Process Optimization of Chemical Vapor Deposition of MAPbBr₃ for Optoelectronic Applications

With growing commercial interest in halide perovskite (HP) thin-film applications, the question which deposition techniques can meet the requirements of industrial scalability and reproducibility becomes increasingly important. At the current time, solution-based methods which are prominent in research are limited to small areas, low reproducibility and are not able to meet special requirements such as deposition on structured substrates. Promising alternatives are vapor-based techniques like vacuum thermal evaporation (VTE) or chemical vapor deposition (CVD) which are well established in the field of semiconductor manufacturing.<br/>At CST (RWTH Aachen University), we have designed and built a custom showerhead-based CVD tool for the deposition of HP thin films by thermal evaporation of halide salts into a heated carrier gas stream. The usage of an inert carrier gas such as N₂ at low vacuum around 5-10 hPa allows for the simultaneous or alternating evaporation of both metal-halide as well as organo-halide salts in a single process. The latter materials are typically difficult to use in high-vacuum techniques like VTE due to their large vapor pressures and low sticking coefficients. In addition, unlike quartz furnace CVD setups found in research literature, our tool features separate sources for different precursor salts. Temperature-controlled substrates are mounted in a heated showerhead-based reactor chamber allowing for uniform deposition on areas up to 100 cm².<br/>While different HP materials have been successfully prepared in our tool, benchmarks for both LEDs and solar cells are lagging behind those of devices based on HP films prepared by spin coating. A typical feature of our CVD films is a reduction of photoluminescence (PL) intensity compared to spin-coated reference samples by 2-3 orders of magnitude. Times-resolved PL spectroscopy indicates the presence of a fast, non-radiative recombination channel in CVD samples absent in their solution-processed counterparts, which likely impedes efficient operation of both LEDs and solar cells. By sequential deposition of PbBr₂ and MABr to form MAPbBr₃ using combinations of CVD and solution deposition, we investigated the influence of both precursors and processes. We observed that spin-coated PbBr₂ layers can be easily converted to MAPbBr₃ layers with strong PL intensity using CVD-MABr, while simultaneously grown CVD-MAPbBr₃ as well as CVD-PbBr₂ films converted by either CVD-MABr or MABr solutions share a low PL intensity. This is in accordance with literature which comprises numerous reports on the conversion of lead-halide precursor films from solution or VTE via MABr-CVD, but lacks reports on the usage of lead halide salts as separate precursors under low vacuum. Unlike reports from high-vacuum evaporation, we observe strong evidence for the dissociation of thermally evaporated PbBr₂. Notably after prolonged evaporation of PbBr₂ precursor material, metallic residues can be found in the crucible. This indicates an at least partial dissociation of PbBr₂ into elemental Pb and volatile Br₂. Due to the high vapor pressure of Br_2,the deposition conditions inside our reactor can be expected to be Br-rich. This could facilitate the formation of intrinsic defects related to excess Br, like Br interstitials, and give a reason for the low PL intensity found when evaporated PbBr₂ is part of the process. Nevertheless, we were able to increase the PL intensity of MAPbBr₃ by 1-2 orders of magnitude by thermal annealing of PbBr₂ precursor layers for up to 3 h at 200-310 °C in N₂. A similar effect can be observed by increasing the substrate temperature during PbBr₂ deposition from 100 °C to above 200 °C, supporting the assumption of a volatile species like Br poisoning our HP. The combined results show a clear path to directly avoid the suspected negative impact of PbBr₂ dissociation in CVD of HPs.