Temperature-dependent failure mechanism of SnAg solder joints with Cu metallization
after current stressing: Experimentation and analysis
C. K. Lin, Wei An Tsao, Y. C. Liang, and Chih Chen
Citation: Journal of Applied Physics 114, 113711 (2013); doi: 10.1063/1.4821427
View online: http://dx.doi.org/10.1063/1.4821427
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/114/11?ver=pdfcov Published by the AIP Publishing
Articles you may be interested in
The failure models of Sn-based solder joints under coupling effects of electromigration and thermal cycling J. Appl. Phys. 113, 044904 (2013); 10.1063/1.4789023
Effect of contact metallization on electromigration reliability of Pb-free solder joints J. Appl. Phys. 99, 094906 (2006); 10.1063/1.2193037
Electromigration failure mechanisms for SnAg 3.5 solder bumps on Ti Cr - Cu Cu and Ni ( P ) Au metallization pads
J. Appl. Phys. 96, 4518 (2004); 10.1063/1.1788837
Mechanism of electromigration-induced failure in the 97Pb–3Sn and 37Pb–63Sn composite solder joints J. Appl. Phys. 94, 7560 (2003); 10.1063/1.1628388
Mean-time-to-failure study of flip chip solder joints on Cu/Ni(V)/Al thin-film under-bump-metallization J. Appl. Phys. 94, 5665 (2003); 10.1063/1.1616993
Temperature-dependent failure mechanism of SnAg solder joints with Cu
metallization after current stressing: Experimentation and analysis
C. K. Lin, Wei An Tsao, Y. C. Liang, and Chih Chena)
Department of Materials Science and Engineering, National Chiao Tung University, Hsin-chu 30010, Taiwan (Received 8 May 2013; accepted 2 September 2013; published online 19 September 2013)
Temperature-dependent electromigration failure was investigated in solder joints with Cu metallization at 126C, 136C, 158C, 172C, and 185C. At 126C and 136C, voids formed at the interface of Cu6Sn5 intermetallic compounds and the solder layer. However, at temperature
158C and above, extensive Cu dissolution and thickening of Cu6Sn5occurred, and few voids were
observed. We proposed a model considering the flux divergency at the interface. At temperatures below 131C, the electromigration flux leaving the interface is larger than the in-coming flux. Yet, the in-coming Cu electromigration flux surpasses the out-going flux at temperatures above 131C. This model successfully explains the experimental results.VC 2013 AIP Publishing LLC.
[http://dx.doi.org/10.1063/1.4821427]
I. INTRODUCTION
As microelectronic devices continue to produce high operation speeds and superior performance, the current den-sity in interconnects continues to increase, and electromigra-tion remains a critical reliability issue for interconnects.1,2 Solder joints have been employed for interconnects in high-performance devices.3–5 The diameter range of a solder bump is currently 70–100 lm. The diameter for a microbump has reduced dramatically to 20 lm in 3-dimensional inte-grated circuits (3D IC).6–8 The cross-sectional area of a microbump is only 0.05 times the cross-sectional area of a flip-chip joint. Therefore, electromigration continues to be an important reliability issue for solder joints.9–15
Ni and Cu are the most popular under-bump metalliza-tion (UBM) materials. Cu exhibits superior wettability and a high reaction rate with solders.16 During current stressing, electron flow enhances the dissolution of Cu into solders, causing a rapid consumption of Cu and a significant forma-tion of Kirkendall voids.17–20Conversely, Ni exhibits a slow reaction rate with solders.21–23 Therefore, solder joints with Ni UBMs possess larger electromigration lifetimes.24–26 Electromigration in solder joints with Cu UBMs has been examined extensively.17–20 Two major failure mechanisms have been identified: void formation1,14,27 and the dissolu-tion of intermetallic compounds (IMCs).28–30 Recently, Keet al. reported that void formation dominated the electromi-gration failure mechanism at high temperatures whereas the dissolution of IMCs for solder joints with Cu UBMs occurred at low temperatures.31In this study, we observed the reverse trend: the formation of voids at low temperatures and the dis-solution of IMCs at high temperatures in the Cu-Sn system.
