2026 Volume 21 Article ID: 1405043
A medium-scale tungsten (W)-sheet/stainless steel (SUS) first-wall component with an oxide dispersion strengthened copper (ODS-Cu; GlidCop®) intermediate layer was developed using the advanced brazing technique (ABT) for high-temperature first-wall applications in the Kyushu University Experiment with a Steady-State Spherical Tokamak (QUEST). To ensure its reliability, the component’s thermal response was first evaluated in a vacuum test chamber, demonstrating a uniform surface temperature distribution up to 400°C with no problematic outgassing. Following these tests, the component was installed in the QUEST first-wall and subjected to temperatures up to 350°C during plasma operations from April 2 to 18, 2024. Surface inspections and elemental analyses performed after the campaign confirmed that the brazed interfaces remained robust; no further delamination or cracking occurred beyond the minor exfoliation initially observed after fabrication, despite long-term thermal loads and plasma-wall interactions. These results demonstrate the technical feasibility of using brazed W-sheet components for high-temperature first-wall applications in steady-state tokamaks.
Tungsten (W) is considered one of the most suitable candidate materials for first-wall armour materials in future fusion reactors, including DEMO, owing to its high melting point, low hydrogen isotope solubility, and low sputtering yield against plasma particles [1–4]. Several W-coating techniques, such as vacuum plasma-sprayed (VPS-W), atmospheric plasma-sprayed (APS-W) and explosion welding, would be applicable to covering the entire first-wall surface, which is the largest plasma facing surface among the plasma facing components (PFCs) [5–7]. However, explosion welding requires special geometry and equipment and may not be suitable for covering large areas. Although the VPS-W and APS-W coatings are advantageous for large-area applications, injected hydrogen isotopes on those surfaces during plasma operations, are difficult to desorb even at a high temperature range beyond 800 K [8–10]. This behavior is attributed to an intrinsic porous microstructure of plasma-sprayed W coatings, which acts as an effective trapping site for hydrogen isotopes.
The Kyushu (Q-shu) University Experiment with a Steady-State Spherical Tokamak (QUEST) is a medium-sized spherical tokamak designed for steady-state operation and is equipped with an all-metal plasma facing surface (PFS). The QUEST vacuum vessel is made of stainless steel (SUS), and almost all of the SUS area is covered with APS-W. One of the key research objectives of QUEST is to reduce the amount of retained hydrogen on the surface, i.e., to as close to a hydrogen recycling rate (R) ≈ 1 from the APS-W surface, by maintaining the APS-W coated first wall at high temperature with an embedded heater up to ~500°C [11]. If a recycling rate (R) ≈ 1 is realized, the possibility for achieving a steady-state discharge will be greatly increased with better particle control. Under the above background, although a 6-hour discharge accompanied with a wall temperature of 473 K was achieved in the QUEST long-pulse experiment in 2020, the plasma density became uncontrollable at the latter phase of the discharge because trapped hydrogen particles from the wall were continuously desorbed [12, 13]. One of the possible explanations for such an undesirable desorption is that a certain amount of hydrogen particles would have been trapped in the APS-W coated surface during the early stage of the discharge, and were continuously desorbed until the end of the 6-hour discharge. To suppress such an uncontrollable desorption, covering the first-wall surface with a very thin W plate, i.e., a W-sheet, would be effective in achieving a recycling rate R ≈ 1 at an early stage of discharges. This is because W-sheets are known to have intrinsically low pore density owing to their heavily cold-worked microstructure compared with plasma-sprayed W (e.g., APS-W) [9, 10]. Owing to the low porosity structure, almost all trapped hydrogen isotopes in W-sheets, including polycrystalline W, are desorbed below 500~600 K [9, 10]. Based on these findings, we successfully developed a small-scale sample (brazed area: 40 × 40 mm2) of a new type of first-wall component consisting of a W-sheet (0.254 mm thick)/stainless steel (SUS) with an oxide dispersion strengthened copper (ODS-Cu) intermediate layer via Advanced Multi-Step Brazing (AMSB) in the previous work [14]. The principle of AMSB is a repetitive application of the advanced brazing technique (ABT), which was initially developed to braze W to ODS-Cu (GlidCop®) with the Ni-11%P filler material. Subsequently, we confirmed that the ABT can produce a mechanically robust joint between GlidCop® and SUS (GlidCop®/SUS) joint. The fabricated small-scale sample of W/GlidCop®/SUS with 0.254 mm thickness W-sheet, exhibited very fine joints without any macro-cracks and -voids for both the W/GlidCop® and GlidCop®/SUS interfaces, and the W surface. In addition, almost all of the injected hydrogen isotopes (Deuterium particles) were desorbed below 550 K. If this W/GlidCop®/SUS structure is applied to an entire QUEST first-wall and operated at temperatures above 550 K, the probability of achieving R ≈ 1 in the early stages of a long pulse discharge would be greatly enhanced.
