High-fidelity electronic structure calculations for non-crystalline solids

The continuous random network model has been successful for the first-principles calculation of many mechanical and structural properties of amorphous systems, as well as first approximations to the electronic structure. However, the large cells used to represent an amorphous system in this model (100s of atoms) make it difficult to calculate those properties that rely on theory which scales poorly with the number of atoms. It has recently been shown that the structure and some properties of amorphous solids can also be calculated by averaging over the properties of a large number of relatively small cells (~20 atoms) generated using random structure sampling. Herein, we take advantage of this representation to calculate accurate electronic properties of amorphous solids using current state-of-the-art electronic structure methods. We propose averaging schemes for properties that are founded on the physics of these systems and demonstrate their efficacy. These averaged properties are directly compared with experimental measurements of well-studied amorphous systems and calculations for the crystalline phase at the same level of theory. The scaling of these calculations are discussed to understand future prospects for high-fidelity experiment-independent property calculations in amorphous systems.