In this study, we investigated the electromigration fail-ure mechanism for SnAg solder joints with 20 lm Cu metal-lization on the substrate side. Void formation dominated the electromigration failure mechanism at low stressing temper-atures whereas the dissolution of Cu UBM occurred at high
stressing temperatures. We also proposed a model that con-siders the Cu electromigration fluxes in Cu6Sn5and the Cu
electromigration fluxes in solder. The model successfully explains the experimental results.
II. EXPERIMENTAL
Typical flip-chip solder joints were adopted for the electro-migration tests. Fig. 1(a)shows the schematic drawing for the solder joints. On the chip side, 0.1 lm Ti was sputtered as an ad-hesion layer. Then 0.5 lm Cu was sputtered as a seed layer for the subsequent electroplating of Ni layer. A 2.0 lm layer of Ni was electroplated as the UBM. The metallization on the substrate side is 20-lm-thick Cu. The composition of solder is SnAg (2.6 wt. %). The diameter of UBM and passivation opening is 110 lm and 90 lm, respectively. The contact opening on the substrate side is 110 lm. The Al trace on the chip side is 65 lm wide and 1.5 lm thick whereas the Cu trace on the substrate side is 100 lm wide and 20 lm thick. Pre-solder of Sn-3.0Ag-0.5Cu was used on the substrate side when joining the flip-chip joints.
Four-point probes were used to monitor the resistance change during electromigration tests. Fig. 1(b) presents the schematic structure for the test layout. Currents were applied throng Nodes N2 and N3. Voltage was measured by Nodes N1 and N4. Only bump B2 and B3 were stressed by 1.3 A of current. The resistance measured included the bumps B2 and B3, as well as the Al trace connecting the bumps B2 and B3. The solder joints were stressed by 1.3 A at various tempera-tures, including 100C, 110C, 130C, 140C, and 150C. The calculated current density is 1.4 104A/cm2based on the UBM opening. As the measured resistance increased 10 mX, the current stressing was terminated, and the electro-migration failure mode was examined by a scanning electron microscope (SEM). Compositional analysis was performed by energy dispersive spectrometer (EDS).
III. RESULTS
The microstructure of the fabricated bump was shown in Fig.2. Ternary IMCs of (Cu,Ni)6Sn5formed in the interface
a)Author to whom correspondence should be addressed. Electronic mail:
of Ni and solder on the chip side. The Cu on the chip side came from the substrate side.32,33On the substrate side, bi-nary Cu6Sn5IMCs formed at the interface of Cu
metalliza-tion and the solder. The real temperature in solder joints may be higher than the ambient temperature during current stress-ing due to serious Joule heatstress-ing effect in the stressstress-ing cir-cuit.34,35 Therefore, the real temperature needs to be calibrated. In this study, we employed the temperature coef-ficient of resistivity (TCR) to measure the real temperature in solder joints. The solder joint was placed in an oven, and we measured the resistance using the four point method as a function of the oven temperature. The real temperatures in solder joints during various stressing condition are listed in
Table I. For example, the real temperature is 126C for the joint stressed by 1.3 A at an ambient temperature of 100C. In the following text, we will refer the real temperature as the stressing temperature.
The electromigration failure mode is attributed to void formation at the interface of the Cu UBM and solder when the joint is stressed at 126C. Figs.3(a)and3(b)show a pair of solder joints stressed by 1.3‘ A at 126C for 1600 h. The directions of the electron flow are labeled in the figures. The resistance increased 10 mX after 1600 h. The electromigra-tion damage in both bumps cause the resistance to increase.
FIG. 1. (a) Schematic structure of the solder joint configuration used in this study. (b) Layouts for electromigration tests and Four-point structure for measuring bump resistance.
