|
|
|
|
|
The Effect of Thermal Cycling Profile on Thermal Fatigue Performance of an 84-Pin Thin Core BGA
Analysis Lab |
|
Authored By:Dan Burkholder Intel Corporation, Chandler, AZ, USA Raiyo Aspandiar, Yunfei Wang Intel Corporation, Hillsboro, OR, USA Russ Brown Intel Corporation, Folsom, CA, USA Richard Coyle Nokia, Murray Hill, NJ, USA Babak Arfaei Binghamton University, Binghamton, NY Vasu Vasudevan Dell Technologies, Round Rock, TX Aileen Allen HP, Inc., Palo Alto, CA, USA Keith Howell Nihon Superior Co., Ltd., Osaka, Japan Qin Chen Eunow, Suzhou, China Derek Daily Senju Comtek Corp, Santa Clara, CA Haley Fu iNEMI, Shanghai, China Carol Handwerker Purdue University, West Lafayette, IN, USA Daniel Werkhoven Interflux Electronics NV, Belgium Kei Murayama Shinko Electric Industries Co. LTD., Nagano, Japan Hongwen Zhang, Francis Mutuku, Huaguang Wang Indium Corporation, Clinton, NY, USA Morgana Ribas MacDermid Alpha Electronics Solutions, Bengaluru, India Murali Sarangapani Heraeus Materials Singapore Pte Ltd, Singapore SummaryThere is an increasing interest in many market segments to use solder alloys with lower melting temperatures for electronics assembly. Low temperature solders can provide manufacturing, economic, and environmental benefits. Since 2015, the International Electronics Manufacturing Initiative (iNEMI) Low Temperature Solder Process and Reliability (LTSPR) Project has been evaluating Low Temperature Solder (LTS) paste formulations based on the Bi-Sn system. This paper summarizes the findings from a thermal cycling test to evaluate the effect of thermal cycling profile on thermal fatigue performance of low temperature solders. The study uses a daisy chained printed circuit board and two daisy chained ball grid array (BGA) test vehicles, a 192-pin chip array BGA (CABGA192) and an 84-pin thin core BGA (CTBGA84). The test matrix includes multiple LTS solder alloys designated by code names. The alloys are down selected from the larger project alloy matrix based on assembly effectiveness and mechanical test performance. There are two types of solder alloys, so-called ductile metallurgies that employ alloy modifications to improve the properties of the basic Bi-Sn alloy, and joint reinforced pastes (JRP) that employ resin additions that generate in situ fillets during reflow to provide joint support. Components manufactured with the established SAC305 (Sn3.0Ag0.5Cu) composition are used as the baseline for the study. Three types of LTS solder joints are evaluated, homogeneous, hybrid (heterogeneous), and hybrid formed with joint reinforced pastes (JRP). Hybrid joints have a SAC BGA soldered with a ductile metallurgy LTS solder paste. The resultant solder joint consists of an unmelted SAC region at the package side of the joint and a melted region at the PCB side containing Bi from the solder paste. A hybrid joint also may be described as heterogeneous because it contains two regions with clearly distinct microstructures, compositions, and properties. Homogeneous joints are created when a BGA manufactured with LTS solder spheres is soldered to the PCB using a LTS solder paste with matching composition. A JRP joint is a special type of hybrid joint that has resin fillets that form during reflow to enhance joint support. To address the resources required for a comprehensive evaluation of low temperature solder technology, the International Electronic Manufacturing Initiative (iNEMI) launched the BiSn-based Low Temperature Soldering Process and Reliability (LTSPR) Project in 2015. Now in this third iNEMI LTSPR phase, testing is conducted with two distinct Accelerated Temperature Cycling (ATC) profiles, 0/100° C (IPC-9701B, TC1) and -15/85° C (selected due to homologous temperature considerations between Sn-Ag-Cu and Bi-Sn solder and maintaining a similar delta T [100° C] across both ATC profiles). This interim paper reports the thermal cycling results for the 84-pin thin core BGA component (CTBGA84) tested with the -15/85° C profile and compares these results (as a companion paper) to the CTBGA84 tested with the 0/100° C profile, presented in a previous publication [61]. Weibull statistics, microstructural characterization, and failure mode analysis are used to compare the differences in alloy performance and to compare the performance of hybrid and homogeneous solder joint configurations. At the time of writing this