inemi pb-free alloy characterization...

iNEMI Pb-FREE ALLOY CHARACTERIZATION PROJECT REPORT: PART IV - EFFECT OF ISOTHERMAL PRECONDITIONING

ON THERMAL FATIGUE LIFE

Richard Coyle1, Richard Parker2, Gregory Henshall3, Michael Osterman4, Joe Smetana5 Elizabeth Benedetto6, Donald Moore7, Graver Chang8, Joelle Arnold9, and Tae-Kyu Lee10

1Alcatel-Lucent, Murray Hill, NJ, USA 2Delphi, Kokomo, IN, USA,

3Hewlett-Packard Co., Palo Alto, CA, USA 4CALCE, College Park, MD, USA 5Alcatel-Lucent, Plano, TX, USA

6Hewlett-Packard Co., Houston, TX, USA 7Henkel Corp., Irvine, CA, USA

8IST Inc., HsinChu, Taiwan, ROC 9DFR Solutions, College Park, MD, USA

10Cisco Systems, San Jose, CA, USA [email protected]; [email protected]

ABSTRACT The Pb-Free Alloy Characterization Program sponsored by International Electronics Manufacturing Initiative (iNEMI) is conducting an extensive investigation using accelerated temperature cycling (ATC) to evaluate ball grid array (BGA) thermal fatigue performance of a significant number of commercial and experimental Sn based Pb-free solder alloys. The temperature cycling test matrix includes a subset of test cells designed to evaluate the influence of aging or isothermal preconditioning on temperature cycling performance. The aging tests consist of a comparison of two commercial Sn based Pb-free alloys, SAC305 and SAC105, with and without aging for 10 days at 125 °C. Results are presented for multiple temperature cycles in order to compare the relationship between cyclic temperature ranges (T) and temperature extremes on the thermal fatigue life and microstructural evolution. The test data and failure analyses are discussed in terms of the relationship to Ag content and to the initial as well as evolving microstructures during temperature cycling. Key words: Pb-free solder, thermal fatigue, aging, thermal preconditioning INTRODUCTION Solder joints age and degrade during service and eventually fail by the common wear out mechanism of solder fatigue [1]. For Sn-Ag-Cu (SAC) Pb-free solders, there are reports that the solder fatigue life, unlike that of Sn37Pb eutectic solder, degrades faster if the joints are subjected to isothermal preconditioning, often referred to as aging [2-4]. Thermal fatigue and creep performance is critically important for high reliability applications because it is a major source of failure for surface mount (SMT)

components [5]. In Sn based Pb-free alloys, the processes of microstructural aging, creep, and thermal fatigue are known to be dependent on the alloy composition, particularly the Ag content. Some studies have identified aging factors that affect microstructure, creep, and fatigue of SAC alloys. Research has included limited work on commercial components [1, 6-9] in addition to basic solder property studies [10-15]. These studies indicate that there is a relationship between the bulk SAC solder microstructure (not interfacial intermetallic structures that limit mechanical properties [16]) and thermal fatigue performance. In contrast, with Sn-37Pb eutectic surface mount solder joints, there is little evidence that the initial solder microstructure has a strong influence on the thermal fatigue resistance [17-18]. The difference between the Sn based Pb-free and Sn-37Pb alloys appears to be due to the difference in creep behavior of the alloys. For SAC alloys, studies indicate that the creep and fatigue behavior can be dependent strongly on microstructure. In the case of SAC alloys, higher Ag content results in higher Ag3Sn particle density and prolonged thermal fatigue life. This observation supports the hypothesis that the fatigue resistance in SAC alloys is controlled by the Ag content and Ag3Sn intermetallic particle density [6, 19-22]. The microstructures of SAC alloys evolve continually by coarsening of Ag3Sn and Cu6Sn5 particles and recrystallization of Sn grains in areas of stress concentration during thermal aging or thermal cycling [23-26]. The combined effect appears to be one of reducing creep resistance and thermal fatigue resistance of the solder joint. In the cases where the initial solder microstructure is more resistant to microstructural evolution during thermal

376

exposure, the solder joint resistance to thermal fatigue is much less likely to change over time [23]. Some of the data also indicate that the effect of aging on the thermal cycle performance, as simulated by preconditioning at constant temperature, can vary by component package type [1]. The Pb-Free Alloy Characterization Program sponsored by International Electronics Manufacturing Initiative (iNEMI) is conducting an extensive investigation using accelerated temperature cycling (ATC) to evaluate BGA thermal fatigue performance of a significant number of commercial and experimental Pb-free solder alloys [27]. The temperature cycling test matrix for the iNEMI program includes test cells designed to evaluate the influence of aging or isothermal preconditioning on temperature cycling performance. This paper presents some of the initial results of the study comparing thermal fatigue performance of SAC305 (Sn3.0Ag0.5Cu) and SAC105 (Sn1.0Ag0.5Cu) alloys with and without aging at 125 °C. Results are presented for multiple temperature cycles in order to compare the effects of cyclic temperature ranges (T) and temperature extremes on the thermal fatigue life and microstructural evolution. The test results and failure analyses are discussed in terms of the relationship to Ag content and to the initial as well as evolving microstructures during temperature cycling. EXPERIMENTAL Detailed descriptions of the iNEMI Pb-Free Alloy Characterization program goals, experimental plan, and test protocols are provided in previous publications [27-28]. The test program remains in progress and includes 12 commercial or developmental Pb-free solder alloys, a eutectic Sn-Pb solder baseline, two different ball grid array (BGA) components, and 10 different temperature cycles for studying thermal fatigue behavior. The results for the experimental aging subset presented here include only SAC305 and SAC105 alloys, both of the BGA components, and only 4 of the temperature cycles. Test Vehicle Figure 1 shows the fully populated test vehicle. Each test board contains 16 of the larger 192CABGA and 16 of the smaller 84CTBGA BGA components. The BGA component and PCB test vehicle characteristics are shown in Table 1.

Figure 1: A fully populated iNEMI Alloy Characterization test vehicle. Table 1: Ball grid array (BGA) and printed circuit board (PCB) test vehicle attributes.

