Predictions of thermal cycling life of PBGA microprocessor components with variation of mechanical properties of the packaging materials

Abstract

Plastic Ball Grid Array (PBGA) is one of the most popular microprocessor components used in today’s electronic industries. The primary objective is to transmit electronic signals from the chip to the PCB of the circuit. The package is made of several components with different mechanical properties. When PBGA is subjected to harsh environments, such as temperature fluctuation in different applications, thermal stress develops between package components due to mismatch in co-efficient of thermal expansions (CTE). Again, the processors operate at high BUS speed and generates high amount of heat. This heat should be dissipated to the surroundings in a short time. But if the heat is not properly dissipated, thermal stress develops in the IC components. As a result, solder joints between package substrate and PCB fails after a certain number of thermal cycles leading to interruption of I/O signal transmission. Number of thermal cycles upto failure is termed as thermal cycling life of the package. In the study, quarter symmetry finite element model of a PBGA package is developed to predict the location of critical solder ball with maximum damage accumulation during thermal cycling. Life prediction of the package has been made based on the failure of the critical solder ball. The same model is also used to investigate the effects of mechanical properties of some of the major component materials on the life of the package and to predict the thermal cycling life of PBGA package for different mechanical properties of its components. Thermal cycling life has been found as strong functions of mechanical properties of substrate materials, solder materials and molding compounds. Using the results from the analysis, maximum life or reliability of PBGA package has been predicted using optimum mechanical properties of the components. Finite element results have been validated using available data from experiments and numerical study in literature. ABSTRACT