Microelectronics Reliability 54 (2014) 1253–1273

Contents lists available at ScienceDirect

Microelectronics Reliability
journal homepage: www.elsevier.com/locate/microrel

Review paper

A review: On the development of low melting temperature

Pb-free solders
Hiren R. Kotadia a,⇑, Philip D. Howes b, Samjid H. Mannan a
a
Department of Physics, King’s College London, Strand, London WC2R 2LS, UK
b
Department of Materials, Imperial College London, Exhibition Road, London SW7 2AZ, UK

a r t i c l e i n f o a b s t r a c t

Article history: Pb-based solders have been the cornerstone technology of electronic interconnections for many decades.
Received 21 August 2013 However, with legislation in the European Union and elsewhere having moved to restrict the use of Pb, it
Received in revised form 20 October 2013 is imperative that new Pb-free solders are developed which can meet the long established benchmarks
Accepted 25 February 2014
set by leaded solders and improve on the current generation of Pb free solders such as SAC105 and
Available online 6 April 2014
SAC305. Although this poses a great challenge to researchers around the world, significant progress is
being made in developing new solder alloys with promising properties. In this review, we discuss
Keywords:
fundamental research activity and its focus on the solidification and interfacial reactions of Sn-based
Soldering
Lead-free solder
solder systems. We first explain the reactions between common base materials, coatings, and metallisa-
Intermetallic compounds tions, and then proceed to more complex systems with additional alloying elements. We also discuss the
Interfacial reactions continued improvement of substrate resistance to attack from molten Sn which will help maintain the
Nanocomposite solders interface stability of interconnections. Finally, we discuss the various studies which have looked at
employing nanoparticles as solder additives, and the future prospects of this field.
Ó 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1253
2. Sn–Ag–Cu (SAC) solder alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1255
2.1. Solidification of SAC alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1255
2.2. Interfacial reactions between the SAC solder and its substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1257
2.3. Effect of alloying elements on solder and IMCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1260
3. Sn–Cu solder alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1264
4. Sn–Ag solder alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1265
5. Sn–Bi solder alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1267
6. Sn–Zn solder alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1267
7. Nanocomposite solders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1269
7.1. Challenges in the processing and manufacturing of nanocomposite solders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1269
8. Outlook and summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1269
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1271
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1271

1. Introduction the electronic age, and it is anticipated that it will remain the pri-
mary assembly and interconnection technology for some time to
The soldering process has been a fundamental aspect in the come. As the durability and reliability of solder joints is absolutely
realisation of all electronic products since the commencement of essential to the functionality and lifespan of an electronic product,
it is key that the solders used are optimised in terms of their phys-
⇑ Corresponding author. Tel.: +44 01895266408. ical and chemical properties to provide robust interconnections. It
E-mail address: hkotadia@gmail.com (H.R. Kotadia). has long been recognised that solder joints embody a potential

http://dx.doi.org/10.1016/j.microrel.2014.02.025
0026-2714/Ó 2014 Elsevier Ltd. All rights reserved.
1254 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273

point of weakness in all electronic products: regardless of the ever
increasing sophistication of modern electronic systems, they will
not function if their component interconnections fail. To date, the
most successful solders have been based on Sn–Pb alloys, and in-
deed their unique combination of chemical, physical, thermal,
and mechanical properties have provided durable and reliable
functionality for many decades [1]. However, due to the environ-
mental and health concerns associated with the use of Pb, legisla-
tion in the European Union and elsewhere has moved to restrict its
use. Despite much resistance on the grounds of cost and reliability
concerns, the transition from traditional Sn–Pb eutectic alloys to
Pb-free solder alloys is occurring [1–3]. As Pb is one of the least
expensive elements on Earth, its replacement will have an intrinsic
cost increase while the incidental costs of developing viable alter-
natives are non-trivial. However, the overriding concern is one of
reliability, and whether the new Pb-free alloys can live up to the
performance level of their traditionally used counterparts.
In this development process, the new solder alloys will be com-
pared with traditional eutectic or near-eutectic Sn–Pb systems, and
will be expected to perform equally well or better in most respects.
Thus, new elemental additions to Sn-based systems should fulfil
the following basic requirements [4–6]:
 Reduce the surface tension of pure Sn to improve the
wettability.
 Enable quick formation of intermetallic compounds (IMCs)
between the solder and substrate by diffusion.
 Improve upon the ductility of Sn.
 Prevent the transformation of b-Sn to a-Sn, which causes
unwanted volume change and degrades structural integrity
and reliability of solder.
 Keep the melting temperature around 183 °C with eutectic or
near-eutectic composition, in instances where the liquid phase
can transform into two or more solid phases.
 Improve mechanical properties (e.g. creep, thermo-mechanical
fatigue, vibration and mechanical shock, sheared and thermal
ageing).
 Prevent the occurrence of excessive tin whisker growth.
The Pb-free solder alloys fulfilling these requirements are more
likely to be multi-component alloys rather than a binary
composition like Sn–Pb, given the greater range of possibilities
with the former. A striking problem with Pb-free Sn-based alloys
is that the liquidus and solidus temperatures are generally either
too high or too low or far apart [1]. Additionally, the mechanical
properties and wettability of such alloys are a concern and must
be improved upon. The addition of further alloying elements pro-
vides an opportunity to control and tune these thermal and
mechanical properties, although this task is non-trivial and re-
quires much research and development. To date, various alloying
elements have been studied, including Ag, Bi, Cd, Cu, In, Sb, Zn,
Al [5,7]. A key point in this work is to ensure that the melting tem-
perature of the solder alloy is not too high, as higher temperatures
would compromise their applicability in constructing electronics
packaging, which often include polymeric materials with low heat
tolerance. If this condition is not met along with those listed above,
a substantial and unreasonable change in soldering practice would
be required [8]. From these considerations, the basic systems of
Sn–Cu, Sn–Ag, and Sn–Ag–Cu have emerged as the front runners
in the replacement of Sn–Pb solders. However, these high-Sn-con-
tent alloys (of around 95–99.3 wt.% [3]) have proved to be prob-
lematic in terms of void formation, large undercooling during
solidification, overly rapid IMC formation and spalling of interfacial
IMCs during high temperature storage. These concerns are listed,
along with the emergent solder alloys, in Table 1 [9,10].
There is a continuous effort to understand the mechanism of the
interfacial interactions – including the formation and growth of
IMCs – occurring at solder/substrate interfaces. IMC layer forma-
tion consists of several distinct physical processes, such as nucle-
ation, growth and coarsening. Among these processes, IMC
nucleation is one of great importance as the particular sequence
of IMCs formed has a significant effect on the morphology and evo-
lution of IMC layers, consequently affecting the reliability of sol-
dered assemblies [5]. The IMC growth problem becomes acute
for applications where an electronics package will experience se-
vere temperature gradients and cyclic mechanical loading through
vibration or shock [9–11]. The interfacial IMC formation between
component/solder/substrate and its failure after aging is shown
schematically in Fig. 1. One frequently utilized way to influence
the interfacial reactions and the properties of product layers in a
given system is to alloy either metallization (conductor) or solders
with small amounts of additional elements. It is noted that the
presence of impurities in the interconnection system may also
have marked effects on the properties and growth of IMC layers

Table 1
Example of Pb-free solder alloys melting range between 109 and 226 °C and concerns.

Alloy system Composition Melting range (°C) Application remarks

Melting temperature below 180 °C
Bi–In Bi–33In (eutectic) 109 Bi content, melting point too low for some applications
Sn–In Sn–52In 118 In adds to cost. Specialized applications for wetting ceramics and glasses
Sn–50In 118–125
Sn–Bi Sn–58Bi (eutectic) 138 Low melting point eutectic solder. Potential segregation problems. Low melting phase with Pb traces
Melting temperature range 180–200 °C
Sn–Bi–In Sn–20Bi–10In 143–193 Replacement candidates for near-eutectic SnPb alloys. Potential segregation and cracking problems
with increasing Bi content. Low melting phase with Pb traces
Sn–Zn–Bi Sn–8Zn3Bi 189–199 Zn imparts poor corrosion resistance and reduced wettability
Sn–Zn Sn–9Zn (eutectic) 198.5
Melting temperature range 180–230 °C
Sn–Ag Sn–3.5Ag (eutectic) 221 Primary replacement candidates for neareutectic SnPb alloys. High melting point
Sn–2Ag 221–226
Sn–Ag–Cu Sn–3.8Ag–0.7Cu (SAC387) 217
(near eutectic)*
Sn3.9Ag0.6Cu 217
Sn–1Ag–0.5Cu (SAC 105) 217
Sn–3Ag–0.5Cu (SAC305) 217
Sn–4Ag–0.5Cu (SAC405) 217
Sn–Cu Sn–0.7Cu (eutectic) 227 Low cost. Plumbing alloy. Poor mechanical properties. Application for wave soldering
*
The exact eutectic composition is unknown.
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
due to their ability to form thin and void-free IMCs at the interface
[7,9,10] and new interfacial IMC barriers that decrease the reac-
tion/growth rate.
The undercooling behaviour of Sn-rich Pb-free solders is also an
important issue. One of the distinct properties of Sn-rich solders is
a tendency for a large amount of undercooling of the b-Sn phase
during solidification [9]. The undercooling required for solidifica-
tion of Sn-rich solder is much larger than high Pb solders or Sn–
Pb eutectic solders. This large undercooling is also attributed to
the growth of large primary phases such as Ag3Sn (in the case of
Sn–Ag–Cu solders). It was also reported that the amount of the
b-Sn undercooling in Sn-rich solders is inversely proportional to
sample size [12]. A large undercooling in flip chip solder bumps
can have a serious bearing on the reliability of solder joints, since
random solidification among many solder bumps can cause a situ-
ation of some bumps being solidified while others not, which can
lead to a stress concentration in some bumps and possibly to early
mechanical failures within solder joints [13]. Impurities in the melt
can act as nucleation sites for crystallisation, which reduces the de-
gree of undercooling and results in finer grains. It has been shown
that trace element additions can be successful in reducing und-
ercooling [14]. For example, it has been suggested that the nucle-
ation of b-Sn due to impurity elements (such as Zn, Fe and Co)
can reduce the undercooling of Sn-rich alloys [14,15]. Hence, a bet-
ter understanding of the solidification behaviour of Sn-rich solder
joints in a small solder volume is essential to successfully imple-
ment Pb-free flips chip technology.
There is naturally a balance which needs to be struck when
introducing trace alloying elements into the Sn-based systems.
Introducing an element which improves one characteristic of the
system might have a detrimental effect upon other characteristics.
For example, adding Zn has the benefit of suppressing excessive
growth of the intermetallic layer during ageing, which improves
the strength of the solder joint. However, the Zn also causes the
wettability of the solder to reduce, which increases surface tension
and lowers the interfacial contact area, which is not good. There is
a danger here that such systems can quickly become very complex
as more elements are added to control specific characteristics of
the solder, thus making the ensemble harder to analyse.
Given the pressing need to develop new high performance Pb-
free solders, it may be time to move the traditional boundaries of
metal alloy investigation into new areas. An opportunity now ex-
ists to use the many advances made in the field of nanotechnology
to benefit the search for the next generation of solders, and may
even allow us to develop new alloys with unprecedented joint reli-
ability, even in harsh conditions such as high-temperature and
high-impact loading. The presence of fine second-phase non-reac-
tive particles can strengthen metal alloys, increasing fatigue resis-
tance and creep strength by serving as obstacles to grain growth
and coarsening of the solder microstructure by exerting a pinning
force on migrating grain boundaries, and by retarding grain bound-
ary sliding.
This review concerns recent fundamental research activity in
Pb-free solders. Much research has been focussed on understand-
ing the solidification and interfacial reactions of Sn-based solder al-
loys upon addition of trace alloying elements to the solder or the
substrate. We explain the reactions between common base materi-
als, coatings, and metallisations such as Cu, Ni, Ni–P (ENIG), Ag,
Ag–Pd, and Au, which all form IMCs with Sn, and then proceed to
more complex systems with the additional alloying elements. Fur-
thermore, we discuss the continued improvement of substrate
resistance to attack from molten Sn which will help maintain the
interface stability of interconnections. Finally in brief, we discuss
the various studies which have looked at employing nanoparticles
as solder additives, and we discuss the future prospects of this
field.
1255
2. Sn–Ag–Cu (SAC) solder alloys
2.1. Solidification of SAC alloys
From amongst the various Sn-rich Pb-free solders, eutectic and
near-eutectic Sn–Ag–Cu (SAC) alloys have been identified as pri-
mary replacement candidates for near-eutectic SnPb alloy. They
are the most widely used solder filler metals because of their ease
of use and relatively slow creep rate, while displaying very prom-
ising joint strength and ductility [16]. SAC alloy performance im-
proves under thermal fatigue tests as the thermal cycling
temperature range decreases [17], and outperform Sn–Pb alloys
under less extreme cycling conditions. Additionally, SAC alloys ap-
pear to be more resistant to gold embrittlement than Sn–Pb solder
[18]. The reliability of soldered assemblies depends on the micro-
structure of interconnections, the evolution of which is strongly
influenced by temperature, stress, and electric current density.
Knowledge of the solidification structure formed in the soldering
operation is, thus, of vital importance in understanding how inter-
connection microstructures evolve over time [19].
The eutectic composition of the SAC ternary system is around
Sn–3.8Ag–0.7Cu [16]. SAC solders are located in the Sn-rich corner
of the Sn–Ag–Cu phase diagram, laying within the composition
range Sn–(2.0–4.0 wt.% Ag)–(0.5–1.0 wt.% Cu) [20]. There are vari-
ous works of interest here [21–26], including by Loomans and Fi-
nen [23] who reported a ternary eutectic temperature of
217.2 ± 0.2 °C, and a composition of 3.5 wt.% Ag and 0.9 wt.% Cu,
and Moon et al. [22] who reported the eutectic composition to be
3.5 wt.% Ag ± 0.3 wt.% Ag and 0.9 wt.% Cu ± 0.2 wt.% at
217.2 °C ± 0.2 °C. The projected ternary phase diagram, with a de-
tailed phase temperature relationship at the Sn-rich corner, is
shown in Fig. 2 [22,27]. A majority of current Pb-free solder alloys
are based in this Sn-rich corner [22]. In the case of near-eutectic al-
loys, several different microstructures can develop during equilib-
rium solidification, dependent upon the overall alloy composition
and the specific topology of the phase diagram. For SAC alloys
these microstructure are [28]: (i) primary Ag3Sn, (ii) primary
Cu6Sn5, (iii) primary b-Sn, (iv) monovariant Ag3Sn + b-Sn, (v)
monovariant Cu6Sn5 + b-Sn, (vi) monovariant Ag3Sn + Cu6Sn5, and
(vii) eutectic b-Sn + Ag3Sn + Cu6Sn5.
Under equilibrium conditions solidification will start when the
alloy cools to a temperature where its overall composition inter-
sects the liquidus projection of the system, at which point the solid
microconstituents described above will form. This process will
shift the liquid to a new composition and lower temperature,
continuing with a changing composition, and decreasing volume
fraction and temperature. During this process, the formation of
microconstituents evolves up to a point where the composition
of the remaining liquid will correspond to the eutectic composition
at the eutectic temperature. Further cooling of this liquid will lead
to solidification as a ternary eutectic.
The microconstituents present in SAC alloys commonly exhibit
morphologies such as Ag3Sn plates, hollow hexagonal Cu6Sn5 nee-
dles, and nonfaceted b-Sn dendrites for the primary phases and
sometimes in the monovariant microconstituents. Additionally,
the b-Sn matrix can be embedded with rod-like Ag3Sn and/or
Cu6Sn5 particles for the eutectic or monovariant microconstituents
(Fig. 3a) [29]. Monovariant two phase microconstituents are com-
monly eutectic-like, an indication of coupled growth. However,
such coupled growth is inhibited by kinetic factors when both
phases exhibit strong faceting tendencies. This would be expected
for monovariant Ag3Sn + Cu6Sn5. For un-coupled growth the mono-
variant microconstituents can often resemble primary phases [28].
With significant undercooling, the plates exhibit strong growth in
the liquid phase before the solder joint finally forms. In contrast,
1256 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273

Fig. 1. A schematic drawing showing: (A) flip-chip solder joints, (B) solder joint formation and failure, and (c) formation of barrier layer after addition of elements (X).