Test Vehicle Assembly and Aging The test vehicles were assembled using Type 3 no-clean SAC305 paste. The peak temperature for the SAC305 reflow profile was set at 245°C. Following assembly, all the cards were subjected to electrical test and x-ray inspection. Isothermal aging of the test boards was done in an ambient air environment at 125 °C for 10 days (240 hours). Accelerated Temperature Cycling The components and the test circuit boards were daisy chained to allow electrical continuity testing after surface mount assembly and in situ, continuous monitoring during thermal cycling. Thermal cycling was done in accordance with the IPC-9701A industry test guideline [29] and additional details are provided in another publication [28]. The solder joints were monitored using either an event detector or a data logger set at a resistance limit of 1000 ohms, as described elsewhere [28]. The failure data are reported as characteristic life (the number of cycles to achieve 63.2% failure) and slope from a two-parameter Weibull analysis. The temperature cycling matrix reported in this paper is shown in Table 2; results for other profiles will be reported as data become available

377

Table 2: The various temperature cycling conditions used in the aging experimental test matrix.

1 02 -40 103 -40 104 25 10

Cycle Number

Minimum Temp.

Maximum Temp. (°C)

Temp. Range (°C)

Dwell Time (min.)

100 100125 165100 140125 100

10

Microsturctural Characterization and Failure Analysis A baseline characterization was performed on representative board level assemblies from each test cell. These baselines were performed to document the microstructures before temperature cycling and to enable comparisons to samples removed from the temperature cycling chambers for failure analysis. Microstructural characterization and failure analysis was done using optical metallography (destructive cross-sectional analysis) and scanning electron microscopy (SEM). Once the initial phase identification and elemental analysis was completed using energy dispersive spectroscopy (EDS), the SEM operating in the backscattered electron imaging (BEI) mode was used to differentiate phases in the SAC microstructures. Backscattered electron imaging (BEI) has been shown to be particularly useful in previous studies for differentiating phases in the SAC microstructures [6, 8, 9, 19]. Optical metallography (destructive cross-sectional analysis) was the primary method used to verify the failure mode. RESULTS AND DISCUSSION Microsturctural Characterization Figures 2 and 3 show high magnification backscattered images (original magnification 1000X) comparing the baseline (prior to thermal cycling) BGA solder joint microstructures for the 192CABGA and the 84 CTBGA packages. These micrographs enable the microstructural comparisons between surface mount (SMT) assembly and subsequent isothermal preconditioning (aging) for each of the two package test vehicles and two solder ball alloys used in this study. Lower magnification optical images of assembled and aged samples are shown in Appendix A. Analysis at relatively high magnification is required to resolve the fine microstructural features of the SAC alloys. The 192CABGA microstructures shown in Figure 2 consist of primary Sn cells with Ag3Sn particles (lighter phase) that most likely result from binary eutectic decomposition at the cell boundaries. As expected, the SAC305 alloy with its higher Ag content contains more Ag3Sn particles and those particles often precipitate as wider networks along the Sn boundaries [6, 19]. The Ag3Sn precipitate density is much lower in SAC105 alloy and the Ag3Sn particles are arranged more linearly rather than in wider networks or bands. During isothermal aging, the Ag3Sn precipitates coarsen or ripen [1-4], which results in fewer but larger particles. The primary Sn grain size does not seem to be altered by the isothermal aging.

The microstructures for the 84CTBGA package shown in Figure 3 differ significantly from those of the larger 192CTBGA package. The Ag3Sn precipitate density again is higher in the SAC305 but the Sn cell size in the smaller diameter 84CTBGA balls of both alloys is noticeably smaller compared to the Sn cell size in the larger diameter 192CABGA balls. The Sn cell boundaries in this case most likely are low angle sub-grain boundaries, are decorated by the Ag3Sn precipitates. Isothermal aging coarsens the Ag3Sn precipitates in both alloys and this does appear to produce significant growth in Sn cell size, although the boundaries are no longer well-defined due to coarsening. In addition to having fewer Ag3Sn particles, SAC105 differs from SAC305 in that it contains more Cu6Sn5 intermetallic precipitates (darker phase). The presence of Cu6Sn5 is not a critical factor because Ag3Sn particles are considered to be the primary microstructural feature that influences thermal fatigue resistance in SAC alloys [6, 19-22, 27, 30]. Compared to the Ag3Sn particles, the number of Cu6Sn5 particles is low and they are distributed randomly. During thermal cycling, SAC microstructures experience additional aging due to the combined in situ influence of strain and temperature. In both of the packages and alloys studied here, isothermal aging has resulted in obvious Ag3Sn particle coarsening before thermal cycling began.

Figure 2: Backscattered electron micrographs of baseline microstructures of SAC305 and SAC105 192CABGA solder joints. The Ag3Sn particle density is higher in the SAC305 solder joints and Ag3Sn particle coarsening is evident in both alloys due to isothermal aging.

378

Figure 3: Backscattered electron micrographs of baseline microstructures of SAC305 and SAC105 84CTBGA solder joints. The primary Sn cell size is much smaller in the 84CTBGA compared to the 192CABGA shown in Figure 2. Ag3Sn particle coarsening due to isothermal aging is evident with both alloys. ATC Test Results Summary The thermal cycling results are summarized in Tables 3 through 6 and shown graphically in the Weibull plots presented in Appendix B. The tables include the characteristic lifetimes and Weibull slopes (β) to facilitate comparisons of components and temperature cycles for the two alloys, with and without aging. Table 3: Thermal cycling statistics for the 192CABGA assembled with SAC305 solder balls, with and without isothermal aging.

Table 4: Thermal cycling statistics for the 192CABGA assembled with SAC105 solder balls, with and without isothermal aging.

Table 5: Thermal cycling statistics for the 84CTBGA assembled with SAC305 solder balls, with and without isothermal aging.

Table 6: Thermal cycling statistics for the 84CTBGA assembled with SAC105 solder balls, with and without isothermal aging.

