The Joint European Torus, known universally as JET, was the world’s largest and most powerful tokamak for nearly four decades, and the only operational fusion device ever to run experiments with the actual fuel mix planned for a commercial reactor: deuterium and tritium. Located at the Culham Centre for Fusion Energy in Oxfordshire, England, JET operated from 1983 until December 2023, producing landmark results that shaped how physicists understand burning plasmas and how engineers design the machines meant to contain them. Its record-breaking fusion experiments, material science breakthroughs, and role as the primary testbed for ITER make it one of the most consequential scientific instruments ever built.
What JET Was Built to Do
JET began operating in 1983 as a joint project of European fusion laboratories, and its mission evolved over four decades. The original goal was straightforward: build a tokamak large enough to approach the plasma conditions needed for fusion energy, and study the physics that govern plasma behavior at those scales. Over time, JET became much more than a physics experiment. It served as the proving ground for technologies, materials, and operational strategies that future reactors would need. During its 40 years of operations, JET consistently pushed both physics and engineering boundaries, providing the fusion community with a testing ground for theories and innovative ideas across transport, plasma heating, plasma-wall interaction, and deuterium-tritium experiments.1Plasma Physics and Controlled Fusion. 40 years of JET operations: a unique contribution to fusion science
A tokamak works by confining hot plasma in a doughnut-shaped magnetic field. The plasma needs to reach temperatures of over 100 million degrees for the nuclei inside it to fuse. JET’s vacuum vessel, roughly 6 meters across and 4 meters tall, was designed to hold plasmas carrying several million amperes of electrical current. That current, combined with powerful external magnets, creates the magnetic cage that keeps the plasma from touching the vessel walls. Getting this cage to work reliably at extreme temperatures and pressures was the core engineering challenge JET spent decades refining.
Fusion Records and the Deuterium-Tritium Campaigns
JET’s most celebrated achievement was running actual fusion fuel. Most tokamaks operate with pure deuterium or hydrogen because those fuels are easier and safer to handle. JET was the first tokamak to use a substantial deuterium-tritium (D-T) mix, and it did so in three major campaigns: the Preliminary Tritium Experiment in 1991, DTE1 in 1997, and DTE2 in 2021-2023.
The 1997 DTE1 campaign set records that stood for over two decades. Peak fusion power reached 16.1 MW, and the machine sustained about 4 MW of fusion power for 4 seconds. The ratio of fusion power to input power hit 0.62, meaning the plasma was producing energy at a rate approaching two-thirds of what was being pumped in.2Journal of Nuclear Materials. D–T experiments in the JET tokamak That campaign also provided the first clear demonstration of alpha particle heating, where the helium nuclei produced by fusion reactions deposited their energy back into the plasma, further raising its temperature. Studies confirmed that the alpha particle heating matched classical theoretical predictions, with maximum central temperature and energy content lining up with the peak in fusion yield as the tritium fraction was varied.3Nuclear Fusion. Alpha particle studies during JET DT experiments
The DTE2 campaign, which ran more than two decades later, broke JET’s own record by producing 59 megajoules of fusion energy in a single pulse and sustaining fusion power over a five-second window. One notable experiment at 3.0 MA plasma current generated 18 MJ of fusion energy over 5 seconds, providing lessons about operating at high current with the newer tungsten wall environment.4Plasma Physics and Controlled Fusion. Development of a high-current, high fusion performance scenario on JET: physics insights and operational challenges The DTE2 results were especially valuable because JET had by then been fitted with an ITER-like wall, making the data directly relevant to ITER’s planned operating conditions.
The Isotope Effect and Why Fuel Choice Matters
Running different fuels through the same machine allowed JET to answer a question that cannot be studied any other way: how does the mass of the fuel ions affect plasma confinement? This is called the isotope effect, and it has practical consequences for ITER and future reactors.
