Preliminary development of a conceptual first wall for DEMO

The conceptual first wall has a typical sandwich-like structure, as shown in figure 1, which consists of the PFM CVD-W, a bulk substrate of CLF-1 and an interlayer between them.

Zoom In Zoom Out Reset image size

As the PFM of the first wall for DEMO, it is necessary to have a high melting point, high thermal conductivity, low sputtering yield and low fuel retention, as well as good performance under high steady state and transient heat loads. Many investigations have proved that W is considered to be the only candidate for PFMs so far [7, 8]. Due to the low sputtering yield of W and the intermediate heat load lower than 5 MW m⁻² on the first wall by reference to the blankets of ITER, CFETR and others [9–12], it is fitting to apply a thick layer of W onto the first wall. Thick CVD-W with a high purity of 99.9999 wt% and a density of 19.2 g cm⁻³ has been manufactured by SWIP and the Xiamen Tungsten Company, and presents a high thermal conductivity and good thermal shock resistance [2, 3]. It can be deposited up to 2 mm thick at a deposition rate of 0.5–1 mm h⁻¹. Based on the fluence of energetic particles and heat loads expected on the first wall, a CVD-W layer of 1–2 mm is also what is required, while alleviating the dead load and electromagnetic load of the first wall. In this work, samples with a CVD-W layer of 1 mm in thickness are fabricated.

Many kinds of RAFM steel have been proposed and investigated as the structural material for fusion devices. Alloying additions in RAFM are controlled to bolster structural properties while minimizing activation by reducing the Co, Nb, Ni and Mo contents. CLF-1 (Cr 9.0, C 0.11, Mn 0.6, W 1.6, V 0.3, Ta 0.1, and Fe) has been developed by SWIP and has similar mechanical properties to other RAFMs, such as EUROFER97 in the European Union (EU) and F82 H in Japan [4]. For the purpose of validating the design only, the thickness of the bulk CLF-1 in the conceptual first wall is 5 mm.

Since there is a big difference between the thermal expansion coefficients of CVD-W and CLF-1, which are 4–5 $\times$ 10⁻⁶/°C for CVD-W and 10.9–13.2 $\times$ 10⁻⁶/°C for CLF-1 in the temperature range of 200 °C–1000 °C [2], an interlayer is required to mitigate the stress between them. To achieve good adhesion without any delamination or cracks, it is necessary to use a material with an intermediate thermal expansion coefficient in comparison to those of CVD-W and CLF-1.

As the first wall of DEMO, it will work within the temperature range of 300 °C–500 °C [1, 9, 11–14]. The PFM also experiences a high fluence of particles and transient heat loads due to transient events during the operation of a fusion reactor. Therefore, the interlayer must be stable at the working temperature and have good performance under a high temperature gradient and thermal cycling.

Another concern for the interlayer is the capability of prevention of tritium permeation due to the great importance of controlling the tritium buildup in the first wall, improving the fuel efficiency and conforming to the safety regulations of tritium. The requirements of the tritium barrier are quite different from those for the tritium permeation barrier inside the blanket module or the tritium plant. The interlayer here has to suffer high neutron radiation in the order of tens of dpa [1] and the implantation of energetic deuterium and tritium, which could cause excessive retention and permeation due to the cracks and defects induced by neutron radiation and thermal cycles.

Overall, the interlayer is required to have good bonding properties and tritium prevention, which is crucial for assembling the first wall, controlling the inventory buildup and maximizing the fuel efficiency. The candidate interlayer must satisfy: (1) a suitable thermal expansion coefficient which matches the expansion coefficients of CVD-W and CLF-1; (2) good bonding properties for both W and CLF-1; (3) good performance under the working temperature and thermal cycles; (4) excellent properties for prevention of the tritium permeation; (5) acceptable thermal conductivity; (6) good properties after neutron irradiation.

The interlayer is designed in terms of the requirements described above. Since investigations on tritium barriers have been performed for years and many results can be found easily [5, 15–18], it will be efficient if we select those materials with suitable thermal expansion coefficients from the tritium barriers which have been validated by previous works.

A lot of materials, including metals, oxide, nitride, carbide and so on, can be used as a tritium barrier on a steel bulk. Usually, the interlayer consisting of Ti or Ni has good adhesion strength with the W and steel. Ti is not good as an interlayer since it will react with WF₆ during the following fabrication of W by CVD. Ni is not a good tritium barrier due to its high activation after neutron irradiation.

