Lithium battery replacement is one step closer with help of Microsoft's AI running on its Azure Quantum Elements platform.
Microsoft, in collaboration with the U.S. Department of Energy's Pacific Northwest National Laboratory (PNNL), has made a significant breakthrough in battery material research using artificial intelligence (AI). Their joint effort led to the discovery of a new material that could potentially replace lithium in batteries, requiring 70% less lithium than current batteries. This discovery was made possible through the use of AI and high-performance computing (HPC), which accelerated the research process dramatically.
The process began with a massive pool of 32.6 million potential materials. Using AI models trained on molecular science data and traditional scientific supercomputers, the team was able to rapidly narrow down the possibilities. The initial AI screening reduced the pool to 500,000 materials predicted to be stable. Further screening for functional properties like redox potential and band gap brought the number down to about 800. This was followed by additional physics simulations and AI models, ultimately leading to the selection of 18 top candidates.
The new material, a sodium-lithium compound, was synthesized and tested by PNNL scientists. In tests across various temperatures, the compound exhibited viable ionic conductivity, showing potential as a solid-state electrolyte material. The new electrolyte uses approximately 70% less lithium, replacing some of it with sodium, a more abundant and cheaper element.
This discovery is not only significant for its potential to create more sustainable and efficient batteries but also demonstrates how AI can dramatically speed up scientific research. Traditional methods for developing new battery materials could take years, involving extensive calculations and chemistry. The use of AI and HPC in this case reduced the process to a matter of days, showcasing a new paradigm in materials science research.
For the future, the next steps involve further validation and testing of this new material to assess its suitability for mass production. This will likely include more extensive laboratory tests and possibly pilot manufacturing processes to determine the material's performance in real-world applications, such as in electric vehicles or energy storage systems. The success of this project opens the door to further AI-assisted discoveries in various fields of science and technology.
The AI used in the process of discovering the new battery material was a combination of AI models and high-performance computing (HPC) systems. Microsoft's Azure Quantum Elements platform played a crucial role in this research.
AI Models: Microsoft employed AI models that were trained on millions of data points from materials simulations. These models were designed to estimate characteristics of materials related to energy, force, stress, electronic band gap, and mechanical properties. By leveraging these AI models, the researchers were able to predict materials properties much faster than traditional density functional theory (DFT) calculations, approximately 1,500 times faster.
High-Performance Computing (HPC): The AI models were integrated with conventional HPC-based simulations to effectively screen the vast number of material candidates. Microsoft Azure's HPC capabilities were utilized for DFT calculations, which are necessary to confirm the properties of materials as identified by the AI screening. The HPC system was also used for molecular dynamics simulations to model structural changes and further refine the selection of potential materials.
The combination of AI and HPC enabled the researchers to rapidly narrow down the original pool of 32.6 million potential materials to just 18 viable candidates in a significantly shorter time frame than traditional research methods would allow. This approach exemplifies the power of integrating AI with high-performance computing to accelerate scientific discovery and innovation in materials science
Reference: here
1. This article is augmented by my custom AI assistant.
2. FOLLOW ME: https://twitter.com/dmitristern
Comments
Post a Comment