the case of Sn-base solders, thin layers of Au and Ni on the Cu substrate are
indispensable to improve wettability and to prevent growth of Sn–Cu IMCs,
respectively 关19–21兴. However, Bi-base solders can be directly soldered on Cu
substrates, which have an advantage compared with the conventional soldering
process.
Thermal Cycling Test

In practical applications of power semiconductor devices, a chip of Si is sol-
dered on an insulator, both of which are plated with other metallic elements. To
evaluate the reliability of solder joints after a thermal cycle test, some soldered
assemblies consisting of a semiconductor chip 共insulated gate bipolar transis-
tor 共IGBT兲: Si/Ni, 12⫻ 9 mm2兲 or its imitation 共copper-invar-copper 共CIC兲: Cu/
Fe–Ni Invar alloy/Cu, 12⫻ 9 mm2兲, a Bi-base composite solder 共0.1 g兲 and a
substrate 共direct bonded aluminum 共DBA兲: Ni/Al/AlN/Al/Ni, CIC or Cu, 17
⫻ 34 mm2兲 were prepared on a hot-plate under an argon atmosphere. The
coefficients of thermal expansion 共CTEs兲 of each component are listed in Table
2.
In advance of the thermal cycle test, the soundness of the solder layer was
checked by X-ray radiography. Figure 8 shows a transmitted X-ray image.
Bright areas represent voids, the total area fraction of which is about 5.2 %. No
structural failures were observed at the corners and edges of the soldered as-
sembly from which cracks could open preferentially during thermal cycles.
After a thermal cycle test, the assemblies were cut along their diagonal line,
and the microstructural evolution of the cross sections was examined by an
optical microscope and an EPMA. Figure 9共a兲 shows an optical micrograph of a
CTE-mismatched sample, which consists of a CIC, Bi-base solder 共No.1兲, and a
Cu substrate after 34 thermal cycles between −40 and 105° C. No cracks were
(a)
10μm
(b)
10μm
FIG. 6—Fracture surface of 共a兲 pure Bi and 共b兲 No. 1 solder after tensile test at 25°C.
(a)
CuAlM n
Bi
Bi 3 Ni
Ni
10 μ m
(b)
CuAlM n
Bi
Cu
10 μm
m
FIG. 7—Microstructure around interface between No. 1 solder and 共a兲 Ni and 共b兲 Cu
substrates soldered at 330°C for 1 min.
TABLE 2—CTEs of each component.
CTEs
Component Structure 共ppm· ° C−1兲
IGBT-Si chip Si/Ni 4
CIC imitation chip Cu/Fe–Ni Invar alloy/Cu 4
DBA substrate Ni/Al/AlN/Al/Ni 4
CIC substrate Cu/Fe–Ni Invar alloy/Cu 4
DBC substrate Ni/ Cu/ Si3N4 / Cu/ Ni 4
Mo Mo/Ni 5.1
Cu Cu 17
Bi-base solder ¯ 10–12
Zn–Al base solder ¯ 27
1mm
FIG. 8—Transmitted X-ray image of a soldered assembly of 关Si/Ni chip兴/关Bi-base com-
posite solder 共No. 1兲兴/关DBA substrate兴.
observed within small cycles and small temperature hysteresis even though the
CTE-mismatch is large, which causes a large thermal stress. Cracks open and
extend in the solder layer with increasing cycles and increasing hysteresis after
204 thermal cycles between −40 and 195° C, as shown in Fig. 9共b兲. However, the
direction of the crack extension changed several times, which indicates that
reinforced particles play a role as obstacles to the propagation of cracks. Pb-
10Sn and Sn-0.7Cu conventional solders were also subjected to the thermal
cycle test of a CTE-mismatched combination of an IGBT 共Si/Ni/Au兲 and a Cu
substrate 共Au/Ni/Cu兲. Results are summarized in Table 3. In both cases, brittle
IMCs formed and grew at the interface between the solder and the substrate,
which resulted in fracture of the assemblies.
In the case of a CTE-matched assembly consisting of a CIC imitation chip,
Bi-base solder 共No.1兲 and a CIC substrate, the micrograph of assembly of was
examined after 3003 thermal cycles between −40 and 195° C, as shown in Fig.
9共c兲. Figure 10 shows a BSE image taken after 500 thermal cycles between −40
and 250° C. In spite of the maximum thermal cycles or temperature hysteresis,
each soldered assembly was connected. Although no cracks were observed in-
side samples, Bi-oxide formed at the interfaces and around particles. Therefore,
Bi-base solders need to be protected from the atmosphere when they are ex-
posed to a high-temperature environment.
FIG. 9—Optical micrographs of soldered assemblies; 关imitation chip of CIC兴/关No. 1

solder兴/关Cu substrate兴 after 共a兲 34 cycles between −40 and 105°C and 共b兲 204 cycles
between −40 and 195°C, and 共c兲关CIC兴/关No. 1 solder兴/关CIC substrate兴 after 3003 cycles
between −40 and 195°C.
TABLE 3—Crack length under CTE-mismatched 共chip/solder/Cu substrate兲 soldered assem-

blies after thermal cycling test at −40 and 195°C.
Crack Length
Solder 共mm兲
Pb-10Sn 2.6 @ 300 cycles
Sn-0.7Cu 2.0 @ 200 cycles
Bi composite solder 1.0 @ 200 cycles
Zn–Al Based Solder
Alloy Design
Figure 11共a兲 shows a calculated phase diagram of the Zn–Al system using the
ADAMIS database. About 6 mass % Al alloy exhibits a eutectic reaction at
381° C. This melting temperature is rather high compared with conventional
Pb-bearing high-temperature solders. Since unified numbering system 共UNS兲
Z35530 共AG40A: Zn-4Al兲 and UNS Z33521 共AG41A: Zn-4Al-1Cu 共mass %兲兲 al-
loys, which are hypoeutectic alloys, are commercially used as die casting ma-
terials, these are applied for high-temperature solders. Figure 12 shows the
results of the DSC measurement of Zn-4Al and Zn-4Al-1Cu alloys 关22兴 during
heating. The first endothermic peaks, which appeared at 282 and 285° C, cor-
BSE image @ -40oC/250oC 500cycles
CIC chip
Bi-oxide
No. 1 solder
CIC substrate
100μm
FIG. 10—BSE image of an assembly; 关imitation chip of CIC兴/关No. 1 solder兴/关CIC sub-

strate兴 after 500 cycles between −40 and 250°C.
FIG. 11—Calculated phase diagrams of the 共a兲 Zn–Al and 共b兲 Cu–Zn binary systems
using the ADAMIS database.
Zn4Al1Cu o
285 C o
380 C
o o
282 C 381 C
Zn4Al
Endo.
Al rich fcc + hcp(Zn)

/ Zn-rich fcc
200 250 300 350 400 450
FIG. 12—Heating curves of DSC analysis measured for Zn-4Al and Zn-4Al-1Cu alloys.
respond to the monotectoid reaction: Al-rich fcc+ hcp 共Zn兲 = Zn-rich fcc. The
second ones at 381 and 380° C correspond to the eutectic melting reaction:
Zn-rich fcc+ hcp 共Zn兲 = liquid. The effect of Cu on both of these reactions is
small. Below 282° C, no phase transformation occurs in these alloys. This indi-
cates that Zn–Al base alloys are potential candidates for Pb-free high-
temperature solders, which could withstand the severer thermal stress and the
wider temperature hysteresis between −40 and 250° C, which is required for
packaging of the next-generation power semiconductor devices.
Interfacial Reaction during Soldering and Aging

Zn-4Al-1Cu alloy was soldered on a Cu substrate and heat-treated at 420° C for
5 min. The microstructure of the soldered interface is shown in Fig. 13. During
the heat-treatment, Cu dissolved into the molten Zn-4Al-1Cu alloy and IMCs
formed and grew at the interface. Two dashed lines are superimposed on the
optical micrograph; the upper one is the initial surface of the Cu substrate, and
the lower one represents the dissolved front. IMC layers were identified by
EPMA measurement across the soldered interface. From the Cu substrate to
the Zn-base solder, thin ␤共CuZn兲, thick ␥ 共Cu5Zn8兲, and thin ␧ 共CuZn4兲 layers
formed, which is consistent with the phase diagram of the Zn–Cu system
shown in Fig. 11共b兲. The thickness evolution of each IMC layer during heat-
treatment at 420° C is shown in Fig. 14. Linear relationships between the thick-
ness, d, and the square root of time, t1/2, was confirmed, which suggests that the
growth of each IMC follows the parabolic law controlled by the volume diffu-
sion
d = k · t1/2 共1兲
where:
k represents the growth rate constant, which can be estimated by the slope
of the regression lines in Fig. 14.
IMC
Cu
dissolution
Cu
20μm
FIG. 13—Microstructure of Zn-4Al-1Cu alloy soldered on Cu substrate at 450°C for 5

min.
120
o
Cu / Zn-4Al at 420 C
100
IMCs' thickness /μm
80
γ (Cu5Zn8)
60
40 ε(CuZn4)
20
β(CuZn)
0
0 10 20 30 40 50 60 70
Holding time, t1/2 / s1/2
FIG. 14—Thickness evolution of IMCs formed at interface of Zn-4Al alloy soldered on
Cu substrate at 420°C.
TABLE 4—The growth rate of IMCs between Cu or Ni/molten Zn–Al based solder.
k
共m · s−1/2兲
Substrate/Solder IMC 420° C 450° C 500° C 530° C

Cu/Zn4Al ␤ 共CuZn兲 6.0⫻ 10−8 6.6⫻ 10−8 1.5⫻ 10−7 2.7⫻ 10−7
␥ 共Cu5Zn8兲 1.7⫻ 10−6 2.6⫻ 10−6 3.6⫻ 10−6 4.9⫻ 10−6
␧ 共CuZn4兲 6.5⫻ 10−7
9.7⫻ 10−7 1.0⫻ 10−6 1.1⫻ 10−6
Cu/Zn4Al1Cu ␤ 共CuZn兲 7.2⫻ 10−8
8.5⫻ 10−8 2.6⫻ 10−7 9.3⫻ 10−7
␥ 共Cu5Zn8兲 1.8⫻ 10−6
2.6⫻ 10−6 3.4⫻ 10−6 9.5⫻ 10−6
␧ 共CuZn4兲 7.1⫻ 10−7
9.3⫻ 10−7 9.5⫻ 10−7 1.1⫻ 10−6
Ni/Zn4Al Al3Ni2 1.1⫻ 10−8
¯ ¯ ¯
Ni/Zn4Al1Cu Al3Ni2 1.3⫻ 10−8 4.4⫻ 10−7 共until 60s兲 ¯ ¯
Estimated values of rate constant at temperatures between 420 and 530° C

are listed in Table 4, where k increases with increasing soldering temperature.
Next, a Zn-4Al-1Cu alloy was soldered on a Ni substrate at 420 and 450° C
for 5 min; the microstructure of BSE images at the interface is shown in Fig. 15
关23兴. A very thin IMC layer formed at 420° C. After longer heat-treatment, the
IMC was identified to be a Al3Ni2 phase 共64Al-33Ni-3Zn 共at. %兲兲 by EPMA. The
crystal structure of this IMC was reported to be a hexagonal type with lattice
constants of a = 0.4 nm and c = 0.49 nm 关24兴. Recently, according to a revision
by Yadav et al., it is a vacancy-ordered phase with a rhombohedrally distorted
B2 structure 共hexagonal兲关25兴. It seems, therefore, that the Al3Ni2 layer easily
forms and rapidly grows in comparison with other IMCs in the Ni–Al and
Ni–Zn systems due to intrinsic vacancies. Growth rates of the Al3Ni2 layer
formed between Ni substrates and Zn-4Al or Zn-4Al-1Cu alloys at 420° C are
shown in Fig. 16 in comparison with those of the Cu–Zn IMC layers. Appar-
ently, their growth also follows the parabolic law. As summarized in Table 4, the
values of k are much smaller than those of the Cu–Zn IMC layers, which sug-
gests that the stability and reliability of the solder joint below 420° C can be
expected to be high. On the other hand, the formation and morphology of the
IMCs formed at the interface between a Ni substrate and a Zn-4Al-1Cu alloy
soldered at 450° C changed drastically, as shown in Fig. 15共b兲. A Zn3Ni layer
formed on the Ni substrate and thin layers of the Al3Ni2 seemed to peel off
from the interface and separate into the molten Zn-4Al-1Cu alloy over and over,
spreading up to 4 ␮m within 60 s. The apparent growth rate constant of the
Al3Ni2 multi-layers at 450° C reached 4.4⫻ 10−7 m · sec−1/2, which is about 40
times larger than that of 420° C, as listed in Table 4 关23兴. Therefore, the upper
limit of the operating temperature exists between 420 and 450° C, above which
interfacial stability might be lost.
After soldering at 420° C for 5 min, the assemblies of Ni/Zn-4Al and Ni/Zn-
4Al-1Cu were further heat-treated at 300° C to simulate the growth of IMCs in
high-temperature operations. Taking into account the initial thickness of the
Al3Ni2 layer, d0, Eq 1 is revised as
(a) Zn-4Al-1Cu/Ni : 420°C - 5min.
Zn-4Al-1Cu
Al3Ni2
Ni
5μm
(b) Zn-4Al-1Cu/Ni : 450°C - 5min.
Zn-4Al-1Cu
gray : Al3Ni2
white: Zn
Zn3Ni
200μm Ni
FIG. 15—Microstructures of Zn-4Al-1Cu alloy soldered on Ni substrate 共a兲 at 420°C for
5 min and 共b兲 at 450°C for 5 min.
d = d0 + k · t1/2 共2兲
This relationship is confirmed in Fig. 17 and the growth rate constants, k,
of the IMCs are summarized in Table 5 关22,23兴. The growth of the Al3Ni2 layer
is the slowest among all the IMCs. Consequently, a Ni layer at the surface of
soldered components is preferable for high-temperature solders.
Power Module and Reliability

A Zn–Al base solder sheet can be easily prepared by cold rolling. However, it is
difficult to remove the oxide layer formed on the surface of the sheet even
under a reducing atmosphere at temperatures between 300 and 400° C.
γ (Cu5Zn8) β (CuZn)
0.9
ε (CuZn4)
0.8
Zn-4Al-1Cu
Thickness of Al3Ni2 /μm
0.7
0.6
0.5
Zn-4Al
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60
Holding time t1/2 /s1/2
FIG. 16—Thickness evolution of IMC 共Al3Ni2兲 formed at interface of Zn-4Al and Zn-
4Al-1Cu alloys soldered on Ni substrate at 420°C in comparison with that of Zn-4Al
alloy soldered on Cu substrate at 420°C.
A pretreatment method of the oxidative Zn–Al base solders is proposed,

which can be applied for a conventional reflow process in the soldering of
multiply layered power module consisting of a chip/solder/an insulator/solder/a
heat sink plate. Before the reflow process, each surface of Zn–Al solder sheets
was exposed to radio frequenct 共RF兲 plasma etching to remove oxide layers.
After that, a thin layer of Cu with a thickness of 50 nm was deposited on each
surface by the direct current 共DC兲 sputtering method to prevent the Zn-base
solder sheets from being oxidized. These two processes were performed con-
secutively in a vacuum chamber. Then multiply layered power modules illus-
trated in Fig. 18共a兲 were prepared at 430° C for 5 min in an electrical furnace
under a gas flow of 5 % H2 mixed with N2. During the soldering process,
modules were pressed by a 4 g of Cu plate on the top. The soundness of the two
soldered layers in each module was examined by the scanning acoustic micro-
scope 共SAM兲. Figure 18共b兲 shows SAM images of the solder layers of a soldered
module, which consists of an 关IGBT chip 共Ni兲兴/关共Cu兲, a Zn–Al base solder sheet
共Cu兲兴/关共Ni兲 a DBC insulator 共Ni兲兴/关共Cu兲 a Zn–Al base solder sheet 共Cu兲兴/关共Ni兲 a Mo
heat sink兴. 共Elements in parentheses represent plated layers.兲 Two joints of this
module were soldered simultaneously. In both images, dark and bright con-
trasts represent desirable joints and some failures, respectively. Most area of
t(days)
3 7 21 42
2.0
Thickness of Al3Ni2 / μm
1.5
Zn-4Al-1Cu
1.0
Zn-4Al
0.5
0
0 500 1000 1500 2000 2500
1/2 1/2
Holding time t /s
FIG. 17—Thickness evolution of Al3Ni2 layers during heat-treatment at 300°C.
the solder layers looks dark, which suggests that the soundness of the soldered
joints is satisfactory. Figure 19 shows cross sections of the soldered joints, after
1000 thermal cycles, 共a兲 at chip side and 共b兲 at heat sink side, and after 2000
thermal cycles, 共c兲 at chip side, between −40 and 230° C. Only a thin Al3Ni2
layer was recognized at each soldered interface, whose thickness is kept very
TABLE 5—The growth rate of IMCs between Cu or Ni/solid Zn–Al based solder.
k
共m · s−1/2兲
Solid/Solid Diffusion Couples IMCs 200° C 300° C

Ni/Zn-4Al Al3Ni2 4.12⫻ 10−10 6.27⫻ 10−10
Ni/Zn-4Al-1Cu Al3Ni2 6.13⫻ 10−10 8.88⫻ 10−10
Cu/Zn-4Al ␤⬘ 共CuZn兲 7.64⫻ 10−10 2.90⫻ 10−9
␥ 共Cu5Zn8兲 9.32⫻ 10−8 6.20⫻ 10−7
␧ 共CuZn4兲 1.37⫻ 10−8 1.08⫻ 10−7
Cu/Zn-4Al-1Cu ␤⬘ 共CuZn兲 1.21⫻ 10−9 2.03⫻ 10−9
␥ 共Cu5Zn8兲 1.37⫻ 10−7 3.90⫻ 10−7
␧ 共CuZn4兲 3.60⫻ 10−8 1.17⫻ 10−7
(a)
IGBT-chip
Zn-Al solder
Zn-Al solder DBC insulator
Mo heat sink plate
(b) SAM image

heat sink plate side
Chip side
10mm
FIG. 18—共a兲 A schematic structure of power semiconductor module. 共b兲 SAM images of
solder layers.
thin even after the severe and a large number of thermal cycles. The soundness
of the soldered interfaces remained even though the surfaces of the DBC insu-
lator were waved and a small number of cracks opened in the solder layer
around the edge of the IGBT chip during the thermal cycles. These results
suggest that the Zn–Al base solders have great potential to be replaced with the
conventional Pb-base high-temperature solders when the surface coating as
well as the CTE matching of each component is appropriately chosen.
Conclusion
Bi-base composite alloys 共m.p. = 270° C兲 and Zn–Al base alloys 共m.p. = 380° C兲
are proposed for use as Pb-free high-temperature solders. Reinforcement par-
ticles of a Cu–Al–Mn martensitic alloy in a Bi-base matrix are expected to relax
thermal stress and to prevent the propagation of cracks in the Bi-base matrix.
Heat resistance and reliability of assemblies soldered with the Bi-base compos-
ite were confirmed against 3000 thermal cycles between −40 and 195° C and
more than 500 cycles between −40 and 250° C.
Interfacial reactions between Zn–Al based alloys and Cu or Ni substrates
during soldering and heat-treatment were also investigated. With the Cu sub-
strate, ␥ 共Cu5Zn8兲 and ␧ 共CuZn4兲 phases formed and thickened rapidly. With
FIG. 19—Cross section microstructures of soldered modules, which consist of an

关IGBT chip 共Ni兲兴/关共Cu兲, a Zn–Al base solder sheet 共Cu兲兴/关共Ni兲, a DBC insulator 共Ni兲兴/
关共Cu兲 a Zn–Al base solder sheet 共Cu兲兴/关共Ni兲 and a Mo heat sink兴, after 1000 thermal
cycles, 共a兲 chip side and 共b兲 heat sink side, and after 2000 thermal cycles, 共c兲 chip side,
between −40 and 230°C.
the Ni substrate, on the other hand, only a thin Al3Ni2 layer formed during
soldering at 420° C. It hardly thickened during heat-treatment below 270° C
nor during 2000 thermal cycles between −40 and 230° C, which suggests that
the Al3Ni2 layer prevents the formation of other IMCs and that the heat resis-
tance and reliability of the soldered assemblies are very high.
Acknowledgments
This work was supported by a Grant-in-Aid for Scientific Research from the
Japan Society for the Promotion of the Science and the Global COE program.
References
关1兴 Ueda, H., Sugimoto, M., Uesugi, T., Fujisima, O., and Kachi, T., “High Current
Operation of GaN Power HEMTS,” Proc. of the 17th Inter. Symp. Power Semicond.
Devices & IC’s, Santa Barbara, CA, May 23–26, IEEE, 2005, pp. 311–314.
关2兴 Asano, K., Hayashi, T., Takahashi, D., Sugawara, Y., Ryu, S.-H., and Palmour, J. W.,
“Temperature Dependence of On-State Characteristics, and Switching Character-
istics of 5kV class 4H-SiC SEJFET,” IEEJ Trans. IA, Vol. 125, 2005, pp. 147–152.
关3兴 Pietranico, S., Pommier, S., Lefebvre, S., Khatir, Z., and Bontemps, S., “Charac-
terization of Power Modules Ceramic Substrates for Reliability Aspects,” Micro-
electron. Reliab., Vol. 49, 2009, pp. 1260–1266.
关4兴 Karakaya, I. and Thompson, W. T., Bull. Alloy Phase Diagrams, Vol. 8, 1987, pp.
340–347.
关5兴 Saunders, N. and Miodownik, A. P., Phase Diagrams of Binary Copper Alloys, P. R.
Subramanian, D. J. Charkrabarti, and D. E. Laughlin, Eds., ASM International,
Materials Park, OH, 1994, pp. 412–418.
关6兴 Suganama, K., Kim, S.-J., and Kim, K.-S., “High-Temperature Lead-Free Solders:
Properties and Possibilities,” JOM, Vol. 61, 2009, pp. 65–71.
关7兴 Kim, S., Kim, K.-S., Suganuma, K., and Izuta, G., “Interfacial Reaction of Si Die
Attachment with Zn–Sn and Au-20Sn High temperature Lead-Free Solders on Cu
Substrate,” J. Electron. Mater., Vol. 38, 2009, pp. 873–883.
关8兴 Takahashi, T., Komatsu, S., Nishikawa, H., and Takemoto, T., “Improvement of
High-Temperature Performance of Zn–Sn Solder Joint,” J. Electron. Mater., Vol.
39, 2010, pp. 1241–1247.
关9兴 Jiang, J., Lee, J.-E., Kim, K.-S., and Suganuma, K., “Oxidation Behavior of Sn–Zn
Solders Under High-Temperature and High-Humidity Conditions,” J. Alloys
Compd., Vol. 462, 2008, pp. 244–251.
关10兴 Okamoto, H. and Massalski, T. B., Phase Diagrams of Binary Gold Alloys, H. Oka-
moto and T. B. Massalski, Eds., ASM International, Materials Park, OH, 1987, pp.
278–289.
关11兴 Okamoto, H. and Massalski, T. B., “The Au–Si 共Gold-Silicon兲 System,” Bull. Alloy
Phase Diagrams, Vol. 4共2兲, 1983, pp. 190–198.
关12兴 Liu, X. J., Ohnuma, I., Wang, C. P., Jiang, M., Kainuma, R., Ishida, K., Ode, M.,
Koyama, T., Onodera, H., and Suzuki, T., “Thermodynamic Database on Microsol-
ders and Copper-Based Alloy Systems,” J. Electron. Mater., Vol. 32, 2003, pp. 1265–
1272.
关13兴 Liu, X. J., Kinaka, M., Takaku, Y., Ohnuma, I., Kainuma, R., and Ishida, K., “Ex-
perimental Investigation and Thermodynamic Calculation of Phase Equilibria in

the Sn–Au–Ni System,” J. Electron. Mater., Vol. 34, 2005, pp. 670–679.
关14兴 Takaku, Y., Ohnuma, I., Yamada, Y., Yagi, Y., Nishibe, Y., Kainuma, R., and Ishida,
K., “Development of Bi-Base High-Temperature Pb-Free Solders with Second
Phase Dispersion: Thermodynamic Calculation, Microstructure and Interfacial Re-
action,” J. Electron. Mater., Vol. 35, 2006, pp. 1926–1932.
关15兴 Shimizu, T., Ishikawa, H., Ohnuma, I., and Ishida, K., “Zn–Al–Mg–Ga Alloys as
Pb-Free Solder for Die-Attaching Use,” J. Electron. Mater., Vol. 28, 1999, pp. 1172–
1175.
关16兴 Rettenmayr, M., Lambracht, P., Kempf, B., and Tschudin, C., “Zn–Al Based Alloys
as Pb-Free Solders for Die Attach,” J. Electron. Mater., Vol. 31, 2002, pp. 278–285.
关17兴 Yamada, Y., Takaku, Y., Yagi, Y., Nishibe, Y., Ohnuma, I., Sutou, Y., Kainuma, R.,
and Ishida, K., “Pb-Free High Temperature Solders for Power Device Packaging,”
Microelectron. Reliab., Vol. 46, 2006, pp. 1932–1937.
关18兴 Dybkov, V. I. and Duchenko, O. V., “Growth Kinetics of Compound Layers at the
Nickel-Bismuth Interface,” J. Alloys Compd., Vol. 234, 1996, pp. 295–300.
关19兴 Lin, C.-Y., Jao, C.-C., Lee, C., and Yen, Y.-W., “The Effect of Non-Reactive Alloying
Elements on the Growth Kinetics of the Intermetallic Compound Between Liquid
Sn-Based Eutectic Solders and Ni Substrates,” J. Alloys Compd., Vol. 440, 2007,
pp. 333–340.
关20兴 Chan, Y. C., Chiu, M. Y., and Chuang, T. H., “Intermetallic Compounds Formed
During the Soldering Reactions of Eutectic Sn-9Zn with Cu and Ni Substrates,” Z.
Metallkd., Vol. 93, 2002, pp. 95–98.
关21兴 Takaku, Y., Liu, X. J., Ohnuma, I., Kainuma, R., and Ishida, K., “Interfacial Reac-
tion and Morphology Between Molten Sn Base Solders and Cu Substrate,” Mater.
Trans., Vol. 45, 2004, pp. 646–651.
关22兴 Takaku, Y., Felicia, L., Ohnuma, I., Kainuma, R., and Ishida, K., “Interfacial Reac-
tion Between Cu Substrate and Zn–Al Based High Temperature Pb-Free Solders,”
J. Electron. Mater., Vol. 37, 2008, pp. 314–323.
关23兴 Takaku, Y., Makino, K., Watanabe, K., Ohnuma, I., Kainuma, R., Yamada, Y., Yagi,
Y., Nakagawa, I., Atsumi, T., and Ishida, K., “Interfacial Reaction Between Zn–Al-
Based High-Temperature Solders and Ni Substrate,” J. Electron. Mater., Vol. 38,
2009, pp. 54–60.
关24兴 Ellner, M., Kattner, V., and Predel, B., “Konstitutionelle und Strukturelle Untersu-
chungen im Aluminiumreichen Teil der Systeme Ni–Al und Pt–Al 关Constitutional
and Structural Investigation of Al-Rich Portion in the Ni-Al and Pt-Al Systems兴,” J.
Less-Common Met., Vol. 87, 1982, pp. 305–325.
关25兴 Yadav, T. P., Mukhopadhyay, N. K., Tiwari, R. S., and Srivastava, O. N., “Formation
of Al3Ni2-Type Nanocrystalline ␶3 Phases in the Al–Cu–Ni System by Mechanical
Alloying,” Philos. Mag. Lett., Vol. 87, 2007, pp. 781–789.
doi:10.1520/JAI103052
Satyanarayan1 and K. Narayan Prabhu2
Wetting Behaviour and Evolution of

Microstructure of Sn–Ag–Zn Solders on
Copper Substrates with Different
Surface Textures
ABSTRACT: The effect of surface roughness on wetting behaviour and evo-

lution of microstructure of two lead-free solders 共Sn-2.625Ag-2.25Zn and
Sn-1.75Ag-4.5Zn兲 on copper substrate was investigated. Both solders exhib-
ited good wettability on copper substrates having rough surface and lower
wettabilty on smooth surfaces. The contact angles of solders decreased lin-
early with increase in surface roughness of the substrate. The exponential
power law, ␾⫽exp共⫺K␶n兲, was used to model the relaxation behaviour of
solders. A high intermetallic growth was observed at the interface particularly
on copper substrates with rough surface texture. A thin continuous interface
showing scallop intermetallic compounds 共IMC兲 was obtained on smooth sur-
faces. With an increase in surface roughness, the IMC morphology changed
from scallop shaped to needle type at the Sn-2.625Ag-2.25Zn solder/
substrate interface and nodular to plate like IMCs for Sn-1.75Ag-4.5Zn solder
matrix.
KEYWORDS: lead-free solders, wetting, contact angle, EPL, IMCs
Introduction
Soldering is defined as a metallurgical joining method using a filler metal
known as solder with melting point below 400° C 关1兴. Eutectic Sn-37Pb solder
alloy is the most common solder material used in electronics industry because
Manuscript received February 25, 2010; accepted for publication July 31, 2010; pub-
lished online September 2010.
1
Karnataka, Surathkal, Srinivasnagar 575 025, India, e-mail: satyan.nitk@gmail.com
2
Karnataka, Surathkal, Srinivasnagar 575 025, India, e-mail: prabhukn_2002@
yahoo.co.in
Cite as: Satyanarayan and Prabhu, K. N., ‘‘Wetting Behaviour and Evolution of
Microstructure of Sn–Ag–Zn Solders on Copper Substrates with Different Surface
Textures,’’ J. ASTM Intl., Vol. 7, No. 9. doi:10.1520/JAI103052.
50
SATYANARAYAN AND PRABHU, doi:10.1520/JAI103052 51
of its low melting point, better wettability, good mechanical, fatigue, and creep
properties 关1,2兴. But the lead present in the solder material is highly toxic and
considered to be hazardous to the environment 关3兴. This has led to the devel-
opment of new lead-free solders like Sn–Cu, Sn–Ag–Cu, Sn–Ag, Sn–In, and
Sn–Bi for electronic applications in which Sn is a major element 关3–5兴.
Among Sn–Ag solders, Sn-3.5Ag solder has a higher melting point 共221° C兲
and poor wettabilty compared to Sn-37Pb solder with a freezing temperature of
183° C 关5兴. By the addition of alloying elements such as Cu, Zn, In, Ni, and Bi to
Sn-3.5Ag solders, it is possible to reduce the melting temperature and simulta-
neously improve wettabilty and mechanical properties 关3,5–7兴. McCormack and
Jin 关8兴 reported that addition of 1 wt % Zn to the Sn-3.5Ag solder alloy de-
creases the melting point of the solder alloy from 221 to 217° C. However,
Chang et al. 关9兴 noted that prolonging soldering time and increasing tempera-
ture are advantageous for the adhesion strength improvement of the Sn-9Zn-
1.5Ag/Cu and Sn-9Zn-2.5Ag/Cu interfaces, but it is detrimental to the Sn-9Zn-
3.5Ag/Cu interface because of microvoid formation due to increase in the Ag
content. Therefore, the wettabilty between solders and substrates is an impor-
tant parameter during soldering process and it plays a vital role in bond forma-
tion for improvement in the adhesion strength 关10,11兴. Wetting of liquid solder
on the substrate is an example of reactive wetting and is enhanced by the
addition of alloying elements to the solder 关11兴.
In soldering, how well the liquid solder wets the substrate is of fundamen-
tal importance and influences the quality and reliability of the solder joint. The
surface energy of the reacting liquid/solid interface is affected by surface char-
acteristics like surface roughness of the substrate. Mayappan et al. 关12兴 re-
ported that by increasing the roughness of the substrate, additional surface
area is produced, which causes an increase in its surface energy. Wenzel 关13兴
examined the effect of surface roughness on wetting behaviour and concluded
that the apparent contact angle decreases with the roughness ratio if the con-
tact angle is less than 90°. However, Shuttleworth and Bailey 关14兴 indicated that
the apparent contact angle increases with surface roughness. Chen and Duh
关10兴 reported that solder wettability degrade as substrates become rough.
The reactive wetting of a solder on a substrate is characterized not only by
the degree and rate of wetting but also by the formation of intermetallic com-
pounds 共IMCs兲关11兴. The degree of wetting is indicated by the contact angle
formed between the solidified solder and substrate at the interface 关11,15兴. The
contact angle formed at the interface is determined by using Young’s equation
␥sv − ␥sl = ␥lv cos ␪, where ␥ is the interfacial tension and subscripts s, l, and v
indicate the solid, liquid, and vapour phases 关11兴. ␪ is the contact angle of
solder as shown in Fig. 1.
The formation of IMC at the interface between the solidified solder and
substrates indicates good metallurgical bond and wettability 关10,11,15,16兴. The
type and morphology of IMCs significantly affect the solder joint reliability.
Kamal and Gouda 关17兴 reported that increase in Ag–Zn IMC at the interface
results in decrease in the adhesive strength. The bulk Ag3Sn plates affect the
plastic deformation properties of the solder and cause plastic-strain localiza-
tion at the boundary between Ag3Sn and ␤-Sn 关18兴. The knowledge of IMC
FIG. 1—Schematic sketch of liquid solder droplet resting on substrate 关11兴.
formation at the interface is therefore very much essential for predicting the
reliability of solder joints.
Most of the previous investigations were focused on the study of formation
of interfacial structure during soldering of Sn–Ag–Cu, Sn–Cu, and Sn–Ag–Zn
alloys solidifying against Cu substrates. A review of literature on Sn–Ag–Zn
solders suggests that there is no general agreement on the exact composition of
the solder that will improve all the properties 关19–22兴. It is found that a higher
Ag content degrades the adhesion strength of the solder. On the other hand, a
higher Zn content decreases the oxidation resistance and wettability. Hence, in
the present investigation, the Ag and Zn contents were varied to investigate
their effect on wettability of solders. Sn-2.625Ag-2.25Zn and Sn-1.75Ag-4.5Zn
alloys were selected for investigation. Further, the effect of surface texture of
the substrate on wetting behaviour and evolution of microstructure of these
lead-free solders on copper substrates is investigated.
Experimental Procedure
Sn-2.625Ag-2.25Zn and Sn-1.75Ag-4.5Zn alloy were prepared using commer-
cially procured ingots of pure Sn-3.5 Ag and Sn-9Zn alloys with 99.9+ % pu-
rity. The ingots were mixed in appropriate proportions depending on the type
of the alloy and melted in an electric resistance furnace to produce Sn–Ag–Zn
solder alloys. The melt was poured into a metallic mould to obtain solder in-
gots. The ingots were then analyzed for chemical composition by wet chemical
method. Table 1 gives the composition of solders used in the present work.
The ternary Sn–Ag–Zn alloy ingots were cut and drawn into solder wires
having a diameter of about 1.4 mm. Solder wire was melted using solder sta-
TABLE 1—Composition of solder alloys.
Constituent
Zn Ag Fe Sn
Percent by wt 2.25 2.62 ⬍0.005 Balance
4.5 1.75 ⬍0.005
tion 共KLAPP 920D兲 and solidified as balls having a weight of about 0.080 g. The
solder balls were then used for the assessment of wettability and kinetics of
spreading on copper substrates of different surface textures.
The surface roughness of substrates was measured using Form Talysurf 50
surface profiler. Contact angle measurements were carried out using FTA 200
dynamic contact angle analyzer. An environmental chamber with heating ele-
ment and temperature controller form the accessory for melting the solder ball
on the substrate for wetting studies. The system can capture both static and
dynamic spreading phenomena. The initial heating rate obtained with the
chamber is about 3 – 4 ° C / min, which eventually reduces as the chamber tem-
perature approaches the set value. Spherical balls of solder alloy were kept on
the substrate and the solder/substrate system was kept inside the environmen-
tal chamber after coating with proprietary flux 共inorganic acid, Alfa Aesar,
USA兲. The chamber was heated to a temperature higher than the liquidus tem-
perature of the solder alloy 共241° C for Sn-2.625Ag-2.25Zn and 244° C for Sn-
1.75Ag-4.5Zn solder兲 and maintained at that temperature during the entire pro-
cess of spreading. Images were captured at regular time intervals after
spreading has started. Initially the images were captured at a rate of 0.0167 fps,
and then the time of interval of image capture was incremented by 0.5 %. The
spreading process is recorded for ⬃2420 s. The captured images were ana-
lyzed using FTA 32 Video 2.0 software to determine the wetting behaviour.
The solder drop bonded to the substrate was sectioned along the axis and
polished using SiC papers of different grit sizes. The final polishing was carried
out on velvet cloth disk polisher using 1 ␮m lavigated alumina and then etched
with 5 % nital. There was no indication of embedded polishing particles in the
tested surfaces. The solder/substrate interfacial region was micro-examined
using Zeiss Axio-imager optical microscope as well as Jeol JSM 6380LA scan-
ning electron microscope. X-ray diffraction 共XRD兲 study was carried out to
identify and characterize the IMC at the solder/substrate system. A Jeol JDX-
8P-XRD system was used for this purpose.
Results and Discussions
The typical relaxation curves for spreading of Sn-2.625Ag-2.25Zn and Sn-

1.75Ag-4.5Zn solder on copper substrates with different surface roughness are
shown in Figs. 2 and 3, respectively. The decrease in contact angle relaxation
was sharp at the initial stages, and it became gradual as the solidifying solder
approached equilibrium. The liquid solder spreads rapidly in the beginning at a
time of about 100–120 s with a sharp increase in base radius. The spreading of
solder ceased after a time of about 900–1000 s. In the present study, the contact
angle at which relaxation rate becomes 0.01°/s was taken as the stabilized or
equilibrium contact angle. Each spreading experiment of solder on substrate is
carried out at least three times.
Equilibrium contact angle values in the range of 28°–41° were obtained on
disk polished Cu substrates having smooth surfaces for Sn-2.625Ag-2.25Zn. For
Sn-1.75Ag-4.5Zn solder, the contact angles were slightly higher and were in the
range of 42°–47°. SiC polished Cu substrates resulted in contact angles in the
FIG. 2—Relaxation behaviour of Sn-2.625Ag-2.25Zn solder on SiC paper polished cop-

per substrates.
range of 25°–33° for Sn-2.625Ag-2.25Zn and 35°–45° for the Sn-1.75Ag-4.5Zn

solder. The corresponding stabilized contact angles obtained with rough sur-
face textured substrates 共belt polished兲 for Sn-2.625Ag-2.25Zn and Sn-1.75Ag-
4.5Zn solder were in the range of 19°–35° and 38°–46°, respectively. Table 2
gives the values of static contact angles for copper substrates having different
surface textures. The contact angle decreases monotonically with the increase
in surface texture for Sn-1.75Ag-4.5Zn solder. Kumar and Prabhu 关2兴 reported
static contact angles in the range of 27°–34° on copper substrates for Sn-3.5Ag
solder. Arenas and Acoff 关23兴 obtained contact angles in the range of 17°–30° for
the same alloy using different types of fluxes.
For qualitative discussion and interpretation of the above results, dimen-
sionless parameters, namely, dimensionless contact angle 共␾兲 and dimension-
less time 共␶兲, are defined as follows:
FIG. 3—Relaxation behaviour of Sn-1.75Ag-4.5Zn solder on belt polished copper

substrates.
dimensionless contact angle, ␾ = 共 ␪ − ␪ r兲 / 共 ␪ i − ␪ r兲
dimensionless time, ␶ = 共 t − t i兲 / 共 t r − t i兲
where:
␪i = initial contact angle from which the relaxation is measured and
␪r = reference contact angle 共equal to 50 % of ␪i兲.
ti and tr are the corresponding values of time at the initial and reference
conditions.
Kinetics of Spreading
The following kinetic equation is proposed on the basis of relaxation behaviour
of the solders:
TABLE 2—Static contact angles under different surface texture of Sn-2.625Ag-2.25Zn and
Sn-1.75Ag-4.5Zn solder on Cu substrates.
Sn-2.625Ag-2.25Zn Sn-1.75Ag-4.5Zn
Ra ␪ Ra ␪
Type of Surface Treated 共␮m兲共 °兲共␮m兲共 °兲
Disk polish 共1 ␮m lavigated alumina兲 0.0212 39.89 0.0767 47.68
0.1346 41.30 0.1195 42.3
0.1380 28.14 0.171 45.13
SiC polish 共4/0-1000 grade兲 0.2017 25.04 0.2004 41.74
0.2387 32.63 0.2433 35.66
0.2906 25.04 0.6193 45.72
Belt polish 0.9886 35.41 1.555 46.8
1.1822 23.65 1.7607 44.8
2.3594 19.33 3.0307 38.56
␾ = exp共− K␶n兲
In order to find the parameter K and exponent n, the exponential power law
共EPL兲 equation is rearranged as
ln共− ln ␾兲 = ln K + n ln ␶
This equation is in the form of y = mx + c. Hence, the plot of ln共−ln ␾兲 versus
ln共␶兲 yields a straight line with slope “n” and y-intercept “ln K.” Typical EPL
plots for spreading of Sn-2.625Ag-2.25Zn and Sn-1.75Ag-4.5Zn solders on sub-
strates are shown in Figs. 4 and 5, respectively. The slope and y-intercept were
determined from the best fit equations. A reasonably good fit 共R2 ⱖ 0.9兲 was
found in most of the experiments indicating that the EPL equation could suc-
cessfully represent the spreading kinetics. A similar best fit equation was re-
ported by Kumar and Prabhu for Sn-3.5Ag and Sn-9Zn solders 关2兴. The calcu-
lated EPL parameters are presented in Table 3 for both solder materials.
High value of “K” results in rapid spreading in the initial stages of relax-
ation and a small value of n indicates that the liquid quickly spreads and attains
equilibrium value of contact angle in a short period of time. The effect of
roughness on EPL parameters was investigated. Figures 6 and 7 show the varia-
tion of EPL parameters 共K and n兲 with substrate roughness 共Ra兲 for spreading
of Sn-2.625Ag-2.25Zn, and the corresponding variations for Sn-1.75Ag-4.5Zn
solder are shown in Figs. 8 and 9. It was observed that both parameters 共K and
n兲 exhibited a decreasing trend with increasing substrate surface roughness.
The following best fit equations are used to correlate EPL parameters 共K
and n兲 with surface roughness 共Ra兲
K = A共Ra兲−B
n = C共Ra兲−D
where:
FIG. 4—Typical EPL plot for Sn-2.625Ag-2.25Zn spreading on belt polished Cu

substrates.
A, B, C, and D = regression constants.

