DIII-D Screening Narrows Fusion Wall Material Choices

Fusion pilot plants need plasma-facing materials that can survive more than a clean heat-load test. The wall and divertor surfaces must tolerate transient ELMs, particle fluence, neutron damage, redeposition, mixed-material chemistry, and geometry-driven edge effects in the same operating window. This paper reports a coordinated DIII-D qualification campaign that puts 44 candidate materials from 12 institutions through comparable tokamak exposures rather than evaluating each material in a separate laboratory setting.

The result is not a single winner. It is a screened map of which material classes failed by cracking, localized melting, mass loss, deuterium retention, or particulate release, and which ones remain credible enough for component-scale follow-up.

Core Contribution

The main contribution is the comparative design of the campaign. The authors used the Divertor Materials Evaluation System, or DiMES, to expose flush samples and 10° angled samples under Ohmic, L-mode, and H-mode conditions with ELMs. The heat-flux range spans 0.2–2.5 MW m$^{-2}$ on flush geometries and 10–15 MW m$^{-2}$ on angled geometries, with a 2 Hz rastered outer-strike-point H-mode scenario used to broaden the exposure profile.

Figure 1 is useful because it shows the experimental constraint behind the comparison: the sample response depends not only on nominal scenario labels, but on where each DiMES button sits relative to the strike-point sweep and heat-flux peak.

That matters for interpretation. The paper does not claim that a short DiMES exposure reproduces a pilot-plant lifetime. It claims that a shared tokamak exposure can reveal first-order failure modes across materials that would otherwise be hard to rank.

Technical Approach

The candidate set spans engineered tungsten, modified tungsten, refractory multi-principal-element alloys, ultra-high-temperature ceramics, low-Z ceramics, and renewable boron concepts. The analysis combines post-exposure microscopy, mass change, spectroscopy, deuterium release, and in-situ erosion measurements. Several samples were deliberately tested in geometries that amplify edge heating, because future divertor components will not see ideal flat-plate boundary conditions.

For engineered tungsten, the most informative comparison is among fiber-reinforced and micro-castellated architectures. Figure 2 shows that these designs did not all fail in the same way after ELMy H-mode exposure: long-fiber Wf/W showed crack arrest at fiber interfaces, while microstructured tungsten tended to localize damage in segmented melting regions. That is a more useful engineering signal than a binary pass/fail label.

The campaign also tested manufacturing route effects. Additively manufactured tungsten retained near-reference integrity in some conditions, but the paper distinguishes high-density LPBF material from EB-PBF material and AM W-Ta. Figure 5 shows the relatively modest surface evolution in LPBF and conventionally manufactured pure tungsten under L-mode and H-mode exposures, while the text reports stronger degradation for AM W-Ta under angled high-flux conditions.

Results and Analysis

The clearest positive results are concentrated in a few classes. Engineered tungsten architectures largely retained macroscopic integrity under angled ELMy H-mode loading. Long-fiber Wf/W showed the clearest crack-arrest behavior, and WfSiCf/W preserved overall integrity despite edge melting and erosion. W-Re and K-doped W behaved close to ITER-grade tungsten under high-heat-flux H-mode exposure, with no major macroscopic damage and comparatively low mass loss.

The additively manufactured results are more conditional. High-density AM-W showed minimal surface evolution, while the approximately 97% dense LPBF material remained viable but showed possible deuterium blistering. EB-PBF AM-W kept its grain texture but developed localized roughening and microcracking. AM W-Ta was the warning case: apparent mass loss increased from 0.64 mg on the flat sample to 2.19–2.87 mg on 10° angled samples, with localized coral-like morphology. The paper’s interpretation is that density, texture, and processing route are not secondary details for tungsten qualification.

Among refractory alloys, VTaHfMo was the strongest multi-principal-element alloy result, remaining compositionally stable with minimal morphology change after flush H-mode exposure. Zr-Ti-based alloys showed voiding, grooves, melting, or segregation. Chromium received a narrower test: the initial in-situ tokamak measurement produced gross sputtering values of order 10$^{-2}$, which is mainly a benchmark for future erosion modeling rather than a full qualification.

The ceramic picture is mixed. CVD SiC retained macroscopic integrity and was the best low-Z ceramic in this set, but spectroscopy indicated preferential silicon loss with an effective Si erosion yield near 0.5, about 5–10× above prior DIII-D trends. TiB$_2$ showed the lowest D$_2$ release in the Ohmic set at 1.28 × 10$^{-7}$ mol and limited blistering, but other ceramics were less convincing: Si$_3$N$_4$ failed after the first discharge, B$_4$C lost boron preferentially, and ZrC lost about 7 mg under H-mode exposure. NbC and (Nb$_{0.5}$Ta$_{0.5}$)C were stronger carbide candidates, with only 0.02–0.03 mg mass loss.

The renewable boron pebble-rod concept demonstrated controlled recession rather than catastrophic loss, but recovery was incomplete. About 13% of released boron was ionized near the outer strike point and up to 50% was recovered locally; the rest was inferred to deposit as dust or redistribute nonlocally. That makes the concept interesting for replaceable, low-Z surfaces, but the paper is clear that particulate transport and recovery control remain the gating issues.

Evidence Reading

The evidence is strong as a comparative tokamak screening study: many materials, shared DIII-D scenarios, and multiple diagnostics. It is weaker as final pilot-plant qualification. Exposure durations are short, neutron damage is only partly represented, and several findings depend on local geometry and strike-point alignment. The useful takeaway is therefore a down-selection: engineered tungsten, high-density AM-W, W-Re/K-doped W, VTaHfMo, Nb-containing carbides, and CVD SiC deserve the next round; AM W-Ta, Zr-Ti alloys, ZrC, Si$_3$N$_4$, B$_4$C, and boron pebble rods have specific failure modes to resolve before they can be treated as component candidates.