Search

A small fraction of the α-particles produced in a future (D, T)-fusion plasma will escape directly and hit some areas of the surrounding solid walls with their full energy, while the majority will be confined in the magnetic field, slowed down in the plasma and reach the first wall with about the same energy distribution as the other plasma particles. During bombardment of a solid by energetic α-particles the implanted surface layers undergo several modifications leading finally to a rough surface with a spongy microstructure. Some of these modifications, i.e. blistering and flaking, are only observed if the critical gas concentration builds up first at some depth below the surface. This condition is met only if simultaneous erosion by sputtering stays below a certain limit. In fusion devices with loss fractions of ≲ 1% high energy α-particles and a plasma temperature near the first wall above some 10 eV, blistering and flaking are likely to be avoided.

The trapping of 5 keV deuterium in ∼ 165 and ∼ 4000 nm thick BeO films grown by thermal oxidation on Be substrates was investigated at different temperatures using the D(3He,H)4He nuclear reaction. The ratio of implanted D to BeO molecules obtained at saturation is 0.24 to 0.34. The D migrates from its end of range location and distributes itself uniformly in the BeO film. With increasing implant temperature the BeO layer flakes from the Be substrate. The distribution of D in BeO films at high concentrations is not consistent with diffusivity measurements at low concentrations of T in BeO.

The computer simulation program MARLOWE which follows the trajectories of energetic ions and recoiling target atoms in solids has been used to calculate sputtering yields for low energy (0.1–10keV) light ions (H, D, T,4He). Recoil energy densities were calculated for comparison with analytical theories. The sputtering yields obtained for amorphous Fe agree within a factor of two with experimentally measured values for polycrystalline stainless steel, while the calculated yields for protons on amorphous molybdenum are more than twice the experimental values on polycrystalline material. The calculations show that in the parameter range investigated, ions backscattered in the solid contribute a major part to sputtering. This result confirms earlier calculations of the threshold energy for sputtering which are in agreement with recent measurements.

The microstructure and chemical composition of dropletlike deposits formed on the surface of graphite limiters due to plasma discharges in the axis symmetric divertor tokamak experiment were analyzed using 57Fe Mossbauer transmission and conversion electron spectroscopy. It is found that the droplets consist mostly of alloy cementite, (Fe,Cr,Ni)3C, with an admixture of austenite smaller than ≂20%. The amount of cementite and the contribution of austenite depend considerably on the location of the deposits relative to plasma‐drift direction, which reflects the varying conditions of metal and carbon redeposition and agglomeration in the tokamak chamber.

The chemical composition of iron-containing droplets formed on the graphite tiles of the poloidal ASDEX limiter after exposure to plasma discharges was analysed using 57Fe Mossbauer spectroscopy. Measurements in both transmission and conversion electron Mossbauer spectroscopy modes indicate that iron is present predominantly in the form of an alloyed carbide phase of the cementite type (Fe, Cr, Ni)3C.

After having been in use during about 150 discharges the graphite limiter of the ASDEX tokamak was removed and analyzed by means of SEM, EIXE, FIXE, RBS and Mossbauer Spectroscopy. The surface was seen to be covered by a nonuniform layer of 10 16 –10 18 metal at/cm 2 , primarily in the form of droplets with diameters up to a few μm. Simulation experiments suggest that the formation of these droplets is caused by heating. The relative metal concentrations in the droplets are close to those for stainless steel. The Mossbauer spectroscopy analysis indicates, however, that the major part of the metal is present in carbide phases, most probably (Fe, Cr, Ni) 3 C. On parts of the limiter the metal coverage shows a sharp maximum at the points nearest the main plasma, flanked by two minima. The minimum on the ion drift side is much broader than the one on the electron drift side. However, these features are not common to all parts of the limiter.

