A5028907288- Machine Learning in Materials Science
- Inorganic Chemistry and Materials
- Catalysis and Oxidation Reactions
- Chemistry and Chemical Engineering
- CO2 Reduction Techniques and Catalysts
- X-ray Diffraction in Crystallography
- Computational Drug Discovery Methods
Northwestern University
2022-2024
Abstract A grand challenge of materials science is predicting synthesis pathways for novel compounds. Data-driven approaches have made significant progress in a compound’s synthesizability; however, some recent attempts ignore phase stability information. Here, we combine thermodynamic calculated using density functional theory with composition-based features to train machine learning model that predicts material’s synthesizability. Our the synthesizability ternary 1:1:1 compositions...
Predicting the synthesizability of a new molecule remains an unsolved challenge that chemists have long tackled with heuristic approaches. Here, we report method for predicting using simple yet accurate thermochemical descriptor. We introduce Emin, energy difference between and its lowest constitutional isomer, as predictor is accurate, physically meaningful, first-principles based. apply Emin to 134,000 molecules in QM9 data set find when used alone reduces incorrect predictions...