On the reliability of using reverse Monte Carlo simulations to construct the atomic structure model of metallic glasses

Reverse Monte Carlo (RMC) simulations have been widely utilized to generate three-dimensional structural models for amorphous materials. One practical case is the modeling of atomic-level structure in metallic glasses, based on, for example, X-ray scattering and extended x-ray absorption fine structure (EXAFS) data. However, to what extent this approach can faithfully mimic the atomic structure of real metallic glasses has not been quantitatively evaluated. Using the "known" configurations from molecular dynamics simulations as the benchmark, we evaluate the reliability of RMC modeling with the total structure factor and EXAFS as the input for RMC fitting. The results indicate that the RMC-generated configuration lacks accuracy in reconstructing the local atomic packing, which tends towards the most random (and thus more energetically unfavored) state that can reproduce the reference data. To help guide towards a physically stable and meaningful atomic configura- tion, we advise the incorporation of additional constraints and validation check points.