The United States, in recent years, has been struggling with a supply shortage of critical materials needed for advanced materials discovery and manufacturing. Widespread delays have impacted sectors ranging from the defense sector to the industrial technology domain.

A deep reliance on overseas sources sheds light on vulnerabilities in the global material supply chain and introduces roadblocks to the development of new materials, innovative manufacturing solutions and commercialization.

With these challenges in mind, Lenore Dai, a professor of chemical engineering and the vice dean of faculty administration in the Ira A. Fulton Schools of Engineering at Arizona State University, is leading an interdisciplinary team to optimize the use of artificial intelligence tools and machine learning models in fundamental materials research.

The project is a collaboration with researchers at the University of Missouri and Brewer Science, a microelectronics materials manufacturer.

“This collaborative effort is a perfect example of why Arizona State University is committed to conducting research of public value,” says ASU President Michael Crow. “Embracing the use of AI and machine learning as mechanisms for advancing research in a way that leads to important advances in manufacturing processes is the kind of impact we seek to have in the work we do at ASU. This is fantastic progress by Dr. Dai, and we are excited to be a part of the team that is making it happen.”

University of Missouri President Mun Choi says, “I am so pleased that our world-class faculty researchers are partnering with distinguished colleagues at ASU and Brewer Science on groundbreaking AI and materials science research. This innovative collaboration continues our incredible momentum while meeting a critical need for our nation.”

Dai is the principal investigator for a new contract with the Engineer Research and Development Center (ERDC) of the U.S. Army Corps of Engineers called Accelerate Materials Design and Process Optimization through Artificial Intelligence and Machine Learning.

Leaders from the ERDC emphasize the significance of the new collaboration, highlighting how AI and machine learning will revolutionize materials science research and innovation.

“We’re excited to get this new partnership underway between Arizona State University, the University of Missouri, Brewer Science and ERDC,” says Robert Moser, director of the ERDC’s Information Technology Laboratory. “The nexus of materials science and artificial intelligence is an important one that can shape a range of applications of interest to ERDC and the Army. I’m looking forward to seeing the outcomes from this impactful work and expert team.”

Edmond Russo, director of the Environmental Laboratory at the ERDC, says, “Artificial intelligence and machine learning are transforming how we discover new materials by allowing us to quickly analyze complex scientific data to find materials with the exact properties we need. I’m excited about this project because it not only uses advanced AI techniques but also brings together the expertise of Arizona State University, the University of Missouri and Brewer Science. This partnership enhances our skills in AI and machine learning for optimizing materials and automating labs, helping us innovate faster in materials science.”

The project will focus on using AI and machine learning to enhance the development of new materials and optimize manufacturing processes. It will advance the design and discovery of novel materials systems using large language models, or LLMs, to assist with generating hypotheses and integrating novel experimentally validated computational tools for materials discovery, design and manufacturing.

“AI tools, such as LLMs like ChatGPT, can be used for accelerating scientific research and covering a wide range of knowledge that is challenging for a single human to obtain. However, there is a high error rate associated with LLMs, which have issues such as hallucination,” Dai says. “One of our goals is to develop prompting and fine-tuned methodologies and software modules to significantly reduce these errors.”

Additionally, the team will integrate novel experimentally validated machine learning models that carefully investigate whether the specific learning architecture is appropriate to capture the physical causal relations across key chemical, microstructural and physical features in materials design and development processes.

“The work that Professor Dai and her team is doing highlights the innovative use of AI in ways that matter to industry and the public,” says Kyle Squires, senior vice provost of engineering, computing and technology at ASU and dean of the Fulton Schools. “These tools can help researchers accelerate the discovery process, create more agile manufacturing processes and, ultimately, deliver innovations that will transform society.”

Dai highlights the expansive scope of the project, emphasizing how collaboration across multiple institutions and sectors amplifies its potential.

“It is important to acknowledge the reach and potential this project has as a result of the collaboration with the University of Missouri, Brewer Science and various faculty across disciplines within the Fulton Schools,” Dai says. “This creates the landscape to capture a large audience and produce an incredible impact, which reaches across institutions, private entities and government sectors.”

LeackStat 2024