Fig. 2. (a) Ternary phase diagram showing the Sn–Ag–Cu ternary eutectic reaction and (b) Sn-rich corner [22,27].

Fig. 3. Optical micrographs of Sn–3.8Ag–0.7Cu (SAC) solder alloy: (A) overview of the microstructure of ingot, and (B) on Cu substrate.
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
the Cu6Sn5 phase tends to grow as fibres, which generally form
with a pure Sn core and a hexagonal cross-section. As a result,
the amount of liquid that can transform to solid per unit time
(for a given undercooling) is limited. This arises as the Ag3Sn and
Cu6Sn5 phases act to restrict one another’s growth, which increases
the solidification time of the solder [24]. It is well know that b-Sn
has difficulty nucleating during cooling from the process tempera-
ture, which leads to formation of a metastable phase. Consequently
the kinetics of b-Sn formation in most Sn-based liquid metal alloys
are unique among common metal alloys [22].
The final solidification of the SAC microstructure is dependent
upon the cooling rate, solder volume and solder composition. The
relationship between these parameters is very complex as they
are interdependent. For example, experimental results show that
the microstructure becomes finer as solder volume is decreased,
and experiments with bulk solder ingots have shown that signifi-
cant changes in the microstructure as the cooling rate is varied.
Increasing the cooling rate reduces the size of intermetallics, de-
creases the secondary dendrite arm spacing of b-Sn and increases
the area of small IMC precipitations [30]. The influence that the
cooling rate exerts upon the grain structure is less significant as
solder volume is decreased, although the formation of small, ran-
domly orientated grains becomes more frequent [30].
The cooling rate is a critical factor in the formation of b-Sn den-
drites and Ag3Sn plates in SAC solder joints. If the cooling rate is
high, the Ag3Sn plates do not have sufficient time to achieve signif-
icant growth even if they had nucleated. Large plates are not com-
monly seen in as-received solder balls, or in rapidly cooled solder
joints (at a rate of 1 °C/s or higher). Therefore, one can see that,
by controlling the cooling rate, the formation of Ag3Sn plates can
be kinetically controlled as well. In slowly cooled alloys, the forma-
tion of large Ag3Sn plates reduces the Ag content to approximately
2 wt.% regardless of the initial Ag content [31]. In rapidly cooled al-
loys the formation of large Ag3Sn plates is avoided. However, using
a high cooling rate is not always practical, especially when a sub-
strate with a high thermal mass is used. Additionally, a high cool-
ing rate can have some unwanted side effects such as thermal
stress or strain in the substrate. It is interesting to note here that
the microhardness of slowly cooled alloys is more or less indepen-
dent of the SAC alloy content.
The solder composition is an important factor in determining
solidification behaviour. For example, the composition has a strong
influence on the growth of primary intermetallics. If the Ag content
is above 3.5 wt.%, large primary Ag3Sn intermetallics form. Below
3 wt.% Ag, no primary Ag3Sn plates form, regardless of the cooling
rate. If the Cu content is above 1 wt.%, the solder solidifies with
large primary Cu6Sn5 intermetallics [30]. Experiments designed
to study directional solidification have revealed that the growth
of b-Sn dendrites can be directed by temperature gradients inside
the solder [30]. Solder composition also has an effect on the grain
structure of the solidified alloy. Even small changes in composition
will determine whether the solder is composed of a smaller num-
ber of large grains, or with many smaller grains. A second strategy
for refining microstructure of Sn-based solder is to promote rapid
heterogeneous nucleation by addition of inoculants.
2.2. Interfacial reactions between the SAC solder and its substrate
The reaction between a molten solder and its metal substrate is
of paramount importance for all solder technologies. This reaction
gives rise to IMCs whose nucleation and growth define the overall
reliability of the solder joint. Many studies have focussed on the
chemical reactions of Sn, specifically with Cu and Ni substrates, as
it is of major importance in Pb-free solders [3,4,6,32]. The reaction
of Sn with such compounds induces the formation of IMCs such as
Cu6Sn5, Cu3Sn and Ni3Sn4, by heterogeneous nucleation and growth
1257
at the substrate–liquid interface [3]. As Cu is the most common
metal conductor used in electronics, it is imperative that all solder
alloys can form strong and stable joints with it. Cu generally exhibits
very good solderability characteristics provided that it is protected
from oxidation by a thin organic layer or a Sn coating.
Fig. 4a and b shows an SEM image of a typical SAC/Cu interfacial
joint, and the IMCs which form during reflow can be clearly seen.
Upon diffusion of Cu into the liquid solder, scallop shaped
g-Cu6Sn5 IMCs form at the Cu/liquid interface by heterogeneous
nucleation during reflow. Theoretically, one would expect to ob-
serve a layer of Cu3Sn between the Cu and Cu6Sn5 IMC, but this
layer is often too thin to experimentally observe directly after re-
flow. The Cu6Sn5 IMC forms first because its formation has a lower
activation energy than the Cu3Sn mechanism. The Sn/Cu reaction
can be predicted through consideration of the thermodynamic
behaviour of the system (Fig. 4d), specifically from the Gibbs free
energy curves for the liquid and the FCC Cu phase, and the Gibbs
energy of formation of the Cu–Sn compounds at 235 °C [33]. The
primary factors which determine the thickness and morphology
of the reaction layer are the rate of dissolution of Cu into the liquid,
and the chemical reaction between the Sn and Cu. The rate of dif-
fusion of the Cu through the liquid also plays an important role
here, and therefore its metastable solubility in the liquid solder is
a crucial factor. This rate ultimately determines the extent and
morphology of the two-phase zone on top of the uniphase layer.
As the SAC–Cu reaction continues, the Cu6Sn5 IMC layer be-
comes increasingly thermodynamically unstable on the Cu sub-
strate. As this instability increases, the Cu3Sn IMC layer begins to
form (Fig. 4c) by consuming the Cu6Sn5 IMC layer, proceeding as
Cu6 Sn5 ! 2Cu3 Sn þ 3Sn. During ageing the Cu3Sn layer grows sig-
nificantly between the Cu and Cu6Sn5, and therefore plays a critical
role in determining the reliability of the solder joint. The behaviour
of Cu3Sn during its growth is strongly dependent on the diffusion
of the dominant reactive species and on the reaction pathway.
Over time, a large amount of voids form in the Cu3Sn layer, either
by the Kirkendall effect or by solute segregation [3].
As time proceeds, the Cu6Sn5 phase becomes the majority com-
ponent of the solder joint as it appears in the initial stages of IMC
formation and grows rapidly. It has been suggested by some
researchers that a process of coarsening is taking place here,
[34,35] however the IMC coarsening process is still to be fully
understood. A two-flux nonconservative Ostwald ripening model
was proposed by Kim et al. [34] and Kim and Tu [36]. This model
stems from the idea that the rapid growth of the Cu6Sn5 in the li-
quid is due to the dissolved Cu precipitating in the liquid as the
scallop-type Cu6Sn5 and then coarsening by Ostwald ripening
[34]. A further model was proposed by Schaefer et al. [37] who de-
scribed the growth kinetics of an IMC layer in a saturated solder.
Their model assumed that grain boundary diffusion is the primary
transport mechanism through the IMC layer, and that the IMC
grains are equiaxed throughout the growth layer. However,
although elements of this approach were in agreement with exper-
imental observation, [38] their ascertion that the scallops grains
are equiaxed did not address the driving force for coarsening. Over-
all, an advanced description of the IMC structure coarsening is
impossible without a clear picture of how the scallop IMCs form
and grow, which has led many researchers to study this particular
process [39–41]. Unfortunately there has been no solid proof pre-
sented to date that this mechanism has been truly understood.
Some studies have dismissed the suggestion that the IMC interfa-
cial growth process is rate-limiting, based on the kinetics of the
thickening of the IMC phase [38]. Control of diffusion yields a time
exponent of 1/2, while experimental results have shown results
close to 1/3 [38]. Literature values range from 1/5 to 1 [42] depend-
ing on temperature, time interval, and solder composition. The dif-
fusion coefficients can be expresses as a function of temperature as
1258 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273

Fig. 4. SEM images SAC/Cu substrate: (A) after reflow 260 °C or 60 s, (B) scallop shaped IMC morphology etched with 17% concentrated nitric acid, (C) ageing at 150 °C for
100 h, and (D) Gibbs energies of formations of the Cu–Sn intermetallic compounds at 235 °C [33].

Fig. 5. SEM images SAC/ENIG substrate: (A) after reflow 260 °C or 60 s, (B) needle-shaped IMC morphology etched with 17% concentrated nitric acid, (C) ageing at 150 °C for
1000 h, and (D) Gibbs free energy diagram at 235 °C with respect to the Ni mole for formation of various IMCs [5].
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
 
Q leads to an higher rate of diffusion of Sn atoms in the system. A de-
D ¼ D0 exp
RT
where D is the diffusion coefficient, D0 is the intrinsic diffusivity of
the material, Q is the activation energy for diffusion, R is the gas
constant and T is the temperature in Kelvin. The consequence of this
T dependence is that IMC growth is greatly accelerated at increased
temperature.
Although Cu is the most common substrate for solder joints,
there are some alternative substrate compounds which offer better
performance. A Ni (Sn coated) or an electroless Au/Ni(P)/Cu (ENIG)
substrate can be used as Printed Circuit Board (PCB) and compo-
nent metallisation as they slow down the interfacial reactions be-
tween the molten solder and component and substrate
metallization layers. When a Sn rich solder reacts with the Ni sub-
strate, generally below 260 °C, heterogeneous Ni3Sn4 IMC nucle-
ates at the interface and proceeds to grow [4]. Initially, the Ni
dissolves from the solid substrate into the liquid Sn solder, result-
ing in the reaction which forms the Ni3Sn4, [43,44] and then disso-
lution ceases. The rate of dissolution of the Ni into the liquid solder
is much lower than, for example, those of Cu or Au [45]. This means
that the Ni3Sn4 IMC layer is relatively thin compared to analogous
systems [4,5]. There are two other Ni–Sn IMCs formed in this pro-
cess – Ni3Sn2 and Ni3Sn – but these grow much slower, with Ni3Sn
particularly struggling to nucleate at the solder/Ni interface. For Sn
solders containing a small amount of Cu (e.g. SAC), their reaction
with the Ni substrate results in IMCs of either (Cu,Ni)3Sn4 or
(Cu,Ni)6Sn5 (Fig. 5a). The (Ni,Cu)6Sn5 IMC, with its distinctive nee-
dle-shaped morphology, is shown in Fig. 5b. At a Cu concentration
above 0.7 wt.%, a (Ni,Cu)6Sn5 IMC compound forms at the Ni inter-
face, based on the Cu6Sn5 structure. Fig. 5C shows Ni–Sn IMCs
growth with time and temperature, whilst Fig. 5d shows the Gibbs
free energy diagram for the formation of various IMCs, with respect
to Ni, at 235 °C. It is important to note that the Ni–Sn IMCs are
more stable than the corresponding Cu–Sn IMCs.
Electroless Ni(P) substrate has a relatively slow reaction rate
with solders compared with the pure Ni substrate, therefore the
addition of other alloying elements to the reaction between Ni
and Sn is of technological importance [3,4,46–48]. However, as
the presence of P complicates the reactions, the details of the rele-
vant reaction mechanisms are still unclear. In general, a layer of
stable Ni3P forms between the Ni–Sn IMC and the substrate, which
effects the growth of the Ni–Sn IMC phase. It was reported by Hung
et al. [46,49] that the Ni3P layer can act as a barrier to the diffusion
of Ni, and consequently suppress the growth of Ni3Sn4, which in
turn stops the growth of Ni3P. Consequently, the P liberated in this
selective reaction becomes enriched at the liquid–substrate inter-
face, leading to a recrystallisation reaction which forms Ni3P, a sta-
ble compound [50]. The implication of this is that the Ni
concentration in the molten Sn is much lower than the amount re-
quired for the formation of Ni–Sn IMCs. However, it was suggested
by He et al. [50] that the Ni–Sn layer acts as a protective barrier
against direct attack from molten Sn. This would cause the Sn to
penetrate along microstructural defects, eventually leading to the
decomposition of the Ni3P which crumbles as particles into the
molten Sn. This would lead, in turn, to the formation of a ternary
layer of Ni–Sn–P, approximately 0.5 lm thick. An alternative
explanation for the formation of the Ni–Sn–P layer is that Sn diffu-
sion along the grain boundaries between the fine Ni3P particles,
This would provide a rapid path for diffusion, especially when all
of the Ni–P has reacted and therefore ceases to supply Ni atoms
from underneath the Ni3P layer to react with the Sn in the solder
[51]. Experimental studies have revealed that the initial growth
of IMCs on the Ni–P substrate is lower compared to Ni substrates.
However, if the molten reaction time is long, breaking and diffu-
sion of the IMCs and an increased formation of Kirkendall voids
1259
ð1Þ
tailed explanation of the phase formation, with the relevant ther-
modynamic calculations, is given in Ref. [4].
It is desirable that the resistance of electroless Ni–P to attack
from molten Sn is improved in order to maintain interface stability
of liquid solder joints. A possible option here is to introduce some
high melting point alloying elements into the substrate surface
layer, such as Cr, Pt, Ti, V, Nb, Ta, and W [11]. As the reaction be-
tween the liquid solder and the substrate proceeds, these alloying
elements become enriched at the surface and form tightly bound
barrier layers to slow down the diffusion of Ni and Sn atoms. For
example, it has been demonstrated that the use of electroless Ni–
W–P substrates greatly extends the service life of the barrier
metallisation, and therefore the solder joint [52]. Recently, a Au/
Ni–Zn/Ti and Au/Ni/Ti UBM was developed for Pb-free solders by
Hae-Young et al. [53], who observed that Ni3Sn4 IMC forms at
the interface. Their results showed that the growth rate of the
interfacial IMC which formed on the Ni–Zn layer was slower than
on the Ni layer.
A SAC/Ag system, studied by Wu et al. [54], showed an Ag3Sn
IMC layer at the substrate interface and formation of Cu6Sn5 parti-
cle precipitates in the solder matrix. Through kinetic analyses it
was determined that growth of the Ag3Sn IMC was diffusion con-
trolled (Fig. 6a).
Kim et al. [55] studied a SAC/(Pt/Ti/Si) system and showed that
PtSn4 IMC forms at the substrate interface, and that the intermetallic
formation at the solder/Pt interface is diffusion controlled. From the
Pt–Sn phase diagram [56], it can be seen that five stable IMCs (Pt3Sn,
PtSn, Pt2Sn3, PtSn2, and PtSn4) are likely to form at the Pt-based sub-
strate. However, the main reaction product observed experimen-
tally is PtSn4, which even forms when the solder contains 1.7 wt.%
Cu. The dissolution rate of Pt into the molten Sn (between 250 and
310 °C), was smaller than that of Ni into Sn. However, although
the wettability on Pt is inferior to that on Ni, it is still acceptable
for UBM and surface finish applications (Fig. 6b).
For pure Co substrates, the IMC CoSn3 is formed at the substrate
interface [57,58]. Wang et al. [57] have observed that the growth
rate of the CoSn3 phase is reaction controlled in the early stages,
and then becomes diffusion controlled when the layer thickness
reaches a critical value. The Cu6Sn5 phase (generally labelled as
the (Cu,Co)6Sn5 phase for Co substrates) only forms when Cu atoms
are available in the system. Considering the thermodynamics of the
system, they see that the (Cu,Co)6Sn5 is significantly stabilised by
Co solubility, so it has a relatively low Gibbs free energy compared
to Cu6Sn5. The (Cu,Co)6Sn5 layer, which is formed between the Sn
and CoSn3, acts as a diffusion barrier against the diffusion of Sn,
the dominant diffusing species in the formation of CoSn3. It is still
unclear why the CoSn3 growth is inhibited by the Cu content of the
solders, and more studies are required to properly resolve this is-
sue. Substrates manufactured from Co and Ni have been studied
by Huang et al. [59] For Ni-63Co substrates, (Co,Ni,Cu)Sn3 is the
only interfacial compound phase formed at the solder/substrate
interface. Increasing Cu content in the solder suppresses the
growth of this interfacial IMC (Fig. 6c).
In SAC/Au systems, the Au dissolves and diffuses throughout the
molten SAC solder whilst reacting with the Sn to form a diffusion
layer of AuSn2, AuSn4, or a combination of the two [60]. During
solidification the saturated Au solution precipitates as AuSn4, as
can be predicted from the Au–Sn binary phase diagram. Interest-
ingly, no traces of Ag3Sn or Cu are found in the IMC.
For Fe substrates, the reaction between the SAC and the Fe leads
to the formation of FeSn2 IMC after reflow. The thickness of this
IMC layer has been observed to be lower than the Ni–Sn IMCs on
a Ni/P substrate (Fig. 6d) [61].
A substrate of Cu–Mn was studied by Tseng et al. [62] with suf-
ficient amounts of Mn deposited on the Cu substrate, the interfacial
1260 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273

