As the electronics industry converts to lead-free solder worldwide, fundamental research on the physical and mechanical metallurgy (mechanisms of deformation, recovery, recrystallization) of tin (Sn) lags far behind its implementation. The substitution of Sn-based solders for Sn-Pb solder has led to failures in modes and locations never encountered before, and the fundamental cause of these failures is still not understood. Hence, lead-free solder joints lack a physically based (rational) foundation for reliability prediction. As the electronics infrastructure transitions to lead-free solder, high reliability products will become increasingly vulnerable. The research objectives of this program are to identify rules for operation of slip systems, hardening characteristics and recovery processes that can be implemented into crystal plasticity finite-element constitutive models. This analytical capability, combined with statistically large amounts of experimental characterization, will enable identification of fundamental microstructural evolution mechanisms and their interrelationship with dislocation generation and recovery in Sn based solders. A second major objective is to establish a database of observations on damaged and undamaged joints that also quantifies how locally active slip systems are correlated with crystal orientations, orientation gradients, grain boundaries, and recrystallization mechanisms so that interrelationships between these variables can be discovered. Success will be enabled by analytical capabilities present in the PI's group at MSU leveraged with Cisco's ability to provide high quality specimens manufactured in a repeatable industrially relevant process, augmented by the information gained in related research and development efforts. Insights gained will assist constitutive model development, and identify criteria that govern microstructural evolution and damage nucleation. These models will allow designers to evaluate worst-case microstructures in worst-case solder joint locations computationally, which cannot be done empirically.
NON-TECHNICAL SUMMARY: Historically, more than half the failures in electronic systems can be traced to solder joints. Because fundamental research on the metallurgy of tin (the basis for lead-free solder) lags far behind its implementation, this failure rate will increase as the worldwide electronics industry transitions to environmentally friendly lead-free solder. At present, manufacturing decisions are based upon costly empirical studies that are limited to the product and service conditions tested. In this project undergraduate and graduate students will work together with industrial partners at Cisco Systems, Inc., to develop specimens, experiments, and analyze data that will enable physically-based 3-D material models to be built. Students and the PIs will work at Cisco via internships, and 2-4 weeks faculty visits in summers to develop synergistic research thrusts at the interface between industrial product development and basic science. Existing collaborations with colleagues at Max-Planck-Insititut fur Eisenforschung in Dusseldorf will contribute to analysis and modeling capabilities. These results will be publicized in archival journals, conferences, workshops, and K-12 outreach to help the public gain appreciation for how materials engineering affects electronic system reliability, and hence, modern life (why do electronics quit working?). As models are developed, they will be evaluated by electronic system design engineers at Cisco in order to establish the ability to computationally predict the likelihood of damage in particular designs.
Thomas R. Bieler, Farhang Pourboghrat, Michigan State Univerisity Tae-kyu Lee, Cisco Systems, Inc. San Jose, CA One of the most critical locations in electronic assemblies are the solder joints that provide electrical and mechanical connectivity; it is not an exaggeration to state that these solder joints hold our entire electronic infrastructure together. Solder joints enable development of highly integrated devices using a relatively simple and economical assembly process. Figure 1 shows how miniaturization driven by mobile, transport, and extreme environments has led to lifetime decreases as the solder joint size and package size shrinks, where the lifetime is measured using a Weibull plot to identify a characteristic lifetime for a set of 10-20 packages subjected to accelerated thermal cycling. On the left side, the higher lifetimes for very small packages are good, but the size severely limits functionality. Weibull plots of package lifetimes show multiple slopes or data clusters that imply operation of multiple mechanisms that lead to failure. Current industrial practice requires expensive statistical qualification to close a trial and error package design iteration loop. Conventional finite element analysis is not able to predict microstructural evolution and failure, so existing modeling strategies cannot get beyond this packaging roadblock. The research objectives and intellectual merit of this program was to identify rules for operation of slip systems that account for deformation and hardening characteristics, microstructural recovery and recrystallization processes, and to identify relationships between locally active slip systems and microstructural evolution. From identifying these interrelationships, criteria for microstructural evolution and damage nucleation were identified that can be incorporated into multi-scale models. The primary focus was on Sn-3Ag-0.5Cu, the most commonly used solder alloy, with some comparisons to Sn-1Ag-0.5Cu. The property variability results from the complex geometry of a joint combined with the highly anisotropic thermal and mechanical properties of Sn (the thermal expansion of Sn in the c-axis direction is twice that of the a-axis direction), implying that the local thermomechanical stress–strain history for each joint is unique. This explains why failures in packages have been observed at any location in a package, not only in the most highly stressed corners, which is the case for traditional Sn-Pb joints. In Sn based joints, failure is correlated to the crystal orientation; joints with the c-axis nearly parallel to the interface are most vulnerable, due to being in tension at high temperature (Figure 2). In contrast, joint orientations with the c-axis nearly perpendicular to the interface show the least amount of cracking due to being in compression at high temperature. Damage begins near the package interface where stresses are high, starting with dislocation recovery processes that increase the misorientation of low angle boundaries by a continuous recrystallization process (cRx). At some point, primary recrystallization events (Rx) take place in the corners, and these highly misoriented grains grow preferentially into the cRx volume. Figure 2 shows that subgrain boundaries have higher strain energy (measured using differential aperture x-ray microcopy), that provides a driving force for the growth of primary Rx grains. This is also shown spatially in both the c-axis and local area misorientation (LAM) map in Figure 2. The recrystallized region in the LAM map is the gold area surrounded by the red line, where all of the grains have very small orientation gradients associated that go with a more perfect crystal. However, Rx grain boundary misorientations tend to be random, and their low cohesive strength makes these boundaries susceptible to cracking. This NSF grant has also led to the first identification of critical resolved shear stress values for slip of Sn in solder joints, based on measurements of slip activity in 32 shear deformed joints, and several single crystal solder joints deformed in tension. Figure 3 shows how the newly developed crystal plasticity model is able to predict the deformation and shape change of a crystal orientation that was not used to calibrate the model. This is the first model of its kind, and its potential to predict the variability that is commonly found in lead-free solder joints will have a broad impact, as it is developed further, and implemented into finite element codes used by package designers. This will enable prediction of variability, and enable rational design compromises before qualification experiments are started, which will save time and accelerate development of new electronic systems.
