A stellarator is a type of fusion reactor that confines superheated plasma using magnetic fields generated entirely by external coils, rather than relying on electric current flowing through the plasma itself. First conceived in the 1950s, the design fell out of fashion for decades as the simpler tokamak dominated fusion research. But advances in computational optimization and superconducting magnets have revived the stellarator as a serious contender for steady-state fusion power, and the concept is now attracting both public research investment and private startup money.
The Basic Idea Behind the Twisted Shape
Fusion requires holding a plasma of hydrogen isotopes at temperatures exceeding 100 million degrees, hot enough for atomic nuclei to collide and fuse. No physical material can withstand that temperature, so the plasma has to be suspended in a magnetic cage. The simplest magnetic cage would be a straight tube, but charged particles would escape out the ends. Bending the tube into a doughnut shape, called a torus, solves that problem but introduces a new one: the magnetic field is stronger on the inside of the doughnut than the outside, causing particles to drift sideways and escape.
A stellarator fixes this drift by twisting the magnetic field lines so that particles spend equal time on the inner and outer sides of the torus, averaging out the drift. The twist is produced entirely by the shape of the external coils, which are contorted into complex, non-planar geometries. This is the defining feature of a stellarator: the confining field comes from the hardware outside the plasma, not from current running through the plasma.
How Stellarators Compare to Tokamaks
The tokamak, the world’s most studied fusion design, also uses a toroidal magnetic field but adds a large electric current running through the plasma itself to create the necessary twist. That plasma current is typically driven by a transformer, which by its nature works in pulses. Sustaining the current continuously requires additional systems, such as radio-frequency wave launchers or neutral beam injectors, which consume substantial power and add engineering complexity.
A stellarator sidesteps this entirely. Because the magnetic field and its twist come from external three-dimensional coils, there is no need for a large toroidal plasma current and therefore no need for current-drive systems. This eliminates the enormous external power otherwise required to maintain steady-state operation, greatly lowering recirculating power and cost.1PubMed Central. A promising approach to steady-state fusion: High-temperature superconducting strong-field stellarator with precise omnigenity It also removes the risk of disruptions, sudden losses of plasma current that can dump huge amounts of energy into the reactor wall. Disruptions are one of the most serious engineering threats facing tokamak designs, and stellarators are essentially immune to them.
The tradeoff is complexity. A tokamak’s coils are relatively simple, flat rings arranged symmetrically around the torus. A stellarator’s coils twist and bend in three dimensions, requiring extreme precision in manufacturing. That difficulty is why tokamaks pulled ahead in the 1970s and 1980s: they were simply easier to build and study with the technology available at the time.
From the Figure-8 to Modern Optimization
The stellarator was invented by physicist Lyman Spitzer at Princeton in 1951. His original design was literally a tube of plasma twisted into a figure-8 shape and closed on itself. The figure-8 geometry naturally introduced the field-line twist needed to cancel particle drifts. But early experiments showed that plasma stability in these devices was poor, and the concept was shelved in favor of more promising approaches.
Researchers have recently revisited the figure-8 idea using modern computational tools, re-imagining it as an optimized stellarator. Their analysis suggests the geometry actually has the potential for a high degree of stability and exceptionally simple construction compared to the baroque coil shapes used in contemporary designs.2Plasma Physics and Controlled Fusion. Back to the figure-8 stellarator Whether this retro approach leads anywhere practical remains to be seen, but it illustrates how dramatically the field’s computational capabilities have changed since the 1950s.
The real transformation of the stellarator concept came in the 1980s and 1990s with the advent of numerical optimization. Researchers realized they could use computers to search through vast spaces of possible coil shapes, hunting for configurations with desirable plasma properties. Instead of accepting whatever magnetic field a given coil geometry happened to produce, they could specify the physics they wanted and work backward to find coils that would deliver it.
Quasi-Symmetry and Why It Matters
In a perfect tokamak, the magnetic field strength on any surface inside the plasma depends on only one angle as you go around the torus. This symmetry is what makes tokamak plasmas relatively well-confined: particles bounce along predictable paths and don’t drift outward quickly. A generic stellarator lacks this symmetry, and its particles can wander out of the plasma much faster, a problem known as neoclassical transport.
