Deep Learning-Based Increment Theory for Formation Enthalpy Predictions

Authors

Abstract

Quantitative predictions of molecular thermochemistry, such as formation enthalpy, have been limited for complicated species due to a lack of available training data. Such predictions would be important in predicting reaction thermodynamics and constructing kinetic models. We introduce a graph-based deep learning approach that can separately learn the enthalpy contribution of each atom in its local environment and consider the effect of overall molecular structure. Because this approach follows the additivity scheme of increment theory, it can be generalized to larger and more complicated species not present in the training data. We demonstrate that our model can outperform the conventional increment theory in the predictions of formation enthalpy. We expect this approach will also enable rapid prediction of other extensive properties of large molecules that are difficult to derive from experiments or ab initio calculation.

Content

Supplementary material

Supporting Information_Deep Learning-Based Increment Theory for Formation Enthalpy Predictions
Data analysis and feature selection of this work.
SI high-quality data
CCSD(T)-F12a dataset and NIST-TRC experimental dataset.