Keeping the fire of a star burning here on Earth. One of the greatest obstacles to realizing fusion energy is not the generation of plasma itself, but the durability of the vessel that must confine such extreme heat. In massive projects such as the International Thermonuclear Experimental Reactor (ITER), currently under construction in southern France, a component called the divertor is installed beneath the reactor core plasma, which reaches temperatures of hundreds of millions of degrees. The heat load on the divertor's surface reaches up to 20 megawatts per square meter under steady-state conditions. And when sudden plasma eruptions known as edge-localized instabilities occur, the instantaneous heat load can spike to several times that figure. This far exceeds the punishing loads experienced by spacecraft surfaces during atmospheric reentry.
Tungsten has an extremely high melting point of 3,422°C, but its thermal conductivity and workability have their limits. A common design approach, therefore, is to place tungsten as an armor material in direct contact with the plasma, and bond it to a copper alloy heat sink behind it—one with excellent thermal conductivity—to rapidly channel heat away into cooling water. If the heat sink's cooling capacity is exceeded and the copper melts and collapses, the entire high-vacuum system faces catastrophic failure.
However, precisely understanding how copper atoms respond to sudden heat loads, and at what stage they reach their limit, has long been difficult. At the frontier of materials development, efforts to pinpoint the limits of metals through both experimentation and computation have continued for years.
A Half-Century-Old Consensus Overlooked the Limits of Post-Hoc Evaluation
Post-Hoc Evaluation That Reveals Only Frozen Traces
Melting in metals is a phase transition phenomenon in which the order of a crystal lattice—atoms arranged in a regular pattern—is broken apart by thermal vibration, causing a shift to a disordered liquid state. To verify this in a laboratory setting, conventional test methods have had to rely on post-hoc evaluation known as the "cook-and-look" approach: subjecting a metal sample to intense heat, then examining the traces of melting left behind after it has cooled and solidified.
This approach only allows researchers to examine the resolidified lump of metal after melting has finished. What happens during the trillionths of a second while temperature is rising—how the crystal lattice distorts, and from where it begins to collapse—remained a complete black box. There was no way to directly observe the decisive moment when atoms depart from their original positions and transform into liquid.
The Superheating Limit Scenario Derived from Simulation
As a result, materials scientists have created virtual copper inside computers and used simulation methods such as molecular dynamics to infer behavior under extreme conditions. Molecular dynamics is a technique that calculates the motion of millions of individual atoms according to Newtonian mechanics. However, tracking the interactions of vast numbers of atoms on a computer requires simplifying the conditions to some degree.
The scenario produced by computational models was remarkably clear. When copper is subjected to a sudden heat load, the surface first begins to melt near its normal melting point of 1,085°C. As heat propagates inward, the surrounding solid material attempts to suppress thermal expansion, causing internal pressure to rise sharply. This pressure raises the melting point, allowing the core to remain solid and continue withstanding the heat.
However, the models predicted that the moment the temperature reached 1,424°C, the thermal kinetic energy of the atoms could no longer be contained, and the entire crystal structure would collapse instantaneously. This temperature, roughly 1.25 times the normal melting point, has been defined as the theoretical upper limit at which copper can maintain its crystalline form—the superheating limit. Recent AI-driven screening methods for exploring new materials have also been built on this assumption of instantaneous collapse.
A Femtosecond Electron Camera Captures a Delayed Collapse
Seeing Through Matter with Electron Waves via MeV-UED
An international research team centered at the U.S. Department of Energy's SLAC National Accelerator Laboratory has demonstrated, through a study published in Nature Communications (DOI: 10.1038/s41467-026-75970-1), that this firmly held assumption diverges from actual physical reality.
To directly observe what happens inside a metal during heating, the team employed MeV-UED (Megaelectronvolt Ultrafast Electron Diffraction). While X-ray-based diffraction techniques are widely used, electron-based diffraction has its own distinct strengths. Because electrons carry an electric charge, they interact strongly with atomic nuclei and electron clouds in matter, making them well suited for capturing structural changes in extremely thin, nanometer-scale samples.
This instrument functions as a giant electron camera: it fires a high-energy electron beam, accelerated to over 99 percent of the speed of light, at a sample and records on a screen the scattering pattern produced as electrons pass through the gaps between atoms. Accelerating electrons to such extremes shortens their quantum-mechanical wavelength dramatically, allowing precise discrimination of individual atomic arrangements. Its greatest strength lies in its temporal resolution. The pulse width—equivalent to a shutter speed—is set at the femtosecond scale (one-quadrillionth of a second). Within this vanishingly small time frame, in which light travels a mere 0.3 micrometers, the device captures continuous, blur-free images of atoms vibrating intensely under heat.
Gradual Liquefaction That Proceeds Beyond the 1,424°C Wall
The research team irradiated an extremely thin pure copper film, only tens of nanometers thick, with a powerful short-pulse laser to artificially create extreme heating conditions. Upon femtosecond laser irradiation, electrons inside the metal first absorb intense energy, and that heat is subsequently transferred, with a delay, to the atomic lattice. Immediately after this non-equilibrium state was induced, the team fired the electron beam and tracked changes in the diffraction pattern.
