China’s “artificial sun” is the informal name for the Experimental Advanced Superconducting Tokamak, or EAST, a doughnut-shaped fusion reactor in Hefei that heats hydrogen plasma to temperatures many times hotter than the core of the actual sun. The nickname stuck because the machine attempts to recreate on Earth the same nuclear fusion process that powers stars. EAST is the most prominent piece of a broader Chinese fusion program that now includes multiple tokamaks, a planned engineering test reactor, and even a private-sector entrant building the world’s first fully high-temperature-superconducting fusion device.
What EAST Actually Is
EAST is a tokamak, a machine that uses powerful magnetic fields to confine a ring of superheated plasma so that hydrogen nuclei can collide and fuse together, releasing energy. What made EAST notable when it began operating in 2006 was its use of superconducting magnets. Conventional copper magnets heat up quickly and drain enormous amounts of electricity, limiting how long a plasma can be sustained. Superconducting coils, cooled to extremely low temperatures, carry current with virtually no resistance, which means EAST can attempt to hold plasma for far longer stretches than most older machines. Thermal modeling of the device shows that its superconducting magnets operate at about 4.2 Kelvin, just a few degrees above absolute zero, with thermal shield components maintained near 80 Kelvin to insulate the coils from the searing plasma inside.
The plasma itself, meanwhile, can reach electron temperatures above 100 million degrees Celsius during high-performance experiments. That is roughly seven times hotter than the sun’s core, which is why the “artificial sun” label resonates. The sun can fuse hydrogen at comparatively lower temperatures because its immense gravitational pressure does part of the work. On Earth, without that pressure, you need far higher temperatures to get nuclei moving fast enough to overcome their electrical repulsion and stick together.
How the Plasma Gets So Hot
Heating plasma to fusion-relevant temperatures requires more than just an initial spark. EAST uses a combination of methods, two of which stand out. Lower hybrid current drive (LHCD) injects microwave-frequency radio waves into the plasma, pushing electrons along the magnetic field to sustain the ring-shaped current that keeps the plasma stable. Electron cyclotron resonance heating (ECRH) fires millimeter-wave beams at a frequency tuned to the natural spin rate of electrons in the magnetic field, dumping energy directly into them. Together, these two systems have produced electron temperatures up to about 120 million degrees Celsius, with sustained periods above 100 seconds at temperatures around 12 keV, which translates to roughly 140 million degrees.
Getting these heating systems to work in concert over long pulses is harder than it sounds. The lower hybrid waves face a density ceiling: above a certain plasma density, the waves can no longer penetrate deep enough to drive current efficiently. Research on EAST found that at higher magnetic fields, wave accessibility is the main bottleneck for higher-frequency lower hybrid waves, while at lower frequencies, energy gets siphoned away by parasitic interactions before it reaches the plasma core.
Taming Edge Instabilities
One of the trickiest problems in tokamak physics is managing a class of plasma eruptions called edge localized modes, or ELMs. When a tokamak operates in its high-performance mode, the plasma edge can periodically belch out bursts of heat and particles. A single large ELM in a future reactor could dump enough energy to damage the inner wall in milliseconds. EAST has become one of the leading testbeds for a technique called resonant magnetic perturbations, or RMPs, which apply small deliberate ripples to the confining magnetic field to calm the plasma edge.
Experiments on EAST demonstrated a nonlinear transition where RMPs shifted from merely reducing ELM size to fully suppressing them altogether.
Subsequent work showed that when an RMP field with a particular pattern was applied, the ELM frequency could increase by a factor of five while the energy released per burst dropped by a similar factor, converting dangerous large eruptions into frequent small ones that the wall can handle.
A practical wrinkle emerged, though. Suppressing ELMs with RMPs tends to cause the plasma to slowly lose density, a phenomenon called pump-out. Researchers on EAST addressed this by injecting frozen pellets of deuterium fuel during ELM-suppressed operation. The pellets partially compensated for the density drop and, in an outcome rarely seen on other machines, raised the plasma’s stored energy back to pre-RMP levels.
A New Confinement Regime
Beyond just suppressing instabilities, EAST researchers recently achieved something that could change how future reactors operate. By applying very small resonant magnetic perturbations, at just a tenth of a percent of the main magnetic field strength, they triggered the formation of what physicists call an internal transport barrier. This is essentially a zone inside the plasma where heat and particles stop leaking outward as fast, dramatically improving the core’s ability to retain energy. The key was using perturbations with a high toroidal mode number so that the effect stayed localized at the plasma edge rather than scrambling conditions deeper inside.
What made this result particularly striking was that the heating power needed to create the internal barrier dropped by half compared to previous approaches. If this regime proves reproducible and scalable, it could mean future reactors need less auxiliary heating to reach and sustain burning-plasma conditions, which is a big deal for the economics of fusion energy.
