Baltimore researcher tests supersonic spinning plasma fusion concept using salvaged hospital scanners
Inside a University of Maryland lab, a fusion reactor cobbled together from salvaged hospital MRI scanners and repurposed glass beer bottles sits quietly at the center of one of energy science’s more unlikely bets. The setup looks nothing like the sleek, billion-dollar machines that dominate today’s fusion industry.
But this isn’t a hobbyist project. It’s a serious attempt to resurrect a technology the scientific establishment wrote off nearly 40 years ago — one that most researchers stopped pursuing long before the current fusion boom began.
A technology the fusion world left behind
Magnetic mirrors are one of the oldest ideas in fusion science. The concept is straightforward: place powerful electromagnets at each end of a cylindrical vacuum chamber, and the resulting magnetic fields act like mirrors, bouncing charged particles back and forth and keeping them trapped long enough to fuse.
Equipment arrived as donations from the Air Force Research Laboratory, and summer interns funded by the Maryland Space Grant Consortium helped assemble the machine.
The physics is elegant. Making it work was another matter entirely.
During the Cold War, research teams invested heavily in the approach. The plasmas they created proved stubbornly unstable, and the mirrors leaked — energetic particles slipped through faster than researchers could compensate for. Everything culminated in the world’s largest magnetic mirror reactor at Lawrence Livermore National Laboratory, shut down the day after it was completed in 1986. That was effectively the end of serious mirror research.
Funding and attention shifted to tokamaks — donut-shaped machines that confine plasma in a closed loop. Far more complex and expensive to build, they hold plasma more reliably, and decades of international investment followed. Eventually that momentum produced projects like ITER, the multinational megaproject that has absorbed billions of dollars and decades of engineering effort. Magnetic mirrors, by contrast, were quietly filed away.
The core appeal of mirrors never disappeared, though. As advisor Jason Cassibry, a professor at the University of Alabama in Huntsville, puts it: “If they want something that’s cheaper to build and easier to maintain, they might go with a mirror concept.” in a report from IEEE Spectrum.
Spinning plasma back to life
The idea that rotation might rescue magnetic mirrors had been circulating for years before anyone tested it seriously. At the University of Maryland, a small team did exactly that. From 2004 to 2010, researchers operated the Maryland Centrifugal Experiment — a reactor designed to spin plasma at supersonic speeds, fast enough that centrifugal force would stabilize the plasma and plug the leaks that had doomed earlier mirror designs.
Carlos Romero-Talamas was a postgraduate student on that project. The results surprised even the team running it.
“With only a few megawatts of power input, we were getting similar results to the Livermore spheromak with hundreds of megawatts,” he says. A small, underfunded university experiment matching the output of a far more powerful machine suggested the centrifugal approach had genuine potential.
It didn’t translate into continued funding. In 2010, the University of Maryland applied to the Department of Energy for support to build a larger follow-up device and was rejected. Tokamaks were dominant, ITER was consuming attention and resources, and there was little appetite for revisiting an approach the field had already moved past.
Building a reactor on a budget
A decade later, the landscape had shifted enough that Romero-Talamas secured a second chance. In 2020, ARPA-E awarded his team $5.2 million to build the Centrifugal Mirror Fusion Experiment, or CMFX. That sum is a small fraction of what the U.S. has spent on ITER alone — now approaching $3 billion — which meant every dollar had to stretch.
The most expensive components in most fusion reactors are the superconducting magnets. Romero-Talamas sourced his from two decommissioned Philips MRI scanners, stripping out patient beds, scanning coils, and casings to reach the magnets inside. Glass beer bottles were repurposed as insulators. Equipment arrived as donations from the Air Force Research Laboratory, and summer interns funded by the Maryland Space Grant Consortium helped assemble the machine.
The salvaged MRI magnets created one unexpected complication: the bore — the tunnel designed to fit a human patient — was larger than Romero-Talamas had planned for, forcing a redesign. But the larger vacuum chamber turned out to be an advantage, making it easier to control plasma impurities.
Inside CMFX, a central electrode drives deuterium ions to rotate at more than a million meters per second. Results published recently confirmed that the spinning suppressed instabilities, reduced particle losses, and heated the plasma. At its peak, CMFX reached about one-tenth of the temperature, density, and confinement time needed for thermonuclear fusion — a significant milestone for a machine built largely from salvaged parts.
A startup enters the race
Confident enough in those results, Romero-Talamas took a sabbatical from the university in 2024 and founded Terra Fusion. The company’s next target is TFEC-1, a proof-of-concept reactor using 10 Tesla magnets and operating at ten times the voltage of CMFX. The goal is to reach breakeven by 2029 — a timeline that puts Terra Fusion in direct competition with far better-funded rivals.
Terra Fusion isn’t the only magnetic mirror startup in the field. Realta Fusion, a spin-out from the University of Wisconsin-Madison, is developing its WHAM reactor with a $10 million federal grant and 17 Tesla magnets made from high-temperature superconducting tape. WHAM’s powerful field does most of the confinement work, so Realta doesn’t need internal electrodes — a design choice its chief science officer, Derek Sutherland, considers a key advantage. “That thing’s going to get chewed up in a power plant,” Sutherland says of Terra Fusion’s electrode approach.
Romero-Talamas disagrees. Terra Fusion’s design deliberately uses older, widely available superconducting technology rather than cutting-edge materials, keeping costs lower and the engineering more tractable. “Our machine is much, much simpler,” he says.
What magnetic mirrors could power — and who’s watching
Terra Fusion’s commercial vision centers on small modular reactors generating up to 100 megawatts — compact enough to power data centers, cargo ships, and eventually spacecraft, with a target of reaching market by the mid-2030s. Realta is pursuing a parallel path, aiming for 50-megawatt reactors suited to industrial heat applications like steel and concrete production.
Before either company reaches those milestones, nearer-term revenue opportunities exist. CMFX could be used to produce medical isotopes or serve as a test bed for developing neutron-resistant materials — a service that other fusion companies might pay for as they advance their own reactors. The electrode technology Romero-Talamas’s team is building could also find applications in industrial electric arc furnaces.
Cassibry, who advises several fusion startups, is cautiously optimistic about both approaches. “Most likely, the scaling suggests that confinement and stability will persist all the way up to breakeven,” he says. But he adds a familiar warning: “Whenever you build something, it’s always very humbling because there’s some piece of physics that you didn’t see coming.”
That humility may be the most important thing to watch as Terra Fusion moves from a salvaged lab experiment toward a commercial prototype. The centrifugal mirror concept has already outlasted one era of skepticism. Whether it can survive contact with the engineering realities of a full-scale reactor — and do so faster and cheaper than its rivals — is the question the next few years will answer.
Daniel García is an Editor-in-Chief with strong expertise in structural work and engineering principles. He combines this technical foundation with deep knowledge of energy, spatial design, and emerging technologies, bringing a forward-thinking and analytical approach to editorial leadership.
