, a «3.7..

o .w. Ann “#5..

,
,, .D.
.. .4
r . n

.kfiw

9:!
t, .L ; ;r
:v . Iv

352 m. ..
. . “Eu

I?

a.

u
.k

A

 

.mwmmfirhu away...
.V b 0.4.2.5“? ark».
.61 .flP-‘vrl

. 3:
ill:

v21
: pal-33“.».
first... 1“. £95: 3)
3. it 9;: :r’LI
In .IJOu.’I}!...Li-O1
. 1‘53“) is}! .

.lv.‘
1. 3L

. fin... fluh..._.8-n.w u. . . , LT AWN?»urn...““VHWJTmiflfwwhm. l...

 

This is to certify that the
dissertation entitled

STUDIES ON LEAD-FREE SOLDERS REINFORCED WITH
MECHANICALLY INCORPORATED
Cu, Ag, and Ni PARTICLES

presented by

Fu Guo

has been accepted towards fulfillment
of the requirements for

. ls Scie
EPh.D. degreem Materia nce

 

and Engineering

flo‘fiw

Major professor

 

Date 02 (8702*

SU '3 an Affirma; .v! Action/Equal Opportunity Institution 0‘ 12771
I

 

 

LIB RARY
Michigan State
University

 

 

 

 

PLACE IN RETURN BOX to remove this checkout from y 0 ur reco rd
TO AVOID FINES return on or before date due- '
MAY BE RECALLED with earlier due date if requested,

 

 

DATE DUE DATE DUE ' DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClFiC/DateDue.p65-p. 15

 

 

 

STUDIES ON LEAD-FREE SOLDERS REINFORCED WITH
MECHANICALLY-INCORPORATED
Cu, Ag, AND Ni PARTICLES
By

Fu Guo

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the reCIUil‘ementS
for the degree 0f
DOCTOR OF PHILOSOPHY

Department of Chemical Engineering and Materials Science

2002

ABSTRACT

STUDIES ON LEAD-FREE SOLDERS REINFORCED WITH MECHANICALLY
INCORPORATED Cu, Ag AND Ni PARTICLES

By
Fu Guo
Three types of composite solders were pI‘OdUCed in the current investigation by
mechanically adding nominally 15v% of ~ 6pm sized Cu, ~ 4M1“ sized Ag, or ~ 6pm
sized Ni particles into the eutectic Sn-3.5Ag solder paste. The principal aim of this Study
is to investigate how the microstructure, isothermal aging, reflow as Well as creep
properties of the eutectic Sn-3.5Ag solder will be affected by the lhco’l’OI‘ation of these
particles as reinforcements. Small realistic sized single shear lap SOIder joint design
were used throughout the investigation to better mimic the Condition for solders used in
automobile and microelectronic industries- The initial microstructure 0f as-fabricated
composite solder joints was examined and analyzed using Optical and scanning electron
miCI'OSCOpy (SEM) with Energy Dispersive X-ray (EDX). The effect of refloW and
isothermal aging on the microstructure as well as the mol'PhOlOgical changes in the
interfacial intermetamc (M) layers of the composite solder joints were extensively
analyzed, Effect of solder reflow on the solderability and mechanical pr0perties were
studied. Nanoindentafio“ testing (NIT) was used to obtain mechanical data from
multiple reflowed composite solders. Creep tests were conducted on composite solder
idmts at 25°C, 65°C and 105°C representing hemm’k’g‘ms temperatures r“gins from 0.61

to 0.78. Qualitative and quantitative evaluations of creep behavmr were Obtained from

the distortion of excimer laser induced surface ablation markings on the sol der jam
Various creep parameters such as global and localized creep strain, variation of creep
Strain and strain-rate, activation energy for creep, and the onset of tertiary creep were
determined. The resultant properties of the Composite solders Pmd‘me’d in this
investigation were compared with those of non-COTTIPOSite solders and “‘6 comPOSite
solders produced by in-situ method.

Significant findings in this study revealed that both Cu and Ni composite Solder
joints significantly improved the creep resistance of euteCtic Sn-3.5Ag Solder joints.
Although Ag composite solder joints exhibited comparable creep resistanCe to eutectic
Sn-3.5Ag solder, very uniform deformation features were observed dUrin g creep.
Excessive microstructural evolution in terms of IM growth was observed in Cu and Ni
composite solder joints under isothermal aging at 150°C due to the profuse diffusion of
Cu and/or Ni in Sn. In contrast, the microstructure in the Ag composite solder joint Was
very stable under isothermal aging and reflow conditions. The Ni composite solder joi ht

exhibited a stable microstructure after multiple reflOWS Whereas extensive 11V! growth Was
Still Obseng in Cu composite solder joint under similar reflow conditions. Ag and Ni
have Comparable wetting characteristics to eutectic Sn-3.5Ag solder. Cu composite
solder With 15v% of Cu reinforcements resulted in a high average contact angle due to
the inct-eaSe of effective volume fraction. Responsible mechanisms for the effects of

mlnfomement addition on microstructure, aging, “HOW and creep behaVi‘” are discussed.

ACKNOWLEDGMENTS

I would like to express my sincere thanks to my adVisor Professor James P- Lucas
for all his excellent guidance, encouragement, friendly manner, and PTOfeSSlonal example
throughout my doctoral studies at Michigan State University. Without his support and
patience, this work would not have been possiblc- I Would also like ‘0 3.1““:le thank
Professor K.N. Subramanian and Professor Thomas R. Bieler for their kind help and
instruction throughout the entire work. I also thank Professor P- Duxbury and PTOfessor
D. Liu for serving as my graduate committee and giving me valuable suggestions to both
the thesis and experimental designs. The success of this work is also based 0n the help
from all other members in our research groUP- Special thanks go to J- Lee, C, Brandon, H.
Rhee, and S. Choi for specimen preparation, property testing and information exchange,

I sincerely appreciate the financial support PTOVided by my adViSors though the
Research Excellent Fund administered through the Composite Materials and Structure
Center at Michigan State University. The later part Of this work was supported by the
Visteon Electronic Technical Center, Dearborn, MiChigan, and National SCienCe

Foundation under contract number NSF-DMR-0031796-
My deepeSt thanks are extended to my wife, my parents, and my friends whose
love, Corltinuous support and encouragement have supported me to achieve a PhD.

degree from Michigan State University.

iv

TABLE OF CONTENTS

 

 

 

 

 

 

 

List of Tables ““=‘“ :’ """" X
List of Figures ‘ =“ °°°°°°°°°° Xi
Chapter I Introduction -- ----- " ........... 1
Chapter 11 Literature Review -- - - --.- ‘ “ 4
2.1 Current Research Status on Lead—f1“ee Solders.........---................ 4
2.1.1 Introduction ..................................................................................... 4
2.1.2 Service Requirements ..................................................................... 6
2J3 Prior Studies ............................................................................. 7
2.2 Composite Solders .................................................................................. 10
2.2.1 Composite Approach and Its Merit to Enhance the Behavior of
Solders .......................................................................................... 10

2.2.2 Important Considerations for the Selection of Reinforcements 14

2.2.3 Methods for Introducing the Reinforcements —— in-situ,
Mechanical Mixing (Inert or Reactive Rei nforcements) . - . . 15

2.2.4 Microstructualllnterfacial ASPeCtS 0f Lead-free Composite

Solders ........................................................................................... 16

2.2.5 Resultant Pmperties of Lead-free Composite Solders . . - . - 17
2.2.6 Summary ....................................................................................... 19
2,3 Creep Deformation Behavior of Sn-Based Solders ............................... 20
2.3.1 Overview of Creep Deformation Theory ...................................... 20

2.3.2 Importance of Studying the Creep Behavior of Sn-based Sol d er
Materials ......................................................................................... 31
2.3.3 Materials Used in Creep Tests ...................................................... 31

2.3.4 Set-ups for the Creep Tests .........................................................

 

32
2.3.5 Reported Creep Data and Related Mechanisms for Sn-based Solder
Materials ....................................................................................... 34
2.4 Microstructural Issues in Solder Materials ............................................ 38
Chapter III Microstructure of As-fabricated Cu, Ag, and Ni Particle ReinfOI‘CEd
Composite Solder Joints __ ...... 41
3.1 Preparation of Composite SOlders.. ........................................................ 41
3.2 Solder Joint Fabrication ......................................................................... 42

3.3 Microstructure of As-fabn'cated Cu Particle Reinforced composite Solder
Joints ....................................................................................................... 44

3.4 Microstructure of As-fabricated Ag Particle Reinforced Composite Sold“
Joints ....................................................................................................

3.5 Microstructure of As-fabricated Ni Particle Reinforced Composite Soldct
Joints ....................................................................................... 46

3.5.1 Microstructure under RegUIar Heating and Cooling Conditions -- 46

3.5.2 Microstructure under Difffi‘vrent Heating and Cooling Condi ti ons 5 1

3.6 Summary ............................................................................................... 68

Chapter IV Microstructural Evolution in Cu, Ag, and Ni Particle Reinforced
Composite Solder Joints under Isothermal Aging at 150°C.....---...... 70

4.1 Introduction ............................................................................................ 70
4, 2 Experimental Procedures ................................................................... 72
4.3 Microstructural Evolution in Cu Particle Reinforced Composite Solder

Joints ....................................................................................................... 72
4.3.1 Intermetallic Layer around Cu Particles ........................................ 72

4.3.2 Intermetallic Layer at the Cu Substrate/Solder Interface . - , - 74

4.3.3 Mechanism for IM Layer Growth during Solid-state ISOthermal
Aging ............................................................................................ 77

vi

(it;

4.4 Microstructural Evolution in Ag Particle Reinforced Composite Solder

Joints ....................................................................................................... go
4.5 Microstructural Evolution in Ni Particle Reinforced Composite Solder
Joints ....................................................................................................... 84
4.5.1 1N1 Layer around the Ni Particle Reinforcements ......................... 84
4.5.2 IM Layer at the Cu Substrate/Solder Interface .............................. 92
4.5.3 Possible Mechanism for M Layer Growth during Solid—state
Isothermal Aging ........................................................................... 95
4.6 Summary ................................................................................................ 96

Chapter V Effect of Reflow on the Solderability, Microstructure and

 

Mechanical Properties of Cu, Ag, and Ni Particle Reinforced
Composite Solders -- -- 9%
5.1 Introduction ............................................................................................ 9%
5.2 Experimental Procedures ....................................................................... 99
5.2.1 Materials ....................................................................................... - 99
5-2-2 Wetting Exverimems .................................................................... .99
. 100

5.2.3 Reflow Experiments on Solder Joints..............

5.3 Effect of Reflow on Solderability of C11, Ag, and Ni Particle Rein forced

Composite Solders ............................................................................... . 101

5 .4 Effect of Reflow on Microstrucml‘e 0f Cu, Ag, and Ni Particle
Reinforced Composite Solder Joints .................................................... 107
.. 107

5.4.1 Intermetallic Layer around the Particle Reinforcements... - - - .

5.4.2 Interfacial Intenneta] lic Layer in the Composite Solder Joints... 112

5.4.3 Effect of Reflow on the metastructure of Eutectic Sn-3.5Ag

Solder Joints ................................................................................ 1 17

5.5 Effect of Reflow on Mechanical Pr0perties of Eutectic Sn-35Ag
SOlderS .................................................................................................. 117
5.6 Summary ............................................................................................. . 125

vii

Chapter VI Creep Deformation Behavior of Eutectic Sn-3. 5Ag Solder Joints
with or without Small Alloying Element Additions, and Cu, Ag, Ni

 

 

Reinforcement Particles- --—===— ‘---==-.. 128
6.1 Introduction .......................................................................................... 128
6.2 Experimental Procedures ..................................................................... 130

6. 3 Creep Deformation Behavior of Cu, Ag, Ni Particle Reinforced Solder
Joints as Compared to Eutectic Sn- 3. 5Ag and Sn-4Ag-0. SCu Solder

Joints ..................................................................................................... 135
6.3.1 Typical Creep Data ...................................................................... 135
6.3.2 Secondary Creep Rate for Different Solder Joints at Different
Testing Temperatures .................................................................. 143
6.3.3 Strain at the Onset of Tertiary Creep .......................................... 145
6.3.4 Activation Energy for Creep - .................................................. 1‘“
6.3.5 Deformation Profiles in Composite and Non-composite Solder
Jomts ...................................................................................... . 150
6 3 6 Summary .................................................................................... 155
6.4 Evaluation of Creep Behavior 0f Near Eutectic Sn- 3. m5Ag Solders
Containing Small Amount of Alloy Additions ........... .... .- - - -.... 156
6.4.1 Microstructure ............................................................................ . 157
6.4.2 Comparison of Secondary Creep Strain Rate under Different
Applied Stresses .......................................................................... 157
6. 4. 3 Comparison of Alloys with Normalized Strain Rate vs. N ormalized
Stresses Plot.. ........................................................................... 162
6.4.4 Strain and Time at the Onset of Tertiary Creep .......................... 164

6.4.5 Microstructural Examination of Deformed Solder Joints - - , , 167

 

6.4.6 Summary ..................................................................................... 170
Chapter VII Conclusions and Recommendations --- -- --- """""" “" 172
7. 1 Conclusions .......................................................................................... 172

viii

7.2 Closing Thoughts / Recommendations ................................................

 

 

 

I 78
Appendices .................... _ —==..., 182
Appendix A Procedures for the Calculation of Global and Localized Creep
Parameters ............................................................................... 1 83
Appendix B Procedures for Operating HIT ACI-H-2500C Scanning Electron
Microscope .............................................................................. 219
Appendix C List of Personal Publications & Presentations Directly Related
to the Investigation in This Thesis. ......................................... 221
References ----=--- - --. .............. 223

 

 

ix

lei

Tab.

lat

LIST OF TABLES

Table 2.1 Examples of Lead-free Soldering Project in the World. (p. 8)

Table 2.2 Examples of Lead-free Solder Alloys Currently under Research and
Development. (p.9)

Table 3.1 Comparison of Diffusion Parameters for Ag. CU. and Ni in 311- (13- 50)
Table 3.2 Results Based on the Study on IMC Morphology. (p. 68)

Table 5.1 Wetting Angles as a Function of Reflow Conditions (Angles Measured in
Degrees). (p. 102)

Table 6.1 Ranking of Creep Resistance of Solder Alloys Based upon Best_Fit PQWCI Law
Creep. (p. 161)

Table 6.2 Comparison of Creep Properties of SOlder Joints Made with Eutectic 59-3.51" g,
Sn-4Ag-0.5Cu, Sn-2Ag-1Cu—1Ni, and Sn-Ag-0.5Ni Solders. (p. 167)

LIST OF FIGURES

Figure 2.1 The model shape of a typical creep curve. (p. 21)

Figure 2.2 Schematic diagram of (a) Nabarro-Herring creep and (b) Coble CI'CCP on an
ideal grain. (p. 25)

Figure 2.3 A schematic illustration of accommodating diffusional creep by grain
boundary sliding. (a) Four garains in a hexagonal array before creep
deformation. (b) After deforming by diffusional creep, one dimension of the
grain is increased and the other is decreased, and voids are formed between the
grains. (c)The voids are removed by grain-boundary sliding. The extent of
sliding displacement is quantified by the distance Y’Y”, which is the offset
along the boundary between grains 1 and 3 of the original vertical scribe \‘me
XYZ. (p. 26)

Figure 2.4 A schematic deformation mechanism map. The axes of the diagram are
homologous temperature (T IT m) and stress ( in terms of the Shear modulus)-
The stress-temperature combinatlon determines the primary defom atio“ m 6'
At the boundary lines, the defol'flfation is due equally to two different
mechanisms, and, at the interseCUOn of the lines, to three different mechanisms

(1). 30)

Figure 3- 1 Dimensions of shear-lap solder joint, COpper substrate and alummum fixture.
(p. 43)

Figure 3.2 Temperature profile of single sheaf lap solder joint fabrication. (p. 43)

Figure 3. 3 Comparative dimension of a Si ngle shear lap solder joint used in the
investigation. (P- 44)

Figure 3 .4 Microstructure of as fabricated Cu particle reinforced solder joint - (a) Overall
view of the microstructure in the joint. (b) Cu reinforcement particle and the
IM layer formed around Cu, (O) N layer formed at the Cu substrate/solder

interface. (P- 45)

Figure 3.5 mcmstructure of as fabricated Ag particle reinforced soldeI'JOint- (3) Overall
View of the microstructure in the joint, (b) Ag reinforcement Particle and the
INI layer formed around Ag, (c) [M layer formed at the Cu SUbStrate/solder

interface. (p. 47)

Figure 3.6 Microstructure of as fabricated Ni particle reinforced solder joint. (a) Overall
view of the microstructure in the joint, (1)) Ni reinforcement particle and the

IM layer formed around Ni, (c) IM layer formed at the Cu substrate/solder

interface. (p. 49)

Figure 3.7 The temperature-time reflow profile with four different cooling rates (heating
rates remained the same for all reflow conditions). (1) Cooling conditions
are in water, (2) on aluminum chill block, (3) on fire brick, (4) and on wood.

(p. 53)

Figure 3.8 Temperature-time profiles of the fastest and the slowest cooling rate.
Sunflower shape of INIC prevails around Ni particles after (a) the fasted

cooling rate, and 0)) the slowest cooling rate. The cooling rate apparently
has no significant effect on IMC morphology. (p. 54)

Figure 3.9 The temperature-time profile with four different heating rates (cooling Yams
remained the same for all reflow conditions). A5, A6, and A7 indie ate areas

underneath the heating part of these profiles at temperatures above the melting
point of the solder for heating rates 5, 6, and 7. (p. 55)

Figure 3.10 Temperature-time profiles assmiawd with micrographs of Cu—Ni-S n
intermetallic are shown (a) Sunflower shape for heating profiles of 5 and 8.
(b) mixed sunflower shape and Sinisle-Crystal-faceted morphology occurred
with the medium heating rate (6), and (C) Single-crystal-faceted morphology

accompanied with the slowest heating rate (7). (p. 56)

Figure 3- 1 1 Effect of reflow at 280°C With fastest heating rate on the growth of MC
layers around Ni particles in Ni composite solder joint. (a) first reflow, (b)
second reflow, (c) third reflow and (d) fourth reflow. A5 is the heat input area

above melting temperature to peak temperature. 4A5 and A7 were measured,
The arrows indicate the miCro structure associated with each reflow. (p. 58)

Frgure 3- 1 2 Heat input area comparison Of A 5 and A6 (A65 3A5). (p. 59)

Figure 3- 1 3 Effect of reflow at a peak temperature 0f 250°C With the slowest heating rate
on the gIOWth of IMC layers around Ni particles. The heat inpUt of one
reflow is equivalent to heat input with the reflow that undergone fast heating

rate (A5==A9). The heat input for the four reflows in temperature region
above the melting point is almost the same as that of the slowest heating rate

xii

(A724A9). Arrows indicate IMC morPhOIOgy observed. (p. 60)

Figure 3.14 Heat input approximately equivalent to 3A5 obtained with a hold of 15
seconds at peak temperature-time profile and the resultant
single-crystal-faceted morphology. (p. 61 )

Figure 3.15 Effect of low peak temperature and long hold times at this temperature on the
IMC morphology. These two temperature profiles represent heat i“put
approximately equivalent to two reflows with peak temperature 0f 250°C
(2A9=A10=Al l). Sunflower INIC morphology was observed after both
these conditions. (p. 63)

Figure 4.1 Effect of isothermal aging at 150°C on the growth of Cu-Sn intermetallic layer

around Cu particles in Cu reinforced composite solder joint: (a) as-fabricated’
(b) after 100 hours, (c) after 500 hours, ((1) after 1000 hours. (p. 73)

Figure 4.2 Effect of isothermal aging on the growth of Cu-Sn intermetallic layer aro‘md
the Cu particles and Cu particle consumption. (p. 75)

Figure 4.3 Effect of isothermal aging at 150°C on the growth of Cu-Sn intermetallic We:

at Cu substrate-solder interface in Cu reinforced composite solder joint: (3)

as-fabricated, layer thickness= 1-81 um, (b) after 100 hours, (c) after 500
hours, ((1) after 1000 hours, layer thickness =7.76p.m. The development of

Kirkendall voids is clearly evident between the Cu substrate/[M layer after
long aging times in (c) and (d)- (P- 75)

Figure 4-4 Proposed mechanism for Cu-Sn intermetallic layer grOWth during aging for a
Cu/Sn solder diffusion couple. (p. 79)

Figure 4, 5 Effect of isothermal aging at 1 50°C on the growth of Ag38n intermetallic layer

around Ag particles in Ag reinforced composite solder joint: (a) as-fabricated,
(b) after 100 hours, (0) after 500 hours, ((1) after 1000 hours. (p. 8 1)

Figure 4.6 Effect of isothermal aging at l 50°C on the growth of Cu-Sn interrnetallic layer
at Cu substrate-solder interface in Ag reinforced composite solder joint: (a)

as-fabricated, layer thickness: 1. 59 pm, (b) after 100 hours, (0) after 500
hours, ((1) after 1000 hours, layer thickness =6.95um. (p. 82)

Figure 4-7 Effect of isothermal aging on the intermetallic layer growth at the copper
substrate/solder interface. (p. 83)

xiii

Figure 4.8 Effect of isothermal aging at room temperature(25°C) on the growth or
Cu-Ni-Sn intermetallic layer around Ni Particles in Ni reinforced composite
solder joint. (a) as fabricated, (b) after 100 hours, (c) after 500 hours, (d) after
1000 hours. Note: No growth of the N layer with aging time at 25°C. (p. 85)

-Sn intermetallic

joint. (a) as

1000 hOUfS. Note:

at 50°C on the growth of Cu-Ni
Ni reinforced composite solder
00 hours, ((1) after
50°C. (p. 86)

Figure 4.9 Effect of isothermal aging
layer around Ni particles in
fabricated, (b) after 100 hourS, (c) after 5

No growth of the IM layer with aging time at
100°C on the growth of Cu-Ni-Sn intermetallic

layer around Ni particles in Ni reinforced composite solder joint. (a) as
fabricated, (b) after 100 hours, (c) after 500 hours, ((1) after 1000 hours. Note:

th of the 11V! layer with aging time at 100°C. (p, 87)

Figure 4.10 Effect of isothermal aging at

Insignificant grow
rmal aging at 150°C on the growth of Cu-Ni-sn intermetemc

layer around Ni particles in Ni reinforced composite solder joint. (a) as ‘6.
fabricated, (b) after 100 hours, (0) after 500 hours, (d) after 1000 how’s’fio '

Profuse growth of the IM layer with aging time at 150°C. (p. 88)

Figure 4.11 Effect of isothe

at 1 50°C on the growth of Cu-Ni-Sn intermemmc
layer around one Ni particle. (a) as fabricated, (b) after 100 hours , (c) after

500 hours, (d) after 1000 hours- Note: profuse [Ni layer growth and
consumption of Ni reinforcement particle. (p, 90)

Figure 4.12 Effect of isothermal aging

Figure 4-13 Effect of isothermal aging at 150°C 0“ the growrh of IM layers around the Ni
particles and Ni particle consumption. Data for Cu composite solder and Ni
composite solder aged at 100°C, 50°C, and room temperature are plotted for
comparison. (a) 1le layer evolution around the particle reinforcement, (b)
reinforcement partic 1c consumption. (p. 91)

0C on the growth of IM layer at the Cu

ed, layer thickness=4 pm, (b) after

Figure 4- 14 Effect of isothermal aging at 150
layer thickness=56 um.

substrate/solder interface. (a) as fabricat
100 hours, (0) after 500 hours, (d) after 1000 hours.
(p. 93)

0°C on the growth of IM layers at the Cu
substrate/ solder interface in the Ni composite solder joint. (a) Comparison of
interfacial [M layer growth in Ni, Cu and Ag comp0site solder joint under

aging at 150°C Interfacial N layer growth in the Ni composite solder joint
aged at 100°C and less is plotted as well for comparison. (b) Comparison of

Figure 4. 1 5 Effect of isothermal aging at 15

xiv

 

 

 

   

Cussn [Ni layer growth at the Cu substrsate/ solder interface in Ni and Cu
composite solder joint under aging at 150°C. (p. 94)

Figure 5. 1 Temperature profile of solder/COPE)er Wettability experiment. (p. 100)

Figure 5.2 Schematic drawing of wetting properties of different solder materials observed
in the reflow experiment. (p. 103)

Figure 5.3 Changes of wetting angles with the volume fraction of the copper particles in
the Cu particle reinforced composite solders. (p. 105)

Figure 5.4 Effect of reflow on the growth of Cu-Sn intermetallics around Cu particle
reinforcements: (a) first reflow, (b) second reflow, (c) third reflow, (d) fourth
reflow. (p. 108)

Figure 5.5 Effect of reflow on the growth of Ag3Sn intermetallics around Ag particle
reinforcements: (a) first reflow, (b) second reflow, (c) third reflow. (d) fourth
reflow. (p, 109)

Figure 5.6 Effect of reflow on the Cu-Sn intermetallic layer growth around Cu particles
and Cu particle consumption. (p. 110)

Figure 5.7 Effect of reflow on the growth of Cu-Ni-Sn M layers around the Ni particles
in the Ni composite solder joint. (a) first reflow, (b) second reflow, (c) third
reflow, (d) fourth reflow. (p. 111)

Figure 5.8 Influence of reflow on the growth of Cu-Sn IM layer at Cu substrate/solder
int(erface in Cu reinforced composite solder: (a) first reflow, layer thickness:
1-85 um, (b) second reflow, (c) third reflow, (d) fourth reflow, layer thickneSS
=4- 95pm. (p. 113)

Figure 5-9 Influence of reflow on the growth of CD'S" N layer at Cu substrate/solder
interface in Ag reinforced composite 801C161”! (a) first reflow, layer thickness:
1.2 3 11m, (b) second reflow, (c) third reflow, (d) fourth reflow, layer thickness
=2-78pm. (p. 114)

F“
lgure 5.10 Comparison of the effect of reflow on the IM layer growth at Cu
Substrate/solder interface in Cu and Ag composite solders. The layer
thickness was divided by its initial thickness for comparison. (p. 115)

Fi
Elite 511 Effect of reflow on the growth of 1M layers at the Cu substrate / solder

interface in the Ni composite solder joint. (a) first reflow, (b) second reflow,
(c) third reflow, (d) fourth reflow. (p. 116)

Figure 5.12 Effect of reflow on the microstructure of eutectic Sn-3.5Ag non-composite
solder joint: (a) first reflow, (b) second reflow, (c) third reflow, (d) fourth
reflow. (p. 118)

Figure 5.13 Effect of reflow on the growth of Cu-Sn intermetallic layer at Cu
substrate-solder interface in eutectic Sn-3.5Ag non-composite solder joint:
(a) non-reflow, layer thickness = 1.74 pm, (b) first reflow, (c) second reflow,
((1) third reflow, layer thickness =4.27p.m. (p. 119)

Figure 5.14 Comparison of interfacial intermetallic layer growth due to reflow between
Cu composite solder, Ag composite solder, Ni composite solder and eutectic
Sn-3.5Ag non-composite solder joint. (p. 120)

Figure 5.15 Change in hardness of eutectic Sn-3.5Ag solder as a function of reflow.
(p. 121)

Figure 5.16 Change in yield strength of eutectic Sn-3.5Ag solder as a function of reflow.
Yield strength decreases with multiple reflows. (p. 122)

Figure 5.17 Microstructure of eutectic Sn-3.5Ag solder around indent: (a) non-reflow
solder; (b) three reflowed solder. The size of the Sn cells is larger for
reflowed materials, correspondingly, the yield strength is lower. (p. 124)

Figure 5.18 The effect of reflow on steady-state creep strain rate of eutectic Sn-3.5Ag
solder. The stress exponent is slightly higher for multiple reflowed solder.
(p. 125)

Figure 6.1 Laser ablation patterns imprinted on solder joints used in creep testing. (a)
before creep; and (b) after creep. (p. 132)

Figure 6.2 Traditional dead load creep testing frame with an electrical resistance furnace.
(p. 132)

Figure 6.3 New design of the dead weight loading miniature creep testing frame. (p. 134)
Figure 6.4 (a) Localized and global creep strain vs. time for the non-composite eutectic

Sn-3.5Ag solder joint. (b) 3-D plot of the creep strain rate vs. time and
position across the solder joint. (p. 136)

xvi

 

 

Figure 6.5 (a) Localized and global creep strain vs. time for the non-composite
Sn-4.0Ag-0.5Cu solder joint. (b) 3D plot of creep strain rate vs. time and
position across the solder joint. (p. 138)

Rigure 6.6 (a) Localized and global creep strain vs. time for the Cu particle reinforced
composite solder joint. (b) 3-D plot of strain rate vs. time and position across
the solder joint. (p. 140)

Figure 6.7 (a) Localized and global creep strain vs. time for the Ag particle reinforced
composite solder joint. (b) 3-D plot of creep strain rate vs. time and position
across the solder joint. (p. 142)

Figure 6.8 The secondary creep rate for composite and non-composite solder joints as a
function test temperatures. The Ni particle reinforced composite solder joint
exhibited the best creep resistance at all test temperatures. (p. 144)

Figure 6.9 Onset of tertiary creep for different solder joint materials as a function of
secondary creep rate. Composite solders made by mechanical mixing
method generally showed a lower strain for the onset of tertiary creep than
non-composite solders and composite solder made by in-situ method. (p. 146)

Figure 6.10 Micrographs showing creep deformation of an in-situ Cu6Sn5 reinforced
composite solder joint. (a) solder joint before creep, where there is no
damage (interfacial debonding, voids, cracks, etc.), (b) solder joint after creep,
showing multiple damage sites ( particle rotation, particle/matrix interfacial
debonding, voids formation, etc.) across the thickness of the solder joints, as
indicated by arrows. A multitude of deformation sites would tend to promote
homogenization of the deformation over the joint, as shown in (b). (p. 148)

Figure 6.11 Plot of log strain-rate vs. inverse absolute temperature. The activation
energy for creep is listed in the legend. The activation energy for creep for Ni
particle reinforced composite solder is higher than all other solders listed.

(p. 150)

Figure 6.12 The deformation profile for (a) non-composite solder joint before creep; (b)
non-composite solder joint after creep; (c) Cu particle reinforced composite
solder joint before creep; and (d) Cu particle reinforced composite solder

joint after creep. (p. 151)

Figure 6.13 Example of uniform creep deformation in Ag particle reinforced composite

xvii

 

solder joint: (a) before creep; (b) after creep. (p. 152)

Figure 6.14 SEM micrographs showing non-uniform deformation in the crept Ni
composite solder joint as exhibited by the distortion of the laser ablation
patterns. (a) whole joint, before creep, (b) whole joint, fracture due to creep,
(c) one part of the joint, before creep, ((1) one part of the joint, after creep.
(p. 154)

Figure 6.15 Microstructures of solder joint materials: (a) Eutectic Sn-3.5Ag solder alloy,
(b) Sn-4Ag-0.5Cu solder alloy, (c) Sn-3.5Ag—0.5Ni solder alloy (inset
indicating Cu-Ni-Sn intermetallics), and (d) Sn-2Ag-1Cu-1Ni solder alloy.
(p. 158)

Figure 6.16 Comparison of steady-state creep rates of the four solder alloys. (a) room
temperature data, (b) 85°C data. (p. 159)

Figure 6.17 Normalized steady-state creep strain rates versus normalized stresses for
eutectic Sn-3.5Ag, Sn-4Ag-O.5Cu, Sn-2Ag-lCu-1Ni, and Sn-3.5Ag-0.5Ni
solder joints along with Darveaux’s data for aged eutectic Sn-3.5Ag solder
joints. (p. 163)

Figure 6.18 (a) Average strains for the onset of tertiary creep at room temperature, (b)
average strains for the onset of tertiary creep at 85°C, (c) time for the onset of
tertiary creep at room temperature, (b) time for the onset of tertiary creep at
85°C. (p. 166)

Figure 6.19 Creep deformation in the solder joints: (a) Eutectic Sn-3.5Ag solder, (b)
Sn-4Ag-0.5Cu solder, (c) Sn-Ag-0.5Ni solder, (d) Sn-2Ag—1Cu-1Ni solder,

and (e) a magnified view of the deformation in a Sn-Ag-0.5Ni solder
joint.(p.l69)

Figure A1 Example of selection of creep images. Totally 15 out of 38 images were
selected from a creep testing on in-situ Cu68n5 particle-reinforced eutectic

Sn-3.5Ag solder joints. These 15 images represent creep deformation in the
solder joint at different time intervals. (p. 185)

Figure A2 Open the selected images in PhotoShop®. (p. 186)
Figure A3 The last image is the first one to edit. (p. 187)

Figure A4 Pick a line that is clear throughout to trace. (p. 187)

xviii

 

Figure A5 Invert the original image to edit. (p. 188)
Figure A6 Select the line to trace. (p. 189)
Figure A7 The image is cropped to the line region. (p. 189)

Figure A8 Carefully trace the line. Use 200% magnification when tracing the line.(p.190)

Figure A9 Canvas size change. (a) select “canvas size” in “image” menu. (b) Change the
width and height in the box. (c) A white margin was formed due to canvas size

change. (p. 191)

Figure A10 Save the edited image to “.git” file type for use in DataThief® PC version.
(p. 192)

Figure All A transparency with grid is taped on the computer screen. A coordinate
system is set up on the transparency. (p. 193)

Figure A12 Launch the DataThief® software. Don’t maximize the DataThief® window.
(p. 194)

Figure A13 Choose the reference point from the image and mark on the transparency. (p.
195)

Figure A14 Set up the coordinate system in DataThiefD. (p. 195)

Figure A15 Illustration of data extraction. (a) Scattered data is collected. (b) Data is saved
in “.txt” format. (p. 196)

Figure A16 Each image should match the reference point on the transparency. (p. 197)
Figure A17 Irnport the extracted data from DataThief® to Microsoft Excel®. (p. 198)

Figure A18 Normalized the y position with respect to the initial y position before creep.
(p. 199)

Figure A19 Data used to generate displacement vs. unnonnalized position curve. (p. 200)

Figure A20 A sample displacement-position curve plotted in KaleidaGraph®. (p. 200)

xix

 

incl)"
at

 

Figure A21 Example showing how global strain data were obtained. (p. 20 1)
Figure A22 Illustration of how to prepare data for global strain-time plot. (p. 202)

Figure A23 A sample plot of global creep strain vs. time using the data from Figure A22.
(p. 202)

Figure A24 Steady-state creep rate obtained using linear regression with the data from the
secondary stage in the global creep strain-time plot. (p. 203)

Figure A25 Copy the raw data and paste it to row l-row 20 in a new spreadsheet for
normalized displacement-normalized position plot. (p. 204)

Figure A26 Illustration of normalizing x values. (p. 204)
Fligure A27 Make the first data in the last y column zeroed. (p. 206)

Figure A28 Normalize all the y position with respect to the initial y position in the first y
column. (p. 206)

Figure A29 Illustration of how to make all the data greater than 0. (p. 207)

Figure A30 A sample plot of normalized displacement vs. normalized position through
joint thickness curve. (p. 207)

Figure A31 Curve fit all the displacement lines with 3rd order polynomial. (p. 208)

Figure A32 Record the coefficients M1, M2, and M3 form Figure A31 for all the
displacement curves (M0 is not needed) and fill in a new spreadsheet. (p. 208)

Figure A33 Illustration of how to obtain localized creep strain. (p. 209)

Figure A34 Manipulation of data prepared for plotting localized creep strain with time.
(p. 210)

Figure A35 Variation of localized creep strain with time. (p. 210)
Figure A36 Plot of localized and global strain with time. (p. 211)
Figure A37 Curve fit the “localized creep strain vs. time” curves in KaleidaGraph® with

3rd order polynomial. (p. 212)

XX

 

 

 

Figure A38 Record the coefficients M 1. M2, and M3 for all the localized strain Curves (M0
is not needed) and fill in the spreadsheet. (p. 212)

Figure A39 Illustration of how to obtain localized creep strain rates. (p. 213)

Figure A40 Manipulation of data prepared for localized strain rate vs. time/position plot.
(p. 214)

Figure A41 A sample plot of localized creep strain rate vs. time. (p. 214)

Figure A42 A sample plot of localized creep strain rate vs. position through solder joint
thickness. (p. 215)

I“‘igure A43 Reorganization of the data in Excel® prepared for use in DeltaGraph®.
(p.215)

Figure A44 Illustration of data directly copied from Microsoft Excel®. (p. 216)

Figure A45 Illustration of how to generate three-dimensional graph in DeltaGraph®.
(p. 217)

Figure A46 Variation of localized strain rate with normalized position and time. (p. 217)

Figure A47 Variation of localized creep strain with normalized position and time. (p. 218)

xxi

 

CHAPTER I
INTRODUCTION

Lead-free solders have been the main focus of significant research activity for the
Past decade due to the concerns of toxicity and health hazard of lead present in the
commonly used lead-bearing solders. Under this global trend toward a lead-free
environment, the acceptable lead-free solders should function as a complete altemative to
traditional lead-bearing solders in terms of electrical interconnectability, structural
integrity, and reliability. Although government legislation and regulations have been
driving force to eliminate the use of leaded solders and there are lead-free solder
cv'ilndidates already in the market [1,2], still no seemingly outstanding substitute has
emerged to fully replace the eutectic and near eutectic Sn—Pb solders because of the
stringent demands for excellent structural performance of these alternative solders for use

in current industries for electronic packaging and assembly.

Foremost among all the lead-free solder candidates, eutectic Sn-3.5Ag solder has
received attention worldwide as a potential substitute because of its non-toxic nature as
well as its comparable wetting and mechanical properties to eutectic Sn-37Pb solder [3-6].
Its higher melting point, 221°C, makes it more suitable for higher temperature
applications [5]. Use of such lead-free solders in electronic industries is relatively new
and the knowledge base of alternative solders is not well established due to the long-term
dominance of leaded solders over centuries. An extensive property database is not yet
available and some questions still remain concerning data that currently exist. Therefore,
significant research activities have been involved in an effort to investigate the properties

of eutectic Sn-3.5Ag solder and meanwhile to improve its comprehensive properties.

 

Solders used in various applications, such as automobile under-the-hOOd, aerospace
and defense, microelectronics, power generation and distribution etc., are SUbjected to
Severe service environments including multiple loading scenarios over a range of

Operating temperatures [5]. Coefficient of thermal expansion (CTE) mismatch between
substrate, solder and chip leads, environmental thermal fluctuation (as much as —40°C to
160°C), and the complex loading conditions (such as creep and mechanical fatigue) are
the main contributing factors for solder joint failure, which is considered as the primary
reason for the failure of electronic components [7,8]. These low melting point solder
alloys are typically used at temperatures well above half of their absolute melting points,
80 recrystallization, superplasticity, creep/relaxation, and creep-fatigue are operative
under normal service conditions [9,10]. Solder materials functioning at such high
homologous temperatures were also found to undergo extensive microstructural evolution,
primarily including interfacial intermatallic layer growth and microconstituent phase
coarsening, which will eventually degrade the solder joint [4, 11-14].

Two major avenues are being pursued to mitigate the detrimental effects already
alluded to. Several studies involve the addition of such alloying elements as copper,
nickel, antimony, bismuth, etc. in order to reduce the melting points of eutectic Sn-3.5Ag
solder as well as to improve its mechanical properties [15-21]. Another way to achieve
improvement is to introduce a dispersion of second phase reinforcements into the solder
matrix to stabilize the microstructure and thereby increase its mechanical strength [22-32].
This approach, the so-called composite approach, was designed for the purpose of

improving the solder’s intrinsic vulnerability to cyclic and creep deformation.

 

 

In this investigation, composite solders are made primarily by incorporating

micron-sized Cu, Ag or Ni particulate reinforcements into the eutectic Sn-3,5 Ag sol der-

 

matfix, The focus of this study is to charaCterize and investigate how different types of

reinforcements would alter microstructure, aging» wetting, and reflow properties of the

CUtCCtiC Sn-3.5Ag solder matrix. QuantitatiVe Understanding the Preliminary creep
behavior of these composite solders is anOthe'r major Part Of this inVeStigation, Global
and localized creep deformation as We“ as the onset 0f tema’)’ 076613 of each comp0site
solder material were quantified and analyzed using small realistic Sized solder 30““
specimens typically used in microelectronics. In addition to the composite apPYOaCh

. _ . 14")
employed primarily in the study. the effects 0f alloying Clement additions (Q11 and!“

0“ the creep behavior of eutectic Sn-3.5Ag solder were investigated,

 

 

CHAPTER II

LITERATURE REVIEW

2.1 Current Research Status on Lead-free Solders
2.1.1 Introduction

Several challenges are faced in the development of lead-free
solders Since the
Y are

not just drop-in substitutes for traditionally used leaded SOIdCrs. These chaue be
nges may

related to the solder melt temperature, P‘oce’ ssrng t‘flnl’erature, wettability ehanical
. me

and merino-mechanical fatigue behaviorS etc. Knowledge base on leaded SQIdCfS gained

by experience over a long period time is not directly applicable to lead-free S iders' AS a
0

. - - . . 0‘
result, a database for modeling for reliabrlrty predictions of lead-free s 1d61'5 is o .0
91
currently available [33]. Most of the lead-free solder developments er 6160‘“,
. . 10‘!
applications are aimed at arriving at suitable alloy composrtrons [3-6] - Sty—Ag 3‘

system, with or without small alloy additions such as Cu, is believed to have significant
potential [345,151. The binary Sn-Ag eutectic temperature is 221°C: and the ternary Sn-
Ag-Cu eutectic temperature is 217°C, both being reasonably higher than the SIP—Pb bina'y
entectic temperature of 183°C. Although the processing parameters have to be modified
to mommate this increase in eutectic temperature, use of such soldel‘s also Provide
higher service temperature capability to the solder joints. Results Of se Vera! studies On
such solders “.3 recently reported in the published literature [3’6’15 1’
Although these approaches on high temperature lead'f’ee some“ tend to eVaIuate
the alloy systems with a melting temperature of about 220°C, significant increase in the

. , f the t I‘m
Sew ‘90 temperature capability may not result as a consequence 0 he O‘mechanical

 

 

 

5U

behavior of such systems. Current Nation?ll Center for Manufacturing Systems (NCMS )

project deals with lead-free solders for applications with a service temperature of 150°C

[34]. For example in the automotive under-the-hood applications,

there is significant
interest from designers to mount the electronic circuit boards on the '
engine manifold.