The interaction of the boundary plasma with the limiters has been analyzed for two 5 MA, 20 s Ohmic discharges in JET by Langmuir probes and by an all carbon collector probe. The deposition of D, C, Ni and Cr and the erosion on the probe have been measured as functions of the distance to the last closed magnetic flux surface. The values are compared to the particle fluences and their energies as calculated from the measured plasma parameters. The calculated deuterium fluences are about a factor of 5 to 10 larger than the deposition on the probe and the calculated e-folding length is steeper than the measured values. The carbon deposition and erosion at the probe has been determined with the aid of 13 C implants as well as from the depth of the metal and deuterium depositions and the interference colours of the deposited films. The distribution of deposited carbon is similar to the trapped deuterium with a ratio D : C of about 0.3, indicating that D trapping is dominated by saturation of the deposited C layer. The metal deposition is small; it amounts to a few times 10 −3 of the carbon deposition, indicating the dominance of carbon in the boundary layer.

The cathode spots from vacuum arcs on 316 stainless steel are compared with the tracks found on the same material after exposure to the plasma of the tokamak TFR 600. Further the erosion yields of vacuum arc cathodes of 316 stainless steel and titanium are determined from experiments and the measured values are compared with theoretical estimates. The velocity of the arc is investigated as a function of the applied magnetic cross-field. The scatter of both, the velocity data and the erosion yields is substantial. Improved experiments are planned.

Gas re-emission and surface structure modifications measured by diffuse scattering of light were observed simultaneously during the implantation of 40 keV helium in poly- and single crystalline nickel at room temperature. At a critical fluence of 5.2 × 1017 He+/cm2 the He concentration reaches a maximum of 0.4 He atoms/Ni atom at the most probable range and gas re-emission and blistering start simultaneously. With implantation up to 2 × 1018 He+ /cm2 the re-emission approaches 100% of the incident flux, the scattered light intensity reaches a maximum and the He concentration reaches about 0.3 He atoms/Ni atom over the whole range. For further implantation the intensity of scattered light decreases probably due to the blister covers becoming removed by sputtering, while the gas re-emission remains unchanged. It can be inferred from the measurements that the blisters are not the dominant source of gas re-emission, but that both effects are due to a common mechanism.

The erosion yields by vacuum arcs of 16.6 A in a magnetic field of 0.05 T have been measured by weight loss as a function of temperature for A1, A1 + 3 % Mg (proposed for INTOR) [1] and TiC which was deposited by chemical vapour deposition (CVD) in a 5 μm thick layer on Mo and stainless steel. For Al the erosion increases steeply at a temperature above 0.8 T m , arc velocity decreases with temperature and the arc craters become larger.

This paper presents the results of a round‐robin neutron‐sputtering experiment in which six independent research groups have participated. Targets of cold‐rolled niobium prepared by two standard techniques were used by all participants. Three participating groups used (d,t) neutrons of about 14.8 MeV, one group used (d,Be) neutrons produced by 40‐MeV deutrons on a thick Be target, and two groups used 16‐MeV protons in transmission. The emission of micrometer‐sized particles was observed by two groups and not observed by the other four. This particle emission was observed by two of the groups who used (d,t) neutrons in their irradiations. Chunk emission was observed for only one of the two standardized target types used in the round robin. The round‐robin results support the concept that the neutron‐sputtering yield of Nb, including atomic and chunk emission, has an estimated probable value no larger than 10−4 atoms per neutron. However, sputtering yields in the range 1.4×10−5 to ≲2.6×10−3 are reported here. The probable value for the sputtering yield was found to be independent of the two target preparations and the method of analysis, whether or not chunks were observed during analysis.

A procedure is developed for determining the stoichiometry of a sample as function of depth from ion-beam analysis energy spectra. The approach is in principle equally applicable to back scattering experiments, experiments involving nuclear reactions with known cross sections and experiments combining these techniques. The procedure is especially suitable for routine computer evaluation of energy spectra. It is more straightforward and/or involves less rigid assumptions and approximations than alternative approaches. If all elemental signals are measured, an accurate beam dose is not needed. The limitations are basically those inherent to ion beam analysis in general.

