Halide organic-inorganic perovskites (HOPs) offer a wide range of bandgaps and low-cost fabrication making them an ideal candidate for tandem solar cells. Perovskite-perovskite based tandem solar cells have exceeded efficiencies of 23% [1] for the four-terminal (4T) and of 18% [2,3] for the two-terminal configuration. The maximum theoretical efficiency limit for the two cell tandem structure is about 42% [4]. 

It is well known that both the bandgap and the thickness of the individual cells must be optimized to achieve maximum efficiencies for a tandem cell. Furthermore, it is important to understand charge carrier generation, transport and recombination within the individual cells as well as when they are combined. Numerical simulations based on finite element drift-diffusion methods have been successfully used to explore various phenomena in single junction perovskite solar cells. However, there is lack of studies on how the fundamental phenomena of the single subcells as well as the combining layers impact the final performance of perovskite in tandem. Also, so far only few combinations of perovskite materials have been experimentally used.

In the present work we investigate the performance of perovskite-perovskite based two terminal (2T) tandem solar cells using finite element drift diffusion calculations. We investigate the effect of bulk and interface recombination, mobility, electrode work function and doping of the transport layers, on the final performance of the tandem solar cell. Based on the parametrized single junction cells, we study how the final performance of the 2T tandem device is affected with the changes in material bandgaps and thicknesses of the individual subcells. The subcells are coupled optically using the outcoupled light from the top cell as an input for the bottom cell. The electrical connection is achieved by considering a series circuit configuration of the two independent subcells [5]. The study aims to select the most efficient combination.

Moreover, thicknesses can be easily optimized to achieve maximum output, but it is always hard to find practical materials especially low bandgap perovskites, corresponding to the optimized efficiency. Recently we have used a Machine learning (ML) approach as a powerful tool to predict hidden trends for optimal bandgap based on available data.  Thus, the combination between DFT, ML and drift-diffusion can indicate (i) the optimal parameters for the tandem and (ii) the selected materials.