A Machine Learning Study to Explore the Structural Basis of Non‐Conjugated Compounds for Their Optical Activity Features

Abstract Various investigations have been aimed to elucidate that why certain molecules exhibit an enhanced nonlinear optical ( NLO ) response than others. Such kind of knowledge is advantageous to design of new NLO switches where their ON / OFF states can display their NLO behavior. Based on the significance of nonconjugated crystal systems, the current study aims at identifying the driving forces of the polarizability/first hyperpolarizability through machine learning ( ML ) analysis by growing crystal of 2,4‐diaminotrotriazole. The study shows that how input of a Simplified Molecular Line Entry System ( SMILES ) of compound into a large polarizability/hyperpolarizability data can display its NLO response. The results show a good correlation between their polarizability and the HOMO–LUMO energy gaps. Their SHapley Additive exPlanations ( SHAP ) analysis reveals that the transition dipole moment () between the ground and first excited state is one of the primary contributors for such molecular systems. It is also illustrated that, besides various non‐conjugated related redox states, the ML model can adequately characterize these NLO responses.