Fig. 6. SEM image of the reaction between SAC solder and various substrates: (A) Ag [54], (B) Pt [55], (C) Co [57], (D) Fe [61], (E) Cu–Mn [62], and (F) Cu–Zn [63].

Table 2 Cu3Sn IMCs, with a small amount of Zn diffused within, were seen
Interfacial IMC layers between SAC/Substrate after reflow. to form at the interface, while Kirkendall void formation was com-
Substrate IMCs Reference pletely absent. The growth rate of these IMCs was relatively slow
Cu Cu6Sn5 [3,4,6,32]
because of their large activation energies and low layer growth
Cu3Sn coefficients, with the effect becoming more prominent at higher
Ni (Cu,Ni)3Sn4 or [4,43,44,50] ageing temperatures (Fig. 6f). Table 2 lists the formation of interfa-
(Cu, Ni)6Sn5 cial IMCs on various substrates.
Au/Ni–P/Cu (ENIG) (Cu,Ni)3Sn4 or (Cu, Ni)6Sn5 [22,25,50–52]
Ni3P
Ni–Sn–P
Ag Ag3Sn [54] 2.3. Effect of alloying elements on solder and IMCs
Pt/Ti/Si PtSn4 [55,56]
Co CoSn3 [57,58] SAC solder alloys show promising ductility and joint strength,
Au AuSn2 or AuSn4 or both [60] and are thus quickly becoming the preferred Pb-free solders for
Fe FeSn2 [61]
Cu–Mn Cu6Sn5 and Cu–Mn [62]
electronic assemblies, especially for surface mount technology
Cu–Zn (Cu,Zn)6Sn5 [63] [3,31,64,65]. It has been suggested that high temperature ageing
(Cu,Zn)3Sn (approx. 1000 h at 150 °C or above) may promote void formation
and coalescence in Cu3Sn IMCs, leading to crack formation at the
solder joint interface [9,16,66,67]. If this coalescence occurs at low-
IMCs are Cu6Sn5 and Cu–Mn. The formation of Kirkendall voids and er temperatures with longer exposure times, joint embrittlement
growth of Cu3Sn was significantly suppressed during high temper- may occur, particularly if conditions favour electromigration. The
ature storage by the creation of an Mn enriched Cu3Sn diffusion result is that there is an immediate requirement for novel high per-
barrier. These results suggested that this Cu–Mn substrate, with formance SAC solders with an added solute (denoted X hereafter)
a low Mn concentration, may be useful for slowing down the con- to effectively suppress void formation. It has been shown that min-
sumption of the Cu substrate in the formation of solder joints or alloy substitutions for Cu (e.g. X = Co, Fe, Zn, Ni, Al, etc.) in these
(Fig. 6e). SAC + X alloys promotes IMC segregation and effectively eliminates
The growth kinetics of IMCs formed between SAC solders and a joint embrittlement [16]. However, it is known that embrittlement
Cu–Zn substrate was studied by Young Min et al. [63] Cu6Sn5 and can be attributed to other phenomena. For example, it can occur in
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
as-solidified SAC solder joints that are cooled from the peak reflow
temperature at a rate slower than approx. 1 °C/s, a rate typical of
surface mount technologies [31]. Such a cooling rate is common-
place in ball-grid array (BGA) reflow, and is attributable to the fact
that the pro-eutectic Ag3Sn intermetallic phase can nucleate and
grow, commonly from the Cu6Sn5 IMC layer of the joint. The forma-
tion of these large plates can lead to crack formation and joint fail-
ure at the early stages of cyclic loading due to the weak interfacial
binding between the Ag3Sn and the Sn matrix [31].
In an attempt to thermodynamically suppress the formation of
large Ag3Sn plates in SAC alloys, Ag content has been reduced to
2 wt.% with Cu fixed at 0.9 wt.% [31]. The formation of large Ag3Sn
plates has been substantially reduced in SAC alloys with an Ag con-
tent of less than 3 wt.%, even in extremely after slow cooling (i.e.
around 0.02 °C/s) [31]. Consideration of the thermodynamics of
this system can help us explain this apparent dependence on Ag
content. According to the phase diagram of Sn–XAg–0.7Cu, [31]
the position of the Ag3Sn liquidus is in the metastable region,
which we deduce by direct extrapolation of the liquidus from the
ternary eutectic temperature. As mentioned previously, the solidi-
fication of Sn–3.8Ag–0.7Cu requires an undercooling of 15–30 °C
before the Sn phase can solidify. The availability of nucleation sites
dictates the metastability of the liquid Sn phase. With Cu at
0.7 wt.% the composition of liquidus which intersects at 20 °C und-
ercooling is approximately 2.7 wt.% Ag. This Ag content defines a
threshold below which the Ag3Sn cannot nucleate and grow,
assuming undercooling is less than 20 °C. This ultimately means
that suppression of Ag3Sn plate formation can be performed inde-
pendently of cooling rate.
It appears that the formation of large Ag3Sn plates is linked with
the level of Cu in the solder. Although this effect is not all together
well understood, a key piece of evidence was reported by Moon,
et al. [22] who found that the Cu content of the large Ag3Sn plates
exceeds that of b-Sn dendrites (about 0.36 wt.% Cu versus 0.16 wt.%
Cu). This is a potential explanation for why there is no difference in
the population of large Ag3Sn plates observed for different Cu com-
positions (such as 0.35 vs. 0.7 wt%), given that only a low number
of Cu atoms are required to nucleate an Ag3Sn plate in an under-
cooled SAC alloy [68]. Additionally, increasing the amount of Cu
in the system can cause excessive precipitation of Cu6Sn5 particles
near the interface, which may then harden the solder joint.
The Cu content has additional effects upon the solidification
range of SAC alloy solder joints. We define the solidification range
as the difference between the liquidus (TL) and the solidus temper-
atures (TS). In Table 3, DT is shown for two Cu concentrations (0.7
and 0.9 wt.%) and five Ag concentrations (2.1–2.9 wt.%) [31]. We
can see that DT is highly dependent upon Cu content, but less so
on Ag, in the hypoeutectic Ag composition range. A possible conse-
quence of the large solidification range is likely to be a significant
increase in the solder defect rate (depending upon the specific
application), thus we may conclude that the Cu content of SAC al-
loys should be no higher than 0.9 wt.% [31].
This work by Kang et al. [31] nicely illustrates the benefits of re-
duced Ag content in SAC alloys (below 3 wt.% Ag), especially with
regard to minimising the growth of large Ag3Sn plates. However,
Table 3
A calculation of solidification range variations as a function of Ag and Cu in SAC [31].
Alloy composition (wt.%) Solidification range (DT = Tl  Ts)
0.7% Cu 0.9% Cu
Sn–2.1Ag–Cu 3.5 15.7
Sn–2.3Ag–Cu 3.3 17.3
Sn–2.5Ag–Cu 2.6 17.3
Sn–2.7Ag–Cu 2.0 17.3
Sn–2.9Ag-Cu 1.6 17.3
1261
3.5
Pt Mo W
Ir
S Ru Rh
To Re
Ge
3.0
Cr
Mn(7) Pd Au
Electronegativity
2.5
P Sn
2.0 Ni
Co
Fe
1.5
Nb
1.0 Mn(5)
Si V Zn
Co Ga Al Ti
0.5
0.75 0.95 1.15 1.35 1.55 1.75 1.95
Atomic radius (metallic)
Fig. 7. A Darken–Gurry ellipse plot with Cu as the central atom, where the elements
within the ellipse are likely to exhibit extensive solid solubility in Cu (up to 5 at.%)
[67].
although it is observed that growth of large Ag3Sn plates appears
to be less sensitive to Cu content than Ag content, it is additionally
noted that high Cu content can lead to the growth of Cu6Sn5 rods.
This is particularly prevalent when Cu substrates are used. Further-
more, a high Cu content has an additional effect of increasing the
solidification range, as noted above. Therefore, by reducing both
Cu and Ag content, one can expect to observe the beneficial effects
of suppressing the growth of both Ag3Sn plates and Cu6Sn5 parti-
cles, and decreasing the solidification range. This will have the
overall effect of increasing reliability of the SAC solder joint.
Considering the eutectic SAC solder alloy it is important to iden-
tify the selection criteria for the addition of a fourth metallic ele-
ment (beyond Co and Fe) into a SAC + X system. Anderson et al.
[67] have discussed including the electronegativity of the elements
as an essential criterion, in addition to choosing elements with a
similar atomic radius to Cu. Fig. 7 shows a ‘Darken–Gurry ellipse’
which provides a combined analysis of electronegativity and atom-
ic radius, allowing one to predict which elements would dissolve in
solid solution to around 5 at.% in a solvent element. In Fig. 7, the
solvating element is Cu, and is therefore placed at the centre of
the ellipse, and it is seen that all of the elements used in the anal-
ysis lie within the ellipse (Si, Ti, Cr, Mn, Ni, Ge, and Zn) with the
exception of Ti. However, it is known from other sources (specifi-
cally, the Cu–Ti phase diagram) that Ti will dissolve in Cu up to
8 at.%, so it can be included in the list of potential alloying ele-
ments. The resultant list of elements then includes ones with
atomic radii both larger and smaller than that of Cu. The solubility
of these alloying elements in Cu is significant because it is consid-
ered that by inhibiting the interdiffusion of Sn and Cu, void forma-
tion and coalescence at the Cu3Sn/Cu interface in joints after high
temperature ageing can be suppressed. By closely matching Cu
with the alloying element, one can promote substitution of X into
the Cu3Sn and Cu6Sn5 IMCs, a move which is intended to increase
lattice strain and reduce vacancy diffusion, therefore slowing the
layer growth and associated void formation. In order to test the
assertions made here one would need to make detailed observa-
tions of the time-dependent concentration gradients of Cu and Sn
at various isothermal ageing temperatures. However, in practice
this is a complicated measurement to make as there exists only a
very narrow interface region [67].
It is interesting to note here the influence of Zn upon addition
to the SAC solder system, particularly with regard to the forma-
tion of the IMC interface layer. In SAC/Cu joints, one typically ob-
serves Cu6Sn5 IMC at the interface, with Cu3Sn arising after
thermal ageing [7,9,15,69,70]. The growth of the Cu3Sn IMC is
suppressed after the addition of 0.2 wt.% Zn to the SAC system,
with the Cu6Sn IMC being the only IMC to form [15]. Increasing
1262 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
Fig. 8. SEM images showing the interfacial microstructures for different systems on Cu (FR4) substrate, after reflow at 260 °C for 60 s: (A) SAC, (B) SAC–1.5Zn, (C) SAC–3Zn,
and (D) SAC–5Zn. The solder mass was 0.01 g [10].
Fig. 9. The Sn-rich corner of the ternary phase diagram of the Sn–Cu–Zn system
[72,10]. Interfacial phase changes with Zn concentration are marked by an arrow.
the Zn to 1.5–5 wt.% sees the formation of Cu–Zn interfacial IMCs
at various locations across the Cu substrate [10]. The thermody-
namics of this sytem indicate that, upon addition of the Zn, the
driving force for the formation of these Cu–Zn IMCs is likely to
be higher than for the Cu–Sn. When a molten solder wets a Cu
substrate, there is a metastable equilibrium state at the solder/
Cu interface. The first IMC to form should be the one with the
highest driving force for formation, and in the SAC–Zn system
(when Zn is less than 1.5 wt.%), the Cu6Sn5 phase possesses the
largest driving force. However, if the wt.% of Zn is increased,
the formation of Cu–Zn compounds becomes more energetically
favourable and they form first [9,10].
The stability of Cu–Zn IMCs depends on the availability of Zn
from the solder during reflow [7,9,10,71]. At this point, the con-
sumption of Zn rapidly changes the local equilibrium at the inter-
face, from Sn + Cu–Zn to Sn + Cu6Sn5, and the Cu–Zn/Cu interface
becomes unstable (Fig. 8). This instability arises because the nature
of the interface changes: it goes from being chemically bonded to
being physically contacted, resulting in the IMC spalling off and a
new thermodynamically stable Cu6Sn5 IMC growing at the interface.
This phenomenon is called massive spalling, an example of which
can be seen in Fig. 8b; an effect which is also dependent on the vol-
ume of the solder joint [10]. When the Zn content is above around
2 wt.% the molten solder equilibrates with Cu5Zn8: below 2 wt.%,
it equilibrates with Cu6Sn5. This behaviour is illustrated in Fig. 9,
which shows the Sn–Zn–Cu isothermal section at 250 °C [72]. Look-
ing at Fig. 9, we see that the reaction between the molten solder and
the Cu substrate should produce different IMCs at the interface if the
Zn content of the solder is changed. The solid phase (Sn) has tie-lines
with the CuZn5, Cu5Zn8 and Cu–Sn phases, respectively, with
decreasing Zn concentration in the Sn phase, and it was found by
Kotadia et al. [9] that the Cu–Zn IMC layer acts as a barrier layer
for Cu6Sn5 IMC (Fig. 8c). It has been shown that the formation of
interfacial IMCs are different than that of the solder ball (for small
volumes of solder), even though the solder compositions are the
same (Fig. 10). Fig. 11 shows the thickness kinetics of Cu3Sn and
(Cu,Zn)6Sn5 plotted as the average thickness versus storage time
Fig. 9, and we can see that the Cu3Sn IMC layer in SAC–Zn/Cu sys-
tems appears after 200 h in high temperature storage. In such sys-
tems, the Cu3Sn IMC layer grew only up to 1 lm (even after
1000 h). Furthermore, many Kirkendall voids are observed at the
Cu3Sn/Cu interface, however these appear to have been completely
suppressed by the addition of Zn [9].
Various studies have observed that Cu6Sn5 and Cu3Sn IMC for-
mation is significantly suppressed when the concentration of Zn
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273 1263

