Substrates influencing thin-film geometry through charge-transfer: a computational study with DFT and Machine Learning

Organic thin-films are known for their versitile applications due to their tuneable properties. Most properties are strongly dependent on the polymorph geometry the film assumes, which is mainly determined by processing conditions and by the employed substrate. Traditional ab-initio studies of thin films geometries remain computationally prohibitive, due to the immense number of possible configurations. However, recent developments in machine-learning assisted structure search have made structure-to-property investigations accessible. In this contribution we demonstrate the impact that different substrates can have on the geometry of the first two layers of a thin film using the prototypical systems of benzoquinone on Ag(111) and on graphene. We identify the energetically most favourable geometries for both systems and compare them. While the polymorphs formed in the first layer of benzoquinone are very similar, we find differences in the second layer. Further analysis reveals the origin of the deviations to arise from the substrate-induced charge transfer into the first layer: Silver, which transfers more charge to the organic molecules, systamtically favors bilayers with larger electronic interactions than graphene, for which the charge-transfer is negligible.