AI and X-ray Studies Unveil Secrets to Enhancing Zinc-Ion Battery Efficiency

A collaborative team from the U.S. Department of Energy's (DOE) Brookhaven National Laboratory and Stony Brook University (SBU) has harnessed artificial intelligence (AI) to model and understand the intricate behavior of electrolytes in zinc-ion batteries. This research, published in the journal PRX Energy, focuses on water-based electrolytes that shuttle charged zinc ions between electrodes during charging and discharging cycles. The team aimed to unravel why high concentrations of zinc chloride (ZnCl2) in the electrolyte enhance both stability and conductivity, crucial factors for efficient battery performance. Esther Takeuchi, chair of the Interdisciplinary Science Department (ISD) at Brookhaven Lab and the William and Jane Knapp Chair in Energy and the Environment at SBU, highlighted the significance of AI in scientific advancements. "AI is a powerful tool that can greatly accelerate our understanding of complex systems," she said. Amy Marschilok, manager of the Energy Storage Division at ISD and a professor of chemistry at SBU, emphasized the potential impact of their findings. "Our work could pave the way for more robust and cost-effective zinc-ion batteries, ideal for large-scale energy storage due to their safety and affordability," she added. The primary challenge in studying these electrolytes is observing the atomic-scale interactions between zinc and chloride ions and water, which are typically too intricate and time-consuming for conventional computational methods. To overcome this, the team utilized a machine learning (ML) approach, combining traditional simulation data with advanced AI algorithms to predict these interactions accurately and efficiently. Deyu Lu, a staff scientist in the Theory and Computation Group at Brookhaven Lab’s Center for Functional Nanomaterials (CFN), led the research. He explained that conventional simulation methods cannot handle the vast number of atomic interactions required to capture the evolving nature of these systems. Instead, they generated a small "training set" of interaction data using conventional simulations and fed it into an AI program. Chuntian Cao, a scientist from the Computing and Data Sciences (CDS) directorate and the paper's first author, described the process as an iterative "active learning" cycle. The AI model would make predictions, which were then validated and refined using additional conventional calculations. This method significantly reduced the computational load, allowing the team to simulate interactions involving thousands of atoms over hundreds of nanoseconds, a feat un achievable with traditional methods alone. The AI model revealed that high concentrations of zinc chloride stabilize water molecules by disrupting the hydrogen-bond network in water. In pure water, each oxygen atom forms hydrogen bonds with hydrogen atoms in neighboring water molecules, creating a network that makes water highly reactive and susceptible to splitting, a process that degrades battery performance. As the zinc chloride concentration increases, the number of hydrogen bonds drops, leaving only about 20% of the original network. This stabilization prevents water from splitting and enhances the electrolyte's performance. Beyond stabilizing water, the high salt concentration also maintains efficient zinc ion transport. At low concentrations, zinc and chloride ions move independently and in opposite directions due to their charges. However, at higher concentrations, the ions and water molecules form negatively charged clusters, which can impede ion mobility. Fortunately, at very high concentrations, large, negatively charged aggregates form, reducing the number of such clusters and allowing smaller, positively charged clusters to move freely, thus ensuring high conductivity. To validate the AI predictions, the team conducted experiments at the National Synchrotron Light Source II (NSLS-II), another DOE user facility at Brookhaven. Using the Pair Distribution Function (PDF) beamline, they measured the atomic distances and confirmed the AI model's predictions. Shan Yan, a co-author from ISD, noted the importance of these measurements in understanding the solvation structure of ions. Milinda Abeykoon, the lead scientist for the PDF beamline, emphasized the high resolution and flexibility of the facility, which allowed the team to cross-check their theoretical findings with empirical data. The successful combination of AI and experimental validation has important implications for the development of zinc-ion batteries. According to Takeuchi, this research showcases the interdisciplinary capabilities of Brookhaven Lab and the potential of AI in materials chemistry. Marschilok highlighted the crucial collaboration with graduate students from SBU, underscoring the importance of combining theoretical and experimental expertise to generate high-quality data and train future scientists. Industry insiders and experts in energy storage technologies view this breakthrough as a significant step towards making zinc-ion batteries viable for large-scale applications. The insights gained from this study not only enhance our understanding of the fundamental chemistry but also provide practical guidelines for optimizing electrolyte compositions. Brookhaven National Laboratory, a world-renowned institution for interdisciplinary research, continues to lead in the application of cutting-edge techniques to solve complex scientific problems. This work exemplifies the lab’s commitment to advancing sustainable energy solutions through innovative collaborations and state-of-the-art facilities.