| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Computational Discovery and Design of Materials
|
| Presentation Title |
Discovery of Surfaces with Extreme Work Functions and High Stability by Machine Learning |
| Author(s) |
Peter Schindler |
| On-Site Speaker (Planned) |
Peter Schindler |
| Abstract Scope |
The identification of thermally stable materials possessing surfaces with an ultra-low work function holds the potential to revolutionize the direct conversion of heat into electricity through thermionic energy conversion with high efficiencies. Conversely, surfaces characterized by ultra-high work functions are indispensable for devices requiring substantial contact barriers to suppress electron leakage. Recently, data-driven methodologies have emerged as a groundbreaking approach to efficiently explore vast chemical spaces in search of novel materials with tailored properties. In this study, we present a high-throughput workflow employing density functional theory (DFT) to calculate the work functions and cleavage energies of more than 55,000 surfaces. Furthermore, we devised a physics-based methodology to design surface descriptors and established a surrogate machine learning model for predicting the work function. By leveraging this surrogate model, we can rapidly predict the work function across an extensive chemical space, offering a considerable speed enhancement with near-DFT accuracy. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Machine Learning, Surface Modification and Coatings |