The calculated values of regression constants are given in Table 4. A de-
creasing value of K implies slower kinetics during spreading and decreasing n
with increasing roughness indicates faster kinetics The values of regression
constants clearly indicate that the surface roughness affects the EPL parameter
n more significantly than K.
Figures 10 and 11 show microstructures of Sn-2.625Ag-2.25Zn and Sn-
1.75Ag-4.5Zn solder alloy solidified on copper substrate. The thickness of the
interface increases gradually with increase in surface roughness.
A continuous thin layer of IMC was observed at the interface 共Figs. 10共a兲
and 11共a兲兲. The thicknesses of the IMCs were in the range of 5 – 8 ␮m and
4 – 7 ␮m for Sn-2.625Ag-2.25Zn and Sn-1.75Ag-4.5Zn solders, respectively.
FIG. 5—Typical EPL plot for Sn-1.75Ag-4.5Zn solder spreading on SiC polished Cu
substrate.
TABLE 3—EPL parameters for spreading of solder Sn-2.625Ag-2.25Zn and Sn-1.75Ag-

4.5Zn on Cu substrates.
Sn-2.625Ag-2.25Zn Sn-1.75Ag-4.5Zn
Roughness, Ra Roughness, Ra
共␮m兲 K n R2 共␮m兲 K n R2
0.0212 2.24 0.60 0.90 0.0767 3.06 2.05 0.97
0.1346 1.07 0.61 0.95 0.1195 1.6 0.53 0.89
0.1380 1.20 0.72 0.65 0.171 2.95 1.69 0.83
0.2017 1.29 0.32 0.87 0.2004 1.31 0.89 0.94
0.2387 1.78 0.40 0.91 0.2433 2.48 1.58 0.96
0.2906 1.96 1.53 0.95 0.6193 1.68 0.62 0.90
0.9886 1.54 0.22 0.92 1.555 0.73 0.44 0.68
1.1822 1.12 0.20 0.86 2.359 0.73 0.15 0.95
2.3594 0.79 0.63 0.87 3.0307 1.69 0.6 0.86
FIG. 6—Variation of EPL parameter K with substrate roughness 共Ra兲 for spreading of
Sn-2.625Ag-2.25Zn.
Disk polishing on velvet cloth resulted in near uniform asperities, which results
in the diffusion of a small amount of solder into the copper substrate that
reacts with Cu to form a thin layer of IMC at the interface. Thin layer of IMC at
the interface indicates poor wetting of solder. The uniform asperities of smooth
surface act as barriers to the diffusion of liquid solder into the base metal.
Hence, the solder material shows poor wettability on disk polished Cu sub-
strates.
Figures 10共b兲 and 11共b兲 show microstructures at the solder/substrate inter-
faces of the Sn-2.625Ag-2.25Zn and Sn-1.75Ag-4.5Zn alloys, respectively. At
higher surface roughness, the IMC layer became thicker for both solders. Sn-
2.625Ag-2.25Zn solder/Cu exhibited growth stage of scallop IMC at the inter-
face 共Fig. 10共b兲兲. Sn-1.75Ag-4.5Zn/Cu showed nodular, globular shape of IMCs
in the matrix of the solder. The microscopic surface asperities on SiC polished
Cu substrates are responsible for extensive wetting. The dissolution of solder
into the substrate is faster compared to that for smooth polished substrates.
This dissolution of more amount of liquid solder into the base metal leads to an
FIG. 7—Variation of EPL parameter n with substrate roughness 共Ra兲 for spreading of
Sn-2.625Ag-2.25Zn.
increase in the thickness of IMC at the interface. Hence, the IMCs become
thicker with values of about 10 and 7.3 ␮m for Sn-2.625Ag-2.25Zn and Sn-
1.75Ag-4.5Zn solders on SiC polished Cu, respectively.
The interfacial microstructures of Sn-2.625Ag-2.25Zn and Sn-1.75Ag-4.5Zn
on belt polished Cu are shown in Figs. 10共c兲 and 11共c兲, respectively. The rough
surface texture is associated with a more contact area. The rough surfaces may
act as preferable sites for reaction, diffusion and nucleation 关11兴. Higher rough-
ness on the substrate also helps in the removal of the formation of oxide layer
of the liquid solder. An increase in surface roughness acts as non-barrier for the
diffusion of solidifying solder into the substrates, which leads to the dissolution
of more amount of molten solder into the substrate. Because of this dissolution
of liquid solder in larger quantity at higher surface texture, the size of the IMC
becomes sufficiently thick, larger and coarser. Sn-2.625Ag-2.25Zn exhibited
needle shaped IMC morphologies growing into the solder field 共Fig. 10共c兲兲 and
FIG. 8—Variation of EPL parameter K with substrate roughness 共Ra兲 for spreading of
Sn-1.75Ag-4.5Zn.
Sn-1.75Ag-4.5Zn shows plate like IMC in the solder matrix 共Fig. 11共c兲兲. Tables 5
and 6 show the thickness and interfacial characteristics of Sn-2.625Ag-2.25Zn
and Sn-1.75Ag-4.5Zn solder, respectively, at different substrate surface textures.
Yeh et al. 关24兴 reported the dissolution rate of Cu as a function of temperature
in various molten solders. Cu dissolution rate in Sn–Ag solder is 5.83
⫻ 10−1 ␮m / s and Sn–Zn solder is 2.50⫻ 10−1 ␮m / s at a temperature of
300° C.
Figure 12 shows the XRD pattern obtained for Sn-2.625Ag-2.25Zn on belt
polished Cu substrate system. XRD analysis clearly indicated the formation of
IMCs like Ag3Sn, Cu5Zn8, and Cu6Sn5 formed at the solder/Cu substrate inter-
face. Figure 13 shows the scanning electron microscopy 共SEM兲 micrograph of
the Sn-2.625Ag-2.25Zn on belt polished Cu substrate interface. The composi-
tion analysis at the interface clearly indicates the formation of Cu6Sn5 共+003
region, Fig. 13兲 IMC at the interface. Cu atoms from the substrate will combine
with Sn atoms, which have diffused from the solder matrix to form stable
FIG. 9—Variation of EPL parameter n with substrate roughness 共Ra兲 for the spreading
of Sn-1.75Ag-4.5Zn.
TABLE 4—Calculated values of regression constants for variation of EPL parameters “K”
and “n” with surface roughness.
Constants
Solder Alloy A B C D
Sn-2.625Ag-2.25Zn 1.2047 0.1855 0.2437 0.3148
Sn-1.75Ag-4.5Zn 1.0485 0.3879 0.4154 0.6943
FIG. 10—共a兲 Microstructure of Sn-2.625Ag-2.25Zn on Disk polished Cu substrate. 共b兲

Microstructure of Sn-2.625Ag-2.25Zn on SiC polished Cu substrate. 共c兲 Microstructure
of Sn-2.625Ag-2.25Zn on Belt polished Cu substrate.
FIG. 11—共a兲 Microstructure of Sn-1.75Ag-4.5Zn on Disk polished Cu substrate. 共b兲

Microstructure of Sn-1.75Ag-4.5Zn on SiC polished Cu substrate. 共c兲 Microstructure of
Sn-1.75Ag-4.5Zn on Belt polished Cu substrate.
TABLE 5—Thickness and interfacial characteristics for Sn-2.625Ag-2.25Zn solder on substrate.
Interfacial Layer Thickness
Mean Max
Type of Roughness Treated on Cu 共␮m兲共␮m兲 Remarks
Uniform thin continuous interface and in
Disk polished 5.18 8.38 some area diffusion of Cu into the solder
Thick continuous interface and exhibited
growth stage of rod/sharp IMCs into the
SiC polished 5.67 10.3 solder field

Thick interface and needle/sharp IMCs grown
Belt polished 7.98 11.54 into the solder field
TABLE 6—Thickness and interfacial characteristics for Sn-1.75Ag-4.5Zn solder on substrate.
Interfacial Layer Thickness
Mean Max
Type of Roughness Treated on Cu 共␮m兲共␮m兲 Remarks
Disk polished 4.19 6.59 The continuous thin interface
Thick continuous interface, nodular shaped
IMCs observed in the solder matrix; in a few
cases, the interface exhibited growth stage of
SiC polished 5.09 7.38 scallop IMC particles
Thick interface and scallop IMC particles
grown into the solder field; in a few cases,
blocky protrusions were grown into the solder
matrix; plate like IMCs occurred in the solder
Belt polished 8.15 12.94 matrix
FIG. 12—XRD pattern of Sn-2.625Ag-2.25Zn on belt polished Cu substrate.
Cu6Sn5 IMC layer. Table 7 gives the qualitative analysis of elements at the
interface. It indicates that region +003 is composed of Cu and Sn atoms in the
ratio of 6:5.
Figure 14 shows the XRD pattern obtained for Sn-1.75Ag-4.5Zn solder on
belt polished Cu substrate. Apart from IMCs Ag3Sn, Cu5Zn8, and Cu6Sn5 iden-
tified at the Sn-1.75Ag-4.5Zn solder/substrate interface, Ag-Zn IMCs are also
identified in the solder matrix system. Region P in Fig. 15 共SEM-energy-
dispersive X-ray spectroscopy micrograph兲 is composed of Ag and Sn in the
atomic ratio of 3:1, clearly indicating the presence of Ag3Sn. Table 8 indicates
the elemental analysis at the interface. The SEM micrograph of Sn-1.75Ag-
4.5Zn on belt polished Cu substrate is shown in Fig. 16. The area analysis 共Fig.
16兲 shows that the plates like IMCs in the solder matrix were composed of the
Ag and Zn elements. Table 9 gives the elemental analysis at the solder matrix,
indicating the formation of Ag and Zn rich phase IMCs have formed.
The morphologies of IMC at the interface and solder matrix for the two
solder alloys were found to be different. Sn-1.75Ag-4.5Zn on SiC and belt pol-
ished copper substrates exhibited nodular and plate shape Ag–Zn and Cu–Zn
IMCs in the solder matrix. A number of Cu-Sn IMCs in the form of scallop
morphologies occurred at the Sn-2.625Ag-2.25Zn solder /Cu substrates. The
rate of spreading was different for two alloys. The Sn-1.75Ag-4.5Zn solder has
FIG. 13—SEM micrograph of Sn-2.625Ag-2.25Zn on belt polished Cu substrate.
TABLE 7—EDS analysis results of 共+003 mark in Fig. 13兲 Sn-2.625Ag-2.25Zn solder on belt
polished Cu substrate interface.
Positions
Elements Region +003,at %

Cu K 14.63
Zn K 6.66
Ag L 5.18
Sn L 73.53
FIG. 14—XRD pattern of the Sn-1.75Ag-4.5Zn on belt polished Cu substrate.
shown poor wettabilty compared to Sn-2.625Ag-2.25Zn solder on copper sub-

strates. This is attributed to the formation of number of plate like Ag–Zn IMC at
the interface and matrix of Sn-1.75Ag-4.5Zn solder. Kamal and Gouda 关17兴
reported that the addition of Zn above 1 – 1.5 wt % decreases the wettabilty of
the solder. The other reasons for the variation in the results of spreading of
both solder alloys could be due to additional factors such as viscosity of solder,
chemical reaction, diffusion, solidification, etc. However, the wettabilties of
both solder alloys are within acceptable limits and can satisfactorily be used in
soldering applications.
Conclusion
Based on the results and discussion the following conclusions are drawn.
共1兲 High spreading rates 共2–5°/s兲 were observed during spreading of sol-
ders in the initial 10–15 s, whereas the relaxation rates were negligible
共⬍0.01° / s兲 after 1000 s.
共2兲 Sn-2.625Ag-2.25Zn solder material exhibited better wettability com-
pared to Sn-1.75Ag-4.5Zn solder on copper substrates.
共3兲 The EPL ␾ = exp共−K␶n兲 is proposed to represent the kinetics of spread-
FIG. 15—SEM micrograph of Sn-1.75Ag-4.5Zn on SiC polished Cu substrate.
TABLE 8—EDS analysis of 共region P兲 Sn-1.75Ag-4.5Zn on SiC polished Cu substrate.
Positions
Elements Region P, at %
Cu K ¯
Zn K ¯
Ag L 79.67
Sn L 20.33
FIG. 16—SEM micrograph of Sn-1.75Ag-4.5Zn on belt polished Cu substrate.
ing. A high value of K and a smaller value of n imply faster wetting

kinetics.
共4兲 The effect of surface roughness 共Ra兲 on EPL parameters 共K and n兲 can
be represented by the following best fit equations:
K = A共Ra兲−B
n = C共Ra兲−D
TABLE 9—EDS 共area兲 analysis of region square for Sn-1.75Ag-4.5Zn on belt polished Cu
substrate.
Positions
Elements Region square, at %

Cu K ¯
Zn K 65.82
Ag L 33.63
Sn L 0.55
共5兲 The contact angles obtained during spreading exhibited a decreasing

trend with increasing surface roughness for both solder/Cu substrate
systems. High contact angles were observed on disk polished sub-
strates, and low contact angles were observed on belt polished copper
surfaces. Intermediate values of contact angles were obtained on SiC
paper polished surfaces.
共6兲 Sn-2.625Ag-2.25Zn solder exhibited a thin continuous interface on disk
polished and thick interface on SiC polished Cu substrates. Belt pol-
ished Cu substrates exhibited coarser interface with needle shaped
IMCs growing into the solder field.
共7兲 On disk polished Cu substrates, Sn-1.75Ag-4.5Zn solder exhibited a
thin interface. A thick interface with nodular IMC in the solder matrix
is observed in SiC polished Cu substrate. Theses nodular IMCs were
transformed into plate like IMCs as the surface texture on belt polished
Cu substrates increased.
共8兲 The variation in the wetting behaviour of two solders is caused by the
change in IMC thickness at the interface and transformation in its mor-
phology.
Acknowledgments
The writers acknowledge the help received from Dr. Girish Kumar, former
Ph.D. scholar, NITK, and now Associate Professor, Department of Mechanical
Engineering, St. Joseph Engineering College, Mangalore, during contact angle
measurements. The writers also thank Ms. Kripa Suvarna, Research Scholar in
the Department of Metallurgical and Materials Engineering, NITK, for the help
received for SEM characterization. One of the writers 共K.N.P.兲 thanks the De-
fence Research Development Organization, Government of India, New Delhi,
for providing financial assistance for the procurement of the Dynamic Contact
Angle Analyzer.
References
关1兴 Manko, H. H., Solder and Soldering, 3rd ed., McGraw-Hill, Inc., New York, 1979,
pp. 1–153.
关2兴 Kumar, G. and Prabhu, K. N., “Wetting Behaviour and Evolution of Microstruc-
ture in Sn-37Pb and Sn-3.5Ag Solders,” Proceedings of the International Conference
on Advanced Materials and Composites 共ICAMC兲, Oct. 24–26, 2007, National Insti-
tute for Interdisciplinary Science and Technology, CSIR, Trivandrum, India, pp.
535–540.
关3兴 Wang, X., Liu, Y. C., Wei, C., Yu, L. M., Gao, Z. M., and Dong, Z. Z., “Effects of
Composition and Cooling Rate on the Microstructure of Sn-3.7Ag-0.9Zn-Bi Sol-
ders,” Appl. Phys. A: Mater. Sci. Process., Vol. 96, 2009, pp. 969–973.
关4兴 Vianco, P. T., Martin, J. J., Wright, R. D., and Hlava, P. F., “Dissolution and Inter-
face Reactions Between the Sn-3.9Ag-0.6Cu, Sn-0.7Cu and Sn-37Pb Solders on
Silver Base Metal,” Metall. Mater. Trans. A, Vol. 37A, 2006, pp. 1551–1561.
关5兴 Xu, R. L., Liu, Y. C., Han, Y. J., Wang, C. X., and Yu, L. M., “The Formation and
Evolution of Intermetallic Compounds Formed Between Sn–Ag–Zn–In Lead Free
Solder and Ni/Cu Substrate,” J. Mater. Sci.: Mater. Electron., Vol. 20, 2009, pp.
675–679.
关6兴 Liu, Y. C., Wan, J. B., and Gao, Z. M., “Intermediate Decomposition of Metastable
Cu5Zn8 Phase in the Soldered Sn–Ag–Zn/Cu Phase Interface,” J. Alloys Compd.,
Vol. 465, 2008, pp. 205–209.
关7兴 Wan, J. B., Liu, Y. C., Wei, C., Jiang, P., and Gao, Z. M., “Effect of the Soldering
Time on the Formation of Interfacial Structure Between Sn–Ag–Zn Lead-Free Sol-
der and Cu Substrate,” J. Mater. Sci.: Mater. Electron., Vol. 19, 2008, pp. 1160–
1168.
关8兴 Mccormack, M. and Jin, S., “Improved Mechanical Properties in New, Pb-Free
Solder Alloys,” J. Electron. Mater., Vol. 23, No. 8, 1994, pp. 715–720.
关9兴 Chang, T. C., Hon, M. H., and Wang, M. C., “Adhesion Strength of the Sn-9Zn-
xAg/Cu Interface,” J. Electron. Mater., Vol. 32, No. 6, 2003, pp. 516–522.
关10兴 Chen, Y. Y. and Duh, J. G., “The Effect of Substrate Surface Roughness on the
Wettability of Sn–Bi Solders,” J. Mater. Sci.: Mater. Electron., Vol. 11, 2000, pp.
279–283.
关11兴 Kumar, G., and Prabhu, K. N., “Review of Non-Reactive and Reactive Wetting of
Liquids on Surfaces,” Adv. Colloid Interface Sci., Vol. 133, 2007, pp. 61–89.
关12兴 Mayappan, R., Ismail, A. B., Ahmad, Z. A., Ariga, T., and Hussain, L. B., “Wetting
Properties of Sn–Pb, Sn–Zn and Sn–Zn–Bi Lead-Free Solders,” J. Teknologi, Vol.
46共C兲, 2007, pp. 1–14.
关13兴 Wenzel, R. N., “Resistance of Solid Surfaces to Wetting by Water,” Ind. Eng.
Chem., Vol. 28, No. 8, 1936, pp. 988–994.
关14兴 Shuttleworth, R. and Bailey, G. L. J., “The Spreading of a Liquid over a Rough
Solid,” Discuss. Faraday Soc., Vol. 3, 1948, pp. 16–22.
关15兴 Zhao, H., Nalagatla, D. R., and Sekulic, D. P., “Wetting Kinetics of Eutectic Lead
and Lead-Free Solders: Spreading over the Cu Surface,” J. Electron. Mater., Vol. 38,
No. 2, 2009, pp. 284–291.
关16兴 Islam, R. A., Chan, Y. C., Jillek, W., and Islam, S., “Comparative Study of Wetting
Behavior and Mechanical Properties 共Microhardness兲 of Sn–Zn and Sn–Pb Sol-
ders,” Microelectron. J., Vol. 37, 2006, pp. 705–713.
关17兴 Kamal, M. and Gouda, El. S., “Effect of Zinc Additions on Structure and Proper-
ties of Sn–Ag Eutectic Lead Free Solder Alloy,” J. Mater. Sci.: Mater. Electron., Vol.
19, 2008, pp. 81–84.
关18兴 Shen, J., Chan, Y. C., and Liu, S. Y., “Growth Mechanism of Bulk Ag3Sn Interme-
tallic Compounds in Sn–Ag Solder During Solidification,” Intermetallics, Vol. 16,
2008, pp. 1142–1148.
关19兴 Zou, H. F. and Zhang, Z. F., “Effect of Zn Addition on Interfacial Reactions Be-
tween Sn-4Ag Solder and Ag Substrates,” J. Electron. Mater., Vol. 37, No. 8, 2008,
pp. 1119–1129.
关20兴 Lin, K. L. and Shih, C. L., “Microstructure and Thermal Behavior of Sn–Zn–Ag
Solders,” J. Electron. Mater., Vol. 32, No. 12, 2003, pp. 1496–1500.
关21兴 Lin, K. L. and Shih, C. L., “Wetting Interaction Between Sn–Zn–Ag Solders,” J.
Electron. Mater. Vol. 32, No. 2, 2003, pp. 95–100.
关22兴 Wei, C., Liu, Y. C., and Wan, J. B., “Formation of Interfacial Structure of Sn–3.7Ag–
0.9Zn Eutectic Solder with Different Al Additions,” J. Mater. Sci.: Mater. Electron.,
Vol. 20, 2009, pp. 861–866.

关23兴 Arenas, M. F. and Acoff, V. L., “Contact Angle Measurements of Sn–Ag and Sn–Cu
Lead-Free Solders on Copper Substrates,” J. Electron. Mater., Vol. 33, No. 12, 2004,
pp. 1452–1458.
关24兴 Yeh, P. Y., Song, J. M., and Lin, K. L., “Dissolution Behavior of Cu and Ag Sub-
strates in Molten Solders,” J. Electron. Mater., Vol. 35共5兲, 2006, pp. 978–987.
doi:10.1520/JAI102939
Weiqiang Wang,1 Michael Osterman,1 Diganta Das,1 and

Michael Pecht2
Solder Joint Reliability of SnBi Finished

TSOPs with Alloy 42 Lead-Frame
under Temperature Cycling
ABSTRACT: Tin-bismuth 共SnBi兲 is a lead-free alternative to pure tin 共Sn兲

lead-frame finish. SnBi finish is considered by some to be a tin-whisker miti-
gation strategy. In selecting a SnBi finish, the interfacial strength and reliabil-
ity of solder interconnects formed with select assembly solders must be con-
sidered. To characterize the solder interconnect reliability of SnBi finished
parts, sets of test assemblies were created with Sn and SnBi finished thin-
small-outline-package 共TSOP兲 parts using SAC305 and SnPb solder. Test
assemblies were subjected to temperature cycling and interconnect strength
tests. It was found that SnBi finish caused TSOP solder joints to have a
shorter fatigue life than Sn finish under temperature cycling testing.
KEYWORDS: reliability, tin-bismuth, solder joint, temperature cycling,
shear test
Introduction
The electronics industry has transitioned to lead-free electronics both to com-
ply with government legislation and to be compatible with supply chain infra-
structure 关1兴. Pure tin 共Sn兲 finish has been used to replace traditional eutectic
tin-lead 共SnPb兲 finish for component terminations in most lead-free electronics.
However, tin finish has the potential to grow tin whiskers out from its surface,
which jeopardizes the reliability of electronics 关2,3兴. Thus, tin finish is prohib-
ited from being used in mission and life critical electronics 关4,5兴. Tin-bismuth
Manuscript received December 28, 2009; accepted for publication July 9, 2010; pub-
lished online August 2010.
1
Center for Advanced Life Cycle Engineering 共CALCE兲, Univ. of Maryland, College Park,
MD 20742.
2
Dept. of Electronics Engineering, City Univ. of Hong Kong, Tat Chee Avenue, Kowloon,
Hong Kong SAR.
Cite as: Wang, W., Osterman, M., Das, D. and Pecht, M., ‘‘Solder Joint Reliability of SnBi
Finished TSOPs with Alloy 42 Lead-Frame under Temperature Cycling,’’ J. ASTM Intl.,
Vol. 7, No. 8. doi:10.1520/JAI102939.
74
WANG ET AL., doi:10.1520/JAI102939 75
共SnBi兲 finish is an alternative lead-free termination finish that has been cited to
be more resistant to tin-whisker growth than Sn finish 关6–9兴. However, the
reliability of solder joints formed with SnBi finished terminations, particularly
with tin-lead solder, is a concern.
SnBi finish with a Bi content of 2–4 % by weight is recommended by the
International Electronics Manufacturing Initiative as a viable practice for miti-
gating tin-whisker risk 关10兴. During the reflow soldering process, solder joints
containing Bi are formed due to that the SnBi finish dissolves into the molten
solder. Bi dissolves into the Sn matrix of solders to form a solid solution. The
solubility of Bi in Sn is about 2.5 % by weight at room temperature 关11兴. If the
Bi percentage is above the solid solubility of Bi in Sn at a particular tempera-
ture, additional Bi would precipitate out as Bi phase 关12,13兴. The alloying of Bi
into Sn–Ag or Sn–Ag–Cu solders increases the strength of the solders due to the
solid solution strengthening effect of Bi in Sn and the dispersion strengthening
effect of precipitated Bi phase 关12,14兴. Kariya et al. 关12兴 showed that the alloy-
ing of Bi into Sn-3.5 % Ag solder caused the fatigue life of the solder in tension-
tension mode to decrease with the increased Bi percentage ranging from 2 % to
10 %. Kanchanomai et al. 关15兴 found that the alloying of Bi with a percentage of
1 % and 3 % caused the isothermal low cycle fatigue life of Sn-3Ag-0.5Cu to
decrease. Bradley et al. 关16兴 showed that the reliability of Sn–Ag–共3 %, 4.8 %, or
7.5 %兲Bi solder was lower than Sn–Ag–Cu or Sn-36Pb-2Ag solders under tem-
perature cycling testing for solder joints formed with SnPb solder plated thin-
small-outline-packages 共TSOPs兲. Park et al. 关17兴 found that SnBi finished
TSOPs with Alloy 42 leads had a 15 % shorter fatigue life than SnPb finished
TSOPs with Alloy 42 leads when they were assembled with SnPb solder paste,
although the solder joints had statistically the same solder joint strength. Cop-
per lead-frames had a much longer solder joint fatigue life than Alloy 42 lead-
frame for SnBi finished TSOPs assembled with SnPb solder paste. SnAgCu
solder paste assembled SnBi finished TSOPs had a longer fatigue life than SnPb
solder paste assembled SnBi finished TSOPs. Yoon et al. 关18兴 found that the
Sn-3.4Ag-3Bi-0.7Cu solder showed the same fatigue life as the Sn36Pb2Ag sol-
der in chip scale packages under temperature cycling. Other temperature cy-
cling studies conducted on solder joints formed with SnBi finished components
关19–23兴 found similar reliability for SnBi finish compared to SnPb finish and
other lead-free finishes 共e.g., Sn and SnCu兲. However, these studies did not
obtain enough failure data to perform a precise comparison between different
termination finishes.
In the study presented in this article, the reliability of SnBi finished com-
ponents with Alloy 42 lead-frames was studied under temperature cycling. The
strength of solder joints was also compared between terminals with SnBi finish
and Sn finish to study the effect of SnBi finish on the mechanical strength of
solder joints.
Experimental Materials and Procedures
The integrity of solder interconnects formed with SnBi finished terminals were
investigated under temperature cycling test conditions. The specifications of
the test specimens and test procedures are provided in the following sections.
TSOP 50 Resistor 2512
FIG. 1—Optical picture of test board.
Component
TSOPs manufactured by Amkor Technology were used in this study. Each pack-
age had 50 leads, 25 per side, with a pitch size of 0.8 mm. The package body
dimension was 20.95⫻ 10.36⫻ 1.00 mm3. Each package contained an encap-
sulated silicon die measuring 3.70⫻ 3.00⫻ 0.14 mm3. The die was non-
functional not connected to the lead-frame. The lead-frame material was Alloy
42 共42 % NiFe兲. The leads were electro-plated with Sn共2–4 %兲Bi finish or Sn.
Adjacent lead-frames were connected with wirebonds.
Board Assembly
Individual printed circuit boards 共PCBs兲 were created with either SnBi finished
TSOPs or Sn finished TSOPs. The parts were mounted on the custom designed
PCBs through a surface mount reflow process. The board included eight TSOP
positions and eight positions for 2512 resistors. No-clean solder pastes were
used during the reflow process. A sample test specimen is shown in Fig. 1. The
board material was Polyclad 370HRm and the exposed copper lands were
coated with organic solderability preservative 共OSP兲 finish applied on it. The
measured glass transition temperature of the board material was 153° C.
A lead-free solder paste, Sn3.0Ag0.5Cu 共SAC305兲共NC-SMQ230 produced by
Indium Corp.兲, and a non-clean eutectic SnPb 共SnPb兲共NC-SMQ51SC produced
by Indium Corp.兲 solder paste were used for the assembly. Different reflow
profiles were applied for the lead-free SAC305 solder paste and the eutectic
SnPb solder paste, as shown in Table 1.
Solder Joint Strength Evaluation

The integrity of the solder attachment was evaluated through a shear test. To
conduct the shear tests, individual SAC and SnPb mounted TSOP parts were
cut from the test assemblies. The influence of isothermal aging was considered
WANG ET AL., doi:10.1520/JAI102939 77
TABLE 1—Assembly matrix and reflow conditions.
Part Lead Finish Solder Paste Reflow Profile

Sn SnPb 70–80 s above 183° C; the peak temperature was
SnBi 215° C
Sn SAC 50–60 s above 217° C; the peak temperature was
SnBi 240° C
by subjecting the sets of individual TSOP assembly pieces to either 100° C for
24 h or at 125° C for 350 h. Non-aged test specimens were also tested.
The shear tests were conducted on a DAGE 2400 test system. The test
specimens were clamped to a fixture, and the shear force was applied onto the
lead using the shear tool as shown in Fig. 2. The standoff height of the shear
tool head was the distance from the copper pad on the board to the lower edge
of the shear tool head, as shown in Fig. 3. The standoff height was maintained
at 30 ␮m, which was higher than the solder joint height between the lead and
the copper pad to ensure that the shear force was applied to the solder joint
through the lead during the test. A gap between the vertical part of the lead and
the shear tool head was maintained to minimize the impact of the lead-frame’s
strength on the test results of solder joint strength. The shear test was
displacement-controlled with a constant shearing speed of 200 ␮m / s. The
shearing speed was based on recommendations for the ball grid array shear test
in JEDEC Standard JESD22-B117A 关24兴. The shearing was stopped when the
lead was sheared off of the solder joint. The maximum shear force during the
FIG. 2—Picture of test specimen and shear head on DAGE 2400 test system.
FIG. 3—Diagram of shear test setup.
test was recorded as the shear strength for the solder joint. Fourteen solder
joints on each test part were sheared. Average values and standard deviations
were calculated.
Temperature Cycling Test

To test solder interconnection reliability, three boards of each assembly combi-
nation were subjected to a temperature cycling test. Test boards were precon-
ditioned at 100° C for 24 h. The temperature cycle test profile was ⫺55 to
125° C with a 15 min dwell at each temperature extreme resulting in a 1 h
cycle, as shown in Fig. 4. Solder joint interconnects were monitored through
low resistance paths formed with board metallization and each mounted TSOP
part. The failure criterion was defined according to IPC Standard 9701 关25兴 in
combination with IPC SM-785 关26兴: The first interruption of electrical continu-
ity 共⬎300⍀兲 that is confirmed by nine additional interruptions within an addi-
tional 10 % of the cyclic life. Times to =failure data was collected for reliability
analysis. Failure analysis was performed on the assemblies after the tempera-
ture cycling test was completed.
Results and Discussion

The maximum forces used to shear the solder joints of SnBi finished TSOPs
were compared with those of Sn finished TSOPs. The temperature cycling test
was conducted to 100 % failure for the TSOP parts. The solder joint reliability
of SnBi finished parts was compared with that of Sn finished components.
Solder Joint Strength of SnBi Finished Thin-Small-Outline Packages

The measured maximum shear force for the test specimens is plotted in Fig. 5.
Test results indicate that SnPb solder joints formed with SnBi finished termi-
nals have similar strength to Sn finished terminals. For SAC soldered terminals,
the SnBi finished terminals showed a slightly higher mean maximum shear
WANG ET AL., doi:10.1520/JAI102939 79
15 min
125oC
Temperature
Time
15 min
-55oC
1 hour
FIG. 4—Temperature profile of temperature cycling test.
force compared with Sn finished terminals. However, only the 125C/350 h aging
condition was found to be statistically significant. Table 2 presents the analysis
of variance 共ANOVA兲 analysis between measured maximum shear force be-
tween SnBi finished and Sn finished terminals for different solder combina-
tions and aging conditions. From the test data, it was found that aging reduced
the shear force for the SAC soldered terminals. The maximum shear force in-
creased for the SnPb soldered terminals.
1800
1600
Shear force (gram-force)
1400
1200
1000
800
600 Sn finish-SnPb solder SnBi finish-SnPb solder
400 Sn finish-SAC solder SnBi finish-SAC solder
200
0
Non Aged 100C@24h 125C@350h
FIG. 5—Maximum forces used to shear solder joints during shear test.
TABLE 2—Single factor ANOVA with a significance level of ␣=0.05.
Data Group F Value Fcritical Value

Non-aged: 1.49 ⬍ 4.26
Sn finish–SAC solder
SnBi finish–SAC solder
100° C @ 24 h: 0.28 ⬍ 4.26
125° C @ 350 h: 6.11 ⬎ 4.23
Selected sheared solder joints were cross-sectioned to investigate the crack

paths. It was found that the shear test only damaged part of the solder joints, as
shown in Fig. 6. The crack path was found to be similar on all inspected shear
sites. So the solder joint forces measured only accounted for the forces used to
shear the solder under the leads but not the inner fillets.
The Bi content in the termination finish was measured by X-ray fluores-
cence to be 2.8+ 0.5 %. After reflow soldering, the Bi content in the solder
joints should have been much lower than this value since the solder paste vol-
ume was much larger than the termination finish volume. Although the alloy-
ing of Bi in the solders has been shown to increase the strength of the solder
alloys 关12,14兴, the percentage of the Bi in the solder joints in this study was too
low to reveal such effects. Also, since the failure sites were under the toes of the
leads, the interfacial adhesion strength between the solder and the alloy 42
leads may have played an important role in the shear strength measured, which
Alloy 42 lead
Solder
Copper pad Board
FIG. 6—Cross-section picture of a sheared TSOP solder joint.

