As the country accelerates its shift toward energy dominance, the demand for innovative battery technologies has never been greater.

“Whether it’s transportation on land, in the sky or on water, or it’s the electric grid, portable electronics and other devices, batteries will be a key technology for powering the modern world,” said Venkat Srinivasan, director of the Argonne Collaborative Center for Energy Storage Science (ACCESS) at the U.S. Department of Energy’s (DOE) Argonne National Laboratory.

This transition will require improved battery performance, longer life, shorter charging time, greater safety, and lower cost, as well as less reliance on critical minerals that could pose supply chain issues.

Argonne’s cutting-edge tools: APS and Aurora

Central to Argonne’s efforts in this area are two powerful research tools — the upgraded Advanced Photon Source (APS) and the Aurora exascale supercomputer at the Argonne Leadership Computing Facility (ALCF). The APS and ALCF are DOE Office of Science user facilities located at Argonne.

“This one-two punch should be game changing for the future of battery research at Argonne and elsewhere,” Srinivasan said.

Aurora is an extremely powerful state-of-the-art machine.

“Aurora has more than 60,000 GPUs,” said Argonne computational scientist Chris Knight, referring to graphical processing units, a type of computer processor. ​“It opens a realm of possibilities where researchers can now write code that has more of the physics they need and still obtain insightful results in a timely manner. This way, they can really start to answer some of the profound questions that battery researchers in both academic and industry groups are trying to tackle.”

At the APS, battery research is getting a dramatic boost from the improved brightness (up to 500 times) of the facility’s X-ray beams following a comprehensive upgrade. Brightness corresponds to how many X-rays are delivered per square centimeter on the sample in a particular amount of time. High brightness allows for study of smaller structures within a material.

This improved brightness means that researchers are able to look at batteries in operando — that is, as the batteries are charging and discharging in real time — with much improved precision. Additionally, researchers are able to look deep inside a battery’s cathode, anode and electrolyte — the three main components of a battery. Finally, they are able to look at extremely detailed and fine-grained structures within the battery, such as small defects that could impair the battery’s performance or contribute to its potential failure.

While the original APS had the capacity to image small structures — down to about 20 nanometers — the increased brightness of the upgraded APS will dramatically increase imaging speed, said Stefan Vogt, associate division director of Argonne’s X-ray Science division.

“These techniques with the upgraded APS became a whole lot faster, so we can now look at things much more operationally than ever before,” he said. ​“If you’re charging or discharging the battery, ideally you’re going to want to see the whole thing, and you’re going to want to see it in real time.”

The upgraded APS, according to Vogt, will provide a unique combination of both temporal and spatial resolution, allowing for one of the most realistic and accurate measurements of a battery ever taken with X-rays.

Data deluge: Managing massive information

The result of all these high-tech battery studies will be a mountain of data, and that’s where Aurora comes in. Vogt estimated that with its upgraded capabilities, the APS as a whole will produce over 100 petabytes of data in a year, with battery research representing a small percentage of that number, maybe several petabytes.

A single petabyte is equivalent to the amount of data that can be stored in 10,000 to 100,000 human brains. Doing the data analysis will require Aurora’s exascale capabilities — able to perform over a quintillion calculations per second.

Argonne has also built a terabit (a trillion bits) streaming connection between the APS and Aurora to enable fast and seamless data transfer, giving researchers the ability to process data while an experiment is running.

“There’s the potential for quick feedback and adjustments to experiments, which is extremely exciting,” Vogt said. ​“We can save valuable beam time.”

One aspect of Aurora that Vogt said is particularly exciting is its artificial intelligence (AI) capabilities.

“We can think about building data sets designed to be analyzed using machine learning to identify new areas in material science for battery materials,” he said. ​“If you look at enough data and have something like a well-trained AI, maybe it finds new areas to explore.”

One future, Vogt envisions, involves using an autonomous laboratory to make new battery materials and feed them to the X-ray beamline, which would then send the data to Aurora. There, a foundation or large language model, such as Argonne’s AuroraGPT, could use the data to make new predictions on other materials to investigate.

Advanced techniques: Imaging and structural analysis

One way the APS and Aurora could create synergy involves using advanced X-ray spectroscopy — a technique that initially produces a visual image that is mathematically converted into a spectrum. A spectrum is essentially a graph that depicts the spin and valence states of each element in the sample. Common elements in a battery cathode include nickel, cobalt, and manganese.

Perhaps more importantly and interestingly, an additional theoretical and mathematical step can model spectra to determine the electronic spins in the battery materials, showing its fundamental state of charge or discharge. This last step typically involves machine learning and could be performed by Aurora, said Argonne beamline scientist Chengjun Sun.

“We can see the electronic structure of cobalt, manganese and nickel at each state of the charge status,” he said. ​“That way, we can really understand how the electrons are moving around.”

The increased focused flux of the upgraded APS is the key to enabling X-ray spectroscopy’s success, according to Sun.

“The flux makes everything possible,” he said. ​“It enables us to stare directly into the heart of the battery.”

An additional X-ray technique that will leverage capabilities of both the APS and Aurora is called ptychography. Ptychography can generate highly detailed images of a sample in real time by creating a pattern of light and dark regions that reveal critical information, allowing researchers to understand molecular structures or defects in materials.

Argonne scientists have recently created a new machine learning model called PtychoNN to do this process in real time. In a recent experiment, PtychoNN streamed ptychographic data to Polaris, Aurora’s predecessor, which successfully analyzed the data.

“With ptychography, you actually don’t have a conventional lens, and instead you use the fact that the beam is coherent. Then you extract the image from a complicated interference pattern on the detector,” said Argonne X-ray scientist Mark Wolfman. ​“You don’t really see the thing you’re looking at on the detector. You see a bunch of patterns and then you work backward to figure out what the image was.”

Using Aurora will enable researchers to dramatically speed up the ptychographic process, Wolfman said.

“We’re in a place where plugging that into Aurora means you can have a closed loop where you can see what you’re doing as you’re doing it,” he said.

Aurora and the upgraded APS will offer major benefits for battery research in the next few years.

“These two facilities have enormous potential to enable important research,” Vogt said. ​“It’s a great time for science at Argonne.”

Battery research at Argonne is largely funded by the DOE Office of Basic Energy Sciences and the Vehicle Technologies Office of the Office of Energy Efficiency and Renewable Energy.