Reichardt’s Dye-Based Solvent Polarity and Abraham Solvent Parameters: Examining Correlations and Predictive Modeling

The concept of “solvent polarity” is widely used to explain the effects using different solvents in various scientific applications. However, a consensus regarding its definition and quantitative measure still lacking, hindering progress solvent-based research. This study hopes add conversation by presenting development two linear regression models for solvent polarity, based on Reichardt’s ET(30) polarity scale, Abraham parameters transformer-based model predicting directly from molecular structure. first incorporates standard descriptors s, a, b, extended ionic j+ j−, achieving impressive test-set statistics R2 = 0.940 (coefficient determination), MAE 0.037 (mean absolute error), RMSE 0.050 (Root-Mean-Square Error). second model, covering more extensive chemical space but only achieves 0.842, 0.085, 0.104. applicable any with an associated SMILES string, 0.824, 0.066, 0.095. Our findings highlight significance parameters, especially dipolarity/polarizability, hydrogen-bond acidity/basicity, descriptors, polarity. These offer valuable insights researchers interested parameter general.