Eurofer 97 loses most hardness under the shortest disruption pulses

Magnum-PSI held Eurofer 97 at a steady 500°C while layering millisecond laser pulses on top, closer to a real disruption than earlier lab methods

(Image courtesy of DIFFER)

Eurofer 97 is the candidate structural steel for the first wall and breeder blanket of Europe’s DEMO fusion reactor. New experimental work from the University of Bristol and DIFFER shows the material loses up to 13% of its surface hardness after exposure to disruption-representative heat pulses. Researchers used the Magnum-PSI linear plasma device to replicate the millisecond transients a first wall faces during a mitigated disruption, and the damage extends 110 to 150 micrometres into the material.

How the magnum-PSI test replicated a plasma disruption

The team worked with DIFFER, testing Eurofer 97 sourced via the UK Atomic Energy Authority from the Karlsruhe Institute of Technology. They exposed multiple samples to steady-state argon plasma at 500°C, using a superconducting magnet with a field strength of 0.872 tesla to hold the plasma beam in place. A high-power laser then struck the surface for 0.5, 1 or 3 milliseconds at a time, running for 1, 100 or 1000 pulses and pushing the surface to roughly 950 to 1000°C before it cooled back to baseline. These durations bracket the range predicted for mitigated disruptions, around 0.5 milliseconds for ITER and 2 to 4 milliseconds for EU-DEMO, where such events could recur more than 2500 times over a 30-year operating life.

Previous attempts to replicate this damage relied on standalone lasers with heating rates far slower than a real disruption, and pulse lengths that ran up to a hundred times longer than the events they modelled. Magnum-PSI’s combination of a steady plasma beam and a millisecond-scale laser gave the team a closer proxy for the plasma and heat-flux conditions a first wall actually experiences. That distinction mattered, because the team wanted to know not just whether Eurofer 97 degrades under these conditions, but how the exact duration of a heat pulse changes the way it degrades.

Eurofer 97’s hardness fell regardless of pulse length

After 1000 pulses, hardness fell by 13% for the 0.5 millisecond exposures, 6.3% for the 1 millisecond exposures and 11.2% for the 3 millisecond exposures, each measured against the plasma-only baseline. The shortest pulses were not the gentlest, and the reason lies in how the microstructure responded to different exposure times.

Grain diameters shrank by 25% at 0.5 milliseconds and by 43% at 1 millisecond, a pattern the authors attribute to dynamic recrystallisation, where thermal stress breaks up existing grains and reforms them smaller. Smaller grains should normally make a steel harder, through Hall-Petch strengthening, since more grain boundaries make it harder for dislocations to move through the material.

Hardness mapping of Eurofer 97 after disruption-scale pulse testing at 0.5, 1 and 3 milliseconds

(Image courtesy of University of Bristol and Materials & Design, Photo: H.E. Tipping et al.)

That relationship held only briefly. By 3 milliseconds, grain diameters had reduced by just 5.5%, as the recrystallised grains had time to grow back together, a process called grain coarsening that works against Hall-Petch strengthening. It’s a plausible reason why the 3 millisecond samples lost more hardness than the 1 millisecond samples, despite starting from the same recrystallisation process.

Electron microscopy of the 1000-pulse, 3 millisecond sample found chromium, tantalum and vanadium precipitates dissolving to a depth of around 110 micrometres, a figure the authors cross-checked against an independent carbon diffusion calculation. These precipitates normally pin dislocations in place, adding a second source of strength on top of the grain boundaries, so their disappearance removes hardening capacity that grain refinement alone cannot replace. Hardness mapping placed the affected zone slightly deeper, between 120 and 150 micrometres, a gap the team attributes to the resolution limits of the micro-indenter rather than a genuine difference in damage extent.

The dissolved precipitates point to something more fundamental happening beneath the surface. As the steel cycled between 500°C and roughly 950°C, it stayed within a temperature band where iron transforms to its ferrite and austenite phases, never crossing back down through the 383°C threshold where martensite normally reforms on cooling. Without that transition, the tempered martensitic structure Eurofer 97 starts with could not reassert itself, and carbon that the precipitates had locked in simply diffused into the bulk rather than reforming them near the surface. A separate calculation of that carbon diffusion length, run independently of the microscopy, arrived at a combined figure of around 112 micrometres, matching the microscopy-based damage depth closely enough to support the same underlying explanation.

What the disruption damage means for first wall procurement

Eurofer 97 is the European reference structural material for DEMO’s first wall and breeder blanket, and it is already destined for Europe’s Test Blanket Modules inside ITER, where it will help demonstrate tritium breeding for future reactor fuel cycles. The steel normally sits behind tungsten armour, and this experiment deliberately removed that layer to isolate the structural material’s own response to a disruption. The authors note that cracking or erosion of the tungsten tiles could still expose Eurofer 97 directly during a fault, which keeps the underlying damage mechanism relevant even under normal shielding assumptions.

This result also marks a shift from earlier laser-only ageing studies on the same steel. Hargreaves et al. found that slower, longer pulses at 850°C caused severe grain coarsening alongside new martensite formation and a 32% hardness drop, roughly double what any pulse duration produced here. The faster, more disruption-realistic heating in this study strongly suggests a predominantly ferritic microstructure formed through the mechanism above instead, with a smaller but still significant hardness penalty split between two separate weakening mechanisms rather than one dominant one.

For those thinking ahead to component lifetimes, the sensitivity to pulse duration underscores how tightly disruption mitigation performance will need to be controlled. A shift of half a millisecond changed grain size reduction by close to 20 percentage points.

Industrial scaling adds further uncertainty. The laser produced a damage spot around 2 millimetres across, well below the square-metre wetted area a full-scale first wall would see during a real disruption. The authors flag that cooling behaviour and total damaged area could both differ at reactor scale.

Tungsten armour and combined irradiation are next

The Bristol team plans to test how a tungsten armour layer changes this damage profile, since the current results describe an unshielded scenario that approximates a fault condition. The researchers deliberately excluded neutron irradiation from this study to isolate the thermal effects, and combining the two is a stated interest for future work. They also want to establish how much transient exposure the material can withstand before recrystallisation gives way to sustained grain coarsening, a threshold that would matter directly for predicting when first wall components need replacing.

Stay ahead in the fusion revolution explore more breakthroughs from leading innovators in clean energy technology.