A "Glimpse" at CT scanning
We spoke to Dr Peter Attia, CTO and cofounder of Glimpse, to bring you an understanding of why we need CT image scanning and why Glimpse reckons their platform will enable better scanning.
The battery industry is going from strength to strength. It faces three key problems (as described in more depth by Glimpse’s academic paper):
The catastrophic battery safety events that routinely make the news, threatening both public safety and the perception of battery technology.
The punishing physics of battery pack reliability means that in many cases, it only takes one defective cell in a pack of 100s-1000s of cells to cause a failure.
Battery factories are struggling to ramp quickly and profitably, largely due to the intrinsic challenges of making millions of cells per day at micron-scale precision.
All three of these problems stem from battery quality.
The challenge of battery quality is closely related to the challenge of battery failure. Battery failure comes in three flavors:
First, performance degradation is defined as, well, degradation in the performance of the battery. Side reactions on the electrodes that reduce capacity and energy are the most well-known root cause of performance degradation. Performance degradation occurs in nearly every battery. While it’s certainly important, especially for next-generation batteries, nobody writes headlines about how their battery capacity decreased from 90% to 85%.
Second, functional failure is defined as when the cell can no longer meet its functional requirements. Open-circuit and short-circuit failure are the most common categories. Depending on the pack design, a single functional failure can bring down the whole pack.
Third, safety events occur when batteries cause harm to humans or the environment. Of course, the most vivid example is thermal runaway.
While battery failure is a super messy can of worms, the highest-severity battery failures are unfortunately often driven by tiny manufacturing defects such as cathode overhang, metallic contaminants, and separator pinholes. For instance, the Chevy Bolt issues that warranted a $2B recall were caused by the simultaneous presence of “two manufacturing defects known as a torn anode and a folded separator.“ These challenges are only exacerbated by the intense (and intensifying!) pressure that battery producers face on metrics such as yield, throughput, ramp-up time, utilization, and - oh yea - profitability.
Clearly, we need solutions to solve the challenge of battery quality. A key component of quality control is quality inspection. While we’ll need a full selection of techniques for battery quality control, an ideal inspection technique would be nondestructive, highly spatially resolved, and super fast to keep up with the millions of cells made per day at a gigafactory.
Glimpse aims to leverage X-ray computed tomography (CT) scanning at scale as it checks many of the boxes for ideal inspection techniques. Historically, CT has been considered too slow—both in terms of scan time and analysis time—for high-throughput battery production.
Glimpse, however, believes in the power of CT. It’s nondestructive and highly spatially resolved (10–100 µm), so if you could build, say, a software platform to speed up the analysis, it could be a real cornerstone of cell manufacturing in gigafactories.
Their goal is “to build the world's most scalable and powerful battery quality monitoring platform to save customers billions of dollars from battery field failures and slow battery factory ramp ups.” Glimpse believes that CT scanning shouldn’t be a pass/fail tool, and their software is designed to enable comparisons of cell populations across days, months, and years. You can see everything from 10-micron current collectors to the automatically-measured mean anode overhang across hundreds of cells.
They recently released a 1000-cell demo (available here) and made the data publicly available in order to mobilise as many folks as possible to work on the problem and understand their vision.
In discussion with Peter Attia
What's your background and what led you to founding Glimpse?
I moved out west in 2014 to start my PhD in Will Chueh’s lab at Stanford. My PhD focused on battery lifetime. “The” paper of my PhD was one where we used machine learning to predict cycle life using data from the first 100 cycles. My time as a grad student gave me a strong appreciation for the power of combining battery science with software to yield insights from battery data.
After graduating in 2019, I moved down the road to work at Tesla on the cell qualification team. I had the opportunity to work on various cell programs, including Tesla’s in-house 4680s. I learned so much at Tesla and was constantly amazed by its world-class team tackling problems from nanoscale battery degradation to high-throughput battery factory design. I think it’s fair to say I learned that manufacturing high-quality batteries at scale is a super hard problem—and CT could help. At the same time, new battery factories were being announced weekly, and battery safety issues, such as e-bike fires in NYC, routinely made headlines. Plus, I needed to start something big so that I didn’t become completely overshadowed by the other Dr. Peter Attia. Thus, we started Glimpse in March 2023 with the mission of enabling battery quality at scale.
Why is CT important in battery manufacturing? Why CT over other techniques?
I’ll be the first to tell you that I don’t see much value in pitting battery quality control techniques against each other. Battery quality control is super duper hard, and I think we should throw everything we have at it to make safe and reliable batteries at the global scale. From the bottom of my heart, I hope everyone working on battery quality control is wildly successful because that means our industry will have learned to efficiently build safe and reliable products.
That said, I think most folks in the industry agree that CT offers an unparalleled level of insight into battery quality. CT is the only technology available today that is both nondestructive and spatially resolved to tens of microns. We need this level of spatial resolution for battery inspection to detect electrode-scale latent defects and understand manufacturing variability.
Of course, CT has a reputation for being slow, on the order of hours per cell. That’s where Glimpse comes in: Using a tailored combination of hardware and software, we can scan many cells in two minutes today while maintaining high image quality. Our next generation hardware will enable us to drive this time down to tens of seconds next year.
How would you measure Glimpse’s success?
At a high level, Glimpse’s success will be measured by the billions of dollars that our customers save due to drastically reduced battery pack field failures, shorter battery factory ramp-up times, and faster qualification of new battery designs and new battery manufacturing equipment & processes.
One of our key internal metrics for making technical design decisions is to minimize the “time to insights” between scanning a batch of cells and understanding their quality. Many factors determine the “time to insights”: the time to (un)load a cell, scan it, process the data, view the images, perform automated inspection, and ultimately to evaluate battery quality across a factory or fleet. The faster our customers can gain insights into their battery quality, the faster they will ramp production, achieve profitability, and enable widespread electrification.
What's special about Glimpse?
Don’t get me started! I love how Glimpse works on the cutting edge of all sorts of technologies: X-ray imaging, high-performance computing, computer vision, data analysis & visualization, and battery science. Plus, because we work closely with partners for our hardware but focus most of our efforts on software, we’re able to move at the speed of a software company. I’m very proud of what we were able to accomplish during our first year, and the pathway to achieving our long-term vision appears clear.
I’d be remiss not to mention my cofounders, Eric and Patrick. Eric and I worked closely together on a particularly tough failure mode at Tesla. Eric gets much of the credit for convincing me to take the leap into entrepreneurship, and I’m very glad he did. Eric combines business savviness with strong technical understanding, and I’m thrilled to have him build up a business from this technology. Patrick and I worked together on battery lifetime prediction while I was a grad student. I’m confident that there are very few, if any, folks on the planet with Patrick’s deep understanding of both batteries and software systems—expertise that shows up every day when designing and building our tech stack. I’m honored to work with both of them, as well as the rest of the small but mighty Glimpse founding team.
What's next in 2024?
We’ve built the fundamentals of our technology — now it’s time to build up our business. We’re seeking customers for our two product offerings: scan-on-demand and on-premise scanning. We’re also continuing to push the boundaries of our technology in every direction, and we have some exciting announcements coming soon re the next generation of CT technology for battery inspection. Finally, we’re hiring! Visit https://glimp.se/careers to learn more.
🌞 Thanks for reading!
📧 For tips, feedback, or inquiries - reach out