JET experiments showed that the power needed to enter the high-performance H-mode regime is inversely proportional to the effective isotopic mass of the plasma. Heavier fuel ions make the transition easier.5Plasma Physics and Controlled Fusion. H-mode power threshold and confinement in JET H, D, D-T and T plasmas In dedicated comparisons between pure deuterium and pure tritium L-mode plasmas, the tritium pulses showed roughly 28% higher scaled energy confinement than their deuterium counterparts.6Nuclear Fusion. Isotope mass scaling and transport comparison between JET Deuterium and Tritium L-mode plasmas Even the way the plasma rotates on its own changes with fuel mass: increasing isotope mass increased thermal energy confinement time while shifting certain transport transitions to higher densities.7Nuclear Fusion. Isotope effects on intrinsic rotation in hydrogen, deuterium and tritium plasmas
These findings matter because ITER will operate with a 50-50 D-T mix but will spend years commissioning in hydrogen and deuterium before introducing tritium. Understanding exactly how performance scales with fuel mass is essential for predicting ITER’s behavior from its early, non-nuclear phases.
The ITER-Like Wall
One of JET’s most significant upgrades happened in 2010-2011, when its entire interior was replaced with new plasma-facing materials chosen to match what ITER plans to use: beryllium tiles in the main chamber and tungsten in the divertor, the region at the bottom of the vessel where exhaust plasma is directed.8Fusion Engineering and Design. Overview of the JET ITER-like Wall Project This was the same combination of materials foreseen for ITER, and testing it in a real tokamak environment was considered critical.9Fusion Engineering and Design. The JET ITER-like wall experiment: First results and lessons for ITER
Before the upgrade, JET had carbon fibre composite walls. Carbon is forgiving in a plasma environment, but it has a serious drawback: it absorbs and traps tritium fuel at high rates, which is both a safety concern and a waste of expensive radioactive fuel. The switch to beryllium and tungsten dramatically reduced fuel retention. After the new wall was installed and campaigns began in 2011, researchers investigated retention properties and developed new operational scenarios suited to the metal wall environment.10Nuclear Fusion. Overview of the JET results with the ITER-like wall
Tungsten brought its own headaches, though. It is extremely resistant to erosion and melting, but tiny amounts that get sputtered into the plasma can radiate enormous amounts of energy, cooling the core and degrading performance. Modeling identified neoclassical transport enhanced by poloidal asymmetries as the dominant mechanism responsible for tungsten accumulating in the central core of the plasma, while turbulent diffusion and temperature screening effects could sometimes prevent it.11Plasma Physics and Controlled Fusion. Theoretical description of heavy impurity transport and its application to the modelling of tungsten in JET and ASDEX upgrade Experimentally, JET found that when only neutral beam heating was used, tungsten accumulated dangerously in the core, approaching concentrations high enough to degrade performance. But adding central ion cyclotron resonance heating caused the tungsten concentration to decay, an effect linked to changes in internal plasma instabilities that altered how impurities were transported.12Plasma Physics and Controlled Fusion. Observations on the W-transport in the core plasma of JET and ASDEX Upgrade This finding gave ITER operators a practical tool: use central radio-frequency heating to flush tungsten out of the plasma core.
Fuel recovery also became an active area of study. During the D-T campaigns, researchers tested methods to remove fuel trapped in the wall materials. One technique using dedicated plasma configurations achieved a 79% reduction in deuterium measured in the most targeted wall region after a series of cleaning plasmas.13Nuclear Fusion. Overview of fuel retention and recovery in JET deuterium–tritium operation For ITER, which will use kilograms of tritium that must be tracked and recycled, these results are directly applicable.
Heating the Plasma
Getting a plasma to fusion temperatures requires more than just running electrical current through it. JET used three main heating methods, each contributing something different. Neutral beam injection (NBI) fires high-energy atoms into the plasma, where they collide with plasma particles and transfer their kinetic energy. Ion cyclotron resonance heating (ICRH) uses radio waves tuned to the frequency at which ions spiral around magnetic field lines, pumping energy directly into the ions. Ohmic heating, generated by the plasma current itself, provides the initial temperature rise but becomes less effective at higher temperatures.
