Genetic Algorithm-assisted optimization of partially dyed-TiO2 for room-temperature printable photoanodes of dye-sensitized solar cells

Abstract

The processing parameters for photoanodes that are printable in a room-temperature (RT) continuous process for dye-sensitized solar cells (DSCs) were optimized by the use of Genetic Algorithm (GA). The photoanodes were prepared at RT from mixtures of partially dyed-TiO₂ (PDT) with various sizes and dye-loadings and were subsequently compressed for the fabrication of PDT films with different compositions, nanoporous structures, and thicknesses. According to our decision parameter design, there were 2²⁰ different cases, from which only 30 chromosomes for each generation were selected for the testing of photovoltaic performance to determine the global optimal point in power conversion efficiency (η). After 7 generations (i.e., only 210 chromosomes tested out of 2²⁰), η reached 6.02%, which was 22% higher than the value reported previously without the aid of a GA. The photoanode composed of a mixture of 14, 21, and 40 nm TiO₂ with different dye-loadings and compressed to a thickness of 8.0 μm under 94 MPa showed the highest η (short circuit current = 12.03 mA cm⁻², open circuit voltage = 685 mV, fill factor = 72.8%). A comparison of the photovoltaic parameters and characteristic charge transport properties revealed that sparsely dyed small TiO₂ provided an efficient electron transport route to increase the short circuit current. The role of large TiO₂ dyed up to monolayer coverage, on the other hand, was also disclosed to partially increase the short circuit current by light scattering. We hope that the GA-assisted photoanode optimization, presented here, will encourage new directions in research while providing an efficient RT printing process for the continuous fabrication of DSCs.