 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

DATE DUE DATE DUE

0920“
10

 

 

INFLUENCE OF NANO—STRUCTURED CHEMICALS ON THE
MICROSTRUCTURES AND MECHANICAL RELIABILITY OF
LEAD-FREE TIN-BASED SOLDERS

BY
DEEP CHOUDHURI

A DISSERTATION

Submitted to
Michigan State University
In the partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Materials Science and Engineering

2009

ABSTRACT

INFLUENCE OF NANO-STRUCTURED CHEMICALS ON THE
MICROSTRUCTURES AND MECHANICAL RELIABILITY OF PB-FREE SN-BASED
SOLDERS

BY

DEEP CHOUDHURI

Nanostructured Chemicals utilized in this study are polyhedral oligomen’c
silsesquioxanes (POSS) trisilanol. The crucial Chemical entity of POSS trisilanols
are the silanol (Si-OH) groups, which can form thermodynamically feasible bonds
with metals (e.g Sn, Cu). This study capitalizes on such bonding ability of POSS
trisilanols by incorporating them in Pb-free Sn-based solder alloys (e.g. eutectic
Sn-Ag, near eutectic Sn-Ag-Cu). These solder alloys are used for fabricating
solder joints used in microelectronic interconnects. During service solder joints
are subjected to thermal excursions due to which they experience
thennomechanical fatigue (TMF). As a result, mechanical reliability of solder
joints undergoes deterioration. This study investigated the inﬂuence of POSS
trisilanol addition on the mechanical reliability of Pb-free Sn-based single shear-
lap solder joints subjected to TMF between -55°C to 125°C. Effect of TMF was
evaluated through observation of surface damage evolution and measurement of
residual mechanical shear strength as a function of number of TMF cycles. It
was found that POSS trisilanol addition improved the mechanical reliability of Pb-
free Sn-based solder joints by enhancing interfaCial bonding, and the ability of
the solder matrix to accommodate strains generated during TMF. To further

evaluate the inﬂuence of nanostructured Chemical addition separate set of

T

studies were performed on POSS trisilanol containing bulk and single shear-lap
solder joint specimens. Effect of POSS trisilanol on bulk specimens were
investigated through observation of as-solidified microstructure and thermal
analysis. On the other hand studies involving single shear-lap solder joints were
conducted by isotherrnally aging them at different temperatures. Results from
the studies on bulk and joint specimens indicate that POSS trisilanols are present

at the Sn—Sn grain boundaries and between interfacial regions of different phases.

Acknowledgements

First of all I would like to express my sincere gratitude to my advisor Dr. Andre
Lee for pnoviding intellectual guidance and helping me to see the broader picture
in research and life in general. I am extremely grateful to my co-advisor Dr. K.N.
Subramanian for providing insightful suggests during the course of PhD work. l
have gained a lot from my discussions with him, individually and during group
meetings (which I occasionally missed). I would also especially like to thank Dr.
Chen En Ho for mentoring me in experimental and metallographic techniques.

I also want to convey my appreciation to my committee member Dr
Melissa Baumann and Dr Timothy Hogan for very insightful and constructive
criticisms.

I would like to thank my friends Devesh, Tithi, Shishir, Ke-bin Low, Ajay,
DeepB, Madhusudhan, Madhu, Surabhi, Raylene, Shahana, Sumit, Sachin,
Naveen, Sutty, Varun, Kashi, Kavita, Bhushan, Shreya , and many more for
helping both academically and philosophically. Life here at MSU would have
been dull without them. Finally, I would like to Ma and Baba, and family
members Mithu didi ,Babuda, and Boromama for their steady advice and support

during my journey.

iv

Table of contents

List of Tables .................................................................................................... viii
List of Figures ..................................................................................................... x
Chapter 1 Introduction ........................................................................................ 1
1.1 Sn-based Pb-free solders ............................................................................ 1
1.2. Inﬂuences of service conditions on Sn-based solder joints ..................... 3
1.2.1 Inﬂuence of high Sn content and interrnetallic compound phases on the
service reliability of Sn-based alloys .............................................................. 4
1.2.2 Inﬂuence of service parameters on the reliability of solder joints .......... 8
1.3 Development of Pb-free Sn-based solders ............................................ 12
1.3.1 Alloying approach ............................................................................... 13
1.3.2 Composite approach ........................................................................... 16
1.3.3 Development of nano-composite solders using nano-structured
Chemicals ..................................................................................................... 19
1.4 Research strategy ..................................................................................... 24
Chapter 2 Experimental Procedure and methods .......................................... 26
2.1 Materials and processing ........................................................................... 26
2.2 Preparation of Solder Joints ...................................................................... 27
2.3 Preparation of bulk alloy specimen ............................................................ 27
2.4 Procedure for polishing joint and bulk specimens ...................................... 32
2.5 Setup for TMF experiments ....................................................................... 33
2.6 Mechanical and microstructural Characterization ....................................... 36

Chapter 3 Effect of temperature regimes on the thermomechanical fatigue

behavior of eutectic Sn-Ag joints ................................................... 39
3.1. Background .............................................................................................. 40
3.2. Experimental details ................................................................................. 42
3.3 Results and discussion .............................................................................. 43
3.3.1 Microstructural Evaluation of Damages resulting from Low Temperature
Regime TMF ................................................................................................ 43
3.3.2 Effect of temperature regime of TMF on surface damage accumulation
..................................................................................................................... 52
3.3.3 Inﬂuence of temperature regime of TMF on the residual mechanical
properties ..................................................................................................... 57
3.4 Conclusions ............................................................................................... 61

Chapter 4 Low temperature regime thermomechanical fatigue behavior of
nano—structured polyhedral oligomeric silsesquioxanes

 

reinforced Sn—based solder joints ................... _ ........ 63

4.1. Background .............................................................................................. 64
4.2. Experimental details ................................................................................. 65
4.3 Results and discussion .............................................................................. 67
4.3.1 Inﬂuence of POSS reinforcements on the residual mechanical behavior
of Sn-based solder joints ............................................................................. 67
4.3.2 Microstructural features contributing to the low temperature behavior of
POSS reinforced solder joints ...................................................................... 73
4.4. Conclusions .............................................................................................. 86

Chapter 5 Inﬂuence of polyhedral oligomeric silsesquioxane additions on

 

Sn-Ag-Cu bulk specimens -- 87
5.1. Background .............................................................................................. 88
5.2. Experimental details ................................................................................. 90
5.3. Results and discussions ........................................................................... 95
5.3.2 Inﬂuence of POSS additions on the microstructure of Sn—3.0wt%Ag-
O.5wt%Cu ..................................................................................................... 95
6.3.3 Thermal analysis ............................................................................... 109
5.4. Summary ................................................................................................ 113

Chapter 6 Inﬂuence of polyhedral oligomeric silsesquioxane additives on
the coarsening kinetics of Agasn interrnetallic compound in Sn-

 

Ag-Cu solderjoints......................................... -._--..119

6.1. Background ............................................................................................ 120
6.2. Experimental details ............................................................................... 122
6.3 Results and discussion ............................................................................ 123
6.3.1 Coarsening theories .......................................................................... 123
6.3.2 Inﬂuence of POSS addition on Aggsn particle size evolution ............ 127
6.3.3 Growth kinetics ................................................................................. 138
6.4 Conclusions ............................................................................................. 140
Chapter 7 Summary and recommendations ................................................. 150
7.1 Summary ................................................................................................. 150
7.2 Recommendations ................................................................................... 151

Appendix A Inﬂuence of polyhedral oligomeric silsesquioxanes trisilanol
chemistry on the low temperature regime thermomechanical

 

fatigue behavior of nanocomposite Sn-based solderjoints ...... 154
A1 Experimental details ................................................................................. 156
A2 Results and discussion ............................................................................ 156
A21 Effect of phenyl trisilanol POSS on the residual mechanical behavior of
Sn-based solder joints ............................................................................... 157
A22. Inﬂuence of phenyl trisilanol POSS reinforcements on microstructural
damages .................................................................................................... 163
A3. Conclusions ............................................................................................ 170
Appendix 3 Strain assisted allotropic transformation of B-Sn ................... 172
B1 . Introduction ............................................................................................. 1 73
B2. Experimental details ................................................................................ 1 75
B21 Mechanical Pre-Straining .................................................................... 175
32.2 Characterization and measurements ................................................. 175
B3. Results and discussion ........................................................................... 176
B3.1 Effect of mechanical pre-straining on the allotropic transformation of [3-
Sn .............................................................................................................. 176
83.2 Microstructural and mechanical Characterization ............................... 181
B3.3 Effect of stress on the growth kinetics of a-Sn ................................... 188
B4. Summary ................................................................................................. 193
References ...................... . .............................. 194
vii

""——I

List of Tables

Table 1.1 Listing of single crystal elastic modulus “E” and CTE “a” of Sn as a
function of ﬁve different crystallographic orientations. [25] .................. 6

Table 1.2 Bond energy values at 25°C [61]. ................................................... 22

Table 4.1 Statistical analysis of residual mechanical behavior of SA, and SA2iB
solder joints (Figure 4.3.1 (a)). Single tail t-test is used for this
purpose. Analysis is carried out on the basis of p values. This is
evaluated by comparing residual mechanical strength data of two
Chosen cycles. If there is a difference in the residual strength at the
Chosen number of cycles, then the p value is less than 0.05, otherwise
the two Chosen data sets are not statistically different. In this analysis,
residual strength data for 100 cycles is compared with 400 and 500
cycles. ................................................................................................ 71

Table 4.2 Statistical analysis of residual mechanical behavior of SAC305, and
SACZiB solder joints (Figure 4.3.1). Methodology for analysis is
exactly the same as described in Tables 4.1. In this analysis, residual
strength data after 100 cycles is compared with 400 and 500 cycles
for both the types of specimens and additional 1000 cycles for
SACZiB .............................................................................................. 72

Table 5.3.1 A AT values for all specimens .......................................................... 99

Table 5.3.2 Effects of trisilanol POSS additions on the heat of fusion of SAC305
......................................................................................................... 1 10

Table 5.3.3 R2 from linear least square ﬁt and activation energy of melting
resulting from Kissinger’s analysis (Figure 5.3.10 (b)) ..................... 112

Table 6.1 list of activation energies of diffusion of Ag and Cu along “a” and “C”
axes of Sn ........................................................................................ 121

Table 6.2 Summary of mechanisms controlling coarsening of secondary phase
particles and the respective phase size exponents .......................... 126

viii

Table 6.3 Statistical analysis for determining phase size constant “n” in <D>n -
<Do>n = Kt through non-linear regression ........................................ 142

Table 6.4 Linear regression analysis to determine “K” in <D>n - <Do>n = Kt for
the data presented in Figures 7 ....................................................... 146

Table 6.5 Inﬂuence of POSS trisilanols on the activation energies of diffusion 149

Table AI Comparison of single tail p-values obtained from t-test. Here residual
strength data after 1 and 100 cycles are compared with 100, 400, and
500 cycles for SA2phe and 100, 400, and 1000 cycles for SAC2phe.
P-values <0.05 would indicate that the compared data sets are
statistically different .......................................................................... 162

Table BI Hardness values of B-Sn at different locations and distances from the
Bid Sn interface ................................................................................ 186

Table Bll Values showing the growth rate for a given strain state of the
specimens and the corresponding correlation coefﬁcients for linear
least square ﬁts ................................................................................ 192

ix

List of Figures

Figure 1.2.1 Schematic showing the body-centered-tetragonal crystal structure
of Sn. ................................................................................................... 6

Figure 1.2.2 Schematic showing (a) different service parameters present during
thermal excursions, and (b) High and low temperature regimes
experienced during thermal excursions. Adapted from Subramanian
[18]. .................................................................................................... 11

Figure 1.3.1 Schematic showing two different types POSS anatomies (a)
Trisilanol POSS with three Si-OH groups, (b) Octa POSS without any
Si-OH groups. .................................................................................... 21

Figure 1.3.2 SEM images showing microstructural damages after 1000 cycles of
TMF in (a) eutectic Sn-3.5wt%Ag (adapted from Lee at.al. [21]), and
(b) eutectic Sn-3.5wt%Ag reinforced with 2wt% (10 vol%) POSS with
Si-OH groups (adapted from Lee et.a|. [55]). ..................................... 22

Figure 2.2.1 Apparatus used for fabricating solder joints (a) aluminum ﬁxture
used for positioning the half dog bone copper strips, (b) schematic of a
solder joint with dimensions shown. (Adapted from J G Lee thesis [1]).
........................................................................................................... 29

Figure 2.2.1 (o) Temperature proﬁle used for fabricating solder joints . (Adapted
from J G Lee thesis [1]) ...................................................................... 30

Figure 2.3.1 Schematic showing the experimental set-up used for the
preparation of bulk specimens. .......................................................... 31

Figure 2.4.1 Schematic showing the jig used for holding solder joints during
polishing. The wider ends of the joint are backed up with Cu strips of
similar thickness as that of the half dog-bone pieces. This prevents
bending or twisting of the joint (at the center) during polishing. ......... 31

Figure 2.5.1 Image showing the environmental Chamber used for conducting
TMF experiments ............................................................................... 34

Figure 2.5.2 Image showing the custom built stage used for placing solder
specimens inside the environmental Chamber. .................................. 34

Figure 2.5.3 Temperature proﬁles used for thermally cycling the solder joints...35

Figure 2.6.1 Set-up used for mechanical Characterization (a) RSA Ill used, and
(b) location of the solder joint inside the specimen holder of RSA III .38

Figure 3.3.1 Schematic of the regions on the solder joint observed for
microstructural evolution. Region-1 is at the center, region-2 is
~100pm below region-1 and region-3 is at the IMC / Solder interface.
........................................................................................................... 45

Figure 3.3.2 SEM images of region 1 showing the evolution of surface damage
within the solder and near the IMC layer and at the Cu68n5 IMC /
solder interface after (a) 0, (b) 50, (c) 100, (d) 200, (e) 300, (f) 400, (g)
500 cycles of TMF. Scale bar indicates 50pm. ................................. 48

Figure 3.3.3 SEM images of region 2 showing the evolution of surface damage
near the Cu68n5 IMC layer and at the Cu68n5 IMC lsolder interface
after (a) 0, (b) 50, (C) 100, (d) 200, (e) 300, (f) 400, (g) 500 cycles of
TMF. Scale bar indicates 50pm. ....................................................... 49

Figure 3.3.4 SEM images of region 3 showing the evolution of surface damage
at the IMC/solder interface after (a) 0, (b) 50, (c) 100, (d) 200, (e) 300,
(f) 400, (g) 500 cycles of TMF. ........................................................... 50

Figure 3.3.5 High magniﬁcation images showing debonding at the Cu68n5 IMC
layer / solder interface at two different regions after 500 cycles (a)
Location near region 1 and (b) is the corresponding BE image. SEM
image of region 3 is shown in (c) and its corresponding BE image in
(d). ..................................................................................................... 51

Figure 3.3.6 SEM images showing surface damage on Sn-3.5wt%Ag Solder
joints after 1000 cycles of high temperature regime TMF with long
dwell at (a) high temperature extreme, and (b) low temperature
extreme. Adopted from Subramanian et.al. [21,42] .......................... 55

Figure 3.3.7 SEM images showing the effect of high temperature regime TMF on
surface damage features near the IMC layer / solder interface, (a) Sn-
Sn grain decohesion and (b) grain boundary sliding / relief and shear
banding. Adopted from Choi et.al.[19] ................................................ 56

Figure 3.3.8 Plot showing the variation of residual shear stress of Sn-3.5wt%Ag
solder joints as a function of TMF cycles. Error bars represent 95%
conﬁdence intervals. .......................................................................... 58

Figure 3.3.9 (a) Cross sectional view of a mechanically failed Sn-3.5wt%Ag
solder joint;(b) Enlarged view on the of the region inside the dotted
rectangle in (a). This joint had undergone 500 cycles of TMF prior to
mechanical testing. ............................................................................ 59

Figure 3.3.10 Plot comparing the effect of high and low temperature regime TMF
on average shear strength of Sn-3.5wt%Ag single shear-lap solder
joints after 500 cycles of TMF. Residual strength data used for
comparison was obtained from [18, 21, 42]. ...................................... 60

xi

Figure 4.3.1 (a) Comparison of residual shear strengths of (a) SA and SA2iB
Error bars show 95% conﬁdence intervals. Second order exponential
non-linear least square ﬁts have also been shown ............................ 69

Figure 4.3.1 (b) Comparison of residual shear strengths of SAC305 and SACZIB
solder joints. Error bars show 95% conﬁdence intervals. Second order
exponential non-linear least square ﬁts have also been shown. ........ 70

Figure 4.3.2 SEM images showing surface damage evolution in eutectic Sn-
3.5Ag (SA) solder joint due to low temperature regime TMF between -
55°C to 125°C. Surface damages after 100 cycles of TMF has been
indicated with arrows. ........................................................................ 74

Figure 4.3.3 SEM images showing surface damage evolution in SA + 2wt% Iso-
Butyl trisilanol (SA2iB) solder joint due to low temperature regime TMF
between ~55°C to 125°C. Surface damages after 100 cycles of TMF
has been indicate with arrows. ........................................................... 75

Figure 4.3.4 SEM images showing surface damage evolution in Sn-3.0wt%Ag-
0.5wt%Cu (SAC305) solder joint due to low temperature regime TMF
between -55°C to 125°C. Surface damages after 100 cycles of
TMF has been indicate with anows. .................................................. 76

Figure 4.3.5 SEM images showing surface damage evolution in SAC305+2wt%
lso-Butyl trisilanol (SACZIB) solder joint due to low temperature
regime TMF between -55°C to 125°C. Surface damages after 100
cycles of TMF has been indicate with arrows. ................................... 77

Figure 4.3.6 High magniﬁcation SEM images comparing the effects of low
temperature regime TMF on surfaces damages within the solder
region after 500 cycles. Nano—composite (SA2iB and SACZiB) solder
joints show signiﬁcant damages within the solder region as compared
to Sn-based solder (SA and SAC305) joints. ..................................... 80

Figure 4.3.7 High magniﬁcation SEM images comparing the effects of low
temperature regime TMF on the Cu68n5 layer - solder interfacial
regions in Sn-based and nano-composite solder joints. Cu68n5 layer
-solder decohesion is evident in both SA and SAC305, while damage
is primarily located within the solder region in SA2iB and SAC2iB.
Scale bar indicates 10pm ................................................................... 83

Figure 4.3.8 High magniﬁcation SEM images comparing Cu68n5 particle - solder
interfaces in SA and nano-composite SA2iB after 500 cycles of low
temperature regime TMF. lnterfacial decohesion is evident in SA
while shear bands within the solder matrix near the Cu68n5 - solder
interface is observed in SA2iB. .......................................................... 84

xii

Figure 4.3.9 BE image of SAC2iB showing shear bands within the solder region
near the Cu68n5 - solder interface ..................................................... 85

Figure 5.2.1 Schematic showing two different types POSS anatomies (a) Iso-
Butyl Trisilanol POSS with three Si-OH groups, (b) lso-Butyl Octa
POSS without any Si-OH groups. Chemical formula for organic Iso-
Butyl(R) group has also been shown. ................................................ 92

Figure 5.2.2 Comparison of cooling rates utilized for solidifying molten bulk
solder specimens. Solid line indicates the faster cooling rate at
53°C/min, while the dashed line indicates slower cooling at a rate of
16 Clmin. ........................................................................................... 93

Figure 5.2.3 Left hand side showing the 30 views of the “button” shaped as-
solidiﬁed bulk specimens. The right hand side shows the regions
observed for microstnrctural characterization. The square boxes at
the center of each view i.e. planar and cross-sectional, are the
locations from where images for analysis were acquired. .................. 94

Figure 5.3.1 Optical micrographs showing the effects of 2wt% trisilanol POSS
and Octa POSS additions on the as-solidiﬁed microstructures of
SAC305. All the specimens have been cooled at 53°C lsec. .......... 100

Figure 5.3.2 Plot showing the quantitative microstructural analysis of the
secondary dendrite arm spacing. ..................................................... 101

Figures 5.3.3 Optical micrographs of showing the effects of 2wt% Trisilanol
POSS additions to SAC305 specimens cooled at (a)—(b) 16'C/seC
(slower cooling rate), and (C)—(d) 53'C/sec (faster cooling rate). ...... 102

Figure 5.3.4 Schematic showing the dendritic morphologies of (a) SAC, and (b)
after trisilanol POSS additions. ........................................................ 103

Figure 5.3.5 (a) Plots showing measured cooling curves of SACSOS, SAC305 +
2wt% lso-Butyl trisilanol POSS, and SAC305 + 2wt% lso-Butyl Octa
POSS. .............................................................................................. 104

Figure 5.3.5 (a) Plots showing effects of trisilanol POSS and Octa POSS
additions on undercooling. All the specimens were cooled at 53°C lsec.
......................................................................................................... 105

Figure 5.3.6 High magniﬁcation BEI showing sizes and morphology of Cu6$n5

and Ag3Sn lMCs in the ternary eutectic regions in (a) SAC305, and
(b) SAC305 + 2wt% Iso-Butyl trisilanol POSS s ............................. 1087

xiii

Figure 5.3.7 (a) Initial as-solidiﬁed microstructure of SAC305 + 2wt% lso-Butyl
trisilanol bulk specimen, (b) microstructure of the same specimen after
isothennally aging at 100°C for 5 months. ....................................... 108

Figure 5.3.8 Typical DSC therrnogram observed during melting. Area within the
peak (minima) provides the heat of fusion. The negative y-axis
represents endothermic heat input ................................................... 114

Figure 5.3.9 Typical DSC themiograms showing the increase in peak
temperature Tp with increase in heating rate during melting. The
negative y-axis represents endothermic heat input .......................... 115

Figure 5.3.10 (a) Plots showing the effects of 2wt% trisilanol POSS additions on
peak temperature vs. heating rate. Slopes in (a) were obtained by
ﬁtting a straight line. ......................................................................... 116

Figure 5.3.10 (b) Plots showing the effects of 2wt% trisilanol POSS additions on
Arrhenius plot for melting. . ............................................................. 117

Figure 5.3.11 BEI comparing microstructures of (a) SAC305 and (b) SAC305 +
2wt% iso-Buty trisilanol POSS. POSS containing specimen show
considerably more Sn grain boundaries than SAC305. Furthermore,
size of Sn grains in (b) is less than 20pm. ....................................... 118

Figure 6.3.1 Backscattered electron images (BEI) showing the initial
microstructures of (a) Sn-3.0wt%Ag-0.5%Cu (SAC305), and (b)
SAC305 + 2wt% lso-Butyl trisilanol POSS (SAC2iB). Different
microconstituents within the microstructure have been indicated. High
magniﬁcation SEM images showing the size and morphology of Aggsn
IMC particles in (C) SAC305, and (d) SAC2Ib. ................................. 129

Figure 6.3.2 Distribution of Agasn particle sizes in SAC305 and SACZIB
specimens. Two-tailed p value equals 0.54, suggesting that there is
no statistical difference between the A933" particle diameters in
SAC305 and SACZiB specimens. This analysis was carried out using
six specimens for each type of solder joint. ..................................... 130

Figure 6.3.3 BEI showing the inﬂuence of isothermal aging at 107°C, 150°C, and
180°C on (a)-(C) SAC305 and (d)—(f) SACZiB specimens. ............... 131

Figure 6.3.4 (a) <D> v.s. t plots comparing A9330 particle sizes in SAC305 and
SAC2iB specimens isothennally aged at 80°C. ............................... 132

Figure 6.3.4 (b) <D> v.s. t plots comparing Aggsn particle sizes in SAC305 and
SAC2iB specimens isothermally aged at 90°C. ............................... 133

Figure 6.3.4 (C) <D> v.s. t plots comparing Ag3Sn particle sizes in SAC305 and
SAC2iB specimens isothennally aged at 107°C. ............................. 134

xiv

Figure 6.3.4 (d) <D> v.s. t plots comparing Aggsn particle sizes in SAC305 and
SAC2iB specimens isothennally aged at 120°C. ............................. 135

Figure 6.3.4 (9) <D> v.s. t plots comparing Agssn particle sizes in SAC305 and
SACZIB specimens isothennally aged at 150°C. ............................. 136

Figure 6.3.4 (f) <D> v.s. t plots comparing Aggsn particle sizes in SAC305 and
SAC2iB specimens isothennally aged at 180°C. ............................. 137

Figure 6.3.5 (a) Plot of <D>4 - <Do>4 v.s t for determining the “K” for SAC305
specimens isothennally aged at 80°C, 90°C, and 107°C. ................ 143

Figure 6.3.5 (b) Plot of <D>3 - <Do>3 v.s t for determining the “K” for SAC305
specimens isothennally aged at 120°C, 150°C, and 180°C. ............ 144

Figure 6.3.5 (C) Plot of <D>5 - <DO>5 v.s t. for determining the “K” for SACZiB
specimens isothennally aged at 80°C, 90°C, 107°C, and 120°C. ....145

Figure 6.3.6 (b) Arrhenius plots for calculating activation energies for SAC2iB.
......................................................................................................... 148

Figure A1 Schematic showing the anatomy of POSS trisilanol. Chemical
formula of the organic phenyl (R) group is also shown. ................... 159

Figure A2 (b )Plots comparing inﬂuence of pheTriPOSS on the residual
mechanical behavior of SAC305. ..................................................... 161

Figure A3 SEM images showing microstructural evolution in SA2phe solder
joints after (a) 0, (b) 100, and (C) 500 cycles .................................... 165

Figure A3 SEM images showing microstructural evolution in SACthe solder
joints after (a) 0, (b) 100, and (C) 500 cycles .................................... 166

Figure A5 High magniﬁcation SEM images comparing surface damage within
solder region after 500 cycles of TMF in (a) SA2phe and (b) SACthe
specimens ........................................................................................ 167

Figure A6 SEM images comparing the interfacial damage after 500 cycles in (a)
SA2phe and (b) SAC2phe. Cu68n5 / Sn decohesion in SACthe have
been indicated by black arrows. ....................................................... 168

Figure A7 Compares microsthctural evolution of the Cu68n5 layer / solder
interfacial region for (aHb) SA2phe and (C)-(d) SACthe after 100
and 500 cycles. These images show phenyl trisilanol reduces the
intensity of interfacial damage. ........................................................ 169

Figure B1 Stereo-optical images showing the location of d-Sn on the (a) tensile
side of the non-uniformly bent specimen with 6% outer ﬁber strain, (b)

XV

compressive side of the uniformly bent specimen with 7.7% outer ﬁber
strain (shown as Region-1 and Region-2) ........................................ 178

Figure 82 XRD of the a-Sn detached from the non-uniformly bent specimen..179

Figure 83 SEM images showing the microstructural evolution of o-Sn on the
non-uniformly bent specimen aged at -15°C (a) as observed after 2.1
months, (b) after 2.6 months. .......................................................... 182

Figure 84 SEM images showing the microstructural evolution of o-Sn on the
uniformly bent specimen aged at -15°C (a) after 3.5 months in Region-
1 (b) after 4 months in Region-1, (C) after 3.5 months in Region-2 (d)
after 4 months in Region-2 ............................................................... 183

Figure B5 SEM micrographs showing the inter-granular and trans-granular
cracks within a-Sn region on (a) non-uniformly bent, and (b) uniformly
bent specimens. ............................................................................... 184

Figure 86 SEM micrographs showing surface relief features with in B-Sn region
adjacent to o-Sn (a) non-uniformly bent, and (b) uniformly bent
specimens ........................................................................................ 185

Figure B7 (a) Schematic showing the location of the indents on the broad side of
the non-unifome bent specimen (b) SEM image showing the location
and size of the indents. A load of 1 Kg was applied to the broad side
of the specimen for 10 seconds. ...................................................... 187

Figure 88 (a) Stereo-optical image showing the location of the a-Sn wart (inside
the white Circle) on the tensile side of the non-uniformly bent specimen
used for measuring growth kinetics .................................................. 190

Figure 88 (b) plot showing the Change in measured area of the same a-Sn wart
with number of days. ........................................................................ 190

Figure 89 Plot showing the change in the measured area of Region-1 as a

function of days. Region-1 was located on the compressive side of
the uniformly bent specimen. ........................................................... 191

xvi

Chapter 1

Introduction
1.1 Sn-based Pb-free solders

Eutectic Sn-37wt%Pb solder alloys have been used as interconnects in
electronic packaging, automotive, and aircraft industries for a long time [1].
However, the presence of Pb in solders has raised serious environmental and
health concerns. This has led to concerted efforts by industries all over the worid
to ﬁnd a viable substitute for Pb [1]. It was also realized that a replacement Pb-
free solder should have reasonable wetting behavior, ease of availability, and low
cost [1]. Additionally, replacement solders should be able to withstand
increasingly demanding service conditions, while simultaneously retaining
mechanical reliability of an electronic package. This necessitated solder
materials with both high strength and ductility [2-4].

