Abstract:Accurately predicting activation energies is crucial for understanding chemical reactions and modeling complex reaction systems. However, the high computational cost of quantum chemistry methods often limits the feasibility of large-scale studies, leading to a scarcity of high-quality activation energy data. In this work, we explore and compare three innovative approaches—transfer learning, delta learning, and feature engineering—to enhance the accuracy of activation energy predictions using graph neural networks, specifically focusing on methods that incorporate low-cost, low-level computational data. Using the Chemprop model, we systematically evaluated how these methods leverage data from semiempirical quantum mechanical (SQM) calculations to improve predictions. Our findings reveal that delta learning, which adjusts low-level SQM activation energies to match high-level CCSD(T)-F12a targets, offers the greatest accuracy improvement. However, this approach also requires more computational resources due to the complexity of transition state searches for each query of activation energy during model application. Transfer learning, which pretrains models on large datasets of low-level data, provided mixed results, particularly when there was a mismatch in the reaction distributions between the training and target datasets. Feature engineering, which involves adding computed molecular properties as input features, showed modest gains, particularly when incorporating thermodynamic properties. Our study highlights the trade-offs between accuracy and computational demand in selecting the best approach for enhancing activation energy predictions. These insights provide valuable guidelines for researchers aiming to apply machine learning in chemical reaction engineering, helping to balance accuracy with resource constraints.

Application of the Golden Section Search Algorithm in the Nonlinear Isoconversional Calculations to the Determination of the Activation Energy from Nonisothermal Kinetic Conversion Data

A DNA Based Genetic Algorithm for Parameter Estimation in the Hydrogenation Reaction

A New DNA Genetic Algorithm and Its Application in Parameter Estimation

New Isoconversional Method for Evaluating Unambiguous Activation Energy Values for Solid Decomposition

Experimental Analysis of Intrinsic Kinetics and Rate Profiles for Coal Char Conversion under Different Atmospheres

Improved Method for the Determination of Kinetic Parameters from Non-isothermal Thermogravimetric Analysis (TGA) Data

Application of Genetic Algorithm to Estimating Kinetic Model Parameters of Propylene Ammoxidation

Isoconversional methods as single-step kinetic approximation

New Incremental Isoconversional Method for Kinetic Analysis of Solid Thermal Decomposition

Determination of the kinetic parameters of complex condensed phase reactions using a nonlinear model

Isoconversional methods: A powerful tool for kinetic analysis and the identification of experimental data quality

Determination of the apparent activation energy surface from isothermal data of char combustion and gasification

A New Kinetic Method of Processing Ta Data

Comparison of Two Methods for Analyzing the Activation Energy of Coal Combustion

Can varying activation energy be determined reliably from thermogravimetric experiments?

Kinetic Analysis Based on the Kinetic Compensation Effect and Optimization Calculation

Advanced isoconversional kinetic analysis: insight in mechanisms and simulations. Successes and future

A New Method to Explain the Model Dependence of Apparent Activation Energy Derived from A Single Nonisothermal Dynamic Curve

Analysis of Two Methods for Estimating the Temperature Dependence of the Gas-Phase Chemical Reaction Rate Constants

Enhancing Activation Energy Predictions under Data Constraints Using Graph Neural Networks

A simple and linear isoconversional method to determine the pre-exponential factors and the mathematical reaction mechanism functions