The interplay between these methods turned out to be more interesting than simply adding their power levels together. In experiments heating deuterium at the fundamental cyclotron frequency, adding roughly 25% more power through ICRH on top of NBI boosted fusion power output by 30-50%, depending on beam configuration. Ion temperatures jumped from about 4 keV to 5.5 keV, and a clear synergistic effect appeared: the combined heating produced more neutrons than the sum of each method operating alone.14Plasma Physics and Controlled Fusion. Fundamental ion cyclotron resonance heating of JET deuterium plasmas A more advanced technique called the 3-ion ICRF scenario, which exploits a resonance layer between two different ion species, achieved strong core-localized power deposition, efficiently generating energetic deuterium ions and driving up the neutron rate.15Nuclear Fusion. Plasma heating and generation of energetic D ions with the 3-ion ICRF + NBI scenario in mixed H-D plasmas at JET-ILW These heating schemes are directly relevant to ITER, where understanding how different heating methods interact at reactor scale is a priority.
Taming Plasma Instabilities
A fusion plasma in H-mode performs well overall, but it comes with a price: periodic bursts of energy called edge-localized modes, or ELMs, that erupt from the plasma edge and slam into the vessel walls. Left uncontrolled, ELMs in a reactor-scale machine would erode wall components unacceptably fast. JET devoted substantial effort to understanding and controlling them.
One approach used external magnetic perturbation fields generated by error field correction coils mounted between JET’s transformer limbs. Applying a static low-order perturbation field during H-mode operation quadrupled the ELM frequency while shrinking the energy dumped per ELM to less than 2% of the total stored energy, down from about 7%. Energy confinement time degraded by at most 20%.16PubMed. Active control of type-I edge-localized modes with n=1 perturbation fields in the JET tokamak Follow-up experiments tested both n=1 and n=2 perturbation fields, confirming that ELM frequency could be pushed from around 30 Hz to about 120 Hz while keeping the energy loss per ELM below 2%.17Nuclear Fusion. Active control of type-I edge localized modes with n = 1 and n = 2 fields on JET The basic idea, making many small ELMs instead of a few large ones, is now the leading ELM control strategy planned for ITER.
JET also studied how plasma shape affects the edge. Increasing the triangularity of the plasma cross-section improved pedestal confinement, primarily by raising pedestal density while reducing ELM frequency. Intriguingly, this improvement persisted even when researchers deliberately destabilized the pedestal using vertical kicks to pace the ELMs, suggesting that higher triangularity favorably affects the transport processes between ELMs, not just the stability boundary itself.18Plasma Physics and Controlled Fusion. Pedestal particle balance studies in JET-ILW H-mode plasmas
Beyond ELMs, the sudden and catastrophic collapse of the entire plasma, known as a disruption, remains one of fusion’s hardest unsolved problems. JET tested shattered pellet injection (SPI) as a disruption mitigation tool, firing frozen pellets of material into the plasma at high speed to spread the energy over a larger area and reduce localized damage. These experiments were conducted at stored energies exceeding 7 MJ, providing data to inform ITER’s disruption mitigation system.19Nuclear Fusion. Disruption thermal load mitigation with shattered pellet injection on the Joint European Torus (JET)
Remote Handling and AI for Plasma Control
Working inside a tokamak that has run tritium is not something humans can do. The vessel becomes activated by neutron bombardment, making hands-on maintenance impossible. JET pioneered remote handling technology that allowed all in-vessel work to be done robotically. The system used two articulated booms entering through opposing mid-plane ports: one to transfer tools and components, the other to position a two-armed, force-feedback manipulator called Mascot that could perform delicate tasks inside the vessel.20Fusion Engineering and Design. Development and application of high volume remote handling systems in support of JET and ITER The entire ITER-like wall upgrade, involving thousands of individual tiles, was carried out using this remote handling system. The experience informed the design of ITER’s own remote maintenance strategy.