On the basis of above mentioned criterion, binary Sn-based Pb—free solder
alloys have been identiﬁed as possible substitutes. Among several candidates,
(e.g. Sn-Ag, Sn—Cu, Sn-Sb, Sn-Zn, Sn-Bi etc.), eutectic Sn-3.5wt%Ag solder was
recognized as a viable replacement through extensive research [1, 4—6]. Sn-

3.5wt%Ag offered the best combination of high strength and ductility [4, 7].

Melting point of Sn-3.5wt%Ag (221°C) is higher than Sn-37wt%Pb (183°C).

Consequently, Sn-3.5wt%Ag is suitable to automobile under-the-hood application,
where solder joints regularly experience high service temperatures. However,
mechanical behavior of Sn—3.5wt%Ag is comparable to Sn-37wt%Pb, but

compares less favorably in terms of solderability. As a result, elements like Cu,

8i, In, Zn etc. have been added to the base Sn—Ag to improve solderability and
depress melting point [7]. Elemental additions were also observed to inﬂuence
the mechanical behavior of the resulting Sn-based ternary alloys systems [7-10].
Cu, 8i, In and Zn additions improved the tensile strength of base Sn-Ag alloy [7-
10]. However, Sn-Ag-Cu (resulting from Cu additions) exhibited comparatively
better ductility and mechanical fatigue behavior. Furthermore, mechanical
properties of Sn—Ag-Cu solders were found comparable to that of Sn-3.5wt%Ag
[7, 9, 10]. Consequently, improved mechanical properties along with ease of
availability, low cost, and good wetting behavior has made Sn-Ag-Cu a popular
choice for study and application [8, 11-15].

Although majority of these studies were carried out on bulk Sn-based Pb-
free solders specimens, but in real applications solders are used in joint
conﬁguration. These joints are found in surface mount devices such as small
outline package, quad ﬂat package, and ball grid array package [1]. These
surface mounts involve joining integrated Chip (IC) packages on a printed circuit
board (PCB) [1]. Consequently, joints form not only the electrical connections,
but also mechanical connection between the substrates and the components (IC
package). Since Sn-based Pb-free solder alloys are mechanically softer than
other components (eg. Cu substrate, lead frames) [1], strength of soldered joints
form the weakest link in terms of mechanical reliability of the electronic package.
As a consequence it is necessary to design Sn-based solders that can withstand
harsh service conditions without compromising their mechanical reliability.

These service conditions require solder joints to experience thermal excursions

resulting in thermomechanical fatigue (TMF) of those joints. Additionally, solder
joints are also subjected to external mechanical stresses during service eg.
mechanical vibrations and impact loading [1, 7, 10, 16-18].

In view of the above mentioned requirements and issues literature review
presented in this chapter serves the following purpose:

1. Identify service conditions related reliability issues with Sn-based Pb-free
(hence forth “Sn-based”) solder joints and,

2. Present progress made in the development of Sn-based solder materials
for combating reliability issues, and problems associated with such solder
materials.

1.2. Inﬂuences of service conditions on Sn-based solder joints

As mentioned in the previous section, solder joints experience TMF during
service. TMF occurs during thermal excursions encountered by joints in
automotive under-the-hood and power cycling type applications[1, 14, 18-24].
Temperature difference arising from thermal excursions results in cyclic
displacements between materials with dissimilar coefﬁcient of thermal expansion
(CTE). Such CTE mismatches arise between the constituents present within the
solder joint (i.e. solder region, the interfacial interrnetallic compound (IMC) layer,
and the substrate )[1, 14, 18-26]. Stresses arising from such CTE mismatches
result in TMF of joints. Additionally, creep / stress relaxation may also occur
within solder joints during thermal excursions [3, 23, 27-30]. So the combination
of stresses (generated by CTE mismatch), and creep / stress relaxation result in

temperature dependant deformation within the solder joints. The resulting

deformation causes microstructural damages within the solder region, leading to
deterioration in the mechanical reliability. However, the extent of such damages
would also depend on the material used for fabricating those joints, in addition to
service temperatures.

Sn-based alloys contain over 95wt% B—Sn. Consequently, properties of
Sn will signiﬁcantly inﬂuence the mechanical reliability of Sn—based solder joints.
So it is necessary to identify the contribution of high B-Sn content in Sn-based
solder joints subjected to TMF.

Material independent service parameters may also inﬂuence the
mechanical reliability of soldered joints. For example, during thermal excursions
different stress level would result from dissimilar “temperature difference”. Such
differences in service conditions would lead to different creep I stress relaxation
response in Sn-based solder joints. Consequently, different service parameters
will elicit different material response. Such coupling of service parameter and the
solder material used may eventually determine joint service life.

1.2.1 Inﬂuence of high Sn content and interrnetallic compound phases on
the service reliability of Sn-based alloys

At room temperature B—Sn exists as white tin which has a Body-Centered-
Tetragonal (BCT) crystal structure (Figure 1.2.1) [25, 26]. Henceforth Sn would
refer to B-Sn unless stated othenrvise. Lattice parameters of the BCT crystal are
C = 2.18A, and a = b = 5.83A [25, 26]. Differences in the lattice parameters result
in different mechanical and thermal properties along different crystallographic

directions. Table 1.1 lists elastic modulus and CTE of Sn single crystal as a

function of crystallographic direction. These values indicate the high anisotropic
nature of Sn grains. Anisotropy in mechanical and thermal properties becomes
particularly important during thermal excursions. Under such service conditions,
stresses are generated at Sn-grain boundaries due to anisotropic CTE
mismatches between adjoining grains [25, 26]. This causes different stress
levels depending on the relative orientations of anisotropic Sn grains with respect
to each other. Consequently, thermal excursions lead to inhomogeneous stress
states within the solder region of a joint [23, 24, 26].

Additionally, Sn-based solders, due to their low melting, experience high

homologous temperature during service. Temperature extremes encountered

during thermal excursions range between 0.5Tm, and 0.8Tm. Under those

conditions, solder joints would experience temperatures higher than the
equicohesive temperature of grain boundaries [39]. At equicohesive temperature
grain boundaries become weaker than the grain interiors. As a result, Sn-Sn
grain boundary sliding occurs upon imposition of stresses resulting from CTE
mismatches between Sn grains [1]. The occurrence of grain boundary sliding
during TMF also indicates that joints experience creep / stress relaxation like
conditions during thermal excursions [1, 21, 23, 24, 26]. Moreover, several
studies have shown that decohesion / cracking also occurs at Sn grain
boundaries during TMF of Sn-based solder joints [1, 21, 23, 24, 26]. Decohesion
at such grain boundaries have been attributed to the presence large stresses at

those boundaries during thermal excursions [25, 26].

       

Cl

> a [100]

 

 

Figure 1.2.1 Schematic showing the body-centered-tetragonal crystal structure

of Sn.

Table 1.1 Listing of single crystal elastic modulus "E" and CTE “a” of Sn as a

function of ﬁve different crystallographic orientations. [25]

a(10 K'

Crystallographic orientation E (GPa)

15.

 

Microstructural damage resulting from such grain boundary events (sliding
ldecohesion) deteriorates mechanical reliability of Sn-based solder joints [1, 21,
24-26]. In essence, Sn—based solder joints accumulate additional microstructural
damage with the solder region due anisotropic CTE mismatches between Sn
grains.

Moreover, presence of fewer and larger Sn grains could further
exacerbate such damages, since TMF behavior of joints will be dependent on the
relative orientation of Sn grains [31]. During TMF, large stresses between
unfavorably oriented Sn grains will damage grain boundaries and might cause
early failure of joints [31]. The possibility of large grains in Sn—based solders
arises due the fact that Sn requires a large undercooling (~40°C) prior to
solidiﬁcation [8, 31]. This means as-solidiﬁed Sn -based solder may contain
large grains. Puttlitz et.al. showed that damage between large Sn grains in Sn-
Ag-Cu solder joints signiﬁcantly reduces their TMF resistance [14].

Additionally, microstructures of Sn-based ‘solders also have interrnetallic

compound (IMC) phases. Chemical composition of the lMCs depends on the

minor element in Sn—based alloys. For example Ag3Sn is present in Sn-Ag alloys,
while Sn-Ag-Cu contains Ag3Sn, and Cuasns IMCs. Role of these IMCs is to
strengthen the solder matrix through different means. CU5Sn5 have been shown
to pin Sn-grains boundaries [32], while micron sized Ag3Sn in solder joints act as

obstacles to dislocation motion [33]. However, during service both Cu63n5 [21]

and Ag3Sn [3436] have been reported to coarsen. It should be mentioned that

coarsening of CueSn5 is a bigger issue that of AggSn. This is so because Cu

diffusion though Sn is considerably faster than that of Ag [37, 38]. Consequently,

coarsening of CueSn5 would result in fewer pinning sites at the Sn-Sn grain

boundaries. Whereas larger Ag3Sn IMC size would lead to ineffective blockage

of dislocations [39]. Therefore, coarsening of lMCs during service has the
capacity to compromise mechanical reliability of Sn—based solder joints.

Service conditions may also require Sn-based Pb-free solder joints to
operate at sub-zero temperatures for long durations. These conditions may arise
for solder joints used in devices for space and aircraft applications. At subzero
temperatures Sn undergoes allotropic transformation from a ductile to a brittle
phase (a-Sn). This phenomenon is called tin pest. Such a transformation occurs
below 13°C. At present tin-pest has not been reported to occur during service,
but is only observed under laboratory settings [24, 40, 41].
1.2.2 Inﬂuence of service parameters on the reliability of solder joints
To determine the inﬂuence of service parameters, it'is important to identify the
type of temperature proﬁle appropriate for an application. This study focuses on
thermal excursions prevalent in automobile under-the-hood situations and during
power cycling. Figure 1.2.2 (a) shows model temperature proﬁle experienced by
solder joints under previously mentionedcircumstances [1].

Service parameters associate with such thermal excursion (Figure 1.2.2),

such as heating / cooling rates, dwell times at temperature extremes, and

temperature differences (AT), have been reported to signiﬁcantly inﬂuence the
reliability of Sn-based solder joints [21, 22, 42]. Furthermore, reliability of solder
joints is directly related with their service life time. Based on such parameters, a

simple parametric formulation was proposed by Subramanian [22]:

. . -m -n -p <7 -r s
IIfetlme ~ [A r] m [(dT/dt)H] [(dT/dt)C] [A tH/Atc] [AM], (1)

where, the service parameters AT, f, (dT/dt)H, (dT/dt)c, AIL/Alb, and AM are

temperature difference between two extremes, frequency of TMF cycles, heating
rate, cooling rate, ratio of dwell times at high and low temperature extremes, and
microstructural Changes during TMF respectively, while m, n, p, q, r, s are
material dependant parameters. However, this model does not take into account
the effect of individual temperature extremes but their difference, i.e AT.

Rhee et.al. demonstrated through isothermal monotonic testing that
deformation modes in eutectic Sn-3.5wt%Ag solder joints are highly temperature

sensitivity [6]. That study revealed that deformation mode Changes from Sn grain

boundary sliding to shear banding in the neighborhood of 125°C. Similar

observations were reported by Choi et.al. through creep studies on Sn-3.5wt%Ag
solder joints [43]. These studies strongly suggest that deformation modes of Sn-
rich (<90wt%) solders are highly susceptible to temperature extremes
encountered during service. Further, microstructural damages caused by TMF,

depends on deformation modes. This suggests that reliability of solder joints,

 

experiencing TMF, will not only depend on the mentioned service parameters but
also on the temperature regimes in which AT is imposed (Figure 1.2.2(b)).
In essence the proposed parametric model does not include the effects of

temperature regime on the TMF service life of Sn-based solder joints.

 

10

T Dwell time at high
i temperature
extreme, At

    
  
   

Heating Cooling
rate . rate , AT
(dT/dt) (CIT/doc

Dwell time at low

 

 

 

 

temperature

extreme, Atc t

(a)

T
High
temperature
regime
\- Low
temperature
regime

_ t

(b)

Figure 1.2.2 Schematic showing (a) different service parameters present during
thermal excursions, and (b) High and low temperature regimes experienced

during thermal excursions. Adapted from Subramanian [18].

11

1.3 Development of Pb-free Sn-based solders

Sn-based solders joints experience thermal excursions during service. Thermal
excursions occur during automotive under-the-hood applications and device tum
on—off type situations (power cycling) [1]. As a result of those thermal excursions
solder joints experience TMF. Additionally, accidental dropping of devices result
in impact loading of soldered joints [44-47]. Microstructure of an effectively
designed solder should be able to handle such service conditions. Normally a
ﬁne grained microstructure provides a material with enough strength to withstand
fatigue due to mechanical vibrations [2, 17]. However, solder alloys, due to their
low melting points, inevitably encounter high homologous temperatures during
service. At high homologous temperatures grain boundary sliding occurs in ﬁne
grained microstructure during TMF or mechanical vibrations [2]. Additionally,
solder joints experience creep/stress relaxation during TMF resulting in a
requirement for a stronger solder matrix. However, the solder matrix should be
compliant enough to withstand stresses generated due to imposed AT. This

presents a demand for a solder material that has both high strength and ductility.

Furthermore, high temperature aging causes coarsening of lMCs (e.g.Ag3Sn

and Cu68n5) present in the microstructure. These IMCs are known to provide

strength to the solder matrix. Microstructural instability resulting from coarsening
of lMCs would further lead to weakening of the solder matrix. Consequently,
development of Sn-based solder must balance the need for strength and

compliance, while simultaneously retarding service induced microstructural

12

Changes. Additionally, implementation of such solders should require minimal

changes in the established industrial processes. To summarize, a well designed

Sn—based Pb—free solder must perform the following roles:

1.

3.

4.

Engineer microstructure: A ﬁne grained microstructure while
simultaneously retarding grain boundary sliding.

Accommodate imposed stress/strain without losing strength. In other
words potential Sn-based solder candidates must have both ductility and
strength.

Stabilize microstructure: Retard IMC coarsening.

Amenable to ease of processing and implementation

1.3.1 Alloying approach

As mentioned in section 1.1, Sn—Ag and Sn-Ag-Cu family of solder alloys are

seen as possible replacements for Sn-Pb alloys. Recently, Sn—Ag-Cu alloys are

experiencing more commercial use and subsequently being studied extensively

[8, 11-13, 48]. In particular near eutectic Sn-3.9wt%-0.6wt%Cu and Sn-3.0wt%-

0.5wt%Cu have been recommended by National Electronics Manufacturing

Initiative (NEMI) and Japan Electronics lnforrnation Technology Industries

Association (JEITA) respectively [49, 50]. However, reﬂowed microstructure of

those Sn-Ag-Cu alloys contains either large Sn primary dendritic structure or

large IMC particles [8, 11 -1 3, 48]. Such a coarse microstructure results due to

the large undercooling experienced by Sn rich matrix during solidiﬁcation [31].

Consequently, there is a need for reﬁning Sn-Ag-Cu microstructure.

13

One of the approaches taken was to Change the compositions of Ag or Cu

in Sn—Ag-Cu [8]. This approach was based on the assumption that CusSns IMC

particles promote heterogeneous nucleation of primary Sn. Studies carried out
by varying the concentrations of both Cu and Ag, Anderson et.al. found that Sn-
3.6wt%Ag-1.0wt%Cu had the most reﬁned microstructure [8, 11, 48]. It was also
shown that Sn-3.6wt%Ag-1.0wt%Cu had a higher shear and yield strength than
eutectic Sn—Ag. However, Sn-3.6wt%Ag—1.0wt%Cu had a lower ductility than
eutectic Sn-3.5wt%Ag as compared to other Sn-Ag-Cu alloys studied. To
simultaneously reﬁne the microstructure and retain ductility (as compared to
eutectic Sn-Ag) minor concentrations of Co and Fe were added to Sn—Ag-Cu
alloys resulting in quaternary Sn—Ag-Cu-X (X=Co, Fe) alloys. It was found that
both additions reﬁned the microstructure, but microstructure of Sn-Ag-Cu-Co was
comparatively more reﬁned. The resulting reﬁnement was attributed to the

presence of Co-Sn and Co-Sn-Cu lMCs. It was suggested that during

solidiﬁcation those IMCs promoted the heterogeneous nucleation of Cu68n5,

which in turn assisted in the nucleation of Sn. Futherrnore, Co and Fe additions
resulted in similar ductility and strength as that of eutectic Sn-Ag [8, 11, 12, 48].

Although Co and Fe additions resulted in desired improvements, presence of

Cu68n5 IMC within the microstructure is a source of concern. CusSn5 in Sn-Ag-

Cu solder joints has been known to coarsen during TMF [21]. Such coarsening
contributed to the deterioration in the mechanical reliability of Sn-Ag-Cu solder

joints subjected to TMF [21]. Similar coarsening of IMCs during TMF was also

14

observed in quaternary Sn-Ag-Cu-Bi and Sn-Ag-Cu-ln alloys, even though Bi and
In additions improved the fatigue life of solder joints [51].

Later Subramanian et,al developed Sn-Ag-Cu-Ni quaternary alloy systems

that reduced CusSn5 IMC coarsening during TMF. The improved microstructural

stability was due to the presence of (CuNi)6Sn5 IMC in those alloys, which

prevented diffusion of Cu through Sn [1, 2, 52]. Furthermore, (CuNi)6Sn5 IMC

pinned Sn grain boundaries thereby retarding Sn-Sn grain boundary sliding and
improving TMF reliability. However, those studies were more focused on the
improving the TMF reliability of Sn-Ag-Cu-Ni solder joints, rather than
determining inﬂuence of Ni additions on solidiﬁcation behavior of Sn-Ag-Cu.
Later, Anderson et.al. showed that Ni additions to Sn-Ag-Cu results in
microstructural reﬁnement very similar to Co and Fe additions [13].
Consequently, it seems that minor additions to Sn-Ag-Cu do have the potential to
improve the service life of solder joints. However, implementation quaternary Sn-
Ag-Cu-X (X = Co, Fe, Bi, In Ni) solders may depend on the ﬂexibility of the.
existing processes in the industry.

Another alloying approach that has received much attention is the addition
of rare earth (RE) elements to existing Sn-based alloys. Typical RE used lie
within the lanthanide series of the periodic table. Among those La and Ce are
most commonly used elements [2, 53]. Such RE additions are found to improve
both ductility and strength [53]. Furthermore, RE additions also result in a

reﬁnement of Sn dendrites and IMO particles [53]. In Sn-Ag-Cu alloys RE

15

additions decrease the size of both CueSn5 and AggSn IMCs [53]. Furthermore,

reﬁned Sn dendritic structure observed in Sn-Ag-Cu-RE alloys also suggests that
RE elements act as nucleating agents for Sn [53]. Wu et.al. suggested that such
reﬁnement is caused by the presence of RE at different interfaces (namely

IMC/Sn , Sn-Sn grain boundaries etc.) [53]. This was further conﬁrmed through

isothermal aging studies by Pei et.al. They showed that coarsening of A938n

IMC is retarded by La additions [54]. Clearly alloying with RE improves
mechanical properties and microstructural stability of the base alloy. However,
cost and availability of RE might be an issue in their implementation.
1.3.2 Composite approach
Composite solder is another technique which is being vigorously pursued by our
group and other researchers [2]. The prime motivation for this approach is to
engineer microstructure of commonly used solder alloys so as to improve creep
properties and TMF resistance [2]. This approach involves incorporation of
reinforcements in the solder matrix without increasing the melting point of the
material [2]. Additionally, such reinforcements have the potential to pin grain
boundaries and consequently improve the TMF resistance of Sn-based solder
joints.

Presently, there exists a variety of techniques which fuIﬁIls the

aforementioned requirements. One such technique involves incorporation of

compatible IMC, e.g Cussns, FeSn / FeSn2, and Ni3Sn4, reinforcements by in-

situ or mechanical mixing methods [2, 3, 52]. Isothermal aging studies carried to

16

evaluate microstructural stability of such IMCs showed that FeSn / FeSn2, and

Ni3Sn are relatively more stable than Cu68n5 [2, 3, 52]. Nonetheless, IMC

reinforcements have the tendency to coarsen, which would be detrimental to the

mechanical reliability of solder joints. On the other hand, Cussn5 reinforcement

improved creep resistance of Sn-3.5wt%Ag solder joints as compared to FeSn /

FeSn2, and Ni3Sn4, [2, 3]. Even though Cu68n5 improves creep resistance, its

applicability as a potential reinforcement was limited by the fact that it coarsens

during TMF [2, 3, 21].

These studies on IMC reinforcements (Cu68n5, FeSn I FeSnz, and

Ni3Sn4) showed that microstructural stability is a problem, since the

reinforcements tend to coarsen rapidly during isothermal aging. Consequently,

other types of inert, and non-coarsening reinforcements have been utilized. Fe

and ceramic (Al203, TiOz) reinforcements fall in such a category. These

reinforcements improved the strength and creep resistances of solders [2, 55].
However, viability of such an approach is limited by the fact that those
reinforcements do not bond with the Sn-rich matrix, and tend to agglomerate
during the reﬂow process [2, 55]. In an attempt to reduce agglomeration,
reinforced solders were rolled to CI'USI'I and disperse the particulates. In addition
to adding an extra processing step, this approach resulted in interfacial cracking

between the reinforcement and the matrix [55].

17

Another approach proposed by Subramanian et.al. was the incorporation
of micro-sized Ag, Cu and Ni metallic particles to Sn-Ag based solders [2, 3, 55].

