| Abstract [eng] |
The decrease of laser-induced damage threshold (LIDT) of materials when irradiated with multiple laser pulses, the fatigue effect, is observed in most types of solids. However, the physical nature of fatigue and its extrapolation is still an active area of research. We are generally no longer limited by the temporal resolution of time-resolved experiments used to investigate fatigue, but are often faced with the dilemma of how to properly interpret the experimental data and apply the findings in practice. Therefore, the aim of this thesis was to merge time-resolved and statistical laser-induced damage fatigue research techniques in order to improve lifetime prediction of dielectric coatings irradiated with femtosecond laser pulses. The thesis investigates existence of catastrophic and color change damage modes using unsupervised machine learning and identifies their unique fatigue behavior. A separate chapter is dedicated to the quantitative investigation of catastrophic laser-induced damage mode fatigue. A link was established between LIDT fatigue and mechanical fatigue crack growth which allowed construction of a fatigue model constrained by both S-on-1 and time-resolved experiments. The last chapter of the thesis is dedicated to the metrology of laser-induced damage fatigue. An improved analysis technique based on Bayesian inference is introduced which accounts for both complete and censored test data. |
