High throughput calculations for a dataset of bilayer materials

Abstract

Bilayer materials made of 2D monolayers are emerging as new systems creating diverse opportunities for basic research and applications in optoelectronics, thermoelectrics, and topological science among others. Herein, we present a computational bilayer materials dataset containing 760 structures with their structural, electronic, and transport properties. Different stacking patterns of each bilayer have been framed by analyzing their monolayer symmetries. Density functional theory calculations including van der Waals interactions are carried out for each stacking pattern to evaluate the corresponding ground states, which are correctly identified for experimentally synthesized transition metal dichalcogenides, graphene, boron nitride, and silicene. Binding energies and interlayer charge transfer are evaluated to analyze the interlayer coupling strength. Our dataset can be used for materials screening and data-assisted modeling for desired thermoelectric or optoelectronic applications.

Engineering piezoelectricity at vdW interfaces of quasi-1D chains in 2D Tellurene

“…Unstable phonon spectra for AC, AC ′′ , AA b ′′ , AB b ′′ and AC b ′′ stackings with negative modes are shown in figures S1 and S3 (ESI). Next, the interlayer binding (B.E) and adhesion (A.E) energies of only dynamically stable stackings are calculated by employing the following equations [46,47],…”

Frictional Properties of Two-Dimensional Materials: Data-Driven Machine Learning Predictive Modeling

“…The variation in adhesion energy is typically considered by constructing a potential energy surface tracking the energy for different stacking configurations at equilibrium separation between the materials. In this study, the reported stacking patterns for the BMDS with their equilibrium interlayer separations and corresponding DFT adhesion energies are calculated and taken to initiate the ML study of the bilayers tribological properties. , The stacking patterns are obtained by using combinations of different relative slidings and/or 180° rotations of the various lattices (see Figures S1–S11 in the Supporting Information and in ref and ).…”

“…Bilayered materials classes are denoted as CL-1 for 1T-AX 2 ( P true 3 ̅ m 1), CL-2 for 2H-AX 2 ( P true 6 ̅ m 2), CL-3 for 1T-AX 1 X 2 ( P 3 m 1), CL-4 for 2H-AX 1 X 2 ( P 3 m 1), CL-5 for 1T-A 2 X 2 ( P true 3 ̅ m 1), CL-6 for 2H-A 2 X 2 ( P true 6 ̅ m 2), CL-7 for Sq-AX 2 ( P true 4 ̅ m 2), CL-8 for Sq-A 2 X 2 ( P 4/ nmm ), and CL-9 for Hex-AX or A 2 . Nomenclature of stacking configurations, chemical formulas, and monolayer space groups can be found in the Supporting Information. , …”

“…Nomenclature of stacking configurations, chemical formulas, and monolayer space groups can be found in the Supporting Information. 23,35 extra tree regression (ETR), light gradient boosting machine (LightGBM), GBR, and XGBR are used to develop the prediction models for E adh , E cor , and E vdW separately. After comparing the regression metrics in terms of the coefficient of determination (R 2 ) and root-mean-square error (RMSE) for the different ML models, the best models with the highest R 2 and lowest RMSE are reported for each property.…”

“…In this work, we present statistical machine learning (ML) modeling of frictional and adhesive properties based on our recently reported data set of bilayer materials (BMDS). 23 The BMDS includes 760 bilayer systems comprised of chemically inert, dynamically stable monolayers. DFT simulations supplemented with vdW interactions are used to identify the various stacking configurations of nine classes of bilayers in the BMDS data set.…”