Dissertation

國立台灣科技大學
材料科學與工程研究所博士論文
學號：D10004808
以金屬玻璃薄膜阻障層來抑制電子封裝中錫鬚
晶生長之研究
Thin Film Metallic Glass as an Underlayer for Tin
Whisker Mitigation in Electronic Packaging
研究生：I Made Wahyu Diyatmika
指導教授：朱瑾博士
中華民國103年01月10日

摘要
一般於微電子封裝中，人們會加入一層阻障層來抑制錫鬚的生長，這錫鬚之生
成會導致電子器件之短路及失效，故這層阻障層可以防止銅錫之間反應進而產生介金
屬化合物，而此銅錫介金屬化合物為錫鬚生長的主要驅動力之一。目前於研究及工業
應用上厚度達數個微米之鎳金屬已被廣泛的被使用為阻障層，然而因為鎳層屬於多晶
結構，其晶界仍可能提供了銅錫之間反應擴散的途徑。因此於本研究中，以
Zr46Ti26Ni28 及 Zr51.7Cu32.3Al9Ni7 兩種不同成分的金屬玻璃薄膜作為阻障層來阻止銅錫
之間的反應，於本實驗中，以有鍍金屬玻璃膜及無鍍膜之試片作為比較，使其在恆溫
及循環模式下進行熱處理，我們發現，僅有25奈米的金屬玻璃薄膜已經可以有效阻止
銅錫之間的反應。並於加速測試下，在鍍有金屬玻璃阻障層之試片上我們沒有觀察到
任何錫鬚，相反的，在無鍍阻障層之試片，我們發現錫鬚的數量隨著時效時間增加及
溫度的上升而增加，此外，當於熱循環加熱測試時，由於金屬玻璃阻障層非常薄(僅25
奈米)，因此只會產生微量的壓應力。本研究發現，金屬玻璃阻障層能有效的抑制錫鬚
的生長，其厚度薄且具有非晶的結構使金屬玻璃薄膜可以成為有效抑制錫鬚產生的擴
散阻障層材料。
關鍵字：錫鬚、電子封裝、金屬玻璃薄膜、阻障層
ii

Abstract
Introduction of underlayer is one of the mitigation methods commonly used for the
suppression of the Sn whiskering phenomenon in electronic packaging. Sn whiskers have
been found to result in detrimental short circuits and arcing in electrical devices and
eventually the failure of device. The presence of a proper underlayer is used to prevent the
intermetallic compound formation resulting from a Cu/Sn interaction, which is believed to be
the major driving force of Sn whisker growth. Plated µm-thick Ni as an underlayer has been
widely studied and industrially accepted. However, Ni underlayer suffers from its
polycrystalline grain structure where grain boundaries can potentially act as a diffusion path
for the Cu/Sn interaction to take place. In this study, Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 thin
film metallic glasses (TFMGs) underlayers are introduced to alleviate the Cu/Sn interaction.
Samples with and without TFMGs underlayers were subjected to various heat treatments at
elevated temperatures in monotonic and cyclic modes. TFMG underlayer effectively blocks
the Cu/Sn interaction, even with the thickness as thin as 25 nm. No Sn whisker is observed in
the sample with TFMG underlayer after acceleration tests. In contrast, Sn whiskers are found
in the absence of the underlayer and the whisker density increases with increasing aging time
and temperature. In addition, with the concept of using very thin underlayer, the introduction
of TFMG underlayer is expected to yield insignificant degrees of compressive stress, which is
anticipated to occur when the samples are exposed to thermal cycling. It is found that TFMG
underlayer plays an important role in effectively suppressing Sn whisker growth. Their thin
thickness and amorphous nature are considered beneficial to make TFMGs as a promising
diffusion barrier for Sn whisker mitigation.
Keywords: Sn whiskers, electronic packaging, thin film metallic glass, underlayer
iii

Acknowledgments
I would like to express my greatest gratitude and appreciation to my advisor Professor
Jinn P. Chu for his guidance, valuable discussion and financial support during my study. A
thank-you is extended to my co-advisor Professor Y. W. Yen for introducing me to electronic
packaging technology. I am also indebted to Professor Joe Greene from University of Illinois
for opening my mind in doing research. I appreciate National Taiwan University of Science
and Technology for sponsoring me to pursue my Ph. D degree.
In my daily work I have been blessed with a friendly lab mates. So, I thank all lab
mates in E1-141 (Metallic Glasses and Thin Films Lab) for the wonderful time we spent
together, kind assistance and support during my experiments. I would also like to thank
Professor J. H. Huang, Professor J. S. C. Jang, Professor Albert Wu, Professor C. H. Hsueh,
Professor C. M. Chen, Professor C. R. Kao, members of oral defense committee for their
constructive advices.
Special thanks to Advanced Optoelectronic Device Fabrication Laboratory, the share
cleanroom facility at National Taiwan University of Science and Technology for the Sn layer
depositions. Last but not least, I would like to thank my parents, my brothers and Lya for all
the love, encouragement and understanding throughout all these years.
iv

Table of Contents
摘要 ...........................................................................................................................................ii
Abstract................................................................................................................................... iii
Acknowledgments ...................................................................................................................iv
List of Tables ........................................................................................................................ viii
List of Figures..........................................................................................................................ix
Chapter 1 Introduction............................................................................................................1
1.1 Background of study ........................................................................................................1
1.2 Objectives of study...........................................................................................................3
Chapter 2 Literature review ...................................................................................................5
2.1 Characteristics of Sn whisker...........................................................................................5
2.2 Mechanisms of Sn whisker growth..................................................................................6
2.3 Driving force of Sn whisker growth...............................................................................10
2.4 Cu-Sn thin film couples..................................................................................................11
2.5 Sn whisker mitigations...................................................................................................14
2.6 Ni underlayer..................................................................................................................16
2.7 Amorphous diffusion barrier..........................................................................................20
2.8 Thin film metallic glass..................................................................................................20
2.9 Wettability of metallic glass...........................................................................................22
2.10 Grain refinement in Cu alloy thin film.........................................................................27
2.11 Physical Vapor deposition (PVD)................................................................................28
2.11.1 Sputter deposition..................................................................................................28
2.11.2 Magnetron sputtering.............................................................................................30
2.11.3 Electron beam (e-beam) evaporation.....................................................................32
2.12 Electroplating deposition..............................................................................................34
v

Chapter 3 Experimental procedures....................................................................................36
3.1 Cu-Sn bulk couples ........................................................................................................36
3.1.1 Sample designations ................................................................................................37
3.1.2 Substrate preparations..............................................................................................37
3.1.3 Cu alloy thin film deposition...................................................................................37
3.1.4 Pre-annealing...........................................................................................................39
3.1.5 Sn layer deposition ..................................................................................................40
3.1.6 Aging treatment.......................................................................................................40
3.1.7 Surface morphology and interfacial observation.....................................................41
3.1.8 Chemical composition analysis ...............................................................................42
3.2 Thin film metallic glass characterizations......................................................................43
3.2.1 Thermal analysis......................................................................................................43
3.2.2 Crystallographic analysis.........................................................................................44
3.2.3 Microstructure analysis............................................................................................44
3.2.4 Electrical resistivity measurement...........................................................................45
3.2.5 Surface roughness analysis......................................................................................46
3.2.6 Adhesion evaluation ................................................................................................46
3.3 Cu-Sn thin film couples..................................................................................................47
3.3.1 Sample designations ................................................................................................48
3.3.2 Substrate preparations..............................................................................................48
3.3.3 Ti and Cu thin film depositions...............................................................................48
3.3.4 Thin film metallic glass depositions........................................................................50
3.3.5 Sn layer depositions.................................................................................................51
3.3.6 Aging treatment.......................................................................................................52
3.3.7 Thermal cycling.......................................................................................................52
vi

3.3.8 Thermal reflow ........................................................................................................52
3.3.9 Surface morphology and Sn whisker observation...................................................52
3.3.10 Crystallographic analysis.......................................................................................52
3.3.11 Microstructure analysis..........................................................................................53
Chapter 4 Results and discussion.........................................................................................54
4.1 Sn whisker formation in Cu-Sn bulk couples.................................................................54
4.1.1 Effect of Cu(Ru) underlayer on Sn whisker formation ...........................................55
4.1.2 Effect of pre-annealing on Sn whisker formation ...................................................58
4.2 Sn whisker formation in Cu-Sn thin film couples..........................................................61
4.3 Thin film metallic glass characterizations......................................................................65
4.3.1 Crystallographic analysis.........................................................................................65
4.3.2 Thermal analysis......................................................................................................67
4.3.3 Surface roughness analysis......................................................................................68
4.3.4 Electrical resistivity measurement...........................................................................69
4.3.5 Adhesion evaluation ................................................................................................69
4.4 Thin film metallic glass as an underlayer for Sn whisker mitigation.............................71
4.4.1 Thermal stability of Zr46Ti26Ni28 TFMG underlayer aged at room temperature.....71
4.4.2 Thermal stability of Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs underlayer aged at
various temperatures................................................................................................83
4.4.3 Thermal cycling stability of Zr46Ti26Ni28 TFMG underlayer ..................................94
4.4.4 Thermal reflow stability of Zr46Ti26Ni28 TFMG underlayer....................................96
Chapter 5 Conclusions & Future Works.............................................................................98
5.1 Conclusions........................................................................................................................98
5.2 Future works ......................................................................................................................99
References.............................................................................................................................101
vii

List of Tables
Table 2-1 Total number of Sn whiskers, nodules, and hillocks on the samples with DC- and
pulse-plated Ni barriers for different storage times in an ambient of 60°C and 93% RH [9]. 19
Table 3-1 The main deposition parameters of Cu(Ru) films...................................................38
Table 3-2 The main deposition parameters of pure Cu films. .................................................38
Table 3-3 The main deposition parameters of Ti film. ............................................................49
Table 3-4 The main deposition parameters of Cu films. .........................................................49
Table 3-5 The main deposition parameters of Zr46Ti26Ni28 TFMG.........................................50
Table 3-6 The main deposition parameters of Zr51.7Cu32.3Al9Ni7 TFMG................................50
Table 3-7 The main deposition parameters of Sn layer...........................................................51
Table 4-1 Classification of adhesion test results......................................................................70
viii

List of Figures
Figure 2-1 Various shapes of Sn whisker (adapted from http://nepp.nasa.gov/whisker/) [30]. 5
Figure 2-2 A schematic drawing of the top-view of regularly spaced whiskers on the Sn
surface. The whiskers have a diameter of 2a and a spacing of 2b [55]..................8
Figure 2-3 (a) Cross-sectional SEM images of a leg of the Sn–Cu finished leadframe (b) a
higher magnification image of the interface between the Sn–Cu finish and the Cu
leadframe and (c) a cross-sectional SEM image of pure Sn finish on Cu leadframe
prepared by focused ion beam, [57]. ....................................................................11
Figure 2-4 Micrographs (approximately 24 μm tall x 63 μm wide) of as-deposited tin films
fabricated by (a) matte electroplating (“thin”), (b) matte electroplating (“thick”),
(c) bright electroplating (“thick”), (d) DC sputtering, (e) resistive evaporation, (f)
electron beam evaporation, (g) electroless plating, and (h) bright electroplating
(“thin”) [59]..........................................................................................................13
Figure 2-5 SEM image showing whiskers on a sample deposited at a pressure of 0.17 Pa and
annealed initially at 323 K for 34 days. The sample was then aged at room
temperature for 15 months [26]............................................................................14
Figure 2-6 Top-view image of a SOIC-8 lead. The circled area corresponds to the most bent
area of the lead [12]..............................................................................................17
Figure 2-7 Top-view images of the as-reflowed leads with a Ni barrier thickness of (a) 1 μm,
(b) 2 μm, and (c) 4 μm [12]..................................................................................18
Figure 2-8 Surface SEM micrographs of the laminated Cu/Ni/Sn samples with (a) the DC-
plated Ni barrier and (b) the pulse-plated Ni barrier [9]. .....................................19
Figure 2-9 (a) Bending stress vs. surface strain curves for uncoated and coated BMG samples,
together with 316L stainless steel for comparison. The curves are offset along the
ix

x-axis for ease of viewing, (b and c) are photographs of uncoated and MG/Ti
bilayer-coated BMG [85]. ....................................................................................22
Figure 2-10 Variation curves of contact angles with time at 473 K for molten Bi–Sn on
Fe78B13Si9 substrates in (a) amorphous and (b) crystalline states [90].................24
Figure 2-11 Cross section EPMA of Bi–Sn/ Fe78B13Si9 substrate at 473 K (a) near Bi–
Sn/amorphous Fe78B13Si9 interface and (b) near Bi–Sn/crystalline Fe78B13Si9
interface [90]. .......................................................................................................25
Figure 2-12 TEM results from as-deposited (a) Cu(Ru), (b) Cu(RuNx) and (c) pure Cu films
[94]. ......................................................................................................................27
Figure 2-13 Events that occur on a surface being bombarded with energetic atomic-sized
particles [96].........................................................................................................29
Figure 2-14 Focused electron beam (e-beam) evaporation with a bent beam source [97]......33
Figure 3-1 Flowchart of experimental procedures for Cu-Sn bulk couples.............................36
Figure 3-2 Sample designations for Cu-Sn bulk couples used in this experiment. .................37
Figure 3-3 Magnetron Sputtering System................................................................................38
Figure 3-4 Vacuum furnace used for pre-annealing treatment. ...............................................39
Figure 3-5 Schematic drawing of electroplating setup. ...........................................................40
Figure 3-6 Laboratory oven used for various heat treatments.................................................41
Figure 3-7 A dual-beam focused ion beam (FIB, FEI Quanta 3D FEG). Inset is EDS (Oxford,
X-Max) detector. ..................................................................................................42
Figure 3-8 Flowchart of experimental procedures for TFMG characterizations.....................43
Figure 3-9 Differential scanning calorimetry (DSC, Netzsch 404 F3 Pegasus)......................43
Figure 3-10 X-ray diffractometry (D8 Discover SSS).............................................................44
Figure 3-11 Transmission electron microscopy (Philips Technai G2)....................................45
Figure 3-12 Four-point-probe apparatus (Laresta-EP MCP-T360). ........................................45
x

Figure 3-13 Atomic force microscopy (Bruker Icon)..............................................................46
Figure 3-14 Flowchart of experimental procedures for Cu-Sn thin film couples....................47
Figure 3-15 Sample designations for Cu-Sn thin film couples used in this experiment..........48
Figure 4-1 (a) Top-view SEM image of the sample without seed layer (b) top-view ion-
induced secondary electron image in higher magnification.................................54
Figure 4-2 Top-view SEM images of the sample with (a) 400-nm (b) 800-nm-thick Cu(Ru)
seed layers after aging at 60°C for 20 hr..............................................................55
Figure 4-3 Cross-sectional backscattered electron images of the sample (a) with Cu(Ru) seed
layer (b) without seed layer after aging at 60°C for 20 hr....................................56
Figure 4-4 EDS spectrum of the Cu6Sn5 IMC. ........................................................................57
Figure 4-5 Top-view SEM images of the samples aged at 60°C for 20 hr and after pre-
annealing at (a)-(c) 100°C (d)-(f) 400°C with Cu(Ru), pure Cu seed layers and
without seed layer, respectively. ..........................................................................58
Figure 4-6 Whisker density as a function of pre-annealing temperature.................................59
Figure 4-7 SEM micrograph showing various common shapes of Sn whiskers observed on Sn
layer after aging for 33 days at room temperature. ..............................................61
Figure 4-8 Various whisker morphologies found in the samples without underlayers. ..........62
Figure 4-9 Preparation of TEM sample in a specific location by FIB: (a) Sn whiskers grown
on the surface of Sn layer (b) deposition of a thin Pt protective layer (c) etching
of two rectangular trenches on the surface (d) a thin piece of layered sample is
left standing between the two holes. ....................................................................63
Figure 4-10 Cross-sectional TEM images of a Sn whisker: (a) in low magnification revealing
locations A and B for further analyses at high magnification in (b) and (c). (b)
and (c) HRTEMs and SADPs taken from different locations, showing the growth
direction along [220]. ...........................................................................................64
xi

