Overview of nitrogen-15 application as a tracer gas for material migration and retention studies in tokamaks

Two experiments were performed in TEXTOR. In both cases nitrogen was injected toroidally symmetrically, while—in the second experiment—other marker gases (WF₆ and ¹³CH₄) were introduced locally through test limiters. The first experiment was performed with a double roof-shaped test limiter composed of a graphite holder and four stripes: two made of tungsten and two made of graphite. It was inserted into the machine from the top and placed 1.5 cm deep into the scrape-off layer plasma. The last-closed flux surface was defined by the position of the main toroidal limiter ALT-II (advanced limiter test) at minor radius 46 cm. The system was exposed to discharges with edge cooling by injecting first ¹⁴N and then ¹⁵N isotopes. The total amount of injected ¹⁵N₂ was 6.4 Pa m³, i.e. 3.4 × 10^{21 15}N atoms, in 18 discharges of a total duration of 112 s. The exposed limiter is shown in figure 1(a). One may distinguish a shiny-looking erosion area on the top and blackish deposit on the lower part of plates.

Zoom In Zoom Out Reset image size

An assembly of the test limiter in the second experiment is shown in figure 1(b): a single roof-shaped block on which a polished plate (both made of graphite) with a hole for WF₆ puffing was attached. The image was taken after the exposure which led to the formation of a co-deposit near the gas inlet. The limiter assembly was placed 2 cm from the last-closed flux surface from the bottom of TEXTOR. The injection of 3.5 × 10^{21 15}N atoms in 14 shots with neutral beam heating was accompanied by puffing of 1.9 × 10²⁰ W atoms and ¹³CH₄ (1.8 × 10^{21 13}C atoms) from the upper test limiter [20]. The experiment was performed on the last operation session before the shut-down and the main plasma parameters were: line averaged plasma density n_e = 3.5 × 10¹⁹ m⁻³, plasma current I_p = 0.35 MA and the magnetic field strength B_t = 2.25 T.

The experiment in AUG was also performed during the last operation session of the 2010–2011 campaign before the major shut-down. In total  ∼ 4.6 × 10²² atoms of ¹⁵N were puffed into hydrogen plasma during 10L-mode discharges. The main parameters were: line averaged plasma density n_e = 5.8 × 10¹⁹ m⁻³, plasma current I_p = 0.8 MA and the magnetic field strength B_t = −2.5 T. The discharges were heated by 1.8 MW neutral beam injection and 0.96 MW electron cyclotron resonance. The puffing was performed from a single location: a port in the outer midplane in sector 9. Due to the inlet construction the gas portion was reaching plasma like a plume. The same port was also used for puffing ¹³CH₄ during the same discharges in equal quantity of ¹³C atoms. The migration of ¹³C is discussed elsewhere [21]. Following the experiment 35 tiles were removed from AUG and then analyzed: 12 from the lower and 8 from the upper divertor, 4 from the inner heat shield; 9 from protection limiters of the ion cyclotron resonance frequency (ICRF) antenna and two limiter plates from the main chamber.

The exposures were followed by comprehensive surface analyses using accelerator-based ion beam methods to determine all species co-deposited on PFC during the injection experiments. Nuclear reaction analysis (NRA), Rutherford backscattering spectroscopy, time-of-flight heavy ion elastic recoil detection (ToF-HIERDA) [22] were done on a large number of tiles retrieved from the tokamaks. Tandem accelerators and particle detection facilities at the Uppsala University (5 MV) and at the Max Planck Institute for Plasma Physics (3 MV) were used.

For ToF-HIERDA a 26 MeV ¹²⁷I⁷⁺ beam was used. The method is very well suited for light isotopes, including nitrogen. The beam impinges on the sample at the angle of 22.5°, whereas the recoils are detected at 45° with respect to the direction of the incoming beam. The ToF tube consists of two very thin carbon foils and an implanted silicon solid-state energy detector. When the recoiled species passes through the foils a signal is triggered and the flight time can be measured. To reduce the noise, the signal from the second foil is used as the start signal for the timing and, the signal from the first detector is delayed and then used as the stop signal [22]. Finally the total remaining energy of the recoil is collected by a solid-state detector. The reversed flight time and the energy gives banana-shaped traces for each recoiled mass from which the depth distribution can be calculated for smooth surfaces [23]. For reliable results smooth surfaces is preferable, therefore, they were restricted to polished graphite or tungsten plates.

The study of the spatial distribution of ¹⁵N in co-deposits from AUG was performed by means of NRA using a 900 keV proton beam and detecting both reaction products of the ¹⁵N(p,γα)¹²C reaction: γ-rays and α-particles. The measurements in the complex mixture of light and heavy species (e.g. tungsten) required proper calibration of the analysis system. α-particles were detected by a silicon solid state detector (150° at relative to the incoming beam) protected by a 17 μm Mylar foil to eliminate the background from protons backscattered from tungsten-coated tiles. The choice of the foil thickness is crucial: thick enough to stop the scattered protons but still thin to allow α-particles from the nuclear reaction to pass through. Another serious issue is to distinguish effects related to nuclear reactions with boron [24], used in AUG for wall conditioning. The presence of this element results in a continuum of alpha particles in the spectrum at energies overlapping with the peak from the ¹⁵N reaction. At low boron contents the ¹⁵N peak can be separated from the background. For detecting the γ-spectrum a NaI crystal detector with a photomultiplier tube was used to obtain sufficient energy resolution to separate the 4.43 MeV ¹⁵N peak from other high energy signals. The calibration of the entire analysis system required reference samples with implanted ¹⁵N to properly assess γ-ray and α-particle conversion factors and to adjust the beam energy for maximum γ-ray yield.