For this type of thermal cycling evaluation, the IPC 9701 test method guideline [29] mandates a sample size of 32, in order to minimize the impact that experimental deviations or outliers have on the Weibull statistics. This study uses an initial sample size of only 16 components per cell resulting from resource limitations imposed by the large number of alloys and cycles required in the experimental plan [28]. There are variations in slope (β) across the data sets in this study and these β variations should be taken into consideration when making characteristic lifetime comparisons between data sets. The most striking and pertinent trend in the current data is that the 10 day/125 °C isothermal aging generally does not result in a significant reduction in the thermal fatigue life in the majority of the test cells. This outcome could have been anticipated based on previous publications [1, 6]. Likewise, it is not startling to see that the influence of aging on thermal fatigue life can be dependent on temperature cycling parameters variables such T and temperature extremes [1, 31]. This is expected because the critical underlying factors that influence microstructural aging in SAC alloys are the solder joint strain generated during thermal cycling and the higher temperature extreme used in the cycle. It is well known that strain and temperature, in conjunction with the coefficient of thermal expansion (CTE) mismatch, drive thermal fatigue failures in solder joints. With respect to package type, the most obvious trend in these data is that the 84CTBGA outperforms the 192CABGA by 30-40% as indicated by characteristic lifetime. However, unlike some previous findings [1, 31], there is not a strong link in the current study between aging, characteristic lifetime, and package type. In a few of the test cells in the current study, the 192CABGA is slightly more

379

sensitive to the effects of isothermal aging than the 84CTBGA. Presumably this is because the package construction generates higher shear strains in the solder joints resulting in accelerated microstructural evolution (coarsening) and subsequent fatigue cracking. This view is consistent with results from other studies suggesting that strain-enhanced coarsening during temperature cycling dominates the microstructural evolution [6, 32] in components with longer inherent lifetime. In those situations the negative contribution of microstructural evolution due to pre-aging is relatively insignificant compared to the microstructural evolution due to strain-enhanced, in situ evolution. However, it must be noted that significant microstructural differences are observed in the two packages (Figures 2 and 3) and this could be as equally important as the inherent strain of the two packages. Although aging generally did not degrade the fatigue life in most of the test cells, there are a few exceptions. In the 0/100°C cycle, the 192CABGA with SAC105 loses about 20% of its fatigue life with aging. The other two cycles where aging lowered the characteristic lifetime measurably were the -40/125 °C and the 25/125 °C. However, in those cases the aging affect was observed with both packages but only with the SAC305 alloy. The common factor in those two test cells is the 125 C upper temperature extreme. Aging is seen to affect the reliability more often in the test cells using the 125 C upper temperature extreme but even here the results are not dramatic or consistent. The data from Tables 3-6 show that -40/125 °C testing drives failures several times faster than any of the other cycles. Because failure occurs very rapidly during the -40/125 °C testing, the microstructural contribution from pre-aging may be significant. It is interesting that aging reduces the reliability of the 192CABGA by about 30% with the SAC305 alloy but has no effect on the SAC105. Aging reduces the reliability of the 84CTBGA by 15% with the SAC305 and produces no change with SAC105. Regardless of the isothermal aging, the SAC305 outperforms the SAC105 with either the 192CABGA or the 84CTBGA. A similar deviation in the general aging trend is observed in the data for the 25/125 °C temperature cycle. With this cycle, the data show a reduction in characteristic lifetime with aging for both components but only with the SAC305. Again, the fatigue performance of the SAC305 measured by characteristic lifetime is equal to or better than the SAC105 despite any reduction due to pre-aging. The most significant aging effects are found with the two most severe temperature cycles and with the SAC105 192CABGA in the 0/100 °C temperature cycle. While the observations for the severe cycles can perhaps be reconciled with an argument based on rapid in situ aging, the 0/100 °C data for the SAC105 are anomalous and challenging to interpret in a manner self-consistent with typical phenomenological arguments based on Ag content and particle coarsening. However, it must be reiterated that there are variations in slope (β) across the data sets and these β

variations should be taken into consideration when making characteristic lifetime comparisons between data sets. The SAC305 exhibits better overall fatigue resistance regardless of the aging phenomenon although in a few of the more severe test cells the statistical separation between the two alloys arguably is minimal. This result is not so surprising considering the recent results reported by Lee and Ma, in which they observed a minimal difference in performance between SAC305 and SAC105 using a high-stress test vehicle configuration [31]. They concluded that Ag3Sn precipitate coarsening occurred so rapidly under their test conditions that the Ag content and particle density were no longer the primary factors controlling thermal fatigue life. The current -40/125 °C results are consistent with this hypothesis because Ag content is less of a factor in this severe temperature cycle characterized by a combination of the highest strain (T) and the highest upper temperature extreme. Although the current results may be explained with the Lee and Ma hypothesis, the reduction in fatigue life in the SAC305 components due to aging in the two temperature cycles using the 125 °C temperature extreme are worthy of further study. Special microstructural analysis may be required in order to explain these findings. It must be noted that all the temperature cycles reported in this study used 10 minute dwell times. Some studies have shown that testing with a longer dwell time further diminishes the effect of isothermal aging or eliminates it completely [1, 6, 8, 30]. The iNEMI Alloy Program includes testing with longer dwell times and those results will be published at a later date. FAILURE ANALYSIS Failure analysis was performed to document the solder joint thermal fatigue failures. The restricted sample size of 16 per test cell made it necessary to cycle each test board to 100% failure (N=100) to satisfy the statistical requirements of the acceleration factor Design of Experiment [28]. Each board is populated with both components (Figure 1) and the mean cycles to failure of the 84CTBGA is substantially longer than that of the 192CABGA. Therefore, the 192CABGA components were exposed to a large number of cycles beyond the final failure in that set, which results in significantly more damage accumulation throughout the solder ball array of the 192CABGA packages. The optical photomicrographs in Figures 4 and 5 show fatigue cracking resulting from thermal cycling of BGA solder joints assembled with SAC305 and SAC105, with and without thermal preconditioning (aging). Representative photomicrographs are shown comparing results from the least aggressive (0/100 °C) and most aggressive (-40/125 °C) cycles. Additional failure analysis for the other temperature cycles will be presented in future publications. The Pb-free failures are characterized by failures at the package side of the solder joint with a variety of crack paths and fracture features such as crack branching,

380

recrystallization and cavitation. These fracture characteristics are seen routinely in SAC thermal fatigue failures and are consistent with those reported much earlier by workers such as Dunford [33]. The observed fatigue damage typically is greater in the 192 CABGA due to the extended cycling beyond the final failure for those packages. In both components, the most significant amount of damage was observed at the die shadow as expected for these types of packages. It is assumed that the initial failures occur at the package corners or at the die shadow but that can not be confirmed by metallographic analysis. Optical photomicrographs confirming thermal fatigue cracking in samples from the -40/125 °C cycle are shown in Figures 6 and 7. The extent of the fatigue damage observed in -40/125 °C typically is more severe than in 0/100 °C due to the more aggressive temperature cycling, particularly in the SAC105 samples.

