Proceedings of the International Symposium on Plasma Wall Interaction

WALL

RECENT RESULTS ON PLASMA-WALL INTERACTION IN THE T.F.R. TOKAMAK

T.F.R. group*,     ASSOCIATION EURATOM-CEA SUR LA FUSION, Departement de Physique du Plasma et de la Fusion Contrôlée, Centre d’Etudes Nucléaires, Boíte Postale n° 6. 92260 FONTENAY-AUX-ROSES (FRANCE)

Publisher Summary

This chapter discusses the results of the survey of the phenomena related to the plasma–wall interaction in the T.F.R. Tokamak. Quantitative space and time resolved vacuum ultraviolet measurements allowed to obtain the initial plasma composition, thus, revealing that oxygen is, from the begin-ning, the dominant impurity. Photodesorption and photoelectron induced desorption were of the right order of magnitude to account for the oxygen fluxes. The chapter describes the technological advances made to the above phenomenon that deal with the dynamics of impurities as revealed by various injection experiments. To increase the ion temperature, fast hydrogen or deuterium neutral beams were injected at the current plateau for several discharge conditions. In both the cases, all the detected impurity signals were increased during the injection time, thus, showing strong increment in line radiation; bolometry has independently arrived at the same conclusion.

1 INTRODUCTION

A survey of phenomena related to plasma-wall interaction in the T.F.R. tokamak has been reported in January 1974 [1] and, more recently, in February 1976 [2]. After a brief review of these results, we shall describe in this paper the technological modifications made after February 1976, and present our latest results. They deal with the dynamics of impurities as revealed by various injection experiments, with the effect of reducing the limiter radius, and with the consequences of using boron carbide and graphite limiters. Finally, we shall try to draw some physical conclusions from the results described.

2 REVIEW OF PREVIOUS RESULTS [2]

Quantitative space and time resolved vacuum ultraviolet measurements allowed to obtain the initial (~7 ms) plasma composition, thus revealing that oxygen is, from the beginning, the dominant impurity. During the current plateau the VUV radiation emitted in the plasma periphery was interpreted in terms of a recycling flux (atoms/cm²sec entering into the discharge) and of radiated power per unit length along the plasma column (W/cm). The oxygen flux was the largest one, and, from particle balance considerations, it was concluded that all the incoming oxygen atoms could not reach the hot plasma core. Two molybdenum lines characteristic of the plasma center were detected and, from the corona equilibrium hypothesis, a central molybdenum density of the order of 10¹¹ cm−3 was deduced.

Several mechanisms were thought to be responsible for the plasma pollution. Photodesorption and photoelectron induced desorption were of the right order of magnitude to account for the oxygen fluxes. Surface evaporation of the limiter and sputtering due to fast protons or hydrogen atoms were thought to be responsible for the release of iron and molybdenum.

3 RECENT MODIFICATIONS

The vacuum chamber has been described in [2]; it must, however, be remember that, due to evaporation of the limiter, the walls are presumably covered with a thin molybdenum layer. The only important modification concerned the limiter, which was formerly made of a whole solid molybdenum ring. It has how been changed to three sectors (fig. 1), which can be moved radially in order to vary the plasma radius (from 12.5 cm to 20 cm) without breaking the vacuum.

Fig. 1 The limiter made of three sections

Rather than the baking procedure formerly employed, we have now systematically used cleaning discharges with the following characteristics: 50 kA, 3 kG, 0.2 Hz, 20 ms, H2(D2), 2.5 × 10−4 torr for two hours. This technique allowed a fast recovery to the best experimental conditions after each opening of the vacuum chamber.

4 IMPURITY BEHAVIOUR DURING PARTICLE INJECTION

4.1 Cold deuterium injection

Several attempts to increase the electron density were performed by injection deuterium gas (by means of either slow or fast valves, or by a combination of the two) during the discharge. In these experiments the electron density increased smoothly to the limit value. All the ultraviolet signals due to light elements (H or D, N, P, and O) also increased, whereas those coming from heavy elements (Fe, Ho) decreased slowly, both for peripheral and central ions. According to the usual interpretation of the VuV signals, this is due to the fact that there is an increased recycling flux of the light elements, but both a decreased incoming flux and a reduced central density for the heavy impurities.