WANG ET AL., doi:10.1520/JAI102939 81
(a) SnBi Finish-SnPb Solder Joint after aging at (b) SnBi Finish-SAC Solder Joint with aging
100oC for 24 hours
IMC
IMC
(c) SnBi Finish-SnPb Solder Joint after aging at (d) SnBi Finish-SnPb Solder Joint after aging at
125oC for 350 hours 100oC for 24 hours
FIG. 7—Microstructure of solder joints of SnBi finished TSOPs: 共a兲 SnBi finish-SnPb
solder joint after aging at 100°C for 24 h; 共b兲 SnBi finish-SAC solder joint with aging; 共c兲
SnBi finish-SnPb solder joint after aging at 125°C for 350 h; and 共d兲 SnBi finish-SnPb
solder joint after aging at 100°C for 24 h.
would have made the effects of Bi alloying less obvious. However, without
knowing the exact failure sites of the solder joints, it was not clear whether the
interfacial strength had played an important role or not.
For SAC solder joints, along with thermal aging, there was microstructure
coarsening and an increase in the volume fraction of intermetallic compounds
共IMCs兲 in the solder joints, as shown in Fig. 7共b兲 and 7共d兲. Right after reflow, a
fair amount of IMCs was distributed in the solder matrix since during reflow
process, Sn formed IMCs with Ag and Cu, which were already in the solder
pastes. After aging at 100° C for 24 h, the size of IMCs increased. Even large
scale IMCs with a length of around 50 ␮m were found. Microstructure coars-
ening and volume fraction increase of IMCs in the solder joints have been
shown to increase the embrittlement of the solder joints and to cause faster
TABLE 3—Single factor analysis of variance 共ANOVA兲 of shear strength of SnPb solder
paste assembled solder joints with a significance level of ␣=0.05.
Data Group F Value Fcritical Value

Sn finish–SnPb solder: 27.00 ⬎ 4.23
100° C @ 24 h
125° C @ 350 h
SnBi finish–SnPb solder: 32.82 ⬎ 4.23
100° C @ 24 h
125° C @ 350 h
crack propagation 关27,28兴. So the shear strength of the solder joints decreased
for solder joints assembled with SAC solder paste after thermal aging. However,
for solder joints assembled with SnPb solder paste, the solder joint shear
strength increased after aging at 125° C for 350 h comparing to aging at 100° C
for 24 h confirmed by ANOVA, as shown in Table 3. The difference between the
aging effects on the solder joint strength change of SnPb solder joints and SAC
solder joints was because of the IMC volume ratios are different between them.
There was no Ag or Cu from the SnPb solder paste to form any IMC with Sn
during the reflowing process. The IMCs were mainly formed at the interface
between the solder and the copper pad and between the solder and the alloy 42
lead. Although Cu can be dissolved into the solder both during reflow process
and during aging process to form IMC with Sn, the amount of IMCs formed
was smaller comparing to the IMCs in SAC solder joints. After relatively long
term aging at 125° C for 350 h, more IMCs were formed in the solder matrix
and the size of IMCs increased because of their growth, as shown in Fig. 7共a兲
and 7共c兲. The IMCs increased the solder joint strength because of second phase
strengthening effect. So the shear strength of SnPb solder joints increased after
aging at 125° C for 350 h.
Reliability of SnBi Refinished Thin-Small-Outline Packages Under Temperature

Cycling Test
For both SAC305 and SnPb soldered assemblies, SnBi finished TSOPs showed
lower reliability than Sn finished TSOPs under temperature cycling conditions,
as shown in Fig. 8. When assembled with the same solder paste, SnPb or
SAC305, SnBi finished TSOPs and Sn finished TSOPs had close Weibull plot
slopes. Compared to Sn finish, SnBi finish showed a 24 % shorter mean-time-
to-failure 共MTTF兲 for SAC305 solder paste and an 11 % shorter MTTF for SnPb
solder paste, as shown in Fig. 9. SAC solder paste assembled solder joints had
longer fatigue life than SnPb solder paste assembled solder joints.
During the temperature cycling test, cracks were initiated and propagated
through the solder joints. This finally caused the daisy chains to fail. The fa-
tigue life of a solder joint depends on the crack initiation and propagation
process. The material combination of lead-frames, solder alloys, and copper
pads, the geometry of solder joints, the stress conditions, defects, and the mi-
crostructure of the solder joints all had effects on the crack initiation and
WANG ET AL., doi:10.1520/JAI102939 83
99
Weibull
SnBi finish-SnPb solder
90
ESTIMATED FAILURE PROBABILITY (%)
W2 RRX - SRM MED
F=24 / S=0
Sn finish-SnPb solder
50
W2 RRX - SRM MED
F=24 / S=0
SnBi finish-SAC solder
W2 RRX - SRM MED
F=24 / S=0
10
Sn finish-SAC solder
W2 RRX - SRM MED
5
F=24 / S=0
1
1000 4000
Cycles
β1=10.7 η1=1770 ρ1=0.96
β2=9.1 η1=2353 ρ1=0.97
β3=8.3 η1=2594 ρ1=0.99
β4=7.5 η1=2941 ρ1=0.98
FIG. 8—Failure probability of TSOP solder joints under temperature cycling testing.
3000
.
2500
Mean-Time-to-Failure (Cycles)
2000
1500
1000
500
0
Sn finish-SnPb SnBi finish-SnPb Sn finish-SAC SnBi finish-SAC
solder solder solder solder
FIG. 9—MTTF of solder joints under temperature cycling test.

Crack path
Heel
Inner fillet
Lead
Copper pad Board
FIG. 10—Crack path in solder joints assembled with SnPb solder paste.
propagation process. In SnPb solder joints, the crack propagated through the
inner fillets of the solder joints and along the interface between the bulk solder
and the Alloy 42 leads during the temperature cycling test, as shown in Fig. 10.
A discontinuous layer of solder was found on the alloy 42 lead side from the
magnified picture in Fig. 10. This means that the crack went through both the
bulk solder and the interface. Similar crack paths have been observed by litera-
ture studies 关29–31兴 when Sn finished TSOPs assembled with SnPb solder paste
were temperature cycled.
When the TSOPs were assembled with SAC305 solder paste, the SnBi fin-
ished TSOPs also had a shorter solder joint fatigue life than the Sn finished
TSOPs. The crack paths essentially followed the same route as the cracks in
SnPb solder assembled solder joints, as shown in Fig. 11. However, the differ-
ence was that the cracks propagated through the bulk solder near the Alloy 42
leads but not along the interface.
The observed crack paths resulted from the combined effects of global and
local coefficient of thermal expansion 共CTE兲 mismatches during the tempera-
ture cycling test 关32兴. Global CTE mismatch refers to the CTE mismatch be-
tween the TSOP component and the PCB, while local CTE mismatch is between
the Alloy 42 lead and the solder alloy. As the CTE difference between Alloy 42
共4–5 ppm/K兲关30兴 and solder alloy 共21–25 ppm/K兲关33兴 is large, literature studies
have shown that local CTE mismatch could be a reliability concern under tem-
perature cycling conditions 关32,34兴. The crack propagation at the interface be-
tween the Alloy 42 lead and the bulk solder was due to the combined effects of
global and local CTE mismatch. Lee et al. 关35兴 showed through finite element
simulation and temperature cycling testing that the cracks initiated at the inner
fillet surface of the bulk solder located beside the heels of the leads. The study
showed that the crack initiation was due to the higher local stress concentra-
tion at the site. Thus, in the current study, during the temperature cycling
testing, cracks probably initiated at the inner fillet surface of the bulk solder
located beside the heel of the leads due to global CTE mismatch. It may also
have initiated under the toe of the leads, as the solder fillet was thinner at this
WANG ET AL., doi:10.1520/JAI102939 85
Lead
Crack path
Copper pad Board
FIG. 11—Crack path in solder joints assembled with SAC solder paste.
location and the stress concentration was also higher there than at other loca-
tions. After crack initiation, the crack propagated along the interface between
the Alloy 42 lead and the bulk solder due to the combined effects of global and
local CTE mismatches.
While the addition of Bi does not appear to affect SnPb solder attach
strength, the addition of Bi does result in a measurable decrease in temperature
cycling fatigue life for SnPb solder joints. Similar reports have been made by
Park et al. 关17兴 comparing SnPb finish and SnBi finish. The alloying of Bi in the
SnPb solder joint decreased the fatigue life.
For SAC305 solder interconnects, the addition of Bi has also reduced the
temperature cycle fatigue life. However, the impact was lower than observed for
SnPb solder interconnects. Similar results have been SnAg solder interconnects
关36兴. For SAC305 solder interconnects, the addition of Bi increases the solder
joint strength. It has been reported that the addition of Bi to tin-silver-copper
solder increased solder strength and decreased elongation 关14兴. This effect may
result in higher inelastic damage to solder interconnect and reduced tempera-
ture cycling fatigue life.
Conclusions
Sn共2.0–4.0 %兲Bi and Sn finished Alloy 42 lead-frame TSOPs were assembled
with eutectic SnPb solder paste and Sn3.0 %Ag0.5 %Cu onto PCBs with OSP
finish. Shear tests were conducted to evaluate the impact of Bi on solder joint
strength. The addition of Bi did not impact the maximum shear force for SnPb
solder joints but resulted in a slight increase for SAC 305 solder joints. A tem-
perature cycling testing was conducted to determine the reliability of solder
joints formed with Sn共2.0–4.0 %兲Bi finished TSOPs in comparison with Sn
finished TSOPs. The presence of Bi was found to reduce the temperature cy-
cling fatigue life of both SnPb and SAC305 solder joints. The reduction in
temperature fatigue life was found to be greater for SnPb solder joints. The
results are consistent with other studies and indicated that designer should
take the reduction in temperature cycling fatigue life into consideration when
using parts with SnBi finished terminations.
Acknowledgments
The writer would like to thank Jeff Kennedy and Celestica for performing the
surface mount assembly. The writers would also like to thank the members of
the CALCE Electronics Products and Systems Consortium for their support of
this study.
References
关1兴 Ganesan, S. and Pecht, M., Lead-Free Electronics, John Wiley and Sons, Inc., New
York, 2006.
关2兴 Brusse, J., “Tin Whisker Observations on Pure Tin-Plated Ceramic Chip Capaci-
tors,” Proc. AESF SUR/FIN, Chicago, IL, June 25, 2002, AESF Society, Orlando, FL,
pp. 45–61.0002-7820
关3兴 NASA, Tin Whisker Failures, 2007, http://nepp.nasa.gov/WHISKER/failures/
index.htm 共Last accessed Aug. 1, 2010兲.
关4兴 McDowell, M. E., “Tin Whisker: A Case Study,” IEEE Proceedings, Aerospace Ap-
plication Conference, Steamboat, Co., Jan. 31–Feb. 5, 1993, IEEE, New York, pp.
207–215.
关5兴 Pinsky, D., Osterman, M., and Ganesan, S., “Tin Whiskering Risk Factors,” IEEE
Trans. Compon. Packag. Technol., Vol. 27, No. 2, 2004, pp. 427–431.
关6兴 Kim, K. S., Yu, C. H., and Yang, J. M., “Tin Whisker Formation of Lead-Free Plated
Leadframes,” Microelectron. Reliab., Vol. 46, 2006, pp. 1080–1086.
关7兴 Kim, K. S., Yu, C. H., and Yang, J. M., “Behavior of Tin Whisker Formation and
Growth on Lead-Free Solder Finish,” Thin Solid Films, Vol. 504, 2006, pp. 350–
354.
关8兴 Hillman, D., Margheim, S., and Straw, E., “The Use of Tin/Bismuth Plating for Tin
Whisker Mitigation on Fabricated Mechanical Parts,” Proceedings of CALCE Inter-
national Symposium on Tin Whiskers, College Park, MD, April 24–25, 2007,
CALCE, College Park, MD, http://www.calce.umd.edu/tin-whiskers/ISTW2007.htm
共Last accessed Aug. 1, 2010兲.
关9兴 Mathew, S., Osterman, M., Shibutani, T., Yu, Q., and Pecht, M., “Tin Whisker: How
to Mitigate and Manage the Risks, Proceedings of High Density and Microsystem
Integration, Shanghai, China, June 26–28, 2007, IEEE, New York, pp. 1–8.
关10兴 International Electronics Manufacturing Initiative 共iNEMI兲, Tin Whisker User
Group, “iNEMI Recommendations on Lead-Free Finishes for Components Used in
WANG ET AL., doi:10.1520/JAI102939 87
High-Reliability Products,” Dec. 2006, pp. 1–23, http://thor.inemi.org/

webdownload/projects/ese/tin_whiskers/Pb-Free_Finishes_v4.pdf 共Last accessed
on Feb. 10, 2009兲.
关11兴 Reinikainen, T. and Kivilahti, J. K., “Mechanical and Microstructural Characteris-
tics of Dilute SnBi and SnBiIn Alloys,” IEEE Proceedings, International Sympo-
sium on Advanced Packaging Materials, Braselton, GA, 1998, IEEE, New York, pp.
170–174.
关12兴 Kariya, Y., and Otsuka, M., “Effect of Bismuth on the Isothermal Fatigue Proper-
ties of Sn-3.5mass%Ag Solder Alloy,” J. Electron. Mater., Vol. 27, No. 7, 1998, pp.
866–870.
关13兴 Wu, C. M. L., and Huang, M. L., “Microstructural Evolution of Lead-Free Sn–Bi–
Ag–Cu SMT Joints During Aging,” IEEE Trans. Adv. Packag., Vol. 28, No. 1, 2005,
pp. 128–133.
关14兴 Zhao, J., Qi, L., Wang, X., and Wang, L., “Influence of Bi on Microstructure Evo-
lution and Mechanical Properties in Sn–Ag–Cu Lead-Free Solder,” J. Alloys
Compd., Vol. 375, 2004, pp. 196–201.
关15兴 Kanchanomai, C., Miyashita, Y., and Mutoh, Y., “Low-Cycle Fatigue Behavior of
Sn–Ag, Sn–Ag–Cu, and Sn–Ag–Cu–Bi Lead-Free Solders,” J. Electron. Mater., Vol.
31, No. 5, 2002, pp. 456–465.
关16兴 Bradley, E., “Lead-Free Solder Assembly: Impact and Opportunity,” IEEE Proceed-
ings, Electronic Components and Technology Conference, New Orleans, LA, May 30,
2003, IEEE, New York, pp. 41–26.
关17兴 Park, K., Kim, N., and Oh, S., “Reliability of the SnBi Lead Finished Pb-Free
Product,” IEEE Proceedings, 4th International Symposium on Electronic Materials
and Packaging, Kaohsiung, Taiwan, Dec. 4–6, 2002, IEEE, New York, pp. 150–156.
关18兴 Yoon, S. W., Park, C. J., Hong, S. H., Moon, J. T., Park, I. S., and Chun, H. S.,
“Interfacial Reaction and Solder Joint Reliability of Pb-Free Solders in Lead
Frame Chip Scale Packages 共LF-CSP兲,” J. Electron. Mater., Vol. 29, No. 10, 2000,
pp. 1233–1240.
关19兴 Han, J., Kim, N., Kim, M., and Lee, W., “Pb-Free Memory Module with Sn-Bi Lead
Finish and Sn–Ag–Cu Solder Paste,” AESF SURFIN2000, Chicago, IL, June 2000,
AESF Society, Orlando, FL, pp. 1–5.
关20兴 Henshall, G., Roubaud, P., and Chew, G., “Impact of Component Terminal Finish
on the Reliability of Pb-Free Solder Joints,” Journal of Surface Mount Technology,
Vol. 15-4, 2002, pp. 1–10.
关21兴 Wulfert, F.W. and Vo, N.D., “Assessment of Pb-Free Finishes for Leadframe Pack-
aging,” IPC Proceedings, Electronic Circuits World Convention, Cologne, Germany,
Oct. 7–9, 2002, IPC, Bannockburn, IL, IPC156.
关22兴 Sriyarunya, A., “Manufacturability and Reliability of Lead-Free Packages,” IEEE
Proceedings, International IEEE Conference on Asian Green Electronics, Hong Kong
and Shenzhen, China, Jan. 5–9, 2004, IEEE, New York, pp. 105–109.
关23兴 Nakadaira, Y., Vo, N. D., Sundram, B., Matsuura, T., Kangas, R., Lee, K., Tsuriya,
M., Conrad, J., and Arunasalam, S., “Pb-Free Plating for Peripheral/Leadframe
Packages,” IEEE Proceedings EcoDesign: Second International Symposium on En-
vironmentally Conscious Design and Inverse Manufacturing, Tokyo, Japan, Dec.
11–15, 2001, IPC, Bannockburn, IL, pp. 213–218.
关24兴 JEDEC, JESD22-B117A Standard, “Solder Ball Shear.”
关25兴 IPC Standard 9701, January 2002, “Performance Test Methods and Qualification
Requirements for Surface Mount Solder Attachments,” IPC, Bannockburn, IL.
关26兴 IPC Standard SM-785, November 1992, “Guidelines for Accelerated Reliability
Testing of Surface Mount Solder Attachments,” IPC, Bannockburn, IL.
关27兴 Sheen, M. T., Chang, C. M., Teng, H. C., Kuang, J. H., and Hsieh, K. C., “Influence
of Thermal Aging on Joint Strength and Fracture Surface of Pb/Sn and Au/Sn
Solders in Laser Diode Packages,” J. Electron. Mater., Vol. 31, No. 8, 2002, pp.
895–902.
关28兴 Kim, K. S., Huh, S. H., and Suganuma, K., “Effects of Intermetallic Compound on
Properties of Sn–Ag–Cu Lead-Free Soldered Joints,” J. Alloys Compd., Vol. 352,
2003, pp. 226–236.
关29兴 Yoon, S., Hong, J., Kim, H., and Byun, K., “Board-Level Reliability of Pb-Free
Solder Joints of TSOP and Various CSPs,” IEEE Trans. Electron. Packag. Manuf.,
Vol. 28, No. 2, 2005, pp. 168–175.
关30兴 Noctor, D. M., Bader, F. E., Viera, A. P., Boysan, P., Suresh, G., and Foehringer, R.,
“Attachment Reliability Evaluation and Failure Analysis of Thin Small Outline
Packages 共TSOP’s兲 with Alloy 42 Leadframes,” IEEE Trans. Compon., Hybrids,
Manuf. Technol., Vol. 16, No. 8, 1993, pp. 961–971.
关31兴 Seyyedi, J., Iannuzzelli, R., and Bukhari, J., “Reliability Evaluation of TSOP Solder
Joints for PC Card Application,” Soldering Surf. Mount Technol., Vol. 8, No. 2,
1996, pp. 29–32.
关32兴 Clech, J., John, M., Noctor, D., Bader, F., and Augis, J., “A Comprehensive Surface
Mount Reliability Model Covering Several Generations of Packaging and Assembly
Technology,” IEEE Trans. Compon., Hybrids, Manuf. Technol., Vol. 16, No. 8, 1993,
pp. 949–960.
关33兴 NIST, Database for Solder Properties, http://www.boulder.nist.gov/div853/
lead_free/part1.html#%201.23. 共Last accessed Jan. 3, 2009兲.
关34兴 Baker, D., Gupta, V., and Cluff, K., “Solder Joint Crack Initiation and Crack Propa-
gation in a TSOP Using Strain Energy Partitioning,” Proc., Advances in Electronic
Packaging, ASME, Int. Electronic Packaging Conf., Binghamton, NY, 1993, ASME,
New York, Vol. 4–2, pp. 943–949.
关35兴 Lee, S. and Lee, K., “Thermal Fatigue Life Prediction of Gull-Wing Solder Joints in
Plastic Thin Small Outline Packages,” Jpn. J. Appl. Phys., Vol. 35, 1996, pp. L1515–
L1517.
关36兴 Kariya, Y., Hirata, Y., and Otsuka, M., “Effect of Thermal Cycles on the Mechanical
Strength of Quad Flat Pack Leads/Sn-3.5Ag-X 共X⫽Bi and Cu兲 Solder Joints,” J.
Electron. Mater., Vol. 28, No. 11, 1999, pp. 1263–1269.
doi:10.1520/JAI103064
Konstantina Lambrinou1 and Werner Engelmaier2
The Microstructural Aspect of the

Ductile-to-Brittle Transition of Tin-Based
Lead-Free Solders
ABSTRACT: This work focuses on specific aspects of the ductile-to-brittle

transition in the fracture behavior of tin-based lead-free solders. This transi-
tion is essentially associated with the crystal structure of ␤-Sn, which is the
main constituent of these solders. Moreover, the transition is affected by
many factors, including the ambient temperature, the applied strain rate, the
mechanical constraint, and certain solder microstructural features such as
the shape, size, and spatial distribution of intermetallic particles. Since the
mechanical constraint in the solder is related with the specimen dimensions,
this work compares the fracture behavior of two different sizes of specimens
made of tin-based solders: Rectangular beams and solder joints. Both types
of specimens were tested in impact, while the produced fracture surfaces
were studied using scanning electron microscopy. The detailed fractography
analysis allowed the correlation of the overall solder fracture behavior with
certain features in the solder microstructure. This study used also the addi-
tional insight into the embrittlement mechanism of tin-based solders to ex-
plain previous results from the thermal cycling of eutectic tin-lead solder
joints.
KEYWORDS: lead-free soldering, tin-based solders, ductile-to-brittle
transition, fracture behavior, fractography, solder joint reliability
Introduction
In the framework of a worldwide awakening with respect to all industrial ac-

tivities that can potentially harm the environment, various legislations restrict-
ing the use of lead 共Pb兲 in electrical and electronic equipment were introduced.
Manuscript received March 3, 2010; accepted for publication June 8, 2010; published
online July 2010.
1
imec, Kapeldreef 75, B-3001 Leuven, Belgium.
2
Engelmaier Associates L.C., 7 Jasmine Run, Ormond Beach, FL 32174.
Cite as: Lambrinou, K. and Engelmaier, W., ‘‘The Microstructural Aspect of the Ductile-
to-Brittle Transition of Tin-Based Lead-Free Solders,’’ J. ASTM Intl., Vol. 7, No. 7.
doi:10.1520/JAI103064.
89
The desired restriction in the amount of Pb necessitates the replacement of the

traditional near-eutectic tin-lead solder 共Sn-37Pb兲 by a lead-free 共Pb-free兲 solder
that meets the requirements imposed by the end application. The most promi-
nent Pb-free solders that are currently used in several mainstream electronic
applications are tin-based 共Sn-based兲 solders that contain different amounts of
one or more alloying elements, like copper 共Cu兲, silver 共Ag兲, nickel 共Ni兲, etc.
One of the main reliability concerns related to the performance of commer-
cial Sn-based solders is the fact that they are more prone to impact failures in
comparison to the Sn-37Pb solder 关1–11兴. The superior impact reliability of
Sn-37Pb is attributed to the fact that this solder is more compliant than its
Sn-based potential replacements with an about 40 % lower modulus of elastic-
ity, thus accommodating the stresses at the solder/bond pad interface more
efficiently. The failure modes that have been widely identified in literature for
Sn-based solders in conventional electronic applications are ductile failures in
the solder bulk, brittle failures at the intermetallic 共IMC兲 layers that form at the
solder/bond pad interface, and mixed ductile-brittle failures, i.e., failures that
occur partly in the solder bulk and partly in the interfacial IMCs 关1,2,7–15兴. The
above wide perception of the possible failure modes in Sn-based solders is
consolidated in two standards from the Joint Electron Devices Engineering
Council: 共a兲 JESD22-B117A on “Solder Ball Shear” 关16兴 and 共b兲 JESD22-B115
on “Solder Ball Pull” 关17兴. Other failure modes that compromise the impact
reliability of Pb-free solder joints, e.g., “pad cratering” 关18兴, but cannot be di-
rectly related to solder failure will not be addressed here.
The impact reliability of Sn-based solders is further compromised when the
service temperature decreases due to the ductile-to-brittle transition in the frac-
ture behavior of ␤-Sn. One of the first persons to observe the negative effects of
this ductile-to-brittle transition on the performance of Sn-based solders was
Wild of IBM 关19兴. The embrittlement of ␤-Sn, which is the main solder con-
stituent, is responsible for a new brittle failure mode that occurs in the solder
bulk, in contrast with the typical brittle failures occurring in the interfacial
IMCs. This novel mechanism of solder embrittlement and the factors affecting
it have recently been addressed in the literature for solder joints subjected to
impact 关20兴. To be more precise, the factors affecting the ductile-to-brittle tran-
sition in the fracture behavior of Sn-based solders include the temperature, the
applied strain rate, the mechanical constraint of the material, and the distribu-
tion of IMC particles in the solder 关20兴. Actually, the effect of IMCs on the
mechanical properties of Sn-based solders has been recognized very early in
the advent of surface mount soldering 关21,22兴. Today, the electronics industry
openly acknowledges that adding gold 共Au兲, Ag, and palladium 共Pd兲 to the sol-
der has a negative effect on solder joint reliability due to the embrittlement
associated with the formation of Au–Sn, Ag–Sn, and Pd–Sn IMCs 关23,24兴.
Since the mechanical constraint changes with the specimen dimensions,
thus affecting the solder fracture behavior, this article will attempt a compari-
son of the fracture behavior of bulk solder specimens with the previously re-
ported fracture behavior of solder joints 关20兴. The comparison is aided by the
fact that both bulk specimens and solder joints have been tested in impact
under similar strain rate conditions. Studying the fracture behavior of bulk
solder specimens is also expected to provide a better understanding of the in-
LAMBRINOU AND ENGELMAIER, doi:10.1520/JAI103064 91
trinsic fracture behavior of Sn-based solders since bulk solder specimens have
fewer constraints than solder joints. In fact, solder joints are not just confined
between the metal surfaces to which they are soldered, they are also confined
between the IMC layers forming at the bond pad/solder joint interface; this
confinement influences greatly their fracture behavior. For example, the low-
temperature impact reliability of Sn-based solders can be seriously compro-
mised by the exaggerated occasional growth of interfacial IMCs 共IMC lengths
⬎ 100 ␮m兲, which act as detrimental “external notches” for these notch-
sensitive solders 关20,25兴. This article correlates the intrinsic fracture behavior
of Sn-based solders with the solder microstructure by performing an extensive
fractography study on the tested solders using scanning electron microscopy
共SEM兲. The fractography results are presented separately for each solder so as
to better correlate solder-specific microstructural features with fracture behav-
ior.
Experimental Procedure
All solders studied in this work were produced by conventional casting in a

small mold to ensure fast solidification and produce solder microstructures
that were comparable to the microstructure of commercial solder joints
关26,27兴. The test specimens were Charpy V-notch 共CVN兲 rectangular bars of two
different dimensions: 10⫻ 10⫻ 55 mm3 共standard dimensions兲 and 5 ⫻ 5
⫻ 55 mm3 共smaller dimensions due to limited solder availability兲, as previously
reported 关26,27兴. The solder alloys used to make the standard-size specimens
were Sn-3Ag-0.5Cu 共SAC 305兲, Sn-4Ag-0.5Cu 共SAC 405兲, Sn-5Ag, and Sn-37Pb,
while the solders used to make the smaller specimens were Sn-0.7Cu, Sn-0.7Cu-
0.1Ni, and nearly pure 共99.99 %兲 Sn; the numbers in the solder alloy designa-
tion express the mass percent of the element they precede. The depth of the
V-notch was 2.5 mm for all solders, except for the 1.3-mm-deep notch intro-
duced in the Sn-0.7Cu solder specimens. All impact tests were performed ac-
cording to the test methodology described in ASTM E23-06 关28兴. The solder
CVN specimens were tested in impact between −190 and 100° C so as to study
the temperature dependence of the solder fracture behavior. Heating of the test
specimens was done in hot water, while cooling was ensured with the aid of dry
ice 共CO2 pellets兲 or ethanol cooled in liquid N2. After having stabilized the
sample temperature for 10 min at the desired level, the specimens were imme-
diately tested, as previously reported 关26,27兴. The impact velocity during the
testing of these solders was 3.8 m/s 关26,27兴, which was comparable with the
impact velocity 共3.16 m/s兲 during the impact testing of Sn-based solder joints in
Ref 20.
The fractography study on the tested specimens was performed using the
following scanning electron microscopes: JSM-5600 LV 共JEOL, Tokyo, Japan兲,
XL 30 FEG, and Quanta Inspect D8334 共both from FEI, Eindhoven, The Neth-
erlands兲. Qualitative elemental analysis of selected microstructural features
was performed using 共a兲 beryllium-window energy dispersive X-ray spectrom-
eter 共EDS兲 and dedicated analysis software 共RÖNTEC, Berlin, Germany兲 on the
JSM-5600 LV and 共b兲 INCA PentaFET-x3 detector and INCA Energy analysis
SAC 305 SAC 405 Sn-37Pb Chart Title 99.99% Sn

Sn-5Ag Sn-0.7Cu-0.1Ni Sn-0.7Cu
-200 -180 -160 -140 -120 -100 -80 -60 -40 -20 0 20 40 60 80 100 120
50
50
)J
ΔE (J)
(


50
Temperature (°C)
FIG. 1—Impact energy, ⌬E, as a function of the test temperature for all tested solders.
Open markers indicate data produced by testing the standard-size CVN specimens
共10 ⫻ 10 ⫻ 55 mm3兲, while filled markers indicate data resulting from testing the smaller
CVN specimens 共5 ⫻ 5 ⫻ 55 mm3兲. Due to the variation in specimen size, two energy
scales are used: A 50-J scale for the standard specimens 共left兲 and a 5-J energy scale for
the smaller ones 共right兲. The shaded areas superimposed on the impact resistance
curves of solders SAC 305, SAC 405, and Sn-5Ag emphasize the fact that the ductile-to-
brittle transition in their fracture behavior occurs over a whole temperature range and
not at a single DBTT value.
software 共Oxford Instruments, Oxfordshire, United Kingdom兲 on the XL 30

FEG.
Results
Impact Test Results

The plot in Fig. 1 shows the energy absorbed during impact as a function of
temperature for all tested solders. Since the energy absorbed during impact is
equal to the change in the potential energy of the hammer with which the
specimens were struck, the y-axis in this plot gives energy changes rather than
absolute energy values. Moreover, the energy absorbed during impact depends
on the specimen size, as a large specimen will unavoidably absorb more energy
upon failure than a small specimen. Therefore, the presentation of all impact
test data in a single plot requires the use of two different energy scales: A 50-J
scale for the standard-size CVN specimens and a 5-J scale for the small CVN
specimens. Numerical values of the change in the impact energy during solder
embrittlement are given in Table 1 for all solders.
Figure 1 suggests that all tested Sn-based solders undergo a ductile-to-
TABLE 1—Max impact energy change, ⌬Emax, during solder embrittlement.
⌬Emax a, J
Specimen Dimensions, mm3
Solder Alloy 10⫻ 10⫻ 55 5 ⫻ 5 ⫻ 55

SAC 305 40.31
SAC 405 33.89
Sn-5Ag 46.55
Sn-37Pb 11.90
99.99 % Sn 3.10
Sn-0.7Cu 4.42
Sn-0.7Cu-0.1Ni 3.44
a
⌬Emax = ⌬Eductile共max impact energy in ductile regime兲 − ⌬Ebrittle 共minimum impact
energy in brittle regime兲. ⌬Emax gives an indication of the impact energy change during
solder embrittlement.
brittle transition in their fracture behavior since their impact resistance de-
creases substantially as the test temperature decreases. Figure 1 also shows
that the onset of the ductile-to-brittle transition for the smaller-size CVN speci-
mens occurs ⬃40° C below the completion of this transition for the standard-
size CVN specimens. Moreover, one may observe that the ductile-to-brittle tran-
sition starts at progressively higher temperatures as the silver 共Ag兲 content of
the solder increases. In fact, the plot of Fig. 1 suggests that the onset of the
transition for the SAC 305 solder is at around −75° C, while that for the Sn-5Ag
solder is at around −35° C.
Another interesting observation is that the embrittlement of the solders
containing a small amount of alloying elements 共i.e., Sn-0.7Cu, Sn-0.7Cu-0.1Ni,
and 99.99 % Sn兲 occurs within a narrow temperature range that is identical for
all three solders, i.e., between −110 and −130° C. On the other hand, the more
heavily alloyed solders 共i.e., SAC 305, SAC 405, and Sn-5Ag兲 become brittle over
a broader temperature range, the width and exact position of which on the
temperature scale depend on the solder composition.
Both Fig. 1 and Table 1 show that over the same range of test temperatures
and for specimens of identical dimensions, the impact resistance of the Pb-
containing solder of reference 共i.e., Sn-37Pb兲 decreases appreciably less than
the impact resistance of all Pb-free solders. For standard-size CVN specimens,
cooling down to about −80° C decreases the impact energy of Sn-37Pb by 71.8
%, while the impact energy of SAC 305, SAC 405, and Sn-5Ag decreases by 89.4
%, 92.2 %, and 94.5%, respectively. One might point out, however, that the
room temperature impact resistance of Sn-37Pb is poorer than that of the Ag-
containing Pb-free solders considered in this study. Still, the fracture behavior
of the Sn-37Pb solder appears less influenced by temperature changes than
these Pb-free candidate replacements. Nevertheless, it is precisely this tempera-
ture range that was identified as a “caveat” transition temperature range for the
accelerated temperature cycling of Sn-37Pb solder joints 关29兴. On the other
hand, it is interesting to note that the ductile-to-brittle transition occurs at
(a) (b)
15 min
o
125 C
100 μm 20 μm
Temperature
100 μm
Time
15 min
-55 C
20 μm
(c) (d)
1 hour
FIG. 2—SAC 405 specimens tested in impact at room temperature 关共a兲 and 共b兲兴 and at
−40°C 关共c兲 and 共d兲兴. 共a兲 Ductile failure: The fracture surface is characterized by the
presence of crater-like cavities known as “dimples.” 共b兲 Closer inspection of the fracture
surface and inset idealized drawing of the dimple-formation mechanism. 共c兲 Mixed
ductile-brittle failure: Parts of the fracture surface are dimple-rich 共arrows兲, and other
parts are flat. 共d兲 The flat parts of the fracture surface are related to the presence of IMC
particles 共crosses兲.
temperatures that are relatively close to the onset of significant creep for the
Sn-37Pb solder but not for the other solders. The onset of significant creep is
assumed to occur at a homologous temperature of TH = 0.5, which is around
−45° C for Sn-37Pb and around −28° C for SAC solders.
Fractography Study Results
Solder 405—Since the fractography analysis performed on solder SAC 405

gave very similar results with the analysis on solder SAC 305, results obtained
from the latter will be omitted for the sake of brevity. Suffice it to mention that
the similarity in the fracture behavior of these two solders on a micrometer
level can be attributed to comparable microstructural features.
The most important results of the fractography study on solder SAC 405
are shown in Figs. 2–4. These results suffice to understand the temperature
dependence of the fracture behavior of this solder and of the microstructurally
similar solder SAC 305. Ductile failures are characterized by the presence of
numerous crater-like cavities on the fracture surface. These cavities are com-
monly referred to as “dimples,” while images depicting them are shown in Fig.
2共a兲 and 2共b兲. Mixed ductile-brittle failures are characterized by the alternation
of dimple-rich areas 共ductile failure兲 and flat areas 共brittle failure兲 on the frac-
50 μm (a) (b)
+
+ + 20 μm
(d) 20 μm
+
+
+
+
+
+
40 μm
(c)
FIG. 3—SAC 405 specimens tested in impact at −50°C 关共a兲 and 共b兲兴 and at −60°C 关共c兲
and 共d兲兴. 共a兲 Mixed failure: A substantial part of the fracture surface consists of “pla-
teaus” related to the presence of IMCs in the solder 共crosses兲. 共b兲 The river patterns
formed during the fracture of the Sn matrix enveloping this IMC particle 共cross兲 are
typical for catastrophic types of failure. Further away from the IMC particle, one may
discern dimple-rich areas indicative of ductile localized failure. 共c兲 Mixed failure show-
ing signs of advanced embrittlement: The accidental agglomeration of IMC particles
created a very large “plateau” on the fracture surface. 共d兲 The high stress intensity in the
immediate vicinity of a large IMC particle is responsible for the fast fracture of the Sn
matrix around it 共arrows兲.
ture surface, as shown in Figs. 2共c兲, 2共d兲, and 3. Brittle failures are recognized
by the stepwise, cleavage-like appearance of the fracture surface and the con-
current absence of dimple-rich areas 共Fig. 4兲.
It would be an oversimplification of the solder fracture behavior to con-
sider that all the fracture surface parts indicating brittle failure are formed in
the same way. In fact, careful inspection of the fractured specimens reveals that
there are different types of brittle failure, each one of which reflects another
aspect of the overall solder embrittlement mechanism. The importance of each
embrittlement aspect depends on the solder service conditions and is solder-
specific; for a specific solder composition, fixed specimen dimensions, and ap-
plied strain rate, the contribution of each embrittlement aspect is primarily
defined by the ambient temperature. The above ideas become clearer by con-
sidering that at intermediate test temperatures, the flat areas on the fracture
surface are basically related to the presence of IMC particles in the solder 共Figs.
2共c兲, 2共d兲, 3共a兲, and 3共b兲兲. As the temperature decreases, however, the fracture
surface parts indicating brittle failure result more and more from the embrittle-
(a) 100 μm (b)

