Halide perovskite (HaP)-silicon tandem solar cells received considerable interest from the scientific community, which resulted in a substantial increase in power conversion efficiency (PCE) up to 34.9 %,[1] considerably higher than the individual single-junction cells. The increase in efficiency has led to an enhanced interest in the industrialization of this technology. However, the most efficient devices are fabricated using deposition techniques that are not compatible with large scale production, making the technology transfer more challenging. Co-evaporation is an up-scalable technique, which offers good thickness control and conformal coverage of micro-size pyramidal textures on silicon bottom cells,[2] making it particularly suitable for HaP-silicon tandem devices. It was shown that the growth of co-evaporated HaP films can be modified using seed layers [3,4] or by rinsing the self-assembled monolayer (SAM) hole-extraction layer (HTL) interface with ethanol.[5,6] Still, it remains unclear how these modifications influence the early HaP film formation. Previous studies have shown that incomplete SAM coverage can lead to variations in HaP film quality and, ultimately, to degraded device performance.[7] In the present study, we use synchrotron-based surface-sensitive techniques, in particular x-ray photoemission electron microscopy (XPEEM) and infrared scattering type scanning near-field optical microscopy (IR s-SNOM). Both XPEEM and IR s-SNOM techniques enable experiments with a lateral resolution of 20 nm, allowing us to analyze the influence of SAM inhomogeneities and CsCl seed layer on the composition of 5 and 20 nm thick co-evaporated FA_0.8Cs_0.2PbI_2.7Br_0.3 (FA⁺ = formamidinium cation, C₆H₅N₂⁺) HaP films. The measurements are complemented with cathodoluminescence, photoluminescence, and x-ray diffraction analyses of both thin (5, 20 nm) and thick (100, 500 nm) films, as well as the characterization of the buried interface by film delamination. We use a model system to study the influence of inhomogeneities in [2-(3,6-Dimethoxy-9H-carbazol-9-yl)ethyl]phosphonic acid (MeO-2PACz) SAM coverage where we identified regions with monolayer MeO-2PACz, as well as regions with thicker MeO-2PACz layer containing unbound molecules. The incorporation of FA⁺ is impeded in regions with a monolayer MeO-2PACz, unless a CsCl seed layer is employed. We show how inhomogeneities in MeO-2PACz typically occur when spin-coating SAM on textured silicon bottom cells, and their influence on HaP formation can be mitigated using CsCl seed layer, allowing for ~30 % (certified) efficient HaP-silicon tandem solar cells, the highest value for fully vacuum-processed HaP films to the best of our knowledge. Our study provides a deep understanding of co-evaporated HaP film formation at the nanoscale, allowing for future evidence-based buried interface optimization.