This W111 significantly decrease t t i . '8’ and minimize a1
SCVCI‘
complications in the electrical CifCUltl y- 81.111 ] :"dfilOliS also eXiSt in aer e and
ospac

defense applications.

A more severe environment experienced by the solder joint is in a high“
current/high temperature application such as 1 11 an automotive alternator 0r cfifier. 1’“

present there is no suitable and economical SUbstitute for the high lead 8016613 used {or

such applications- Solders based on Sn-A“ alloys are cost prohibitive -

-se
1 ‘1 131' g6
be 606
automotive type manufacturing situations [35]. It is believed that this ma will
fl
focus of significant number of investigatlons 1n the near future. Since ragular elec

Solders, such as in electronic compOnents or computers. (10 “0t cxPefiehCe SUCh sever 6
environments, this aspect of lead-free solders has not been addressed so far _
Incorporation of dispersoids to improve the mechanical and themomecmnicu
behavior of the solders, even in leaded-solder systems, has been an aven ue that has been
Pursued to improve the service temperature capability without Significantly altemaung
the processing parameters- Since there have been several studies dealing with diSPeI'soidS

in leaded soldgrs a brief review illustrating the knowledge gained from them Will be

PFOVided prior to addressing the lead-free solders.

H

 

2.1.2 Service Requirements

Solders in general Operate at high homologous temperature ranges During turn‘
' mg

on and off operations of the electrical circuitry when heat up or cool down th 1
’ ey a SO
experience low cycle thermo-mechanical fatigue due to stresses

that develop as a
consequence of CTE mismatch between the SOlder/substrate/componems

Mechanical
vibration of other entities to which the electromc components are mechanicau tta bed
y a c ’

such as automotive engines, can create higher frequency Vibrational fatigue conditions'
When an automobile/tank hits a pot-hole/malor obstruction. or landing of an airplane ea“

impose impact loading on the solder joint- Although fine-gl‘ained microstr. 1 Quite may be
beneficial for mechanical fatigue considerations, it may not be ideal for Cr 9 1.esist'a“°¢

since creep deformation at the service temperature (high homologous ternl)e

. ‘é
. . . . 9‘“
solder alloys) will be by grain boundary sliding. In addition to thQse 099°

ees
requirements, the highly in-homogenous as-joined solder joint tnicrostrue tare 603‘s
during service, This aging process causes growth of some” substrate interface
intel‘metallic layer and coarsening of microstructural constituents within the 801‘“?r joint.
Such evolving merostructure continuously alters the mechanical proper-ti es of the solder
joint resulting in significant hurdles in reliability prediction modeling. Presence Of fine,
Stable, compatible dispersoids at the grain boundaries can retard cowefliflg,

e”hilIICe
mechanical fatigue behavior, decrease creep rate by decreasing gram boundary Sliding

tendency by k§ 3mg the grain boundaries.
Although one would prefer a strong solder joint for creep ”S‘smce <301181'de1‘atiomg,
. - til ' ‘ .
It may “0t be ideal in electronic applications. If the solder m e Jo‘nt 18 not able to

. . nent ~
dlSsipate the Stresses that develop. failure of the electronic compo s “’1" result. One

 

wi-

would prefer a reasonably strong and Pliable solder joint. Although these two
exclusive, both of them can be satisfied by
appropriate microstructural engineering of solders with SUitable dispersoids

requirements appear to be mutually

The method010gy that will be employed to enhance the service Performance of the
solder joint should not affect the well laid 00‘ cuTrent metallurgical processes in the
electronic manufacturing. It is essential that the

significantly alter the solderability,

temperature, etc.

. - - , “e

cause significant decrease in solder reliability could probably be contrgned by fi
. . . 05‘
dispersoids in the grain boundaries. Fine dISPel‘Slon of copper atoms In a1 uwnufi‘ ca
significant decrease in gleam-migration in computer circuitry [36, 37].

\
However, 5
' - orma

aSpects will not be addressed in this review due to a lack of available 1nf

relevant to Sn in the literature.

2.13 Prior smdm [38-40]

There h been significant worldwide active research aetiVltles of f“
ave
soldering durin the past decade Examples of such research and devel Opment pro} t5
8 ° ec

can be listed in Table 2.1.

. . 'ed ..
In US a. sortium of 11 industrial corporations In US cam on! the lead-free
9 Con
. ' es NC .
solder PIOJCCt” led by National Center for Manufacturing Selene ( 1‘43) in order to
, emerged f
evaluate alter-n atives to eutectic Sn-Pb solder. Five alloy3 to“) the down-

Seleetion as Vi able candidates to replace 80% of the currently "tectic Sn-Pb

solder. It was found during the NCMS project that Sn-58Bi eutectic, Sne3, 4 Ag 4.8Bi and
Sn.3,5 Ag eutectic solders performed substantially better than the eutecuc Sn

'Pb Solder in
certain surface mount applications. The project also showed that Sn-SgBi and Sn 3 4A
_ . g-
4.8Bi had fatigue lives comparable to or better than eutectic 811-131,

, possessing similar
bulk properties at room temperature and very high tensile and yield strengths combined

with moderate to high elongation. However. the NCMS concluded that a

“drop-ins,
replacement for eutectic Sn-Pb solder was St)" “Qt identified at the end Of the “1992’

1996” four-year project.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.1 Examples of deeMrojxt in the World [38'4“l
Country wization Project N e
NCMS (AT&T/Lucent am
Technologies, Ford Motor . L
US Company, GM-Hughes Lead‘fm" Sol Clef PM“;
Aircraft, NIST, etc.) \/
“'13“ part funded project
(GEC, BNR Europe, Lead-free ring
UK and Europe \ Multicore Solders and ITRI Solde
Ltd.)

 

 

 

N
JIPE Project ( Senju Metals,

Alpha Metals Japan, Nihon Lead-fr IE Soldering
Handa, Ishikawa Metals,

etc.)

Japan

 

 

 

 

 

’

/

w/A

The DH funded “lead-free soldering” project in UK was focused on the Swtabil,

for electronic assembly and supply potential during their lead-free alloy 851503.01” FiVi
solder alloys. namely Bi-428n, Sn-9Zn, Sn-SSb, Sn-3.5Ag, and Sn-O.7Cu’ were tested
and Bi-423n. S n-9Zn, Sn-SSb solders were mjected due ‘0 different reast)?” such as poor

meclianical Properties, corrosron of zinc phase, 01' melting “3 pe mg too h18h.

amon
Sn'3-5Ag and Sn-O.7Cu solder alloys showed better Performance g the fi
allOys.

ve selected

The JIPE project in Japan focused on the Sn-Ag—Bi (Bi 745%

) alloy. It was found
that any solder alloy containing more than 7% bismllth were Very brittle and fillet liftin
g

became a serious concern even though the melting temperature of the

solder alloy is
relatively low. The Sn-Ag—Cu alloys with or without a few percent of bismuth proved to

be useful. It was also discovered that Sn—Zn and Sn

-C
“ “1103’s have a great potential for
use as solder alloys. Studies dealing With these alloys are currently in progress

As a result of recent research activities, a large number of lead-free solders have
been developed and number of patent applications have been filed for V arious alloy
compositions. Although not all these alloys are commercially available, there is still a

wide range to choose from. The most convenient way to separate the avail able toad.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. , tee
alloys is to consider their melting temperatures. Some typical examples of leao f g
6‘“
solders under research and development are listed in Table 2.2 along with their me‘
temperature ranges.
Table 2.2 EXamples of Lead-free Solder Alloys Currently under meh and
Development [38-40] 1 . (OC)
Category Alloy System Composition (wt%) Me ting Range
Low meltin temp. Sn-Bi Sn-SSBi # 131:
(<1 80°C) Sn-In Sn-SgZZIn 1 :3 5
Melting Tem (183- Sn-Zn Sn- 0 . #_ -
200°C)cquiv1;1ent to Sn-Bi-Zn Sn-SZn-3Bi : 8 9-199
euiCCtiC Sn‘Pb solder Sn-Bi—In Sfl-ZOBl-IOID 43-193
Mid Sn-Ag Sn-3.5Ag #4 221
to -range melting Sn-Cu Sn-O.7Cu 227
mp. (200~2 30°C) Sn-Ag-Cu Sn-3.8Ag-0.7Cu 21 7
High melong temp. Sn-Sb Sn-SSb %
(Bo-350°C) Sit-Au Sn-80Au

'lable,
In sum ary, there are several lead-free alloys that are aval hoWever there is

. far.
no universal drop-in replacement for leaded solders identlf16d so At the moment the

 

WEE

Salt

\jt

 

most promising alloys for general electronic soldering appears to be those based on 311..

3 5Ag and Sn-Ag-Cu. Other alloys with potential are Sn-0.7Cu and Sn-Ag—Bi. The need

for higher process temperatures is another important technological issue that has to be
addressed when companies change over to lead-free soldering because higher process

temperatures will impact existing solden'ng teChnology in such key areas as materials

stability/reliability. equipment reliability and h‘ghe’ energy cost issues. No applicable

lead-free solder alloys can be used at high SC] Ice temperature except eXPensiVe Sn-SOA"
solder which will not be practical in large scale Soldering ope ration. How “3 increase me
service temperature of existing lead-free ”Idem for sum applications 01‘ alfive 3‘ “6

a"6
solder compositions is one of the serious problems that lead-free 801d“ r 38% 310‘th h

yet to address.

2.2 Composite Solders
ce the Behavior of Solders

2.2.1 Composite Approach and Its Merit to Enhan
' 11

2.2.1.1 Pam 0f Composlte Approac
rove the serv ice pedomanCC

Composite approach was developed mainly to lmp f
. 0 .
In other words, the has“: purpose this

including service temperature capabilitY-
methodology is to engineer and stabilize a fine grained microStrUCture' homogenize
solder joint deformation, so as to improve the mechanical properties of the solder jOint,
especially cl‘aep and thermomechanical fatigue ICSiStancc. . A1 80’ the added
1' matrix, but may effectively

reinforcements do not change the melting point of the solde
- improvi
increase the Sc ice temperature of the base solder materials by 11g the creep or
IV

thermomecbmi cal fatigue properties of the solder matrix.

 

10

 

2.2.1.2 Prior Studies of Composite Solders \

Several efforts have been made to improve the comprehensive properties of lead.
bearing solders using composite approach [22-32]. Microstructural analysis as well as
mechanical testing of such composite solders have been reported. Certain composite
solders did show improved mechanical properties sought by electronic/automobile
industries.

Marshall et al. have carried out studies in microcharacterization of composite

solders [22-26]. Their composite solders were primarily prepared by mixing Cu68n5 (10,
20, 30wt%), CU3Sn (10,20, 30wt%), Cu (7.6wt%), Ag (4wt%), or Ni (4 wt%) particles
with the eutectic Sn-37Pb solder paste. The microstructure features of these bulk
composite solder specimens showed Cu-Sn, Ag-Sn, and Ni-Sn intermetallics were
developed in the composite solders around Cu, Ag, and Ni particles respectively. Cu68n5
layer formed around Cu3Sn particles in the Cugsn reinforced composite solder, while no
more new intermetallic formed in the Cu68n5 particle reinforced composite solder. The
microstructural analysis showed good bonding of the particulate reinforcements to the
solder matrix suggesting that the resulting composite solders might exhibit enhanced
Strength.

Intermetallic formation at the solder/copper interface was studied for the above
composite solder samples aged at 140°C for 0 to 16 days, as reported by Pinizzotto et
al, [41]. Intermetallic formation near the Cu substrate was greatly affected by these
particle additions. Ag and Au retarded and Ni suppressed the formation of CU3Sn and
enhanced the growth of Cu6$n5 as compared to pure solder with the Cu substrate.

Addition of Cu containing particles to the solder results in a decrease of both the Cu68n5

ll

 

ll}

1
p.

 

and Cu3Sn interface intermetallic thickness relative to the pure solder. This effect w as
believed to be due to the particles acting as Sn sinks. Similar studies were carried out by
Wu et al. with aging temperatures of 110°C to 160°C for 0 to 64 days [42]. The Cu
containing reinforcements resulted in increased activation energies for Cu68n5 formation
and decreased activation energies for Cu3Sn formation as compared to pure solder. The
activation energy for Cu3Sn formation decreased relative to the eutectic solder for the Ag
and Au composite solders even though less CU3Sn was formed at the substrate interface.
Ni and Pd drastically reduced the CU3Sn thickness and increased the Cu68n5 thickness.
Two mechanisms were proposed for the effects of Cu—containing particles and Ag
particles on the kinetics of intermetallic formation. First, the particles act as Sn sinks
which remove Sn from the solder and decrease the amount of Sn for reaction at the
interface. Second, the particles reduce the solder cross-sectional area available for Sn
diffusion, which also reduces the amount of Sn available at the interface for reaction.
Dispersion strengthened in-situ composite solders of Sn-Pb-Ni and Sn-Pb-Cu
alloys containing 0.1-1.0ttm dispersoids/reinforcements were produced by induction
melting and inert gas atomization, reported by Sastry et al. [27]. It was found that, upon
I‘eflow of the solder specimens, the fine spherical dispersoids in rapidly-solidified Sn-Pb-
Cu alloys coarsen to > 1pm platelets, however, the dispersoids in Sn—Pb-Ni alloys remain
Spherical and remain stable with a size of < 1 pm. The difference in stability of
djSpersoids in Cu— and Ni-containing solders was explained on the basis of the difference
in solubilities and diffusivities of Cu and Ni in Sn-Pb matrix. These composite solders
shOwed an increase of 25-180% in yield stress and 20-80% in the modulus values

compared to eutectic Sn-37Pb solder.

12

 

Another type of dispersion strengthened composite solder was formulated by
Betrabet er al. by adding 2.2wt% of Ni38n4 intermetallic particles into the Sn-40Pb solder
matrix [28]. Mechanical alloying, a solid-state high-energy milling process deveIOped
for superalloy manufacture, provided the means to process such dispersion strengthened
solders. The presence of Ni3Sn4 dispersoids resulted in a smaller grain size in the as-cast
microstructure and after aging at 100°C for 29 hours. Their subsequent study of
CugNiSn3 intermetallic particles reinforced Sn-40Pb composite solder showed an increase
of the strain to failure in shear by 40% while the ultimate shear strength essentially
remained unchanged [29]. They claimed this as an indication of improved fatigue
resistance because it was believed that fine, uniformly dispersed phases would stabilize
microstructures by pinning grain boundary dislocations and by restricting grain boundary
motion.

Mavoori and Jin prepared their composite solders by mixing 3v% of 10nm sized
A1203 powders or 3v% of 5nm sized TiOz powders with 35m sized eutectic Sn-37Pb
solder powder [30]. Nanosized, non-reacting, non-coarsening oxide particles formed

uniform coatings of solder after repeated plastic deformation for rearrangement of the
Particles. Three orders decrease in the steady state creep rate was achieved by this
approach. Such composite solder was found to be much more creep resistant than their
Control sample, the eutectic Sn-80Au solder. This has great significance in replacing the
Conventional high melting point (278°C) Sn-80Au solder for its creep resistant
applications such as optical or optoelectronic packaging.

Clough et al. (with G. Lucey) reported that, with properly controlled porosity,

CUGSns particle reinforced eutectic Sn-37Pb solders exhibited twice the yield strength

13

 

without significant ductility loss [31]. It Was also shown in their study that the creep rate it
of the composite solder was nearly an order of magnitude less than that of unreinforced
solders. The boundary layer fracture behavior was studied using single shear lap
specimens using the same composite solder. The specimens failed as shear fracture ran in
from opposite edges about 10m inside of the interfaces. These boundary layer fractures
were characterized and a fracture model was developed. Composite strengthening was
shown to significantly improve the ductility, creep life and properties associated with
improved reliability and creep-fatigue life [43].

The effects of phase additions on the microstructure, wettability and other
mechanical properties of the composite solders have also been reported in other studies
[44-45]. In general, composite solders tend to render improved properties. All the
reported investigations were basically exploratory in nature, and the extent of
improvement must be weighed against environmental and economic factors before
widespread adoption can be realized. However, studies on lead-free Sn-Ag based

Composite solders have received attention only recently.

2-2.2 Important Considerations for the Selection of Reinforcements

Reinforcements added to the solder matrix should satisfy certain conditions for
enhancing the solder performance [22]. Such conditions include: (1) the reinforcing
Phases should bond to the solder matrix, and the bonding could be weak or strong
depending on the need for specific considerations, (2) the reinforcements must have
acCeptable solubility in molten solder under normal reflow temperatures so as to maintain

the stability of the reinforcements during reflow or aging process, (3) the density of the

14

 

7““?- am...

we

reinforcements should be close to that of Solder matrix, so that a uniform distribution of
the reinforcing phase could be promoted, (4) the size of the reinforcing phases should be
optimal in order to stabilize the microstructure (It has been found that reinforcement
particles of ~lp.m or smaller tend to stabilize the microstructure [46].), (5) the particles
should not be prone to significant coarsening due to high interfacial energies of fine
particles during service, (6) the reinforcement should not significantly alter the
processing temperature, (7) the reinforcement should not alter the solderabilty by
changing wetting characteristics with the substrate.
2.2.3 Methods for Introducing the Reinforcements — in-situ, Mechanical Mixing
(Inert or Reactive Reinforcements)
Usually, reinforcement particles used in the composite solder can be grouped into

two categories. One kind of reinforcement addition involves the intermetallic particles.

 

These intermetallic reinforcements can be incorporated to the solder matrix either by
adding preformed intermetallic particles (like Cu68n5, Cu38n, or Ni38n4 [22-26,27-29,31])
01' by converting from elemental particles (like Cu, Ni, or Ag [22—26]) by their reaction
With Sn during fabrication or the subsequent aging and reflow process. Another kind of
r'E-‘vinforcement addition involves those having a low solubility and diffusivity in Sn, or
eVen non-reactive with Sn. Examples of such reinforcements could be Fe particles [32] or
OXide particles like A1203 or TiOz [30]. Choices of such reinforcements serve several
Pul‘poses. Proper choice of foreign reinforcement additions could desirably introduce
urliforrnly distributed intermetallic hard particles or non-coarsening particles. Well-

diSpersed reinforcements can serve as obstacles to grain growth, crack growth and

15

 

dislocation motion so as to strengthen the solder against creep and fatigue deformati 0,,
[28].

Two possible ways to introduce the desired reinforcements to the solder matrix are
the in-situ method and the mechanical mixing method. In-situ method refers to the
technique by which reinforcing phases, like Cu68n5 or Ni3Sn4 intermetallic particles, are
readily formed upon processing the bulk solder. The mechanical mixing method is more
related to extrinsically adding reinforcement particles into the solder matrix (usually
solder paste, sometimes molten solder). In the latter, composite solder paste is usually
prepared by mechanically blending the mixture for certain length of time to achieve

uniform distribution of the reinforcements.

2.2.4 Microstructural/Interfacial Aspects of Lead-free Composite Solders

 

Subramanian et al. have reported the microstructural evolution in the Cu68n5

particle reinforced eutectic Sn-3.5Ag based composite solders made by in-situ method
[47]. The effect of the intermetallic reinforcements on the growth of the intermetallic
interfacial layer between the solder and substrate was studied by carrying out aging
Studies ranging from few hours to few thousand hours. In these studies the average
interface thickness was measured after chosen aging times. The presence of the
il'ltermetallic reinforcements retards the growth of the interfacial intermetallic layer, even
after several thousand hours. The presence of Cu68n5 reinforcements also decreases the
coarsening kinetics of the Ag38n phase that normally forms in the solder. Microstructural
stuclies on the coarsening of Ag38n and Cu68n5 indicate that during short-term aging of a

few hundred hours Cu68n5 does not coarsen measurably. However, the Ag38n particles

16

 

- _.
A A:

‘1..-
‘f'f

did coarsen during these aging studies. The composite solder showed a lower activati 012
energy but a slower growth rate [13,48] when comparing the coarsening kinetics between
the composite and the non-composite eutectic Sn-3.5Ag solder. The intentionally added

Cu6Sn5 particles showed coarsening after several thousand hours long-term aging.

2.2.5 Resultant Properties of Lead-free Composite Solders

Certain composite solders have exhibited enhanced strength and other desired
properties sought by the electronics industry. McCormack et al. have developed
composite solders by adding 2.5wt% of ~2p.m magnetic Fe powders into pure Sn and
eutectic Sn-Bi solder [32]. The idea of adding Fe powders lies in the fact that Fe has low
solubility and diffusivity in Sn-based solder and thus is impervious to coarsening. A fine,

uniform dispersion of particles was obtained by imposing a magnetic field during the

 

solidification process. The composite solder made by adding 2.5 wt% of Fe powders to

pure Sn exhibited ~60-100% higher ultimate tensile strength than the dispersion-free

Solder materials. More importantly, its creep resistance at 100°C showed an increase by
a factor of 20. Fe particles reinforced eutectic Sn-Bi composite solder exhibited 10%
higher tensile strength and 5 times improvement in creep resistance under the similar
Conditions.

Subramanian et al. conducted creep tests with unaged and aged, non-composite and
Composite solder joints with various loads at several temperatures [47]. The 20v% in-situ
CugSns particle reinforced composite solder has about two to three orders of magnitude
botter creep resistance as compared to non-composite solder at room temperature and

l“)Vver strain-rates, representing conditions that will exist during cold hold time in a

17

 

thermomechanical cycle. Although aging reduces the creep resistance of these solders,
the composite solder possesses better creep resistance as compared to the non—composite
solder even under aged conditions. At higher temperatures and higher strain rates region,
the composite solder approaches the non-composite solder in creep behavior. At higher
temperatures, the ability of dislocations to climb over particles is apparently fast enough
to render the particles as ineffective dislocation barriers. At lower temperatures, the time
needed for a dislocation to climb around a particle (for a given stress) is much longer, so
the creep resistance is significantly improved.

The isothermal mechanical fatigue fracture behavior of the composite solder
containing 20% Cu68n5 was reported by Gibson et al. and compared with that of non-
composite solder [49]. The fracture surface of the composite eutectic Sn—3.5Ag solder

containing 20% Cu68n5 exhibited cleavage of the Cu6Sn5 particulate reinforcement and

 

ductile, Mode I fracture of the eutectic matrix with no single origin of initiation
corresponding to homogeneous ductile fracture. Meanwhile, the fracture surface of non—
composite eutectic Sn-3.5Ag solder joints exhibited ductile, mixed mode (I and H)
fracture behavior and step-type fatigue striations that originated at a local region.
Nanoindentation experiments by Lucas et al. have shown that the interfacial
Strength can be estimated from indenting particles that rotate about their initial position
[50]. Weak interfaces between Cu68n5 reinforcement particles and eutectic Sn-3.5Ag
n'latrix were detected for the in-situ composite solder from the nanoindentation testing. It
was shown that in-situ composite solder increases the ductility of the solder matrix
without increasing the strength significantly. Similar results were also reported by

Betrabet et al. for their Sn-Pb based composite solders [28, 29], as stated in 2.2.1.2. It is

18

 

‘jffi‘rr... .

believed that the Cu6Sn5 reinforcements create heterogeneities in the strain sites for k
inhibiting deformation and thus to promote homogeneous deformation. Therefore, an

increased ductility was achieved in the in-situ Cu68n5 reinforced composite solder.

2.2.6 Summary

Use of dispersoids is a viable means to improve the properties and service
temperature capabilities of solders. Such an approach can provide these improvements
without significantly affecting the currently used processing parameters to make the
solder joints. These dispersoids need to be inert or compatible with the solder so that
they will be relatively stable when solder joints are placed in service. They could be
introduced in the solder by in-situ methods or by converting the mechanically mixed
metallic particles into stable intermetallic compounds either by melting the solder prior to

making the joint or during reflow. Presence of these dispersoids aid in stabilizing the

 

solder joint microstructure by retarding the aging process. All dispersoids tend to

improve the creep strength of the solder by several orders of magnitude. An ideal
dispersoid should enhance the ductility without significantly strengthening it. Since
SOlder joint is highly inhomogeneous the deformation within the joint is highly localized.
with the presence of weakly bonded heterogenities due to dispersoids, deformation can
begin at several locations within the joint and cause homogeneous deformation. The
latter will aid in improving the ductility of the solder joint. Such features will make the
solcler more compliant and forgiving to accommodate the stress by relaxation while

delaying the onset of tertiary creep. This will improve the thermomechanical fatigue

19

 

...—eel

resistance of the solder joint. Reinforcements introduced by in-situ methods appear to be

more suitable for achieving this goal.

2.3 Creep Deformation Behavior of Sir-Based Solders

2.3.1 Overview of Creep Deformation Theory [51-55]

2.3.1.1 Introduction to Creep

Creep is defined as a phenomenon Of continuous deformation that materials

undergo when subjected to a constant load at an elevated temperature (Usually
Tappfiflm>05). Accumulation of strain with respect of time is he basic information
resulting from deformation under constant stress and temperature. The essential feature
of the characteristic time behavior at constant load and temperature Can be Shown on the
typical creep curve, as illustrated in Figure 2-1- Three regimes 0f the creep process; the
primary creep, the secondary creep or steady state creep and the tertiary creep may
usually be distinguished. At the primary creep, characterized by Strain hardening, the
strain rate decreases and at the secondary creep, characterized by the balance between
strain hardening and recovery, the strain rate remains approximately constant. Final 1),, at
the tertiary stage, the increase in strain rate is observed as a result of the change of the
dimension of the cross-section and the materials deterioration preceding the creep rupture,
Among these three stages, secondary creep rate is generally used for comparing the creep

resistance of materials, while the onset of tertiary creep is one of the criteria used to

predict the service life of the material.

26

 

   
  
 

 

   

 

ll
Primary Secondary Tertiary
Fracture
‘_ 1 fi- z II M“ m
u)
i:
‘3
1::
m
(18
dt Creep Rate
18“
M

 

Time t

Figure 2.1 The model shape of a typical creep curve [51‘54, 56],
Creep data are often presented in the form of the empirical e(Elation of power law
creep. Solder materials also follow the same power—law relationship between applied
stress and strain rate, as reported from many StUdieS [57-651 In evaluating the creep

behavior of solder materials, a generally accepted theory is expressed as follows [51 - 55

58. 64].
Steady_state creep is generally expressed by a relationship of the form [10, 54, 58]
CGb b P a " - Q) (2-1)
= —— — — D ex —— ’
3‘ kT (d) (a) 0 kT

where 5's is the steady-state strain rate, G is the shear modulus, b is the Burgers vector, k

is the Boltzmann’s constant, T is the absolute temperature, d is the grain size, 01's the

applied stress, Do is a frequency factor, Q is the activation energy for the deformation

21

 

 

process, n is the stress exponent. P is the grain size exponent, and C is a constant
characteristic of the underlying micromechanism.

The stress exponent n is dependent on the rate controlling mechanism. Though
there is considerable theoretical controversy as to the exact mechanism for creep
deformation, there exists at least four different types of generally accepted creep
deformation mechanisms. These include: (i) diffusional creep, characterized by n=l and
activation energy corresponding to self difoSion or grain boundary diffusion; (ii) grain
boundary sliding, characterized by n=2 and activation energy corresponding to grain
boundary diffusion; (iii) dislocation glide, characterized by n=3-4 and activation energy
corresponding to solute atom diffusion; and (iv) dislocation climb, characterized by ":5-

7 and activation energy corresponding to self diffusion and dislocation Pipe diff .
usron [58,

66]. Both dislocation glide and dislocation climb belong to the dislocation Cree p
PI'OCCSS.

2.3.1.2 Diffusional Creep

Diffusional creep occurs under low stress, high temperature conditions, which
describes viscous flow where n=l. Diffusional creep involves the flow of vacancies and
interstitials through a crystal under the incluence of applied stress. All matter deforms by

this mechanism if sufficiently slow times are taken. Diffusional creep occurs under stress
levels around 0/0<10'4. Nabarro-Herring creep and Coble creep are included under this
category.

When a polycrystalline metal creeps by the diffusion of atoms through the crystal
lattice, the process is known as the Nabarro-Herring creep [67, 68]- Nabarro and Herring

pr0posed that the creep process was controlled by stress-directed atomic diffusion. Stress

22

 

 

me

to

bec

ihe

 

changes the chemical potential of the atoms on the surfaces of the grains in a polycrysta]
in such a way that there is a flow of vacancies from grain boundaries experiencing ten 81-16
stresses to those which have compressive stresses, Simultaneously, there is a
corresponding flow of atoms in the opposite direction, and this leads to elongation of the
grain. The steady-state creep for Nabarro-Herring Creep is

~140b3Dv
N" deZ

 

’ (2-2)

where d is the grain diameter and Dv is the lattice diffusion coeffitient We nOte th t
' a
increasing the grain size reduces the creep rate.
Cable creep has the same idea, except it is based upon grain boundary
diffusion f
or
conditions where diffusion in the lattice is comparatively slower [69]. This ki
nd of creep
happened at lower temperatures than N abarro-Herring creep. The Steady Stat
' ‘ e Creep rate
can be expressed by

~ 5001;4ng ,

8 ~ 2-
c de3 ( 3)

A schematic diagram showing Nabarro-Herring and Coble creep on an ideal grain
is illustrated in Figure 2.2. Even though both forms of creep are favored by high
temperature and low stress, it is expected that Coble creep will dominate the creep rate in

very fine grained materials. In general case, the creep rate due to diffusional flow should
be considered as a sum of 6" NH and 3C , since the mechanisms operate in tandem, i.e.,

thfly are parallel creep processes.

23

d:

 

2.3.1.3 Grain Boundary Sliding

To prevent the formation of internal voids or cracks during diffusional creep of a
polycrystal, additional mass-transfer Processes mu“ C>Ccur at the grain boundaries. These
results in grain boundary sliding and the diffusional creep rate must be balanced exactly
by the grain boundary sliding creep rate if internal Voids are not to be formed. Diffusional
flow and grain boundary sliding, therefore, can be CODSidered as sequential processes in
which mass is first transported by Nabarro-Herring and/or Coble creep and a grain shape
change and separation is resulted. This is followed by “crack healing” via grain boundary
sliding. A schematic illustration of accommodating diffusional cream by grain boundary
sliding is shown in Figure 2.3 [70]. Raj and Ashby [71] considered the role of grain
boundary sliding and found that both Coble and Nabarro-Herring Cmep tend to make a
continuum. Given a boundary With ledges occurring periodically, Under shear stress,
vacancies move away from the compressed parts 0f boundary and move towards tensile
parts. Beere [72], Speight [73] and Aigeltinger [74] also found the fact that diffusional
creep can be considered as diffusion which is accommodated by grain boundary
sliding, or grain boundary sliding which is accommodated by diffusion. Any
consideration of mutually independent contributions of the grain boundary sliding and
diffusion is meaningless [71]. Some actual data from Zn~22A1 provided n=2 under this
creep Incohanism [75, 76]. It is important to note that the role of grain boundary sliding in
diffusional creep differs from that in dislocation creep. In the former, grain boundary
sliding is an indispensable prerequisite of diffusion, while in the latter, dislocation creep,

grain boundary sliding need not occur at all if a sufficient number of glide and climb

system Operate for the Von Mises criterion to be met.

24

0'

z

Flux

\\ //&_°

 

 

b

 

 

(a)
o

...... i ....

Flux

 

O'-——° —G

Wu

0
(b)

 

 

 

Figure 2.2 schematic diagram of (a) Nabarro-Herring creep and (b) COble creep on an
Ideal grain [67-69]-

X (1)
.1. s B .
D A (3)
AY(3) ‘
2' Z

(a) (b)

 

 

 

Figure 2.3 A schematic illustration of accommodating diffusion
boundary sliding. (a) Four garains in a hexagonal a”

.5. .. ......
. . . f
deformation. (b) After deforming by diffusronal creep, One dimensg: 53:5
grain is increased and the other is decreased, and v01ds are (mm d between
the grains. (c)The voids are removed by grain-boundary sliding. The extent
of sliding displacement is quantified by the distance Y’Yn, Which

is the
offset along the boundary between grains 1 and 3 of the original Vertical
scribe line XYZ [55, 70].

26

tit

 

.‘

2.3.1.4 Dislocation Creep

Dislocation creep involves the movement of dislocations, which overcome barriers
by thermally assisted mechanisms involving the diffusion of vacancies or interstitials. It

-4
occurs at intermediate to high stress levels (10 <o7’G<10’2) under a temperature range
between 0.4Tm and 0.7Tm, where the dominant mechanism of creep is the rate of

dislocation motion by glide or climb. This I‘egl'me is known as dislocation creep. The
steady-state creep rate under such dislocation creep represents a balance between the
competing factors of the strain hardening rate and the rate of themal recovery by the
rearrangement and annihilation of dislocations [51-55]. In other words, disiOCation creep
starts from the idea that creep can be conveniently Viewed as a manlfeStation of
competitive work hardening and recovery. The work hardening aSPCCIS of creep are
related to factors involving dislocation glide. Recovery aspficts are related to
nonconservative dislocation motion, e.g.. dislocation Climb in Which obstacles to
dislocation motion are circumvented and/or by which dislocations are annihilated \ i.e.,
removed from the structure.

(i) Dislocation Glide. This kind of creep occurs when dislocation motion is limited
by solute atmosphere (in pure metals and alloys), which pin the dislocation due to the
reduced strain energy associated with a dislocation core [75]. This mechanism occurs at

relatively high stresses, o/G>10'2, as compared to other creep mechanisms. The creep

rate is establi shed by the ease with which dislocations are impeded by obstacles including
precipitates, Solute atoms, and other dislocations [51]. Weetman presented the first theory

of creep controlled by viscous dislocation glide in solid solution alloys and assumed not a

27

 

specific type of interaction of solute atoms with dislocations [77]. Creep rate is described

by the following equation,

 

6 ~ Usbz ___ MKET, (24)

" GAB A G
where A is a constant depending on temperature and the mechanism of interaction of

solute atoms with dislocations, B=Gb2/271(1 " V)- Thus the stress exponent is n=3. Other

models, developed by Mott [78], Raymond [79], and Barrett [30]. etc., involve the
dislocation creep controlled by the glide of screw dislocation with jogs in pure metals
where similar stress exponent values were found.

(ii) Dislocation Climb. The earliest models of dislocation creep were advanced by
Weertman [81-84], which were based on a mechanism in which disl Ocation Climb plays a
major role. At elevated temperature, if a gliding dislocation is held up by an obstacle, a
small amount of climb may permit it to surmount the obstacle, allowing it to glide to the
next set of obstacles where the process is repeated- Almost all Of the Strain is produced by
the glide step, however, the climb step controls the rate. Since diffusion of vacancies or

interstitials is necessary for dislocation climb, the rate-limiting factor is atomic diffusion.
This model again predicts an equation for creep rate in which stress is raised to the third
power. HOWever, creep experiments with a range of metals show that the stress exponent

varies from 3 to 8, with a value of 5 most common. Thus, for intermediate to high stress

levels at temperature above 0.4Tm, the steady-state creep rate is described by a power-

law relation

’1
£3 = ADva (Z) , (2-5)

28

 

 

where A and n are materials constants. Since the diffusion coefficient Dv can be described
by
Dv = Do exp(—QlkT) . (2-6)
we can rearrange Equation (2—5) into
as = Ba" exp(—Q/ kT) , (2-7)
Which results in a simplified form of Equation (2—1). The modification [85] to Equation
(2‘5) allows it to be used for both high temperature creep, where lattice diffusion

predOmi nates, and for low temperature creep, where diffusion along dislocation cores is

the Predorninant mechanism.

2.3.1.5 Power-law Breakdown [10, 58, 85-88]
At 6: ven higher stress, 0/G>10'3, the strain rate is an exponential function of stress.
The intemiediate-to-high stress power-law breakdown region can be described by a

single expression

It
£3 = C1 g[sinh(a%]] apt-25g} (2'8)

where 0‘ prescribes the stress level at which the power law dependence breaks down, and
C1 is a cotlstant.

To Obtain the true activation energy, the temperature dependence of the shear

modulus mum be incorporated

G = G0 — G,T', (2-9)

 

29

 

Where Go is the modulus at 0°C, G1 giVeS the temperature dependence, and 7" is the
temperature in °C.
According to this theory, it was approved that all the creep data can be fitted on the
Same general form of constitutive equations [58], which provides a general criterion for
the Comparison of the creep properties of different solder materials. Deformation maps,
deVelop<3d by Ashby et al.[89], schematically shown in Figure 2.4, provided a practical

Way of illustrating and utilizing the constitutive equation for the various creep

deformati on mechanisms.

 

 

 
    

 

 

 

 

 

 

 

0'4 , .............................................................................
\ Dislocation glide
Dislocation
cncep

A 63 .. - _________________________________________________________________________
‘2
D
v 1 t'cit
.5 E asr y

52 — -- ..................................................................

01 -— -- ___________________________________________________________________________

Tl T, T,
T/Tm—’

 

Figure 2-4 A schematic deformation mechanism map. The axes of the diagram are

homologous temperature (T IT m) and stress (in terms of the shear modulus).
The stress-temperature combination determines the primary deformation
mode. At the boundary lines, the deformation is due equally to two different
mechanisms, and, at the intersection of the lines, to three different

mechanisms [55, 89].

 

30

 

 

 

Bllnw'
Su-b:
omen"
whim
scnicc
defom
these a

350 K
dominz
temper

perfom

alloys‘
Itsu-ltg

meet

rep]:

2.3.2, “mortance of Studying the Creep Behavior of Sir-based Solder Materials

Sn-based solder alloys are widely used for interconnection and packaging of

COmmer'cial microelectronic assemblies. High reliability and good mechanical

perfomiance are required for solders, especially solder joints, because during normal
sel‘Vice conditions, the low melting solder alloys are subjected to creep and fatigue
defol‘mation. Even at ambient temperature of operation, the homologous temperature for
theSe alloys is high, around 0.6, while the interconnect temperatures can reach as high as

350 K due to local heating. In such environment, high temperature creep processes

dominate the deformation. It is imperative for researchers that knowledge of high

temperature creep behavior be gained in order to predict reliability and overall
Performance of solder materials.

There have been many reports about the creep deformation behavior of solder
alloys, especially Sn-based solder alloys. The following review presents the set-up and
results 0f creep tests performed on Sn-based solder alloys in some important studies.
Creep Properties of the materials tested are presented and compared. Controlling

meChanisms for creep are summarized. Due to the adoption of “green” manufacturing
practices, there has been new interest in developing lead-free solders. The following

revrew also focuses on the creep properties reported in the recent literature for lead-free

solder :11 10y s

2‘33 Materials Used in Creep Tests
But‘i‘vctic Sn-3.5Ag solder has been broadly targeted as the foremost candidate to

replace lead-bearing solders. Creep deformation behavior of Eutectic Sn-3.5Ag solder

 

31

 

 

‘mi

has been most frequently reported in the recent literature [57-62]. Other lead-free some“
Often used for creep study include Sn-SSb [57], Sn-9Zn [59,60], eutectic 811-31 (62 1, and
Sn‘4.OAg-0.5Cu [63] solder alloys. Examples of lead-bearing solders used for creep
Study often include 6OSn-40Pb [58, 64], 6ZSn-36Pb-2Ag [58], 97Pb-3Sn [58], 95Pb-SSn
[58]. 6ZSn-38Pb [64], etc. The creep properties for pure Sn were investigated for
Compari son in some cases [57]. Other solder alloy systems, like In-Ag solder alloys,
Were also tested for creep properties in an effort to explore suitable substitutes for the

existing lead—bearing solders [65].

2.3.4 Set—ups for the Creep Tests
Some creep tests are set up for testing bulk solder materials. As part of an
investigati on of using eutectic Sn-3.5Ag solder for flip chip interconnection, reported by
Yang et al., constant load creep tests were performed at high homologous temperatures
from 25°C to 180°C (0.6Tm to 0.92Tm) using a dead load creep machine, and the
displacement was monitored using a linear variable differential transformer (LVDT) with
an accuracy of i5x10’5 in [61]. Mavoori et al. also reported a constant load creep testing
on bulk, cast, dog-bone shaped specimens on a servo-hydraulic MTS machine [59].
Constant l oad and temperature were controlled by a 458.20 programmable microconsole.
Until recently, most of the creep tests were performed on bulk solder materials.
Recent Studies have pointed out that creep behavior of bulk solders does not truly
represent the creep behavior of solder joints functioning in real engineering applications.
Bulk solders do not behave in similar ways under creep conditions that thin solder joints

With . . . . . . .
cons‘l‘l‘arnts do. As an initial consrderation we know that the rnrcrostructure of bulk

32

 

wed

some“ is significantly different than microstructure of routinely fabricated solder joints

because solidification parameters and process variables are quite different, particularly

Cooling rates [90, 91]. Unlike the microstructure of bulk solder, the solder joint

micI‘ostructure consists of an interfacial intermetallic layer that forms between the solder

and the substrate and intermetallics particles found within the solder matrix consisting of

ConStituent elements of all materials comprising the solder joint. The interfacial

intel‘metallic layer inherent in solder joints can act as a constraint to plastic flow. Thus,
the interrnetallic layer can alter the stress state in the joint, which can be significantly
different than that in bulk solders under creep conditions. So, creep data obtained using
blflk solder materials will not necessarily correlate well to creep data gathered using thin
Solder joints. Due to these reasons, recent studies are conducted with solder joints that
are representative to real conditions for solders used in industrial practices.

