Polyphenolic compounds are now widely studied using computational chemistry approaches, the most popular of which is Density Functional Theory. To ease this process, it is critical to identify the optimal level of theory in terms of both accuracy and resource usage-a challenge we tackle in this study. Eleven DFT functionals with varied Hartree-Fock exchange values, both global and range-separated hybrids, were combined with 14 differently augmented basis sets to calculate the reactivity indices of caffeic acid, a phenolic acid representative, and compare them to experimental data or a high-level of theory outcome. Aside from the main course, a validation of the widely used Janak's theorem in the establishment of vertical ionization potential and vertical electron affinity was evaluated. To investigate what influences the values of the properties under consideration, linear regression models were developed and thoroughly discussed. The results were utilized to compute the scores, which let us determine the best and worst combinations and make broad suggestions on the final option. The study demonstrates that M06-2X/6-311G(d,p) is the best fit for such research, and, curiously, it is not necessarily essential to include a diffuse function to produce satisfactory results.
Keywords: Hartree–Fock exchange; Janak’s theorem; basis set; benchmark; caffeic acid; computational chemistry; density functional theory; polyphenols; theoretical chemistry.