Advanced materials are considered to be core enablers for technologies aimed at broad global concerns, such as climate change, sustainable manufacturing, critical-materials supply chains, environmental mitigation and remediation and others [1]. The process by which novel advanced materials are designed and developed is accelerating through the use of a raft of technologies, including those in artificial intelligence (AI) and robotics, and has been enabled by the expansion of high-performance computing and cloud-based data infrastructure. This one-page reference provides an overview of how machine learning and other AI approaches are accelerating materials development.
Advanced materials are aimed at improving the performance of a product or process. Properties targeted include weight, strength, durability, conductivity, stability, self-healing capability and more [2]. Advanced materials may include alloys, coatings, catalysts, composite materials and two-dimensional materials like graphene, as well as nanoscale materials, like quantum dots.
AI-enabled materials development
Purely experimental approaches to discovering new materials has been relatively slow, so researchers have turned to computational approaches to keep up with demand for new materials. Although it is a tool rather than a panacea, AI provides a number of exciting opportunities for materials research. In general, a machine-learning model of a material can be developed to provide predictive capabilities where traditional, physics-based models either do not yet exist, or remain so challenging (and slow) as to be too expensive in time or other resources [1]. AI models can aid in the development of new physical models by elucidating previously hidden, complex relationships. Similarly, sophisticated use of AI tools will open opportunities for understanding increasingly complex systems.
AI has been employed in advanced-materials in a number of ways, including the following [3]:
• Materials property and performance prediction, given a set of input parameters (including processing history and service conditions, for example)
• Discovery of new material compositions and processing routes for achieving application-oriented targets in terms of desired material properties
• Image-based analysis methods for automating materials characterization
To support the use of machine learning in the design and development of advanced materials, a number of broad initiatives have been launched. Several examples are described in Table 1.
References
1. Materials Genome Initiative Strategic Plan, National Science and Technology Council, OSTP, Executive Office of the President, Materials Genome Initiative, 2021.
2. Yankovitz, D. and others., The future of materials, Deloitte Report, 2023. www2.deloitte.com/us/en/insights/industry/oil-and-gas/the-future-of-materials.html.
3. Pryzer-Knapp, E.O., Pitera, J.W., Staar, P.W. and others, Accelerating materials discovery using artificial intelligence, high-performance computing and robotics, Nature, npj Computational Materials, 8 (84), April 2022.
4. National Institute of Standards and Technology, Gaithersburg, Md., https://jarvis.nist.gov/
5. Citrine Informatics, Chemical and Materials Development Platform, Redwood City, Calif., citrine.io.
6. Sasidhar, K.N., Siboni, N. and others, Enhancing corrosion-resistant alloy design through natural language processing and deep learning, Science Advances, Vol. 9 (32), 2023.