DISCOVERY OF NOVEL METAL-THIOSEMICARBAZONE COMPLEXES USING IN SILICO MODELS IN ENVIRONMENTAL ANALYSIS

Abstract

Twenty novel metal-thiosemicarbazone complexes were discovered by in silico techniques. The stability constants (logb₁₂) of complexes were also predicted by using the quantitative structure and property relationships (QSPR) models. The models were created using the multiple linear regression (MLR) and artificial neural network (ANN) approaches. The structure characteristics of complexes consist of molecular and quantum properties. The published literature is used to collect the stability constants with experimental parameters. The best model, MLR2-QSPR (k = 4), consisted of molecular descriptors such as S⁶, Dipole, xv1, and N’⁴. Statistical metrics such as R²_train = 0.913, Q²_LOO = 0.903, and SE = 0.408 were used to validate the quality of this MLR-QSPR. The statistical data for the ANN4-QSPR model I(4)-HL(10)-O(1) were also reported: R²_train = 0.972, Q²_test = 0.975, and R²_CV = 0.985. In addition, the work used the results of variables from the QSPR models for developing new thiosemicarbazone ligands and based-ligand complexes. As a result, novel metal-thiosemicarbazone complexes were newly outlined and predicted the stability constants by two developed QSPR models. The results obtained from models can be applied to develop novel chemicals that can be administrated for use in analytical chemistry and environmental evaluation monitoring.