Accelerating Computational Materials Discovery with Machine Learning and Cloud High-Performance Computing: from...

Area of Science:

• Materials Science
• Computational Chemistry
• Electrochemistry

Background:

• High-throughput computational materials discovery promises accelerated innovation but faces bottlenecks due to computational resource limitations.
• Experimental validation of computationally discovered materials, especially for product applications, remains limited.

Purpose of the Study:

• To demonstrate a viable pathway for large-scale computational materials discovery and experimental validation.
• To identify novel solid-state electrolyte materials for advanced battery applications.

Main Methods:

• Combined state-of-the-art machine learning (ML) models with traditional physics-based simulations.
• Utilized cloud high-performance computing (HPC) resources to screen over 32 million material candidates.
• Synthesized and experimentally characterized promising candidate materials, focusing on solid-state electrolytes.

Main Results:

• Predicted approximately half a million potentially stable materials from over 32 million candidates.
• Identified 18 promising novel solid-state electrolyte candidates for battery applications.
• Synthesized and validated the NaLi_3-YCl₆ (0≤ x≤ 3) series as potential solid electrolytes.

Conclusions:

• Advanced ML and HPC methodologies can overcome traditional bottlenecks in materials discovery.
• The integrated computational and experimental approach significantly accelerates the identification and validation of functional materials.
• This work paves the way for a new era of efficient and innovative materials discovery, particularly for energy storage solutions.