Cylindrical microlasers: emission research and machine learning-assisted analysis between ASE and lasing phenomena

Abstract Miniature polymeric cylindrical lasers, fabricated using SU-8 photoresist doped with rhodamine B at a fixed concentration of 0.5%wt, can generate amplified spontaneous emission, optical lasing, and random lasing (RL). The tunability and identification between these is a complex challenge, primarily stemming from their geometric properties and experimental conditions, including operating thresholds and the emitted intensity behaviors. We utilize machine learning (ML) algorithms to intelligently analyze data and identify the most critical parameters for classifying cavities. Our algorithm successfully predicted an unusual lasing behavior characteristic of the RL phenomenon, which is not part of the training dataset, highlighting the robustness and accuracy of the ML classifier in precisely identifying and differentiating between the various emission mechanisms in our microlasers. Additionally, we investigated the radiation characteristics of our cavities from two perspectives: internally, by analyzing the individual emission pattern as the sum of multiple individual whispering-gallery modes sustained within the microcavity, and externally, at the far-field distance, modeling as an emission pattern produced by multiple point sources arranged along the cavity edges. Our results facilitate the development and in-depth understanding of a new generation of polymeric miniature lasers, revealing novel perspectives to analyze gain emission mechanisms by harnessing the potential of ML towards a deep comprehension of their underlying phenomena.