Ag, Cu, and Ni reinforcements bonds with Sn-rich matrix though the formation of

interfacial IMCs Ag3Sn, Cu-Sn (CU3Sn, and CU58n5). and (CuNi)58n5,

respectively. In terms of microstructural stability Ag reinforcements proved to be
better than Cu and Ni [2]. However, Ni reinforced Sn-3.5wt%Ag solder joints
demonstrated improved creep resistance as compared to Cu and Ag
reinforcements [2, 3]. Although Ni particles improves the creep resistance but it
is difﬁcult to control their size and distribution during the reﬂow process [55]. This
could be a source of concern for microelectronics industry which is increasingly
driven towards miniaturization [55]. Consequently, a requirement for sub-micron
size reinforcements arises.
Based on the literature review, it is found that a potential reinforcement

should satisfy the following minimum requirements:

1. Does not coarsen during service,

2. Bonds with the Sn-rich solder matrix

3. Pin grain boundaries

4. Provide ease of processing without introducing additional costly steps, and

5. Sub-micron sized and have relatively uniform distribution throughout the

matrix

18

 

1.3.3 Development of nano—composite solders using nano-structured
chemicals
Polyhedral oligomeric silsesquioxanes (POSS) are nano-structured Chemicals

with an average diameter of 1.5nm [55]. These Chemicals are cage like structure

with repeated monomer units of SiO1.5 with a hydrocarbon (R) group, eg. Iso-

butyl, phenyl etc., attached to the Si atoms [55]. Examples of POSS have been
shown in Figures 1.3.1. These ﬁgures show that POSS can be either a complete
cage (Figure 1.3.1 (a)) or an incomplete cage with silanol (Si-OH) bonds
replacing a Si atom at the corner (Figure 1.3.1 (b)). Schematic shown in Figures
1.3.1(b) are among of the several structural variations of POSS [56, 57]. Utilizing

appropriate chemical synthesis routes one can tailor the number of Si-OH bonds

and Si atoms attached to the Sin5 cage [58]. However, for this review and the

studies presented in the following sections (and Chapters), role Si-OH bonds are
ofconsequence.

POSS with silanol groups are known to form thermodynamically feasible
bonds with Sn (Si-O-Sn) [55]. Essentially, these bonds are formed between
oxygen and metal atoms following breakage of O-H bonds, while Si-O bond
remains intact. This is evident from Table 1.2, which compares different oxygen-
metal bond energies with that of Si-O and O-H bonds present within POSS.
Consequently, POSS with surface active Si-OH presents itself as a viable
alternative reinforcement, since it can form bonds with Sn-rich solder matrix.

Furthermore, bonded POSS cages will not coarsen during service since that

19

would involve breaking very strong bonds. This suggests that bonded POSS

cages are inert during service. Additionally, POSS with silanol groups are

thermally stable till 300°C [55, 59, 60]. This temperature is considerably higher

that the melting points of most Sn-based solder alloys and processing
temperatures used for fabricating soldered joints with such alloys. As a result
degradation of the POSS structure is unlikely during processing.

Effect of POSS reinforcements on the thermo-mechanical behavior and
TMF service mechanical reliability of Sn-3.5wt%Ag solder joints has been
investigated by Lee et.al [55]. Their experiments strongly suggested that POSS
with surface active Si-OH bonds with the solder matrix, and does not

agglomerate during the reﬂow process indicating ease of processing.

20

 

 

: :i (a)

R Organic Group

 

R Organic Group

R

  
   

   

.0

Figure 1.3.1 Schematic showing two different types POSS anatomies (a)
Trisilanol POSS with three Si-OH groups, (b) Octa POSS without any Si-OH

groups.

21

Table 1.2 Bond energy values at 25°C [61].

 

Figure 1.3.2 SEM images showing microstructural damages after 1000 cycles of

TMF in (a) eutectic Sn-3.5wt%Ag (adapted from Lee at.a|. [21]). and (b) eutectic

22

Sn-3.5wt%Ag reinforced with 2wt% (10 vol%) POSS with Si-OH groups (adapted

from Lee et.al. [55]).

 

23

Monotonic shear testing revealed that POSS reinforced solder joints were
relatively insensitive to strain rates, suggesting that the reinforced matrix is able
to accommodate imposed stress/strain. Finally, TMF studies showed that, POSS
pins Sn grain boundaries from sliding or debonding [55]. Figures 1.3.2 compares
the microstructures of solder joints with and without POSS reinforcements after
1000 cycles of TMF. Figure 1.3.2(a) shows surface damage manifestation in
form of grain boundary sliding in Sn-3.5wt%Ag, while reinforcing with POSS
results in lack of such damages (Figure 1.3.2(b)).

Evidently, incorporation of POSS reinforcements within solder matrix
presents an attractive potential option for implementing composite solder. To
establish the viability of POSS as reinforcement, the following studies on Sn-
based Pb-free solders are warranted:

1. Effect of P088 additions on the TMF behavior of solder joints
2. Inﬂuence of P088 additions on the microstructural features and its

stability during service.

1.4 Research strategy

Based on this literature review it was noted that temperature regime, within which
solder joints experience thermal excursion, might inﬂuence the mechanical
reliability of Sn-based solder joints. To validate this hypothesis, Sn-based solder
joints have been subjected to a lower temperature regime as compared to
previous studies [1, 20, 21, 32, 42]. For this purpose, model Sn-3.5\Art%Ag solder

was utilized. Effect of low temperature regime TMF was evaluated through

24

 

microstructural and mechanical characterization techniques. Results of this
investigation and the corresponding discussion are presented in chapter 3.
These results would serve as a baseline for evaluating the effects of POSS
reinforcements on mechanical reliability of Sn—based solders.

Chapter 4 presents the study on the effects of POSS additions on the TMF
behavior of Sn-based solder alloys. However, only the effect of low temperature
TMF is studied, since Lee et.al. has already investigated the effect of high

temperature regime TMF on POSS reinforced Sn-3.5wt%Ag solders [55]. The

methodology for characterization in this chapter is similar to the ones in Chapter 3.

As pointed out earlier, any additions (alloying elements or reinforcements)
that can bond with the metal matrix can inﬂuence the microstnrcture of the base
alloy. Consequently, Chapter 5 assesses the effects of POSS additions on the
microstructure of bulk specimens.

Microstructural stability during service is one of the requirements in the

development of Sn-based solder. This aspect is investigated in chapter 6 by

observing the effects of POSS on the coarsening behavior of Ag3Sn IMC.

25

'L

Chapter 2

Experimental Procedure and methods

2.1 Materials and processing

Eutectic Sn—3.5wt%Ag (SA) and Sn-3.0wt%Ag-0.5wt%Cu (SAC305) solder

pastes (Kester ®) were used in this study. Each variety of paste contained of

~89 wt % alloy while the rest was water soluble ﬂux. Commercial brand of the
solder pastes was R520A. Nanostructured Chemical used were (i) Iso-Butyl
trisilanol POSS, and (ii) lso-Butyl Octa trisilanol POSS. These Chemicals were
obtained from Hybrid Plastics®. All the pastes and Chemicals were kept in a
refrigerator when not used.

Nanostructured POSS was added into the solder pastes through

mechanical mixing at room temperature (~25°C). Glass petri dish and stainless

steel spatula were utilized for mixing, and were thoroughly Cleaned with isopropyl
alcohol prior to usage. The mixing procedure involved two steps. First, ~10
grams of solder paste was weighed in a petri dish using a precision microbalance.
This was followed by addition of ~0.2 grams of a particular type of nanostructured
POSS Chemical (usually white colored powder) to the paste. This quantity of
POSS was 2% of the total weight of the solder paste used. Finally, the solder
paste and the chemical were thoroughly mixed in the glass petri dish for ~20
minutes with a stainless steel spatula till the whitish powder is not visible. The

methodology adopted for adding POSS to solder pastes offers both ease of

26

processing and economically viability. Following preparation all the solder pastes

were stored in a refrigerator.

2.2 Preparation of Solder Joints

Copper dog-bone halves (500pm thick) were initially washed with 30% volume
HN03 to remove surface oxide. This was followed by rinsing them in distilled

water and ﬁnally isopropyl-alcohol. Two “Cleaned” Cu dog-bone halves were

placed in an aluminum ﬁxture (Figure 2.2.1(a)) with the narrower ends

2
overlapping an area of ~1 mm with solder paste between them and spaced

vertically by ~100pm by 600pm glass spacers (Figure 2.2.1 (a)). Geometry and
dimensions of the fabricated solder joint have been shown in Figure 2.2.1 (b).
The ﬁnal assembly had appropriate amount of solder paste sandwiched between

the overlapping Cu dog-bones. The whole assembly was then heated up to
250°C on a hot plate which was preheated to 500°C. Heating / cooling proﬁle

used for this purpose are shown in Figure 2.2.1(C). After preparation all the

specimens were stored in a commercially available refrigerator.
2.3 Preparation of bulk alloy specimen

Bulk specimens were prepared by placing solder paste in an aluminum cup of

ﬁxed diameter then heating the whole setup at 280°C. Temperature in all the
experiments was controlled with a commercially available digital temperature
controller. Figure 2.3.1 shows the experimental setup used for melting the bulk

solder. Specimens were kept on the hot plate for approximately 15 minutes to

ensure the all the ﬂux has evaporated, leaving behind just the molten alloy inside

27

the cup. In order to solidify the molten alloy, the cup was quickly removed from
the hot plate and placed on a surface which was maintained at a constant
temperature. Details of the temperatures used would be provided in Chapter 5.
After solidiﬁcation bulk specimens were washed with water and isopropyl alcohol
to remove the excess ﬂux from the specimen surface. The solidiﬁed specimens

were then stored in a commercially available refrigerator.

28

 

Cu strips

Glass plates

     
 
 

 

 

 

 

Thermocouple
wire

 

 

Heated aluminum ﬁxture

(a)

Solder joint

Figure 2.2.1 Apparatus used for fabricating solderjoints (a) aluminum ﬁxture
used for positioning the hatf dog bone copper strips, (b) schematic of a solder

joint with dimensions shown. (Adapted from J G Lee thesis [1]).

29

 

300 I l I I l

 

 

 

 

 

250 — -
A . A Melting point (SA), 221°C
,9 200 ' ‘ Melting point (SAC305), 217°C
v T ..
Q
E
g 150 — J
0
E“
q, 100 — _
{-1
50 .
0 I

 

 

0 50 100 150 200 250 300
Time (sec)

(0)
Figure 2.2.1 (C) Temperature proﬁle used for fabricating solderjoints . (Adapted

from J G Lee thesis [1]).

30

I Power source

A

 

  
   
 
     

Heater coil
Temperature
controller

Thermal diffuser
Ove/ri (steel plate)

Figure 2.3.1 Schematic showing the experimental set-up used for the preparation

of bulk specimens.

    
  
 
 
 
 
    

. Solderjoint Fixed SUPPOI"
Backrng c0pper stage
strip
Backing copper
strip
Movable pIece ovable piece

Screws
Figure 2.4.1 Schematic showing the jig used for holding solder joints during
polishing. The wider ends of the joint are backed up with Cu strips of similar
thickness as that of the half dog-bone pieces. This prevents bending or twisting

of the joint (at the center) during polishing.

31

2.4 Procedure for polishing joint and bulk specimens

Prior to polishing as-fabricated solder joints were placed on a custom built jig

(Figure 2.4.1) so as to keep it steady during the polishing procedure. All joints

were polished sequentially with 800-2400 grit papers, followed by 0.3pm Al203

and 0.05pm colloidal-Si02 solutions.

However, bulk specimens were polished sequentially with 200, 800-2400

grit papers, followed by 0.3pm Al203 solution. Polishing of bulk specimens with

0.3pm AI203 solution was carried out in two stages. The ﬁrst stage involved

polishing the specimen vigorously to remove SIC particles embedded on the
surface from the previous polishing step. This additional step decreases the time
required for the polishing process, but leaves surface scratches. Consequently,

in the next step, the polishing cloth was again Cleaned and then

the specimen was polished in a slow and gentle manner with 0.3pm Al203

solution. This step ensured a mirror like ﬁnish to the specimen surface. Final

polishing step was carried out with 0.05pm colloidal SiOz solution, where the

specimen was polished as gently as possible so as to avoid any scratches.

To avoid debris from the previous polishing steps, specimens (joint or bulk
specimens) were ultrasonically cleaned after each step of polishing. After
polishing all the specimens were stored in a commercially available refrigerator

maintained at 4°C.

32

2.5 Setup for TMF experiments

Solder joint specimens were thermally cycled inside a microclimate

environmental chamber (Cincinnati Sub-zero®) (Figure 2.5.1). This Chamber

has an inbuilt air Circulatory system, which maintains uniform temperature
throughout the chamber. Prior to thermal cycling, solder joints were placed on
groves of a custom built stage (Figure 2.5.2). This set-up prevented the joints
from displacing under the inﬂuence of the Circulating air inside the Chamber.

Finally a programmed temperature proﬁle was executed to thermally cycle the

specimens between -55°C and 125°C. Figure 2.5.3 shows the comparison

between the programmed and the actual temperature proﬁle. Each cycle had an

average heating rate of 3.9°C/min with a dwell time of ~10 min at 125°C; along
with an average cooling rate was 2.4OC/min and the dwell time was ~45 min at -

55°C. Consequently each thennal cycle would take ~3 hours. To characterize

the effects of TMF on solder joints due to such thermal cycling a Chosen number

of specimens were taken out the Chamber after 1, 25, 50,100, 200, 300, 400, 500,

33

T

 

Figure 2.5.1 Image showing the environmental chamber used for conducting

TMF experiments

 

Figure 2.5.2 Image showing the custom built stage used for placing solder

specimens inside the environmental Chamber.

34

 

  
   
   

L 1 l n l l l 1 l
140 - ~10 min — Chamber temperature
120_ proﬁle with specimens
. inside
100 1 -- -- Programmed temperature
A 4 proﬁle
9 80 - o .
ﬁg 60‘ 3.9 C I mIn
E 40; -2.4°C / min
3 .
E 20 -
l2 .
o -
-20-
-40 -
-60 -

 

 

 

O 50 1 00 1 50 200

Time(Minutes)

Figure 2.5.3 Temperature proﬁles used for thermally cycling the solderjoints.

35

and 1000 cycles. These specimens were temporarily stored within a commercial

refrigerator at 4°C prior to characterization.

2.6 Mechanical and microstructural Characterization

Mechanical characterization of solder joints were carried out using Rheometric
Solids Analyzer (RSA-Ill) which is a highly force sensitive tensile testing type
machine. Figure 2.6.1 shows the set-up used for mechanical testing. Batches of
solder joints, which were thermally cycled for 1, 10, 25, 50, 100, 200, 300, 400,

and 500 TMF cycles, were strained to failure for this mechanical characterization

at 25°C with a simple shear strain rate of 10-33-1. This provided the peak force

(in Newtons) required for breaking the solderjoints. Optical microscope was used
to measure the surface area of the fracture surface. This was necessary because
there are voids present in the as-fabricated solder joints, whose area had to be
subtracted from the total joint surface area. Measurements of areas were carried
out by processing of digitally acquired optical images with |mageJ® software.
The resultant area of the fracture surface (excluding void) was used to calculate
the peak stress.

Microstructural Characterization was carried out with a JOEL 6400
Scanning Electron Microscope (SEM). For this purpose, representative solder
joints were metallurgically polished (c.f. section 2.4) on one side, prior to placing
them in the environmental Chamber. These specimens were taken out of the

chamber at periodic intervals and the resultant surface damage was documented

36

using SEM. Microstructures of bulk specimens were characterized with both

SEM and Optical Microscope (OM).

37

 

(b)

Figure 2.6.1 Set-up used for mechanical Characterization (a) RSA lll used, and

(b) location of the solder joint inside the specimen holder of RSA III

38

Chapter 3

Effect of temperature regimes on the thermomechanical
fatigue behavior of eutectic Sn-Ag joints

Rationale and objectives

Service parameters associated with thermal excursions are known to inﬂuence
the thermomechanical fatigue (TMF) behavior of solder joints. Parameters
related to thermal excursions, like temperature difference, heating / cooling rates,
dwell times at temperature extremes etc. are known to affect the mechanical
reliability of Sn-based solder joints. However, temperature regimes, between
which solder joints experience TMF, may also inﬂuence their residual mechanical
behavior. The reason for such inﬂuences is due to the fact that deformation
modes of Sn-based solders are a strong function of the ambient temperature.

Consequently, the present study investigates the inﬂuence of low temperature
regime TMF between -55°C and 125°C on eutectic Sn-3.5Ag solder joints. This
study is carried out using identical specimen geometry as those used for studying

high temperature regime TMF between -15°C to 150°C. This Chapter presents

microstructural evolution with number of TMF cycles, and mechanical
Characterization of residual strength realized under low temperature regime TMF,

and compares them with those observed under high temperature regime TMF.

39

3.1. Background

Mechanical reliability of Sn-based solder joints is known to deteriorate due to
thermal excursions experienced during service [23]. This deterioration of
properties is referred as thermomechanical fatigue (TMF). The degree and
manner of property degradation attributed to TMF are inﬂuenced by thermal
service environmental parameters such as temperature difference between
extremes, dwell time at temperature extremes, heating and cooling rates, etc [21,
42]. The roles of each of these service parameters on different Pb-free Sn-based
alloys have been investigated using joints of same geometry, and varying a
single service parameter while holding all other parameters ﬁxed [22]. The extent
of degradation of these individual parameters are evaluated by comparing
residual properties, such as strength and electrical conductivity after exposing
these joints to varying number of TMF cycles [22]. The manner of degradation is
evaluated by monitoring surface damages accumulation of a single specimen
after exposing it to different number of TMF cycle [19-21, 42]. In addition to these
service parameters, the mechanical testing conditions used to evaluate residual
properties can also inﬂuence the ﬁndings since the mechanical responses of Sn-
based alloys are temperature and strain-rate sensitive [6]. This temperature and
strain-rate sensitivity inﬂuences on defamation of Sn-based alloys suggest that
the TMF behavior of these alloys may be affected by the temperature regimes
and extreme temperatures encountered during thermal cycling.

There are reported studies on effect of thermal cycling of various solder

alloys at temperature regimes of -40°C to 125°C, and -15°C to 150°C [24, 51,

4o

r—'--I

62]. Although the imposed temperature difference in both these studies was the
same, specimen geometry, type of solders and substrates used, dwell time at
temperature extremes and heating/cooling rates, etc. were different from each
other [24, 51, 62]. Therefore, it is difﬁcult to use these results to assess the
solder joint reliability due to TMF as affected by the temperature regimes. In the

past several years, systematic studies using identical single shear-lap solder joint

geometry have been carried out in the temperature regime of -15°C to 150°C to

evaluate the effects of heating rate and dwell times on the TMF behavior of
several Sn—based solders including eutectic Sn-3.5wt%Ag solder [3, 18-24, 42].

The present study investigates the responses of the same eutectic Sn-3.5wt%Ag

solder joints subjected to thermal cycling at temperature regime of -55°C and

125°C, while using the same joint geometry with similar dwell time at

temperature extremes and ramp rates used in prior studies, so that effects of
temperature regime can be compared. The temperature difference used here is

slightly different from prior studies due to the following reasons. It is known that

the deformation behavior of Sn Changes in the temperature range of 100°C to
125°C [7]. So it is essential to keep the upper temperature extreme at least at
125°C to avoid unnecessary complications from different deformation modes. In

addition, thermal cycling between ~55°C and 125°C is currently used as the most

severe thermal testing standard established by the Joint Group on Pollution

41

Prevention for Pb-free solder alloys. Hence the lower temperature extreme of

-55°C was Chosen for this study.

3.2. Experimental details

Single shear-lap solder joints were prepared using commercially available Sn-
3.5wt%Ag solder paste purchased from a commercial source. Machined spacers
used during fabrication of joints helped to maintain a relatively constant joint
thickness of about 100 :I: 5pm. The exact thickness of joints was measured with
an optical microscope for determining the displacement rates and shear strains
needed for mechanical Characterization. Fabrication details of these solder joint

specimen are described in chapter 2. All solder joints were thermally cycled from
-55°C to 125 C using a bench top model of MicroClimate® test Chambers by

Cincinnati Sub-Zero with a programmable controller. Test specimens were

placed in the environmental chamber that was initially at -55°C. It is then heated

with an average rate of 39°C per minute to 125°C, and held at this temperature

for 10 minutes. Following this high temperature dwell specimens were cooled to

-55°C with an average cooling rate of about 2.400 per minute. The dwell time at

-55°C was set for 45 minutes. This thermal proﬁle was cycled during TMF testing.

Programmed and the actually measured temperature proﬁles experienced by

specimens have been described in chapter 2.

42

Microstructural Characterization was carried out using a Scanning Electron
Microscope (SEM). For this purpose, representative Sn-3.5wt%Ag solder joints
were metallographically polished on one side, prior to placing them in the
environmental Chamber. These specimens were taken out of the chamber at
periodic intervals and the resultant surface damage was documented using SEM.
Mechanical Characterization was carried out using a mechanical solids analyzer
(RSA-III) by TA Instruments Batches of Sn—3.5wt%Ag solder joints, which were

thermally cycled for 1, 10, 25, 50, 100, 200, 300, 400, and 500 TMF cycles, were

strained to failure for this mechanical characterization at 25°C with a simple

- -1 ,
shear strain rate of 10 3s . Eight to ten specimens were used to study each

condition so that the results obtained provide realistic trends. Fractured surfaces
of the mechanically tested specimens were examined to identify the facture path,
and to obtain the actual joint area that contributed to the strength of the joint. This

methodology is well documented in prior publications [19, 20]
3.3 Results and discussion

3.3.1Microstructural Evaluation of Damages resulting from Low
Temperature Regime TMF

To assess the evolution of surface damage, three regions were Chosen on
the polished side of the representative solder joint. Schematic provided in Figure
3.3.1 indicate the locations of these regions, which, hereafter will be referred to

as regions 1, 2 and 3. Microstructural evolution in these regions with increased

43

number of TMF cycles is presented in Figures 3.2, 3.3 and 3.4. All these images

reveal that damage accumulates from very early stages of TMF, i.e. within the

  
 
  
  

IM Layer

ours no

Region -1

 

 

 

Region -2

Figure 3.3.1 Schematic of the regions on the solderjoint observed for
microstructural evolution. Region-1 is at the center, region-2 is ~100pm below

region-1 and region-3 is at the IMC / Solder interface.

45

ﬁrst 50 cycles, unlike in specimens subjected to TMF in -15°C to 150°C

range [21, 42]. In that high temperature regime TMF, surface manifestation of
damage was ﬁrst noted only after about 250 cycles [21 , 42].

The surface manifestation of damage noted in the current study occurs in
two ways. First type of surface manifestation is in the form of relief features within
the solder region. Such features were observed in regions 1, and 2 (Figures 3.3.2
and 3.3.3). In region 3, which is very near the corner of the joint, this feature was
absent throughout the entire duration of this study (Figure 3.3.4). In region 1,
such damage appears to proceed from the center of the joint towards the IMC /
solder interface at an angle of approximately 30° with the IMC layer, whereas in
region 2, damage runs almost parallel to the IMC/solder interface. These
observations suggest that surface relief features within the solder matrix may be
due to thermal stresses arising from the anisotropy of Sn during TMF.
Furthermore, the localized nature of surface damage is due to the inability to
dissipate these stresses by plastically deforming the Sn grains. Hence grains
move relative to each other by mode-II or mode-III types of fracture.

The second type of surface manifestation noted is the debonding between

the Cussn5 IMC and Sn grains. Such debonding occurs both at the solder matrix
and the CU5Sn5 IMC layer it forms with the substrate, as well as within the solder

matrix and the Cu6$n5 particles embedded in it. This type of surface

46

manifestation is observed prominently after 100 cycles in all three regions of the

joint.

47

I
.‘

1

._3_,____._— IMC Layer

3. .
/ —-/~" '

'f. :1 . Solder

 

Figure 3.3.2 SEM images of region 1 showing the evolution of surface damage
within the solder and near the IMC layer and at the Cu68n5 IMC / solder interface
after (a) 0, (b) 50, (c) 100, (d) 200, (e) 300, (f) 400, (g) 500 cycles of TMF. Scale

bar indicates 50pm.

48

50 Cvcles

7

 

Figure 3.3.3 SEM images of region 2 showing the evolution of surface damage
near the Cu68n5 IMC layer and at the Cu53n5 IMC Isolder interface after (a) 0, (b)
50, (C) 100, (d) 200, (e) 300, (f) 400, (g) 500 cycles of TMF. Scale bar indicates

50pm.

49

 

‘ A
(d) 200 males (e) 300 Cycles (f) 400 Cycles (9) 500 Cycles

Figure 3.3.4 SEM images of region 3 showing the evolution of surface damage
at the IMC/solder interface after (a) 0, (b) 50, (C) 100, (d) 200, (e) 300, (f) 400, (g)

500 cycles of TMF. Scale bar 10pm.

50

Debonding
IMC layer ‘
BE Image ’ m

, ~ BE Image

Debonding

Cu t" IMC Particle

 

_ I: 4, IMC layer .

Figure 3.3.5 High magniﬁcation images showing debonding at the Cu6$n5 IMC

layer I solder interface at two different regions after 500 cycles (a) Location near
region 1 and (b) is the corresponding BE image. SEM image of region 3 is shown

in (C) and its corresponding BE image in (d).

51

Micrographs shown in Figures 3.3.2, 3.3.3 and 3.3.4, depict the progression of

interfacial debonding between Cuesn5 lMCs and Sn matrix with increased

number of thermal cycles. High magniﬁcation SEM micrographs of these

interfaces after 500 TMF cycles suggest that fractures occur in two different

modes. In one case, Sn grains protrude over the surface of CueSn5 IMCs by

mode-II or mode-III type of fracture at their interface, while in other case cracks
form in between solder and IMCs by mode-I type fracture and open up. It is
important to note that there are no indications of plastic deformation in the Sn
grains adjacent to these lMCs. These features can be observed in the

micrographs presented in Figures 3.3.5. Such observations are indicative that

the thermal stresses generated between the Cuesn5 IMC and the adjoining Sn

grains could not be accommodated by deformation within the Sn grains prior to
crack formation, unlike in high temperature regime TMF [19, 21, 42].