Figure 4-11 XRD patterns of the TFMGs showing typical broad humps around 30°- 45° of
2θ..........................................................................................................................65
Figure 4-12 HRTEM of (a) Zr46Ti26Ni28 (b) Zr51.7Cu32.3Al9Ni7 TFMGs. Insets in (a) and (b)
are the corresponding SADPs...............................................................................66
Figure 4-13 DSC curves of the Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs. .........................67
Figure 4-14 Plane-view SEM images of Zr46Ti26Ni28 TFMG deposited onto a Cu layer in low
(a) and high (b) magnifications. (c) 2-D (d) 3-D AFM images............................68
Figure 4-15 SEM micrographs of the as-deposited Sn layer in the samples: without the
underlayer in low (a), high (b) magnifications in plane-view and (c) 50°-tilted
view; with the Zr46Ti26Ni28 TFMG underlayer in low (d), high (e) magnifications
in plane-view and (f) 50°-tilted view. ..................................................................71
Figure 4-16 SEM micrographs of samples with and without the Zr46Ti26Ni28 underlayer after
aging at room temperature for 5, 12 and 33 days. Whiskers appear as small bright
spots in the images. ..............................................................................................72
Figure 4-17 Whisker density as a function of aging time for the sample without TFMG
underlayer.............................................................................................................74
Figure 4-18 (a) and (b) Cross-sectional TEM images of the sample without TFMG underlayer
after aging at room temperature for 33 days. White dashed lines in (a) indicate
the approximate location of IMC region. .............................................................75
Figure 4-19 Schematic of Sn whisker formation during aging: (a) as-deposited condition, (b)
early stage of IMC formation, (c) IMC thickening and Kirkendall void formation
and (d) Sn whisker growth. ..................................................................................76
Figure 4-20 Schematic of four-zone structure proposed by Galyon et al. [16]. ......................78
Figure 4-21 Four-zone structure estimated for Sn/Cu. Red dashed-lines are for ease of
viewing for distinguishing different zones...........................................................79
xii

Figure 4-22 SEM micrographs showing the presence of IMC formed in the sample without
the TFMG underlayer in (a) plane and (b) 50°-tilted views, after aging for 33
days and Sn layer removal by chemical etching. .................................................80
Figure 4-23 Cross-sectional TEM image of the sample with Zr46Ti26Ni28 TFMG underlayer
after aging for 33 days..........................................................................................81
Figure 4-24 EDS line scans in the region with the inset revealing the location where the line
scans are performed..............................................................................................83
underlayer in low (a), high (b) magnifications in plane-view and (c) 50°-tilted
view; with the Zr46Ti26Ni28 TFMG underlayer in low (d), high (e) magnifications
in plane-view and (f) 50°-tilted view; with the Zr51.7Cu32.3Al9Ni7 TFMG
underlayer in low (g), high (h) magnifications in plane-view and (i) 50°-tilted
view. .....................................................................................................................84
Figure 4-26 SEM micrographs of samples with and without the TFMG underlayers after
aging at room temperature for various lengths of time. .......................................85
aging at 40°C for various lengths of time.............................................................86
aging at 60°C for various lengths of time.............................................................87
Figure 4-29 Whisker density as a function of aging time at various temperatures for the
samples without TFMG underlayer......................................................................88
Figure 4-30 XRD patterns of the Sn layers in the samples without the underlayer aged at
different temperatures...........................................................................................89
Figure 4-31 XRD patterns of the Sn layers in the samples with the Zr46Ti26Ni28 TFMG
underlayer aged at different temperatures............................................................90
xiii

Figure 4-32 XRD patterns of the Sn layers in the samples with the Zr51.7Cu32.3Al9Ni7 TFMG
underlayer aged at different temperatures............................................................91
Figure 4-33 Cross-sectional TEM image of the sample with Zr51.7Cu32.3Al9Ni7 TFMG
underlayer after aging at 60°C for 6 days.............................................................92
scans are performed..............................................................................................93
Figure 4-35 SEM micrographs of samples with and without the Zr46Ti26Ni28 TFMG
underlayers after thermal cycling for 500 cycles. ................................................94
Figure 4-36 XRD patterns of the Sn layers in the samples with and without the Zr46Ti26Ni28
TFMG underlayer after thermal cycling for 500 cycles.......................................95
TFMG underlayer after thermal reflow at 260°C.................................................97
xiv

Chapter 1 Introduction
1.1 Background of study
In electronic packaging, the old problem of spontaneous Sn whisker has come back
due to the restriction on use of lead (Pb) in the microelectronic industries. Sn whisker is a
serious cause of failure in electronic devices as they create short circuits. Sn whisker growth
is a phenomenon of stress relaxation where the compressive stress developed within Sn layer
is the driving force of Sn whisker growth [1-3]. Compressive stress generally originates from
the deposition process, mechanical machining, thermal expansion mismatch during the
thermal cycling, diffusion of Cu into the Sn, and the intermetallic compound (IMC)
formation. Moreover, Cu diffusion and IMC formation exhibit repetitive compressive stress
for Sn whisker growth [1]. Sn whisker formation was also observed under ambient conditions
[4-6]. It has been reported that at 25°C, the diffusion coefficient of Cu along the
crystallographic a and c axes of Sn is about 0.5 x 10-8
and 2 x 10-6
cm2
/sec, respectively [7].
This indicates that Cu diffusion into Sn occurs very quickly at room temperature.
One might consider that the grain size of the Cu as a diffusion species plays a role in
IMC formation. Therefore, it is important to investigate the effect of Cu grain size on IMC
formation. In this study, the grain size of Cu is varied with two approaches. One is by adding
minor amounts of insoluble element into Cu thin film prepared by magnetron sputtering. Due
to the grain refinement effect, the grain size of Cu alloy is expected to be smaller than that of
pure Cu film. The other one is by annealing, which is believed to exhibit the grain growth so
that the grain size of Cu is expected to be bigger after heat treatment. Revealing the effect of
grain size on diffusion behavior of Cu towards Sn layer would provide a better understanding
to look for a proper Sn whisker mitigation.
1

The most important concern of this study is how to prevent the growth of Sn whisker.
In order to mitigate this detrimental phenomenon, various methods have been proposed.
These methods include annealing of the Sn deposit, incorporating additives into the Sn
deposit, using a thick layer of large-grained Sn, optimization of the reﬂowing process, and
introducing an underlayer that act as a diffusion barrier [8].
Introducing plated µm-thick Ni underlayer as the diffusion barrier has been well
studied and industrially accepted [9-15] owing to its ability to block Cu diffusion and
mitigate the Sn whisker growth by generating built-in tensile stress [16]. A Ni underlayer
with large grain sizes may be an effective diffusion barrier due to a decrease in the grain
boundaries compared with smaller grains, which is believed to enhance the atomic diffusion
as major diffusion paths [9]. However, an underlayer with the polycrystalline grain structure
gives rise to grain boundaries for Cu/Sn interactions. Consequently, an underlayer with
amorphous structure with the absence of grain boundaries is beneficial to prevent diffusion
reactions.
Amorphous thin films have been studied and found to be excellent diffusion barriers
in integrated circuits applications [17]. These kinds of barriers block the interaction between
the interconnect materials and the silicon. In particular, refractory metal nitrides, deposited by
reactive sputtering have shown to be promising diffusion barriers [18-20]. Based on these
studies, it is suggested that amorphous thin films are potentially useful as the diffusion barrier
in other systems, in addition to those between the interconnect materials and the silicon.
In this study, thin film metallic glass (TFMG) with its amorphous nature is introduced
as the underlayer to alleviate Cu/Sn interaction and thus to reduce Sn whisker formation.
TFMG is of great interest in recent decades due to its unique properties adopted from its bulk
form such as high strength, large elastic limits, smooth surface and excellent corrosion/wear
resistance [21, 22]. There are some systems of thin film metallic glasses such as Cu-, Fe-, Zr-,
2

Pd-, Pt-based and etc. Particularly, in respect to the underlayer for Sn whisker mitigation, the
thermal stability and the electrical resistivity have to be considered. The production cost is
another important aspect that needs to take into account. In this study, we use the ternary
system of Zr46Ti26Ni28 (in atomic percentage) and quaternary system of Zr51.7Cu32.3Al9Ni7
TFMGs for their ease of fabrication and good thermal stability.
Many published studies have investigated Sn whisker formed on typical thick films
prepared using electroplating. Electroplating is a common method used by most
microelectronic industries. On the other hand, in some cases [23, 24], using thick
electroplated Sn layers could take up to several years for Sn whiskers to grow. However,
there are also many studies [25-28] working on thin film cases; for instance, one of the
pioneer studies [7] on Sn whisker used the thin film scheme prepared by vapor deposition. In
the present study, Cu-Sn thin film couples prepared by vapor depositions are chosen to
shorten the time for the Sn whisker growth. In addition, the grain sizes of vapor-deposited Cu
and Sn thin layers are much smaller than those of the thick electroplated layer counterpart
[29]. Therefore, faster and more massive diffusion are expected to occur in the fine-grained
structures. Furthermore, in this study, TFMGs alloyed with and without the presence of Cu
are studied to show to effectively block the Cu/Sn interactions after being subjected to
various acceleration tests at elevated temperatures.
1.2 Objectives of study
In general, the objectives of the present study are to prevent Cu/Sn interaction that
results in IMC formation. IMC formation is believed to generate compressive stress, which is
one of major driving forces for Sn whisker growth. In addition, the specific objectives of this
study are as follows:
1. To investigate the effect of grain size on diffusion behavior between Cu and Sn.
3

2. To find an alternative underlayer to replace the conventional polycrystalline Ni
underlayer in Cu-Sn couples.
3. To have a feasibility study on the use of TFMG as a potential underlayer for Sn
whisker mitigation.
4. To reveal the role of Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs in mitigating the
growth of Sn whisker after being subjected to heat treatments in monotonic and cyclic
modes.
5. To offer different perspectives to academia and industry in designing the proper
underlayer in Cu-Sn couples.
4

Chapter 2 Literature review
To study how Sn whiskers grow and how to mitigate them, it is important to
understand the properties of Sn whiskers. A review of some major theories regarding to
whisker growth will be given further in this chapter and the state-of-art of Sn whisker studies
will be discussed. In addition, the overview of thin film metallic glass which is a mitigation
method proposed in this study will be also described.
2.1 Characteristics of Sn whisker
A Sn whisker is characterized as a single crystal, growing spontaneously from
the Sn finish or Sn-rich lead free solder and has various shapes, as shown in Fig. 2-1 [30].
Figure 2-1 Various shapes of Sn whisker (adapted from http://nepp.nasa.gov/whisker/) [30].
Sn whiskers can yield a serious reliability risk to electronic assemblies. The general
risks include short circuits, low-pressure-induced metal vapor arcing (plasma) and debris
contamination [31]. The Sn whisker growth is believed to be affected by some factors such
as, mechanical force, temperature, residual stress, intermetallic compounds (IMC) formation,
5

oxidized layer, electric field, etc. The preferred growth direction of Sn whiskers was
reported as [001] [32]. Beta (β)-Sn possesses a body centered tetragonal (bct) crystal
structure with close packed direction of [001] and the crystallographic dimensions are a =
0.5819 nm and c = 0.3175 nm (c/a = 0.5456) [33]. The oxide layer might be readily broken in
this direction, which affects the preference orientation for Sn whisker growth [34].
2.2 Mechanisms of Sn whisker growth
Some important studies with respect to Sn whisker growth mechanisms are discussed
here based on those mostly adapted from review articles by Galyon and Osenbach [34-36].
The possible mechanisms are dislocation, recrystallization, grain boundary diffusion and
interface fluid flow. Peach first proposed the screw dislocation mechanism to explain whisker
growth [37], followed by Koonce and Arnold [38]. They stated that whisker growth was
from the base of the whisker and not from the top by observing that the tip morphology was
unchanged while the whisker grew longer [34]. Frank and Eshelby brought up a diffusion-
limited mechanism by which dislocations developed a whisker [39]. Moreover, Frank
proposed a dislocation slide process that depended on the self-diffusion of Sn [40, 41].
Amelinckx et al. explained a helical dislocation model for whisker growth [42]. Inconsistent
with the proposed dislocation mechanisms was a point made by Baker [34]. He indicated that
the base of the whisker was typically a grain boundary [43]. Meanwhile, Lindborg employed
XRD to determine stresses in electroplated zinc (Zn) films and showed a minimum stress
level was required for the initiation of whisker growth [44]. Fisher et al. found that a
compressive stress of 52 MPa could accelerate the whisker growth to 1 micron per sec at
room temperature. That rate was too fast for lattice diffusion. The point of his work was to
prove that the driving force for whisker growth was the compressive stress [45]. An
important observation from Key was that the lack of depression area around whiskers
6

indicated that Sn atoms travelled long distances to create a whisker [46]. Another significant
finding was that not all whiskers grew with the directions of low-indices planes. This fact
thus limited the applicability of the dislocation slip model. Ellis proposed a new mechanism
of recrystallization [47], which was further explained by Kudryavtsev [48], Kakeshita [49],
and Dunn [50]. Boguslavsky and Bush discussed recrystallization again and suggested that
the driving force for recrystallization was microstresses due to dislocations, the grain
boundary network stresses working for the grain growth [51].
Vianco et al. conducted a series of elegant experiments to propose a dynamic
recrystallization theory (DRX). They cast pure Sn into a hollow cylinder with a diameter of
10 mm, wall thickness of 2 mm and length of 20 mm, then put pressure on the two ends of
the cylinder to introduce compressive stress into Sn. They measured the creep behavior under
test conditions that would generate whiskers in a Sn film. A low activation energy was
obtained to indicate an ultrafast mass transport process, that supported the cyclic DRX model
[52, 53].
Tu proposed a different explanation for whisker growth. He suggested that the driving
force compressive stress was induced by the copper-tin (Cu6Sn5) intermetallic compound
formation between the Sn film and the Cu-based substrate [54]. In addition, the oxide layer
surrounding the whisker must be broken to allow the whisker to grow as well as a grain
boundary diffusion provided the diffusion path [55]. Tu also developed a mathematical model
based on a grain boundary diffusion mechanism to calculate the rate of whiskers growth.
Kinetic model of Sn whisker growth he developed is as follows [55].
7

Figure 2-2 A schematic drawing of the top-view of regularly spaced whiskers on the Sn
surface. The whiskers have a diameter of 2a and a spacing of 2b [55].
To consider the growth of a whisker, Tu assumed that the whiskers have a diameter of
2a and a separation of 2b as shown in Fig. 2-2 [55]. They have a steady-state growth in a two-
dimensional stress field. The diffusional process is formulated in cylindrical coordinates. The
driving force of movement is:
𝐹 = −
𝜕𝜎
𝜕𝑟
Ω ............................................................................. (1)
where σ is the stress, Ω is the atomic volume of Sn, and r is the radial coordinate. To
calculate the stress, it can be regarded as an energy density, and a density function obeys the
continuity equation [56]. Hence in a steady-state process,
∇2
𝜎 = 0 ................................................................................... (2)
The solution of the equation is:
𝜎 =
𝜎0
ln (𝑏 𝑎⁄ )
𝑙𝑛
𝑟
𝑎
..................................................................... (3)
where σ is the stress in the Sn film induced by the compound formation. The flux to grow the
whisker is evaluated at r = a to be
𝐽 = 𝐶
𝐷
𝑘𝑇
𝐹 ............................................................................... (4)
where C = 1/ Ω in a pure metal. Then the volume of materials transported to the base of the
whisker in a period of Δt results in a growth of Δh of the whisker. Let A = 2πas be the
8

peripheral area of the growth step at the base where s is the step height. The growth rate of
the whisker is
Δℎ
Δ𝑡
=
2
ln (𝑏 𝑎)⁄
𝜎0Ω𝑠𝐷
𝑘𝑇𝑎2
................................................................. (5)
As an example, to calculate the whisker growth rate given by the last equation, they took the
measured data to be a = 3 µm, b = 0.1 mm, σΩ = 0.01 eV (at σ = 0.79 x 109
dyn/cm), kT =
0.025 eV at room temperature, s = 0.3 nm, and D = 10-8
cm2
/s (the self-grain boundary
diffusivity of Sn at room temperature). They obtained a growth rate of about 1 x 10-9
cm/s.
Tu et al. proposed a new model of the grain boundary fluid flow transport to explain
the growth of whiskers [57]. Grain boundary fluid flow might be faster than “crystalline”
grain boundary diffusion. This idea originated from a comparison between self-diffusion and
viscous flow in liquids [32]. The model developed by Tu et al. as follows:
ℎ 𝑓
̇ =
(𝑝0−𝑝𝑖)𝛿3
6𝜂𝑎2 ln(𝑏 𝑎)⁄
.................................................................... (6)
where h is the height of the whiskers (m), 𝛿 is the step height, 𝑝0 is the stress at radius = b
(MPa), 𝑝𝑖 is the stress at radius = a (MPa), η is the viscosity of the atomic fluid flow, while
the meanings of the other parameters remain the same as equation (5).
The comparison of growth rate ℎ 𝑓
̇ based on the fluid flow mechanism and ℎ 𝑑
̇ based
on grain boundary diffusion was obtained as equation (7):
ℎ 𝑓̇
ℎ 𝑑̇
=
𝜋𝛿2
6Ω2 3⁄ ................................................................................ (7)
9

