As fossil fuel resources become increasingly depleted, it is quintessential that alternative sustainable sources of energy have been rigorously investigated. Hence, it is very desirable to widen the spectrum of energy resources available given increasing demand of energy in modern civilization. Among various energy resources, photovoltaic technology is one of the most promising candidates providing consistent and steady source of electrical energy.
In this study, a protocol was developed for exploring the vast chemical space of possible perovskites and screening the promising candidates. Using computational quantum mechanical principles (Density Functional Theory), the band-gaps of the candidates were computed to filter for perovskites with optimal solar conversion efficiencies. Machine learning algorithms, namely artificial neural networks, were utilized in order to assess the significance of input parameters that affect the band gap of the perovskites.