| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
2024 Technical Division Student Poster Contest
|
| Presentation Title |
SPU-6: Optimization of VaspSol Solvation Free Energy Predictions |
| Author(s) |
Sean Florez, Eric Fonseca, Richard Hennig |
| On-Site Speaker (Planned) |
Sean Florez |
| Abstract Scope |
Density Functional Theory (DFT) is essential for predicting material properties and reaction barriers. However, computational costs increase when solvent molecules surround the system of interest. Continuum models like VASPsol approximate solvation effects using the Possion-Boltzmann model. This study improved VASPsol's accuracy by developing artificial neural networks to predict solvation energy error with the PBE functional. COSMO sigma profiles, computed with Gaussian09 and the cc-pvTZ basis setwere used as molecular descriptors.
The predicted errors were input into a Nelder Mead optimization process to identify optimal parameters for minimizing VASPsol solvation energy errors.
We evaluated the estimated optimized parameters using VASPsol and repeated this process to reduce errors. VASPsol's error was reduced from 1.17 kcal/mol to 1.05 kcal/mol for all 274 solutes solvated in water in the Truhlar set. Our work enhances the accuracy and reliability of VASPsol, contributing to the advancement of continuum models in DFT. |
| Proceedings Inclusion? |
Undecided |
| Keywords |
Computational Materials Science & Engineering, Modeling and Simulation, Energy Conversion and Storage |