Separately, studies of long-duration improved-confinement plasmas lasting over a thousand seconds on EAST revealed complex internal behavior including a stable kink mode in the core region. Researchers reconstructed the mode’s structure using soft X-ray imaging, revealing what they described as a twisted pattern resembling a Taichi symbol. The mode was driven by strong central heating and generated unusual current patterns on the magnetic axis, broadening the overall current profile in ways that both helped and complicated confinement.
Where Plasma Meets Metal
No matter how well you confine the core, some fraction of plasma energy eventually reaches the machine’s inner surfaces. The component that bears the worst of this is called the divertor, a specially shaped target at the bottom of the tokamak that intercepts escaping plasma and channels it away. EAST uses tungsten monoblock divertors, the same basic design planned for ITER, the large international fusion experiment under construction in France.
Measurements on EAST’s lower divertor during high-power discharges with about 10 megawatts of heating found peak heat fluxes around 17 megawatts per square meter. Projecting forward to EAST’s full planned heating capacity of roughly 33 megawatts, the peak could reach about 56 megawatts per square meter, an intense thermal load that pushes the limits of tungsten’s endurance. Managing that heat is one of the central engineering challenges for any future power-producing reactor.
One creative alternative being explored on EAST is replacing solid surfaces with flowing liquid lithium. Two generations of continuously flowing liquid lithium limiters have been tested in the machine. A thin circulating lithium layer, less than a tenth of a millimeter thick with a flow rate of about 2 cubic centimeters per second, was sustained during high-performance plasma discharges. The liquid lithium surface is self-healing in the sense that damage from plasma bombardment gets carried away and replaced by fresh material. Experiments showed that the controllable lithium emission from the limiter reduced impurities, lowered the heat load on the divertor, and in some cases improved overall plasma stored energy. If scaled up, flowing liquid metal surfaces could offer a path around the lifetime limitations of solid components in a reactor environment exposed to intense neutron bombardment.
HL-3 and China’s Other Tokamak
EAST is not the only fusion device operating in China. The HL-3 tokamak (formerly known as HL-2M), located in Chengdu, achieved its first plasma in 2020 and is designed to reach plasma currents up to 2.5 million amperes, significantly higher than EAST. Its focus is somewhat different: HL-3 is being used to study advanced divertor configurations and high-performance plasma scenarios that are directly relevant to future reactor designs.
One recent line of research on HL-3 involves a novel divertor geometry called a closed V-shaped slot X-divertor. Simulations showed this design traps recycling neutral particles and impurities far more efficiently than conventional open divertor layouts, which allows it to achieve a cooled, detached plasma state at much lower densities. That matters because divertor detachment is how engineers protect the target from being melted, and achieving it more easily expands the range of conditions under which a reactor could safely operate.
CFETR and the Road to a Power Plant
China’s long-term fusion roadmap runs through EAST and HL-3 to a much larger machine called the Chinese Fusion Engineering Testing Reactor, or CFETR. Where EAST and HL-3 are primarily physics experiments, CFETR is intended to be an engineering demonstration reactor that bridges the gap between ITER and a commercial power plant.
The design calls for two phases of operation. Phase I targets fusion power in the range of 50 to 200 megawatts. Phase II aims to push beyond 1 gigawatt, which is in the range of a commercial fission reactor, while validating the full set of technologies needed for a demonstration power plant. To accommodate both phases in a single machine, the design was scaled up from an initial plasma major radius of 5.7 meters to 7.2 meters.
A critical goal for CFETR is tritium self-sufficiency. In a deuterium-tritium fusion reactor, one of the two fuel species, tritium, is radioactive, rare, and must be bred inside the reactor itself by surrounding it with lithium-containing blankets that capture escaping neutrons and convert lithium into fresh tritium. The design target for CFETR is a tritium breeding ratio greater than 1.2, meaning the reactor would produce at least 20 percent more tritium than it consumes. China is developing a helium-cooled solid breeder blanket for this purpose, with the design optimized through detailed three-dimensional neutronics simulations to meet both tritium production and neutron shielding requirements. A test version of this blanket system is slated to be tested in ITER to verify the technology before CFETR construction.
BEST and the Burning Plasma Frontier
More recently, plans have emerged for an additional facility called the Burning Plasma Experimental Superconducting Tokamak, or BEST, also to be built in Hefei. BEST is designed to operate under reactor-relevant conditions with a fusion gain factor above 5 and a fusion power exceeding 180 megawatts. A “gain factor” above 5 means the reactor produces more than five times the energy that is pumped in to heat the plasma, which is the threshold at which the fusion reactions themselves become the dominant source of heat inside the machine, a state known as a burning plasma.