2.3 Driving force of Sn whisker growth
Whisker growth is a phenomenon of stress relaxation that exists within the pure Sn or
Sn alloy plating. In general, the driving force is compressive stress, which may results from
chemical, mechanical and thermal factors [34]:
1) Residual stresses within the Sn finish are caused in the plating process by the
factors of impurities, grain size, plating thickness, current density. The internal
stress is the main reason to induce the growth of crystals [57].
2) Intermetallic compound formation (the diffusions between the materials of tin plating
and substrate lead to formation of intermetallic compounds and cause compressive
stress within tin plating) [58].
3) External stress (or applied stress, which can be introduced by torquing of a nut or a
screw, bending or stretching of the surface, or any other inappropriate handling and
probing).
4) Coefficients of thermal expansion mismatch: the mismatch of thermal expansion
between the substrate and Sn plate creates the thermal stress at the interface that drive
whisker growth [34].
5) Diffusion (zinc and copper atoms from brass substrate both have high diffusivity
when they diffuse into tin film).
6) Oxide layer: stress is released from breaking or nicking the oxide layer of Sn film,
facilitating the whisker growth [58].
The origin of the compressive stress can be mechanical, thermal, and chemical. The
mechanical and thermal stresses, however, tend to be finite in magnitude so that they cannot
sustain a spontaneous or continuous growth of whiskers for a long time. The chemical force
is essential for spontaneous Sn whisker growth, but not obvious. The origin of the chemical
force is due to the room temperature reaction between Sn and Cu to form the intermetallic
10

compound of Cu6Sn5 as shown in Fig. 2-3 [57]. The chemical reaction provides a sustained
driving force for spontaneous growth of whiskers as long as the reaction keeps going with
unreacted Sn and Cu [57].
Figure 2-3 (a) Cross-sectional SEM images of a leg of the Sn–Cu finished leadframe (b) a
higher magnification image of the interface between the Sn–Cu finish and the Cu leadframe
and (c) a cross-sectional SEM image of pure Sn finish on Cu leadframe prepared by focused
ion beam, [57].
2.4 Cu-Sn thin film couples
Many published studies have investigated Sn whisker formed on typical thick films
prepared using electroplating. Electroplating is a common method used by most
microelectronic industries. On the other hand, some cases [23, 24], using thick electroplated
Sn layers could take up to several years for Sn whiskers to grow. However, there are also
many studies [25-27] working on thin film cases. For instance, one of the pioneer studies on
Sn whisker used the thin film scheme prepared by vapor deposition. Tu [7] investigated the
interdiffusion and intermetallic compound formation in Cu-Sn thin film couples by X-rays
using a Seemann-Bohlin diffractometer. In his study, the films were prepared by consecutive
evaporation at room temperature on fused quartz substrates and subsequently annealed
between -2 and 100°C. The thickness of Sn layer was ranged from 350 to 2500 nm and the
Cu layer was ranged from 180 to 600 nm. He stated that the advantage of using thin film
11

specimens was that the reactants could be detected by diffraction method at early stage of
reaction and his technique detected new phases formed and also measured lattice parameter,
grain size, degree of ordering and the rate of reaction. The rate of reaction was measured by
the change of the integrated intensities. Therefore he obtained both structural and kinetic data
about the reaction.
Pinol et al. studied the influence of Sn deposition methods on Sn whisker formation
[59]. Deposition methods employed include matte and bright electroplating, electroless
plating, sputtering, and evaporation (resistive and electron beam). Whiskers were found to
form fastest (at age = 0) when the films had been applied using electron beam evaporation
and DC sputtering as shown in Fig. 2-4 [59]. Resistive evaporation was the only method of
the three PVD techniques they employed which did not produce whiskered samples within
the observation period. “Thin” matte electroplated and electroless plated films whiskered
next, after an incubation period of 9 weeks. After more than a year of observation, no whisker
activity was observed on the “thick” matte electroplated, “thin” bright electroplated, “thick”
bright electroplated or resistively evaporated films.
12

Figure 2-4 Micrographs (approximately 24 μm tall x 63 μm wide) of as-deposited tin films
fabricated by (a) matte electroplating (“thin”), (b) matte electroplating (“thick”), (c) bright
electroplating (“thick”), (d) DC sputtering, (e) resistive evaporation, (f) electron beam
evaporation, (g) electroless plating, and (h) bright electroplating (“thin”) [59].
Another study using thin film scheme was done by Winterstein et al. They
investigated the effect of long-term sample aging on the formation of tin whiskers on sputter
deposited tin films with the thickness of 300 nm [26]. They controlled the Ar pressure during
the sputter deposition so that either compressive or tensile stress could be obtained within the
Sn films. Most importantly, they suggested that after long-term aging, the sample with the
thickness of 300 nm produced significant amount of whiskers with the length of several
hundred micrometers as illustrated in Fig. 2-5 [26]. They also suggested the maximum
growth rates could be estimated to be about 0.01 nm/s.
13

Figure 2-5 SEM image showing whiskers on a sample deposited at a pressure of 0.17 Pa and
annealed initially at 323 K for 34 days. The sample was then aged at room temperature for 15
months [26].
2.5 Sn whisker mitigations
Mitigation refers to processes that greatly enhance resistance to whisker formation.
Mitigation is not elimination; it is rather reduction in severity. Whiskers may still form after
application of a mitigation strategy. Effective mitigation strategies resultantly reduce risk of
whisker formation and/or whisker-induced harm. Recent research dedicated to the mitigation
of Sn whiskers is described here. Mathew et al. [60] summarized several mitigation methods
as follows:
1) Conformal coating is a thin layer of materials coated on the top of the Sn layer to
prevent whiskers from penetrating out of the layer. Uralane 5750 layer, epoxy,
acrylic, a potting material, silicone, and RTV are such coatings that can mitigate Sn
whiskers.
14

2) Various new plating electrolytes are developed to control the grain size of Sn films
and reduce the internal stress that form Sn whisker. The most popular electrolyte is
methane-sulfonic acid (MSA) with various additives.
3) Different surface treatments are applied to restrain whisker penetration such as
double Ni layers, surface roughening, polyhedral oligomeric silsesquioxanes (POSS)
which can break the Sn oxide layer to reduce the stress on the Sn film, etc. The
addition of magnesium (Mg), bismuth (Bi) or rare-earth (RE) elements might have
different mitigation effects as well.
4) Underlayer is a thin metal layer pre-coated underneath Sn layer. The most popular
underlayer is Ni, though silver (Ag)-Ni is investigated as well.
5) High-temperature annealing is applied to relieve residual stress which might restrain
whisker growth [60].
An underlayer is mainly a coating applied onto a substrate material, which is followed
by a second coating of a different material. Underlayers used with tin coatings are typically
nickel, copper, and silver. Nickel is the most commonly utilized underlayer for tin coatings.
Copper has been utilized for brass and iron based substrates. Silver is less commonly utilized.
Underlayers are used to enhance the corrosion resistance of the surface coating, to act as
diffusion barriers between the substrate material and the surface coating, and to change the
basic state of stress in the surface film.
Xu et al. [61], from Cookson Electronics Corporation, published stress measurements
for tin coatings on copper substrates with a nickel underlayer. They found that nickel
underlayer resulted in tin coatings with tensile residual stresses, whereas the tin films without
a nickel underlayer usually had residual compressive stresses. Furthermore, Xu reported that
whiskers did not form on tin coatings with nickel underlayer after 4 months of observation at
15

storage conditions of 25 and 50°C. These results would be the first known evidence in the
known scientific literature that nickel underlayer changed the tin coating stress state from
compressive to tensile. Zhang, et al. [62] showed data indicating that tin diffused into the
nickel underlayer in greater quantities than did the nickel into the tin. This was the first such
statement in the published technical literature. Whitlaw and Crosby [63] from Shipley Corp.
reported that nickel underlayer, on tin-coated copper-based alloy substrates with the nominal
composition of 2.4% iron, 0.03% phosphorus, and 0.1% zinc (C194), effectively eliminated
all evidences of whisker formation for observation periods of 4 months at storage conditions
of 52°C at 98% RH.
2.6 Ni underlayer
A Ni barrier layer plated between the tin coating and the copper substrate has been
proven to be effective to prevent whisker formation by changing the stressing level [9, 64-
66]. The thickness, porosity, and ductility of the Ni films seem to be very important to ensure
an effective barrier layer because the mechanical damage or stress of the surface films is
believed to cause ineffective mitigation. A minimum Ni barrier thickness of 0.5 μm was
recommended to guarantee a pore-free film because a porous Ni barrier could not prevent the
diffusion of Cu from the C194 substrate to the Sn film [67]. Ni barriers with thickness up to
several micrometers have been reported for Sn/Ni-coated flat strips [9], connectors [68], and
IC packages [69]. For connector applications, the deformation level of the plated leads during
manufacture or application may not be severe. Therefore, Ni barrier-induced mechanical
failure of matte Sn films seldom occurs. However, for the bended leads of IC package
application, a sound design of the Sn/Ni films is required to guarantee that the mechanical
stability is good because considerable plastic deformation could happen at the most bent
areas of the leads. This inevitably results in large residual stresses in that area.
16

Liu et al. has reported that the thick Ni barrier could induce crack in matte Sn film
[12]. They used a small out-line eight-lead integrated circuit (SOIC-8) package with a pitch
of 1 mm as the substrates and Ni barrier layer was introduced before Sn layer deposition by
electroplating as shown in Fig. 2-6 [12]. This study found that during the reflow process of
the lead-free package, thermal stress usually occurs as a result of coefficient of thermal
expansion mismatch between the coatings and the copper alloy substrate (Sn: 22 x 10−6
K−1
,
Ni: 13.4 x 10−6
K−1
, Cu: 16.5 × 10−6
K−1
) and a thinner Ni barrier with a thickness of 1 or 2
μm could help the matte Sn films to withstand reflow-induced stress without causing cracks.
Yet, the thicker Ni barrier could induce considerable mechanical damage to the matte Sn
films of the IC package, as can be seen in Figure 2-7 [12].
Figure 2-6 Top-view image of a SOIC-8 lead. The circled area corresponds to the most bent
area of the lead [12].
17

Figure 2-7 Top-view images of the as-reflowed leads with a Ni barrier thickness of (a) 1 μm,
(b) 2 μm, and (c) 4 μm [12].
Another work on the reliability of Ni layer as diffusion barrier was done by Chen et
al. [9]. They compared the reliability of DC- and pulse-plated Ni layers in term of Sn whisker
formation after aging at 60°C with 93% relative humidity (RH) as shown in Table 2-1 [9]. It
was observed that the pulse-plated Ni barrier resulted in a denser and smoother Sn layer than
the DC-plated Ni barrier. In terms of the DC-plated Ni barrier, the resulting Sn film surface
was composed of polygonal grains that piled loosely; however, the pulse-plated Ni barrier led
to a Sn layer with tightly connected and cobblestone-like grains on the surface, as presented
in Fig. 2-8 [9]. This indicated that the surface morphology of the plated Sn layer was
dependent on the characteristics of the Ni barrier.
18

Figure 2-8 Surface SEM micrographs of the laminated Cu/Ni/Sn samples with (a) the DC-
plated Ni barrier and (b) the pulse-plated Ni barrier [9].
They also suggested that the pulse-plated Ni layer seemed to have a better reliability
to prevent the Sn whisker formation owing to its bigger crystallite size and smaller
interplanar spacing compared to that of DC-plated Ni layer. In other words, a Ni underlayer
with large grain sizes might be an effective diffusion barrier due to a decrease in the grain
boundaries compared with smaller grains, which was believed to enhance the atomic
diffusion as major diffusion paths.
Table 2-1 Total number of Sn whiskers, nodules, and hillocks on the samples with DC- and
pulse-plated Ni barriers for different storage times in an ambient of 60°C and 93% RH [9].
Storage time With DC-plated Ni With pulse-plated Ni
7 days 39 whiskers 4 small hillocks
40 days 40 whiskers + 30 nodules 10 small hillocks
19

2.7 Amorphous diffusion barrier
Amorphous thin films have been studied and found to be excellent diffusion barriers
in integrated circuits applications [17]. These kinds of barriers block the interaction between
the interconnect materials and the silicon. In particular, refractory metal nitrides, deposited by
reactive sputtering have been shown to be promising diffusion barriers [18-20]. Based on
these studies, it is suggested that amorphous thin films are potentially useful as the diffusion
barrier in other systems, in addition to those between the interconnect materials and the
silicon.
2.8 Thin film metallic glass
Bulk metallic glass (BMG) is a multi-component alloy in amorphous state produced
by rapid solidification that exhibits unique set of properties which are could not be seen in
conventional crystalline materials due to the absence of crystalline defects [70]. The most
important feature of BMGs that distinguish them from general amorphous materials is the
glass transition that transforms supercooled liquid into a glassy state when cooled from high
to low temperature and vice versa [71].
The mechanical properties of metallic glasses are superior to their crystalline
counterparts in many cases. In tensile loading, the elastic strain limit of metallic glasses is
about 2%, much higher than that of common crystalline metallic alloys. Thus, the yield
strength of amorphous alloys is relatively high in tension and compression [72]. Upon
yielding at room temperature, metallic glasses often show plastic flow in absence of work-
hardening and a tendency towards work-softening leads to shear localization. Under tensile
condition, the localization of plastic flow into shear bands limits dramatically the overall
plasticity, so that metallic glass specimens usually fail catastrophically [70].
20

The first metallic glass was discovered in 1960 by Duwez and co-workers by rapid
quenching of liquid Au80Si20. A few years later, Chen and Turnbull were able to make
amorphous powders of ternary Pd–Si–N with N=Ag, Cu or Au. During the late 1980s,
Inoue’s group in Sendai, Japan, investigated rare-earth materials with Al and ferrous metals.
They found that alloying with a lanthanides and Al yields to excellent glass-forming ability
(GFA). Since then, research in the area of bulk metallic glasses has been growing
significantly [70]. As a result, several hundreds of kinds of metallic glasses such as Zr-, Cu-,
Ti-, Fe-, Pd-, Pt-, Ni-, Mg-, and Au-based systems have been discovered [22].
Thin film metallic glass (TFMG) is of great interest in recent decades due to its
unique properties adopted from its bulk form such as high strength, large elastic limits, and
excellent corrosion and wear resistance [21, 22]. Increasing interest in developing and
understanding this new family of materials has also led to making TFMG processing
possible, which was not readily achieved in the past when MGs were available only as
powder or ribbons [22].
Thin films prepared by vapor-to-solid deposition are expected to be farther from
equilibrium than those prepared by a liquid-to-solid melting/casting process. Thus, the GFA
can be further improved and composition ranges for amorphization are wider when formed
by thin film processing such as sputtering [73, 74]. In fact, sputter deposition technology has
been used for GFA determination of metallic glass systems, by varying the film composition
and density when co-sputtered with Zr and Cu elemental targets [75].
Owing to their unique properties, TFMGs have been reported to have many potential
applications including biomedical tools [76-78], recording layer [79], micro-actuator [80],
stability improvement of field emission cathode [81], fatigue property improvements of
particular substrates [82-84]. In addition, Chu et al. has reported that TFMG exhibited
bending ductility improvement of brittle substrate [85]. In this study, the bending ductility of
21