+ + +
+
+
+
+ 20 μm
(c) (d)
50 μm 10 μm
FIG. 4—SAC 405 specimens tested in impact at −70°C 关共a兲 and 共b兲兴 and at −75°C 关共c兲
and 共d兲兴. 共a兲 Brittle failure: The combination of Sn embrittlement with the effect of IMCs
共arrows兲 produces a very flat fracture surface. 共b兲 The failure of the solder is brittle
everywhere, not only in the vicinity of the IMC particle. 共c兲 Brittle failure: Parts of the
fracture surface show unusual patterns of parallel lines. 共d兲 Closer inspection of these
“patterns” reveals that they result from the embrittlement of solder areas consisting of
the ternary eutectic Sn– Ag3Sn– Cu6Sn5.
ment of the ␤-Sn solder matrix. One of the last solder constituents to become
brittle is the ternary Sn– Ag3Sn– Cu6Sn5 eutectic structure 共Fig. 4共c兲 and 4共d兲兲.
Solder Sn-5Ag—The impact testing of solder Sn-5Ag did not result in frac-
ture between room temperature and −31° C 共Fig. 5兲. This is attributed to the
satisfactory solder ductility at room temperature and moderate sub-zero tem-
peratures. The study of the test specimens that were not fractured revealed the
presence of “stretch marks” indicative of plastic deformation 共Fig. 5共a兲–5共c兲兲.
Careful scrutiny of these marks revealed the disruption of solder continuity at
several places along their length; it also showed the formation of small heaps of
solder between the places where the solder continuity was lost 共Fig. 5共c兲兲. The
small heaps of solder were associated with the presence of sub-surface IMC
particles, as shown in Fig. 5共d兲 and 5共e兲.
Once started, the fracture of this solder becomes rapidly brittle as the test
temperature decreases, in agreement with Fig. 1. The mixed ductile-brittle fail-
ure of the solder at −42° C 共Fig. 6共a兲–6共c兲兲 turned predominantly brittle at
−55° C 共Fig. 6共d兲–6共f兲兲 and completely brittle at −65° C 共Fig. 7兲. An interesting
observation of the fractography study on solder Sn-5Ag is the existence of bi-
nary Sn– Ag3Sn eutectic “colonies” on the fracture surface 共Figs. 6共d兲, 6共e兲, 7共c兲,
and 7共d兲兲. Since these colonies appear often intact, one may safely assume that
(a) 200 μm (b) 50 μm
(c) (d) 20 μm
20 μm 50 μm
(e)
FIG. 5—Sn-5Ag specimens tested in impact at room temperature 关共a兲 and 共b兲兴 and at
−31°C 关共c兲–共e兲兴. 共a兲 No fracture: The specimen exhibits parallel lines that are thought to
be stretch marks resulting from the plastic deformation of the solder. 共b兲 Closer exami-
nation reveals the disruption of material continuity at many locations along these
“striae.” 共c兲 No fracture: The plastic deformation of the solder results again in the for-
mation of “striae.” Careful scrutiny of these lines reveals the alternation of sites where
material continuity is lost with sites of solder pile-up 共stars兲. 共d兲 The locations of solder
pile-up are caused by the fact that hard IMC particles, like the ones indicated by crosses
in this image, push the soft solder as they try to follow the overall specimen deforma-
tion. The IMC motion seems to be responsible for the small solder heaps appearing on
the specimen surface. 共e兲 Closer examination of one of the IMCs in 共d兲 and of the
plastically deformed solder in its vicinity.
the crack responsible for solder failure prefers to separate them from the solder
matrix than propagate through them. Another important observation is that the
large amount of Ag3Sn platelets in this solder affects its fracture behavior to a
great extent, as shown in Figs. 7共a兲, 7共b兲, and 8. As the temperature decreases,
these IMCs act as critical-size flaws that are often responsible for the cata-
strophic 共i.e., brittle兲 failure of the other two solder constituents, i.e., the Sn
grains and the Sn– Ag3Sn eutectic grains 共Fig. 8兲.
Solder Sn-0.7Cu—Solder Sn-0.7Cu exhibited a highly satisfactory fracture

resistance between room temperature and −125° C 共Fig. 9兲 since no fracture
was observed in this temperature interval. Once started, however, fracture be-
came brittle within a very narrow temperature range: At −135° C, the solder
behaved essentially as a brittle material 共Fig. 10兲. As will be demonstrated later
on, the same holds for the other two lightly alloyed Sn-based solders studied
here, i.e., the Sn-0.7Cu-0.1Ni and the 99.99 % Sn.
The fracture surface of Sn-0.7Cu specimens that failed in a brittle manner
(a) (b) (c) 20 μm
+ +
+
200 μm 100 μm
(d) (e) (f)
20 μm 10 μm 50 μm
FIG. 6—Sn-5Ag specimens tested in impact at −42°C 关共a兲–共c兲兴 and at −55°C 关共d兲–共f兲兴. 共a兲
Mixed failure: Parts of the fracture surface are flat and parts are dimple-rich. The flat
areas are mostly related to the presence of IMCs 共crosses兲. 共b兲 The fracture of the solder
in the immediate vicinity of the IMC particle 共star兲 is different than further away. 共c兲
Closer view of the IMC particle of 共b兲: Fast 共brittle兲 fracture of the solder next to the IMC
particle, as opposed to the ductile 共dimple-rich兲 fracture of the solder away from the
IMC. 共d兲 Rather brittle failure: The fracture surface shows colonies made of the binary
Sn– Ag3Sn eutectic structure 共star兲 surrounded by embrittled Sn. 共e兲 Closer examination
of a Sn– Ag3Sn eutectic colony. 共f兲 Typical area of embrittled Sn; in this case, surround-
ing an IMC particle 共star兲.
showed the co-existence of two types of fracture 共Figs. 10共c兲, 10共d兲, 11共b兲, and
11共e兲兲: A stepwise, cleavage-like 共transgranular兲 fracture and an intergranular
fracture. The cleavage-like type of fracture is associated with the fracture of Sn
共Figs. 10共a兲, 10共b兲, 11共a兲, and 11共c兲兲, while the intergranular type of failure re-
lates more to the separation of Sn– Cu6Sn5 eutectic grains from the Sn solder
matrix 共Fig. 11共b兲, 11共e兲, and 11共f兲兲.
Solder Sn-0.7Cu-0.1Ni—Similar to solder Sn-0.7Cu, no fracture resulted

from the impact testing of specimens made of the Sn-0.7Cu-0.1Ni solder be-
tween room temperature and −125° C. Specimens that were tested in this tem-
perature range exhibited severe plastic deformation. The study of the specimen
surface revealed numerous sites of solder pile-up, the formation of which was
attributed to the displacement of sub-surface IMC particles during impact test-
ing 共Fig. 12兲.
Brittle failure started at −135° C and continued down to −190° C, showing
the same basic trends. The careful investigation of the fractured specimens
revealed some features that are unique for this solder and can so far only be
explained by the addition of Ni to the solder. First, the fracture surface was
highly inhomogeneous: In fact, it was characterized by the alternation of two
(a) (b)
+
+
+
+
100 μm + 10 μm
50 μm 10 μm (d)
(c)
FIG. 7—Sn-5Ag specimen tested in impact at −65°C: Brittle failure. 共a兲 The local ag-
glomeration of IMC particles 共crosses兲 facilitates the solder embrittlement at such low
temperatures. 共b兲 The cleavage-like failure of the Sn matrix results in the formation of
sharp ledges on the fracture surface. The catastrophic failure of Sn appears to have
initiated from the edges of an Ag3Sn IMC particle 共cross兲. 共c兲 Some parts of the fracture
surface reveal colonies of the binary Sn– Ag3Sn eutectic structure. The fact that the
outline of these colonies is clearly visible indicates an easy solder separation at the
Sn/eutectic interface. 共d兲 Closer view of one of the eutectic colonies in 共c兲.
areas with remarkably different appearances. One showed clear signs of

cleavage-like failure of the Sn matrix 共Figs. 13共c兲, 14共a兲, 14共b兲, and 15共d兲兲, while
in the other, the Sn matrix remained smooth down to the extreme temperature
of −190° C 共Figs. 13共a兲 and 14共c兲兲. The smooth parts of the fracture surface
revealed the exaggerated growth of IMC particles 共Figs. 13共b兲, 15共a兲, and 15共b兲兲,
which was not observed for any of the other Sn-based solders. Moreover, frac-
ture induced the clean separation of the IMC particles from the Sn solder ma-
trix 共Figs. 13共d兲, 14共d兲–14共f兲, and 15共c兲兲, irrespective of whether the IMCs were
in the smooth or rough parts of the fracture surface. Another interesting obser-
vation is that the failure-inducing crack travelled in the solder by going from
one IMC particle to the next 共Fig. 15共e兲兲, indicating that crack propagation
along the Sn/IMC interface is energetically more favorable than IMC fracture.
The assumption of a rather weak Sn/IMC interface is corroborated by the fact
that the broken IMC particles were often sticking out of the fracture surface by
several micrometers 共Fig. 15共f兲兲. The clean separation of the IMC particles from
the Sn matrix that envelops them allows a better perception of the three-
dimensional growth morphology of the IMCs in this solder 共Figs. 13共b兲, 14共e兲,
14共f兲, 15共a兲, and 15共b兲兲.
Solder 99.99 % Sn—Similar to the two previous solders, the impact testing
of specimens made of the 99.99 % Sn solder did not result into fracture be-
(a) 200 μm 50 μm
+ (b)
100 μm 50 μm
(c) (d)
FIG. 8—Sn-5Ag specimen tested in impact at −75°C: Brittle failure. 共a兲 Catastrophic
failure of a Sn grain initiated by an Ag3Sn IMC platelet 共arrow兲 that was present in the
grain center. The grains of Sn in Sn-based solders tend to grow assuming a dendritic
habit, a fact that may clearly be visualized here, as the fracture of this Sn grain stopped
at its boundaries with the adjacent grains. 共b兲 Catastrophic failure of a Sn– Ag3Sn
eutectic colony; this failure is probably related to the presence of an IMC platelet 共cross兲
close to the nucleation site of the colony. The colony grows by fanning out from its
nucleation site due to the continuous branching of one or both of the phases in the
eutectic. 共c兲 The fast fracture of a eutectic colony seems to succeed the brittle failure of
a neighboring Ag3Sn platelet 共cross兲. 共d兲 Closer inspection of the eutectic colony sug-
gests that probably the cause of its failure was also the cause of its formation: The
Ag3Sn IMC has probably facilitated the nucleation of the eutectic colony in 共c兲, as
indicated by the direction of the eutectic lamellae in the fractured colony.
200 μm
500 μm (a) (b)
FIG. 9—Sn-0.7Cu specimen tested in impact at −125°C: No fracture. 共a兲 The specimen
has essentially only deformed plastically during testing, as indicated by the formation of
stretch marks on the specimen surface. 共b兲 Closer inspection of the specimen surface
shows the occasional disruption of material continuity 共arrow兲.
(a) (b) 10 μm
20 μm
(d)
(c) 100 μm 20 μm
FIG. 10—Sn-0.7Cu specimen tested in impact at −135°C: Primarily brittle fracture.

Careful investigation revealed two different areas on the fracture surface: Areas charac-
terized by the cleavage-like fracture of Sn 关共a兲 and 共b兲兴 and areas characterized by inter-
granular separation 关共c兲 and 共d兲兴. 共a兲 Part of the fracture surface with many ledges
共arrows兲 formed during the cleavage-like failure of Sn. 共b兲 Closer inspection of one of the
ledges in 共a兲; the sharp ridges on the ledge are associated with the separation of ␤-Sn
along specific atomic planes. 共c兲 Part of the fracture surface revealing areas of inter-
granular fracture 共crosses兲 separated by areas of transgranular fracture 共stars兲. 共d兲
Closer inspection of the border between intergranular 共cross兲 and transgranular 共star兲
fractures. Intergranular fracture is usually associated with the binary Sn– Cu6Sn5 eu-
tectic structure in the solder 共cross兲.
tween room temperature and −108° C. The severe plastic deformation of the
specimens that were tested in this temperature range was expressed in the
formation of stretch marks on the specimen surface 共Fig. 16兲.
Brittle failure started at −130° C and continued to −190° C, showing two
basic types of fracture: A cleavage-like fracture 共Figs. 17共c兲, 17共d兲, and 18兲 and
an intergranular fracture 共Fig. 17共e兲 and 17共f兲兲. Quite interestingly, twin crystals
of ␤-Sn were observed at several places on the fracture surface 共Fig. 18兲.
Solder Sn-37Pb—The fracture behavior of the Sn-37Pb solder showed a

propensity towards intergranular failure, irrespective of the test temperature
共Figs. 19共a兲, 19共d兲, 20共c兲, and 20共d兲兲. Moreover, differences were observed in the
fracture behavior of the two components of the binary Sn–Pb eutectic grains:
First, the Pb-rich phase exhibited greater ductility than the Sn-rich phase at all
temperatures 共Figs. 19共b兲, 19共e兲, 20共a兲, and 20共e兲兲. Second, the ductility of the
Sn-rich phase decreased as the temperature decreased, while the ductility of
the Pb-rich phase remained practically unaltered over the considered tempera-
ture range.
(a) 20 μm
10 μm
(b)
(d)
20 μm 5 μm
(c)
50 μm 5 μm
(e) (f)
FIG. 11—Sn-0.7Cu specimens tested in impact at −155°C 关共a兲 and 共b兲兴 and at −170°C
关共c兲–共f兲兴. 共a兲 Brittle failure: Parts of the fracture surface reveal the cleavage-like fracture
of Sn. 共b兲 Apart from the cleavage-like failure of Sn 共cross兲, areas of intergranular
separation of Sn– Cu6Sn5 eutectic grains 共star兲 occupy a large part of the fracture sur-
face. The clean separation of the two areas indicates a poor strength for the Sn/eutectic
interface at these service conditions. 共c兲 Brittle failure: The extensive cleavage-like fail-
ure of Sn indicates severe embrittlement. 共d兲 Close inspection of the border between an
area of cleaved Sn 共cross兲 and a Sn– Cu6Sn5 eutectic grain 共star兲. The fibrilar phase in
the eutectic consists of ultra-fine rods of Cu6Sn5 IMC. 共e兲 Part of the fracture surface
with obvious intergranular fracture 共star兲 occurring between grains of the Sn– Cu6Sn5
eutectic. 共f兲 Closer inspection of the interface between two adjacent Sn– Cu6Sn5 eutectic
grains. Failure followed the interface between the two eutectic grains, while the zigzag-
ging of the failure-inducing crack suggests a rather strong grain boundary.
Discussion
Sn-based solders consist of ␤-Sn dendrites interspersed with IMC particles that
appear either as single-phase precipitates or as constituents of binary or ter-
nary eutectic structures. The IMCs present in the solders studied here are:
Cu6Sn5 in Sn-0.7Cu, Ag3Sn in Sn-5Ag, Cu6Sn5 and Ag3Sn in Sn–Ag–Cu 共SAC兲
solders, and Cu6Sn5 in Sn-0.7Cu-0.1Ni 共Fig. 21兲. The eutectic structures in
(a) 20 μm (b) 5 μm (c)
*
*
100 μm
FIG. 12—Sn-0.7Cu-0.1Ni specimen tested in impact at −125°C: No fracture. 共a兲 The

overview of the fracture surface shows numerous sites of solder pile-up. These small
heaps of solder are formed when the hard IMC particles push the soft solder matrix in
an effort to follow the overall plastic deformation of the specimen during impact testing.
共b兲 Closer inspection of the specimen surface reveals the presence of IMC particles
共stars兲 under a solder heap. This heap of solder was formed as near-the-surface IMC
particles pushed the solder lying ahead of them in the direction of motion. The wake of
IMC particles is observed at the basis of the solder heap, while the direction of IMC
motion is represented by arrows. 共c兲 Closer view of the IMC particles responsible for the
creation of the solder heap shown in 共b兲.
these solders are the Sn– Cu6Sn5 binary eutectic in Sn-0.7Cu, the Sn– Ag3Sn
binary eutectic in Sn-5Ag, the Sn– Cu6Sn5 – Ag3Sn ternary eutectic in SACs,
and 共possibly兲 the Sn– Cu6Sn5 binary eutectic in Sn-0.7Cu-0.1Ni 共Fig. 21兲. Since
␤-Sn is the primary constituent of all Sn-based solders, it is obvious that the
properties of these solders will be governed by the properties of ␤-Sn, which
crystallizes in the body-centered tetragonal 共bct兲 crystal system. It might be
useful to mention that ␤-Sn is thermodynamically stable above 13.2° C; below
that temperature, ␤-Sn transforms to ␣-Sn, which has the diamond cubic
structure 关30兴. This transformation, also known as “tin pest,” is highly undesir-
able because ␣-Sn is more brittle than ␤-Sn; moreover, this transformation is
accompanied by a volume increase of ⬃26 % that leads to the complete mate-
rial disintegration 关30兴. Fortunately, however, the formation of tin pest is very
sluggish and is characterized by a long incubation period; for example, 18
months of ageing at −18° C was reported necessary to transform 40 % of the
specimen surface of Sn-0.5Cu ingots to ␣-Sn 关30兴. Based on the above, the
crystalline form of reference for Sn-based solders used in conventional elec-
tronic applications is ␤-Sn.
The ductile-to-brittle transition in the fracture behavior of ␤-Sn is associ-
ated with its bct crystal structure because the ductility of any material is inex-
tricably linked with its crystal structure 关20,31,32兴. The ductile-to-brittle tran-
sition experienced by body-centered cubic 共bcc兲 and bct metals at low
temperatures is caused by certain phenomena that are not manifested in face-
centered cubic 共fcc兲 metals but will not be discussed here as they have been
extensively addressed elsewhere 关20兴. It is sufficient to remember that the
ductile-to-brittle transition of bcc/bct metals is accompanied by an increase in
their yield strength, i.e., the stress at which plastic deformation starts. If the
yield strength becomes very high, it is probable that the material breaks before
20 μm 2 μm
(a) (b)
(d)
5 μm 2 μm
(c)
FIG. 13—Sn-0.7Cu-0.1Ni specimen tested in impact at −135°C: Brittle fracture. 共a兲 The
solder fracture surface is inhomogeneous, showing areas with distinctly different frac-
ture behaviors. This part of the fracture surface is characterized by the smooth appear-
ance of the solder matrix around the various IMC particles. 共b兲 Closer inspection of the
fracture surface reveals the exaggerated dendritic growth of IMC particles, which is not
observed in other Sn-based solders. EDS analysis of this particle showed that it is a
Cu6Sn5 IMC, as could also be guessed by the hexagonal symmetry of its hollow primary
branch. The clean separation of the IMCs from the matrix suggests a poor matrix/IMC
interfacial strength. 共c兲 This part of the fracture surface shows signs of Sn embrittle-
ment 共cross兲. 共d兲 Closer inspection shows the clean separation of the long IMC particle
from the embrittled solder matrix.
it yields. This happens when the level of tensile stresses exceeds the material
cohesive strength, resulting into a cleavage-like 共brittle兲 failure along specific
atomic planes in the material 关28,20,31兴. The temperature where embrittlement
occurs is known as the ductile-to-brittle transition temperature 共DBTT兲.
The ductile-to-brittle transition in the fracture behavior of bcc/bct metals is
affected by the applied strain rate, the mechanical constraint, and the distribu-
tion of brittle second-phase particles in the material 关20兴. First, the transition
shifts to higher temperatures as the loading rate changes from slow to dynamic
共impact兲关20,33兴. Second, the transition shifts to higher temperatures as the
mechanical constraint of the material increases from plane-stress conditions
共minimum constraint兲 to plane-strain conditions 共maximum constraint兲
关20,31,33兴. A fact that, if neglected, might lead to erroneous conclusions with
respect to the DBTT of a specific material is that the mechanical constraint
depends on the specimen dimensions: Plane-stress conditions are more likely
to occur in very thin specimens, while plane-strain conditions are easier to
establish in really thick specimens 共the terms “thin” and “thick” being relative
and material-specific兲关20,31,33兴. Third, the distribution of brittle second-phase
5 μm 2 μm
(a) (b)
20 μm (d)
(c) 5 μm
5 μm (f)
2 μm
(e)
FIG. 14—Sn-0.7Cu-0.1Ni specimen tested in impact at −145°C: Brittle fracture. 共a兲

Parts of the fracture surface give strong evidence of the cleavage-like failure of Sn. 共b兲
Closer view of the cleavage-like failure of Sn. 共c兲 IMC particles are observed sticking out
of smooth parts of the fracture surface. This type of failure is similar to the failure of
certain types of fiber-reinforced composites. 共d兲 Closer inspection of the area in 共c兲
shows the clean separation of the solder at the IMC/matrix interface. The appearance of
the fracture surface suggests that the interface between IMCs and solder matrix is quite
weak, thus allowing the deflection of cracks. 共e兲 Another proof of the weak matrix/IMC
interface: The morphology of fragmented IMCs can be accurately deduced from the IMC
imprints on the solder matrix. 共f兲 The fracture behavior of the solder in the smooth
fracture surface parts offers unique information on the growth of IMCs, like the three-
dimensional star-like IMC depicted here.
particles, such as the IMCs in the Sn-based solders under consideration here,
affects greatly the process of solder embrittlement on a microscopic level. Apart
from the volume fraction of the second-phase particles, important aspects of
this distribution are the geometrical shape, acuity, size, and spatial distribution
of these particles. The IMCs in Sn-based solders promote the ductile-to-brittle
transition of the ␤-Sn solder “matrix” in two ways: First, by dislocation “pin-
ning” and, second, by creating high stress intensity in their immediate vicinity.
Dislocation pinning limits the solder capability for plastic deformation 关20,31兴
(a) (b)
5 μm 5 μm
5 μm 10 μm
(c) (d)
(e)
2 μm 2 μm
(f)
FIG. 15—Sn-0.7Cu-0.1Ni specimens tested in impact at −180°C 关共a兲–共d兲兴 and at −190°C

关共e兲 and 共f兲兴. 共a兲 Brittle fracture: Smooth part of the fracture surface, showing the flower-
like dendritic habit of an IMC crystal. A cross indicates the nucleation site for this IMC.
共b兲 Another flower-like IMC dendrite 共nucleation site indicated by a star兲, the size 共di-
ameter兲 of which is estimated to exceed 50 ␮m. 共c兲 Clean separation of IMC particles
from the solder matrix in a smooth fracture surface part. 共d兲 Clean separation of the
IMC particles from the solder matrix in areas of severe matrix embrittlement. 共e兲 Brittle
fracture: Smooth part of the fracture surface, revealing crack deflection from the solder
matrix to the solder/IMC interface. The crack appears to travel from one IMC particle to
the next presumably because crack propagation along the solder/IMC interface is ener-
getically more favorable than IMC fracture. 共f兲 The fact that the IMC particles stick out
of the solder fracture surface by several micrometres is another proof of the poor solder/
IMC interfacial strength.
and may lead to dislocation coalescence and crack nucleation 关20兴. It must be
mentioned, however, that dislocation pinning is achieved by distributions of
IMCs with the “correct” size and spacing 关20,31兴. Changes in the IMC size and
spacing due to coarsening, for example, will not ensure dislocation pinning, a
fact that will be reflected in a decrease of the DBTT, as has already been re-
ported for the SAC 405 solder 关34兴. The high stress intensity next to faceted
IMCs makes them behave as “internal notches” in the solder 关35兴, and the tri-
50 μm 20 μm
(a) (b)
100 μm 50 μm
(c) (d)
FIG. 16—99.99 % Sn specimens tested in impact at room temperature 关共a兲 and 共b兲兴 and
at −108°C 关共c兲 and 共d兲兴. 共a兲 No fracture. Stretch marks formed during to the plastic
deformation of the solder are visible everywhere on the specimen surface. 共b兲 Closer
investigation of the specimen surface reveals that the stretch marks are parallel within
each grain, changing orientation as they go from one grain to the next. The grain
boundaries are visible. 共c兲 No fracture: Stretch marks indicative of plastic deformation
may be discerned on the specimen surface. 共d兲 Closer inspection of the specimen surface
shows a finer pattern of “striae,” the orientation of which changes from one grain to the
next.
axial stress state in the IMC vicinity facilitates locally the solder embrittlement
关20兴.
Practical reliability consequences of the ductile-to-brittle transition are
shown in Fig. 22, which is a Manson–Coffin plot of thermal cycling data from
eutectic Sn–Pb solder joints subjected to one of the following three temperature
cycles: 共−50° C ↔ +100° C兲, 共−50° C ↔ +25° C兲, and 共+25° C ↔ +100° C兲关19兴. All
thermal cycling tests were performed at a heating or “strain rate” of
3.75° C / min with 20-min dwells at the temperature extremes. The plot of Fig.
22 shows that at a given level of shear strain, the damage accumulated during
the 共−50° C ↔ +100° C兲 thermal cycle is more severe than the damage accumu-
lated during either of the other two thermal cycles, and this holds for the larg-
est part of the applied shear strain range 共i.e., between 0.1 % and 30 %兲. It is
believed that the significantly earlier failures observed for the 共−50° C ↔
+100° C兲 cyclic loading are caused by the synergy of low-temperature em-
brittlement with high-temperature creep-fatigue. First, one must consider that
in accelerated thermal cycling, cyclic loading to lower temperatures results in a
more incomplete creep process—and hence a smaller creep-fatigue damage—
than cyclic loading to higher temperatures. This is supported by the fact that
the Sn–Pb solder fatigue life is longer for the 共−50° C ↔ +25° C兲 thermal cycle
500 μm (a) 10 μm
(b)
(c) (d)
10 μm 5 μm
100 μm (e) 20 μm (f)
FIG. 17—99.99 % Sn specimens tested in impact at −130°C 关共a兲 and 共b兲兴 and at −147°C
关共c兲–共f兲兴. 共a兲 Brittle failure: The fracture surface is a mixture of intergranular failure
共stars兲 and cleavage-like failure 共crosses兲. 共b兲 Part of the fracture surface where the
共transgranular兲 fracture shows tendency towards cleavage-like failure. 共c兲 Brittle failure:
Cleavage-like failure of Sn. 共d兲 Closer inspection of cleaved Sn. 共e兲 Triple point 共arrow兲
between three Sn grains. 共f兲 Closer inspection reveals the onset of intergranular failure
at the triple point of 共e兲.
than for the 共+25° C ↔ +100° C兲 one. Furthermore, cracks nucleating during the
high-temperature creep-fatigue stage of solder damage will propagate more
easily in the solder toward the low-temperature end of the 共−50° C ↔ +100° C兲
thermal cycle due to the notch sensitivity of the partially embrittled Sn-rich
phase in the eutectic Sn–Pb solder.
The consequences of the ductile-to-brittle transition are most keenly felt in
accelerated reliability testing. Prior studies pointed out that some of the popu-
lar, highly accelerated thermal cyclic tests produce results that may be con-
founded and leading to conservative extrapolations of the product reliability
关29兴. Highly accelerated thermal cycling test conditions reported to produce
confounded and conservative results are the following: TC3 共−40° C ↔
+125° C兲, TC4 共−55° C ↔ +125° C兲, and TC5 共−55° C ↔ +100° C兲关19兴.
10 μm
20 μm
(a) (b)
FIG. 18—99.99 % Sn specimens tested in impact at −190°C: Brittle fracture. 共a兲 The
fracture is predominantly cleavage-like 共arrows兲. 共b兲 Fracture reveals the formation of
twin crystals of ␤-Sn in the solder.
(a) 200 μm (b) (c)
10 μm 10 μm
(d) (e) (f)
200 μm 10 μm 10 μm
FIG. 19—Sn-37Pb specimens tested in impact at room temperature 关共a兲–共c兲兴 and at

−40°C 关共d兲–共f兲兴. 共a兲 Overview of the fracture surface: The fracture appears predominantly
intergranular. 共b兲 Secondary electron 共SE兲 detector image of the Sn–Pb binary eutectic.
共c兲 Backscattered electron 共BSE兲 detector image of the Sn–Pb eutectic: This image forms
a pair with 共b兲, as they are both taken from the same place on the fracture surface. The
bright areas are the Pb-rich phase in the binary Sn–Pb eutectic, while the gray areas are
the Sn-rich phase. 共d兲 Overview of the fracture surface. 共e兲 SE detector image of the
Sn–Pb eutectic structure. The Pb-rich areas in the eutectic exhibit greater plastic defor-
mation than the Sn-rich areas and are clearly responsible for the relief of the fracture
surface while the Sn-rich areas are flatter. The observation of the Pb-rich areas in the
eutectic suggests that the stress level in them has reached the tensile strength of the
material during testing, as plastic deformation resulted into necking 共indicated by the
ridges on top of these areas兲. 共f兲 BSE detector image of the fracture surface part depicted
in 共e兲.
(a) 10 μm (b) 10 μm
(c) 50 μm (d)
200 μm
(e) 20 μm (f) 20 μm
FIG. 20—Sn-37Pb specimens tested in impact at −60°C 关共a兲 and 共b兲兴 and at −75°C
关共c兲–共f兲兴. 共a兲 SE detector image of the Sn–Pb eutectic structure. The Pb-rich areas in the
eutectic have undergone plastic deformation, as suggested by their enhanced topogra-
phy, while the Sn-rich areas appear flat. 共b兲 The BSE counterpart of 共a兲: The bright areas
are Pb-rich, while the gray areas are Sn-rich. 共c兲 Overview of the fracture surface: Frac-
ture still appears predominantly intergranular. 共d兲 Closer inspection of the fracture sur-
face shows that solder failure prefers to occur at the boundaries between Sn–Pb eutectic
grains 共arrows兲. 共e兲 SE detector image of the Sn–Pb eutectic: The surface relief of the
Pb-rich areas is indicative of ductile behavior, while the flatness of the Sn-rich areas is
an expression of brittle behavior. 共f兲 The BSE equivalent of 共e兲.
Impact Test Results

As already mentioned, the onset of the ductile-to-brittle transition for the
smaller-size CVN specimens occurs at around −120° C, while the same transi-
tion for the standard-size CVN specimens is completed ⬃40° C higher, i.e., at
around −80° C 共Fig. 1兲. Two reasons could account for that temperature differ-
ence: 共a兲 The smaller amount of IMC particles in the solders used to produce
the smaller specimens and 共b兲 the possible difference in the solder stress state
due to the variation in specimen dimensions. The solders used to make the
smaller specimens were lightly alloyed and, therefore, contained a smaller
FIG. 21—共a兲 The Cu–Sn equilibrium phase diagram 关37兴, where the Sn-0.7Cu solder
composition is represented by a star. 共b兲 The Ag–Sn equilibrium phase diagram 关38兴,
where the Sn-5Ag solder composition is indicated with a star. 共c兲 The Sn-rich corner of
the SAC equilibrium phase diagram 关39兴, where the light gray star represents the SAC
305 solder composition and the dark gray star represents the SAC 405 solder composi-
tion. 共d兲 The 220° C isothermal section of the Sn–Cu–Ni equilibrium phase diagram,
axes in at. % 关40兴. A star indicates the Sn-0.7Cu-0.1Ni solder composition.
amount of IMCs than the Ag-containing solders used to make the standard-size
specimens. Since IMCs act as internal notches in a notch-sensitive ␤-Sn matrix,
an increase in the amount of IMC particles in the solder is expected to increase
its DBTT 关20,35兴. Moreover, the stress state in the smaller specimens might be
closer to plane-stress conditions than the stress state in the standard-size speci-
mens. The transition from plane-stress to plane-stain conditions is expected to
shift the DBTT to higher values, which would agree with the impact data.
Another interesting observation is that the DBTT increases as the Ag con-
tent in the solder increases from 3 wt % 共SAC 305兲 to 5 wt % 共Sn-5Ag兲. This is
associated with the amount of Ag3Sn IMCs in the solder, which increases with
FIG. 22—Manson–Coffin plot of thermal cycling data from eutectic Sn–Pb solder joints
关19兴. Thermal cycling was carried out at a ‘‘strain rate’’ of 3.75°C/min with 20-min
dwells at the temperature extremes. For the most important part of the applied shear
strain range below 30 %, the damage accumulated during the 共−50°C↔+100°C兲 thermal
cycle is more severe than the damage accumulated during either of the other two ther-
mal cycles at a given shear strain level. This results from the synergy of low-temperature
embrittlement with hight-temperature creep-fatigue, whereupon cracks nucleating dur-
ing the high-temperature creep-fatigue stage of damage propagate more easily in the
solder during its exposure to low tempertures due to the partial embrittlement of the
Sn-rich phase in the Sn-Pb eutectic.
the Ag content. The detrimental effect of Ag3Sn IMCs on the fracture behavior
of Sn-based solders has been reported in previous studies 关20,23兴 and is further
confirmed by the fractography study performed here 共Figs. 6共a兲, 7共a兲, and 8兲.
The effectiveness of Ag3Sn IMCs in their role as internal notches lies most
likely in their shape: They are faceted platelets with very sharp edges 共Figs. 6共c兲,
7共b兲, and 8共d兲兲, which results in very high stress intensity in the solder around
them.
Another impact test result that merits attention is that the more heavily
alloyed, Ag-containing solders do not become brittle at a well-defined DBTT
value but over a whole temperature range. The spread in DBTT values is shown
by superimposing shaded areas on the impact curves of SAC 305, SAC 405, and
Sn-5Ag in Fig. 1. It is believed that the observed spread in DBTT values is
related to the distribution of IMCs in the solder. To be more precise, this dis-
tribution cannot be controlled during conventional soldering, as it is impos-
sible to control IMC nucleation. What can be controlled is the IMC volume
fraction, as this is defined by the solder composition and can be predicted with
the help of equilibrium thermodynamics. However, thermodynamics cannot

predict the number, size, and spatial distribution of the IMC precipitates, espe-
cially under the non-equilibrium conditions of soldering. The reason is that
IMC precipitation is a stochastic process that cannot exclude the local agglom-
eration of IMCs, which may have detrimental effects on the solder fracture
resistance 共Figs. 3共c兲 and 7共a兲兲. The unpredictable aspects of the IMC distribu-
tion in the solder, like the occasional occurrence of IMC agglomeration, explain
partly why the embrittlement of Sn-based solders containing high amounts of
IMC-forming alloying elements occurs over a whole temperature range and not
at a single DBTT value.
The limited temperature dependence of the fracture behavior of the Sn-
37Pb solder can be explained by considering the solder microstructure, which
is made of the binary Sn–Pb eutectic structure. This structure is based on the
alternation of Pb-rich lamellae with the fcc crystal structure and Sn-rich lamel-
lae with the bct crystal structure. Since the bct Sn-rich phase undergoes a
ductile-to-brittle transition, while the fcc Pb-rich phase remains ductile at low
temperatures, the fracture behavior of the Sn-37Pb solder is obviously a com-
promise of these two behaviors.
Fractography Study Results
Solder 405—Ductile failures of strongly alloyed solders 共e.g., SAC 305, SAC
405, and Sn-5Ag兲 are characterized by the presence of dimples on the fracture
surface 共Fig. 2共a兲 and 2共b兲兲. Dimples are fracture surface cavities resulting from
the loss of cohesion between the hard, brittle IMC particles and the soft, ductile
Sn matrix during testing. This loss of cohesion results from the coalescence of
microvoids formed at the Sn/IMC interface, which eventually leads to crack
propagation through the test specimen 关20,31兴. The mechanism of dimple for-
mation is presented in the inset idealized drawing of Fig. 2共b兲.
Mixed ductile-brittle failures occur at intermediate temperatures and are
characterized by the alternation of dimple-rich areas 共ductile failure兲 and flat
areas 共brittle failure兲 on the fracture surface 共Figs. 2共c兲, 2共d兲, and 3兲. Upon
further cooling, the fraction of brittle failure on the fracture surface increases,
resulting into primarily brittle failures 共Fig. 4兲. Even though these observations
agree with the concept of gradual embrittlement, they neither identify the in-
dividual events contributing to solder embrittlement, nor describe their succes-
sion in the embrittlement process.
At moderately low temperatures, for instance, the fracture surface parts
that indicate brittle failure are associated with IMC failures 共Figs. 2共c兲, 3共a兲, and
3共c兲兲. Close inspection of these IMC failures reveals that the failure of the Sn
matrix changes from brittle in their immediate vicinity to ductile a few mi-
crometers away from the IMC particles 共Figs. 2共d兲, 3共b兲, and 3共d兲兲. This local-
ized solder embrittlement is facilitated by the triaxial stress state in the Sn
matrix next to IMC particles, as already mentioned. At more aggressive tem-
peratures, on the other hand, brittle failure extends also in the solder matrix
between IMC particles due to the embrittlement of ␤-Sn. In such conditions,
the overall appearance of the fracture surface is flat, including the solder parts
that are made of the ternary Sn– Ag3Sn– Cu6Sn5 eutectic structure 共Fig. 4兲.
Solder Sn-5Ag—The good ductility of the Sn-5Ag solder resulted in appre-

ciable plastic deformation and no fracture down to −31° C 共Fig. 5兲. Along the
stretch marks indicative of the plastic deformation around the notch, small
heaps of solder were observed 共Fig. 5共c兲兲. These solder heaps are believed to
have formed by IMC particles that pushed the solder around them into small
heaps as they tried to follow the overall specimen deformation 共Fig. 5共d兲 and
5共e兲兲. Similar features were also observed in solders Sn-0.7Cu 共Fig. 9兲 and Sn-
0.7Cu-0.1Ni 共Fig. 12兲 tested in impact at temperatures that did not promote
fracture.
Fractography analysis on this solder provided strong evidence of the detri-
mental effect of Ag3Sn platelets on the fracture resistance of the solder on a
microscopic level. Due to their shape and edge sharpness, these IMCs are high
stress intensity sites in the host solder, facilitating local solder embrittlement
even at moderately low temperatures 共Fig. 6共b兲 and 6共c兲兲. Moreover, the in-
creased volume fraction of Ag3Sn IMCs in this solder results more often in
agglomeration, a phenomenon that causes the embrittlement of large solder
areas, especially as the temperature decreases 共Fig. 7共a兲兲. At −75° C, a single
Ag3Sn platelet is capable of causing the catastrophic failure of a Sn dendrite or
a Sn– Ag3Sn eutectic colony many times its size 共Fig. 8兲.
Solder Sn-0.7Cu—The ductile-to-brittle transition in the fracture behavior

of the Sn-0.7Cu solder was completed within a narrow temperature range, as
indicated by both fractography analysis and impact test data. The embrittle-
ment of this lightly alloyed solder is primarily attributed to the embrittlement
of the ␤-Sn matrix, since the volume fraction of IMCs that could contribute to
the solder embrittlement process is very limited. The same holds for the other
two lightly alloyed solders, i.e., the 99.99 % Sn and the Sn-0.7Cu-0.1Ni.
The embrittlement of ␤-Sn results in a stepwise, cleavage-like failure 共Figs.
10共a兲, 10共b兲, 11共a兲, and 11共c兲兲. This type of fracture is transgranular and occurs
along certain atomic planes in ␤-Sn. As mentioned earlier, cleavage-like failure
occurs when the tensile stresses exceed the cohesive forces between certain
atomic planes, leading to fracture before the crystal gets the chance to deform
plastically.
Another type of failure that is common in all lightly alloyed solders is the
intergranular separation of Sn-Cu6Sn5 eutectic grains 共Figs. 10共d兲, 11共b兲, and
11共c兲兲. Preferential crack propagation at the Sn/eutectic interface indicates a
poor interfacial strength, most likely due to the dissimilarity of the two mate-
rials.
Solder Sn-0.7Cu-0.1Ni—The inhomogeneous appearance of the fracture

surface is thought to be associated with the inhomogeneous distribution of
nickel, which is an alloying element used only in solder Sn-0.7Cu-0.1Ni. It is
possible that the exaggerated growth of IMC particles in the smooth parts of
the fracture surface is also related to the segregation of Ni in these solder areas,
as it has not been observed in any of the other Sn-based solders. The rougher
parts of the fracture surface were characterized by the cleavage-like failure of