FIG. 2. Cross-sectional SEM image showing the microstructure of a as-fabricated solder bump. Ni UBM was adopted on the chip side, and Cu met-allization was used on the substrate side.
TABLE I. Calibration of the real temperatures in solder joints at different hotplate temperatures. Applied current (A) Hotplate Temp. (C) Real Temp. (C) Joule heating (C) 1.3 150 185 35 1.3 140 172 32 1.3 130 158 28 1.3 120 146 26 1.3 100 126 26
FIG. 3. The microstructure of solder joints after current stressing by 1.4 104
A/cm2at 126C for 1600 h. (a) With a downward electron flow. (b) With an upward electron flow. Void formation occurred at the Cu-Sn interface on the substrate side.
For the joints with a downward electron flow, some voids formed in the upper right corner of the joint, which is current crowding region. The voids located between (Ni,Cu)3Sn4
IMCs and the solder as shown in Fig. 3(a). A layer type (Cu,Ni)6Sn5 formed on the substrate side. For the solder
bump with an upward electron flow, a large amount of voids formed on the substrate side, which is the current crowding region as shown in Fig.3(b). In fact, the voids are located in the interface of (Cu,Ni)6Sn5IMCs and the solder. The results
indicate that solder was migrated to the chip side, and vacan-cies accumulated at the interface. On the other hand, voids also formed in the interface of Cu metallization and solder on the substrate side for the bump with an upward electron flow.
As the stressing temperature increase to 136C, void for-mation also dominates the electromigration failure mecha-nism. Figures 4(a) and 4(b) show the cross-sectional SEM images for another pair of solder joints stressed at 136C for 1702 h. The resistance increased by 10 mX after the current stressing. Void formation occurred in the Ni/solder interface, as shown in Fig.4(a). For the solder bump with an upward electron flow, serious void formation was also observed at the interface of the (Cu,Ni)6Sn5IMCs and the solder, as presented
in Fig. 4(b). This microstructure change indicates that void formation dominates the failure mechanism at 136C.
However, the failure mode switches to Cu dissolution when the stressing temperatures increase beyond 158C. Figure 5(a) shows the microstructure after current stressing at 158C for 260 h. The Ni UBM was visibly consumed. The Ni UBM becomes discontinuous as the Ni UBM on the right-hand side completely migrated to the solder to form (Cu,Ni)6Sn5IMCs. However, no obvious void formation was
observed at the interface of the Ni UBM and the solder. The results for the Cu/solder interface, which was stressed at 158C, are presented in Figure5(b). Extensive dissolution of Cu on the cathode occurred, and a significant amount of (Cu,Ni)6Sn5IMCs accumulated on the anode/chip end. Few
voids formed at the Cu/solder interface. The formation of the Cu-Sn and Ni-Sn IMCs also contributed to the resistance increase of the solder joints. The electrical resistivity of Cu, Ni, and SnAg solder is 1.7, 6.8, and 12.3 lX cm, respec-tively. Nevertheless, the resistivity of Cu6Sn5 and Ni3Sn4
IMCs is 17.5 and 28.5 lX cm, respectively.36
As the stressing temperature increased to 172C, the Ni UBM was almost completely consumed after 140 h. As
FIG. 4. The microstructure of solder joints after current stressing by 1.4 104
A/cm2at 136C for 1702 h. (a) With a downward electron flow. (b) With an upward electron flow. Thickening of Cu-Sn IMC took place at the Cu-Sn interface on the substrate side.
