Genetic algorithm-assisted optimization of nanoporous TiO₂ for low-temperature processable photoanodes of...

Genetic algorithm (GA), a promising optimization process in Heuristics, has proven to be a powerful tool in controlling the nanostructure of low-temperature processable photoanodes in dye-sensitized solar cells (DSSC). For photoanodes that are composed of various sizes of TiO₂ nanoparticles and multiwalled carbon nanotubes in a double-layer configuration, the best composition was determined based on the objective functions of cell efficiency (η) and its variance. The latter function was chosen since TiO₂ dispersions with no organic binders often aggravated the efficiency of reproducibility. From a total of 64,536 cases, 24 different cases (6 samples prepared for each composition) per generation were selected, and their objective functions were compared. GA was effective in the optimization of photoanodes, and resulted in a cell efficiency of 7.3 ± 0.2% with a short circuit current of 13.8 ± 0.4 mA cm(-2), an open circuit voltage of 0.737 ± 0.006 V, and a fill factor of 71.8 ± 0.6% after 3 generations. The η of 7.3 ± 0.2% is the highest value for low-temperature processable dye-sensitized solar cells prepared without further treatment of TiO₂ films to enhance interparticle connections.