Single shear lap solder joint specimens were most frequently used in the creep
tests. An example of using single shear lap solder joint for creep testing can be referred
to the Paper reported by Raeder et al. [62]. These specimens were prepared by joining

tWO pieces of commercially pure copper with 0.25mm thick foil solder. Solder mask on
the copper limited the spread of the solder during soldering process. Spacers, with the

same thicl<ness as the solder preform, were placed on either side of the solder to confine

the joint thickness.

A double shear lap configuration used for creep testing was reported by Darveaux
6’ al. [58] ~ Such a configuration minimizes the bending component during the tests. All
these tests were conducted in a screw driven Instron 4501 with a load cell. LabVIEW

dat ~ .
a acqul sitron software was used to control the Instron and to record load and

33

 

N

displacement readings. The temperature Chamber was capable of maintaining a constant

temperature within i0.5°C.

2‘35 Reported Creep Data and Related Mechanisms for Sn-based Solder Materials

23.5.1 Creep Deformation Behavior of Sn, Sn-3.5Ag and Sn-SSb Solder Alloys [57]
Creep deformation mechanisms in pure Sn, Sn-3.5Ag and Sn-SSb solder alloys

(bulk) was studied by Mathew et al. in the temperature ranges of ambient to 473K [57].
P °Wer-l aw relationships between strain rate and stress were generally observed. Stress
e"Ponent n=7.6i0.2, 50:02, 5.0103 were obtained for Sn, Sn-3.5Ag and Sn-SSb
respectively. Activation energies for creep Qc =60.3i3.8, 60.7i6.6, and 44.7i3.7
kJ/mole for Sn, Sn—3.5Ag, and Sn-5Sb respectively. From phenomenological analyses,
activation energies for self-diffusion through the lattice and dislocation core for Sn were
estimated as 65 and 46 kJ/mole respectively. Based on the experimentally determined
stress ex1)()nent and activation energy values, it was suggested by the investigators that
the creep deformation mechanisms are dislocation climb controlled by lattice diffusion in
Pure 5“ and Sn-3.5Ag alloy, and viscous glide controlled by pipe diffusion in Sn-SSb
alloys ($01 id solution alloy).
2'3‘5'2 COnipai'ison of Creep Behavior between Sn-3.5Ag, 6OSn-40Pb, and 628n-
b-2Ag [58]

The creep defamation behavior of Sn-3.5Ag solder was studied and compared

With ”108% of 6OSn-40Pb and 6ZSn-36Pb—2Ag solder alloys, reported by Darveaux et al.
[581' F r0111 their plots of normalized strain rate versus normalized shear stress, they have

proved “1 at all of the creep data can be fitted to the same general form of constitutive

34

 

wmi

relat‘ions','i.e.. only the constants depend on the solder alloy type. Comparison shows that
the Sn-3.5Ag has better high temperature creep resistance than 6OSn-40Pb or 62$n~36pb_
2A8 solder alloys. It is evident that the power law break down regime is approximately
10‘3 t/G for all alloys tested. The 6OSn-40Pb and 62Sn-36Pb-2Ag alloys showed a stress
exPonent of 3.0, indicating dislocation viscous glide controlled deformation. The Sn-
3-5Ag alloy had a stress exponent of 5.5, indicating dislocation climb controlled
deformation mechanism. A study of alloy ductility was conducted by plotting the onset
of tel'tr'ary creep versus normalized shear stress [58]. They found that the Sn-3.5Ag and
97p 1338 b alloys absorbed more strain before the onset of failure than the 6OSn-40Pb and

623n-36Pb-2Ag alloys. The 95Pb-SSn alloy was the least ductile, but, as suggested by

these inve stigators, this could be due to the constraining effect of the smaller joint size in

their creep tests.

2.3.5.3 Creep Deformation Behavior of Eutectic Sn-3.5Ag and Sn-9Zn Solders [59]

Creep tests were performed to both eutectic Sn-3.5Ag and eutectic Sn-9Zn solders
under load control with initial stresses ranging from 10 to 22 MPa at two different
t‘flrll’eri‘l‘lll‘es, 25 and 80°C, by Mavoori et al.[59]. The stress exponent for Sn-3.5Ag was
found to be higher than that for the Sn-9Zn. It was observed that there is a crossover
between the plots for eutectic Sn-3.5Ag and Sn-9Zn, which indicated that at lower
stresses, cl‘eep rates for Sn-3.5Ag are found to be smaller than for Sn-9Zn; however, as

the Stress i 8 increased, the creep rate of Sn-3.5Ag became faster than that of Sn-9Zn. The

25 and 80°C creep data were compared with data derived from Dom and Norton’s

35

 

e(Nation at the stress ranges tested and a close agreement between theoretical and

experimental results was found.

235.4 Creep Deformation Behavior of Eutectic Sn-3.5Ag and Sn-Bi Solder and
Solder Joints [62]

The isothermal tensile steady state creep rates of cast Sn-Bi and Sn-Ag eutectic
solders and their joints were reported by Raeder et al. at temperatures between 20°C and
160°C [62]. A stress exponent of n=8.4 for room temperatures and n=6.7 for high
temperatures were reported for eutectic Sn-Ag solder. A stress exponent of n=3.l was

l'ePOrted for eutectic Sn-Bi solder at all temperatures under intermediate strain rate. At
l'Oom temperature, the creep rates for eutectic Sn-Ag solder joints were found to be
higher than those for Sn-Bi solder joints when the stress level was less than ~27MPa.
The tensil e behavior of bulk solder was compared With shear behavior of solder joints for
each solder material. The shear lap joint specimens for eutectic Sn-Ag solder were found
to be much stronger than the bulk specimens, while eutectic Sn—Bi solder did not show
SUCh a Si gnificant difference. The reason for such a behavior is attributed to the
PmceSSing variables; the Sn-Bi joints were reflowed for a shorter time period at 200°C
and the Sn-Ag joints were reflowed at 250°C. At these temperatures copper dissolves
more quickly in molten Sn-Ag solder than in Sn-Bi solder. Copper present in the form of
intermetéLllies in the Sn-Ag joints may strengthen the joints. It was also suggested that

quenching the Sn-Ag joints might strengthen them in comparison to the slow cooled bulk

solders.

36

23.5.5 Creep Deformation Behavior 0f Eutectic Sn-3.5Ag Solder and It? Solder
Joints [61]

Yang et al. have reported constant-load creep test data on bulk Sn-3.5Ag solder
alloy at high homologous temperatures from 25°C to 180°C [61]. At all testing
temperatures, the strain rate increased with stress following a power-law relationship with
a Stre$s exponent n=10. This exponent is much higher than that usually noted for creep
dUe to dislocation climb where generally the value varies from about 4 to 7. The
aCti Vation energy they obtained was Q=12.1Kcal/mole (about 50kJ/mole). This value is
abOUt one half of the activation energy for self-diffusion in pure Sn and thus is
identifiable with either grain boundary or dislocation-pipe diffusion. The relatively large
Value for the stress exponent (n=10) along with this observation points to low-
temperature climb of edge dislocation to be the rate-controlling mechanism [61].

Upon comparing the creep data between bulk and joint specimens, it has been
found that the joint specimens were more creep resistant than bulk specimens. Several
possible reasons regarding the discrepancy in creep behavior between solder joints and
bulk 801dE=r materials have been suggested [46]: (l) The substrate metal may dissolve into
the mOIIGn solder during soldering process altering the chemical composition of the
SOIdCIJOint. (2) The substrate metal may also react with the solder to form intermetallics
at the interface. Sometimes, the intermetallic layer(s) may break away from the interface
and disperse into the bulk solder resulting the dispersion strengthening mechanism. (3)
The grain size of the solder joint is process-dependent and may differ from that of a bulk
Solder Specimen. (4) Geometry constraints (eg. thickness-to-aspect ratio) may also

Influence the creep behavior of the solder joints. (5) Microstructural difference in thin

37

 

joint (usually a finer microstructure) due to substantial difference in cooling rates during

fabrication could also be a possible factor to be accounted for.

2.3.5-6 Summary

It should be noted from the above review that there is no clear agreement in the
eXperimental data, which resulted in the discrepancy for the operating creep mechanism.
Such discrepancy may be due to the use of different type of specimens (bulk or joint),
different processing conditions (cooling rate, aged or unaged), and/or different testing
methods (tensile or compressive), etc. But it should be noted that eutectic Sn-3.5Ag

Solder generally exhibits superior creep resistance as compared to other solder materials

discussed above.

2—4 Microstructural Issues in Solder Materials

Solders are known for their evolving microstructures especially when aged and

r ef] owed at high temperatures and during thermal and isothermal mechanical fatigue

pr‘(><>¢E:sses. Microstructure evolution of solder materials has been known for a long time

[1 3 -4932]. Even at room temperature, the microstructure of solder materials is prone to

coarsening, as noted by Tien et al. from their microstructural observation of Sn-Pb

Solders [93]. Similar microstructure coarsening occurs also in lead-free solders, like in
&\%ctic Sn-3.5Ag solders, as reported by Gibson et al. [13], and Yang et al. [14].

The microstructure of as-fabricated Sn-Ag solder joints usually consists of Ag38n

preCipitate in the Sn matrix with Cu-Sn intermetallics in both the bulk solder in the form

0f dendrites and at the solder/copper substrate interface in the form of layers [13,14].

38

 

The Cu-Sn intermetallic phases in this layer are Cu68n5 (n-phase) and C1133" (8~phase),
where Cu68n5 is adjacent to solder and Cu3Sn is adjacent to Cu [94, 95]. (In the cases of
using Ni substrates, the intermetallic phases will primarily be Ni3Sn4, Ni38n2 and also a
metastable phase, NiSn3 [96-98].) These intermetallics are formed as a result of the
diffusion and dissolution of Cu in molten solder during the soldering process [99-101].
Thus the microstructural evolution of Sn-Ag solder joints during aging and reflow usually
consists of Ag3Sn precipitate coarsening, Cu-Sn intermetallic dendrite coarsening in the
bulk solder, and Cu-Sn intermetallic layer growth at the interface.

There have been a significant number of in-depth studies of the formation and
gTOWth of intermetallic layers during aging and reflow. It was reported that the growth
rate of the intermetallic layer is initially much higher when the solder is molten and it is

related to the rate at which the reactants can diffuse to and/or through the existing layer
(frequently, the growth rate is proportional to t"2 or tm’, where t is time) [96]. The
kin(Etics of intermetallic formation at the liquid solder/Cu interface appears to be similar
for eutectic Sn-Pb, Sn-Ag and pure Sn, when comparing the work reported by London
[1 02] and Warwick [97]. It was also reported that the intermetallic layer formation
tefnded to proceed much faster on Cu substrate than on Ni substrate [103]. Choi et al.
have found that the initial thickness of the Cu68n5 intermetallic layer was consistently
Smaller after solidification of the Sn-Pb composite solder joint, but subsequent growth
\‘e‘ics of the interfacial thickness were similar with Sn-Pb non-composite solder [104].
5uflwoo and Mei et al. have studied the growth of Cu-Sn intermetallics at a pre-tinned
copper/solder interface and they constructed a multiphase diffusional model to analyze

the 8- and n-phase at a plane Cu-Sn interface in a semi-infinite diffusion couple [105,

39

106]- Using their diffusion model, Mei et.al. were able to compute the position of the 8-
and n-phase with time by determining the interdiffusion coefficient in the 8 and n-layers.
An advantage of Mei’s analytical model was that precise knowledge of the concentration
profile at the growing interface was not required. Wu et.al. studied the formation and
growth of intermetallics in the composite solders and suggested that tin was the
predominant diffusing species that controlled formation and growth of the Cu—Sn
internretallic layer during aging and reflow [42]. The effect of Ag, Au, Ni, and Cu
particle additions on the formation and growth of the intermetallic layer was discussed
and compared to pure solder.
Solder joint failure has been found to occur during room temperature and high
temperature aging. Lampe’s aging study of nine Sn-Pb solder alloys showed dramatic
decreases in shear strength within the first twenty days of room temperature aging and
CXtel’lsive microstructural coarsening [107]. Yang et al. observed the failure of eutectic
Sn‘Ag solder joints made by infrared reflow after three days of aging at 190°C [14].
meir joints made by laser soldering failed after seven days at the same temperature. The
fajlllre was due to crack initiation and growth cause by the stresses from thermal
I“isliaatch at local areas as well as void formation in the CU3Sn phase, as explained by

these investigators.

40

CHAPTER III
IVIICROSTRUCTURE OF AS-FABRICATED Cu, Ag, AND Ni PARTICLE

REINFORCED COMPOSITE SOLDER JOINTS

3.1 Preparation of Composite Solders

The composite solders were prepared by mechanically adding ~6 urn size Cu, ~4
Mm size Ag, or ~6 um size Ni particles to the Sn-3.5Ag eutectic solder paste for some
Studies. Copper, silver, and nickel powders used as the reinforcements were obtained
from Atlantic Equipment Engineers, INC., Bergenfield, New Jersey. The purity of all
materials was reached at 99.9%. The morphology and size of the as received Cu, Ag, or
Ni particles were studied using scanning electron microscopy. Particle sizes were
In(easured directly from scanning electron microscopy (SEM) micrographs using Adobe

PhOtoshop 5.0®, an imaging processing software package.
One way of making the composite solder material involved mechanically blending
the reinforcing particles and solder paste in a ceramic crucible for at least 15 minutes to
p“(>Iljote uniform reinforcement distribution. The composite solders used in this study
noll'linally contained 15 volume percent Cu, Ag, or Ni reinforcements. In order to have a
‘1“an initial examination of the microstructure, after mechanical mixing, usually, a
bthI:<)n-shaped composite solder specimen was prepared by melting and solidifying a
Wen] amount (~5 grams) of the composite solder paste on a copper substrate or in a
ceramic crucible at 280°C. The composite solder was cooled from the molten state by
placing the substrate or crucible on an aluminium plate. Solidified samples were cleaned,

Seetioned and polished for SEM and energy dispersive x-ray (EDX) examination. The

41

composite solders were also examined using optical microscopy. Energy dispersive x-ray
analysis was carried out to identify element phases within the solder matrix and IM
layers.

Alternatively, composite solders were prepared by slowly adding the reinforcement
particles into a molten Sn-3.5Ag solder at 280°C while stirring the mixture. The mixture

was cooled from the molten state using the solidification procedure described above.

3.2 Solder Joint Fabrication
Single shear lap, dog-bone shaped solder joint specimens were used thoughout the
study for microstructural examination or mechanical tests. The specimen consisted of Cu
SUbstrate arms which were fabricated by electro-spark discharge machining (EDM). The
Sl-lblerate dimensions are shown in Figure 3.1. The Cu substrates were chemically
ckframed with a solution of 50% Nitric acid and 50% H20. Next, a solder masking
Cc>I‘Itqnound was applied on the tip ends of the Cu substrate to limit solder joint size to an
area of ~ 1 m2. Composite solder paste or several solder foil preforms ~ 30 pm thick
by 1 mm2 in area were then sandwiched between the two Cu substrates as shown in
Fi glare 3.1 in order to fabricate solder joints that were typically 100 pm thick. In the
cases when solder foil preforms were used, Alpha ZOO-L flux was applied to the
$01 derable areas of the Cu substrate prior to applying the solder foils. The prepared joints
fit‘e placed in a holding fixture and subsequently heated to a temperature of 280°C on a
pot plate to achieve melting. The melted joints were then removed from the heat source
and quickly placed on an aluminum chill block which promoted cooling rates similar to

01' Slightly faster than those in industrial practice of reflowed solder joints. The thermal

42

history profile for solder joint fabrication is shown in Figure 3.2. The comparative

dimensions of

Solder Foil

\R’Qn Alpha 200L Flux
Solder Mask

  
  
 

11_ 5 mm Thermocouple Wires

0.5 mm

3.0 mm
Heated Aluminum Fixture

Figure 3.1 Dimensions of shear-lap solder joint, copper substrate and aluminum fixture.

 

3.. ...I. wi..,va—v,iiy.l.,VV‘..ij—ryrw

    

Temperature ('C)
E
l

 

 

 

 

Time (sec)

Figure 3.2 Temperature profile of single shear lap solder joint fabrication.

43

the solder joint are illustrated in Figure 3.3. The solder joint area is ~ 1mm2 and its
thickness is nominally 100m. The joint size is representative of solder joints used in
microelectronics industrial applications. An essential step in our investigative procedures
involves metallographical polishing one side of the solder joint for microstructural

characterization and documentation prior to testing.

Solder Joint

Cu
Substrate

 

Fi gun 3.3 Comparative dimension of a single shear lap solder joint used in the
investigation.
3’3 Mcrostructure of As-fabricated Cu Particle Reinforced Composite Solder Joints
A typical microstructure in the as-fabricated Cu particle reinforced composite
«EQGers joint is shown in Figure 3.4. The Cu-Sn 1M layer that formed at the periphery of
cu reinforcement particles is clearly shown in Figure 3.4(a) and (b). The IM layer
thiCkness was approximately 1.5-2 mm thick. In fact, the IM layer that forms around the

C" reinforcement is an intermetallic co-layer consisting of a CU3Sn layer (ta-phase) that

44

('u

it .

__ _ . _ a
'( u-Sn IM Layer 0" A.—
Solder

Matrix

n

0

20um (C) C u (til [II

Pi gure 3.4 Microstructure of as fabricated Cu particle reinforced solder joint. (a) Overall
view of the microstructure in the joint, (b) Cu reinforcement particle and the
IM layer formed around Cu, (c) IM layer formed at the Cu substrate/solder

interface.
Shares an interface with the Cu particle (Cu substrate) and of a Cu68n5 layer (n-phase)
th at shares an interface with CU3Sn and the solder. During the first reflow of the sample,
the Cugsn layer is fairly thin and virtually undetectable as noted by others [42, 51, 105].
The dark core region of the particulate reinforcement is pure Cu, and the light exterior
labier surrounding the Cu particle is the Cu58n5/Cu38n IM co-layer. The Sn-Ag solder
ll'lElthx microstructure is characterized by eutectic Ag38n phase residing between Sn
Vains. Energy dispersive x-ray analysis confirmed the chemical elements present in the

matrix microstructure and in the IM layer. The IM layer at the Cu substrate/solder

interface is also a Cu-Sn co-layer, as shown in Figure 3.4(c).

45

 

.1
.9
L

 

3.4 Microstructure of As-fabricated Ag Particle Reinforced Corn P081245. So Id” Jam;
The microstructure that is representative of the Ag particle reinforced composite
solders is shown in Figure 3.5 (a)-(b). The light contrast 110th around the perimeter of
the Ag particles was identified as a thin layer of Ag35n. The darker core is non—reacted
Ag. In the first reflow of the composite solders, the thickness of the Ag3Sn [Ni layer that
formed in the Ag particle reinforced composite solder was much smaller than the CuSn
1M layer that formed in the Cu particle reinforced composite solder. An EDX scan across

. he Ag
3 number of Ag particles indicated that, except for a thin layer at the very edge, t

the Ag
reinforcement was pure Ag. The Ag3$n 1M layer that formed around

Ag
_ of th6
reinforcements is only ~ 0. l-O.3 pm thick. The solder matrix Inlcrostructure e“
o c

. d at
composite ShOWS the usual pure Sn cells with eutectic Aggsn partrcles (mm

-“\\\a!
. 5 $‘
boundaries. The 1M laYcr at the Cu substrate/solder interface (Figure 3.50) Show

. . ia\
Cu-Sn IM co-layer as in Cu composite solder. No Ag was detected 11) the Interfac

layer.

3-5 Microstructure of As-fabricated Ni Particle RBinfol‘ced comPOSite Sold

Qp J .
. 0
3.5.1 Merostmcture under Regular Heating and Cooling Condition lnts

Microstructure that is representative of the as-fabricated Ni particle r .

Info
co - . - ' “Ce
mpOSI t6 SOIder JOInt IS Shown in Figure 3,6(3). The euteCtlc Sn—3.5Ag SOld d

er m .
micrmt . . . . . x
PuQture is characterized by eutectic Ag3Sn phase “331de between Sn grains Th
' e

dark c -
Ore region of the particulate reinforcement rs pure Ni, and the light eXterior layer

sum) -
until tag the Ni particle is the IM layer formed upon first reflow of the solder

materi . .
a1 ‘ The IM layers developed around the Ni particles were approxrmatejy 1—2 pm

46

v‘Sn—AgAg
'Mat'rix”

 

) magi;

. - a

Figure 3.5 Microstructure of as fabricated Ag particle reinfofced solder 3° “title dxdel
view of the microstructure in the joint, (13) Ag reinforcement Pan 6150\

IM layer formed around Ag, (c) IM layer formed at the Cu 5‘1 3

interface.

47

tick all

“We
m
if
part1
Ener
press;
m'nfor
ptriphc
Compou

N [are

maCu

lhal shu
Slim
phase)
Wider

CU sul

differe
mink

Sn, it

thick and can be characterized by a “sunburst” pattern Surroun (:11);g the M paint/e
reinforcements. The “sunburst” INIC patterning around the Ni particles may also be
referred to as “sunflower”. It is also clearly shown that there is a significant number of
~0.5 um sized small I'M particles scattered, mostly, along periphery of Ni reinforcement
particles. A few of these small IM particles appear in the eutectic Sn-3.5Ag matrix-
Energy dispersive X-ray (EDX) analysis was performed to verify chemical elements
present in the particle and [Ni layer. The composition of IM layer surrounding the Ni

. . - t the
reinforcement consisted of Cu-Ni-Sn. EDX analysrs to identify small particles a

M
t rnar)’
periphery 0f the IM layer revealed also that these particles were also a e

d
. . . . le an d the surrounde
compound of Cu, Ni and Sn. A higher magnified View the N1 Pan“:
IM layer is shown in Figure 3.6(b). o étsttng
C
EDX analysis of Cu substrate interfacial 1M layer reVealed M c0433“
ar‘i met
of a Cu-Sn layer that shares an interface with Cu substrate and a C‘l‘bli—Sn tom
that shares an interface with solder matrix. The Cu-Sn IM layer that is adjacent to the CU
substrate was mostly Cu68n5 (rt-phase) in the as-fabricated solder jOinr, The
C1135!) (8-

phase) is virtually undetectable in the first reflow but the e-phase thickness

under subsequent aging or reflow processes [42, 51, 105]. The interfacial 1M ’61 386s

Cu substrate is shown in Figure 3.6(c). yer at the
171% diffusion results listed in Table 3-1 [13’ 108] can be used to exPlain th

difference: in the extent of [Ni layer formation observed around the Cu, Ag and Ni particl:
[ninth-cements. In Table 3.1, which alludes to the extent of Cu, Ag, and Ni diffusion in

3n, it ' . .
ls e=\.;ident that the diffusing distance fer Cu in Sn perpendicular to its c~axis of the

48

Interfacial IM Layer .

. _ @u-hli-Sn’lM Layer

CwNi-SnzParticle’s '

' Cu—Ni-Sn IM Lax/er ' '
Cu-Sn IM Lax/‘3r

Cu substrate

 

) 0‘163‘:

. - a

Figure 3.6. Microstructure of as fabricated Ni particle reinforced Wider lo‘nt'.( e “Xx
afi‘d home

view of the microstructure in the joint, (b) Ni reinforcement P
1M layer formed around Ni, (c) [M layer formed at the Cu S‘1 S

interface.

49

unit cell is ~ 400 pm after 100 hours at 25°C. In contrast, the diffusi 17g distanccf A .
or gm

Sn perpendicular to the c—axis is ~ 0.5 um under similar COUditiOUS- Cu diffuses nearly

103 times faster in Sn perpendicular to its c-axis than does Ag. In similar calculations, Cu

diffuses more than 105 times faster in Sn parallel to its c-axis than Ag, which indicates

that the diffusion couple of Sn—Cu is significantly more active than the diffusion couple

of Sn-Ag. Ni and Cu are similar in chemical characteristics. such as atomic number,

similar atomic radius, and electronegativity. Thus, their diffusion behavi

CXpected to be similar. Consequently, much thicker initial IM layers were obse

. forcemcnt
around Cu and Ni particles as compared to the IM layers around the Ag rein

 

 

 

 

 

 

 

 

  

 

  

 

 

 

 

 

 

 

 

 

 

or in Sn is

rved

 

 

 

 

 

 

 

 

 

 

particles.
30%\
Table 3.1 Comparison of Diffusion Parameters for Ag, C“, and Ni in S“ ‘13,
1)0 Q wt)"2 in Hm. I 00 homs
cmzls kJ/ mol 25°Cv 75°C 1 00°C 1 500C
Ag in Sn.Lc* 0.18 77 0.454 4.2 \m 45
Ag in Sn l/c** 0.0071 55.1 7.5 37 ___
Cu in Sn_LC* 0.0024 33.1 369 964 1414 200
C" in Sn // c** D=2><10*i c3177?” 8485 --- --- 2660
M in Snjc’“ 0.0137 54.2 --- “‘
M in S?” 0’” 0.0192 18.1 21549 36421 44915 :36
L: :L indicates diffusion of species transverse to the c-axis of the unit cell of tin
- I] indicates difquion of species parallel to the C-aXlS of the unit cell of tin. '

*ak

 

50

* : D0 and Q values are not available in the references individually.

3.5.2 Microstructure under Different Heating and Cooling Candi tions
3.5.2.1 Introduction

Intermetallic compound (IMC) formation around meChanically introduced metallic
Particles due to the joining operation is directly related to reflow temperature—time profile
of the solder while it is in the liquid state [109, 110]. During reflow, reaction between the
incorporated metal particles and the molten solder occurs, and IMC may nucleate and
grow at the particle/solder interface. Two different IMC morphologies around the Ni
reinforcement Particles have been observed resulted from reflow operation. The Cffcc“

' 1e
' - . . - and P31 “c
of the reinforcement shape on the fracture behavror, duetility, resrdual stress.

h
wgvefq sue
cracking have been observed in metal matrix composites [1 1 l. 112-1- H0

in swaat’
findings have not been reported for both lead-bearing and bad—free Somets. Qamt
o o . so
spherical shaped reinforcements would provide the compOSItC material w‘m 0t
0

mechanical properties as compared to angular shaped reinforcements. The object“
the study in this section is to investigate the thermal history dependence for the
development of different IMC morphologies during reflow.
35°23 Experimental
To assess the effect of thermal history on the INIC formation and mo
“Dholqgic

de
velopment, several heating and COOling thermal prom“ were used for reflow'
m

particleS g N

reinforced composite Sn-3.5Ag solder joints. Following are the 61368
of

ex '
Penmehts that were conducted.

’51

( 1) Cooling Rates Experiments

In order to evaluate the effects of cooling rate, four different Coo/1'1; g rates were
attempted. This was achieved by reflowing all the Specimens using a preheated hot plate
as the heating source. The preferred cool down segments of the reflow profile were
achieved by placing the assembly of specimens after the solder is molten either (1) in
water, (2) on a chill alurniurn block, (3) on a firebrick, or (4) on a wood block. It should
be noted that the heat-up rates for these reflow profiles remained the same for a“

. 13 of
Specimens used in this part of study; only the cooling rates differed to study 1’0

- ed
' . s expenenc
different cooling rates. The corresponding temperature-time reflow profile

ufe'
ares tempera“

in this series of experiments are given in Figure 3.7. Figure 3.8 comp {as‘es‘

- . . . - till“. fig
time profiles corresponding to fastest and slowest cooling rates. while mm“

heating rate.
(2) Heating Rates Experiments

This series of experiments were carried out to evaluate the influence of heating rate.
Three different heating rates were used. The desired heating rates Were ac .
hleved by

placing the specimen assembly (a) on the preheated hot plate With the high
es

. 1‘ po
setting to provide the fastest heating rate, (b) on the preheated hot “at Wet
e

intemjed - . . . ° at the
late power setting to attain medlllm heating rate, and (C) by turning on

plate only after Placing the assembly on it to pmVide the SIOWCSt heating rate Th
. e

tem . . . . .
Perathl-c profiles corresponding to this series of experiments are given in Figures

3.9

Also to assess the effect of higher joining temperature, Specimens were

ref!
owed at the fastest heating rate bet to a higher peak temperature (330°C) Normal

ak .
pe tell‘lperature used during study was 280°C. In these heating rate experiments where

52

 

 

 

 

 

 

 

 

 

 

 

350 I I T l I I
5 E 5 —— Fastest cooling rate (1)
A _-----------L ............ . ............. : ________ -
U 300 I f E g -— Fast cooling rate (2)
‘3 ‘\ 5 i 3 —' "" Medium cooling rate (3) ________ ,.
a 250 ”"\"”:""“"""'E‘-----------J.
In \ s 5 : i ---- Slowest cooling rate (4)
= I ; - , - -; ........... —-
H 200 ------ ------------ - ---------------------
N I ;
s: 2 a
a. 150 ‘ ----------- 5"'*':'.';"_j;“_ """"
ca; 100 ------- \ --------- x ............. -------------------
a . g a.
50 ................... \__~“-—_—-—I—;_’
2
o J l
0 300 400

 

Time (see)

Figure 3.7 The temperature-time reflow profile with four different Coolin g rates (heati
rates remained the same for all reflow conditions). (1) Cooling cond' - ,
in water, (2) on aluminum chill block, (3) on fire brick, (4) and “10115 a

on w(J I

53

 

l‘ I‘fl‘vll

 

 

 

 

 

350 I E a

  
       
      
 

.Shnflowcr

:- I'sh 31’”
l‘ 3" ¢’. '

v
I

300 _ ..................... .............. Slowest ............. I

cooling rate

200

 

150

Temperature (oC)

 

 

’-
.
.....

100

 
 

 

Fastest
‘5— : '
.. ....g---.--.._...cooling-rate...-i ....................
é Sunflower: 3 E f
' shape "

 

 

o 100 200 300 400 500 coo
Time (sec) 7””

Egan: 3 ‘3 Temperature-time profiles of the fastest and the slowest Cooling rate,
Sunflower shape of IMC prevails around Ni particles after (a) the fasted
cooling rate, and (b) the slowest cooling rate. The cooling rate apparently
has no significant effect on IMC morphology.

S4

 

 

  
   
   
    

    

350 T m
I‘\ ' 8 i ------ . I #‘
300 _ 5.4.4-}. ______ : _ ‘7.” : : ”Fastestheatmgxate (5)
= 6 ' “ z" ' ----------- YT.M99£9a.9e.tipg.-rete.(§)._
250 _ 't- ........... ' fiSIMjestheating'r-ate (7)
221 . . E ' ".':'--'--'--'-Sam,eheating.rateas5.-__
zoo -- ---, ........ W! 3.153.935.1555!!! ...........
peak (8) ......... g ........... .. Tm

 

Temperature (°C)

 

 

 

 

Time (sec)

Figure 3 9 The tern ‘
. perature-tlrne profile with four different he ' 00‘ “g rates
. a \
relalalned the same for all reflow conditions). A5, Agng tai: (.c dicate areas
un emeath the heatln g part of these profiles at temperahland ‘1]: the melting
I'CS abo

point of the solder for heating rates 5, 6, and 7.

55

   

   
 
  
    

l.

300 .......... Fast‘hgatiiillggtgee. .6) E
59250 ;
Q)

5 200 ............
E 150
8.
E 100 ........
0
El 50 ..........................
0o 100 200 300 400 500

Time (sec)

5

................................................

ure (°C)
§

5‘.
a

5

Temperat

7. $.19??? bee

§

998. Feet?) .,

(°C)

§

3.
e

Temperature

0 100 200. 300
Time (sec) 500

Figure 3 10 Tom - ' f 168 associ t d ' '
. perature time pro 1 a e wrth microgra h _
intermetallic are shown (a) Sunflower shape for heating $0212: ofg‘NI-Sn
(b) mixed sunflower shape and single-crystal-faceted morpholo y and 8,
with the medium heating rate (6), and (c) Single-crystal- g OCCun'ed

fa
accompanied with the slowest heating rate (7). Ceted morphOIOgy

56

the heat-up rates were deliberately changed, the cooling rates were held the same by
cooling all the specimens used in this Part of the study on an aluminum chill block. The
corresponding temperature profiles are given in Figure 3.9. Attention is called to the
shaded features presented in Figure 3.9. The Shaded areas of reflow profiles designated
as A5, A6, and A7, are areas bounded by the heat-up curve between the melting and peak
temperatures of the solder and a vertical line from the peak temperature to the melting
temperature (T m) line. The areas, A5, A6, and A7, define the amount of heat irlp'»1t
involved in reflowin g the solder joint. It should be noted that a faSt heating rate to the
Peak reflow temperature gives the smallest area, A5, hence, a low heat input, While areas

be a
A6 and A7 represent medium and high heat min”, respectively. There appears to

. , ve and
direct correlation between the extent of the so-called “heat lnput” as defined abO

_ - goldfi-
the IMC morphology observed around the Ni reinforcements in the (toml’osfie

Whether “sunflower” or “blocky” IMC mOTPhOIOgieS prevails seems to depend on the
“cumulative” heat input as defined by the area 0f the heat up portion of the thermal
reflow profile.

(3) Multiple Reflow Experiments

Studies dealing with multiple-reflow thermal profiles Were carried 0" u .
fastest heating rate to the peak temperature of 280°C. The heat input designatedng the
was maintained for each reflow as Show“ in Figures 3'11 and 3-12- These Inn]: f5
reflow Studies were helpful to compare the MC morphological changes du: :
differences in amount of heat input.

Another set of multiple-reflow exPeriment was conducted by manipulating th
e

peak temperature vs. heating rate so that the heat input of each reflow C(1uals the 83
me

57

Temperature (°C)

 

 

Figure 3.11 Effect of reflow at 280°C with fastest heating rate on the
0

:ESunfloW 8.11 ape
. 'EAfter lst and
'E'Zn’d'reflow‘ -

|I
EH
.0
l:
:e
f!

 

Single-CrystaI-Faceftedsup.
After 3rd and 4th reflow '

 

 

L

40 60 so 100
Time (sec)

120 140

layers around Ni particles in Ni composite SOlder joint (a) 533:1 of 1MC

second reflow, (c) third reflow, and (d) fOUITh reflow A5 - ow

b , t t ak ' IS the ’ (b)
area a ove melting temper?l ure 0 pe temperature, 4A5 eat input
measured. The arrows indicate the microstructure asson and Were
reflow. ated with each

58

320

.......

Temperature (° C)

 

 

20

 

1st- ‘2»n$l>fp_‘é_l"!1,'_'.e.t.19.‘§ ,,,,,,, --------------
.. | ......

...“

lst lien0W (A5)i - Sunflower shape .
2nd§reflow (2A55) - Sunflower shape;
3rd fi'eflow (3A5) - Single-.crytal-facegted shaPe

 

 

60 so 100 120
Time (see)

Figure 3.12 Heat input area comparison of A5 and A6 (A63 3A5)

amOUnt 0f heat inpm represented by area A5. This I'CinW temper-attire profile

A9, Was

carefully designed and successfully achieved by heating the solder jOint t
0 Peak

temperature of 250°C under the slowest heating rate, as ShOWn in Figure 3.13.

Fast cooling rate was employed for both these multiple~reflow experiments b
y

cooling the solder joints on an aluminum chill block after each reflow,

59

320

 

Sunflhwer 81131:)e Single-crystal;

      

300 "

Temperature (°C)

 

Time (Sec) 140

Figure 3- 13 Effect of reflow at a peak temperature 0f 250°C with the
on the growth of IMC layf’JS around Ni Particles. The he
reflow is equivalent to heat 1nputw1th the reflow that un
rate (A5zA9). The heat input for the four Ieflows in
above the melting point is almOSt the same as that of the
( A724 A9). Arrows indicate IMC morphology observed.

SIOWest heating
d t input Of :3:
“Bone fast hea
terllperatune re
SIOWest heatin

tin g
giOn
g rate

(4) Heat to Routine Peak Temperature with Long Hold Time Experiment;

To evaluate the effect of long h01d time in the molten state, the reflow Prom“:
which is equivalent to heat input 31'63 as three times A5 (3A5) was achieved by keeping
the molten solser for 15 more seconds at 280°C with the fastest heating rate as shown in

Figure 3.14. Again, the cooling rate was maintained by cooling on the aluminum Chi"

block.

  

 

 

 

 

 

300
Total area = 3 A5
280 5 ;
A
U
a, 260 ...............................
i”.
= ,
H I
e .
a 240 ,,,,,,,,,,,,, -.§ ._
E 1
Q)
h
220
200 l J
0 10 20 3o 40

50

Time (sec)

Figure 3-14 Heat input approximately equivalent to 3A5 obtained With a
seconds at peak temperature—

time profile and the resnltam Sinhlold of 15
faceted morphology. g e‘c

l"ystal.

61

(5) Heat to Low Peak Temperature with Long Hold Time EXperimen ts

To study the influence of long hold time at just above melting tempemture,
experiments were carried out at a peak temperature of 230°C for 126 seconds and 225°C
for 258 seconds using the slowest heating rate as shown in Figure 3.15. The amount of
heat input of both thermal reflow pI‘OfileS was about 2 times of A5 (2A5). Again b0th
reflows used the same fast cooling rate.

Temperature profiles of all reflowS were obtai ned with a thermocouple attached to
the specimen fixture at a region very close to the solder joint.

Metallographic polishing was carried out after reflow at side surface of the ”ma

. . . . tron
lornts. The polished sides of solder joints were observed With a scanning 6‘“

cles aftef each set

microscope (SEM) to reveal the intermetallic formation around Ni parti

of reflow experiments.

3.5.2.3 Results and DiSCIISSiOB
It has been reported that the mOFPh0108Y 0f IMC layer is strongly influ d by
61206
heating and cooling rates [109], Thus, in order to evaluate the relative 1'
"fluence
of

heating and cooling rates on the IMC morphology around Ni Particles in th
e

00m .
SOidCr, Several different thermal I'Cfiow profiles were investigated as (1380 ' poslte
"bed .
. . In th
exPeflmental procedures. The rfisultant IMC morphologies obtained due t e
0 Various

reflow profiles are discussed as follows.

62

270

 

Temperature (“C)

 

 

 

 

     

150 200
Time (Sec) 25" 300

Figure 3 _ 15 Effect of 10w peak temperature and long hold times at this tern

C morphology. These two temperature profiles represent heat in
equivalent to two reflows with peak temperature of 250°C (2A9.~.A1
IMC mOrphology was observed after both these conditions,

perature 0n the
PM approximately
OzAl l). Sunflowfl

63

(1) Effect of Cooling Rates on IMC Morphology
Examination of the IMC layer using the reflow profiles shown in Figure 3.7
revealed that regardless of the cooling rate, the resultant IMC morphology observed Was
the “sunflower” shape. Figure 3.8, indeed, shows sunflower morphology was obtained
for fastest as well as slowest of cooling rates employed. The inset micrographs clearly
illustrate that cooling rate has virtually no effect on the resultant IMC morphology.
Therefore, it can be deduced that the cooling segment of the thermal reflow profile does
not significantly affect the morphological development of the IMC layer around the Ni
particles in the composite solder. Since the cooling segment of the reflow profile did not
have an apparent influence on the IMC morphology, more attention was focused on the
investigation of heating segment in the reflow profiles to ascertain the possible effects it
may have on the IMC morphology.
(2) Effect of Heating Rates on IMC Morphology
In contrast to the above findings that the cooling rate does not affect the IMC
morphology, two different morphologies of the IMC were observed around Ni particles
depending on the nature of the heating rate segment of the reflow curve, as shown in
Figure 3.10. A sunflower shape IMC morphology shown in Figure 3.10a was observed
with the fastest heating rate (reflow profile # 5 in Figure 3.9), and fastest heating rate
with higher peak temperature (reflow profile # 8 in Figure 3.9). Figure 3.1% illustrates a
mixture of both sunflower and single-crystal-faceted (blocky) morphology of IMC with
medium heating rate (reflow profile # 6 in Figure 3.9). Pure single-crystal-faceted
morphology is shown in Figure 3.10c and was observed under the slowest heating rate

(refl ow profile # 7 in Figure 3.9). Since the cooling segment of the reflow profiles in

 

Figure 3.9 and Figure 3.10 are the same, it is reasonable to suggest that it is the “heating A
segment” of these reflow profiles that influences the IMC morphology observed. When

the heat input, as defined earlier by the area related to the heating segment, is sufficiently

large, the preferred IMC morphology is “blocky” single-crystal-faceted in shape. When

the heat input is smaller, as is the case with a fast heating rate, the appearance of the IMC

morphology is sunflower. There must exist a critical heat input where the transitions of

the IMC morphology from sunflower shape to a blocky shape will occur.

(3) Comparison of IMC Morphology Obtained after Multiple-Reflows with
Those from Different Heating Rates

In an attempt to ascertain whether just temperature or amount of heat input

(represented by area under the heating segment of reflow profile curve above melting

 

temperature) is influencing the IMC morphology, multiple reflow studies were carried
out using the fastest heating rate to 280°C peak temperature followed by cooling on the
aluminum chill block (reflow profile # 5 in Figure 3.9). Although the initial morphology
is the sunflower shaped as shown in Figure 3.11a and b, the single—crystal-faceted
“blocky” morphology develops after about the third reflow.

The area underneath the heating segment curve can represent the heat input
received by solder during each reflow from the melting point (221°C) to peak
temperature (280°C). Figure 3.11 illustrates that a small heat input which normally gives

sunflower morphology will give a blocky IMC morphology if the cumulative heat input
as result of multiple reflows is at least equal to the critical transition heat input required
for blocky morphology. 3A5 in Figure 3.12 represents the critical heat input required to

achieve “blocky” IMC morphology. It is shown that the heat input of three such reflows

65

provides critical heat input reqllired to change the IMC morphology from sunflower to
single-crystal-faceted blocky shape.

To validate the 3A5 criterion for causing the change in IMC morphology, the
amount of heat input with medium heating (A6) was compared to that of three reflows (3-
A5, three fastest heating rate reflows), as shown in Figure 3.12. It was found that A6
which is less than 3A5 (but close to it) did not provide enough heat input to completely
change the sunflower shaped IMC morphology into the blocky single-crystal-faceted
morphology. However, the sunflower morphology tends to change into blocky single-
crystal-faceted morphology due to this heat input of A6. Since A6 is slightly smaller than
3A5, the transition is not complete as shown in Figure 3.10b. The cumulative heat input
represented by four reflows (4-A5 areas) was approximately equivalent to the heat input
of A7 in Figure 3.11 that results in the single-crystal-faceted IMC morphology.

