Data mining of physical properties and non-linear optical parameters for materials enhancement

Abstract

Mid-infrared (IR) nonlinear optical (NLO) materials with high performance have a great importance for technological applications in many civil and military fields. However, it still a big challenge to discover new mid-IR NLO materials that achieve a good balance between the NLO effect and band gap. In this work, we study the correlation between the position of the atoms in the periodic table (period and group) and the band gap and nonlinear optical coefficients for the NLO compounds took advantage of the principal component analysis (PCA) .where 91 ternary and 141 quaternary compounds from experimental and DFT calculations data are selected. Based on these results and machine learning models, we employed partial least square (PLS) to predict the band gap of the NLO crystals. Since we get a linear relation that correlate the band gap and the position of the atoms in periodic table.