Several causes can explain to increase of the oxygen flux: enhanced photodesorption, due to excitation of the injected deuterium; desorption due to deuterium or hydrogen atoms produced in the dissociation of the incoming molecules; enhanced flux of charged particles.

4.2 Cold oxygen injection

In order to get a better insight into the opposite behaviour of light and heavy impurities, oxygen was injected during the current plateau of a deuterium discharge (200 kA, 50 kG, a = 17 cm). The plasma parameters for discharge are shown in the first two lines of Table I. As a consequence of the oxygen injection there is a slight increase of the electron density, accompanied by broadening of its radial profile, on the other hand the electron temperature does not change. The impurity behaviour is shown in fig. 2 for oxygen and in fig. 3 for molybdenum.

TABLE I

Fig. 2 Oscillograms of 0 VI and 0 VII during the injection of cold oxygen and radial profiles 200 kA, 50 kG, a = 17 cm, deuterium, 1st June 1976.

Fig. 3 Oscillograms of Mo XXXI and Mo XXXII same conditions.

In fig. 2 we have shown the radiance of the OVI 1032 Å (photons cm−2 s−1 s−1r) and of the OVII 1623 Å (arb. unit) lines, along with the radial profiles of the OVI emission E(photons (photons cm−3s−1). As we have already said, a radiance increase indicates an increased flux; therefore, from the difference in the two oxygen signals it is clear that the increased oxygen flux deduced from the OVII line (i.e., the oxygen flux crossing the radius Te = 250 – 300 eV) is much lower than the oxygen flux deduced from the OVI line (i.e., the oxygen flux crossing the radius where Te = 50 eV). This is the first experimental evidence that the peripheral plasma shell has a much higher recycling rate than the hotter central plasma. The origin of this partial decoupling is still obscure.

In fig. 3 we have shown the line electron density (cm−2), the MoXXXII radiance (photons cm−2 s−1 sr−1), and the radial profiles of the MoXXXI and MoXXXII emission E (photons cm−3 s−1). Since E = ne nzQ (where nz is the molybdenum ion density, and Q the excitation rate coefficient almost independent of Te for the lines observed), the decreased emission, associated with the constancy of Te (0) (i.e., no displacement of the coronal equilibrium) indicates that molybdenum is expelled from the plasma center. Note that this would also imply that the density ratio for the two central molybdenum ions is constant in time. This is not the case in fig. 3 for the profiles at 260msec, but we think that the errors due to the measurements and to the Abel inversion are too large to draw any conclusion from this discrepancy. The time delay between the increase of the OVI emission and decrease of central molybdenum is of the order of 15-20msec.

fig 4 shows the radial profiles of the total power emission per unit volume measured by space-resolved bolometry. The oxygen injection results in a large decrease of the central emission. Since oxygen radiates mainly at the plasma periphery and molybdenum radiates at all radii, this result agrees with the spectroscopic observation, i.e., with a molybdenum expulsion.

Fig. 4 Bolometric profiles along a radius before (196 and 226 ms) and after (262 ms) injection of cold oxygen 165 kA, 50 kG, a = 17 cm, deuterium, 2nd June 1976.

Numerical simulations, not based on experimental data have indicated that the diffusion of a large oxygen puff towards the plasma center results in an inverted oxygen gradient which pumps out molybdenum. It is, however, not yet clear if this explanation applies to the present experiment, which could be probably also explained by a decrease of the invard molybdenum flux, due to the injected oxygen.

4.3 Injection of a powerful fast neutral beam

In order to increase the ion temperature fast hydrogen or deuterium neutral beams were injected at the current plateau for several discharge conditions[³]. In both cases all the detected impurity signals increased during the injection time, thus showing a strong increase in line radiation; bolometry has independently arrived at the same conclusion.

As compared to the experiments described in sections 4.1 and 4.2, heavy impurity behaved differently. This is probably due to the sputtering caused either by the increased ion temperature or by the unconfined and/or uncaptured fast particles. In order to discriminate between these two possibilities we performed a series of experiments with a co-and counter injection of deuterium on a deuterium plasma with co-injection of hydrogen on a hydrogen plasma. From the radial profiles of the molybdenum line emission, the central molybdenum density nMo (0) and the ratio nMo (0)/ne(0) were obtained. When co-injecting deuterium, both quantities increased during the injection time, whereas in the other two cases nMo(0) was constant (or only slightly increasing) and nMo(0)/ne(0) was constant (or only slightly decreasing).