Ceramics are typically considered to be good tritium barriers due to their hydrogen isotope permeabilities, which are typically much lower than those in metals. Oxides are not chosen here due to the activation after neutron radiation and the bad compatibility with plasmas. Among different carbides, titanium carbide has been tested as a tritium barrier [15, 19, 20]. But it is bad here because of the adhesion problems with direct deposition on a steel [5, 15, 19, 20]. Silicon carbide is a promising coating layer under neutron radiation for fusion reactors [21] due to its high radiation resistance, high thermal conductivity and excellent mechanical properties, as well as stability at high temperatures [16, 17, 22, 23]. Although the hydrogenic diffusivity in SiC is many orders of magnitude lower than W [5, 24–26], the thermal expansion coefficient of SiC is 3.4–5 $\times$ 10⁻⁶/°C within the temperature range of 200 °C–1000 °C, which is lower than those of both CVD-W and CLF-1. Therefore, it is not suitable for mitigating the stress between CVD-W and CLF-1.

For nitrides, TiN has been studied widely as a barrier due to its good adhesion and the ease of deposition [18]. Its thermal expansion coefficient is 9.35 $\times$ 10⁻⁶/°C, which pairs with the coefficients of CVD-W and CLF-1 very well. Additionally the TiN coatings on the 316 l steel and F82 H have been shown to give good adhesion [15, 27, 28]. As a result, TiN is selected as the material of the interlayer, as shown in figure 1.

The thickness of the interlayer has a strong effect on the residual stress inside the coating layer. The higher the thickness, the higher the stress. To mitigate the residual stress and thermal stress of the interlayer under working conditions, the thickness of the interlayer was chosen to be 1 µm.

With this test, the samples were periodically heated to check the adhesion of the interlayers. Tests were carried out in a box-type high-temperature sintering furnace (type KSL-1200x) and the samples were sealed into a quartz tube with a pressure of 8 $\times$ 10⁻³ Pa of the air gas. The thermal cycling test was only performed for the three good samples: samples 6, 7 and 8. Since the wall for DEMO will work in the temperature range of 300 °C–500 °C, the thermal cycling tests were carried out at 700 °C, which is higher than the maximum working temperature of 500 °C. During the test, the samples were heated up to 700 °C at the rate of 10 °C min⁻¹ and then held at that temperature for 12 min. After that, the tube was taken out of the furnace and dipped into water to cool down to room temperature. The entire process above was repeated eleven times. After the thermal cycling tests, the samples were inspected and all three test samples were intact. The results show that the TiN interlayers of these three samples have sufficient adhesion as an interlayer between W and CLF-1 so far.

Samples 7 and 8 were tested with hydrogen plasma exposure in the Magnum-PSI machine [29]. The Magnum-PSI is a linear plasma generator with the capabilities of a particle flux of several 10²⁴ m⁻² s⁻¹ and steady-state heat flux of greater than 10 MW m⁻². A transient plasma load can also be superimposed on the steady-state plasma by a pulsed plasma, which is given by discharging a set of capacitor banks. The pulsed plasma is applied to simulate the transient heat flux during ELMs in H-mode plasmas [30]. The profiles of electron density n_e and electron temperature T_e were monitored by a Thomson scattering system about 30 mm in front of the target, while the plasma parameters of the pulsed plasma were not diagnosed during experiments. The surface temperatures of the tested samples were measured by a spectropyrometer (FAR SpectroPyrometer model FMPI) in the wavelength range of 900–1700 nm and a fast-framing infrared camera (FLIR SC7500MB) in the wavelength range of 3.97–4.01 μm at a frame rate of 5 kHz. During the plasma exposure tests, the test samples were clamped on a sample holder, which was actively cooled by flowing water at a room temperature of 25 °C.

The experimental conditions of samples 7 and 8 can be seen in figure 2 and table 2. With the conditions of case 1, sample 7 was firstly exposed into a hydrogen plasma with the particle flux of $2\, \times$ 10²⁴ m⁻² s⁻¹ and the heat flux of 6.8 MW m⁻² determined with methods in [31], which is phase 1 of sample 7 with the electron density peak of $2\, \times$ 10²⁰ m⁻³ and the electron temperature peak of 1.4 eV, as shown in figure 2. After a 15 s exposure, the sample surface temperature increased up to 1350 °C and it is too high for the first wall sample. Hence, the plasma parameters had to be lowered to the electron density peak of $0.78\, \times$ 10²⁰ m⁻³ and the electron temperature peak of 0.83 eV with the particle flux of $6\, \times$ 10²³ m⁻² s⁻¹ for 200 s, which is called phase 2 in figure 2. The plasma parameters of phase 2 for the exposure of sample 7 were selected by controlling the surface temperature. The reason for this is that the temperature of the TiN interlayer is critical for evaluation of its bonding properties and the feasibility of the sandwich-like structure. Although the plasma parameters have important effects on the W surface, the investigations on the behavior of the CVD-W are not within the scope of this research, and have been previously evaluated [2, 3]. During phase 2, the surface temperature of the sample was kept at 650 °C. It is easy to calculate that the steady-state temperature of the TiN interlayer was 622 °C with the W thermal conductivity of 125 W/m/K, the TiN conductivity of 29.3 W/m/K and the CLF-1 conductivity of 29 W/m/K at around 500 °C by assuming the room temperature of the sample holder due to its active cooling. Since the working temperature of the first wall of DEMO is in the range of 300 °C–500 °C [1, 9, 11–14], it is enough to validate the function of the interlayer under the conditions of phase 2 in case 1.

