Studies on methods for evaluating and selecting photochemical compounds for Dye-Sensitized Solar Cells
In response to the growing demand for renewable and sustainable energy sources, this thesis investigates computational methods for evaluating and selecting photochemical compounds with potential application in Dye-Sensitized Solar Cells (DSSCs). The research is structured into two main fronts: (i) the photophysical characterization of curcumin and its metal complexes with Cu(II) and Zn(II), using electronic structure methods based on Density Functional Theory (DFT), Time-Dependent DFT (TD-DFT), and wavefunction-based methods (NEVPT2 and LCCSD(T)); and (ii) the development of machine learning models for large-scale dye screening from molecular databases, focusing on the prediction of the first electronic transition energy. The results indicate that curcumin-metal complexes exhibit promising photophysical properties for light-harvesting applications, particularly the curcumin-copper(II) complex, which shows distinct charge-transfer transitions (MLCT and LMCT). Additionally, supervised learning models trained on millions of molecules from the PubChemQC database, using diverse molecular descriptors, demonstrated performance comparable to traditional quantum chemistry methods, offering an efficient alternative for large-scale molecular screening. This thesis contributes to the advancement of computational chemistry applied to energy materials by proposing an integrated approach that combines quantum mechanical methods and artificial intelligence to accelerate the discovery of new photosensitive compounds.