Title: Computational Synthesis of Substrates by Crystal Cleavage

Abstract: The discovery of novel substrate materials has been dominated by trial and error, opening the opportunity for a systematic search. To identify stable crystal surfaces, we generate bonding networks for materials from the Materials Project database with one to five atoms in the primitive unit cell. For three-dimensional crystals in this set, we systematically break up to three bonds in the bonding network of the primitive cell. Successful cleavage reduces the bonding network to two periodic dimensions, creating a layer of the cleaved crystal. We identify 4,693 unique cleavage surfaces across 2,133 bulk crystals, 4,626 of which have a maximum Miller index of 1. To characterize the likelihood of cleavage and the thermodynamic stability of the cleaved surfaces, we create monolayers of these surfaces and calculate the work of cleavage and the partially-relaxed surface energy using density functional theory to discover 3,991 potential substrates, 2,307 of which do not contain f-valence electrons and 2,183 of which are derived from a bulk precursor with an entry in the Inorganic Crystal Structure Database. Following, we identify distinct trends in the work of cleavage of these layers and relate them to metallic and covalent/ionic bonding of the three-dimensional precursor. We also assembled a database of commercially available substrates and show that the database of predicted substrates significantly enhances the diversity and range of the distribution of electronic properties and lattice parameters, providing opportunities for the epitaxial growth of many materials. We illustrate the potential impact of the substrate database by identifying several new epitaxial substrates for the transparent conductor BaSnO3, which exhibit low cleavage energies and result in strains an order of magnitude lower than currently used substrates.