Two arguments are in favour of sputtering caused by the increase of the ion temperature:

- the variation of the central molybdenum ion radiances as a function of the injected fast neutral current has a tendency to show a nonlinear increase of the radiance, this fact by itself would rule out sputtering caused by uncaptured fast particles.

- the order of magnitude of the measured radiances can be explained by sputtering due to thermal deuterium neutrals or by sputtering due to the oxygen ions. Since the sputtering yield is in both cases at approximately the threshold energy, it increases considerably by increasing the ion temperature (by neutral injection) thus explaining the molybdenum radiance mentioned above.

A typical growth rate for the molybdenum density in the case of deuterium on deuterium injection is

Assuming that this growth rate is uniform over the entire plasma, the incoming flux is then

Note that since the molybdenum density at the center is about 10¹¹ cm−3, the molybdenum particle confinement time in the hot central plasma is approximately 50 msec.

5 INFLUENCE OF THE LIMITER RADIUS

In order to discriminate between walls and limiter as the dominant source of plasma pollution, a series of experiments was performed by reducing the limiter radius from 20 cm down to 12,5 cm, while keeping constant the safety factor q(a) and the toroidal field. The plasma parameters for this series of experiments are shown in Table I (lines 3 to 5). fig. 5 shows the radial profiles of the OVI emission (photons cm−3 sec−1) for three different values of the limiter radius. Since the emission integrated along the line of slight is directly connected to the flux, it can be seen that decreasing the limiter radius from 19cm to 15cm. reduces considerably (approximately a factor of 5) the recycling oxygen flux (and, therefore, also the radiated power); this in agreement with the fact that oxygen comes mainly from the walls.

Fig. 5 Profiles of OVI along a radius for different limiter radii, March 1976.

As for molybdenum, a comparison can be made only between the first two discharges, since the third one (with a = 12.5 cm) has a central electron temperature which is just at the threshold for the appearance of MoXXXI. The radiance of the MoXXXI line for the first two discharges was approximately the same, but since the electron density and the radial emission profiles are different, we have deduced that the central molybdenum density increases by a factor of 1.5–2 when reducing the limiter radius from 19 cm to 12.5 cm. This shows that, by increasing the scrape-off distance, it is possible to obtain a discharge where the influence of the limiter is dominant over the influence of the walls. This conclusion must, howewer, be moderated by the fact that reducing the limiter radius modifies the discharge conditions and plasma parameters.

6 EXPERIMENTS WITH BORON GARBIDE AND GRAPHITE LIMITERS [4]

6.1 Limiters characteristics

Boron carbide and graphite 5890-CL limiters have been recently tested in T.F.R. Both materials were produced by Carbone Lorraine. They were studied, machined, and outgassed by CEA[⁵]. The thermal treatment consisted of 4 cycles, baking the pieces up to 1100°C under a pressure of 1 × 10−6 Torr. Both limiters were built in three sectors, as already described.

6.2 Gross behaviour of the limiters

The B4C limiter was submitted to 100 discharges at 80 kA in H2. Its mechanical behaviour was good and the plasma was macroscopically as good as the one obtained with a molybdenum limiter. However when the current was increased to 200 kA for 100 discharges, the plates were rapidly damaged, the most severe destructions being in the equatorial plane. This damage has been attributed to the poor ability of B4C to support thermal shocks.

On the other hand the C limiter behaved well mechanically up to 380 kA in H2 for more than 1000 discharges, most of them in the 200 – 300 kA range. This included a few severe disruptive discharges. The experiments with this limiter, extended over 10 working days, during which the vacuum chamber was opened three times. Due to the standard discharge cleaning procedure described in section 3 the recovery to the best experimental conditions after each opening was very fast. After dismantling the machine, a light irisation could be seen on the edges of the sectors and on their supports indicating that the same kind of metal transport occured with this limiter as with the previous molybdenum limiters. The middle sector was eroded around its center over a surface area of 4 cm² and a maximum depth of .5 mm. A few microfeailures, .1 mm thick, could be seen in this area (fig. 6). These observations show that this area was submitted to a thermal load higher than 10 kW cm−2 and that the eroded surface reached locally a temperature higher than 3000°C. The walls around the mimiter were covered with a thin layer of carbon powder which is easy to remove.