Sn and an IMC growth similar to that observed in the Sn-based solders with no
Ni addition. If the above hypothesis were true, a more systematic investigation
of the effects of Ni on the fracture behavior of ␤-Sn might be worthwhile.
The clean separation of the IMC particles from the Sn solder matrix leads
to a fracture resembling strongly that of fiber-reinforced composites with a
weak fiber/matrix interface. In such composites, even when both matrix and
fibers are brittle, the crack prefers to propagate along the fiber/matrix interface
than to break the reinforcing fibers, as interface debonding is energetically
more preferable than fiber failure 共Fig. 15共c兲 and 15共e兲兲. Of course, interface
debonding is followed eventually by fiber failure, as soon as the stress level in
the composite exceeds the fiber strength. The final stage in this type of compos-
ite material behavior, which is typical for certain ceramic matrix composites
reinforced with ceramic fibers, is fiber pull-out, whereupon the debonded part
of the fiber sticks out of the fracture surface. This is exactly what happens with
the broken IMC particles in the smooth fracture surface parts of solder Sn-
0.7Cu-0.1Ni 共Figs. 14共d兲, 14共f兲, and 15共f兲兲.
Solder 99.99 % Sn—The fracture surface of the embrittled 99.99 % Sn sol-

der demonstrates both the cleavage-like 共transgranular兲 failure 共Fig. 17共c兲 and
17共d兲兲 and intergranular failure 共Fig. 17共e兲 and 17共f兲兲 of Sn. The discovery of
twin crystals of ␤-Sn on the solder fracture surface 共Fig. 18兲 is a strong indica-
tion of material separation along certain crystal planes since twin crystals are
portions of the same crystal with specific crystallographic orientations relative
to each other 关36兴.
Solder Sn-37Pb—The Sn-37Pb solder is based on the binary Sn–Pb eutectic

structure; therefore, its fracture behavior may be explained by considering the
differences in the fracture behavior of the two phases in the eutectic. The Pb-
rich phase is more compliant 共E ⬇ 16 GPa兲 than the Sn-rich phase 共E
⬇ 50 GPa兲, a fact that justifies the more pronounced plastic deformation of the
Pb-rich phase at room temperature 共Fig. 19共b兲 and 19共c兲兲. As the temperature
decreases, the fcc Pb-rich phase maintains its ductility, while the ductility of the
bct Sn-rich phase decreases gradually due to the ductile-to-brittle transition in
the fracture behavior of ␤-Sn 共Figs. 19 and 20兲. Moreover, the solder tendency
towards intergranular failure at all temperatures suggests rather weak cohesion
between eutectic grains 共Figs. 19共a兲, 19共d兲, 20共c兲, and 20共d兲兲.
This work showed that the fracture behavior of bulk solder specimens
made of Sn-based solders is very similar to the previously reported fracture
behavior of joints made of the same solders 关20兴. Differences in the specimen
size and shape, as well as small differences in the solder microstructure, did not
prove capable of drastically changing the fracture behavior of these Pb-free
solders. Similar to the prior study on the solder joints, the distribution of IMCs
in the bulk solder specimens affected greatly the solder fracture behavior since
the IMCs are large defects that become critical as the ␤-Sn solder matrix un-
dergoes a ductile-to-brittle transition. Figures 23 and 24 illustrate the role of
IMCs on the fracture behavior of Sn-based solder joints and their potentially
detrimental effect on the solder joint impact reliability.
(a) (b)
10 μm 20 μm
100 μm (c) Solder (d)

mask
10 μm Bond pad 10 μm
(e) (f)
10 μm 20 μm
FIG. 23—SE detector images of the fracture surfaces of SAC 405 关共a兲–共d兲兴 and SAC 305
关共e兲 and 共f兲兴 solder joints tested in impact. 共a兲 Impact test at −22°C: The large Ag3Sn
platelet 共cross兲 has facilitated the embrittlement of the Sn matrix in its immediate vi-
cinity, as indicated by the local flatness of the fracture surface. 共b兲 Impact test at −74°C:
The agglomeration of Cu6Sn5 IMC particles 共arrows兲 has promoted locally the brittle
failure of the solder. 共c兲 Impact test at −78°C: Large IMC particles, such as the Cu6Sn5
IMC in the figure inset, act as critical-size flaws that initiate the solder joint brittle
failure at temperatures where the Sn matrix is embrittled. 共d兲 Impact test at −98°C: The
large Ag3Sn platelet 共cross兲 acts as critical-size defect that triggers the failure of the
embrittled solder. It is thought that the failure resulted from the fast propagation of a
crack that had nucleated at the Ag3Sn platelet and later deflected towards the solder
joint/bond pad interface. 共e兲 Impact test at −32°C: Indicated by a cross, a large Cu6Sn5
IMC particle 共length>100 ␮m兲 facilitates solder embrittlement in its vicinity. 共f兲 Impact
test at −85°C: A cross marks a large Cu6Sn5 IMC particle 共length>150 ␮m兲 that is
associated with the onset of brittle failure over a large part of this solder joint.
Conclusions
This work focused on the microstructural aspect of the ductile-to-brittle tran-
sition in the fracture behavior of Sn-based Pb-free solders. Moreover, the effect
of specimen size on the solder fracture behavior was considered, since two
(a) (b)
30 μm 10 μm
(c) (d)
10 μm 20 μm
FIG. 24—SEM images of different IMCs compromising the solder joint impact reliabil-
ity by acting as severe “notches.” The solder joints in 共a兲, 共b兲, and 共d兲 have been deeply
etched to reveal the IMCs, while 共c兲 shows the cross-section of a solder joint. The
observed IMCs are 共a兲 AuSn, AuSn3, and AuSn4; 共b兲 Cu6Sn5 and AuSn3Sn; 共c兲
AuSn3Sn; and 共d兲 Pd3Sn, Pd2Sn, PdSn, PdSn2, PdSn3, and PdSn4. These images are
courtesy of 共a兲 I. Hernefjord, Ericsson Microwave Systems, Sweden; 共b兲 R. Ghaffarian,
JPL, United States; 共c兲 H. Walter et al., AMIC GmbH, Germany; and 共d兲 L. Hyun-Kyu,
Duksan HI-Metal Co., South Korea.
different sizes were tested in impact, and the resulting fracture surfaces were
studied using SEM. The detailed fractography analysis showed that the distri-
bution of IMC particles in the solder interacts with the embrittlement process
of the ␤-Sn solder matrix, and that IMCs with rather sharp edges, like the
Ag3Sn platelets in Ag-containing solders, promote locally the solder embrittle-
ment even at temperatures where the embrittlement of ␤-Sn is not yet com-
plete. Moreover, it was observed that the stages in the embrittlement mecha-
nism of Sn-based solders are essentially the same, irrespective of the specimen
size; however, the fracture resistance of bulk solder specimens is usually supe-
rior to that of solder joints, due to the confinement of the latter between two
failure-prone, brittle IMC layers. The understanding of the ductile-to-brittle
transition in the fracture behavior of Sn-based solders was also employed to
explain previous results from the thermal cycling of eutectic Sn–Pb solder
joints.
Acknowledgments
K. Lambrinou would like to acknowledge IWT Flanders, Belgium, which sup-
ported financially the study of the fracture behavior of Pb-free solders in the
framework of the ALSHIRA Research Project 共Contract No. IWT 040373兲. K.
Lambrinou would also like to thank Dr. G. Papavassiliou of the Institute of
Materials Research of the National Centre of Scientific Research “Demokritos,”

Greece, for allowing access to the FEI Quanta Inspect D8334 SEM, and Dr. C.
Schmetterer of the Technische Universität Bergakademie Freiberg, Germany,
for providing a high-resolution copy of the 220° C isothermal section of the
Sn-Cu-Ni equilibrium phase diagram shown in Fig. 21共d兲. W. Engelmaier would
like to acknowledge the many contributors to the understanding of the behav-
ior of solder joints over th last 50 years. While, of course, it is impossible to
mention even the major contributors by name, special tribute needs to be paid
to R. Wild, IBM, for his now classical work on the understanding of solders. In
addition, he is in all likelihood the first to actually observe the effects of the
ductile-to-brittle transition in Sn-based solders.
References
关1兴 Newman, K., “BGA Brittle Fracture—Alternative Solder Joint Integrity Test Meth-
ods,” Proceedings of the 55th Electronic Components and Technology Conference,
Lake Buena Vista, FL, May–June 2005, IEEE, Piscataway, NJ, pp. 1194–1201.
关2兴 Newman, K., “Board-Level Solder Joint Reliability of High Performance Comput-
ers Under Mechanical Loading,” Proceedings of the Ninth Conference on Thermal,
Mechanical and Multiphysics Simulation and Experiments in Micro-Electronics and
Micro-Systems, Delft, The Netherlands, April 2008, IEEE, Piscataway, NJ, pp. 672–
686.
关3兴 Reiff, D. and Bradley, E., “A Novel Mechanical Shock Test Method to Evaluate
Lead-Free BGA Solder Joint Reliability,” Proceedings of the 55th Electronic Com-
ponents and Technology Conference, Lake Buena Vista, FL, May–June 2005, IEEE,
Piscataway, NJ, pp. 1519–1525.
关4兴 Siviour, C. R., Williamson, D. M., Palmer, S. J. P., Walley, S. M., Proud, W. G., and
Field, J. E., “Dynamic Properties of Solders and Solder Joints,” J. Phys. IV, Vol.
110, 2003, pp. 477–482.
关5兴 Song, F., Lee, S. W. R., Newman, K., Sykes, B., and Clark, S., “Brittle Failure
Mechanism of SnAgCu and SnPb Solder Balls During High Speed Ball Shear and
Cold Ball Pull Tests,” Proceeding of the 57th Electronic Components and Technology
Conference, Reno, NV, May–June 2007, IEEE, Piscataway, NJ, pp. 364–372.
关6兴 Tee, T. Y., Ng, H. S., Lim, C. T., Pek, E., and Zhong, Z., “Board Level Drop Test and
Simulation of TFBGA Packages for Telecommunication Applications,” Proceedings
of the 53rd Electronic Components and Technology Conference, New Orleans, LA,
May 2003, IEEE, Piscataway, NJ, pp. 121–129.
关7兴 Tsai, K. T., Liu, F.-L., Wong, E. H., and Rajoo, R., “High Strain Rate Testing of
Solder Interconnections,” Soldering Surf. Mount Technol., Vol. 18, No. 2, 2006, pp.
12–17.
关8兴 Wong, E. H., Rajoo, R., Seah, S. K. W., Selvanayagam, C. S., van Driel, W. D.,
Caers, J. F. J. M., Zhao, X. J., Owens, N., Tan, L. C., Leoni, M., Eu, P. L., Lai, Y.-S.,
and Yeh, C.-L., “Correlation Studies for Component Level Ball Impact Shear Test
and Board Level Drop Test,” Microelectron. Reliab., Vol. 48, 2008, pp. 1069–1078.
关9兴 Wong, E. H., Seah, S. K. W., and Shim, V. P. W., “A Review of Board Level Solder
Joints for Mobile Applications,” Microelectron. Reliab., Vol. 48, 2008, pp. 1747–
1758.
关10兴 Wong, E. H., Seah, S. K. W., van Driel, W. D., Caers, J. F. J. M., Owens, N., and Lai,
Y.-S., “Advances in the Drop-Impact Reliability of Solder Joints for Mobile Appli-
cations,” Microelectron. Reliab., Vol. 49, 2009, pp. 139–149.

关11兴 Caers, J. F. J. M., Wong, E. H., Seah, S. K. W., Zhao, X. J., Selvanayagam, C. S., van
Driel, W. D., Owens, N., Leoni, M., Tan, L. C., Eu, P. L., Lai, Y.-S., and Yeh, C.-L., “A
Study of Crack Propagation in Pb-Free Solder Joints under Drop Impact,” Proceed-
ings of the 58th Electronic Components and Technology Conference, Lake Buena
Vista, FL, May 2008, IEEE, Piscataway, NJ, pp. 1166–1172.
关12兴 Seah, S. K. W., Wong, E. H., Mai, Y. W., Rajoo, R., and Lim, C. T., “High-Speed
Bend Test Method and Failure Prediction for Drop Impact Reliability,” Proceedings
of the 56th Electronic Components and Technology Conference, San Diego, CA, May–
June 2006, IEEE, Piscataway, NJ, pp. 1003–1008.
关13兴 Date, M., Shoji, T., Fujiyoshi, M., Sato, K., and Tu, K. N., “Impact Reliability of
Solder Joints,” Proceedings of the 54th Electronic Components and Technology Con-
ference, Las Vegas, NV, June 2004, IEEE, Piscataway, NJ, pp. 668–674.
关14兴 Date, M., Shoji, T., Fujiyoshi, M., Sato, K., and Tu, K. N., “Ductile-to-Brittle Tran-
sition in Sn–Zn Solder Joints Measured by Impact Test,” Scr. Mater., Vol. 51, 2004,
pp. 641–645.
关15兴 Ou, S., Xu, Y., and Tu, K. N., “Micro-Impact Test on Lead-Free BGA Balls on
Au/Electrolytic Ni/Cu Bond Pad,” Proceedings of the 55th Electronic Components
and Technology Conference, Lake Buena Vista, Fl, May–June 2005, IEEE, Piscat-
away, NJ, pp. 467–471.
关16兴 JESD22-B117A, October 2006, “Solder Ball Shear,” Joint Electron Devices Engi-
neering Council, Arlington, VA.
关17兴 JESD22-B115, May 2007, “Solder Ball Pull,” Joint Electron Devices Engineering
Council, Arlington, VA.
关18兴 Engelmaier, W., “‘Pad Cratering’ & ‘Trace Buckling’—New Failure Modes Created
by Pb-Free Soldering,” Global SMT & Packaging, Vol. 10, No. 6, 2010, pp. 36-
38.0002-7820
关19兴 Wild, R., “1974 IRAD Study—Fatigue Properties of Solder Joints,” IBM Technical
Report No. M45-74-002, Contract No. IBM 4A69, Jan. 5, 1975.
关20兴 Lambrinou, K., Maurissen, W., Limaye, P., Vandevelde, B., Verlinden, B., and De
Wolf, I., “A Novel Mechanism of Embrittlement Affecting the Impact Reliability of
Tin-Based Lead-Free Solder Joints,” J. Electron. Mater., Vol. 38, No. 9, 2009, pp.
1881–1895.
关21兴 Wild, R., “Intermetallic Compounds and Their Effect on Solder Joint Strengths,”
IBM Technical Report No. 65-581-141, 1965.
关22兴 Wild, R., “Effects of Gold on Solder’s Properties,” Proceedings of INTERNEPCON
UK, Brighton, United Kingdom, 1968, pp. 27–32.
关23兴 Engelmaier, W., “Solder Joint Formation & Intermetallic Compounds 共IMCs兲,”
Global SMT & Packaging, Vol. 3, No. 4, 2003, pp. 36–38.
关24兴 Engelmaier, W., “Soldering Pad Surfaces–Gold, Silver & Their IMCs: Solder Joint
‘Embrittlement’,” Global SMT & Packaging, Vol. 3, No. 6, 2003, pp. 29–32.
关25兴 Lal, A., Bradley, E., and Sharda, J., “Effect of Reflow Profiles on the Board Level
Drop Reliability of Pb-Free 共SnAgCu兲 BGA Assemblies,” Proceedings of the 55th
Electronic Components and Technology Conference, Lake Buena Vista, FL, May–
June 2005, IEEE, Piscataway, NJ, pp. 945–953.
关26兴 Ratchev, P., Loccufier, T., Vandevelde, B., Verlinden, B., Teliszewski, S.,
Werkhoven, D., and Allaert, B., “A Study of Brittle to Ductile Fracture Transition
Temperatures in Bulk Pb-Free Solders,” Proceedings of the 2005 IMAPS European
Microelectronics Packaging Conference, Brugge, Belgium, June 12–15, 2005,
IMAPS, Washington, DC, pp. 248–252.
关27兴 Ratchev, P., Vandevelde, B., and Verlinden, B., “Brittle to Ductile Fracture Transi-
tion in Bulk Pb-Free Solders,” IEEE Trans. Compon. Packag. Technol., Vol. 30, No.
3, 2007, pp. 416–423.
关28兴 ASTM E23-06, 2006, “Standard Test Methods for Notched Bar Impact Testing of
Metallic Materials,” Annual Book of ASTM Standards, ASTM International, West
Conshohocken, PA.
关29兴 Engelmaier, W., “Guidelines for Accelerated Testing of Surface Mount Solder At-
tachments,” IPD-SM-785, IPC-Institute for Interconnecting and Packaging Elec-
tronic Circuits, Lincolnwood, IL, November 1992.
关30兴 Kariya, Y., Gagg, C., and Plumbridge, W. J., “Tin Pest in Lead-Free Solders,” Sol-
dering Surf. Mount. Technol., Vol. 13, No. 1, 2001, pp. 39–40.
关31兴 Hertzberg, R. W., Deformation and Fracture Mechanics of Engineering Materials,
John Wiley & Sons, Inc., New York, 1996.
关32兴 McMahon, C. J., Jr. and Graham, C. D., Jr., Introduction to Engineering Materials:
The Bicycle and the Walkman, Merion Books, Philadelphia, 2000.
关33兴 Barsom, J. M. and Rolfe, S. T., Fracture and Fatigue Control in Structures: Applica-
tions of Fracture Mechanics, ASTM Manual Series MNL41, ASTM International,
West Conshohocken, PA, 1999.
关34兴 Ratchev, P., Vandevelde, B., and Verlinden, B., Proceedings of the IPC/JEDEC Tenth
Int. Conf. Lead-Free Electron. Comp. Assemblies 共CD-ROM兲, Brussels, Belgium, Oc-
tober 2005, Joint Electron Devices Engineering Council, Arlington, VA.
关35兴 Askeland, D. R., The Science and Engineering of Materials, Brooks/Cole, Monterey,
CA, 1984.
关36兴 Barrett, C. S., and Pearson, W. B., “Metallurgy and Microstructures,” ASM Hand-
book, ASM International, Materials Park, OH 44073-0002, 2004, Vol. 9.
关37兴 Cu–Sn Equilibrium Phase Diagram, Second Edition of Binary Alloy Phase Dia-
grams, NIST Scientific and Technical Databases.
关38兴 Ag–Sn Equilibrium Phase Diagram, Second Edition of Binary Alloy Phase Dia-
grams, NIST Scientific and Technical Databases.
关39兴 http://www.metallurgy.nist.gov/phase/solder/agcusn.html 共Last accessed Feb. 28,
2010兲, NIST, Phase Diagrams & Computational Thermodynamics.
关40兴 Schmetterer, C., Flandorfer, H., Luef, Ch., Kodentsov, A., and Ipser, H., “Cu–Ni–
Sn: A Key System for Lead-Free Soldering,” J. Electron. Mater., Vol. 38, No. 1,
2009, pp. 10–24.
doi:10.1520/JAI103021
Feng Gao,1 Jianping Jing,2 Frank Z. Liang,3

Richard L. Williams,3 and Jianmin Qu4
Loading Mixity on the Interfacial Failure

Mode in Lead-Free Solder Joint
ABSTRACT: In this paper, single solder joints 共SSJs兲 were subjected to mod-
erate speed loading 共5 mm/s兲 in different directions, from pure tensile mixity
mode to pure shear. Fracture surfaces from different loading directions were
examined both experimentally and numerically. The intermetallic compound
共IMC兲 is formed between the solder alloy and the Cu pad, and the failure
typically occurs at or near the solder/IMC/Cu interfaces of the board side.
Pure tensile loading typically leads to interfacial fracture along the IMC/Cu
interface. Mixity mode loading usually results in a mixture of interfacial and
cohesive failure with damage propagating in a zigzag fashion between the
solder/IMC interface and the solder alloy. Loading with higher shear compo-
nent tends to result in more cohesive failure of the solder alloy near the
solder/IMC interface. Under pure shear loading, failure is almost always co-
hesive within the solder alloy near the solder/IMC interface.
KEYWORDS: lead-free solder, single solder joint, damage
propagation, plastic deformation, interface, finite element analysis
1
George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technol-
ogy, Atlanta, GA 30332-0405 and McCormick School of Engineering and Applied Sci-
ence, Northwestern Univ., Evanston, IL 60208, e-mail: feng-gao@northwestern.edu
2
The State Key Laboratory of Mechanical System and Vibration, Shanghai JiaoTong
Univ., Shanghai 200000, China.
3
Intel Corporation, Hillsboro, OR 97124.
4
George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technol-
ogy, Atlanta, GA 30332-0405 and McCormick School of Engineering and Applied Sci-
ence, Northwestern Univ., Evanston, IL 60208.
Cite as: Gao, F., Jing, J., Liang, F. Z., Williams, R. L. and Qu, J., ‘‘Loading Mixity on the
Interfacial Failure Mode in Lead-Free Solder Joint,’’ J. ASTM Intl., Vol. 7, No. 5.
doi:10.1520/JAI103021.
121
Introduction
Due to the stiffer and more brittle characteristics of lead-free solder alloys, the
solder joints of portable electronic products are prone to drop and impact dam-
age 关1–6兴. This is further compounded by packaging miniaturization, which
reduces the amount of solder material available to absorb shock energy. It has
been found that when a portable device drops on the floor, the local strain rate
within a solder joint may vary between 1 and 1000 s−1, depending on the drop
height, orientation, and the properties of the floor surface 关7兴. The failure be-
havior of solder joints subjected to high strain rates has been studied exten-
sively 关8–12兴. The tests are typically the ball grid array 共BGA兲 component drop/
impact tests at the board level, while the failure usually initiates at the solder
joint level. Thus understanding the failure process of a single solder joint 共SSJ兲
may lead to a more detailed damage mechanism. In the meantime, the high-
speed pull and shear impact tests have also been utilized to evaluate the failure
mode of the solder joints 关13–17兴. In reality, solder ball interconnections may be
subjected to the combined tensile, shear, and peeling stresses. Therefore a re-
alistic assessment of solder ball integrity should consider the loading compo-
nents simultaneously. However, there is very little study on the failure behavior
of solder joint under different loading mixities at an intermediate strain rate
range between 1 and 100 s−1. The larger loading mixity indicates a greater
shear component but a less normal component. Therefore, there has been a
critical need to understand failure modes and mechanism of a SSJ subjected to
dynamic loading mixity at intermediate strain rate.
In this study we report some results regarding failure mode under a mod-
erate strain rate and how the failure mode changes under different combina-
tions of normal and shear loading. The SSJs were subjected to velocity con-
trolled loading. The optical microscopy on fracture surface was conducted to
verify the failure mode. To interpret the experimental observations, the finite
element analysis was performed to understand the failure mechanism during
the dynamic loading process.
Experimental Procedures
The SSJ samples used in this study were laser-cut from a BGA package as-
sembled on a printed circuit board 共PCB兲. A schematic of the finished SSJ is
shown in Fig. 1共a兲. The commercial Sn-4.0Ag-0.5Cu 共SAC405兲 solder alloy was
used with the SSJ failure to occur along the PCB interfaces. This was accom-
plished by designing the BGA package–solder ball interface area greater than
the solder joint–PCB interface area, commonly referred to as the solder joint
aspect ratio. The SSJ samples are loaded using a high-speed loading frame
equipped with a specially design test apparatus. Samples can be gripped in
different orientations so that the loading angle ␪ between the loading direction
and the PCB surface can vary with 0° corresponding to pure tension and 90°
corresponding to pure shear.
Another unique feature of the test apparatus is that the load is not applied
to the SSJ sample until the grip has reached the desired speed. This removes
GAO ET AL., doi:10.1520/JAI103021 123
FIG. 1—The SSJ testing: 共a兲 The schematic diagram of the SSJ; 共b兲 experimental force-
displacement curves versus loading mixity.
the inertia of the load frame and applies a true impact load to the SSJ sample
with known velocity.
In this study, tests were conducted under four loading angles of 0°, 30°, 60°,
and 90° to investigate the effect of loading mixity. The board substrate of the
SSJ was fixed, while the substrate at the package side was subjected to the
velocity loading. The grip was set to move at 5 mm/s. The reaction force mea-
sured by the load sensor attached to the grip was recorded every 2 ⫻ 10−4 s.
The corresponding grip displacement was also recorded to obtain the force
versus time or force versus displacement curve. The cross-sectional optical mi-
crocopy was conducted on the SSJ samples both before and after the dynamic
test.
Figure 1共b兲 illustrates the measured force-displacement curves due to different

loading mixity at 5 mm/s. Basically, the peak force continues to decrease with
the larger loading mixity, namely, the greater shear component. On the con-
trary, the time of the peak force occurrence increases with the larger loading
mixity. In addition, the full failure displacement for shear test is much larger
than that of pure tensile test. These results indicate that the different failure
mode may take place under different loading mixity, which will be stressed in
detail below based on the fracture surface observations.
Figure 2共a兲 shows the microstructure of a SSJ before testing. The interme-
tallic compound 共IMC兲 was formed at both PCB board and package sides, act-
ing as the metallurgical interconnection. The non-homogeneous microstruc-
ture of solder alloy consists of ␤-Sn, 共␤-Sn+ Ag3Sn兲 eutectic, and 共␤-Sn
+ Ag3Sn+ Cu6Sn5兲 eutectic phases. Large-needle-shaped Ag3Sn particles are
also observed, which is attributed to the high initial Ag content in SAC405 and
the solidification process 关18兴. Figure 2共b兲 is a polarized image showing that
there are only a few grains in a SSJ. The different contrast of these grains
represents different grain orientations. Such high non-homogeneous grain
structure will partially affect the material property of the small size lead-free
solder ball.
In order to investigate the effect of loading mixity, four different loading
directions were used, that is, 0° 共pure tensile兲, 30°, 60°, and 90° 共pure shear兲. It
is found that the majority of SSJ samples failed at the interfaces of the board
side. Figures 3–6 illustrate the failure behavior of the SSJ samples under the
loading rate of 5 mm/s at different loading angles.
It is seen from Fig. 3共a兲 that under pure tensile loading, the damage devel-
ops along the IMC/Cu interface of the board side. Almost all the IMC is at-
tached with the solder ball, while only little IMC residue is probed on the Cu
pad, as shown in Fig. 3共b兲 and 3共c兲. Thus a brittle interfacial fracture along the
IMC/Cu interface of the board side is suggested. Figure 4 shows a SSJ sample
failed under a loading angle of 30°. Again, failure occurs at the board side; see
Fig. 4共a兲. However, the fracture surface is no longer at the IMC/Cu interface. Its
zigzag path alternates between the solder alloy and the solder/IMC interface, as
illustrated in Fig. 4共b兲. Under a higher loading angle of 60°, the facture path
GAO ET AL., doi:10.1520/JAI103021 125
FIG. 2—The microstructure of SSJ: 共a兲 Optical micrograph; 共b兲 the corresponding po-
larized image.
shows the similar zigzag form with more cohesive failure within the solder
alloy; see Fig. 5共a兲 and 5共b兲. Under pure shear loading at 90°, failure occurs
almost entirely within the solder alloy near the solder/IMC interface, as shown
in Fig. 6共a兲 and 6共b兲. In addition, Figs. 4共a兲, 5共a兲, and 6共a兲 also show the plastic
deformation behavior of the solder ball due to the shear component of the
FIG. 3—Failure occurs along the IMC/Cu interface at the board side at 0° loading: 共a兲
Solder joint at package side; 共b兲 close-up of the fracture interface; and 共c兲 residue Cu pad
at board side.
GAO ET AL., doi:10.1520/JAI103021 127
FIG. 4—Failure occurs along the path of 共solder/IMC+solder matrix兲 at board side
under the loading mixity of 30°: 共a兲 Solder joint at package side; 共b兲 residue Cu pad at
board side.
angular loading. The higher the loading angle, the more severe the shear defor-
mation.
In summary, under the loading speed of 5 mm/s, pure normal tension leads
to a brittle interfacial failure of the IMC/Cu pad interface. Higher loading angle,
FIG. 5—Failure occurs along the path of 共solder/IMC+solder matrix兲 at board side
under the loading mixity of 60°: 共a兲 Solder joint at package side; 共b兲 residue Cu pad at
board side.
which corresponds to a larger shear component, leads to a more cohesive fail-

ure within the solder alloys, while pure shear loading results in almost entirely
cohesive failure. At the high drop/impact loading rate, the failure of lead-free
solder joint is usually brittle and occurs at the IMC/substrate interface regard-
GAO ET AL., doi:10.1520/JAI103021 129
FIG. 6—Failure occurs along the solder ball near the solder/IMC interface of board side
under the pure shear loading 共90°兲: 共a兲 Solder joint at package side; 共b兲 residue Cu pad
at board side.
less of the loading mixity 共e.g., tension or shear兲关1–4,8–10兴. Our experimental
results show that at the moderate strain rate as the solder joint studied herein,
the loading mode will be sensitive to the loading mixity.
To better understand and interpret the experimental observations discussed
FIG. 7—Numerical simulation model of SSJ: 共a兲 3D geometry of a SSJ; 共b兲 close-up of
the interfaces configuration at both package and board sides.
above, the testing under different loading mixities was simulated using the
finite element method. The simulation was conducted using the commercial
software ABAQUS®. The three-dimanional 共3D兲 geometry of a typical SSJ is
shown in Fig. 7共a兲. A 3D finite element model is then constructed for a SSJ. The
following components are included in this model: Substrates at package and
GAO ET AL., doi:10.1520/JAI103021 131
TABLE 1—Dimensions of SSJ specimen.
PCB Cu pad interface diameter 共␮m兲 350

Substrate pad interface diameter 共␮m兲 450
Solder ball diameter 共␮m兲 550
Solder joint height 共␮m兲 330
Substrate/PCB 共width⫻ depth兲共␮m兲 1400⫻ 1120
board sides, Ni finish at package side, Cu pad at board side, SAC405 lead-free
solder ball, solder mask, and IMC layers between solder and Cu pad at board
sides. Figure 7共b兲 depicts a close-up configuration of the interfaces at both
package and board sides. All relevant geometric dimensions and materials
properties are listed in Tables 1–3. The intermetallic compound is regarded as
Cu6Sn5 at board side. Their properties are determined based on the first-
principles calculation 关19兴. In particular, the SAC405 solder is modeled as
elastic-plastic using classic metal plasticity law, which is extracted inversely by
fitting the experimental force-displacement curve. The results are illustrated in
Table 4. To simulate a dynamic loading, the bottom surface of the finite element
model is constrained in x-, y-, and z-directions, which mimic the situation
where the bottom of the sample is glued to a rigid substrate. A velocity of 5
mm/s is prescribed for all the nodes on the top surface of the finite element
model.
In order to reveal the plastic deformation or stress fields clearly, cross-
sectional illustrations are presented below. The stress field is expected to indi-
cate the potential site for the damage initiation, while the equivalent plastic
strain 共PEEQ兲 at the solder alloy is employed to show the possible damage
propagation path qualitatively. The corresponding simulation results are pre-
sented in Figs. 8–11.
Figure 8共a兲 and 8共b兲 shows the plastic deformation and von Mises-stress
contours under pure tensile loading 共0°兲, respectively. It can be seen that the
maximum stress concentration is formed at the edge of solder/IMC/Cu pad
interfaces of the board side. Figure 8共c兲 illustrates the close-up of the von
Mises-stress field at the interface area of the board side. At the package side, no
severe stress concentration is observed. This indicates that the solder/IMC/Cu
interfaces at the board side is the dangerous site for the damage initiation,
which is consistent with the experimental observations. The maximum plastic
deformation is mainly located at the edge of the interface between the solder
alloy and the IMC layer and expands towards the solder alloy. Interestingly, at
the board side, the solder alloy adjacent to the IMC layer does not suffer a
remarkable plastic deformation, as shown in Fig. 8共a兲. Since the plastic defor-
TABLE 2—Isotropic material parameters.
Solder 共GPa兲 Copper 共GPa兲 SM 共GPa兲 IMC 共GPa兲关19兴

E 53 117 24 119
v 0.3425 0.34 0.4 0.29
TABLE 3—Anisotropic material parameters.
E1 , v1 E2 , v2 E3 , v3 G12 G13 G23

共GPa,/兲共GPa,/兲共GPa,/兲共GPa兲共GPa兲共GPa兲
PCB 22, 0.28 22,0.28 4.8,0.18 8 4 4
Substrate 21,0.3 21,0.3 6,0.2 8 4 4
mation in the adjacent ductile layer 共SAC405 lead-free solder alloy兲 has remark-
able toughening effect on the interface fracture 关2,13兴, it can be concluded that
the SSJ is more susceptible to the brittle interfacial fracture along the IMC/Cu
interface under high pure tensile loading.
Under the loading mixities of 30° and 60°, the maximum stress concentra-
tion still exists at the edge of solder/IMC/Cu interfaces. However, due to the
shear stresses, the asymmetry stress contours are formed, as shown in Figs.
9共b兲 and 10共b兲. In Figs. 9共b兲 and 10共b兲, the maximum stress concentration is
located at the right edge of the board side interfaces, which corresponds to the
damage initiation site. In addition, the shear stress also leads to an asymmetry
plastic deformation contour of the solder alloy, as shown in Figs. 9共a兲 and 10共a兲.
A relatively severe plastic deformation at the left edge of the package side is also
formed. The maximum plastic deformation occurs at the right edge of the
board side, which may also engender the damage initiation at that location.
It is interesting to notice that the plastic deformation of the solder alloy
adjacent to the IMC layer is also altered. That is, the plastic deformation area of
solder ball adjacent to solder/IMC interface tends to be enhanced with loading
angle 共or larger shear component兲. Obviously, this will make the damage propa-
gation shift up to the solder/IMC interface or even the solder alloys. Due to the
different magnitudes of the plastic deformation along the solder/IMC interface,
as shown in Figs. 9共a兲 and 10共a兲, zigzag damage propagation along the path
共solder/IMC interface+ solder matrix兲 may occur. This simulation result is con-
sistent with the microstructure observations shown in Figs. 4共b兲 and 5共b兲.
Under the pure shear loading 共90°兲, as shown in Fig. 11共a兲, the maximum plas-
tic deformation lies on the solder ball area adjacent to IMC layer, which is more
effective to release the solder/IMC interfacial energy by the solder alloy. This
will result in an entire cohesive failure within the solder alloys, which is also
consistent with the experimental results shown in Fig. 6共b兲.
Conclusions
The damage behavior of a SSJ subjected to different loading mixities at 5 mm/s
rate is investigated in this work. It is found that the failure typically occurs at or
near the solder/IMC/Cu interfaces on the board side. Simulation result also
TABLE 4—Elastic-plastic property of SAC405 solder alloy.
Flow stress 共MPa兲 26 60 80 120 150

Plastic strain 0 0.005 0.01 0.03 0.05
GAO ET AL., doi:10.1520/JAI103021 133
FIG. 8—Stress and equivalent plastic deformation 共PEEQ兲 contours under pure tensile
loading 共0°兲: 共a兲 PEEQ; 共b兲 von Mises stress; and 共c兲 close-up of von Mises stress at
board side.
FIG. 9—Stress and equivalent plastic deformation contours under loading mixity of
30°: 共a兲 PEEQ; 共b兲 von Mises stress; and 共c兲 close-up of von Mises stress at board side.
GAO ET AL., doi:10.1520/JAI103021 135
FIG. 10—Stress and equivalent plastic deformation contours under loading mixity of
60°: 共a兲 PEEQ; 共b兲 von Mises stress; and 共c兲 close-up of von Mises stress at board side.
FIG. 11—Stress and equivalent plastic deformation contours under pure shear loading
共90°兲: 共a兲 PEEQ; 共b兲 von Mises stress; 共c兲 close-up of von Mises stress at board side.
GAO ET AL., doi:10.1520/JAI103021 137
shows that the maximum stress concentration occurs at the solder/IMC/Cu in-
terfaces on the board side, which corresponds to the dangerous sites for the
damage initiation. Pure tensile loading typically leads to interfacial fracture
along the IMC/Cu interface. Mixed mode loading usually results in a mixture of
interfacial and cohesive failure with damage propagating in a zigzag fashion
between the solder/IMC interface and the solder alloy. Loading with higher
shear component tends to result in more cohesive failure of the solder alloy
near the solder/IMC interface. Under pure shear loading, failure is almost al-
ways cohesive within the solder ball near the solder/IMC interface. The failure
mode transition is attributed to the plastic deformation alteration of solder
alloy adjacent to the IMC layer on the board side.
Acknowledgments
The financial support from Intel Corporation is greatly acknowledged. Also the
writers would like to thank Mr. Carter Ralph for the sample preparation and
test setup.
References
关1兴 Wong, E. H., Rajoo, R., Seah, S. K. W., Selvanayagam, C. S., van Driel, W. D.,
Caers, J. F. J. M., Zhao, X. J., Owens, N., Tan, L. C., Leoni, M., Eu, P. L., Lai, Y.-S.,
and Yeh, C.-L., “Correlation Studies for Component Level Ball Impact Shear Test
and Board Level Drop Test,” Microelectron. Reliab., Vol. 48, 2008, pp. 1069–1078.
关2兴 Suh, D., Kim, D.-W., Liu, P. L., Kim, H., Weninger, J. A., Kumar, C. M., Prasad, A.,
Grimsley, B. W., and Tejada, H. B., “Effects of Ag Content on Fracture Resistance
of Sn–Ag–Cu Lead-Free Solders Under High-Strain Rate Conditions,” Mater. Sci.
Eng., A, Vol. 460–461, 2007, pp. 595–603.
关3兴 Wong, E. H. and Mai, Y.-W., “Advances in the Drop-Impact Reliability of Solder
Joints for Mobile Applications,” Microelectron. Reliab., Vol. 49, 2009, pp. 139–149.
关4兴 Wong, E. H., Selvanayagam, C. S., Seah, S. K. W., van Driel, W. D., Caers, J. F. J.
M., Zhao, X. J., Owens, N., Tan, L. C., Frear, D. R., Leoni, M., Lai, Y.-S., and Yeh,
C.-L., “Stress-Strain Characteristics of Tin-Based Solder Alloys for Drop-Impact
Modeling,” J. Electron. Mater., Vol. 37, 2008, pp. 829–836.
关5兴 Liu, Y. L., Gale, S., and Johnson, R. W., “Investigation of the Role of Void Forma-
tion at the Cu-to-Intermetallic Interface on Aged Drop Test Performance,” IEEE
Trans. Electron. Packag. Manuf., Vol. 30, 2007, pp. 63–73.
关6兴 Mattila, T. T., Marjamaki, P., and Kivilahti, J. K., “Reliability of CSP Interconnec-
tions Under Mechanical Shock Loading Conditions,” IEEE Trans. Compon. Packag.
Technol., Vol. 29, 2006, pp. 787–795.
关7兴 Long, X., Dutta, I., Sarihan, V., and Frear, D. R., “Deformation Behavior of Sn-
3.8Ag-0.7Cu Solder at Intermediate Strain Rates: Effect of Microstructure and Test
Conditions,” J. Electron. Mater., Vol. 37, 2008, pp. 189–200.
关8兴 Yeh, C.-L., Lai, Y.-S., and Kao, C.-L., “Evaluation of Board-Level Reliability of
Electronic Packages Under Consecutive Drops,” Microelectron. Reliab., Vol. 46,
2006, pp. 1172–1182.
关9兴 Luan, J.-E., Tee, T. Y., Pek, E., Lim, C. T., and Zhong, Z. W., “Dynamic Responses
and Solder Joint Reliability Under Board Level Drop Test,” Microelectron. Reliab.,
Vol. 47, 2008, pp. 450–460.
关10兴 Wong, E. H., Seah, S. K. W., and Shim, V. P. W., “A Review of Board Level Solder
Joints for Mobile Applications,” Microelectron. Reliab., Vol. 48, 2008, pp. 1747–
1758.
关11兴 Li, J., Mattila, T. T., and Kivilahti, J. K., “Computational Assessment of the Effects
of Temperature on Wafer-Level Component Boards in Drop Tests,” IEEE Trans.
Compon. Packag. Technol., Vol. 32, 2009, pp. 38–43.
关12兴 Zaal, J. J. M., van Driel, W. D., Kessels, F. J. H. G., and Zhang, G. Q., “Correlating
Drop Impact Simulations with Drop Impact Testing Using High-Speed Camera
Measurements,” J. Electron. Packag., Vol. 131, 2009, pp. 011007.
关13兴 Yeh, C.-L. and Lai, Y.-S., “Effect of Solder Alloy Constitutive Relationships on
Impact Force Response of Package-Level Solder Joints Under Ball Impact Test,” J.
Electron. Mater., Vol. 35, 2006, pp. 1892–1901.
关14兴 Lai, Y.-S., Yeh, C.-L., Chang, H.-C., and Kao, C.-L., “Characterizations of Ball Im-
pact Responses of Wafer-Level Chip-Scale Packages,” J. Alloys Compd., Vol. 450,
2008, pp. 238–244.
关15兴 Morita, T., Kajiwara, R., Ueno, I., and Okabe, S., “New Method for Estimating
Impact Strength of Solder-Ball Bonded Interfaces in Semiconductor Packages,”
Jpn. J. Appl. Phys., Vol. 47, 2008, pp. 6566–6568.
关16兴 You, T., Kim, Y., Kim, J., Lee, J., Jung, B., Moon, J., and Choe, H., “Predicting the
Drop Performance of Solder Joint by Evaluating the Elastic Strain Energy from
High-Speed Ball Pull Tests,” J. Electron. Mater., Vol. 38, 2009, pp. 410–414.
关17兴 Liu, D.-S., Kuo, C.-Y., Hsu, C.-L., Shen, G.-S., Chen, Y.-R., and Lo, K.-C., “Failure
Mode Analysis of Lead-Free Solder Joints Under High Speed Impact Testing,”
Mater. Sci. Eng., A, Vol. 494, 2008, pp. 196–202.
关18兴 Gao, F., Nishikawa, H., and Takemoto, T., “Intermetallics Evolution in Sn-3.5Ag
Based Lead-Free Solder Matrix on an OSP Cu Finish,” J. Electron. Mater., Vol. 36,
2007, pp. 1630–1634.
关19兴 Lee, N. T. S., Tan, V. B. C., and Lim, K. M., “First-Principle Calculations of Struc-
tural and Mechanical Properties of Cu6Sn5,” Appl. Phys. Lett., Vol. 88, 2006, pp.
031913.
doi:10.1520/JAI103044
Phil Geng1
Ball Grid Array Lead-Free Solder Joint

Strength under Monotonic Flexural Load
ABSTRACT: This work compared ball grid array 共BGA兲 lead-free solder joint
strengths to eutectic lead 共Sn–Pb兲 solder joint strengths under monotonic
bend load at room temperature. Flexural test methodologies for evaluating
solder joint strength are presented. Various effects on solder joint strength
were summarized systematically into three parts. The first part focused on
the effect of solder joint geometries. BGAs with Sn-4Ag-0.5Cu and 63Sn-
37Pb solders were tested, respectively. The effects of package side solder
resist opening sizes, solder ball diameters, and board side metal defined/
solder mask defined pads were investigated with 0.062 in printed circuit
board 共PCB兲. The results showed that the solder joint strength of Sn–Ag–Cu
solder is lower than that of the traditional Sn–Pb solder under room tempera-
ture board flexural load and similar dynamic load. The second part investi-
gated the effects of type of package 共plastic BGA 共PBGA兲 versus ceramic
BGA 共CBGA兲兲, board thickness 共0.093 in. versus 0.135 in.兲, and the effect of
rework 共reworked versus non-reworked兲 with Sn-3.9Ag-0.7Cu and 63Sn-
37Pb solders. The joint strength of Sn–Ag–Cu solder is consistently lower
than that of eutectic Sn–Pb solder for both board thicknesses, both CBGA
and PBGA packages, and both non-reworked and reworked packages. The
third part explored the feasibility of alternative low temperature solders as
board-level interconnects. In addition to the traditional 63Sn-37Pb solder and
the lead-free Sn-4Ag-0.5Cu solder, four other lead-free solders 共Sn-52In,
Sn-58Bi, Sn-57Bi-1Ag, and Sn-9Zn-0.006Al兲 were tested with 0.044 in PCB.
Effects of board surface finishes with immersion silver 共ImAg兲 or organic
solderability preservatives, and pads with via-in-pad 共VIP兲 or non-VIP pads
were investigated. Test results showed that most of the BGAs with non-VIP
pads performed better than those with VIP pads, except Sn–In solder with
ImAg surface finish. The Sn–In solder showed the lowest performance, while
Manuscript received February 20, 2010; accepted for publication May 18, 2010; pub-
1
Intel Corporation, 2111 NE 25th Ave., Hillsboro, OR 97124, e-mail:
phil.geng@alum.mit.edu
Cite as: Geng, P., ‘‘Ball Grid Array Lead-Free Solder Joint Strength under Monotonic
Flexural Load,’’ J. ASTM Intl., Vol. 7, No. 6. doi:10.1520/JAI103044.
139
Sn–Bi and Sn–Bi–Ag solder compositions showed better performance. Sn–

Zn–Al solder joint strength performs better than others.
KEYWORDS: ball grid array 共BGA兲, lead-free solder, solder joint
strength, bending, flexural load
Introduction
Ball grid array 共BGA兲 solder joint strength under mechanical loads plays a
significant role in printed circuit board 共PCB兲 reliability 关1–3兴. With current
adoption of the lead-free solder, solder joint reliability under dynamic loads is a
key factor for successful product development. For example, heavy heatsinks in
a desktop or server 共500⬃ 1000 g兲 can cause solder joint failure under shock
and vibration load during shipping or handling conditions. Also, drop and im-
pact loads in a laptop or handheld device can cause solder joint failure.
Solder joint strength depends on many factors, such as load and strain
rates 关4,5兴, system dynamic characteristics 关6–8兴, BGA orientation, PCB layout
and test configurations 关8–10兴, joint failure modes 关11,12兴, and solder alloy and
interface properties 关13–15兴.
This work compared BGA lead-free solder joint strengths to the traditional
eutectic lead 共Sn–Pb兲 solder joint strength under monotonic bend load at room
temperature. The design of experiments 共DOE兲 was summarized systematically
into three parts. The first part focused on the effect of solder joint geometries.
The second part investigated the effects of package type, board thickness, and
rework. The third part investigated the effects of alternative solders alloys.
Part I. Effect of Solder Joint Geometries

This section investigates the 0.8 mm flip-chip BGA solder joint strength under
bending load. A four-point bend comparative study of Sn–Pb and Sn–Ag–Cu
共SAC兲 solder joints is performed with a 0.8 mm flip-chip BGA assembled to a
PCB. The DOE included eutectic 63Sn37Pb and Sn4Ag0.5Cu solder joints, dif-
ferent BGA package side solder resist openings 共SROs兲, and different solder ball
diameters. Effects of PCB board side metal defined and solder mask defined
pads were also examined.
Experimental Setup
A four-point bend test was designed 关4,9兴 to evaluate the solder joint strength
under relatively high strain rate load. The supporting span is 203.2 mm 共8 in.兲
and the loading span is 101.6 mm 共4 in.兲, as shown in Fig. 1. The crosshead
loading speed is 50 mm/s. This is a speed achievable with a regular hydraulic
Instron or MTS tester. Higher speed may induce significant dynamic effect and
complication to the test data.
The test board is designed for the four-point bend test, as shown in Fig. 2.
The board size is 345⫻ 101.6 mm2 共13.6⫻ 4 in.2兲. The board thickness is
1.575 mm 共0.062 in.兲. The relatively wider board provides sufficient room for
various BGA size including large BGAs 关4兴. It also minimizes the anticlastic
GENG, doi:10.1520/JAI103044 141
FIG. 1—Four-point bend test setup.
curvature effect, as shown in the finite element analysis 关4兴. The PCB surface
finish is immersion silver 共ImAg兲. The BGA size is 37.5 mm, and the solder joint
pitch is 0.8 mm. The PCB pad size is 254 um 共0.010 in.兲. The solder joint
standoff is at 292⬃ 485 ␮m.
Solder joint continuity of the test boards was monitored during the bend
test. Deflection and load values were extracted at the time of daisy chain
discontinuity—“first electrical failure” on a test board. A detailed test procedure
can be found in Refs 9 and 11.
The DOE evaluated the effect of solder ball materials 共Sn–Pb and SAC兲,
package side SROs 共normalized to pad sizes of 0.9, 1.2, and 1.6兲, and solder ball
Primary Side View
Secondary Side View
FIG. 2—Test board.

