Metal Halide perovskites (MHPs) emerged as promising materials for photovoltaics and other technological applications. MHPs have an ABX₃ composition: one A⁺ ion type, an organic or inorganic monovalent cation – e.g., CH₃NH₃⁺, (NH₂)₂CH⁺, Cs⁺ - per B²⁺ type cation - typically Pb²⁺ or Sn²⁺ - and three X^- ions - I^-, Br^- or Cl^-. These MHPs are characterized by their structure, where BX₆ corner sharing octahedra form a sort of 3D checkerboard with “white” boxes occupied by A⁺ ions (Figure 1). Due to the presence of heavy B²⁺ type cation and iodide ions, these perovskites absorb a large fraction of the solar light impinging on them which renders them favourites over silicon and dyes. However, what makes halide perovskite-based solar cells appealing is their high efficiency (converting more than 1/4 of the solar energy into electric current) despite the very simple preparation procedure involving rough and ready wet chemistry. This is due to the peculiar electronic characteristics of halide perovskites, which makes this class of material resilient/tolerant to the presence of impurities or “defects” in the crystalline structure. This, of course, does not mean that the efficiency and other properties of halide perovskite-based devices, first and foremost their operational stability, do not depend on the quality of the deposited film.

The improvement of the quality of perovskite films and the development and engineering of deposition procedures for mass production of perovskite solar cells require a detailed understanding of the thermodynamics and kinetics of the fabrication process, and how it depends on the chemical composition of the material. Investigating the deposition process is very challenging as it involves disparate length and time scales, from the atomistic scales –  10^-10 m, and femtoseconds, 10^-15 s – to perovskite crystallite sizes of hundreds of nanometers, 10^-7-10^-6 m, and to tens to thousands of seconds, 10-10³ s. Typical structure elucidating techniques, e.g., X-ray diffraction (XRD), allows to detect the presence of ~10 nm crystallites in a sample, making this technique suitable to follow the conversion reaction of the PbX₂ precursor into APbX₃ perovskite upon dripping of the second reactant, the AX salt. In our recent article published by Communication Materials, an intense X-ray beam at beamline ID10b (ESRF) has been used to detect the fingerprint of the formation and growth of as small as 10 nm perovskite crystallites among the pre-existing species, e.g., the PbX₂precursor. In our experiments, X-ray diffraction  images under grazing incidence were recorded at a frequency of 1 per second while the conversion reaction was taking place, which allowed us to closely follow the process in real time. To understand how atomic, and molecular species interact during the conversion reaction during the formation of perovskites.  we used atomistic simulations, of atoms and molecules on powerful supercomputers. We were able to follow the local evolutions of the conversion reaction from the PbX₂ precursor and AX₂ salt – PbI₂ and PbBr₂, and CH₃NH₃I, CH₃NH₃Br, (NH₂)₂CHI and (NH₂)₂CHBr  – into the corresponding perovskite in the so-called sequential deposition method. The comparison between experimental data and theoretical atomistic simulations results enabled us to shed some light on the conversion reaction mechanism of MHPs.

We found that despite the apparent difference in the structure of the precursor, the conversion mechanism of PbI₂ and PbBr₂ is relatively similar. PbI₂ is a layered structure consisting of slabs aligned along the c crystallographic direction (Figure 2a). PbBr₂ does not present such a layered structure (Figure 2b). Nevertheless, in both cases CH₃NH₃⁺ and X^-ions arising from CH₃NH₃X salts solutions percolate into the PbX₂ precursor structures, forming an intercalated intermediate structure (Figure 3). These intermediate structures are more stable than the PbX₂ and CH₃NH₃X reactants but less stable than the final perovskite phase, and the reaction proceeds further to completion. Both steps of the conversion reaction, intercalation of CH₃NH₃⁺ and X^- into the precursor to form the intermediate phase and conversion of the latter to perovskite, involve the breaking and formation of bonds. Hence, the formation of perovskite according to the sequential deposition method cannot be interpreted as a simple nucleation and growth process.

Together with the effect of halide on the mechanism, kinetics and energetics of the perovskite conversion reaction, we also evaluated the effect of the nature of the A⁺ ion. We analyzed how the conversion mechanism changes when using (NH₂)₂CHX instead of CH₃NH₃X. With (NH₂)₂CHPbI₃ the stable phase at ambient conditions not being perovskite-type, the 3D checkerboard-like structure shown above, but a different phase, the so-called d-phase, in which PbI₃ pillars are immersed in an (NH₂)₂CH⁺ embedding charge-compensating medium (Figure 4). However, going from PbI₂ to the d-phase is difficult as they are structurally very different, and the process proceeds by initially forming the perovskite (NH₂)₂CHPbI₃ phase, which is the result of intercalation of (NH₂)₂CH⁺ and I^- into the PbI₂ precursor, like the process leading to the formation of CH₃NH₃PbI₃. On the contrary, the reaction of (NH₂)₂CHBr with PbBr₂ leads to the (NH₂)₂CHPbBr₃. Interestingly, the formation of (NH₂)₂CHPbBr₃ is a single-step process: the complete intercalation of (NH₂)₂CHBr in PbBr₂ does not exhibit an intermediate phase but readily proceeds to the final perovskite phase. This is due to the bulkier (NH₂)₂CH⁺ cation, which does not fit into the PbBr₂ precursor. Second and perhaps even more interesting is that the energetics and kinetics of the conversion reaction are strongly determined by the first intercalation phase. This does not depend only on how well A⁺ and X^- fit into the PbX₂ precursor but also on the solvation of the ions in the isopropyl alcohol solution. Our simulations reveal that CH₃NH₃⁺ is better solvated in the alcohol than the less polar (NH₂)₂CH⁺, which on the one hand explains the faster kinetics of formation of (NH₂)₂CHPbBr₃ over CH₃NH₃PbBr₃ and on the other hand suggests a novel tuning strategy based on the selection of the solvent based on the solubility of the AX salt into it, perhaps the key conclusion of our work.