Abstract:Despite the rapid development of computational methods, including density functional theory (DFT), predicting the performance of a catalytic material merely based on its atomic arrangements remains challenging. Although quantum mechanics-based methods can model 'real' materials with dopants, grain boundaries, and interfaces with acceptable accuracy, the high demand for computational resources no longer meets the needs of modern scientific research. On the other hand, Machine Learning (ML) method can accelerate the screening of alloy-based catalytic materials. In this study, an ML model was developed to predict the CO 2 and CO adsorption affinity on single-atom doped binary alloys based on the thermochemical properties of component metals. By using a greedy algorithm, the best combination of features was determined, and the ML model was trained and verified based on a data set containing 78 alloys on which the adsorption energy values of CO 2 and CO were calculated from DFT. Comparison between predicted and DFT calculated adsorption energy values suggests that the extreme gradient boosting (XGBoost) algorithm has excellent generalization performance, and the R -squared ( R 2 ) for CO 2 and CO adsorption energy prediction are 0.96 and 0.91, respectively. The errors of predicted adsorption energy are 0.138 eV and 0.075 eV for CO 2 and CO, respectively. This model can be expected to advance our understanding of structure–property relationships at the fundamental level and be used in large-scale screening of alloy-based catalysts.

What problem does this paper attempt to address?

The paper aims to address the following issues: 1. **Adsorption Energy Prediction Challenge**: Despite the rapid development of computational methods such as Density Functional Theory (DFT), predicting the performance of catalytic materials solely based on the atomic arrangement of materials remains challenging. Although quantum mechanical methods can simulate "real" materials (including dopants, grain boundaries, and interfaces) with acceptable accuracy, these methods no longer meet the demands of modern scientific research due to their high computational resource requirements. 2. **Machine Learning Accelerated Screening**: The paper utilizes machine learning (ML) methods to accelerate the screening process of alloy-based catalytic materials. Specifically, an ML model was developed to predict the adsorption affinity of carbon dioxide (CO₂) and carbon monoxide (CO) on single-atom doped binary alloys, and it was trained based on the thermochemical properties of the constituent metals. 3. **Feature Selection and Model Optimization**: By using a greedy algorithm to determine the optimal feature combination, the ML model was trained and validated on a dataset containing 78 alloys. Experimental results show that the Extreme Gradient Boosting (XGBoost) algorithm has excellent generalization performance, with coefficients of determination (R²) of 0.96 and 0.91 for the adsorption energy predictions of CO₂ and CO, respectively, and prediction errors of 0.138 eV and 0.075 eV, respectively. 4. **Understanding Structure-Performance Relationships**: The model is expected to advance our understanding of structure-performance relationships at a fundamental level and be used for large-scale screening of alloy-based catalysts. In summary, the goal of the paper is to improve the prediction capability of adsorption performance of catalytic materials through machine learning methods, thereby accelerating the discovery of novel alloy catalytic materials.

A prediction model for CO2/CO adsorption performance on binary alloys based on machine learning

Applied Machine Learning to Analyze and Predict CO2 Adsorption Behavior of Metal-Organic Frameworks

Machine Learning Prediction of CO Adsorption Energies and Properties of Layered Alloys Using an Improved Feature Selection Algorithm

Machine-Learning-Augmented Chemisorption Model for CO2 Electroreduction Catalyst Screening.

Predicting binding motifs of complex adsorbates using machine learning with a physics-inspired graph representation

Accelerated prediction of Cu-based single-atom alloy catalysts for CO2 reduction by machine learning

Machine-Learning-Driven High-Entropy Alloy Catalyst Discovery to Circumvent the Scaling Relation for CO2 Reduction Reaction

A Transferable Machine-learning Scheme from Pure Metals to Alloys in Predicting Adsorption Energies

Machine-Learning Adsorption on Binary Alloy Surfaces for Catalyst Screening

Efficient mapping of CO adsorption on Cu_(1-x)M_x bimetallic alloys via Machine Learning

Adaptive Design of Alloys for CO2 Activation and Methanation Via Reinforcement Learning Monte Carlo Tree Search Algorithm

A Machine-Learning-assisted Study of Propylene Adsorption Behaviors on Transition Metals and Alloys: Beyond the Dewar-Chatt-Duncanson Model

Developing Cheap but Useful Machine Learning based Models for Investigating High-Entropy Alloy Catalysts

A Machine Learning Model to Predict CO2 Reduction Reactivity and Products Transferred from Metal-Zeolites

Developing Cheap but Useful Machine Learning-Based Models for Investigating High-Entropy Alloy Catalysts

Machine learning approach for screening alloy surfaces for stability in catalytic reaction conditions

Machine Learning Prediction of Monatomic Adsorption Energies with Non-First-principles Calculated Quantities

Machine Learning Prediction of H Adsorption Energies on Ag Alloys

Machine Learning-Assisted Screening of Stepped Alloy Surfaces for C 1 Catalysis

Comparative Study on the Machine Learning-Based Prediction of Adsorption Energies for Ring and Chain Species on Metal Catalyst Surfaces

Prediction of the Enhanced Performance of Cu(I)-modified Porous Materials Towards CO Adsorption by Using Tree-Based Machine Learning Models