As an extension of the previous work [14], in this study, therefore, a medium-scale sample of the W-sheet/SUS first-wall component with a GlidCop® intermediate layer; (W/GlidCop®/SUS) was developed using ABT for a high-temperature first-wall component in QUEST. After fabrication, a heating test up to ~400°C was conducted in a test chamber under vacuum conditions. The component was then installed at the first-wall position of QUEST, and was exposed to the QUEST plasma for two weeks experiment. This paper presents an initial report of fabrication test of the medium-scale W/GlidCop®/SUS sample and the results of its inspection for application in the plasma confinement device (QUEST).
Figure 1(a) shows the interior of the QUEST vacuum vessel [15], in which most of the first-wall panels (SUS liner) are covered with APS-W. The positions corresponding to the installation of the medium-scale W/GlidCop®/SUS sample, equivalent to 1/6 of the area of a single liner, and the SUS liner are highlighted in red. Figure 1(b) shows an enlarged CAD image of the medium-scale W/GlidCop®/SUS sample occupying 1/6 area of a single SUS liner. The covered area of the medium-scale sample is approximately 190 × 150 mm2.
Figure 2 shows the schematic views and fabrication procedures of the medium-scale sample of the W/GlidCop®/SUS first-wall component. Ni-11%P filler materials with a thickness of 38 μm were inserted at both the W-sheet/GlidCop® and GlidCop®/SUS interfaces, respectively. To ensure high joint quality, a compressive load was applied by placing a 20 kg weight on the W-sheet during the heat treatment process. The heat treatment program is also shown in this figure. In this study, a single-step brazing process was employed. All materials used for joining were prepared slightly larger than their final dimensions prior to brazing. After heat treatment, several machining processes were carried out to adjust the component to its final dimensions, as shown in Fig. 2(b). Figure 2(c) shows a photograph of the W/GlidCop®/SUS first-wall component after completion of all fabrication steps. Slight exfoliation was observed at two locations on the W-sheet surface, as indicated by the yellow arrows. Such delamination was not observed in the previous work with a smaller joint area (40 × 40 mm2) [14].
Although cross-sectional observations with scanning electron microscope (SEM) were not performed on the present medium-scale sample to avoid destructive damage prior to its installation in QUEST, the microscopic integrity of joints fabricated under the almost identical ABT conditions has been thoroughly characterized in our previous work [14]. According to the SEM and energy dispersive X-ray spectroscopy (EDS) analyses in Ref. [14], the Ni-11%P filler material diffuses into the base metals (especially GlidCop®), producing uniform and robust interfaces free of micro-voids or cracks. The uniform thermal distribution subsequently observed (as discussed in Sec. 3.2) provides indirect but clear evidence that such high-quality bonding was successfully achieved across the entire medium-scale area.
