Fuel retention and erosion-deposition on inner wall cladding tiles in JET-ILW

Table 2 presents the Be and D contents measured on a number of samples sectioned from the unexposed and two plasma-exposed tiles. Figures 2(a) and (b) show the toroidal profiles of Be and D, respectively, measured on the whole (not cut) exposed tiles. The Be layer thickness is given for an assumed material density of 1.85 g/ cm ³. However, with the increasing content of co-deposited elements the material density increases and the actual layer thickness after exposure could be different than listed in table 2 and figure 2(a). We emphasize that data from the tile central area, i.e. the funnel-shaped dip (toroidal position 82–118 mm) near the bolting hole, will not be discussed in the following. Any attempt of interpretation may be strongly misleading because of possible non-uniformity of the original coating in that region and analytical difficulties related to the change of beam-target geometry when measurements are done on inclined surfaces.

Table 2. Summarized NRA results for the unexposed and exposed tiles measured with a 4.5 MeV ³He²⁺ beam. The Be layer thickness is given for an assumed material density of $1.85\displaystyle \frac{g}{{{\rm{cm}}}^{3}}.$

Tile	Region	Be [${10}^{18}\displaystyle \frac{at.}{{{\rm{cm}}}^{2}}$]	Be [$\mu m$]	D [${10}^{15}\displaystyle \frac{at.}{{{\rm{cm}}}^{2}}$]
Unexposed	average	110–115	9.0	—
106	A	94–96	7.6–7.8	210–335
	B	98–120	7.9–9.7	120–150
	average	102	8.4	199
412	A	125–130	10.1–10.5	505–605
	B	136–149	11.0–12.0	303–452
	average	134	10.8	481

Zoom In Zoom Out Reset image size

The analyses done for three samples located a few cm apart on the not exposed tile have revealed high uniformity of the initial Be layer thickness (9.0 μm). The Be thickness on the two exposed plates (data in table 2) is different: around 8.5 μm and 11.0 μm on samples from Tiles 106 and 412, respectively. On a given tile, the thickness varies approximately within ±1.3 μm (10%–16%). The same range of quantitative variations on those two plates are seen on the entire toroidal scans, figure 2(a), while much more pronounced differences are on Tile 403 from 7 μm near the center of the IWC column to 12 μm in the limiter shadow. It is difficult to state whether the increased Be content in the limiter shadow on 403 is related to the deposition by material eroded from another area of the IWC coating or from the adjacent or even from more distant overheated limiters, i.e. upper dump plates (UPD). The first two options are equally probable, while the latter hypothesis is least plausible, because the strongest erosion of UPD occurred at the end of ILW-2 [19]. Therefore, one would have expected a similar or even stronger effect on Tile 106 retrieved after ILW-2.

In summary, all observed variations in the Be layer thickness clearly indicate local material erosion and deposition. The results also provide crucial information that in neither case the coating thickness has dropped below 7 μm, i.e. below the value specified as the minimum initial thickness. However, the exact erosion-deposition balance of Be cannot be determined because the layer thickness on respective tiles was not measured before their installation in JET. At that time, there were technical limitations for the pre-characterization of those large and heavy tiles. The research and development (R&D) process performed on several witness samples (e.g. 2×12 cm Inconel stripes) located next to the coated tiles had shown high uniformity of the initial Be layers. This fact has been confirmed recently when analyzing samples cut from the initial tile. Secondly, the decisions on the retrieval of IWC plates became viable once spare tiles could have been manufactured. For that reason, it is not intended to judge whether the results for actual IWC components are in agreement with the data for wall inserts. These were Inconel 600 buttons (10 mm in diameter) having one half coated with nominally 40 nm thick W film obtained by physical vapor deposition and the other half with 2.5–3.0 μm thick Be layer obtained by TVA. The erosion from these pre-characterized Be marker coatings had been reported at the level of 0.7–0.9 μm (the sum for three campaigns) with erosion rates dependent on the insert location [11]. Taking into account the variations in the Be thickness (±1.3 μm) on the exposed IWC, it would be difficult to draw conclusions about erosion at the level below 1 μm. However, It should be added that the production methods of Be layers on IWC [8] and those on wall inserts [12] were not alike thus resulting in significant structural differences.

