Accelerating computational materials discovery with artificial intelligence and cloud high-performance computing: from large-scale screening to experimental validation

Chi Chen,Dan Thien Nguyen,Shannon J. Lee,Nathan A. Baker,Ajay S. Karakoti,Linda Lauw,Craig Owen,Karl T. Mueller,Brian A. Bilodeau,Vijayakumar Murugesan,Matthias Troyer

2024-01-09

Abstract:High-throughput computational materials discovery has promised significant acceleration of the design and discovery of new materials for many years. Despite a surge in interest and activity, the constraints imposed by large-scale computational resources present a significant bottleneck. Furthermore, examples of large-scale computational discovery carried through experimental validation remain scarce, especially for materials with product applicability. Here we demonstrate how this vision became reality by first combining state-of-the-art artificial intelligence (AI) models and traditional physics-based models on cloud high-performance computing (HPC) resources to quickly navigate through more than 32 million candidates and predict around half a million potentially stable materials. By focusing on solid-state electrolytes for battery applications, our discovery pipeline further identified 18 promising candidates with new compositions and rediscovered a decade's worth of collective knowledge in the field as a byproduct. By employing around one thousand virtual machines (VMs) in the cloud, this process took less than 80 hours. We then synthesized and experimentally characterized the structures and conductivities of our top candidates, the Na$_x$Li$_{3-x}$YCl$_6$ ($0 < x < 3$) series, demonstrating the potential of these compounds to serve as solid electrolytes. Additional candidate materials that are currently under experimental investigation could offer more examples of the computational discovery of new phases of Li- and Na-conducting solid electrolytes. We believe that this unprecedented approach of synergistically integrating AI models and cloud HPC not only accelerates materials discovery but also showcases the potency of AI-guided experimentation in unlocking transformative scientific breakthroughs with real-world applications.

Materials Science,Computational Physics

What problem does this paper attempt to address?

This paper mainly discusses how to use artificial intelligence (AI) and cloud high-performance computing (HPC) to accelerate material discovery and validate it through experiments. The research team screened over 32 million candidate materials and predicted approximately 500,000 potentially stable materials, particularly in the field of solid-state electrolytes (battery applications). By combining AI models with cloud HPC, they were able to complete a large number of calculations within 80 hours, reducing the time from theory to experiment.

First, the study generated a large number of structural candidates by ion substitution in known crystal structures, expanding the search space beyond specific cations. Then, machine learning models were used for preliminary screening, combined with quantum mechanical calculations, to evaluate the thermodynamic stability, electronic, and electrochemical properties of the materials. In particular, they were looking for solid-state electrolyte materials with high lithium ion conductivity.

During the screening process, the researchers identified 18 promising candidate materials with new compositions, including knowledge from the past decade of research. Through synthesis and characterization experiments, they confirmed the potential of the NaxLi3-xYCl6 series materials as solid-state electrolytes. Additionally, they are exploring other potential candidate materials to discover more novel lithium and sodium conductive solid-state electrolytes.

The paper emphasizes that the integration of AI and cloud HPC not only accelerates material discovery but also demonstrates the powerful capability of AI-guided experiments in achieving scientific breakthroughs and practical applications. Future research will relax the screening criteria, consider more complex material structures, and directly generate materials with specific properties. This approach is expected to significantly reduce the time and resources required for traditional experimental discovery of new materials.