Reliability estimation of BGA and CSP assemblies using degradation law model and technological parameters deviations

Very low failure rates, down to 10 FITs (one FIT is equivalent to one failure per 10(9) h x component), are now currently expected for electronics devices. To assess this challenge even for complex assemblies, process control, design and technological solutions must be optimised to guarantee almost no defect during the operating life. Reliability evaluation must also evolve from classical accelerated tests, towards the use of more sophisticated laws based on physics of failure and process parameters. We present results from BGA and CSP thermocycling measurements where expected cumulative failure distributions have been modelled using computed degradation law and process dispersion data. Starting points are: An already validated evolution law models the observed degradations on test samples of two different assembly technologies. Numerous physical simulations of technological attributes deviations, using process data are performed. Statistical computations using the previous results lead to a theoretical lifetime distribution under various stress conditions. The extrapolated results related to reliability functions are compared to experimental tests. (C) 2003 Elsevier Ltd. All rights reserved.