JET also became a testing ground for artificial intelligence applied to plasma control. Disruption prediction models based on deep learning, specifically the Fusion Recurrent Neural Network (FRNN), were developed using large datasets from JET. These predictors now produce both a “disruption score” indicating the probability of an imminent collapse and a “sensitivity score” that identifies the physical reasons behind the warning, adding a layer of interpretability that helps operators decide how to respond.21Contributions to Plasma Physics. Implementation of AI/DEEP learning disruption predictor into a plasma control system In parallel, real-time bolometry tomography and generative topographic mapping algorithms were developed and integrated into JET’s plasma control system for disruption avoidance during the D-T campaigns.22Fusion Engineering and Design. Progress in preparing real-time control schemes for Deuterium-Tritium operation in JET
Experiments with Alternative Fusion Fuels
While D-T fusion was JET’s headline act, the machine also explored D-helium-3 reactions, which produce charged alpha particles and protons rather than neutrons. A largely aneutronic reaction like D-³He is attractive in principle because it would cause far less neutron activation of the reactor structure. JET used the 3-ion radio-frequency heating scenario to accelerate deuterium beam ions to MeV-range energies in mixed D-³He plasmas, generating core-localized alpha particles at rates of about 2 × 10¹⁶ per second at moderate input power levels.23Nuclear Fusion. Generation and observation of fast deuterium ions and fusion-born alpha particles in JET D-3He plasmas with the 3-ion radio-frequency heating scenario These experiments produced 3.7 MeV alpha particles similar in energy to those from D-T fusion, along with 14.6 MeV protons, giving researchers a way to test fast-ion diagnostics in conditions relevant to future burning plasmas.24Nuclear Fusion. Fusion product studies via fast ion D–D and D–3He fusion on JET
Nobody expects D-³He reactors anytime soon because the temperatures and confinement requirements are far more demanding than for D-T, but these experiments were valuable for validating diagnostic instruments and theoretical models of energetic particle behavior. ITER will produce large populations of fast alpha particles in its D-T plasmas, and the diagnostic techniques tested in JET’s D-³He runs will help ITER researchers measure how those particles behave.
Decommissioning a Fusion Machine
JET’s final plasma pulse occurred in December 2023, and the machine has entered decommissioning. This process is itself a first: no major D-T-capable tokamak has ever been decommissioned before, so JET is writing the playbook. The vessel is radioactive from decades of neutron bombardment, and the interior surfaces carry residual tritium embedded in wall materials. Accurate predictions of the nuclear waste inventory and shutdown dose rates based on actual operations were required to guide waste management, and improved computational methods increased the precision of those predictions for a more rigorous decommissioning process.25Fusion Engineering and Design. Nuclear waste analysis of JET, utilising shutdown dose rate code enhancements
The decommissioning is expected to take years and will rely heavily on the remote handling expertise JET developed over its operational life. Every activated component must be characterized, removed robotically, and sorted by waste classification. For the fusion industry, the lessons learned here about managing activated materials, recovering residual tritium, and safely dismantling a large tokamak after decades of nuclear operation will be directly useful when ITER and future demonstration reactors eventually reach end of life. JET’s contributions, in other words, did not stop when its plasma went cold.
The H-Mode Pedestal and Confinement Under the Metal Wall
When JET switched from carbon to its beryllium-tungsten ITER-like wall, one of the first questions was whether the new materials would change the plasma confinement properties that physicists had spent years characterizing. The answer was yes, at least partially. The high-resolution Thomson scattering diagnostic mapped the pedestal structure, the steep-gradient region at the plasma edge that sustains H-mode performance, across a range of plasma shapes after the new wall was installed.26Nuclear Fusion. The H-mode pedestal structure and its role on confinement in JET with a carbon and metal wall Early ITER-like wall campaigns showed that achieving the same level of confinement as the carbon wall required more gas fueling and active control of impurities, because the beryllium wall produced fewer radiating impurities to help distribute heat at the edge. Researchers had to adjust their playbooks, developing new operational recipes that eventually recovered much of the lost performance.
The magnetic separatrix configuration, which creates an X-point at the bottom of the plasma cross-section to divert exhaust particles into the divertor, was itself an early JET achievement. Experiments with separatrix plasmas carrying up to 3 MA of current demonstrated high recycling regimes in ohmic discharges and, when neutral beam heating was applied, achieved the energy confinement improvements characteristic of H-mode.27Nuclear Fusion. Magnetic separatrix experiments in JET This early work laid the groundwork for the diverted plasma operation that became standard in nearly every tokamak built since.