Effect of cooling rate, silver composition, dwell time and solder joint size on the reliability of tin-silver-copper solder joints

Thermo-mechanical fatigue, as a consequence of the coefficient of thermal expansion mismatch (CTE), is considered a significant wear-out failure mode for solder joints in electronics applications. For years, accelerated temperature cycling (ATC) has been the preferred technique for evaluating the fatigue performance of eutectic Sn-Pb solder joints and a considerable amount of test data and relevant field experience exists for eutectic Sn-Pb solder.;Experimental investigation was performed to evaluate the effect of cooling rate, dwell time during ATC, solder joint size and Ag composition on the microstructure and consequently on the fatigue life of ceramic ball grid array (CBGA) lead free solder joints. The key experimental variables that define the initial microstructure are the solder joint cooling rate during surface mount (SMT) assembly and the Ag content of the CBGA solder balls. Two Pb-free alloys were considered in this research, Sn3.0Ag0.5Cu, commonly called SAC305 with Ag composition of 3.0 wt. % and Sn1.0Ag0.5Cu, commonly called SAC105, with Ag composition of 1.0 wt. %. The time-temperature reflow profile was configured with a cooling rate of 1.0°C/s, and classified as "low cooling" and a cooling rate of 3.0°C/s, and classified as "fast cooling". The range of cooling rates considered were representative of the reflow soldering process for moderately complex printed circuit assembly. Furthermore, two solder joint sizes with diameters of 0.381 mm and 0.203 mm, and classified as "large" and "small" size respectively, were evaluated to study the effect of solder joint size on the solder joint microstructure.;ATC was performed from 0°C to 100°C in accordance with the IPC 9701 guideline using both 10 minute dwell time and an extended 60 minute dwell time that had been shown to promote microstructural coarsening in SAC alloys. All cells were tested to 63% or greater failure rate with a test duration exceeding 1500 cycles. Baseline characterization was performed on representative board-level assemblies to document the microstructures before ATC and to enable comparisons to samples removed from temperature cycling for failure analysis. Solder joint wetting conditions and fillet formations were evaluated using an optical microscope and two dimensional X-ray imaging. The solder joints were then cross sectioned and fine polished. The main techniques used for microstructural characterization by failure analysis were (i) optical microscopy under both bright field and cross polarized imaging conditions, (ii) scanning electron microscopy (SEM) using backscattered electron imaging (BEI), and (iii) energy dispersive spectroscopy (EDS) for elemental analysis. The ATC test data and failure analysis results were discussed in terms of the relationship to the evolving microstructure and fatigue behavior that resulted from temperature cycling.;The experimental results revealed that, at a 95% confidence level, dwell time at extreme temperatures during ATC testing to be the most critical factor that determined the number of cycles to fail, which corresponds to the fatigue resistance of the solder joints. Effects of cooling rate and Ag composition on fatigue life of SAC105 and SAC305 solder joints were found to be not significant. While the difference in microstructure as a consequence of cooling rate both for SAC305 and SAC105 solder joints was not significant, slow cooled SAC solder joints exhibited slightly better fatigue life than fast cooled SAC solder joints. At the 10 minute dwell time, the fatigue resistance of SAC305 was found to be slightly better than SAC105 solder joints. However, at the 60 minute dwell time, the fatigue resistance of both SAC105 and SAC305 was found to be nearly identical suggesting that the reliability of the solder joints might be insensitive to Ag content when the dwell time during ATC was longer. Furthermore, both SAC305 and SAC105 solder joints exhibited superior fatigue resistance than the eutectic Sn-Pb solder joints independent of cooling rate and dwell time. The fatigue resistance of the eutectic Sn-Pb solder joints was found to be significantly affected by both cooling rate and dwell time. At higher cooling rate and 10 minute dwell time, the solder joints exhibited better fatigue life while at slow cooling rate and 60 minute dwell time, the fatigue life of the eutectic Sn-Pb solder joints were found to have significantly inferior fatigue resistance. Solder joint size was found to have no effect on the reliability of solder joints. (Abstract shortened by UMI.)...