Zr50Cu30Al10Ni10 BMG substrate was improved by Zr53Cu26Al15Ni6 TFMG coating as shown
in Fig. 2-9 [85]. Accordingly, this study suggests that TFMG can be applied in electronic
packaging technology where the mechanical stability is needed; for instance, the bended
leads of IC package application [12].
Figure 2-9 (a) Bending stress vs. surface strain curves for uncoated and coated BMG samples,
together with 316L stainless steel for comparison. The curves are offset along the x-axis for
ease of viewing, (b and c) are photographs of uncoated and MG/Ti bilayer-coated BMG [85].
2.9 Wettability of metallic glass
Wetting properties of liquid alloy on solid substrates are important in various practical
applications, such as brazing, soldering, plasma spraying, moulding of steel and alloy [86-
89]. The amorphous alloys are in thermodynamically metastable states and they will transfer
into more stable states under appropriate circumstances. The amorphous alloys obtained at
22

higher quenching rates would relax structurally at low temperature. At higher temperatures,
the amorphous alloys may crystallize into polycrystalline phases. These changes will make
the wetting and diffusion behaviors of amorphous alloy substrates differ from that of
crystalline metal substrates. The wettability and diffusion processes are specific to metallic
glasses and differ from the well-established point defect mechanics in crystalline systems
[90].
Ma et al. [90] had done the contact angle and diffusion distance measurements of
molten Bi–Sn alloy on amorphous and crystalline Fe78B13Si9 at 423 K in order to test the
difference in wettability and diffusion mechanisms between the two materials. They found
that the contact angles for the amorphous Fe78B13Si9 substrate decreased monotonically with
increasing time which was similar to the variation tendency of the contact angles for
polycrystalline alloy substrate as shown in Fig. 2-10 [90]. The spreading rate of contact angle
on amorphous substrate is larger than that on the crystalline one. The mean equilibrium
contact angle (θeq) on amorphous substrate was 38.1°, which was smaller than that on
crystalline substrate. Therefore, the wettability of Bi–Sn alloy melt on Fe78B13Si9 amorphous
substrate was superior to that on crystalline one.
23

Figure 2-10 Variation curves of contact angles with time at 473 K for molten Bi–Sn on
Fe78B13Si9 substrates in (a) amorphous and (b) crystalline states [90].
24

Figure 2-11 Cross section EPMA of Bi–Sn/ Fe78B13Si9 substrate at 473 K (a) near Bi–
Sn/amorphous Fe78B13Si9 interface and (b) near Bi–Sn/crystalline Fe78B13Si9 interface [90].
They also stated that the content of Sn increased gradually high toward the interface
because highly active Sn tended to segregate on the interface as can be seen in Fig. 2-11 [90].
However, accumulation of Sn atom at the interface between the molten Bi–Sn and amorphous
Fe78B13Si9 was higher than between the molten Bi–Sn and crystalline Fe78B13Si9. Moreover,
the width of diffusion layer of Bi–Sn alloy melts on the amorphous substrate was above 1
μm, while the width of diffusion layer of Bi–Sn alloy melts on the crystalline substrate was
above 5 μm. They suggested two major reasons resulted in such different widths of diffusion
25

layer. One was because no grain-boundaries and interfaces existed in the amorphous alloy,
the diffusive activation energy of active atoms (e.g. Sn) in molten alloy should be increased.
The other one was that the homogenous atomic arrangement configuration in amorphous
alloy limited the nucleation of new phase with long incubation period, which makes the
diffusion of active atoms (e.g. Sn) more difficult. Although there are no grain boundaries and
interfaces in amorphous alloy, diffusion of atoms (e.g. Sn) in amorphous alloy can be realized
through the exchange between atoms and atoms or atoms and cavities (free volume) at
temperatures below Tg [90].
Another work on the wettability of Fe78B13Si9 metallic glass was also done by Ma et
al. [91]. They suggested another concept on the wettability of metallic glass by estimating
qualitatively the effect of structural relaxation and crystallization reaction on the surface
tension of amorphous Fe78B13Si9 substrate. Surface tension was mainly determined by the
atomic character and the spatial atomic configuration, which would lead to the difference in
surface tension between the amorphous alloy and polycrystalline alloy. Amorphous alloys
might be considered as solids with a frozen-in liquid structure, which indicated that the
atomic character and the spatial atomic configuration of amorphous alloys were similar to
those of molten alloys. Usually, the surface tension of molten alloys was lower than that of
the solid crystalline alloys. Therefore, it was reasonable to think that the surface tension of
the amorphous alloy was smaller than that of the crystalline alloy. Due to structural relaxation
and crystallization reaction, the amorphous alloys might crystallize into polycrystalline
phases at high temperatures. So the surface tension of amorphous Fe78B13Si9 alloy would
increase with the proceeding of structural relaxation and crystallization reaction [91].
26

2.10 Grain refinement in Cu alloy thin film
The additive elements are found to be beneficial for refining the grain structures [92].
In principle, the mechanism of grain refinement in thin film is quite straight forward.
Particularly, the grain refinement can be achieved by adding additives into the film. Alloying
effects have a substantial contribution to the microstructural characteristics such as defects,
small grain size, and strain energy. For the Cu thin film, the addition of small amount of
insoluble elements has been reported to inhibit the grain growth of the film [93]. These
additives often have little or negligible solubility in Cu. Therefore, the effective crystallinity
of alloy films decreases with the increase in alloying content. Chu et al. [94] demonstrated
the effect of adding small amount of insoluble element (Ru) and nitrogen on the grain size of
Cu films. Figure 2-12 [94] shows typical TEM results from as-deposited pure Cu, Cu(Ru),
and Cu(RuNx) films. Columnar structures with various crystallite sizes were clearly revealed.
The crystallites size were ~8–12 nm for Cu(Ru), ~5–9 nm for Cu(RuNx), and ~25–35 nm for
pure Cu. Crystallite refinement effects due to Ru and RuNx were thus apparent.
Figure 2-12 TEM results from as-deposited (a) Cu(Ru), (b) Cu(RuNx) and (c) pure Cu films
[94].
27

2.11 Physical Vapor deposition (PVD)
PVD processes are atomistic deposition processes in which material is vaporized from
a solid or liquid source in the form of atoms or molecules, transported in the form of a vapor
through a vacuum or low pressure gaseous (or plasma) environment to the substrate where it
condenses. Typically, PVD processes are used to deposit films with thicknesses in the range
of a few nanometers to thousands of nanometers; however they can also be used to form
multilayer coatings, graded composition deposits, very thick deposits and freestanding
structures [95].
2.11.1 Sputter deposition
Sputter deposition [96], which is often being called just sputtering, is the deposition of
particles whose origin is from a surface (target) being sputtered. Sputter deposition used to
deposit films of compound materials by sputtering from either a compound target or an
elemental target in a partial pressure of argon, nitrogen, etc.
The physical sputtering process involves the physical vaporization of atoms from a
surface by momentum transfer from bombarding energetic atomic-sized particles. The
schematic drawing of sputtering process is shown in Figure 2-13 [96]. The energetic particles
are usually ions of a gaseous material accelerated in an electric field. Sputtering was first
observed by Grove in 1852 and Pulker in 1858 using von Guericke-type oil-sealed piston
vacuum pumps. The terms “chemical sputtering” and “electro-chemical sputtering” have
been associated with the process whereby bombardment of the target surface with a reactive
species produces a volatile species. This process is now often termed “reactive plasma
etching” or “reactive ion etching” and is important in the patterning of thin films.
Sputter deposition can be done in:
• A good vacuum (< 10-5
Torr) using ion beams.
28

• A low pressure gas environment where sputtered particles are transported from the
target to the substrate without gas phase collisions (i.e., pressure less than about 5
mTorr) using a plasma as the source of ions.
• A higher pressure gas where gas phase collisions and “thermalization” of the
ejected particles occurs but the pressure is low enough that gas phase nucleation is
not important (i.e., pressure greater than about 5 mTorr but less than about 50 mTorr).
Figure 2-13 Events that occur on a surface being bombarded with energetic atomic-sized
particles [96].
The advantages of sputter deposition include:
(1) The sputtering enables to form uniform thin film, even over large areas.
(2) Easy to control the surface smoothness and uniformity of thin films.
(3) Deposition of films with predictable and stable properties nearly bulk.
(4) Excellent adhesion of films with the substrate.
(5) Deposition of the films having the nearly same composition as the target.
29

2.11.2 Magnetron sputtering
In 1935 Penning found that superimposition of the magnetic field can increase the
deposition rate of sputtered films. In the early 1960's, Gill and Kay proposed an inverted
magnetron sputtering system and demonstrated that the sputtering gas pressure was as low as
10-5
Torr, which were two orders lower than conventional sputtering systems. Magnetron
sputtering can be done either in direct current (DC) or radio frequency (RF) modes [96].
2.11.2.1 DC Magnetron Sputtering
In DC diode sputtering, the electrons that are ejected from the cathode are accelerated
away from the cathode and are not efficiently used for sustaining the discharge. By the
suitable application of a magnetic field, the electrons can be deflected to stay near the target
surface and by an appropriate arrangement of the magnets; the electrons can be made to
circulate on a closed path on the target surface. This high flux of electrons creates high
density plasma from which ions can be extracted to sputter the target material producing a
magnetron sputtering configuration.
The principal advantage to the magnetron sputtering configuration is that dense
plasma can be formed near the cathode at low pressures so that ions can be accelerated from
the plasma to the cathode without loss of energy due to physical and charge-exchange
collisions. This allows a high sputtering rate with a lower potential on the target than with
the DC diode configuration. This configuration allows the sputtering at low pressures (<5
mTorr), where there is no thermalization of particles from the cathode, as well as at higher
pressures (>5 mTorr) where thermalization occurs. DC sputtering is done with conducting
targets. If the target is a non-conducting material, the positive charge will build up on the
material and it will stop sputtering.
30

2.11.2.2 RF Magnetron Sputtering
RF sputtering can be done both conducting and non-conducting materials. Here,
magnets are used to increase the percentage of electrons that take part in ionization of events
and thereby increase the probability of electrons striking the argon atoms, increase the length
of the electron path, and hence increase the ionization efficiency significantly [96].
At frequencies above 50 kHz, the ions do not have enough mobility to allow
establishing a DC diode-like discharge and the applied potential is felt throughout the space
between the electrodes. The electrons acquire sufficient energy to cause ionizing collisions in
the space between the electrodes and thus the plasma generation takes place throughout the
space between the electrodes. When an RF potential, with a large peak-to-peak voltage, is
capacitively coupled to an electrode, an alternating positive/negative potential appears on the
surface. During part of each half-cycle, the potential is such that ions are accelerated to the
surface with enough energy to cause sputtering while on alternate half-cycles, electrons reach
the surface to prevent any charge buildup. RF frequencies used for sputter deposition are in
the range of 0.5–30 MHz with 13.56 MHz being a commercial frequency that is often used.
Rf sputtering can be performed at low gas pressures (<1 mTorr).
Since the target is capacitively coupled to the plasma, it makes no difference whether
the target surface is electrically conductive or insulating although there will be some
dielectric loss if the target is an insulator. If an insulating target material, backed by a metal
electrode is used, the insulator should cover the whole of the metal surface since exposed
metal will tend to short-out the capacitance formed by the metal-insulator-sheath-plasma. RF
sputtering can be also used to sputter electrically insulating materials, although the sputtering
rate is low [96].
31

2.11.3 Electron beam (e-beam) evaporation
Focused high energy e-beams are necessary for the evaporation of refractory materials
such as most ceramics, glasses, carbon, and refractory metals [97]. This e-beam heating is
also useful for evaporating large quantities of materials. When vaporizing solid surfaces of
electrically insulating materials, local surface charge buildup can occur on the source surface,
leading to surface arcing, which can produce particulate contamination in the deposition
system.
In the deflected electron gun, the high energy e-beam is formed using a thermionic-
emitting filament to generate the electrons, high voltages (10–20 kV) to accelerate the
electrons, and electric or magnetic fields to focus and deflect the beam onto the surface of the
material to be evaporated as indicated in Fig. 2-14 [97]. E-beam guns for evaporation
typically operate at 10–50 kW though some operate as high as 150 kW. By using high power
e-beam sources, deposition rates as high as 50 microns per second have been attained from
sources capable of vaporizing material at rates of up to 10–15 kilograms of aluminum per
hour. Electron beam evaporators may be made compatible with UHV processing. Generally,
e-beam evaporators are designed to deposit material in the vertical direction, but high rate e-
beam sources have been designed to deposit in a horizontal direction.
32

Figure 2-14 Focused electron beam (e-beam) evaporation with a bent beam source [97].
In many designs, the e-beam is magnetically deflected through >180° to avoid
deposition of evaporated material on the filament insulators. The beam is focused onto the
source material, which is contained in a water-cooled copper hearth “pocket”. The e-beam
may be rastered over the surface to produce heating over a large area. Electron gun sources
may have multiple pockets so that several materials can be evaporated by moving the beam
or the crucible, so that more than one material can be vaporized with the same multipocket
electron source.
The high-energy electron bombardment produces secondary electrons that are
magnetically deflected to ground. The electrons ionize a portion of the vaporized material and
these ions or the emission from excited atoms may be used to monitor the evaporation rate. If
they are not removed, the secondary electrons can create an electrostatic charge on
electrically insulating substrates. If the fixture is grounded, the electrostatic charge may vary
over the substrate surface, particularly if the surface is large, affecting the deposition pattern
33

and properties of the deposited film. This can be averted by electrically floating the substrate
fixture so that it assumes a uniform electrically floating potential.
2.12 Electroplating deposition
Electroplating is an electrodeposition process for producing a dense, uniform, and
adherent coating, usually of metal or alloys, upon a surface by the act of electric current [98].
The coating produced is usually for decorative and/or protective purposes, or enhancing
specific properties of the surface. The surface can be conductors, such as metal, or
nonconductors, such as plastics. Electroplating products are widely used for many industries,
such as automobile, ship, air space, machinery, electronics, jewelry, defense, and toy
industries. The core part of the electroplating process is the electrolytic cell (electroplating
unit). In the electrolytic cell (electroplating unit), a current is passed through a bath
containing electrolyte, the anode, and the cathode. In industrial production, pretreatment and
post treatment steps are usually needed as well.
The physical embodiment of an electroplating process consists of four parts:
1. The external circuit, consisting of a source of direct current (DC), means of conveying
this current to the plating tank, and associated instruments such as ammeters,
voltmeters, and means of regulating the voltage and current at their appropriate
values.
2. The negative electrodes or cathodes, which are the material to be plated, called the
work, along with means of positioning the work in the plating solution so that contact
is made with the current source.
3. The plating solution itself, almost always aqueous, called by platers the "bath".
34

4. The positive electrodes, the anodes, usually of the metal being plated but sometimes
of a conducting material which serves merely to complete the circuit, called inert or
insoluble anodes.
The work to be plated is the cathode (negative electrode) of an electrolysis cell
through which a direct electric current is passed. The work is immersed in an aqueous
solution (the bath) containing the required metal in an oxidized form, either as an aquated
cation or as a complex ion. The anode is usually a bar of the metal being plated. During
electrolysis metal is deposited on to the work and metal from the bar dissolves:
at cathode: Mz+
(aq) + ze-
→ M(s)
at anode: M(s) → M z+
(aq) + ze-
Faraday's laws of electrolysis govern the amount of metal deposited. Works are
electroplated to (i) alter their appearance; (ii) to provide a protective coating; (iii) to give the
work special surface properties; (iv) to give the work engineering or mechanical properties.
35

Chapter 3 Experimental procedures
In this chapter, the experimental procedures are described in detail. Since there were
two kinds of substrates used in this study, the experimental procedures are described in two
parts. One is Cu-Sn bulk couples experiment in which the Cu foil was used as the substrate.
This experiment was aimed to investigate the effect of Cu grain size in the Cu/Sn
interdiffusion. The other one is Cu-Sn thin film couples in which the Si wafer was used as the
substrate and the TFMG underlayers were introduced between Cu and Sn layers.
3.1 Cu-Sn bulk couples
Releasing residual stress
(pre-annealing)
Substrate preparations
Cu alloy and pure Cu thin
films depositions
(magnetron sputtering)
Sn layer deposition
(electroplating)
Aging at 60°C for 20 hr
Characterizations
Surface & interfacial
observation
(SEM, FIB)
Chemical composition analysis
(EDS)
Figure 3-1 Flowchart of experimental procedures for Cu-Sn bulk couples.
36