The breakthrough insight was that a stellarator doesn’t need to look like a tokamak to have tokamak-like symmetry. If the magnetic field strength varies in the right pattern on each flux surface, even in a geometrically complex, twisted device, the particles will behave as if they are in a symmetric field. This property is called quasi-symmetry, and it comes in several flavors depending on which direction the hidden symmetry lies. When the field strength depends only on a particular angle, even highly energetic particles remain confined over long time scales and collisional transport is minimized.3PubMed Central. Precise stellarator quasi-symmetry can be achieved with electromagnetic coils
The Helically Symmetric Experiment (HSX) at the University of Wisconsin was built specifically to test one flavor of quasi-symmetry, called quasihelical symmetry. Measurements on HSX showed that the thermal diffusivity, essentially how fast heat leaks out of the plasma, was reduced in the quasihelically symmetric configuration compared to a deliberately broken-symmetry configuration. For the same heating power, electron temperatures were measurably higher, and the density profile was more favorably peaked in the center rather than hollow.4PubMed. Experimental demonstration of improved neoclassical transport with quasihelical symmetry Further work on HSX confirmed that turbulence-driven transport in these plasmas matched theoretical predictions to within about 10% of measured energy confinement times.5PubMed. Effect of quasihelical symmetry on trapped-electron mode transport in the HSX stellarator These experiments were among the first to prove that quasi-symmetry was not just a theoretical curiosity but a practical engineering target.
Wendelstein 7-X
The flagship stellarator experiment in the world today is Wendelstein 7-X (W7-X), located in Greifswald, Germany, and operated by the Max Planck Institute for Plasma Physics. W7-X was designed from the ground up as an optimized stellarator, using a different optimization strategy called quasi-isodynamic optimization that minimizes particle losses through careful shaping of the magnetic field. The machine took about two decades to design, build, and assemble, with its 50 non-planar superconducting coils representing one of the most complex engineering achievements in fusion history.
Results from W7-X have been encouraging. During its 2017 and 2018 experimental campaigns, the device achieved record values of the fusion triple product, the combined measure of temperature, density, and confinement time, for stellarator plasmas. Crucially, these results provided evidence of reduced neoclassical energy transport, meaning that the optimization strategy was working as predicted and that the plasma profiles achieved could not have been obtained in a stellarator lacking a comparable level of optimization.6PubMed. Demonstration of reduced neoclassical energy transport in Wendelstein 7-X
More recent W7-X campaigns have tackled turbulent transport, the other major channel through which energy escapes the plasma. Researchers observed that strong density gradients in the plasma core, achieved through pellet injection or wall conditioning with boron, led to significantly reduced turbulent transport. In neutral-beam-heated discharges, the main-ion particle flux dropped sharply, causing the density profile to peak in a way that could not be explained by the particle source alone.7Nuclear Fusion. Turbulence-reduced high-performance scenarios in Wendelstein 7-X Controlling turbulence has historically been one of the hardest problems in stellarator physics, so these findings are seen as a significant step forward.
The device’s heating capability has also grown. During the first divertor operation phase, electron cyclotron resonance heating was extended to 7 megawatts, and hydrogen pellet injection was implemented to fuel the plasma at higher densities. Stable operation above a key density cutoff was demonstrated, which is essential for reaching the high densities a future power plant would need.8Physics of Plasmas. Performance of Wendelstein 7-X stellarator plasmas during the first divertor operation phase
The Confinement Database
W7-X is the best-known stellarator, but it is not the only one. The international stellarator community has compiled an energy confinement database containing more than 3,000 data points from nine major experiments worldwide. Analysis of this database has confirmed robust dependencies of the energy confinement time on plasma density and heating power, and it revealed a systematic offset in confinement between different magnetic configurations that correlates with a property of the external field called the effective helical ripple.9Nuclear Fusion. Characterization of energy confinement in net-current free plasmas using the extended International Stellarator Database In plainer terms, how tightly twisted and how smooth the external field is makes a measurable difference to how well the device holds onto energy, and the effect is consistent across machines built by different teams in different countries.