The high neutron fluxes expected inside BEST create an intense nuclear environment that will allow researchers to study materials degradation, blanket performance, and other reactor-relevant issues that cannot be fully tested in machines that run at lower power. BEST occupies a strategic niche in China’s fusion roadmap: it is not a power plant, but it would produce conditions close enough to a real reactor that engineering lessons learned there would directly transfer to CFETR and beyond.
A Private-Sector Entrant
China’s fusion ambitions are no longer exclusively state-run. Energy Singularity, a Shanghai-based startup, designed and built the HH70, which is currently the first tokamak in the world to be fully integrated with high-temperature superconducting coils. Conventional superconductors, like those in EAST, require cooling to near absolute zero with liquid helium. High-temperature superconductors can operate at somewhat warmer temperatures (still very cold by everyday standards, but far cheaper and simpler to maintain), and they can carry far higher current densities, enabling more compact and powerful magnets.
The significance of HH70 is less about its plasma performance, which is modest, and more about proving that a full set of high-temperature superconducting magnets can be manufactured, assembled, and operated in a working tokamak. Several Western fusion startups are pursuing the same approach, but Energy Singularity got there first with a fully integrated device. If high-temperature superconductors deliver on their promise, future fusion reactors could be substantially smaller and cheaper than designs based on older magnet technology.
Using Machine Learning to Predict Disruptions
One of the most feared events in a tokamak is a disruption, a sudden, violent loss of plasma confinement that dumps enormous energy onto the machine’s walls and structure. Disruptions can physically damage the device, and in a reactor-scale machine, they could be catastrophic. Predicting them far enough in advance to trigger a mitigation response, like injecting a burst of gas to spread the energy over a wider area, is an active area of research worldwide.
Chinese researchers have been applying machine learning to this problem across multiple machines. On the J-TEXT tokamak in Wuhan, years of iterative work brought disruption prediction from a narrow capability covering only limited types of disruptions with short warning times to a system that can predict all kinds of disruptions with high success rates and long enough warning times to trigger a mitigation response.
Perhaps more important for the future is the finding that models trained on one tokamak can be transferred to a very different machine with remarkably little additional data. A deep transfer learning model trained on J-TEXT was successfully applied to EAST, despite the two machines differing substantially in size, operating regime, and magnetic configuration. The transferred model, using only about 20 EAST discharges for fine-tuning, reached performance comparable to a model trained directly on roughly 1,900 EAST discharges. This result suggests that future machines like ITER, which will not have thousands of disruptions to learn from during their early operating years, could leverage disruption knowledge accumulated on existing devices.
How China’s Fusion Program Fits Into the Global Picture
China is a major participant in ITER, contributing components and personnel to the international project, while simultaneously building its own domestic pipeline of increasingly capable machines. The logic is straightforward: ITER will demonstrate that fusion can produce a net energy gain at scale, but it is not designed to generate electricity. The step from ITER to a power plant requires someone to build and operate an engineering test reactor. China’s CFETR is explicitly designed to fill that gap, with EAST, HL-3, and BEST serving as stepping stones that develop the physics understanding and engineering solutions needed along the way.
The Chinese helium-cooled ceramic breeder test blanket system, currently under development, will be tested in ITER before being deployed in CFETR. This kind of technology pipeline, using international collaboration to validate components that then feed into a domestic program, is a model that other fusion-pursuing nations have proposed but that China has advanced further in concrete hardware terms.
The addition of private companies like Energy Singularity adds another dimension. Where the state program follows a methodical progression through ever-larger tokamaks, private ventures can take riskier bets on disruptive technologies like high-temperature superconducting magnets. If those bets pay off, they could compress the timeline not just for China but for the global fusion effort, since the physics and engineering lessons transfer across borders even when the specific hardware does not.
Why Liquid Lithium Might Matter More Than It Sounds
Of all the individual technologies being tested on EAST, flowing liquid lithium is one worth watching closely. The fundamental problem it addresses is deceptively simple: the inner wall of a fusion reactor gets hammered by neutrons, plasma particles, and heat, and solid materials degrade over time. Tungsten is the current best option for solid surfaces, but even tungsten erodes, develops cracks, and changes its material properties under intense neutron irradiation. In a commercial reactor expected to run for years, replacing damaged wall components would mean shutting down and opening the radioactive vacuum vessel, an operation measured in months and millions of dollars.
A flowing liquid metal surface sidesteps this entirely. Damage is carried away with the flow. Fresh material continuously replaces what was lost. The liquid cannot crack or develop the kind of microstructural damage that degrades solids. EAST’s experiments with flowing lithium limiters demonstrated that such a system can operate stably during real tokamak discharges and showed beneficial effects on plasma purity and divertor heat loading. The technology is still far from reactor-ready, but it is no longer purely theoretical either. The two generations of flowing lithium limiters tested on EAST have moved the concept from laboratory curiosity to something with real tokamak operating hours behind it.