Data in table 2 and figure 2(b) provide insight into the amount and toroidal distribution of deuterium on the tiles. The most important is that the retention on most of the studied areas does not exceed 1×10¹⁸ D cm⁻². This is highly consistent with D contents measured on deposition areas of IWGL [4]. Such retention level is considered as low, especially when compared to the inventory in JET with carbon wall (JET-C) where over two orders of magnitude greater values were measured [20]. Besides that similarity, some other features are perceived. The overall deposition patterns on Tiles 106 and 412 are rather uniform despite some toroidal variations. The absolute D contents are different: 0.1–0.4×10¹⁸ cm⁻² and 0.3–1.1×10¹⁸ cm⁻² on 106 and 412, respectively. It is predominantly attributed to the ILW-2 finish with around 300 discharges fueled with H₂ thus leading to in situ H-D isotope exchange. Two other aspects, the differences in the total exposure time and the plasma contact time with IWGL blocks adjacent to respective IWC, may be consequently considered less important (data in table 1).

Another point is in the comparison of the D retention data for the whole (not cut), figure 2, and cut tiles, table 2. While the Be thickness results for Tile 106 and 412 agree fairly well in both cases, the deuterium data are different by a factor of around two. The average amount of D on the not cut Tile 106 is 0.32×10¹⁸ cm⁻², which is 50% higher compared to the data in table 2 for the cut samples. In the case of Tile 412 the D content is nearly twice greater on the not cut plate. No comparison is available for Tile 403 with the greatest D content because there were no second round measurements. The most probable reason for the diverging D content may be either the release of D from the layer or the isotope D-H exchange during the long time (around 2 years) separating the measurements. While the temporal decrease of D content in C-based tokamak co-deposits was studied, up to 30% in 4–5 years [21], nothing is known yet about the long-term behavior of fuel in Be-rich co-deposits or coatings from JET-ILW.

The greatest retention is on Tile 403, where the average D amount is 1.97×10¹⁸ cm⁻² and the highest values in the limiter-shadowed region (approximately on 30% of the tile area) reach up to 5.82×10¹⁸ cm⁻², i.e. at the same level as D inventory in the upper part of the inner divertor [5]. When compared to the adjacent IWGL tile, the highest D retention recorded in the deposition zone of the limiter was 1.4×10¹⁸ cm⁻² [4], hence significantly below the amounts measured in the limiter-shadowed area of Tile 403.

Taking into account available data and the ILW operation history, a tentative attempt is made to extrapolate to the D retention in the entire Be-coated area after ILW1–3. Results for Tile 106 are excluded because of the H operation at the end of ILW-2 campaign. In the first approximation data from Regions B (not shadowed areas) on Tiles 403 and 412 are taken into account. With the average D content of 1.0×10¹⁸ atoms cm⁻² the D retention limit is assessed at the level of 5.3×10²² D atoms equal to 176 mg of deuterium in the IWC beryllium coating. This value is very close to the sum of D atoms on IWGL and OPL (4.4×10²²), as extrapolated after ILW1–3 [4] and, it is five times smaller than that determined for the divertor: 25×10²² D atoms [10]. The retention of 176 mg D in the Be coating is considered as a lower limit, because there are areas with several times greater fuel content (5.8×10¹⁸ cm⁻²) than the mean value. For comparison, the maximum D contents in in JET-ILW, up to 12×10¹⁸ cm⁻², have been detected in the inner divertor on the High Field Gap Closure Tile (Tile 0) and the apron of Tile 1.

