Evaluation of Quantum Chemical Methods and Basis Sets Applied in the Molecular Modeling of Artemisinin

Abstract

In this paper, we evaluate semiempirical methods (AM1, PM3, and ZINDO), HF and DFT (B3LYP) in different basis sets to determine which method best describes the sign and magnitude of the geometrical parameters of artemisinin in the region of the endoperoxide ring compared to crystallographic data. We also classify these methods using statistical analysis. The results of PCA were based on three main components, explaining 98.0539% of the total variance, for the geometrical parameters C3O13, O1O2C3, O13C12C12a, and O2C3O13C12. The DFT method (B3LYP) corresponded well with the experimental data in the hierarchical cluster analysis (HCA). The experimental and theoretical angles were analyzed by simple linear regression, and statistical parameters (correlation coefficients, significance, and predictability) were evaluated to determine the accuracy of the calculations. The statistical analysis exhibited a good correlation and high predictive power for the DFT (B3LYP) method in the 6-31G** basis set.