Zoom In Zoom Out Reset image size

For case 2, sample 8 was exposed into a plasma with the electron density peak of $0.93\, \times$ 10²⁰ m⁻³ and the electron temperature peak of 0.64 eV for 200 s, as shown in figure 2 and table 2. Moreover, a transient heat load of 0.16 GW m⁻² given by a pulsed plasma was superimposed onto the sample simultaneously, which was used to evaluate the effect of heat fatigue on the interlayer due to ELM-like events. The pulse was in a triangle shape with the rise time of 0.65 ms and the decay time of 0.6 ms at the frequency of 5 Hz. A total of 400 pulses were applied to this sample during this experiment. Under the conditions of case 2, the surface temperature was kept at 519 °C, while the steady-state temperature of the TiN interlayer was 497 °C, which is at the upper limit of the possible working temperature of the first wall for DEMO.

The sandwich-like structure can be seen easily by SEM. Figure 3 shows the microstructure of the cross section of sample 5, which is typical for all of the samples. The sandwich-like structure, including CLF-1, CVD-W and the TiN interlayer, is shown clearly in figures 3(a), (b) and (c), where figures 3(b) and 3(c) give the details of the area B in figure 3(a) and the area C in figure 3(b), respectively. The thickness of the TiN interlayer is 0.96 µm in figure 3(c), while the distributions of different elements, including Ti and N, can be seen in figure 3(d), which indicates the apparent TiN interlayer between Fe and W. To deposit TiN well onto the CLF-1 substrate with good adhesion, the CLF-1 surface must have a certain roughness level. Therefore, a roughness level of about 1 µm is applied to the CLF-1 surface by polishing before the TiN coating fabrication, as shown in figure 3(c).

Zoom In Zoom Out Reset image size

As shown in table 1, samples 1–5 delaminated during the wire-electrode cutting for preparation of the samples for testing and material analysis. By checking the microstructures of these samples by SEM and mapping the element distributions by EDX, the results show that all of these five samples were broken near the CLF-1 side. It is found that the fracture surface of the sandwich-like structure is always located in the CLF-1 material close to the interlayer or partially in the TiN layer near the CLF-1 substrate. A typical result of material analysis by EDX can be seen in figure 4, which shows the mapping of different elements on the fracture surface of the W side of sample 3. In figure 4(a), the species and percentages in area for different elements are given in the legend. Table 3 shows the weight and atomic percentages of different elements on this surface. Oxygen appears since the sample was analyzed days after being broken, and as a result, some particles on the surface were oxidized. From figure 3(a) and table 3, there is a lot of Fe, Ti and N on the surface but a lot less W, although this surface is on the W side of sample 3, which clearly implies that sample 3 was broken near the CLF-1 side. An unexpected phenomenon is that a large amount of fluorine (F) appears on the fracture surface. And F is always distributed in the same zone with Fe, different from the distributions of Ti, N and W. This fact can also be seen obviously from the mapping of different elements on the cross section of sample 6 in figure 7 later in this section. The F is introduced only by WF₆ during the CVD of W, which indicates that the F penetrates into CLF-1 during the fabrication of CVD-W. Since TiN is a good barrier for tritium and other hydrogenic particles, as described in section 2.2, it would also be difficult for F to pass through TiN and then penetrate into the CLF-1 substrate. A reasonable explanation is that there are some defects in the TiN interlayer, and thus F penetrates into the substrate. The defect of the interlayer and the F penetration into CLF-1 cause the sample to break easily. As a comparison, the mapping of element distributions on the fracture surface of sample 7 is also checked. Sample 7 was not broken during cutting, the thermal cycling test and the plasma exposure test, but it was forced apart for material analysis after the exposure test. The percentages of different elements on the fracture surface of the CLF-1 side of sample 7 are presented in table 4. The results show that there is no F at all on the fracture surface. Since a large amount of Fe exists on this surface and F is usually distributed in the same zone as Fe, the data in table 4 provide evidence that no F penetrates and diffuses in the CLF-1 substrate of sample 7. Moreover, the material analysis is performed immediately after the sample is broken; as a result, no oxygen can be found on the fracture surface.

