A Foundational Model for Reaction Networks on Metal Surfaces

Santiago Morandi,Oliver Loveday,Tim Renningholtz,Sergio Pablo-García,Rodrigo A. Vargas-Hernández,Ranga Rohit Seemakurthi,Pol Sanz Berman,Rodrigo García-Muelas,Alán Aspuru-Guzik,Núria López

DOI: https://doi.org/10.26434/chemrxiv-2024-bfv3d

2024-07-01

Abstract:Process optimization in heterogeneous catalysis relies on the control of competing reactions. The reaction mechanisms based on chemical knowledge can be evaluated via density functional theory unveiling experimental catalytic trends. However, this approach finds its limits when applied to complex reaction networks or large molecules, disregarding alternative paths and rare events. Here we present CARE, a foundational model for catalysis on metal surfaces with a rule-based reaction network generator for CxHyOz species built with GAME-Net-UQ, a graph neural network with uncertainty quantification targeting thermodynamic and kinetic parameters, coupled to microkinetic modeling. CARE reproduces experimental activity trends in methanol decomposition, selectivity to C3 products in electrochemical reduction processes, and models the Fischer-Tropsch synthesis to C6 products, including 370k reactions, breaking the current limits of network exploration. This comprehensive model opens the path towards the exploration of thermal and electrocatalytic surface processes previously not amenable to atomistic simulations.

Chemistry

What problem does this paper attempt to address?

This paper proposes a solution to the problem of complex catalytic reaction network control, especially in catalytic processes on metal surfaces. Current methods have limitations in dealing with large reaction networks or large molecules and cannot consider all possible reaction paths and rare events. The research team developed a basic model called CARE (Catalytic Automated Reaction Evaluator), which includes: 1. Rule-based reaction network generator for CxHyOz species. 2. A graph neural network called GAME-Net-UQ for quantifying uncertainty in thermodynamic and kinetic parameters, applied to 42 metal surfaces. 3. Microkinetic modeling capability for analyzing reaction activity. CARE predicts the energy of chemical species and transition states using templates and graph neural networks, enabling the handling of large-scale reaction networks, such as the process of methanol decomposition, selective production of C3 products in electrochemical reduction, and Fischer-Tropsch synthesis to C6 products, including 370,000 reactions. This method improves the exploration of network integrity while reducing reliance on atomic simulation, thereby addressing the computational burden issue of existing methods in dealing with complex chemical spaces.

A Foundational Model for Reaction Networks on Metal Surfaces

Kinetic Modeling with Automatic Reaction Network Generator, an Application to Naphtha Steam Cracking

Deep reaction network exploration at a heterogeneous catalytic interface

Data-efficient modeling of catalytic reactions via enhanced sampling and on-the-fly learning of machine learning potentials

Exploring catalytic reaction networks with machine learning

ReNeGate: A Reaction Network Graph-Theoretical Tool for Automated Mechanistic Studies in Computational Homogeneous Catalysis

Autonomous Reaction Network Exploration in Homogeneous and Heterogeneous Catalysis

Automatic graph representation algorithm for heterogeneous catalysis

General Screening of Surface Alloys for Catalysis

Improving Molecule-Metal Surface Reaction Networks Using the Meta-Generalized Gradient Approximation: CO2 Hydrogenation

Ideation and Evaluation of Novel Multicomponent Reactions via Mechanistic Network Analysis and Automation

General Reaction Network Exploration Scheme Based on Graph Theory Representation and Depth First Search Applied to CO2 Hydrogenation on Pd2Cu Catalyst

Reaction Networks Resemble Low-Dimensional Regular Lattices

A human-machine interface for automatic exploration of chemical reaction networks

Heuristics-Guided Exploration of Reaction Mechanisms

Automated Exploration of Reaction Networks and Mechanisms Based on Metadynamics Nanoreactor Simulations.

Pathfinder -- Navigating and Analyzing Chemical Reaction Networks with an Efficient Graph-based Approach

A Systematic Technique for Kinetic Parameter Estimation in Heterogeneous Solid Catalytic Reaction Networks with Applications

Perspective on Computational Reaction Prediction Using Machine Learning Methods in Heterogeneous Catalysis.