The trapping of 7 keV deuterium in 316 stainless steel at room temperatures has been measured as a function of current density, fluence and time. The depth distribution of the deuterium which diffuses into the stainless steel is obtained using the D( 3 He, p) 4 He reaction. The measured yields have been deconvoluted to obtain the concentration profile.

The exchange of helium isotopes 3 He, 4 He implanted into polycrystalline niobium at 30 and 50 keV was investigated by measuring the decrease of 3 He during postimplantation of 4 He and the increase of 3 He implanted after preimplantation of 4 He using the 3 He(d, α)H reaction. At low doses both isotopes are trapped with nearly 100% efficiency until a saturation concentration of the isotope mixture has been obtained. Thereafter the exchange proceeds to a first approximation according to a simple mixing law. From this we conclude that at saturation concentration the binding energy of the helium to traps is low and the helium diffuses readily within the implanted layer.

Trapping of 4, 8, 16 keV 3 He and 4, 8 keV D up to saturation and release of implanted 3 He by D bombardment and vice versa was investigated in polycrystalline Mo targets at 160 K using the 3 HeCd, p) 4 He nuclear reaction. In a saturated sample the initial release does not depend on whether the range of the first implant is smaller, nearly equal, or larger than the second implant. This behaviour cannot be explained by a replacement process. In unsaturated samples the release of the first implant occurs only after the second implant has reached saturation.

The surface layers of tiles from the carbon limiters in octant 1, ocant 4 and the RF-antenna limiters were analysed after the end of the first three experimental sessions of JET, 1983, 1984 and 1985. On all tiles metallic contaminants were found from the vessel walls and the RF antennae. In general on all areas deposition and erosion was observed. At the part of the tiles in direct contact with the plasma the erosion dominates, whereas deposition dominates on the outer parts of the limiter radially further away from the plasma. In the small gaps between adjacent limiter tiles, deposited layers of carbon having a thickness of several μm were found, which were partly flaked and contained metallic impurities. Electrical are tracks were observed on the limiter tiles and in the area between limiter tiles.

The modifications of the vessel wall and limiter surfaces in today's large plasma machines are reviewed with special emphasis on the investigations at JET concerning material erosion, transport and redeposition. This is correlated with the problems of impurity release and hydrogen recycling. The specific roles of the limiters and the vessel walls are discussed and some consequences on the use and the distribution of different materials for in-vessel components in tokamaks are indicated.

A proposal has been prepared for three facilities for diagnosing the plasma surface interactions and the plasma edge parameters in JET. The principal system proposed will be for plasma edge diagnostics. A carriage driven by linear motors and containing up to five probes will be transferable into the torus from a surface analysis station. After exposure to a discharge the probes can be returned to the analysis station and analysed before the next discharge. The second system will be for short-term specimens. It will consist of a magazine of samples which can be exposed in the torus for one or more well-defined discharges. These samples will be introduced and removed from the torus via a “vacuum suitcase”. The third system, termed “long-term” samples, consists simply of regions on the inner wall of the torus where samples can be fixed at different toroidal and poloidal locations. These samples will be removed in stages during the lifetime of the machine and examined to see the long-term integrated effects on surface damage, mechanical properties, and hydrogen isotope implantation.

In high temperature plasma experiments arc tracks are found on all wall areas in close contact with the plasma. The craters of such arc tracks, which are the same as for vacuum arcs, were investigated in more detail. The investigation was done with an optical microscope and a scanning electron microscope. Crater diameter and crater depth from vacuum arcs on clean cathode surfaces were measured for Cu, W, Al and Fe. Crater dimensions were determined from probes exposed to one single arc discharge moving across the probe surface due to a cross magnetic field. Controlled parameters were arc current and cathode temperature; their influence on crater size was investigated. Cleaning of the tungsten cathode surface was achieved by heating in vacuum. In other cases such as with copper cathodes simultaneous heating and exposure to a glow discharge was necessary to obtain reproducible surface conditions. Crater size was determined by statistical methods. The most probable diameter of the measured distribution curve is representative for the processes involved.