The ³He backscattered data gathered during measurements of the total D and Be contents have indicated the presence of impurity species: C, O, Ni, W. With ToF-HIERDA the composition of the surface region was determined up to the depth of 1 μm. Particular emphasis was on low-Z species such as hydrogen isotopes, C, N, O. This also included the search for ¹⁵N and ¹⁸O used in JET as tracers. Data for Ni (Inconel) and W are presented, while the content of aluminum (Al) is not shown, because that element was most probably transferred to the coating by operation of the remotely handled equipment used during shutdowns. The average amounts of hydrogen isotopes and impurity atoms in Regions A and B on each tile are collected in table 3. The atomic concentrations are shown for two depths, 1500 and 9000×10¹⁵ atoms cm⁻², to discriminate between the surface/near-surface layer and the bulk of the coating. Detailed depth profiles of several species identified in Region A are displayed in figure 3. Impurities with very small concentrations (<0.1 at.%) are left out of the plot for clarity.

Table 3. Collected ToF-HIERDA results for hydrogen isotopes and plasma impurity species on the initial and exposed tiles. Results are given for 1500 and 9000 at. cm⁻², respectively.

				Atomic concentrations [${10}^{15}\displaystyle \frac{at.}{{{\rm{cm}}}^{2}}$]
Tile	Region	Depth		Be	O	2D	1H	C	N	Ni	W
		[${10}^{15}\,$at.cm−2]	[nm]
unexposed	—	1500	135	1473	66	—	11	16	—	13	—
		9000	785	8669	245	—	46	68	2	30	—
106	A	1500	175	1092	325	10	79	40	27	11	1.1
		9000	945	7414	1120	68	177	156	81	50	3.0
	B	1500	165	1199	249	—	82	24	25	7	1.0
		9000	905	7753	913	44	160	110	63	37	2.6
412	A	1500	185	1019	410	38	14	36	27	39	2.3
		9000	1025	6725	1835	119	50	123	61	116	6.0
	B	1500	165	1230	285	24	7	14	14	10	0.8
		9000	915	7800	1044	88	24	52	32	31	1.7

Zoom In Zoom Out Reset image size

As expected, oxygen is the major alien species, but its content—especially in the surface region—has increased by nearly one order of magnitude after plasma exposure. It informs about: (i) the low oxygen contamination of the initial coating (5 atomic % at the very surface down to 1.5–2 at.% in the bulk) and (ii) efficient gettering of O plasma impurities. In general, the distribution of impurities, both depth and lateral, is not uniform. The results in table 3 prove greater impurity contents in Regions A than in B on the two exposed tiles. The impurity accumulation in Region A on Tile 106 is up to 7 at.% greater than on Region B, while for Tile 412 the difference amounts to nearly 13 at.%. With the information that Region A is adjacent to the protruding limiters, the increased impurity deposition may be explained by the fact that the IWGL tiles partly shadow Region A from plasma impact; the magnetic field direction is marked in figure 1(a). This leads to a lower erosion of the IWC surfaces and, simultaneously, to a higher net impurity deposition.

Data in table 3 and depth profiles in figure 3 show that the impurity content in the surface and near surface layer (1500×10¹⁵ atoms) is relatively higher than if the entire analyzed depth of 9000×10¹⁵ atoms is considered. The amount of impurities clearly decreases with depth. This applies to all detected impurities on the initial and exposed tiles. The long-term diffusion of impurities into the porous coating cannot be excluded, but it should be considered that ToF-HIERDA provides results only for the very surface layer, less than 1 μm. The atomic content of impurity species constitute around 28 at.% (106) and 29 at.% (412) of the surface layer composition. Averaged contents of H, D, Be, C, N, O and Inconel for the depth of 1500×10¹⁵ at.cm⁻² plotted logarithmically for vividness are in figure 4. In bar charts for the exposed tiles, the orange-hatched part corresponds to the impurity content in the initial sample. The total height of respective bars is the total content of the element in the surface. No distinction is done for Be, as it is the main element.