FIG. 5. The microstructure of solder joints after current stressing by 1.4 104A/cm2at 158C for 260 h. (a) With a downward electron flow. (b) With an upward electron flow. Serious Cu dissolution and Cu-Sn IMC for-mation were observed at the Cu-Sn interface on the substrate side.
shown in Figure6(a), the Ni UBM migrated to the solder and formed significant amounts of (Cu,Ni)6Sn5IMCs in the
solder. Few voids were observed at the original Cu/solder interface. Conversely, extensive Cu dissolution and IMC for-mation also occurred in the Cu/solder interface, as illustrated in Figure6(b). Although some voids formed at the Cu/solder interface, they were not significant, as demonstrated in Figure3(b). When the temperature increased to 185C, rapid dissolution of the Cu and Ni UBMs occurred. Figures7(a)
and 7(b) depict the microstructures of the two bumps stressed at 185C for 56 h. The 2-lm Ni UBM almost com-pletely migrated to the solder through electromigration, resulting in the formation of large amounts of (Cu,Ni)6Sn5
IMCs on the substrate side, as shown in Figure 7(a). Although large amounts of Ni atoms migrated, no discernible voids formed in the vicinity of the original Ni/solder inter-face on the chip side. For the electromigration in Cu metalli-zation in Figure 7(b), significant Cu dissolution also occurred at the cathode end on the substrate. The Cu line on the substrate side exhibits a thickness of 20 lm. Although Cu was almost completely consumed at some locations, few voids were generated at the Cu/solder interface.
IV. THEORETICAL ANALYSIS OF FLUX DIVERGENCE AT THE Cu6Sn5/SOLDER INTERFACE
To explain the temperature-dependent failure mode, we developed a model to calculate the electromigration flux of Cu and Sn at the Cu6Sn5/solder interface. Fig.8(a)includes
a schematic drawing of the solder joint that was subjected to an upward electron flow. Due to electromigration, Cu atoms migrated to the Cu6Sn5 (g0) layer and the Sn2.6Ag solder.
Three distinct fluxes were assumed due to electromigration: Cu flux in the g0 IMC layer,JCu in g0; Cu flux in the solder, JCu in Sn;and Sn flux in the solder,JSn in Sn. We ignore the Cu
and Sn fluxes due to chemical potential. If JCu in g0 is larger than JCu in Sn, the thickness of the interfacial IMC increases.
Conversely, if JCu in Sn is larger thanJCu in g0, the thickness of the interfacial IMC decreases. Considering the flux diver-gence at the interface between the solder and the IMC, if the net flux of Cu is larger than the Sn flux, the interfacial IMC will expand, and no voids will form at the interface. When the outgoing Sn flux at the interface is larger than the incom-ing Cu flux at the interface, voids form at the interface. In this section, we calculate the three fluxes by quantitatively
FIG. 6. Cross-sectional SEM image showing the microstructure of solder joints after current stressing by 1.4 104A/cm2 at 172C for 140 h. (a) With a downward electron flow. (b) With an upward electron flow. Serious Cu dissolution and Cu-Sn IMC formation were observed at the Cu-Sn inter-face on the substrate side. Some voids formed at the interinter-face.
FIG. 7. Cross-sectional SEM image showing the microstructure of solder joints after current stressing by 1.4 104
A/cm2at 185C for 56 h. (a) With a downward electron flow. (b) With an upward electron flow. Extensive Cu dissolution and Cu-Sn IMC formation were found at the Cu-Sn interface on the substrate side. Almost no voids were observed at the interface.