These findings established that a critical amount of heat input is required for IMC
morphological change. It is also clear that path by which critical heat input is achieved is
immaterial as long as this critical value is reached cumulatively. In Figure 3.13, a reflow
profile to a peak temperature of 250°C was used. Again, when the cumulative areas of
A9 are greater than the critical heat input, single-crystal-faceted IMC morphology was
observed.

(4) Routine Peak Temperature with Hold Time

The reflow profile shown in Figure 3.14 again confirms that as long as the area

under the heating segment is equivalent to the critical heat input, the IMC morphology
Will be altered. T emperature-time profile and micrographs for this routine peak

temperature (280°C) with long hold time, equivalent to 3-reflows, is shown in Figure 3.14.

66

 

 

‘u |

Since again the single-crystal-faceted shape was found to occur, and it corresponded well
with 3A5 criterion in this study.

(5) Low Peak Temperature with Long Hold Time

In order to evaluate the effect of low temperature with longer hold time, the heat
input profiles corresponding to heating to 225°C for 258 seconds, and to 230°C for 126
seconds, were imposed as illustrated in Figure 3.15. The heat input of both reflows was
approximately the same heat input provided by the sum of 2A5 areas (i.e., 2A5 z A10 z
Al 1). Both reflow profiles gave the sunflower IMC morphology as shown in Figure 3.15.
Such observations once again confirming that there exists a critical heat input to facilitate
the development of blocky single-crystal-faceted morphology.

Presently, it is unclear as to the precise nature and mechanisms that influence the
morphological change and patterning of the IMC formed around the Ni particles
reinforcements in the composite solder. However, it is well understood from this study
that the heat input of the heating segment of the reflow profile must reach some critical
value before the IMC morphology changes from “sunflower” to “blocky” single-crystal-
faceted shape.

Albeit speculative, a final comment related to the formation and growth of IMC is
offered. From results of this study, it can be observed that a scallop shaped IMC was
possibly formed initially and then grew to the sunflower morphology. Under the slowest
heating rate, the channel between each “sunflower petal” would provide an easy path for
the interdiffusion between Sn, Ni and Cu species, which resulted in a morphology change

to single-crystal-faceted shape in a gradual manner with sufficient heat input. Longer

SOldering time and high temperature holds in the molten state also causes generally

67

 

 

thickened IMC and a gradual morphological change from the initial shape by a similar
mechanism.

The salient findings of this investigation are given in Table 3.2. The results
presented in this table strongly suggested that amount of heat input in the heating part of
the reflow at temperatures above the melting point of the solder has significant influence

in the development of the IMC morphology.

Table 3.2 Results Based on the Study on IMC Morphology

 

Area iii 3:221; 1:; used Sunflower shaped IMC
A5 and A9 x]
2A5 and 2A9 \]
3A5 and 3A9
4A5 and 4A9
A6 (Medium heating) \/
A9 (Slowest heating)
3A5 (Fastest heating rate
with 15 sec hold time)
A10 (Heat at 230°C for \I
126 sec)
All (Heat at 225°C for \j
258 sec)

Single-crystal-faceted
shaped IMC

 

 

 

 

 

 

 

4 44.44

 

 

 

 

 

 

 

3.6 Summary

1. Extensive formation of scallop-shaped Cu-Sn IM layers around the Cu
reinforcements was observed, whereas the formation of Ag-Sn around the Ag
reinforcement was minimal in the as-fabricated composite solder joints. The IM
layers developed around the Ni particles were also significant and can be

characterized by a “sunburst” pattern surrounding the Ni particle reinforcements. A

68

significant number of ~0.5 um sized small 1M particles were observed scattering,
mostly, along periphery of Ni reinforcement particles.

The morphology of IM layer around Ni particles was affected due to different
heating rate, but not cooling rate. Sufficient heat input to cause a “blocky” single-
crystal-faceted shape IMC morphology can be achieved by multiple reflows of the
solder provided the cumulative heat input reaches the critical value.

The [M layer around the Cu reinforcement was a co-layer consisting of a CU3Sn layer
close to Cu particles and a Cu68n5 layer close to the solder. The [Ni layer formed
around Ag particles was thin Ag3Sn layer. The IM layers surrounding the Ni
reinforcements were detected as a Cu-Ni-Sn ternary IM layer.

The [M layers formed at the Cu substrate/solder interface were found to be Cu-Sn 1M
layer in both the Cu and the Ag particle reinforced composite solder joints. This Cu-
Sn IM layer is also a co-layer consisting of a Cu3Sn layer close to Cu substrate and a
Cu68n5 layer close to the solder. However, the IM layers formed at the Cu
substrate/solder interface in the Ni composite solder joints was determined as a co-
layer consisting of a Cu—Sn IM layer close to Cu substrate and a ternary Cu-Ni-Sn
IM layer close to Ni composite solder.

The thick IM layer formed around the Cu or Ni particles is due to the fast diffusion
behavior of Cu or Ni atoms in Sn, whereas the diffusion of Ag in Sn is much slower

which results in a significantly thinner Ag-Sn intermetallics around Ag particles.

69

 

CHAPTER IV
MICROSTRUCTURAL EVOLUTION IN Cu, Ag, AND Ni PARTICLE
REINFORCED COMPOSITE SOLDER JOINTS UNDER ISOTHERMAL AGING
AT 150°C

4.1 Introduction [95]

When two metals parts are joined by solder, a metallic continuity is established as
the result of two interfaces that form where the solder is bonded to both metal parts [113].
This metallic continuity, or joining interface, is referred to as the intermetallic (IM) layer.
Intermetallic compounds grow at the interface of the solder and the substrate during long
term storage at ambient temperatures and more rapidly at high temperatures [114, 115].

The effect of IM growth within solder joints is not clearly understood. While the
presence of IM compounds are an indication that a good metallurgical bond has formed,
the fact that these compounds are brittle may also make them deleterious to a joint’s
mechanical integrity. When these compounds form as continuous layers at the
solder/substrate interface, the intermetallics can interrupt electrical currents with their
high resistivity, effectively isolating the metals that were to be electrically joined [116].
In addition, if these 1M compounds become too thick, the reliability of the joint can be
jeopardized due to cracking. This will be a problem especially if the joint is exposed to
any mechanical forces, such as expansion or contraction of the printed wiring board
(PWB) laminate caused by variations in temperature [115]. Exposure of these IM
compounds to the atmosphere via microtunnels formed through the solder during the
tinning or aging process has also been recently postulated as allowing for the

environmental degradation of the intermetallics located on the surface of the base metal

70

 

 

[117]. The interdiffusion processes, which produce the intermetallic layers, can also
produce Kirkendall voids which can also degrade the mechanical properties of the
connection [94].

The intermetallic is more brittle than the bulk solder, and one might predict that
failure therefore would occur at the IM layer. However, the intermetallic does not play a
role in the failure of a solder joint. This point has been proved by many researchers. For
example, Morris et al. [118] has shown that low-cycle thermomechanical fatigue and
ultimate failure of a properly soldered solder joint occurs though the bulk of the solder.
However, if the solder joint is subjected to extended periods of high temperature or long
term aging even at ambient temperature, the IM layer formed in the solder joint will grow
excessively which will degrade the solder joint due to its brittle nature and the voids
formed in the joint. In the composite solder joint with mechanically incorporated .
reinforcements, the 1M layers are formed not only at the solder/substrate interface, but
also around the reinforcement particles. Therefore, it is very important to study the
formation and growth of these IM layers in order to predict the lifetime of the solder joint
under service conditions.

The objective of the aging study in this chapter is to investigate whether
mechanically added Cu, Ag, or Ni particles would alter the microstructure as well as the
aging characteristics of eutectic Sn-3.5Ag solder, especially in small solder joints that are
more representative of solder joints used in microelectronics. This work is focused on the
microstructural characterization of mechanically-introduced Cu, Ag, and Ni particle
reinforced composite solders. The effects of isothermal aging on (i) substrate/solder and

(ii) particle/solder IM interfaces and on the overall microstructure are also examined.

71

Possible mechanisms are proposed for IM layer growth due to solid-state isothermal

aging of solder joints at 150°C.

4.2 Experimental Procedures

Aging experiments were performed by placing the solder joints on an aluminium
plate maintained at a constant temperature for up to 1000 hours. The aging temperature
used in the experiments were room temperature (25°C), 50°C, 100°C, and 150°C for the
Ni composite solder joints, and 150°C only for the Cu and Ag composite solder joints.
The polished solder joints were slightly repolished to remove the oxide layer at aging
times Of 100, 500, and 1000 hours to reveal the change in microstructure. SEM
micrographs were taken before and at set aging time intervals during the experiment.
Particular attention was given to the investigation of the formation and growth of 1M
layers around the reinforcement particles and at the copper substrate-solder interface.
Each of these studies was carried out by monitoring the same targeted area of the solder
joint to provide a reliable measurement of the changes that took place during the aging

process.

4.3 Microstructural Evolution in Cu Particle Reinforced Composite Solder Joints
4.3.1 Intermetallic Layer around Cu Particles
The effect of isothermal aging at 150°C on the IM layer growth around the Cu
particles in the Cu composite solder joint is illustrated in Figure 4.1. Note that the
l'ITchrograph series of Figure 4.1 represents the same field of view imaged at different

aging times. Both Cu68n5 and Cugsn intermetallics grew rapidly with isothermal aging.

72

.I:_C-.iu-SD. Intermetall'
yer .

..
a
u

Q

 

(e) (d)

Fi gure 4.1 Effect of isothermal aging at 150°C on the growth of Cu-Sn intermetallic layer
around Cu particles in Cu reinforced composite solder joint: (a) as-fabricated,
(b) after 100 hours, (c) after 500 hours, (d) after 1000 hours.

73

After aging for 100 hours, the layer had grown to nearly twice its initial thickness.
Growth of e and n intermetallics is quite apparent, where the volume of the Cu-Sn IM
layer formed is ~ 1.4 times the initial volume of the Cu particle before aging (cf., Figures.
4.1a and 4.1b). With continued aging (Figure 4.1d), the Cu particles were completely
consumed and converted to 8 and n intermetallics within 1000 hours. A thin layer of
Cu3Sn adjacent to the Cu substrate (Cu particle) is shown in Figure 4.1(c) after aging for
500 h at 150°C. The contrast of the 8 phase appears darker than the 1] phase in SEM
secondary electron and backseattered electron images. Also, with aging, the extent of
porosity increases between the Cu3Sn IM layer and the Cu particles.

Aging effects on the evolution of Cu-Sn intermetallics around Cu particles and Cu
particle consumption during aging are depicted graphically in Figure 4.2. There was an
initial rapid growth rate of Cu-Sn IM and expeditious consumption of Cu particles during
the first 100 h of aging. With subsequent aging, the growth rate of Cu—Sn intermetallics
and the consumption rate of Cu particles were slower. IM layer growth around the Cu
particle fitted a parabolic equation, i.e., the layer growth exponent was 0.5, suggesting

that the IM layer growth is diffusion controlled [119].

4.3.2 Intermetallic Layer at the Cu Substrate/Solder Interface

The influence of isothermal aging at 150°C on the IM layer growth at the Cu
substrate/solder interface in the Cu composite joints is presented in Figure 4.3. The
initial layer thickness of 1.8 pm grew to 7.8 pm after 1000 hours (Figures 4.3a—4.3d). The
coarsening of matrix Ag3Sn intermetallics was dramatic and is clearly shown when

comparing Figure 4.3(a) and (d).

74

 

   
 

 

 

 

9 P- 1 fT T T fit I I I I 1 I I I f hi j T I l I I I I I if 4
t g 5 5 .41 :
r i i 0' i "i
8 — ----------------------------------------------------------- q --------------- .t----_’4 -------------------- —~
A L‘ : 0 D

8 - r A. 1
_ l i ' i i
8 L. - i 0' ' 4

. ._.---------.---'. .......................................... 1---- f .....................................
g 7 : : r J." j
: a i ‘
3 I : ‘ : 2
g 6 .... ............ .2. ............ r ................... 0' g ..................... .. ............... I. ............. ..r
o 5 I 4 —e—Cu Consumption _:
§ C 5 --a--Cu-Sn evolution aoundCu 4
§ : E i ' o 'i
< 4 _.------------; ...... at. ................... . ........... . ............................................ 1
r : I : i
b r- : . I -1
r- , ' ' "‘
g 3 L' .......................... , ............. J
_ ; l ' ..
E - : l .
.9 ~ : I ~
.1: h ' I -.
r— 2 _— ------------ iii ------------- . --------------------------------------------- . ----------- . ------------- j
1 '- r r r i 41 r r i r r r i r r r i r r r l r r r L g r r J

 

 

 

Time (hours)

Figure 4.2 Effect of isothermal aging on the growth of Cu-Sn intermetallic layer around
the Cu particles and Cu particle consumption.

75

. lntcirfacial
intermetallic
layer

x2 28K13.6um. ’

Cu68n5
lay er’

(halite?)

‘3 3’ layer;
(dagger)

P‘E.E8l’ 13.6un )2. 28K 13.6w»

 

(C) ((1)

Figure 4.3 Effect of isothermal aging at 150°C on the growth of Cu-Sn intermetallic layer
at Cu substrate-solder interface in Cu reinforced composite solder joint: (a) as-
fabricated, layer thickness: 1.81 pm, (b) after 100 hours, (c) after 500 hours,
((1) after 1000 hours, layer thickness =7.76p.m. The development of Kirkendall
voids is clearly evident between the Cu substrate/1M layer after long aging
times in (c) and (d).

76

4.3.3 Mechanism for IM Layer Growth During Solid-state Isothermal Aging

Formation of IM compound layers is inherent to soldering. It is of great importance
to understand how such IM layers evolve in solder joints, particularly since the IM layer
is considered to be a potential source of joint failure due to its brittleness and the
tendency to grow excessively during hi gh-temperature aging.

We describe a phenomenological model for Cu-Sn IM layer growth inspired by
experimental results obtained during careful monitoring of the IM layer growth around
several Cu particles in a composite solder joint isothermally—aged at 150°C for up to 1000
h. From earlier work by Wu et al. [42], it is suggested that tin is the predominant
diffusing species that controlled formation and growth of the Cu-Sn 1M layer. During
reflow Sn is perhaps the dominant diffusing species contributing to IM layer growth.
However, during solid-state aging, it is apparent that interdiffusion of both Sn and Cu
through pre-existing IM layers is responsible for growth. Initial soldering produces layers
of both Cu68n5 and Cu3Sn, although the Cu3Sn layer may be barely detectable. The 11-
phase forms adjacent to the solder and the e-phase forms adjacent to the copper substrate.

The Cu-Sn co-layers thicken by the movement of the various interfaces in both
directions normal to the layer surfaces. Interfacial motion occurs both by inward
movement towards the Cu substrate and by outward movement towards the SnAg solder
matrix. It is proposed that Sn and Cu interdiffuse through 11 and 8 phases and the
following reactions occur to promote growth of the IM co-layers upon aging (cf., Figure
4.4):

1. Formation of Cu68n5 occurs by the reaction of Sn with Cu3Sn at the UT] interface as a

consequence of Sn diffusing through the n-phase.

77

3Sn + 2 Cu3Sn ==> Cu6Sn5 (4-1)

This reaction results in growth of the n-phase by movement of the £111 interface
towards the Cu substrate. Consequently, diminution of CU3SI'I (8) occurs.

2. Formation of more Cugsn results due to the reaction of Sn with Cu at the Cu/e
interface as Sn diffuses through the n-phase.
Sn + 3Cu ==> Cu3Sn (4-2)

As a result of this reaction, growth of the e-phase occurs and dissolution of the Cu
substrate is observed. The Gale interface moves into the Cu substrate.

3. Further thickening of the n-phase (Cu68n5) takes place as Cu atoms diffuse through
both 8 and 1] phases to react with Sn at the n/Sn interface.
6Cu + SSn ==> Cu6Sn5 (4-3)

The reaction produces growth of the n-phase into the SnAg solder matrix, with
consequent movement of the n/Sn interface to the right.

Another observation is that without variance Kirkendall voids occur at the Cult:
interface. Kirkendall voids were seen after prolonged isothermal aging at 150°C. The
voids were readily observed after 100 h of aging (see Figures. 4.1, 4.3, and 4.4b). Such
findings suggest that net mass flow of Cu through the 8- and n—phase is greater than mass
flow of Sn in the opposite direction. Kirkendall porosity was however never observed at

the n/Sn interface.

78

 

Cu Substrate fe'ph "'phase 1 Sn-Ag Solder

  
 

./ 1:-

interface

(a) Shematic of IM layers prior to isthermal aging.

Sn

 

'— Kirkendall voids
(b) Growth of e and 1] IM layers during isothermal aging at 150°C.

 

 

 

Proposed Reaction:

(a) 3Sn + 2Cu3$n => Cu(,Sn5
(reaction at the alt] interface - conversion of C0331! to Cu68n5)

(b) Sn + 3Cu :> Cu3Sn
(reaction at the Cult: interface — conversion of to Cu38n)

(c) 6Cu + SSn 2 Cu68n5
(reaction at the n/Sn interface - formation of Cu6Sn5)

Figure 4.4 Proposed mechanism for Cu-Sn intermetallic layer growth during aging for a
Cu/Sn solder diffusion couple.

79

4.4 Microstructural Evolution in Ag Particle Reinforced Composite Solder Joints

The effect of isothermal aging at 150°C on the IM layer growth around the Ag
particles in the Ag composite solder joint is shown in Figure 4.5. In sharp contrast to the
Cu reinforced composite solder joint, Ag particles were only “partially” converted to
Ag38n intermetallics after 1000 hours aging. The thickness of Ag3Sn IM layer around the
Ag particles did not increase significantly with aging in comparison with Cu68n5/CU3Sn
IM layer around the Cu particles. This phenomenon indicates that during solid-state
isothermal aging at 150°C, interdiffusion of Ag and Sn through the AgSn N layer is
orders of magnitude slower in comparison to Cu and Sn through the Cu—Sn IM layer.

The influence of isothermal aging at 150°C on the IM layer growth at the Cu
substrate/solder interface in the Ag composite solder joints is presented in Figure 4.6. As
stated in 4.3.2, in Cu reinforced composite solder joints (Figures 4.3a-4.3d), the initial
layer thickness of 1.8 pm grew to 7.8 pm after 1000 hours, while in the Ag reinforced
composite solder joint (Figures 4.6a-4.6d) the initial layer thickness of 1.6 pm grew up to
7 pm after 1000 hours. The growth rates are quite similar for the two composite solder
joints, with both exhibiting a final Cu-substrate/solder interfacial IM layer ~ 4.3 times the
initial layer thickness after aging for 1000 hours. A comparison of the aging effect on
interfacial layer growth between Cu and Ag particle reinforced composite solder joints is
shown in Figure 4.7. As illustrated in previous micrographs, the overall thickness of the
interfacial 1M layer that evolves in Cu composite solder joints is greater than in Ag
composite solder joint after aging for 1000 hours. However, the growth rates of

interfacial layers in both Cu and Ag composite solders are comparable.

80

J...

...
ix at fi ..
D a

.1.
flu...
r

 

)

(d

(C)

Figure 4.5 Effect of isothermal aging at 150°C on the growth of Aggsn intermetallic layer

-fabricated,

(c) after 500 hours, ((1) after 1000 hours.

’

around Ag particles in Ag reinforced composite solder joint: (a) as

(b) after 100 hours

81

(2.88K 13.6um

 

(C) (d)

Figure 4.6 Effect of isothermal aging at 150°C on the growth of Cu-Sn intermetallic layer
at Cu substrate-solder interface in Ag reinforced composite solder joint: (3)
as-fabricated, layer thickness: 1.59 pun, (b) after 100 hours, (c) after 500
hours, (d) after 1000 hours, layer thickness =6.95p.m.

82

 

 

8 frli fi"
l- .
O
l- ; . i
O
>- i g 'i
u— : . 4
7— ------------------------------ . ------------------------------- ET ------------ —
I- I ‘ a-I
' 1
b- : . -l
I- I .1
I
l- : -4
67— -------------- L -------------------------------------------- *n ------------- L --------------- v -------------- -—-I
i
.

    

 

""" """"""" —O—lntermetallic in Cu Composite
‘ --B- - lntermetdlic in Ag Composite

 

Thickness or Average Diameter (m'cron)
h
7

 

 

 

Time (hairs)

Figure 4.7 Effect of isothermal aging on the intermetallic layer growth at the copper
substrate/ solder interface.

83

4.5 Microstructural Evolution in Ni Particle Reinforced Composite Solder Joints

Isothermal aging studies were performed using single shear lap solder joint
specimens at different aging temperatures (25, 50, 100, and 150°C) for up to 1000 hours.
In this aging study, it was revealed that significant microstructural changes only occurred
in Ni reinforced composite solder joints aged at temperatures above 100°C. Therefore,
detail discussions will be presented only for isothermal aging at 150°C where profuse
microstructural changes were observed. However, SEM micrographs showing the
microstructure of joints aged at 25, 50, 100°C are presented in Figure 4.8(a)-(d), Figure
4.9(a)-(d), and Figure 4.10(a)-(d), respectively. Figures 4.8, 4.9, and 4.10 of targeted
areas of the solder joint are shown mainly to illustrate that the microstructure was
extremely stable under isothermal aging conditions below 100°C for 1000 hours. The

most noticeable change in the microstructure during isothermal aging below 100°C was

coarsening of Ag3Sn particles.

4.5.1 IM Layer around the Ni Particle Reinforcements

The effect of isothermal aging at 150°C on the IM layer growth around the Ni
particles in the Ni composite solder joint is illustrated in Figure 4.11. This series of
micrographs represent the same field of view (targeted area) imaged at different aging
times. The Cu-Ni-Sn intermetallics around the Ni particle reinforcements grew very
rapidly with isothermal aging at 150°C. After aging for 100 hours (Figure 4.11b), the IM
layer had grown to nearly four times its initial thickness. Some smaller Ni particles were
completely converted to Cu-Ni-Sn intermetallics. The IM layer continued to grow

significantly and even the larger Ni particles were completely consumed and converted to

84

- Cu-N_i-Sn
Cu-Ni-Sn . ' IM Layer
IM Layer '

.‘r ~ » Cu-Ni-Sn
Cu-NI-Sn ' g . IM Layer

IM Layer , .
/‘ _ a. _ /

 

(c) ((1)

Figure 4.8. Effect of isothermal aging at room temperature(25°C) on the growth of Cu-
Ni-Sn intermetallic layer around Ni particles in Ni reinforced composite
solder joint. (a) as fabricated, (b) after 100 hours, (c) after 500 hours, (d) after
1000 hours. Note: No growth of the IM layer with aging time at 25°C.

85

j ‘ Cu-Ni-Sn
_’", IM Layer

. V"5._wbu:Ni-Sn
‘ '. » IM Layer

9.
.. Ce-Ni-Sn
." IM Layer_ -
' ‘7 .

" .

 

(c) (d)

Figure 4.9 Effect of isothermal aging at 50°C on the growth of Cu-Ni-Sn intermetallic
layer around Ni particles in Ni reinforced composite solder joint. (a) as
fabricated, (b) after 100 hours, (c) after 500 hours, ((1) after 1000 hours. Note:
No growth of the IM layer with aging time at 50°C.

86

Cu-Ni-S‘n’h ‘
IM Layer r.‘ ’

A

Cu-Ni-Sn
IM Layer

Cu-Ni-Sn
IM Layer

\; Ni

 

(C) (d)

Figure 4.10 Effect of isothermal aging at 100°C on the growth of Cu—Ni-Sn intermetallic
layer around Ni particles in Ni reinforced composite solder joint. (a) as
fabricated, (b) after 100 hours, (c) after 500 hours, (d) after 1000 hours.
Note: Insignificant growth of the IM layer with aging time at 100°C.

87

V ".r . '.
-'V. H‘ ' )_ _ ,-
‘ 1r 2": " -
/ _,,, '1’ 1'! _ ~'

, Cu-ili-SnlM Layer V

‘ ' 7 Cu-Ni43nlMLayér‘ : " / \
. ‘_ .. ._ , ‘ ' . a. ;

'Cu-Ni-Sn Intermetalliés
. / ,,

r I, )/ if
g, .’/ .‘ r‘ . _' ’\_

J1.»
..,, Cu-Ni-Sn Inlerrnelallics l” ,
w J ‘I . \

= r

 

(e) ((1)

Figure 4.11 Effect of isothermal aging at 150°C on the growth of Cu-Ni-Sn intermetallic
layer around Ni particles in Ni reinforced composite solder joint. (a) as
fabricated, (b) after 100 hours, (0) after 500 hours, (d) after 1000 hours.
Note: Profuse growth of the IM layer with aging time at 150°C.

88

Cu-Ni-Sn ternary intermetallics after aging for 500 hours, as shown in Figure 4.11(c).
IM layer that formed around Ni particles grew and developed a faceted morphology after
500 hours. After 500 hours, the volume fraction of the intermetallics represented ~ 60%
of the total volume of the solder joint material. No significant growth of the
intermetallics was observed between the aging time of 500 to 1000 hours. Slight growth
and some rearrangement of the intermetallics were noted in the solder matrix, as shown
in Figure 4.11(d). Coarsening of eutectic Ag3Sn particles and an increase in the amount
of porosity in the solder matrix are evident with aging when comparing Figure 4.11(a)
and 4.11(d). 1M layer growth and Ni particle consumption are clearly illustrated in
Figure 4.12(a)-(d) where the targeted area followed throughout the aging process was an
individual Ni reinforcement particle.

Effect of aging at 150°C on the evolution of Cu—Ni-Sn 1M layer around Ni particles
is illustrated graphically in Figure 4.13(a) and compared with previously reported Cu
particle reinforced composite solder. It is evident that the grth rate and thus the net
thickness of the IM layers around Ni particle reinforcements are much higher than those
in Cu composite solder joint. The IM layer formed around the Ni particles was ~13 times
thicker than its initial layer thickness after aging for 1000 hours, while the IM layer
formed around the Cu particles was only 5 times thicker than its initial layer thickness
under the same aging condition. It is also shown that the 1M layer around the Ni particles
did not show any significant growth when aged at temperatures lower than 100°C. The
lack of IM growth suggests that Ni reinforced composite solder joints would be desirable

in industrial applications where the temperature environment is below 100°C. The

consumption of Ni and Cu reinforcement particles during aging at 150°C is shown in

89

, Cu-NiTSnlntermetallics ' :l' Cu-Ni—Snlntermetallics

' . S 44“,“: ‘- " ‘3 \

Cu-Ni-Sn lntermetallics

 

.r“,.,/ 10am

(6) ((1)

Figure 4.12 Effect of isothermal aging at 150°C on the growth of Cu-Ni-Sn intermetallic
layer around one Ni particle. (a) as fabricated, (b) after 100 hours, (c) after
500 hours, ((1) after 1000 hours. Note: profuse IM layer growth and
consumption of Ni reinforcement particle.

 

 

 

  
   
 

 

 

 

 

 

 

 

 

 

 

 

 

20 r r T 1 T r fir L r f 1 I r r 1 T —r r r T r r r I f r
r ‘ ' E i
i —‘—m “150°C : .
E E E A-‘P
“'— Cu «150°C 5 : . 5 i
-15 - “’NI-novc ______________ é ............ _‘
E . --~-- ° - i
:1. "Int 50 C : :
v ” -- -- o ' , a
I + NIOIZS C . :
. u- - - 4
' ~ ' - ' i
5 I E
2 1o ._ ...................................................................... :3 .............. : ............. _
1: r : : ’ r
’- r- "I ‘0 -4
. ‘I .
__ . I ‘/ I I J
3 A, or?" E E
’ i ‘ 3 E E ‘ a
5 _ ____________ , ............. ..., ...... , ............... . ......... ............... i ............. _ < >
o 1 l 1 1 l l l l l l 1 l_L_J #1 l I 1 _1 J. l 1 1
-200 0 200 400 600 800 1000 1200
Time (hours)
E
5
b
o
‘6
E —°—Normalized Ni Consumption
a -B- Normalized Cu Consumption
o
E
t
a
a.
E
r:
2
fl
0
a
2
r: (b)
'o
o
.2
T:
E - .
3 ’ L 1 q
z -0.2 l l l l l l l 1 l l l 1 l _L 1 l l l l l 1 L 1 l 1
-200 0 200 400 600 800 1000 1200

Time (hours)

Figure 4.13 Effect of isothermal aging at 150°C on the growth of 1M layers around the Ni
particles and Ni particle consumption. Data for Cu composite solder and Ni
composite solder aged at 100°C, 50°C, and room temperature are plotted for
comparison. (21) IM layer evolution around the particle reinforcement, (b)
reinforcement particle consumption.

91

Figure 4.13(b). Much more expeditious dissolution rate of Ni particles was noted than of
Cu particles during the first 100 hours aging. With subsequent aging, the dissolution rate
for both Ni particles and Cu particles were similar, resulting in complete or near complete

conversion of the reinforcement particles into intermetallics.

4.5.2 IM Layer at the Cu Substrate/Solder Interface

The influence of isothermal aging at 150°C on the IM layer growth at the Cu
substrate Isolder interface in the Ni composite joint is presented in Figure 4.14. The
initial layer thickness of ~4 tun grew to 56 um after 1000 hours, over an order of
magnitude greater than the initial IM layer thickness. Note that the micrographs in
Figure 4.14 were still chosen from the same targeted area of the solder joint interface, but,
in order to fit the breath of the [M layer into one micrograph, the magnification was
lowered in Figure 4.14(c) and (d). The relative interfacial IM layer growth is presented
in Figure 4.15(a). The growth rate of the interfacial 1M layer in Ni composite solder joint
is much faster than that in .Cu and Ag particle reinforced composite solder where the final
interfacial 1M layer thickness was only about 4 times the initial IM layer thickness. It is
also shown in the figure that isothermal aging at 100°C, or at lower temperatures, did not
cause significant modifications in the microstructure at the Cu substrate/solder interface
in terms of IM layer thickness. As stated previously, the interfacial IM layer is a co-layer
of Cu-Ni-Sn layer adjacent to solder and Cu-Sn layer adjacent to Cu substrate, which is
composed of a Cu(,Sn5 and a Cu38n IM layer. Significant growth of CU3Sn layer was
observed during aging, as seen in Figure 4.l4(a)~(d). The thickness of the Cu3Sn layer

increased from 0.5um in the as-fabricated condition to 12 um after aging for 1000 hours,

92

Cu

Substrate V .
.. I; Cu-Nr—Sn

'4) \IM Layer A

I

,=
f“ fv

Cu-Sn H _ .3 l .' ; .;A\‘ 4

(close to Cu)
IM Layer . , 0
j . flame

Substrate

9.51;,“

CU3Sn ; / cussns ‘c ‘
IM Layer _ “ Cu3Sn ‘

IM Layer

 

_ 50pm '3
_.,___

4 I-

(C) (d)
Figure 4.14 Effect of isothermal aging at 150°C on the growth of 1M layer at the Cu

substrate/solder interface. (a) as fabricated, layer thickness=4 mm, (b) after
100 hours, (c) after 500 hours, ((1) after 1000 hours, layer thickness=56 um.

93

 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

I I I I I I I I I I I I I I I I I I I rI I I I I I r1 I I rr I I I Ij’ I I I I I I I I I I I I I I I I I I I IT I I I I I I I Tr‘r
__ lM layer in Ni Composite (150°C)
‘5_ IM lsyerln Cu Composite (156C)
3 "° '- IM layer in A9 Composite (150C)
2 1o " ’ " " ' ' IM layer in Ni Composite (106C) “
f, ” ° ' +' ' m layer in Ni Composite (50°C) ‘
E r - fi— lH layer in Ni Composite (RT) I
l- i «
b
O L ..
>.
I
J b -1
E
3
0
a t r
'l:
O
on
E
. l- .1
Z
‘3
z (a)
I
1 C L
1 1 L L l l l l l L L 1 l L L L L L L L L L L l L L l l 1 L I l J 1 L L l l l L L L L l L L L l l L l L L l L l l L I l l L L Li L L L 1
-200 0 200 400 600 800 1000 1200
Tlme (hour)
14 r f r I r r r l r r r I f 1 I r r r T r r 1 T If r r
l- ; . .
r ; «
i- Y -1
12 h.........................—.-.....a......................-......-..--.-. s..-.......; .............. _
l— , -<
A >- -<
E r- 4
5 10 h— -------------------------------------------------------------------------- ‘- ............................. j
C '- .
: - . -. .
5 ‘ —0—Cu38n thickness in Ni Composite ‘
g 8 F -a—— CuJSn thickness in Cu Composite 4
I- ~ i
g r E E E -
> _. ........... t ............................................. z ............... A: ............... f ............. ...
a 6 _ : : : .
-' ~ I I E ,
5 _ : : : l
n ........................................................... Z ................ : .......... “A ............. ...
o >- "a ‘a ' 4
2 y— ------------------------------- /---..--'-.------.-.-...r ............................... I .............. J
o L L l L l L 1 L 1 L l l L L ;L L L
-200 0 200 400 600 800 1000 1200

Time (hours)

Figure 4.15 Effect of isothermal aging at 150°C on the growth of IM layers at the Cu

substrate! solder interface in the Ni composite solder joint. (a) Comparison of
interfacial 1M layer growth in Ni, Cu and Ag composite solder joint under
aging at 150°C. Interfacial 1le layer growth in the Ni composite solder joint
aged at 100°C and less is plotted as well for comparison. (b) Comparison of
Cu3Sn IM layer growth at the Cu substrsate/ solder interface in Ni and Cu

composite solder joint under aging at 150°C.

94

as depicted in Figure 4.15(b). The growth rate for Cu38n layers was also much faster in
Ni composite solder joint than that in Cu composite solder joint under similar aging
conditions. As is illustrated in Figure 4.15(b), only an ~ 4pm thick CU3Sn layer was
developed after 1000 hours aging at 150°C in Cu composite solder joint which had a

similar initial layer thickness as the Ni composite solder joint.

4.5.3 Possible Mechanism for [M layer Growth during Solid-State Isothermal Aging

In the previous studies on Cu and Ag particle reinforced composite solders, the
formation and growth of IM layers both around the reinforcements and at the Cu
substrate/solder interface exhibited diffusion controlled behavior under isothermal aging
conditions. However, diffusion data for these metallic particles in solder matrix
(primarily Sn) are not readily available at all temperature ranges. Table 3.1 [13, 108]
gives a comparison of diffusion behavior between Ni in Sn and Cu in Sn. The diffusing
distance for Ni in Sn parallel to its c-axis of the unit cell is ~21550 um after 100 hours at
25°C. In contrast, the diffusing distance for Cu in Sn parallel to its c-axis of the unit cell
is ~8500 pun under similar conditions. Ni diffuses 2.5 times faster in Sn parallel to its c-
axis than Cu at room temperature. Although we can expect that Ni diffuses much faster
than Cu in Sn parallel to its c-axis at 150°C, there are data showing that Cu diffuses 7
times faster than Ni in Sn perpendicular to its c-axis at 150°C. Ni and Cu are similar in
chemical characteristics, such as atomic number, similar atomic radius, and
electronegativity. Thus, their diffusion behavior in Sn is expected to be similar. The data
in Table 3.1 show that both Ni and Cu diffuse rapidly in Sn in the solid state. The

simultaneous presence of Ni and Cu in Sn-3.5Ag solder apparently accelerates the growth

95

of Cu-Ni-Sn intermetallics. Wu et al. [42] suggested that the presence of Ni in Sn-Pb
solder tended to suppress the e-phase (Cu3Sn) by raising the activation energy for layer
growth, whereas, the presence of Ni, alternatively, enhanced the growth of the n-phase
(Cu63n5) IM layer by lowering the activation energy for growth of the n-phase. The
presence of Ni and Cu in Sn seems to alter the thermodynamics of the Ni composite
solder such that profuse IM layer occurs, particularly, under solid state aging condition at
150°C and above. The Cu-Ni-Sn intermetallics in Ni composite solder is about 60%, as
compared to about 45% Cu-Sn intermetallics in Cu composite solder, after aging for 1000
hours at 150°C. This is probably due to the presence of Cu, in combination with Ni, in

the Ci-Ni-Sn intermetallics formed in the Ni composite solder.

4.6 Summary

1. Isothermal aging studies showed that more intermetallics formed in Cu composite
solder joints (both around the reinforcements and at the substrate/solder interface)
than in Ag composite solder joints after aging for 1000 hours.

2. Cu reinforcements were totally converted to Cu68n5 intermetallic with sufficient
aging, whereas Ag particles were only partially converted to Ag38n intermetallic.

3. A proposed phenomenological model shows that the growth of IM layers around the
Cu particle reinforcements and at the substrate/solder interface is produced by
interdiffusion of both Sn and Cu atoms through the 8 and 1] IM layers.

4. Solid state isothermal aging studies at 25, 50, and 100°C for ~1000 hours induced
only slight modifications in the microstructure of Ni composite solder joint. However,

aging at 150°C revealed that the microstructure was unstable with profuse growth of

96

the 1M layer and coarsening of the Ni reinforcement as a result of Cu-Ni-Sn
intermetallics formation and growth.

. The large volume fraction of intermetallics formed in the Ni composite solder joint
after aging at 150°C for 1000 hours is probably due to the presence of Cu, in addition

to Ni and Sn, in the intermetallics formed in the Ni composite solder.

97

CHAPTER V

EFFECT OF REFLOW ON THE SOLDERABILITY, NHCROSTRUCTURE AND
MECHANICAL PROPERTIES OF Cu, Ag, AND Ni PARTICLE REINFORCED
COMPOSITE SOLDERS

5.1 Introduction

Typically, the electronics industry involves multiple-pass solder operations in
which the solder is reflowed. In such operations, wettability of the solder materials is of
extreme importance. Investigations dealing with wettability issues of Pb-free solders
have been conducted [120-124] and ongoing research is moving at a rapid pace. Vianco
and others [120,121] have found the wetting characteristics of Sn-Ag based lead-free
solder to be comparable to leaded solder on copper substrate. The wettabilty of gold
substrate with eutectic Sn-Ag solder has been shown to be commensurate to that of Pb-Sn
solder alloys [6]. The wetting characteristics of any particular solder alloy are strongly
influenced by the flux used [122,124]. In particular, the wetting parameters, such as
interfacial tension and wetting rate, of eutectic Sn-Ag solder can be highly variable
depending on the choice of flux [122].

In this chapter, the effects of solder reflow on wettability characteristics,
microstructure and mechanical properties were investigated. Multiple-reflow
experiments were conducted using eutectic Sn-3.5Ag non—composite and eutectic Sn-
3.5Ag based composites with mechanically-added Cu, Ag, and Ni particles on a copper
substrate. The motivation to conduct this investigation on reflow characteristics of

composite solders was due to limited amount of published literatureon this subject and

98

this type of solder material. It is also believed that the composite approach to engineered

solder properties should be advanced [47].

5.2 Experimental Procedures
5.2.1 Materials

The composite solders used in this study nominally contained 15 volume percent
Cu, Ag, or Ni reinforcements. In order to study the effect of different volume fraction of
the reinforcing phase on the wettability of composite solders, Cu particle reinforced
composite solders with 6 and 12 volume percent Cu particles were also prepared and used
in the present study. Eutectic Sn-3.5Ag solder paste was used as a baseline for
comparison.

5.2.2 Wetting Experiments

The copper substrates with dimension of 2 cmx2 cm were chemically cleaned with
a solution of 50% nitric acid and 50% H20 before each experiment. Reflow specimens
were made first by preparing disc-shaped solder paste preform. This was accomplished
by flowing the solder paste through an orifice of 5 mm in diameter and then cut to size
using a razor blade. With this method, consistent size discs weighting approximately 0.3
grams were achieved for the subsequent melting and reflow process.

All the disc-shaped solder preforms were placed on copper substrates for melting
on a hot plate. Reflow experiments were carried out by remelting solder materials on the
same copper substrate for at least 3 times without refluxing to better mimic industrial
prictice. The thermal history profile for initial melting and subsequent reflow

experiments is shown in Figure 5.1. Specimens were heated to a maximum temperature

99

of 280°C and then placed on a chill aluminum block to cool to room temperature. Upon
solidification, all specimens exhibited a spherical-cap shape. As-melted and reflowed

samples were cleaned, sectioned, polished and examined using optical and scanning

 

300 I T I I I r I I 7 I T I 7 I I I I I I I I I f T r I I I
_ I I l I l I ' l
' . _ ' I

 

 

 

” 5 3 § 5 ——Tem erature I:
250 ._ ........... . ...................... : ............. i ....... p r

200 _
150

100

Temperature (Degree)

>-

 

 

 

 

-50 0 50 100 1 50 200 250 300 350

Time (sec)

Figure 5.1 Temperature profile of solder/copper wettability experiment.

electron microscopy to determine the wetting angles. Image processing software was
used to measure the wetting angles from the SEM micrographs. This technique for
determining contact angles provided consistent results similar to those previously
published [47].
5.2.3 Reflow Experiments on Solder Joints

Effects of reflow on the microstructure of Ni composite solder were studied using
solder joint specimens considering that the microstructure of bulk solder is significantly
different from the routinely-fabricated solder joint. Also, reflow of solder joints is a

common manufacturing process. Such reflow experiment will enable the assessment in

100

performance of composite solders. Reflow analysis was carried out by remelting the
solder joints in aluminum fixture on a hot plate for at least 3 reflows at 280°C. The
thermal history profile for the reflow experiment on the solder joints is exactly the same
as the profile for solder joint fabrication, as shown in Figure 3.2. The reflowed solder
joints were metallographically polished and microstructural features were examined
using SEM. In the SEM analysis of all these joint samples, particular interest was focused
on the formation and growth of IM layers around Cu and Ag particle reinforcements in

the composite solder and at the copper substrate/solder interface.

