How Artificial Intelligence Is Advancing Zinc-Ion Battery Efficiency and Safety

Understanding Zinc-Ion Batteries with AI

Scientists at the U.S. Department of Energy’s Brookhaven National Laboratory and Stony Brook University applied artificial intelligence (AI) to analyze zinc-ion batteries. Their focus was on the water-based electrolyte that transports zinc ions during battery operation. The study, published in PRX Energy, examined how zinc ions interact with water at varying concentrations of zinc chloride (ZnCl2), a highly soluble salt. Using AI models, validated by experiments at Brookhaven Lab’s National Synchrotron Light Source II (NSLS-II), they uncovered why higher salt concentrations enhance battery performance.

Esther Takeuchi, chair of the Interdisciplinary Science Department at Brookhaven Lab, highlighted the collaboration of experiment and AI theory as key to gaining these insights. Amy Marschilok, manager of the Energy Storage Division and chemistry professor at SBU, noted that zinc-ion batteries with water-based electrolytes are safer and rely on abundant, affordable materials, making them promising candidates for large-scale energy storage.

Water in Salt

Zinc-ion batteries generate electricity through chemical reactions, but side reactions like water splitting reduce efficiency by wasting energy and producing hydrogen gas. The research team focused on a “water-in-salt” electrolyte, where zinc chloride concentration is so high that water molecules behave differently compared to typical “salt-in-water” solutions.

The goal was to capture atomic-level details on how zinc and chloride ions interact with water and influence electrolyte conductivity at different salt concentrations. Directly observing these interactions is challenging, so the team used AI-enhanced computational modeling.

Developing AI Vision

Conventional simulation methods require enormous computing power and time to track atomic interactions accurately on relevant timescales. Instead, the researchers first generated a limited set of simulation data, which served as a training set for an AI model.

Using facilities at Brookhaven Lab, they trained machine learning (ML) models iteratively. The AI generated predictions, which were cross-checked by an ensemble of ML models—similar to consulting multiple experts to validate an answer. When discrepancies appeared, the team ran traditional calculations to correct the dataset, improving the model progressively.

This active learning approach drastically reduced computational needs. Eventually, the AI could simulate thousands of atoms over hundreds of nanoseconds—scales impossible for conventional methods. This allowed detailed insights into the electrolyte’s behavior.

Stabilizing Water

The AI-driven analysis showed that at high zinc chloride concentrations, water molecules become more stable and less prone to splitting. Normally, water molecules form hydrogen bonds creating a reactive network that facilitates splitting. As salt concentration increases, these hydrogen bonds break down, reducing the network’s density to about 20% in the water-in-salt electrolyte.

This stabilization of water molecules is a key reason why high-concentration electrolytes perform better in zinc-ion batteries.

Shuttling Zinc

Efficient zinc ion transport is critical for battery cycling. At low salt concentrations, zinc and chloride ions move independently in opposite directions, which is favorable. However, at intermediate concentrations, ions form negatively charged clusters that move contrary to the desired zinc ion direction, harming conductivity.

At very high concentrations, large negatively charged clusters form but are sparse, so they don’t significantly impede ion movement. Smaller positively charged clusters remain mobile, maintaining good conductivity and battery function.

Validating Experiments

The team validated AI predictions with real-world experiments at NSLS-II. Using X-ray Pair Distribution Function (PDF) measurements, they mapped atomic distances within the electrolyte samples. This high-resolution data allowed direct comparison between predicted and observed atomic structures.

Their experimental results closely matched AI predictions, confirming the model’s reliability. This combination of theory, AI, and experiment demonstrates a pathway for optimizing battery electrolytes and advancing energy storage technologies.

Amy Marschilok emphasized the collaborative nature of the work, highlighting the involvement of graduate students in preparing samples and conducting experiments. This hands-on approach ensures the next generation of scientists gains experience with advanced techniques.

• Key Takeaways:
• High zinc chloride concentration stabilizes water molecules, preventing energy loss from water splitting.
• AI models enable detailed simulation of atomic interactions at scales unreachable by traditional methods.
• Efficient zinc ion transport depends on the formation and behavior of ion clusters at different concentrations.
• Experimental validation confirms AI-driven insights, supporting electrolyte design improvements.

For professionals interested in exploring AI applications in materials science and battery research, training resources are available at Complete AI Training.