In recent years power conversion efficiencies for lab-scale metal halide perovskite solar cells have risen rapidly to over 23.7% [1]. These perovskite semiconductors are advantageous due to their low Shockley-Read-Hall recombination rates, high absorption coefficients across much of the solar spectrum and high charge-carrier diffusion lengths and mobilities [2]. The planar heterojunction solar cell is a simple and popular design with the first efficient perovskite solar cell of this architecture fabricated by co-evaporation of lead iodide and methylammonium iodide (MAI) [3]. Co-evaporation of perovskite thin-films is carried out in a vacuum chamber in which two or more precursors are heated simultaneously until they evaporate. The vapours rise up and condense on the substrate which is mounted above on a rotating sample holder. Co-evaporation is a scalable process which is already used in a number of other industries and films grown via co-evaporation are uniform, have no pin-holes and are smooth over large areas [4]. This process is also solvent free which makes it fully additive and avoids the washing off of underlying layers. This is especially useful when complex layer stacks are being fabricated, for example in tandem solar cells.

An important factor that will influence the commercial success of metal halide perovskite solar cells is the scalability of the deposition processes. To be able to scale a process, the reproducibility and yield of that process are both critical. In the past, several groups have encountered challenges with the evaporation of methylammonium iodide, including a variability of the chamber pressure and problems with the rate control of MAI deposition. The established method for the control of thermal evaporation processes, such as co-evaporation, is the use of quartz micro balances (QMBs). Unfortunately, it has frequently been reported, that these do not yield reliable results when used to monitor MAI evaporation. To resolve some of these challenges, we studied the role of impurities during the co-evaporation and their influence on the control of the evaporation process. To do this, we first characterised the precursor itself using nuclear magnetic resonance (NMR) and mass spectroscopy. We then evaporated perovskite films and monitored the evaporation process using quartz micro balances which are the established approach for rate control in thin-film deposition. Additionally we employed a residual gas analysis system to perform mass spectroscopy of the gas in the vacuum chamber during the evaporation. After the evaporation we characterised the deposited thin-films using a variety of methods, including scanning electron microscopy and X-ray diffraction. Finally we fabricated solar cells using MAI of different purity and measured their power conversion efficiency. Using this approach we are able to shed light on the influence of impurities in MAI on the co-evaporation of methylammonium lead iodide and the final solar cell performance. We are furthermore able to make suggestions on how to improve the control of the MAI evaporation and therefore the control over and reproducibility of the co-evaporation of methylammonium lead iodide thin-films.