CLCC Solder Joint Life Prediction under Complex Temperature Cycling Loading, Michael Osterman, and Michael Pecht Center for Advanced Life Cycle Engineering (CALCE) University of Maryland College Park, MD 20742 Abstract Life time of electronic products is often limited by competing failure mechanisms brought on by the product s use and its surrounding environment. Solder interconnect failure is a known life limiting failure mechanism that is induced by repeated temperature excursions. Thermal fatigue reliability of solder is conventionally assessed by simple temperature cycling tests, which apply a constant temperature range, fixed dwell times and ramp rates during the test. From these tests, fatigue models and model constants have been developed to predict life expectancy of solder joints. However, simple temperature cycling tests do not necessarily provide a good representation of field use. In the field, electronic devices can experience more complex temperature cycle due to user controlled on and off cycles, non-constant workloads, and changes in surrounding environment. Under these conditions, electronic equipment does not experience constant dwell and ramp. Instead, varing temperature excursions often occur. To address concerns related to complex temperature cycles, an experiment was conducted and several modeling strategies were examined. The most effective modeling approach using the Engelmaier model was not the most effective modeling approach when using a partitioned strain energy model. 1. Introduction Very limited work in the past has been conducted on the modeling approaches for complex temperature cycles[1][2]. Lai et al [1] conducted a numerical study to analyze the thermal mechanical response of BGA solder joint under combined thermal and power cycling. Thermal profile under the combined loading was modeled first, and the thermal result was used as boundary condition of subsequent thermal mechanical analysis. The two step modeling approach gives more possibility of inaccuracy, compared with a direct modeling approach that uses measured temperature for modeling input. Darveaux s energy model was then used to calculate cycles to failure under the combined loading. It was found the involvement of power cycling reduces the fatigue life of the test vehicle by about 50% as compared to pure thermal cycling. Pei et. al [2] modeled solder joint creep energy under a complex temperature cycle loading, by adding creep energy of a primary cycle and a secondary cycle, which were two segments of the complex cycle. Both studies failed to provide any validation to predictions of cycles for failure and neither discussed the rational for how the complex cycle was partitioned. To address the issue of modeling complex temperature cycles, a physical test was conducted and the life expectancies of test samples were estimated using the Engelmaier and energy partitioning fatigue models.
In the modeling effort, various methods of segmenting the complex temperature cycle were examined. Comparisons between life predictions and life from test were used to identify the best modeling strategy 2. Test For this study, test assemblies with ceramic leadless chip carriers (CLCC) were used. Each test assembly had two 84 IO CLCCs and two 68 IO CLCCs. Dimension of 68 I/O and 84 I/O CLCCs are 24mm 24mm and 30mm 30mm respectively. The test boards are 2.3 mm thick and constructed of FR4 laminate. The test boards were assembled with Sn62Pb36Ag2 solder and there are two boards in test..once mounted on the test board, each CLCC part creates a resistance network that can be monitored for failure during an applied temperature cycle loading condition. Only the 8 corner terminals were included into the daisy chain for each part. Failure is defined as a maximum of 20% nominal resistance increase in the next 5 consecutive reading scans. The test was terminated when all components met the failure criteria [3]. The tested complex temperature cycle profile is depicted in Figure 1. The maximum cycle temperature is 75 o C and the minimum temperature is -25 o C. At the maximum cycle temperature dwell, six additional temperature cycles between 55 o C and 75 o C are applied. The overall cycle time was 110 minutes and the rate of temperature change was approximately 10 o C/min. Figure 1: Complex Test Profile The number of cycles to failure of the complex test was recorded for each part. N 50 of 84 I/O and 68 I/O CLCC solder joint are 254 and 315 cycles, respectively. Optical examination of the failed parts confirmed fatigue cracks in the solder resulted in the electrically detected failure. 3. Modeling For fatigue life modeling, both analytic and finite element modeling approaches were used. For the analytic approach, the Engelmaier model, documented in IPC9701 [3], was used. For the finite element modeling approach, the strain energy partitioning model proposed by Dasgupta et al [4] was used. For the strain energy partitioning approach, the total energy dissipation is partitioned into elastic, plastic and creep energy, each derived an individual fatigue life. Life expectancies corresponding to elastic, plastic and creep are superimposed by Miner s rule, so as to derive the fatigue life under the total energy.,