Fig. 10. SEM images showing the interfacial microstructures for (A) SAC–1.5Zn/Cu, as reflowed and (B) SAC–1.5Zn/Cu after 1500 h storage at 150 °C, (C) SAC/Cu as reflowed,
and (D) SAC/Cu after 1500 h storage at 150 °C. Solder reflow was carried out in a solder bath (30 g) at 260 °C for 60 s [10].

Fig. 11. Plots of mean IMC thickness versus storage time: (A) thickness of Cu3Sn IMC layer, and (B) thickness of Cu6Sn5 and (Cu,Zn)6Sn5 IMC layers [9].

is around 1 wt.% [7,9,73] This effect may be attributable to the for-
mation of a Zn-rich band, as suggested by Cho et al. [69], Yu et al.
[74] and Pan and Lin [75]. This group showed that this band would
form near the Cu substrate as a result of the interaction between
the Zn containing solder and the Cu substrate. This band would
then restrict the interdiffusion of Sn and Cu atoms, and conse-
quently suppress growth of the Cu–Sn IMC layer.
It is seen that the thickness of the IMC layer will decrease signif-
icantly with the addition of only a small amount of Zn [9]. If larger
amounts of Zn are added, SAC alloys exhibit problems with wetta-
bility, discontinuity of the IMC and large amounts of Cu–Zn parti-
cles forming in the matrix, amongst others. High temperature
storage data indicate that Zn is actually present at small Zn concen-
trations in the Cu6Sn5 layer, so it may be affecting the diffusion of
Sn, even with only trace amounts of Zn (2–3 wt.%).
In contrast to the SAC–Zn/Cu system, the SAC–Zn/Ni(P) system
does not exhibit massive spalling of an Ni–Zn IMC. The presence of
Zn in the solder increases the wetting angle, which can be attrib-
uted to the oxidation of Zn atoms at the molten solder interface.
On the Cu substrate the wetting angle increases by approximately
10° for every 0.5% incremental increase of Zn concentration. On the
Ni–P substrate, in contrast, the increase in wetting angle is only
1.5° (Fig. 12) [9].
Solder wettability results for SAC–Ni systems have revealed
that varying the Ni content does not adversely affect the spread
area, in comparison with the non-composite solders [76]. Further-
more, the SAC–0.5Ni and SAC–1Ni exhibited improved wettability,
and they spread well on the substrate. The addition of Ni produced
much thinner layers of Cu3Sn at all Ni concentrations: above
0.01 wt.%, the growth of Cu3Sn is effectively slowed, even after
1264 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
Fig. 12. SAC–Zn solder wetting angle with respect to Zn concentration in the SAC
solder, on Cu and Ni–P substrates.
2000 h ageing, so 0.01 wt.% is considered the minimum effective Ni
concentration. The addition of Ni led to much thinner layers of
Cu3Sn at all Ni concentrations studies. A Ni concentration higher
than 0.01 wt.% was shown to effectively slow the Cu3Sn growth,
even after 2000 h of ageing, therefore the 0.01 wt.% addition can
be considered the minimum effective amount of Ni [76].
The addition of Co and Ni into the SAC solder system yields
changes in both the bulk mechanical and microstructural properties
[77]. Adding only Co leads to the formation of CoSn2 IMCs and sup-
presses undercooling during the solidification process. Adding only
Ni increases the formation of rod-shaped (Cu,Ni)6Sn5 precipitates,
and mildly suppresses undercooling. Neither of these single addi-
tions results in effective refinement of the b-Sn grains. However,
dual addition of Co and Ni results in refined b-Sn grains, suppresses
undercooling, and suppresses the formation of CoSn2 IMCs.
The addition of 1 wt.% Bi into a SAC solder system acts to control
the nucleation of Cu6Sn5 and Ag3Sn IMCs in the solder matrix and
at the solder/substrate interface during ageing [78]. The growth
rate is controlled by the reaction rate, for a Sn–2.8Ag–0.5Cu solder
system following linear growth. If the IMC growth follows para-
bolic growth kinetics, then it is presumed to be volume diffusion
controlled. For the diffusion-controlled mechanism the growth of
the IMC layer is defined by
d ¼ ðktÞ
1=2
ð2Þ
where d is the thickness of the IMC layer at time t, and k the
intermetallic growth rate constant. For the Sn–2.8Ag–0.5Cu solder
system, the growth rate constant of IMC layer formation is
Fig. 13. SEM images showing the interfacial microstructures for SAC–2Al/Cu: (A) solder ball (0.01 g), (B) solder bath (30 g), and (C) solder bath sample stored at 150 °C, 1500 h
[10].
2.21  1017 m2 s1, and the addition of Bi at 1 wt.% into this sys-
tem reduces the growth rate to 1.91  1017 m2 s1. This amount
of Bi also reduces the melting temperature and the consumption
rate of Cu.
Al has been added to SAC systems to study the interfacial reac-
tion between the substrate and the solder, with the specific inten-
tion of initiating the growth of a Al2Cu IMC layer at the Cu
substrate interface, respectively [7]. At low concentrations of Al
(0.5 wt.%), microstructural analysis revealed the formation of Ag3Al
and Al–Cu IMCs in the solder matrix (Fig. 13a). By carrying out this
experiment using a solder bath (Fig. 13b), a thin layer of Al2Cu was
seen to form after reflow of the SAC–2Al on Cu substrates, respec-
tively. Subsequently this Al2Cu layer acts as barrier layer for Cu6Sn5
IMC growth (Fig. 13c). When a small volume of solder was used,
without the solder bath, Al2Cu or Al3Ni IMC particles are formed in-
stead of stable barrier layers. Once the Al has reacted to form an
IMC layer between the solder and the joining surfaces, this layer
remains unstable. After increasing the wt.% of Al in the solder,
the concentration of Al2Cu and Al3Ni particles significantly in-
creases, mostly due to the reaction with the Cu or Ni dissolved
from the substrate.
3. Sn–Cu solder alloys
Sn–0.7Cu can be used as an alternative to Pb containing solders
as it exhibits some excellent physical and mechanical properties,
and is less expensive than Pb-free solders [79–83]. Its main appli-
cation is in wave soldering in electronic packaging, for which it has
been in use for over a decade, but like SAC solder, its higher melt-
ing point renders it also suitable for high temperature applications
such as those in the automotive industry [3]. In general terms it is
considered to perform on the same level as Sn–Pb, although it has
proven to be superior in terms of creep and fatigue .
The binary alloy of Sn–Cu has a eutectic composition of Sn–
0.7Cu (wt.%) and a eutectic temperature of 227 °C. A hypoeutectic
Sn–Cu solder microstructure composed of primary b-Sn dendrites
and grey eutectic colonies. If cooling is slow, the Sn–0.7Cu alloy fol-
lows the equilibrium solidification path, L ! b-Sn þ Cu6 Sn5 , with
Cu precipitating as hollow rods of Cu6Sn5 [84]. The microstructure
of the Sn–0.7Cu alloy is shown in Fig. 14a, where it can be seen that
primary b-Sn dendrites dominate. On a Cu substrate, the solder
forms with large Sn-rich dendrites interspersed with fine Cu6Sn5
intermetallics. At the interface with a Cu substrate, a thin layer
of Cu3Sn and a thick scallop layer of Cu6Sn5 were observed. At
the eutectic temperature, the solid solubility of Cu in Sn is only
0.006–0.01 wt.%, with the intermediate phase corresponding to
44.8–46% Sn [79].
The thin layer of IMC formed during the reflow reaction be-
tween Sn–0.7Cu and Ni was reported to be Ni3Sn4 [85]. It was
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
A Primary Sn dendrites
Eutectic matrix
Sn-0.7Cu
Fig. 14. Optical micrograph of (A) Sn–0.7Cu alloy, and (B) effect of Ni addition on Sn–0.7Cu [93].
found that the IMCs exhibited a high sensitivity to Cu concentra-
tion on Sn–Cu solder. There was a total of three reactions identi-
fied, representing distinct dependencies upon Cu concentration.
When the Cu concentration was high (Sn–0.7Cu to Sn–1Cu),
the IMC formed as (Cu1y,Niy)6Sn5. With an intermediate
concentration of Cu, around Sn–0.4Cu, both (Ni1x,Cux)3Sn4 and
(Cu1y,Niy)6Sn5 formed. Finally, at low concentration of Cu (around
Sn–0.2Cu), only (Ni1x,Cux)3Sn4 IMC was seen to form [85].
The interfacial reaction between eutectic Sn–Cu solder and
Au/Ni–P/Cu (ENIG) substrate was investigated by Yoon et al. [86].
After reflow, a duplex IMC structure of (Cu,Ni)6Sn5 and (Ni,Cu)3Sn4
was observed after high temperature ageing (170 °C for 1000 h),
but only the (Cu,Ni)6Sn5 IMC formed at lower temperatures (70–
150 °C). Huang et al. [87] have studied the solid/solid interfacial
reactions of Sn–0.7Cu solder with an electroless Ni–15%P substrate,
and observed the formation of needle shaped (Cu,Ni)6Sn5 IMC par-
ticles at the solder/substrate interface and a P enriched layer of
Ni3P in between the Ni–P substrate and the IMCs.
Co is of interest as a substrate in electronic applications as it
exhibits improved wettability and mechanical joint strength, com-
pared to Ni–P substrates [88]. However, it has been observed that ra-
pid reactions between Sn-rich solders and Co substrates can yield
joints with low mechanical strength [88]. There is evidence to sug-
gest that the addition of Ni to the Co layer effectively suppresses
the interfacial reaction between the solders and the substrate, as re-
ported by Huang et al. [59] This group observed the interfacial reac-
tions between Sn–0.7Cu/Co and electroplated Ni–Co alloy layers at
various reflow times, and despite the presence of Cu, the formation
of CoSn3 IMC was apparent. (Co,Ni,Cu)Sn3 compound layers were
seen at the interface of eutectic Sn–Cu solder/Ni–63 at.% Co.
Pt is often used as a base conductor in hybrid PCBs as it shows a
high resistance to oxidation and a low dissolution rate (comparable
to Ni) [89]. Pt has been successfully employed as a top surface layer
in under bump metallurgy (UBM), and exhibits a satisfactory wet-
tability for use in industrial soldering applications [90]. The inter-
facial reaction between Sn–0.7Cu solder and Pt has been
investigated by Kim et al. [55]. PtSn4 was found to form at the
interface with a slow growth rate, which is a favourable property
of this metallisation layer. The growth rate of the PtSn4 was seen
to follow the parabolic growth law, which implied that IMC growth
was controlled by bulk diffusion of elements to the reactive inter-
face [55].
1265
B
Uniform eutectic structure
Sn-0.7Cu + 600ppm Ni
Adding Ni to Sn–0.7Cu solder acts to suppress the growth of
pro-eutectic Sn dendrites and encourage solidification as a true eu-
tectic [91,92]. Nogita et al. [93] observed improved soldering prop-
erties for Sn–0.7Cu eutectic alloys doped with Ni at concentrations
up to 1000 ppm. The group saw a significant alteration in the
nucleation patterns and solidification behaviour of the Sn and
Sn–Cu6Sn5 eutectic. A micrograph of the Sn–0.7Cu alloy after Ni
addition is shown in Fig. 14b, where a disappearance of Sn den-
drites can be seen relative to the alloy without Ni addition. Nogita
et al. [93] suggested that the solidification (a semisolid region with
a liquid fraction) front was advancing from the edges of the sam-
ples towards the centre in the bulk solder alloy, a process which
they referred to as the ‘wall mechanism’. Above an Ni concentra-
tion of 600 ppm in the alloy, the microscopic growth process at
the interface changed, resulting in the solid evolving from a large
number of nucleation sites throughout the alloy melt.
The addition of Zn into Sn–0.7Cu solder alloy, and its reaction
with a Cu substrate, was investigated by Wang et al. [15]. The addi-
tion of a small amount of Zn into the solder refines the microstruc-
ture of the bulk alloy, and suppresses growth of the IMC interface
layer during thermal ageing. Above 1 wt.% Zn, the group observed
a complete change in the interface composition from Cu–Sn to
Cu–Zn, and through thermodynamic analysis (shown in Fig. 15a)
attributed this to a variation of the IMC driving force for formation
on the Zn concentration. These results suggest that the addition of
Zn to Sn–0.7Cu solder alloy should provide a viable method for
improving the mechanical properties and reliability of solder joints.
Fig. 15b shows a plot of IMC thickness versus ageing time for Sn–Cu
and Sn–Cu–Zn at 150 °C, and indicates that a small amount of Zn de-
creases the IMC layer thickness at the solder/substrate.
4. Sn–Ag solder alloys
The eutectic alloy of Sn–3.5Ag is an exciting Pb-free solder can-
didate due to its good mechanical properties and excellent wetting
behaviour on Cu substrates [94]. The microstructure of this system
consists of a primary b-Sn phase and a Sn–Ag eutectic mixture.
Fig. 16 shows an optical micrograph of a Sn–Ag alloy [95]. Hypoeu-
tectic Ag3Sn IMC flakes are observed here, which can be explained
by the variation of solidification path and formation of a couple
zone. Nucleation and growth of the Ag3Sn crystals occurs before
1266 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
Fig. 15. (A) Calculated results of the driving force for IMC formation between the molten solders and Cu substrate at 250 °C [15], and (B) IMC thickness versus storage time.
Fig. 16. An optical micrograph showing the microstructure of the eutectic Sn–3.5Ag
solder alloy [95].
the eutectic reaction commences, with the leading Ag3Sn phase
nucleating adjacent to the primary Ag3Sn crystal, which would re-
sult in bulk Ag3Sn IMCs. This means that the solidified fraction of
bulk Ag3Sn IMCs is actually larger than would be expected from
the equilibrium phase diagram. When acting as the primary phase
in hypoeutuectic Sn–Ag solder, b-Sn requires a larger degree of
undercooling, and there is a gradual increase in the fraction of bulk
Ag3Sn IMCs with increasing cooling rates [96]. An increase in the
cooling rate decreases the secondary dendrite arm size and spacing
Fig. 17. SEM images showing interfacial reactions between (Sn–3.5Ag)–1Zn solder/Cu: (A) after reflow, and (B) after 1000 h of storage time at 150 °C [7].
in the Sn-rich phase, giving a fine dispersion of Ag3Sn, whilst a de-
creased cooling rate results in needle-like Ag3Sn [97].
It is well known that soldering with Sn–3.8 wt.%Ag on Cu sub-
strates involves the formation of an IMC layer composed of Cu3Sn
and Cu6Sn5, with the dissolution substrate Cu enriching the molten
solder [98,99]. On an Ni–P substrate, the Sn–Ag solder forms Ni3P,
NiSnP, and Ni3Sn4 IMCs after reflow. In the absence of Cu, the bin-
ary system of Sn and Ni, or in the ternary system Sn–Ag on Ni,
forms Ni3Sn4. However, when Cu is present it suppresses Ni3Sn4
formation, instead forming (Cu,Ni)6Sn5 with some solution of Ni
[100].
Vianco et al. observed that the addition of Bi can improve wet-
tability on Cu substrates, with a minimum contact angle of 31 ± 4°
seen with 4.83 wt.% Bi addition [101]. Kotadia et al. [7] studied Sn–
Ag solder alloys, adding 0.5–1.5 wt.% Zn, studied on Cu and Ni–P
substrates, and found that 0.5 wt.% Zn addition did not result in
considerable suppression of Cu6Sn5 IMC growth when Sn–Ag–
0.5Zn/Cu samples were stored at 150 °C for 1000 h. However, the
IMC growth was significantly suppressed by 1 wt.% and 1.5 wt.%
Zn addition to Sn–Ag solder. The Cu3Sn IMC layer almost disap-
peared from the interface. Accumulation of Zn was found at the
Cu substrate/Cu3Sn interface. Massive spalling of the Cu–Zn IMC
layer acts as a barrier layer to Cu6Sn5 IMC growth (Fig. 17).
With the addition of 0.1 wt.% Ni and Co into the Sn–3.5Ag solder
system, the interfacial IMC at the Cu substrate changed from
Cu6Sn5 to (Cu0.8Ni0.2)6Sn5 and (Cu0.93Ni0.07)6Sn5, respectively. Addi-
tionally, the Cu3Sn IMC was suppressed, and the overall IMC
growth rate is reduced for the Sn–Ag–Ni/Cu and Sn–Ag–Co/Cu sys-
tems [102]. Tsai et al. [103] studied Sn–Ag with Ni (at concentra-
tions of 0 to 1 wt.%). They observed that the type of IMC that
forms on the Cu substrate is independent of the Ni concentration.
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
In the Ni-doped solder, the Cu6Sn5 phase contains only a small
amount of Ni. The addition of Al into the Sn–Ag solder does not
lead to the formation of any barrier layer or dissolution into the
interfacial IMC. However, Al based intermetallic particles have
been observed in the solder matrix [7]. There are various elements
known to have a similar effect, as we note in the SAC and Sn–Cu
sections of this review.
5. Sn–Bi solder alloys
Sn–Bi alloys show promise as Pb-free solder systems. The addi-
tion of 58 wt.% Bi into Sn form a eutectic with a relatively low melt-
ing temperature of 139 °C, and at room temperature Sn–Bi solder
alloys exhibit improved yield and fracture strength in comparison
with Sn–Pb alloys [104]. A reduction in the electronic assembly
temperature allows current or lower cost PCBs and components
to be used, thus reducing costs and reduces possible damage to
PCBs and popcorning of plastic packages through overheating
[104]. Sn–Bi exhibits a negative volume change upon melting,
and is cheaper than the comparable In–Sn solders which have sim-
ilarly low melting points [104]. An additional advantage is that the
properties of Sn–Bi alloys are well understood and characterised
due to their common use in soldering applications, and they are
commercially available in paste form [105]. The main drawback