2.2. Procedures for performance verification: From preliminary tests to QUEST installation and plasma exposureHeating tests were conducted in a vacuum test chamber with a base pressure of approximately 2.8 × 10−5 Pa to evaluate the thermal response and outgassing characteristics of the medium-scale W/GlidCop®/SUS first-wall component (sample). The heating test was performed once up to a target temperature of 400°C. Furthermore, a separate reference heating test was performed using a blank SUS plate to provide a baseline for the quadrupole mass spectrometer (QMS) measurements, as discussed in Sec. 3.2.
Prior to the actual installation in QUEST, an ‘installation test’ was conducted on a real-scale dummy first-wall substrate under atmospheric conditions. Subsequently, the sample was installed at the QUEST first-wall position (LC B-10). The sample was then subjected to QUEST plasma discharges from April 2 to 18, 2024, which included 20,743 s of discharge-cleaning plasma and 4,016 s of tokamak discharges. During this campaign, the sample was maintained at temperatures up to approximately 350°C using a resistive heater.
After the plasma exposure campaign, the sample was extracted from the vacuum vessel, and its surface morphology and elemental composition were characterized using a laser microscope (KEYENCE VK-X1100) and laser-induced breakdown spectroscopy (LIBS). The LIBS used in this study was the “KEYENCE Laser-based Elemental Analyzer (EA-300 Series)”, which is capable of performing compositional analysis under atmospheric conditions. Furthermore, because the LIBS system is integrated with a digital microscope (KEYENCE VHX-8000), it enables precise micro-level positional analysis. The analysis was conducted with a laser wavelength of 355 nm, a spot diameter of ~10 μm, and a single laser pulse per measurement. The detection depth limit per measurement was estimated to be approximately 0.7 μm, supported by our separate focused ion beam (FIB) assisted cross-sectional observations of a reference W sample irradiated under the same conditions. Note that some laser parameters, such as pulse width and fluence, are proprietary to the manufacturer and were not disclosed.
As mentioned in Sec. 2, the slight exfoliation of the W-sheet surface observed in the medium-scale sample, which was notably absent in the smaller 40 × 40 mm2 samples from our previous work, provides critical insights into the scaling effects of the ABT. The primary mechanism for this delamination is attributed to the accumulation of residual thermal stresses during the cooling phase of the brazing process. Tungsten and GlidCop® exhibit a significant mismatch in their coefficients of thermal expansion (CTE); W has a CTE of approximately 4.3 × 10−6/K, whereas that of GlidCop® is much higher, around 16.6 × 10−6/K. As the component cools from the brazing temperature (approximately 960°C), the GlidCop® layer undergoes much greater thermal contraction than the W-sheet. This mismatch generates substantial shear stresses around the joint interfaces including bulks. In larger bonding areas, such as the 190 × 150 mm2 area used in this study, the integrated stress across the interface increases significantly. When these accumulated stresses exceed the local bonding strength provided by the Ni-11%P filler, delamination occurs, typically at the edges or locations of stress concentration. These findings suggest that while the current ABT is robust for small-scale applications, further optimization such as stress-relief patterning of the W-sheet or the introduction of more compliant interlayers will be essential for scaling up to even larger first-wall components in future fusion devices.
3.2. Thermal integrity and outgassing evaluation in a vacuum test chamberPrior to installing the W/GlidCop®/SUS sample in the QUEST tokamak, entire heating tests were conducted in a vacuum test chamber to evaluate the uniformity of the surface temperature distribution and to analyze the composition of the gases desorbed during heating. Figure 3 shows a top-view photograph of the W/GlidCop®/SUS sample and the positions of thermocouples C2, C3, E2, E3, E5 and E6 embedded in the SUS substrate. The embedded depths and orientations are indicated next to each thermocouple label.
The sample temperature was controlled using a resistive heater placed directly beneath the SUS substrate; this heater-sample configuration was identical to that used in the actual QUEST first-wall. Gases desorbed during heating were analyzed in real time using a QMS installed in the test chamber.
