Scientific Achievement
An AI-guided quantum mechanical simulation framework was developed to identify homobenzylic ethers (HBEs) of desired oxidation potentials with 5-fold efficiency improvement over a brute force approach.
Significance and Impact
The highly efficient AI-driven computational framework accelerate discovery of materials with desired properties coupled with high-throughput simulations.
Research Details
- Developed a computational workflow that enumerates HBE candidates and performs Density Functional Theory (DFT) calculations of oxidation potential
- AI Tool: Developed a Bayesian Optimization (BO) model using Gaussian Process Regression and Expected Improvement acquisition function
- Demonstrated the capability to identify desired candidates from a DFT-evaluated data set of 1,400 HBEs.
- Applied the AI Tool to a data set of 112,000 HBEs, from which 42 new desired candidates were discovered after evaluating only 100 HBEs (100 virtual experiments).