Perovskite solar cells currently enter a stage, where market introduction is within reach. For serious upscaling, control over the quality of the perovskite material is most critical. To this end, the community currently relies heavily on laboratory experience, engineering, and fine-tuning approaches. A key challenge is the controlled growth of perovskite crystallites while processing thin films. Several strategies are currently in use, such as like anti-solvent or additive engineering. While the impact of these strategies is well-evidenced in the resulting layers, the underlying mechanisms that govern the crystallization process are still subject to a vigorous debate. A frequently cited theory is that the nuclei for the perovskite crystallization evolve from intermediate Pb²⁺-MA⁺-I^--solvate clusters. [1] A dominant role of solvate clusters implies a strong impact of complexing and coordination in the precursor solution on the crystallization process. Reports of the colloidal structures and Lewis-base-acid interactions in the perovskite precursor ink seemingly support this theory. [2] As of yet, however, insights that unambiguously link the complex formation in the precursor inks to the perovskite formation are lacking.

We present a holistic approach in which we study the entire course of perovskite formation. We begin with as study of lead complexation in the precursor stage (using NMR and electrical conductivity measurements in the solution) and proceed with in-situ GIWAXS investigations during thin film deposition and along the way to thin film formation (bare layers and solar cells). We systematically study the impact of common solvents like DMF, DMSO and NMP. As an exemplary additive, to study the influence of Lewis-base additives, we chose thiourea, which is a strong sulphur donor and an effective crystallization mediator. With ²⁰⁷Pb NMR and conductivity studies, we found a strong and systematic impact of the choice of solvents on the formation of lead complexes in the precursor solution, as well as indications, that suggest the presence of 3D corner sharing structures already in the precursor ink. Importantly, the differences in lead complexation depending on the solvent apparently diminish with increasing the concentration of the precursor ink; the final grain sizes remained largely unaffected by even strong variations found in the diluted precursors. On the other hand, the addition of thiourea did not affect the nature of lead complexes in the precursor solution. By in-situ GIWAXS, we are finally able to identify the annealing step as the decisive stage, where the presence of the additive affects the formation of perovskite grains and their crystallographic orientation. We could further substantiate our interpretation by FTIR studies. As such, for the first time, we provide a convincing link (or the lack thereof) between precursor chemistry and final thin film formation.