Figure 4(a) shows the temperature profiles measured by the thermocouples (C2, C3, E2, E3, E5, E6) embedded in the SUS substrate of the W/GlidCop®/SUS sample. Thermocouple A1, located near the sample, was used as a reference. The maximum target temperature was set to 400°C, and all thermocouples except E2 reached values close to the target temperature of approximately 400°C. Although the reason why thermocouple E2 saturated at around 280°C is still under investigation, a homogeneous temperature distribution was confirmed by infrared camera imaging (see Fig. 4(c)). Therefore, the relatively low temperature indicated by E2 is considered to have been caused by poor thermal contact between the thermocouple and the SUS substrate. After the overall sample temperature reached approximately 400°C, coolant was sequentially circulated through the cooling water and cooling air pipes installed at the bottom of the heater to rapidly cool down the sample towards room temperature. No signs of delamination or other defects were observed at the joint interfaces of the W/GlidCop®/SUS sample, even after the rapid cooling process.
In-situ QMS spectra of desorbed gases during the heating and cooling phases are shown in Fig. 4(b). All the detected gas components originated from the atmosphere, and no unexpected components that could cause issues in the use of the W/GlidCop®/SUS sample as a plasma-facing component were detected. For reference, an additional heating test up to 400°C was conducted using a blank SUS plate without the W/GlidCop®/SUS sample. The results showed that the trends in desorbed gas components from the blank SUS plate were very similar to those from the W/GlidCop®/SUS sample.
3.3. Installation test on a real-scale dummy first-wall substrate for mechanical interference checkTo ensure a smooth installation in the QUEST tokamak, an ‘installation test’ was performed on a real-scale dummy first-wall substrate of QUEST under atmospheric conditions. The primary objective of this test was to inspect potential mechanical interferences resulting from manufacturing tolerances between the medium-scale W/GlidCop®/SUS sample, the specially shaped SUS liner, and the Molybdenum (Mo) alloy bolts (Fig. 5). Figure 5 shows a photograph of the installation test. A specially shaped SUS liner covering the remaining 5/6 area was also installed with the medium-scale W/GlidCop®/SUS sample. Mo alloy bolts with a very flat head shape were used to mount the SUS liner and the sample. Such low-profile bolts are advantageous for material studies of plasma-facing components because they do not form undesirable shadows in plasma-wall interactions. This installation test using the dummy first-wall substrate confirmed that no interference issues were observed among the SUS liner, the medium-scale W/GlidCop®/SUS sample, and the Mo alloy bolts.
Based on the results of the aforementioned installation test, the medium-scale W/GlidCop®/SUS sample was installed at a QUEST first-wall position to perform initial exposure tests during QUEST discharges under high-temperature conditions up to approximately 300°C. Figure 6 shows photographs of the medium-scale W/GlidCop®/SUS sample and the specially shaped SUS liner installed at the lower position (LC B-10) of the QUEST first-wall. The mounting method using Mo alloy bolts was identical to that shown in Fig. 5. A thermocouple for temperature monitoring was installed behind the central part of the SUS liner. From the viewpoint of mounting this medium-scale sample onto the QUEST first-wall substrate, no interference issues were observed.
Plasma exposure experiments in QUEST were performed during the period from April 2 to 18, 2024. The objective of this experiment was to examine whether the medium-scale W/GlidCop®/SUS sample could withstand long-term heating rather than plasma exposure. During plasma discharge, the temperature of the medium-scale W/GlidCop®/SUS sample and the SUS liner were controlled by a resistive heater placed directly beneath the SUS substrate. Owing to the structure of the QUEST first wall temperature monitoring system, temperature measurement was available at only a single location behind the center of the SUS liner. However, it is considered that no significant temperature difference existed between the SUS liner position and that of the medium-scale W/GlidCop®/SUS sample.
