Statistical Method and Non-Destructive Analytical Tools in the Failure Analysis of LED Array

Abstract

When several units failed during operation or reliability test, and their failure modes are the same, it is difficult to examine if there are multiple failure mechanisms unless destructive analysis is performed on every units. However, this can be very time and resource consuming. This work demonstrates a method that combine statistical method and non-destructive analytical tools to determine various failure mechanisms where their failure modes are the same. The sample examined in this work is an array of LEDs assembly which consists of 20 lED chips. The sample is subjected to a standard 85 °C and 85%RH test for 1000 h, and all of them have the same failure mode of high forward voltage. We determine the number of possible degradation mechanisms using statistical method and find that there are three different mechanisms. Series resistance increase is found to be the dominant mechanism behind this high forward voltage rise. Non-destructive tools are used to confirm the three different mechanisms and identify their underlying physics.