The intensive research on perovskite solar cells (PSC) in the last years has allowed the technology to reach exceptional results, with a record power conversion efficiency (PCE) of 25.5% [1] for single junction cells. Due to low temperature processing, band gap tunability and flexible composition, this class of materials is also highly employed in multijunction, or tandem, photovoltaic applications. With so many demonstrated advantages, one of the remaining challenges for the industrialization and commercialization of the technology is the upscaling of fabrication of perovskite films. Large-area deposition methods such as (ultrasonic) spray coating, inkjet printing, slot dye coating, and blade coating provide great avenues for high throughput fabrication, but often come with a significant decrease in PCE when compared to spin coated devices [2].

This study proposes a design of experiment approach, aiming to optimize the fabrication of perovskite thin films by ultrasonic spray coating. Initially the entire process was reviewed, from substrate cleaning and solution preparation to deposition and post treatment of the film, and key parameters were drawn out. Based on this analysis, and previous studies [3], experiments were designed to gain further insight on two important steps of the process: the deposition of a wet layer of precursor solution by ultrasonic spray coating, and the gas-quenching of the solvent for perovskite crystallization.

The deposition step was examined by a one factor at a time (OFAT) approach, with the establishment of a set of standard parameters and the variation and analysis of each parameter individually. By doing so it was possible to verify that the spray coating parameter are paramount to the formation of a homogeneous wet layer and good coverage of the substrate, yet the fine tuning of the perovskite crystallization, and consequentially the layer quality, depends mostly on the evaporation step.

The gas quenching-assisted evaporation was then thoroughly studied by a full factorial analysis of three key parameters: gas (nitrogen) pressure, distance between nitrogen gun and substrate, and solvent volatility, determined by the ratio between a low volatile and a high volatile solvent. The produced films were then characterized in regard to their morphology and performance as the active layer of a solar cell.

This full factorial analysis provided insight on the complex mechanism of perovskite crystallization in a thin film. It also allowed the assessment of interactions between the parameters and shed light on how they influence the evaporation of solvent in different regimes. Moreover, the use of design of experiments resulted in higher efficiencies than the previously trial-and-error approach, with the best performing cell reaching a PCE of 18.1%, which is in line with spin coated perovskite solar cells with similar architecture, i.e. nickel oxide (NiOx) as hole transport layer, no additives on the perovskite solution, and no anti-reflection or passivation layers.