Multi-Task Mixture Density Graph Neural Networks for Predicting Cu-based Single-Atom Alloy Catalysts for CO2 Reduction Reaction
This addresses the problem of costly first-principles calculations for material scientists by enabling reliable catalyst discovery.
The paper tackled predicting CO adsorption energy on Cu-based single-atom alloy catalysts from unrelaxed structures, achieving a mean absolute error of 0.087 eV and improved generalization to out-of-domain configurations.
Graph neural networks (GNNs) have drawn more and more attention from material scientists and demonstrated a high capacity to establish connections between the structure and properties. However, with only unrelaxed structures provided as input, few GNN models can predict the thermodynamic properties of relaxed configurations with an acceptable level of error. In this work, we develop a multi-task (MT) architecture based on DimeNet++ and mixture density networks to improve the performance of such task. Taking CO adsorption on Cu-based single-atom alloy catalysts as an illustration, we show that our method can reliably estimate CO adsorption energy with a mean absolute error of 0.087 eV from the initial CO adsorption structures without costly first-principles calculations. Further, compared to other state-of-the-art GNN methods, our model exhibits improved generalization ability when predicting catalytic performance of out-of-domain configurations, built with either unseen substrate surfaces or doping species. We show that the proposed MT GNN strategy can facilitate catalyst discovery.