| Abstract [eng] |
Laser-induced damage threshold (LIDT) defines optical components. However, multiple pulse LIDT depends on number of pulses used for radiation due to optical fatigue effect. LIDT measurements are usually done by doing S-on-1 tests described in ISO 21254-2 standard. However, due to measurement time limitations S-on-1 tests are performed for small number of pulses, therefore, extrapolation model is used to predict LIDT value at fluence and pulse number of interest. Extrapolation model defined in ISO 21254-2 standard can only be applied for statistical fatigue effect, however, material fatigue usually dominates at longer radiations. So, in order to find proper LIDT value at large number of pulses one must find a model that defines material fatigue. Prognostics of fatigue effect on components used in other fields (e.g. electronics) are done by measuring distribution of time to failure and acquiring parametric models of distribution parameters. Therefore, the main goal of this work is to use methods used in other fields to predict optical components fatigue. In order to do that, lifetime distributions were investigated and new extrapolation models were found. Lifetime distributions of dielectric Al2O3 and metallic Au mirrors were measured for different pulse duration, different polarization and different fluence values. Then, measured lifetime distributions were approximated with logarithmic logistic model and parameters dependence on fluence were found. Lastly, cumulative distribution models for different fluence and number of pulses were created. It was shown that lifetime distribution σ parameter do not depend on fluence for both samples if samples are exposed to femtoscond pulses. However, if samples are exposed to nanosecond pulses σ parameter starts to follow exponential dependence on fluence which can be greatly influenced by pulse energy stability. Lifetime distribution parameter μ can be defined by power law dependence for all measured lifetime distributions. Large difference between femtosecond and nanosecond lifetime distributions are seen for Al2O3 sample, due to difference in material-light interaction strength. Different lifetime distributions might be a consequence of different optical damage source. Nonetheless, optical component lifetime can be extrapolated by using logarithmic logistic distribution model (if one damaging mode exists) or by mixing two logarithmic logistic distributions (if two damaging modes exist). So, in order to create prognostic model for optical component fatigue one must take into account pulse duration, light polarization and material that was used to manufacture component and in this work it was proven that laser damage fatigue can be modeled by using lifetime distribution parameters dependencies on fluence. |