Figure 4: A series of photomicrographs comparing thermal damage fatigue in 192CABGA and 84CTBGA packages assembled with SAC305 solder balls and SAC305 solder paste, with and without isothermal aging. The temperature cycle was 0/100 °C.

Figure 5: A series of photomicrographs comparing thermal damage fatigue in 192CABGA and 84CTBGA packages assembled with SAC105 solder balls and SAC305 solder paste, with and without isothermal aging. The temperature cycle was 0/100 °C.

Figure 6: A series of photomicrographs comparing thermal damage fatigue in 192CABGA and 84CTBGA packages assembled with SAC305 solder balls and SAC305 solder paste, with and without isothermal aging. The temperature cycle was -40/125 °C.

381

Figure 7: A series of photomicrographs comparing thermal damage fatigue in 192CABGA and 84CTBGA packages assembled with SAC105 solder balls and SAC305 solder paste, with and without isothermal aging. The temperature cycle was -40/125 °C. The backscattered electron micrographs in Figures 8 through 12 illustrate the microstructural evolution process resulting from accelerated temperature cycling. These higher magnification images show the intermetallic particle coarsening and Sn recrystalization that precede fatigue failure in these Pb-free SAC alloys. Figure 8 shows images from SAC305 192CABGA components with and without 0/100 °C temperature cycling. Figure 9 shows comparable SAC305 192CABGA images for the -40/125 °C temperature cycling. These images demonstrate the dramatic Ag3Sn particle coarsening in the region that contains the fatigue cracking. The wider bands of Ag3Sn particles present originally along the Sn cell boundaries in the SMT assembled SAC305 sample (also see Figure 2) no longer exist due to particle coarsening during thermal cycling. With both temperature cycles, there are very few Ag3Sn particles present in the strain-localized region surrounding the fatigue crack. Despite the significant differences in their thermal cycles, the resultant microstructures in this region are very similar.

Figure 8: Backscattered electron micrographs illustrating the microstructural evolution of the 192CABGA component with SAC305 solder balls with 0/100 °C thermal cycling.

Figure 9: Backscattered electron micrographs illustrating the microstructural evolution of the 192CABGA component with SAC305 solder balls and with -40/125 °C thermal cycling. The images in Figure 10 and Figure 11 show examples of microstructures from the high strain regions surrounding fatigue cracks created during 0/100 °C temperature cycling. When viewed at even higher magnification in Figure 12, only a few large (coarse) Ag3Sn particles can be seen in the fatigued region. The strain-enhanced aging that occurs during temperature cycling is known to accelerate particle coarsening and promote fatigue crack propagation. Note the similarity in the microstructures of all four samples shown in Figures 10 and 11 regardless of component differences, Ag content and pre-aging of the microstructures. The characteristic lifetimes may be different for each combination of alloy and component but by the time the fatigue crack has begun to propagate to failure, the evolution and extent of the particle coarsening in the strain-localized region is roughly equivalent for all conditions tested. Further, the inherent characteristic lifetimes for both components (with either alloy) are sufficiently high that the initial microstructural degradation imparted by isothermal pre-aging often has minimal influence on the number of

382

cycles to failure. All these observations are consistent with the hypothesis that SAC fatigue life generally is controlled by the Ag content and Ag3Sn particle density and that the effects of pre-aging on fatigue life are minimal for most components under most temperature cycling conditions. While there are a few exceptions to these observations within the data sets reported, most notably in the -40/125 °C data set, it also remains to be determined if those exceptions would translate into an impact on actual service life. With respect to microstructural characterization and failure analysis, there is one potentially important aspect of the deformation and fatigue failure of SAC alloys that is not addressed in this paper. The fatigue resistance of SAC solder joints depends on the interaction between the Ag3Sn intermetallic particles and the primary Sn cell or dendrite boundaries. There is evidence from the literature that the Sn grain morphology and orientations of the anisotropic Sn grains can influence the fatigue behavior [24, 34, 35]. Two specialized analytical methods can be applied to study the Sn orientation problem: polarized light microscopy (PLM) and orientation imaging microscopy (OIM) using electron backscattered diffraction (EBSD). These analytical methods were beyond the scope of the current study and may not yield useful results in this case given the sample damage caused by extended cycling. However, even if the fatigue damage is so severe that it precludes effective microstructural analysis, there may be some value to characterizing the baseline microstructures using each of these methods.

Figure 10: Backscattered electron micrographs illustrating the microstructural evolution in the 192CABGA component during 0/100 °C temperature cycling.

Figure 11: Backscattered electron micrographs of the microstructural evolution in the 84CTBGA component during 0/100 °C temperature cycling.

Figure 12: An example of accelerated intermetallic particle coarsening in the strain-localized region of a 0/100 °C SAC305, 192CABGA sample. The combination of strain and temperature in this region promotes recrystallization and fatigue crack propagation. In the absence of higher strain, coarsening is much slower as evidenced by the smaller particles and higher particle density in the region adjacent to (below) the crack.