Fig. 6 Microphotographies of the failures of the graphite limiter.

6.3 Plasma performances with the carbon limiter

The plasma performances obtained with the carbon limiter are compared in Table I (lines 6 to 12) with those obtained in a few typical cases with the molybdenum limiter. The obvious conclusion to be drawn from this comparison if that there is no spectacular difference between the performances obtained with the two limiters.

All the results reported below have been obtained (unless otherwise specified) with a limiter radius of 19 cm.

Ultraviolet vacuum spectroscopy-measurements showed that:

a. all carbon signals (for example the CIV 1548 Å line) were very large for several discharges whenever there was a long interuption (< 15 mn) between two discharges. They came down to a stationary level, apparently representative of the discharge under study, within ten shots provided that a constant repetition rate, faster than 15 minutes, was maintained.

b. As for every other light impurity, the radiance of the carbon lines showed an initial peak and an approximately stationary plateau. The plateau is interpreted as an inward impurity flux[⁶]. Once a constant repetition rate was maintained, large fluctuations of the initial peak (up to a factor of 4) did not influence the plateau value. This behaviour is similar to that of oxygen in discharges with a molybdenum limiter.

c. Whenever the carbon radiance increased the molybdenum radiance decreased. This relation between light and heavy impurities was also observed when using a molybdenum limiter. (The comparison was done in this later case between 0 and Mo; see section 4).

d. Once a stationary regime was obtained the radiance of the CIV 1548 Å line, at the plateau, was the same with the carbon limiter as with the molybdenum limiter. The radiance of 0 and Mo were also the same order of magnitude in both cases. A comparison of the radiation losses is shown on Table II. One notices that 0 is still the dominant light impurity.

TABLE II

Power radiated for the whole plasma, KW, by various lines

e. The radiance of the different lines did not depend much on the value of the plasma current.

Very few measurements have been performed with a distance between the plasma and the wall increased to 3 cm. A decrease of oxygen was seen, similar to the one obtained in discharges with a molybdenum limiter when the scrape-off region was increased. However, with the carbon limiter a decrease of molybdenum was also observed, unlike the case in which a molybdenum limiter was used.

A radial exploration of the local emission of neutrals and photons was carried out with a bolometer. The results, given in fig. 7 show the local flux, W, in Watt cm−3 as a function of the radius.

Fig. 7 Bolometric profiles under the following conditions: a) 200 kA, 40 kG, a = 20 cm, Hydrogen, Mo limiter, 16 July 1976. b) 200 kA, 50 kG, a = 19 cm, Hydrogen, C limiter, 4 Augst 1976. c) 300 kA, 50 kG, a = 19 cm, Hydrogen, C limiter, 5 Augst 1976. d) 200 kA, 50 kG, a = 17 cm, Hydrogen, C limiter, 11 Augst 1976.

The following conclusions have been deduced from bolometry:

- with a Mo limiter the bolometric profile is initially hollow and then levels off (fig. 7a). This behaviour is in agreement with the spectroscopic observations which show that oxygen the main light impurity, is present early in the discharge. After the initial ionisation phase it radiates only at the periphery while molybdenum, the main heavy impurity, enters progressively into the plasma and radiates everywhere and hence at the center.

- with a C limiter, although there are a few exceptions (fig. 7b), the profiles remain hollow even for long discharge times (fig. 7c).

- the total radiation level is less with a C limiter than with a Mo limiter although it remains of the same order of magnitude (fig. 7a and fig. 7b).

- reducing the radius of a C limiter from 19 to 17 cm reduces the total radiation level by more than a factor two (fig. 7d). This reduction is not observed with a Mo limiter, and agrees with the spectroscopic observation of a molybdenum reduction.

