Understanding the XYB14 Crystal Family: A Data Science Solution for DFT Limitations

Understanding the XYB<sub>14</sub> Crystal Family: A Data Science Solution for DFT Limitations


  • Scott P. Beckman, Washington State University, Pullman
November 15, 2017 - 11:00am to 12:00pm


The boron-rich XYB14crystal family has a wide band gap, high hardness, excellent thermal stability, and a high Seebeck coefficient, making it a potential candidate as a high-temperature thermoelectric material. The AlYB14composition is of particular interest. In this work, first-principles calculations are performed to investigate the bonding in AlYB14 and related members of the crystal family as it relates to the thermoelectric and thermomechanical properties. These calculations, combined with data archived in the literature for similar compounds, provide a basis for investigating 1,661 chemical variations in the family. Adopting this data science approach represents a 2,000,000× speedup compared to “traditional” DFT methods. The extended data set can be immediately analyzed using unsupervised and supervised regressions and machine-learning algorithms. 

Sponsoring Organization 

Materials Science and Technology Division Materials Theory Group


  • Chemical and Materials Sciences Building
  • Building: 4100
  • Room: J-302

Contact Information