The magnetic confinement of a hot hydrogen plasma as investigated in fusion research is limited and the vessel walls are subjected to bombardment with ions, electrons and energetic neutral atoms. This load causes hydrogen implantation, erosion predominantly by sputtering, and redeposition of eroded material. At areas closer to the plasma, erosion dominates while deposition dominates at more distant areas. The surface layers are finally transferred into a mixture of all elements from the vessel walls up to depth of microns. The walls are additionally heated by the particles and the radiation coming from the plasma. At areas of high heat load the deposited layers can agglomerate to small droplets and surface melting as well as cracks may develop. The large fusion experiments under operation today as JET, TFTR or JT-60 are designed to reach plasma parameters close to those in a fusion reactor. Here the control of the plasma surface interaction processes represents a problem in respect to reducing the wall damage and minimizing the tritium inventory.

The surfaces of two Be limiters and of a test sample from the stainless steel torus wall used in UNITOR have been analyzed by RBS, PIXE, AES, and SEM/EIXE. The main foreign atoms found on the limiters are O (2 × 10 17 to 2 × 10 18 /cm 2 ), Fe, Ni Cr, and Sn (2 × 10 15 to 2 × 10 16 /cm 2 ). There are no large differences of the foreign atom coverage on the different surface areas of the limiters. Generally the front surface of the limiters shows a slightly lower coverage than the sides. On limiter 2, the Sn coverage is about a factor of 4 larger than on limiter 1, indicating that the Sn source was close to this limiter. The stainless steel test sample from the torus wall shows islands of BeO, which can be attributed to splashes or droplets eroded from the limiters during disruptions. Major erosion of the limiters occurred only during the first 100 shots, when many discharges were terminated by disruptions. Redeposited wall material is found on the wall sample which partially covers the BeO deposits. The surface modifications of limiters and wall samples were discussed by taking into account the particle fluxes from and to the walls and limiters.

Ohmic heating studies in hydrogen and deuterium up to currents of Ip = 5 MA have been completed and additional heating experiments by means of ICRH and NBI are now in process. With ICRH powers up to 7 MW, the global energy confinement time τE is observed to deteriorate with increasing heating power. NBI experiments are at a more preliminary stage, but also show deterioration in τE. Detailed studies of confinement time scaling in ohmically heated discharges show a strong dependence of plasma size, τE ∼ R1.7a1.3, but a much weaker dependence on q and ne than seen in smaller experiments. Boundary phenomena have been studied from early 1984. Data on the impurity coverage of limiters and wall have been collected during three successive experimental campaigns. Substantial amounts of wall material have consistently been found on the limiters and a large inventory of hydrogen isotopes concentrated on the colder edges. The wall in the shadow of protective elements shows non-uniform erosion. Langmuir probe measurements of edge plasma parameters have been taken in OH, RF and NB discharges. Additional heating leads generally to a temperature increase of the scrape-off layer. Model calculations using data from these diagnostics seem to produce a consistent picture of impurity production in the scrape-off layer which agrees reasonably well with spectroscopic observations.

The velocity of vacuum arcs in a magnetic field and the erosion at the cathode have been investigated for Al + 3% Mg and stainless steel. The arc currents were adjusted between 16.6 A and 70 A and a magnetic induction of 0.02 to 0.1 T was applied. The arc current and the arc velocity can be measured reproducibly only after conditioning of the cathode surface by 20 to 40 arc operations. For conditioned cathodes at room temperature the arc velocities increase with arc current and magnetic induction. For higher cathode temperatures the arc velocity decreases while the erosion yields and the crater size increases.

