Validation of different SAC305 material models calibrated on isothermal tests using in-situ TMF measurement of thermally induced shear load

In the past, a large number of material models for Sn-based solder alloys have been proposed, which are usually calibrated based on the material testing under isothermal conditions. However, their ability to map the lifetime differences depending on the temperature rate under field and test-lab conditions, as well as on the mean operating temperature, is still not completely investigated and validated. The novel thermo-mechanical fatigue (TMF) measurement set-up described in Metasch et al. (2017, 2014) is employed for in-situ measurement of the material degradation driven by temperature cycles. The experimental system involves different materials, which impose thermally induced displacements onto the solder connections. The acceleration of test duration can be controlled by placing the sample in loading positions with different level of thermally induced displacement. The measurement enables monitoring of the force-reduction and the concurrent change of displacement. In the current study, the samples comprising a real-scale geometry of the four Ball Grid Array (BGA) connections were stressed with the temperature cycles relevant for typical lab-tests and field conditions. The level of the thermally induced shear displacement in the solder joints was significantly higher than in an Engine Control Unit ECU (as investigated in Schafet et al., 2012). Since the experimental set-up includes various geometrical and material features, an extensive FE-based sensitivity study has been performed. The simulation of the free-expanding system as well as of the system with different pre-characterized dummy samples (without solder joints) revealed the capabilities and specific mechanical behavior of the experimental set-up. Finally, for Sn96.5Ag3.0Cu0.5 solder alloy the ability of the different material formulations to reproduce trends of measured force-displacement hysteresis was analyzed: for double power-law creep model (DPL), unified inelastic strain formulation by Anand, and unified visco-plastic model proposed by Chaboche. Their accuracies in predicting of the acceleration factor between the different temperature profiles are summarized and discussed.