to the system is the low operating temperatures that limit Sn–Bi
solders (even 100 °C corresponds to a homologous temperature
of 0.90).
The Sn–Bi alloy is a simple eutectic system, exhibiting phases of
Sn rich, Bi rich, and SnBi eutectic. Fig. 18a shows an SEM image of a
Sn–Bi/Cu sample, where the microstructure of the Sn–Bi alloy con-
sists of a primary b-Sn phase and a Sn–Bi eutectic mixture [106].
The limit of solid solubility of Sn in Bi is not completely clear,
and the values reported in the literature vary, [107] ranging be-
tween 0.2 and 4.1 at.% Sn [108]. The Sn–Bi system initially forms
Cu6Sn5 IMC on a Cu substrate, followed by Cu3Sn during subse-
quent ageing. On Ni–P substrates, Ni3P and Ni3Sn4 layers, as shown
in Fig. 18b. The NiBi3 phase only occurs when the Bi content is over
98 wt.%, whereas the Ni3Sn4 phase is formed in most Sn–Bi/Ni
systems.
It has been shown that even trace amounts of some alloying ele-
ments can form a barrier layer which helps to suppress the forma-
tion of interfacial IMCs at high temperatures. However, the
elements Al, Cr, Si, Nb, Pt and Cu (at 1–2 wt.%) appear to have no
significant effect on the IMC growth under high temperature stor-
age, [105] either failing to form an appropriate barrier or failing to
slow down IMC growth. Nowottnick et al. [109] observed the for-
mation of an effective Cu–Al IMC barrier layer at the molten sol-
der/Cu interface after addition of trace amounts of Al to Sn–58Bi
Fig. 18. Representative SEM images of (A) Sn–Bi/Cu system [105] and (B) Sn–Bi/Ni–P system [106].
1267
under vacuum conditions. However, in air migration of Al to the
surface, and subsequent oxidation, reduced the amount of Al in
the solder bulk. This meant that no IMC contained any Al, showing
that the thermodynamically favourable migration and oxidation of
Al at the molten solder surface prevented the Al atoms from reach-
ing the solder/Cu interface. Cr, Si and Nb have very low solubilities
in Sn–58Bi solder and therefore precipitate out of the alloy. The
addition of 1 wt.% Cu had no significant effect on the IMC growth,
despite the fact that grains of Cu6Sn5 precipitated out of the solder
before the interfacial reaction commenced [105].
By adding 1 wt.% Ag into the basic Sn–Bi solder it is possible to
slightly reduce the consumption rate of the Cu substrate. This may
be due to Ag3Sn groups forming and being captured at the surface
of the Cu6Sn5 grains in order to reduce the interfacial energy,
which slows down IMC growth. Yu et al. [110] reported that
nanoparticulate Ag3Sn was seen to precipitate on the surfaces of
Cu6Sn5 grains in the Sn–3.5Ag/Cu system, and on the Ni3Sn4 grains
in the Sn–3.5Ag/Ni. It has been shown that even small additions of
Ag or Au into 58Bi–42Sn can significantly improve isothermal fati-
gue resistance due to grain refinement. Increasing Au content also
rapidly increases melting temperature in Bi–Sn–Au solders, but no
significant difference is observed with Ag addition [104].
The addition of 0.05–0.5 wt.% of Co to the Sn–Bi solder system
forms an Cu6Sn5 IMC interfacial layer with low Co solubility, an ef-
fect which is not seen with Zn addition. The Co suppresses the for-
mation of Cu3Sn, and the growth rate of Cu6Sn5 increases with
increased Co addition. This Cu6Sn5 IMC layer exhibits a porous
structure with solder filled voids [111].
6. Sn–Zn solder alloys
Sn–Zn solder systems are also potential candidates for replacing
Pb based solders [112–114]. Eutectic Sn–Zn solder exhibits a melt-
ing temperature similar to Sn–Pb solder (198 °C compared to
183 °C) whilst demonstrating greater mechanical strength and a
lower cost. However, questions still remain over its unsatisfactory
wettability due to Zn being easily oxidised, which leads to an oxide
layer forming. This means that special fluxes and additional pre-
cautions must be used during the manufacture of electronic inter-
connections with this solder. Fig. 19 shows an optical micrograph
of a Sn–9Zn solder alloy, exhibiting a microstructure consisting
of an Sn–Zn eutectic mixture, with the fine Zn-rich phases being
uniformly dispersed in the Sn-rich matrix.
Various attempts have been made to improve the properties of
Sn–Zn solder (mechanical properties and oxidation resistance) by
the addition of trace amounts of alloying elements [2,115–119].
For example, it has been found that the addition of Bi reduces
the surface tension of Sn–Zn alloys, [120] and that Nd, La and In
1268 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
Fig. 19. An SEM micrograph of Sn–9Zn solder alloy.
Fig. 20. SEM images of the interfaces between eutectic Sn–9Zn solder/Cu
substrates, after reflow.
effectively increase the wetting area of Sn–Zn–Bi solders [114]. The
effect of adding Ag, Al and Ga to Sn–9Zn solders has been investi-
gated by Wang et al. [121], who concluded that the optimal addi-
tion of the elements for improving wettability is 0.3, 0.005–0.02
and 0.5 wt.%, respectively. An additional way of improving
wettability of Sn–9Zn–X is to use an N2 atmosphere, and to
use ZnCl2–NH4Cl flux, which performs even better than
Sn–3.5Ag–0.5Cu solder under the same conditions [122].
Additionally, rare earth metals such Ce, Bi, and In can improve
the wetabillity of eutectic Sn–Zn alloys, as well as mechanical
properties and oxidation resistance.
On a Cu substrate, Sn–9Zn solder has been observed to form
Cu5Zn8 IMCs which fracture after solid-state ageing, allowing diffu-
sion of Sn from the solder into the IMCs [123]. This resulted in the
formation of Cu6Sn5 and Cu3Sn phases at the Cu5Zn8/Cu interface,
as shown in Fig. 20, a process which was observed to significantly
weaken the joint strength. If a Au/Ni/Cu substrate was used instead
of pure Cu, a AuZn3 IMC layer formed at the interface due to a rapid
reaction between the Au and Zn, and then detached from the
interface during reflow. This detachment was explained by the
mismatch in the coefficients of thermal expansion and weak adhe-
sion between the AuZn3 IMC layer and Ni layer caused by depletion
of the Au layer. After extending the ageing time to 100 h, a Ni5Zn21
IMC was seen to form on the Ni substrate [123]. The addition of
2 wt.% Cu into eutectic Sn–9Zn solder has been seen to improve
wettability on a Ni substrate [116]. A continuous Ni5Zn21 IMC layer
at the (Sn–9Zn)–2Cu/Ni interface after soldering was observed by
Zhao et al. [116].
The effects of adding Cu to Sn–9Zn solder, and its reaction with
an Ag substrate, was investigated by Yen et al. [124] as they at-
tempted to overcome the oxidation of Zn in Sn–Zn solder alloys.
Above a Cu content of 3 wt.%, this group observed massive spalling
of the Ag–Zn interfacial IMCs. During soldering, they observed
Cu5Zn8 compounds forming at the Sn–Zn/Cu interface, with both
Cu6Sn5 and Cu5Zn8 forming if the Cu content was between 2 and
6 wt.%. However, beyond 8 wt.% Cu a single Cu6Sn5 scallop layer
has been observed to form at the interface [115].
Das et al. [125] have studied the addition of Ag microparticles to
a Sn–Zn eutectic solder on an ENIG substrate up to 4 wt.%. The
group observed the initial formation of an Au–Zn IMC layer, fol-
lowed by an Ag–Zn interfacial IMC layer on the ENIG substrate.
Chen et al. [126] have studied the Sn–9Zn solder alloy with a
0.044 wt.% addition of Cr, on a Cu substrate, and found IMCs of
Zn17Cr and/or Zn13Cr, the growth of which was suppressed during
ageing. Additionally, Shen et al. [127] investigated the effect of Cr
on the interfacial reactions of Sn–8Zn–3Bi solders with Cu sub-
strates. They observed that the Cu5Zn8 IMC was much thinner than
that of Sn–8Zn–3Bi solder joints, and no layers deficient in Zn were
found. The Cr rich phase was seen to react with the Zn during long-
term ageing, transforming the Sn–Cr phase into an Sn–Zn–Cr
phase. The diffusion of Zn atoms at the interface was indirectly
controlled by this solid phase transition, and also slowed down
the growth rate of the Cu–Zn IMCs. Additionally, Sn–Zn–Al solder
forms an Al4.2Cu3.2Zn0.7 IMC layer, [128] and the addition of Al or
Cr can significantly improve the oxidation resistance of Sn–Zn
based solder [129].
It is known that the wettability of Sn–Zn can be improved by
doping with Ag, [130] and that it can also improve ductility
[117]. Additionally, the addition of Ag up to 2 wt.% to Sn–9Zn eu-
tectic alloy increases the shearing strength [125]. Sn–Zn–Ag solder
forms Ag, Zn, Ag–Zn, or Ag–Sn compounds during equilibrium
solidification, depending on the composition. With increasing Ag
content the morphology of the Zn-rich phase transforms from nee-
dle-like to rod-like as solidification rate is increased [131]. The
addition of Ag to Sn–9Zn alloy yields a hyper-eutectic alloy, which
is near the Sn–Zn binary eutectic alloy. The solidification of
Sn–9Zn–xAg ternary alloys commences with precipitation of the
Ag–Zn compounds, followed by the b-Sn phase, and then finally
growth of the eutectic Sn–Zn structure. The solidification process
of Sn–9Zn–xAg alloys can elucidated as follows:
L ! L þ c  Ag5 Zn8
L þ c  Ag5 Zn8 ! L þ c  Ag5 Zn8 þ e  AgZn3
L þ c  Ag5 Zn8 þ e  AgZn3 ! L þ c  Ag5 Zn8 þ e  AgZn3 þ b  Sn
L þ c  Ag5 Zn8 þ e  AgZn3 þ b  Sn
! L þ c  Ag5 Zn8 þ e  AgZn3 þ b  Sn þ eutectic ðSn  ZnÞ
It is seen that the proportion of the Ag–Zn compound and the
b-Sn phase increases as Ag content is increased, and that of the
eutectic Sn–Zn structure decreases [132].
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
7. Nanocomposite solders
There is also the opportunity to use advances in nanoparticle
technology to make a major improvement in joint reliability for
harsh environments such as high-temperature and high-impact-
loading situations. The addition of fine second-phase particles gen-
erally increases the strength of metal alloys, can increase fatigue
resistance and creep strength by limiting grain boundary sliding,
and they can act as obstacles to grain growth and coarsening of
the solder microstructure by exerting a pinning force on migrating
grain boundaries [133]. In this scenario, the dispersed particles
need to be uniform in size and distributed homogenously through-
out the solder to obstruct grain boundary sliding and dislocation
movement. Generally the properties of a nanocomposite are
greatly influenced by the size scale of its component phases, and
by the degree of mixing between these phases. Depending on the
nature of the components used, and the method of preparation,
significant differences in composite properties may be obtained.
Micro-sized metal, ceramic and IMC particles have commonly been
added to composite solders to enhance their mechanical proper-
ties. A possible avenue for decreasing this high melting tempera-
ture is to decrease solder particle size in the paste to the
nanoscale [134].
There are three general methods for introducing nanoparticles
into solder matrices in the preparation of nanocomposite solders:
1) Mechanical mixing of ceramic nanoparticles,
2) Core-shell (metallic) nanoparticles and
3) Metallic nanoparticles.
The most simple and direct approach taken in the preparation
of nanocomposite solder pastes is to mechanically mix reinforce-
ment nanoparticles into the paste e.g. nano-sized TiO2, SiC [135–
139]. However, such additions do not normally stay in the solder
during reflow, and are almost completely expelled with the flux.
Powder metallurgical synthesis routes have been used by various
authors in order to obtain relatively homogeneous dispersions of
multi-wall carbon nanotubes (MWCNT), Single-wall carbon nano-
tubes (SWCNT), and nano-Al2O3 [138–140]. Another approach is
to add various amounts of ceramic nanoparticles to solder when
it is either in its melt or semi-solid stage, and to mix the whole
slurry by mechanical stirring, followed by casting at room temper-
ature. Precise amounts of alloy can be cut from an ingot fabricated
in this fashion, and placed onto a metallic substrate. The disadvan-
tage of these schemes for solder paste applications is that once the
solder alloy is prepared, the ingots must be further processed into
solder particles. The distribution of nanoparticles in the solder
after this final step is unlikely to be homogenous and may interfere
with the solder particle formation process also. To overcome these
problems, Mannan and co-workers have been working on the syn-
thesis of core–shell nanoparticles consisting of ceramic cores (e.g.
SiO2) coated with metallic nanoparticles (e.g. Au, Ag and Pd) form-
ing a fragmented shell [133,141–143]. The object of this appraoch
was to overcome the issue of particles expulsion due to them not
wetting into the liquid solder.
The advances in reactive nanoparticles (e.g. Zn, Al, Cu–Zn) can
be harnessed to radically improve solder reliability in harsh envi-
ronments [144]. Thus far it has proven difficult to develop such
systems because of the interference of the reactive metals in the
solder wetting process, because they oxidise at the solder surface
and prevent wetting to the metal contacts. However, by using suit-
ably coated nanoparticles, it should be possible to delay alloying of
these reactive particles until after wetting has occurred. The
advantages of reactive metal systems can thus be realised without
the previously insurmountable disadvantages. Recent work has
1269
shown however that if the particles are sub-micron then they have
a large chance of being expelled with the flux before coming into
contact with the solder liquid, and that the optimal particle size
may be in the 1–5 lm size range, which still allows the particles
to dissolve. A variety of nanoparticle inclusions (Co, Ni, Pt, Al, P,
Cu, Zn, Ge, Ag, In, Sb and Au) were studied, adding them to an sol-
der in order to evaluate their potential benefits in terms of the
solidification of base solder, formation of interfacial IMC, and
mechanical properties [145]. Much research has been directed to-
wards developing in situ methods where reinforcement particles
are formed upon bulk processing of the solder alloys themselves,
by precipitating a new phase from the solder e.g. Cu6Sn5, Ag3Sn
reinforced composite solder was developed [146,147].
7.1. Challenges in the processing and manufacturing of nanocomposite
solders
It is hoped that both the performance and potential applications
of solder materials can be greatly enhanced by the development of
advanced nanocomposites, and there is little doubt that they hold
great market potential. However, the big challenge of developing
the processing–manufacturing technologies, in terms of quantity
and value for commercialisation, offers a significant hurdle. There
are various issues that need to be addressed, for example the dis-
persion or chemical compatibility of nanoparticles with matrix
materials is key. Fine particles exhibit a strong tendency to aggre-
gate, therefore it is very difficult to obtain a homogeneous disper-
sion in the solder matrix using traditional dispersion techniques.
Such agglomeration leads the particles to be expelled during reflow
due to problems with wetting, such that premature failure in the
final product is almost inevitable. The rheological properties of sol-
ders are also a key consideration for the printing of nanocomposite
solders, as clogging of the stencil while printing with highly vis-
cous materials results in defective PCBs.
Introduction of novel nanomaterials into industry requires
extensive safety evaluation, which includes understanding the im-
pact of nanomaterials on the environment, human health and on
other biological species. It is of current general concern that nano-
scale particles may have negative health and environmental im-
pacts, and chemical legislation is lagging behind technological
development. For the application of nanomaterials in biological
and medical sciences these questions are obviously of utmost con-
cern, and it can be said that their application in metallurgy and sol-
der development does not carry such pressing restrictions.
However, it is still vitally important to understand the risks from
early on in the development of nanocomposite solders, especially
considering the use of ultrafine particles during scaled-up
manufacture.
8. Outlook and summary
In this article, we have presented an overview of the current
state-of-the-art in Pb-free solder, with an emphasis of the effects
of various substrates and alloying elements, before moving onto
cover how the advent of nanoparticle additions has opened up
new opportunities for novel ways to improve solder strength and
reliability. Forming reliable solders without Pb is a formidable task,
and we still need further insight from physical metallurgy to ad-
vance our knowledge of ternary and quaternary Pb-free solder sys-
tems to allow us to truly move on from the binary Sn–Pb solder of
the past. Furthermore, increasing pressure is being placed on the
quality of solder by advanced packaging architectures which de-
mand ever better performance from interconnections. Hence,
continuing research into understanding the complex interactions
1270 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273

