In order to identify novel substrate materials, we developed a high-throughput bond breaking algorithm. This algorithm takes a three-dimensional crystal as input, systematically breaks bonds, and checks if the bonding network has been reduced to two periodic directions. We apply this algorithm to Materials Project database and identify 4,693 symmetrically unique cleaved surfaces across 2,133 crystals. We then characterize the thermodynamic stability of these cleaved surfaces using the DFT software VASP, characterizing 3,991 surfaces as potential substrates with energy comparable to the experimentally used substrates (0001) AlN, ZnO, and CdS.

This repository contains the structure files, setting files, pseudopotential choices, bulk precursor structure and MaterialsProject ID, and thermodynamic data for the substrates considered in this work.