Work on smaller devices continues to generate useful physics as well. At the TJ-II stellarator in Spain, researchers have shown that carefully controlling the plasma current with external coils can modify the twist profile of the magnetic field, producing large changes in edge density gradients and stored plasma energy. The explanation lies in how certain features of the magnetic geometry interact with turbulence at the plasma edge, facilitating the formation of natural flow structures that act as transport barriers.10Journal of Plasma Physics. The rotational transform and enhanced confinement in the TJ-II stellarator
Building Twisted Coils
The single greatest engineering hurdle for stellarators is manufacturing the coils. Unlike tokamak coils, which are essentially flat rings, stellarator coils bend out of plane in precise, complex curves. The tolerances are tight: small deviations from the designed shape can produce error fields that degrade plasma confinement or introduce unwanted particle losses. These requirements can lead to increased costs and assembly delays, and failure to meet them can compromise the device’s performance.11arXiv. Manufacturing Tolerances of Non-Planar Coils for an Optimized Tabletop Stellarator
Two technological developments are changing this picture. The first is high-temperature superconducting (HTS) tape, which can carry enormous currents and generate very strong magnetic fields. Stronger fields allow the entire device to be made smaller, because the plasma can be confined in a tighter space. A team at MIT and the startup Commonwealth Fusion Systems has already demonstrated a large-bore HTS magnet for a tokamak, and similar technology is being adapted for stellarators. The first non-planar stellarator coil wound with HTS REBCO cable has been developed, demonstrating that the technology can handle the complex geometries involved.12Superconductor Science and Technology. Development of the first non-planar REBCO stellarator coil using VIPER cable
The second development is the use of permanent magnets to supplement or replace some of the electromagnetic coils. The MUSE stellarator at Princeton, a small tabletop device, was built using rare-earth permanent magnets arranged around a simple set of planar coils. This dramatically simplifies construction. Measurements on MUSE have confirmed that the magnetic field topology matches predictions, and that error fields can be identified and corrected.13Plasma Physics and Controlled Fusion. Error field measurements and correction on MUSE permanent magnet stellarator Permanent magnets cannot be switched off, which limits operational flexibility, but for a device that is meant to run in steady state anyway, that may be an acceptable trade.
Handling the Heat Exhaust
Any fusion reactor has to get rid of the heat that leaks out of the plasma edge without melting the walls. In a tokamak, this is handled by a divertor, a specially shaped region that channels escaping particles to reinforced target plates. Stellarators use a conceptually different approach called the island divertor, which exploits magnetic islands, regions where the twisted field lines form isolated loops at the plasma edge, to guide exhaust particles to target plates.
W7-X was designed specifically to test this concept. During its first island divertor campaign, with 5 to 6 megawatts of heating, the device achieved full detachment of the divertor heat flux, meaning that the plasma near the target plates cooled enough for the incoming energy to be radiated away rather than striking the surface directly. Significant neutral gas pressures and compression ratios were measured, both important for pumping away helium ash in a future reactor.14Nuclear Fusion. Stable heat and particle flux detachment with efficient particle exhaust in the island divertor of Wendelstein 7-X
In a separate set of experiments, radiative power exhaust was demonstrated using impurity seeding, deliberately puffing small amounts of nitrogen or neon into the edge plasma to increase radiation. High radiative power losses, reaching about 80%, reduced the divertor heat loads by roughly two-thirds. The heat flux detachment came at the cost of only a modest drop in stored plasma energy.15Nuclear Fusion. First demonstration of radiative power exhaust with impurity seeding in the island divertor at Wendelstein 7-X Achieving this controlled reduction of heat fluxes within the complex three-dimensional edge geometry is considered a very promising finding for future stellarator reactors.
The role of divertor geometry on density behavior continues to be studied, as the island divertor concept remains central to W7-X’s mission of demonstrating reactor-relevant plasma confinement for steady-state timescales of up to 30 minutes during its upcoming high-performance campaigns.16Nuclear Materials and Energy. Investigating the role of divertor geometry on density build-up in the island divertor
The Economics of a Stellarator Power Plant
Building a fusion power plant based on a stellarator raises questions that go beyond plasma physics. Can the device be made compact and affordable enough to compete with other energy sources? Recent design studies have explored this by scanning across different device sizes, magnetic field strengths, and engineering assumptions to find cost-minimized configurations.