Table 4
Summary of the alloying and nanoparticles effect, application state of various solders systems.

Solder alloy Alloying elements/nanoparticles effect Application state Future work

system
Sn–Ag–Cu (SAC) Forth alloying elements (e.g. Co, Fe, Zn, Ni, Al,
Bi,) added to SAC solder, Zn has a significant
effect on interfacial IMCs and nucleation
during solder alloy solidification
Sn–Cu Zn and Ni alloying elements have a significant
effect on interfacial IMC and the solidification
of solder. Addition of small amounts of Ni
increases solder fluidity
Sn–Ag Bi additions improve wettability on Cu
substrates and improve strength by
precipitation hardening. Reduced solder cost.
Zn has a similar effect as SAC and Sn-Cu solder
Sn–Bi The elements Al, Cr, Si, Nb, Pt and Cu (at 1–2
wt.%) appear to have no significant effect on
the IMC growth under high temperature
storage
Sn–Zn Wetabillity improved by Cu and Ag addition.
Trace rare earth elements have significant
effect on wetting and mechanical properties
Pb–free solder Reactive nanoparticles (e.g. Co, Ni, Pt, Al, P, Cu,
with Zn, Ge, Ag, In, Sb and Au) change
nanoparticles microstructure, melting temperature,
wettability and mechanical properties of
solder. However, there is no consistency in the
literature concerning the distribution of
alloying elements in the solder without
expelling with the flux
Non-reactive nanoparticles (e.g. silica, TiO2,
Al2O3, ZrO2, carbon nanotube) can improve
mechanical properties if homogeneously
distributed in the solder matrix. Core–shell
nanoparticles can be effective
Primary replacement candidates for near
eutectic Sn–Pb alloys for SMT applications.
High melting point and high solder cost
Low cost and specific application of wave
soldering. Poor mechanical properties
Primary replacement candidates for near
eutectic Sn–Pb alloys. High melting point.
However, not suitable for ENIG substrates.
High solder cost. Used as wire for hand
soldering rework, solder spheres for BGA/CSP
components, step soldering and die
attachment in high power devices
Low melting point eutectic. Potential
segregation problems and poor bonding
properties. Use widely in through-hole
technology assemblies
Zn imparts poor corrosion resistance and
reduced wettability specifically in Cu
substrate. Zn has low vapour pressure leading
to segregation in wave container and
inconsistency in process
To the authors knowledge, not currently used
in industrial applications
It is now obvious in both academia and
industry that it will be difficult to replace all
Sn–Pb solders with any one solder alloy due to
the high or low melting temperature, high
interfacial IMC growth, poor wetting, low
corrosion resistance and cost problems
Therefore, it is necessary to find a more
realistic solution, possibly by combining with
other alloying elements for suitable
applications. Additionally, by understanding
the physical metallurgy and processing
conditions of the solder alloys, we can possibly
to improve microstructures and reliability
properties of the solder
Current requirement is to find industrial scale
synthesis routes with good reproducibility

of the metallurgy, mechanical behaviour and reliability, for all clas-
ses of solders, is essential to drive innovation in the electronics
industry.
Several Pb-free solder alloys have been investigated for surface-
mount technology (SMT) assembly. SAC near-eutectic alloys are
perhaps the leading replacement technology due to their similarity
to Sn–Pb and their high performance, and they are currently used
for the majority of SMT applications. At 217 °C, they have a higher
melting temperature than eutectic Sn–Pb, introducing the neces-
sity for compatible board designs and component technology. It
should be noted that SAC alloys can cause problems in wave sol-
dering, and due to the solder volume and solder joint configuration
they often suffer from hot tears and shrinkage porosity. Further-
more, the cost of SAC solder is higher than many other alternative
alloys due to the higher Ag content. Because of these limitations,
the electronic packaging industry is actively seeking alternative al-
loys for non-SMT applications. Currently, a common alternative al-
loy is eutectic Sn–0.7Cu solder, which has demonstrated success in
production. The most important criteria for this choice were avail-
ability and cost, however Sn–Cu has been shown to exhibit inferior
wettability and mechanical properties in comparison to Sn–Ag al-
loys. Furthermore, binary Sn–Zn eutectic solder alloys are not gen-
erally ideal for wave soldering applications as they exhibit poor
drossing rates, which is related to the low vapour pressure and oxi-
dation characteristics of Zn. Zn alloys are also known to exhibit
poor corrosion resistance compared to Sn–Pb. The main classes
of solder systems are briefly summarized in Table 4.
In this review, we have aimed to provide a detailed explanation
of the formation of interfacial IMCs, which are known to play a
pivotal role in joint structure, and growth between various solders
and substrates. As an interfacial reaction takes place, one generally
expects the substrate to dissolve into the liquid solder, followed by
a chemical reaction between the solder and substrate alloying
elements, and finally solidification. The importance of each stage
varies between systems depending on the solubility of elements.
Substrates and minor alloying elements can have a significant
effect on the microstructure evolution of IMCs, with increased or
decreased IMC reaction/growth rates, formation of additional reac-
tion interface layers, displacement of the binary phases, and al-
tered formation of reaction products. The characteristics of the
interfacial IMCs are fundamental in defining the reliability of sol-
der joints because of the effects that the different IMCs have on
the mechanical properties. The effects of impurities on intermetal-
lic reactions are also significant in all systems. Additional trace ele-
ments can have a major influence on solidification of the solder
and on interfacial IMC growth, with mechanical properties either
improving or degrading depending on the phase formation. From
the point of view of reliability, slow growth rate of interfacial IMCs
can be achieved by providing a good diffusion barrier layer be-
tween solder and substrate.
Nanoparticle additions to Pb-free solder alloys have been devel-
oped to take advantage of the unique properties of nanomaterials
to overcome current limitations and future challenges of solder
development. Recent research has shown the benefit that nanopar-
ticles can bestow on a system. Approaches include adding reactive
nanoparticles to solder or mixing ceramic nanoparticles homoge-
neously throughout the solder. Current research suggests that the
size and surface properties of the nanoparticles are very important
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273 1271

