Wesley Chang, PhD, is listening for trouble.
An assistant professor of mechanical engineering and mechanics, Chang and his students are sending sound waves through lithium batteries, looking for the kinds of internal changes that signal failure before it happens. Using scanning acoustic microscopy, the system captures shifts in mechanical properties like stiffness and density. These affect how quickly and cleanly ultrasound pulses move through the battery’s internal layers, revealing the presence of swelling, gas formation or other subtle mechanical deformations that can lead to breakdowns.
What makes the platform powerful is not just what it detects, but how it’s built. The entire system is designed to be open-source, modular and affordable. It uses commercially available parts, custom software, and a simple benchtop configuration that any lab can assemble. Chang’s goal is not to compete with high-end diagnostic tools, but to offer an accessible alternative that makes battery testing faster, more affordable and more scalable.
“The goal is to provide meaningful information to battery engineers, without requiring a six-figure instrument or a PhD to operate it,” Chang said.
The system works by emitting high-frequency sound waves into a battery cell and measuring how those waves deform as they pass through different materials. Solid components like electrodes, liquid electrolytes and gas bubbles all interact with sound in different ways. By comparing the original waveform to the reflected signal, researchers can infer changes in the internal structure of the battery. One of the most valuable aspects of this method is its sensitivity to gas pockets, which often form as a result of electrolyte decomposition or mechanical stress and are a key early indicator of failure.
Chang first recognized the urgency of this approach through conversations with industry partner SES AI, a lithium-metal battery startup. While academic labs had access to cutting-edge tools, many battery manufacturers did not.
“The lack of comprehensive diagnostic tools available to the battery industry really surprised me,” Chang said. “It made clear that the biggest gaps aren’t always in materials or chemistry — they’re in our ability to see what’s actually happening during production.”
That diagnostic blind spot is especially concerning as battery production scales rapidly. Most factories rely on optical and laser-based methods to catch defects, along with spot performance testing. X-rays or magnetic field-based imaging offer better resolution, but they are slow, expensive and not practical for in-line use. Ultrasound offers a promising complement. It is fast, low-energy, and capable of identifying changes in mechanical properties that other methods miss.
But speed remains a major constraint. In the lab, scans can be run slowly to maximize resolution. On a production line, the system must keep up with speeds exceeding 100 meters per minute. Chang’s team is exploring ways to deploy ultrasound in slower segments of the process, such as electrode drying or electrolyte injection, where even brief pauses might allow for a scan. They are also developing real-time data analysis techniques to reduce the computational burden.
“We can scan very slowly to get higher resolution images and resolve finer features,” Chang said. “But electrode and cell manufacturing has to be extremely fast to get to production levels that are profitable.”
The data challenge is significant. A single scan generates a large volume of high-frequency waveform data that must be interpreted quickly to be useful in a factory setting. Chang’s lab is collaborating with researchers in data science to create algorithms that can flag anomalies on the fly, without slowing down the line.
“If we can collect a large amount of data but it’s too expensive or energy intensive to process all of it, then it remains inappropriate for industrial use,” he said.
The lab is also working to expand the capabilities of the system from two-dimensional cross-sections to three-dimensional reconstructions. Commercial pouch cells are composed of dozens of stacked electrode layers. A 2D scan can show that something is wrong, but not where it is within the stack. With advanced signal processing, Chang’s group is trying to build tomographic maps that would allow engineers to pinpoint which layer is causing problems.
New work from Chang’s lab pushes these checks even earlier in the build, onto the electrode films themselves. The team tested an air-coupled scan, meaning the ultrasound travels through air with no gel or water and no contact with the film, which is friendlier to fast, in-line use. It turns the raw signal into simple gain maps and compact distributions, so process errors show up quickly. The same signal provided a rough “acoustic density” estimate that gives engineers a quick read on how evenly packed the film is, without destructively cutting samples.
“Air scans give you feedback right on the film,” Chang said. “If the picture shifts, an engineer can adjust a setting or pull a roll before it becomes waste. The density readout shows how even the film is, and a contaminant check catches the small things that matter later.”
This adds an early checkpoint to the lab’s ultrasound toolkit while the original system continues to image assembled pouch cells and track electrolyte wetting during build. Taken together, the approaches span film to finished cell and point to lower scrap, higher yield and more confidence in safety as production scales.
Even as the research progresses, Chang is thinking about how to translate it to the broader battery community. He recently co-developed a national online course on electrochemical diagnostics through The Electrochemical Society. The project is aimed at preparing engineers and graduate students for jobs at battery manufacturers and EV companies.
“This is the first time they’ve done something like this,” Chang said. “Usually, they organize conferences and webinars, but have not developed a full interactive online course before.”
The curriculum is hands-on, practical and grounded in industry needs. Chang’s role included shaping the content and building a team with complementary expertise. The course is part of a broader effort to address a skills gap in the fast-growing battery workforce, and to help new engineers start their careers with a solid foundation in the tools and techniques that are most urgently needed.
That same motivation drives the way Chang shares his own research. His lab’s ultrasound system comes with documentation, user guides and regularly updated software, making it easy for others to adopt. Rather than locking the technology behind patents or commercial licenses, he wants it to become part of the standard toolkit for battery testing and research.
“Battery scientists want to build better batteries, not develop new tools,” Chang said. “So we’re building the instruments for them.”