5.3 Effect of Reflow on Solderability of Cu, Ag, and Ni Particle Reinforced
Composite Solders

Results of wetting angles measured as a function of reflow conditions are shown in
Table 5.1.

Results in Table 5.1 indicate that eutectic Sn-3.5Ag solder paste has the best
wettability of all non-reflow solders with an average contact angle of about 105°. The
wetting angle determined in this study using solder paste preform was compared to that
determined using solid solder preform of a previous study [47]. The results of this study
indicate that the wetting angle of eutectic Sn-3.5Ag solder paste is ~8° smaller than the
eutectic Sn-3.5Ag solid solder preform, where the nominal wetting angle was typically
around 18°. The wetting angle for first reflowed composite solder with 15 v% Cu
particles was significantly higher, approximately 47°. However, composite solder with
comparable volume fraction of Ag reinforcements exhibited a wetting angle of about 21°.
Ni particle reinforced composite solder exhibited a wetting angle of 13° under the same

experimental conditions.

101

As indicated in Table 5.1, there is no statistically meaningful difference in the
wetting angles between first reflowed (as-melted) and multiple reflowed solder
materials.

Table 5.1. Wetting Angles as a Function of Reflow Conditions
(Angles Measured in Degrees)

 

Materials Conditions Anglel Angle 2 Angle 3 Angle4 Aver. 1* Aver. 2**
Reflow 1 13.6 8.1 9.5 10.6 10.45 10.512
Eutectic 311-35 Ag Reflow 2 11.4 7.9 6.1 10.9 9.08
(mm) Reflow 3 21.3 9.9 15 15.40 12113.9
Reflow 4 14.4 8.6 14.7 9.8 11.88
reflow 1 43.3 51.1 52.6 42.2 47.30 47.3:1:5.3
reflow 2 57.1 48.7 55.8 28.1 47.43
reflow 3 38.2 37.9 37 44.6 39.43 44.1i10
reflow 4 33.4 49.1 65 34.8 45.58
reflow 1 25.7 17.3 22.3 19.2 21.13 21.1132
reflow 2 19.4 22.6 12.1 11.4 16.38
reflow 3 23.5 21.0 19.3 9.7 18.38 18.5:46
reflow 4 21.4 23.1 14.7 23.9 20.78
reflow 1 12.5 14.8 12.2 12.3 12.95 13.0:tl.2
reflow 2 12.5 9.1 10.4 10.9 10.73
reflow 3 12.0 11.5 15.4 15.8 13.67 12.1:t1.5
reflow 4 12.3 12.8 13 10.0 12.03
reflow 1 18.2 17.8 19.2 16.8 18.00 18.0103

Eutectic Sn-3.5Ag reflow 2 22.5 18.9 20.7 20.7 20.70
(preform) reflow 3 25.3 23.7 19.7 21.3 22.50 19.6i3.4
reflow 4 16.4 15.7 15.3 14.6 15.50

* Aver. 1 is the average of all the angle measurement under each different
condition.

** Aver. 2 basically averages the angles in the multiple reflow experiment to
compare with the first reflow conditions.

 

 

 

 

 

 

 

Cu composite solder

 

 

 

 

Ag composite solder

 

 

 

 

Ni composite solder

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

The wettability ranking of these materials is shown in the last column of Table 5.1 and
schematically illustrated in Figure 5.2. The first reflowed and multiple reflowed solder

all produced spherical caps of the solder on the substrate.

102

It is contended that the eutectic Sn-3.5Ag solder paste has the best wettability due

in part to the way in which the solder paste is manufactured. The self-fluxed paste would

 
  
 

 

  

Cu
Composite
Solder

0: 47.3

 

 

 

    
  
 
 

 

Cu Substrate

 

 

I 0: 21.1

 

 

Ag Composite Solder
Cu Substrate ]

   

 

 

 

0: 12.5

 

 

 

 

Ni Composite SMI

u Substrate

 

 

0: 10.5

 

 

 

 

Sn-3.5Ag 301W

Cu Substrate I

 

 

Eutectic Sn-Ag Solder Preform: 9: 18.00

 

 

 

Figure 5.2 Schematic drawing of wetting properties of different solder materials observed
in the reflow experiment.

103

tend to prevent formation of metal oxides that can be detrimental to wettability. A rough
estimation of the amount of flux which encompasses the particles of the solder paste was
approximately 12 wt%. Apparently, fluidity is therefore enhanced giving a significant
lower contact angle.

As noted previously, the Cu composite solder showed the highest contact angle and
hence the least wettable. To ascertain the variation in contact angle with reinforcement
content, we conducted a collateral experiment where the volume fraction of mechanically
added Cu particle reinforcements ranged from 6 to 18 v%. For first reflowed Cu
composite solder, as shown in Figure 5.3, the wetting angle increased with the increase of
the volume fraction of Cu reinforcements. At the lowest volume fraction of 6 v%, the
wetting angle was 18° whereas at 18 v% the wetting angle increased to 47°. The
propensity for the Cu reinforcement to form Cu68n5 intermetallics apparently affects the
flow properties of this composite solder. From out microstructural analysis, we find that
~ 6 micron size Cu particle reinforcements will be converted to Cu68n5 intermetallics
after 3 - 4 reflows. Due to the prolific growth of the Cu-Sn intermetallic layer around the
Cu reinforcements, the effective volume fraction is enhanced even for the as-melted
composite solder. For example, after four reflows, the effective volume fraction becomes
approximately twice that of the as-mixed reinforcement volume fraction of the solder
paste. One might envisage that the Cu reinforcements could act to increase the interfacial
surface tension with the molten solder (chemically constrained) thus reducing flow on the
substrate. Also, there can be a geometric effect where the Cu particles prevent contact of
the molten solder with the Cu substrate and/or inhibit flow of the leading edge of the

spreading spherical cap which can lead to decrease in the wettability. Also, Cu particles

104

modified by the growth of Cu-Sn intermetallic layer will tend to constrain the advancing

edge of the molten solder cap.

 

 

 

 

  

 

 

 

so I I I I Y—T T I 17" TTTTTTTTT '1 Y T Y 1 ft TY r YYYYYYYYY I IIIIIIIII
r- ; : Z I 4
1 z ' '
I— -J
I
- —0—Wettmg angle (degree)J 1
l- , f I
40 b .................. 1 .................... 1 .. ................................... _.
I l I t
A - : : : : .
o
e -«
D '1
e _ . . . . .
'D _ : : :
V 3° _. ........................................................................................... _,
o : : : :
'5 ' ' ' ‘ ' ‘
I:
a
O .
5 ' : : : : ‘
g 20 .... ...................................................................................... . ..... _.
s . : : : : q
" . . . . '4
‘0 ... .................................. I .................... ; .................... : .................. .1
_ : : ; ;
J L L 4L ll 1 l 1 1_L L l I l l 1 L1 iiiii l 111111111 I. 11111111
-5 0 5 10 15 20

Volume percent of Cu pertlclee

Figure 5.3 Changes of wetting angles with the volume fraction of the copper particles in
the Cu particle reinforced composite solders.

It is also significant that Ag particle reinforced composite solder at 15 v% did not
show high wetting angle as observed in Cu particle reinforced composite solder. In
contrast to Cu reinforcements that form a thick Cu-Sn IM layer, Ag reinforcements form
only a thin Ag38n 1M layer. Moreover, there is no significant increase in the effective
volume fraction due to the reflow process. Apparently, wettability of composite solder is

not significantly affected when the effective reinforcement volume fraction is

105

approximately 20 v% or less. Similar wetting angles of around 18°-20° were observed in
a composite solder containing 20 v% non-coarsening in situ Cu68n5 reinforcements
fabricated in our laboratory [47].

Also, it was found that reflow had no noticeable influence on contact angle of the
Cu and Ag composite solders. However, it is worthy noting that upon reflow the
substrate being rewetted is the Cu68n5 1M layer.

Results in Table 5.1 indicate that the wettability of the nominally 15v% of Ni
composite solder (0=12.5°) is comparable with the wettability of eutectic Sn-3.5Ag
solder paste (0=10.5°). The average wetting angle was smaller than that of Ag (0=21.1°)
or Cu composite solder (0=47.3°). The wettability of Ni composite solder paste ranked
the first among three composite solders investigated. There was also no significant
meaningful difference in the wetting angles between first reflowed and multiple reflowed
Ni composite solder. This is primarily due to the fact that the extent of intermetallics
formed both around the Ni particles and at the Cu substrate/solder interface was sparse
during subsequent multiple reflow conditions. The micrographs proving this fact will be
introduced in the later part of this chapter. Thus, the total effective volume fraction of the
reinforcements remained nearly unchanged during the reflow process. As stated above,
the wettability of composite solder was not significantly affected when the effective
reinforcement volume fraction is approximately 20% or less. Therefore, the wettability
was not affected by the reflow of the 15v% of Ni particles in the Ni composite solder.
Evaluating the effective volume fraction is very important because the reinforcement and
its concomitant IM layers around them could act to increase the interfacial surface

tension with the molten solder, and thus, affects the wettability of the solder.

106

5.4 Effect of Reflow on Microstructure of Cu, Ag, and Ni Particle Reinforced
Composite Solder Joints

5.4.1 Intermetallic Layer around the Particulate Reinforcements

The effect of multiple reflow on the growth of Cu—Sn intermetallics formed around
Cu particle reinforcements is shown in Figure 5.4. The 1M layer around the Cu particles
grew dramatically with each reflow condition at the expense of the Cu particles. After the
fourth reflow, the Cu reinforcement particles were almost completely converted to Cu-Sn
intermetallics. Also, significant voiding was observed due to Kirkendall effects. Figure
5.5 shows the effect of multiple reflow on the growth of Ag-Sn intermetallics around Ag
particulate reinforcements. In stark contrast to extensive Cu-Sn IM layer formation
around the Cu particles, significant Ag-Sn IM layer formation was not observed after four
reflows around the Ag reinforcement. However, there were slight changes in Ag particle
morphology after the fourth reflow, as shown in Figure 5.5d.

We have used the diffusion results listed in Table 3.1 earlier to explain the
difference in the extent of [Ni layer formation observed around the Cu and Ag
reinforcement particles. At reflow temperatures, diffusion kinetics are extremely rapid as
the reflow process represents diffusion between transport species due to melting of the
solder materials. Consequently, the difference in the initial IM layer thickness around
particulate reinforcements and the extent of IM layer growth is greater around Cu
particles of Cu composite solders after multiple reflows.

Figure 5.6 summarises the effect of multiple reflow on the growth of the Cu-Sn IM
layer around the Cu particle, and concurrently, consumption of the Cu particle. It is also
evident that after four reflows full conversion of the Cu particles to Cu-Sn intermetallics

has been achieved. Micrographs illustrating formation of Cu-Sn intermetallics and

107

Cu particles have been convurtgd
.to Cudng lgtermctallrcs. ' '
‘fi . ’0' '_ .

 

(C) (d)

Figure 5.4 Effect of reflow on the growth of Cu-Sn intermetallics around Cu particle
reinforcements: (a) first reflow, (b) second reflow, (c) third reflow, (d) fourth
reflow.

108

. .

flu

.25..

35..

fl,

.

¢
5. Fr
\ _ . u

.-

01 f '
w. '.

a
.grfiv’a; n

A

e

: ‘ifiqa 31‘

A

.5

1d »\ .4.-

'5’.

Nflai

51:;
go“ it, I

\)'

f ..
.)7

 

durumw

(C)

Figure 5.5 Effect of reflow on the growth of Ag38n intermetallics around Ag particle

(c) third reflow, (d) fourth

reinforcements: (a) first reflow, (b) second reflow,

reflow.

 

a r—«w-r ’r— '—l '—r -— r77: 2, —v— v-~~.7j —y—-fi—,A«—;

l CuOcrsunptm h
m9inam¢lcsaanmrum01

 

0
Fur, ij'T—r‘v rV‘T-r‘

 

Thidrneee or Name Dim (mla’on)
A

N
v if . fVT‘ r‘V WT1—1‘w—v 7*Vr

 

 

 

 

O
r.
r
1

Figure 5.6 Effect of reflow on the Cu-Sn intermetallic layer growth around Cu particles
and Cu particle consumption.

growth around Cu particle along with the concurrent depletion the Cu particle are shown
in Figure 5.4.

The effect of multiple reflow on the microstructure of the Ni particle reinforced
composite solder joint is shown in Figure 5.7. The Cu-Ni-Sn ternary “sunburst”-shaped
IM layer circumscribes the dark core region representing unconverted Ni particles in the
as-fabricated solder joint in Figure 5.7(a). It is observed that even with four reflows of
the Ni composite solder joint, the Ni particles were not completely converted into
intermetallics. As a result of multiple reflow, the most significant microstructural
changes in the joint were a slight increase in the growth of 1M layer around the Ni

reinforcement and a significant increase in the size of matrix Ag3Sn precipitates.

110

cu-Ni-Sn 5‘ _» . - .- -- i
,lutermetallics - ‘. . 7 " ‘ * '5 OWE—5°53?»
- . 7 . - . intermetalties . “

 

(c) (d)
Figure 5.7. Effect of reflow on the growth of Cu-Ni-Sn IM layers around the Ni particles

in the Ni composite solder joint. (a) first reflow, (b) second reflow, (0) third
reflow, (d) fourth reflow.

111

5.4.2 Interfacial Intermetallic Layer in the Composite Solder Joints

The influence of multiple reflow on the growth of the Cu-Sn IM layer at the copper
substrate-solder interface in the Cu reinforced composite solder is shown in Figure 5.8
where significant interfacial layer growth was noted. The initial IM layer thickness due to
the first reflow of the specimen was 1.9 um. However, after only four reflows the layer
thickness increased up to 5 pm, that is about 2.7 times its initial thickness. For the Ag
reinforced composite solders shown in Figure 5.9, the initial layer thickness was 1.2 pm.
After four reflows, the layer thickness increased to 2.8 um, that is about 2.3 times its
initial thickness. The final interfacial layer thickness for the Ag composite solder was
about 1/2 the interfacial layer thickness of the Cu composite solder after four reflows.
Nevertheless, the ratio of the final IM interfacial layer thickness and the initial 1M layer
thickness for the Ag and Cu composite solders, are comparable (Figure 5.10) indicating
that the grth kinetics are controlled by similar diffusion mechanisms. Figure 5.10
graphically depicts the interfacial layer growth in both Cu and Ag composite solder as a
function of the number of reflows.

As mentioned above, a thicker IM layer tends to form in the Cu composite solder
than in Ag composite solders. Why this is so, can be rationalised by incorporating the
findings of Wu et.al. [42]. While investigating eutectic Pb—Sn solder reinforced with
various types of particle reinforcements, they found that the addition of Ag particle
reinforcement increased the activation energy for formation of the Cu68n5 (1)-phase) layer.
This increase in the activation energy of the n-phase makes nucleation more difficult, and
as such, formation of Cu6Sn5 is suppressed. Consequently, the initial 1M layer thickness

of Ag reinforced composite solder tends to be smaller than that of Cu reinforced

112

Cu
Substrate

.Interfacial

 

Figure 5.8 Influence of reflow on the growth of Cu-Sn IM layer at Cu substrate/solder
interface in Cu reinforced composite solder: (a) first reflow, layer thickness:
1.85 pm, (b) second reflow, (c) third reflow, (d) fourth reflow, layer thickness
=4.95u.m.

113

Cu
Substrate

 

(C) (d)

Figure 5.9 Influence of reflow on the growth of Cu-Sn IM layer at Cu substrate/solder
interface in Ag reinforced composite solder: (a) first reflow, layer thickness:
1.23 pm, (b) second reflow, (c) third reflow, (d) fourth reflow, layer thickness
=2.78p.m.

114

 

     

 

 

 

 

 

 

 

o .w IV» V b e g r
m m m w m. m .m m
m ..m .m m m o m M

.qjfi...._1_.:./:34114_:_. mu m m .w 8» W W;
n . / y n ..u H... e a M
y “3 .m u a m d m l t
. CH3. a n w .1 M h
v " Md/ 1 . m .l S d m C We
W\\\. 4 anm o w. a m p m a 0
.mm m m .m .t e e ..m 6
”MD. a .m .m m h a W m
r rsm .m. r r m t m m. an
eec . e c .m d C rm

. Y...“ n .0 Y n n Y
. . hsm f 1 31¢ .... u .1 t. 0
,Su m.f o m .m m t m ......
. Mmm y .... m .w m m w m
e r .m .m m 6 fi 6 o
mmm m m ”m o u. .m n w.
de m m e m. . M a L b
08.]. .U d m m 11. m
fl .1. a c ........
mam .w w v. m m i e
fnm w 0 My 0 m m h m .m
m.1.l m h 0 we m
mam. e .m n f ..m m H e
firmd n h m. v. Y. m o n .w
new ). .c m .m h ...... n .m m
.m.mum 9 m o N r, d w o
frd 5 D. c M k d 0 t
0&3 m u d m .m m .m w

n .m t S 1

www g C m m .n e .w a m
warm mu m m .m w .m a r, m
mmm a m m a m m. ..m m m

I1 S 1
Cam m a .m. m m m e m .w.
0 S .9 b .m e .1 .h H
J m ..m P g N f m m
5 .m A .m nnnnn
m m. .lw. .m m I w m .lP
mu m Ms M w ”me fl m w
n w is. w 1 m m .m m

115

1 ' ‘ Cu
C“ ' i ' ' Substrate
Substrate .

Cu-Nj-Sn ' ._
; IM Layer

/(close'to soldeg)
. - ‘ 1 i2").

Cu-Sn
IM Layer -'
(close to Cu) Can

M Layer

(close to Cu)

Cu ' -- _
Substrate . . ' Cu . '
Substrate ’ - ‘ .
. 1. 3e". Cu-Ni-Sn
41411211341
'closeqlp 501031;?)

Cu-Sn : ~ I. 1 . Cu-Sn
IM Layer : . ‘ . lM Layer

(close to Cu) . . . .. ~ (close to Cul

 

(C) ((1)
Figure 5.11. Effect of reflow on the growth of IM layers at the Cu substrate / solder

interface in the Ni composite solder joint. (a) first reflow, (b) second reflow,
(c) third reflow, (d) fourth reflow.

116

within the short time period the specimen is held (5 seconds) at the reflow temperature of
280°C. The fact that the microstructure remains stable under reflow conditions is
beneficial for manufacturing Operations where reflow of solder joints is a required

process.

5.4.3 Effect of Reflow on the Microstructure of Eutectic Sn-3.5Ag Solder Joints

The microstructure of eutectic Sn-3.5Ag solder joints as a function of reflow is
shown in Figure 5.12. The microstructure in Figure 5.l2(a) for first reflowed solder is
characterized by Sn cells separated by wide bands of eutectic Ag3Sn. With multiple
reflows, the Ag38n bands separating the Sn cells became narrower and simultaneously
the size of the Sn cells increased, as exhibited in Figure 5.12(b)-(d).

The effect of reflow on the interfacial intermetallic layer in the eutectic Sn-3.5Ag
solder joint is illustrated in Figure 5.13. The interfacial layer grew with reflow from the
initial layer thickness of 1.7 pm to a final layer thickness of 4.3 11m. The interfacial layer
grew by a factor of 2.5 after reflow. This is very similar to the interfacial intermetallic
layer growth behavior that occurred during reflow in the Cu particle reinforced composite
solder joints as shown in Figure 5.14. Figure 5.14 compares the grth of interfacial

layers at the copper substrate in all the solder joint materials studied.

5.5 Effect of Reflow on Mechanical Properties of Eutectic Sn-3.5Ag Solders
Reflowing of the solder materials, particularly solder joints on circuit boards, is
required in manufacturing processes. Two or three reflows are considered normal for a

typical wave soldering process. To assess whether multiple-reflow of solder materials

117

Sn cells.

/ .

   

(a) (b)

A 2381] bands

\

Increased Sn cells

   

(C) ((1)
Figure 5.12 Effect of reflow on the microstructure of eutectic Sn-3.5Ag non-composite

solder joint: (a) first reflow, (b) second reflow, (c) third reflow, (d) fourth
reflow.

118

1 Cu Substrate
Cu-Sn Interfacial '
1M Layer ’

 

.,-, ..,,, ,.
7.1:. (121%. 1:3.

(C) ((1)

Figure 5.13 Effect of reflow on the growth of Cu-Sn intermetallic layer at Cu substrate-
solder interface in eutectic Sn-3.5Ag non-composite solder joint: (a) non-
reflow, layer thickness = 1.74 um, (b) first reflow, (c) second reflow, ((1)

third reflow, layer thickness =4.27|.tm.

119

 

   
  

 

 

 

 

 

 

4.5 _ 1 r ,
’s‘ ’ ' 12‘
i5 4 ._ .................. <5 ..___.¢-____¢_...a_¢ ................. _
: - ' : -4
5 35 I— —0—Cu Composite Solder .-e ___________ 3 _________________ J
2 ’ ; -B— Ag Composite Solder ' ' 3
is t -‘>- Ni Composite Solder . 3
i 3 :- --X--EutecticSn-3.5Ag Solder If -------------- j
I l- . - 3 ' 4
..l . j 3 1 j
g 2.5 :..................§ .................. Z ..... ' ............. . .......... ; ...... '5] .................. ._:
a ' = =./ '
E . : "’4
.............. .1 ..... (.....-J ..............................
h 2 — . O a ' -------- —1
‘3 5:" Jr ' J
.7: 15 : 3 '/ E i 5 3
‘5 ' r- E'/ g f 5 4
*- : «'31 3 E 5 j
1 l 1 l 1
0 1 2 3 4 5

Number of reflowe

Figure 5.14 Comparison of interfacial intermetallic layer growth due to reflow between
Cu composite solder, Ag composite solder, Ni composite solder and eutectic
Sn-3.5Ag non-composite solder joint.

will affect the resultant mechanical properties, nanoindentation testing (NIT) was

performed on eutectic Sn-3.5Ag solder. NIT was conducted on eutectic Sn-3.5Ag solders
reflowed up to 4 times using the Nano Indenter® XP mechanical properties microprobe

(MPM) from MTS Corporation. With the MPM, simultaneous measurement of indenter
penetration depth and load makes it possible to determine certain mechanical properties.
Mechanical properties assessed were hardness, modulus and creep properties. The most
significant finding with regards to mechanical properties change turned out to be the
difference in hardness as a function of reflow as shown in Figure 5.15. With reflow of
the solder, the hardness decreased. Also, the yield strength can be estimated according to

Tabor [125] from hardness data. In Figure 5.16, the relative change in yield strength is

120

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.5 . . . . . . . . . . r 4 . f . . r . r . . . as
l.
0.4 1000 O
. A O .
5 f. V O ’ O o ‘53“? ‘
8,, .. . . 0 ‘3 fl
g g 3 ..m..........s.m'..-.‘ Tl RMK .. .. .....1
u l .3 i .
u. i o g
g <50 0 Q. ’ ‘
. O l .
8 .05 <2} A e c- 9 Ream-4 .
E 0.2 O X 0 0’ 52..‘..m..'_..m' Q g
g 4) V o Reflow-3
5- L .~ . 0 O
r O .
0.1
U 1 1 1 1 r 1 1 1 r r r .1 L r a 1 1 1 1 r 1 1 n n
0.1 0.2 0.3 0.4 0.5 0.5 0.7
Load (ml!)

Figure 5.15 Change in hardness of eutectic Sn-3.5Ag solder as a function of reflow.

121

 

 

 

 

 

 

 

 

 

 

shown as a function of reflow. On average, the yield strength of eutectic Sn-3.5Ag solder
is reduced by ~ 30 %. The reduction in hardness/yield strength motivated by the change
in solder chemistry and microstructure due to the reflow process. It is suggested that the
composition of the solder change from the eutectic composition (96.58n-3.5Ag) to an off-
eutectic composition containing less Sn due to reflow. The reduction in the amount of Sn
is due to its consumption in the formation of the intermetallic layer consisting of Cu63n5
and Cu38n compounds [126]. Change in solder microstructure is expected for off-
eutectic stoichiometry. Proeutectic Ag38n will precipitate out of the liquid first followed
by solidification at the eutectic composition. As indicated in Figure 5.12, the Sn cells
became larger and the necklace of eutectic Ag3Sn became narrower subsequent to
multiple reflows. The microstructure of the first reflowed and 4-reflowed solder around
indent is shown in Figure 5.17. In Figure 5.l7(a) for first reflowed solder, smaller Sn
cells and wider Ag3Sn bands were evident. Whereas, Figure 5.l7(b) showed larger Sn
cells and thinner Ag3Sn bands after four reflows. The lower hardness and yield strength
observed is consistent with materials having a larger grain size.

Nanoindentation creep tests were conducted. Assuming steady-state behavior,

{3‘ = A - 0'" , the stress exponent for indentation creep was determined for multiple
reflowed specimens. The stress exponent for eutectic Sn-3.5Ag solder materials after
four reflows was n=8, whereas the first reflowed solder exhibited a value of n=7.1. The
stress exponent data are shown in Figure 5.18. The difference in n-values suggests that
creep deformation in the solder is influenced by reflow history. From n values obtained,
the steady-state creep rate of reflowed solder materials is higher due to reflow-induced

microstructure and composition changes.

123

 

(b)

Figure 5.17 Microstructure of eutectic Sn-3.5Ag solder around indent: (a) non-reflow
solder; (b) three reflowed solder. The size of the Sn cells is larger for
reflowed materials, correspondingly, the yield strength is lower.

124

Strain Rate (llsec)

 

 

 

 

 

 

 

 

1 b T YWTT f T Y T r fr V’— T r T 1“
r i 5 ‘
I 1
P r 1
f . : 1
r “
4-refljows
0.1 .. - ......
t 1 1
«l
I l
r l
l i 1
;
0.01 1:”..- MUM“... ....................-...?.....- .."J 4
’ i 5 ‘
b I
- J
» 4
1
q
o.m1 L 1 .L L_ A A l—LJ A 1 1._.1 A L 1 A 1 l A + _L ..A__L l_1_1
0.01 0. 1 ‘l 10

Stress (GPa)

Figure 5.18 The effect of reflow on steady-state creep strain rate of eutectic Sn-3.5Ag

solder. The stress exponent is slightly higher for multiple reflowed solder.

5.6 Summary

1.

Eutectic Sn-3.5Ag solder paste showed the best wettability on Cu substrate among
all the solder materials studied. The 15 v% Ni composite solder showed comparable
wettability to eutectic Sn-3.5Ag solder paste. The wettability of 15 v% Ag-
reinforced composite solder was significantly better than 17 v% Cu-reinforced

composite solder. Because of a profuse intermetallic layer growth around Cu

125

reinforcements, the effective volume fraction doubled. Wettability can be improved
by lowering the volume fraction of the reinforcing phase.

No significant changes in contact angles were observed with multiple reflow of all
the solder materials studied.

The formation of Cu-Sn intermetallics around the Cu reinforcements was
considerable, whereas the Ag3$n IM layer thickness around the Ag reinforcements
was minimal. Growth of the IM layer around the Cu particle reinforcements was
excessive leading to total consumption of the Cu particles after 3-4 reflows. No
significant coarsening of the Ag reinforcements was evident after multiple reflows.
Multiple reflow does not significantly change the microstructure in Ni composite
solder joints. The M layer growth was minimal, although coarsening of Ag3Sn
particles was observed.

Reflow studies showed that the 1M layer at the Cu substrate/solder interface also
grew much faster in Cu composite solder joints than in Ag composite solder joints.
However, multiple reflow did not cause significant growth of interfacial IM layers in
Ni composite solder joints.

The microstructure of first reflowed eutectic Sn-3.5Ag solder can generally be
characterized by small Sn cells surrounded by wide-banded eutectic Ag3Sn phase. In
comparison, the microstructure of multiple reflowed solder is characterized by large
Sn cells circumvented by a thin necklace of eutectic Ag3Sn precipitates.

The hardness and yield strength of multiple reflowed eutectic Sn-3.5Ag solder were

reduced by 30% after three reflows. This finding is commensurate with the

126

increasing size of Sn cells produced by multiple reflow as a larger grain/cell size
gives a lower yield strength.

The stress exponent, n, determined using indentation creep testing was 7 for first
reflowed solders and 8 for multiple reflowed solders. From the stress exponents
observed, the steady-state creep rate for the multiple reflowed solder will be higher

compared to first reflowed solder.

127

CHAPTER VI

CREEP DEFORMATION BEHAVIOR OF EUTECTIC Sn-3.5Ag SOLDER
JOINTS WITH OR WITHOUT SMALL ALLOYING ELENIENT ADDITIONS,
OR Cu, Ag, Ni REINFORCEMENT PARTICLES

6.1 Introduction

Eutectic Sn-3.5Ag solder has received attention worldwide as a potential substitute
due to its non-toxic nature as well as its comparable wetting and mechanical properties to
eutectic Sn-37Pb solder [1-6]. In order to be an acceptable substitute for lead-bearing
solders it has to satisfy both the process requirements and the reliability requirements that
commonly used Sn-Pb solder possesses. In terms of process requirements, the most
important factor is the melting point. The melting temperature of eutectic Sn-3.5Ag
. solder (221°C) is ~ 40°C higher than eutectic Sn-Pb solder (183°C), which causes
manufacturing difficulties and cost increase compared to processing methods used for
Pb-bearing solders, although the advantage of this higher melting point has proven to be
more suitable for higher temperature applications [5]. In terms of reliability issues,
recent growth of surface mount technology, and especially the miniaturization in
microelectronics industry have made the structural role of solder joints as mechanical
support to components become ever more critical compared to their traditional function
of just providing electrical contact. Solder alloys are typically subjected to harsh
environments and are used at temperatures well above half of their melting points in
degrees absolute. So microstructural evolution, recrystallization, superplasticity, creep,
and creep-fatigue are operative under normal service conditions, among which creep is

the most common and important micromechnical deformation phenomena [4, 7 -14].

128

Since eutectic Sn-3.5Ag solder is considered as a leading candidate of Pb-bearing
solders, several approaches have been used to improve its comprehensive properties such
as creep resistance, thermomechanical fatigue resistance, mechanical strength, and solder
joint reliability [13, 20, 47-50, 91, 104, 127]. Among these approaches, composite
approach was designed mainly to improve its service performance including service
temperature capability. Composite solders, whether tested as bulk or as joint specimens,
generally showed improved properties [13, 22-32, 47-50, 91, 104, 127]. However,
systematic studies of the creep defamation behavior of composite solders, especially
lead-free composite solders, have been rarely reported [127]. Addition of alloying
elements to eutectic Sn-3.5Ag solder is another major avenue attempting to combat the
above stated issues [15-21]. Such additions usually include copper, nickel, antimony,
silver, or bismuth, etc. This methodology was initially designed to achieve a suitable
melting temperature range primarily for processing purposes. The improvement of
mechanical properties is closely related to the amount and type of the alloying elements,
so some alloying additions may not improve the properties to a desirable extent, and
sometimes even deteriorate them [20, 21]. Although significant effects of alloying
element additions on the mechanical properties of binary systems are expected, these
alloying effects have not been systematically investigated, i.e., no fundamental work on
these areas is found for lead-free solders in published literature.

In this chapter, creep studies were carried out to solder materials based on the two
major approaches discussed above. The first research effort was aimed at a systematic
parallel study to characterize the creep deformation behavior of lead-free eutectic Sn-

3.5Ag based composite solders produced by mechanically adding Cu, Ag, or Ni particles.

129

Various creep parameters are quantified and compared with the non-composite eutectic
Sn-3.5Ag and Sn-4Ag-0.5Cu solder alloys. Possible mechanisms for creep are suggested
for these composite solders. As a second research effort, we report on near-eutectic Sn-
Ag based ternary alloys, two ternary alloys (Sn-4Ag-O.5Cu and Sn-Ag-O.5Ni) and a
quaternary alloy (Sn-2Ag-lCu-1Ni) which were developed in an attempt to depress the
melting temperature, improve the creep properties, and reduce the cost of the eutectic Sn-
3.5Ag solder. Choices of Cu and/or Ni as alloying elements are primarily due to
formation of additional intermetallic phases that could improve mechanical properties of
the solder. Also, small additions of Cu were found to decrease the melting temperature
and improve wetting of the eutectic Sn-3.5Ag solder [15-21]. These solder alloys
represent some of the leading lead-free solder candidates that are being considered for
use in the automotive industries in the near future. Preliminary evaluations of such
solders using reliability test vehicles in accelerated thermal cycling tests indicated
significantly improved performance as compared to eutectic Sn-3.5Ag solder. Motivated
by these findings, room and high temperature creep studies were carried out on the solder
joints made of these solder alloys to investigate the effects of Cu and/or Ni additions on

the creep deformation behaviors of eutectic Sn-3.5Ag solder joints.

6.2 Experimental Procedures
To facilitate creep strain measurement, a straight-line laser ablation pattern was
intentionally formed on the surface of the solder joint spanning the thickness and length

of the solder joint. This geometric pattern was created by near-surface ablation using an

inert gas Kr-F excimer laser from Lamda Physik® with a wavelength of 248 nm and a

130

maximum pulse power of 1,500 m]. The pulse duration is 25 ns. By exposing the
masked solder joint to a single excimer laser pulse a distinctive pattern dictated by the
mask geometry is imprinted on the surface of the solder joint due to highly localized
surface melting and ablation. The pulse energy was 300 m1 and the energy density for a
spot size of 0.5 cm x 1.1 cm was 0.55 J/cmz. An excimer laser ablation mapping pattern
imprint on the solder joint is shown in Figure 6.1. Figure 6.1(b) illustrates how these line
patterns change their shape due to creep deformation. SEM images of the whole joint
were taken before and after creep testing to assess the extent of creep deformation in 2-D.

Creep testing was carried out primarily on the micro-sized solder joints using dead
weight loading on a miniature creep testing frame equipped with an electrical resistance
furnace, as shown in Figure 6.2. The creep tests were conducted at 25°C, 65°C and
105°C to the solder joints made from the particulate (Cu, Ag, or Ni) reinforced composite
solder materials, representing homologous temperatures of 0.6, 0.68, and 0.78,
respectively. Alternatively, creep tests were conducted at 25°C and 85°C only for Sn-Ag
based solder alloys with different Cu and Ni alloying element additions. For elevated
temperature tests, the solder joints were heated by conduction. The testing temperature
was measured by a thermocouple kept in contact with the solder joint. The temperature
fluctuation observed in the joint area was maintained at i 1°C for the duration of the
creep test.

During creep testing, the deformation of the laser—ablation line pattern was tracked
by capturing digital images at regular time intervals with a LECO® image analyzer. The
loads used in the creep tests ranged from 1 to 5 pounds (OAS-2.3 Kg) and solder joints

were typically loaded until tertiary creep stage started.

131

 

(b) after creep

Figure 6.1. Laser ablation patterns imprinted on solder joints used in creep testing.
(a) before creep; and (b) after creep.

   

<—— Creep frame

Variable voltage
power supply

Tested solder joint

Figure 6.2 Traditional dead load creep testing frame with an electrical resistance furnace.

132

Creep data were obtained by mapping the time-sequence images of the distorted
laser-ablation line patterns on the solder joint, as shown in Figure 6.1(b). By this laser-
ablation mapping and data extraction technique developed previously [56, 127], a variety
of creep-related parameters such as (i) unnormalized and normalized displacement of the
solder joint vs. time, (ii) global and localized strain vs. time, (iii) secondary creep strain
rate, (iv) variation of local strain rate with time and joint thickness, and (v) strain for the
onset of tertiary creep, etc. were obtained.

A new setup for creep testing was recently developed. Creep tests on Ni composite
solder joints were mostly preformed using this setup. Under this new creep setting, creep
tests were canied out on the small, excimer-laser-marked solder joints still using dead
weight loading on a new miniature creep testing frame fixed on an optical microscope, as

shown in Figure 6.3. The deformation of the polished side of the solder joint was
monitored using a Kodak® CCD camera connected to a microscope with a computer.

Time-sequence images of the distorted laser-ablation patterns were captured
automatically by the camera at set time intervals. This creep testing setup has many
advantages over the previously introduced setup. The present design effectively
eliminated vibration of the solder joint during creep testing. Also, through-lens lighting in
the microscope and the auto-balance function of the imaging software significantly
improved the image quality. Such improvement facilitated the subsequent image analysis
as well as promoted accuracy in data acquisition. Finally, the auto-snap function of the
software made it possible to automatically acquire time-sequenced images and data

during creep testing.

133

Specimen

EU“

 

Figure 6.3 New design of the dead weight loading miniature creep testing frame.

134

Microscopy on undeformed specimens was examined after the joint was mounted
in a fixture to polish one side using conventional metallographic techniques. Scanning
electron microscopy was conducted on a Hitachi S-2500C SEM with a Link SIS Energy
Dispersive Spectroscopy analysis system. The prior polished side of specimens was

observed after deformation to observe heterogeneous strain effects.

6.3 Creep Deformation Behavior of Cu, Ag, Ni Particle Reinforced Solder Joints as
Compared to Eutectic Sn-3.5Ag and Sn-4Ag-0.5Cu Solder Joints

6.3.1 Typical Creep Data
6.3.1.1 Creep of Non-composite —-Eutectic Sn-3.5Ag Solder Joints

Variation of the localized and global creep strain with time for a non-composite
eutectic Sn-3.5Ag solder joint is illustrated in Figure 6.4(a). The solder joint was loaded
at 13 MPa at 25°C for about 6 hours. The global shear strain is indicated with a solid
heavy line, which exhibits classical creep behavior with typical regions of primary,
secondary and tertiary creep. Other lines represent the shear strain at different positions
throughout the solder joint thickness, where one side of the solder joint is defined as
position 0 while the other side of the joint is defined as position 1.0. The localized shear
strain at each position followed the same trend as the global strain. It is shown that the
highest localized shear strain occurred at position 0 and position 1.0, which indicates that
intensive creep deformation was primarily along the substrate-solder interface region in
this specimen. In the middle region of the solder joint, position 0.5, the localized creep

strain was found to be the lowest. The secondary creep rate measured from the global

. - -1
strain curve was 2.0 x 10 5 s .

135

Non-Composlte Sir-Ag Solder Joint at 25 °C. 13 Ian

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    

 

 

 

 

 

 

 

 

2 C t Loclillond sat-tin ; Grlotelgtraln'vc'flm'o : V ' I I E ' 1
_ —e—0 -—x--0.3 --o- 0.0 "Nb-"0.9 l " :
-E}— 0.1 --+--0.4 --I— 0.7 -—¥—1 -
-0— 0.2 ' b— 0.5 --O'- 0.8 -O—G|obll
1.5 ' . . ,_ ------------------------------ —
= r
:73.
0.5
0
0 5000 1 10‘ 1.510‘ 210‘ 2.510‘
Time (see) (a)
3/P‘\.
r’ \N\
/”/K //\\‘- \\~
r’ ‘\ 8.
0900251 // KN, ‘\\ 004.00025
7; 00002“ f,/” \\ \Wx Hmooooz E?
3 000015- \‘m, \ N 3'
a L/ W\\ % \r 0.00015 E
a; 00001“ ’ ' Ext-0.0001 8
0: .. \ 1.2
g 0.00005 Fw- 0.00005 g
.. K 8
5’: 0 #>v0
-0. 5 i )3"- 0.00005

 

 

(b)

Figure 6.4.(a) Localized and global creep strain vs. time for the non-composite eutectic
Sn-3.5Ag solder joint. (b) 3-D plot of the creep strain rate vs. time and position across
the solder joint.

136

The variation of creep strain rate with respect to time and position is plotted in a 3-
D format in Figure 6.4(b). The plot indicates that the creep rates are highest at the
Opposing interfaces of the solder joint throughout the duration of the experiment. The
mid-thickness position has the lowest creep rate at all times. It is also evident that the
creep rate reached the highest value at the start of the creep test and near the end of the
creep test representative of primary and tertiary creep deformation respectively. The
lowest creep rate was reached at about 1><104 seconds (2.8 h) after creep test started,

which corresponds to the steady-state indicated by the global creep strain curve.

6.3.1.2 Creep of Non-composite —Sn-4.0Ag-0.5Cu Solder Joints

The localized and global creep strain for a non-composite Sn-4.0Ag-0.5Cu solder
joint is plotted as a function of time in Figure 6.5(a). The solder joint was loaded at a
stress of 16.2 Mpa at 25°C for about 5 hours. The localized and global shear strain were
comparable to the eutectic Sn-3.5Ag solder joints except the strains were slightly lower in
some locations. The localized shear strain at each position still followed a trend similar
to the global shear strain with slight variation. The highest localized creep strain
(damage accumulation) was observed on one side of the specimen near the interface at
position 0, whereas the lowest localized creep strain was found at position 1. Creep
deformation was inhomogeneous and the deformation was dominant on one side of the
solder joint (Figure 6.5a). The global strain curve was quite similar to the localized creep

strain at positions 0.4 and 0.5, which is near the mid-thickness region of the solder joint.
The secondary creep rate was measured to be 2.06x10'5 5']. Notably, the secondary creep

strain rate was lower in Sn-4.0Ag-0.5Cu solder compared to Sn-3.5Ag solder despite the

137

Non-Composite 95.58n-4.0Ag-0.50u Solder Joint at 25°C, 16.2MPa

 

 

 

 

 

 

   

   

   

2.5 b V I f T y I U T ff 1 I I I I T T r T I 17 I -‘
C Localized Strain a Global Strain vs. Tillie 3 I
F ‘ * ‘ : <
2 : —o-—0 --~--0.3 --o— 0.6 ....,...0_9 """"" § """""""" “s
I -a— 0.1 --+--0.4 --I- 0.7 —v—1 3
» -o—0.2 - a—0.5 --+-- 0.8 -o—Globa1 a
1.5 - --.---/1 ----------- ~
.5 t ' ‘ 3 /3’
g : s . 2 .
m 1 .... ...... /..-)..¢..’. ..............
ii ’ ° 6 ’ 3
.. ; .... --——a—=—:: /.-----s---<
('5 * I/o—""'3.:~"".xb—':"§"
0.5 >-- ------------------------ ‘ -"~.H~~ avfi‘RT--'--J-od-.‘";6; """" 5“”! -------- —1
' , :d-7.-:'-Z_'——'L:' 9 [Cu—o. ..-
' 31-3 27:. 7:2“ 93.27““ “559-
0 _ .................. f ’5}; ................
-05 1 1 1 1 1
-5000 0 5000 1 10‘ 1.510‘ 210‘
Time (sec)
‘ 0.00035 to
A
8 a! 0.0003 g
g ‘I 0.00025 =‘
‘0’ " 0.0002 g?
*5 " 0.00015 8
a: ‘g 0.0001 3
.= f 0.00005 3
a r0
1:! fl. 0 o
m f .0. V

 

 
  

    

(b)

Figure 6.5 (a) Localized and global creep strain vs. time for the non-composite Sn-4.0Ag-
0.5Cu solder joint. (b) 3D plot of creep strain rate vs. time and position across
the solder joint.