incorporating available data from the literature. The electro-migration flux,JEM, is typically expressed as37
JEM¼ C
D RTZ
eqj; (1)
where C is concentration, D is diffusivity, T is tempera-ture, R is gas constant, Z is effective charge number, e is electron charge, q is resistivity, and j is current den-sity. The Cu flux in the g0 IMC layer can be expressed as follows: JCu in g0 ¼ CCu in g0DCu in g 0 RT Z Cu in g0eqg0j: (2)
Because the diffusivity of Cu in (Cu,Ni)6Sn5IMCs was
not available, we used the diffusivity of Cu in g0IMC, which is expressed as follows:38 DCu in g0 ¼ 6:2 108 exp 80500 RT ðm2=sÞ: (3)
TheZof Cu in the Cu6Sn5IMC is 87.39The qg0is the resis-tivity of Cu6Sn5.40 The Cu flux in solder is expressed as
follows: JCu in Sn¼ CCu in Sn DCu in Sn RT Z Cu in SneqSnj: (4)
Because the Sn weights exceed 97% in the SnAg solder and the Cu diffusivity in Sn DCu in Snwas measured, we take
DCu in Snas the diffusivity of Cu in solder,41which is expressed
as follows: DCu in Sn¼ 2:4 107 exp 33020 RT ðm2=sÞ: (5)
The Z for Cu in Sn is 3.25 (Ref.42), and the resistivity of Sn is 1.23 107X m. The concentration of Cu in Sn, as a function of temperature, can be expressed as follows:43,44
CCu in Sn ¼ 0:9 exp 37500 RT ðat %Þ ¼ 5:61 106exp 37500 RT ðmol=m3Þ: (6)
In the solder near the Cu-Sn IMC, Sn will migrate toward the chip side by electron flow. The Sn flux is expressed as
JSn¼ CSn
DSn
RT Z
SneqSnj; (7)
where theZof Sn is 9.6,45the self-diffusivity of Sn is46
DSn¼ 1:2 109 exp 43890 RT ðm2=sÞ: (8) Thus, JSn¼ ej 1:09 1011 1 T exp 43890 RT : (9)
Considering the interface between the Cu6Sn5and the solder,
the incoming Cu flux is JCu in g0 whereas the outgoing flux is (JCu in Snþ JSn). We substitute the parameters and obtain the
incoming and outgoing fluxes as a function of temperature. The unit for the electromigration flux is mol/m2s, and the unit for concentration is mol/m3. Therefore, the density of Sn is 6.24 104mol/m3.
The Cu concentration in Cu6Sn5 is also dependent on
temperature. We obtain the solubility limit from the Cu-Sn phase diagram47 and solve the temperature dependence of Cu in g0. The Cu in g0can be obtained as follows:
CCu in g0¼ 5:50 106 exp 242
RT
ðmol=m3Þ: (10)
With these parameters, we can plot the electromigration flux at the Cu6Sn5/solder interface as a function of temperature.
Fig.8(b)displays the curves for the incoming and outgoing fluxes in the temperature range of 20C–200C. The results indicate a crossover at 131C, below which the outgoing flux is larger than the incoming flux. Therefore, voids may form at the interface. In contrast, the incoming flux is larger than the outgoing flux, which results in the dissolution of Cu
FIG. 8. (a) Schematic drawing of a solder joint subject to an upward electron flow. Three electromigration fluxes at the Cu-Sn interface were also labeled. (b) The calculated curves for the in-coming and out-going fluxes at tempera-ture range from 20C to 200C.
and extensive IMC formation, but without void formation. The theoretic calculations successfully explain the experi-mental results.
V. DISCUSSION
Several experimental results obtained by other research-ers support the results of this study. Linet al. reported exten-sive Cu-Sn IMC formation when the interface of Cu and Sn was stressed by 3.28 103A/cm2 at 160C.48 However, when the interface was stressed by 5.3 103A/cm2at 55C,
numerous large voids formed at the interface.49Xuet al. per-formed electromigration tests under conditions of 2.0 104A/cm2 and 135C, in which both mechanisms
were observed.19 Therefore, temperature-dependent electro-migration failure occurs in the Cu and Sn interfaces.