in achieving desirable solder interfacial IMCs and mechanical prop- [25] Takamatsu Y, Esaka H, Shinozuka K. Formation mechanism of eutectic Cu6Sn5
and Ag3Sn after growth of primary b-Sn in Sn–Ag–Cu alloy. Mater Trans
erties. However, there is currently no systematic information about
2011;52:189–95.
the nanoparticle (ceramic, reactive or core–cell) effects on solder [26] Gebhardt E, Petzow G. The constitution of the silver–copper–tin system. Z
melting temperature, wettability, mechanical and rheological Metallkd 1959;50:597–605.
properties, suggesting that extensive investigations still need to [27] Liang J, Dariavach N, Shangguan D. Solidification condition effects on
microstructures and creep resistance of Sn–3.8Ag–0.7Cu lead-free. Metall
be carried out. There is also a need to find stabilisers for reactive Mater Trans A 2007;38:1530–8.
and core–shell nanoparticles to reduce agglomeration, and to de- [28] Swenson D. The effects of suppressed beta tin nucleation on the
velop engineering processes for industrial scale synthesis. microstructural evolution of lead-free solder joints. In: Lead-free electronic
solders. US: Springer; 2007. p. 39–54.
[29] Miller CM, Anderson IE, Smith JF. A viable tin-lead solder substitute Sn–Ag–
Cu. J Electron Mater 1994;23:595–601.
Acknowledgments [30] Mueller M, Wiese S, Roellig M, Wolter KJ. Effect of composition and cooling rate
on the microstructure of SnAgCu-solder joints. In: Electronic components and
technology conference, 2007. ECTC ‘07. Proceedings. 57th; 2007. p. 1579–88.
This research was funded by the Engineering and Physical Sci- [31] Kang SK, Choi WK, Shih DY, Henderson DW, Gosselin T, Sarkhel A, et al. Ag3Sn
ences Research Council (Grant no. EP/G054339/1). plate formation in the solidification of near-ternary eutectic Sn–Ag–Cu. Jom –
J Miner Metals Mater Soc 2003;55:61–5.
[32] Tu KN, Zeng K. Tin–lead (SnPb) solder reaction in flip chip technology. Mater
Sci Eng R – Rep 2001;34:1–58.
References [33] Kivilahti JK. Soldex 2.0—the thermodynamic databank for interconnection
and packaging materials. Espoo, Finland: Helsinki University of Technology;
1996.
[1] Loomans ME, Vaynman S, Ghosh G, Fine ME. Investigation of multicomponent
[34] Kim HK, Liou HK, Tu KN. Three-dimensional morphology of a very rough
lead-free solders. J Electron Mater 1994;23:741–6.
interface formed in the soldering reaction between eutectic SnPb and Cu.
[2] Rahn A. The basics of soldering. New York: John Wiely & Sons; 1993.
Appl Phys Lett 1995;66:2337–9.
[3] Tu KN. Solder joint technology: material, properties, and reliability. NY
[35] Ghosh G. Coarsening kinetics of Ni3Sn4 scallops during interfacial reaction
(USA): Springer; 2007.
between liquid eutectic solders and Cu/Ni/Pd metallization. J Appl Phys
[4] Laurila T, Vuorinen V, Kivilahti JK. Interfacial reactions between lead-free
2000;88:6887–96.
solders and common base materials. Mater Sci Eng R – Rep 2005;49:1–60.
[36] Kim HK, Tu KN. Kinetic analysis of the soldering reaction between eutectic
[5] Laurila T, Vuorinen V, Paulasto-Krockel M. Impurity and alloying effects on
SnPb alloy and Cu accompanied by ripening. Phys Rev B 1996;53:16027–34.
interfacial reaction layers in Pb-free soldering. Mater Sci Eng R – Rep
[37] Schaefer M, Fournelle R, Liang J. Theory for intermetallic phase growth
2010;68:1–38.
between Cu and liquid Sn–Pb solder based on grain boundary diffusion
[6] Abtew M, Selvaduray G. Lead-free solders in microelectronics. Mater Sci Eng R
control. J Electron Mater 1998;27:1167–76.
– Rep 2000;27:95–141.
[38] Lord RA, Umantsev A. Early stages of soldering reactions. J Appl Phys 2005;98.
[7] Kotadia HR, Mokhtari O, Bottrill M, Clode MP, Green MA, Mannan SH.
[39] Hayashi A, Kao CR, Chang YA. Reactions of solid copper with pure liquid tin
Reactions of Sn–3.5Ag-based solders containing Zn and Al additions on Cu
and liquid tin saturated with copper. Scripta Mater 1997;37:393–8.
and Ni(P) substrates. J Electron Mater 2010;39:2720–31.
[40] Kawakatsu I, Osawa T, Yamaguchi H. On the growth of alloy layer between
[8] Chawla N. Thermomechanical behaviour of environmentally benign Pb-free
solid copper and liquid tin. J Jpn Inst Met 1970;34:539–46.
solders. Int Mater Rev 2009;54:368–84.
[41] Ma D, Wang WD, Lahiri SK. Scallop formation and dissolution of Cu–Sn
[9] Kotadia HR, Mokhtari O, Clode MP, Green MA, Mannan SH. Intermetallic
intermetallic compound during solder reflow. J Appl Phys 2002;91:3312–7.
compound growth suppression at high temperature in SAC solders with Zn
[42] Chao B, Chae S-H, Zhang X, Lu K-H, Im J, Ho PS. Investigation of diffusion and
addition on Cu and Ni–P substrates. J Alloy Compd 2012;511:176–88.
electromigration parameters for Cu–Sn intermetallic compounds in Pb-free
[10] Kotadia HR, Panneerselvam A, Mokhtari O, Green MA, Mannan SH. Massive
solders using simulated annealing. Acta Mater 2007;55:2805–14.
spalling of Cu–Zn and Cu–Al intermetallic compounds at the interface
[43] Nash P, Nash A. The Ni–Sn (Nickel-Tin) system. J Phase Equilibr 1985;6:350–9.
between solders and Cu substrate during liquid state reaction. J Appl Phys
[44] Massalski T. Binary alloy phase diagrams. ASM; 1996.
2012;111.
[45] Bader W. Dissolution and formation on intermetallics in the soldering
[11] Mannan SH, Clode MP. Materials and processes for implementing high-
process. In: Proceedings of the conference on physical metallurgy, metal
temperature liquid interconnects. IEEE Trans Adv Packag 2004;27:508–14.
joining, St. Louis, MO, October 16–17, TMS/AIME, Warredale, USA, 1980.
[12] Kinyanjui R, Lehman LP, Zavalij L, Cotts E. Effect of sample size on the
[46] Hung KC, Chan YC, Tang CW, Ong HC. Correlation between Ni, Sn,
solidification temperature and microstructure of SnAgCu near eutectic alloys.
intermetallics and Ni3P due to solder reaction-assisted crystallization of
J Mater Res 2005;20:2914–8.
electroless Ni–P metallization in advanced packages. J Mater Res
[13] Kang SK, Moon-Gi C, Lauro P, Da-Yuan S. Critical factors affecting the
2000;15:2534–9.
undercooling of Pb-free, flip-chip solder bumps and in situ observation of
[47] Alam MO, Chan YC. Effect of 0.5 wt% Cu in Sn–3.5%Ag solder to retard
solidification process. In: Electronic components and technology conference,
interfacial reactions with the electroless Ni–P metallization for BGA solder
2007. ECTC ‘07. Proceedings. 57th, 2007, p. 1597–603.
joints application. IEEE Trans Compon Packag Technol 2008;31:431–8.
[14] Cho MG, Kim HY, Seo SK, Lee HM. Enhancement of heterogeneous nucleation
[48] Lee C-Y, Lin K-L. The interaction kinetics and compound formation between
of b-Sn phases in Sn-rich solders by adding minor alloying elements with
electroless NiP and solder. Thin Solid Films 1994;249:201–6.
hexagonal closed packed structures. Appl Phys Lett 2009;95.
[49] Hung KC, Chan YC, Tang CW. Metallurgical reaction and mechanical strength
[15] Wang F, Ma X, Qian Y. Improvement of microstructure and interface structure
of electroless Ni–P solder joints for advanced packaging applications. J Mater
of eutectic Sn–0.7Cu solder with small amount of Zn addition. Scripta Mater
Sci: Mater Electron 2000;11:587–93.
2005;53:699–702.
[50] He M, Lau WH, Qi G, Chen Z. Intermetallic compound formation between Sn–
[16] Anderson IE, Walleser J, Harringa JL. Observations of nucleation catalysis
3.5Ag solder and Ni-based metallization during liquid state reaction. Thin
effects during solidification of SnAgCuX solder joints. JOM 2007;59:38–43.
Solid Films 2004;462–463:376–83.
[17] Qi Y, Zbrzezny AR, Agia M, Lam R, Ghorbani HR, Snugovsky P, et al.
[51] Kumar A, Chen Z, Mhaisalkar SG, Wong CC, Teo PS, Kripesh V. Effect of Ni–P
Accelerated thermal fatigue of lead-free solder joints as a function of reflow
thickness on solid-state interfacial reactions between Sn–3.5Ag solder and
cooling rate. J Electron Mater 2004;33:1497–506.
electroless Ni–P metallization on Cu substrate. Thin Solid Films
[18] Bradley E, Ieee I. Lead-free solder assembly: impact and opportunity. In: 53rd
2006;504:410–5.
Electronic components & technology conference, proceedings; 2003. p. 41–6.
[52] Tsai Y-Y, Wu F-B, Chen Y-I, Peng P-J, Duh J-G, Tsai S-Y. Thermal stability and
[19] Yu H, Kivilahti JK. Nucleation kinetics and solidification temperatures of
mechanical properties of Ni–W–P electroless deposits. Surf Coat Technol
SnAgCu interconnections during reflow process. IEEE Trans Compon Packag
2001;146–147:502–7.
Technol 2006;29:778–86.
[53] Hae-Young C, Tae-Jin K, Young-Min K, Sun-Chul K, Jin-Young P, Kim Y-H. A
[20] Anderson IE, Foley JC, Cook BA, Harringa J, Terpstra RL, Unal O. Alloying
new Ni–Zn under bump metallurgy for Pb-free solder bump flip chip
effects in near-eutectic Sn–Ag–Cu solder alloys for improved microstructural
application. In: Electronic components and technology conference (ECTC),
stability. J Electron Mater 2001;30:1050–9.
2010 proceedings 60th; 2010. p. 151–5.
[21] Anderson IE, Walleser JW, Harringa JL, Laabs F, Kracher A. Nucleation control
[54] Wu R, Tsao L, Chang S, Jain C, Chen R. Interfacial reactions between liquid
and thermal aging resistance of near-eutectic Sn–Ag–Cu–X solder joints by
Sn3.5Ag0.5Cu solders and Ag substrates. J Mater Sci: Mater Electron
alloy design. J Electron Mater 2009;38:2770–9.
2011;22:1181–7.
[22] Moon KW, Boettinger WJ, Kattner UR, Biancaniello FS, Handwerker CA.
[55] Kim TH, Kim YH. Sn–Ag–Cu and Sn–Cu solders: interfacial reactions with
Experimental and thermodynamic assessment of Sn–Ag–Cu solder alloys. J
platinum. JOM 2004;56:45–9.
Electron Mater 2000;29:1122–36.
[56] Okamoto H. Pt–Sn (platinum–tin). J Phase Equilibr 2003;24:198.
[23] Loomans M, Fine M. Tin–silver–copper eutectic temperature and
[57] Wang C-H, Kuo C-Y. Coupling effect of the interfacial reaction in Co/Sn/Cu
composition. Metall Mater Trans A 2000;31:1155–62.
diffusion couples. J Electron Mater 2010;39:1303–8.
[24] Snugovsky L, Snugovsky P, Perovic DD, Sack T, Rutter JW. Some aspects of
[58] Zhu W, Wang J, Liu H, Jin Z, Gong W. The interfacial reaction between Sn–Ag
nucleation and growth in Pb free Sn–Ag–Cu solder. Mater Sci Technol
alloys and Co substrate. Mater Sci Eng, A 2007;456:109–13.
2005;21:53–60.
1272 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
[59] Huang K, Shieu F, Huang T, Lu C, Chen C, Tseng H, et al. Study of interfacial
reactions between Sn(Cu) solders and Ni–Co alloy layers. J Electron Mater
2010;39:2403–11.
[60] Teck Kheng L, Zhang S, Wong CC, Tan AC, Hadikusuma D. Interfacial
microstructures and kinetics of Au/SnAgCu. Thin Solid Films
2006;504:441–5.
[61] Hutter M, Schmidt R, Zerrer P, Rauschenbach S, Wittke K, Scheel W, et al.
Effects of additional elements (Fe Co, Al) on SnAgCu solder joints. In:
Electronic components and technology conference, 2009. ECTC 2009. 59th,
2009. p. 54–60.
[62] Tseng C-F, Wang K-J, Duh J-G. Interfacial reactions of Sn–3.0Ag–0.5Cu solder
with Cu–Mn UBM during aging. J Electron Mater 2010;39:2522–7.
[63] Young Min K, Hee-Ra R, Sungtae K, Young-Ho K. Kinetics of intermetallic
compound formation at the interface between Sn–3.0Ag–0.5Cu solder and
Cu–Zn alloy substrates. J Electron Mater 2010;39.
[64] Tu KN, Gusak AM, Li M. Physics and materials challenges for lead-free solders.
J Appl Phys 2003;93:1335–53.
[65] Anderson I, Harringa J. Elevated temperature aging of solder joints based on
Sn–Ag–Cu: effects on joint microstructure and shear strength. J Electron
Mater 2004;33:1485–96.
[66] Vianco P, Rejent J, Hlava P. Solid-state intermetallic compound layer growth
between copper and 95.5Sn–3.9Ag–0.6Cu solder. J Electron Mater
2004;33:991–1004.
[67] Anderson I, Harringa J. Suppression of void coalescence in thermal aging of
tin–silver–copper–X solder joints. J Electron Mater 2006;35:94–106.
[68] Kang SK. Effects of minor alloying additions on the properties and reliability
of Pb-free solders and joints. In: Lead-free solders: materials reliability for
electronics. John Wiley & Sons Ltd.; 2012. p. 119–59.
[69] Cho MG, Kang SK, Shih DY, Lee HM. Effects of minor additions of Zn on
interfacial reactions of Sn–Ag–Cu and Sn–Cu solders with various Cu
substrates during thermal aging. J Electron Mater 2007;36:1501–9.
[70] Wang CH, Chen HH. Study of the effects of Zn content on the interfacial
reactions between Sn–Zn solders and Ni substrates at 250 °C. J Electron Mater
2010;39:2375–81.
[71] Yang SC, Ho CE, Chang CW, Kao CR. Massive spalling of intermetallic
compounds in solder-substrate reactions due to limited supply of the active
element. J Appl Phys 2007;101.
[72] Chou Cy, Chen Sw. Phase equilibria of the Sn–Zn–Cu ternary system. Acta
Mater 2006;54:2393–400.
