Nanoporous TiO2-Water Interface Training Data - Metadata

Training data for the nanoporous TiO2-Water Deep Potential

The primary goal of this dataset is to provide high-quality training data for the development of Deep Potential (DP) models, specifically for the study of nanoporous TiO2-water interfaces. This dataset is relevant to HydroGEN's mission of advancing hydrogen production technologies through material innovation. Machine learning models for the potential energy of multi-atomic systems, such as the Deep Potential (DP) model, make molecular simulations with the accuracy of quantum mechanical density functional theory (DFT) possible at a cost only moderately higher than that of empirical force fields. The nanoporous TiO2-water interface is critical in several applications, including photocatalysis and PEC water splitting. We aim to train the model to predict the proton transfer and water splitting reaction of TiO2-water interfaces accurately. Our broader objective is to learn how these interfaces behave under different conditions, which can inform the design of more efficient hydrogen production systems. The data focuses on specific phases and structures (anatase) of TiO2 that are relevant to its applications in PEC and electrocatalysis.

The dataset includes atomic configurations, energies, and forces. The data is provided in standard units such as angstroms (Å) for distances, electron volts (eV) for energies, and eV/A for forces. The data can be opened and analyzed using any molecular dynamics simulator, such as LAMMPS.

The DeepMD-kit code was used to train the DP. Information about data format, and how to use DeepMD-kit can be found at https://docs.deepmodeling.com/projects/deepmd/en/master/

Authors

Hyuna Kwon (Lawrence Livermore National Laboratory (LLNL)), Marcos Calegari Andrade (LLNL), Daniel V. Esposito (Columbia University), Shane Ardo (University of California, Irvine), Tuan Anh Pham (LLNL), Tadashi Ogitsu (LLNL)

Dataset Metadata

Additional Info