It is interesting that some Sn atoms migrated against electron flow and back filled the original position of the Cu metallization on the substrate side, as shown in Figs. 3(b),
5(b), and7(b). This phenomenon may be attributed to back stress of electromigration.50,51The Cu and Sn were migrated to the chip side due to electromigration. Yet, the solder joints were confined by the underfill, and there were no hillocks or extrusion on the chip side to release the compressive stress on the chip side. On the other hand, tensile stress was devel-oped on the substrate side because the deficiency of Sn and Cu atoms. Therefore, a stress gradient was built up across the solder joints, which triggered the migration of Sn to the original location of Cu metallization. In addition, chemical
potential may also cause the diffusion of Sn atoms to the Cu end. Sn atoms tend to form Cu-Sn IMCs on the Cu-Sn inter-face to lower the free energy. As the Cu atoms were migrated away from the substrate side by the electron flow, Sn atoms may diffuse back to the Cu surface. This chemical potential force may increase as the temperature increases because the Cu-Sn reaction rate is higher at high temperatures. As shown in Figs. 5(b)and7(b), extensive Cu dissolution took place, and most of them were migration to the chip side. Nevertheless, Sn atoms diffused against electromigration and filled the original Cu position, resulting in almost no voids formed in the substrate side.
It is noteworthy that voids may form at later stages of electromigration even at high temperatures. Especially, when the Cu is consumed completely, voids will occur because the in-coming Cu flux becomes zero.24
The electromigration behavior at the Ni3Sn4/solder
interface appears similar to that at Cu6Sn5/solder interface.
As presented in Figs.3(a)–7(a), voids formed at the interface of (Ni,Cu)3Sn4/solder at 126C–185C. Yet, Ni dissolution
becomes more significant as the temperature increases. However, the diffusion parameters Ni in (Ni,Cu)3Sn4IMCs
are currently not available. It deserves more study and analysis.
Chemical potential may also affect the Sn flux at the IMC/solder interface because Sn atoms tend to move to the Cu to form Cu-Sn IMCs. Therefore, the Sn flux due to chem-ical potential diffuses against electron flow for the joints with upward electron flow. In the following, we will
FIG. 9. The calculated Sn flux attributed to chemical potential and electromigration-induced Sn flux by 1.4 104A/cm2at (a) 120C; (b) 150C; (c) 170C; and (d) 200C.
calculate the Sn flux at the stressing conditions adopted in this study. We will take the IMC growth rate constants published in literature and estimate the Sn flux due to the chemical poten-tial. The Sn flux attributed to the chemical potential is
JSn in IMC ¼
k A VSn dSn
VIMC MSn A t
; (11)
wherek is IMC growth rate constant, A is area, VSnis molar
volume of Sn,dSnis density of Sn,VIMCis molar volume of
IMC,MSn is atomic weight of Sn, and t is time. The IMC
growth rate constant for Cu6Sn5IMC in a Sn-Cu-Ni system
is 0.2232 lm/h0.5at 150C.52With these parameters, we can plot the electromigration-induced Sn flux using Eq.(7)and the Sn flux due to chemical potential as a function of time. Figures9(a)through9(d)show the calculated results at tem-perature 120C, 150C, 170C, and 200C, respectively. The results indicate that the Sn flux due to chemical potential is smaller than the electromigration flux in the stressing con-ditions in this study. Therefore, it is reasonable to neglect the flux due to chemical potential in this work.
VI. CONCLUSIONS
In summary, we investigated the electromigration failure mechanism in solder joints with Cu UBM at temperatures ranging from 126C to 185C. Void formation at the Cu6Sn5/solder interface caused failure at low temperatures.
However, the dissolution of Cu UBM and the formation of Cu-Sn IMC dominated the failure mechanism at high tem-peratures. By considering the flux divergence at the Cu6Sn5/
solder interface, we proposed a model to calculate the elec-tromigration flux at the interface. The results indicate that the outgoing flux is larger than the incoming flux at tempera-tures below 131C. However, the reverse trend is observed for temperatures above 131C. This model successfully explains the observed experimental results.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial sup-port of the National Science Council of the Republic of China (Grant No. NSC 99-2221-E-009-040-MY3).