3.3.2 Effect of temperature regime of TMF on surface damage accumulation
In order to compare the effects of temperature regimes on the TMF behavior Sn-
3.5wt%Ag solder joints, SEM images from published literature have been utilized

in this section.

High temperature regime TMF between -15°C and 150°C, irrespective of

whether the long dwell was at the high or low temperature extreme, causes

intense microstructural damage within the solder region near the Cu68n5 layer

present at the solder/substrate interface as shown in Figure 3.3.6. This type of

52

microstructural damage within the solder is completely absent in specimens

subjected to the low temperature regime TMF between -55°C and 125°C.

Surface damage features resulting from high temperature regime TMF are

similar, irrespective of whether it was from TMF with long dwell at low

temperature extreme (45°C) or with long dwell at high temperature extreme

(150°C). Under such conditions, microstructural damage is localized within the

Sn matrix present at a region Closer to the solder/substrate interface IMC layer
(Figures 3.3.6 (a) and 3.3.6 (b)). Such damages resulting from high temperature
regime TMF consists of surface manifestations of grain boundary sliding and/or
relief, and grain-boundary decohesion, all of which associated with signiﬁcant
extents of plastic deformation within the Sn grains. Illustrations of these
manifestations are presented in Figures 3.3.7. In Figure 3.3.7 (a), Sn grain
boundary decohesion is observed at approximately one grain diameter away
from the IMC layer and solder interface, while Figure 3.3.7 (b) presents surface
damage due to grain boundary sliding and/or relief.

On the other hand, microstructural damage due to low temperature regime
TMF is marked by the lack of observable plastic deformation within Sn grains.
This is consistent with the observation reported by Rhee et.al. that mode of
deformation and hence microstructural damage in Sn is strongly dependent on
temperature [7]. As a consequence, at the low temperatures encountered in the
present study absence of signiﬁcant plastic deformation within the Sn grains

plays an important role. The thermal stresses arising from TMF are basically

53

causing grain boundary debonding within Sn, and interfacial debonding between

lMCs and Sn grains.

54

1M
‘\‘.

 

I
i
i
it.
It .
I
l
r;
I

(a) (b)

Figure 3.3.6 SEM images showing surface damage on Sn-3.5wt%Ag Solder
joints after 1000 cycles of high temperature regime TMF with long dwell at (a)
high temperature extreme, and (b) low temperature extreme. Adopted from

Subramanian et.al. [21, 42].

55

’.
In
.3
I.“
” I
It
.3
0.
II
'4'

.g'

 

Figure 3.3.7 SEM images showing the effect of high temperature regime TMF on
surface damage features near the IMC layer/ solder interface, (a) Sn-Sn grain
decohesion and (b) grain boundary sliding / relief and shear banding. Adopted

from Choi et.al.[19].

56

3.3.3 Inﬂuence of temperature regime of TMF on the residual mechanical
properties

Effect of low temperature regime TMF on the residual mechanical strength is
shown in Figure 3.3.8. Residual shear strength of joints was found to drop
rapidly by ~35% within ﬁrst 100 cycles, and ~ 50% after 500 cycles, as compared
to the initial shear strength.

This drastic decrease in the residual shear strength can be attributed to
the appearance of cracks at the interface of IMC layer and solder matrix in the
very early stages of TMF. Evidence of such interfacial debonding on mechanical
properties comes from the fracture features in a mechanically sheared specimen
as shown in Figures 3.3.9. The fracture path of this Sn—3.5wt%Ag solder joint,
subjected to 500 cycles of TMF in the low temperature regime, seems to have
initiated along the IMC layer and solder interface near the comer (region 3), and
proceeds predominantly along the interface. Sn-Sn grain boundary decohesion
observed in region 1 may cause some deviations of the crack path. This ﬁnding
suggests that these interfacial cracks developed at interfaces during TMF cycles,
had caused the failure to take place at those locations during mechanical testing.
Since the imposition of thermal stresses arising from this low temperature regime
TMF are not accommodated by plastic deformation, to a greater degree as
compared to high temperature regime TMF, within Sn grains, relaxation of
thermal stresses that one normally expects during dwell times [29, 30] at extreme

temperatures may not be a contributing factor.

57

 

-I
-l
q

E? 40' '
g .I
s 3.. -
E

g 304 E .
.2 \I

.2 \i g

g 25- \* i
“a

a? 2,- \I _

 

 

 

f r T T I ' I ' I ' I

0 100 200 300 400 500
erber of TNF cycles

Figure 3.3.8 Plot showing the variation of residual shear stress of Sn-3.5wt%Ag

solder joints as a function of TMF cycles. Error bars represent 95% conﬁdence

intervals.

58

 

0)
+4
(15
L-
IA—I
U)
.Q
3
(D
3
O

 

Figure 3.3.9 (a) Cross sectional view of a mechanically failed Sn-3.5wt%Ag
solder joint;(b) Enlarged view on the of the region inside the dotted rectangle in

(a). This joint had undergone 500 cycles of TMF prior to mechanical testing.

59

8 S 8
l L l I

N
O
I

Average Residual Shear Strength (MPa)
F

 

O
I

 

0 Cycles

Figure 3.3.10 Plot comparing the effect of high and low temperature regime TMF
on average shear strength of Sn-3.5wt%Ag single shear-lap solderjoints after

500 cycles of TMF. Residual strength data used for comparison was obtained

 

Low Temperature Regime

- Long Dwell at -55°C
High Temperature Regime

- Long Dwell at -15°C
|:] Long Dwell at 150°C

 

 

500 Cycles

from [18, 21, 42].

60

Figure 3.3.10 compares effect of low and high temperature regime TMF
on the residual strength of Sn-3.5wt%Ag solder joints after 500 cycles. This plot
also shows the inﬂuence of other service parameters such as dwell times and
. heating rate on residual strength of solder joints subjected to high temperature
regime TMF. Evidently, low temperature regime TMF has a more detrimental
effect on the residual mechanical properties of solder joints as compared to high
temperature regime TMF. These results, in conjunction with microstructural
observations, suggest that interfacial and grain boundary fractures, with lack of
signiﬁcant observable plastic deformation (causing localized surface damage)
within the Sn grains during low temperature TMF, may be the major contributing
factors for the signiﬁcant drop in strength under such conditions. Under these
Circumstances cracks continue to form and propagate with each successive cycle
of TMF. Hence, residual strength does not seem to stabilize even after 500
cycles of low temperature TMF. This is in contrast with ﬁndings of high
temperature regime TMF studies, whereas residual mechanical properties seem

to stabilize after 250 TMF cycles [21, 42].

3.4 Conclusions

The following summarizes the effect of low temperature regime (-55°C and

125°C) TMF on eutectic Sn-3.5wt%Ag solder joints.

1. Surface damage is mainly in the form of debonding between Cu68n5

IMCs and solder, and with some Sn grain boundary fractures, without any

61

observable plastic deformation within the Sn grains. Such minimal plastic
deformation results in localized surface damage within the solder matrix.

These observations are completely different from those noted due to high

temperature regime (-15°C and 150°C) TMF. Such conditions result in

grain boundary events with extensive plastic deformation within Sn grains

at regions Close C6$n5 IMC layer.

Residual mechanical properties continued to drop drastically during the
early stages of low temperature regime TMF, without stabilizing even after
500 cycles, whereas in high temperature regime TMF residual shear
strength of solder joints stabilize after 250 cycles. In addition, magnitude
of the loss in mechanical shear strength is much greater for low

temperature TMF.

62

Chapter 4

Low temperature regime thermomechanical fatigue
behavior of nano—structured polyhedral oligomeric
silsesquioxanes reinforced Sn-based solder joints

Rationale and objectives

Thermomechanical fatigue (TMF) experienced by Sn—based solder joints during
service adversely affects their mechanical reliability. Deterioration in the
mechanical strength of such joints occurs primarily due to interfacial damage.
However, the mode of interfacial damage depends on the temperature regimes
within which solder joints are subjected to TMF. High temperature regime TMF
causes Sn-Sn grain boundary (GB) events such as GB sliding I decohesion,
while interrnetallic compound - solder interfacial decohesion results during low
temperature regime TMF. Consequently, to improve the mechanical reliability of
Sn-based joints, enhanced bonding of the aforementioned boundaries /
interfaces must be realized. The nanostructured chemical polyhedral oligomeric
silsesquioxanes (POSS) with surface active silanol (Si-OH) groups are known to
form thermodynamically feasible bonds with the solder matrix. As a result, POSS
presents a viable solution towards improving the mechanical reliability of Sn-
based solder joints experiencing TMF. This study explores such a possibility by
subjecting trisilanol POSS reinforced solder joints to low temperature regime

TMF, followed by mechanical and microstructural Characterizations of such joints.

63

4.1. Background

In microelectronic applications solder joints are required to experience cycling
between temperature extremes [20, 21, 42, 55]. In Sn-based solder joints,
thermal cycling causes anisotropic strains between the micro-constituents (Sn
grains, intermetallic compounds (IMC) and substrate) resulting in
thermomechanical fatigue (TMF) of those joints [18, 20, 21, 23, 42, 55]. Damage
occurring as a consequence of TMF deteriorates the mechanical reliability of
solder joints [20, 21, 42, 55]. This has motivated the development of several
strategies to improve material properties for enhancing solder service
performance. For that purpose the commonly used approaches are composite
solders and alloying. The former involves incorporation of reinforcements into
the solder matrix, while the later requires minor element additions to base Sn—Ag
or Sn—Ag-Cu alloys. In the present state, both the techniques result in the
formation of lMCs within the solder matrix, which tend to coarsen during service
(section 1.3). Additionally, depending on the type and availability of the
reinforcements / elemental additions there could be either processing or
economic implications.

The latest methodology for improving service reliability of Sn-based
solders involve incorporation of nano-structured, surface active and non-
coarsening reinforcement polyhedral oligomeric silsesquioxanes (POSS) to the

solder matrix [55]. POSS consists of a thermally stable inorganic core made up

of Si and O framework (SiO1,5) [55]. Study carried out by Lee et.al. have shown

that POSS with surface active silanol (Si-OH) ) groups have the ability to bond

64

with the Sn matrix and does not agglomerate during the reﬂow process [55].

TMF studies between -15°C and 150°C revealed that surface damage has been

greatly reduced upon addition of POSS to eutectic Sn—3.5wt%Ag solder joints.
This suggested that Sn grain boundary events, responsible for deteriorating
mechanical reliability, was diminished upon POSS reinforcement [2, 18, 55].
However, the inﬂuence of POSS reinforcement on the lower temperature regime

TMF behavior of Sn-based solder joints is yet to be ascertained. Studies

presented in Chapter 3 have indicated that Cu68n5 layer-solder decohesion

resulting from lower temperature regime TMF between -55°C and 125°C. That

kind of damage had a more detrimental effect on solder joint mechanical

reliability than higher temperature regime TMF (-15°C and 150°C).

Consequently, this study investigates the inﬂuence of POSS reinforcements to

Sn—based solder joints subjected to lower temperature regime TMF between
-55°C and 125°C

4.2. Experimental details

Materials used in this study were commercially available eutectic Sn-
3.5Ag (SA) and Sn—3.0wt%Ag-0.5wt%Cu (SAC305) solder pastes, and nano-

structured chemical Iso-Butyl POSS trisilanol POSS was obtained from Hybrid

Plastics®. 2wt% of lso-Butyl trisilanol POSS was then mechanically mixed with

SA and SAC305 solder paste to prepare two additional pastes of composition SA

65

+ Iso-Butyl trisilanol (SA2iB), and SA0305 +2wt% lso-Butyl trisilanol (SACZiB).
Finally, single shear lap solder joints (Figure 2.2.1 (b)) were prepared using the
four solder pastes, i.e. SA, SA2iB. SAC305, and SAC2iB. Details of specimen

preparation have been provided in chapter 2.

To conduct TMF studies solder joints were thermally cycled from -55°C to

125°C using a bench top model of MicroClimate® test chambers by Cincinnati

Sub-Zero with a programmable controller. Test specimens were placed in the

environmental Chamber that was initially at -55°C. It is then heated with an
average rate of 39°C per minute to 125°C, and held at this temperature for 10
minutes. Following this high temperature dwell, specimens were cooled to -55°C

with an average cooling rate of about 2.400 per minute. The dwell time at -55°C

was set for 45 minutes. This thermal proﬁle was cycled during TMF testing. The
programmed and the actually measured temperature proﬁles experienced by
specimens have been presented in Chapters 2 and 3.

Microstructural Characterization was carried out using a Scanning Electron
Microscope (SEM). For this purpose, representative solder joints were
metallographically polished on one side, prior to placing them in the
environmental chamber. These specimens were taken out of the Chamber at
periodic intervals and the resultant surface damage was documented using SEM.

Mechanical characterization was carried out using a mechanical solids analyzer

66

(RSA-Ill) by TA lnstruments®_ Batches of solder joints prepared with types of

pastes, which were thermally cycled for 1, 10, 100, 200, 300, 400, and 500 TMF

cycles, were strained to failure for this mechanical Characterization at 25°C with

-3 -1
a simple shear strain rate of 10 s at 25°C. Eight to ten specimens were used

to study each condition so that the results obtained provide realistic trends.
Fractured surfaces of the mechanically tested specimens were examined to
obtain the actual joint area that contributed to the strength of the joint. This

methodology has been described in chapter 2.
4.3 Results and discussion

4.3.1 Inﬂuence of POSS reinforcements on the residual mechanical
behavior of Sn-based solder joints

Figures 4.3.1 plots the residual mechanical strength as a function of number of
TMF cycles imposed on Sn-based solder joints with and without POSS
reinforcement. Figures 4.3.1(a), and 4.3.1(b) show the effects of such
reinforcement on residual mechanical behavior eutectic SA, and SAC305 solder
joints respectively.

Regardless of the composition, all solder joints experience drop in their
residual strength within 100 cycles of TMF. However, reinforced solder joints
experiences a lesser drop than their corresponding control SA and SAC305
solder joints. After 100 cycles, effects of POSS reinforcement on the residual
mechanical behavior of solder joints become markedly prominent. Nano-

composite (SA2iB and SACZiB) solder joints retain higher residual strength after

67

100 cycles as compared to their controls. Moreover, the residual strength of
nano-composite solder joints stabilizes after 100 cycles of TMF, while that of SA
and SAC305 drops signiﬁcantly after 500 cycles. Statistical analysis supporting
this observation has been provided in Table 4.1-4.2. Trends comparing the
effects of POSS reinforcement on the average residual shear strengths of SA
and SAC305 solder joints have been presented in Figures 4.3.1(C) and 4.3.1(d)
respectively. These trends were determined through second-order non-linear

least square analysis. Equation used for this purpose is of the form,

N

2
y(N) = yn+ AiZexp(—E)
‘=1 (1)

where, y(N) is the residual strength after N cycles of TMF, “yn” corresponds to
residual strength after 1000cycles, while “tr" and “AI” are ﬁtting constants.
Exponential ﬁt was Chosen because residual strength shows an initial drop
followed by stabilization with progression of TMF cycles. Such a behavior is

indicative of specimens undergoing fatigue [39].

68

 

I r l ' I ' I ' I ' I

I SA

o SA2iB
350‘ ---I\IorI-linearleastsquareﬁtforSA(R2=0.993) ‘
a ° —Non-linear least square ﬁt for SA2iB (R2= 0.976) _'
s 45' ‘
U,
C
.2 40- .
(D
$-
8
.c 35' ~
m 1
E
.E 301 ‘
CD .
GI
a: 25. _

20

 

 

 

 

#TNFcyCIes

Figure 4.3.1 (a) Comparison of residual shear strengths of (a) SA and SA2iB
Error bars show 95% conﬁdence intervals. Second order exponential non-linear

least square ﬁts have also been shown

69

81

 

' I ' I ' I ' I j l

I SAC305
m- 0 MB 4
- .. - hbrI-Iinear'leastsqraeﬁtforSACBOS(R2=0.982).
45 — Non-linea least sqrare ﬁt for SAC2iB (R2=0.943)
40- _

 

Iii 3.3 if
7‘4

II-V

I

III.

I-I-l
I-—-.—l
l-—.—-I

Residual shear strength (MPa)

 

 

B

 

#Tllchycles

Figure 4.3.1 (b) Comparison of residual shear strengths of SAC305 and SACZiB
solder joints. Error bars show 95% conﬁdence intervals. Second order

exponential non-linear least square ﬁts have also been shown.

70

Table 4.1 Statistical analysis of residual mechanical behavior of SA, and SA2iB

solder joints (Figure 4.3.1 (a)). Single tail t-test is used for this purpose.

Analysis is carried out on the basis of p values. This is evaluated by

comparing residual mechanical strength data of two chosen cycles. If

there is a difference in the residual strength at the Chosen number of

cycles, then the p value is less than 0.05, otherwise the two Chosen

data sets are not statistically different. In this analysis, residual strength

data for 100 cycles is compared with 400 and 500 cycles.

 

 

 

 

 

p values
Number of TMF cycles
SA SA2iB
400 0.1554 0.3337
.1
500 6.4786 x 10 0.8766

 

 

 

The analysis shows that in case of SA2iB there is no statistically signiﬁcant

difference between residual strength measured after 100 and 500 cycles.

However, residual strength for SA drops after 500 cycles TMF as compared to

100 cycles.

71

 

Table 4.2 Statistical analysis of residual mechanical behavior of SAC305, and

SACZiB solder joints (Figure 4.3.1). Methodology for analysis is exactly

the same as described in Tables 4.1. In this analysis, residual strength

data after 100 cycles is compared with 400 and 500 cycles for both the

types of specimens and additional 1000 cycles for SAC2iB..

 

 

 

 

 

 

Number of TMF p values
cycles SAC305 SAC2iB
400 0.1213 0.6313
500 0.002 0.123
1000 - 0.223

 

 

 

The analysis shows that in case of SACZiB there is no statistically signiﬁcant

difference between residual strength measured after 100, 500 and 1000 cycles.

However, residual strength for SAC305 drops after 500 cycles TMF as compared

to 100 cycles.

72

 

ResuIts from mechanical Characterization demonstrate that POSS
trisilanol reinforcement improves the residual mechanical behavior of both SA
and SAC305 solder joints within the number of TMF cycles Chosen for this study.
Such an improvement indicates that POSS reinforcements retard the
microstructural damages responsible for the deterioration in the mechanical
strength of control joints. Studies Characterizing the effects of POSS
reinforcement on the microstructural damage due to low temperature regime
TMF have been presented in the following section.

4.3.2 Microstructural features contributing to the low temperature behavior
of POSS reinforced solder joints

This section compares the effects low temperature regime TMF on the
microstructural damage in both nano-composite and unreinforced solder joints.
Based on this comparison, microstructure - residual mechanical strength
relationship is established.

SEM images presented in Figures 4.3.2(a)-(C), Figures 4.3.3(a)-(C),
Figures 4.3.4(a)-(C), and Figures 4.3.5(a)—(c) shows the surface damage
evolution after 0, 100 , and 500 cycles in SA, SA2iB. SAC305, and SAC2iB
representative solder joints respectively. After 100 TMF cycles, surface damage
is evident in all the specimens. This corresponds to the observed drop in the
mechanical strength after 100 cycles in both nano—composite and control solder
joints (Figures 4.3.1). However, as noted earlier, after 100 TMF cycles nano-
composite solder joints have a higher residual strength, and that subsequent

thermal cycling does not have any effect on their residual strength.

73

laISAOCycIes ' ‘4 \ (”SA-100 Cvcles \f "

-
,lﬂ

 

(CI SA. 500 Cvcles

 

Figure 4.3.2 SEM images showing surface damage evolution in eutectic Sn-

3.5Ag (SA) solder joint due to low temperature regime TMF between -55°C to

125°C. Surface damages after 100 cycles of TMF has been indicated with

arrows.

74

   

 

 

  
  

(a) SA2iB. 0 Cycles

lb) SA2iB. 100 Cvcles

 

 

 

Figure 4.3.3 SEM images showing surface damage evolution in SA + 2wt% Iso-

Butyl trisilanol (SA2iB) solderjoint due to low temperature regime TMF between

-55°C to 125°C. Surface damages after 100 cycles of TMF has been indicate

with arrows.

75

 

 

(a) SAC305, 0 Cvcles f (b) SAC305, 100 Cvcles

 

 

Figure 4.3.4 SEM images showing surface damage evolution in Sn-3.0wt%Ag-

0.5wt%Cu (SAC305) solder joint due to low temperature regime TMF between

-55°C to 125°C. Surface damages after 100 cycles of TMF has been indicate

with arrows.

76

 

 

(a) SACZIB. 0 Cvcles lb) SAC2iB, 100 Cvcles

 

 

Figure 4.3.5 SEM images showing surface damage evolution in SAC305+2wt%

Iso-Butyl trisilanol (SACZIB) solder joint due to low temperature regime TMF

between -55°C to 125°C. Surface damages after 100 cycles of TMF has been

indicate with arrows.

77

This suggests that POSS reinforcements alter the damage mechanisms
operative during low temperature regime TMF.

Figures 4.3.2 and 4.3.4 show that damages within the solder region of
control specimens appear to be localized in nature, suggesting the inability of the
solder matrix to accommodate anisotropic strains generated during TMF through
plastic deformation of the solder matrix. In comparison, POSS reinforced joints
appear to experience signiﬁcantly more damage in the solder region (Figure
4.3.3, and 4.3.5). To further elucidate the effects of POSS reinforcements, high
magniﬁcation SEM images of the solder region after 500 cycles of TMF have
been presented in Figures 4.3.6. It is apparent from the SEM images that nano-
composite solder joints have accumulated signiﬁcantly more damage than the
control joints. Such extensive damage within the solder region suggests that
POSS reinforcements effectively disperse damage over a larger volume. In other
words, solder matrix in nano-composite solder joints can effectively
accommodate such anisotropic strains through plastic deformation of the solder
matrix. Thereby, diminishing the detrimental effects of TMF cycles on the
mechanical strength of nano-composite solder joints, and possibly contributing to
the stabilizing of residual mechanical strength after 100 cycles.

The second type of surface manifestation observed in control joints are

that of Cu6Sn5 layer I Sn interfacial decohesion. SEM images of SA (Figure

4.3.2(C)) and SAC305 (Figure 4.3.4(C)) after 500 TMF cycles show such

interfacial damage. As mentioned earlier, low temperature regime TMF studies

on SA joints (Chapter 3) have shown that Cu6$n5 layer/ Sn interfacial damage

78

signiﬁcantly contributes to the deterioration in the residual mechanical strength of

such joints. Consequently, it becomes important to assess the inﬂuence of

POSS reinforcement on the Cu6$n5 layer I solder interface during low

temperature regime TMF. SEM images (Figure 4.3.3(b)—(c), and Figures

4.3.5(b)-(C)) of POSS reinforced solder joints incur damage mainly within the

solder region near the Cu6$n5 IMC layer / solder interface. However, such

images do not show the ﬁner microstructural details required for comparison.

Consequeme high magniﬁcation SEM images of the CueSn5 I solder interfacial

region have been presented in Figures 4.3.7. Figures 4.3.7(a)—(b) show the

effects of low temperature regime TMF on CusSns layer l solder interface SA

and SA0305 solder joints respectively after 500 cycles. Cu68n5 layer - solder

interfacial decohesion is apparent in both types of joints, indicating the
vulnerability of such interfaces during low temperature regime TMF. However, a
very different damage feature results upon reinforcing with POSS. High

magniﬁcation images presented in Figures 4.3.7(c)—(d) reveal cracks nrnning

parallel to the Cuesn5 IMC layer, with signiﬁcant surface damages (unlike control

specimens) within the solder region of both SA2iB (Figure 4.3.7(C)), and SAC2iB

(Figure 4.3.7(d)). Additionally, there seems to be minimal cracking at the

Cu68n5 layer - solder interface.

79

.‘

(b) SA2iB I‘, --~

(a) Sn-3.5Ag

 

.“3.

 

Figure 4.3.6 High magniﬁcation SEM images comparing the effects of low

temperature regime TMF on surfaces damages within the solder region after 500
cycles. Nano-composite (SA2iB and SACZIB) solder joints show signiﬁcant
damages within the solder region as compared to Sn-based solder (SA and

SACSOS) joints.

80

Such observations suggest that anisotropic strains due to TMF are

accommodated by the solder region though plastic deformation in nano-

composite solder joints (Figures 4.3.7(C)—(d)), while Cu65n5 layer I solder

interfacial decohesion is responsible for such accommodation in control

specimens (Figures 4.3.7(a)-(b)). Furthermore, such accommodations in nano-

composite joint can only occur if there is a strong bonding between Cu6$n5 and

surrounding solder matrix. In other words, POSS improves CueSn5 I Sn

interfacial bonding in addition to reinforcing the solder matrix. Further evidence

of improved CueSn5 I Sn interfacial bonding due to POSS additions have been

presented in Figures 4.3.8. These high magniﬁcation SEM images show the

inﬂuence of low temperature regime TMF on Cu68n5 particle (within solder

region) / Sn interface in SA (Figure 4.3.8(a)) and nano—composite SA2iB solder

joints (Figure 4.3.8(b)) after 500 cycles. lnterfacial decohesion between Cu58n5

and solder matrix is apparent in SA. However in case of SA2iB. shear bands

along with cracks are observed within the solder region, even though there

seems to be few cracks present at the CU5Sn5 particle - solder interface. Shear

bands in SA2i8 suggests that the neighboring solder region undergoes
considerable plastic deformation due to the imposed anisotropic strains. Similar

shear banding has also been observed within the solder region of SAC2iB near

the Cuasn5 l Sn interface (Figure 4.3.9). It may also be noted that such

81

improved Cu6Sn5 — Sn interfacial bonding also alludes to the presence of POSS

nano-particles at those interfaces. Nonetheless, enhanced bonding between

Cu68n5 layer and solder matrix will contribute to the higher residual strength

demonstrated by POSS reinforced solder joints (Figures 4.3.1).

