Reliability evaluation of a novel metal oxide-aluminum glycerol film capacitor using nonlinear degradation modeling with dependency considerations

Abstract

Capacitors are components predominantly used to stabilize voltage, store energy, and lower electrical ripples. To improve the flexibility and capacitance of capacitors, we were motivated to develop hybrid film capacitors using two emerging thin film techniques, atomic and molecular layer deposition (ALD and MLD). Validation of its operational reliability in a power module or an electrical system is critical to facilitate its adoption. Equivalent series resistance (ESR) and capacitance are key performance characteristics (PC) of a capacitor, whose degradation path and process are nonlinear and dependent. Few studies have implemented the Wiener process (WP), to capture the dependency between the PC, based on the assumption of a normally distributed performance loss, notwithstanding that the performance loss may not be normally distributed. To address this concern as well as improve the accuracy of reliability prediction, a reliability framework is proposed. The gamma distribution is found to better fit the incremental PC degradation used in this work. The derived reliability model captured the non-linearity in the PC degradation path as well as its dependency using a selected Copula function. Akaike information criterion (AIC) was used to determine the most suitable Copula. Reliability estimation based on dependency consideration shows the least mean square error (MSE) compared to reliability estimate using a model that considers only one PC or assumes independent PCs. In addition, the hybrid capacitor was compared with an aluminum oxide dielectric layer capacitor, and the result was discussed.