← Back to Explore

DiSCoVeR: an Attention and Density-based Machine Learning Algorithm for Discovering Novel Superhard Materials

JSHS · 2023

Overview

Superhard materials are crucial in a host of modern technology industries from construction to transportation, medicine, and energy. While diamond is currently the hardest known material, its high cost makes it impractical for mass industrial applications, necessitating a more abundant and cheaper material. The Descending from Stochastic Clustering Variance Regression algorithm (DiSCoVeR) is presented in this research: a machine learning (ML) framework that combines four computational data and ML methods to discover novel superhard materials. The project was divided into four phases. First, hyperparameter optimization is performed on CrabNet, an attention-based ML algorithm that makes materials property predictions and part of DiSCoVeR, and state-of- the-art performance is achieved. DiSCoVeR is then used to screen ~70,000 compounds for superhard materials. Next, three of the top candidates (CrBMo2, Cr22MoC6, and Cr21(WC3)2) are synthesized and their hardness measured to experimentally validate DiSCoVeR. All three candidates exhibited significantly higher Vickers hardness than any hard materials used in industry today. In the final phase, DiSCoVeR is employed to discover new superhard materials from over 7 million unique candidates and 11 million calculations, including completely theoretical compounds never synthesized before. It finds several candidates predicted to have a higher bulk modulus than diamond. This research is the first time an automated screening method is created and performed with special emphasis on high-performing, novel materials. Not only is DiSCoVeR proven to be effective at screening for superhard materials, but it has enormous potential for discovering high-performance materials of other properties as well.

Competition history

  • JSHS 2023 Category not listed

Resources

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Source: Junior Science and Humanities Symposium

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google