138

applied stress being 25% greater during creep testing. The 3-D plot of the variation of

creep strain rate with time and position is shown in Figure 6.5(b).

6.3.1.3 Creep of Cu Particle Reinforced Composite Solder Joints

The creep history of a Cu particle reinforced composite solder joint subjected to
creep deformation under a stress of 17 MPa at 25°C for about 8 hours is presented in
Figure 6.6(a). The localized and global strain was significantly lower in this specimen in
comparison to the non-composite solder joints presented above. The localized strain
varies dramatically at different position throughout the solder joint thickness. Similar to
the non—composite solder joints, the creep defamation occurred mostly along one side of
the specimen. The joint reached the highest localized strain near one Cu substrate/solder
interface at position zero, while the lowest localized creep strain was shown at position 1,
the opposite side of the solder joint. Figure 6.6(a) shows that the localized creep strain
was nil at position 1, indicating intensive deformation at only one interface of the joint.
In the center region of the specimen, position 0.4~0.5, the localized strain curve was

found to be very similar to the global strain curve. From the global strain curve, the
secondary creep rate was determined for this sample to be 4.08x10'6 s], which is 5 times

lower than that of the non-composite eutectic Sn-3.5Ag solder and Sn-4.0Ag-0.5Cu
solder joints in spite of higher applied stress levels imposed on the composite solder. The
3-D plot, Figure 6.6(b), shows the variation of creep strain rate with time and position for

a Cu particle reinforced composite solder joint.

139

Cu Particle Reinforced Composite Solder Joint at 25°C, 17M pa

 

 

 

 

 

 

 

 

 

  
      

 

0.8 "I'T"‘V1f'VIII—IITIIYIIIfiv—rflrrvrlrru
Localized smlnga Global Strain ve.1;lme ' ;
: —+—0 --~--0.3 --o- 0.6 ----A----0.9 I . .
0.5 L— -a— 0.1 --+--0.4 --I— 0.7 +1.0 ............. ............ _
-o—0.2 - 0—05 --+-- 0.0 —0-Global ' fig
: I
'5 0.4
171
«‘5
2 02
m .
0
« . : - 3 I - 1
-0.2 111111..111545114111.1111;11911111111.
-5000 0 5000 110‘ 1.510‘ 210‘ 2.510‘ 310‘ 3.510‘
Tlme(aec)
(a)
0.0004 f 0.0004
0.00035 0.00035 U?
’6‘ 1:1
8 0.0003 f “-0003 ea.
\ 0.00025 fl 0.00025 :2
3 0.0002 ' 0.0002 g!
*3 0.00015 f 0.00015 3
0: 0.0001 A
a 0.0001 ' 0 5 z
'3 0.00005 ( .0000 g
1: 0 \ 0 3
m 000005 5 0.00005

0))

Figure 6.6 (a) Localized and global creep strain vs. time for the Cu particle reinforced
composite solder joint. (b) 3-D plot of strain rate vs. time and position across
the solder joint.

140

6.3.1.4 Creep of Ag Particle Reinforced Composite Solder Joints

The variation of localized creep strain with time for the Ag particle reinforced
composite solder joints, shown in Figure 6.7(a), is quite different from the nan-composite
and Cu particle reinforced composite solder joints. This specimen was stressed with 10.6
MPa at 25°C for 3.6 hours. It is evident that the localized creep strain was closely
following the same path as the global strain curve. The localized creep strain at every
location through the solder joint thickness was virtually the same as the global strain,
which indicates a uniform defamation in this solder joint throughout the creep testing
process. Unlike the prior specimens, significant variations in the local creep strain were

not found at any location of the joint region. Moreover, extremely unifam defamation
was exhibited throughout the solder joint. The secondary creep rate was 1.20x10'5 s",

approximately the same as nan-composite eutectic Sn-3.5Ag and Sn-4.0Ag-0.5Cu solder
joints. Compared to the Cu particle reinforced composite solder joint, the secondary
creep strain was higher for the Ag particle reinforced composite solder joint.

The 3-D plat, Figure 6.7(b), shows the variation of creep strain with respect to time
and position through the joint thickness. The creep strain rate correlated well with the
creep stain behavior shown in Figure 6.7(a). The plot indicates that the creep rate is quite
unifam over the solder joint thickness for the duration of the test. Figure 6.7(b) shows

that the creep rate reached the highest value at the start of the creep test and near the end
of the creep test. The lowest creep rate was reached at about 0.7x104 seconds (1.9 hours)

after creep test started, which corresponds to the steady state in the global creep strain

curve.

141

Ag Particle Reinforced Composite Solder Joint at 25°C, 10.6 Mpa

Streln & ve. Time

—o—0 —-*--0.3 -—o— 0.6 -------Ar 0.9
-a— 0.1 --+--0.4 --e— 0.7 +1.0
—o — 0.2 - 15— 0.5 --+-- 0.8 -0— Global

Shear Strain

 

0 2000 4000 6000 8000 1 10. 1.2 10‘ 1.4 10‘

Time (see)

(a

   

    
 
 

 
 
 
  
 
 
  

 

 

 
   
    
   
    
   
  
  

    
  
  

. *0.00045
,3 0.0004w ~o_0004 U:
a 0.00035~ ~0.00035 g.
5 0.0003». ~0.0003 =‘
8 0.000254 -0.00025 E
g 0.0002- ~0.0002 f:
1: 0.00015-. e» ~000015 <
'5‘ 0.0001. ‘J'i' f 3
is, lie-me... _. ...... a
- W451"! ~0.00005
— fi%' 3;
°°. M . 5 E 8
° ‘3 v. *9. g 8 o
p . O ‘ 15) 0
“70¢? 8 mass)

(b)
Figure 6.7 (a) Localized and global creep strain vs. time for the Ag particle reinforced

composite solder joint. (b) 3-D plot of creep strain rate vs. time and position across the
solder joint.

142

6.3.2 Secondary Creep Rate for Different Solder Joints at Different Temperatures

Secondary creep rate was determined using linear regression from the global creep
strain curve. Steady-state creep rates under a nominal stress at 17 MPa are plotted as a
function of temperature in Figure 6.8. The secondary creep rates were much higher for
all solder joint materials at 105°C (0.78 Tm) due to diffusion mechanisms and grain
boundary sliding. Among these, Ni composite solder joints exhibited the lowest steady-
state creep rate. Cu composite solder joints exhibited much lower secondary creep rate
than the Ag composite solder and both non-composite solders. At room temperature, Ni
composite solder joints was 5 times more creep resistant than Cu composite solder joints
and ~30 times better than Ag composite and non-composite solder joints. Even at 105°C,
the Ni composite solder joints were ~3 times more creep resistant than Cu composite
solder joints and more creep resistant than Ag composite and non-composite solder joints
by a factor of 12. The creep perfomance of Sn-4.0Ag-0.5Cu solder joints was only
slightly better than eutectic Sn-3.5Ag solder joints. Comparing the secondary creep rates
of Cu and Ag reinforced composite solder joints, Cu particle reinforced composite solder
joints exhibit a lower steady state creep rate than Ag particle reinforced composite solder
joints. Despite extremely unifom defamation of Ag reinforced composite solder joints,
no significant improvement in creep resistance was found in comparison to non-
composite solder joints. The creep property of Sn-4.0Ag-0.5Cu solder alloy is slightly
better than the eutectic Sn-3.5Ag solder alloy at all test temperatures.

Cu particle reinforcement was successful in improving the creep resistance. The
inprovement is primarily due to an increase in the efiective volume fraction of the

reinforcement with reflow. The effective volume fraction of reinforcement intrinsically

143

 

0.01 1 1 l l
--8— Ni Composite 1 :
-B-— Cu Composite

 

 

0.0001

10'

Secondary Creep Rate (leeo)

10

 

_ -0—AgCompasite ............. ............ '
0'001 --X--Sn-3.5Ag 5 g g/ .
- - +- - Sn-4Ag-0.5Cu f o'i'l' f
- ' '0']: / ’.

 

 

 

10'
20 40 60 80

Temperature (°C)

Figure 6.8 The secondary creep rate for composite and non-composite solder joints as a
function test temperatures. The Ni particle reinforced composite solder joint

100

exhibited the best creep resistance at all test temperatures.

increases due to profuse fomation of a Cu6Sn5/Cu3Sn intemetallic (IM) layer around the
Cu particle reinforcement. Essentially, the volume of hard (Cu + Cu-Sn 1M)
reinforcement phase increases at the expense of the much softer solder matrix. It is
believed that the IM-modified Cu particle reinforcements act to restrain plastic flow on
the solder matrix due to elastic constraints, thus, enhancing creep resistance. 1M layer
fomation (Ag3Sn) around Ag particles in Ag composite solder joints was minimal
compared to Cu composite solder joints. Since the IM layers famed around the Ag

reinforcements were very thin, the volume fraction of the reinforcing phase remained

144

120

essentially the same. The Ag3Sn 1M layer around the Ag particles did not grow
significantly during creep testing. The introduction of 15 v% of ~4 um Ag particles into
the solder matrix tended to hamagenize the strain and promote unifom defamation in
the joint. However, the steady state creep resistance was not significantly affected by the
addition of the Ag particle reinforcements.

The reason that the Ni composite solder joint exhibited the best steady-state creep
resistance may be also due to the increase in volume fraction of fine Cu-Ni-Sn
intemetallic particles famed in the joint, which act as additional reinforcement. These
individual Cu-Ni-Sn particles in the solder matrix can constrain plastic flow during creep
defamation, thus, enhancing creep resistance. Provided the defamation mechanism for
creep in this material is by dislocation climb/glide, the Ni reinforcement and the fine Cu-
Ni-Sn [M particles can act as obstacles to the motion of dislocations resulting in

improved steady-state creep behavior.

6.3.3 Strain at the Onset of Tertiary Creep
Strain for the onset of tertiary creep (OTC) is plotted versus secondary creep rate in

Figure 6.9. Figure 6.9 contains data obtained for joints made with solder materials of the
current investigation and the OTC data for Cu6Sn5 reinforced composite solder produced

by in-situ method (@ 25°C) for comparison. Generally, the OTC occurred at a

significantly higher shear strain level for non-composite solder alloys in comparison to
the composite solders. This was the finding except for the in-situ Cu68n5 particle
reinforced composite solder joint. A shear strain level of ~0.5 observed for the OTC of

in-situ Cu6Sn5 reinforced composite solder joints was similar to non-composite solder

145

 

 

 

 

 

 

 

 

:, o_5 _. ................... ........ ‘ ........... ..................... ................... __
2 : -, . .
O 0 ; 0 Cu composite (25°C)
3' E E
.3 0.4 _ ............ ............... D Cucomrrosnuwcl ...... L
t 5 3
'2 + Ag compoelte (25C)
.6 : : X Ag compoelte (BS’C)
0- 0.3 1.. .................... I. ..................... f. ..................... __
g ; 5 E Ni compoelte (25°C)
0 51 Ni composite (05C)
0 : I
‘5 0.2 L. ..................... .............. 1 C 8n-4Ag-O.5¢u(25°C) ...... _
= 2 [S] X :
c 5 I all-41190500 (05°C)
- 83
g 30 + 3 A Eutectic (25°C)
a, 0.1 _.. ......... . ........ ..................... ._
; ; V Eutectic (65°C)
C] 9 ill-elm Cu.Sn. compoelte (25°C)
0 l 1 fl 1
.0 -e
-5 10 0 5 1 0 0.0001 0.0001 5 0.0002

Global Secondary Creep Bate (lleec)

Figure 6.9 Onset of tertiary creep for different solder joint materials as a function of
secondary creep rate. Composite solders made by mechanical mixing method
generally showed a lower strain for the onset of tertiary creep than non-
composite solders and composite solder made by in-situ method.

joints. In contrast, the OTC occurred at strain levels between 0.1 and 0.15 for Cu and Ag
particle reinforced composite solder joints in which the reinforcements were
mechanically added. The average strain for OTC in the Ni composite solder joint was
0.11, which is comparable to the strain for the OTC in the Cu and Ag composite solder

joints. Strain at the OTC can be considered as a failure criterion to predict the service life
and thus the reliability of the solder joint. As an example, the in-situ Cu58n5 reinforced

composite solder joint starts tertiary creep at a much higher strain level (~0.4) as

compared to the Ni composite solder joint (~0.18). The strain for the OTC as a criterion

would tend to suggest that in-situ Cu68n5 reinforced composite solder joints would

146

perhaps be more reliable in comparison to Cu, Ag, or Ni reinforced solder joints. The
higher strain value for the OTC of in-situ Cu68n5 reinforced composite solder joints, as
compared to the other composite solder joints with mechanically-incorporated Cu, Ag
and Ni particle reinforcements, could be due to the difference in interfacial bonding

strength between the reinforcements incorporated in-situ (intrinsically) and by

mechanically mixing (extrinsically). Nanoindentation tests by Lucas et. al. have shown
that interfacial shear strength of Cu6$n5 reinforcements added in-situ is much weaker [50]
than the interfacial shear strength of Cu, Ag, or Ni reinforcements added mechanically.
The defamation features in an in-situ Cu68n5 reinforced composite solder joint are
shown in Figure 6.10(a)-(b). It is assumed that the weak interfacial bonding of
homogeneously distributed Cu6Sn5 particles provides multiple nucleation sites for
defamation throughout the entire solder joint. A multitude of defamation sites would
promote homogenization of the defamation over the joint while simultaneously reducing
the prOpensity for intense defamation at a single site which would result in the OTC
occurring at significantly lower strain levels. Admittedly, other factor may affect the
strain levels at which OTC occurs, but their contributions are believed to be minimal in
comparison to effect associated with interfacial bonding strength of the reinforcement

types in the composite solder.

6.3.4 Activation Energy for Creep

The secondary creep strain is described quite well by the constitutive equation

proposed by Darn [54]:

147

 

Figure 6.10 Micrographs showing creep defamation of an in-situ Cu68n5 reinforced
composite solder joint. (a) solder joint before creep, where there is no
damage (interfacial debonding, voids, cracks, etc.), (b) solder joint after
creep, showing multiple damage sites ( particle rotation, particle/matrix
interfacial debonding, voids fomation, etc.) across the thickness of the
solder joints, as indicated by arrows. A multitude of defamation sites would
tend to promote homogenization of the defamation over the joint, as shown
in (b).

n —
y = [1(3) exp(_Q), (6-1)
G kT
where 7 is the steady—state creep rate, dis the shear stress, G is the shear modulus, Q is

the activation energy, k is the Boltzmann’s constant, and A and n are constants. The
activation energies for creep for the four solder joint materials were obtained by plotting

the variation of log strain rate vs. I/I' as shown in Figure 6.11. The activation energy for

148

these nan-composite and Cu or Ag composite solder joint materials were in the range of
0.52 eV and 0.56 eV. The Cu and Ag reinforced composite solders revealed slightly
higher activation energies than the non-composite solders. The activation energies
obtained in this study are somewhat less than the activation energy (0.62 eV) obtained by
Raeder et al. [62] for Sn-3.5Ag solder using single shear lap solder joint specimen.
However, the activation energy for creep in Sn-3.5Ag obtained by Yang et al. [61] was
similar to those obtained in the present study, Q = ~ 0.54 eV. The activation energy for
Ni composite solder was 0.64 eV, which was higher than that for all the other solder
materials (0.52 eV-0.56 eV) tested. The higher activation energy for creep for the Ni
composite solder corresponds well with its better creep resistance. The extrinsically-
incorporated particle-reinforced composite solders all revealed somewhat higher
activation energies than the non-composite solders. This suggests that energy barrier for
creep in the composite solders is raised and dislocation motion is hindered by the
reinforcements and by microstructural modifications induced by reinforcement additions.

As reported by Renalds et al. [65], a typical eutectic material that is sufficiently
fine-grained shows a low-stress creep mechanism, which is dominated by grain boundary
sliding. The grain boundary sliding mechanism is frequently seen in fine-grained eutectic
Pb-Sn and other fine-grained solder joints, and is responsible for superplastic creep in
fine-grained materials [65]. This creep mechanism is characterized by an activation
energy for grain boundary diffusion, which is about half of the activation energy for bulk
diffusion: QGB=0-5QBULK- The activation energies obtained for the solders studied

suggest that grain boundary or dislocation pipe diffusion is likely the controlling

149

mechanism governing their creep behavior. At lower temperature and with a high stress

exponent dislocation climb can also be a contributing factor for creep in these solders.

 

.2'5 IIIlll‘tl‘rjlllllvvrtjrrfljlllllrlvllti‘r

 

' I 3 —9— Ag Comp.. 0:0.559V
""" " —El— Cu Comp..o=0.ssev

' 5 —<>- Sn4AgO.5Cu, o=o,ssov
- -)( - - Sn3.5Ag. o=o.szev
- - +- - Ni Comp., 0:0.649V

 
 
 
 
 

11111111111111

 

 

  

 

1111111111

Log Strain Rate,(1leec)
1'.
0|
1111111
'.
l
1
dxé
11111 '

 

 

'65 illlilliilllll1|L111|11I111111111|1111|

 

0.0026 0.0027 0.0028 0.0029 0.003 0.0031 0.0032 0.0033 0.0034

Temperature,ilK

Figure 6.11 Plot of log strain-rate vs. inverse absolute temperature. The activation
energy for creep is listed in the legend. The activation energy for creep for
Ni particle reinforced composite solder is higher than all other solders listed.

6.3.5 Deformation Profiles in Composite and Non-composite Solder Joints

Typical creep defamation profiles of non-composite and Cu particle reinforced
composite solder joints are shown in Figure 6.12(a)-(d) before and after creep
defamation. The SEM micrographs reveal clearly the distortion of the laser-ablation line

patterns due to creep defamation. Significantly more defamation occurred on one side

150

 

Figure 6.12 The defamation profile for (a) non-composite solder joint before creep; (b)
non-composite solder joint after creep; (c) Cu particle reinforced composite
solder joint before creep; and ((1) Cu particle reinforced composite solder
joint after creep.

151

(the left side in the figure) of the solder joint than the other side (the right side in the
figure). The distortion of the laser ablation patterns tends to show that intense
defamation is localized to a particular side of the specimen, but there is no significant
evidence to suggest fracture of the Cu substrate/solder interface. Actually fracture
occurred in the solder near the Cu-Sn interfacial IM layer. Such a defamation behavior
was common in solder alloys and also in Cu reinforced composite solder joints.

Figure 6.13 shows the typical defamation profile of the Ag reinforced composite

 

Figure 6.13 Example of unifom creep defamation in Ag particle reinforced composite
solder joint: (a) before creep; (b) after creep.

152

solder joints. In contrast to non-unifom defamation noted in solder alloys and Cu
composite solder, very unifom defamation was most often exhibited by the Ag
reinforced composite solders. Moreover, unifom creep defamation occurred throughout
the thickness of the joint. The line patterns deformed linearly across solder joint
thickness. The straight-line creep defamation pattern was maintained during the
secondary creep stage for the Ag reinforced composite solder joint. Such highly unifom
shear defamation suggests the strain in Ag reinforced composite is homogeneous [47].
We contend that Ag particle stimulate plastic defamation at numerous sites, and
therefore mitigate localized intense strains in just one or two regions of the solder joint
that eventually develop into defamation bands.

A close examination of the creep-fractured Ni composite solder joint revealed that
most joints deformed heterogeneously during creep. While defamation occurred in the
solder close to one of the copper substrates, still intense defamation and fracture always
occurred in the solder and is not accommodated by fracture in the Cu substrate/solder
interface. Typical creep defamation features in the Ni composite solder are illustrated in
Figure 6.14(a)-(d). Generally, as is seen in Figure 6.14(d), very little lifting of Sn grains
was revealed in the defamation and fracture of Ni composite solder joints, though this
type of defamation features were often observed in the Sn-Ag based solder alloys [7].
Grain boundary sliding as a mechanism for creep was not as prevalent in Ni composite
solder joint. Others [57, 58, 61] reported that dislocation climb is the likely controlling

mechanism for creep of eutectic Sn-3.5Ag solder/joint based on experimentally

153

a.

. 1‘{~)$¢f ._

. \
l~|
".5

"I
‘.

l‘ :{z‘mfi'fi‘

.,‘.

' ' u “" 55.1‘.L.td‘,'—1..j;lfi_ 1. Z I' .
71

240nm

 

(C) ((1)

Figure 6.14 SEM micrographs showing non-unifom defamation in the crept Ni
composite solder joint as exhibited by the distortion of the laser ablation
patterns. (a) whole joint, before creep, (b) whale joint, fracture due to creep,
(c) one part of the joint, before creep, ((1) one part of the joint, after creep.

154

determined values of the stress exponent and the activation energy. Judging from the

activation energy obtained, we contend that dislocation climb/glide is the mostly likely

controlling mechanism for creep in the Ni composite solder joint.

6.3.6 Summary

1.

The creep resistance was improved significantly for Cu particle reinforced composite
solder joints at all test temperatures.

The creep resistance of Ag particle reinforced composite solder joints was
comparable to the steady state creep rate observed for eutectic Sn-3.5Ag and Sn-
4.0Ag-0.5Cu solder joints. Addition of 0.5Cu to eutectic Sn-Ag solder did not
significantly improve the creep properties.

The creep resistance was significantly improved for Ni particle reinforced composite
solder joints at all test temperatures. Ni composite solder joint had the best steady-
state creep resistance of the particle reinforced composite solder joints. Similarly, the
steady-state creep perfomance of the Ni composite solder joints was better than
eutectic Sn-3.5Ag and Sn-4Ag-0.5Cu non-composite solder joints.

The onset of tertiary creep for Ni composite solder joints was reached at a similar
strain level as Cu and Ag particle reinforced composite solder joints. Generally, the
onset of the tertiary creep stage occurred at a lower strain level for composite solders
with mechanically added reinforcement than the non-composite solders and the in-
situ Cu68n5 reinforced composite solder.

The activation energy for creep for Ni composite solder was 0.64 eV, which was

higher than the activation energy values obtained for Cu, Ag composite solders and

155

the eutectic Sn-3.5Ag, Sn-4.0Ag-0.5Cu non-composite solders (0.52eV-0.56eV).
The activation energies for Cu, Ag composite solders and the eutectic Sn-3.5Ag, Sn-
4.0Ag-0.5Cu non-composite solders were very similar which suggests that the
controlling mechanism for creep for these materials is grain boundary or dislocation
pipe diffusion.

6. Defamation in Ag particle reinforced composite solder joints was very unifom
throughout the joint thickness, whereas, the defamation patterns for other types of
solder joints were localized mostly along one side near the Cu substrate/solder
interface. Defamation in Ni particle reinforced composite solder joints was not
unifom throughout the joint thickness. The defamation also occurred close to one
side of the solder joint in the solder. Grain boundary sliding defamation was not as
prevalent in creep of Ni composite solder joints based on microstructural analysis.
Dislocation climb/glide is the mostly likely controlling mechanism for creep in the

Ni composite solder joint.

6.4 Evaluation of Creep Behavior of Near Eutectic Sn-3.5Ag Solders Containing
Small Amount of Alloy Additions

The solder materials used to make the solder joints for the present study are: (1)
Eutectic Sn-3.5Ag solder alloy, (2) Sn-4Ag-0.5Cu and (3) Sn-3.5Ag-0.5Ni ternary solder
alloys, and (4) Sn-2Ag—1Cu-1Ni quaternary solder alloy. The Sn-3.5Ag-0.5Ni solder
alloy was prepared by in-situ methods in our laboratory, and other solder materials were

provided as solder paste by Visteon Technical Center, Dearbom, MI.

156

6.4.1 Microstructure

The microstructures of the solder joints made with these solder alloys are shown in
Figure 6.15. The microstructure of the eutectic and the two ternary alloys are similar.
The background microstructure of all three can be characterized by Sn cells separated by
wide bands having about 15 vol% Ag38n precipitates, as shown in Figure 6.15(a)-(c).
For the 0.5 Cu ternary alloy in Figure 6.15(b), small amounts of globular Cu-Sn
intemetallics were observed that were slightly larger than the Ag38n precipitates. In the
0.5Ni ternary alloy shown in Figure 6.15(c), Cu-Sn intemetallics were somewhat larger,
more rod-like, and more evenly distributed than in the 0.5 Cu alloy. In addition, some
larger blocky Ni3Sn4 intemetallics were observed having a layer of a Cu-Ni-Sn
intemetallic at the interface with Sn (Figure 6.15(c) inset), indicating that Cu partially

invaded the Ni3Sn4 intemetallic. The microstructure of the Sn-2Ag-1Cu-1Ni solder joint

shown in Figure 6.15(d) shows multiple Cu-Ni-Sn intemetallic strengthening phases.

Since there is less Ag in this alloy, the Ag3Sn bands were much narrower than the other

alloys.

6.4.2 Comparison of Secondary Creep Strain Rate under Different Applied Stresses

Secondary creep strain rates were determined for these four solder joint alloys
under different applied stresses, as shown in Figure 6.16. In this figure (as well as
following ones), eutectic Sn-3.5Ag is represented by circles and bold lines, Sn-4Ag-
0.5Cu by triangles and dashed bold lines, Sn-3.5Ag-0.5Ni by diamonds and narrow lines,

and Sn—2Ag-1Cu-1Ni by squares and narrow dashed lines. The solid symbols represent

157

(rsAg‘ssn‘ .'

SI] I. if. :r ' --

Cu-Sn lMCs - Sn

.'. _ .: .‘
' ,e , . . /.:T-'.. ,
‘ ‘ I . ' . .
. Ll \' .‘ _. J

r. “\ r' . ' . . ’u
' ‘ ' ' -- ,2" Cil-Ni-Sn' was,“

“4," , «I. ~".
:- (Han. ...
L _.
. ‘ " ,N e.
.~ _ "_':
’. a .
l_; ‘ "I\ . a
.f' ‘
.‘_
. I

\ ...
e' ' .
”—- -

I

 

(C) (d)
Figure 6.15 Microstructures of solder joint materials: (a) Eutectic Sn-3.5Ag solder allay,

(b) Sn-4Ag-0.5Cu solder alloy, (c) Sn-3.5Ag-0.5Ni solder alloy (inset
indicating Cu-Ni-Sn intemetallics), and (d) Sn-2Ag-1Cu-1Ni solder alloy.

158

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.001 E r r T r I Y
- Sn-3.5Ag, RT
0 0°01 F‘“ -' - Sn-4A9-0.5Cu, RT
f -o— en-3.5Ag-0.5Nl, RT
,0. 1 -e- Sn-2Agl1N11Cu,RT
O .5 fl
5 10 E— ..................................... ; ..................... I ...........
; E
a: .
c -0 ,__ ........................................................
ti 1° 5
= t
10 .
10-1 E ............................................... ' ............................................... _
10'. l A
5 e 7 e a 10 20
Stress (MPa)
(3)
0.001 a
0.0001 -=
g 1
5 10" ‘2
3; a
. .
m -n
C 4 _ .......................................................................................... _
2 ‘° "
10 C 1
F 5 -e- all-3.5119. sec 1
10-7 E'""I ................................................. a - Sn-CAg-OJCuJSC .3
5 A— en-s.5Ag-0.5Nl.esc 5
: l’ -13- Sn-2Agl1NliCu.050 :
10'. 1 1 m 1 4
5 e 7 e e 10 20

Streee (MPa)

(b)

Figure 6.16 Comparison of steady-state creep rates of the four solder alloys. (a) room

temperature data, (b) 85°C data.

159

room temperature creep experiments and the open symbols represent the 85°C creep
experiments.

Figure 6.16(a) compares the secondary creep rates of room temperature
experiments for all four alloys. In contrast to the eutectic and quartemary lNi-lCu alloy
that exhibit a single stress exponent, the ternary alloys with 0.5 Cu or 0.5 Ni show a
higher (and similar) stress exponent at lower stresses and a lower stress exponent at
higher stresses. Such transitions in the stress exponent suggest the operation of a
threshold stress. The ternary alloy with 0.5 Ni is the most creep resistant alloy (at
stresses below about 15 MPa). Both of these ternary alloys have a crossover point in
common with the eutectic alloy near 13 MPa, where the creep resistance is similar. The
lCu-lNi quaternary alloy and the eutectic alloy have similar stress exponents, but the
quaternary alloy has secondary strain rates about 10 times higher than the eutectic over
the tested range of stresses. Due to the crossover behavior, the relative creep resistance
of the alloys changes from low stress to high stress is summarized in Table 6.1 (using
extrapolation of some data).

Figure 6.16(b) shows the 85°C creep data, which show similar strain rates for
stresses that are about half of the room temperature values. Similar trends are observed
as at room temperature, except the lCu-lNi quaternary alloy is the least creep resistant
because it is about 100 times less creep resistant than the eutectic alloy. Due to cross
over effects, creep resistance is similar for the eutectic and the ternary alloys at a stress
near 8 MPa.

In Figure 6.16, both eutectic data sets, and the 85°C lCu-lNi set have apparent

outlier datum points that are stronger than the rest of the specimens. From related work,

160

this variability is not due to experimental errors, but is usually correlated with more
homogeneous strain across the joint than is commonly observed, where a preferential
shear band develops that is often near one of the two interfaces.

Table 6.1 Ranking of Creep Resistance of Solder Alloys Based Upon Best Fit Power

 

 

 

 

 

 

 

 

 

 

Law Creep
Room Temperature ~10 MPa low stress rank ~20 MPa high stress rank
Least Sn-2Ag-lCu-1Ni Sn-4Ag-0.5Cu
. eutectic Sn-3.5Ag Sn-3.5Ag-0.5Ni
Creep resrstance
Sn-4Ag-0.5Cu Sn-2Ag-lCu-1Ni
MOSt Sn-3.5Ag-0.5Ni Eutectic Sn-3.5Ag
85°C ~6 MPa low stress rank ~12 MPa high stress rank
Least Sn-2Ag-1Cu-1Ni Sn-2Ag—1Cu-1Ni
. Sn-4Ag-O.5Cu Sn-3.5Ag-0.5Ni
Creep resrstance
eutectic Sn-3.5Ag Sn-4Ag-0.5Cu
MOSt Sn-3.5Ag-0.5Ni eutectic Sn-3.5Ag

 

 

 

 

 

Sn-3.5Ag-0.5Ni solder joints thus exhibited best creep resistance among all the
joints tested. The reason for the difference in creep resistance among these solder alloys
may be explained from the microstructural point of view. Both the intemetallic
reinforcing phase famed in the matrix as well as the small matrix Ag3Sn particles play
important roles in the improvement of the creep behavior of these solder alloys. There is
not much nricrostructural difference between eutectic Sn-3.5Ag and Sn-4Ag-0.5Cu
solder. But the small amount of Cu-Sn intemetallics present in the Sn-4Ag-O.5Cu
improved the creep resistance of this alloy to a little extent. That’s why comparable but
slightly higher creep resistance was observed in this alloy compared to eutectic Sn-3.5Ag
solder. The presence of Ni in the Sn-3.5Ag—O.5Ni solder seems to stimulate the fomation

of Cu-Sn intemetallic phases in the matrix though the Cu comes only from the substrate

161

during soldering process. On the contrary, much less number of Cu-Sn intemetallic
particles were observed in the Sn-4Ag-0.5Cu solder alloy though Cu is already contained
in the solder prior to joint fabrication. It is reasonable to relate more intemetallic
reinforcements to the improved creep resistance, as exhibited by Sn-3.5Ag-0.5Ni. In
addition to the increased number of the reinforcement in the Sn-3.5Ag-0.5Ni alloy, they
are also larger and more rod-like than in the 0.5 Cu ternary alloy. The larger size and rod
shape may constrain plastic flow during creep defamation because the height of the
climb barrier is larger [128], which hinders dislocation climb and improves secondary
creep behavior. However, the shape of the intemetallics tends to become more globular
with an increase of Cu, as is shown in Figure 6.15(d) for the Sn-2Ag—1Cu-1Ni solder
alloy. This quaternary alloy also has lower silver content, which implies that fewer
Ag3Sn particles exist, and this may also account for the poorer creep resistance.
Therefore it may be possible to improve creep resistance of the Sn-Ag solder alloy by

increasing the Ni content carefully.

6.4.3 Comparison of Creep Behavior of Alloys in Normalized Strain Rate vs.
Normalized Stresses Plot

Nomalized secondary creep strain rates are plotted versus nomalized stresses for
these solder joint materials in Figure 6.17 along with Darveaux’s data for aged eutectic
Sn-3.5Ag solder joints [58]. Darveaux' data show the effect of power law breakdown at

high creep stress in the shape of the curve that is often modeled using

yzsinh(aa)" exp(-Ql RT). Thus, creep data are expected to have slight upward

curvature in the stress and strain rate space examined in this paper. It is evident that the

eutectic and the lCu-lNi quaternary alloy creep data obey the power law relation, but the

162

 

O
I
................................................................ k......... s . . --...-.....
0.1 ,__ : .......... 1
9

0.01 :— ------------------------------------------------------------- 1 -------------------------------- 1

 

  
 

 

1
...........................................................................................
I
i
9
e

Darveeuaetal

e 004.511.,111

; 0 0114.: .es'c

0.0001 """" DWI-8m:.SCu,RT

: ' 'V' ' smug-0.11m. ee'c

. —.—- Sn4.3A¢-0.5Nl. at

10 : .................................................. ------- —0—en4.sAg-0.sul.ee‘c
~ : I 804MC01NIJ‘T

Normalized Strain Rate

 

 

_ 5 U all-2m curlil. ea'c

 

 

0.0004 0.0006 0.0008 0.001

Normalized Streee

Figure 6.17 Nomalized steady-state creep strain rates versus nomalized stresses for
eutectic Sn-3.5Ag, Sn-4Ag-0.5Cu, Sn-2Ag-lCu—1Ni, and Sn-3.5Ag-0.5Ni
solder joints along with Darveaux’s data for aged eutectic Sn-3.5Ag solder
joints [58].

ternary alloys exhibit high stress exponents at lower stress due to the operation of a
temperature dependent threshold stress, above which a particular defamation mechanism
is able to operate [129-131]. Temperature dependent threshold stresses are known to
operate in alloys with solute atom strengthening or incoherent particle strengthening.
The higher threshold stresses in the nickel ternary alloy account for the higher creep
resistance observed at lower stress. Over the range of stresses examined in the eutectic
and lCu-lNi quaternary alloy, there is no apparent threshold stress phenomena.

Also, the eutectic data is generally more creep resistant than the Darveaux’s data.
The creep resistance in Darveaux’s specimens may have been reduced by the aging

treatment, in contrast to the present data, where all the solder joints were deformed in the

163

as-fabricated condition within a week after they were fabricated. Compared to the other
alloys, the eutectic data are rather noisy. We have observed that compared to the other
alloys, the eutectic alloy is most likely to develop a shear band close to one or the other
interface, and this tendency to develop a plastic instability may account for the range of
scatter. Figure 6.17 also illustrates that Sn-2Ag-lCu-1Ni solder joints are the least creep
resistant among all the tested solder joints. This alloy had significantly less Ag than the

others, so this may account for the weakness of creep resistance of this alloy.

6.4.4 Strain and Time at the Onset of Tertiary Creep

The strain at which the tertiary creep stage starts may be useful for modeling and
predicting the service life of solder joints. A comparison of the strain for the onset of
tertiary creep for these solder alloys is shown in Figure 6.18(a) and (b) for room
temperature and 85°C, respectively. At room temperature, the onset of tertiary creep
stage is delayed to strains near 0.6 for the ternary 0.5Cu solder joints, whereas, it is ~0.3
for eutectic, the 0.5Ni ternary, and the lCu-lNi quaternary solder joints.

At 85°C, the shear strain at the onset of tertiary creep is lower, and tends to show
the reverse phenomena as at room temperature. Tertiary creep is delayed the most for the
quaternary 1Ni-1Cu (~0.5) solder joints, compared to eutectic Sn-3.5Ag (~0.3), and near
~0.2 for the ternary 0.5Cu and 0.5Ni solder joints. Thus the quaternary lCu-lNi alloy
may provide beneficial compliance for some applications. This enhancement of shear
strain for the onset of tertiary creep can be attributed to the ability of the material to resist
heterogeneous defamation, analogous to necking in tensile tests. Alloys with a high
stress exponent (low strain rate sensitivity) are intrinsically less able to resist localized

deformation [132]. Also, spherical intemetallics would be less likely to stimulate

164

Strain at the Oneet of Tertiary Creep

Strain at the Oneet of Tertiary Creep

 

 

 

1.2 1 ' T T T l ‘ i
- O
. Sn-3.5Aa. RT

1 -. ' Sn-4Ao-0.5Cu. RT .
” ’ Sn-Ao-0.5Ni. RT
h I .

 

 

 

 

 

 

 

 

 

 

 

 

5 10 15 20
Streee (MPa)
(3)
1 .2 L fi . r I I r 1' fl 1 r T T i !
. o 311-35119. 850 1
i v 311411190500. 850 J
1 _ ...... _ ................................................................. a
o Sn-Ag-0.5Nl. 05c .
: C1 SnoAg/1Ni10u, 850 4
0.0 '

0.6

0.4

0.2

 

 

 

 

 

 

 

15
Streee (MPa)

0))

165

25

 

 

  
  
  
   

 

 

 

 

 

 

 

 

 
   
   
 

 

 

 

 

 

\ V * Sn—SJAg. RT
3. 10. _ _____________ _\ _____________________________________________ .1 - emu-Men, er __
g —e— smug-0.0m. er
0 an-zAg-1Nl-1cu. RT
e
.2
t
l! a
.5 10 ”Wt; """"""""""""" y """"""""""""""""""""""""""""" ‘
e-o “‘~
e
e
c
O
2
z 10‘ .--.... ..... ...-.. ..- . . ................ fl
e
o
E
1...
1000 1 m a r 1 1 1 1 m A 1 1 1 1 1 L 1 1 1
10 20
Streee (MPa)
(e)
10' -
1'; Sn-SJAQ. are
3’: M4A9-OJCII. ee'c
g- an-aaao-oam. ee‘c
3 1o5 ell-zAg-ml-rce. ee'c '*
Z‘
.2
t:
.2
4 ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, d
3 10
‘6
e
c
O
o 3
5 1000 """ ’13 """"""""""""""""""""""" T
‘6 E
o :
E :
l- s
1 .
100
5 6 7 8 9 10 20
Streee (MPa)
((0

Figure 6.18 (a) Average strains for the onset of tertiary creep at room temperature, (b)
average strains for the onset of tertiary creep at 85°C, (c) time for the onset of
tertiary creep at room temperature, (b) time for the onset of tertiary creep at

85°C.

166

localized defamation gradients than rod-shaped reinforcements. The 0.5Ni ternary alloy
having more rod-shaped reinforcements has the lowest strains to tertiary creep of the
alloys tested. The relative merits of these alloys in the strain to the onset of tertiary creep
is summarized by rank in Table 6.2, and compared with secondary creep resistance.
Figures 6.18(c) and ((1) illustrate the time to tertiary creep instead of strain to
tertiary creep. It is clearly shown that the ternary Sn-4Ag-0.5Cu and Sn-3.5Ag-0.5Ni
alloys exhibit longer defamation time before the onset of tertiary creep than the eutectic
Sn-3.5Ag alloy and the lCulNi alloy at a similar stress level at bath room temperature
and 85°C at lower stress levels. This proves the improved creep resistance for the ternary
Sn-4Ag-0.5Ni and Sn-3.5Ag-O.5Cu alloys because it takes longer time for these alloys to
reach the final creep stage where fracture usually occurs as soon as tertiary stage started.

Table 6.2 Comparison of Creep Properties of Solder Joints Made with Eutectic Sn-
3.5Ag, Sn-4Ag-0.5Cu, Sn-2Ag-1Cu-1Ni, and Sn-Ag-0.5Ni Solders

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Property Steady-state Onset of
Solder Material Creep Rate Tertiary Creep
Room Temp. + + + + + +
RoomTemp. ++++ ++++
Sn-4Ag—0.5Cu 8 5°C + + + + +
. Room Temp. + + + + + +
Sn-2Ag-1Cu-1Nr 85°C + + + + +
. RoomTemp. +++++ ++
Sn-Ag-0.5Nr 85°C + + + + +
Note:

+ + + + +: Excellent; + + + +: Very good; + + +: Good; + +: Marginal; +: Poor.

6.4.5 Microstructural Examination of Deformed Solder Joints
Representative micrographs that illustrate defamation on previously polished
surfaces in the crept solder joints are presented in Figure 6.19(a)-(e) for the four alloys.