3.1.1 Sample designations
This experiment was done actually with the idea of having different grain sizes of Cu
as a diffusion species in sputtered Cu(Ru) and pure Cu thin films and comparing with the
sample without any underlayer. The sample designations in the Cu-Sn bulk couples
experiment are shown in Fig. 3-2.
Figure 3-2 Sample designations for Cu-Sn bulk couples used in this experiment.
3.1.2 Substrate preparations
Commercial Cu foil used as the substrate was cut into a dimension of 1.5 cm x 1.5
cm. The substrate was ultrasonically cleaned in acetone, rinsed in de-ionized water and then
dipped in diluted HCl solution. Finally, the substrate was ultrasonically rinsed in de-ionized
water again to remove the contamination on the substrate surface.
3.1.3 Cu alloy thin film deposition
Approximately 400- and 800-nm-thick Cu films added with 1.1 at. % Ru, denoted as
Cu(Ru) hereafter, were deposited onto the substrates by RF magnetron sputtering of Cu(Ru)
alloy target. As comparison, pure Cu films were also deposited with the same thickness by
using DC power. Table 3-1 shows the sputtering parameter that has been employed for both
thin films.
37

Figure 3-3 Magnetron Sputtering System.
Table 3-1 The main deposition parameters of Cu(Ru) films.
Parameter Value
Working Pressure 3 mTorr
Base Pressure < 5 x 10-7
Torr
RF Power 100 W
Substrate Bias 50 V
Ar Flow Rate 20 sccm
Target – Substrate Distance 100 mm
Pre-sputtering 5 min
Substrate Rotation 20 rpm
Deposition Rate 0.11 nm/s
Table 3-2 The main deposition parameters of pure Cu films.
Parameter Value
38

Torr
DC Power 100 W
Substrate Bias 50 V
3.1.4 Pre-annealing
In order to release the residual stress, before Sn layer deposition, samples were
annealed, denoted as pre-annealing hereafter, in vacuum furnace at elevated temperatures
such as, 100°C, 150°C, 200°C for 20 min and 300°C for 15 min, and also at 400°C for 10
min.
Figure 3-4 Vacuum furnace used for pre-annealing treatment.
39

3.1.5 Sn layer deposition
Prior to electroplating, each sample was taped with clear platers tape on its back side
so that the plating was applied to one side only. The Solderon ST-380 (Rohm and Haas
Electronic Materials) as the electroplating method was used in this work. The matte Sn layer
was electroplated at 5 ASD for 2 min to grow a Sn layer with thickness of 4.5 µm. The
temperature of the electroplating condition was controlled at 27°C. The electroplating station
was equipped with a power supply connected in series to a digital volt-ohm meter (VOM) to
precisely measure the applied cell current.
Figure 3-5 Schematic drawing of electroplating setup.
3.1.6 Aging treatment
All of samples were aged in a laboratory oven under ambient atmosphere at 60°C for
20 hr to accelerate the whisker formation. This temperature was chosen because this
temperature was thought to be the optimum temperature for Sn whisker to grow.
40

Figure 3-6 Laboratory oven used for various heat treatments.
3.1.7 Surface morphology and interfacial observation
A dual-beam focused ion beam (FIB, FEI Quanta 3D FEG) equipped with scanning
electron microscope (SEM) mode operated at an accelerating voltage of 30 kV was used to
observe the surface morphology and interfacial reaction as shown in Fig. 3-7. SEM was also
used to observe the Sn whiskers. The cross-sections were prepared using FIB. The samples
were milled at a 52° tilt angle with a 30-kV gallium (Ga) ion beam operating at a current of
0.1 nA. Initial trench milling of the sample was done at 15 nA and the final face milling at 1-
3 nA. In SEM, images acquired with both the secondary and backscattered electrons are
produced from the interaction of the electron beam with the specimen.
41

Figure 3-7 A dual-beam focused ion beam (FIB, FEI Quanta 3D FEG). Inset is EDS (Oxford,
X-Max) detector.
3.1.8 Chemical composition analysis
The chemical composition of the intermetallic compound (IMC) formed after aging
treatment was determined with an SEM using energy dispersive spectrometer (EDS). The
measurements were performed with an accelerating voltage of 30 kV, a beam current of 62
pA and a 30 s acquisition time.
42

3.2 Thin film metallic glass characterizations
3.2.1 Thermal analysis
Figure 3-9 Differential scanning calorimetry (DSC, Netzsch 404 F3 Pegasus).
Substrate preparation
Thin film metallic glass
depositions
Thermal analysis
(DSC)
Surface
morphology
and roughness
analyses
(SEM & AFM)
Chemical
composition
(EDS)
Microstructure
&
Crystallographic
Analysis
(XRD & TEM)
Adhesion evaluation
(Elcometer 107)
Figure 3-8 Flowchart of experimental procedures for TFMG characterizations.
43

The glass transition and crystallization temperatures of Zr46Ti26Ni28 and
Zr51.7Cu32.3Al9Ni7 TFMGs were evaluated using differential scanning calorimetry (DSC,
Netzsch 404 F3 Pegasus). DSC thermal analysis of TFMGs was carried out up to 700˚C at a
heating rate of 40 K/min in argon environment.
3.2.2 Crystallographic analysis
Figure 3-10 X-ray diffractometry (D8 Discover SSS).
X-ray diffractometry (XRD) was used to identify crystal structure characterization
(BRUKER, D8 DISCOVER SSS) using monochromatic Cu Kα radiation (λ=1.5406 Å) at 40
kV and 200 mA as shown in Fig. 3-10. The angle between X-ray incident beam and sample
surface was 1°. The analyses were performed in a 2θ range from 20° to 100° with a step
speed of 0.05° per second.
3.2.3 Microstructure analysis
A transmission electron microscope (TEM, Philips Tecnai G2) with an accelerating
voltage of 200 kV was used for microstructural characterizations as shown in Figure 3-11.
44

FIB with a Ga ion source and an accelerating voltage of 30 kV was used for TEM sample
preparation.
Figure 3-11 Transmission electron microscopy (Philips Technai G2).
3.2.4 Electrical resistivity measurement
Figure 3-12 Four-point-probe apparatus (Laresta-EP MCP-T360).
The electrical resistivity of the TFMGs was measured using a four-point probe
apparatus (Laresta-EP MCP-T360) at 100 mA as illustrated in Fig. 3-12. The electrical
resistivity was the average value of at least 10 measurements taken on each sample.
45

3.2.5 Surface roughness analysis
Figure 3-13 Atomic force microscopy (Bruker Icon).
The surface roughness of the TFMG deposited on the Cu layer was measured using
atomic force microscopy (AFM, Bruker Icon) in tapping mode with a tip radius curvature of
2 nm and a force constant of 0.4 N/m and a resonant frequency of 70 kHz.
3.2.6 Adhesion evaluation
The adhesion of the TFMG to the Cu layer was evaluated by a cross-hatch cut method
(Elcometer 107) in accordance with ASTM D3359-02. The Elcometer 107 Cross Hatch
Cutter for adhesion tests provides an instant assessment of the quality of the bond to the
substrate. Due to its rugged construction this cross hatch gauge is ideal for thin, thick or
tough coatings on all surfaces.
46

3.3 Cu-Sn thin film couples
Thin film metallic glass
depositions
Substrate preparation
Ti and Cu thin films
depositions
Sn layer deposition
(e-beam evaporation)
Aging treatment
(laboratory oven)
Characterizations
Surface & interfacial
observation
(FIB, SEM)
Chemical composition
analysis
(EDS)
Thermal cycling
(500 cycles)
Crystallographic
analysis
(XRD & TEM)
Thermal reflow
(260°C)
Figure 3-14 Flowchart of experimental procedures for Cu-Sn thin film couples.
47

3.3.1 Sample designations
In this experiment, three kinds of layered samples, with and without the underlayers,
were prepared on the Si wafer (100) substrate.
Figure 3-15 Sample designations for Cu-Sn thin film couples used in this experiment.
3.3.2 Substrate preparations
A Si wafer (100) was used as the substrate. The substrates were cleaned by acetone
and ethanol. Prior to depositions of film layers, the substrates were initially sputter-etched in
the sputtering chamber by using a DC power supply with an applied bias of -500 V in 35
mTorr of Ar for 10 min. Sputter etching was done in order to remove the oxide layer formed
on the Si wafer surface.
3.3.3 Ti and Cu thin film depositions
The experiment was then continued by the deposition of a 20 nm Ti adhesive layer.
The Cu layer with the thickness of 500 nm was then sputter deposited in the same chamber
without breaking vacuum. The deposition parameters for both Ti and Cu films are listed in
Tables 3-3 and 3-4, respectively.
48

Table 3-3 The main deposition parameters of Ti film.
Parameter Value
Torr
RF Power 100 W
Substrate Bias 50 V
Deposition time 0.06 nm/s
Table 3-4 The main deposition parameters of Cu films.
Parameter Value
Torr
RF Power 100 W
Substrate Bias 50 V
49

3.3.4 Thin film metallic glass depositions
TFMGs of Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 (in atomic percent) with the thickness
ranging from 25 to 100 nm were deposited prior to the Sn deposition. The deposition
parameters for both Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs are listed in Table 3-5 and
3-6, respectively.
Table 3-5 The main deposition parameters of Zr46Ti26Ni28 TFMG.
Parameter Value
Torr
RF Power 100 W
Substrate Bias 50 V
Deposition Time 0.11 nm/s
Table 3-6 The main deposition parameters of Zr51.7Cu32.3Al9Ni7 TFMG.
Parameter Value
Torr
RF Power 100 W
Substrate bias 50 V
50

3.3.5 Sn layer depositions
The Sn layer with the thickness of 400 nm was deposited by electron beam
evaporation. The Sn layer deposition was performed in Advanced Optoelectronic Device
Fabrication Laboratory, the share cleanroom facility on campus. The Sn layer deposition was
done with 5 keV electron gun and 130 mA average beam current. The deposition rate was
monitored by quartz crystal microbalance (QCM). The deposition parameters for the Sn layer
are listed in Table 3-7.
Table 3-7 The main deposition parameters of Sn layer
Parameter Value
Operating voltage 5 keV
Beam current 130 mA
Base pressure < 5 x 10-7
Torr
Working pressure 2 x 10-6
Torr
Target – Substrate Distance 15 cm
51

3.3.6 Aging treatment
Aging of the samples were carried out in ambient atmosphere at room temperature,
40°C and 60°C for various periods of time. The aging treatment was done in a laboratory
oven.
3.3.7 Thermal cycling
Thermal cycling was also carried out for 500 cycles in the temperature range of -35°C
to +85°C. This kind of heat treatment was performed in a laboratory oven under ambient
atmosphere.
3.3.8 Thermal reflow
Thermal reflow was also conducted in a laboratory oven under ambient atmosphere.
The samples were reflowed for 3 times at a peak temperature of 260°C and a holding time of
30 s. According to JEDEC J-STD-020, soldering temperature must be greater than 230°C to
ensure proper melting of Sn solder.
3.3.9 Surface morphology and Sn whisker observation
After aging, thermal cycling and thermal reflow treatments, the surface morphology
of the samples was examined using SEM. To assess the whisker formation, the whisker
density was taken by measuring the whisker numbers observed in at least 10 different SEM
image areas.
X-ray diffraction (XRD) measurements at the glancing angle of 1° were done to
detect the IMC formation present at the film/substrate interface. The analyses were performed
52

in a 2θ range from 20° to 100° with a step speed of 0.05° per second. XRD BRUKER, D8
DISCOVER is capable of depth-controlled phase identification.
3.3.11 Microstructure analysis
The microstructure of layered samples was analyzed by TEM (Philips Technai G2).
FIB was used to prepare TEM samples, in particular to prepare the cross sections of the root
of whiskers, to obtain structural and morphological information on the Sn whiskers and the
grains surrounding them.
53

Chapter 4 Results and discussion
4.1 Sn whisker formation in Cu-Sn bulk couples
This part of study investigates the effects of the Cu alloy thin film as a seed layer and
annealing on Sn whisker formation. In order to study those effects, Cu thin films, added with
minor concentration of insoluble Ru with various thicknesses prepared by radio magnetron
sputtering, were introduced as the seed layers. Pure Cu thin films with the same thicknesses
were also introduced for comparisons. Pre-annealing of the samples was also carried out prior
to Sn electroplating. To accelerate Sn whisker growth, all of samples were aged at 60°C for
20 hr. Their effects in term of whisker density are revealed in this section.
Figure 4-1 (a) Top-view SEM image of the sample without seed layer (b) top-view ion-
induced secondary electron image in higher magnification.
54

Figure 4-1(a) shows the surface morphology of Sn layer of the sample without seed
layer. It can be seen that there is no Sn whisker formation in as-deposited condition indicating
that the internal stress developed during Sn layer deposition may not be enough for Sn
whisker to grow. Figure 4-1(b) shows the grain morphology of Sn layer in as-deposited
condition. It reveals the average of the grain size is about 4 µm, which is comparable with the
thickness of Sn layer.
4.1.1 Effect of Cu(Ru) underlayer on Sn whisker formation
Figure 4-2 Top-view SEM images of the sample with (a) 400-nm (b) 800-nm-thick Cu(Ru)
seed layers after aging at 60°C for 20 hr.
55

The surface morphology of Sn layer in the sample with 400-nm and 800-nm thick of
Cu(Ru) seed layers are presented in Fig. 4-2. It is found that the Sn whiskers become more
populous and the whiskers grow longer when a thicker Cu(Ru) seed layer is introduced. It
suggests that the compressive stress induced by IMC formation in the sample with the thicker
seed layer may be higher than that of the thinner one.
Figure 4-3 Cross-sectional backscattered electron images of the sample (a) with Cu(Ru) seed
layer (b) without seed layer after aging at 60°C for 20 hr.
The IMC is found to be thicker in the sample with 400-nm-thick Cu seed layer
compared with the IMC formed at the interface of the sample without seed layer as shown in
Figs. 4-3 (a) and (b). With the presence of the Cu(Ru) seed layer, Cu atoms likely diffuse
more dramatically into the Sn layer. It is considered to cause a thicker IMC formation when
the seed layer is introduced. Likely, the Kirkendall voids are found as an indication of the
56

rapid and unbalanced interdiffusion between Cu and Sn [99-101]. Figure 4-4 shows a typical
EDS spectrum obtained from the IMC and its chemical composition is determined to be Sn-
39.5 wt. % Cu, which is thought to be Cu6Sn5. This Cu6Sn5 is a well-known major product
that results from Cu/Sn interactions after solid-state aging [102].
Figure 4-4 EDS spectrum of the Cu6Sn5 IMC.
57

4.1.2 Effect of pre-annealing on Sn whisker formation
Figure 4-5 Top-view SEM images of the samples aged at 60°C for 20 hr and after pre-
annealing at (a)-(c) 100°C (d)-(f) 400°C with Cu(Ru), pure Cu seed layers and without seed
layer, respectively.
Figure 4-5 shows the surface morphology of the samples after pre-annealing and
aging. Figures 4-5 (a)-(c) present the surface morphology of the samples pre-annealed at
100°C for 20 min prior to the Sn layer deposition and followed by aging afterwards.
Meanwhile, the samples pre-annealed at 400°C for 10 min are shown in Figs. 4-5 (d)-(f). As
shown in Fig. 4-5 (a), the sample with the Cu(Ru) seed layer seems to have a greatest number
58

of Sn whiskers among others. Even after pre-annealing at 400°C, a significant amount of
whisker is still observed as revealed in Fig. 4-5(d). Surprisingly, Sn whiskers are absent in the
sample without seed layer after pre-annealing at 400°C as can be seen in Fig. 4-5(f).
Figure 4-6 Whisker density as a function of pre-annealing temperature.
The whisker density is quantified and plotted as a function of pre-annealing
temperature, as shown in Fig. 4-6. The whisker density is an average value obtained by
measuring the number of whiskers observed from 10 different SEM image areas. For the
samples with the Cu(Ru) seed layer, pre-annealing at 100°C and 150°C appears not to
decrease the Sn whisker formation. Further, pre-annealing at 200°C and 400°C slightly
decrease the Sn whisker growth. For the samples with the pure Cu seed layer, pre-annealing
59

at temperatures up to 200°C does not inhibit or decrease the growth of Sn whiskers.
However, after annealing at 400°C, the whisker density is found to be about a half of the
whisker density found after pre-annealing up to 200°C. Compared with the samples with pure
Cu seed layer, the samples with Cu(Ru) seed layer exhibit much higher whisker density. One
can consider that the addition of 1.1 at. % Ru causes the Cu film to undergo the grain
refinement. It is believed that during the film deposition, the presence of minor concentration
of Ru might hinder the grain growth of Cu [103]. As a result, Cu(Ru) film grows with smaller
grain sizes and higher amount of grain boundaries compared with those of the pure Cu film.
These grain boundaries act as the atomic diffusion paths. Consequently, a faster Cu diffusion
occurs through grain boundaries, leading to a thicker IMC. A thicker IMC formed at the
interface induces a greater compressive stress for the whisker to grow. In other study [104],
when the crystalline quality of the Cu seed layer was inferior, a large lattice mismatch
developed at the interface between the electroplated Sn ﬁlm and the Cu seed layer resulted in
many lattice defects and stress might exist at the interface between the electroplated Sn ﬁlm
and the Cu seed layer. This lattice mismatch was considered to be related to the high-
diffusion velocity of Cu from the Cu seed layer.
For the samples without the seed layer, after pre-annealing at 100°C, a small amount
of whiskers is observed. Interestingly, the Sn whisker formation seems to be effectively
prevented upon pre-annealing at 150°C. In addition to the release of the residual stress, the
grain size of the Cu foil is anticipated to be much bigger than both Cu(Ru) and pure Cu seed
layers due to its bulk form and grain growth after pre-annealing. Therefore, with the much
bigger grain size and less amount of grain boundaries, the Cu diffusion towards the Sn layer
should be minimized. In this respect, no or thin thickness of IMC layer is expected to form at
the interface between Cu and Sn layers so that the internal stress may not be sufficient to
form Sn whiskers.
60

