Predicting Early Failure of Quantum Cascade Lasers During Accelerated Burn-in Testing Using Machine Learning

Abstract

Device life time is a significant consideration in the cost of ownership of quantum cascade lasers (QCLs). The life time of QCLs beyond an initial burn-in period has been studied previously; however, little attention has been given to predicting premature device failure where the device fails within several hundred hours of operation. Here, we demonstrate how standard electrical and optical device measurements obtained during an accelerated burn-in process can be used in a simple support vector machine to predict premature failure with high confidence. For every QCL that fails, at least one of the measurements is classified as belonging to a device that will fail prematurely—as much as 205 hours before the actual failure of the device. Furthermore, for devices that are operational at the end of the burn-in process, the algorithm correctly classifies all the measurements. This work will influence future device analysis and could lead to insights on the physical mechanisms of premature failure in QCLs.