With the boom in electric vehicles (EVs) and renewable energy, global demand for batteries is surging. However, existing lithium batteries face challenges such as resource scarcity, supply chain concentration, and high water consumption in mining, which limit the sustainable development of the energy transition. To overcome these bottlenecks, scientists are leveraging the power of supercomputers and artificial intelligence to accelerate the search for new generation battery materials based on abundant metals like magnesium, zinc, and aluminum.
Table of Contents
Challenges in the Lithium Battery Supply Chain
Lithium, hailed as “white gold,” is currently the most critical battery material. But its scarcity makes the supply not only expensive but also vulnerable to geopolitical influence. This forces countries worldwide to face supply chain risks when promoting EV adoption and the energy transition.
Furthermore, lithium mining places a heavy burden on the environment. Extraction methods using evaporation ponds alone require millions of gallons of water, severely impacting ecosystems in arid regions. These real-world problems make the search for alternatives to lithium an inevitable direction in energy technology development.
Seeking Alternatives to Lithium
Scientists have begun studying other more common metals, such as magnesium, zinc, and aluminum. These metals are extremely abundant on Earth, widely distributed, and relatively inexpensive. If they can be applied to next-generation batteries, it could not only reduce reliance on lithium but also make energy storage technology more sustainable.
The potential of these metal batteries comes from their multivalent ion characteristics. Compared to lithium, multivalent ion batteries may store more energy per unit volume, which holds significant implications for EV range and grid stability.
The Role of Supercomputers: SDSC Expanse
Supercomputers are indispensable tools in this material exploration. The EXPANSE system, provided by the U.S. National Science Foundation (NSF) and located at the San Diego Supercomputer Center (SDSC), allows researchers to simulate and screen a massive number of potential material structures.
These high-performance computing platforms can process enormous amounts of data, rapidly simulating the stability and performance of materials under different conditions. This helps scientists shorten experimental cycles and significantly speed up the R&D process for new battery materials.
Advantages and Challenges of TMOs
The research team focused on Transition Metal Oxides (TMOs). These materials have high structural diversity, offering good ionic conductivity and the capacity to accommodate multiple charge carriers, making them ideal candidates for multivalent ion batteries.
However, precisely because of the complex chemical composition and crystal structure of TMOs, traditional exploration methods are often like “searching for a needle in a haystack,” making it difficult to quickly find the most suitable combination within the vast material space. This is where artificial intelligence can play a major role.
Application of AI Models
Researchers combined various generative AI models to conduct unprecedented material exploration:
- Crystal Diffusion Variational Autoencoder (CDVAE): Capable of generating a large number of candidate crystal structures, covering a wide chemical space.
- Large Language Models (LLM): Add ideal properties to the generated structures, screening for materials closer to thermodynamic equilibrium and easier to synthesize.
- Atomic Graph Neural Networks (ALIGNN): Precisely predict the electronic and thermodynamic stability of materials, helping to narrow the screening range.
Through the collaborative work of these models, the team can effectively pinpoint candidate materials with the greatest application potential, avoiding the high cost and low efficiency of traditional experiments.
The One-in-Ten-Thousand Screening Process
Initially, the AI system generated approximately 20,000 candidate structures (about 10,000 from CDVAE and 10,000 from LLM). After multi-layered screening and simulation, the final count was reduced to 42 structures from CDVAE and 13 from LLM.
Notably, both types have advantages. Structures generated by LLM are easier to synthesize in the lab, while CDVAE provided more “breakthrough” materials under non-equilibrium conditions, potentially leading to key advancements in future novel batteries.
The Most Promising Candidate Structures
The research team ultimately confirmed five of the most stable TMO structures. The common features of these structures are:
- Possessing an open tunnel framework
- Capable of effectively accommodating multivalent ions
- Exhibiting good structural and thermodynamic stability in simulation tests
Using Density Functional Theory (DFT) calculations, the researchers further verified their feasibility in reality, proving that these materials indeed hold promise for application in the next generation of batteries.
Support and Outlook
NJIT Associate Professor Datta emphasized that the NSF’s ACCESS program is vital to the research, describing it as “like oxygen.” Without these computational resources, these groundbreaking studies would be difficult to achieve.
As AI and supercomputers continue to advance, we are steadily moving toward a future no longer reliant on lithium. This will not only reduce the environmental burden but also potentially provide safer, more stable, and more sustainable solutions for the global energy transition.
Data Source:
- Battery Crisis Solved? AI Helps Discover Mystery Metal Material to Replace Lithium
- SDSC: Exploring Next-Generation Battery Materials Using GenAI
We offer customized adjustments to the grinding process, tailored to meet processing requirements for maximum efficiency.
Feel free to contact us and we will have specialist available to answer your questions.
If you need customized quotations, you’re also welcome to contact us.
Customer Service Hours: Monday to Friday 09:00~18:00 (GMT+8)
Phone: +8867 223 1058
If you have a subject that you want to know or a phone call that is not clear, you are welcome to send a private message to Facebook~~
Honway Facebook: https://www.facebook.com/honwaygroup
You may be interested in…
[wpb-random-posts]