One such study found that combining next-generation HTS magnets, flowing liquid metal walls for heat removal, and compact plasma configurations could yield an ignited stellarator power plant with a major radius of 4 meters or less, a magnetic field above 9 tesla, and a net electric output of about 1 gigawatt, enough to power a midsize city. This design would minimize both the capital cost and the levelized cost of electricity.17Nuclear Fusion. Economically optimized design point of high-field stellarator power-plant For comparison, early stellarator reactor concepts from the 1990s assumed devices with major radii of 15 to 20 meters, so the size reduction enabled by stronger magnets is dramatic.
One practical complication for any stellarator power plant is tritium breeding. A fusion reactor burns deuterium and tritium, and since tritium is radioactive with a short half-life, it must be bred inside the reactor itself by surrounding the plasma with lithium-containing blankets. The problem is that stellarators need space for their divertor structures, and every square centimeter of wall given to the divertor is a square centimeter not available for breeding blankets. A recent study of the HELIAS stellarator reactor concept found that the resulting loss in the tritium breeding ratio ranged from 4% to 35%, depending on how large the divertor was and what coolant it used. Water-cooled divertors were worse than helium-cooled ones, because they displaced more breeding material.18Fusion Engineering and Design. Effect of divertor on tritium breeding in HELIAS Whether the breeding ratio can be kept above the self-sufficiency threshold of 1.0 in a practical stellarator design is still an open question and one that will shape the divertor and blanket layout of any future machine.
The Growing Private Sector Interest
For most of their history, stellarators were purely a government-funded endeavor, pursued at national laboratories and universities. That has changed in the past few years. Several private fusion companies have announced stellarator-based programs, attracted by the inherent steady-state capability and the absence of disruption risk. The most prominent is Type One Energy in Wisconsin, which is designing a stellarator using HTS magnets and modern optimization. Renaissance Fusion in France is pursuing a design with liquid metal walls and simplified coils. Proxima Fusion, a spin-off from the Max Planck Institute, is building directly on W7-X’s quasi-isodynamic optimization approach.
These companies are betting that the computational and manufacturing advances of the past decade have tipped the balance. In the 1980s, designing a quasi-symmetric stellarator required months of supercomputer time; today, optimization algorithms running on clusters can explore thousands of candidate configurations in days. New single-stage optimization approaches, which simultaneously optimize the plasma shape and the coil geometry rather than treating them sequentially, are being integrated into experimental design tools for devices like the Columbia Stellarator Experiment.19Nuclear Fusion. Integrating novel stellarator single-stage optimization algorithms to design the Columbia stellarator experiment The speed of this design loop matters because it makes iterative prototyping feasible. A startup can design, simulate, and begin building a small stellarator in a timeframe compatible with venture capital funding cycles, something that would have been unthinkable a generation ago.
Turbulence and the Remaining Physics Gaps
Neoclassical transport, the particle-drift problem that quasi-symmetry was designed to solve, is largely understood and increasingly well controlled. The harder remaining problem is turbulence. Turbulent eddies in the plasma can transport heat and particles outward much faster than collisional processes alone, and predicting turbulence in the complicated three-dimensional geometry of a stellarator is computationally demanding. Gyrokinetic simulations, which track the motion of particles in the fluctuating fields, are the primary theoretical tool for this work and have been applied to realistic W7-X magnetic geometries to study ion-temperature-gradient-driven turbulence in the plasma core.20PubMed Central. Statistical Analysis of Plasma Dynamics in Gyrokinetic Simulations of Stellarator Turbulence
The encouraging news from W7-X is that turbulence can be reduced by manipulating the density profile, as the pellet injection experiments described earlier showed. Whether this turbulence reduction can be sustained in steady-state, reactor-grade conditions is the question that W7-X’s upcoming long-pulse campaigns are designed to address. If turbulence turns out to be manageable, the path from optimized stellarator experiment to power plant becomes considerably shorter. If it remains a stubborn source of energy loss, the optimization toolbox will need to expand beyond neoclassical physics to include turbulence suppression as an explicit design target, a frontier that several research groups are already working on.