paper, two of the seven -15/85 °C profile legs had not yet reached the N63 failure criteria per IPC-9701B (homogeneous Sultan 2 at 19% failure rate with 25,000 cycles run and hybrid Beserah JRP at 58% failure rate with 28,500 cycles run). However, as many cycles have already been completed for this experiment and the Weibull slope (β) and Correlation Coefficient (r2) for these two legs are healthy, only a relatively small change in the final Weibull statistics are expected once N63 is reached. The Weibull statistics data will be updated again when this experiment finishes completely. ConclusionsThe thermal fatigue resistance of six Low Temperature Solder (LTS) alloys was assessed using a CTBGA84 ball grid array test vehicle and accelerated thermal cycling profiles of 0 to 100 °C (IPC-9701B, TC1) and -15 to 85 °C. The original concern with the accelerated 0/100 °C thermal profile was that the Bi-Sn might not perform as well with the higher acceleration provided at an upper temperature extreme of 100 °C. As 100 °C is close to the onset of the low temperature solder melting temperature, it may result in a decrease in the strength of the solder, possibly lowering the characteristic lifetime. Additionally, most computer products, used by consumers, do not operate continuously at 100 °C. For both reasons, another profile with a lower upper temperature extreme was also used for this study. Thus the -15/85 °C accelerated profile was selected due to homologous temperature considerations between Sn-Ag-Cu and Bi-Sn solder and maintaining a similar delta T [100° C] across both ATC profiles. From the Weibull characteristic lifetime analysis of the CTBGA84 component, all LTS legs in the -15/85 °C profile had comparable or exceeded the characteristic lifetime performance of the SAC305 baseline, when adding 90% confidence intervals. In the more aggressive 0/100 °C profile, all LTS legs had comparable or exceeded the characteristic lifetime performance of the SAC305 baseline, when adding 90% confidence intervals, except for hybrid Red Flesh which had a lower characteristic lifetime performance (as presented in a previous publication [61]). From the Weibull 1% cumulative failure analysis of the CTBGA84 component, when adding 90% confidence intervals in the -15/85 °C profiles, all LTS legs had comparable 1% cumulative failure performance except for hybrid Golden Pillow 2 JRP, which had a very different slope than the other legs (β=1.1), with first failures starting as early as 507 cycles. These early 1% cumulative failures indicate a time zero SMT quality defect (HoP-like solder joints). These defects are believed to form by the premature gelling of the resin in the JRP paste before the solder has become molten and has wet the BGA SAC sphere. Increasing the initial reflow profile ramp rate to >3°C/sec (from 1-2°C/sec) for the Golden Pillow 2 leg is one suggested method to eliminate these defects [22]. In the more aggressive 0/100 °C profile, all CTBGA84 component LTS legs had comparable or exceeded the 1% cumulative failure performance of the SAC305 baseline, when adding 90% confidence intervals, except for homogeneous Red Flesh which had a lower 1% cumulative failure performance (presented in previous publication [61]). Metallographic cross-sectional failure analysis on the CTBGA84 component, from both ATC profiles, showed that fatigue cracking near the package side within the bulk solder was the predominant failure mode. However, some mixed mode interfacial/fatigue failures were detected in some of the hybrid test legs. Non-fatigue failures in hybrid solder joints may be related to Bi diffusion into the SAC region during thermal cycling, which strengthens the SAC solder. Fatigue cracking at the package side in the SAC region is found to initiate at boundary triple points and propagate in solder material, which is commonly characterized by local recrystallization, global recrystallization, crack branching, and cavitation at boundary triple points [37, 41, 42, 43, 54, 55, 56, 61, 62]. A much smaller amount of fatigue cracking was detected in the Bi-mixed region of hybrid and JRP solder joints. The fatigue cracking at the PCB side in the Bi-mixed region proceeds typically along phase boundaries, recrystallized Sn grains, and occasionally by cleavage through Bi precipitates. Initially Published in the SMTA Proceedings |
|
Comments
|
|
|
|
|
|
|
|
|