Hybrid perovskite photovoltaics (PVs) are poised to impact future’s renewable energy production and contribute significantly to the reduction of global CO2 emissions. Power conversion efficiencies (PCEs) of perovskite solar cells (PSCs) were increased almost up to the level of silicon PV, while stable power outputs of over 10,000h of operation were demonstrated. Further, the bandgap of hybrid perovskites can be tuned to match optimal values for tandem PVs. However, the scalability of the technology still falls behind the industry standard for concurrent PV technologies, that is the PCE of perovskite modules falls off steeply with increasing aperture area.

There are two fundamentally different routes for fabricating perovskite thin-films, coating of solution films that are subsequently dried and crystallized or deposition from a vapor at low pressures. While solution coating, for example by spray coating or slot-die coating, is more cost-effective due to its operation at atmospheric pressure and its compatibility with roll-to-roll machines, trials on vacuum evaporation indicate superior conformity and adhesion on small- to medium-sized substrates, which are important characteristics when depositing on rigid, fixed-size silicon solar cells for tandem application. However, due its high economical prospects, solution processing on small to medium-sized substrates via intermittent coating should not be discarded per se.

In this work, we will investigate slot-die coating on medium-sized substrates (25cm-300cm) from a practical and theoretical viewpoint. The main challenge for this process arises from two aspects: 1) The difficulty to achieve homogeneous coating over the whole substrate length and 2) the difficulty to dry the precursor solution in homogeneous way over the limited substrate length. We first address the latter issue by making use of state-of-the-art drying models in a simulative sandbox. In particular, we show that Bayesian optimization can find effective parameters for drying of perovskite precursor films on medium substrate sizes. Furthermore, we show that by making use of pre-existing knowledge of the coated thickness inhomogeneity from ex-situ or in-situ measurements, the same algorithms can be exploited for drying films with an inhomogeneous thickness distribution in an optimal way such that the perovskite crystallization occurs a the optimal time and position during the drying. We further outline the interdependencies of the different coating parameters and demonstrate how these interdependencies can be exploited for achieving more homogenous perovskite coatings. In a final perspective, we showcase how these insights can be leveraged in automatically controlled perovskite coaters and dryers.