4.2 Sn whisker formation in Cu-Sn thin film couples
SEM micrograph in Fig. 4-7 shows various shapes of Sn whiskers detected in the
sample without underlayer after aging for 33 days at room temperatures. The typical whisker
observed on the Sn layer is the striated whisker, with some needle-like whiskers that
originated from the small Sn grains. It is believed that the shape of Sn whisker heavily
depends on the size of the grain, from which the whisker starts to grow.
Figure 4-7 SEM micrograph showing various common shapes of Sn whiskers observed on Sn
layer after aging for 33 days at room temperature.
In addition to those common appearances of Sn whisker in Fig. 4-7, Fig. 4-8 shows a
montage of images illustrating some of the different whisker conformations observed by
SEM. Figures 4-8 (a) and (b) show examples of long whiskers. A long whisker in Fig. 4-8(a)
is found to be ~80 µm in length. It thus suggests that the whisker length does not depend on
the thickness of Sn layer. Sn whisker in Fig. 4-8(c) kinks close to its nucleation site then
61

continues to grow in a different direction. A kinked whisker in Fig. 4-8(d) can be bent during
the growth so that the growth direction after kinking would not be that straight. Figure 4-8(e)
shows a relatively rare example of a bent whisker. Some of the whiskers show grooved or
striated surface structures, as illustrated in Fig. 4-8(f). These grooves run along the length of a
whisker whether it is straight or kinked. The striations are aligned with the surface porosity.
The grooves or striations in the whiskers are related to the surface morphology of the Sn and
appear to result from voids on the Sn surface having either intraganular or intergranular
porosity [105].
Figure 4-8 Various whisker morphologies found in the samples without underlayers.
62

Figure 4-9 Preparation of TEM sample in a specific location by FIB: (a) Sn whiskers grown
on the surface of Sn layer (b) deposition of a thin Pt protective layer (c) etching of two
rectangular trenches on the surface (d) a thin piece of layered sample is left standing between
the two holes.
To provide a better understanding on the cross-sectional TEM sample, the sample
preparation procedure is given briefly. One of short whiskers in Fig. 4-9(a) was chosen for
TEM study. Before the FIB etching, a protective layer of a thin stripe Pt was deposited onto
the surface to protect the thin slice below the stripe from ion beam etching as shown in Fig.
4-9(b). In Fig. 4-9(c) FIB etching was started by making two rectangular trenches on both
sides of Pt-deposited stripe. FIB etching was then continued until a thin piece of layered
63

sample was left standing between the two trenches. The thin slice in Fig. 4-9(d) was prepared
for TEM and it could be cut off and taken out, which was then ready for TEM observation.
Figure 4-10 Cross-sectional TEM images of a Sn whisker: (a) in low magnification revealing
locations A and B for further analyses at high magnification in (b) and (c). (b) and (c)
HRTEMs and SADPs taken from different locations, showing the growth direction along
[220].
64

A cross-sectional TEM image of a Sn whisker is shown in Fig. 4-10(a). It is likely that
during FIB etching and thinning, the thin slice was damaged and defects were introduced.
Fortunately, the root of whisker with some surrounding Sn grains is clearly seen. However,
the presence of IMC is hard to determine probably due to the sample damage by FIB.
Although this TEM analysis is not able to represent all of observed whiskers, the HRTEMs
and SADPs in Figs. 4-10 (a) and (b) reveal that this particular Sn whisker is a single crystal
with the growth direction of [220]. Both SADPs were taken with the zone axis of [222]. This
finding agrees with the well-known concept of the Sn whisker, which found to be a single
crystal [35].
4.3 Thin film metallic glass characterizations
Figure 4-11 XRD patterns of the TFMGs showing typical broad humps around 30°- 45° of
2θ.
65

Figure 4-11 shows the typical broad humps around 2θ of 30°-45° in XRD spectra of
Zr51.7Cu32.3Al10Ni7 and Zr46Ti26Ni28 TFMGs. No obvious peaks corresponding to the
crystalline structure indicate large amount of amorphous phase in both samples. Figure 4-12
shows high-resolution TEM (HRTEM) images and selected area electron diﬀraction patterns
(SADPs) of Zr51.7Cu32.3Al10Ni7 and Zr46Ti26Ni28 TFMGs. The SADPs show only a halo ring,
revealing the amorphous structure of both Zr-based TFMGs. Moreover, the absence of
nanocrystalline phase in the diffraction pattern confirms that these TFMGs consist of only a
glassy structure in amorphous phase. The HRTEM images further show a highly disordered
structure without any detectable nanocrystals or ordered clusters, which again confirm their
amorphous nature.
Figure 4-12 HRTEM of (a) Zr46Ti26Ni28 (b) Zr51.7Cu32.3Al9Ni7 TFMGs. Insets in (a) and (b)
are the corresponding SADPs.
66

4.3.2 Thermal analysis
Figure 4-13 DSC curves of the Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs.
DSC measurements in Fig. 4-13 demonstrate that the TFMGs undergo a glass
transition followed by an exothermic peak, indicating the transformation from a supercooled
liquid state to a crystalline phase. Tg of Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs are
determined to be 466°C and 465°C, respectively. Meanwhile, their Tx are 502°C and 510°C,
respectively. A quaternary system of Zr51.7Cu32.3Al9Ni7 seems to have a better glass forming
ability, which leads to a wider range of ΔT (ΔT = Tx - Tg), compared with that of ternary
system of Zr46Ti26Ni28 TFMG. Based on these characteristics, the TFMGs are confirmed to be
amorphous. In addition, it is suggested that the amorphous structure should remain within
TFMGs when the working temperature is well below Tx in the present study. Since Sn is
67

widely used in the electronic packaging, the working temperatures would not be much higher
than its melting temperature of 232°C. It is advantageous that the high-Tx TFMGs are
considered to be reliable as the amorphous underlayer to prevent the interactions between Cu
and Sn.
4.3.3 Surface roughness analysis
Figure 4-14 Plane-view SEM images of Zr46Ti26Ni28 TFMG deposited onto a Cu layer in low
(a) and high (b) magnifications. (c) 2-D (d) 3-D AFM images.
The smooth surface of TFMG deposited onto the Cu layer is revealed in Fig. 4-14.
The featureless SEM images in Figs. 4-14 (a) and (b) have no visible contrast as a result of
68

the smooth surface due to the amorphous nature of TFMG. The smooth surface is also clearly
demonstrated in AFM topographic 2-D and 3-D images in Figs. 4-14 (c) and (d). The surface
property of the underlayer is also believed to influence the Sn layer morphology.
4.3.4 Electrical resistivity measurement
The electrical resistivity of 100-nm-thick Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs
are measured to be 1.84 x 10-4
Ω•cm and 2.31 x 10-4
Ω•cm, respectively. Their electrical
resistivity are comparable with that of Ni thin film with the thickness of 100 nm reported in
other study [106]. It can be considered that with such a thin underlayer, the electrical
performance of electronic packaging would not be downgraded.
4.3.5 Adhesion evaluation
Based on the adhesion evaluation by cross-hatch cut method, the adhesions of the
Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs to the Cu layer are categorized as Class 5B, which
indicates very good adhesion of that interface as shown in Table 4-1. However, due to the
softness of the Sn layer, the adhesion evaluation of the Sn layer to the TFMG underlayer was
not able to be carried out.
69

Table 4-1 Classification of adhesion test results
70

4.4 Thin film metallic glass as an underlayer for Sn whisker mitigation
4.4.1 Thermal stability of Zr46Ti26Ni28 TFMG underlayer aged at room temperature
underlayer in low (a), high (b) magnifications in plane-view and (c) 50°-tilted view; with the
Zr46Ti26Ni28 TFMG underlayer in low (d), high (e) magnifications in plane-view and (f) 50°-
tilted view.
Figure 4-15 shows the surface morphology of as-deposited Sn layer in samples with
and without an underlayer. The Sn layer deposited on TFMG is smoother and denser with
fewer voids than that of the sample without the underlayer. The good surface condition for
the Sn layer is presumably due to the presence of smooth Zr46Ti26Ni28 TFMG. The surface
property of the underlayer is thus believed to influence the Sn layer morphology. In this
study, the amorphous and smooth characters of underlayer appear beneficial for forming the
relatively flat and dense Sn overlay. No observable Sn whiskers are seen in as-deposited
condition for both samples. While further experimental confirmations are needed, the absence
of whiskers in the as-deposited samples suggests that the residual stress generated during the
71

Sn layer deposition may not be significant for the whisker growth. In other study [107], Sn
whiskers were found right after Sn layer deposition were reported to be mainly caused by the
compressive stress developed in the Sn layer.
Figure 4-16 SEM micrographs of samples with and without the Zr46Ti26Ni28 underlayer after
aging at room temperature for 5, 12 and 33 days. Whiskers appear as small bright spots in the
images.
72

Figure 4-16 shows the surface morphology of the Sn layer in the sample with and
without the TFMG underlayer after aging at room temperature. In the sample with the TFMG
underlayer, no whisker can be observed even after aging for 33 days. On the other hand, in
the sample without Zr46Ti26Ni28 TFMG underlayer, numerous whiskers are evident as they
appear as small bright spots in the SEM images. In addition, the whiskers become more
populated with aging time. To quantify the number of whiskers, the whisker density
measured in SEM images taken from the sample without TFMG underlayer after aging is
plotted in the Fig. 4-14 and the details are listed in Table 4-2.
Table 4-2 Whisker density observed in samples with and without TFMG underlayer after
aging for various days at room temperature
Aging time (day)
Whisker density (number/mm2
)*
No underlayer With TFMG underlayer
As-deposited n/o n/o
1 1474 n/o
5 9738 n/o
7 11420 n/o
12 18958 n/o
33 28027 n/o
*: n/o indicates no whisker observed.
73

Figure 4-17 Whisker density as a function of aging time for the sample without TFMG
underlayer.
According to Fig. 4-17 and Table 4-2, the whisker density increases with aging time.
The increase of whisker density is more pronounced at the early stage of aging (within the
first 12 days), and then slowly afterwards. This may be due to a decrease of compressive
stress in the Sn layer after lengthy periods of aging.
74

Figure 4-18 (a) and (b) Cross-sectional TEM images of the sample without TFMG underlayer
after aging at room temperature for 33 days. White dashed lines in (a) indicate the
approximate location of IMC region.
75

Figures 4-18 (a) and (b) show typical cross-sectional TEM micrographs of the sample
without underlayer after aging at room temperature for 33 days. IMC is found present, which
is confirmed to be Cu6Sn5. As a result of Cu/Sn interactions and IMC formation, the Cu/Sn
interface is not well defined. Instead, the interfacial region consists of IMC and unreacted or
displaced Sn. The Kirkendall void is also observed in the region. It is found more obvious in
Fig. 4-18(b). The unbalanced interdiffusion of Cu and Sn results in a Kirkendall effect [35]
with a vacancy-rich zone in the vicinity of the Cu/Sn interface.
Figure 4-19 Schematic of Sn whisker formation during aging: (a) as-deposited condition, (b)
early stage of IMC formation, (c) IMC thickening and Kirkendall void formation and (d) Sn
whisker growth.
Figure 4-19 is presented to further schematically illustrate a mechanism of the
formation of IMC and Kirkendall void. The sample is found free from intermetallic reaction
76

in as-deposited condition, as shown in Fig. 4-19(a). Cu diffuses readily into the Sn layer at
the ambient conditions over the time. Cu6Sn5 formation in Fig. 4-19(b) requires that six Cu
atoms diffuse into space occupied by eleven Sn atoms and combine with five of the eleven Sn
atoms. This would be a 45% reaction of the 11 Sn atoms as shown in equation (8) [16]. This
intermetallic reaction may continue to higher degrees of completion, as indicated in Fig. 4-
19(c). Equation (9) shows a 90% reaction [16]. The molar volume of Cu6Sn5 plus the
displaced Sn atoms is much larger than the molar volume occupied by the original eleven Sn
atoms. It is the molar volume increase for the combined Cu and Sn atoms in that region
initially occupied only by the Sn atoms, which yields a compressive stress state within the
entire intermetallic region.
11 Sn atoms → Cu6Sn5 + 6Sn
(molar vol. = 176) → (molar vol. = 118) + (molar vol. = 96) ............................ (8)
11 Sn atoms → 2Cu6Sn5 + 1Sn
(molar vol. = 176) → (molar vol. = 236) + (molar vol. = 16) ............................ (9)
The molar volume increases for (8) and (9) are 21% and 43% respectively. However,
the intermetallic region expansion will be limited due to the restraining influences of the
overlying Sn layer and the underlying Cu layer. The underlying Cu layer is particularly
restraining due to its shrinkage zone located immediately adjacent to the Cu/Sn interface. It is
a direct result of a Kirkendall effect.
The unbalanced interdiffusion of Sn and Cu results in the well-known metallurgical
phenomena, called the Kirkendall effect. The necessary Kirkendall vacancies are located
within the Cu layer at the Sn/Cu interface as shown in Fig. 4-19(c). This vacancy
concentration causes shrinkage within the Cu lattice structure and establishes a tensile stress
state due to the restraining influences of the overlying intermetallic region (an active
expansion zone) and the underlying non-vacancy-rich Cu (an inactive zone).
77

Figure 4-20 Schematic of four-zone structure proposed by Galyon et al. [16].
Galyon et al. [16] also proposed so-called four-zone structure for a Sn/Cu couple after
IMC formation, as shown in Fig. 4-20. In the present study, this four-zone structure is found
to be similar to the one proposed in their study. Hence, it is necessary to understand this
structure. According to their review article [16], a Sn layer on Cu layer will establish a four-
zone structure during aging. Zone-1 is the Sn layer immediately above, and adjacent to, the
intermetallic region. Zone-1 may, or may not, have its own internal stress generating
mechanisms. Zone-2 is an intermetallic region consisting of intermetallic (Cu6Sn5) and
displaced/unreacted Sn. The zone-2 region is always active and expansive due to
intermetallic formation. Zone-3 is the Kirkendall vacancy-rich zone located within the Cu
layer and it is always active and contractive due to the shrinkage effects of vacancy
formation. Zone-4 is the Cu layer and it is always inactive (i.e., no internal stress generating
mechanisms). The four-zone structure is basic to understanding the time-dependent stress
development for Sn/Cu couples.
78

