The Use of Physics-of-Failure Analysis to Predict the Reliability of Semiconductor Devices 1999-01-0163
The automotive electronics market has seen and will continue to see unimpeded growth due to the substitution of mechanical and electromechanical devices with electrical devices wherever feasible for increased reliability. In addition, automakers are increasingly looking to incorporate advanced electronics technology into their vehicles to satisfy customer demands for more and innovative features. Examples of this are the use of global positioning system (GPS) for directions and roadside assistance and increased integration of the engine and powertrain to provide smoother, more fuel-efficient operation. Despite this growth, however, the automotive electronics market continues to shrink as a percentage of the total market due to the phenomenal growth of the computer and telecommunications markets. Combined with the shrinkage of the military and aerospace electronics markets, electronics qualifications for customers are increasingly driven by markets whose usage requirements (e.g. environmental extremes, power consumption) are more benign than that seen in the automotive market. Thus, electronics manufacturers are increasingly reluctant to run accelerated qualification tests and screens for automotive customers unless they are reimbursed for the added effort. This drives up costs to the automotive market. In addition, the outgoing quality and reliability of electronics is improving to the point where potential defects are increasingly difficult to observe during standard environmental stress testing. A simulation tool is needed that can estimate the failure rate of a device using a particular design, set of materials and processes in order to enable the customer to intelligently assess the risk of failure to a given subassembly. By using a Physics-of-Failure approach and working with the Computer Aided Life Cycle Engineering/Electronic Packaging Research Center (CALCE/EPRC) at the University of Maryland, we are developing such a software tool. Using material and design parameters of the device, reliability estimates based on models developed in the industry can then be obtained and used to assess the inherent risk to the subassembly without environmental stress testing. The customer can, from this assessment, decide if and how much more further testing may be required to satisfy their end users that their product meets their quality and reliability requirements. In this article, we will describe how the models work, along with examples relating their results with what is seen in real life situations. Anticipated savings in cost will be reviewed along with caveats that must be understood in order to prevent the incorrect use of these models.