82

(b) SA2iB
«If; Cussns layer

in

(d) SAC2iB

g I! CusSn5layer
. l

\.

 

Figure 4.3.7 High magniﬁcation SEM images comparing the effects of low

temperature regime TMF on the Cu68n5 layer — solder interfacial regions in Sn-

based and nano-composite solder joints. Cu68n5 layer —solder decohesion is

evident in both SA and SAC305, while damage is primarily located within the

solder region in SA2iB and SACZIB. Scale bar indicates 10pm.

83

Shear bands,

7 4, ﬂ 4 W; .

 

Figure 4.3.8 High magniﬁcation SEM images comparing Cu68n5 particle — solder
interfaces in SA and nano-composite SA2iB after 500 cycles of low temperature

regime TMF. lnterfacial decohesion is evident In SA while shear bands within the

solder matrix near the Cu68n5 - solder interface is observed in SA2iB.

84

t \

m s"°°"°a"d°

3M 3':

5

 

‘5; E

Figure 4.3.9 BE image of SACZIB showing shear bands within the solder region

near the Cu6$n5 — solder interface.

85

4.4. Conclusions

Trisilanol POSS reinforcements improve the mechanical reliability of Sn-based
solder joints experiencing low temperature regime TMF. Such improvements
result due to:
1. Dispersion of damage within the solder region over a larger volume and
thereby facilitating the solder joints to effectively accommodate anisotropic
strains generated during TMF. Such damage dispersion causes plastic

deformation of the solder matrix.

2. Improvement of CueSn5 layer / solder interfacial bonding. This further

suggests the presence of POSS particles at such interfaces.

86

Chapter 5

Inﬂuence of polyhedral oligomeric silsesquioxane
additions on Sn-Ag-Cu bulk specimens

Rationale and objectives
Polyhedral oligomeric silsesquioxanes (POSS) improves mechanical reliability of

Sn-based solder joints by pinning Sn-Sn grain boundaries, and improving

Cussns layer - Sn interfacial bonding. This suggests that POSS additions may

inﬂuence the microstructure of Sn-based solders. However, due to the small
volume of the solder region (in a joint conﬁguration), additional effects of POSS
additions on the microstructure may not be detected. Furthermore, additional
input of Cu within the solder region, due the utilization of Cu substrates, may
further hinder in any assessment of such effects. Consequently, effect of POSS
on bulk solder is investigated. This study was carried out through the

microstructural observation and thermal analysis.

87

5.1. Background

Reinforcing Sn-based solder with nanostructured polyhedral oligomeric
silsesquioxanes (POSS) with surface active silanol (Si-OH) groups has been

shown to retard Sn grain boundary events [55]. Furthermore, results presented

in Chapter 4 indicate improved intermetallic compound (IMC) Cuesn5 layer I

solder interfacial bonding upon reinforcing with POSS trisilanols. Additionally,
such reinforcements also increased the load bearing capacity of the solder matrix
by accommodating stresses generated during thermomechanical fatigue (TMF)

(Chapter 4). These studies suggest that POSS (with Si-OH groups) are present

within the solder matrix and also at the Sn-Sn and Cu6Sn5 I Sn interfaces. This

further suggests that POSS may have segregated at those interfaces during
solidiﬁcation. Such a behavior has been Classically observed during minor
elements additions to alloy systems [1, 32]. However, in majority of the systems,
minor elements form secondary / tertiary phases, which are then found at the

grain boundaries and within the matrix. For example, adding minor

concentrations of Ni to Sn—Ag-Cu results in the formation of (Cu Ni)6$n5 IMC,

which pins not only Sn-grain boundaries but also are found within the solder
matrix [1, 32]. The difference between such minor element additions and
nanostructured POSS trisilanols is that the former results in an easily observable
secondary I tertiary phase within the microstructure, while the later does not.
Average diameter of POSS with Si-OH is 1.5nm, and after bonding with the

matrix, does not coarsen during service [55]. Consequently, it is rather difﬁcult to

88

observe POSS within the solder microstructure. This necessitates other methods
for indirectly conﬁrming the presence of POSS within the solder. Furthermore,
such methods should avoid the geometric constraints posed by the substrates
used in a solder joint.

Studies, thus far, were carried out on POSS reinforced Sn-based solder
joints were prepared with Cu substrates. Dimensions of those joints were
approximately 1mm x 1mm x 0.1mm (chapter 2) [1, 20, 21, 25, 32, 42, 55].

Microstructure of a typical joint consists of approximately 3-5 pm thick (12-20

vol%) interfacial Cussns layer. Further, chapter 4 showed that POSS may prefer
to be present at the Cu68n5 layer I Sn interface, which has a total interfacial area

of ~ 2 mm2 (considering both sides of the joint). So it is expected that signiﬁcant

fraction of POSS additions may be present at those interfaces. This means
additional effects resulting from the interaction of that fraction of POSS with the
Sn-based solder may not be detected. As a result it is necessary to study the
inﬂuence of POSS additions on bulk specimens.

Additionally, Cu substrates, used in joint fabrication, also increase the Cu

concentration within the solder region of the joint by 05-08 wt% [8, 63]. This is

evident from the presence of Cuasn5 particles within the solder region of the Sn-

3.5wt%Ag joints (Figures 3.3.1). Any addition of POSS will deﬁnitely result in its

segregation at the Cu68n5 particles I Sn interface. This could be a very likely

situation for Sn-Ag-Cu solders, which has gained signiﬁcant popularity in recent

89

years [7, 8, 10, 12, 15, 48, 50, 64]. Hence studies on observing the additional
effects of POSS should incorporate this fact for a more realistic assessment.

In view of the above mentioned considerations bulk Sn-3.0wt%Ag-
0.5wt%Cu (SAC305) have been Chosen for this study. Inﬂuence of POSS
additions to SAC305 has been investigated through analysis of microstructure

and differential scanning calorimeter (DSC) therrnograms.
5.2. Experimental details

POSS added solder pastes were prepared by mechanically mixing 2wt% nano-
structured POSS Chemicals to commercially available Sn-3.0wt%Ag-0.5%Cu
(SACSOS) solder paste. Two varieties of POSS were used in this study (i) Iso-
Butyl trisilanol POSS and, (ii) lso-Butyl Octa POSS. Compositions of the POSS
containing specimens were SAC305 + 2wt% lso-Butyl trisilanol POSS, and
SAC305 + 2wt% lso-Butyl Octa POSS. It is to be noted that trisilanol POSS is an
open cage structure consisting of three surface active silanol (Si-OH) groups
(Figure 5.2.1 (a)), whereas Octa POSS is a Closed caged structure with no Si-OH
groups attached (Figure 5.2.1 (b)). Procedural details of POSS addition to solder
paste have been provided in Chapter 2.

Bulk solder specimens were prepared by ﬁrst heating solder pastes

(placed in aluminum cups) at 280°C to evaporate the ﬂux in the paste. The

resulting melt was then solidiﬁed by placing the aluminum cups on surfaces

maintained at two different temperatures for obtaining different cooling rates.

Faster cooling rate of 53°C/min was achieved by placing the melt on a

90

perforated brass block maintained at room temperature, while cooling at a slower

rate, Le 16 Clmin, was carried out solidifying the melt on hot plate maintained at

100°C. Figures 5.2.2 shows the measured temperature proﬁles for cooling at

room temperature and at 100°C. These measurements were carried out by

inserting a K-type thermocouple (with resolution 122°C) inside the molten solder.

Mass of “button shaped” specimens obtained after solidiﬁcation was ~6 grams
(Figures 5.2.3).

Microstructural Characterization was carried out using secondary electron
and optical microscopes. Optical microscope observations are based on two
types of views which have been described in Figures 5.2.2. The planar view is
the ﬂat face which was Closest to the cooling surface, while the cross-sectional
view is obtained by cutting the button shaped specimen perpendicular to the
planar view. Polishing techniques used for microstructural observation have
been described in Chapter 2. For quantitative analysis of the optical micrographs,
ImageJ© software was used.

Thermal analysis was performed using Metller Toledo DSC-1® with
mechanical inter-cooler in sealed crucible. For this purpose bulk specimens
were rolled into thin sheets and small strips (weighing 15 mg :l: 0.1) were cut from
such sheets. The small strips were then encapsulated in aluminum pans and

were placed inside DSC-1 for thermal analyses.

91

 

Figure 5.2.1 Schematic showing two different types POSS anatomies (a) Iso-

Butyl Trisilanol POSS with three Si-OH groups, (b) lso-Butyl Octa POSS without
any Si-OH groups. Chemical formula for organic Iso-Butyl(R) group has also

been shown.

92

 

300 I | I I ' I ' 1
—Air cooled on a perforated brass block
- - Air cooled on a hot plate maintained at 100°C

250 —

200 -

Temperature (°C)

—L

0|

0
I

 

 

 

 

100 . . . . . , . .
0 1 2 3 4 5
Time (mins)

—I

Figure 5.2.2 Comparison of cooling rates utilized for solidifying molten bulk

solder specimens. Solid line indicates the faster cooling rate at 53 C/min, while

the dashed line indicates slower cooling at a rate of 16 Clmin.

93

3D view 0f 35'035t bulk Planar (polished) view : surface in
speCImen contact with hot plate

 

Bottom face

 

 

Cross-sectional (polished) view

an

Cut

 

Figure 5.2.3 Left hand side showing the 30 views of the “button” shaped as-

solidiﬁed bulk specimens. The right hand side shows the regions observed for

microstructural characterization. The square boxes at the center of each view i.e.

planar and cross-sectional, are the locations from where images for analysis

were acquired.

94

I

5.3. Results and discussions

This section presents results from microstructural observations and thermal
analysis. On the basis of those results, effects of POSS additions on bulk
SA0305 specimens have been discussed.

5.3.2 Inﬂuence of POSS additions on the microstructure of Sn-3.0wt%Ag-
0.5wt%Cu

Optical micrographs (planar view) comparing the effects of 2wt% trisilanol and
Octa POSS additions on the bulk microstructure of SAC305 have been

presented in Figures 5.3.1. Specimens shown in these micrographs were cooled

at 53°C/min during solidiﬁcation. It is observed that lso-Butyl trisilanol POSS

addition reﬁnes the as-solidiﬁed microstructure, whereas the lso-Butyl Octa
POSS addition does not produce any observable Changes. The differences in
the microstructure suggest that silanol (Si-OH) groups attached to the POSS
cage interacts with the solder.

Further, trisilanol POSS addition results in a smaller sized dendritic arms
(Figure 5.3.1 (b)), whereas, large dendritic arms are observed in the control
(SAC305) and Octa POSS added specimens (Figures 5.3.1 (a), 5.3.1 (C)).
Quantitative microstructural analysis carried out by measuring the secondary
dendrite arm spacing (SDAS) further substantiates the microstructural reﬁnement
brought about by trisilanol POSS addition (Figure 5.3.2). SDAS measurement is
directly related to the number of dendritic arms shooting out of a primary dendrite

during solidiﬁcation. If there are more secondary and tertiary arms present, then

95

SDAS will decrease, since more number of dendritic arms has to be
accommodated within a given space/volume.

Further evidence of microstructural reﬁnement in trisilanol POSS added
specimens have been presented in optical micrographs of cross-sectional views
in Figures 5.3.3. These images show the microstructures of SA0305, and
SAC305 + 2wt% lso-Butyl trisilanol POSS specimens solidiﬁed at two different

cooling rates. Figures 5.3.5(a), and 5.3.3 (b) show the optical micrographs at a

slow cooling rate (16°C / min), while Figures 5.3.3 (C), and 5.3.3 (d) shows the

optical micrographs at a faster cooling rate (53°C / min). Large primary Sn

dendritic arms are present in SAC305 specimens solidiﬁed at both fast and slow
cooling rates (Figures 5.3.3 (a) and 5.3.3 (C)). However, primary Sn dendritic
arms are barely noticeable in the microstructure of specimens containing Iso-
Butyl trisilanol POSS (Figures 5.3.3 (b) and 5.3.3 (d)). These observations
suggest the presence of highly branched dendritic structures within the
microstructure. Schematic shown in Figures 5.3.4 elucidate the difference
between a dendrite with fewer arms (extended dendrite) and a highly branched
dendrite (equiaxed “condensed" dendrite). This study indicates that such
equiaxed dendritic structure is present within the microstructure of specimens
containing trisilanol POSS (Figures 5.3.3 (b), and 5.3.3 (d)). Similar dendritic
structure has also been reported by Rappaz et.al in their study of Al-Si alloys
inoculated with Ti [65]. Their observation was attributed to the presence of Ti,

which promoted heterogeneous nucleation of primary Al. Such equiaxed

96

dendrites form when primary phase (Sn, in this study) dendrite has sufﬁcient time
to grow and form secondary and tertiary branches before the ﬁnal solidiﬁcation
stage. Such an event can only occur when there is a reduction in undercooling
required for solidifying the liquid melt [66]. A lower undercooling is indicative of
the ease with which a primary phase nucleates from the melt. Consequently,
following nucleation, Sn has more time to grow dendritic arms within the melt.
Figures 5.3.5 compare the measured cooling curves of specimens containing
different types of 2wt% POSS. Specimens containing iso-Butyl trisilanol POSS
have lower undercooling as compared to the control (SAC305). In comparison,
Octa POSS, as expected, did not alter the undercooling required to solidify
SAC305. Furthermore, results presented thus far Indicate that POSS with Si-OH
groups (lso-Butyl trisilanol POSS) are responsible for the heterogeneous
nucleation of Sn during the solidiﬁcation of SAC305 from liquid state.

Trisilanol POSS additions are also found to inﬂuence the morphology and

size of IMC phases (A938n and Cu68n5) present in microstructure of SAC305.

Figure 5.3.6 (a) shows the backscattered electron Image (BEI) temary eutectic
region in the as-solidiﬁed microstructure of SAC305, while Figure 5.3.6 (b)
present the BEI of the ternary eutectic regions in SAC305 + lso-Butyl trisilanol

POSS. Specimens used for microstructural evaluation have been cooled at

53°C Imin. In SAC305, large (~30pM) rod and needle shaped Cuesn5 and

Ag3Sn lMCs are observed in the ternary eutectic region, while specimens

containing trisilanol POSS show a reﬁned ternary eutectic microstructure with

97

relatively spheroidal morphology. In general addition of trisilanol POSS results in
a more uniform microstructure. This suggests that such addition inﬂuences the
growth of lMCs within the liquid phase during solidiﬁcation. According to Kingery,
growth of secondary phases (lMCs) in a liquid phase involves two steps [67].
The ﬁrst step involves movement of atoms though the melt to the IMC I liquid
interface, while the next step required reaction between atoms at the interface
[67]. In other words, growth of IMCs in liquid melt depends on the diffusivity of
atoms though the liquid melt and IMC l liquid interfacial energy. Addition of
POSS may affect both the parameters. Empirically, Trivedi et.al. showed that
diffusivity (D) of the liquid melt is related to the SDAS (A) through the following
relationship [68, 69],
A = K (DIv) (1 ),

where, K , and v are constants and solidiﬁcation velocity respectively [68].
According to Reed-Hill, v is related to undercooling (AT) as follows,

v = 8 AT (2),
where B is a constant of the order of 10'6 (mIs°C) [66]. Substituting for v in

equation (1), and rearranging we have,
A AT = (KIB) D (3).
Equation (3) suggests that D is directly proportional to the product A AT. Table

5.3.1 displays values of the product, A AT, for all the specimens studied.

98

Table 5.3.1 A AT values for all specimens

 

 

 

 

 

Specimen A AT

SAC305 233.1

SAC305 + 2wt% lso-Butyl trisilanol POSS 17.7
SA0305 + 2wt% lso-Butyl Octa POSS 337.5

 

 

99

 

A933" 8:
Cu68n5

Sn dendrite

2M% ISO-BUM Trisilanol 2wt% Octa Iso-Butyl Trisilanol

I

 

 

 

      

Figure 5.3.1 Optical micrographs showing the effects of 2wt% trisilanol POSS

and Octa POSS additions on the as-solidiﬁed microstructures of SAC305. All the

specimens have been cooled at 53°C IseC.

(A) (A) 43 15-
o (J1 o 0‘
g I I I

N
01
I

A —L
o 0"
I I

Secondary dendrite arm spacing, A , (Urn)
M
0'1 O

 

   

SAC305 (control) 2wt% Iso-butyl Trisilanol 2wt% Octa Iso-butyl

Secondary dendrite arm spacing (um)

Figure 5.3.2 Plot showing the quantitative microstructural analysis of the

secondary dendrite arm spacing.

101

SAC305 + 2wt% lso-Butyl
Trisilanol

Sn-3.0Ag-0.5Cu (SAC305)

      
 

Figures 5.3.3 Optical micrographs of showing the effects of 2wt% Trisilanol

POSS additions to SACSOS specimens cooled at (a)—(b) 16°C/sec (slower

cooling rate), and (C)—(d) 53°C/sec (faster cooling rate).

102

   

 

Cross secjtiriral view

W
C
2
m ' I I
Extended Sn Dendrite Equraxed condensed Sn
Dendrite
(a) (b)

Figure 5.3.4 Schematic showing the dendritic morphologies of (a) SAC, and (b)

after trisilanol POSS additions.

103

 

 

 

 

 

2m 1 I I I I I I I I L
-O—Sn-3.0wt%Ag-0.5wt%Ag(SA(2305)
' I —I-SAC305+2wt%Iso-butlerisilaiolPOSS
250. % —A—SAC305+2wt%Octalso-buylPOSS
A .
.9 if)?
j; 240- \\
L-
I it
E 230- maﬁa,
CI \
a.
g 220
P _
210-
' (a)
0.0 0.2 0.4 0.6 0.8 1.0
TIme(m'ns)

Figure 5.3.5 (a) Plots showing measured cooling curves of SAC305, SAC305 +

2wt% lso-Butyl trisilanol POSS, and SAC305 + 2wt% lso—Butyl Octa POSS.

104

 

14

(b)

—L
N

A
O
I

21

Undercoollng temperature, AT, (° C)
a:

 

 

Sn-3.0wt%Ag-0.5wt%Cu SAC305 + 2wt% lsobutyl SAC305 + 2wt% Octa Iso-
(SACSOS) Trisilanol POSS butyl POSS

Figure 5.3.5 (a) Plots showing effects of trisilanol P088 and Octa POSS

additions on undercooling. All the specimens were cooled at 53°C /sec.

105

 

Addition of trisilanol POSS reduces SDAS (Figure 5.3.2), and undercooling
(Figure 5.3.5 (b)). Consequently, the listed values (table 5.3.1) show that A AT in
tn'silanol POSS added SAC305 specimen is smaller by at least one order of
magnitude than other specimens. In other words D decreases by at least one
order of magnitude upon trisilanol POSS additions. This suggests that trisilanol
POSS addition restricts diffusivity of atoms though the liquid melt.

Furlhennore, prior studies on the TMF behavior of Sn-based solder joints
have suggested that POSS may be present at the MCI solder interfaces.

Consequently it is possible that such nano-particles may segregate themselves

at Ag3Sn I Sn and Cussns I Sn interfaces during solidiﬁcation. To validate this

hypothesis SAC+2wt%lso-Butyl trisilanol POSS specimen was isothennally aged
at 100°C for 5 months in a parallel experiment. Optical micrographs showing the
effects of such aging conditions are presented in Figures 5.3.7. Evidently, there
is barely any noticeable change within the microstructure. lsotherrnal aging
causes changes in size and morphology of secondary phases [70]. Driving force
required for producing such changes is the necessity to decrease the interfacial

energy of the system [70]. Consequently, the observed microstructural stability

of trisilanol POSS added specimen strongly suggests their presence at AggSn /

Sn and Cuesn5 / Sn interfaces and thereby decreasing the interfacial energy.

106

 

(a) SAC305

, I

 

(b) 2wt% Iso-buty

'\

 

Figure 5.3.6 High magniﬁcation BEI showing sizes and morphology of Cu68n5

and Ag3Sn lMCs in the ternary eutectic regions in (a) SAC305, and (b) SAC305

+ 2wt% lso-Butyl trisilanol POSS.

107

 

Figure 5.3.7 (a) Initial as-solidiﬁed microstructure of SAC305 + 2wt% lso-Butyl

trisilanol bulk specimen, (b) microstructure of the same specimen after

isothennally aging at 100°C for 5 months.

108

In essence, these microstmctural studies reveal that trisilanol POSS

promotes heterogeneous nucleation of Sn and are present at the Sn / IMC

(Ag3Sn, and Cu6$n5) interfaces. These observations further tend to suggest the

presence of Si-O-metal bonds. However, presence of trisilanol POSS at the Sn
grain boundaries could not be detected though these microstructural studies.
6.3.3 Thermal analysis
Thus far inﬂuences of POSS with Si-OH groups on the microstructure of SAC305
have been discussed. However, such inﬂuences on the microstructure occur due
to Si-O-metal bonds, which are most likely to form in the molten liquid prior to
solidiﬁcation. Results investigating the interaction of Si-OH groups with SAC305
during melting are presented in this section. These studies are carried out with a
DSC and employ two methods.

In the ﬁrst method, effects of tn'silanol POSS additions on the heat of

fusion of bulk specimens are investigated. Two heating (melting) / cooling

(solidiﬁcation) cycles were performed at 1000/min and the second (last) heating

cycle was considered for analysis. Specimens for this study were SAC305, and
SAC305 + 2wt% lso-Butyl trisilanol POSS. Typical DSC thermogram of the
second (last) heating cycle are shown in Figure 5.3.8. Analysis of the
thennograms was carried out by integrating the area within the peak. This area
is the measure of heat of fusion (AH) required for melting. Table 5.3.2 lists AH

for the two specimens. It is found that addition of trisilanol POSS increases the

109

AH, i.e. more energy is required for melting a give mass of SA0305 containing
trisilanol POSS.

Several studies have indicated that melting initiates at the grain
boundaries [71-73]. Hseih et.al. conclusively showed through in-situ
transmission electron microscope studies of Al that melting initiates at the grain
boundaries and triple junctions [73]. High AH of trisilanol POSS containing
specimen suggests that melting does not occur easily as compared to control
SACBOS specimens. This observation suggests that trisilanol POSS particles
may be present at the Sn grain boundaries though the formation of Si-O-Sn
bonds.

Table 5.3.2 Effects of tn'silanol POSS additions on the heat of fusion of SAC305

 

 

 

 

Composition AH(JIg)
SACSOS 62.7:l:O.16
SAC305 + 2wt% lso-Butyl trisilanol POSS 66.310.17

 

 

To further conﬁrm the presence of trisilanol POSS at the Sn grain
boundaries, Kissinger’s method of thermal analysis is utilized [74]. The main
objective of Kissinger's analysis is to determine the activation energy of a ﬁrst
order transformation (such as melting). Activation energy is determined by
measuring the variation of peak temperature (Tp) in DSC thennograms with
heating rates (R) while maintaining the same cooling rate. Further, this analysis

assumes that Tp corresponds to the maximum rate of transformation. Solid to

110

 

liquid transition in metals, during melting, is a ﬁrst order transformation [74].
Such a transition results in peak I minima in a DSC thermogram (Figure 5.3.8).
Consequently methodology described Kissinger can be utilized for studying the
effect of trisilanol POSS on the melting behavior of SAC305.

The second set of thermal analysis is carried out by observing the peak

melting temperature (minima of the peak shown in Figure 5.3.8) positions as a

function of four different heating rates i.e. 1000/min, 20°C/min, 30°C/min, and

40°C/min. A typical DSC thermogram showing the variation of peak melting

temperature with heating rate has been presented in Figure 5.3.9. The plot of

peak melting temperature (Tp) v.s. heating rate (R) is shown in Figures 5.3.10 for

both SAC305 + 2wt% lso-butyl trisilanol POSS and SAC305. These plots
indicate that trisilanol POSS addition decreases the slope. Slope of the curves in
Figure 5.3.10 may be related to the grain size. For a given weight, a specimen
with smaller grains will have more number of Sn grain boundaries than the
specimen with larger grains. Consequently there are more sites for melting to

initiate and the melting process would be able to accommodate the heat supplied

more efﬁciently. Tp of trisilanol POSS added SAC305 specimens show a lesser

sensitivity to increase in R or smaller slope as compared to control SAC305.
This suggests the presence of smaller Sn grains in trisilanol POSS containing
specimen. BEI supporting the presence of smaller Sn grains have been in

SAC305 + 2wt% lso-Butyl trisilanol POSS have been shown in Figures 5.3.11.