SUGGESTIONS FOR FUTURE WORK All the temperature cycling reported in the current study was done using 10 minute dwell times. This is a relatively short dwell time and it has been shown in other studies that testing with a longer dwell time further diminishes the effect

383

of isothermal aging or preconditioning in SAC alloys. The test matrix for this iNEMI Pb-Free Alloy Characterization Program includes temperature cycles with longer dwell times and tests with longer dwell times are in progress. The results from those tests will be compared to the current results to further assess the effect of dwell time on characteristic lifetime (fatigue life) measured in temperature cycling tests. Further failure analysis and appropriate microstructural characterization is recommended in order to understand the aging-induced reduction in SAC305 fatigue life measured in the -40/125 °C and 25/125 °C temperature cycles. Specialized techniques for microstructural characterization could be explored including polarized light microscopy (PLM) and orientation image microscopy (OIM) using electron backscattered diffraction (EBSD). Extended thermal cycling may cause sample damage that will limit the effectiveness of either of these techniques. If that happens, some advanced microstructural analysis on the baseline samples may provide some insight relative to the failure behavior in SAC305. SUMMARY AND CONCLUSIONS Accelerated temperature cycling was used to assess the influence of aging or isothermal preconditioning on thermal fatigue performance of two commercial Pb-free alloys, SAC305 and SAC105, with and without aging for 10 days at 125 °C. The experimental test matrix included two different component test vehicles and four different thermal cycles with data for six other cycles to be reported when available. The following observations and conclusions can be drawn from the results of this study.

In almost all the test cells, the isothermal aging parameters used in this investigation (10 days at 125 °C) had minimal measureable effect on the thermal fatigue life. There are exceptions to that observation in the most aggressive temperature cycle, -40/125 °C and one test cell using 0/100 °C cycling. That former cycle drives failures the fastest of any cycle reported here. It is hypothesized that the rapid fatigue failure induced by the severity of the -40/125 °C test conditions allows the microstructural degradation from pre-aging to have a significant affect on the fatigue life.

The data for the 25/125 °C temperature cycle also show an aging effect similar to that observed in the -40/125 °C cycle. With this cycle, the data show a moderate 15% reduction in characteristic lifetime with aging for both components but the reduction due to aging is observed only with the SAC305 alloy. The common factor in these two cycles is the 125 °C upper temperature extreme.

The results of the metallographic analyses show that the failure mode in all cases is solder fatigue at the package side of the solder joints. The first failures appear to occur at the die shadow in both components.

The thermal fatigue performance (characteristic lifetime) of the 192CABGA component may be slightly more sensitive to the effects of isothermal aging than the 84CTBGA. In general, the characteristic lifetime of the 192CABGA is 30-40% less than that of the 84CTBGA. Presumably this is caused by higher inherent solder joint shear strains with a possible additional factor due to microstructure. This may be similar in the case of the -40/125 °C cycle, where microstructural effects from aging have more impact on the shorter life 192CABGA component.

The SAC305 alloy with its higher Ag content outperforms the lower Ag SAC105 alloy in all temperature cycles. However, in the test cells using this most aggressive cycle, the statistical separation between the SAC305 and SAC105 alloys often is minimal. The -40/125 °C results imply that Ag3Sn precipitate coarsening occurs so rapidly that Ag content and particle density are not necessarily the primary factors controlling thermal fatigue life under those test conditions.

ACKNOWLEDGEMENTS The authors want to acknowledge the continued guidance of iNEMI, particularly that of David Godlewski and Jim Arnold. The authors also thank Keith Sweatman and Keith Howell of Nihon Superior and Liang Yin of Universal Instruments for numerous helpful technical discussions. Many thanks to Pete Read and Debra Fleming from Alcatel-Lucent for helpful technical discussions, metallographic characterization, and failure analysis. The authors thank Jim Carrigan, Dave Grant, and the entire staff at Premier Semiconductor Services LLC for their flawless execution of balling the test packages. Thanks to Jason Bragg of Celestica for analyses related to samples from the Premier process to ensure integrity of the ball attachments. Many thanks to Jon Goodbread of Agilent Technologies for his efforts in developing the DAS code for the Agilent data acquisition systems. REFERENCES [1] Joe Smetana, Richard Coyle, Peter Read, Richard Popowich, Debra Fleming, and Thilo Sack, “Variations in Thermal Cycling Response of Pb-free Solder Due to Isothermal Preconditioning,” Proceedings of SMTAI 2011, 641-654, Fort Worth, TX, October 2011. [2] Y. Zhang, Z. Cai, J. C. Suhling, P. Lall, M. J. Bozack, “The Effects of SAC Alloy Composition on Aging Resistance and Reliability,” Proceedings of Electronic Components and Technology Conference, 370-389, San Diego, CA, 2009. [3] H. Ma, J. C. Suhling, Y.Zhang, P. Lall, and M. J. Bozack, “The Influence of Elevated Temperature Aging on Reliability of Lead Free Solder Joints,” Proceedings of Electronic Components Technology Conference 2007, 653-668, Reno, NV, 2007.

