Simulation of optical emission spectra of high temperature plasmas for deep learning studies in laser-induced breakdown spectroscopy

Laser Induced Breakdown Spectroscopy (LIBS) is an analytical method rooted in Atomic Emission Spectroscopy (AES). LIBS employs a short, intense laser pulse directed onto the sample's surface, generating a micro-plasma. The optical emissions from this laser produced plasma are then analyzed to ascertain both the elemental composition and concentrations of the sample. Qualitative and quantitative analysis using LIBS is tedious with conventional approaches. Over the past decade, LIBS elemental analysis integrating with machine learning algorithms have grown significantly. Among the conventional machine learning algorithms, Deep learning Neural Network (DNN) coupled LIBS is a promising analytical tool developed for the efficient compositional analysis. The simulation of optical emission spectra at laserproduced plasma conditions (Te = 1 eV, Ne = 10¹⁷ cm⁻³) allows obtaining synthetic spectra for training the DNN model for different concentration of elements for a range of plasma electron temperature and density. In this work, we have proposed a computational algorithm for simulating the optical emission spectra of different elements in the periodic table and thereby generating datasets (spectrum) needed for training the deep learning neural network models for elemental analysis. This study suggests that employing DNN-supported LIBS is a promising analytical tool for multi-elemental compositional analysis.