Figure 2 shows the HIERDA spectrum recorded on the graphite plate, shown in figure 1(a), which was exposed during the ¹⁴N and then ¹⁵N injection into TEXTOR. The spectrum reveals the presence of light isotopes in the co-deposit: hydrogen, deuterium fuel species, carbon eroded from limiters (including ¹³C injected to the torus on many occasions), boron from wall conditioning and oxygen which has partly been co-deposited as the plasma impurity atom and, partly absorbed by the layer when the exposed probe was stored in ambient atmosphere. Most relevant are the identification of helium, both nitrogen isotopes and fluorine (from WF₆ injections  [20, 25]), which could be detected due to the very high sensitivity of HIERDA. Other features belong to metals (Ni, Cr, Fe) from the Inconel liner and tungsten. Attempts to detect fluorine residual impurities were undertaken previously, but the sensitivity of the applied techniques was lower than that of ToF-HIERDA.

Zoom In Zoom Out Reset image size

The analyzed spots on both materials are marked with the numbered circles. The quantitative data for deuterium, nitrogen isotopes and helium measured in all four points on the C and W plates are collected in graphs in figure 3. As expected, nearly no deposition has been found in the region on the top of the limiter, i.e. point 1 on the graphite and point 5 on the tungsten plate. However, in all other areas the presence of ¹⁴N and ¹⁵N is detected, thus proving that a certain fraction of the gas injected for edge cooling remains in the vessel either as co-deposited atoms or compounds (such as C–N, W–N) formed on PFC under plasma impact.

Zoom In Zoom Out Reset image size

It should be stressed that in earlier experiments with nitrogen injection a small amount of that species (1–3 × 10¹⁵ cm⁻²) was measured with ToF-HIERDA on a tungsten test limiter exposed to discharges seeded with ¹⁴N [14]. Also traces of WN have been observed with x-ray photoelectron spectroscopy but it was difficult to quantify 'tokamak-related' nitrogen knowing that it could also be ad- or absorbed from atmosphere prior to the ion beam analysis measurements. The experiments with puffing a rare ¹⁵N isotope leave no doubts that the puffed nitrogen sticks to PFC surfaces. This also explains plasma memory effect in operation with nitrogen seeding, as described in the introduction. Helium is used for glow discharge cleaning and as a component of a B₂H₆–He mixture for the boronization process with both carbon and metal walls. Its detection in relation to W but not C is consistence with the high He impurity reported by Schmidt [26] for AUG. Its presence may be related either to the diffusion and trapping in tungsten or in mixed compounds such as W_xC, W_xN, WO₂, which were detected earlier by XPS [14].

Graphs in figures 4(a)–(c) represent maps of nitrogen and tungsten distribution on the graphite plate, shown in figure 1(b), with a deposit formed in the second tracer experiment in TEXTOR with ¹⁵N₂ toroidally symmetric injection and WF₆ puffing through the test limiter. The position of the gas inlet is marked with a black dot. All data for the concentration of species are in 10¹⁵ cm⁻². Pattern of the tungsten deposition is shifted due to drifts toward the high field side. The concentration exceeds locally 1 × 10¹⁸ cm⁻² with the maximum around 1 cm from the WF₆ injection hole. The distribution is similar to that measured [27] and modeled [28] earlier for ¹³C after experiments with the injection of ¹³CH₄ through the limiter. On the contrary, the maximum deposition of nitrogen isotopes is extended in the opposite direction. The pattern is nearly the same for ¹⁵N and ¹⁴N though the specific sources of the isotopes were different. While ¹⁵N was injected during the experiment, the other isotope has originated mainly from re-erosion and transport of species accumulated earlier in PFC in various locations. The isotope exchange, ¹⁵N to ¹⁴N, when the sample was stored in air cannot be fully excluded. However, it should be stressed that the measurements performed at different time intervals after the experiment did not show the decrease of the ¹⁵N content.

Zoom In Zoom Out Reset image size

The amount and distribution of the tracer was determined on 35 divertor and limiter tiles. The fraction of the ¹⁵N tracer retained on PFC could be assessed at the level of 14–15% of the total seeded gas. Approximately 10% was deposited on the main chamber wall and 4% in the divertor, as described in detail in [19]. The measurements have shown a global, non-uniform, deposition of nitrogen. The deposition pattern could not be fully reconstructed by early modeling [19] but it encouraged advanced studies. Images and graphs in figures 5(a) and (b) summarize the nitrogen deposition studies near the injection point which was located at the outer midplane on the right side of the antenna for ICRF heating—it is marked with a circle. Toroidal distribution of ¹⁵N measured along several antenna protection tiles is shown on the graphs (figures 5(c)–(f) which demonstrates that the distribution is not uniform. The greatest deposition is found on the protruding parts of the antenna protection tiles.

Zoom In Zoom Out Reset image size

The drawing in figure 5(b) presents the result of modeling. The dissociation chain of the injected gas molecules was modeled with ERO [29] until reaching ¹⁵N⁺ ions with plasma parameters measured during a reference discharge of the experiment. After ionization, the ASCOT code [30] tracked the ions using a magnetic equilibrium from the discharge #27385, a three-dimensional wall geometry of AUG and assuming a background plasma calculated with SOLPS [31].

According to modeling the deposition is not uniform and the greatest concentration of ¹⁵N is expected for the antenna protection tiles closest to the injection location at the middle of the antenna, which is also seen experimentally. On the antenna protection tiles on the opposite side of the antenna the highest concentrations are instead expected in the upper part which also is in accordance with the experimental results. This shows a good qualitative agreement between the experimental and modeled data. However, to elucidate the absolute values and to achieve still better correlation with the concentrations measured in deposits it would be necessary to include in the model also retention and re-erosion mechanisms.