Figure 7 shows the temperature history measured by the thermocouple located behind the central part of the SUS liner shown in Fig. 6. The temperature was gradually increased from April 2 to 18, reaching a maximum value of approximately 350°C. Figure 7(b) shows a magnified view of the selected period highlighted in Fig. 7(a) (hatched region), revealing noticeable temperature fluctuations. These fluctuations consisted of small variations associated with plasma discharges and larger variations caused by local on/off cycling of the heater installed in the panel supporting the medium-scale W/GlidCop®/SUS sample. The plasma discharge history during this period comprised a total of 20,743 s of discharge-cleaning plasma and 4,016 s of tokamak discharges.
After exposure to the QUEST plasma during the period from April 2 to 18, the medium-scale W/GlidCop®/SUS sample was extracted from the first-wall position. Figure 8(a) shows a photograph of the medium-scale W/GlidCop®/SUS sample after QUEST plasma exposure. Compared with the photographs taken after fabrication (Fig. 2(c)) and before installation (Fig. 3), no significant damage, such as additional delamination at the joint or macro-scale cracking, was observed even after the prolonged heating and plasma exposure. To confirm the surface morphology after long-term heating and plasma exposure, surface observation and elemental analysis were performed using a laser microscope and LIBS, respectively. To analyze an entire surface, the coordinate system shown in Fig. 8(b) was used to define the analysis positions.
At all coordinate positions on the W surface, surface morphologies, including sputtering erosion or impurity deposition, could not be observed. The compositional analysis detected almost no elements other than W within the LIBS measurements, with only slight oxygen detected at some locations. A representative laser microscope image is shown in Fig. 8(c).
As mentioned in Sec. 3.1, the objective of this experiment was to examine whether the medium-scale W/GlidCop®/SUS sample could withstand long-term heating rather than exposure to plasma. The irradiation period to QUEST plasma was too short to investigate surface modification due to plasma-wall interaction; therefore, additional irradiation experiments are planned for future long-pulse plasma experiments.
A medium-scale sample of a W-sheet/SUS first-wall component with a GlidCop® intermediate layer; (W/GlidCop®/SUS) was developed using ABT for the high-temperature first-wall component in QUEST at Kyushu University. Accumulated residual stress resulting from the mismatch in the CTE between W and GlidCop®, associated with the relatively large joint area of approximately 100 × 150 mm2, caused small delamination at two locations on the W-sheet surface. To prevent such undesired delamination, improvements in bonding methods will be required in the future.
During heating tests in a test chamber, a uniform temperature increase with a maximum temperature of approximately 400°C was confirmed over almost the entire surface of the medium-scale W/GlidCop®/SUS sample. Furthermore, no additional delamination or damage was observed even after the rapid cooling down phase. No unexpected outgassing was detected by in-situ QMS measurements during both the high-temperature phase (up to approximately 400°C) and the cooling phase.
Following an installation test of the medium-scale W/GlidCop®/SUS sample on a real-scale dummy first-wall substrate of QUEST to inspect potential interferences due to manufacturing tolerance, the sample was installed at a first-wall position (LC B-10) in QUEST. It was then heated from April 2 to 18, reaching a maximum temperature of approximately 350°C, during which the sample was simultaneously exposed to a QUEST plasma discharge. After the QUEST plasma campaign, the sample was extracted from the first-wall. After prolonged heating and plasma exposure, no significant damage, such as delamination at the joint, was observed. In addition, surface analysis revealed no surface morphologies indicative of impurity deposition caused by plasma-wall interactions.
The objective of this experiment was to inspect whether the medium-scale W/GlidCop®/SUS sample could withstand long-term heating rather than exposure to plasma. From this viewpoint, the medium-scale W/GlidCop®/SUS sample demonstrated sufficient tolerance to prolonged heating. Additional irradiation experiments are planned in future long-pulse plasma experiments.
This study was financially supported by KAKENHI (20H01887) from the Japan Ministry of Education, Culture, Sports, Science and Technology. This work was performed with the support and under the auspices of the NIFS Collaboration Research Program (NIFS20KUTR148). The authors would like to thank Metal Technology Co. Ltd. for the use of the furnace for brazing heat treatment. The authors would like to thank Ms. Eri Morishita for performing surface analyses using microscopes.