384

[4] H. Ma, J. C. Suhling, Y.Zhang, P. Lall, and M. J. Bozack, “Reliability of the Aging Lead Free Solder Joint,” Proceedings of Electronic Components Technology Conference, 849-864, San Diego, CA, 2006. [5] Werner Engelmaier “Surface Mount Solder Joint Long-Term Reliability: Design, Testing, Prediction,” Soldering and Surface Mount Technology, vol. 1, no. 1, 14-22, February, 1989. [6] Richard Coyle, John Osenbach, Maurice Collins, Heather McCormick, Peter Read, Debra Fleming, Richard Popowich, Jeff Punch, Michael Reid, and Steven Kummerl, “Phenomenological Study of the Effect of Microstructural Evolution on the Thermal Fatigue Resistance of Pb-Free Solder Joints,” IEEE Trans. CPMT, vol. 1, no. 10, 1583-1593, October 2011. [7] Tae-Kyu Lee, Hongtao Ma, Kuo-Chuan Liu, And Jie Xue, “Impact of Isothermal Aging on Long-Term Reliability of Fine-Pitch Ball Grid Array Packages with Sn-Ag-Cu Solder Interconnects: Surface Finish Effects,” J. Electronic Materials, vol. 39, no. 12, 2564-2573, 2010. [8] R. Coyle, M. Reid, C. Ryan, R. Popowich, P. Read, D. Fleming, M. Collins, J. Punch, I. Chatterji, “The Influence of the Pb-free Solder Alloy Composition and Processing Parameters on Thermal Fatigue Performance of a Ceramic Chip Resistor,” Proceedings of Electronic Components Technology Conference, 423-430, IEEE, Piscataway, NJ, 2009. [9] J. Manock, R. Coyle, B. Vaccaro, H. McCormick, R. Popowich, D., P. Read., J. Osenbach, and D. Gerlach, “Effect of Temperature Cycling Parameters on the Solder Joint Reliability of a Pb-free PBGA Package,” SMT J., vol. 21, no.3, 36, 2008. [10] B. Dompierre, V. Aubin, E. Charkaluk, W. C. Maio Filho, and M. Brizoux, Cyclic mechanical behaviour of Sn3.0Ag0.5Cu alloy under high temperature isothermal ageing, Materials Science and Engineering A, vol. 528, 4812-4818, 2001. [11] Y. Zhang, Z. Cai, J. Suhling, P. Lall, and M. Bozack, “Aging Effects in SAC Solder Joints,” Proceedings of the SEM International Congress and Exposition on Experimental and Applied Mechanics, Albuquerque New Mexico USA, June, 2009. [12] D. Bhate, D. Chan, G Subbarayan, T. C. Chiu, V. Gupta, “Constitutive Behavior of Sn3.8Ag0.7Cu and Sn1.0Ag0.5Cu Alloys at Creep and Low Strain Rate Regimes,” IEEE Trans. CPMT, 2007. [13] Q. Xiao L. Nguyen, W. Armstrong, “Aging and Creep Behavior of Sn3.9Ag0.6Cu Solder Alloy,” Proceedings of Electronic Components Technology Conference, 1325-1332, Las Vegas, NV, 2004. [14] Xiao Q, Bailey H, Armstrong W, “Aging Effects on Microstructure and Tensile Property of Sn3.9Ag0.6Cu Solder Alloy,” J. Electronic Packaging, vol. 126, 208-212, 2004. [15] K. Mysore, D. Chan, D. Bhate, G. Subbarayan, I. Dutta, V. Gupta, J. Zhao, and D. Edwards, “Aging-Informed Behavior of Sn3.8Ag0.7Cu Solder Alloys,” Proceedings Thermal and Thermomechanical Phenomena in Electronic Systems (ITHERM), 870-875, Orlando, FL, May 2008.

[16] Dong Hyun Kim, tae-Kyu Lee, Sang Ha Kim, Han G. Park, and Kuo-Chuan Liu, “Study on Dynamic Shock Performance of SAC305 Solder Joint After Different Aging Conditons,” Proceedings of SMTAI 2008, 182-186, Orlando, FL, August 2008. [17] J. Bartelo, S. Cain, D. Caletka, K. Darbha, T. Gosselin, D. Henderson, D. King, K. Knadle, A. Sarkhel, G. Thiel, C. Woychik, D. Shih, S. Kang, K. Puttlitz and J. Woods, “Thermomechanical Fatigue Behavior of Selected Pb-Free Solders,” Proceedings IPC APEX 2001, LF2-2, Bannockburn, IL, 2001. [18] M. Osterman, A. Dasgupta, and B. Han, “A Strain Range Based Model for Life Assessment of Pb-free SAC Solder Interconnects,” Proceedings of Electronic Components and Technology Conference, 884-890, San Diego, CA, 2006. [19] Richard Coyle, Heather McCormick, John Osenbach, Peter Read, Richard Popowich, Debra Fleming, and John Manock, “Pb-free Alloy Silver Content and Thermal Fatigue Reliability of a Large Plastic Ball Grid Array (PBGA) Package,” Journal of SMT, Vol. 24, Issue 1, 27-33, January-March 2011. [20] Fei Lin, Wenzhen Bi, Guokui ju, Wurong Wang, Xicheng Wei, “Evolution of Ag3Sn at Sn-3.0Ag-0.3Cu-0.05Cr/Cu joint interfaces during thermal aging,” J. Alloys and Compounds, vol. 509, 6666-6672, 2011. [21] S. Terashima, Y. Kariya, Hosoi, and M. Tanaka, “Effect of Silver Content on Thermal Fatigue Life of Sn-xAg-0.5Cu Flip-Chip Interconnects,” J. Electron. Mater. vol. 32, no. 12, 2003. [22] G. Henshall. J. Bath, S. Sethuraman, D. Geiger, A. Syed, M.J. Lee, K. Newman, L. Hu, D. Hyun Kim, Weidong Xie, W. Eagar, and J. Waldvogel, “Comparison of Thermal Fatigue Performance of SAC105 (Sn-1.0Ag-0.5Cu), Sn-3.5Ag, and SAC305 (Sn-3.0Ag-0.5Cu) BGA Components with SAC305 Solder Paste,” Proceedings APEX, S05-03, 2009. [23] Y. Zhang, Z. Cai, J.C. Suhling, P. Lal, M. Bozak, “The Effects of Aging Temperature on SAC Solder Joint Material Behavior and Reliability,” Proceedings of Electronic Components and Technology Conference, 99-112, Lake Buena Vista, FL, 2008. [24] T. Bieler, H. Jiang, L. Lehman, T. Kirkpatrick, E. Cotts, and B. Nandagopal, “Influence of Sn Grain Size and Orientation on the Thermomechanical Response and Reliability of Pb-free Solder Joints,” IEEE Trans. CPMT, vol. 31, no. 2, 370- 381, 2008. [25] R. Darveaux, C. Reichman, C. Berry, W-S Hsu, and A, Syed, “Effect of Joint Size and Pad Metallization on Solder Mechanical Properties,” Proceedings of Electronic Components and Technology Conference, 113-122, Lake Buena Vista, FL, 2008 [26] S.K. Kang, Paul Lauro, Da-Yuan Shih, Donald W. Henderson, Timothy Gosselin, Jay Bartelo, Steve R. Cain, Charles Goldsmith, Karl J. Puttlitz, and Tae-Kyung Hwang, “Evaluation of Thermal Fatigue Life and Failure Mechanisms of Sn-Ag-Cu Solder Joints with Reduced Ag Contents,” Proceedings of Electronic Components and Technology Conference, 661-667, Las Vegas, NV, 2004.