111

lnsensitivity of Tp vs. R slope for SACSO5 + 2wt% lso-Butyl trisilanol POSS

further indicates that Si-OH in trisilanol POSS promotes heterogeneous
nucleation of Sn. Further, activation energy (0) of the melting process is

determined from the equation [74]:

ln(R/Tp2)=-(Ql8.314Tp)+C (1)

where C is a constant. Figure 5.3.10 (b) shows the plot of In ( R / Tp 2 ) vs .1pr

for the all the specimens. Linear least square ﬁt have been overlaid on the plot in

Figure 5.3.10 (b). Table 5.3.3 shows the R2 for the linear ﬁt and activation

energy for the melting process. Evidently specimens with trisilanol POSS have a
higher Q than the control SAC305. These values further suggest that trisilanol
POSS additions retard the initiation of melting at the Sn grain boundaries by

being present at those boundaries.

2 .
Table 5.3.3 R from linear least square ﬁt and activation energy of melting

resulting from Kissinger’s analysis (Figure 5.3.10 (b))

 

2
Composition R Q (KJlmol)

 

SAC305 0.971 628.7

 

 

SAC305 + 2wt% lso-Butyl trisilanol POSS 0.983 1051.7

 

 

 

112

 

5.4. Summary

Addition of POSS with surface active Si-OH group causes microstructural
reﬁnement in Sn-3.0wt%-0.5wt%Ag. Factors contributing to such an effect are:
1. Promotion of heterogeneous nucleation of Sn by surface active Si-OH
groups attached to trisilanol POSS cage, and thereby resulting in smaller
Sn grains.
2. Presence of trisilanol POSS at Sn—Sn grain boundaries and IMC/Sn

interfaces.

113

 

 

 

 

 

 

 

 

 

o l L l l l
13 f P
3
3 -1 - -
2
III-
“
N
w
:r:
.2- -
zoo ' 21'0 ' 250 ' 250 ' 250 ' 250
Temperature (°C)

Figure 5.3.8 Typical DSC thermogram observed during melting. Area within the
peak (minima) provides the heat of fusion. The negative y-axis represents

endothermic heat input.

114

 

Heat Flow (J lg)

 

 

 

 

' I ' l ' | - l I

200 210 220 230 240 250
Temperature (°C)

Figure 5.3.9 Typical DSC thennograms showing the increase in peak

temperature Tp with increase in heating rate during melting. The negative y-axis

represents endothermic heat input.

115

 

500 ,., . , , .

 

 

 

- - A- - Sn-3.0wt% Ag-0.5wt% Cu (SAC305
—I— SAC305 + 2wt% lso-Butyl Trisilanol POSS
. A
A '
x
#31495“ ,AZ] 0.15 min '
A. . ' ' ' O l/-
I/A 0.09 min
I/
(a) -
490 I ' I ' l ' l
10 20 30 40

Heating Rate, R (°K I min)

Figure 5.3.10 (a) Plots showing the effects of 2wt% trisilanol POSS additions on
peak temperature vs. heating rate. Slopes in (a) were obtained by ﬁtting a

straight line.

116

 

-8.5-

-9.0 -

N

“- -9.5-
t

m -10.0-

v

c .
-' -10.5-

A Sn-3.0wt% Ag-0.5wt% Cu (SAC305) '
I SAC305 + 2wt% lso-Butyl Trisilanol POSS

 

k.

 

 

-11.0

0.00201

 

0.00202 ' 0.00203 ' 0.00204

1ITP

Figure 5.3.10 (b) Plots showing the effects of 2wt% trisilanol POSS additions on

Arrhenius plot for melting. .

117

 

Figure 5.3.11 BEI comparing microstructures of (a) SACSO5 and (b) SAC305 +

2wt% iso-Buty trisilanol POSS. POSS containing specimen show considerably
more Sn grain boundaries than SAC305. Furthermore, size of Sn grains in (b) is

less than 20pm.

118

Chapter 6

Inﬂuence of polyhedral oligomeric silsesquioxane

additives on the coarsening kinetics of Ag3Sn
intermetallic compound in Sn-Ag-Cu solderjoints

Rationale and objectives

Prior studies have shown that nanostructured trisilanol polyhedral oligomeric
silsesquioxanes (POSS) with surface active Si-OH groups are present at the
intermetallic compound (IMC) / Sn interfaces. Presence of trisilanol POSS at
those interfaces resulted in minimal microstructural changes during isothermal
aging. This suggests that trisilanol POSS nanostructures may enhance
microstructural stability of solder joints by retarding coarsening of IMC particles in
solder joints. As a result isothermal aging studies on solder joints have been

carried out to investigate the inﬂuence of POSS additives on the coarsening

kinetics of Ag3Sn IMC phase. Although Sn-Ag-Cu contains both Ag3Sn and
Cu6$n5 IMC phases, Ag3Sn is chosen for this study because it coarsens
considerably slower than Cuasns. This means monitoring gradual change in
AggSn size can be easily carried out. Consequently, studying evolution of

Ag3Sn during isothermal aging provides a convenient method for investigating

the effect of POSS with Si-OH on microstructural coarsening. '

119

 

6.1. Background

Microstructural stability of intermetallic compounds (IMC) in Sn-based solder
joints is an important service requirement. The main purpose of lMCs is to
strengthen the solder matrix [32, 33]. Consequently service conditions leading to
their coarsening will deteriorate their mechanical reliability. In automobile under-

the—hood and power cycling applications service temperatures are in the

neighborhood of 150°C [1]. As a result, isothermal aging like conditions will exist

for solder joints encountering high temperatures for long durations in service.

Under such conditions lMCs will tend to coarsen. This has been shown to be

true for Cu68n5 and Ag3Sn lMCs in Sn-Ag-Cu solder joints experiencing thermal

excursions [21, 34-36]. So to improve the service life of joints coarsening of
lMCs must be retarded.

Results presented in chapter 5 showed that nanostructured polyhedral
oligomeric silsesquioxanes (POSS) with surface active silanol (Si-OH) groups are
present at the IMC / Sn interfaces. The presence of POSS at those interfaces
was further shown to retard microstructural changes during isothermal aging.
This means POSS with Si-OH (e.g. trisilanol POSS) have the capacity to retard
coarsening of IMCs. Since coarsening involves movement of atomic species
though different diffusion pathways (dislocation pipe diffusion, grain boundary
diffusion etc.) POSS additions may inﬂuence the diffusion mechanism
predominant during coarsening. In other words addition of POSS with Si-OH

may affect the growth kinetics of IMC particles.

120

 

This chapter investigates such an inﬂuence of POSS. For this purpose

coarsening behavior of Ag3Sn is studied at different temperatures. Selection of
A93$n is motivated by the fact that during isothermal aging Ag3Sn grows
considerable slower that CusSns, consequently temporal changes in Ag3Sn

sizes would be easier to track [37, 38, 75]. Coarsening of Cu6$n5 and Agasn in

Sn based solders occurs by diffusion of Cu and Ag through Sn respectively [74].
Dyson showed that Cu diffuses signiﬁcantly faster that Ag along “a” and “c” axes
of Sn[37, 38]. This has been demonstrated through the listing of activation

energy values in table 6.1 for both Cu and Ag. Furthermore, aging temperatures

chosen for this study are between 80-180°C. This range considers the

maximum temperature extremes that may be experienced by Sn-based solder
joints during service.
Table 6.1 list of activation energies of diffusion of Ag and Cu along “a” and “c”

axes of Sn

 

Diffusing species Axes of Sn Activation energy (KJ/mol) Reference

 

 

 

 

a 77

A9 [38]
c 55
a 33

Cu [37]

 

 

 

 

 

 

121

rune;

6.2. Experimental details

Materials used in this study were commercially available Sn-3.0wt%Ag-
0.5wt%Cu (SA0305) solder paste, and Iso-Butyl POSS trisilanol POSS nano-
structured chemical obtained from Hybrid Plastics®. 2wt% (10 vol%) of lso-Butyl
trisilanol POSS was then mechanically mixed with SA0305 solder paste to
prepare additional pastes of composition SA0305 +2wt% lso-Butyl Trisilanol
(SACZiB). Further, the two solder pastes, i.e. SAC305, and SACZiB were used
for preparing single shear lap solder joints (chapter 2). Details of specimen
preparation have been provided in chapter 2. For microstructural observation,
one side of the joint was metallurgically polished. First step of polishing involved,
using SiC papers with grit sizes ranging from 800 to 2400. The ﬁnal step
consisted of polishing with 0.3um alumina solution and 0.05pm colloidal silica
solution. Isothermal aging was carried out in air circulated ovens which were
maintained at 80°C, 90°C, 107'C, 120°C, 150°C, and 1800. Temperature
variation of these ovens was within 13°C. To observe the temporal evolution of
microstructure, solder joints were taken out of their respective ovens after 0, 7,
14, 28, 42, and 56 days. Microstructural observation was carried out using JEOL
6400 scanning electron microscope (SEM) which has a resolution of 50nm. Prior
to each observation, the polished side of every solder joint was re-polished
following the ﬁnal step of polishing outlined earlier. This procedure provided
clarity to the SEM images required for determining sizes and distribution of
Aggsn particles. For evaluating the average A9380 particle diameter, images

were taken randomly along the solder region of the joint, and over 150

122

 

measurements were made. A938" particles are assumed to be spherical for

measurement purposes. Software used for measuring particle size and

statistical analysis were ImageJ© and Origin® respectively.

6.3 Results and discussion

6.3.1 Coarsening theories

Secondary phases in alloys grow in their dimensions during isothermal aging,
during which larger particles grow at the expense of smaller particles. The
driving force behind this process is the temporal reduction in interfacial area of
the system, as it proceeds towards thermodynamic equilibrium [70]. Such a
phenomenon is commonly known as coarsening or Ostwald ripening. As a
consequence of reduction in interfacial area, mass transfer occurs from region of

higher chemical potential to that of lower chemical potential. Mathematically, the

ﬂux of the species is given by Fick’s 1$t law in 3D,

3 = —M-V’ju (1)
where, J is the ﬂux of the diffusing species, )1 is chemical potential, and M is the
phenomenological constant related to the diffusion coefﬁcient. Further, chemical
potential, )1, is given by ,

p = ZOyH (2)
where, I) is molar atomic volume, V interfacial energy between the secondary
phase and the surrounding matrix and H being the curvature of a particle.

Evidently, diffusional ﬂux of atoms occurs from regions of higher curvature

(smaller particles with large interfacial area) to that of lower curvatures (larger

123

particles with smaller interfacial area). Consequently, the microstructure
evolution results due to the growth of secondary phase particles with time.

In general growth kinetics of such particles is given by the equation,

<D>" - <Do>n = K t (3)
where, K = K1 exp (—Q/R'l7, <D> is the average diameter of the particle at a

given time, t, <Do> initial average sizes , R is the gas constant , T is the

temperature, Q is the activation energy of diffusion, and n is a phase size
constant which is related to rate controlling mechanism responsible for
coarsening of secondary phases.

Traditionally, theories of coarsening are based on lattice diffusion as the

rate controlling mechanism, where the phase size constant n equals 3 [70, 76-78].

Following Senkov et.al, time variation of particle size is given by the equation,

<D>3-<Do>3=K2[(yQCoDL)/R17 t
(4) where, K2 is a constant, Co is the equilibrium solute concentration within the

lattice, and DL is coefﬁcient of diffusion of solutes through the matrix. These

theories assume that the particles are spherical and have small volume fractions
within the matrix. In spite of these assumptions, several alloy systems have

been known to follow equation 4 [70, 75, 79]. Typically lattice diffusion occurs at

higher homologous temperatures (TIT m) since the activation energy required for

diffusion is higher. However, at lower temperatures diffusion of atoms occur

124

through different mechanisms which are also known as high diffusivity paths [80].
These paths are through grains boundaries, dislocation cores (pipe diffusion),
and triple junction boundaries [77, 78, 80].

Several theoretical and experimental studies have been carried out to
determine the growth kinetics of particles dominated by grain boundary diffusion.
Phase size constant (n) for grain boundary diffusion equals 4. The general

formulation for time variation of average particle size is as follows,
<D>4-<Do>4=K3[(waConggb)/R77f (5)

gb

where, K3 is a constant, w is the grain boundary width, Co is the equilibrium

grain boundary concentration, and ng coefﬁcient of diffusion along grain

boundaries [78]. Equation 5 provides the temporal behavior of particle size due
to diffusion along high angle grain boundaries [77]. Small angle grain boundaries
have also been considered by Ardell as a possible path for diffusion [77]. In that
theoretical treatment, low angle grain boundaries were composed of dislocations
networks and secondary phase particles were embedded within such networks
[77]. It was found that phases size exponent (n), under such conditions, equaled
5. Similar value of n was obtained by Senkov et.al. , however, in that study
diffusion along grain boundary triple junctions were considered. Nonetheless,

the general mathematical form is given as,

<D>5'<DO>5=K4[(QYCOdDd)/3RT] (6)

where, K4 is a constant, q is the cross-sectional area of an individual dislocation,

125

 

d
Co is the equilibrium concentration of solutes within dislocation cores or at triple

junction, Dd is the coefﬁcient of diffusion along a dislocation, cores or triple

junctions and ‘a’ the average diameter where <D> >> a.
Coarsening is also known to be controlled by the transfer of solutes across
the particle-matrix interface [78]. Here the temporal behavior of particle growth is

given by,
<D>2-<Do>2=K5[(YOCo)/R77 (7).

Symbol used in equation (7) have already been deﬁned in equation (4). It may
also be noted that coefﬁcient of diffusion is absent in Eq. (7) suggesting that
growth of particles are not diffusion controlled. Finally, Table 6.2 summarizes the

different mechanisms known to control coarsening of secondary phase particles.

Table 6.2 Summary of mechanisms controlling coarsening of secondary phase

particles and the respective phase size exponents

 

 

 

 

 

 

Phase size exponent,
Rate controlling mechanism

n

Transfer of solutes across the particle / matrix
2 interface
3 Lattice diffusion
4 Diffusion along grain boundaries
5 Triple junction diffusion , dislocation diffusion

 

 

126

 

 

6.3.2 Inﬂuence of POSS addition on Ag3Sn particle size evolution

Typical BEI of the initial microstmcture (prior to isothermal aging) in SA0305 and
SACZiB solder joints are shown in Figures 6.3.1(a) and 6.3.1(b) respectively.

These micrographs show different phases present within the initial microstructure.

The brighter phase is that of Ag3Sn. High magniﬁcation SEM images presented

in Figures 632(0) and 6.3.2(d) shows the initial size and morphology of Ag3Sn
IMC particles in SAC305 and SACZiB respectively. Figure 6.3.2 shows the

distribution of initial average diameters (<Do>) of Ag3Sn particle for both types of

solder joints. t-test shows that there is no statistical signiﬁcant difference (two-tail

p values equals 0.54) between the AggSn particle diameters in the SAC305, and

SACZiB solder joints

BEI in Figures 6.3.3 shows the inﬂuence of isothermal aging at 107°C,

150°C, and 180°C for 42 days (1008 hrs) on the microstructure of SAC305 and
SACZiB. In both types of specimens, Ag3Sn particle sizes increase with
temperature. However, Ag3Sn particles appears to be smaller in SAC2iB at
150°C (Figures 6.3.3(b), and 5.3.3(e)), and 180°C (Figures 6.3.3(c), and

5.3.2(f)). This suggests that addition of POSS retards the growth of Ag3Sn

particles during isothermal aging. However, such visual distinction could not be

127

 

made for lower temperature i.e. 107°C (Figures 6.3.3(a), and 5.3.3(d)).

Consequently, micrographs for lower temperatures have not been shown.

Instead plots showing the variation of Ag3Sn average particle diameter (<D>)

with time (t) in SAC305, and SACZiB solder joints at different temperatures have
been presented in Figures 6.3.4(a) - 5.3.4(f). Error bars represent 95%

conﬁdence intervals. It is evident from plots in Figures 6.3.4 that addition of

trisilanol POSS retards the coarsening of Ag3Sn particles. Studies on

isothennally aged Sn-based solders have shown that the growth of Ag3Sn occurs

through diffusion of only Ag atoms [75, 81]. Results presented in this section
further suggest that addition of trisilanol POSS inﬂuences the movement of Ag

atoms within the solder matrix.

128

 

”—‘m

Sn grain
boundary ‘

(b) SAC305 + 2wt%lso-Butyl
trisilanol

 

 

    

(c) SAcaos (d) SAC305 + 2wt%lso-Butyl
trisilanol

 

 

Figure 6.3.1 Backscattered electron images (BEI) showing the initial
microstructures of (a) Sn-3.0wt%Ag-0.5%Cu (SAC305), and (b) SAC305 + 2wt%
lso-Butyl trisilanol POSS (SACZiB). Different microconstituents within the
microstructure have been indicated. High magniﬁcation SEM images showing

the size and morphology of Agasn IMC particles in (c) SAC305, and (d) SACZlb.

129

 

 

 

l
0.351 g
73‘ o
a I
'5
g 0.30-
a
0
a
I! 0.251
9 o
(I
E
‘5
- 0.20-
o
0.15

 

—A— SAC305 .
—o— SAC305 + 2wt% lso-Butyl trisilanol

 

 

Figure 6.3.2 Distribution of Ag3Sn particle sizes in SAC305 and SACZiB

specimens. Two-tailed p value equals 0.54, suggesting that there is no statistical

difference between the Ag3Sn particle diameters in SAC305 and SACZiB

specimens. This analysis was carried out using six specimens for each type of

solder joint.

130

 

SAC305 + 2wt%
SAC305 Iso-Butyl trisilanol

-m

 

Figure 6.3.3 BEI showing the inﬂuence of isothermal aging at 107°C, 150°C,

and 180°C on (a)—(c) SACSOS and (d)-(f) SACZiB specimens.

131

 

 

 

 

r j r 1 r ﬁ r f r ' I
0.60 " 80° . q
0.55 - 5
E 0.50 - (a) _
3 1 .
A. 0.45. "
9 1 .
.. 0.40- -
b
O . .1
3 0 35
E e
.92 * «
'3, 0.30 1 _
E 025 . ..
g .
4 0 20 I SAC305, n = 4
‘ e SACZiB, n = 5 i:
0.15 - -
r ' r ' r r r r r ' r
0 200 400 600 000 1000
Time (Hrs)

Figure 6.3.4 (a) <D> v.s. t plots comparing Ag3Sn particle sizes in SAC305 and

SACZiB specimens isothennally aged at 80°C.

132

 

 

 

 

0'7 I ' I ' I ' I I I
--

05.. -
g (b)
A
D
V_ 0.5- -
5
0
E
g 0.4- -
O
a)
0 i
@
> 0.3~ -
“ - SAC305, n =4

' e SAcziB,n=5 3‘
0.2 1 , . , . , . r . ,
O 200 400 600 800 1000
Tine(Hrs)

Figure 6.3.4 (b) <D> v.s. t plots comparing Ag3Sn particle sizes in SAC305 and

SACZiB specimens isothennally aged at 90°C.

133

 

 

 

 

 

I ' I I I I I l I
1.0-. -
09. -
A J
E
:5 004 (c) —
A.
9 07- -
3" .
'5 0.6~ .
E . l
.2
v 0.5- -
8: .
N
g 04- _
< ' I SAC305, n=4
“3' o SAC2iB,n=5 3'
0.2 ,

' l ' I r l ' I ' I ' l j I
0 200 400 600 800 1 000 1200 1400
Time (Hrs)

Figure 6.3.4 (c) <D> v.s. t plots comparing AggSn particle sizes in SAC305 and

SACZiB specimens isothennally aged at 107°C.

134

 

a
b
I 1
a ‘
1

A
N
1

1

d
b
l
A
D.
v

A
L

a
b
1

1

O
1»
1
1

04- -

Average diameter, <D>, (pm)

I SAC305, n = 3
02‘ I SAC2iB, n = 5 .

 

 

 

l ' l ‘ I ' l ' I ' I ' I ' l
0 200 400 000 000 1000 1200 1400
Trrre (H‘s)

Figure 6.3.4 (d) <D> v.s. t plots comparing Ag3Sn particle sizes in SAC305 and

SAC2iB specimens isothennally aged at 120°C.

135

 

4.0- -' I I I T I I - r r -
35: -
m (e) _
25: .
20: ..

.3
0|
1
FI-I
1

Average diameter, <D>, (um)

 

 

 

1.0- _
0.5; I SAC305, n=3 _
o SACZiB,n=2
0'0 ' r r r ' r ' r ' r ' r ' r ' r
.200 0 200 400 000 300 1000 1200 1400
,. TirrefI-h‘s)

Figure 6.3.4 (e) <D> v.s. t plots comparing Agasn particle sizes in SA0305 and

SAC2iB specimens isothennally aged at 150°C.

136

 

0|
1

Average diameter, <D>, (gm)
1‘ 5° ‘1’

O
1

.5

I I
A
:5

l SAC305, n = 3
o SAC2iB, n = 2

 

 

I

0

Figure 6.3.4 (f) <D> v.s. t plots comparing Ag3Sn particle sizes in SAC305 and

' I ' I ' l ' I
200 4)O 600 8th
Time (Hrs)

j

I ' 1 I
1000 1200 1400

SAC2iB specimens isothermally aged at 180°C.

137

 

 

6.3.3 Growth kinetics

As outlined in section 6.1, coarsening of secondary phases occur through
diffusion of solutes through different mechanisms. Analysis of growth kinetics of
such secondary phases reveals related mechanisms. Consideration of such

mechanisms may further explain why trisilanol POSS additions retard coarsening

of Ag3Sn.

Growth kinetics are evaluated by ﬁnding out phase size constant “n" and
temperature dependent constant “K” (equation (3)). To determine “n”, non-linear
regression was used, where different values of “n” (presented in Table 6.2) were
substituted in equation (3) and the best-ﬁt curve for a given “n” was obtained.
Such curves have been plotted overiaying the data points in Figures 6.3.4 (e)-

6.3.4(f). For all the ﬁts, R2 is greater than 0.9 (Table 6.3). After determining “n”,

<D>n - <Do>n v.s. t is plotted (Figures 6.3.5) to obtain “K”. Table 6.4

. ,, 2 . . . .
summarizes “K values and R for linear regressron analyses. Further, activation

energies of diffusion were obtained by' plotting K v.s. 1/T on a logarithmic scale
(Figures 6.3.6). Values of activation energies of diffusion for SAC305, and
SACZiB solder joints have been tabulated in Table 6.5. This table also integrates
the results obtained from the analysis of growth kinetics by including rate
controlling mechanisms at different temperatures of aging. Table 6.4 clearly

shows that trisilanol POSS additions to SAC305 indeed change the rate

controlling mechanisms contributing to coarsening of Ag3Sn. Furthermore, It is

138

 

also apparent from Table 6.5 that, regardless of trisilanol POSS additions, there
is dichotomy in rate controlling mechanisms depending on temperature regimes
of isothermal aging.

At lower temperatures, grain boundary diffusion (n=4) dominates the

growth of Ag3Sn in SAC305 between 8000-10700. However, in SACZiB a very

different rate controlling mechanism is operative in a similar temperature regime,
since n equals 5 in this case. Such a value of n suggests that diffusion of Ag
atoms occurs either through the grain boundary triple junctions or the dislocation
cores. Furthermore, the change in diffusion mechanism (n=4 to n=5) may be
explained by considering the effects of trisilanol POSS (with surface active Si-
OH) on SACSOS microstructure. In chapter 5 it was shown that trisilanol POSS
particulates are present at Sn-Sn grain boundaries. This means that movement
of Ag atoms may be impeded by the presence of trisilanol POSS at the grain
boundaries. Additionally, such an impediment may also exist even at the triple
junction boundaries, since trisilanol POSS prefers to be present at the interfacial
regions (chapter 5). Consequently, if interfacial diffusion is retarded by trisilanol
POSS, then Ag atoms can diffuse only through the dislocation cores. Activation
energy of such a process was experimentally determined by Herziq et.al. to be
76.02 KJ/mol [82]. This value is very close to 80.3 KJ/mol, determined in this
study for SACZiB. Additionally, activation energy of SAC2iB, at lower

temperatures, is higher than that of SAC305 (60.3 KJ/mol), further indicating the

retarding effect of trisilanol POSS addition on Ag3Sn coarsening.

139

A similar comparison of activation energies could not be made for higher
temperatures due to lack of data on SACZiB (Table 6.5). However, trisilanol

POSS addition results in the change of rate controlling mechanism also at higher

temperatures (Table 6.5). Ag3Sn coarsening in 80305 occurs by lattice diffusion
(n=3) of Ag atoms through Sn, while growth of Ag3Sn in SACZIB is controlled by

the transfer of Ag atoms across the Ag3Sn / Sn interface (n=2). It may be noted

that there are two mechanisms operating in series during coarsening. At higher

temperatures Ag atoms have to ﬁrst diffuse though the lattice before crossing

over the Ag3Sn / Sn interface. So the slowest step in the series will control the

growth rate. It has been shown that trisilanol POSS is present at the IMC / Sn

interfaces. Consequently, trisilanol POSS will hinder transfer of Ag across

AggSn I Sn interface making it the slowest step, and thereby retarding

coarsening of Ag3Sn even at temperatures as high as 180°C.

6.4 Conclusions

Addition of trisilanol POSS improves the microstructural stability of Sn-3.0wt%Ag-

0.5%Cu solderjoints. Such improvements are brought about by:

1. Retarding coarsening of Ag3Sn particles within the solder matrix

140

2. The presence of trisilanol POSS at different interfaces (Ag3Sn I Sn, and

Sn / Sn grain boundaries). Their presence at those interfaces results in

altering the rate controlling mechanisms responsible for Ag3Sn coarsening.

