Accurate and informative analysis of positron annihilation lifetime spectra by using Markov Chain Monte-Carlo Bayesian inference method

Abstract

Abstract Positron annihilation lifetime (PAL) spectroscopy is one of the most powerful methods to quantitatively characterize atomic-scaled lattice imperfections in condensed materials. Generally, one needs to fit a PAL spectrum by solving a local non-linear optimization problem which is more or less affected by initial guesses. Therefore, by using a traditional analysis program for the same PAL spectrum, different results with close goodnesses of fits are yielded by different researchers. Thus, it is difficult to judge the qualities of different results. To overcome this shortage, an efficient Markov Chain Monte-Carlo Bayesian Inference (MCMC-BI) method based on CosmoMC package is applied to analyze both simulated and experimental PAL spectra in the present work. The same level of accuracy of traditional analysis programs is firstly acquired in this work by using MCMC-BI method, which demonstrates that it can be directly used to analyze PAL spectra. Furthermore, the dependence on the initial guesses of PAL analysis is significantly alleviated. Additionally, more precious information is provided by MCMC-BI method, including different lifetime uncertainties in different confidence intervals and the correlations between annihilation parameters.