Figure 4-21 Four-zone structure estimated for Sn/Cu. Red dashed-lines are for ease of
distinguishing different zones.
The fundamental understanding on the stress state development can be also explained
from the four-zone structure. Zone-1 may have internal stress-generating mechanisms such as
during the Cu layer deposition. Temperature fluctuations can induce stresses between zones
due to expansion coefficient differences. In addition, zone-1 will react to the contraction
and/or expansion of the underlying layer and develop internal stresses accordingly. If
79

underlying layer expands, then zone-1 will act as a constraining layer and develop internal
tensile stresses. If underlying layer contracts, then zone-1 will act as a constraining layer and
develop internal tensile stresses. Zone-2 is always an active stress-generating zone due to
intermetallic formation. The Cu6Sn5 intermetallic is always “horned” in appearance. Both the
Sn and intermetallic located within zone-2 are compressively stressed due to the expansive
action of the intermetallic formation.
Zone-3 is the Kirkendall vacancy/void zone. Vacancy formation and coalescence (i.e.,
voids) will impose shrinkage resulting in a residual tensile stress due to constraints from the
overlying intermetallic region (zone-2) and the underlying, non-vacancy-rich Cu layer (zone-
4). Zone-4 is the underlying Cu layer and it is an inactive zone that follows the expansion or
contraction of the overlying layer (zone-3). To visualize the zone structure in this study,
Figure 4-21 shows a TEM image explicitly revealing these four zones.
Figure 4-22 SEM micrographs showing the presence of IMC formed in the sample without
the TFMG underlayer in (a) plane and (b) 50°-tilted views, after aging for 33 days and Sn
layer removal by chemical etching.
Figure 4-22 shows the morphology of IMC formed in the sample without underlayer
after aging for 33 days and Sn removal. The unreacted Sn is chemically etched away to
80

expose the IMC. The IMC is found to be Cu6Sn5 as determined by EDS results. This Cu6Sn5
is a well-known major product that results from Cu/Sn interactions after solid-state aging
[102]. This result suggests that Cu/Sn interactions can take place even at room temperature to
form intermetallic compounds. The IMC layer has a relatively lower density than Cu and this
density change causes the volume expansion within the IMC layer. Such a volume expansion
leads to the formation of compressive stress toward the Sn layer in the vertical direction of
the Cu and Sn interface [11]. The IMC may pin the Sn grain boundaries, which inhibits the
relaxation of compressive stress in the Sn layer and promotes the formation of Sn whiskers.
Figure 4-23 Cross-sectional TEM image of the sample with Zr46Ti26Ni28 TFMG underlayer
after aging for 33 days.
81

Figure 4-23 is a typical cross-sectional TEM micrograph of the sample with a 100
nm-thick ternary system of TFMG underlayer after aging at room temperature for 33 days,
which shows the region of the Sn, TFMG underlayer, Cu, and Ti adhesive layer on Si
substrate. The thickness of Sn layer is in the range of 390 ± 28 nm, which is typical when
deposited by electron beam evaporation [108]. The adhesion of the evaporated Sn layer to the
Zr46Ti26Ni28 thin film metallic glass seems relatively good and the Zr46Ti26Ni28 thin film
metallic glass is well adhered to the sputtered Cu layer. Noticeably, TFMG appears to serve
as a reliable underlayer to block the Cu/Sn interaction and thus prevent the formation of IMC.
Consequently, no IMC is observed at the interface and the layered structure of Sn/TFMG/Cu
is still clearly seen. Figure 4-23 also reveals no crystallinity or grain structure in the TFMG
layer due to its amorphous nature, which agrees with the XRD pattern in Fig. 4-11 and a
diffuse halo diffraction pattern in the inset of Fig. 4-12(a). In contrast, the typical sputtered
columnar structure is observed in the crystalline Cu layer. The EDS elemental line-scans in
Fig. 4-24 further verify no apparent Cu/Sn interaction. The presence of Zr, Ti and Ni in the
ternary system of TFMG underlayer is also confirmed in the EDS result in Fig. 4-24.
82

scans are performed.
4.4.2 Thermal stability of Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs underlayer aged at
various temperatures
The thermal stability of TFMGs underlayers in the Cu-Sn thin film couples has been
studied. In this work, thin film metallic glasses with and without the presence of Cu
effectively block the Cu/Sn interactions after being subjected to various temperatures of
aging. Three kinds of layered samples, with and without the underlayers, were prepared. For
the sample containing the underlayers, Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs with the
thickness of 100 nm were deposited prior to the Sn deposition.
83

underlayer in low (a), high (b) magnifications in plane-view and (c) 50°-tilted view; with the
Zr46Ti26Ni28 TFMG underlayer in low (d), high (e) magnifications in plane-view and (f) 50°-
tilted view; with the Zr51.7Cu32.3Al9Ni7 TFMG underlayer in low (g), high (h) magnifications
in plane-view and (i) 50°-tilted view.
In addition to Zr46Ti26Ni28 TFMG, the introduction of Zr51.7Cu32.3Al9Ni7 TFMG also
exhibits a better surface property of Sn overlay as shown in Fig. 4-25. This again suggests
that the amorphous underlayer provides a relatively flat and dense surface for Sn to grow.
Moreover, no observable Sn whiskers are seen in as-deposited condition for all samples. The
surface morphology of all samples after aging at room temperature, 40°C and 60°C are
84

presented in Figs. 4-26, 4-27 and 4-28, respectively. The samples with TFMGs underlayers
are found free from Sn whisker formation after aging at temperature up to 60°C. However, in
the sample without TFMG underlayers, numerous whiskers are found as they appear as small
bright spots in the SEM images.
aging at room temperature for various lengths of time.
85

aging at 40°C for various lengths of time.
86

aging at 60°C for various lengths of time.
87

Figure 4-29 Whisker density as a function of aging time at various temperatures for the
samples without TFMG underlayer.
To visualize whisker formation at various aging temperatures, the whisker density
measured in SEM images taken from the sample without TFMG underlayer is plotted as
functions of aging temperatures and times in Fig. 4-29. Increasing the aging temperature
accelerates the Cu/Sn interdiffusion and thus the formation of the IMC, yielding extensive Sn
whisker formation. The stress caused by the IMC formation is then expected to be greater. In
other study [109], the amount of Sn whiskers decreased with the increase in the aging
temperature because of the thicker oxide layer. The high aging temperature also led to an
88

increase in the thickness of the oxide layer on the Sn layer surface. However, in this study,
the oxide layer does not likely play a role in Sn whisker formation.
Figure 4-30 XRD patterns of the Sn layers in the samples without the underlayer aged at
different temperatures.
Fig. 4-30 presents the XRD patterns of Sn layer in the sample without underlayer after
aging at various temperatures. The patterns are indexed according to JCPDS 4-0673 (Sn) and
45-1488 (Cu6Sn5). It is shown that in as-deposited condition only β-phase of Sn is detected.
Moreover, the η-phase of Cu6Sn5 IMC is then detected after aging at elevated temperatures.
After Sn stripping, only Cu6Sn5 IMC remains without any Sn β-phase, indicating Cu6Sn5
IMC formed at the interface.
89

Figure 4-31 XRD patterns of the Sn layers in the samples with the Zr46Ti26Ni28 TFMG
underlayer aged at different temperatures.
90

Figure 4-32 XRD patterns of the Sn layers in the samples with the Zr51.7Cu32.3Al9Ni7 TFMG
underlayer aged at different temperatures.
Figures 4-31 and 4-32 show the XRD patterns of the Sn layers in the samples with
Zr46Ti26Ni28 and Zr51.7Cu32.3Al9Ni7 TFMGs underlayers, respectively. There is no η-phase of
Cu6Sn5 detected either in as-deposited condition or after aging, suggesting no or insignificant
Cu/Sn interaction. It is worth noting that even with the presence of Cu in the quaternary
system of TFMG underlayer, the absence of IMC implies no apparent diffusion of Cu from
TFMG towards the Sn layer. In addition, very minor β(220), β(211) and β(400) peaks are
observed in the samples with TFMG underlayers, revealing the preferred orientation in the Sn
91

layer grown on the amorphous underlayer. This is also believed to cause the relatively
smooth and less of voids of Sn layer deposited on the TFMG underlayers.
Figure 4-33 Cross-sectional TEM image of the sample with Zr51.7Cu32.3Al9Ni7 TFMG
underlayer after aging at 60°C for 6 days.
The reliability of Zr51.7Cu32.3Al9Ni7 TFMG underlayer after aging at 60°C is presented
in Fig. 4-33. Zr51.7Cu32.3Al9Ni7 TFMG plays an important role to serve as a promising
underlayer even with the presence of Cu, to effectively prevent the interactions between Cu
and Sn, thus hindering the formation of IMC. Therefore, no IMC is visible in the region.
Figure 4-33 also shows no grain structures in the TFMG layer due to its amorphous nature,
which agrees with the XRD pattern in Fig. 4-11 and a diffuse halo diffraction pattern in the
inset of Fig. 4-12(b). The average grain sizes of the Cu and Sn layer are measured to be 169
92

nm and 124 nm, respectively. No obvious Cu/Sn interaction is proven in EDS elemental line-
scans in Fig. 4-34. It also confirms the presence of Zr, Cu, Al and Ni. The Cu signal at the
interface comes from the Zr51.7Cu32.3Al9Ni7 TFMG.
scans are performed.
93

4.4.3 Thermal cycling stability of Zr46Ti26Ni28 TFMG underlayer
Thermal cycling is one of the harsh environments to which many electronic devices
are inevitably submitted. Thermal cycling is thus becomes one of important reliability test in
electronic packaging. In the present study, thermal cycling was also carried out in the
temperature range of -35°C to +85°C.
Figure 4-35 SEM micrographs of samples with and without the Zr46Ti26Ni28 TFMG
underlayers after thermal cycling for 500 cycles.
While the significant amount of Sn whiskers are observed in the sample without
underlayer, the samples with the Zr46Ti26Ni28 TFMG underlayers are free from the Sn
whisker formation, even when the thickness of the underlayer is reduced to 25 nm as
illustrated in Fig. 4-35.
94

TFMG underlayer after thermal cycling for 500 cycles.
Figure 4-36 shows the XRD patterns of the Sn layers in the samples with and without
ternary system of TFMG underlayer after 500 thermal cycling (TC) up to 85°C. XRD patters
reveal no detectable Cu6Sn5 IMC formation in the samples with the underlayers having
thickness in a range of 25 to 100 nm. On the other hand, the η-phase of Cu6Sn5 IMC is
detected in the sample with the absence of the underlayer. In addition to the IMC formation,
the compressive stress as a driving force for Sn whisker growth might have been also
generated due to the thermal mismatch between the Sn and Cu layers during thermal cycling
[110-112]. The thermal mismatch is mainly caused by the difference in the coefficient of
thermal expansion (CTE) of both Cu and Sn. In this study, it is ambiguous to determine
95

whether the compressive stress generated during thermal cycling originates only from the
IMC formation or a combination with the CTE mismatch. However, when considering the
compressive stress originates from the CTE mismatch, the thermal mechanical strain in the
Sn layer can be expressed by the following equation [113]:
𝜀 𝑥 = ∆𝐶𝑇𝐸×𝐸 𝑢×𝑑 𝑢
�𝐸 𝑆𝑛×𝑑 𝑆𝑛+𝐸 𝑢×𝑑 𝑢�
.............................................................. (10)
where 𝜀 𝑥 is the strain along x-direction in a plane, ΔCTE is the CTE mismatch between Sn
and underlayer, 𝐸𝑆𝑛and 𝐸 𝑢 are Young’s modulus, 𝑑 𝑆𝑛 and 𝑑 𝑢 are thicknesses of Sn and
underlayer, respectively. The stress 𝜎𝑥 can be determined using Hooke’s law:
𝜎𝑥 = 𝐸 × 𝜀 𝑥 ................................................................. (11)
Although the Young’s modulus and the CTE of TFMGs used in this study have not been
determined, the above two equations imply that when the thickness of the underlayer is
reduced, the compressive stress generated by CTE mismatch should be minimized. In this
respect, introduction of thin thickness of TFMG underlayer appears beneficial to decrease the
thermal stress and thus mitigate Sn whisker growth.
4.4.4 Thermal reflow stability of Zr46Ti26Ni28 TFMG underlayer
While TFMG underlayers are found stable after heat treatments up to 85°C, it is
interesting to study the thermal stability of Sn/TFMG/Cu stacking system at a higher
temperature. Thermal reflow is one of common heat treatments conducted in electronic
packaging technology. The thermal reflow stability of Zr46Ti26Ni28 TFMG underlayer at
260°C is demonstrated in Fig. 4-37. In as-deposited condition, the Sn layer possess a
preferred orientation of β(200). However, after thermal reflow, the preferred orientation no
longer exists. Since the melting temperature of Sn is 232°C, the Sn layer melts and
recrystallizes when the sample is subjected to thermal reflow at 260°C. Most importantly, the
Sn layer is found free from any IMC formation. In principle, as long as the working
96

temperature, in this case reflow temperature, is well below its Tx (502˚C), the amorphous
structure of TFMG should remain. Therefore, there is no observable interaction between the
TFMG underlayer and Sn overlay.
TFMG underlayer after thermal reflow at 260°C.
97

Chapter 5 Conclusions & Future Works
5.1 Conclusions
The very early work in this study demonstrates the effect of the grain refinement due
to the addition of minor concentration of Ru into the Cu layer as a seed layer on the Cu
diffusion behavior towards the Sn layer. It is found that the samples with Cu(Ru) seed layer
exhibit a greatest number of whisker among other samples even after pre-annealing prior to
the Sn layer deposition, which is expected to release the residual stress and yield the grain
growth. In contrast, the annealed Cu foil seems to be effective to prevent the IMC resulted
from interdiffusion between Cu and Sn, which is believed to be a driving force of Sn whisker
formation. Therefore, this study suggests the grain size and the presence of the grain
boundaries play a role in the Cu diffusion, although further investigations are needed,
particularly to reveal the exact grain size of the samples used in this work. Based on this
result, introduction of amorphous diffusion barrier is then considered to be the best mitigation
strategy to prevent the interdiffusion between Cu and Sn.
Most importantly, the present study demonstrates the thermal stability of Zr46Ti26Ni28
and Zr51.7Cu32.3Al9Ni7 thin film metallic glasses to prevent the formation of Cu6Sn5
intermetallic compound as a result of Cu/Sn interaction. The TFMGs in as-deposited
condition have been proven to have amorphous structures as indicated by XRD and DSC
measurements. TFMG underlayers are found to exhibit relatively flat and dense Sn overlay.
A good quality of evaporated Sn layer can be potentially used as a seed layer for an
electroplated layer in flip chip or ball grid array processes.
There is no Sn whisker observed in the sample with Zr46Ti26Ni28 TFMG underlayers
in as-deposited condition and even after aging for 33 days at room temperature. In contrast,
Sn whiskers are found in the sample without the underlayer and the whisker density increases
98

with increasing aging time. In addition, TFMG underlayers, with or without the presence of
Cu, have been demonstrated to prevent the formation of intermetallic compound. Amorphous
TFMG underlayers with thicknesses of 25 nm to 100 nm are shown to block the
interdiffusion of Cu and Sn upon various heat treatments. In particular, the internal stress
developed during thermal cycling is thought to be minimized when the thin TFMG
underlayer is used. Based on these findings, TFMG is considered to be a promising diffusion
barrier for Sn whisker mitigation.
5.2 Future works
There are three works have to be completed in the near future in order to further
explore the advantage of thin film metallic glass as the useful material in electronic
packaging. One is to study the wettability behavior of thin film metallic glass. The wettability
behavior can be determined either by measuring the contact angle of particular Sn molten
when the molten is dropped on the substrate (sessile drop) or by measuring the wetting force
using wetting balance method. Wetting balance provides information of both the speed and
extent of wetting during the entire dipping period, providing much more useful information.
In principle, as described in Chapter 2, the wettability of amorphous materials is relatively
better than their crystalline counterparts. However, the presence of Zr and Ti in the Zr-based
TFMG proposed in the present work is thought to deteriorate the wettability owing to their
ability to form stable oxides. Therefore, another TFMG system such as Ni-based will be also
introduced for comparison.
Another interesting work is the measurement of the residual stress within the Sn layer
after various heat treatments when TFMG is introduced as an underlayer. It can be done by
laser curvature method, which is based on the characteristic curvature radius of the specimen
before and after heat treatments. In addition to the laser curvature method, the residual stress
99

can be also measured by measuring the stress gradient obtained by cos2
αsin2
Ψ XRD method.
The results from both methods will be compared to have a better understanding. The last but
not least, the activation energy of interdiffusion reaction in this particular Cu-Sn thin film
couples will be determined in order to provide a fundamental understanding on the diffusion
behavior of Cu into the Sn layer.
100