Zoom In Zoom Out Reset image size

The mechanics of the broken sample and the defects of the TiN interlayer can be further analyzed by material analysis of sample 5. Figure 5 shows photographs of the fracture surface of the W side of sample 5. The TiN interlayer is displayed distinctly, and it covers the CVD-W in general. By mapping the element distributions with EDX, it is easy to identify the bright zones as the W material, as shown in figure 5. It implies that there are a lot of defects in the TiN interlayer where CVD-W was deposited and F penetrated during the CVD of W.

Zoom In Zoom Out Reset image size

Sample 6 was kept intact during cutting and the thermal cycling test and then broken accidentally before material analysis, which is a little different from samples 1–5. It seems that sample 6 has better bonding properties than samples 1–5. When checking the element distributions on the fracture surface of the W side of sample 6 by mapping with EDX, it is hard to find F within a random small zone, as shown in figure 6(a). But F can be identified when a wider area in the scale of a few square millimeters is checked. It implies that there are some defects in the TiN layer and the number of defects may be considerably fewer than those in samples 1–5. Then, the W side of sample 6 was cut in the region containing F, as shown in figure 6(b), to check the cross section of this specimen and understand the F behavior during the penetration and diffusion. The microstructures and the element distributions of this cross section are presented in figure 7. The CLF-1, the TiN interlayer and CVD-W can be seen clearly in figures 7(a), (b), (c) and (d). The results also show that F is distributed in the same region as Fe, which is consistent with the mapping photographs of the fracture surfaces of other samples. Moreover, the defect of the TiN layer is recognized in the blue circle in figure 7(g), and the patterns of the TiN interlayer and CVD-W in figures 7(g) and (h) are consistent with the results in figure 5. On the other hand, the F penetrates easily and diffuses uniformly into the CLF-1 substrate, which is evidence of the importance of a tritium barrier for the first wall of DEMO. Additionally, it is also verified that the sample was broken in the CLF-1 substrate close to the TiN interlayer according to the element distributions of Fe, Ti, N and W in figure 6 and the microstructures of the cross section in figure 7.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

After the thermal cycling test at the temperature of up to 700 °C, samples 7 and 8 were exposed to hydrogen plasmas under the conditions in figure 2 and table 2. The surface temperature of sample 7 reached 1350 °C in phase 1 and then was kept at 650 °C in phase 2. During phase 2, the temperature at the TiN interlayer was 622 °C, which was a little bit higher than the upper limit of the working temperature of 300 °C–500 °C for the first wall of DEMO [1, 9, 11–14]. The temperature of the interlayer is the critical parameter for this plasma exposure test since it is important for evaluating the bonding property and the feasibility of the sandwich-like structure, while the behaviorof CVD-W under plasmas and heat loads has been investigated previously [2, 3]. For sample 8, it was exposed to a steady-state plasma as well as a pulsed plasma, which was used to simulate the heat fatigue due to ELMs. The temperature of the interlayer of sample 8 was set at 497 °C, right at the upper limit of the working temperature of the first wall. After the plasma exposure tests, the structural integrity of samples 7 and 8 still appeared good from visual inspection. The cross sections of these two samples were analyzed with SEM and are shown in figure 8. There is no obvious change seen at the interlayer, and no detachment or crack can be found in the structures of the samples, which implies that the bonding property of the interlayer is good enough under the working conditions for the first wall of DEMO. These encouraging preliminary test results support further examination of TiN as a potential interlayer material for W on RAFM steel as a first-wall material. For instance, improvement of the interlayer quality and the tritium permeation test may be required in the future.

Zoom In Zoom Out Reset image size

After plasma exposure, the top W surface of sample 7 shows clear surface evolution. The surface roughness changes since the surface temperature is higher than the recrystallization temperature (RCT) of W due to overheating in phase 1. As a result, the grain growth can be seen clearly in figure 9(a), in which the red dashed line marks the boundary of the region with overheating. There are some particles deposited on the region without overheating, which is shown in figure 9(b). By mapping with EDX, it can be identified that the deposited particles are Fe due to the physical sputtering, and the details of the deposition are given in figure 9(c). For sample 8, there is no obvious morphological change except for some localized grain growth due to the overheating with the transient heat load of 0.16GW m⁻² by pulsed plasmas. But there is no deposition of Fe at all since the ion temperature equal to the electron temperature in case 2 is much lower than that in phase 1 of case 1. More discussion on the CVD-W surface evolution is out of the scope of this work.

Zoom In Zoom Out Reset image size