6.4 Qwass experiments

The QWASS apparatus [⁷] has been recently connected to T.F.R. The results of these experiments are reported in detail at this conference[⁸]; here we shall only recall the main points.

The experiments consist in exposing a target to the plasma, varying its position in the scrape off region. 0, C, and Mo deposits were analysed. For a limiter radius of 20cm there was a reduction of a factor of two of the molybdenum deposit when going to the carbon limiter, thus showing that in these conditions a significant part of the molybdenum pollution comes from the walls, the rest coming from the limiter in the Mo limiter case. Such a reduction has not been observed for oxygen, probably as a consequence of the fact that oxygen comes mainly from the walls (see section 5). When the carbon limiter radius was decreased there was a further reduction of the molybdenum deposit, in agreement with the spectroscopie observations (section 6.3).

6.5 Mass spectroscopy

The results of mass spectroscopy measurements are reported in detail at this conference here we shall only recall the main points. The carbon limiter results showed less H2O, more CO2 and more CO than with the molybdenum limiter. However, the CH4 pressure was the same with the three limiter materials.

Moreover, with the carbon limiter the species abundance dit not depend on the value of the plasma current.

6.6 Summary of the results obtained with low Z limiters

The most important conclusions are summarized in the following:

a. The B4C limiter behaved well with low current plasmas but was unable to sustain high current discharges.

b. The carbon limiter behaved well mechanically up to the highest performances of T.F.R. (380 kA, 60 kG, 300ms).

c. The erosion of the carbon limiter is located around the equatorial plane but it is not too important, as for the Mo limiters.

d. The nature and relative concentration of the impurities are roughly the same for a Mo and a C limiter, when the distance between the plasma and the walls is about 1 cm. In particular the amount of carbon in the plasma is as low with a C limiter as with a Mo limiter, provided a high enough discharge repetition rate is maintained throughout.

e. In the same conditions, when the repetition rate is too low, the relative concentration of C inside the plasma is much larger with a C limiter than with a Mo limiter.

f. With a C limiter there are strong indications that the Mo content of the plasma is much reduced when the distance between the plasma and the walls is increased to 3 cm. The total power radiated is less than with a Mo limiter. These observations agree with the reduction of Mo deposits observed by QWASS.

g. The bolometric profiles agree with the spectroscopic observations, the radiation from the center and from the edges being dominated by the heavy and the light impurities, respectively.

h. The amount of impurities, with a C limiter, seems unrelated to the value of the plasma current. Moreover, also the abundance of the species detected by mass spectroscopy does not depend on the value of the current.

i. The use of a C limiter results in a spectacular decrease in the hard × ray and photo-neutron production[¹⁰].

Finally it must be remembered that:

j. In all cases τEe and Zeff with the C limiter are not much different from those obtained with the Mo limiters.

7 DISCUSSION

7.1 Edge diffusion

Looking at the experimental results presented above, it seems justified to introduce two characteristic lengths.:

– the critical thickness λ of the scrape off region, defined as the plasma-wall distance at which wall effects dominate over limiter effects from the point of view of plasma pollution. We have seen that λ ≅ 3cm.

– the thickness, d, of the peripheral shell in which confinement is poor, since most of the inward oxygen flux is reflected within it; this shell is also the origin of an important amount amount of line radiation. We have seen that d ≅ 4–5 cm.

From these two figures, a few calculations on the effects of plasma-wall interaction can be performed.

a) λ can be interpreted as the diffusion length charaeteristic of the particles in the scrape-off layer; it can, therefore, be related to a diffusion coefficient, D⊥

where L is the length of the torus, v// the thermal speed of protons with 10 ev < Ti < 100 ev, and α the coefficient of re-emission. It follows then that

b. If we assume that all the particles have the same transport coefficient and that the transport is the same in the scrape off layer and in the peripheral shell with poor confinement, we can deduce the particle life-time in this latter region

from which

the electron or proton flux ϕ at the plasma edge is, therefore,

and taking n = 10¹³ cm−3, one has

the flux of energy W for the whole torus is then

the surface of the torus being S = 8 × 10⁴ cm², one has

This low value for W seems to be in agreement with the spectroscopic and bolometric results, which show that most of the energy is evacuated to the walls by radiation.

