Analysis of lifetime distributions of dielectric and metallic mirrors

Abstract

When discussing laser-induced damage phenomena, the main parameter of interest is the laser-induced damage threshold (LIDT). Since LIDT is a function of irradiation time (or number of pulses), its characterization is of critical importance when designing reliable industrial or medical laser systems or even planning long-term space missions involving high power lasers. Within laser-induced damage community, decrease of LIDT with increase in irradiation time (the so-called fatigue effect) is often estimated by using the S-on-1 test procedure described in the ISO 21254-2 standard. However, due to measurement limitations, S-on-1 tests are usually carried out for relatively small numbers of pulses, therefore additional extrapolation methods must be used in order to predict lifetime of optical components. The simple extrapolation procedure provided by the ISO 21254-2 standard no longer meets the demands of the community as it ignores existence of multiple failure modes and data censoring resulting from different damage detection techniques, therefore a new approach for LIDT extrapolation is much needed. In our previous work [1], we explored application of accelerated lifetime testing (ALT) techniques on S-on-1 test data. ALT approach treats S-on-1 experiments as sets of lifetime (or time to failure) distributions at different fluence levels instead of damage probability curves at different numbers of pulses. We have shown that this approach, combined with Bayesian inference and Markov chain Monte Carlo (MCMC) sampling, is well suited for extrapolation and uncertainty evaluation of S-on-1 experiments. However, proper application of these methods require empirical knowledge about lifetime distributions for different material types and irradiation parameters. Therefore, in this work an attempt is made to experimentally characterize lifetime distributions of highly reflective dielectric and metallic laser optics at both nanosecond and femtosecond pulse durations. UV and IR wavelengths as well as influence of irradiation fluence and polarization state are also considered. The results of this study provided insights into lifetime distributions of laser-induced damage and helped to shape guidelines for extrapolating S-on-1 experiments to longer irradiations. [1] L. Smalakys, A. Melninkaitis, Predicting lifetime of optical components with Bayesian inference, Opt. Express 29, 903-915 (2021).