For the strain energy partitioning approach, a 3-D finite element model with solder, printed circuit board copper pad and component was built to provide the input for the energy model. Simulations were conducted to three cycles in order to make sure the FE analysis stabilized. Elastic, plastic and creep energies accumulated in the third cycle were used as the input of energy partitioning model. Figure 2: 3-D Finite Element Model The model is depicted in Figure 2. Elastic-plastic deformation was modeled by Ramberg-Osgood strain hardening rule, and creep was modeled by generalized Garofalo equation. The Ramberg-Osgood model is (1) in which the hardening exponent n p is 0.056 and strength coefficient K is 7025 MPa. The Garofalo model is: (2) where in C 1 = 6640, C 2 = 0.115, C 3 = 2.2, C 4 = 7130. To model the test condition, complex temperature cycle was segmented into a primary cycle (represents the effect of overall temperature range) and 6 secondary cycles (represent the effect of minor temperature fluctuation at the high end). Fatigue damage induced by the segmented cycles was defined as the inverse of the predicted life. The life expectancy of the complex cycle was determined by linearly superposition of damage imposed by the primary and secondary cycles.
The complex temperature cycle was modeling using two different segmenting approaches. Both approaches defined the secondary cycles to be from 55 o C to 75 o C with 5 minute dwells. The difference in segmenting was in the definition of the primary cycle. In method 1, the primary cycle was defined to be -25 o C to 55 o C with a dwell of 15 minutes at -25 o C and a dwell of 75 minutes at 55C. In method 2, the primary cycle was defined to be -25 o C to 65 o C with a dwell of 15 minutes at -25 o C and a dwell of 75 minutes at 65 o C The final step of modeling was to superimpose the primary and secondary cycles, in order to predict solder joint life under complex temperature cycling. For the Engelmaier model, all segmenting approaches underestimated the actual complex test result. Segmenting method 1 was found to provide a better correlation with the test results. Under method 1, the modeling underestimated the life of 84 I/O SnPbAg soldered CLCCs by 17%. For the energy partitioning approach, profile segmenting method 2 provided a closer prediction to the test results. Method 1 overestimated the life by 27 % and method 2 underestimated the prediction by 11% for SnPbAg assembled 84 I/O CLCCs. It is noted that the direct modeling of complex cycle, in which energy is evaluated by modeling the complex temperature profile, underestimated the tested cycles to failure by 28%. 4. Conclusions This study examined modeling approaches to predict the life of solder interconnects subjected to complex temperature cycling loading. Different profile segmenting methods were applied and results were compared with physical test results. Segmenting the complex temperature cycle into simple cycles, conducting failure estimations on the simple cycles, and applying a linear superposition strategy was used. For the Engelmaier model, defining the cycle with primary segment to have a maximum temperature corresponding to the low temperature of the secondary cycle was found to provide the best correlation to test data. For the energy partitioning model, defining the primary cycle to have a maximum temperature corresponding to the cyclic mean temperature of the secondary segment yielded the best results. Interesting, modeling the complex cycle using the FEA based energy partitioning model did not provide the best correlation with test data. Further, the optimum temperature profile segmentation differed between the Engelmaier and energy partitioning models. Reference 1. Y. Lai, T. Wang, C.Lee, Thermal mechanical coupling analysis for coupled power -and thermalcycling reliability of board-level electronic packages, IEEE Transactions on Device and Materials Reliability, Vol. 8, No. 1, Mar, 2008 2. M. Pei, X. Fan, P. Bhatti, Field condition reliability assessment for SnPb and SnAgCu solder joints in power cycling including mini cycles, Proceedings 56th Electronic Components and Technology Conference, v 2006, p 899-905, 2006 3. IPC Standard 9701
4. A. Dasgupta, Solder creep-fatigue analysis by an energy-partitioning approach American Society of Mechanical Engineers (Paper), p 1-8, 1991