Machine Learning for Analysis of Time-Resolved Luminescence Data
journal contributionposted on 12.11.2018, 00:00 by Nikola Đorđević, Joseph S. Beckwith, Maksym Yarema, Olesya Yarema, Arnulf Rosspeintner, Nuri Yazdani, Juerg Leuthold, Eric Vauthey, Vanessa Wood
Time-resolved photoluminescence is one of the most standard techniques to understand and systematically optimize the performance of optical materials and optoelectronic devices. Here, we present a machine learning code to analyze time-resolved photoluminescence data and determine the decay rate distribution of an arbitrary emitter without any a priori assumptions. To demonstrate and validate our approach, we analyze computer-generated time-resolved photoluminescence data sets and show its benefits for studying the photoluminescence of novel semiconductor nanocrystals (quantum dots), where it quickly provides insight into the possible physical mechanisms of luminescence without the need for educated guessing and fitting.