141

Table 6.3 Statistical analysis for determining phase size constant “n” in <D>n -

<Do>n = Kt through non-linear regression.

 

 

 

 

 

 

 

 

 

 

SAC305 SAC2iB
Temperature ('C) 2 2
n R n R
80 4 0.91634 5 0.9912
90 4 0.9578 5 0.9143
107 4 0.9974 5 0.9668
120 3 0.9869 5 0.998
150 3 0.9611 2 0.982
180 3 0.9979 2 0.9878

 

 

 

 

 

 

142

 

 

 

 

I ' I ' I ‘ I f I ' I ' I T I
0.54 SAC305 1
A
0.4- ‘ .
(a) - ,
1A.: 0-3‘ o r
O ' i
v. 1
'A 0.2- ~ “
9
. . .
0.1- -
. ' I 00°C
/ 0 mac
“0‘ A 107°C ‘
j . r - r r r - r ' r ' I ' r
0 Z!) 400 000 8m 11!!) 12!) 14!)

Tumors)

Figure 6.3.5 (a) Plot of <D>4 - <Do>4 v.s t for determining the “K” for SAC305

specimens isothennally aged at 80°C, 90°C, and 107°C.

143

 

 

 

 

 

 

13‘ 's—A-Q'ﬂl I ' l ‘ r r r r A
m" a
"11° '5‘ - 1
1 l
A 50- -
V
z I 12“:
0 151°C '
. ‘ 1870 ‘
0‘ r A ‘- l P1
0 Z!) 4!) ED 8!) 1GB 12D 14D
TmW15)

Figure 6.3.5 (b) Plot of <D>3 - <Do>3 v.s t for determining the “K” for SAC305

specimens isothermally aged at 120°C, 150°C, and 180°C.

144

0.3-

0.2-

<D>~"’.<DO>5

0.1 -

 

 

0.0-

I ' I

SAC305+ 2wt% Iso-Bgyl Trisilanol

I 80°C
0 90°C
‘ 107°C
'

1 20°C

(C)

' I ' l ' I ' I ' I f l

 

 

Figure 6.3.5 (c) Plot of <D>5 - <Do>5 v.s t. for determining the “K” for SACZiB

specimens isothennally aged at 80°C, 90°C, 107°C, and 120°C.

I
0 2D

 

I r'rrr
4m 6m m1mm1m

145

 

Table 6.4 Linear regression analysis to determine "K” in <D>n - <Do>n = Kt for

the data presented in Figures 7

 

 

 

 

 

 

 

 

 

SAC305 SAcziB
Temperature ('C) 2 2
K R K R
80 90015-05 094‘” 1.001305 0'98“
90 2.00604 “9149 3.001305 09349
107 4001304 0.9795 6.00305 0.961
120 1. 4013-03 0.989 2.00304 0.9933
150 4.82E-02 0'983 8.50E—03 099”
180 1.201%] 0.9965 1.55302 0.9898

 

 

 

 

 

146

 

 

   

 

 

 

l ' l
-2- I 80°C, 90°C, 107°C
0 120°C, 150°C, :
.3- 180 C _
J O = 111.1 KJlmol
-4- 2 -
J R = 0.962 4
n = 3
.5- J
S? .
v .5- _
c
A '1 (a) -1
.7- _
' Q = 60.3 KJImol -
-8- K -
- R2 = 0.9879 I -
_9_ n = 4 _
SAC305
'10 I r I F I r
2.25 25) 27.5 31!
1/Tx1m0 (K')

Figure 6.3.6 (a) Arrhenius plots for calculating activation energies for SACSOS,

147

A _

 

  
 

 

 

 

SAC305+ 2wt% lso-Butyl Trisilanol
I 80°C, 90°C, 107°C,
120°C
.9_ _
g -10- -
C
-' Q = 80.3 KJlmol
R2 = 0.9692
n=5
-11 - ..
'12 I r I l
25) 275 3.0

1ITx1lm (K)

Figure 6.3.6 (b) Arrhenius plots for calculating activation energies for SACZiB.

148

Table 6.5 Inﬂuence of POSS trisilanols on the activation energies of diffusion

 

 

 

 

 

 

 

 

Ag3Sn I solder interface

 

Calculated
Solder Temperature n, rate controlling[75]
activation
joint type regime ('C) mechanism
energy (KJImoI)
80 — 107 4, Grain boundary diffusion 60.3
SAC305
120 — 180 3, Lattice diffusion 111.1
2, Diffusion through
80 — 120 dislocation cores or through 80.27
triple junctions
SACZ'B 5, Transfer of solutes across
150 _ 180 particle / matrix interface i.e. _

 

149

 

Chapter 7

Summary and recommendations

7.1 Summary

Comparison between solder joints subjected to low temperature regime (-55°C
and 125°C) and high temperature regime (-15°C and 150°C) thermomechanical
fatigue (TMF) indicates that the former is signiﬁcantly detrimental to mechanical
reliability of Pb-free Sn-based solder joints (chapter 3). Microstructural damages
features suggested that, at lower temperature regime, limited ability Sn matrix to
dissipate anisotropic strains due to TMF through plastic deformation.
Consequently, during lower temperature regime TMF, cracks formed within
solder during the early stages of TMF continue to propagate with progression of
thermal excursions. In comparison, at higher temperature regimes the solder
matrix is considerable compliant so as to undergo plastic deformation upon
experiencing TMF. As a result cracks are not able to propagate during thermal
excursions.

Another important ﬁnding resulted from the comparison between low and
high temperature regime TMF of Pb-free Sn-based solder joints is that the
interfacial damages play a signiﬁcant role in both the cases. lnterfacial IMC / Sn
decohesion occurred during low temperature regime TMF whereas Sn / Sn grain
boundary damages resulted at higher temperature regimes. These observations
indicated that interfaces (between Sn grain boundaries and inter-phase

interfaces) in Pb-free Sn-based solders are vulnerable during TMF.

150

 

TMF studies carried out on solder joints with iso-Butyl trisilanol POSS
show considerably improved mechanical reliability (chapter 4). Microstnrctural
observations suggested that iso-Butyl trisilanol POSS additions improved the
ductility of the solder matrix, due to which the matrix was able to plastically
deform and dissipate anisotropic stresses generated during TMF. As a result,
cracks formed at the initial stages of TMF are not able to propagate in the later
stages. Furthermore, plastic deformation of the solder matrix during TMF also
suggests that it also experiences strain hardening, which would contribute to
solder joint strength. Additionally, it was observed that the iso-Butyl trisilanol
POSS additions improved the IMC l Sn bonding. This suggested the presence of
trisilanol POSS at different interfaces eg. Sn I Sn grain boundaries and IMC / Sn.

Thermal analysis of bulk specimens (e.g. Kissinger’s analysis) indicated
the presence of iso-Butyl trisilanol POSS at the Sn / Sn grain boundaries
(chapter 5). While observation of coarsened microstructure (chapter 5) and
determination of coarsening kinetics of lMCs (chapter 6) conﬁrmed the presence
of such nanostructured Chemical at the IMC IS interfaces.

As a consequence presence of iso-Butyl trisilanol POSS at different
interfaces in conjunction with enhanced plasticity of the solder matrix improves

the TMF reliability of trisilanol POSS added Pb-free Sn-based solderjoints.
7.2 Recommendations

1) Thus far trisilanol POSS has eluded visual detection within the solder
matrix due to its small size (1.5nm). This issue may be resolved through

utilization of high resolution TEM observations. Such a method would also

151

 

 

2)

3)

4)

be helpful in locating them at the IMC I Sn interfaces and determine if
trisilanol POSS additions result in an orientation relationship between the
lMCs and Sn.

This study has shown that isobutyl trisilanol POSS are present at different
interfaces in solid state. However, the interaction of trisilanol POSS in the
liquid (molten) solder is yet to be ascertained. In-situ studies based on
synchrotron X-ray diffraction may be helpful towards that end.
Alternatively, molecular dynamics simulations may also provide clues.
This knowledge may prove useful in understanding the ﬁnal solidiﬁed
microstructure and possible provide means of engineering the ﬁnal
microstructure.

It must also be pointed out that the present study only leverages the
silanol (Si-OH) groups present in trisilanol POSS for bonding with the
metallic matrix (Si-O-metal). There exists a possibility of utilizing the
organic (e.g. isoButyl) groups attached to the SiO1,5 framework. lf,
through appropriate chemistry, the size of POSS (with silanol groups)
anatomy is increased, and organic groups are made considerably larger
(microns) and thermally stable, then the organic groups might also be able
to act as cushions for accommodating strains generated within the solder
during TMF.

Finally, pertaining to the TMF behavior of Pb-free Sn-based solders,
continuum mechanics based models may be used to predict their life time

reliability. At present there are many such models; however there are very

152

few that incorporate damage cause due to TMF. Continuum damage
mechanics (CDM) provides the means of doing so. CDM incorporates the
effects of submicron sized cracks and the plasticity of materials to
determine their mechanical response. Consequently, CDM framework
may be adopted for our purposes, although it has to be modiﬁed for the
inherent anisotropy of Sn since most continuum based models deals with

isotropy.

153

 

Appendix A

Inﬂuence of polyhedral oligomeric silsesquioxanes
trisilanol chemistry on the low temperature regime
thermomechanical fatigue behavior of nanocomposite
Sn-based solder joints

Rationale and objectives

The solder pastes used in this study are essentially a ﬁne dispersion of solder
alloy (of known composition) particulates in an organic ﬂux. However, the
composition of that ﬂux is difﬁcult to ascertain, since its chemistry is protected
through patent laws. Consequently, during processing it problematic to evaluate
the interaction between the ﬂux and the organic (“R”) group attached to the
polyhedral oligomeric silsesquioxanes POSS cage.

Preparation of nanocomposite solder joints involves mechanically mixing a
particular variety of POSS trisilanol nanostructured chemical with the commercial
available solder paste. Upon solidiﬁcation, depending on the interaction with the
ﬂux, the “R” group may inﬂuence the microstructure, and possibly the
thermomechanical fatigue (TMF) behavior of Sn—based solder joints. Thus far
only one variety of nanostructured chemical was utilized, namely lso-Butyl
trisilanol POSS. It was shown that such reinforcements improved the low
temperature regime TMF reliability of Sn-based solder joints. For POSS
trisilanols to be a viable reinforcement, it is necessary to obtain a consistently
improved TMF behavior irrespective of the type of “R” grou'p attached to the
POSS trisilanol reinforcements. Consequently, this Chapter investigates the

inﬂuence of phenyl trisilanol POSS reinforcements on the low temperature

154

regime TMF behavior of Sn-based solder joints. Phenyl trisilanol POSS was
Chosen, since it has a lower thermal stability than lso-Butyl trisilanol POSS [59,
60, 83]. Consequently, the results presented in this section will also depict
effects two possible extremes of POSS trisilanol reinforcement selection (based
on thermal stability) on the low temperature regime TMF behavior of Sn-based

solder joints.

155

A1 Experimental details

Experimental procedures followed in this study have been described in detail in
chapters 2, 3, and 4. This section provides a summary those procedures.

Eutectic Sn—3.5wt%Ag (SA) and Sn-3.0wt%Ag-0.5wt%Cu (SAC305)
solder pastes were obtained from the same vendor. Brand of the solders pastes
are similar to that of prior studies i.e. R520A. Nanostructured chemical phenyl
trisilanol POSS was procured from Hybrid Plastics®. Following the processing
procedure described section 2.1, phenyl trisilanol POSS was mechanically mixed
with SA and SAC305 to prepare solder pastes of compositions SA + 2wt%
phenyl trisilanol POSS (SA2phe) and SAC305 + 2wt% phenyl trisilanol POSS
(SACthe). The resulting pastes were then utilized for preparing nanocomposite
solder joints according to section 2.2.

To conduct TMF studies, SA2phe and SAC2phe nanocomposite solder

joint were subjected to thermal cycling between -55°C and 125°C in an

environmental chamber. Details of the temperature proﬁle utilized have been
provided in section 2.4. Mechanical and microstructural techniques necessary
for investigating the inﬂuence of TMF have been discussed in section 2.6.

A2 Results and discussion

This section establishes microstructure —mechanical property relationship of
phenyl trisilanol POSS reinforced nanocomposite solder joints subjected to TMF.

Anatomy of phenyl trisilanol POSS (pheTriPOSS) has been shown in ﬁgure A1.

156

In the schematic the organic ‘R" group consists of phenyl (-C5H5) group, while in

the previous studies (chapter 4, 5, and 6) the “R” group was lso-Butyl (-C4H9).

A2.1 Effect of phenyl trisilanol POSS on the residual mechanical behavior
of Sn-based solder joints

Figures A2 plots the residual shear strength as a function of number of TMF
cycles. Figure A2 (a) compares the residual mechanical behavior of SA and
SA2phe solder joints, while Figure A (b) compares the residual mechanical
behavior of SAC305 and SACthe solder joints.

These plots suggest that addition of 2wt% pheTriPOSS stabilized the
residual strength of nanocomposite (SA2phe, and SACthe) solder joints after
100 cycles. Further conﬁrmation was obtained through t-test (Tables Al), which
compares the residual strength data after 100 cycles with that of 400, 500, and
1000(SACthe). Single tail p-values for all such comparisons are less than 0.05,
indicating that the compared data sets are statistically different. Figures A2 also
indicates that residual strength of SA2phe and SACthe solder joints stabilized
at a higher value than the control specimens (SA and SAC205). Similar results
were obtained upon addition of 2wt% Iso-Butyl trisilanol POSS (ibutTriPOSS)
reinforcements to Sn-based solder joints (chapter 4).

However pheTriPOSS reinforced nanocomposite solder joints exhibited
differences in residual mechanical behavior, as compared to ibutTriPOSS
reinforced joints. lbutTriPOSS reinforcements did not change the initial strength

of both SA and SAC305 (Figures 4.3.1), whereas pheTriPOSS reinforcements

157

 

increased the initial strength of SA joints but decreased that of SAC305 joints.
Progressive TMF cycles caused a drop in the residual strength of SA2phe and
ibutTriPOSS reinforced joints, except for SACthe solder joints. In fact
successive thermal excursions do not seem to affect the residual strength of
SACthe joints. This observation is further corroborated by single tail p-values,
which were obtained after comparing residual strength data for 1 cycle with other
chosen number of TMF cycles (Table AI).

Based on this analysis one can infer that pheTriPOSS reinforcements
have different inﬂuences on the SA and SAC305 solder joints subjected to TMF.
Moreover, these results also indicate that pheTriPOSS reinforced Sn-based
solder joints do not exhibit consistent residual mechanical behavior as observed
in ibutTriPOSS reinforced solder joints. This further suggests that processing
with different organic “R” group (phenyl or lso-Butyl) may elicit different TMF
response from the reinforced solder joints.

Nonetheless, it is evident that irrespective of the “R' group, addition of
POSS with silanol (Si-OH) bonds improves the mechanical reliability of Sn—based
solder joints. To correlate the improved residual mechanical behavior (of
SA2phe and SAC2phe solder joints) with resulting microstructural damage due to
TMF, microstructural characterization was carried out. Results of that

characterization have been presented in the following section.

158

 

 

 

Figure A1 Schematic showing the anatomy of P088 trisilanol. Chemical

formula of the organic phenyl (R) group is also shown.

159

F-‘El

 

. 8:
.4
4

é
2r

3.?

/.
J
l.
.l
,_./_.

Residual shear strength (MPa)

 

 

 

J
.j“
1

Nurrber of TMF cycles

Figure A2 (a) Plots comparing inﬂuence of pheTriPOSS on the residual

mechanical behavior of SA.

160

 

 

Residual shear strength (MPa)

 

 

 

55 . T . . . - .

5.: 1m;
.5; ;
.,L _
35'. thy/l t '.
30- i -
25; \__§,§\-
20‘le ‘

o'zbo'4bo'ebo'abo'1o'oo
MnberofTNFcydes

Figure A2 (b )Plots comparing inﬂuence of pheTriPOSS on the residual

mechanical behavior of SAC305.

161

Table Al Comparison of single tail p-values obtained from t-test. Here residual

strength data after 1 and 100 cycles are compared with 100, 400, and

500 cycles for SA2phe and 100, 400, and 1000 cycles for SAC2phe. P-

values <0.05 would indicate that the compared data sets are

statistically different.

 

 

 

 

 

 

 

 

 

 

Material SA2phe SAC2phe
Number of
1 cycle 100 cycles 1 cycle 100 cycles
TMF cycles
100 cycles 0.5 0.19 0.5
400 cycles 0.06 0.352 0.41
< 0.001
500 cycles 0.102 - -
1000 cycles - 0.227 0.244

 

 

 

 

 

162

 

A2.2. Inﬂuence of phenyl trisilanol POSS reinforcements on microstructural
damages
Microstructural evolution of SA2phe and SAC2phe solder joints has been
depicted in Figures A3 and A4 respectively after 0, 100, and 500 cycles of TMF.
It is evident from the scanning electron microscope (SEM) images that SA2phe
accumulates significant more surface damage within the solder region as
compared to SAC2phe. High magniﬁcation SEM images further comparing the
intensity of surface damage within the solder region after 500 cycles of TMF have
been shown in Figure A5. It is observed that SACthe (Figure A5 (a))
experiences minimal surface damage as compared to SA2phe (Figure A5 (b)).
These SEM images imply that anisotropic strains (due to thermal
excursions) are accommodated through plastic deformation of the solder region
in SA2phe solder joint. However in case of SAC2phe, lack of plastic deformation

of the solder matrix suggests that anisotropic strains are accommodated

elsewhere within the joint. In chapter 3 and 4 it was shown that Cu6$n5 I Sn

interfaces (in Sb—based solder joints) are vulnerable during low temperature

regime TMF between -55°C and 125°C. Consequently, Figures A6 compares

the microstructural damage adjacent to the interfacial Cu68n5 layer / solder

region after 500 cycles for both SA2phe (Figure A6 (a)) and SAC2phe (Figure A6

(b)) specimens. The interfacial region in SA2phe shows extensive damage

within the solder matrix as compared to SACthe. Furthermore, CusSn5 layer/

163

 

Sn decohesion is barely noticeable in SA2phe, whereas such a damage is cleariy

visible in SACthe specimen.

164

 

20pm

Figure A3 SEM images showing microstructural evolution in SA2phe solder

joints after (a) 0, (b) 100, and (c) 500 cycles

165

. (a) 0 Cycles

 

Figure A3 SEM images showing microstructural evolution in SACthe solder

joints after (a) 0, (b) 100, and (c) 500 cycles

166

‘ (a) SA2phe

 

(b) SAC2phe

 

Figure A5 High magniﬁcation SEM images comparing surface damage within

solder region after 500 cycles of TMF in (a) SA2phe and (b) SAC2phe specimens.

167

   

 

(a) SA 2phe ,
/’ ‘ K CU6sn5

  

Cu
substrate

 

: (b)SAC3phe

 

Cu substrate

Figure A6 SEM images comparing the interfacial damage after 500 cycles in (a)

SA2phe and (b) SACthe. Cussns / Sn decohesion in SACthe have been

indicated by black arrows.

168

.omemv _m_omtoE_ Co £225 9: 8269 _ocm__m.5 3:93 305 momma: 685. $0.96 cow new 09 8am enamo<m A3

-on ucm ocqm<w 313.8 :06»: _m_omt2c_ .028 :96. 95030 05 .6 co_S_o>o .8303592E mmﬁquo h< 959".

 

 

 

 

 

169

However, SEM images comparing interfacial damage evolution between
SAC2phe (Figures A7 (a)—(b)) and SAC305 (Figures A7 (c)-(d)) suggests that the
intensity of damage and its evolution (due to low temperature regime TMF) in
SACthe is relatively less than the control SAC305 specimen.

In essence these observations tend to suggest that pheTriPOSS

reinforcements improve Cu6$n5 layer/ Sn bonding in SA, whereas such effects

are not so prominent for SAC305. It may be noted that irrespective of alloy

composition, ibutTriPOSS reinforcements resulted in considerable plastic

deformation within the solder region (Figures 4.3.6) and minimized CueSn5 / Sn

decohesion (Figures 4.3.7 and Figures 4.3.8).
A3. Conclusions

Based on residual mechanical strength and microstructural evaluations, the
following may be inferred about the effects of pheTriPOSS reinforcements on the
low temperature regime TMF behavior of Sn-based solderjoints:

1. PheTriPOSS reinforcements improve the mechanical reliability of SA and
SAC305 solder joints by stabilizing the residual mechanical behavior to a
higher strength. Similar improvements have been observed in
ibutTriPOSS reinforced solder joints. This suggests that Si-OH bonds in
POSS trisilanols are more important for improving the mechanical
reliability of Sn-based solder joints than the organic “R" group (lso-Butyl or

phenyD.

170

 

 

2. Microstructural and residual mechanical results show that improvement in
mechanical reliability is brought about in different ways depending on the
composition of solder alloy.

3. In case of SA2phe, during the early stages of TMF, initial drop in residual
strength may be attributed to accumulation of surface damage due to
plastic deformation of the solder matrix. In the later stages of TMF,
residual mechanical behavior stabilizes at a higher strength (as compared

to SA) due to accommodation of anisotropic strains (due to TMF) through

plastic deformation of the solder matrix and improvement in Cuesns layer

/ Sn interfacial bonding.

4. On the other hand, low temperature regime TMF does not seem to affect
the residual mechanical behavior of SACthe solder joints. Lack of
plastic deformation of the solder matrix with progressive TMF cycles may
be major contributing factor.

5. Results presented further show that chemistry of POSS trisilanol also
plays a role. Different "R” groups (attached to POSS trisilanol) may react
with the solder paste in a very different way during processing. Thus
resulting in an as-solidified microstructure, which accommodates

anisotropic strain due to TMF in a different manner.

171

 

 

Appendix B

Strain assisted allotropic transformation of B—Sn

Rationale and objectives

Tin-pest, which involves allotropic transformation of B-Sn to a-Sn at temperatures

below 132°C, has been recognized as a potential reliability concern for Sn-

based lead-free electronic solder joints. Effect of imposed pre-strain on B—m Sn

transformation was investigated by bending rectangular specimens of 99.99%

pure bulk B-Sn to known outer ﬁber strain levels, followed by aging at -15°C. 0-

Sn was observed on the narrow side surface of the specimens bent to non-
uniform and uniform curvatures after 2.1 and 3.5 months, respectively.
Microstructural characterization was carried out to understand the inﬂuences of
(1) imposed strain on the cr-Sn region, and (2) the advancing Bio Sn phase
boundary on the adjoining B-Sn. Growth rate studies revealed that a-Sn grew
signiﬁcantly faster under the inﬂuence of tensile pre-strain as compared to

compressive pre-strain.

172

 

B1 . Introduction
At temperatures of 13.200 and below, pure Sn can transform from B-Sn (body-

centered-tetragonal) to a—Sn (diamond-cubic). This allotropic transformation

involves a change of density from 7.265 gm/cm3 to 5.769 gm/crn3 [61], a

volumetric change of almost 26 %. This transformation is also known to be an
autocatalytic event, where the transformation will speed up once it starts.
Therefore, to reduce this incubation time, a large bulk B-Sn specimen is often
inoculated with a small piece of a—Sn [40]. The region surrounding this small
inoculated a—Sn will quickly undergoes the B—ea Sn transformation. In addition,

the reverse transformation, a—>B Sn, will take place when the temperature is

raised to above 13.200. However upon cooling back to below 13.200, the

previously infected region will quickly convert back to a-Sn with little or no
incubation requirement. For these reasons, a-Sn is referred as tin-pest or tin-
disease. The corresponding volumetric increases upon transition and the brittle
nature of a-Sn will resulted in a complete disintegration of specimens made of Sn
and alloys with high Sn content [84, 85]. Therefore, the occurrence of tin-pest
has been identiﬁed as a serious potential reliability concern for microelectronics
industry where Sn-based lead—free alloys are expected to be used extensively in
electrical interconnects [40, 41].

Solder joints are expected to experience various levels and gradients of

stress and/or strain by virtue of their geometry and the imposed service

173

 

environments [86, 87]. Consequently factors such as temperature, mechanical
loading, strain imposed due to the mismatch of thermal expansion coefﬁcients
between substrate and solder, and concentration of certain impurity elements
can effect the tin-pest formation at sub-zero environment. One of the earlier
studies on the effects of mechanical deformation on B—+d Sn transformation was
carried out by Roger et.al. [88]. However, neither the magnitude of the imposed
strain nor the kinetics of a-Sn growth on the strained specimen was reported. A
more quantitative assessment of the inﬂuence of mechanical deformation on tin-
pest formation was conducted by Joo et. al.[89]. They observed a-Sn regions on

cold-rolled Sn-0.8%Cu specimens within few days during isothermal aging at

-30°C [89]. It was suggested that the rate of transformation was directly

proportional to the imposed reduction ratio. Plumbridge’s group had also

reported u-Sn formation on the machined surfaces of Sn-0.5%Cu specimens,

when stored at -18°C for more than a year. They attributed the transformation to

the residual stresses resulting from the machining operation [40, 85]. In any
case, both cold-rolling and machining produce complicated strain states on the
surface of the specimen. Hence, it is more difﬁcult to assess the effect of
individual components of strain tensor on the transformation. In addition, there
have been very few studies that address the microstructural aspects of this
transformation[40, 84, 85, 90]. The objective of this work is to understand the
fundamental issues related to the inﬂuence of strain on the tin-pest formation and

the relative kinetics by the level of imposing strain. As describe below, well-

174

characterized deformation were imposed on 99.99% pure bulk B-Sn specimens,
store at a constant sub-zero temperature and periodically removed to room
temperature for observation of any B—m transformation. In addition, for
specimens underwent tin pest formation, we report the kinetics as well as the

mechanical property near the transformed region.
B2. Experimental details

32.1 Mechanical Pre-Straining
Commercially available 99.99% pure bulk B-Sn specimens of rectangular

geometry were subjected to three types of mechanical pre-straining followed by

aging at -1 5°C. In the ﬁrst set, ﬁve metallurgically polished [B-Sn specimens (35

mm x 8 mm x 2.3 mm) were pressed around semi-circular channels to achieve
4.8% 5.8%, 7.7%, 11.5%, and 23.0% outer ﬁber strains. In the second set, a
specimen (20 mm x 12 mm x 1.2 mm) was bent to non-unifonn curvatures with
outer ﬁber strains ranging from 3.6% to 6%.
82.2 Characterization and measurements

X-ray diffraction was used to verify the transformed region is that of a-Sn
phase. SEM and stereo-optical microscopy were utilized for microstructural
characterization of transformed regions. Mechanical characterization was carried
out with Vickers hardness testing with a load of 1 Kg applied for 10 seconds.
Kinetics of [3—>a transformation was studied by measuring the temporal evolution
of the transformed areas by periodically recording images of transformed regions
with a calibrated stereo-optical microscope. The transformed area was then

measured directly from the digital image with a resolution of 5 pm per pixel.