385

http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=4538254

http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=4538254

[27] G. Henshall, R. Healy, R. S. Pander, K. Sweatman, K. Howell, R. Coyle, T. Sack, P. Snugovsky, S. Tisdale, and F. Hua, “iNEMI Pb-free Alloy Alternatives Project Report: State of the Industry, SMTJournal, vol. 21, no. 4, 11, 2008. [28] Gregory Henshall, Jian Miremadi, Richard Parker, Richard Coyle, Joe Smetana, Jennifer Nguyen, Weiping Liu, Keith Sweatman, Keith Howell, Ranjit S. Pandher, Derek Daily, Mark Currie, Tae-Kyu Lee, Julie Silk, Bill Jones, Stephen Tisdale, Fay Hua, Michael Osterman, Bill Barthel, Thilo Sack, Polina Snugovsky, Ahmer Syed, Aileen Allen, Joelle Arnold, Donald Moore, Graver Chang, and Elizabeth Benedetto, “iNEMI Pb-Free Alloy Characterization Project Report: Part I – Program Goals, Experimental Structure, Alloy Characterization, and Test Protocols for Accelerated Temperature Cycling,” Proceedings of SMTAI 2012, Orlando, FL, October 2012. [29] IPC-9701A, “Performance Test Methods and Qualification Requirements for Surface Mount Solder Attachments,” IPC, Bannockburn, IL, 2006. [30] Richard Coyle, John Osenbach, Peter Read, Heather McCormick, Debra Fleming, Richard Popowich, Michael Reid, Jeff Punch, Maurice Collins, Robert Kinyanjui, and Steven Kummerl, “Dwell Time, Microstructural Dependencies, and the Interpretation of Thermal Fatigue Test Data of SnPb and Pb-free Solders,” Proceedings of SMTAI 2009, 384-392, San Diego, CA, October 2009. [31] Tae-Kyu Lee and Hongtao Ma, “Aging Impact on the Accelerated Thermal Cycling performance of lead-Free BGA Solder Joints in Various Stress Conditions,” Proceedings of Electronic Components and Technology Conference, 477-482, San Diego, CA, May 29- June 1, 2012. [32] I. Dutta, P. Kumar, and G. Subbarayan, “Microstructural Coarsening in Sn-Ag-based Solders and Its Effects on mechanical Properties,” JOM, Vol. 61, No. 6, 29-38, June 2009. [33] S. Dunford, S. Canumalla, and P. Viswanadham, “Intermetallic Morphology and Damage Evolution Under Thermomechanical Fatigue of Lead (Pb)-Free Solder Interconnections,” Proceedings of Electronic Components Technology Conference, 726-736, Las Vegas, NV, June 1-4, 2004. [34] B. Arfaei, N. Kim, and E.J. Cotts, “Dependence of Sn Grain Morphology of Sn-Ag-Cu Solder on Solidification Temperature,” J. Electronic Materials, Published online Oct 19, 2011. [35] Liang Yin, Luke Wentlent, LinLin Yang, Babak Arfaei, Awni Oasaimeh, and Peter Bargemen, “Recrystallization and Precipitate Coarsening in Pb-Free Solder Joints During Thermomechanical Fatigue,” J. Electronic Materials, Vol. 41, No. 2, 241-252, 2011.

386

APPENDIX A

Optical Photomicrographs of SMT Assembled and Aged 192CABGA and 84CTBGA Samples

Figure A-1: Low magnification optical photomicrographs showing cross sectional views of the SAC305 192CABGA and 84CTBGA test vehicles. The photomicrographs show each package test vehicle with and without isothermal aging for 10 days at 125 °C.

Figure A-2: Low magnification optical photomicrographs showing cross sectional views of the SAC105 192CABGA and 84CTBGA test vehicles. The photomicrographs show each package test vehicle with and without isothermal aging for 10 days at 125 °C.

387

Figure B-1: This Weibull plot shows the influence of aging on the thermal fatigue life of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC305 alloy solder balls and tested using the 0 to 100 °C temperature cycle. Aging has no measureable effect on the fatigue life of either component.

Figure B-2: This Weibull plot shows the influence of aging on the thermal fatigue life of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC105 alloy solder balls and tested using the 0 to 100 °C temperature cycle. Aging reduces the fatigue life of the 192CABGA by about 20% but has no measureable effect on the fatigue life of the 84CTBGA.

Figure B-3: This Weibull plot shows the influence of aging on the thermal fatigue of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC305 alloy and tested using the -40 to 125 °C temperature cycle. Aging reduces the fatigue life of the 192CABGA by about 30% and the 84CTBGA by about 15%.

Figure B-4: This Weibull plot shows the influence of aging on the thermal fatigue of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC105 alloy and tested using the -40 to 125 °C temperature cycle. Aging has no appreciable affect on the characteristic lifetime of either component. Note that the Weibull slope (β) for the 192CABGA component with SAC105 and no isothermal aging (β=10.16) is significantly higher than the other three test cells (β~4-6).

ReliaSoft Weibull++ 7 - www.ReliaSoft.com

Probability - Weibull

Delphi Data\Cell5-BGA84-SAC105SAC305: Delphi Data\Cell5-BGA192-SAC105SAC305: Delphi Data\Cell13-BGA84-SAC105 agedSAC305: Delphi Data\Cell13-BGA192-SAC105 agedSAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000

99.000 Probability-Weibull

Delphi Data\Cell13-BGA192-SAC105 agedSAC305Weibull-2PRRX SRM MED FMF=16/S=0

Data PointsProbability Line


Data PointsSusp PointsProbability Line

Delphi Data\Cell5-BGA192-SAC105SAC305Weibull-2PRRX SRM MED FMF=16/S=0


Delphi Data\Cell5-BGA84-SAC105SAC305Weibull-2PRRX MF=

SRM MED F16/S=0


Richard ParkerDelphi8/3/201211:09:32 AM


Probability - Weibull

Delphi Data\Cell7-BGA84-SAC305SAC305: Delphi Data\Cell7-BGA192-SAC305SAC305: Delphi Data\Cell14-BGA84-SAC305 agedSAC305: Delphi Data\Cell14-BGA192-SAC305 agedSAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000










