In the past decade, major progress has taken place in the field of metal halide perovskite photovoltaics (PV), as witnessed by the recently achieved conversion efficiency of 25.7% [1] and the concrete opportunity to go beyond the thermodynamic limit of single junction PV by means of perovskite/ crystalline silicon tandems, with the very recent record of 33.2% [2]. Among the most recent efforts towards further development of perovskite PV and subsequent commercialization, strategies ranging from perovskite surface passivation and compositional engineering to thin film encapsulation, contribute to suppress the perovskite absorber and device instability to moisture ingress. In parallel, research efforts presently investigate the role of inorganic charge transport layers (CTLs) next to the one of the most commonly used organic CTLs, such as poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine (PTAA), [6,6]-phenyl-C61-butyric acid methyl ester (PCBM) or C60. The inorganic layers are acknowledged to impart thermal and mechanical stability to the device, as well as to improve device efficiency yield.In this contribution, our recent research on atomic layer deposited (ALD) selective CTLs for metal halide perovskite PV [3] is reviewed by addressing two case studies, namely, SnO2 thin films, selective towards electron transport, and NiO thin films, selective towards hole transport. ALD is selected as deposition technology because of its merit of ultimate control over film thickness and conformality.ALD SnO2 [4]. based on cycles of tetrakis(dimethylamido)-tin and H2O as precursor and co-reactant, respectively, is the state-of-the art in both p-i-n perovskite single junction as well as tandem perovskite/crystalline silicon and perovskite/CIGS PV. Its presence is key to the device thermal and mechanical stability, to prevent humidity ingress in the device and suppress damage to the fullerene (PCBM or C60) during the sputtering of the transparent front contact. Moreover, SnO2 plays a key role in perovskite/perovskite tandem PV, where it serves as solvent barrier to prevent damage of the wide-band gap perovskite sub-cell during solution processing of the narrow-band gap perovskite top cell. In parallel, our in situ studies by means of IR spectroscopy during ALD of SnO2 on perovskite/fullerene disclose the reason behind the difference in device conversion efficiency when SnO2 is processed either on PCBM or C60, i.e. the devices based on PCBM underperform those based on C60 by 3% absolute difference. Specifically, the Sn-precursor is responsible for the modification of the ester group in PCBM thereby affecting the PCBM/perovskite interface and device efficiency.ALD NiO thin (7 nm) films [5], based on cycles of bis(methylcyclopentadienyl)nickel and H2O as precursor and co-reactant, respectively, are found to impart stability to the perovskite device under acceleration test at 85°C, with 80% retention of the initial conversion efficiency after 300 hours at 1 sun illumination. At the same time, the presence of the phosphonic acid-based self-assembled monolayer (SAM), is essential to engineer an almost lossless (i.e., in terms of suppression of charge recombination) NiO/SAM/perovskite interface. Also, the SAM homogeneity and surface coverage on NiO improve with respect to direct SAM processing on ITO (see figure), as witnessed by transmission electron microscopy (TEM) and conductive atomic force microscopy. This result is explained in terms of chemisorption reactions between SAM phosphonic acid groups and NiO hydroxyl groups. The SAM homogeneity on NiO leads to higher shunt resistance in the device with respect to the one with SAM directly processed on ITO. Finally, the combination of NiO and SAM results in a narrower distribution of device performance reaching more than 20% efficient champion device. In parallel, two-terminal perovskite/crystalline silicon tandem devices with an ITO/NiO/SAM tunnel recombination junction (TRJ) exhibit a better device yield with respect to tandem devices with an ITO/SAM TRJ, i.e. the standard deviation decreases from 4.6% with ITO/SAM to 2.0% with ITO/NiO/SAM.[1] Green et al., Prog. Photovolt. Res. Appl. 30, 687 (2022)[2] https://www.kaust.edu.sa/news/kaust-team-sets-world-record-for-tandem-solar-cell-efficiency[3] Zardetto et al., Sustainable Energy & Fuels 1, 30 (2017)[4] Bracesco et al., J. Vac. Sci. Technol. A, 38(6), 063206-1 (2020)[5] Phung et al., ACS Applied Materials and Interfaces. 14(1), 2166 (2022)[6] Phung et al., submitted to Sol. Mat. (2023) Figure 1