Preferential defamation on one side was observed for all the solder joint materials. The

167

 

(C) (d)

168

 

Figure 6.19 Creep defamation in the solder joints: (a) Eutectic Sn-3.5Ag solder, (b) Sn-
4Ag-0.5Cu solder, (c) Sn-Ag-0.5Ni solder, (d) Sn-2Ag-1Cu-1Ni solder, and
(e) a magnified view of the defamation in a Sn-Ag-0.5Ni solder joint.

surface features of all the tested solder joints are similar, as shown in Figure 6.19 The
most important microstructural characteristic is the decohesion at Sn-Sn grain boundaries
that leads to ultimate failure of the solder joint. This feature was often observed along
the side that had intense defamation. The grain boundary decohesion is apparently due to
the grain boundary sliding effect on particular boundaries which makes the Sn grain (or
cluster of grains) to either slide up or sink down with respect to neighboring grains within

the solder. A magnified view of grain boundary decohesion is illustrated in Figure

169

6.19(e). These nricrographs suggest that the defamation mechanism that leads to failure
in creep of these solder joints is grain boundary sliding. The fact that the stress
exponents are high and variable (and not close to 2, which is typical for superplasticity),
implies that this defamation mechanism does not operate uniformly throughout the joint,

nor is it the rate limiting process of defamation.

6.4.6 Summary

Creep tests were conducted on solder joints made with eutectic Sn-3.5Ag, two
ternary alloys, Sn-4Ag-0.5Cu, Sn-3.5Ag-0.5Ni, and a quaternary Sn-2Ag-1Cu-1Ni solder
alloy. The relative perfomance of these alloys using secondary creep rate and strain to
tertiary creep is summarized in Table 6.2.

1. The ternary alloys showed improved creep resistance at low stresses due to threshold
stress phenomena, but similar defamation behavior as the eutectic solder at higher
stresses.

2. The ternary alloys with 0.5 Ni showed substantially better creep resistance at lower
stresses. The quaternary alloy with about half the silver as the other alloys had
uniformly poorer creep resistance.

3. The average strain for the “onset of tertiary creep” at room temperature in Sn-4Ag-
0.5Cu solder joints was higher than solder joints made with the other three alloys, but

it was low at elevated temperature, where the quaternary alloy had the largest strain

before tertiary creep.

170

4. Sn-4Ag—0.5Cu and Sn-3.5Ag-0.5Ni alloys exhibit longer defamation time before the
onset of tertiary creep than the eutectic Sn-3.5Ag alloy and the lCulNi alloy at a
similar stress level at bath room temperature and 85°C.

5. Microstructural examination showed significant creep defamation along particular '

Sn-Sn grain boundaries, and fracture nucleation probably developed from boundaries

that were sliding excessively.

171

CHAPTER VII
CONCLUSIONS AND RECOMMENDATIONS

7.1 Conclusions

In the current investigation, composite solders were prepared by mechanically
incorporating metallic Cu, Ag, or Ni particle reinforcements to the eutectic Sn-3.5Ag
solder paste. Microstructural characterization of composite solder joints before and after
solid—state isothemal long term aging and multiple reflow were carried out. Effects of
solder reflow on wettability and mechanical properties of these solder materials were
studied. Creep tests were performed to study the creep defamation behavior of the
composite solder joints as well as solder joints made with near eutectic Sn-Ag solders
with small alloying element addition.

The major results of this investigation are concluded as follows.

1. In the as fabricated composite solder joints, extensive fomation of Cu-Sn IM layers
around the Cu reinforcements was observed, whereas the fomation of Ag-Sn around
the Ag reinforcement was minimal. The IM layers developed around the Ni particles
were also significant and can be characterized by a “sunburst” pattern surrounding the
Ni particle reinforcements. There is a significant number of ~0.5 um sized small 1M
particles scattered, mostly, along periphery of Ni reinforcement particles. A few of
these small IM particles appear in the eutectic Sn-3.5Ag matrix. The morphology of
IM layer around Ni particles was affected due to different heating rate, but not
cooling rate. With the heat input equivalent to three reflows, the sunburst shaped IM
layer around Ni particles can be completely transfomed into a single-crystal-faceted

shape. The M layer around the Cu reinforcement was a co-layer consisting of a

172

Cu38n layer close to Cu particles and a Cu6Sn5 layer close to the solder. The [M layer
famed around Ag particles was thin Ag3Sn layer. The 1M layers surrounding the Ni
reinforcements were detected as a Cu-Ni-Sn ternary IM layer. The IM layers famed
at the Cu substrate/solder interface were found to be Cu-Sn IM layer in both the Cu
and the Ag particle reinforced composite solder joints. This Cu-Sn 1M layer is also a
co-layer consisting of a CU3Sn layer close to Cu substrate and a Cu6Sn5 layer close to
the solder. However, the IM layers famed at the Cu substrate/solder interface in the
Ni composite solder joints was determined as a co-layer consisting of a Cu-Sn [Ni
layer close to Cu substrate and a ternary Cu-Ni-Sn 1M layer close to Ni composite
solder. The thick IM layer famed around the Cu or Ni particles is due to the fast
diffusion behavior of Cu or Ni atoms in Sn, whereas the diffusion of Ag in Sn is
much slower which results in a significantly thinner Ag-Sn intemetallics around Ag
particles.

. Isothemal aging studies showed that more intemetallics famed in Cu composite
solder joints (both around the reinforcements and at the substrate/solder interface)
than in Ag composite solder joints after aging for 1000 hours. Cu reinforcements
were totally converted to Cu6Sn5 intemetallic with sufficient aging, whereas Ag
particles were only partially converted to Aggsn intemetallic. A proposed
phenomenological model shows that the growth of 1M layers around the Cu particle

reinforcements and at the substrate/solder interface is produced by interdiffusion of
both Sn and Cu atoms through the e and ’1] IM layers. Solid state isothemal aging
studies at 25, 50, and 100°C for ~1000 hours induced only slight modifications in the

microstructure of Ni composite solder joint. However, aging at 150°C revealed that

173

the microstructure was unstable with profuse growth of the IM layer and coarsening
of the Ni reinforcement as a result of Cu-Ni-Sn intemetallics fomation and growth.
The large volume fraction of intemetallics famed in the Ni composite solder joint
after aging at 150°C for 1000 hours is probably due to the presence of Cu, in addition
to Ni and Sn, in the intemetallics famed in the Ni composite solder.

. Wettability studies showed that eutectic Sn-3.5Ag solder paste has the best wetting
characteristics on Cu substrate among all the solder materials studied. The 15 v% Ni
composite solder showed comparable wettability to eutectic Sn-3.5Ag solder paste.
The wettability of 15 v% Ag reinforced composite solder was significantly better than
17 v% Cu reinforced composite solder. The poor wettability of Cu composite solder
is primarily due to the increase in the effective volume fraction due to reflow. Such
increase in volume fraction was not observed in the Ag and Ni composite solder after
multiple reflow. Wettability can be improved by lowering the volume fraction of the
reinforcing phase. No significant changes in contact angles were observed with
multiple reflow for all the solder materials studied.

. The effects of multiple reflow on the microstructure of the composite solder joints
were investigated and compared with eutectic Sn-3.5Ag solder. The most significant
microstructural changes were observed in the Cu composite solder joints. Growth of
the 1M layer around the Cu particle reinforcements was excessive leading to total
consumption of the Cu particles after 3-4 reflows. No significant coarsening of the
Ag reinforcements was evident after multiple reflows. Multiple reflow did not
significantly change the microstructure in Ni composite solder joints. The IM layer

growth was minimal, although coarsening of Ag38n particles was observed. Reflow

174

studies also showed that the 1M layer at the Cu substrate/solder interface also grew
much faster in Cu composite solder joints than in Ag composite solder joints.
However, multiple reflow did not cause significant growth of interfacial IM layers in
Ni composite solder joints. Therefore, both Ag and Ni composite solders exhibited
much more stable nricrostructure in their joints under the same reflow conditions than
Cu composite solder. The microstructure of first reflowed eutectic Sn-3.5Ag solder

can generally be characterized by small Sn cells surrounded by wide-banded eutectic

Ag3Sn phase. In comparison, the microstructure of multiple reflowed solder is

characterized by large Sn cells circumvented by a thin necklace of eutectic Ag3Sn

precipitates.

. The effects of reflow on the mechanical properties of eutectic Sn-3.5Ag solder were
evaluated from the data of hardness, yield strength, steady-state creep rate, and stress
exponent for creep using nanoindentation testing. The hardness and yield strength of
multiple reflowed eutectic Sn-3.5Ag solder were reduced by 30% after three reflows.
This finding is commensurate with the increasing size of Sn cells produced by
multiple reflow as a larger grain/cell size gives a lower yield strength. The stress
exponent, n, determined using indentation creep testing was 7 for first reflowed
solders and 8 for multiple reflowed solders. From the stress exponents observed, the
steady-state creep rate for the multiple reflowed solder will be higher compared to
first reflowed solder.

. In terms of creep properties of the composite solder joints, it was found that the creep
resistance was significantly improved for Ni particle reinforced composite solder

joints at all test temperatures. Ni composite solder joint had the best steady-state

175

creep resistance of the particle reinforced composite solder joints. Similarly, the
steady-state creep perfomance of the Ni composite solder joints was better than
eutectic Sn-3.5Ag and Sn-4Ag-0.5Cu non-composite solder joints. The creep
resistance was also improved significantly for Cu particle reinforced composite solder
joints at all test temperatures. However, the creep resistance of Ag particle reinforced
composite solder joints was comparable to the steady state creep rate observed for
eutectic Sn-3.5Ag and Sn-4.0Ag-0.5Cu solder joints. Addition of 0.5Cu to eutectic
Sn-Ag solder did not significantly improve the creep properties.

. The onset of tertiary creep for Ni composite solder joints was reached at a similar
strain level as Cu and Ag particle reinforced composite solder joints. Generally, the
onset of the tertiary creep stage occurred at a lower strain level for composite solders
with mechanically added reinforcement than the non-composite solders and the in-
situ Cu68n5 reinforced composite solder.

. The activation energy for creep for Ni composite solder was 0.64 eV, which was
higher than the activation energy values obtained for Cu, Ag composite solders and
the eutectic Sn—3.5Ag, Sn-4.0Ag-0.5Cu non-composite solders (0.52eV-0.56eV). This
data corresponds well with the improved creep resistance of Ni composite solder
joints. The activation energies for Cu, Ag composite solders and the eutectic Sn-
3.5Ag, Sn-4.0Ag-0.5Cu non-composite solders were very similar and close to the
activation energy for grain boundary diffusion, which suggests that the controlling
mechanism for creep for these materials is grain boundary or dislocation pipe

diffusion.

176

9.

10.

Although the improvement of creep resistance for Ag composite solder joint was
insignificant, defamation in Ag particle reinforced composite solder joints was very
unifom throughout the joint thickness, whereas, the defamation patterns for other
types of solder joints were localized mostly along one side near the Cu
substrate/solder interface. The defamation in Ni composite solder joints also
occurred close to one side of the solder joint in the solder. Grain boundary sliding
defamation was not as prevalent in creep of Ni composite solder joints based on
microstructural analysis. Dislocation climb/glide is the mostly likely controlling
mechanism for creep in the Ni composite solder joint.

Creep tests were conducted on solder joints made with eutectic Sn-3.5Ag, two ternary
alloys, Sn-4Ag-0.5Cu, Sn-3.5Ag-0.5Ni, and a quaternary Sn-2Ag-lCu-1Ni solder
alloy to study the effects of alloying element addition on the creep behavior of
eutectic Sn-3.5Ag solder. The ternary alloys showed improved creep resistance at
low stresses due to threshold stress phenomena, but similar defamation behavior as
the eutectic solder at higher stresses. The ternary alloys with 0.5 Ni showed
substantially better creep resistance at lower stresses. The quaternary alloy with
about half the silver as the other alloys had unifomly poorer creep resistance. The
average strain for the “onset of tertiary creep” at room temperature in Sn-4Ag-0.5Cu
solder joints was higher than solder joints made with the other three alloys, but it was
low at elevated temperature, where the quaternary alloy had the largest strain before
tertiary creep. Sn-4Ag-O.5Cu and Sn—3.5Ag—0.5Ni alloys exhibit longer defamation
time before the onset of tertiary creep than the eutectic Sn-3.5Ag alloy and the

lCulNi alloy at a similar stress level at bath room temperature and 85°C.

177

Microstructural examination showed significant creep defamation along particular
Sn-Sn grain boundaries, and fracture nucleation probably developed from boundaries

that were sliding excessively.

7 .2 Closing Thought I Recommendations

There has been a lot of research effort involved in the present investigation for
composite solders with mechanically incorporated Cu, Ag, and Ni particle
reinforcements. Many tests, such as aging, reflow, nanoindentation and creep etc., have
been performed to evaluate these composite solders. Since this batch of the experiment
was just the starting exploration of composite solders made with extrinsical incorporation
method, there were several places that the author feels need to have more input and
further investigation due to lack of experience or experimental conditions. Some key
knowledge or specific mechanism is still unclear and more effort should be pursued to
make the current investigation more complete and comprehensive.

1. The volume fraction of the composite solders was 15% used in the current
investigation. This cause a problem that the effective volume fraction of Cu or Ni
composite solders will be much more than 15 v% under aging or reflow conditions at
elevated temperatures, which leads to poor wettability and brittle solder joint. In the
future investigation, much lower initial volume fraction should be used in order to
control the ultimate volume fraction of the reinforcement to be around 20%. An
initial 5 v% of reinforcement addition may be a good starting point. The effect of
different volume fraction on aging, reflow, and creep characteristics of the composite

solder can be pursued if possible.

178

The creep testing on these composite solder joints was focused primarily on the
defamation analysis of the solder joints where microstructural features due to creep,
global and localized strain distribution, strain for the onset of tertiary creep etc., were
evaluated. In the future investigation, a systematic study on the creep defamation
mechanism of these composite solder joints can be pursued using a wide range of
applied stresses at four different temperatures. The effect of the particles added to
the eutectic Sn-3.5Ag solder on the responsible creep mechanisms can be discussed
based on a series of systematic creep tests. Since Cu and Ni composite solder joints
showed improved creep resistance in terms of steady-state creep rates, it would be
meaningful to start with these two solder materials. Also, creep tests can be carried
out on one solder joint with fixed stress level while the testing temperature being
varied. This is a possible means to have a quick evaluation of activation energy for
creep for the materials tested.

As for the activation energy for creep, our data showed that the activation energies
for these composite solders are about half of the self-diffusion of Sn, which are
typical grain boundary diffusion activation energies for grain boundary sliding type
of defamation mechanism. Our surface damage due to creep also showed significant
grain relief effect. If grain boundary sliding is the controlling mechanism, does
adding reinforcements still contribute to the improved creep resistance? Which is the
most contributing factor need to have further thought. However, there is no clear
agreement in the experimental data that clearly defines the Operating creep
mechanisms in the literature just for the eutectic Sn-3.5Ag solder. This aspect needs

to be further clarified, with possible repeated experiments, by extensively studying

179

stress exponent, activation energy as well as microstructural features of creep
defamation.

It is also advisable to study the residual mechanical properties (eg. creep, simple
shear pr0perties) of these composite solder joints after aging and reflow when
significant microstructural evolution has taken place. Alternatively, changes of
mechanical properties with aging or reflow can also be pursued.

From out laser pattern and established creep data extraction technique, both global
and localized creep parameters can be evaluated. We always use global secondary
creep rate and global strain for the onset of tertiary creep as parameters to compare
creep resistance and service life. Are the global parameters the controlling factors or
the localized ones? This aspect needs to be further pursued possibly by using one
parameter/variable to represent comprehensive effects on the creep behavior caused
by both the global and local parameters.

The reflow temperature profile can be modified in the future investigation where
different heating/cooling rate and even different dwell time at peak temperature can
be pursued. The current profile has very short time above melting point where
sometimes it may not cause discernible changes in the microstructure even due to 4
reflows, as reported for Ni composite solder. Alternatively, increasing the number of
reflows under current temperature profile for Ni composite solder may be helpful.
The premise to clarify the above observed disagreements is to consistently have good
quality solder joints ready for the tests. Solder joints with different thickness or fillet
shape/size have to be rejected. Therefore, methods need to be further developed and

completed to improve the integrity of the solder joint though we are making great

180

progress towards that. For example, the fomation of voids should be avoided or
lowered to the minimum extent by choosing proper flux or improving our current
processing method in the lab environment. Proper references from industries may be

taken and adopted.

181

APPENDICES

182

APPENDIX A

PROCEDURES FOR THE CALCULATION OF GLOBAL AND LOCALIZED
CREEP PARAMETERS
1 Image Processing

1.] Image Selection

A series of images were captured during creep testing at certain time intervals.
Usually more than 20 or 30 images were captured for each individual creep testing using
the old creep testing technique. More than 100 images were obtained with the newly
developed creep testing fixture with an CCD camera that can automatically capture at set
time intervals. To assure that reasonably correct creep parameters to be obtained in the
subsequent date extraction steps, usually, 12-15 images were needed, as shown in Figure
A1. In Figure A1, totally 15 out of 38 images were selected from a creep testing on in-
situ Cu68n5 particle-reinforced eutectic Sn-3.5Ag solder joints. We will use this creep test
as an example for the following image process, data extraction/analysis illustration. The
selection of these 15 images should represent all the three creep stages with ~4-5 images
in each stage. Sometimes, more images are needed at the transition point from steady-

state creep to tertiary creep stage in order to assess the strain or time at the onset of

tertiary creep.

1.2 Image Processing Using Adobe PhotoShop0
The purpose of image processing is to provide a series of images with consistent
size and quality for the subsequent data extraction in DataThief®. The following steps are

needed to accomplish the image processing.

183

 

184

   

   

(11) (12)
(13) (14)
(15)

 

Figure Al Example of selection of creep images. Totally 15 out of 38 images were
selected from a creep testing on in-situ Cu6Sn5 particle-reinforced eutectic Sn-
3.5Ag solder joints. These 15 images represent creep defamation in the solder
joint at different time intervals.

(1) Start Adobe PhotoShop®.

(2) Open all the 15 images ready to process. This can be done by going to “File” menu,

choosing “Open”, going to the proper directory, then selecting first image and
clicking the last image while holding the “shift” key. By this way, 15 images are

selected simultaneously, as shown in Figure A2. Click “open” after selection. All 15

images with TIF fomat thus are opened in Adobe PhotoShop®.

185

mm ;
m <-.m.~.rm —*: .

2
y
g
f1

 

Figure A2 Open the selected images in PhotoShop".

(3) The last image is the first one to process because last image represents largest
defamation throughout creep testing. Make the last image (in this example, the last
image is #15) as your current opened file by selecting the corresponding file name
from WINDOW menu, as shown in Figure A3.

(4) The first step to process these pictures is to rotate the canvas so that the Cu
substrate/solder interfaces of the solder joint are exactly perpendicular.

(5) There are many excimer laser marks on the solder joint. Pick a line that is clear
throughout the creep testing, for example, the line as shown in Figure A4. The

defamation of this line will be traced in every picture.

186

 

Figure A3 The last image is the first one to edit.

'1

g
I‘
g
.3
F,

 

Figure A4 Pick a line that is clear throughout to trace.

187

(6) Inverse image of the original image is often used because the line is clearer than the

original image. Invert the image by using Ctrl+I, as seen in Figure A5.

This line is still clear.

‘ mannerisms _.

 

Figure A5 Invert the original image to edit.

(7) Select a rectangular region that contains the line you want to edit, using the rectangle
selection tool. Make sure that rectangle box is small as long as it contains the line.
This step is illustrated in Figure A6.

(8) Crop the image so that it only contains the selected rectangular area, as seen in Figure
A7.

(9) Carefully trace the line from the left side of the solder joint to the right side, using the
line tool with black foreground color. It is better to zoom the area to 200% to
promote accuracy when drawing the line. See Figure A8 for an example of the line

traced.

188

Figure A7 The image is cropped to the line region.

189

l
i
g
:5
9
..4‘
.a
r:
i
8‘.

 

E
5;?
a.

 

Figure A8 Carefully trace the line. Use 200% magnification when tracing the line.

(10) Now, we have had images with different sizes due to the amount of deformation at
different stages. It is necessary to make all the images have the same size so that it is
under the same magnification when opened by DataThiet° in the next step.
(Otherwise, DataThief° will end up blowing a small sized image the same size as a
big image. ) This can be accomplished by setting the same canvas size to all the
images. So, the canvas size should be determined from the biggest image, i.e., the
last image. The canvas size change can be done by selecting “Canvas Size” from the
“Image” menu. Usually, the canvas size for the last image could be just a little bigger
than the image size itself. When you change the canvas size for the last image, write
down the height and width for your record so that they can be entered consistently
when you change the canvas sizes for other images later. This step is illustrated in

Figure A9.

 

Figure A9 Canvas size change. (a) select “canvas size” in “image” menu. (b) Change the
width and height in the box. (c) A white margin was formed due to canvas size
change.

(11) Image analysis can be finalized by saving the image files as “.pct” files for use in
DataThiefO Macintosh version, or alternatively, “.gif” files for use in DataThief’ PC

version, as shown in Figure A10.

 

Figure A10 Save the edited image to “.gif” file type for use in Data'I'hiefIn PC version.

(12) Repeat steps (4)-(l 1) for prcessing other images.

2 Data Extraction Using DataThier’

DataThief° is a program to reverse engineer a set of data from a given plot in a
magazine or journal. This program gives you the opportunity to incorporate somebody
else's data points in your plots. This comes in very handy when f.i. you would like to
compare your data with the data in a published article for which you don't have the data

in table format.

192

The purpose of using DataThiefD is to extract a group of data points across the
solder joint thickness (localized places) along the prescribed deformation lines at
designed time intervals under the same coordinate system.

The following steps are needed to accomplish the data extraction.

(1) Set up a coordinate system. This can be done by using a transparency on which a
square grid pattern is printed. More grid points will facilitate the accuracy of the data.
Stick the transparency firmly on the corner of the computer screen with scotch tape
so that it won’t fall or change position during the data extraction process. Set up the

positive x and y direction on the transparency. This step is shown in Figure A1 1.

transparency with grid

Computer Screen

 

Figure A1 1 A transparency with grid is taped on the computer screen. A coordinate
system is set up on the transparency.

193

(2) Launch the DataThief® software. Don’t maximize the DataThiet‘D window. Make its
window have a medium size so that it can be freely dragged and moved on the

computer screen, as can be seen in Figure A12.

Dala'l’hicl‘
Window

' (Don't

Maximize)

 

Figure A12 Launch the DataThiefID software. Don’t maximize the DataThiefD window.

(3) Open the last image first (#15) in DataThiefD. Make a mark on the transparency at
the point which was chosen as a reference point. This point has to be on one side of
the solder joint preferably at the Cu substrate/solder interface, as shown in Figure
A13.

(4) Set up the coordinate system in DataThieio, which exactly matches the coordinate
system previously set on the transparency, as shown in Figure A14. Enter the
minimum/maximum x and y values in DataThiefD.

(5) Choose the data extraction mode to “scattered data mode”. Carefully Click along the

black line from one side of the solder joint to another to extract data, as shown in

194

Figure A14. The distance of each clicked data point along x direction is designed
such that 15-17 data points are collected in total. Once this distance (between each
data point along x direction) has decided, use exactly the same distance consistently

for all other images.

reference
point

 

'l he coordinate in Ham | hiel‘ has In match the coordinate on 1r;iii~p.lrt‘lie).

Figure A14 Set up the coordinate system in DataThiefs.

195

(6) Save the collected data by selecting “Save” from the “File” menu, as shown in Figure
A15. The data can be saved as text file format (*.txt) for the time being. In each data

file, there will be two columns of x-y data pairs.

scatter
data
mode

 

Figure A15 Illustration of data extraction. (a) Scattered data is collected. (b) Data is saved
in “.txt” format.

196

(7) Open the next image (#14) in DataThiefs. Move the DataThiefD window so that the
same reference point on this image matches exactly the mark on the transparency in
step (3), as illustrated in Figure A16. Repeat steps (4)-(6) for the data extraction.

(8) Repeat steps (4)-(7) for the data extraction from other images.

 

Figure A16 Each image should match the reference point on the transparency.

3 Data Analysis —Creep Parameters

By now we have text files containing the extracted data from all the selected
images. The purpose of data analysis is to calculate various creep parameters from the
collected data points. These parameters usually include the global and localized creep
strain and strain rate, localized displacement vs. solder joint thickness, the strain and time
at the onset of tertiary creep, as well as the three dimensional plot of creep strain/strain

rate vs. time and solder joint thickness, etc.

197

Microsoft Exce10, KaleidaGraphO, and DeltaGraph® are used in this part of the
analysis. The following steps are needed to accomplish the data analysis.
3.1 Displacement vs. Unnormalized Position Though Joint Thickness
(1) Import the extracted data from the text files to a Microsoft Excel® spreadsheet in the
format as shown in Figure A17. The first column is the coordinate position of each
data point along x direction. The next 15 columns are the coordinate position of each
corresponding data point along y direction at different time intervals during the creep

deformation process.

111‘ 137‘ 171‘ 161' , 191‘ , 7‘01‘ ,
msmem Jared‘s “E411 tum v—me-m 110m:
and: men 1‘ .
.noem 3936471 362E111 Raw Data 4345411 . 49154)‘ 614501 ,
. ”Lem ‘ Jagger ‘ j v tmfiom‘H mg-i-anv‘t
‘ 157501 . 535E411 ‘ 41-35531 - . 116641) iiaeom men) 431 a)
.7 1 ,, . 1.17541) ”new I. as

V : Ling; ' afiemrriiaum n
ism . alseai ‘ aunt ‘

or 1795411 .i'ai’

 

Figure A17 Import the extracted data from DataThiefD to Microsoft Excel®.

(2) Normalize all the y position with respect to the initial y position in image 1. This step
can be shown in Figure A18 by looking at cell B27 in the spreadsheet. B27=B3-$B3.
Once 327 is obtained, all the y data in rows 27-45 can be copied from B27. Column 1

is kept unchanged at this step.

198

m seem; ‘ 1m 9 119 293 . a 7 am
#53401, ., A 1 ' . . 1 M .E 7. . 7
. - . , . , . m . 3222 ' , WE
._ sins , 4415 “'4 i, i ;, 2 m
z . ; : f g
’11-]! if”;
- -;

 

00! -ZDIEICD 4415411 “251500) 4&4!) 2

. . ' '. am a) 1
wild) a -- ,
Ling: 2,5254!) Juan) 7 i _ ___,

 

‘hfifii’amfir 'am‘ ciisni' an

 

 

Figure A18 Normalized the y position with respect to the initial y position before creep.

(3) Displacement vs. (unnormalized) position across solder joint thickness can be plotted
using data in row 27-row 45, column A-column P, as shown in Figure A19. A sample
displacement-position curve plotted in KaleidaGraph® is given in Figure A20.

3.2 Global Creep Strain & Steady-State Creep Rate

(4) Global creep strain can be calculated using the data at the end points where x=0 and

x=9 in this example. The global creep strain is defined as y = fl , assuming change

in angle 0 is very small. The manipulation of data in Excel spreadsheet is given in
Figure A21 in cell B49, where B49=(B45-BZ7)/($A45-$A27). B49 is the global
creep strain before deformation, while data in C49 to P49 are the global creep strains

at different time intervals during creep.

199

Displacement

,1111

' 01: 'msnm'o

Use these°§§téfi°fiiétm

1111477711

Displacement vs. (unnormalized) ’ ‘7:

position ”

11
11.

 

Figure A19 Data used to generate displacement vs. unnormalized position curve.

Unnormallzed Displacement ve Joint Thickneee

1 —1 i1 fir—firfi fi'fi- r fV ‘r—Ifir fir‘r ' 1+r~j - fi‘ Vfii

 

 

e_._z_._ W. *i‘mi. __1_r.1_.__;-gi__._;

 

 

-2 0 2 4 6 B 10
Position Through Joint thickneee

 

 

+0
-B— 600
w— 1800
--><--aeoo
~ 5400
~I— 101100
.7... 12600
--A 14400
7' 102-co

—‘0— 18000

‘5‘- — 19800

23400

 

Figure A20 A sample displacement-position curve plotted in KaleidaGraphO.

200

 

. , 1‘-1' . ,
_ , 1.2 .
1111124111 omega) 6M mu 412-1111202101“: 1112113 10115413 011154;;
m. , nasnt 11115471 15501-119541 4m
01.114501 JEEUI JQEOI JMEOL‘VE-ISIEJI ALIEN 6*0l 6W
01:4

 

 

 

 

 

Figure A21 Example showing how global strain data were obtained.

(5) Copy the data from B49 to P49, and transpositively paste only the value (not the
equation) to a column in the spreadsheet, for example in the column from C52. The
detailed operation can be expressed as the following: select the data from B49 to P49,
copy, click on C52, go to “Edit” menu, choose “paste special”, then select “Value"
and “Transpose”. Type in the time, from the lab notebook, for each corresponding
image in the column left to the global strain column. Note if the global strain values
are negative, they can be converted to positive numbers to make the plot looking
better without affecting anything. This step is illustrated in Figure A22.

(6) The “time” and “creep strain” columns thus can be used to generate the global creep
strain curve for this material in either Excel® or KaleidaGrath. A plot of global

strain curve is shown in Figure A23.

201

 

Figure A22 Illustration of how to prepare data for global strain-time plot.

Global Strain ve Time

 

06 1 v 1 1 l 1 1 1 1 v r r 1 r‘rrwiv 1 rr‘r‘r

0.5 f

 

 

0.3 ,

Strain

 

 

 

 

 

 

 

0.1_ E
1/ a
o

5000 1 10‘ 1.510‘ 210‘ 2.510‘

Time (see)
Figure A23 A sample plot of global creep strain vs. time using the data from Figure A22.

202

(7) The steady-state creep rate can be obtained using linear regression with the data in
the secondary stage, as shown in Figure A24.

Strain Rate Determined from Secondary Stage
(Linear regreeelon)

 

 

 

 

 

0.38 '- T I f I Y fi Y T j Y I I lfi I I T j T 3 T T f 1
C y = 0.23512 + 168968-0611 R: 0.99404 /: 1
0.37 :.. ................ .................... ..i ........... 1)./”j .................. _-.:
C . I ; : // : .1
L. _ 1 3 /.- ‘
0.30 . ; ; g /'L'/ ' i
t i 3 f :0 Z
L i 2 z // 3 1
0.35 C“ .......... .................... .............. / ....................................... _:
.5 C i 3 ,P’ , , .
g t__..-....._.....--..; .................... ............. ”/1..:.....................: .............. . ..... ............ ....._..
,7, 0-34 . g i // a s i 1
L 3 f 3/ : : - 4
033 E. ....... ................... Ire/I], ............... .................. ...E.-.........._....-..§ .................. _j
_ . // 1 - . 4
_ ,/......: .................... ....... _ ............ 5 ..... . ............ g
0.32 . j 9 ; ; f 3 j
1- f // I I ; -1
Z / E 3 i E 1
0.31 :n ........ //. ........ . ..... ................... It. ........... ........ . ..... ......... .. ..... _:
1— ’s/ 7 . z z . 4
.- 6 ' ; 1 l ' .4
i g l 1 mi 1 l
0.3
8000 110‘ 1.210‘ 1.410‘ 1.610‘ 1.810‘ 210‘
Time(sec)

Figure A24 Steady-state creep rate obtained using linear regression with the data from the
secondary stage in the global creep strain-time plot.

3.3 Normalized Displacement vs. Normalized Position Through Joint Thickness

The next a few steps are needed to generate normalized displacement vs.
normalized position curves. This curve is important for the calculation of localized creep
parameters. The final normalized displacement vs. normalized position curve should have

all the data points above 0 in y (y>0) and the x values of which are between 0 and 1
(xe [0,1].)

(8) Start a new spreadsheet. Copy the raw data and paste it to row l-row 20 as shown in

Figure A25.

203

arm “Tim Vii

011833 0157 01753.0!“ {JIM Gm] 002“)" 0217" GB
1m emu e357 0777570

4725044413275 me: e291" (5312
7715002561

 

Figure A25 Copy the raw data and paste it to row l-row 20 in a new spreadsheet for
normalized displacement-normalized position plot.

(9) Normalize x values by using A23=(A2-0)/(9-0). 0 and 9 are the unnormalized

extreme values in x. Thus A24 to A41 can be copied from A23. See Figure A26.

3 . . . ~ 3 _ 1 .
655m saem 5107.15 3617 ; 3272 ; 2547 278 25 1256! 213 ‘ 2734 ; 195 In
90164135575411‘5095R . :5 37m 2317:2057256 _7165773n‘7m‘15aa 12911

-- 911131111 alum 099327 0mm 0mm um 90):: 001344 01x13".
00.3179 00031560003: 001711 020375 0091 own err-ms emeu em 7
2157 05413.72 051m email: em ems” ems; emsn

10m CINE DIM OHBD

e1m 1311957 mm 1313511013776

012322 cum ens“ 015622 {NEED

e101: e15211e1a22 017m e107

enmp 017533 mm mm 02mm e021o11

e133"! OISE OZZI 022527 412$“

tmuu 4]! em
““44 43mm 00571 0'“ 2157
09105555 055‘}? ”49021257 0
am am new manomfliaze’a‘x vellum mum a
1757

OEIHII 110103333 01:641101 em? £77773 emu 0204 029111

07777731311515 em enm emu emu
cam 09115566 am my ewes 021578
00153770 ORJQJW ems emne 0:06
09.23“ emm erases e1 1222 em emu emu 011757 41325331)
0155566 ewes emu 017L023: emu 031767 enasl 036w
om ntmw em ecm ecma ’wnfiiifiifiixn e'mm unfii'umm
001566 emu 00755 00157 0029 0mm em 0015 1.104255
0mm ems _e05956 095076 0053713010
e06 'em 43094114110079 ems 011378
e awn 0955a 0075“ 00911 009514 110519 11379 013133 0134
"009533 0975110913522 0 “35.0“” 41111911 014175 07152543161537
emns e
021cm 0717213
025179 017170
e76

022E641
emm emu

CINE 0

 

Figure A26 Illustration of normalizing x values.

204

(10) Find the first data in the last y column (cell P2=S.706 in this example) and
normalize it to 0 w.r.t. x in cell P23 by using P23=(P2-5.706)/(9-O). C0py this cell
and paste it to every cell between row 23 and row 41, as shown in Figure A27.

(11) Normalize all the y position with respect to the initial y position in the first y
column. This step can be seen in Figure A28 by looking at cell B43 in the
spreadsheet. B43=BZ3-$BZ3. Once B43 is obtained, all the y data in rows 43-61 can
be copied from B43.

(12) Now find the last data in the last y column (cell P6l=-0.5421 in this example) and
normalize it to O w.r.t. x in cell P83 by using P83=(P61+0.5421)/(1-0). Copy this cell
and paste it to every cell between row 65 and row 83, as shown in Figure A29. Now
all the data between row 65 and row 83 satisfy the condition that (y >- 0 fl xe [0,1]).

(13) The normalized displacement vs. normalized position through joint thickness curve

can be plotted with the data obtained from (12). An example is shown in Figure A30.

3.4 Localized Creep Parameters
(14) Open the “normalized displacement vs. normalized position through joint
thickness” curve in KaleidaGraph® and curve fit all the displacement lines with 3rd

order polynomial. The relationship between y and x thus can be expressed as
y = M0 +M1x+M2x2 +M3x3, as shown in Figure A31.

(15) Record the coefficients M1, M2, and M3 for all the displacement curves (M0 is not

needed) and fill in a new spreadsheet as formatted in Figure A32.

205

5|!
' 555m ,# . .
5572.00.5107‘413 , :2 .2 na- 1

. 557541). m ; ms 1 , 2 1 2.314 1 am we

 

111m 0
4W" 0%"
ow emu
41151: 91m 01
en: mom m31meimh~e1me1u
447 311 4115“ Tani—me 41
mm em

 

: d- 1 e
1WI10 017511
«11112111 em '1 on %% emigm‘
u 'e ’

 

Figure A27 Make the first data in the last y column zeroed.

e - i l
‘15— an 1
' $231-$923

0 011mm MIN! 0"“ nunu ODIII‘ ODJIM DID“ mm mm ounu MIEIIHMIEII 011)..
emu 00312 00m 0003‘ 0017" «000295 00‘6"
. em 11

219I
(1le OIS?" 01m enm 0187 'flIm 0071111074576 029m 0
{HIE} 01537 JIM mm mm 0W 02'0" (JIIIII 02m 0177" on“ 0
DIE! DIE}! 0194a 0221 071572 023)“ 1323

emu 03175! em on 05— em 0392

0mm 0mm (10137401112 1303157 DWI“ 0mm 0mm 0mm 0mm

Wu OWIIO

 

Figure A28 Normalize all the y position with respect to the initial y position in the first y
column.

206

m6 . or .znguzism . 1
21711121111 am, mama: on. imam
a 111.1

 

 

Figure A29 Illustration of how to make all the data greater than 0.

Campanile SID-A. Solder Joint .125 'c. 25 In

 

+Bolon Cmp

 

‘9— 600 sec
- 1800 sec
E --X--3000aoc
E 9 5400 see
8 r~ 7260 391:
fi ' 7'
5 I— 10300 sec
2 --o~ 12600 sec
«g A 14400 sec
E V 1152130 see
c 1 any 1
2
-C ~ 19800 sec
9 21600 sec
“Y 2.3400 sec

 

 

 

 

2 0.4 0.6 0
Nomalized Penman Through Solder Joint Thicknou

Figure A30 A sample plot of normalized displacement vs. normalized position through
joint thickness curve.

207

emmwmuu'c,um

 

 

 

 

 

 

Figure A31 Curve fit all the displacement lines with 3"1 order polynomial.

ISISS 05]” can 071342 417101 :BIAIS‘ 4)
5&1 051157 «OBI menus “379“
um 07““: com:
01'“

m. m__m
g5; 411753319520

”001113. a!” 01-310

DIS? om «011112 (Hm
DNI‘N onion 0%”

 

 

Figure A32 Record the coefficients M1, M2, and M3 form Figure A31 for all the
displacement curves (M0 is not needed) and fill in a new spreadsheet.

208

(l6) Localized creep strain can be calculated using the equation

7 _—_ it): = M l + 2M 2 x + 3M 3 x2 for whatever x values desired. Usually, x values

are taken from 0 to 1 with 0.1 as the interval representing different positions through
the joint thickness. So localized strain at these positions can be calculated using the
above equation. See cell B7 in Figure A33 for the calculation example, where

7=B$3+2*(B$4*$A7)+3*(B$5*$A7"2). All the cells between row 7 and row 17

can be copied from B7.

0 In1_ V urn
?Di76: 04103 MID"

ommmm 0mm: 011). 01055 011i.

to:

omit 91
one: 'ounm
ufiMIo in

m‘a We I6
7 _ any goon mm! m mama vmo
nmzmimw name name amgvs omgT ma 0:91-41
721771062: ODIID 00105 am
7 km imam o Mimi

 

 

 

Figure A33 Illustration of how to obtain localized creep strain.

(17) The localized creep strain data can be copied and transpositively pasted in the

format as shown in Figure A34 between row 49 and row 62 with positive values. An

example of variation localized strain with time at different location through the solder

209

joint thickness is shown in Figure A35. Localized creep strain can also be plotted

with global creep strain as shown in Figure A36.

 

--. u 01 a: 0.: us' as’ or no
7076415 russas saseas auras sues-cs mm. 4mm; 3mm;
312513715045509 “ask wees

NEE-IA emu 1%” 107643 Ifll}

m 32145 lfl’EG min 175E: 13E“ _
m 23151313455165me 9ME5HJEG;

meas- snw 7950f? sneer. 53543, 53605
IOIEJB Bm‘ IIIEG SIEG 5615: SHE-(5 sass
mans nugas animus: ‘75 -
__I was Hm 104:4! 341:7ng
mm; . 1'”

124m sum serif“ "

I55“
1% 5mg? BIIE-(S SEIEELSO‘E g 5:55

 

 

Figure A34 Manipulation of data prepared for plotting localized creep strain with time.

Localized Strlln VI Tlmo

 

 

 

 

 

 

 

 

 

 

 

1
0.5 +0
'13—01
0.6 -‘>—02
--x--o.3
f‘ 04 ‘ 0“
r7) ,
0,2
Iko7
ones
0 ----- 09
I 1 '7,
i l i . i 1 i l i l i
43.2
o 5000 110‘ 1.510‘ 210‘ 2510‘

Time (.00)

Figure A35 Variation of localized creep strain with time.

210

Compoono Sn-Ag Solder JoInI at 25 ‘c. 25 MP-

- r—IFT’I’ 71—177

I

'7T_—” 7‘1? {7‘1 1'7!" T’j‘fl)V—l'a

 

Locullzod 3mm I Glob-I 317-111 V. Tlme

    

”'1

 

Shear Strlh

 

 

 

 

o 5000 110‘ 1.510‘ 210‘ 2510‘

Two (uc)
Figure A36 Plot of localized and global strain with time.

(18) In order to calculate localized creep strain rate, the localized creep strain curves
have to be differentiated. Open the “localized creep strain vs. time” curve in
KaleidaGraphO and curve fit all the localized strain lines with 3'‘1 order polynomial.
The relationship between y and I thus can be expressed as
y = M0 + M1: + Mztz + M323, as shown in Figure A37.

(19) Record the coefficients M1, M2, and M3 for all the localized strain curves (M0 is not
needed) and fill in the spreadsheet as formatted from row 64 to row 68 in Figure A38.

(20) Fill in A71 to A84 again with time as shown in Figure A39. Localized creep strain

rate can be calculated using the equation 7): % = M1 +2M21+3M372 for whatever r

values desired. An example of such calculation can be seen in cell B71, where

211

B71=B$66+2*(B$67*$A7l)+3*(B$68*$A71A2). All the cells between row 71 and

row 84 can be copied from B71.

curve fit

 

Figure A37 Curve fit the “localized creep strain vs. time” curves in KaleidaGrath with
3rd order polynomial.

