In the field of perovskite solar cells, atomic layer deposition (ALD) is an increasingly important tool for the deposition of metal oxides on top of perovskite. The chemical reactions of the ALD precursors with a surface allow low temperature growth of layers with unsurpassed uniformity. Moreover, the process is facile to scale, offering the choice of deposition under vacuum (conventional) or at atmospheric pressure (spatial ALD).[1] The potential value of adopting ALD for layer growth directly on perovskite was earlier established with the application of an ultrathin Al₂O₃ layer on methylammonium lead iodide.[2] We have recently further corroborated this by deposition of Al₂O₃ on wide bandgap, mixed cation, mixed halide perovskite, boosting the V_oc of cells by over 100mV.[3] Next to the case study of Al₂O₃, several other ALD metal oxides have been successfully implemented in perovskite solar cells, among which is SnO₂. It is well known that the latter, grown on perovskite with an organic interlayer between,[4] operates as both an effective electron transport layer (ETL) and as a barrier to the diffusion of chemical species through the cell. However, when ALD SnO₂ is grown directly on top of perovskite, the solar cell performance is worse, although this can be mitigated to a certain degree by adjustment of the perovskite composition.[5]

In this contribution, we therefore select Al₂O₃ and SnO₂ and report on an in-depth study of their growth behavior on (Cs,FA)Pb(I,Br)₃ perovskite, with the purpose of providing insight on the chemical changes at the perovskite/metal oxide interface, responsible for poor cell performance with SnO₂ but surface passivation with Al₂O₃.

ALD allows controlled monolayer by monolayer growth on a substrate, so through x-ray photoelectron spectroscopy, including angle resolved measurements, it is possible to study the changing composition at the perovskite interface as metal oxide is grown.[6] Exposure of the (Cs,FA)Pb(I,Br)₃ perovskite to water (the shared oxygen source in our ALD processes) at a substrate temperature of less than 100° C, does not significantly affect the surface composition of the perovskite.  Instead, the precursor used as the metal source is found to be of greater consequence. For SnO₂ growth (using tetrakis(dimethylamino)tin), we prove that the bulk of the (Cs,FA)Pb(I,Br)₃ perovskite is unchanged compared to pristine perovskite, but at the interface between the two materials a layer consisting of some perovskite constituents (namely Pb, Br and decomposed FA) is gradually formed. Upon deposition of a layer of thickness representative of an ETL (> 10nm SnO₂), it is found that the interface layer (after exposure using a peel-off technique) consists of approximately 10% decomposed FA, 70% Pb and 20% Br, and is < 1nm thick. For the growth of Al₂O₃ on (Cs,FA)Pb(I,Br)₃ using trimethylaluminum precursor, this interface developed is thinner. To determine the factors affecting the formation of interfaces we vary deposition parameters and also compare metal sources, for example slower Al₂O₃ nucleation is observed with dimethylaluminum isopropoxide than trimethylaluminum precursor. The comparison of the interfaces of SnO₂ and Al₂O₃ with perovskite, along with other contributing factors such as the band alignment, the inherently different thickness of metal oxide and cell configurations (p-i-n vs. n-i-p) employed are discussed with reference to the differing cell performance.