Risk Assessment Methodology for Lead-Free Solder Assembly

摘要

Electronics assembly is being disrupted in large industrial concerns as well as engineering prototype laboratories by the rapid transition to lead-free assembly. At the same time, electronics in military systems is becoming more complex, highly miniaturized and subject to ever harsher treatment. For example, inertial guidance systems are being deployed in many gun launched projectiles, all types of autonomous vehicles are becoming smaller and more capable, and hand carried navigation and communications equipment is critical to the effectiveness of deployed soldiers. High reliability is of paramount concern in all of these applications. rnAs part of our effort to develop and qualify a lead-free assembly process, we are designing a software tool to estimate the failure rate of component assemblies. Our initial effort is focused on temperature cycle induced failures, but our intent is to extend the tool capabilities to include shock and vibration stresses as well. We envision two primary applications, the first of which is to estimate the failure rate of individual devices and aggregates of components for designing systems to meet reliability requirements. The second application is to design experiments for demonstrating the reliability of component assemblies or fabrication processes. rnOur approach to estimating failure rates is to blend accumulated energy density wear out models with Weibull distribution parameters, which have been measured for selected cases. We have adopted an object oriented programming methodology and are implementing the software in the Java programming language. Measured attributes are assigned to component and circuit board objects. Solder joint properties are determined from a combination of measured attributes, such as initial composition and volume, component and circuit board finishes, and estimates of metallurgical changes due to base metal dissolution and intermetallic growth. These properties are used to calculate coefficients for a fatigue model, which is used to estimate mean cycles to failure. A Weibull distribution is applied to project the time dependant failure rate.