 

 

G lfiE-IB 2 > ‘ alt—73 IIIEIB EIEG 5615: GAE: 557E:

coefficients M1,,M2 and M3 values

21“; 2E5 3015-12 3M4! 3K6 3&4! JSZEM 3x45 ZHEIB 21765 19E“ 6*}5
9165

w 304E“ BBIEG Iii-£5 1&4! AM. AVE! 41266 376548 2E6 I

tfi—iin: um

 

 

Figure A38 Record the coefficients M1, M2, and M3 for all the localized strain curves (Mo
is not needed) and fill in the spreadsheet.

212

 

 

ice-u
auto; 211w;-1m
J; 1%? Inf-Iii 1;": ngu sggufiaujz

, . l . - 1
—.___._,
I! 1012:;11Ea; 115“ 2m 1755
;%9s£51%g0s17£g 2:451 _gg
7‘: ' 1.21505 .1!“
1 1 a s

 

 

Figure A39 Illustration of how to obtain localized creep strain rates.

(21) Localized creep strain rate can be plotted with time using the data from row 71 to
row 84 in Figure A40, as shown in Figure A41, or alternatively, with normalized
position through joint thickness using the data from row 87 to row 97 in Figure A40,
as shown in Figure A42.

3.5 Three Dimensional Plots

(22) The variation of localized creep strain rate can be plotted with time and normalized
position through solder joint thickness using DeltaGraph° in a three dimensional
format. Before inputting data in DeltaGrapho, the time, position, and the localized
creep strain/strain rate data should be reorganized in a format as shown in Figure A43.
The three columns of data can be directly pasted to DeltaGraphO, as shown in Figure

A44.

213

 

 

Figure A40 Manipulation of data prepared for localized strain rate vs. time/position plot.

Strain Hate vs Tlmo

 

 

 

 

 

 

 

 

 

710‘ ’ "F‘fifiIflfi‘fi'fif '*I—*fi 'fi'fi—fi‘r’fv
+0
.5
610 +01
4—02
510“
-~X--o.3
E ‘5 . '04
5 no
I T Q
=
(I 4 ‘
s are
E I— or
U)
..
210" 08
A ~09
110" 7’ 7
o
o 5000 110‘ 1.510‘ 210‘ 2510‘
Tlmo(ue)

Figure A41 A sample plot of localized creep strain rate vs. time.

214

Sir-In flute VI Polltlon

 

 

+600
-l3—-1600
-0— 3600
--X--s4oo
--+--7200

' 3" 31:30

Strain Rate (Hue)

O
.— 12600
”’ " 14400

t 15200

 

4— Isaac

 

 

 

 

 

 

 

 

 

 

Figure A42 A sample plot of localized creep strain rate vs. position through solder joint
thickness.

 

Figure A43 Reorganization of the data in Excelo prepared for use in DeltaGraphO.

215

if?

 

 

"3:
E
g
g
C
5
g
E
E
E.
E
E
5‘
.5,
E
is.

l

 

Figure A44 Illustration of data directly copied from Microsoft Excel".

(23) A three dimensional plot of localized creep strain rate with time and normalized
position through solder joint thickness thus can be generated with DeltaGraph'D by
selecting the 3-D surface fill chart in the “chart gallery” menu, as can be seen in
Figure A45. An example of the 3-D plot is shown in Figure A46.

(24) Similarly, the variation of localized creep strain can also be plotted with time and
normalized position through solder joint thickness using DeltaGrapho in a three
dimensional format. The data organization format is also shown in Figure A43 at
columns G, H, I from row 101. Follow steps (22) and (23) to obtain this 3-D plot. A

sample plot is shown in Figure A47.

216

 

Figure A45 Illustration of how to generate three-dimensional graph in DeltaGrath.

 

mmmmmmm
mmmmwmm
000.0000
0000000

 

 

 

 

 

~

Wmmmm.

some E: 525

Figure A46 Variation of localized strain rate with normalized position and time.

217

 

 

5.3m

 

Figure A47 Variation of localized creep strain with normalized position and time.

218

APPENDIX B

PROCEDURES FOR OPERATING HITACHI-2500C SCANNING ELECTRON
MICROSCOPE

1. HITACHI-2500C SEM Start Up

(1) Check the three green lights. Those should be on.

(2) Check x, y, z knobs, those should be at 17, 20 and EX—EX position, respectively.

(3) Open outer chamber (EVAC-AIR), put the sample on the sample stage using carbon
tape. Then EVAC

(4) Wait until the green light is on for the outer chamber, then open the door of specimen
exchange chamber, slide the specimen to the sample stage, unscrew it, pull the long
rod back, close the door.

(5) Wait until all the green lights are on.

(6) Turn on the electron beam by rotating the black knob clockwise 90 degrees, slowly
pull it out, then lock it by rotating back 90 degrees(counterclockwise).

(7) Turn on the display switch. Never turn off the vacuum switch next to it!

(8) Wait until the green “ready” light stops blinking in the main panel.

(9) On screen, check the accelerating voltage and working distance. We usually use
25KV and 15mm. When everything is ok in the menu, hit “enter” to get out of the
menu screen.

(10) Turn on the accelerating voltage by pressing ON in ACC VOLTAGE.

(11) Turn the filament current up too lOOuA with the speed of V2 division every 30

seconds.

(12) Press rapid scanning mode.

(13) Turn up contrast and brightness. Use focus knob to acquire a focused image.

219

2. HITACHI-2500C SEM Shut Down

(1) Turn data display off.

(2) Turn down the brightness and contrast completely.

(3) Magnification set to: LOOK.

(4) Turn down the filament current at a speed of ldivision every 30 seconds.

(5) Turn off the accelerating voltage by pressing OFF in ACC VOLTAGE.

(6) Display switch off. Never turn off the vacuum switch next to it!

(7) Turn off the electron beam by rotating the black handle 90 degrees clockwise, push

in, then 90 degrees counterclockwise.

(8) Set the x, y, z knobs to 17, 20 and EX-EX positions.

(9) Wait until all the green lights are on.

(10) Open the specimen exchange chamber, slide the long rod in, take the specimen out
of the stage, pull all the way out, hold the rod with your right hand, close the door
with your left hand.

(11) Hold the rod, wait until all the green lights are on

(12) Open the outer chamber by pressing EVAC-AIR, unscrew the whole stage from the
rod, take the sample off, then screw back the stage to the rod,

(13) Close the outer chamber, then AIR-EVAC, pressing it for a little while.

(14) Sign in the SEM log book.

(15) Wait until you see all the three green lights are on before you leave.

220

APPENDIX C

LIST OF PERSONAL PUBLICATIONS & PRESENTATIONS DIRECTLY
RELATED TO THE INVESTIGATION IN THIS THESIS

PUBLICATIONS:

l.

J .P. Lucas, F. Guo, J. McDougall, T.R. Bieler, and K.N. Subramanian, “ Creep
Deformation Behavior in Eutectic Sn-Ag Solder Joints Using Novel Mapping
Techniques”, J. Electronic Materials, 28(11), 1268 (1999)

F. Guo, S. Choi, J .P. Lucas, and K.N. Subramanian, “Effects of Solder Reflow on
Wettability, Microstructure and Mechanical Properties”, J. Electronic Materials,
29(10), 1241 (2000)

F. Guo, S. Choi, J .P. Lucas, and K.N. Subramanian, “ Microstructural
Characterization of Reflowed and Isothermally-Aged Cu and Ag Particulate
Reinforced Sn-3.5Ag Composite Solders”, Soldering and Surface Mount Technology,
13(1), 7 (2001)

F. Guo, J.P. Lucas, and K.N. Subramanian, “ Creep Behavior in Cu and Ag Particle-
Reinforced Composite and Eutectic Sn-3.5Ag and Sn-4.0Ag-0.5Cu Non-Composite
Solder Joints”, J. Materials Science: Materials in Electronics, 12, 27(2001)

F. Guo, J. Lee, S. Choi, J .P. Lucas, T.R. Bieler and K.N. Subramanian, “Processing
and Aging Characteristics of Eutectic Sn-3.5Ag Solder Reinforced with Mechanically
Incorporated Ni Particles”, J. Electronic Materials, 30(9), 1073 (2001).

F. Guo, J. Lee, J. P. Lucas, T. R. Bieler, and K. N. Subramanian, “Creep Properties
of Eutectic Sn-3.5Ag Solder Joints Reinforced with Mechanically Incorporated Ni
Particles”, J. Electronic Materials, 30(9), 1222 (2001).

S. Choi, J. Lee, F. Guo, T.R. Bieler, K.N. Subramanian, and JP. Lucas, “Creep
Properties of Sn-Ag Solder Joints Containing Intermetallic Particles”, Journal of the
Minerals, Metals & Materials Society, 53(6), 22(2001)

F. Guo, S. Choi, K.N. Subramanian, T.R. Bieler, J .P. Lucas, A. Achari, and M.
Paruchuri, “Evaluation of Creep Behavior of Near Eutectic Sn-Ag Solders Containing
Small Amount of Alloy Additions”, in review at Visteon Electronic Technical Center,
Dearbom, MI 48121.

K.N. Subramanian, S. Choi, and F.Guo, “Lead-free Solders with Dispersoids for
High Temperature Applications”, submitted for “Interconnections: Solder-based”
chapter in Handbook of Lead(Pb)-Free Technology for Microelectronic Assemblies.
Accepted for publication.

221

10. J .G.Lee, F. Guo, K.N. Subramanian, and J .P. Lucas, “Intermetallic Morphology
around Ni Particles in Sn-3.5Ag Solder”, submitted to Soldering and Surface Mount
Technology, in review.

PRESENTATIONS:

1. “Creep Deformation Behavior in Eutectic Sn-Ag Solder Joints Using Novel Mapping
Techniques”, J .P. Lucas, F. Guo, J. McDougall, T.R. Bieler, and K.N. Subramanian,
Interconnections for Electronics Packaging, TMS Annual Meeting, Feb. 28-Mar. 4,
1999, San Diego, CA, presented by J .P. Lucas.

2. “Microstructural Characterization of Reflowed and Isothermally-Aged Cu and Ag
Particulate Reinforced Sn-3.5Ag Composite Solders”, F. Guo, S. Choi, J .P. Lucas,
and K.N. Subramanian, Pb Free and Pb Bearing Solders, TMS Fall Meeting, Oct. 31-
Nov. 4, 1999, Cincinnati, OH, presented by F. Guo.

3. “Creep Behavior in Cu and Ag Particle-Reinforced Composite and Eutectic Sn-3.5Ag
and Sn-4.0Ag-0.5Cu Non-Composite Solder Joints”, F. Guo, J .P. Lucas, and K.N.
Subramanian, Pb Free and Pb Bearing Solders, TMS Fall Meeting, Oct. 3l-Nov. 4,
1999, Cincinnati, OH, presented by F. Guo.

4. “Effects of Solder Reflow on Wettability, Microstructure and Mechanical Properties”,
F. Guo, S. Choi, J .P. Lucas, and K.N. Subramanian, Packaging and Soldering
Technologies for Electronic Interconnects, TMS Annual Meeting, Mar. 12-16, 2000,
Nashville, TN, presented by J .P. Lucas.

5. “Creep Deformation Behavior and Thermomechanical Fatigue Properties of 95.58n-
4Ag-0.5Cu solder joint”, Lead-free Solder Research Group, Interim Report to Visteon
Electronic Technical Center, Sep. 27, 2000, presented by F. Guo and S. Choi.

6. “Processing and Aging Characteristics of Eutectic Sn-3.5Ag Solder Reinforced with
Mechanically Incorporated Ni Particles”, F. Guo, J. Lee, J .P. Lucas, T.R. Bieler and
K.N. Subramanian, Lead-Free Solder Materials and Soldering Technologies, TMS
Annual Meeting, Feb.11-15, 2001, New Orleans, LA, presented by F. Guo.

7. “Creep Properties of Eutectic Sn-3.5Ag Solder Reinforced with Mechanically
Incorporated Ni Particles”, F. Guo, J. Lee, J .P. Lucas, T.R. Bieler and K.N.
Subramanian, Lead-Free Solder Materials and Soldering Technologies, TMS Annual
Meeting, Feb.l 1-15, 2001, New Orleans, LA, presented by J .P. Lucas.

222

REFERENCES

223

REFERENCES

. Vianco, P.T., “Development of Alternatives to Lead-bearing Solders”, Proceedings of
Surface Mount International Conference, San Jose, CA, Surface Mount International,
Edina, MN 55424, USA, 2, 1993, p. 725.

. Hwang, J .S., and Guo, Z.F., “Lead-free Solders for Electronic Packaging and
Assembly”, Proceedings of Surface Mount International Conference, San Jose, CA,
Surface Mount International, Edina, MN 55424, USA, 2, 1993, p. 732.

. McCormack, M. and Jin, 8., “Progress in the Design of New Lead-Free Solder
Alloys ”, Journal of The Minerals, Metals & Materials Society, 45, 1993, p. 36.

. Gibson, A.W., Choi, S., Bieler, TR. and Subramanian, K.N., “Environmental
Concerns and Materials Issues in Manufactured Solder Joints ”, IEEE 5'“
International Symposium on Electronics and the Environment, IEEE, Piscataway, NJ,

1997,p.246.

. Winterbottom, W.L., “Converting to Lead-Free Solders: An Automotive Industry
Perspective”, Journal of The Minerals, Metals & Materials Society, 45, 1993, p. 20.

. Melton, C., “The Effect of Reflow Process Variables on the Wettability of Lead-Free
Solders”, Journal of The Minerals, Metals & Materials Society, 45, 1993, p. 33.

. Choi, S., Subramanian, K.N., Lucas, J .P, and Bieler, T.R., “Thermomechanical

Fatigue Behavior of Sn-Ag Solder Joints”, Journal of Electronic Materials, 29, No. 10,
2000, p. 1249.

. Frear, D.R., “Thermomechanical Fatigue in Solder Materials”, in Solder Mechanicas:
A State of the Art Assessment, edited by DR. Frear et al., TMS, Warrendale, PA,
1991,p.19l.

. Darveaux, R., Murty, K. L., and Turlik, 1., “Predictive Thermal and Mechanical
Modeling of A Developmental MCM”, Journal of The Minerals, Metals & Materials
Society, 44, 1992, p. 36.

10. Murty, K. L., and Turlik, I., “Deformation Mechanisms in Pb/Sn Alloys: Application

to Solder Reliability in Electronic Packaging”, Proceedings of Joint ASME/JSME
Advances in Electronic Packaging, edited by W.T. Chen and H.Abe, ASME, New
York, NY, EP 1, 1992, p. 309.

11. Morris, J .W. Jr., Goldstein, J .L.F., and Mei, Z., “Microstructure and Mechanical

Properties of Sn-In and Sn-Bi Solders”, Journal of The Minerals, Metals & Materials
Society, 45, 1993, p. 25.

224

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Morris, J .W. Jr., “Research Trends in Electronic Solders and Soldering ”,
Proceedings of the Second Pacific Rim International Conference on Advanced
Materials and Processing, edited by K.S. Shin et al., The Korean Institute of Metals
and Materials, Seoul, Korea, 1995, p. 715.

Gibson, A. W., Choi, S., Subramanian, K.N., and Bieler, T.R., “Issues Regarding
Microstructural Coarsening due to Aging of Eutectic Tin-Silver Solder”, Reliability of
Solders and Solder joints, edited by R.K. Mahidhara et al., TMS, Warrendale, PA,
1997,p.97.

Yang, W., Messler, R.W., and Felton, L.E., “ Microstructure Evolution of Eutectic
Sn-Ag Solder Joints”, Journal of Electronic Materials, 23, No. 8, 1994, p. 765.

Miller, C.M., Anderson, LE, and Smith, J .F., “A Viable Tin-Lead Solder Substitute:
Sn-Ag-Cu”, Journal of Electronic Materials, 23, No. 7, 1994, p. 595.

McCormack, M., and Jin, S., “Improved Mechanical Properties in New, Pb-free
Solder Alloys”, Journal of Electronic Materials, 23, No. 8, 1994, p. 715.

McCormack, M., Chen, H.S., Kammlott, G.W., and Jin, 8., “Significantly Improved
Mechanical Pr0perties of Bi-Sn Solder Alloys by Ag-Doping”, Journal of Electronic
Materials, 26, No. 8, 1997, p. 954.

Jacobson, D.M., and Humpston G., “Depressing the Melting Points of Solders and
Brazes by Eutectic Alloying”, GEC Journal of Research, 12, No. 2, 1995, p. 112.

McCormack, M., and Jin, 8., “The Design and Properties of New, Pb—free Solder
Alloys”, Proceedings of the 16'” IEEE/CPMT International Electronics
Manufacturing Symposium, La Jolla, CA, September, 1994, IEEE, Piscataway, NJ,
l994,p.7.

Kariya, Y., and Otsuka, M., “Effect of Bismuth on the Isothermal Fatigue Properties
of Sn-3.5mass%Ag Solder Alloy”, Journal of Electronic Materials, 27, No. 7, 1998, p.
866.

Tomlinson, W.J., and Bryan, N.J., “The Strength of Brass/Sn-Pb-Sb Solder Joints
Containing O to 10% Sb”, Journal of Materials Science, 21, 1986, p. 103.

Marshall, J .L., and Calderon, J ., “Hard-particle Reinforced Composite Solders, Part 1:
Microcharacterisation”, Soldering and Surface Mount Technology, 9, No.2, 1997, p.
22.

Marshall, J .L., Sees, J., and Calderon, J ., “lVIicrocharacterization of Composite

Solders”, Proceedings of Technical Program —Nepcon West Conference, Anahein,
CA, 1992, Cahners Exhibition Group, Des Plains, IL, 3, 1992, p. 1278.

225

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

Marshall, J .L., Calderon, J ., Sees, J ., Lucey, G. and Hwang, J .S., “Composite
Solders”, IEEE Transactions on Components, Hybrids, and Manufacturing
Technology, 14, No.4, 1991, p. 698.

Marshall, J .L., and Calderon, J ., “Hard-particle Reinforced Composite Solders, Part 2:
Solderability”, Soldering and Surface Mount Technology, 9, No. 3, 1997, p. 6.

Marshall, J.L., and Calderon, J ., “Hard-particle Reinforced Composite Solders, Part 3:
Mechanical Properties”, Soldering and Surface Mount Technology, 9, No. 3, 1997, p.
l l.

Sastry, S.M.L., Peng, T.C., Lederich, R.J., Jerina, KL, and Kuo, C.G.,
“Microstructures and Mechanical Properties of In-situ Composite Solders”,

Proceedings of Technical Program —-Nepcon West Conference, Anahein, CA, 1992,
Cahners Exhibition Group, Des Plains, IL, 3, 1992, pp. 1266.

Betrabet, H.S., McGee, S.M., and McKinlay J .K., “Processing Dispersion-
Strengthened Sn-Pb Solders to Achieve Microstructural Refinement and Stability”,
Scripta Metallurgica ct Materialia, 25, 1991, p. 2323.

Betrabet, HS, and McGee, 3., “Towards Increased Fatigue Resistance in Sn-Pb
Solders by Dispersion Strengthening”, Proceedings of Technical Program —Nepcon
West Conference, Anahein, CA, 1992, Cahners Exhibition Group, Des Plains, IL, 3,
1992,p.1276.

Mavoori, H., and Jin. S., “New Creep-Resistant, Low Melting Point Solders with
Ultrafine Oxide Dispersions”, Journal of Electronic Materials, 27, No. 11, 1998, p.
1216.

Clough, R.B., Pete], R., Hwang, J .S. and Lucey,G., “Preparation and Properties of
Reflowed Paste and Bulk Composite Solder”, Proceedings of Technical Program —
Nepcon West Conference, Anahein, CA, 1992, Cahners Exhibition Group, Des Plains,
IL, 3, 1992, p. 1256.

McCormack, M., J in, S., and Kammlott, G.W., “Enhanced Solder Alloy Performance
by Magnetic Dispersions”, IEEE Transactions on Components, Packaging, and

Manufacturing Technology — Part A, 17, No. 3, 1994, p. 452.

NEMI, NIST, NSF, and TMS, Report on the “Workshop on Modeling and Data
Needs for Lead-Free Solders”, TMS 2001 Annual Meeting, New Orleans Convention
Center, New Orleans, LA, Feb. 15, 2001.

Gayle, F. W. etc., “The NCMS High Temperature Fatigue-Resistant Solder Project”,

Presentation at the TMS 2001 Annual Meeting, New Orleans Convention Center,
New Orleans, LA, Feb. 12, 2001.

226

35. Vianco, P.T., “Solder Materials”, in Book: AWS Soldering Handbook, 3rd Edition,
American Welding Society, Miami, FL 33126, 1999, Chapter 2. p.179.

36. Liu, C.Y., Chen, C., Liao, ON, and Tu, K.N., “l\'1icrostructure-electromigration
Correlation in a Thin Strip of Eutectic SnPb Solder Stressed Between Cu Electrodes”,
Applied Physics Letters, 75, No.1, 1999, p. 58.

37. Liu, C.Y., Chen, C., and Tu, K.N., “Electrornigration in Sn-Pb Solder Strips as
Function of Alloy Composition”, Journal of Applied Physics, 88, No.10, 2000,
p.5703.

38. Richards, B.P., Levoguer, C.L., Hunt, C.P., Nimmo, K., Peters, 8., and Cusack, P.,
“Lead-Free Soldering, An Analysis of the Current Status of Lead-Free Soldering”,
DTI, 1999, p. 9.

39. GEC-Marconi, ITRI, BNR Europe, and Multicore Solders, DTI sponsored project
“Alternative Solder for Electronic Assemblies”, 1991-1993, Final Report MS/20073,
26, No.10, 1993.

40. NCMS, Report on four-year collaborative project “Lead-free Solder Alternatives”,
1998.

41. Pinizzotto, R.F., Wu, Y., Jacobs, E.G., and Foster, L.A., “Microstructural
Development in Composite Solders Caused by Long Term, High Temperature

Annealing”, Proceedings of Technical Program —Nepcon West Conference,
Anahein, CA, 1992, Cahners Exhibition Group, Des Plains, IL, 3, 1992, p. 1284.

42. Wu, Y., Sees, J .A., Pouraghabragher, C., Foster, L.A., Marshall, J .L., Jacobs, E.G.
and Pinizzotto,R.F., “The Formation and Growth of Intermetallics in Composite
Solder”, Journal of Electronic Materials, 22, No. 7, 1993, p. 769.

43. Clough, R.B., Shapiro, A.J., Bayba, A.J., and Lucey,G.K. Jr., “Boundary Layer
Fracture of Copper Composite Solder Interfaces”, Advances in Electronic Packaging,
ASME, New York, NY, EEP 4-2, 1993, p. 1031.

44. Kuo, C.G., Sastry, S.M.L. and Jerina, K.L., “Tensile and Creep Properties of In-situ
Composite Solders”, Microstructures and Mechanical Properties of Aging Material,
edited by P.K. Liaw, R. Viswanathan, K.L. Murty, E.P. Simonen, and D. Frear, TMS,
Warrendale, PA, 1993, p. 409.

45. Kuo, C.G., Sastry, S.M.L. and Jerina, KL. (1993), “Fatigue Deformation of In-situ
Composite Solders”, Microstructures and Mechanical Properties of Aging Material,
edited by P.K. Liaw, R. Viswanathan, K.L. Murty, E.P. Simonen, and D. Frear, TMS,
Warrendale, PA, p. 417.

227

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

Subramanian, K.N., Bieler, T.R., and Lucas, J .P., “Mechanical Shaping of Metal
Matrix Composites”, in Key Engineering Materials, edited by GM. Newaz et al.,
Trans. Tech. Ltd., Switzerland, 104-107, 1995, part 1: p. 175.

Subramanian, K.N., Bieler, T.R., Lucas, J .P., “ Microstructural Engineering of
Solders”, Journal of Electronic Materials, 28, 1999, p. 1176.

Choi, S., Gibson, A.W., McDougall, J .L., Bieler, T.R., and Subramanian, K.N.,
“Mechanical Properties of Sn-Ag Composite Solder Joints Containing Copper-Based
Intermetallics”, Reliability of Solders and Solder Joints, edited by R.K. Mahidhara et
al., TMS, Warrendale, PA, 1997, p. 241.

Gibson, A.W., Subramanian, K.N., and Bieler, T.R., “Comparison of Mechanical
Fatigue Fracture Behavior of Eutectic Sn-Ag Solder with and without Cu6Sn5
Intermetallic Particulate Reinforcement”, Journal of Advanced Materials, 30, 1998, p.
19.

Lucas, J .P., Gibson, A.W., Subramanian, K.N., and Bieler, T.R., “Nanoindentation
Characterization of Microphases in Sn-3.5Ag Eutectic Solder Joints”, in
Fundamentals of Nanoindentation and Nanotribology, edited by N.R. Moody et al.,
Proceedings of Materials Research Society Conference, Materials Research Society,
Pittsburgh, PA, 522, 1998, p. 339.

Dieter, G.E., Mechanical Metallurgy, Third Edition, McGraw-Hill Inc., New York,
NY, 1986, p. 432.

Evans, R.W., and Wilshire, B., Introduction to Creep, The Institute of Materials, 1
Carlton House Terrace, London, 1993.

Garofalo, F., Fundamentals of Creep and Creep Rupture in Metals, MacMillan Series
in Materials Science, MacMillan, NY, 1965.

Bird, J .E., Mukherjee,A.J., and Dorn, J .E., Quantitative Relation Between Properties
and Microstructures, Israel University Press, Jerusalem, 1969, p. 225.

Courtney, T.H., Mechanical Behavior of Materials, McGraw-Hill Publishing
Company, New York, NY, 1990, p. 268.

McDougall, J ., “An Analysis of Global and Localized Creep Strains and Creep
Properties of Non-Composite and Composite Lead-Free Solders at Room and
Elevated Temperatures”, MS. Thesis, Michigan State University, September, 1998.

Mathew, M.D., Movva, S., and Murty, K.L., Key Engineering Materials,

“Deformation Mechanism in Tin and Tin Based Electronic Solder Alloys”, 171-174,
2000. p. 655.

228

58. Darveaux, R., and Banerji, K., “Constitutive Relations for Tin-Based-Solder Joints”,
IEEE, 1992, p. 538.

59. Mavoori, H., Chin, J ., Vaynman, S., Moran, B., Keer, L., and Fine, M., “Creep, Stress
Relaxation, and Plastic Deformation in Sn-Ag and Sn-Zn Eutectic Solders,” Journal
of Electronic Materials, 26, 1997, p. 783.

60. Mavoori, H., Vaynman, S., Chin, J ., Moran, B., Keer, L., and Fine, M., “Mechanical
Behavior of Eutectic Sn-Ag and Sn-Zn Solders”, Proceedings of Materials Research
Society Symposium, Materials Research Society, Pittsburgh, PA, 390, 1995, p. 161.

61. Yang, H., Deane, P., Magill, P., and Murty, K.L., “Creep Deformation of 96.58n-
3.5Ag Solder Joints In A Flip Chip Package”, IEEE/Electronic Components and
Technology Conference, 1996, p. 1136.

62. Raeder, C.H., Schmeelk, G.D., Mitlin, D., Barbieri, T., Yang, W., Felton, L.F.,
Messler, R.W., Knorr, DB, and Lee, D., “Isothermal Creep of Eutectic SnBi and
SnAg Solder and Solder Joints”, IEEE/CPMT Int’l Electronics Manufacturing
Technology Symposium, 1994

63. Guo, Z., Conrad, H., and Pao, Y., “Plastic Deformation Kinetics of 95.58n4Cu0.5Ag
Solder Joints”, International Mechanical Engineering Congress & Exhibition of the
Winter Annual Meeting of ASME, Chicago, IL, 94-WA/EEP-12, 1994, p. l.

64. Darveaux, R., and Murty, K.L., “Effect of Deformation Behavior on Solder Joint
Reliability Prediction”, Proceedings of TMS-AIME Symposium on Microstructure
and Mechanical Properties of Aging Materials, Chicago, Nov. 1992

65. Reynolds, H.L., Kang, SH, and Morris, J .W. Jr., “The Creep Behavior of In-Ag
Eutectic Solder Joints”, Journal of Electronic Materials, 28, 1999, p. 69.

66. Murty, K.L., Yang, H., Deane, P., and Magill, P., Advances in Electronic Packaging,
1, Proceedings of the Pacific Rim/ASME International Intersociety Electronic &
Photonic Packaging Conference, edited by E. Suhir et al., ASME, New York, 1997,
p.1221.

67. Nabarro, F.R.N., Report of a Conference on Strength of Solids, Physical Society,
London, 1948, p. 75.

68. Herring, C., “Diffusional Viscosity of a Polycrystalline Solid”, Journal of Applied
Physics, 21, 1950, p. 437.

69. Coble, R.L., “A Model for Boundary Diffusion Controlled Creep in Polycrystalline
Materials”, Journal of Applied Physics, 34, 1963, p. 1679.

229

70. Evans, A.G., and Langdon, T.G., “Structural Ceramics”, Prog. Matls. Sc., 21, 1976, p.
171.

71. Raj, R., and Ashby, M.F., “Grain Boundary Sliding and Diffusional Creep”, Metall.
Trans, 2, 1971, p. 1113.

72. Beere, W., “Stress Redistribution During Nabarro-Herring and Superplastic Creep”,
Met. Sci., 10, 1976, p. 133.

73. Speight, M.V., “Mass Transfer and Grain-Boundary Sliding in Diffusion Creep”,
Acta. Metall., 23, 1975, p. 779.

74. Aigeltinger, A.E., and Gifldns, R.C., “Grain-Boundary Sliding During Diffusional
Creep”, Journal of Materials Science, 10, 1975, p. 1889.

75. Bieler, T.R., “High Temperature Deformation and Processing”, lecture notes from
1996 version of course MSM 974A “Advanced Physical and Mechanical Properties
of Materials 11”, MSU Printing Document Services, East Lansing, MI, Spring 2000, p.
1.

76. Arieli, A., Mukherjee, A.K., “High-Temperature Diffusion-Controlled Creep
Behavior of the Zn--22% Al Eutectoid Alloy Tested in Torsion”, Acta. Metall., 28,
1980,p.1571.

77. Weertman, J ., “Steady-State Creep of Crystals”, Journal of Applied Physics, 28,
1957,p.1185.

78. Mott, NE, in Creep and Fracture of Metals at High Temperatures, Proceedings.
NPL Symposium, HMSO London, 1956, p. 21.

79. Raymond, L., and Dorn, J .E., “Recovery of Creep-Resistant Substructures”, Trans.
AIME, 230, 1964, p. 560.

80. Barrett, CR, and Nix, W.D., “A Model for Steady-State Creep based on the Motion
of Jogged Screw Dislocations”, Acta. Metall., 13, 1965, p. 1247.

81. Weertman, J ., “Steady-State Creep through Dislocation Climb”, Journal of Applied
Physics, 28, 1957, p. 362.

82. Weertman, J ., “Creep of polycrystalline Aluminum as Determined from Strain Rate
Tests”, J. Mech. Phys. 801., 4, 1956, p. 230.

83. Weertman, J ., “Creep of Indium, Lead, and Some of Their Alloys with Various
Metals”, Trans. AIME, 218, 1960, p. 207.

230

84.

85.

86.

87.

88.

89.

91.

92.

93.

94.

95.

Weertman, J ., Discussion about F. Garafolo’s “An Empirical Relation Defining the
Stress Dependence of Minimum Creep Rate in Metals”, Trans. AIME, 227, 1963, p.
1475.

Robinson, S.L., and Sherby, O.D., “Mechanical Behavior of Polycrystalline Tungsten
at Elevated Temperatures”, Acta. Met., 17, 1969, p. 109.

Garafolo, E, “An Empirical Relation Defining the Stress Dependence of Minimum
Creep Rate in Metals”, Trans. AIME, 227, 1963, p. 351.

Sherby, OD, and Burke, P.M., “Mechanical Behavior of Crystalline Solids at
Elevated Temperatures”, Progr. Mater. Sci., 13, 1968, p. 325.

Wu, M.Y., and Sherby, O.D., “Unification of Harper-Dom and Power Law Creep
Through Consideration of Internal Stress”, Acta. Met., 32, 1984, p.156].

Ashby, M.F., “First Report on Deformation-Mechanism Maps”, Acta. Met., 20, 1972,
p. 887.

. Mei, Z., Morris, J.W. Jr., Shine, M., and Summers, T.S., “Effect of Cooling Rate on

Mechanical Properties of Near-Eutectic Tin-Lead Solder Joints”, Journal of
Electronic Materials, 20, 1991, p.599.

Sigelko, J ., Choi, S., Lucas, J .P., Subramanian, K.N., and TR. Bieler, “Effect of
Cooling Rate on Microstructure and Mechanical Properties of Eutectic Sn-Ag Solder
Joints with and without Intentionally Incorporated Cu68n5 Reinforcement”, Journal of
Electronic Materials, 28, No. 11, 1999, p.1184.

Frost, H]. and Howard, R.T., “Creep Fatigue Modeling for Solder Joint Reliability
Predictions Including the Microstructural Evolution of the Solder”, IEEE
Transactions on Components, Hybrids, and Manufacturing Technology, 13, No.4,
1990,p.727.

Tien, J .K., Hendrix, BC, and Attarwala, A.I., “Creep-Fatigue Interactions in
Solders”, IEEE Transactions on Components, Hybrids, and Manufacturing
Technology, 12, No.4, 1990, p. 502.

Rorning, A.D., Chang, Y.A., Stephens, J .J ., Frear, D.R., Marcotte, V., and Lea, C.,
“Physical Metallurgy of Solder-Substrate Reactions” in The Solder Mechanicas: A
State of the Art Assessment, edited by DR. Frear et al., TMS, Warrendale, PA, 1991,
p. 29.

Marshall, J .L., Foster, L.A., and Sees, J .A., in The Mechanics of Solder Alloy

Interconnects, edited by DR. Frear et al., Van Hostrand Reinhold, New York, NY,
l994,p.42.

231

96. Glazer, J ., “Metallurgy of Low Temperature Pb-free Solders for Electronic
Assembly”, International Materials Reviews, 40, No. 2, 1995, p. 65.

97. Warwick, ME, and Muckett, S.J., “Observations on the growth and Impact of
Intermetallic Compounds on Tin-Coated Substrates”, Circuit World, 9, No.4, 1983, p.
5.

98. Haimovich, J ., “Intermetallic Compound Growth in Tin and Tin-Lead Platings Over
Nickel and Its Effects on Solderability”, Welding Journal — Welding Research
Supplement, 68, No. 3, 1989, p. 1025.

99. Shoji, Y., Uchida, S., and Ariga, T., “ Dissolution of Solid Copper into Molten Tin
Under Static Conditions”, Trans. JIM, 21, 1980, p. 383.

100. Kawakatsu, 1., and Yamaguchi, H., “ On the Dissolution of Copper in Liquid Tin”, J.
JIM, 31, 1967, p. 1387.

101. Kawakatsu, 1., Osawa, T., and Yamaguchi, H., “ On the Growth of Alloy Layer
Between Solid Copper and Liquid Tin”, Trans. JIM, 13, 1972, p. 436.

102. London, J ., and Ashall, D.W., “Compound Growth and Fracture at Copper/1' in-
Silver Solder Interfaces”, Brazing Soldering, Autumn 1986, No. 11, p. 49.

103. Kay, P.J., and McKay, C.A., “Growth of Intermetallic Compounds on Common
Basis Materials Coated with Sn and Sn-Pb Alloys”, Trans. Inst. Met. Finish, 54, No.
1,1976,p.68.

104. Choi, S., Bieler, T.R., Lucas, J.P., and Subramanian, K.N., “Characterization of the
Growth of Intermetallic Interfacial Layers of Sn-Ag and Sn-Pb Eutectic Solders and
Their Composite Solders on Cu Substrate During Isothermal Long-Term Aging”,
Journal of Electronic Materials, 28, No. 11, 1999, p. 1209.

105. Sunwoo, A.J., Morris, J .W. Jr. and Lucey, G.K. Jr., “The Growth of Cu-Sn
Intermetallics at a Pretinned Copper-Solder Interface”, Metallurgical Transactions
A, 23A, 1992. PP. 1323.

106. Mei, Z., Sunwoo, AJ. and Morris, J .W. Jr., “Analysis of Low-Temperature
Intermetallic Growth in Copper-Tin Diffusion Couples”, Metallurgical Transactions
A, 23A, 1992, p. 857.

107. Lampe, B.T., “Room Temperature Aging Properties of Some Solder Alloys”,
Welding Journal —Welding Research Supplement, October 1976, p. 3305.

108. Brandes, E.A., and Brook, G.B., Smithell’s Metals Reference Book, 7th ed., Oxford,
Butterworth/Heinemann, 1992, Section 13.

232

109.

110.

111.

112.

113.

114.

115.

116.

117.

118.

119.

120.

Choi, W.K., and Lee, H.M., “Effect of Soldering and Aging Time on Interfacial
Microstructure and Growth of Intermetallic Compounds Between Sn-3.5Ag Solder
Alloy and Cu Substrate”, Journal of Electronic Materials, 29, No.10, 2000, p. 1207.

Chada, S, Laub, W, Foumelle, R.A., Shangguan, D., “An Improved Numerical
Method for Predicting Intermetallic Layer Thickness Developed During the
Formation of Solder Joints on Cu Substrate”, Journal of Electronic Materials, 28,
No.11, 1999, p. 1194.

Scarber, P.Jr., and J anowski, G.M, “Effect of Reinforcement Shape and Volume
Fraction on Resisual Streaa and Particle Cracking in Al.SiC Sub Composites”,
Mineral, Metal, Materials Society/AIME, 1996, p. 57.

Song, S.G., Shi, N., Gray, GT 111., and Roberts, J .A., “Reinforcement Shape Effect
on the Fracture Behavior and Ductility of Particulate-Reinforced 6061-Al Matrix
Composites”, Metallurgical and Materials Transactions A, 27A, No. 11, 1996, p.
3739.

Manko, H.H., Solders and Soldering, 2nd ed., McGraw-Hill, New York, NY, 1979,
Chapter 1.

Davis, P.E., Warwick, M.E., and Kay, P.J., “Intermetallic Compound Growth and
Solderability”, Plating and Surface Finishing, 69, 1982, p. 72.

Woodgate, R.W., Handbook of Machine Soldering, John Wiley & Sons, New York,
NY, 1983, Chapter 2.

Brothers, E.W., “Intermetallic Compound Formation in Soft Solders”, The Western
Electric Engineer, 25, Spring/Summer, 1981, p. 49.

Marshall, J.L., “Scanning Electron Microscopy and Energy Dispersive X-ray
(SEM/EDX) Characterization of Solder— Solderability and Reliability”, Solder
Joint Reliability, Van Nostrand Reinhold, New York, NY, 1991, p. 173.

Morris, J .W., Jr., Tribula, D., Summers, T.S.E., and Grivas, D., “The Role of
Microstructure in Thermal Fatigue of Pb-Sn Solder Joints”, Solder Joint Reliability,
Van Nostrand Reinhold, New York, NY, 1991, p. 225.

Onishi, M- and Fujibuchi, H., “Reaction Diffusion in the Cu-Sn System”,
Transactions Japanese Institute of Metals, 16, 1975, p. 539.

Vianco, P.T., Hosking, P.M., and Frear, D.R., Proceeding of Materials

Development in Microelectronic Packaging Conference, ASM International,
Materials Park, OH, 1991, p. 373.

233

121.

122.

123.

124.

125.

126.

127.

128.

129.

130.

131.

132.

Vianco, P.T., and Frear, D.R., “Issues in the Replacement of Lead-Bearing
Solders”, Journal of The Minerals, Metals & Materials Society, 45, 1993, p. 14.

Vianco, P.T., Hosking, P.M., and Rejent, J .A., Proceedings of Technical Program
—Nepcon West Conference, Anahein, CA, 1992, Cahners Exhibition Group, Des
Plains, IL, 3, 1992, p. 1730.

Yang, W., Felton, LE, and Messler, R.W., Jr., “The Effect of Soldering Process
Variables on the Microstructure and Mechanical Properties of Eutectic Sn-Ag/Cu
Solder Joints”, Journal of Electronic Materials, 24, 1995, p. 1465.

Melton, C., Skipor, A., and Thome, J ., Proceedings of Technical Program —
Nepcon West Conference, Reed Exhibitions, Norwalk, CT, 1993, p. 1489.

Tabor, D.S., Hardness of Metals, Clarendon Press, Oxford, UK, 1951.

Chada, S., Herrmann, A., Laub, W., Foumelle, R., Shangguan, D., and Achari, A.,
“Microstructural Investigation of Sn-Ag and Sn-Pb-Ag Solder Joints”, Soldering
and Surface Mount Technology, 9, No.2, 1997.

McDougall, J ., Choi, S., Bieler, T.R., Subramanian, K.N., Lucas, J .P.,
“Quantification of Creep Strain Distribution in Small Crept Lead-free in-situ
Composite and Non Composite Solder Joints”, Materials Science and Engineering,

A285, 2000, p.25.

Humphreys, F.J., Kalu, P.N., “Dislocation Particle Interactions During High-
Temperature Deformation of 2-Phase Aluminum-Alloys”, Acta Metall. 3S, 1987, p.
2815.

Mishra, R.S., Bieler, T.R., Mukherjee, A.K., “Overview No. 119, Superplasticity in
Powder Metallurgy Aluminum Alloys and Composites”, Acta Metall. et Mater., 43,
1995,p.877.

Li, Y., Langdon, T.G., “A Unified Interpretation of Threshold Stresses in the Creep
and High Strain Rate Superplasticity of Metal Matrix Composites”, Acta Mater., 47,
1999,p.3395.

Mohamed, F.A., “Interpreatation of Superplastici Flow in Terms of a Threshold
Stress”, J. Mater. Sci. 18, 1983, p. 582.

Nichols, F.A., “Overview No. 7: Plastic Instabilities and Uniaxial Tensile
Ductilities”, Acta Metall. 28, 1980, p. 663.

234

   

   

   

E

MICHIGAN SIAI ‘UNIVERSITY‘LIBRARIES
llllH‘HlMllllllllll*HlllllllIllllllllll
1.. ..Izll l :s:...;
3 1293 02334 2086