Nathan Morgan

Nathan Morgan is a graduate student in chemical engineering addressing a critical gap in kinetic modeling: most industrial and environmental reactions occur in the liquid phase, yet existing software lacks state-of-the-art solvation models needed for accurate predictions. He develops deep learning models for solvation free energy in pure solvents and solvent mixtures using Chemprop, an open-source machine learning package downloaded over 100,000 times monthly. Nathan trains these models on synthetic data from quantum mechanical calculations and experimental data using MATLAB’s Deep Learning Toolbox to improve generalization, extending the work to predict radical solvation energies through thermodynamic cycle analysis. As lead developer of Chemprop v2, Nathan prioritizes feature development and coordinates multiple software developers. As a MathWorks Fellow, Nathan will leverage the Optimization Toolbox in Chemprop’s continued development. He will integrate machine learning models into the Reaction Mechanism Generator software to enable accurate liquid-phase kinetic predictions and will advance this computational framework toward multi-phase systems. His work could enable more reliable chemical reaction modeling for rational design of pharmaceutical and specialty chemical processes.