The JET scrape-off plasma during X-point discharges with L- and H-modes is described. There is a flow of current in the scrape-off layer which can influence equilibrium, transport, heating and impurity generation in the scrape-off layer. A local measurement of D ⊥ is obtained in the divertor region. The saturated flux of ions flowing parallel to the magnetic field, I SAT is ∼10 A/cm2 at the divertor separatrix in ohmic discharges. In L-mode with NBI it increases by a factor ⩽ 2, depending on injected power level ( P NBI MW ) and decreases with H-mode. The e-folding decay length, λ I SAT , is the same in ohmic and L-mode but decreases with H-mode. The electron temperatures at the divertor target are ~ 30 eV (OH) and ~ 50–60 eV (H-mode). Measurements of T e during L-mode are not reported because of the high level of fluctuations. An enhancement in the erosion of carbon due to ion-stimulated desorption at the divertor target is discussed. Flow reversal in the scrape-off layer is demonstrated.

Any radiation causing damage in the bulk of a material also causes sputtering, i.e. the release of atoms at the surface. For neutrons, we have to distinguish between backward, forward and isotropic sputtering, depending upon the direction of the neutron flux. The sputtering yields, defined as the mean number of atoms removed from the surface per incident neutron, are generally expected to be proportional to the specific damage energy deposited in the surface layers or the damage energy cross-section, $ σn(E)v(T). The cross-section, σn(E), refers to elastic and inelastic scattering as well as the different nucle reactions. If the primary knock-on atoms are radioactive their emission yield can be detected separately. Generally, neutron sputtering yields are of the order of 10−5 atoms per neutron, except for materials containing fissionable atoms, where a yield of the order of 1 was reported. Even the sputtering yields measured in some detail, such as for Au and Nb, still give only upper limits. For very rough surfaces a small part of the material was found to be emitted as microparticles.

Total erosion yields by sputtering and blistering for 1 to 15 keV H2+ bombardment at normal incidence have been measured by weight loss of 304 stainless steel, pyrolytic graphite, carbon fibres, glassy carbon and SiC. The erosion yields are in the range of 3 × 10−3 to 2.6 × 10−2 atoms per incident hydrogen atom. Observation in the scanning electron microscope shows that blisters occur in stainless steel and SiC at doses of 5 × 1018 particles/cm2, but disappear at doses of 5 × 10 particles/cm2 . The surface roughening observed depends largely on grain orientation. On carbon no blistering could be found. After bombardment the carbon surfaces are generally more smooth than before.

Absolute measurements of the energy distributions of energetic light ions back-scattered from solids can be used to determine the electronic stopping power of these ions in the solid. Stopping powers of protons (50–150 keV) in different metals and metal oxides (Nb, Ta, Ta2O5) have been measured by this method. The results depend strongly on the cleanliness of the surface region. The experiments show deviations from the additive rule for the stopping power in compounds in this energy range.

For protons of 40 to 120 keV energy impinging on the surface of a copper single crystal, the energy and angular distributions of the backscattered hydrogen atoms were measured, especially in and around the [Formula: see text] direction.

Minimum yield of backscattered protons and H-atoms with primary energies of 50–150 keV from a well annealed Nb single crystal has been measured in single and double alignment as a function of emerging energy at different temperatures. From these data minimum yield as a function of depth has been calculated, assuming electronic energy losses only. The results are compared to analytical equations for minimum yield in single and double alignment. These equations were obtained from Lindhard's theory of electronic and nuclear multiple scattering in single crystals. It was assumed that transverse energy increases steadily with penetration depth, diffusion being neglibible. At higher energies the agreement between theory and experiment is good in single alignment. In single alignment and at lower energies experimental values are about a factor of 2 higher. In double alignment there is a discrepancy of nearly one order of magnitude, which probably must be attributed to lattice defects.