References
[1] K. Tu, C. Chen, A.T. Wu, Journal of Materials Science: Materials in Electronics 18/1
(2007) 269.
[2] B.Z. Lee, D.N. Lee, Acta Materialia 46/10 (1998) 3701.
[3] M. Sobiech, U. Welzel, E. Mittemeijer, W. Hugel, A. Seekamp, Applied Physics
Letters 93/1 (2008) 011906.
[4] A. Baated, K.S. Kim, K. Suganuma, Journal of Materials Science: Materials in
Electronics 22/11 (2011) 1685.
[5] K. Kim, C. Yu, S. Han, K. Yang, J. Kim, Microelectronics Reliability 48/1 (2008)
111.
[6] S.B. Li, G.P. Bei, H.X. Zhai, Z.L. Zhang, Y. Zhou, C.W. Li, Journal of Materials
Research 22/11 (2007) 3226.
[7] K. Tu, Acta Metallurgica 21/4 (1973) 347.
[8] J. Brusse, G. Ewell, J. Siplon, Carts Europe 16 (2002).
[9] M.N. Chen, S.J. Ding, Q.Q. Sun, D.W. Zhang, L.K. Wang, Journal of Electronic
Materials 37/6 (2008) 894.
[10] B. Horváth, B. Illés, T. Shinohara, G. Harsányi, Thin Solid Films 520/1 (2011) 384.
[11] B. Illés, B. Horváth, G. Harsányi, Surface and Coatings Technology 205/7 (2010)
2262.
[12] T. Liu, D. Ding, K.P. Galuschki, Y. Hu, Y. Gong, M. Shen, IEEE Transactions on
Components, Packaging and Manufacturing Technology 2/5 (2012) 731.
[13] L. Panashchenko, M. Osterman, Electronic Components and Technology Conference,
2009. ECTC 2009. 59th, 2009, p. 1037.
[14] Y.W. Yen, W.K. Liou, C.C. Jao, Components, Packaging and Manufacturing
Technology, IEEE Transactions on 1/6 (2011) 951.
101

[15] W. Yiqing, D. Dongyan, L. Ting, K.P. Galuschki, H. Yu, A. Gong, S. Ming, S.
Hongqi, W. Xianfeng, S. Jiangyan, L. Ming, M. Dali, Electronic Packaging
Technology & High Density Packaging (ICEPT-HDP), 2010 11th International
Conference on, 2010, p. 980.
[16] G.T. Galyon, L. Palmer, Electronics Packaging Manufacturing, IEEE Transactions on
28/1 (2005) 17.
[17] M.A. Nicolet, Applied Surface Science 91/1 (1995) 269.
[18] J. Reid, E. Kolawa, C. Garland, M.A. Nicolet, F. Cardone, D. Gupta, R. Ruiz, Journal
of Applied Physics 79/2 (1996) 1109.
[19] X. Sun, J.S. Reid, E. Kolawa, M.A. Nicolet, Journal of Applied Physics 81/2 (1997)
656.
[20] X. Sun, J.S. Reid, E. Kolawa, M.A. Nicolet, R.P. Ruiz, Journal of Applied Physics
81/2 (1997) 664.
[21] J.P. Chu, J.C. Huang, J.S.C. Jang, Y.C. Wang, P.K. Liaw, JOM 62/4 (2010) 19.
[22] J.P. Chu, J.S.C. Jang, J.C. Huang, H.S. Chou, Y. Yang, J.C. Ye, Y.C. Wang, J.W.
Lee, F.X. Liu, P.K. Liaw, Y.C. Chen, C.M. Lee, C.L. Li, C. Rullyani, Thin Solid
Films 520/16 (2012) 5097.
[23] B.D. Dunn, Metals and Materials 34 (1975) 34.
[24] B.D. Dunn, Circuit World 2/4 (1976) 32.
[25] K.N. Tu, R.D. Thompson, Acta Metallurgica 30/5 (1982) 947.
[26] J.P. Winterstein, J.B. LeBret, M.G. Norton, Journal of Materials Research 19/3 (2004)
689.
[27] C.L. Rodekohr, G.T. Flowers, M.J. Bozack, R. Jackson, R. Martens, Z. Zhao, E.R.
Crandall, V. Starman, T. Bitner, J. Street, 2011, p. 205.
[28] F. Pei, N. Jadhav, E. Chason, Applied Physics Letters 100/22 (2012) 221902.
102

[29] L.A. Pinol, J. Melngailis, H.K. Charles, Jr., D.M. Lee, R. Deacon, G. Coles, G.
Clatterbaugh, Components, Packaging and Manufacturing Technology, IEEE
Transactions on 1/12 (2011) 2028.
[30] http://nepp.nasa.gov/whisker/.
[31] H. Livingston, BAE SYSTEMS Information and Electronic Warfare System
(2003).
[32] W. Johler, IEEE Trans. on Comp. and Packaging. Tech. 27 (2004) 30.
[33] F. Yang, J.C.M. Li, Lead-Free Electronic Solders, Springer US, 2007, p. 191.
[34] J. Cheng, Ph. D, Department of Mechanical Engineering, University of Rochester,
Rochester, New York, 2011.
[35] G.T. Galyon, Electronics Packaging Manufacturing, IEEE Transactions on 28/1
(2005) 94.
[36] J. Osenbach, J.M. DeLucca, B.D. Potteiger, A. Amin, F.A. Baiocchi, J. Mater. Sci.
Mater. Electron 18 (2006) 283
[37] M.O. Peach, Journal of Applied Physics 23/12 (1952) 1401.
[38] S.E. Koonce, S.M. Arnold, Journal of Applied Physics 24/3 (1953) 365.
[39] F.C. Frank, Phil. Mag. 44/7 (1953) 854.
[40] J. Franks, Nature 177/4517 (1956) 984.
[41] J. Franks, Acta Met. 6/2 (1958) 103.
[42] S. Amelinckx, Bontinck, W., Dekeyser, W., Seitz, F., Phil. Mag. 2 (1957) 355.
[43] G.S. Baker, Acta Met. 5/7 (1957) 353.
[44] U. Lindborg, Metall. Trans. 6A (1975) 1581.
[45] R.M. Fisher, L.S. Darken, K.G. Carroll, Acta Metallurgica 2/3 (1954) 368.
[46] P.L. Key, Proc. of the IEEE Comp. Conf. (1970) 155.
103

[47] W.C. Ellis, D.F. Gibbons, R.C. Treuting, in: R.H. Doremus, B.W. Roberts,
D.Turnbull (Eds.), John Wiley & Sons, New York, 1958, p. 102.
[48] V.K. Glazunova, N.T. Kudryavtsev, Translated from Zhurnal Prikladnoi Khimii 36/3
(1963) 543.
[49] T. Kakeshita, K. Shimizu, R. Kawanaka, T. Hasegawa, J. Mater. Sci. 17 (1982)
2560.
[50] B.D. Dunn, European Space Agency (ESA) Report STR-223 (1987) 1.
[51] P.T. Vianco, J.A. Rejent, J. Electronic Materials 38/9 (2009) 1815.
[52] I. Boguslavsky, P. Bush, Processing of the 2003 APEX Conference-Anaheim, CA.
(2003) S12.
[53] J.A. Rejent, P.T. Vianco, J. Electronic Materials 38/9 (2009) 1826.
[54] K.N. Tu, Acta Met. 21/4 (1973) 347.
[55] K.N. Tu, Physical Review B 49/3 (1994) 2030.
[56] K.N. Tu, J.W. Mayer, L.C. Feldman, Electronic Thin Film Science, MacMillan, New
York, 1992.
[57] K.N. Tu, J.C.M. Li, Mater. Sci. and Engr. A 409 (2005) 131.
[58] K. Zeng, K.N. Tu, Materials Science and Engineering: R: Reports 38/2 (2002).
[59] L.A. Piñol, J. Melngailis Jr, H.K. Charles, D.M. Lee, R. Deacon, G. Coles, G.
Clatterbaugh, IEEE Transactions on Components, Packaging and Manufacturing
Technology 1/12 (2011) 2028.
[60] S. Mathew, M. Osterman, T. Shibutani, Q. Yu, M. Pecht, Proc. of Intl. Symp. on High
Dens. Packag. and Microsys. Integr. 2007, HDP'07 (2007) 1.
[61] C. Xu, Y. Zhang, C. Fan, J. Abys, L. Hopkins, F. Stevie, Proc. IPC SMEMA APEX
Conference (2002) S06.
104

[62] Y. Zhang, C. Xu, C. Fan, J. Abys, A. Vysotskaya, Proc. IPC SMEMA APEX
Conference (2002) S06.
[63] M. Sampson, H. Leidecker, J. Kadesch, J. Brusse, NASA website
(http://nepp.nasa.gov/whisker/) (2002).
[64] J.C. Lee, Electronics Packaging Technology Conference, 2008. EPTC 2008. 10th,
2008, p. 1060.
[65] A. Dimitrovska, R. Kovacevic, Journal of Electronic Materials 38/12 (2009) 2516.
[66] Y. Wang, D. Ding, T. Liu, K.P. Galuschki, Y. Hu, A. Gong, M. Shen, H. Sun, X.
Wang, J. Sun, M. Li, D. Mao, 2010, p. 980.
[67] Recommendations on Lead-Free Finishes for Components Used in High-Reliability
Products Ver. 4, iNEMI, Herndon, VA, December 2006.
[68] S. Lal, T.D. Moyer, Electronics Packaging Manufacturing, IEEE Transactions on 28/1
(2005) 63.
[69] J. Chang, L. Bing, Electronic Packaging Technology & High Density Packaging,
2009. ICEPT-HDP '09. International Conference on, 2009, p. 1014.
[70] J.F. Löffler, Intermetallics 11/6 (2003) 529.
[71] M. Chen, NPG Asia Materials 3/9 (2011) 82.
[72] A. Castellero, B. Moser, D.I. Uhlenhaut, F.H.D. Torre, J.F. Löffler, Acta Materialia
56/15 (2008) 3777.
[73] C.J. Chen, J.C. Huang, H.S. Chou, Y.H. Lai, L.W. Chang, X.H. Du, J.P. Chu, T.G.
Nieh, Journal of Alloys and Compounds 483/1-2 (2009) 337.
[74] H.S. Chou, J.C. Huang, L.W. Chang, Surface and Coatings Technology 205/2 (2010)
587.
[75] Y. Li, Q. Guo, J.A. Kalb, C.V. Thompson, Science 322/5909 (2008) 1816.
105

[76] J.P. Chu, T.Y. Liu, C.L. Li, C.H. Wang, J.S.C. Jang, M.J. Chen, S.H. Chang, W.C.
Huang, Thin Solid Films (2013).
[77] P.H. Tsai, Y.Z. Lin, J.B. Li, S.R. Jian, J.S.C. Jang, C. Li, J.P. Chu, J.C. Huang,
Intermetallics 31/0 (2012) 127.
[78] P.T. Chiang, G.J. Chen, S.R. Jian, Y.H. Shih, J.S.C. Jang, C.H. Lai, Fooyin Journal of
Health Sciences 2/1 (2010) 12.
[79] M.L. Lee, K.K. Win, C.L. Gan, L.P. Shi, Intermetallics 18/1 (2010) 119.
[80] S. Wang, D. Sun, S. Hata, J. Sakurai, A. Shimokohbe, Sensors and Actuators A:
Physical 153/1 (2009) 120.
[81] B.R. Huang, T.C. Lin, J.P. Chu, Y.C. Chen, Carbon 50/4 (2012) 1619.
[82] P.H. Tsai, J.B. Li, Y.Z. Chang, H.C. Lin, J.S.C. Jang, J.P. Chu, J.W. Lee, P.K. Liaw,
Thin Solid Films (2013).
[83] C.M. Lee, J.P. Chu, W.Z. Chang, J.W. Lee, J.S.C. Jang, P.K. Liaw, Thin Solid Films
(published online).
[84] H.K. Lin, S.M. Chiu, T.P. Cho, J.C. Huang, Materials Letters 113/0 (2013) 182.
[85] J.P. Chu, J.E. Greene, J.S.C. Jang, J.C. Huang, Y.L. Shen, P.K. Liaw, Y. Yokoyama,
A. Inoue, T.G. Nieh, Acta Materialia 60/6-7 (2012) 3226.
[86] H. Fujii, H. Nakae, K. Okada, Metallurgical Transactions A 24/6 (1993) 1391.
[87] E. Saiz, C.W. Hwang, K. Suganuma, A.P. Tomsia, Acta Materialia 51/11 (2003)
3185.
[88] A. Contreras, E. Bedolla, R. Pérez, Acta Materialia 52/4 (2004) 985.
[89] J.C. Ambrose, M.G. Nicholas, A.M. Stoneham, Acta Metallurgica et Materialia 40/10
(1992) 2483.
[90] G.F. Ma, H.F. Zhang, H. Li, Z.Q. Hu, Materials Letters 63/18-19 (2009) 1605.
[91] Q.G. Xu, H.F. Zhang, B.Z. Ding, Z.Q. Hu, Materials Letters 56/3 (2002) 137.
106

[92] J.P. Chu, T. Mahalingam, S.F. Wang, Journal of Physics: Condensed Matter 16/4
(2004) 561.
[93] K. Barmak, C. Cabral, K.P. Rodbell, J.M.E. Harper, Journal of Vacuum Science and
Technology B: Microelectronics and Nanometer Structures 24/6 (2006) 2485.
[94] J.P. Chu, C.H. Lin, V.S. John, Applied Physics Letters 91/13 (2007) 132109.
[95] D.M. Mattox, Handbook of Physical Vapor Deposition (PVD) Processing (Second
Edition), William Andrew Publishing, Boston, 2010, p. 1.
[98] H. Yinlun, H.L. Helen, Encyclopedia of Chemical Processing, vol. null, Taylor &
Francis, 2007, p. 839.
[99] M.G. Cho, S.K. Kang, D.Y. Shih, H.M. Lee, Journal of Electronic Materials 36/11
(2007) 1501.
[100] A. Paul, A.A. Kodentsov, F.J.J. Van Loo, Zeitschrift fuer Metallkunde/Materials
Research and Advanced Techniques 95/10 (2004) 913.
[101] J. Yu, J.Y. Kim, Acta Materialia 56/19 (2008) 5514.
[102] K.N. Tu, Solder joint technology, Springer, 2007.
[103] K. Barmak, A. Gungor, A.D. Rollett, C. Cabral Jr, J.M.E. Harper, Materials Science
in Semiconductor Processing 6/4 (2003) 175.
[104] M. Saito, H. Sasaki, K. Katou, T. Toba, T. Homma, Journal of The Electrochemical
Society 156/5 (2009) E86.
[105] J.B. LeBret, M.G. Norton, Journal of Materials Research 18/03 (2003) 585.
107

[106] Y. Zhang, S.S. Ang, A.A.O. Tay, D. Xu, E.T. Kang, K.G. Neoh, L.P. Chong, A.C.H.
Huan, Langmuir 19/17 (2003) 6802.
[107] M.E. Williams, K.W. Moon, W.J. Boettinger, D. Josell, A.D. Deal, Journal of
Electronic Materials 36/3 (2007) 214.
[108] L.A. Piñol, J. Melngailis, H.K. Charles, D.M. Lee, R. Deacon, G. Coles, G.
Clatterbaugh, Components, Packaging and Manufacturing Technology, IEEE
Transactions on 1/12 (2011) 2028.
[109] Y.W. Yen, C.K. Li, M.Y. Tsou, P.S. Shao, Japanese Journal of Applied Physics 50/1
PART 3 (2011).
[110] A. Skwarek, M. Pluska, J. Ratajczak, A. Czerwinski, K. Witek, D. Szwagierczak,
Materials Science and Engineering: B 176/4 (2011) 352.
[111] K. Suganuma, A. Baated, K.S. Kim, K. Hamasaki, N. Nemoto, T. Nakagawa, T.
Yamada, Acta Materialia 59/19 (2011) 7255.
[112] J.W. Shin, E. Chason, Journal of Materials Research 24/4 (2009) 1522.
[113] C. Xu, Y. Zhang, C. Fan, J.A. Abys, IEEE Transactions on Electronics Packaging
Manufacturing 28/1 (2005) 31.
108

Dissertation

More Related Content

Viewers also liked (7)

Similar to Dissertation (20)

Dissertation