Towards the Future of Battery Research with AI

Next-Gen technologies are sweeping across industries at an unprecedented rate, where AI specifically is re-shaping life as we know it. Its integration into various scientific workflows has seen it speed up processes where every second counts. This is particularly valuable in the race to net zero which depends on the optimization of recyclable and sustainable materials. Medical science and battery research is one area benefiting from AI the most. Some of the brightest minds in the field are making giant strides in the exploration of recycled and new battery materials that will make an important contribution to the energy transition. AI and machine learning have opened up new zones of discovery, where some of these exceptional breakthroughs are made by emerging researchers.

This year’s finalists of the ZEISS Microscopy Young Researcher Award | Germany (YRA) are offering some of the most original proposals in optical analysis in the context of battery research. But how do their future visions aim to connect academic research with the battery industry? And how do they plan to achieve their goals using the ground-breaking possibilities of AI?

The demand for lithium-ion batteries continues to increase due to the demand in electric vehicles and energy storage systems. With it, so does the need for greater efficiency, sustainable materials, and green electrode production. Jan Hettig’s research focuses on electrode development and gaining insights into the electrode microstructure. He says these insights help in understanding “the kinetic effects inside of the electrode and battery performance". Hettig believes that fundamental research should always critically engage with sustainability and environmental concerns. The demand calls for a continuous optimization of the raw materials and the cell’s performance. Batteries for electric vehicles, for example, require high energy density for long distances, fast charging, and a greater life cycle. Hettig proposes that his new application would help to obtain in-depth knowledge about processes which are happening on the micro level. He wants to take a closer look at the electrode microstructure: to visualize and understand how kinetic lithium diffusion processes occur. Hettig says that the issue lies in how data is interpreted, as well as with the electrode’s electrochemical behavior – not to mention time limitations: “We need to read literature, we need to brainstorm new ideas, we need to plan experiments”, he says. AI could help to speed up the analysis which is usually time consuming.

Christin Hogrefe examines the micro-structural changes of lithium batteries within two different single layer pouch cells. She has previously developed an in-situ optical microscope where macroscopic changes are visualized. Whilst she was able to witness the expansion in electrodes coming mainly from the anode side, she says that she was still unable to look into the microstructure. Hogrefe proposes that the Zeiss Xradia Versa would help her to generate high precision information to obtain “a better insight into the microstructure, and the changes which are going on during charging” she says. This would provide essential insights into the effects of electrode volume change, both on the electrode microstructure and the overall layer thickness changes. The findings would help her to gain a deeper understanding of the mechanical degradation process on a particle and electrode level. Overall, the insights are relevant for the understanding of the general behavior of silicon as a component in the anode in lithium-ion batteries. She says the trouble lies in imaging analysis: where one measurement generates thousands of images which cannot be manually analyzed. In fact, AI is her go-to tool for yielding unbiased information. With it, she has a better grasp of the activity occurring inside a battery – a vital factor for designing optimized versions with better performance and a longer lifetime. She says that this win-win result would effectively save resources and contribute to faster charging times.

Sima Hellers specializes in lithium batteries and recycling. She says that the significant growth in the production and use of lithium-ion batteries (LIBs) has increased the need to raise the current recycling rates. Hellers’ research aims to determine the fracture propagation in slag particles under different compressive stress to understand the breakage mechanism of lithium-containing phases. She hopes to develop an innovative method combining light and X-ray microscope systems with ultraviolet waves. She warns that battery recycling has many volatile factors that could go wrong, as researchers are effectively recovering information from waste. “Lithium is very small: it's very difficult to detect with different devices” she says. Her idea is to find the right composition in order to break the slag. The goal is to locate the right particle size without applying too much energy among other factors. “It’s old fashioned because it's what people have been doing for centuries” she says. The difference is that Hellers now has the chance to define and modify her own ores and control how the breakage happens. Hellers claims that we can create models to understand how the particles break in the phases to then separate them. During the last decades, various models based on just mathematics or material properties have emerged, yet all of them lack key factors. Hellers wants to know how computer-based intelligence can define the factors that we don't know. “If we apply this much energy, what phases would we have? How would they look, how big would they be?” she asks. The end goal is to protect the environment from battery waste and secure valuable secondary raw materials.





Research scientist Adil Amin is interested in Lithium-ion batteries (LIBs) – a vital power source for anything from portable electronics to electric vehicles. Amin develops processing routes for different anode materials. He says that “in the transition from fossil fuels to the renewable energies, we need new materials for efficient energy storage systems.” This will help to fulfil the increasing demand for fast-charging batteries and greater storage capacity to support long distance travel. It also requires the research and development of next generation materials that can support key factors. “It is the storage ability, the cycle life, and the energy density” says Amin. For example, silicon has the capacity to store ten times more energy compared to traditional ones. Even so, such materials cannot be implemented until there is an effective way to process them. Amin says that the structure – and how the coated layers are built – is the most important feature in the processing phase. He proposes to study the structure of those important layers, and how a change in state – such as expansion and contraction during charge and discharge cycles – can impact the battery’s performance. This activity breaks the particles and the laminate coated layer within the battery. Amin’s proposal would leverage AI microscopy to contribute to the missing pieces of the puzzle. Amin envisions the copious outcomes that could emerge from collaborations between scientists, machine learning experts and battery engineers. With his proposal, Amin intends to help bring new materials onto the market and support the efficient transition towards renewable energies.

Yaolin Xu studies the physical chemistry of different battery systems. He proposes to carry out research on the electrochemical sulfur precipitation and dissolution behavior in lithium-sulfur (Li-S) batteries. Li-S battery is a promising technology for next-generation electricity storage. However Xu claims there is a limited understanding of the electrochemical reactions involved in its charging and discharging process – which ultimately limits its optimization potential. “We need fundamental insights from the very basic unit of chemical behavior, of lithium, of sulfur” he says. “We are pursuing advanced imaging techniques that are available at ZEISS to help us to do something that is not possible elsewhere.” And whilst fundamental research is intrinsic to knowledge production, Xu acknowledges that it can be demanding. “Based on what I know, many of us are doing quite superficial things because it's still under the initial exploration phase.” He says that a large portion of results generated by machine learning is still not technically AI. “I work on physics, chemistry, but I use more experimental tools to get new insights” he claims. By implementing these theoretical tools into his workflow, he can explore a wider materials pool and accelerate discovery. Ordinarily, he says it may take several years to understand the theoretical mechanism. But if he can use theoretical tools like AI, comparing the properties of new materials with all existing and even undiscovered materials, it would speed up his fundamental research. “I think AI could really help us to accelerate design research” he states. When that happens, we’ll find alternatives that are beyond comprehension.