Zoom In Zoom Out Reset image size

The results prove high purity of the initial Be coatings in which the C content barely exceeds 1at.%. On the exposed tiles, the amount of respective plasma impurities (other than oxygen) stays below 2 at.%. Small amounts of W were detected (≤0.14 at.%). The exposed layers retain nitrogen. It is a general feature of all ILW plasma-facing surfaces with co-deposits; once again, a long-term N retention in the torus is demonstrated. The presence of the ¹⁵N tracer puffed from the divertor at the end of ILW-3 has also been identified on Tile 412, but the amount is at the detection limit of 5×10¹⁴ cm⁻².

At last, results for hydrogen and deuterium are discussed. The initial H content was 0.8 at.% and the increase could have been expected due to the D-H isotope exchange in the co-deposited layer when the tiles after ILW campaigns are stored for years in ambient atmosphere. Such expectations are not supported by either result because: (a) the amount of H detected on Tile 412 is equal to that on the initial tile surfaces; (b) relatively high H content on Tile 106 derives from the ILW-2 finish with around 300 discharges with H₂ fueling and - as discussed earlier - the exchange occurred already during the plasma impact. As a result, the surface layer of Tile 106 contains only 1×10¹⁶ D cm⁻² in the limiter-shadowed Region A, while in Region B no species could be identified with the D detection limit at 5×10¹⁴ atoms cm⁻². In summary, the retention results obtained by NRA and ToF-HIERDA are qualitatively fully coherent showing significantly less D on Tile 106 than on 412. The D amounts and depth profiles determined by ToF-HIERDA clearly indicate the presence of fuel at a greater depth than accessible by that method. For obvious reasons, the total retention data are those measured by NRA and discussed in section 3.1.

High-resolution SEM images in figures 5(a)–(c) show respectively the initial appearance of the Be coating on Inconel and then after high-heat flux testing at 1.8 MW m⁻² and 2.55 MW m⁻² during the R&D process [8]. The initial platelets structure is maintained nearly intact under the 1.8 MW m⁻² load, but the structural changes start occurring under the exposure at the load of 2.55 MW m⁻².

Zoom In Zoom Out Reset image size

Surface topography of the initial and exposed Tiles 106 and 412 is shown in figures 6(a)–(c). The clear platelets visible of the unexposed surface, similar to that in figures 5(a)–(b), became blurred after exposure to ILW1–2 and then characteristic features of single Be flakes nearly disappeared in the course of further exposure to plasma. One may assume the changes occurred by co-deposition of different species, oxygen gettering and long-term impact of charge exchange neutrals. Surface melting by excessive heat loads may also be considered, though there is no clear record on disruptions hitting the lower part of the limiter column. Two features may be perceived on the exposed surfaces: (i) the appearance of granular structures; (ii) rounded rims of the flakes (white in secondary electron images) indicating the decrease of electrical conductivity and, by this, suggesting the change of chemical composition, i.e. formation of a semi-conductive or insulating phase. The granular structures, 10 nm to 40 nm in diameter, occupy the whole sample surface, as illustrated by high magnification images in figure 7. These structures cover samples from Tile 106 and Tile 412, whereas the unexposed samples are mostly clean.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Besides the bright-looking rims and granular structures also some winding tube-like objects were observed. They appear in figure 7(a), while a clear evidence is presented in figure 8 (a)–(b) which show both a detaching rim and an isolated tube. EDX spectra recoded on the coating and on the tube are compared in figure 8(c) where K-alpha lines of Be and O are marked. The Be:O concentration ratios calculated with the AZtec software amount to 9.1 and 5.3 for Spectrum 1 and Spectrum 2, respectively. It proves the increased oxygen content in the tube in comparison to the coating surface. It also indicates the formation of Be-O compound, though the exact stoichiometry of the whole tube-like structure cannot be determined from the EDX data. In summary, SEM and EDX document the erosion of Be layers caused by the formation and spalling-off of oxide structures on the edges of the layers.

Zoom In Zoom Out Reset image size