FIG. 3—Board flexural loads at first electrical failure-alloy: Sn–Pb versus SAC.
diameters 共457 and 508 ␮m / 0.018 and 0.020 in.兲 on solder joint strength under
mechanical loads. The DOE design has 12 legs, which include two solder alloy
materials, three SROs, and two ball diameters.
Test Results
Figure 3 compares board flexural loads at the first electrical failure between
Sn–Pb and SAC. The Sn–Pb solder joints showed better joint strength than SAC
solder joints when PCB is under the flexural bending load. This is true for
different SROs. Figure 4 compares board flexural loads at the first electrical
failure between two solder ball diameters. The ball diameter effect is less sig-
nificant comparing to solder alloy effect. Figures 5 and 6 compare board flex-
ural loads at the first electrical failure with different SROs. The smaller SRO
has lower solder joint strength.
Failure Analysis
The dye-and-peel test was performed after the bend test. Failure types defined
in Fig. 7 were identified separately for solder joints with metal defined pad and
solder mask defined pad, respectively. Type 1 failure is fracture at component
pad/substrate interface, type 2 is at component/solder ball interface, type 3 is at
solder ball/board pad interface, and type 4 is at pad/PCB interface.
For solder joints with metal defined pads, type 2 and type 4 failures are
dominant, as shown in Fig. 8. Type 3 failure is rarely observed. Sn–Pb solder
joints have almost all type 4 failure, and the failure type is not sensitive to SRO.
SAC solder joints have more type 2 failure, especially with smaller SROs at the
package side.
For solder joints with solder mask defined pads, type 2 and type 3 failures
GENG, doi:10.1520/JAI103044 143
FIG. 4—Board flexural loads at first electrical failure-ball diameter: 457.2 ␮m 共0.018
in.兲 versus 508 ␮m 共0.020 in.兲.
FIG. 5—Board flexural loads at first electrical failure with Sn–Pb alloy-SRO effect.
FIG. 6—Board flexural loads at first electrical failure with SAC alloy-SRO effect.
are dominant, as shown in Fig. 9. Type 4 failure is rarely observed. Sn–Pb

solder joints have more type 3 failure, while type 2 failure occurs with the
smallest SRO. SAC solder joints have more type 2 failure with smaller SRO and
more type 4 failure with larger SRO.
In summary, for metal defined 共MD兲 pads, larger SRO changes SAC failure
type from package side inter-metallic compound 共IMC兲 failure to board side
pad cratering. For solder mask defined 共SMD兲 pads, larger SRO changes SAC
failure type from package side IMC failure to board side IMC failure. The tested
BGA needs to increase SRO for SAC solder joints to ensure sufficient solder
joint strength.
Part II: Effect of Package, Board, and Rework

This section investigates the effects of package type 共plastic BGA 共PBGA兲 versus
ceramic BGA 共CBGA兲兲, board thickness 共0.093 in. versus 0.135 in.兲, and rework
共non-reworked versus reworked兲. Similar four-point bend test is performed for
comparative study. Note that under same test board bend load or deflection,
FIG. 7—Failure mode types from dye-and-peel test.

GENG, doi:10.1520/JAI103044 145
FIG. 8—共a兲 Package side failure 共type 2兲 percentage among solder joints with metal
defined 共MD兲 pads. 共b兲 PCB pad cratering 共type 4兲 percentage among solder joints with
metal defined 共MD兲 pads.
FIG. 9—共a兲 Package side failure 共type 2兲 percentage among solder joints with solder
mask defined 共SMD兲 pads. 共b兲 Board side IMC failure 共type 3兲 percentage among solder
joints with solder mask defined 共SMD兲 pads.
GENG, doi:10.1520/JAI103044 147
different BGA packages and board thicknesses will result in significantly differ-
ent solder joint forces. One should be careful when interpreting flexural test
results in this section for solder join strength comparison.
Experimental Setup
A four-point bend test was designed to evaluate the solder joint strength. The
supporting span is 140 mm, and the loading span is 70 mm, which are smaller
than those in the early section due to smaller board size. The cross-head speed
is 0.1 mm/s. The resistance of the daisy chain circuit through the package,
board, and solder joints is monitored to determine the point at which first
electrical failure occurs. With the displacement-controlled loading, the test was
stopped when the first electrical open was registered. The test configuration is
similar to the setup in the early section.
The test coupon was cut out from large test boards 关12兴 to a size of 163 mm
in length and 56 mm in width. The CBGAs 共U27 and U28兲 and the PBGAs 共U30兲
are shown in Fig. 10. The board surface finish is electrolytic nickel gold 共NiAu兲.
The DOE includes eight legs with two solder alloys 共eutectic 63Sn37Pb and
Sn3.9Ag0.7Cu兲, two PCB thicknesses 共0.135 and 0.093 in.兲, and reworked and
non-reworked solder joints.
Test Results
The bending loads and deflections of each test coupon at the first electrical
failure were summarized. The solder joint failure is electronic discontinuity of
the daisy chain. The PCB thickness effect is shown in Fig. 11. The package
effect between CBGA and PBGA is shown in Fig. 12. The bending loads and
deflections when solder joint failed are shown in Figs. 13 and 14, respectively.
Thickness Effect—Figure 15 illustrates the box plots for load and deflection
at first failure for all DOE legs. Thicker boards have higher loads and lower
deflection to failure. This is expected since thicker boards are stiffer. With dif-
ferent board thicknesses of the test coupons, if the solder joint level fracture
energies or strengths are the same, the board level failure loads or deflections at
first electrical opens are generally different. Therefore, subsequent data analy-
sis is separated by board thickness.
Alloy Effect—Eutectic Sn–Pb and SAC alloys are compared for load to first
electrical failure in the four-point bend test for 0.093 and 0.135 in. boards in
Figs. 16 and 17. Loads are significantly lower for SAC solder than for eutectic
Sn–Pb solder. From the stress level point of view, SAC solder is stiffer than
eutectic Sn–Pb, with the dynamic modulus being in the 50–53 GPa range for
SAC and in the 33–35 GPa range for eutectic Sn–Pb at room temperature 关13兴.
Hence, at a particular applied load level, the solder joint may be subjected to a
higher stress for the SAC BGAs than the eutectic Sn–Pb BGAs. From the solder
joint strength point of view, the embrittleness of the PCB during SAC reflow
process at higher temperature or other mechanism might contribute to lower
SAC failure strength.
FIG. 10—Test coupons 共CBGA: U27 and U28; PBGA: U30兲.
Package Type Effect—Ceramic Ball Grid Array versus Plastic Ball Grid
Array—For SAC BGA packages, there is no difference between these packages
for 0.093 in. board. The CBGA is better 共stronger兲 than the PBGA for 0.135 in.
thick boards. For eutectic Sn–Pb BGA packages, the CBGA is worse 共weaker兲
than the PBGA for 0.093 in. board but better 共stronger兲 for 0.135 in thick board.
The fact that CBGA performs better for 0.135 in. board for both Sn–Pb and SAC
solders indicates that the ceramic package stiffness has effect at test coupon
level load. Therefore, one should be careful when comparing solder joint level
stress and strength. Since the scope of this paper is solder joint strength 共failure
load/stress兲, not solder joint stress level due to package stiffness/thickness ef-
fect, no further analysis is pursued here to include the package effect.
Figures 13 and 14 show that SAC solder joint strength is lower than Sn–Pb
solder joint strength within the same package data.
GENG, doi:10.1520/JAI103044 149
FIG. 11—Test board loads and deflection at first electric open-PCB thickness effect.
Effect of Rework—The results were inconsistent and in some cases counter-

intuitive with regard to the effect of rework on the performance of the BGA
packages under four-point bend loads. For SAC CBGAs, reworked components
were better 共stronger兲 than non-reworked components, but for SAC PBGAs, the
non-reworked components were better 共stronger兲 than reworked components.
For Sn–Pb PBGAs, non-reworked components were in most cases better than
reworked components. The differences between reworked and non-reworked
FIG. 12—Test board loads and deflection at first electric open-package effect.
FIG. 13—Four-point bend loads of test coupons at first electrical open.
Sn–Pb CBGAs are less than 20 %.

Comparing the non-reworked and reworked data, Figs. 13 and 14 show
clearly that SAC solder has lower solder joint strength than Sn–Pb solder.
Failure Analysis
Similar dye-and-peel test was performed as in the early section. Failure modes
are defined in Fig. 7. The only significant result was in the predominant failure
modes observed between SAC components and Sn–Pb components. As shown
in Fig. 18, SAC components failed predominantly in the type 3 failure mode
with the failure interface being between the solder ball and the PCB land. On
the other hand, Sn–Pb BGA components, failure in the type 4 failure mode with
the failure interface being between the PCB land and the laminate.
FIG. 14—Four-point bend deflections of test coupons at first electrical open.

GENG, doi:10.1520/JAI103044 151
FIG. 15—Comparison of board thickness effect on flexural loads and deflections at first
electrical open.
In summary, Sn–Pb solder joint strength is generally higher than SAC sol-
der joint strength with different package types, different board thicknesses, and
with/without rework.
Part III: Effects of Alternative Low Temperature Lead-Free Solders

The third part explores the feasibility of alternative low temperature solders as
board-level interconnects. In addition to the traditional 63Sn-37Pb solder and
lead-free Sn-4Ag-0.5Cu solder, four other lead-free solders 共Sn-52In, Sn-58Bi,
FIG. 16—Comparison of solder alloy effect for 0.093 in. thick boards.
FIG. 17—Comparison of solder alloy effect for 0.135 in. thick boards.
Sn-57Bi-1Ag, and Sn-9Zn-0.006Al兲 were tested. The effects of board surface

finishes 共ImAg vs. organic solderability preservative 共OSP兲兲 and via-in-pad 共VIP
vs. non-VIP pads兲 were investigated.
Experimental Setup
A three-point bend test was performed. The test setup is shown in Fig. 19. The
mid-span load is applied on the back side of the test coupon. The supporting
span is 103.2 mm. The load speed is 50 mm/s.
FIG. 18—Type 3 and type 4 failure modes separated for SAC and Sn–Pb BGA
components.
GENG, doi:10.1520/JAI103044 153
FIG. 19—Three-point bend test setup.
The test coupon size is 60⫻ 115 mm2, and thickness is 1.12 mm 共0.044 in.兲.
It has ten surface mounted BGAs, five with VIP pads and another five with
non-VIP pads, as shown in Fig. 20. Surface finishes of the PCB include ImAg
and OSP with a package side surface finish of electrolytic Ni ImAu. All the
alloys tested are listed in Fig. 21 with peak reflow temperatures and melting
points. For all compositions, a rosin-based no-clean solder paste was used for
board assembly. The tested data reflected the solder joint strength between 1
⬃ 2 weeks after the assembly reflow process.
The DOE includes six solder alloys 共Fig. 21兲, two surface finishes 共ImAg and
OSP兲, and two pad types 共VIP and non-VIP兲.
Test Results
The test result showed that BGAs packages closer to the loading points are
always failed first due to higher bending moment. Therefore, data analysis was
focused on the four packages near the center of the test coupon.
The first failure package 共electrical open兲 was always BGAs with VIP pad
and under maximum bending loads. This is expected due to the less solder joint
cross-section with a via in a pad. The only exception is Sn–In with ImAg surface
finish with the first failure package occurred on the BGA with non-VIP pad.
Figures 22–24 summarize the displacement, load, and fracture energy data.
Sn–In has the lowest solder joint bending strength, while Sn–Bi and Sn–Bi–Ag
show better strength. Sn–Zn–Al shows the highest strength. The trends of Sn–
In, Sn–Bi, and Sn–Bi–Ag are consistent with the ball pull/ball shear tests 关14兴.
With Via in Pad Non- Via in Pad
FIG. 20—Test coupon.

Ball material Peak Reflow Melting Point
Temp
Sn-37Pb 221°C 183 °C
Sn-4Ag-0.5Cu 245°C 217°C
Sn-52In 148°C 118°C
Sn-58Bi 168°C 138°C
Sn-57Bi-1Ag 168°C 139°C
Sn-9Zn-0.006Al 225°C 199°C
FIG. 21—List of solder alloys tested.
25
d (mm)
20
15
10
ImAg OSP OSP ImAg OSP ImAg OSP ImAg OSP ImAg OSP
Sn-Ag-Cu Sn-Bi Sn-Bi-Ag Sn-In Sn-Pb Sn-Zn-Al
Surface Finish within Alloy
FIG. 22—Board deflections at solder joint electrical discontinuity.

GENG, doi:10.1520/JAI103044 155
50
45
force (N)
40
35
30
FIG. 23—Bending load at solder joint electrical discontinuity.
900
800
Fracture Energy (mJ)
700
600
500
400
300
200
100
FIG. 24—Fracture energy at solder joint electrical discontinuity.

Figure 24 implies that Sn–Pb and Sn–Zn–Al do not form as strong a bond
on ImAg as on OSP—less energy to fracture, which correlates pretty well to the
fracture force. Also Sn–Bi has a fairly extended range possibly due to brittle
behavior while Sn–Bi–Ag has a much smaller range, which indicates that Ag
affects the material properties.
Failure Analysis
Dye-and-peel test was performed on the bend test boards with Sn–Pb, SAC, and
Sn–Zn–Al alloys. Failure modes at the solder joints are predominantly PCB
failure with pad cratering and pad lifting 共type 4兲, as defined in Fig. 7.
In summary, Sn–Bi and Sn–Bi–Ag alloys have equivalent flexural solder
joint strength comparing with Sn–Pb and SAC. Sn–Zn–Al alloy has better flex-
ural solder joint strength. Sn–In has worse flexural solder joint strength.
Conclusions
Comparing Sn–Pb and SAC solder joints, SAC solder failure strength is lower
than the Sn–Pb strength under room temperature board flexural load and simi-
lar dynamic load.
Also, Sn–Pb solder joint strength under relatively higher load rate is con-
sistently higher than SAC solder joint strength, with different package type
共PBGA or CBGA兲, different board thicknesses, with or without rework.
Among the alternative lead-free solder alloys, Sn–In is clearly the softest
and weakest solder material among the tested alloys. Sn–Bi and Sn–Bi–Ag sol-
der compositions showed comparable performance to SAC. And Sn–Zn–Al is
clearly the strongest of all the solder materials under flexural load.
This work evaluates solder joint strengths using simple monotonic bend
testing methodologies. The test result is a good indicator of overall solder joint
health for component level evaluation. In most applications, shock and vibra-
tion events can produce cyclic response and fatigue strength can be significant
lower. The component solder joint strength depends on the specific board or
system dynamic behavior. A detailed dynamic testing and modeling methodol-
ogy can be found in Refs 6, 8, and 10. By matching a four-point bend test board
dynamic behavior to a computer motherboard, solder joint strength was ex-
tracted. The fatigue effect from a specific product can therefore be accounted
for solder joint evaluation.
Acknowledgments
The author would like to thank Noman Armendariz, Raiyo Aspandiar, Tiffany
Byrne, Jerry Gleason, Alan McAllister, Mitul Modi, and Arnaldo Nazario for
their contributions to some parts of this research. Thanks are due to Chris H.
Hanes and Jim D. Williams for their support in the completion of this manu-
script.
References
关1兴 Geng, P. and Beltman, W. M., “Monitoring Motherboard Shock Response near
GENG, doi:10.1520/JAI103044 157
BGA Solder Joints,” Proc. SMTA International Conf., Chicago, IL, September 2002,
SMTA, Minnesota.
关2兴 Pitarresi, J., Geng, P., Beltman, W. M., and Ling, Y., “Dynamic Modeling and Mea-
surement of Personal Computer Motherboard,” Proc. 52th Electronic Components
and Technology Conf., San Diego, CA, May 2002, IEEE, New Jersey.
关3兴 Pitarresi, J., Roggeman, B., Chaparada, S., and Geng, P., “Mechanical Shock Test-
ing and Modeling of PC Motherboards,” Proc. 54th Electronic Components and
Technology Conf., Las Vegas, June 2004, IEEE, New Jersey.
关4兴 Geng, P., Chen, P. H., and Ling, Y., “Effect of Strain Rate on Solder Joint Failure
Under Mechanical Load,” Proc. 52nd Electronic Components and Technology Conf.,
San Diego, CA, May 2002, IEEE, New Jersey, pp. 974–978.
关5兴 Harada, K., Baba, S., Wu, Q., Matsushima, H., Matsunaga, T., Uegai, Y., and
Kimura, M., “Analysis of Solder Joint Fracture Under Mechanical Bending Test,”
Proc. 53th Electronic Components and Technology Conf., New Orleans, LA, May
2003, IEEE, New Jersey.
关6兴 Geng, P., “Solder Joint Shock Testing and Modeling Methodology Development,”
IPC Annual Conference, IPC 6-10d Committee Meeting, Minneapolis, MN, Septem-
ber 2003, IPC, Illinois.
关7兴 Geng, P., Beltman, W. M., Chen, P. H., Daskalakis, G., Shia, D., and Williams, M.
H., “Modal Analysis for BGA Shock Test Board and Fixture Design,” Proc. Fifth
EPTC Conference, Singapore, December 2003, IEEE, New Jersey.
关8兴 Geng, P. and Maguire, J. F., “Dynamic Testing and Modeling for Solder Joint Reli-
ability Evaluation,” Proc. of IPC Technical Conference, Anaheim, CA, February
2004, IPC, Illinois.
关9兴 Geng, P., Modi, M., McCormick, C., McAllister, A., Nazario, A., and Williams, R.,
“A Comparative Study of BGA Solder Joint Reliability Under Four-Point Bend and
Spherical Bend Tests,” Proc. IMAPS International Conf. Electronic Packaging,
Scottsdale, AZ, March 2005, IMAPS, Washington, D.C..
关10兴 Geng, P., “Dynamic Test and Modeling Methodology for BGA Solder Joint Shock
Reliability Evaluation,” Proc. 55th Electronic Components and Technology Conf.,
Lake Buena Vista, FL, June 2005, IEEE, New Jersey.
关11兴 Geng, P., McAllister, A., McCormick, C., Modi, M., and Nazario, A., “0.8 mm BGA
Solder Joint Strength Under Flexural Load,” Proc. SMTA International’04, Chicago,
IL, September 2004, SMTA, Minnesota.
关12兴 Aspandiar, R., Geng, P., and Armendariz, N., “Lead-Free Solder Reliability Under
Flexural Load,” iNEMI Advanced Pb-Free Assembly and Rework Development
Project Report, iNEMI, Virginia, 2005.
关13兴 NIST Database for Solder Properties with Emphasis on New Lead-Free Solders,
Release 4.0, NIST, 2002, http://www.boulder.nist.gov/div853/lead_free/solders.html
共Last accessed June 7, 2010兲.
关14兴 Geng, P., Aspandiar, R., Byrne, T., Pon, F., Suh, D., McAllister, A., Nazario, A.,
Paulraj, P., Armendariz, N., Martin, T., and Worley, T., “Alternative Lead-Free Sol-
der Joint Integrity Under Room Temperature Mechanical Loads,” Proc. Ninth
Intersociety Conference on Thermal and Thermal Mechanical Phenomena in Elec-
tronic Systems 共ITHERM’04兲, Las Vegas, June 2005, ASME, New York.
关15兴 Kim, Y. M., Oh, C.-Y., Roh, H.-R., and Kim, Y.-H., “A New Cu–Zn Solder Wetting
Layer for Improved Impact Reliability,” Proc. 59th Electronic Components and
Technology Conf., San Diego, CA, June 2009, IEEE, New Jersey.
doi:10.1520/JAI103041
Qiulian Zeng,1 Jianjun Guo,2 Xiaolong Gu,1 Qingsheng Zhu,3

and Xiaogang Liu1
Tensile Properties of Sn-10Sb-5Cu High

Temperature Lead Free Solder
ABSTRACT: The Sn-10Sb-5Cu high temperature lead free solder was devel-
oped, and the mechanical property of such bulk solder and the solder joint
was investigated in the present work. The microstructure of the Sn-10Sb-
5Cu bulk solder was composed of long strip-like Cu6Sn5 and square Sn3Sb2
intermetallic compounds 共IMCs兲. In the solder joint, the IMCs were the same
as in the bulk solder but with a much finer microstructure. The test results
showed that the tensile properties of the Sn-10Sb-5Cu bulk solder were
sensitive to the strain rate. The higher the strain rate, the higher the ultimate
strength. The ductility after fracture was enhanced by lowering strain rates.
The fracture elongation approached 6.5 % when the strain rate was 10⫺5/s.
Compared with that of the bulk Sn-10Sb-5Cu solder, the tensile strength of
the Sn-10Sb-5Cu/Cu solder joint was much lower. The joint showed excel-
lent plasticity with a large nominal engineering strain of 80 % during tensile
tests. The tensile strength of the solder joint decreased after some aging
time. However, the strength remained at about 80 % of the original value
after aging for 15 days at the temperature of 150°C. The difference in the
tensile properties between bulk solder and solder joint resulted from the
difference in microstructures. Coarser Cu6Sn5 IMC in the bulk solder led to
lished online May 2010.
1
Zhejiang Province Key Laboratory of Soldering and Brazing Materials and Technology,
Zhejiang Metallurgical Research Institute, Hangzhou 310011, China.
2
Ph.D., Zhejiang Province Key Laboratory of Soldering and Brazing Materials and Tech-
nology, Zhejiang Metallurgical Research Institute, Hangzhou 310011, China 共Corre-
sponding author兲, E-mail: jjguomail@163.com
3
Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China.
Cite as: Zeng, Q., Guo, J., Gu, X., Zhu, Q. and Liu, X., ‘‘Tensile Properties of Sn-10Sb-
5Cu High Temperature Lead Free Solder,’’ J. ASTM Intl., Vol. 7, No. 5. doi:10.1520/
JAI103041.
158
ZENG ET AL., doi:10.1520/JAI103041 159
the higher ultimate strength, and the finer Cu6Sn5 IMC in the solder joint
resulted in much better fracture ductility of the solder joint.
KEYWORDS: tensile properties, high temperature lead free solder,
solder joint, aging
Introduction
It has been of increasing interest to develop high temperature lead free solders
in chip interconnection packaging, although high temperature Pb-based sol-
ders are exempt from the restriction of hazardous substance directive. Sn–Sb
alloys are considered of great potential 关1兴. Most works that have been done in
this field were mainly focused on the creep property of the Sn-5Sb bulk solder
关2–5兴. It has been found that the Sn-5Sb alloy 关6–8兴 exhibits better creep prop-
erties over eutectic Sn–Pb solder 关9兴.
However, the melting point of the Sn-5Sb alloy is still not enough for the
reliability in the post-processing of electronic packaging of the high tempera-
ture application. It is easily deduced that the Sn-10Sb alloy should perform
better than the Sn-5Sb alloy because of its higher melting temperature 关10,11兴.
Cu is one of the most widely used substrates in electronic packaging. How-
ever, it was reported that the reaction rate of the Sn–Sb alloy with Cu was very
high 关12兴, which means that the Sn-10Sb binary alloy could dissolve Cu sub-
strate rapidly.
In order to avoid the fast dissolution of Cu substrate into the Sn-10Sb
solder, we developed a new Sn-10Sb-5Cu ternary alloy. Instead of decreasing
the melting point of the Sn-10Sb alloy by adding, for example, Ag and Au 关13兴,
the addition of Cu increases the liquidus temperature of the Sn-10Sb alloy
according to the Sn-10Sb-Cu ternary equilibrium phase diagram 关14兴, which is
an advantage for the reliability of solder joint in terms of melting temperature.
In the present work, tensile properties of the Sn-10Sb-5Cu bulk solder and
solder joint were studied. The tensile properties of the Sn-10Sb-5Cu bulk solder
showed strain rate sensitivity. And the effect of aging time on the strength of the
solder joint was also investigated. The bulk solder and the solder joint showed
different properties, and the various microstructures were considered to ex-
plain such differences.
Experimental Procedures
Preparation of the Sn-10Sb-5Cu Bulk Solder and Its Tensile Property

The high purity Sn 共99.99 wt %兲 and Sb 共99.95 wt %兲 were formulated and
melted in the furnace at a temperature of 800° C; then Cu element in the form
of Sn-20 wt %Cu inter-alloy was added into the binary Sn–Sb alloy. After 1/2 h,
the melt was cast into a steel mold to form a Sn-10Sb-5Cu ternary alloy cylin-
drical ingot. The composition of the Sn-10Sn-5Cu solder ingot was analyzed by
atomic absorption spectroscopy and listed in Table 1.
The standard tensile test samples were prepared by electric discharge ma-
chining into a dog-bone type specimen with a gauge length of 50 and diameter
TABLE 1—Composition of the Sn-10Sb-5Cu solder 共wt %兲.
Element
Sb Cu Pb Zn Fe Ag Cd Sn
Mass percent 9.96 5.05 0.015 0.0003 0.0016 0.0009 0.0001 Bal.
of 8 mm. Each specimen was annealed at 100° C for 1 h before the test. Tensile
tests were carried out on the Mechanical Testing & Simulation System Corpo-
ration universal testing machine with a load range of 10 tons. And the strain
rates varied from 10−2 / s to 10−5 / s under the strain control mode.
Tensile Test of the Sn-10Sb-5Cu/Cu Solder Joint

The Sn-10Sb-5Cu solder was compressed to thin sheet, placed between two
polished Cu blocks with a size of 10⫻ 10⫻ 10 mm3 to be a sandwich structure.
Some flux was coated on the two soldering sides of Cu blocks. Then the sand-
wich structure as a whole was placed on hot plate for soldering. The thickness
of the solder was controlled to be 0.5 mm. The cooled sandwich sample was cut
by electric discharge machining to be some small solder joint specimens with a
cross-section area of 0.8⫻ 0.8 mm2, shown in Fig. 1. Some small solder joint
samples were aged at 150° C, the holding time in the range of 2 ⬃ 17 days. The
solder joints surfaces were grounded and polished before the tensile property
test. And the tensile test was performed on a micro tensile tester with a mini-
mum displacement of 0.1 ␮m and a minimum load of 10 mN. The displace-
ment rate was controlled at 0.2 mm/min 共i.e., with a strain rate of 6.67
⫻ 10−3 / s兲. The load and displacement curves data were recorded automatically
by computer. Each case was repeated three times, and the sample size shown in
Table 2.
Tensile Properties of the Sn-10Sb-5Cu Bulk Solder

As shown in Fig. 2共a兲, the microstructure of the bulk Sn-10Sb-5Cu solder was
composed of grayer ␤-Sn substrate, black square or triangle Sn3Sb2 phase, and
FIG. 1—Schematic diagram of the tensile test sandwich specimen.

ZENG ET AL., doi:10.1520/JAI103041 161
TABLE 2—Dimensions of the solder joint samples.
Cross-Section Size Thickness of the

Number of Sample of the Sample 共mm2兲 Solder 共mm兲
0 0.76⫻ 0.75 0.5
2 days 0.73⫻ 0.77 0.5
5 days 0.75⫻ 0.82 0.5
10 days 0.76⫻ 0.75 0.5
15 days 0.75⫻ 0.74 0.5
dark strip-like Cu6Sn5 intermetallic compound 共IMC兲 with a maximum length

of 100 ␮m. The size of Sn3Sb2 phase was about 20⬃ 30 ␮m. Figure 2共b兲 is a
scanning electron microscope 共SEM兲 photograph of the Sn-10Sb-5Cu/Cu solder
joint. Different from the bulk solder, the aspect ratio of Cu6Sn5 phase in the
solder joint was much smaller. And the coarse SnSb IMC in the solder joint
microstructure was much less than in the bulk solder. This meant that most of
the SnSb phase in the solder joint was precipitated in the ␤-Sn substrate as fine
precipitates. The difference between the bulk solder and solder joint resulted
from the different cooling conditions during the solidification process. The
cooling rate of the bulk solder in the steel module was about 2 ⬃ 4 ° C / s, while
the cooling rate of the solder joint was about 2 ° C / s. That is, the bulk solder
had a rapider cooling rate but a coarser microstructure. However, the rapider
cooling rate will generally lead to finer microstructure. So only the cooling rate
cannot explain the difference in the microstructure of the bulk solder and sol-
der joint. It must have some other factors affecting the microstructure. It was
deduced that the larger temperature gradient should account for the longer
strip-like Cu6Sn5 IMC in the bulk solder.
Figure 3 shows the tensile curves of the bulk solder. Figure 3共a兲 and 3共b兲
corresponds to load-displacement curves and stress-strain curves, respectively.
It was observed that the tensile properties were sensitive to the strain rate,
including ultimate strength 共UTS兲, elastic modulus, and elongation after frac-
ture.
FIG. 2—Microstructure of the Sn-10Sb-5Cu bulk solder and solder joint: 共a兲 SEM pho-
tograph of the bulk solder; 共b兲 SEM photograph of the solder joint.
FIG. 3—Tensile curves of the Sn-10Sb-5Cu bulk solder: 共a兲 Load-displacement; 共b兲
stress-strain curves.
At the higher strain rates, the maximum load of the sample was increased,
shown in Fig. 3共a兲. And the sample fracture occurred at a smaller strain with
increasing strain rates and showed greater brittle fracture characteristics than
that at the lower strain rates, which might be related to the interaction between
the dislocations and IMCs. The IMCs became the obstacles of dislocation slip,
restricting plastic deformation so that a large quantity of dislocations accumu-
lates around the IMCs. Dislocation accumulation enhanced the local internal
stress, and finally the cracks initiated near the IMCs. At low strain rate, the
displacement became large when the sample fractured.
As shown in Fig. 3共b兲, the higher the strain rate, the larger the UTS. The
UTS ranged from 76.9 to 105.1 MPa in the strain rates ranging from 10−2 / s to
10−5 / s, listed in Table 3. The UTS sensitivity to the strain rate is very similar to
that of pure Sn or Sn-based solder 关15兴, although the Sn-10Sb-5Cu high tem-
perature solder contains many alloying additions. The UTS sensitivity meant
severer dislocation accumulation, leading to much more work hardening at the
higher strain rate. In the case of the lower strain rate, because more dynamic
recovery was occurring, the work hardening process was weakened.
Some other tensile properties parameters are also listed in Table 3. The
UTS was the most sensitive item to the strain rates. And the elongation after
rupture decreased with the increasing strain rate. The elongation at the lower
strain rate of 10−5 / s was larger than 6 % and about 3 % at the higher strain
rate.
TABLE 3—Parameters of tensile properties of the Sn-10Sb-5Cu bulk solder.
Strain Rate/s
10−2 / s 10−3 / s 10−4 / s 10−5 / s

Elastic modulus, GPa 42.33 42.7 36.03 38.13
Elongation after fracture, % 3.68 2.76 3.86 6.48
UTS, MPa 105.1 94.4 85.7 76.9
ZENG ET AL., doi:10.1520/JAI103041 163
FIG. 4—Microstructure of the solder joints after aging at 150°C: 共a兲 Original micro-
structure; 共b兲 3 days; 共c兲 10 days; and 共d兲 17 days.
Effect of Aging Time on Microstructure of the Sn-10Sb-5Cu/Cu Solder Joint

Figure 4 is the microstructure of the Sn-10Sb-5Cu/Cu solder joint after aging at
150° C for various times. Figure 4共a兲–4共d兲 corresponds to the case of the origi-
nal, 3, 10, and 17 days, respectively. From the as-soldered joint microstructure,
the morphologies of the black Cu6Sn5 IMC included some strip-like and lump-
like phase; the longest dimension of the Cu6Sn5 IMC was about 20 ␮m, with a
much smaller size compared with that in the bulk solder. With the increasing
aging time, the strip-like Cu6Sn5 phase turned round and coarser. And the
scallop Cu6Sn5 IMC layer at the interface between the solder and Cu substrate
was thickened and flattened with the increasing aging time.
Tensile Properties of the Sn-10Sb-5Cu/Cu Solder Joint