175

 

 

B3. Results and discussion

83.1 Effect of mechanical pro-straining on the allotropic transformation of
B-Sn
a-Sn formation was noted on the narrow side surfaces in regions very close to

the broad outer surface of the non-uniformly and the uniformly bent specimens

after 2.1 and 3.5 months of storage at -15°C, respectively. On the non-uniformly

bent specimen, a-Sn was located in the region with 6% tensile outer ﬁber pre—
strain. This observation is consistent with the results published by Roger et.al,
where a-Sn was reported to have formed on the tensile side of bent specimens
[88]. However, Baa Sn transformation was also noted on the narrow side
surface in the compressive region of the uniformly bent specimen with 7.7%
outer ﬁber pre-strain. Stereo-optical images showing the location of o-Sn on the
transformed specimens are presented in Figure B1. Wart-like a-Sn can be
observed on the non-uniformly bent specimen as shown in Figure B1 (a), and
two transformed regions on the uniformly bent specimen are indicated as regions
1 and 2 in Figure B1 (b). Unlike the specimen with 7.7% outer ﬁber pre-strain,
other uniformly bent specimens with outer ﬁber pre—strains of 4.6%, 5.8% and

11.5% and 23.0% did not exhibit any transformation even after 6.5 months of

aging at -15OC.' Similarly, in the non-uniformly bent specimen, tin pest was

observed only in the region with an outer ﬁber tensile strain around 6%. X-ray
diffraction (XRD) was performed using pieces detached from the transformed

region on the non-uniformly bent specimen. A plot of diffraction intensity versus

176

diffraction angle was presented in Figure 82, which conﬁrms the presence of 0-

Sn.

177

 

Figure B1 Stereo-optical images showing the location of a-Sn on the (a) tensile

side of the non-uniformly bent specimen with 6% outer ﬁber strain, (b)
compressive side of the uniformly bent specimen with 7.7% outer ﬁber strain

(shown as Region-1 and Region-2).

178

100'1'r'r-r'r'r'r'r

 

Intensity(AU)

 

 

 

II

80 _ I a—Sn, gray tin
0 B-Sn, gray tin j
60 — -
40 - -

I
20 - I _
O
0 - _
1 I ' I I F

 

' l ' l ' I I 1 v v
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
d-Spacing, Angstrom

Figure 82 XRD of the q-Sn detached from the non-uniformly bent specimen.

179

 

Diffraction peaks associated with of B—Sn were also observed in this Figure,
which was expected since some B—Sn that was attached to o-Sn particles used
for the XRD observation. Due to the slow rate of transformation in the
compressive region of uniformly bent specimen, it was not possible to obtain
enough a-Sn particles for XRD analysis.

These observations clearly indicate that the state (tensile or compressive)
of strain has no effect on the ability of B-Sn to undergo the transformation to q-Sn,
since we observed formation of a-Sn on the tensile and compressive regions of
the non-uniformly and uniformly bent specimens, respectively. Furthermore,
there appears to be a critical range of outer ﬁber strains which will facilitates the
transformation, since o-Sn forms in regions which have relatively close values of
outer ﬁber strain, i.e. 6% and 7.7%. These observations suggest that there exist
a critical strain energy of deformation needed to provide the necessary driving
force for the transformation and such energy is independent of the state of strain.
Moreover, since initiation of tin pest was observed only on the narrow side of the
bent specimens, this suggests that a strain gradient may be important for the
B—m Sn transformation. The necessity of strain gradient for the nucleation of a-

Sn is further substantiated by lack of tin pest formation in uniaxially deformed

specimens after 11 months of storage at -1SOC employed in a parallel study, and

lack of nucleation of a-Sn on the broad side of the bent specimens which have
relatively uniform strains. The observed a-Sn on broad surfaces of bent
specimens resulted only by its growth from the nucleation site present on the

narrow side at locations close to the broad surface.

180

I“

B3.2 Microstructural and mechanical characterization

Microstructural evolution of a-Sn regions on the non-uniformly bent and uniformly
bent specimens are presented in Figures B3 and B4 respectively. After 2.1
months of aging, cracks within the o-Sn region were observed to be parallel to
the edge of the non-uniformly bent specimen (Figure BB (3)). However, after
additional 15 days of aging cracks in same region appeared to spread out radially
(Figure B3 (b)). Such an observation indicates that as d-Sn region grows and
newer cracks are formed while the existing ones becomes larger. Comparison
between Figures B3 (b), B4 (b), and B4 (d) show extensive crack formation, after
additional 15 days of aging, on the region experiencing tensile pre-strains in non-
uniformly bent specimen, while fewer cracks are visible in regions under
compressive pre-strains in the uniformly bent specimen. Furthermore, cracks
present with in a-Sn are both inter-granular and trans-granular in nature (Figures
85). As transformation progresses these cracks combine to form larger cracks
(shown in Figures B3 (b) and B4 (b)) and lead to the spalling of a-Sn (Figure B3
(b))-

Growth of o-Sn also affects the adjoining B-Sn region due to the increase
in volume that needs to be accommodated. Micrographs showing surface relief
features within the B-Sn region adjacent to a-Sn region are provided in Figures
B6 (a) and B6 (b). These observations suggest that compressive stresses
generated during the transformation result in surface relief features within the [3-

Sn region.

181

 

Inﬂuence of volumetric expansion warranted by the transformation on the
B-Sn region was further investigated by conducting Vickers indentation tests on

the broadside of the non—uniformly deformed specimen.

 

Figure B3 SEM images showing the microstructural evolution of a-Sn on the

non-uniformly bent specimen aged at -15 C (a) as observed after 2.1 months, (b)

after 2.6 months. Scale bar indicates 500nm.

182

 

3.5 months 4 months

 

 

Figure B4 SEM images showing the microstructural evolution of o—Sn on the
uniformly bent specimen aged at -15°C (a) after 3.5 months in Region-1 (b) after

4 months in Region-1, (c) after 3.5 months in Region-2 (d) after 4 months in

Region-2.

183

 

m t .
Trans-Granular
Crack

Inter-gran ular Cracks

r
.1

.I.

‘ J
‘k ‘
4x .1

‘.

Figure BS SEM micrographs showing the inter-granular and trans-granular
cracks within a-Sn region on (a) non-uniformly bent, and (b) uniformly bent

specimens.

184

 

Figure 36 SEM micrographs showing surface relief features with in B—Sn region

adjacent to a-Sn (a) non-uniformly bent, and (b) uniformly bent specimens.

185

Schematic presented in Figure B7 (a) shows the location on the specimen

where the indents were made. Figure B7 (b) presents the SEM micrograph of

locations of individual indents within the B-Sn region near the BIG interface.

lndent sizes became smaller at locations closer to the o/B interface, indicating

that B-Sn is harder near the 018 interface. The measured values of hardness are

presented in Table Bl. The apparent increase in hardness in the adjoining B-Sn

region is due to the compressive stresses exerted by the advancing a/B Sn

phase boundary.

Table Bl Hardness values of B-Sn at different locations and distances from the

[BIG Sn interface

 

Location of the

Distance of each indent from the

Vickers hardness

 

 

 

lndent Blu interface (pm) (VPH)
A 100 46
B 400 51 5

 

 

 

186

 

Indents

Joounf
_“

I _.
. “

I .

 

(b)
Figure B7 (a) Schematic showing the location of the indents on the broad side of
the non-uniformly bent specimen (b) SEM image showing the location and size of
the indents. A load of 1 Kg was applied to the broad side of the specimen for 10

seconds.

187

83.3 Effect of stress on the growth kinetics of a—Sn

Growth kinetics of the transformation was studied by measuring the area of a a-
Sn region as a function of days. Figure 88 (a) shows the a-Sn region observed
after 5.1 months on the tensile side of the non-uniformly deformed specimen. In
Figure 88 (b), the square root of the measured surface area for the region
showed in Figure 88 (a) was plotted versus time. Measurements were stopped
after the chosen region impinged with a neighboring a-Sn region. Similar
measurements were also made on a a-Sn region in region-1 on the compressive
side of the unifome bent specimen. Plot showing the variation of square root of
the surface area of region-1 with number of days is shown in Figure 89. Linear
least square ﬁtting on the data shown in Figure 88 (b) and Figure 89 were

employ using the form of:

A1/2=at+b (1)

2
Where, A is the measured area in mm , t is time in days, where a, and b are

ﬁtting constants with units mm/day and mm, respectively.

Similar relationship, as presented in equation (1), has been previously reported
by Rogers et.al [88]. The values of constants a and b appearing in equation (1),
and calculated average areal growth rates, obtained from the present study are
listed in Table Bll. Evidently in the presence of tensile pre-strain, the areal
growth rate of a-Sn region is two orders of magnitude faster as compared to that
noted in regions with compressive pre-strain. This observation suggests that

kinetics of the B—m Sn transformation is inﬂuenced by the state of strain.

188

 

 

Figure 88 (a) Stereo-optical image showing the location of the q-Sn wart (inside
the white circle) on the tensile side of the non-uniformly bent specimen used for

measuring growth kinetics

189

 

[Area, (mm2)]"2
a .1 .3 .3 .s N N
'o is s '0) on 'o to
I n I I I n I n I n I

.0
on
I .

 

 

 

 

P
o:

I ' I ' I ' I ' I I I ' I
150 155 160 165 170 175 180

Time,(days)

(b)

Figure 88 (b) plot showing the change in measured area of the same a-Sn wart

with number of days.

190

 

 

53
a)
o
1

L

[Area, (mm2)]"2
.o .0
8 ‘01”

 

 

 

0.35 . , . , . , . , . , . , .
100 110 120 130 140 150 160 170
Time,(days)

 

Figure 89 Plot showing the change in the measured area of Region-1 as a
function of days. Region-1 was located on the compressive side of the uniformly

bent specimen.

191

Table 8ll Values showing the growth rate for a given strain state of the

specimens and the corresponding correlation coefﬁcients for linear least

 

 

 

square ﬁts
Average aerial
State of imposed b 2
growth rate a (mmlday) R Figures
strain 2 (mm)
(mm I day)
Region with
0.143 0.0502 -6.024 0.982 C8(b)
tensile pre-strain
Region with
compressive pre— 0.005 0.0047 0.963 09
0.1241
strain

 

 

 

 

 

 

 

 

During the transformation a-Sn region relieves the stresses generated due
to 27% volume increase by forming cracks. While tensile stresses facilitate this
process of crack formation, compressive stress retards it. Figures 83 (b) and 84
(b) show such an effect of tensile and compressive stresses on crack formation.
As a result, a-Sn region is able to grow quickly under the inﬂuence of tensile
stresses by effectively relieving the transformational stresses through crack
formation. Such a rationale suggests that Sn-based solder joints which are
under the inﬂuence of compressive stresses will have higher reliability even if q-

Sn formation has already taken place.

192

 

84. Summary

Allotropic transformation of B-Sn to a-Sn in mechanically deformed specimens
stored at a constant sub-zero temperature was studied. In these studies using
pre-strained specimens incubation time was still needed to initiate the
transformation, although, it is not yet clear how the length of incubation time is
related to the mode and the magnitude of mechanical deformations. However,
based on the current study some signiﬁcant inferences can be made:

1. Presence of a strain gradient may enhance nucleation of tin pest.

2. A critical strain range, irrespective of being tensile or compressive,
appears to facilitate B—+a Sn transformation. Such an observation
suggests that critical strain energy may aid the transformation.

3. Growth rate of a-Sn is much faster in regions with tensile strain as

compared to that in regions with compressive strain.

193

 

 

References

[1] J. G. Lee, PhD Thesis (2005).

[2] F. Guo, Journal of Materials Science-Materials in Electronics 18 (2007) 129-
145.

[3] F. Guo, J. Lee, K. N. Subramanian, Soldering & Surface Mount Technology
15 (2003) 39-42.

[4] J. Sigelko, S. Choi, K. N. Subramanian, J. P. Lucas, T. R. Bieler, Journal of
Electronic Materials 28 (1999) 1 184-1 188.

[5] M. McCormack, S. Jin, G. W. Kammlott, H. S. Chen, Applied Physics Letters
63 (1993) 15-17.

 

[6] H. Rhee, K. N. Subramanian, A. Lee, J. G. Lee, Soldering & Surface Mount
Technology 1 5 (2003) 21 -26.

[7] C. Kanchanomai, Y. Miyashita, Y. Mutoh, Journal of Electronic Materials 31
(2002) 456-465.

[8] l. E. Anderson, Journal of Materials Science-Materials in Electronics 18
(2007) 55—76.

[9] Y. Kariya, M. Otsuka, Journal of Electronic Materials 27 (1998) 866-870.
[10] Y. Kariya, M. Otsuka, Journal of Electronic Materials 27 (1998) 1229—1235.

[11] l. E. Anderson, 8. A. Cook, J. Harringa, R. L. Terpstra, J. C. Foley, O. Unal,
Materials Transactions 43 (2002) 1827-1832.

[12] l. E. Anderson, J. C. Foley, 8. A. Cook, J. Harringa, R. L. Terpstra, O. Unal,
Journal of Electronic Materials 30 (2001) 1050-1059.

[13] l. E. Anderson, J. Walleser, J. L. Harringa, Journal of Minerals Metals and
Materials 59 (2007) 38-43.

[14] D. W. Henderson, J. J. Woods, T. A. Gosselin, J. Bartelo, D. E. King, T. M.
Korhonen, M. A. Korhonen, L. P. Lehman, S. K. Kang, P. Lauro, D. Y. Shih,
C. Goldsmith, K. J. Puttlitz, Journal of Materials Research 19 (2004) 1608-
1612.

[15] C. M. Miller, I. E. Anderson, J. F. Smith, Journal of Electronic Materials 23
(1994) 595-601 .

[16] N. Barry, I. P. Jones, T. Hirst, l. M. Fox, J. Robins, Soldering 8 Surface
Mount Technology 19 (2007) 29-38.

194

[17] J. K. Shang, Q. L. Zeng, L. Zhang, Q. S. Zhu, Journal of Materials Science-
Materials ln Electronics 18 (2007) 211-227.

[18] K. N. Subramanian, Journal of Materials Science-Materials in Electronics 18
(2007) 237-246.

[19] S. Choi, J. G. Lee, K. N. Subramanian, J. P. Lucas, T. R. Bieler, Journal of
Electronic Materials 31 (2002) 292-297.

[20] J. G. Lee, F. Guo, S. Choi, K. N. Subramanian, T. R. Bieler, J. P. Lucas,
Journal of Electronic Materials 31 (2002) 946-952.

[21] J. G. Lee, K. N. Subramanian, Journal of Electronic Materials 32 (2003) 523-
530.

[22] K. N. Subramanian, Journal of Electronic Materials 34 (2005) 1313-131 7.

 

[23] K. N. Subramanian, Fatigue & Fracture of Engineering Materials & 1
Structures 30 (2007) 420-431.

[24] K. N. Subramanian, J. G. Lee, Materials Science And Engineering A-
Structural Materials Properties Microstructure & Processing 421 (2006) 46-56.

[25] J. G. Lee, L. Telang, K. N. Subramanian, T. R. Bieler, Journal of Electronic
Materials 31 (2002) 1 152-1159.

[26] K. N. Subramanian, J. G. Lee, Journal of Materials Science-Materials In
Electronics 15 (2004) 235-240.

[27] T. Chen, l. Dutta, Journal of Electronic Materials 37 (2008) 347-354.

[28] S. G. Jadhav, T. R. Bieler, K. N. Subramanian, J. P. Lucas, Journal of
Electronic Materials 30 (2001 ) 1 197-1205.

[29] H. Rhee, K. N. Subramanian, Soldering & Surface Mount Technology 18
(2006) 19-28.

[30] H. Rhee, K. Subramanian, A. Lee, Journal of Materials Science-Materials in
Electronics 16 (2005) 169-176.

[31] D. Swenson, Journal of Materials Science-Materials in Electronics 18 (2007)
39-54.

[32] J. G. Lee, K. N. Subramanian, Soldering & Surface Mount Technology 17
(2005) 33-39.

[33] M. Kerr, N. Chawla, Acta Materialia 52 (2004) 4527-4535.

195

[34] C. Andersson, P. E. Tegehall, D. R. Anderssn, G. Wetter, J. Liu, leee
Transactions on Components & Packaging Technologies 31 (2008) 331-344.

[35] l. Dutta, D. Pan, R. A. Marks, S. G. Jadhav, Materials Science and
Engineering A-Stmctural Materials Properties Microstructure and Processing
410 (2005) 48-52.

[36] A. R. Fix, W. Nuchter, J. Wilde, Soldering & Surface Mount Technology 20
(2008) 13-21.

[37] B. F. Dyson, Journal of Applied Physics 37 (1966) 2375-8.

[38] 8. F. Dyson, T. R. Anthony, D. Tumbull, Journal of Applied Physics 38
(1967) 3408-8.

[39] G. Deiter, Mechanical Metallurgy, 2ed, Prentice Hall (1998).

[40] W. J. Plumbridge, Journal of Materials Science-Materials in Electronics 18
(2007) 307-318.

[41] W. J. Plumbridge, Soldering & Surface Mount Technology 16 (2004) 13-20.
[42] J. G. Lee, K. N. Subramanian, Microelectronics Reliability 47 (2007) 1 18-131 .

[43] J. McDougall, S. Choi, T. R. Bieler, K. N. Subramanian, J. P. Lucas,
Materials Science and Engineering A-Structural Materials Properties
Microstructure and Processing 285 (2000) 25-34.

[44] D. S. Liu, C. Y. Kuo, C. L. Hsu, G. S. Shen, Y. R. Chen, K. C. Lo, Materials
Science and Engineering A-Structural Materials Properties Microstructure
and Processing 494 (2008) 196-202.

[45] Y. T. Chin, P. K. Lam, H. K. Yow, T. Y. Tou, Microelectronics Reliability 48
(2008) 1079-1086.

[46] N. Zhang, Y. W. Shi, 2. D. Xia, Y. P. Lei, F. Guo, X. Y. Li, Journal of
Electronic Materials 37 (2008) 1631 -1 639.

[47] D. Y. R. Chong, F. X. Che, J. H. L. Pang, K. Ng, J. Y. N. Tan, P. T. H. Low,
Microelectronics Reliability 46 (2006) 1 160-1171.

[48] l. E. Anderson, 8. A. Cook, J. Harringa, R. L. Terpstra, Journal of Electronic
Materials 31 (2002) 1 166-1 174.

[49] K. Jonnalagadda, F. J. Oi, J. Liu, leee Transactions on Components and
Packaging Technologies 28 (2005) 430-434.

196

 

[50] K. S. Kim, S. H. Huh, K. Suganuma, Materials Science and Engineering A-
Structural Materials Properties Microstructure and Processing 333 (2002)
106-1 14.

[51] J. C. Suhling, H. S. Gale, R. W. Johnson, M. N. Islam, T. Shete, P. Lall, M. J.
Bozack, J. L. Evans, P. Seto, T. Gupta, J. R. Thompson, Soldering 8 Surface
Mount Technology 16 (2004) 77-87.

[52] S. Choi, T. R. Bieler, J. P. Lucas, K. N. Subramanian, Journal of Electronic
Materials 28 (1999) 1209-1215.

[53] C. M. L. Wu, Y. W. Wong, Journal of Materials Science-Materials in
Electronics 18 (2007) 77-91.

[54] M. Pei, J. M. Qu, Journal of Electronic Materials 37 (2008) 331-338.

[55] A. Lee, K. N. Subramanian, Journal of Electronic Materials 34 (2005) 1399-
1407.

[56] S. H. Phillips, T. S. Haddad, S. J. Tomczak, Current Opinion in Solid State &
Materials Science 8 (2004) 21-29.

[57] R. M. Laine, Journal of Materials Chemistry 15 (2005) 3725-3744.

[58] R. Murugavel, A. Voigt, M. G. Walawalkar, H. W. Roesky, Chemical Reviews
96 (1 996) 2205-2236.

[59] J. K. Kim, K. H. Yoon, D. S. Bang, Y. 8. Park, H. U. Kim, Y. H. Bang, Journal
of Applied Polymer Science 107 (2008) 272-279.

[60] L. Song, Q. L. He, Y. Hu, H. Chen, L. Liu, Polymer Degradation and Stability
93 (2008) 627-639.

[61] L. e. DR, (2003).

[62] X. W. Liu, W. J. Plumbridge, Journal of Electronic Materials 36 (2007) 1111-
1 120.

[6318. Chada, R. A. Foumelle, W. Laub, D. Shangguan, Journal of Electronic
Materials 29 (2000) 1214-1221 .

[64] K. W. Moon, W. J. Boettinger, U. R. Kattner, F. S. Biancaniello, C. A.
Handwerker, Journal of Electronic Materials 29 (2000) 1122-1136.

[65] M. Rappaz, P. Thevoz, Acta Metallurgica 35 (1987) 1487-1497.
[66] R. A. Robert E Reed-Hill, Physical Metallurgy, Prentice Hall (1991).
[67] B. H. K. Kingery W.D., Uhlmann D.R., Introduction to Ceramics (1976).

197

[68] R. Trivedi, W. Kum, International Materials Reviews 39 (1994) 49-74.
[69] R. Trivedi, W. Kurz, Acta Metallurgica Et Materialia 42 (1994) 15-23.

[70] P. W. Voorhees, Annual Review of Materials Science 22 (1992) 197-215.
[71] S. E. Babcock, R. W. Ballufﬁ, Acta Metallurgica 37 (1989) 2357-2365.
[72] S. E. Babcock, R. W. Ballufﬁ, Acta Metallurgica 37 (1989) 2367-2376.
[73] T. E. Hsieh, R. W. Ballufﬁ, Acta Metallurgica 37 (1989) 1637-1644.

[74] H. E. Kissinger, Journal of Research of the National Bureau of Standards 57
(1956) 217-221.

[75] S. L. Allen, M. R. Notis, R. R. Chromik, R. P. Vinci, Journal of Materials
Research 1 9 (2004) 141 7-1 424.

[76] l. M. Lifshitz, V. V. Slyozov, Journal of Physics and Chemistry of Solids 19
(1961) 35-50.

[77] A. J. Ardell, Acta Metallurgica 20 (1972) 601-8.
[78] O. N. Senkov, M. M. Myshlyaev, Acta Metallurgica 34 (1986) 97-106.

[79] A. R. Fix, G. A. Lopez, I. Brauer, W. Nuchter, E. J. Mittemeijer, Journal of
Electronic Materials 34 (2005) 137-142.

[80] P. G. Shewmon, Diffusion in Solids, McGraw Hill (1989).

[81] AW. Gibson, K.N. Subramanian, T.R Beiler, Design Reliability of Solder
and Solder lnterconnections, TMS Annual meeting, vol 1997, p."pp.

[82] J. Sommer, C. Herzig, Journal of Applied Physics 72 (1992) 2758-2766.
[83] Y. Q. Zhao, D. A. Schiraldi, Polymer 46 (2005) 1 1640-1 1647.

[84] Y. Kariya, N. Williams, C. Gagg, W. Plumbridge, JOM-Joumal of The
Minerals Metals & Materials Society 53 (2001) 39—41.

[85] Y. Kariya, C. Gagg. W. J. Plumbridge, Soldering & Surface Mount
Technology 13 (2001) 39-40.

[86] J. Cugnoni, J. Botsis, V. Sivasubramaniam, J. Janczak-Rusch, Fatigue &
Fracture of Engineering Materials & Structures 30 (2007) 387-399.

[87] L. H. Xu, J. H. L. Pang, K. N. Tu, Applied Physics Letters 89 (2006).

198

[88] R. R. Rogers, J. F. Fydell, Journal of The Electrochemical Society 100
(1953) 383-387.

[89] Y. J. Joo, T. Takemoto, Materials Letters 56 (2002) 793-796.

[90] R. C. Farrow, M. T. Asom, K. A. Jackson, L. C. Kimerling, Institute of Physics
Conference Series (1989) 259-264.

199

   

llllllllllllllHllllllllHillllllllllllllllllllllllllllll