[73] Kotadia HR, Panneerselvam A, Sugden MW, Steen H, Green M, Mannan SH,
et al. Electronics assembly and high temperature reliability using Sn–3.8Ag–
0.7Cu solder paste with Zn additives, components, packaging and
manufacturing technology. IEEE Trans 2013;3:1786–93.
[74] Yu C-Y, Wang K-J, Duh J-G. Interfacial reaction of Sn and Cu–xZn substrates
after reflow and thermal aging. J Electron Mater 2010;39:230–7.
[75] Pan C-C, Lin K-L. The interfacial amorphous double layer and the
homogeneous nucleation in reflow of a Sn–Zn solder on Cu substrate. J
Appl Phys 2011;109.
[76] Wang YW, Chang CC, Kao CR. Minimum effective Ni addition to SnAgCu
solders for retarding Cu3Sn growth. J Alloy Compd 2009;478:L1–4.
[77] Cheng F, Nishikawa H, Takemoto T. Microstructural and mechanical
properties of Sn–Ag–Cu lead-free solders with minor addition of Ni and/or
Co. J Mater Sci 2008;43:3643–8.
[78] Rizvi MJ, Chan YC, Bailey C, Lu H, Islam MN. Effect of adding 1 wt% Bi into the
Sn–2.8Ag–0.5Cu solder alloy on the intermetallic formations with Cu-
substrate during soldering and isothermal aging. J Alloy Compd
2006;407:208–14.
[79] Hwang JS. Implementing lead-free electronics. McGraw-Hill; 2005. p. 10–90.
[80] Shen J, Liu Y, Han Y, Gao H. Microstructure and mechanical properties of lead-
free Sn–Cu solder composites prepared by rapid directional solidification. J
Mater Sci: Mater Electron 2007;18:1235–8.
[81] Liu M, Xian AP. Tin whisker growth on the surface of Sn–0.7Cu lead-free
solder with a rare earth (Nd) addition. J Electron Mater 2009;38:2353–61.
[82] Kang SK. Development of lead (Pb)-free interconnection materials for
microelectronics. Met Mater Int 1999;5:545–9.
[83] Alam ME, Nai SML, Gupta M. Effect of amount of Cu on the intermetallic layer
thickness between Sn–Cu solders and Cu substrates. J Electron Mater
2009;38:2479–88.
[84] Braunovic M, Myshkin NK, Konchits VV. Electrical contacts: fundamentals,
applications and technology. CRC Press; 2006.
[85] Li C-Y, Chiou G-J, Duh J-G. Phase distribution and phase analysis in Cu6Sn5,
Ni3Sn4, and the Sn-rich corner in the ternary Sn–Cu–Ni isotherm at 240 °C. J
Electron Mater 2006;35:343–52.
[86] Yoon JW, Kim SW, Jung SB. Interfacial reaction and mechanical properties of
eutectic Sn–0.7Cu/Ni BGA solder joints during isothermal long-term aging. J
Alloy Compd 2005;391:82–9.
[87] Huang ML, Loeher T, Manessis D, Boettcher L, Ostmann A, Reichl H.
Morphology and growth kinetics of intermetallic compounds in solid-state
interfacial reaction of electroless Ni–P with Sn-based lead-free solders. J
Electron Mater 2006;35:181–8.
[88] Satyanarayan KN. Prabhu, Reactive wetting, evolution of interfacial and bulk
IMCs and their effect on mechanical properties of eutectic Sn–Cu solder alloy.
Adv Colloid Interface Sci 2011;166:87–118.
[89] Meagher B, Schwarcz D, Ohring M. Compound growth in platinum/tin–lead
solder diffusion couples. J Mater Sci 1996;31:5479–86.
[90] Yang SC, Chang WC, Wang YW, Kao CR. Interfacial reaction and wetting
behavior between Pt and molten solder. J Electron Mater 2009;38:25–32.
[91] Sweatman K, Nishimura T. Proceedings of the ECWS 10 conference, Anaheim,
USA; 2005. p. 22–4.
[92] Sweatman K, Nishimura T. IPC printed circuits expoÒ, APEXÒ and the
designers summit; 2006.
[93] Nogita K, Read J, Nishimura T, Sweatman K, Suenaga S, Dahle AK.
Microstructure control in Sn–0.7mass%Cu alloys. Mater Trans
2005;46:2419–25.
[94] Yang W, Messler R, Felton L. Microstructure evolution of eutectic Sn–Ag
solder joints. J Electron Mater 1994;23:765–72.
[95] Vianco P, Rejent J. Properties of ternary Sn–Ag–Bi solder alloys: Part I—
thermal properties and microstructural analysis. J Electron Mater
1999;28:1127–37.
[96] Shen J, Liu Y, Gao H, Wei C, Yang Y. Formation of bulk Ag3Sn intermetallic
compounds in Sn–Ag lead-free solders in solidification. J Electron Mater
2005;34:1591–7.
[97] Ochoa F, Williams J, Chawla N. Effects of cooling rate on the microstructure
and tensile behavior of a Sn–3.5 wt.%Ag solder. J Electron Mater
2003;32:1414–20.
[98] Klein RJ. Soldering in electronics. 2nd ed. Ayr (Scotland): Electrochemical
Publications Ltd.; 1989. p. 141–85.
[99] Chada S, Fournelle R, Laub W, Shangguan D. Copper substrate dissolution in
eutectic Sn–Ag solder and its effect on microstructure. J Electron Mater
2000;29:1214–21.
[100] Yoon J-W, Jung S-B. Growth kinetics of Ni3Sn4 and Ni3P layer between Sn–
3.5Ag solder and electroless Ni–P substrate. J Alloy Compd
2004;376:105–10.
[101] Vianco P, Rejent J. Properties of ternary Sn–Ag–Bi solder alloys: Part II—
wettability and mechanical properties analyses. J Electron Mater
1999;28:1138–43.
[102] Gao F, Takemoto T, Nishikawa H. Effects of Co and Ni addition on reactive
diffusion between Sn–3.5Ag solder and Cu during soldering and annealing.
Mater Sci Eng A 2006;420:39–46.
[103] Tsai J, Hu Y, Tsai C, Kao C. A study on the reaction between Cu and
Sn3.5Ag solder doped with small amounts of Ni. J Electron Mater 2003;32:
1203–8.
[104] Hua F, Mei Z, Lavagnino A. Eutectic Sn–Bi as an alternative Pb-free solder. In:
Proceedings of an international summit on lead-free electronics assemblies,
IPC Works’99; 1999. p. S/03/08/01/06.
[105] Li JF, Mannan SH, Clode MP, Whalley DC, Hutt DA. Interfacial reactions
between molten Sn–Bi–X solders and Cu substrates for liquid solder
interconnects. Acta Mater 2006;54:2907–22.
[106] Li J, Mannan SH, Clode MP, Liu C, Chen K, Whalley DC, et al. Interfacial
reaction between molten Sn–Bi based solders and electroless Ni–P coatings
for liquid solder interconnects. IEEE Trans Compon Packag Technol
2008;31:574–85.
[107] Hansen M, Anderko K. Constitution of binary alloys. New York: McGraw-Hill;
1958 [1993].
[108] Kattner U, Boettinger W. On the Sn–Bi–Ag ternary phase diagram. J Electron
Mater 1994;23:603–10.
[109] Nowottnick M, Pape U, Wittke K, Scheel W. Solder joints for high temperature
electronics. In: 2003 SMTA international conference proceedings, Chicago
(IL), 23–26 September; 2003. p. 693–9.
[110] Yu DQ, Zhao J, Wang L. Improvement on the microstructure stability,
mechanical and wetting properties of Sn–Ag–Cu lead-free solder with the
addition of rare earth elements. J Alloy Compd 2004;376:170–5.
[111] Huang Y-C, Chen S-W. Effects of Co alloying and size on solidification and
interfacial reactions in Sn–57 wt.%Bi–(Co)/Cu couples. J Electron Mater
2011;40:62–70.
[112] Islam MN, Chan YC, Rizvi MJ, Jillek W. Investigations of interfacial reactions of
Sn–Zn based and Sn–Ag–Cu lead-free solder alloys as replacement for Sn–Pb
solder. J Alloy Compd 2005;400:136–44.
[113] Shiue RK, Tsay LW, Lin CL, Ou JL. The reliability study of selected Sn–Zn based
lead-free solders on Au/Ni–P/Cu substrate. Microelectron Reliab
2003;43:453–63.
[114] McCormack M, Jin S, Chen H, Machusak D. New lead-free, Sn–Zn–In solder
alloys. J Electron Mater 1994;23:687–90.
[115] Yu DQ, Xie HP, Wang L. Investigation of interfacial microstructure and
wetting property of newly developed Sn–Zn–Cu solders with Cu substrate. J
Alloy Compd 2004;385:119–25.
[116] Zhao C, Ma H, Xie H, Wang L. Wetting behavior and interfacial reactions in
(Sn–9Zn)–2Cu/Ni joints during soldering and isothermal aging. J Mater Sci
Technol 2009;25:410–4.
[117] McCormack M, Jin S. Improved mechanical properties in new, Pb-free solder
alloys. J Electron Mater 1994;23:715–20.
[118] Yoon SW, Soh JR, Lee HM, Lee BJ. Thermodynamics-aided alloy design and
evaluation of Pb-free solder, Sn–Bi–In–Zn system. Acta Mater
1997;45:951–60.
[119] Lin K-L, Wang Y-C. Wetting interaction of Pb-free Sn–Zn–Al solders on metal
plated substrate. J Electron Mater 1998;27:1205–10.
[120] Bukat K, Moser Z, Sitek J, Gasior W, Koscielski M, Pstrus J. Investigation of
Sn–Zn–Bi solders – Part I: surface tension, interfacial tension and density
measurements of SnZn7Bi solders. Solder Surf Mount Technol
2010;22:10–6.
 H.R. Kotadia et al. / Microelectronics Reliability 54 (2014) 1253–1273
[121] Wang H, Xue S, Chen W, Zhao F. Effects of Ga–Ag, Ga–Al and Al–Ag additions
on the wetting characteristics of Sn–9Zn–X–Y lead-free solders. J Mater Sci:
Mater Electron 2009;20:1239–46.
[122] Zhang L, Xue S-B, Gao L-L, Sheng Z, Ye H, Xiao Z-X, et al. Development of Sn–
Zn lead-free solders bearing alloying elements. J Mater Sci: Mater Electron
2010;21:1–15.
[123] Yoon JW, Jung SB. Interfacial reactions and shear strength on Cu and
electrolytic Au/Ni metallization with Sn–Zn solder. J Mater Res
2006;21:1590–9.
[124] Yen Y-W, Jao C-C, Lee C. Effect of Cu addition on interfacial reaction between
Sn–9Zn solder and Ag. J Mater Res 2006;21:2986–90.
[125] Das SK, Sharif A, Chan YC, Wong NB, Yung WKC. Effect of Ag micro-
particles content on the mechanical strength of the interface formed
between Sn–Zn binary solder and Au/Ni/Cu bond pads. Microelectron Eng
2009;86:2086–93.
[126] Chen X, Hu A, Li M, Mao D. Study on the properties of Sn–9Zn–xCr lead-free
solder. J Alloy Compd 2008;460:478–84.
[127] Shen YD, Hu AM, Chen X, Li M, Mao DL. Interfacial reactions and reliability of
Sn Zn–Bi–XCr solder joints with Cu pads. In: Electronic packaging technology
& high density packaging, 2008. ICEPT-HDP 2008. International conference;
2008, p. 1–4.
[128] Lin K-L, Hsu H-M. Sn–Zn–Al Pb-free solder—an inherent barrier solder for Cu
contact. J Electron Mater 2001;30:1068–72.
[129] Chen X, Li M, Ren X, Mao D. Effects of alloying elements on the characteristics
of Sn-Zn lead-free solder. In: Electronic packaging technology, 2005 6th
international conference on; 2005. p. 211–7.
[130] Takemoto T, Funaki T, Matsunawa A. Electrochemical investigation on the
effect of silver addition on wettability of Sn–Zn system lead-free solder. Weld
Res Abroad 2000;46:20.
[131] Song JM, Lan GF, Lui TS, Chen LH. Microstructure and tensile properties of Sn–
9Zn–xAg lead-free solder alloys. Scripta Mater 2003;48:1047–51.
[132] Tsai Y-L, Hwang W-S. Solidification behavior of Sn–9Zn–xAg lead-free solder
alloys. Mater Sci Eng A 2005;413–414:312–6.
[133] Mokhtari O, Roshanghias A, Ashayer R, Kotadia HR, Khomamizadeh F, Kokabi
AH, et al. Disabling of nanoparticle effects at increased temperature in
nanocomposite solders. J Electron Mater 2012;41:1907–14.
[134] Shen J, Chan YC. Research advances in nano-composite solders. Microelectron
Reliab 2009;49:223–34.
1273
[135] Ashayer R, Cobley A, Mokhtari O, Mannan SH, Sajjadi S, Mason T.
Nanoparticle synthesis and formation of composite solder for harsh
environments. In: Estc 2008: 2nd electronics system-integration
technology conference, vols. 1 and 2, Proceedings; 2008. p. 929–33.
[136] Lin DC, Wang GX, Srivatsan TS, Al-Hajri M, Petraroli M. Influence of titanium
dioxide nanopowder addition on microstructural development and hardness
of tin–lead solder. Mater Lett 2003;57:3193–8.
[137] Lin DC, Liu S, Guo TM, Wang GX, Srivatsan TS, Petraroli M. An investigation of
nanoparticles addition on solidification kinetics and microstructure
development of tin–lead solder. Mater Sci Eng A 2003;360:285–92.
[138] Li ZX, Gupta M. High strength lead-free composite solder materials using
nano-Al2O3 as reinforcement. Adv Eng Mater 2005;7:1049–54.
[139] Shen J, Liu YC, Han YJ, Tian YM, Gao HX. Strengthening effects of ZrO2
nanoparticles on the microstructure and microhardness of Sn–3.5Ag lead-
free solder. J Electron Mater 2006;35:1672–9.
[140] Lee A, Subramanian KN. Development of nano-composite lead-free electronic
solders. J Electron Mater 2005;34:1399–407.
[141] Ashayer R, Cobley A, Mokhtari O, Mannan SH, Sajjadi S, Mason T.
Nanoparticle synthesis and formation of composite solder for harsh
environments. In: 2nd Electronics system-integration technology
conference; 2008.
[142] Ashayer R, Mannan SH, Sajjadi S. Synthesis and characterization of gold
nanoshells using poly (diallyldimethyl ammonium chloride). Colloid Surf a –
Physicochem Eng Aspect 2008;329:134–41.
[143] Ashayer R, Mannan SH, Sajjadi S, Clode MP, Miodownik MM.
Nanoparticle enhanced solders for high temperature environments. In:
Electronics packaging technology conference, 2007. EPTC 2007. 9th; 2007,
p. 109–13.
[144] Kotadia HR, Panneerselvam A, Green MA, Mannan SH. Limitations of
nanoparticle enhanced solder pastes for electronics assembly. In:
Nanotechnology (IEEE-NANO), 2012 12th IEEE conference; 2012, p. 1-5.
[145] Amagai M. A study of nanoparticles in Sn–Ag based lead free solders.
Microelectron Reliab 2008;48:1–16.
[146] Lee JH, Park DJ, Heo JN, Lee YH, Shin DH, Kim YS. Reflow characteristics of Sn–
Ag matrix in situ composite solders. Scripta Mater 2000;42:827–31.
[147] Hwang SY, Lee JW, Lee ZH. Microstructure of a lead-free composite solder
produced by an in situ process. J Electron Mater 2002;31:1304–8.