Tungsten is another of the plasma-facing materials, described in Chapter 4.17, Tungsten as a Plasma­Facing Material. Like carbon, it will not be a vacuum barrier. Thus, permeation through the tungsten will not lead to tritium release directly into the environ­ment. It can lead to tritium permeation into the coolant through the coolant tubes inside the tungsten facing materials. Permeation will also affect the tri­tium inventory of the fusion device. Tungsten has excellent thermal properties with a very high melting point of 3683 K. The problem that tungsten presents to the tokamak designer is the deleterious radiation losses if tungsten is present in the plasma. Fortu­nately, the energy threshold for sputtering by
hydrogen ions is quite high, 700 eV for tritium.52 For that reason, tungsten will be used primarily in the divertor region where the energy of the impacting particles can be limited.

There are a limited number of reports on the diffusivity of hydrogen isotopes in tungsten. Frauen — felder53 measured the rate of hydrogen outgassing from saturated rolled sheet samples at temperatures over the wide range 1200-2400 K. His material was 99.95% pure tungsten. Zakahrov and Sharapov54 used 99.99% pure tungsten samples in their permeation techniques to determine the hydrogen diffusivity over a limited temperature range of 900-1060 K. In the Benamati et al55 experiments using tungsten con­taining 5% rhenium, a gaseous permeation technique was also used. These experiments were performed over a very limited temperature range of 850-885 K. Reported diffusivities are shown in Figure 5. There are a couple of reasons why the diffusion coefficient reported by Frauenfelder53 is widely accepted as most correct. The first of these reasons is the wide temperature range over which experiments were per­formed. The second reason is that the experiments were performed at a temperature above that where trapping typically occurs. It can be seen in Figure 5 that Zakahrov’s54 diffusivity agrees quite well with Frauenfelder’s at the highest temperatures, but falls below his values at lower temperatures, where trapping would occur.

The database on hydrogen solubility in tungsten is also limited. The results of the two experimental
studies are shown in Figure 6. In the same experi­ments used to determine the diffusivity, Frauenfelder53 also measured solubility. Over the temperature range 1100-2400 K, samples were saturated at fixed pressures and then heated to drive out all of the hydrogen. Over a more limited temperature range of 1900-2400 K, Mazayev et a/.56 also examined hydrogen solubility in tungsten. The agreement with the Frauenfelder’s53 data is quite good in magnitude, but not good in apparent activation energy. As with his diffasivity, the solubility reported by Frauenfelder is typically the value used in predicting the migration of hydrogen in tungsten.

Hydrogen trapping in tungsten has been studied by several research groups. van Veen et a/.57 used bom­bardment by 2 keV protons in their study of the bond­ing of hydrogen to voids in single-crystal tungsten. Thermal desorption from the samples with appms of voids revealed a broad release peak at 600-700 K. It was stated that the release could be modeled as gas going back into solution from the voids with a trap binding energy of 96.5-135 kJ mol-1 controlling the process. Eleveld and van Veen,58 in a similar study, used a lower fluence of 30 keV D+ ions in desorption experiments. In these samples containing vacancies but no voids, the release occurred at 500-550 K. The authors reported a value of 100 kJ mol-1 for the trap binding energy of vacancies. Pisarev eta/.59 used lower fluences of 7.5 keV deuterons into 99.94% pure tungsten samples. During thermal desorption ramps, peaks in the release rates were seen at 350,480,600, and 750 K. The release at the highest temperature was seen only in the highest

fluences. Garcia-Rosales eta/60 used 100 eV deuterium implantation to study the trapping and release rate of hydrogen isotopes from wrought and plasma-sprayed tungsten. Two broad desorption peaks at 475-612 K and 670-850 Kwere seen in the thermal desorption spectra. Modeling of the release data suggested the lower temperature peak to be controlled by both diffu­sion and trapping at a binding energy of 44 kJ mol — . The second release peak was reported to correspond to trapping at defects with a binding energy of 97 kJ mol-1. In experiments with 99.99% pure tungsten and tungsten with 1% lanthanum oxide, Causey et a/.61 examined tritium retention in plasma-exposed samples. Modeling of the results suggested two traps, one with a binding energy of 97 kJ mol-1 and another with 204 kJ mol-1. The density of the trapped tritium averaged 400-500 appm. Anderl et a/.62 used deuterium im­plantation into polycrystalline tungsten to determine the correlation between dislocation density on cell walls and deuterium trapping. Annealing tungsten at 1673 K reduced the dislocation density by a factor of 7, subsequently reducing the deuterium trapping by a similar factor. The binding energy of these traps was estimated to be 88-107 kJ mol-1. As-received 99.95% pure tungsten was used by Sze et a/.63 in experiments with intense deuterium plasma exposure. Exposure at 400 K resulted in blisters with diameters of tens of microns. Elevating the temperature to 1250 K elimi­nated the blisters. Venhaus et a/.64 used high-purity foils in experiments to examine the effect of annealing temperature on blistering by deuterium plasma expo­sure. An unannealed sample and one annealed at 1473

K both exhibited blisters after the plasma exposure. The sample annealed at 1273 K did not blister. There have been a multitude of other reports on blister for­mation on tungsten samples exposed to various forms of hydrogen implantation.65-68

Anderl eta/.62 used 99.95% tungsten in 3 keV D+ ion implantation to determine the recombination — rate coefficient. Over a temperature range of 690­825 K, the recombination rate coefficient was given as kr = 3.85 x 109 exp(-13 500/T)m4 s-1 per mol of H2. This expression is shown in Figure 7, where it is plotted along with the expression given by the Baskes24 model. It can be seen that there is very little correlation between the measured Anderl value and the calculated Baskes value. This is not entirely unusual. Impurities on the surface, especially oxide layers, can have a very strong effect on this coefficient.

While tungsten has excellent low permeability for gaseous tritium, it will be used only in fusion devices as a plasma-facing material. As a plasma­facing material, tungsten will be exposed to intense fluxes of energetic tritium and deuterium. With traps for hydrogen at binding energies of 97 and 203 kJ mol-1(57-62) at natural and radiation-induced defects, it would appear that a substantial tritium inventory could be generated in divertor tungsten. There are several reasons why this high inventory is not likely to occur. The first reason is the high recombination — rate coefficient given earlier. For a recombination — rate constant of 10-1 m4 s-1 per mol of H2 or higher (see Figure 7), the recombination rate on the surface is so rapid as to generate the equivalent of c = 0 at

the boundary. With the very limited penetration dis­tance of energetic hydrogen in the dense tungsten, most of the implanted material is immediately released back out of the surface. There are also recent reports suggesting that ruptured blisters and very fine cracks near the surface69-71 will even further reduce the inward migration of deuterium and tritium into the tungsten.