Catalyst Promoter Optimization via Neural Network Potential
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for calculating activation energy in catalytic reactions with promoters are computationally expensive and inefficient, requiring extensive searches of promoter arrangements, which often fail to improve catalytic performance unless the reaction significantly affects the desired product yield.
Innovation Solution
An information processing device utilizing a trained Neural Network Potential (NNP) model to optimize promoter element arrangements in catalysts, predicting activation energy and searching for suitable promoter elements to lower activation energy, thereby improving catalytic performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If DFT calculation is used to calculate activation energy by adding a promoter, then the activation energy can be accurately calculated, but the computational cost becomes very high
Solution Approach 1:
The patent applies preliminary action by using Nudged Elastic Band (NEB) calculations to pre-determine the reaction mechanism and identify rate-determining steps before performing promoter addition calculations. This preliminary analysis allows the subsequent DFT calculations to focus only on relevant elementary reactions, significantly reducing the total computational cost while maintaining accuracy in activation energy calculation.
Solution Approach 2:
The patent segments the catalytic reaction into multiple elementary reactions and identifies the rate-determining step. By dividing the complex reaction mechanism into discrete elementary steps, the patent can selectively apply computationally expensive DFT calculations only to the critical steps that most affect product yield, rather than calculating all possible reactions equally.
2Reliability
If the promoter arrangement is changed among many atoms to search for the optimal arrangement, then the activation energy can be minimized, but the search becomes a very difficult operation
Solution Approach 1:
The patent applies local quality by focusing promoter placement on specific active sites and coordination environments rather than uniformly searching all possible atomic positions. By identifying that promoters at certain local sites (e.g., specific metal coordination numbers or surface positions) have disproportionately large effects on activation energy, the patent reduces the search space from all possible atomic arrangements to only locally relevant positions.
Solution Approach 2:
The patent performs preliminary analysis to identify rate-determining elementary reactions and their sensitive atomic sites before conducting the promoter search. This preliminary step allows the subsequent promoter arrangement search to focus only on atoms involved in the rate-determining steps, dramatically reducing the complexity of the search while ensuring that the most impactful promoter positions are evaluated.
3Reliability
If calculations are performed for many promoter arrangements to find the optimal one, then the activation energy can be reduced, but the time required for the search increases significantly
Solution Approach 1:
The patent applies partial action by evaluating a selective subset of promoter arrangements rather than exhaustively searching all possible configurations. By using machine learning models to predict and rank promising promoter positions based on features like coordination number, distance to active sites, and electronic structure, the patent achieves satisfactory catalytic performance with far fewer calculations than a complete search would require.
Solution Approach 2:
The patent changes parameters by using machine learning models that predict activation energy based on key structural and electronic parameters of promoter arrangements. Instead of performing full DFT calculations for every possible promoter configuration, the patent uses trained models that evaluate arrangements based on relevant parameters (coordination geometry, bond lengths, electronic density), enabling rapid screening of many configurations to identify promising candidates for detailed validation.
Data Source
AI summary
An information processing device includes one or more processors. The one or more processors are configured to optimize, for a specific elementary reaction in a reaction using a catalyst including a plurality of elementary reactions, an arrangement of a promoter element in the catalyst based on activation energy acquired using a trained model, and search for the promoter element based on the activation energy acquired using the trained model for each type of the promoter element.