Artificial intelligence is improving the way scientists investigate materials. A machine-learning (ML) model was trained by researchers from Ames Laboratory and Texas A&M University to assess the stability of rare-earth compounds.
Prashant Singh and Yaroslav Mudryk working on the computer. Image Credit: AMES Laboratory.
The current study was financially supported by the Laboratory Directed Research and Development Program (LDRD) program at Ames Laboratory. The framework developed by the researchers builds on current advanced methods for experimenting with compounds and comprehending chemical instabilities.
Ames Laboratory has been the head in rare-earths research since the middle of the 20th century. Rare earth elements have an extensive range of uses such as clean energy technologies, permanent magnets and energy storage. The breakthrough of new rare-earth compounds is part of a bigger effort by researchers to extend access to such materials.
The current method depends on machine learning (ML), a kind of artificial intelligence (AI), driven by computer algorithms enhancing elaborate data usage and experience.
Scientists made use of the upgraded high-throughput density-functional theory (DFT) and Ames Laboratory Rare Earth database (RIC 2.0) to create the basis for their ML model.
High-throughput screening is a computational scheme that enables scientists to test hundreds of models rapidly. DFT is a quantum mechanical technique utilized to analyze the electronic and thermodynamic properties of several body systems. On the basis of this collection of information, the newly-developed ML model makes use of regression learning to evaluate the phase stability of compounds.
Tyler Del Rose, an Iowa State University graduate student, performed the majority of the foundational research required for the database by writing algorithms to search the web for data to supplement the database and DFT calculations.
Also, he worked on experimental validation of the AI predictions and helped to enhance the ML-based models by guaranteeing they are representative of reality.
Machine learning is really important here because when we are talking about new compositions, ordered materials are all very well known to everyone in the rare earth community.
Prashant Singh, Scientist, Ames Laboratory
Singh headed the DFT plus machine learning effort with Guillermo Vazquez and Raymundo Arroyave. “However, when you add disorder to known materials, it’s very different. The number of compositions becomes significantly larger, often thousands or millions, and you cannot investigate all the possible combinations using theory or experiments.”
Singh described that the material analysis relies on a discrete feedback loop in which the AI/ML model has been updated with the help of a new DFT database based on real-time structural and phase information gathered from the experiments.
This process guarantees that data has been performed from one step to the next and decreases the probability of committing errors.
Yaroslav Mudryk, the project supervisor, stated that the framework was developed to use rare earth compounds due to their technological significance, but its application is not restricted to rare-earths research.
The same approach could be utilized to train an ML model to forecast magnetic properties of compounds, process controls for transformative manufacturing and improve mechanical behaviors.
It’s not really meant to discover a particular compound. It was, how do we design a new approach or a new tool for discovery and prediction of rare earth compounds? And that’s what we did.
Yaroslav Mudryk, Project Supervisor, Ames Laboratory
Mudryk stressed the fact that this work is still in its infancy. The team set out to examine the complete potential of this technique. However, they have a positive feeling that there will be an extensive range of applications for the framework in the future.
Journal Reference:
Singh, P., et al. (2022) Machine-learning enabled thermodynamic model for the design of new rare-earth compounds. Acta Materialia. doi.org/10.1016/j.actamat.2022.117759.
Source: https://www.ameslab.gov/