BS>From international conference on ion-surface interaction, sputtering, and related phenomena; Garching, Germany (Sep 1972). Sputtering yields for 33- to 150-keV protons normally incident on Cu-films deposited on different backing materials were measured. The decrease of the Cu-film thickness was determined by Rutherford backscattering. As the base materials used have different backscattering intensities, the contribution to sputtering yield by incident ions and by the ions backscattered to the surface within the material could be separated. It was found that the relative contribution to total sputtering yield by the backscattered ions increases slowly for protons in the energy range from 50 to 150 keV. For energies lower than 50 keV the sputtering yields are determined by the special geometry of the target sandwich used. (auth)

The subject of sputtering by ions and neutrons has been reviewed, and yields for various ions on Nb in the energy range of 100 eV to 100 keV have been assembled. Furthermore, evaporation rates of Nb for stationary conditions and for short time energy pulses are calculated, and a brief summary is given of the blistering phenomena.For a Tokamak or Stellarator-type CTR first-wall bombardment is estimated from the a-energy released in the plasma. First-wall-erosion data as well as the flux of sputtered and evaporated Nb-atoms back to the plasma are calculated for two CTR reactors with a difference of a factor of 10 in first-wall load. It is shown that sputtering can account for wall deterioration at a rate of more than 1 mm per year.

The determination of the thickness and composition of thin layers up to several micrometres is of great importance in many applications. We have investigated the characteristics of two major methods in this field: 1. (1) the back-scattering of protons and helium ions in the energy range 0.060–2.5 MeV, and 2. (2) sputtering combined with AES. Both methods have been applied to identical metal sandwich targets of Be, Cu and Nb. The results show that the methods supplement each other remarkably. The non-destructive back-scattering technique gives quantitative thickness measurements of the high mass components (Nb, Cu) with an accuracy to about 10% of the total thickness. The depth resolution is between 100 and 200 A near the surface. Its increase with depth depends on the energy of the primary ions, e.g. it increases by a factor of two roughly every 4000 A in the case of 2 MeV He+ in Al. The destructive sputtering and AES technique has the advantage that the depth distribution of nearly all components (i.e. including impurities like C and O) can be determined with high depth resolution, especially in surface layers. If the excitation cross section and escape depth of the Auger electrons are taken into account, relative concentrations can be estimated. Non-uniform sputtering may cause serious errors. For absolute thickness measurements the method can be calibrated by the back-scattering method. Sputtering yields measured on Cu and oxygen-contaminated Nb are in rough agreement with the values in the literature. The depth resolution of sputtering combined with AES is approximately 50 A near the surface and its increase with depth depends strongly on the target composition.

A layer of TiD2 (5 × 1019 TiD2 cm−2) was prepared by deuterium implantation into a polycrystalline titanium foil at room temperature. The foil was subsequently installed in front of a large area surface barrier detector in a vacuum of 10−7 mbar to measure the reaction products of the two possible D-D fusion reactions. After a counting period of 30 hours, at most two fusion products were found, which limits the reaction probability to below 8 × 10−24 fusion processes per second and per deuterium pair in the implanted layer.

The depth profiles of 1.5–15‐keV 3He ions implanted into a Nb single crystal at doses of 5×1016–7×1018/cm2 have been measured using the 3He (d,p) 4He reaction. A comparison of the results with theoretical predictions for the range and the damage distribution of 3He in amorphous material shows reasonable agreement. Furthermore, the Deckeldicke (i.e., thickness of the covers of the blisters) was determined by Rutherford backscattering in double alignment. The results indicate that stress release rather than explosion of gas bubbles is the dominant mechanism in blister formation.

With the progress in controlled thermonuclear fusion research, experiments having dimensions approaching those of future fusion reactors and discharge times of the order of 100 to 1000 s have become a necessity. Parameters close to those of ignited plasma are anticipated or achieved in today's four large tokamaks TFTR (Princeton), JET (Culham), JT-60 (Naka-machi) and T-l 5 (Moscow); the burning of a fusion plasma is to be investigated in the next generation of experiments. In these large experiments the load from the plasma on the surrounding vessel walls is considerably increased in duration and intensity compared to that of most previous plasma experiments. The wall load is approaching the values foreseen in a fusion reactor and the lifetime of the wall components represents a major physical and engineering problem.

