Researchers, led by the University of Cambridge, developed a form of hafnium oxide that acts as a highly stable, low‑energy ‘memristor’ — a component designed to mimic the efficient way neurons are connected in the brain. The results are reported in the journal Science Advances.
Current AI systems rely on conventional computer chips that shuttle data back and forth between memory and processing units. This constant movement consumes large amounts of electricity, and global demand is exploding as AI adoption expands across industries.
Brain-inspired, or neuromorphic, computing is an alternative way to process information that could reduce energy use by as much as 70 percent by storing and processing information in the same place, and doing so with extremely low power. Such a system would also be far more adaptable, in the same way our own brains are able to learn and adapt.
“Energy consumption is one of the key challenges in current AI hardware,” said Lead Author Dr. Babak Bakhit, from Cambridge’s Department of Materials Science and Metallurgy. “To address that, you need devices with extremely low currents, excellent stability, outstanding uniformity across switching cycles and devices, and the ability to switch between many distinct states.”
Here is an exclusive Tech Briefs interview, edited for length and clarity, with Bakhit.
Tech Briefs: What was the biggest technical challenge you faced while developing these hafnium-based devices?
Bakhit: There were many challenges, because, technically, I must admit that I didn't expect to get something like that. In the beginning when I started the project, my main aim was to grow the film. These interact with hafnium in a way that they were to tune the nanostructure of hafnium.
The main purpose was getting phase decomposition to grow some sort of nanocomposites. And based on nanocomposites, getting some specific controlled tunability resistive switching. But then we realized, ‘Oh well, it's not really possible that much, at least not in a way that we were thinking about in the beginning.’ So, it took me like long time — almost three years — to get to the point that we have to manage it in a proper way and for getting that specific nanostructure.
And in terms of memristors, there are two main mechanisms behind the performance of these devices. The first one is the formation of little filament. It’s like you have an oxide layer sandwiched between two conductive electrodes. The thickness of the oxide layer is typically thin. When you apply some sort of voltage, you would be able to form a conductive filament into the material. It’s like you make it like a nanoswitch. Well, this is one way. It was the initial idea that we had in the beginning, but we realized it's not that doable.
We then changed that to interface switching. Interface switching is like when you have two different layer widths, you can build a barrier potential between these two layers, and you can tune this barrier potential, lower it down. This means having higher conductance or make it higher or expand the barrier and make it more insulating. That is the whole story, briefly.
It took me a huge amount of work in the lab. Because, specifically speaking again, when you talk about interface memristors or interface switching, this is not a traditional interface switching. This is the p-n junction. This is like a narrow like class of interface switching.
Tech Briefs: The article I read says, “The current manufacturing process requires temperatures of around 700 °C — higher than what standard semiconductor fabrication typically allows.” And you’re quoted as saying, “We’re now working on ways to bring the temperature down to make it more compatible with standard industry processes." My question is: Do you have any updates you can share?
Bakhit: Technically, we are in contact with some companies. Specifically speaking, one of them is from Australia. I can’t at the moment tell you the name of the company, unfortunately, but they are investing in this technology and they are supporting me to recruit two postdocs. So, in the near future, I’m getting two postdocs because the reality is I already know how to lower the temperature while keeping the performance close to what we already have. We just need some people to go to the lab and do the job.
After that, from an industry point of view, we would have a very big step toward industrial devices and fabrication, of course.
Tech Briefs: Besides that, do you have any set plans for further research, work, etc.?
Bakhit: The other thing is, at the moment, I'm thinking about a spin-up company myself. That’s why there are some backup companies helping me to move this project forward.