1K. N. Tu,J. Appl. Phys.94, 5451 (2003). 2
K. N. Tu,Solder Joint Technology (Springer, New York, 2007).
3
K. Zeng and K. N. Tu,Mater. Sci. Eng. R.38, 55 (2002).
4
J. H. Lau,Flip Chip Technology (McGraw-Hill, New York, 1996), p. 123.
5M. Abtew and G. Selvaduray,Mater. Sci. Eng., R.27, 95 (2000). 6
K. N. Tu,Microelectron. Reliab.51, 517 (2011).
7
J. C. Lin, W. C. Chiou, K. F. Yang, H. B. Chang, Y. C. Lin, E. B. Liao, J. P. Hung, Y. L. Lin, P. H. Tsai, Y. C. Shih, T. J. Wu, W. J. Wu, F. W. Tsai, Y. H. Huang, T. Y. Wang, C. L. Yu, C. H. Chang, M. F. Chen, S. Y. Hou, C. H. Tung, S. O. Jeng, and D. C. H. Yu, “High density 3D integration using CMOS foundry technologies for 28 nm node and beyond,” inIEEE International Electron Device Meeting (2010), pp. 2.1.1–2.1.4.
8A. Yu, J. H. Lau, S. W. Ho, A. Kumar, W. Y. Hnin, D. Q. Yu, M. C. Jong,
V. Kripesh, D. Pinjala, and D. L. Kwong, “Study of 15 lm pitch solder microbumps for 3D IC integration,” in Electronic Components and Technology Conference (ECTC) (2009), pp. 6–10.
9C. Chen, H. M. Tong, and K. N. Tu,Annu. Rev. Mater. Sci.40, 531 (2010). 10
Y. C. Chan, A. C. K. So, and J. K. L. Lai, Mater. Sci. Eng., B.55, 5 (1998).
11C. Chen, H. Y. Hsiao, Y. W. Chang, F. Y. Ouyang, and K. N. Tu,Mater.
Sci. Eng. R.73, 85 (2012).
12
S. Brandenburg and S. Yeh, “Electromigration studies of flip-chip solder bump solder joints,” in Proceedings of Surface Mount International Conference and Exhibition, San Jose (1998), pp. 337–344.
13C. Y. Liu, C. Chen, C. N. Liao, and K. N. Tu,Appl. Phys. Lett.
75, 58 (1999).
14
Y. W. Chang, S. W. Liang, and C. Chen,Appl. Phys. Lett.89, 032103 (2006).
15T. L. Shao, Y. H. Chen, S. H. Chiu, and C. Chen,J. Appl. Phys.
96, 4518 (2004).
16
H. K. Kim and K. N. Tu,Appl. Phys. Lett.67, 2002 (1995).
17C. Y. Liu, J. T. Chen, Y. C. Chuang, K. Lin, and S. J. Wang,Appl. Phys.
Lett.90, 112114 (2007).
18
C. Y. Liu, L. Ke, Y. C. Chuang, and S. J. Wang, J. Appl. Phys. 100, 083702 (2006).
19L. H. Xu, J. K. Han, J. J. Liang, K. N. Tu, and Y. S. Lai,Appl. Phys. Lett.
92, 262104 (2008).
20
B. Chao, S. H. Chae, X. F. Zhang, K. H. Lu, M. Ding, J. Im, and P. S. Ho,
J. Appl. Phys.100, 084909 (2006).
21P. G. Kim, J. W. Jang, and T. Y. Lee,J. Appl. Phys.86, 6746 (1999). 22H. F. Hsu and S. W. Chen,Acta Mater.
52, 2541 (2004).
23
C. Schmetterer, H. Flandorfer, and H. Ipser,Acta Mater.56, 155 (2008).
24
H. Y. Chen, M. F. Ku, and C. Chen,Adv. Mater. Res.1, 83 (2012).
25H. Y. Chen and C. Chen,J. Mater. Res.26, 983 (2011). 26H. Y. Chen and C. Chen,J. Mater. Res.
25, 1847 (2010).