When a crystal is regularly ordered, the screen displays clear dots or ring patterns. When the crystal melts and becomes disordered, this pattern blurs into indistinct outlines.
The results produced by real-time observation vividly contradicted the simulation's predictions. Even after copper's temperature reached the theoretical limit of 1,424°C, the sharp dots in the diffraction pattern did not vanish instantaneously. The team meticulously recorded how the core maintained a regular atomic arrangement well beyond the predicted limit temperature, gradually melting in stages over a span of several picoseconds to several tens of picoseconds (trillionths of a second).
| Comparison Axis | Conventional Molecular Dynamics Simulation | Real-Time Observation via MeV-UED |
|---|---|---|
| Behavior at the superheating limit (1,424°C) | Crystal lattice collapses instantaneously into liquid | Crystal order is maintained; melting proceeds gradually in stages |
| Assumed pressure conditions | Static conditions (uniform, fixed pressure from all directions) | Dynamic conditions (non-uniform pressure fluctuations and physical relaxation) |
| Behavior at nanoscale grain boundaries | Assumes a uniform, complete melting process | Localized disordering occurs preferentially at grain boundaries before the melting point is reached |
Dynamic Pressure Creates an Escape Route for Atoms and a Buffering Effect at Grain Boundaries
The Breakdown of Static Assumptions and the Physics of Pressure Relaxation
Why did such a significant gap emerge between the meticulously constructed computational model and the actual experimental results? The research team traced the cause to the assumption of static conditions underlying the simulations.
Conventional models assumed that heated copper experiences uniform, constant pressure from all directions, with atoms remaining fixed within their original framework as they absorb heat. In reality, however, the thermal expansion accompanying laser heating is far from uniform. The extreme temperature difference between the surface region, which directly absorbs the laser's energy, and the deeper region, which heat has not yet reached, generates powerful shock waves and complex, dynamic pressure waves within the material.
Under these dynamic pressure conditions, copper atoms are not fixed in place as they are inside a computer simulation. The team observed atoms shifting from regions of higher pressure to regions of lower pressure, with crystal planes slightly sliding past one another to relieve accumulated stress on their own.
This physical relaxation effect, driven by atomic shifts, lowered localized pressure and allowed the crystal lattice to preserve structural order for far longer, even beyond the limit temperature, without completely disintegrating. Real materials proved to be far more flexible and resilient than the idealized models used in calculations.
The Role of Premelting Originating at Grain Boundaries
Furthermore, the electron diffraction images captured another important phenomenon. Even before the copper's overall temperature had fully reached its melting point, atomic arrangements had already begun to disorder locally along the boundaries between nanometer-sized crystal grains.
Metal is not a single, perfect crystal but a polycrystalline material made up of countless small grains fitted together like a patchwork. The grain boundaries, where individual grains meet, feature irregular atomic arrangements and are energetically unstable. As a result, bonding is weaker here than in the uniform interior structure, and when heat is applied, order begins to break down preferentially at these boundary regions.
This process, known as premelting, acts as a buffer that absorbs the overall thermal energy and the volumetric changes associated with thermal expansion. The analysis suggests that these disordered grain-boundary regions function as a kind of cushion, contributing to the prevention of sudden,全体-wide liquefaction of the crystal.
Revised Modeling That Paves the Way for Unknown Alloy Development
Prospects for Advancing Computational Models
This discovery in the microscopic world provides an opportunity to fundamentally reconsider the process of developing materials for the massive wall structures of fusion reactors. Until now, engineers have relied on prediction models based on static assumptions as an absolute standard for selecting materials capable of withstanding extreme heat loads. If the mechanisms of dynamic pressure relaxation and grain-boundary premelting can be incorporated into computational formulas, the accuracy of predicting thermal response under extreme conditions would improve significantly.
This also opens the possibility that materials previously deemed unsuitable due to their calculated limit temperature may, in fact, be reconsidered as promising candidates with sufficient resistance. In building digital twins for materials science, verifying computational predictions against real physical phenomena observed under extreme conditions is an indispensable step.
Applying These Findings to More Complex Structures for Real-World Reactors
That said, not all conditions have been fully clarified. The subject of this experiment was pure, single-element copper. The components actually envisioned for heat absorption in fusion reactors are complex alloys, enhanced in strength and heat resistance through the addition of trace amounts of chromium and zirconium.
Whether atoms can move smoothly and relieve pressure in the same way within an irregular crystal lattice containing a mixture of different elements—just as they do in pure copper—remains unverified. It is also unknown whether the presence of dissimilar elements might obstruct this relaxation motion, triggering a collapse mechanism entirely different from that of pure copper.
The SLAC research team plans to conduct additional experiments deliberately imposing static conditions—uniform pressure from all directions—to further clarify the contours of the discrepancy between simulation and reality. At the same time, they intend to advance their use of electron diffraction technology to unravel the behavior of more complex copper alloys. Through the patient work of decoding, one step at a time, how atoms behave when exposed to extreme heat, the shape of a vessel capable of safely containing unknown thermal energy should come into view.