Richard ParkerDelphi8/3/201211:10:34 AM


P robability - Weibull

Alcatel-Lucent Data\Cell5-BGA84-SAC105SAC305: Alcatel-Lucent Data\Cell5-BGA192-SAC105SAC305: Alcatel-Lucent Data\Cell13-BGA84-SAC105 agedSAC305: Alcatel-Lucent Data\Cell13-BGA192-SAC105 agedSAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000


Alcatel-Lucent Data\Cell13-BGA192-SAC105 agedS...Weibull-2PRRX SRM MED FMF=16/S=0




Alcatel-Lucent Data\Cell5-BGA192-SAC105SAC305Weibull-2PRRX SRM MED FMF=16/S=0


Alcatel-Lucent Data\Cell5-BGA84-SAC105SAC305Weibull-2PRRX MF=

SRM MED F16/S=0


Richard ParkerDelphi7/20/20122:44:03 PM



Alcatel-Lucent Data\Cell7-BGA84-SAC305SAC305: Alcatel-Lucent Data\Cell7-BGA192-SAC305SAC305: Alcatel-Lucent Data\Cell14-BGA84-SAC305 agedSAC305: Alcatel-Lucent Data\Cell14-BGA192-SAC305 agedSAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 100000.0001000.000 10000.0001.000

5.000

10.000

50.000

90.000











APPENDIX B

Weibull Plots

0/100 °C and -40/125 °C thermal cycles

388

Figure B-5: This Weibull plot shows the influence of aging on the thermal fatigue life of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC305 alloy solder balls and tested using the -40 to 100 °C temperature cycle. There is no appreciable difference in the fatigue life of ither component with or without aging.

the fatigue life of ither component with or without aging.

duces the fatigue life of both components by about 0%.

measureable effect on the fatigue life of either omponent.

e

Figure B-6: This Weibull plot shows the influence of aging on the thermal fatigue life of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC105 alloy solder balls and tested using the -40 to 100 °C temperature cycle. There is no appreciable difference in e

Figure B-7: This Weibull plot shows the influence of aging on the thermal fatigue life of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC305 alloy solder balls and tested using the 25 to 125 °C temperature cycle. Aging re2

Figure B-8: This Weibull plot shows the influence of aging on the thermal fatigue life of the 192CABGA (left) and 84 CTBGA (right) assembled with the SAC105 alloy solder balls and tested using the 25 to 125 °C temperature cycle. Aging has no c



IST Data 7-17-12\Cell5-BGA84-SAC105SAC305: IST Data 7-17-12\Cell5-BGA192-SAC105SAC305: IST Data 7-17-12\Cell13-BGA84-SAC105 agedSAC305: IST Data 7-17-12\Cell13-BGA192-SAC105 agedSAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000


IST ell13-BGA192-SAC105 agedSA...WeRRXF=16/S=0

Data 7-17-12\Cibull-2P SRM MED FM


IST l13-BGA84-SAC105 agedSAC...WeRRXF=16/S=0

Data 7-17-12\Celibull-2P SRM MED FM


IST l5-BGA192-SAC105SAC305WeRRXF=16/S=0

Data 7-17-12\Celibull-2P SRM MED FM


IST l5-BGA84-SAC105SAC305WeRRX SRM MED FMF=16/S=0

Data 7-17-12\Celibull-2P





IST Data 7-17-12\Cell14-BGA84-SAC305 agedSAC305: IST Data 7-17-12\Cell14-BGA192-SAC305 agedSAC305:

IST Data 7-17-12\Cell7-BGA84-SAC305SAC305: IST Data 7-17-12\Cell7-BGA192-SAC305SAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000


IST ell14-BGA192-SAC305 agedSA...WeRRXF=16/S=0



IST ell14-BGA84-SAC305 agedSAC...WeRRXF=16/S=0



IST Cell7-BGA192-SAC305SAC305WeRRF=15/

Data 7-17-12\ibull-2PX SRM MED FM

S=1Data PointsSusp PointsProbability Line

IST Cell7-BGA84-SAC305SAC305WeRRF=15/

Data 7-17-12\ibull-2PX SRM MED FM


rkerDelphi7/20/20122:46:28 PM

Richard Pa



Henkel Data\Cell5-BGA84-SAC105SAC305: Henkel Data\Cell5-BGA192-SAC105SAC305: Henkel Data\Cell13-BGA84-SAC105 agedSAC305: Henkel Data\Cell13-BGA192-SAC105 agedSAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000


Hen BGA192-SAC105 agedSAC305WeRRF=15/

kel Data\Cell13-ibull-2PX SRM MED FM


Hen GA84-SAC105 agedSAC305WeRRXF=16/S=0

kel Data\Cell13-Bibull-2P SRM MED FM


Hen A192-SAC105SAC305WeRRXF=16/S=0

kel Data\Cell5-BGibull-2P SRM MED FM


Hen A84-SAC105SAC305WeRRX SRM MED FMF=16/S=0

kel Data\Cell5-BGibull-2P





Henkel Data\Cell14-BGA84-SAC305 agedSAC305: Henkel Data\Cell14-BGA192-SAC305 agedSAC305:

Henkel Data\Cell7-BGA84-SAC305SAC305: Henkel Data\Cell7-BGA192-SAC305SAC305:

Time, (t)

Unr

elia

bilit

y, F

(t)

200.000 10000.0001000.0001.000

5.000

10.000

50.000

90.000


Hen BGA192-SAC305 agedSAC305WeRRF=15/



Hen BGA84-SAC305 agedSAC305WeRRXF=16/S=0

kel Data\Cell14-ibull-2P SRM MED FM


Hen GA192-SAC305SAC305WeRRXF=16/S=0

kel Data\Cell7-Bibull-2P SRM MED FM


Hen BGA84-SAC305SAC305WeRRF=12/



rkerDelphi7/20/20122:46:13 PM

Richard Pa

APPENDI B

Weibu Pl s

-40/100 °C and 25/125 °C thermal cycles

X ll ot

389

inemi pb-free alloy characterization...

Documents