27
E. C. C. Yeh, W. J. Chos, K. N. Tu, P. Elenius, and H. Balkan,Appl. Phys. Lett.80, 580 (2002).
28J. R. Huang, C. M. Tsai, Y. W. Lin, and C. R. Kao,J. Mater. Res.23, 250
(2008).
29
Y. C. Hu, Y. H. Lin, C. R. Kao, and K. N. Tu,J. Mater. Res.18, 2544 (2003).
30Y. L. Lin, C. W. Chang, C. M. Tsai, C. W. Lee, and C. R. Kao,
J. Electron. Mater.35, 1010 (2006).
31
J. H. Ke, T. L. Yang, Y. S. Lai, and C. R. Kao, Acta Mater.59, 2462 (2011).
32T. L. Shao, T. S. Chen, Y. M. Huang, and C. Chen,J. Mater. Res.19, 3654
(2004).
33
S. J. Wang and C. Y. Liu,Scr. Mater.55, 347 (2006).
34
S. H. Chiu, T. L. Shao, and C. Chen,Appl. Phys. Lett.88, 022110 (2006).
35H. Y. Hsiao, S. W. Liang, M. F. Ku, C. Chen, and D. J. Yao,J. Appl.
Phys.104, 033708 (2008).
36
C. Chen and S. W. Liang,J. Mater. Sci.: Mater. Electron.18, 259 (2007).
37
H. B. Hungtington and A. R. Grone,J. Phys. Chem. Solids20, 76 (1961).
38A. Paul, C. Ghosh, and W. J. Boettinger,Metall. Mater. Trans. A42, 952
(2011).
39
S. Ou and K. N. Tu, “A study of electromigration in Sn3.5Ag and Sn3.8Ag0.7Cu solder lines,” inElectronic Components and Technology Conference (ECTC) (2005), pp. 1445–1450.
40H. P. R. Frederikse, R. J. Fields, and A. Feldman,J. Appl. Phys.
72, 2879 (1992).
41
B. F. Dyson, T. R. Anthony, and D. Turnbull, J. Appl. Phys.38, 3408 (1967).
42M. Y. Hsieh and H. B. Huntington,J. Phys. Chem. Solids
39, 867 (1978).
43
J. H. Shim, C. S. Oh, B. J. Lee, and D. N. Lee, Zeitschrift f€ur Metallkunde 87, 205 (1996), available athttp://cat.inist.fr/?aModele=afficheN&cpsidt=3010941.
44K. W. Moon, W. J. Boettinger, U. R. Kattner, F. S. Biancaniello, and C. A.
Handwerker,J. Electron. Mater.29, 1122 (2000).
45
P. H. Sun and M. Ohring,J. Appl. Phys.47, 478 (1976).
46
W. Seith and T. Heumann, Diffusion of Metals: Exchange Reactions (Springer, Berlin, 1962), pp. 65, 68.
47N. Saunders and A. P. Miodownik, ASM Handbook: Alloy Phase
Diagrams, edited by H. Baker (ASM International, Materials Park, Ohio/ Second Publishing, Inc., 1991), Vol. 3, p. 2.178.
48C. T. Lin, Y. C. Chuang, S. J. Wang, and C. Y. Liu,Appl. Phys. Lett.89,
101906 (2006).
49
H. W. Tseng, C. T. Lu, Y. H. Hsiao, P. L. Liao, Y. C. Chuang, T. Y. Chung, and C. Y. Liu,Microelectron. Reliab.50, 1159 (2010).
50I. A. Blech and C. Herring,Appl. Phys. Lett.29, 131 (1976). 51F. Y. Ouyang, K. Chen, and K. N. Tu, Appl. Phys. Lett.
91, 231919 (2007).
52
J. W. Yoon, Y. H. Lee, D. G. Kim, H. B. Kang, S. J. Suh, C. W. Yang, C. B. Lee, J. M. Jung, C. S. Yoo, and S. B. Jung,J. Alloys Compd.381, 151 (2004).