Figure 5 shows the effect of aging time on the tensile curves of the solder joint.
Aging time ranged from 2 to 15 days. It was found that the maximum stress
共UTS兲 of the solder joint after aging decreased steadily with aging compared
with that of the original as-soldering sample. This result was similar to those in
the systems of the Sn-0.7Cu/Cu 关16兴, Sn-Pb/Au 关17兴, Pb/Sn, and Au/Sn 关18兴
joints after aging, which were related with some defects such as coarsening
microstructure 关19兴, vacancy formation 关20兴, and Kirkendall voids 关21兴.
Compared with the bulk solder, the UTS of the solder joint was much lower.
FIG. 5—Effect of aging time on the tensile curves of the solder joint.
The strength of the as-soldered joint was only 50 MPa, while for the bulk solder
the minimum UTS was 76.9 MPa at the strain rate of 10−5 / s, and the maximum
UTS was 105.1 MPa at the strain rate of 10−2 / s. Such a difference was ex-
plained by the different microstructures. The strengthening effect of the IMCs
in bulk solder was much stronger than that in the solder joint. As mentioned
above, the dimension and the quantity of the IMC in the solder joint were
smaller, which resulted in lower strength. More importantly, the measured final
displacement in each case was larger than 0.4 mm, i.e., comparable to the
engineering strain higher than 80 % 共0.4 /0.5 mm, considering of the length of
the solder section of about 0.5 mm兲. This result suggests that the strength and
ductility of solder joint with such a composition were suitable in the real ser-
vice application.
In addition, the decrease of the strength of the solder joint after aging was
not severe. The strength was still at the 75–86 % of the maximum value of the
original solder joint, which will also be an advantage for the long-term reliabil-
ity of the solder joint.
Figure 6 is the fracture morphology of the solder joint and the bulk solder.
From the whole view of the fractured joint sample in Fig. 6共a兲, it was seen that
the rupture occurred in the solder section and severe necking took place, which
indicate good plasticity of the solder joint. The top view of the fractography of
the solder joint is shown in Fig. 6共b兲; besides some dimples on the fracture,
some particle-like IMCs were observed. The composition of these particles was
identified by energy-dispersive X-ray 共EDX兲 to be Cu6Sn5, as marked by num-
bers 1 and 2; a small amount of Sb was detected in such Cu6Sn5 phase, as
marked by the result of +1 marker, shown in Fig. 6共d兲.
ZENG ET AL., doi:10.1520/JAI103041 165
FIG. 6—Fracture observation of the bulk solder and solder joint: 共a兲 Macrostructure of
the ruptured solder joint; 共b兲 fractography of the solder joint; 共c兲 fractography of the bulk
solder; and 共d兲 EDX analysis of the IMC marked in 共b兲.
Similar to what was observed in the solder joint, a great deal of Cu6Sn5
phase was exposed on the fracture surface of the bulk solder. However, the
morphology of the Cu6Sn5 phase in the bulk solder at the fracture was very
long, as marked by the white arrows. And the dimple structure of the bulk
solder was much less than that of the solder joint fracture. This indicated that
the ductility of the solder joint with a finer microstructure was better than that
of the bulk solder with a coarser microstructure. This result agrees with the
relationship between the tensile properties and microstructure discussed
above, in which the finer microstructure results in the better plasticity but a
lower UTS in the solder joint.
Conclusion
共1兲 The tensile properties of the Sn-10Sb-5Cu bulk solder were sensitive to
the strain rate. The higher the strain rate, the larger the UTS. The
elongation after fracture increased with decreasing strain rate. The
elongation of the bulk solder reached 6.5 % at the strain rate of 10−5 / s.
共2兲 The tensile strength of the Sn-10Sb-5Cu/Cu solder joint was lower than
that of the Sn-10Sb-5Cu bulk solder. The engineering strain of the sol-
der joint after fracture approaching to 80 %.
共3兲 Although the strength of the Sn-10Sb-5Cu/Cu solder joint decreased
with the aging time, the UTS of the solder joint after aging for 15 days
at 150° C kept at about 80 % of the original sample.
Acknowledgments
The writers would like to acknowledge financial support provided by the Sci-
ence and Technology Program of Zhejiang Province 共Grant No. 2008F1024兲.
References
关1兴 Jang, J. W., Kim, P. G., Tu, K. N., and Lee, M., “High-Temperature Lead-Free SnSb
Solders: Wetting Reactions on Cu Foils and Phased-In Cu–Cr Thin Films,” J. Mater.
Res., Vol. 14共10兲, 1999, pp. 3895–3900.
关2兴 El-Bahay, M. M., El Mossalamy, M. E., Mahdy, M., and Bahgat, A. A., “Study of the
Mechanical and Thermal Properties of Sn-5wt% Sb Solder Alloy at Two Annealing
Temperatures,” Phys. Status Solidi A, Vol. 198共1兲, 2003, pp. 76–90.
关3兴 Murty, K. L., Mathew, M. D., and Haggag, F. M., “An Investigation of the Defor-
mation Mechanisms in Sn5%Sb Alloy Using Tensile, Creep and ABI Tests from
Ambient to 473K,” Met. Mater. Int., Vol. 4共4兲, 1998, pp. 799–802.
关4兴 Beshai, M. H. N., Habib, S. K., Yassein, A. M., Saad, G., and Hasab El-Naby, M. M.,
“Effect of SnSb Particle Size on Creep Behaviour Under Power Law Regime of
Sn-10%Sb Alloy,” Cryst. Res. Technol., Vol. 34共1兲, 1999, pp. 119–126.
关5兴 Yassin, A., Reuben, R. L., Saad, G., Beshai, M. H. N., and Habib, S. K., “Effect of
Annealing and Microstructure on the Creep Behaviour of an Sn-10 wt% Sb Alloy,”
Proc. Inst. Mech. Eng., Part L, Vol. 213共L1兲, 1999, pp. 59–68.
关6兴 Murty, K. L., Haggag, F. M., and Mahidhara, R. K., “Tensile, Creep, and ABI Tests
on Sn5%Sb Solder for Mechanical Property Evaluation,” J. Electron. Mater., Vol.
26共7兲, 1997, pp. 839–846.
关7兴 Geranmayeh, A. R. and Mahmudi, R., “Room-Temperature Indentation Creep of
Lead-Free Sn-5%Sb Solder Alloy,” J. Electron. Mater., Vol. 34共7兲, 2005, pp. 1002–
1009.
关8兴 Mahidhara, R. K., Sastry, S. M. L., Turlik, I., and Murty, K. L., “Deformation and
Fracture Behavior of Sn-5Sb Solder,” Scr. Metall. Mater., Vol. 31共9兲, 1994, pp.
1145–1150.
关9兴 Mathew, M. D., Yang, H., Movva, S., and Murty, K. L., “Creep Deformation Char-
acteristics of Tin and Tin-Based Electronic Solder Alloys,” Metall. Mater. Trans. A,
Vol. 36共1兲, 2005, pp. 99–105.
关10兴 Chen, S. W., Chen, C. C., Gierlotka, W., Zi, A. R., Chen, P. Y., and Wu, H. J., “Phase
Equilibria of the Sn–Sb Binary System,” J. Electron. Mater., Vol. 37共7兲, 2008, pp.
992–1002.
关11兴 Zeng, Q. L., Gu, X. L., Zhao, X. B., Chen, C. Z., and Liu, X. G., “Progress of
Lead-Free Solder Replacement for Pb-Rich Solder,” Electron. Compon. Mater., Vol.
27共8兲, 2008, pp. 16–21 共in Chinese兲.
关12兴 Lee, C., Lin, C. Y., and Yen, Y. W., “The 260°C Phase Equilibria of the Sn–Sb–Cu
Ternary System and Interfacial Reactions at the Sn–Sb/Cu Joints,” Intermetallics,
Vol. 15共8兲, 2007, pp. 1027–1037.
关13兴 El-Daly, A. A., Swilem, Y., and Hammad, A. E., “Creep Properties of Sn–Sb Based
Lead-Free Solder Alloys,” J. Alloys Compd., Vol. 471共1-2兲, 2009, pp. 98–104.
关14兴 Villars, P., Prince, A., and Okamoto, H., Handbook of Ternary Alloy Phase Diagrams,
2nd printing, ASM International, The Materials Information Society, Materials
Park, OH, Vol. 8, 1997, p. 10005.
关15兴 Shohji, I., Yoshida, T., Takahashi, T., and Hioki, S., “Tensile Properties of Sn–Ag
ZENG ET AL., doi:10.1520/JAI103041 167
Based Lead-Free Solders and Strain Rate Sensitivity,” Mater. Sci. Eng., A, Vol. 366,
2004, pp. 50–55.
关16兴 Bae, K. S. and Kim, S. J., “Microstructure and Adhesion Properties of Sn-0.7Cu/Cu
Solder Joints,” J. Mater. Res., Vol. 17共4兲, 2002, pp. 743–746.
关17兴 Kim, K. S., Yu, C. H., and Yang, J. M., “Aging Treatment Characteristics of Solder
Bump Joint for High Reliability Optical Module,” Thin Solid Films, Vol. 462–463,
2004, pp. 402–407.
关18兴 Sheen, M. T., Chang, C. M., Teng, H. C., Kuang, J. H., Hsien, K. C., and Cheng, W.
H., “The Influence of Thermal Aging on Joint Strength and Fracture Surface of
Pb/Sn and Au/Sn Solders in Laser Diode Packages,” J. Electron. Mater., Vol. 31共8兲,
2002, pp. 895–902.
关19兴 Kim, D. G., Kim, J. W., and Jung, S. B., “Effect of Aging Conditions on Interfacial
Reaction and Mechanical Joint Strength Between Sn-3.0Ag-0.5Cu Solder and Ni-P
UBM,” Mater. Sci. Eng., B, Vol. 121共3兲, 2005, pp. 204–210.
关20兴 Zhang, L., Wang, Z. G., and Shang, J. K., “Current-Induced Weakening of
Sn3.5Ag0.7Cu Pb-Free Solder Joints,” Scr. Mater., Vol. 56共5兲, 2007, pp. 381–384.
关21兴 Kim, K. S., Kim, N. K., Yu, C. H., Kim, J. J., and Chang, E. G., “Microstructure and
Strength of Bump Joints in Photodiode Packages,” J. Electron. Mater., Vol. 33共1兲,
2004, pp. 70–75.
doi:10.1520/JAI103043
S. Mallik,1 N. N. Ekere,2 and R. Bhatti2
Empirical Modeling of the Time-Dependent

Structural Build-up of Lead-Free
Solder Pastes Used in the Electronics
Assembly Applications
ABSTRACT: Solder paste is the primary bonding material used in the assem-
bly of surface mount devices in electronics industries. It generally has a
flocculated structure, which may break-down on shearing and slowly rebuild
at rest. The proper characterization of the time-dependent rheological behav-
iors of solder pastes is crucial for establishing the relationships between the
pastes’ structure and flow behavior and for correlating the physical param-
eters with paste printing performance. In this paper, we present a novel
method that has been developed for characterizing the time-dependent and
non-Newtonian rheological behavior of solder pastes and flux mediums as a
function of shear rates. The objective of the study reported in this paper was
to investigate the thixotropic build-up behavior of solder paste and flux me-
diums. The stretched exponential model has been used to model the struc-
tural changes during the build-up process and to correlate model parameters
with the paste printing process. As expected, for solder paste samples, the
rate of structural recovery was found dependent on the applied shear rate.
The model parameters, such as equilibrium viscosity and characteristic time,
Manuscript received February 17, 2010; accepted for publication July 31, 2010; pub-
lished online September 2010.
1
Electronic Manufacturing Engineering Research Group, Medway School of Engineer-
ing, Univ. of Greenwich at Medway, Chatham Maritime, Chatham, Kent ME4 4TB,
United Kingdom, e-mail: s.mallik@gre.ac.uk
2
Electronic Manufacturing Engineering Research Group, Medway School of Engineer-
ing, Univ. of Greenwich at Medway, Chatham Maritime, Chatham, Kent ME4 4TB,
United Kingdom.
Cite as: Mallik, S., Ekere, N. N. and Bhatti, R., ‘‘Empirical Modeling of the Time-
Dependent Structural Build-up of Lead-Free Solder Pastes Used in the Electronics
Assembly Applications,’’ J. ASTM Intl., Vol. 7, No. 9. doi:10.1520/JAI103043.
168
MALLIK ET AL., doi:10.1520/JAI103043 169
have been correlated with the shear-thinning and slumping behaviors of sol-
der paste during the stencil printing process.
KEYWORDS: solder pastes, lead-free, stretched exponential model,
flux, rheology, time-dependent behaviors
Introduction
Fluids with internal structure can demonstrate time-dependent behavior upon

the application of shear, which can also continue after the shearing has
stopped. The most common time-dependent behavior in which the fluid viscos-
ity decreases with the time of shearing and in which the viscosity gradually
recovers when shear is removed is known as thixotropy, and such fluids are
called thixotropic fluids. From the structural point of view, thixotropic behavior
takes place when the fluid microstructure changes, under the application of
shear, from one state to another in a reversible way. As suggested by Barnes 关1兴,
this microstructural change in the fluid is mainly the result of the competing
action between break-down due to flow-stresses and build-up due to in-flow
collisions and Brownian motion.
Solder pastes generally have a flocculated structure 共shown aggregation of
solder particles兲 and hence are known to exhibit thixotropic behavior. Difficul-
ties arise in mixing and handling of solder paste materials because thixotropic
structures progressively break-down on shearing and slowly rebuild at rest. For
thixotropic materials, the time-scales for structural breakdown and build-up
can range from a couple of minutes 共for structural breakdown兲 to several hours
共for the rebuilding of the material structure兲关1兴.
For solder pastes, thixotropy is designed into the formulation to meet spe-
cific processing requirements, for example, to make the pastes shear-thinning
under shear application 共e.g., squeegee action during stencil/screen printing兲
and to facilitate structural build-up after the application of shear. However,
thixotropy is not always desirable, especially in the case of mixing, handling,
and experimentation—hence the need for a balance in terms of the rate and
extent of thixotropy, as this can also adversely impact other sub-processes in
the stencil printing process. Because of the presence of thixotropic structure,
solder paste behavior is highly influenced by the previous shear history. It is the
thixotropy, which is the main cause of the “batch-to-batch variation” in solder
paste manufacturing. The poor reproducibility of experimental results of solder
paste is also caused by the thixotropic nature of solder paste.
The objective of the study reported in this paper is to investigate the thixo-
tropic build-up behavior of solder paste and flux mediums. The stretched expo-
nential model 共SEM兲 has been used to model the structural changes during the
build-up process and to correlate model parameters with the paste printing
process. The paper is divided into three main parts. The first part focuses on the
theoretical aspect of the SEM. The second part presents the details of materials
and methods used in this study, and the final part outlines the results from the
investigation.
Build-up of Thixotropic Fluids: Stretched Exponential Model

A review of mathematical models for time-dependent behavior of thixotropic
fluids was reported by Barnes in 1997 关1兴. One of the models outlined in this
review is the SEM 共see Eq 1兲, which was suggested as a suitable model for
predicting the structural build-up in suspensions. Although the SEM has been
recommended to model the time-dependent behavior of suspensions, there are
however no reports of using the model to study the time-dependent flow behav-
ior of solder pastes and flux mediums.
r
␩ = ␩e,0 + 共␩e,⬁ − ␩e,0兲共1 − e−共t/␶兲兲共1兲
In this equation, ␩e,0 is the viscosity at the commencement of shearing, ␩e,⬁ is

the viscosity after shearing for an infinite time, ␶ is a time constant, and r is a
dimensionless parameter with values between 0 and 1 and is referred to as the
stretched parameter. Equation 1 can be used to model both the build-up and
break-down in step-up or step-down tests, with the values of ␶ and r depending
on both the level and the direction 关1兴.
In another study, Heymann et al. 关2兴 used an equation similar to Eq 1 to
investigate the build-up of newsprint inks. Instead of using viscosity, they used
a yield stress ␴y to describe the rebuilding of printing inks
␴y共t兲 = ␴0y + 关␴⬁y − ␴y0兴共1 − e−共t/␶兲兲共2兲
The form of the SEM used in this work is a simplified version of Eq 1 with r
= 1. This has also been used by Maingonnat et al. 关3兴 to describe the build-up
phenomenon of colloidal clay suspensions and can be rewritten in the follow-
ing way:
␩共t兲 = ␩0 + 共␩⬁ − ␩0兲共1 − e−共t/␺兲兲共3兲
where:
␺ = characteristic time.
The term ␩⬁ is the viscosity value at equilibrium, and ␩0 is the viscosity
when the structure is completely broken down. The three parameters 共␺ , ␩0 , ␩⬁兲
are calculated as fitting parameters of the model described by Eq 3. In the work
reported in this paper, the MATLAB software 共version 7兲 system was used to
perform the regression analysis and model fitting.
As stated earlier, the SEM may be used to model both the build-up and
break-down behaviors of suspensions. It assumes that the structural changes
共such as structural build-up and break-down兲 are time-dependent phenom-
enon. The limitation of the model is that it is only valid under constant shear
rate condition. In this study, the SEM is used to model the build-up behaviors
of lead-free solder pastes and flux mediums. The model 共as expressed in Eq 3兲
may be used to model time-dependent structural build-up and break-down of
other types of solder pastes such as lead-based solder pastes and solar pastes.
However, these are not under the scope of this study.
TABLE 1—Test materials.
Particle Size
Paste Sample Flux Type 共␮m兲
P1 F1 25–45
P2 F1 20–38
P3 F2 25–45
P4 F2 20–38
Materials and Methods
Test Materials
Four lead-free solder pastes, P1–P4, prepared from two different fluxes 共F1 and
F2兲 are investigated. The solder particles for all the paste samples are made of
the same tin-silver-copper alloy 共95.5 Sn, 3.8 Ag, and 0.7 Cu兲 with a melting
point of 217° C. All the solder paste samples had the same metal content of 88.5
% by weight. The details of these samples are provided in Table 1.
Flux is a complex system as the composition of a typical flux system could
have some 5–20 ingredients. The main ingredient of flux is naturally occurring
rosin. The other commonly used ingredients are solvents, activators, antioxi-
dants, surfactants, rheological additives, and thixotropic agents. According to
the manufacturer, both F1 and F2 fluxes are classified as water-based, rosin-
containing, no-clean, and halide free. Although the compositions of these flux
systems are not known 共because of the proprietary nature of the information兲,
solder pastes with different flux systems are expected to show different defor-
mation behaviors and flow characteristics.
Experimental Methodology
Experimental Setup: Rheological Measurements—The rheological measure-

ments were carried out using a Bohlin Gemini-150 controlled-stress/strain rhe-
ometer. A roughened or serrated parallel plate geometry 共with serrations on
both upper and lower plates兲 of 20 mm upper plate diameter and 40 mm lower
plate diameter was used in order to minimize the effect of wall-slip. The rough-
ness values 共Ra兲 for the upper and lower plate were 13.7 and 17.6 ␮m, respec-
tively. Prior to loading a sample onto the rheometer, the solder paste samples
were stirred or hand mixed with a plastic spatula for about 30 s. The sample is
loaded on the bottom plate, and the top plate is then lowered to the desired gap
height of 500 ␮m by squeezing the extra paste out from between the plates.
The excess paste at the plate edges is trimmed off neatly with a plastic spatula.
Then the sample is allowed to rest for about 1 min before starting the test.
Identical loading procedures were followed in all the tests. All tests were con-
ducted at 25° C 共±0.1° C兲 with the temperature being controlled by a Peltier-
Plate system. The reproducibility of the experimental results was assured by
doing two replicates for each of the tests, and the results were fairly reproduc-
ible with ⫾5 % variation on average.
!"#$ %$&'(
%$&'( ('/& -340.
*+, %$&'(
/3 /8
5"6& -0&7.
FIG. 1—HSLS test design.
High Shear–Low Shear Test—The primary objective of the high shear–low

shear 共HSLS兲 test is to examine the structural build-up of solder pastes and flux
mediums. In order to investigate the build-up phenomena, the fluid structure
had to be broken down first. Based on the experience from previous experimen-
tal studies, applying a preshear of 10 s−1 for 30 s is adequate for partially
breaking down the paste’s structure. One of the problems of applying high
shear rate is that the sample tends to spill out from between the parallel plate
measuring geometry due to centrifugal force, leaving an undesirable gap be-
tween the upper plate and the sample. This ultimately leads to unreliable mea-
surement data being produced. So, the rate and duration of the preshear were
carefully chosen in such a way that the applied shear would not force the
sample out of the gap and will therefore provide reliable rheological data.
Having decided on the preshear, two rheological HSLS test methods were
designed—one for long-term build-up and the other one for short-term
build-up study. These consist of a preshearing step at 10 s−1 for a duration of
30 s and then applying a low shear rate for another 8 h 共for long-term build-up兲
or 900 s 共for short-term build-up兲. Figure 1 is showing the schematic of the
HSLS test design. The value of low shear rate used for the long-term build-up
study was 0.0005 s−1. For short-term build-up study, a total of five tests was
performed on each sample, corresponding to the five low shear rates: 0.001,
0.0015, 0.002, 0.0025, and 0.003 s−1. Preliminary tests showed that the paste
structure recovers 共builds-up兲 under these low shear conditions.

The result section is comprised of two parts. The results from the preliminary
investigation of long-term build-up behavior of paste samples are reported in
1000000
Apparent Viscosity (Pa s)
800000
600000
400000
200000 Shear rate: 0.0005 1/s

P1
P4
0
10 100 1000 10000
Time (sec)
FIG. 2—Apparent viscosity data of solder pastes as a function of time at a shear rate of
0.0005 s−1.
the first part. The second part outlines the results from the experimental and
modeling studies of the short-term build-up of solder paste and flux samples.
Preliminary Investigation of Long-Term Build-up Phenomenon

In this part of the investigation, solder paste samples were subjected to a con-
stant low shear rate for 8 h after being presheared for 30 s at 10 s−1. Figure 2
shows the results of the investigation for P1 and P4 solder pastes for the appar-
ent viscosity versus time plot. The low shear rate used in this case was
0.0005 s−1. The intention here was to examine the time-dependent build-up of
solder paste 共over a long period of time兲 after breaking down its structure. The
observations made in this preliminary investigation have helped in designing
the experiments for the remainder of the study reported in this chapter.
For solder pastes, when they are allowed to recover their structure follow-
ing preshear 共at low shear rate兲, the paste viscosity will tend to increase with
shearing time. The recovery data shown in Fig. 2 show that the rate of increase
in viscosity is fairly rapid at first and then decreases with further shearing time.
This type of behavior is quite similar to the behavior of bauxite residue 共red
mud兲 suspension as observed by Nguyen and Boger 关4兴. They measured the
yield stress instead of viscosity to observe the recovery behavior. Two important
observations can be made when the data presented in Fig. 2 are analyzed. First,
after breaking down the paste structure with the preshear, a remarkable in-
crease in viscosity occurs in a short time-scale—of the order of several minutes.
Second, the viscosity value starts to decrease slowly in longer term after reach-
ing an equilibrium state. These observations strongly suggest that the solder
paste possesses a yield stress. Lapasin et al. 关5兴 also identified yield stress as one
of the important properties of solder paste. The initial build-up of the solder
paste structure 共as may be seen in Fig. 2兲 can be attributed to the development
FIG. 3—Apparent viscosity of solder paste P1 as a function of time at low shear rates.
of yield stress in the solder paste. The decrease in viscosity value after a short
time period 共after reaching an equilibrium state兲 also suggests that the devel-
oped yield stress is quite weak in nature and was not sufficient to hold the
solder paste structure for a long time even at a very low shear rate.
The short-term build-up of solder paste structure is quite significant from
the application point of view. In the reflow soldering stage of the surface mount
technology 共SMT兲 assembly process, this prevents the slumping of solder paste
deposit after stencil withdrawal and during component placement. The time-
frame involved in this short-term build-up represents the time required for
component placement and the start of the reflow soldering process. Therefore,
a clear understanding of this structural build-up behavior is of immense impor-
tance to both the solder paste manufacturers and the end-users.
Results from the Investigation of Short-Term Build-up Behavior
Solder Paste Samples—This section outlines the results obtained from the
experimental and modeling studies of short-term build-up of solder pastes.
Four different commercially available lead-free solder paste samples 共P1, P2,
P3, and P4兲 were investigated; the details of these pastes are presented in the
Materials and Methods section.
The structural build-up of the solder paste samples at different low shear
rates is presented in Figs. 3–6 in terms of viscosity versus time plot. The solder
paste samples were first broken down with a preshear and then allowed to
build-up at low shear rates. The applied low shear rate values were 0.001,
0.0015, 0.002, 0.0025, and 0.003 s−1. The increase in apparent viscosity value
was used as a measure of thixotropic build-up of solder paste structure. This is
because the viscosity is the most important, widely used, and easily measurable
rheological property of solder paste. In a previous study, Maingonnat et al. 关3兴

also used viscosity to represent the structural build-up of a colloidal suspension
of clay. Nguyen and Boger 关4兴 rather used the yield stress value to represent the
recovery of red mud suspension.
All the solder paste samples have shown similar build-up behavior under
the shear rate range investigated. A careful observation of experimental results
共Figs. 3–6兲 demonstrates that for any applied low shear rate, the rate of struc-
tural build-up was quite rapid at first and then slows down as it approaches an
equilibrium state. Another important point to notice is that the rate of struc-
tural recovery was dependent on the applied shear rate. This is more obvious at
the equilibrium end of the build-up curve. Within the low shear rate range
examined, the equilibrium viscosity was found decreasing with increasing
shear rate.
The observed thixotropic build-up behavior of solder paste samples was
modeled using the SEM 共Eq 3兲. Figures 3–6 show a good correlation of the
model fitted result 共full solid line兲 and the apparent viscosity–time data for all
the solder paste samples. Table 1 presents the calculated values of equilibrium
viscosity 共␩⬁兲, characteristic time 共␺兲, and correlation coefficient 共r兲 as a func-
tion of applied shear rate for the paste samples when fitted to the SEM. The
model was originally fitted with three parameters, including the initial viscosity
共␩0兲. The initial viscosity ␩0 was included as a model parameter because this
viscosity value is strongly influenced by the rheometer inertia-effect 关3兴 as well
as the preshear history. However, in the model fitting, negative values were
obtained for ␩0. This is most likely due to insufficient data at the beginning of
shear 关6兴. Negative values for ␩0 are quite unrealistic and do not make any
sense in this context; hence the values are not shown in Table 2. The equilib-
rium viscosity 共␩⬁兲 and the characteristic time 共␺兲 are presented in Figs. 7 and
8, respectively, as a function of applied shear rates. Both ␩⬁ and ␺ were found
to decrease with an increase in shear rates. Figures 7 and 8 suggested that both
the equilibrium viscosity 共␩⬁兲 and characteristic time 共␺兲 can be fitted by power
functions as given in Eqs 4 and 5
␩⬁ = a␥˙ b 共4兲
␺ = c␥˙ d 共5兲
TABLE 2—Estimated values of the SEM parameters for the build-up of solder paste
samples.
Shear Equilibrium Characteristic

Solder Rate Viscosity ␩⬁ Time ␺ Correlation
Paste 共1/s兲共kPa s兲共s兲 Coefficient, r
P1 0.001 638.503 497.769 0.999
0.0015 440.656 317.360 0.999
0.002 329.605 243.681 0.999
0.0025 248.933 195.962 0.999
0.003 221.490 179.286 0.999
P2 0.001 454.847 610.978 0.999
0.0015 309.198 429.052 0.999
0.002 221.011 323.304 0.999
0.0025 179.572 261.415 0.999
0.003 142.506 215.697 0.999
P3 0.001 681.475 395.402 0.999
0.0015 449.619 273.675 0.999
0.002 323.290 211.967 0.999
0.0025 264.311 159.287 0.997
0.003 215.344 115.022 0.986
P4 0.001 843.855 425.078 0.999
0.0015 512.762 282.108 0.999
0.002 356.786 216.180 0.999
0.0025 291.839 160.530 0.997
0.003 233.795 137.156 0.996
1000
Equilibrium Viscosity (KPa s)
P1
P2
P3
P4
Power law model fit
100
0.0001 0.001 0.01
Shear rate (1/s)
FIG. 7—Equilibrium viscosity from the SEM for the solder paste samples.
1000
P1
P2
P3
Characteristic time (s)
P4
Power law model fit
100
0.0001 0.001 0.01
Shear rate (1/s)
FIG. 8—Characteristic time of the SEM for the solder paste samples.
Tables 3 and 4 present the estimated values of parameters from Eqs 4 and 5,
respectively.
The equilibrium viscosity and the characteristic time are of great impor-
tance to the actual solder paste assembly process. A higher equilibrium viscos-
ity value for a solder paste at a given shear rate would mean that the solder
paste will be less susceptible to slumping. Therefore, paste P4 would show the
highest resistance towards slumping, followed by P3, P1, and P2, according to
Table 3 and Fig. 7. The characteristic time here represents the time window
between the stencil printing and reflow soldering of solder paste when slump-
ing must be avoided. A higher characteristic time means more time for compo-
nent placement and reflowing the circuit board to make solder joints.
While resting, the undisturbed solder paste structure may take the form of
a three dimensional continuous network throughout the whole material vol-
ume 关4兴. The network may be a matrix of aggregates and/or flocs of primary
particles held together by the intermolecular forces, also known as van der
Walls forces 关7兴. This intermolecular bonding is caused by the momentary po-
larization of particles and is quite unstable in nature compared to chemical
TABLE 3—Estimated values of parameters for the equilibrium viscosity model fit for solder
paste samples.
Model Parameters for Equilibrium Viscosity

Solder Square of Regression
Paste a b Coefficient, R2
P1 0.683 ⫺0.992 0.996
P2 0.321 ⫺1.053 0.998
P3 0.490 ⫺1.048 0.999
P4 0.269 ⫺1.163 0.998
TABLE 4—Estimated values of parameters for the characteristic time model fit for solder
paste samples.
Model Parameters for Characteristic Time

Solder Square of Regression
Paste c d Coefficient, R2
P1 0.697 ⫺0.946 0.990
P2 0.893 ⫺0.947 0.998
P3 0.227 ⫺1.087 0.979
P4 0.325 ⫺1.040 0.997
bonding. The intermolecular forces consist of repulsive forces, e.g., electro-

static interaction forces, which prevent the collapse of the molecular structure
and attractive forces due to induction and dispersion forces. When the suspen-
sion is sheared, these weak intermolecular forces break-down causing the net-
work to break-down into smaller flocs. In experimental studies, this breakdown
behavior is generally manifested by the decrease in viscosity value. When the
solder paste structure is allowed to recover at low shear rate, the damaged
inter-particle bonds tend to restore by themselves. The mechanism may involve
reorganization and reflocculation of disrupted structural elements under the
action of diffusion 关1,4兴. The diffusion in turn is the result of Brownian motion.
Brownian motion is the random thermal agitation of atoms and molecules that
results in elements of the microstructure being constantly bombarded, which
causes them to move to a favorable position where they can—given the neces-
sary attractive force—attach themselves to other parts of the microstructure
关1兴. The Brownian rebuilding forces are quite small and weak compared to the
shearing forces. This implies that the recovering process for solder paste struc-
ture could be very long, as this small, random force may take a long time to
rearrange particles into flocs.
Flux Samples—This section presents the results of the investigation of

build-up behavior of the two flux samples 共F1 and F2兲. These fluxes were used
to prepare the four solder pastes. As with the solder paste samples, a preshear
共10 s−1 for 30 s兲 was applied to break-down the flux structure. Then the fluxes
were allowed to recover at low shear rates. The shear rate values applied were
0.001, 0.0015, 0.002, 0.0025, and 0.003 s−1.
Figures 9 and 10 present the results of the investigation of apparent viscos-
ity versus time plots. The build-up of the flux structure is obvious from the
continual increase of apparent viscosity value with shearing time. A careful
observation of Figs. 3–6, 9, and 10 reveals that the build-up phenomenon was
more pronounced for solder paste samples compared to flux mediums. At
0.001 s−1, for P1 solder paste, the viscosity increased from 25.8 to 537.7 kPa s
during the experiment 共Fig. 3兲. Whereas, for F1 flux, the increase in the appar-
ent viscosity was from 82 to 358 kPa s 共Fig. 9兲 for the same shear rate. More-
over, the rate of increase in viscosity was quite gradual for flux samples as
opposed to the initial rapid increase observed for solder paste samples.
It was found that the SEM satisfactorily fits the build-up data of flux
FIG. 9—Apparent viscosity of flux F1 as a function of time at low shear rates.
samples. Figures 9 and 10 illustrate that the SEM provides a very good fit 共solid
lines兲 of the apparent viscosity versus time data for the flux samples. This is
also evident from the values of correlation coefficient being close to one, as
presented in Table 5. Table 5 also presents the estimated values of the three
parameters 共␩0 , ␩⬁ , ␺兲 of the model for the flux samples. Unlike solder paste
samples, the model fitting for flux samples did not produce any negative values
for the initial viscosity 共␩0兲. The parameters—initial viscosity, equilibrium vis-
FIG. 10—Apparent viscosity of flux F2 as a function of time at low shear rates.

TABLE 5—Parameters of the SEM for the flux samples.
Shear Rate Initial Viscosity ␩0 Equilibrium Viscosity ␩⬁ Characteristic Time ␺

Fluxes 共1/s兲共kPa s兲共kPa s兲共s兲 Correlation Coefficient, R
F1 0.001 76.795 481.211 763.963 0.9990
0.0015 46.031 275.110 456.849 0.9995
0.002 36.782 174.420 235.514 0.9996
0.0025 30.670 153.636 263.718 0.9996

0.003 31.271 119.481 262.885 0.9953
F2 0.001 58.085 192.926 319.642 0.9984
0.0015 42.792 145.750 224.385 0.9988
0.002 30.245 102.402 149.606 0.9985
0.0025 22.676 84.414 139.052 0.9990
0.003 19.164 74.361 110.930 0.9970
100
F1
Initial Viscosity (kPa s)
F2
Power function fit
10
0.0001 0.001 0.01
Shear rate (1/s)
FIG. 11—Initial viscosity of the SEM for flux samples.
cosity, and characteristic time—are presented in Figs. 11–13, respectively, as a

function of applied shear rate. All three parameters were found to fit nicely
using power law relations
␩0 = m␥˙ n 共6兲
␩⬁ = k␥˙ l 共7兲
␺ = q␥˙ r 共8兲
The estimated values of the parameters of these equations are given in Tables
6–8, respectively.
The calculated values of the SEM parameters for both solder pastes and
1000
Equilibrium Viscosity (kPa s)
F1
F2
Power function fit
100
10
0.0001 0.001 0.01
Shear rate (1/s)
FIG. 12—Equilibrium viscosity of the SEM for flux samples.

1000
F1
Characteristic Time (sec)
F2
Power function fit
10
0.0001 0.001 0.01
Shear rate (1/s)
FIG. 13—Characteristic time of the SEM for flux samples.
flux samples 共presented in Tables 2 and 5兲 can be used to generate empirical

equations at a corresponding shear rate. Table 9 shows the empirical equations
generated for solder paste and flux samples at the constant shear rate of
0.0001 s−1. Equations for other constant shear rates can be generated in the
similar way. The paste formulators may utilize these equations to get a trend
and prediction for new paste formulations.
TABLE 6—Estimated values of parameters for the initial viscosity model fit for flux
samples.
Model Parameters for Initial Viscosity

Square of Regression
Flux Mediums m n Coefficient, R2
F1 0.202 ⫺0.847 0.938
F2 0.046 ⫺1.041 0.989
TABLE 7—Estimated values of parameters for the equilibrium viscosity model fit for flux
samples.
Model Parameters for Equilibrium Viscosity

Flux Mediums k l Coefficient, R2
F1 0.077 ⫺1.259 0.987
F2 0.384 ⫺0.904 0.990
TABLE 8—Estimated values of parameters for the characteristic time model fit for flux
samples.
Model Parameters for Characteristic Time

Flux Mediums q r Coefficient, R2
F1 0.480 ⫺1.051 0.833
F2 0.413 ⫺0.963 0.983
Conclusions
The time-dependent structural build-up of solder paste and flux mediums has
been investigated in this paper. The objective of the study was to quantify and
model the structural build-up of solder pastes and flux mediums using the
SEM. Experiments were designed to examine both short-term and long-term
build-up behaviors of paste materials. The SEM has been used satisfactorily to
fit the short-term structural rebuilding of solder paste and flux mediums. As
expected, for solder paste samples, the rate of structural recovery was found
dependent on the applied shear rate. The model parameters, such as equilib-
rium viscosity and characteristic time, have been correlated with the shear-
thinning and slumping behaviors of solder paste during the stencil printing
process. A higher equilibrium viscosity of solder paste would mean a higher
resistance towards slumping. The characteristic time, on the other hand, rep-
resents the time-frame from the end of stencil printing to the beginning of the
reflowing process. A higher characteristic time for solder paste would therefore
mean more time for component placement by avoiding slumping of the solder
paste.
The results from these experimental and modeling studies of the build-up
of solder paste structure would be quite useful to both the solder paste manu-
facturers and end-users. The paste manufacturers and formulators can use the
technique developed to predict and quantify the slumping behavior of solder
paste. The end-users, for example, the electronics assemblers/manufacturers,
can also use the technique to optimize their assembly process by minimizing/
preventing slumping of the solder paste.
TABLE 9—Empirical equations for solder pastes and flux samples at a constant shear rate
of 0.0001 s−1.
Solder Paste/Flux Empirical Equations

P1 ␩ = 25.786+ 612.717共1 − e−共t/497.769兲兲
P2 ␩ = 6.027+ 448.82共1 − e−共t/610.978兲兲
P3 ␩ = 39.648+ 641.827共1 − e−共t/395.402兲兲
P4 ␩ = 45.009+ 798.846共1 − e−共t/425.078兲兲
F1 ␩ = 76.795+ 404.416共1 − e−共t/763.963兲兲
F2 ␩ = 58.085+ 138.841共1 − e−共t/319.642兲兲
References
关1兴 Barnes, H. A., “Thixotropy—a Review,” J. Non-Newtonian Fluid Mech., Vol. 70,
1997, pp. 1–33.
关2兴 Heymann, L., Noack, E., Kampfe, L., and Beckmann, B., “Rheology of Printing
Inks—Some New Experimental Results,” Proceedings of the XIIth Interantional
Congress on Rheology, Quebec City, August 18-23, Canadian Rheology Group, Que-
bec, 1996, p. 541.
关3兴 Maingonnat, J. F., Muller, L., and Leuliet, J. C., “Modelling the Build-up of a Thixo-
tropic Fluid Under Viscosimetric and Mixing Conditions,” J. Food. Eng., Vol. 71,
2005, pp. 265–272.
关4兴 Nguyen, Q. D. and Boger, D. V., “Thixotropic Behaviour of Concentrated Bauxite
Residue Suspensions,” Rheol. Acta, Vol. 24, 1985, pp. 427–437.
关5兴 Lapasin, R., Dicamp, Sirtori, V., and Casati, D., “Rheological Characterization of
Solder Pastes,” J. Electron. Mater., Vol. 23, No. 6, 1994, pp. 525–532.
关6兴 Chamberlain, E. K. and Rao, M. A., “Rheological Properties of Acid Converted
Waxy Maize Starches in Water and 90% DMSO/10% Water,” Carbohydr. Polym.,
Vol. 40, 1999, pp. 251–260.
关7兴 Durairaj, R., Ekere, N. N., and Salam, B., “Thixotropy Flow Behaviour of Solder
and Conductive Adhesive Pastes,” J. Mater. Sci.: Mater. Electron., Vol. 15, 2004, pp.
677–683.
doi:10.1520/JAI103009
R. Durairaj,1 Lam Wai Man,2 and S. Ramesh2
Rheological Characterisation and Empirical

Modelling of Lead-Free Solder Pastes
and Isotropic Conductive Adhesive
Pastes
ABSTRACT: Solder pastes and isotropic conductive adhesives 共ICAs兲 are

widely used as an interconnect in the electronic industry. Paste printing pro-
cess accounts for majority of the assembly defects in the electronic manu-
facturing industry. This study investigates the effect of shear rates on the
viscosities of the pastes 共solder pastes and ICAs兲 used for flip chip assembly.
Empirical models such as the power law and the Cross model were used to
quantify the viscosity over a range of shear rates for solder pastes and fit to
the experimental data. The shear rate dependence viscosity of solder pastes
could be used to study the flow behavior experienced by the pastes during
the stencil printing process. From the results, viscosities of all three types of
pastes were said to be dependence on shear rate. In a stencil printing pro-
cess, as if the viscosity of the solder paste was too high, the paste might not
wet the surface of the substrate and more energy was needed to force the
paste penetrate the aperture. This study has revealed that the fitting of the
Cross model is generally of better quality than the power law model because
the qualitative behavior of the Cross model throughout the whole range of
shear rates 共0.001 s⫺1 to 100 s⫺1兲 is essentially the same as the experimen-
tal data.
Manuscript received January 30, 2010; accepted for publication June 17, 2010; published
online July 2010.
1
Dept. of Mechanical and Material Engineering, Faculty of Engineering and Science
共FES兲, Univ. Tunku Abdul Rahman 共UTAR兲, Jalan Genting Kelang, Setapak, 53300 Kuala
Lumpur, Malaysia, e-mail: rajkumar@utar.edu.my
2
Dept. of Mechanical and Material Engineering, Faculty of Engineering and Science
共FES兲, Univ. Tunku Abdul Rahman 共UTAR兲, Jalan Genting Kelang, Setapak, 53300 Kuala
Lumpur, Malaysia.
Cite as: Durairaj, R., Man, L. W. and Ramesh, S., ‘‘Rheological Characterisation and
Empirical Modelling of Lead-Free Solder Pastes and Isotropic Conductive Adhesive
Pastes,’’ J. ASTM Intl., Vol. 7, No. 7. doi:10.1520/JAI103009.
186
DURAIRAJ ET AL., doi:10.1520/JAI103009 187
KEYWORDS: rheology, stencil printing process, power law model,