A review is given of the physical mechanisms involved in the interaction of fast hydrogen ions with solids, their diffusion and trapping in the solid lattice after slowing down, and the surface recombination allowing them to get reemitted. Analytical and computational models are discussed which describe the buildup of hydrogen inventory in a vessel wall and its release through the surfaces. They are applied to characterize wall material candidates for fusion devices with respect to their interaction with hydrogen isotopes, and to compare their suitability as seen from plasma control and safety demands.

The various effects of impurities on hydrogenic plasmas are reviewed. The problems involved in impurity diffusion are explained, and the simplifications necessary for including impurity diffusion in computer studies are discussed, several models are presented as examples. The different contributions to the impurity particle fluxes are compared. Finally, the concept of a “Cold plasma blanket” for impurity flux control is investigated.

The plasma wall interactions for two extreme cases, the ‘vacuum model’ and the ‘cold gas blanket’ are outlined. As a first step for understanding the plasma wall interactions the elementary interaction processes at the first wall are identified. These are energetic ion and neutral particle trapping and release, ion and neutral backscattering, ion sputtering, desorption by ions, photons and electrons and evaporation. These processes have only recently been started to be investigated in the parameter range of interest for fusion research. The few measured data and their extrapolation into regions not yet investigated are reviewed.

The surface structure of a solid bombarded by light ions is determined by the gas accumulation in the near surface region and consequent blistering as well as by sputtering. At bombardment with monoenergetic ions with doses of 10 17 to 10 18 cm −2 the surface structure is mainly determined by blistering and/or exfoliation. At higher doses surface structures generated by blistering are gradually sputtered away. Finally, an equilibrium surface structure develops.

The particularities of graphite sputtering at elevated temperatures are presented. Chemical sputtering, i.e. volatile molecule formation with hydrogen isotopes is extensively studied at energies above 100 eV and at thermal energies. Data on radiation enhanced sublimation of carbon are still scarce and available only for ion energies above 100 eV. The current understanding of both processes is summarized and an attempt is made to quantify erosion yields for fusion relevant conditions.

Electrons are emitted from solid surfaces as a result of heating, of ion, electron and photon bombardment from the plasma, and if a sufficiently high electric field is applied. Thermal electron emission only becomes important at temperatures above 2000° C. With particle bombardment at energies in the 10 to 100 eV range, the yields are typically of the order of 0.1 to 1 except in the case of highly charged ions, where potential emission yields may be well above 1. Field electron emission may be important for the ignition of electrical arcs.

A future tokamak reactor has to fulfill several requirements: good confinement properties to achieve ignition at low external heating power; high stability at minimal toroidal magnetic field strength; and efficient helium removal from the system for continuous burn. These conditions have to be met under circumstances, where the material, surrounding the burning plasma, is loaded to its limits. During steady state burn the first wall will be exposed to neutron fluxes of about 3 MW/m2. Additionally 600 MW is deposited into the plasma by the fusion α-particles, conducted to the plasma boundary and has to be removed from there at a tolerable power density. The wall is further exposed to a He-flux of 1.2 • 1021 s-1 and a deuterium and tritium flux which is about a factor of 20 larger. An implication of these figures is that several tens of tons of wall material are eroded and re-deposited during one year of reactor operation /1/.

An outline is given on the current understanding of physical sputtering with emphasis of the use in fusion reactor technology. This determines the ion species H, D, He, C, 0 and self ions as well as the energy range from the sputtering threshold to several keV. Additionally an attempt is made to present empirical analytic formula where comprehensive theories do not exist. Finally, as an example, the impurity fluxes from charge exchange wall sputtering in ASDEX are calculated.

In the last fifteen years, great strides have been made in controlled fusion research. The plasma temperatures, beta, and confinement needed for a reactor have been approached in recent experiments (Table 1). In the next few years, experiments on JET (EC), TFTR (USA), JT-60 (Japan), and T-15 (USSR) are expected to answer many of the remaining physics questions as to whether it is possible to obtain reactor grade plasmas in large tokamaks. The major physics questions are: (1) confinement (2) beta (3) heating and current drive (4) impurity and particle control.