**2. Theory and computational details**

Theoretical calculations were performed in Gaussian16 (G16) programs suite [23]. Calculations include four solvents (water, ethanol, n-hexane, and methanol) in addition to the gas phase. Selection criteria are mainly based on how often the solvent is used in the laboratory to obtain pigments. The solvation model was PCM (polarizable continuum solvation model) as implemented in G16 program suite. B3LYP/6–311 + g (d,p) is the theoretical method used during geometry relaxation. Open-source databases were used to obtain the first geometry version and then our theoretical methodology was applied to optimize geometric parameters. Functional B3LYP is a widely accepted approach for this kind of molecule, and it was selected for this study mainly for that reason [24]. Basis set 6–311 + g(d,p) as implemented in the Gaussian16 program package [23] complements B3LYP very well according to preliminary calculations. 6–311 + g(d,p) was tested by running a set of calculations with different organic molecules with more than acceptable results. The literature considers B3LYP/ 6–311 + g(d,p) a theoretical method that provides a good level of accuracy for similar molecules [25–29]. A local minimum needs to be reached at the geometric optimization and it was confirmed with the calculation of harmonic vibrational frequencies. The zero-point vibrational energy (ZPVE) scaling was performed as a thermal correction (TC) at 298.15 K. Complementing geometry and frequency calculations, neutral energy, and adiabatic energies were obtained. Thereupon, chemical properties (HOMO, LUMO, gap, ionization potential (IP), electronic affinity (EA), electrophilicity (ω), electronegativity (χ), and hardness (η)) were computed based on the chemical reactivity indexes obtained in energy calculations. The sequence followed during the calculations was: first gas phase and then different solvents, one by one were included such as water, ethanol, n-hexane, and methanol.

Data reported by other research teams were included to compare with our results. So, a good idea is provided on performance against other theoretical methodologies or results obtained experimentally. Discussion is made on whether these molecules may be good dye sensitizers with TiO2 [25–29] for future work. Excited states were calculated (over 10 states) but only the first excited states will be discussed in this document. TDDFT calculations were carried on with B3LYP/6-311 g + (d,p) for consistency with energy calculations. Useful energy graphs and excited states spectra diagrams considering the six more common anthocyanidin variants in the same figure are included in this work for comparison of results under similar theoretical methods. Chemissian code [30] was used to develop most energy graphs included in the results section.