Cross model
Nomenclature
ICAs ⫽ isotropic conductive adhesives

K ⫽ consistency coefficient
m ⫽ rate constant
n ⫽ power law index
␩ ⫽ viscosity
␩a ⫽ apparent viscosity
␩0 ⫽ zero shear viscosity
␩⬁ ⫽ infinite shear viscosity
␶ ⫽ shear stress
␥˙ ⫽ shear rate
␭ ⫽ internal structure
R2 ⫽ regression model
Introduction
The electronic industry has seen a rapid growth in various sectors of the mar-
ket, e.g., the computer, telecommunications, automotive, and consumer sec-
tors. Some of the key drivers for this growth include the consumers demand for
portability, flexibility, and better performance of the final product. As a result,
this imposes tight requirements in terms of size reduction, performance in-
creases, higher reliability, and lower cost. As the current product miniaturisa-
tion trend is set to continue for hand-held consumer products, area array type
package solutions 共such as chip scale packages and flip chip兲 are now being
designed into products. The assembly of these devices requires the printing of
very small paste 共solder paste or ICAs兲 deposits consistently from pad to pad
and from board to board.
Paste materials are dense suspensions, which exhibit complex flow behav-
ior under the influence of stress. The solder pastes have been reported to be
thixotropic 关1–3兴, shear thinning 关4,5兴, and posses yield stress. A few papers
have attempted to correlate rheological measurement to printability 关6–8兴 and
slumping. These earlier works on the visco-elastic behavior of pastes identified
the need for more information on the solid and liquid characteristics of the
pastes, especially the need for further work on linear visco-elastic region of
pastes.
The aim of this study was to investigate the effect of shear rates on the
viscosities of the pastes 共solder pastes and isotropic conductive adhesives
共ICAs兲兲 used for flip chip assembly. The flow curve test method was used to
evaluate the pastes with respect to its printability and printing defects. Empiri-
FIG. 1—Sub-processes in paste printing 关16兴.
cal models such as the power law and the Cross model were used to quantify
the viscosity over a range of shear rates for solder pastes and fit to the experi-
mental data.
Overview of the Paste Printing Process

The paste printing process is paramount in producing high volume, low cost
production; it does account for some 60 % of assembly defects 关9兴, and it is
estimates that up to 87 % of reflow soldering defects are caused by stencil
printing defects 关10兴. As can be seen in Fig. 1, the key sub-processes in the
solder paste printing process include the paste roll in front of the squeegee, the
aperture filling, and aperture emptying stages. During the stencil printing, the
paste develops a rolling action in front of the squeegee, filling the apertures in
the stencil some distance ahead of the squeegee. The squeegee then shears off
the paste in the apertures as it moves over the stencil. It is known that during
the printing process, the squeegee generates hydrodynamic pressures in the
paste roll that injects the paste into the apertures. Once the print stroke is
completed, the board is separated mechanically from the stencil. The separa-
tion of stencil and printed circuit board 共PCB兲 or substrate occurs after the
squeegee move across the stencil, and the substrate is then separated mechani-
cally from the stencil.
The paste printing process is known to be controlled by a number of pro-
cess parameters, which can be divided into four groups 关11兴: Printer, stencil,
environmental, and paste parameters. Some of these parameters are fixed 共e.g.,
stencil兲, while the paste properties such as viscosity are constantly changing
during the print cycle. The key physical sub-processes include 共i兲 paste pre-
print treatment, 共ii兲 squeegee deformation, 共iii兲 paste roll 关12兴, 共iv兲 aperture
filling 关13兴, 共v兲 aperture emptying 关14,15兴, and 共vi兲 past slump. These sub-
processes are linked together by the properties of the pastes such as its flow
history and its rheology. The pressure in the paste during and after aperture
filling helps determine whether the paste will adhere onto the substrate, stencil,
or squeegee after aperture emptying.
Viscosity
Viscosity is defined in Newton’s law as the coefficient of shear stress versus
shear rate
␴
␩= 共1兲
␥˙
where:
␩ = viscosity,
␶ = shear stress, and
␥˙ = shear rate 关17兴.
Viscosity can also be defined as the internal friction of a liquid, caused by
molecular attraction, which makes it resist a tendency to flow. Viscosity is one
aspect of rheology and a very important issue for stencil printing process in
electronic industry. A viscosity test is used to determine the flow characteristics
of a solder paste. Solder pastes are thixotropic fluids. Thixotropic refers to the
quality of certain materials that are paste or gel-like at rest but exhibit fluid
behavior when stress applied 关18兴. In other words, their viscosity changes with
stress. When subjected to a constant rate of shear stress, viscosity of solder
paste will decrease over time 共shear-thinning behavior兲关19兴. The viscosity of the
solder paste decreases as the shear stress on the solder paste increases. There-
fore, solder pastes are more fluid when dispensed with a squeegee 共applied
shear stress兲. But the paste remains thick when no stress is applied. When
shear-thinned, solder pastes are capable of flowing into stencil apertures. When
the shear stress is removed, the viscosity of solder paste increases, allowing the
deposit to maintain its printed geometric shape. If a paste is overly viscous, it
results in insufficient paste volume, resulting in open joints. However, low vis-
cosity results in slumping and bridging. Hence, it is very important to under-
stand the flow properties of solder paste in order to acquire an optimum stencil
printing results 关16兴.
Theoretical viscosity models such as the power law and the Cross model
are required to fit the experimental data to determine the accuracy of the pro-
cessing range and for which data can be obtained 关5,20兴. In this paper, viscosi-
ties of different types of solder pastes are being studied over a range of shear
rates.
Empirical Modelling of Solder Pastes
Power Law Model

The power law model is also known as Ostwald de Waele power law equation
␶ = K␥˙ n 共2兲
where:
␶ = shear stress,
␥˙ = shear rate, and
K = consistency coefficient.
K describes the overall range of viscosities across the region of the flow
curve that is being modelled. If the power law region includes 1 s−1 shear rate,
then K is the viscosity or stress at that point. The n value is the power law
index. For a shear-thinning fluid, 0 ⬍ n ⬍ 1. The more shear thinning of a
sample, the closer n is to zero 关5,21,22兴. On defining the viscosity, ␩ as ␶ / ␥˙
␩ = K␥˙ n−1 共3兲

Shear stress–shear rate plots of many fluids become linear when plotted on
double logarithmic coordinates, and the power law model describes the data of
shear thinning and shear thickening fluids. Taking natural logarithms of both
sides, the equation below is obtained
log共␩兲 = 共n − 1兲log共␥˙ 兲 + log共K兲共4兲

A plot of log ␩ against log ␥˙ shows this relationship to be linear. Nevertheless,
a plot of experimental and predicted values of log ␩ and log ␥˙ is useful for
observing trends in data and ability of the model to follow the experimental
data. One reason that the power law model is useful is because of its ability
over the shear rate range of 101 – 104 s−1 that can be obtained with many com-
mercial visco-metric measuring devices 关5兴. The magnitudes of the consistency
and the power law indexes of certain solder pastes depend on the specific shear
rate range being used so that when comparing the properties of different solder
pastes, an attempt should be made to determine them over a specific range of
shear rates. One drawback of the power law model is that it does not describe
the low shear and high shear rate constant-viscosity data of shear-thinning
fluids 关23兴.
Cross Model
Cross originally derived his equation for particulate suspensions in aqueous
and non-aqueous media, which involved the formation and rupture of struc-
tural linkages between particles during flow 关24兴. The Cross equation has been
adapted to a form more appropriate to suspensions. A simple expression was
assumed as an equation of state
␩a = ␩⬁ + c␭ 共5兲
in which the apparent viscosity, ␩a, of a semisolid suspension is linearly related
to its internal structure, ␭, whose value is zero for the fully broken down con-
dition as ␥˙ , the shear rate, becomes large and unity in the fully built up condi-
tion developed as ␥˙ approaches zero. Under these conditions, the value of the
constant, c, is given by ␩0 − ␩⬁, where ␩0 is the asymptotic viscosity at low shear
rates and ␩⬁ is that at high shear rates. This indicates how the fluid behaves in
very high shear processing conditions. It should be noted that in this simple
model, the structural parameter ␭ is defined as 共␩ − ␩⬁兲 / 共␩0 − ␩⬁兲 and is there-
fore a linear function of viscosity ␩. A generalised kinetic equation for struc-

tural change 关25兴
d␭
= a共1 − ␭兲 − b␭␥˙ m 共6兲
dt
where the first term on the right describes the rate of structural build up being
proportional to the extent of un-built-up structure. The second term describes
the rate of breakdown proportional to the degree to which structure is already
built up and to the magnitude of the shear rate. Equilibrium is achieved at
d␭ / dt = 0, leading to
冉冊
a = ␭共a + b␥˙ m兲 and ␭= 共7兲
b m
1+ ␥˙
a
Hence
␩0 − ␩⬁
冉冊
␩ = ␩⬁ + m 共8兲
b
1+ ␥˙
a
which is the Cross steady-state equation 关23,25兴. The zero shear viscosity, ␩0, is
a critical material property and can prove valuable in making assessments of
suspension and emulsion stability. The parameter m is known as the rate con-
stant. It is dimensionless and is a measure of the degree of dependence of
viscosity on shear rate in the shear-thinning region. When m = 0, this indicates
Newtonian behavior. While m ⬎ 0, this means the viscosity decreases with in-
creasing shear rate, ␥˙ , which is the condition for shear-thinning behavior. The
value of b / a is known as the consistency coefficient. The cross model describes
well the shear dependence of fluids over a wide range of shear rates as shown in
Fig. 2 关20兴.
Experimentation
Apparatus
All the flow curve test measurements were carried out with the Physica MCR
301 controlled stress rheometer. In order to avoid the formation of wall slip at
the interface between the plate and conductive paste, a parallel plate geometry
was chosen with a diameter of 25 mm. A gap height of 0.5 mm was used
between the upper and lower plates, as shown in Fig. 1. Prior to loading the
sample onto the rheometer, the conductive paste was stirred for about 1–2 min
to ensure that the paste structure is consistent with the particles being re-
distributed into the paste. A sample was loaded on the Peltier plate, and the
geometry plate was then lowered to the gap of 0.5 mm. The excess paste at the
plate edges was carefully trimmed using a plastic spatula. Then the sample was
allowed to rest for about 1 min in order to reach the equilibrium state before
FIG. 2—Plot of viscosity against shear rate for shear-thinning fluids identifying three
separate regions: A zero-shear viscosity at low shear rates, a power law region at inter-
mediate shear rates, and an infinite-shear viscosity at high shear rates 关23兴.
starting the test. All tests were conducted at 25° C with the temperature con-
trolled by the Peltier-plate system. Each test was repeated for three times for
stabilisation 共with fresh samples used for each test兲.
Paste Materials
There are two main types of the paste materials used in the assembly flip chip
devices, namely, 共i兲 solder paste and 共ii兲 ICAs.
Solder Paste
Solder paste is one of the most widely used interconnection materials in the
surface mount technology assembly process. Solder paste is a homogenous and
stable suspension of solder alloy particles suspended in a flux/vehicle system, as
can be seen in Fig. 3共a兲. The flux/vehicle is a combination of solvents, thicken-
ers, binders, and fluxing agents 关13兴, as shown in Fig. 3共b兲. Solder pastes con-
sists of three main constituents, namely,
共a兲 Solder alloy particles which forms the base for the metallic bond,
共b兲 The flux system which helps to promote the formation of the metallic
FIG. 3—共a兲 Solder particles and 共b兲 constituent of the flux vehicle system 关13兴.
FIG. 4—ICAs microstructure.
bond by providing a good wetting condition and for cleaning the sur-
faces, and
共c兲 The vehicle carrier system which facilitates the binding together of the
solder powder particles and the flux system together, and for providing
the desired rheological properties for processing and depositing the
paste onto the PCB.
Isotropic Conductive Adhesives

The ICAs consist of 70–80 % metal fillers dispersed in an epoxy resin, as shown
in Fig. 4. During curing, the epoxy resin shrinks, which enables the metal fillers
to come into contact, hence conducting electricity. There are various types of
metal fillers, e.g., silver, nickel, gold, copper, carbon, and metal-coated particles.
The most commonly used metal filler is silver. Silver is preferred to other metal
fillers because of its unique characteristic to conduct electricity even after it
oxidises. When the adhesive is cured, the filler particles are uniformly distrib-
uted and form a network within the polymer structure. From these networks,
electricity can pass through making the mixture electrically conductive and due
to nature of the network, the current can flow in any direction. The ICAs gen-
erally consist of resin such as epoxy 共most commonly兲, polyamides, silicones,
and acrylic adhesives.
TABLE 1—Constituent of solder paste and ICAs investigated.
Particle size
Paste Distribution Particle Paste Solder Alloys/Materials
Samples 共␮m兲 Shape Medium 共Percentage by Weight兲
Flux vehicle Tin 95.5 %/copper
P1 20–45 Spherical system 0.7 %/silver 3.8 %
Flux vehicle
P2 20–45 Spherical system Tin 96.5 %/silver 3.5 %
P3 20–45 Flakes Epoxy resin Silver 88 %
Material and Sample Preparations

Three different types of pastes were used in the study, and the details of the
pastes 共lead-free solder pastes are labeled as P1 and P2, while ICA sample is
labeled as P3兲 are presented in Table 1. The pastes 共P1 and P2兲 contain metal
particles, and P3 contains silver flakes of 88 wt %. In order to minimize sepa-
ration and prolong shelf life, manufacturers prescribe specific storage condi-
tions, for example, the ICA and pastes used in this study are stored in a fridge
unit at ⫺20 and −4 ° C, respectively. As it is important to carry out the experi-
ments on the paste samples at room temperature, the procedure used in the
study is to bring out the ICA and solder paste out of the fridge prior to the
rheological tests and to allow these pastes to attain room temperature. All the
rheological measurements are carried out at 25± 0.1° C.
Rheology Test
Viscosity Test—In the stencil printing process, the viscosity of the paste
must be low enough for the squeegee to force the paste through the stencil
apertures but high enough to recover to its required shape and not flow beyond
its stenciled area. The viscosity test can be used to provide a quick indication of
the viscosity of a solder pastes changes over a wide range of shear rates. The
experimental parameters utilized for the viscosity test is outlined in Table 2.
Thixotropy Test—Thixotropy test aims to simulate the structural breakdown

and recovery of the solder paste and flux system. In the hysteresis loop test, the
shear rate were increased from 0.001 to 100 s−1 and then decreased from 100
to 0.001 s−1. This property maybe essential for understanding the rolling mo-
tion of the paste during the printing process as the squeegee pushes the paste
TABLE 2—Experimental parameters for flow curve test.
Experimental Values
Initial Shear Final Shear Number of Interval Between Overall Duration

Rate 共s−1兲 Rate 共s−1兲 Measuring Point Measuring Point 共s兲共s兲
0.001 100 31 3 100
back and forth. In this test, the pastes are subjected to low and high shear rates
over a period of time, and the recovery of the paste as a function of the viscosity
is noted after the removal of high shear rate.
Viscosity
Figure 5 shows the plots for viscosity against the shear rate for all three paste
samples. As expected, as the shear rate is increased, all the pastes showed a
decrease in the viscosity. The drop in viscosity clearly indicates that the pastes
are shear thinning in nature and the structure of the pastes was undergoing
changes due to destruction of flocculation in the suspensions 关17兴. As men-
tioned previously, the viscosity measured at low shear rates will be useful in
assessing the suspension stability. With respect to this, at a low shear rate of
0.001 s−1, the highest viscosity was recorded for sample P2, followed by P1 and
P3, as shown in Fig. 3. The high viscosity measured for sample P1 could be due
to the strong interaction between tin/silver particles with the flux medium as
opposed to the tin/copper/silver system formulation in sample P2. From the
result, sample P1 showed a good stability at low shear rates, which could indi-
cate the particles and flux medium will not separate. In addition, the high
viscosity will give the paste a good cohesive behavior and prevent slumping
during and after the printing process. In contrast to P1 and P2, the viscosity
measured for sample P3 was the lowest. These results are in line to that re-
ported by Durairaj et al. 关1兴. The low viscosity attributed o sample P3 could be
merely due the poor interface wetting of the epoxy resin and the silver flake. In
addition, the irregular shapes of the silver flakes could have decreased the floc-
culation in the systems, hence reducing the overall viscosity of the conductive
adhesives 关4,5,7,11兴.
The experimental viscosity data was fitted to the power law and Cross
model, as shown in Fig. 5. For all samples, the Cross model showed a better fit
compared to the power law model. Despite the fact that both the model was
designed to evaluate the shear-thinning behavior of suspensions, the results
indicate otherwise. There could be two possible explanations: First, the shear
rate investigated may to wide and fall beyond the range of the power law
model. The second reason could be attributed to the presence of three regions:
First, Newtonian region; second, shear-thinning region; and third, Newtonian
region, which is easily captured by the Cross model. The results from the ex-
periment seem to correlate well with previous studies 关4兴 and also prove that
the pastes 共solder paste and ICAs兲 show the three regions when the samples are
sheared from the low to high shear rates. Hence for a wider shear rates, the
Cross model provides a better experimental fit compared to the power law
model.
A further analysis was carried on the fitted data, as shown in Tables 3 and
4. The power law and the Cross model used to quantify the viscosity/shear rate
profile for the shear-thinning solder pastes and fit to the experimental data. In
a power law model, as the power index, n, lies between zero and one, 0 ⬍ n
FIG. 5—Flow curve of pastes: 共a兲 lead-free solder paste, P1; 共b兲 lead-free solder paste, P2;
and 共c兲 ICA paste, P3.
TABLE 3—Variables used in the power law model.
Consistency Coefficient,
Samples K 共Pa s兲 Power Law Indexes, n Correlation Ratio, R2
P1 454.55 0.409 23 0.871 66
P2 2913.3 0.161 09 0.671 38
P3 72.068 0.353 70 0.726 83
⬍ 1, this indicates that the viscosity of the sample being tested was exhibiting
shear-thinning behavior 关5,21,22兴. Based on Table 3, the n values of P1, P2, and
P3 samples were 0.409 23, 0.161 09, and 0.353 70, respectively. Since the n
values lie between zero and one, the viscosities of all three samples were expe-
riencing shear-thinning behavior. In a Cross model equation, the rate constant,
m, is a measure of the degree of dependence of viscosity on shear rate in the
shear-thinning region 关20兴. When m = 0, this indicates Newtonian behavior.
While m ⬎ 0, this means that the viscosity decreases with increasing shear rate,
␥˙ . Based on Table 4, the rate constant, m, values of all three type of pastes were
more than zero, which indicated shear-thinning behavior 共P1 = 0.699 67, P2
= 0.873 12, P3 = 0.583 89兲.
The correlation coefficient, R2, showed the relationship between the viscos-
ity and the shear rate. R2 is measure of how well the data correlated. The closer
it is to one, the closely correlated the data is 关26兴. Based on Tables 3 and 4, the
correlation ratios, R2, of the Cross model for all three pastes were higher than
that of the power law model. While P3 showed the highest R2 value, 0.993 29,
while P1 and P3 showed 0.981 87 and 0.975 03, respectively. The R2 values of
all three pastes were said to be almost perfect linear relationship between vis-
cosities and shear rates because the R2 value is ⬃1. Often, the magnitudes of
the consistency and the flow behavior indexes of a solder pastes depend on the
specific shear rates range being used so that when comparing the properties of
different solder pastes, an attempt should be made to determine them over a
specific range of shear rates.
From the point of view of approximation of the obtained results, the power
law model is good; however the Cross model can describe the results more
precisely. This follows from the facts that the Cross model provides more infor-
mation on rheological properties of a suspension in a wide range of shear rates.
As mentioned earlier, the power law model does not describe the low shear and
TABLE 4—Variables used in the Cross model.
Infinite Shear
Zero Shear Viscosity, Viscosity, Rate Constant, Correlation Ratio,
Samples ␩0 共Pa s兲 ␩⬁ 共Pa s兲 m R2
P1 90 163 38.359 0.699 67 0.981 87
P2 1 830 000 97.084 0.873 12 0.975 03
P3 9180 8.61 0.583 89 0.993 29
high shear rates constant-viscosity data of shear-thinning fluids. Of these rea-

sons, the power law model does not fit well to the experimental data as the
Cross model did.
Thixotropic Behavior of Pastes

Figure 6 shows the hysteresis loop for all three paste samples. The samples
were constantly subjected to high shear rate, 100 s−1, with time and recover to
their initial shear rate, 0.001 s−1. The overall time interval was 240 s. The effect
of increasing shear rate on the viscosity for the paste samples was being inves-
tigated. The drop in viscosities for all three samples clearly indicates that the
pastes are shear thinning in nature and the structure of the pastes was under-
going changes due to the destruction of flocculations in the suspensions 关14兴.
All three samples show a hysteresis area for which an area between the up and
down curves is observed. The region between the up curve and down curves in
the hysteresis curve is an indication of the thixotropic behavior of the pastes.
Therefore, all three samples studied are thixotropic suspensions. The enclosed
area within the curves indicates the extent of the structural breakdown in the
sample for the applied shear.
The plot of the effect of the shear rate on the viscosity is presented in Fig.
6共a兲 for P1 pastes. As expected, the viscosity of the pastes drops with increasing
shear rate, which indicates shear-thinning behavior of the pastes. The area
between the down curve and up curve indicates that the P1 paste is thixotropic
in nature, which have been confirmed in previous studies on solder pastes
关15,16兴. P1 paste shows the highest degree of thixotropy because of the high
hysteresis area among all three pastes. The large area within the hysteresis loop
in the P1 paste indicates that the sample undergone a large structural break-
down. P1 paste is said to have the weakest structural bonding, which easily is
being broken down by increasing shear rate. The stronger attraction between
the particles in P2 paste leads to a good recovery after the shear rate is removed
and P2 is said to have a strong thixotropic behavior. While for P3 paste, the
particle size ranges from 8 to 10 ␮m, which is smaller than P1 and P2 pastes,
where the particle size is around 20– 45 ␮m. These smaller particles of P3
paste tend to fill up the spaces between the flocs and form stronger bond.
Therefore, P3 paste, which consisted of smallest particle size, is said to be
strongly thixotropic.
Based on Fig. 6共b兲 and 6共c兲, the viscosities of P2 and P3 pastes drop with
increasing shear rates, which is consistent with that of the P1 paste and shows
that the pastes exhibit shear-thinning behavior. For the P2 paste, the downward
curve crosses the ascending curve, the cross over point being situated at
0.01 s−1. On this downward path, the viscosity increase at low shear rate indi-
cates that a network structure is able to be rebuilt when the shear rate goes
under a critical value 关17兴. The smallest hysteresis loop area 共P2 paste兲 corre-
sponds to a thixotropic state for which the inter-particle bonds would be strong
enough to favour a quick rearrangement of the structure. This could be attrib-
uted to the strong interaction of the small Sn particles.
Figure 6共c兲 shows superposed upward and downward curves at high shear
rates for P3 paste. This could just as well correspond to an extremely thixotro-
FIG. 6—Hysteresis loop of pastes: 共a兲 Lead-free solder paste, P1; 共b兲 lead-free solder
paste, P2; and 共c兲 ICA paste, P3.
pic material, capable of rebuilding its structure almost instantaneously 关17兴. To

support this justification, a steady shear rate test was carried out. From the
steady shear rate test, sample P2 recorded the highest viscosity 共94000 Pa.s兲 at
zero shear rate followed by P1 共14 400 Pa.s兲 and P3 共13 100 Pa.s兲. In the middle
interval, which high shear rate is applied, P3 is observed to undergo the largest
structural breakdown followed by P2 and P1. As stated earlier, P3 has the low-
est viscosity at zero shear rate; this suggests that P3 has less resistance to flow
compared to sample P1 and P2. Therefore, P3 paste has the largest structural
breakdown when high shear rate is subjected. Although P3 showed the largest
structural breakdown, its rapid structural build up interpreted that P3 is a
strongly thixotropic paste material.
Conclusion
In this study, the viscosities of several commercial solder pastes 共lead-free and
ICA paste兲 are examined to find the effect of shear rate on the viscosity and to
establish the correlation between paste viscosity and stencil printing process.
Furthermore, the power law and the Cross model used to quantify the viscosity/
shear rate profile for the shear-thinning solder pastes and fit to the experimen-
tal data. From the experimental results, as the shear rates increased, the vis-
cosities of the three pastes 共solder pastes and ICA兲 decreased 共shear thinning兲.
In addition, the lead-free solder pastes exhibited the highest viscosity at low
shear rates, which indicates that the dispersion of the paste is the more stable
ICA paste. In a stencil printing process, a paste of too high viscosity needs more
energy to force the paste through the aperture and leads to poor surface wet-
ting. The statistical data show that the Cross model fits well to the experimental
data than the power law model because it provides information in a wider
range of shear rates. The presence of an area between the down curve and up
curve shows that the paste materials are thixotropic in nature. The findings
from the study show that a smaller particle size leads to a large surface area
and better inter-particle attraction. The structural breakdown and recovery of
the pastes are important parameters that can be used in the development of
new formulation of solder pastes and ICAs.
References
关1兴 Durairaj, R., Ekere, N. N., and Salam, B., “Thixotropy Flow Behaviour of Solder
and Conductive Adhesives Paste,” J. Mater. Sci.: Mater. Electron., Vol. 15, 2004, pp.
677–683.
关2兴 Nguty, T. A., Ekere, N. N., and Adebayo, A., “Correlating Solder Paste Composition
with Stencil Printing Performance,” IEEE/CPMT International Electronics Manu-
facturing Technology Symposium, September 1999, pp. 305–312.
关3兴 Lapasin, R., “Rheological Characterisation of Solder Pastes,” J. Electron. Mater.,
Vol. 23共6兲, 1994, pp. 525–532.
关4兴 Durairaj, R., Jackson, G. J., Ekere, N. N., Glinski, G., and Bailey, C., “Correlation of
Solder Paste Rheology with Computational Simulations of the Stencil Printing
Process,” Soldering Surf. Mount Technol., Vol. 14共1兲, 2002, pp. 11–17.
关5兴 Evans, J. and Beddow, J., “Characterisation of Particle Morphology and Rheologi-
cal Behaviour in Solder Paste,” IEEE Trans. Compon., Hybrids, Manuf. Technol.,
Vol. 10共2兲, 1987, pp. 224–231.
关6兴 Bao, X., Lee, N. C., Raj, R. B., Rangen, K. P., and Maria, A., “Engineering Solder
Paste Performance Through Controlled Stress Rheology Analysis,” Soldering Surf.
Mount Technol., Vol. 10共2兲, 1998, pp. 26–35.
关7兴 He, D., Ekere, N. N., Jackson, G. J., Rajkumar, D., and Salam, B., “Monte Carlo
Study of Solder Paste Microstructure and Ultra-Fine-Pinch Stencil Printing,” J.
Mater. Sci.: Mater. Electron., Vol. 14共8兲, 2003, pp. 501–506.
关8兴 Lapasin, R., “Rheological Characterization of Solder Pastes,” J. Electron. Mater.,
Vol. 27共3兲, 1998, pp. 138–148.
关9兴 Haslehurst, L., Ekere, N. N., “Parameter Interactions in Stencil Printing of Solder
Pastes,” J. Electron. Mater., Vol. 6共4兲, 1996, pp. 307–316.
关10兴 Okuru, T., Kanai, M., Ogata, S., Takei, T., and Takakusagi, “Optimisation of Solder
Paste Printability with Laser Inspection Technique,” IEEE/CPMT International
Electronics Manufacturing Symposium, 1993, pp. 157–161.
关11兴 Haslehurst, L. and Ekere, N. N., “Parameter Interactions in Stencil Printing of
Solder Pastes,” J. Electron. Manuf., Vol. 6共4兲, 1996, pp. 307–316.
关12兴 Ekere, N. N. and He, D., “The Performance of Vibrating Squeegee in the Stencil
Printing of Solder Pastes,” J. Electron. Manuf., Vol. 6共4兲, 1996, pp. 261–270.
关13兴 Ekere, N. N., Ismail, I., Lo, E. K., and Mannan, S. H., “Experimental Study of
Stencil-Substrate Separation Speed in On-Contact Solder Paste Printing for Re-
flow Soldering,” J. Electron. Manuf., Vol. 3共1兲, 1993, pp. 25–29.
关14兴 Mannan, S. H., Ekere, N. N., Ismail, I., and Currie, M. A., “Computer Simulation of
Solder Paste Flow Part II: Dense Suspension Theory,” J. Electron. Manuf., Vol. 4,
1994a, pp. 149–154.
关15兴 Mannan, S. H., Ekere, N. N., Ismail, I., and Currie, M. A., “Computer Simulation of
Solder Paste Flow Part I: Dense Suspension Theory,” J. Electron. Manuf., Vol. 4,
1994b, pp. 141–147.
关16兴 Durairaj, R., Mallik, S., Seman, A., Marks, A., and Ekere, N. N., “Rheological Char-
acterisation of Sn/Ag/Cu Solder Pastes,” Mater. Des., 2008,.
关17兴 Barnes, H. A., “Thixotropic—A Review,” J. Non-Newtonian Fluid Mech., Vol. 70,
1997, pp. 1–33.
关18兴 Mewis, J. and Wagner, N. J., “Thixotropy,” Adv. Colloid Interface Sci., Vol. 147–148,
2009, pp. 214–227.
关19兴 Mewis, J. and Wagner, N. J., “Current Trend in Suspension Rheology,” J. Non-
Newtonian Fluid Mech., Vol. 157, 2009, pp. 147–150.
关20兴 Koszkul, J. and Nabialek, J., “Viscosity Models in Simulation of the Filling Stage of
the Injection Molding Process,” J. Mater. Process. Technol., Vol. 157–158, 2004, pp.
183–187.
关21兴 Bullard, J. W., Pauli, A. T., Garboczi, E. J., and Martys, N. S., “Comparison of
Viscosity-Concentration Relationships for Emulsion,” J. Colloid Interface Sci., Vol.
330, 2009, pp. 186–193.
关22兴 McLelland, A. R. A., Henderson, N. G., Atkinson, H. V., and Kirkwood, D. H.,
“Anomalous Rheological Behavior of Semi-Solid Alloy Slurries at Low Shear
Rates,” Mater. Sci. Eng., A, Vol. 232, 1997, pp. 110–118.
关23兴 Rao, M. A., Rheology of Fluid and Semisolid Foods: Principle and Applications, 2nd
ed., Springer, New York, 2007, pp. 27–58.
关24兴 Cross, M. M., “Rheology of Non-Newtonian Fluids: A New Flow Equation for
Pseudoplastic Systems,” J. Colloid Sci., Vol. 20, 1965, pp. 417–437.
关25兴 Kirkwood, D. H. and Ward, P. J., “Comment on the Power Law in Rheological
Equations,” Mater. Lett., Vol. 62, 2008, pp. 3981–3983.
关26兴 Mongomery, D. C., Peck, E. A., and Vining, G. G., Introduction to Linear Regression
Analysis, 3rd ed., John Wiley & Sons, Inc., New York, 2001, p. 641.
Overview
Lead containing solders are used extensively in the electronic packaging in-
dustry. The lead based solders have excellent wetting characteristics and
provide good electrical, thermal, and mechanical continuities. However the
lead present in these solders poses significant environmental hazards, such
as the problem of disposal of electronic assemblies, landfill contamination,
and toxicity toward human and wild life. To mitigate these problems, a large
number of lead free solders have been developed and introduced. Although
lead free solders are environmentally friendly, there are several technical
issues, such as-wetting, solder joint reliability, solder joint strength, and
other mechanical properties, which are not fully resolved. This special issue
on lead free solders addresses some of these concerns.
The compendium consists of ten research papers. In the first paper, the
factors affecting the wetting behavior of solders and the evolution of inter-
facial microstructures are reviewed and discussed. The development of Pb-
free high temperature solders for power semiconductor devices is reviewed
in the second paper. The effect of surface roughness on the wetting behavior
and the evolution of microstructures of two lead free solders on copper sub-
strates is discussed in the third paper. A paper by Wang et al. on solder joint
reliability compares the fatigue life of SnBi finished thin-small-outline-
package (TSOP) parts under thermal cycling to that of Sn finished parts.
The paper on microstructural aspects of the ductile-to-brittle transition fo-
cuses on specific aspects of the DBTT in the fracture behavior of tin-based
lead-free solders. The loading mixity on the interfacial failure mode in a
lead-free solder joint is discussed in the sixth paper. The paper by Phil Geng
compares the solder joint strengths of BGA (Ball Grid Array) lead-free to
that of eutectic lead (Sn–Pb) solder joint strengths. The effect of the mor-
phology of Cu6Sn5 intermetallic compounds on tensile properties of bulk
solder and solder joint is discussed in a paper on Tensile properties of Sn-
10Sb-5Cu high temperature lead free solder. Empirical modeling and rheo-
logical characterization of solder pastes used in electronic assemblies are
discussed in the last two papers.
I sincerely thank all the authors for their contributions and sharing their
knowledge. I am indebted to the reviewers who have played an important
role in the preparation of this STP by their constructive comments and sug-
gestions. I deeply appreciate the timely assistance and the excellent coordi-
nation of the review work by ASTM and JAI staff members. It was wonder-
ful working and interacting with them. I am grateful to Dr. George Totten of
GE Totten & Associates, LLC, USA who inspired, encouraged, and initiated
this work. As guest editor, I earnestly hope that this STP on Lead free Sol-
ders will encourage and facilitate further research in the wonderful area of
vii
environmentally friendly lead free solders. This compendium of research pa-
pers should serve as a valuable resource for students, researchers, and ma-
terial scientists in the electronics industry to understand the existing lead-
free solders better and initiate the development of newer solders.
K. Narayan Prabhu
Department of Metallurgical & Materials Engineering
National Institute of Technology Karnataka, Surathkal
Mangalore, India
viii

ASTM STP 1530 - Selected Technical Papers - Narayan Prabhu - ASTM Committee D-2 On Petroleum Products and Lubricants - Lead-Free Solders (2011)

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

ASTM STP 1530 - Selected Technical Papers - Narayan Prabhu - ASTM Committee D-2 On Petroleum Products and Lubricants - Lead-Free Solders (2011)

Încărcat de

Drepturi de autor:

Formate disponibile

Journal of ASTM International

Selected Technical Papers STP 1530

JAI Guest Editor:

Printed in the U.S.A.

ASTM Stock #: STP1530

Copyright © 2011 ASTM INTERNATIONAL, West Conshohocken, PA. All rights

Girish Kumar1 and K. Narayan Prabhu2

Wetting Behavior of Solders

TABLE 1—Potential lead-free candidate alloys 关4,6兴.

Alloy Melting Point 共°C兲 Relative Price Index Remarks

than that of Sn–Pb eutectic. As a result, the soldering temperature is

Common Lead-Free Solder Systems

FIG. 1—Photomicrograph of the Sn-3.5Ag eutectic solder 关3兴.

typical eutectic alloy with eutectic composition of Sn-8.8 wt % Zn and tempera-

FIG. 2—Photomicrograph of the Sn-9Zn eutectic solder 关14兴.

on Cu substrate 关10,15,16兴. These ternary alloys are known to possess good

Wetting and Intermetallics

FIG. 3—Sessile drop on a solid substrate.

␥sv = ␥sl + ␥lv cos ␪ 共1兲

FIG. 4—Wetting and non-wetting conditions.

Intermetallic Compound Formation

FIG. 5—Schematic sketch of a solder/substrate interface 关20兴.

FIG. 6—Photomicrograph of a solder/substrate interface 关21兴.

solder is used on Cu substrate 关24兴. When Sn based solder is in contact with Cu

Factors Affecting Solder Wetting and Evolution of Microstructure

Solder/Substrate Condition Contact Angle Reference

Solder/Substrate Condition Contact Angle Reference

KUMAR AND PRABHU, doi:10.1520/JAI103055 13

the conventional Sn-60Pb is found to decrease the surface tension of solder

FIG. 8—Effect of RE addition on contact angle of the Sn–Ag solder 关43兴.

FIG. 9—Effect of Pr addition on spread area of the Sn–Ag–Cu solder 关44兴.

cos ␪w = r cos ␪ 共2兲

FIG. 10—Effect of temperature on contact angle relaxation of solders 关30兴.

FIG. 12—Stages of spreading of the Sn-37Pb solder on the Cu surface 关8兴.

The phase transformation from Cu6Sn5 to Cu3Sn is a large energy-

atmospheres. Solder wetting involves the metallurgical reactions between the

Plating,” Appl. Surf. Sci., Vol. 214, 2003, pp. 243–258.

UBM-Coated Si-Wafer to Sn-3.5 wt% Ag Solder,” IEEE Trans. Compon. Packag.

Y. Takaku,1 I. Ohnuma,2 Y. Yamada,3 Y. Yagi,3 I. Nakagawa,4

A Review of Pb-Free High-Temperature

ABSTRACT: Pb-base high-temperature solders 共mass % Sn⫽5–10, melting

Increased awareness of the seriousness of environmental issues mandates a

Alloy Design and Properties

CuAlM powder (10~50μm）

Interfacial Reaction between Bi-Base Solders and Substrates

Bi Cu–Al–Mn Bi3Ni Melting Temperature

growth rate constant of which is estimated to be 1.1⫻ 10−6 m · s−1/2. Dybkov

Fig. 1共a兲. Consequently, the Cu substrate is appropriate for Bi-base solders. In

Thermal Cycling Test

TABLE 2—CTEs of each component.

FIG. 9—Optical micrographs of soldered assemblies; 关imitation chip of CIC兴/关No. 1

TABLE 3—Crack length under CTE-mismatched 共chip/solder/Cu substrate兲 soldered assem-

Zn–Al Based Solder

BSE image @ -40oC/250oC 500cycles

FIG. 10—BSE image of an assembly; 关imitation chip of CIC兴/关No. 1 solder兴/关CIC sub-

Al rich fcc + hcp(Zn)

200 250 300 350 400 450

Interfacial Reaction during Soldering and Aging

FIG. 13—Microstructure of Zn-4Al-1Cu alloy soldered on Cu substrate at 450°C for 5

Substrate/Solder IMC 420° C 450° C 500° C 530° C

Estimated values of rate constant at temperatures between 420 and 530° C

(a) Zn-4Al-1Cu/Ni : 420°C - 5min.

(b) Zn-4Al-1Cu/Ni : 450°C - 5min.