7.2 Desorption

When the plasma is very close to the walls, i.e., with a limiter radius of 20 cm, most of of the flux ϕ falls on the walls. Both the electron and proton energies are sufficient to cause desorption. The desorption coefficient, ne in the energy range 10–100 eV is[¹¹]

giving a desorption flux

Although the equivalent data for protons are still uncertain, we have estimated from[¹²] that ni is at least ten times ne. The problem here lies in the surface covering. Since the TFR walls do not have very good cleaning, it is reasonable to assume that the working conditions of[¹¹] are realistic. Therefore, if we compare ϕde and ϕdi with the measured value of the inward oxygen flux[²], ϕox ≅ 1.5 × 10¹⁵ cm−2 sec−1, we are led to the conclusion that electron and especially ion desorption could explain the oxygen pollution when the plasma is close to the walls. This phenomenon seems to be a better candidate for hight impurity pollution than the photodesorption formerly considered[²].

7.3 Thermal load

When the plasma-wall distance is large, the heat flux W, leaving the plasma as charged particles, goes mostly to the limiter. The calculations presented above lead to a flux of 1 kW cm−2 in the meridian section. The actual flux received by the limiter surface depends on its shape, i.e., on the angle between the meridian plane and the limiter surface, and on the centering of the plasma. Any outshift decreases the surface in contact with the limiter and therefore increase the local thermal load. Large thermal loads are therefore possible, and could explain the molybdenum increase described in section 5. The experimental results suggest that thermal skocks are indeed responsible for the destruction and erosion of the limiter since the conclusions a, and c of section 6.6 are consistent with the respective capability of B4C, Mo and C to withstand thermal shocks.

The origin of this thermal shock was often thought to be the abrupt release of energy at the end of a disruptive discharge. Indeed sparkles of molted metal were observed with a fast movie camera at the end of every disruptive discharge but they were never seen when the discharge ended gently. This is not enough to single out either the thermal plasma or the runaway electrons as the only cause of the erosion. In favor of the latter is the observation that the region which is eroded is the same, with Mo limiters, as the one which is activated.

From conclusion of section 6.6 it can be thought that when the plasma wall distance is short the thermal load could also affect the protection plates of molybdenum which cover several parts of the walls.

7.4 Sputtering

Section 4.3. has shown that there are experimental facts indicating that sputtering, and not only thermal load, might be responsible for the molybdenum pollution. The experiments show that the thermal part of the plasma might be more important than the non thermal part when neutral injection is used.

This is in reasonable agreement with the following crude calculations which consider successively the neutral flux and the charged particleflux.

If one assumes that ni = 10¹³ cm−3 in a peripheral shell of poor confinement extending radially over b = 4cm, that the ion temperature in this region is about 100 eV and that the neutral density 3 10⁹ cm−3, a confinement time for the ions of τo = 10−2s is found due to charge exchange. The neutral flux, ϕo, impiging on the walls is then:

The sputtering yield for an energy close to the threshold value is not well knwon. It can be estimated to be around 5 10−4. The inward flux of molybdenum would then be 2 10¹² part cm−2s−1 which is a factor of 10 lower than the figure, ϕMo × 2 10¹³ cm−2 s−1, reported in section 4.3. However, due to the uncertainties of this calculation such an explaination for pollution cannot be ruled out.

Using the same sputtering yield applied to the charged particles estimated above (see section 7.1b) an upper limit of 4 10¹⁴ (1-α) cm−2 s−1 is found. This phenomena would then seem to be a better candidate.

Sputtering by oxygen ions, which recycle with a flux of 1.5 10¹⁵ cm−2 s−1 [²] and have a higher yield at these low energies ∼10−3, has already been pointed out as a possible cause of pollution[¹³].

7.5 Chemical processes

Conclusions d and e of section 6.6. show that the carbon release might be dominated by a slow process. Methane formation[¹⁴,¹⁵] is consistent with this observation. However, such a chemical process should not play a major role during the duration of the discharge.

Generally speaking, the results of section 6.5 show that the chemical activity of the walls plays a role in the abundance and the nature of the species ready to be desorbed. This points out the importance of cleaning procedures, as already shown on Alcator.

7.6 Characteristics of the carbon species used

Previous experiments with a limiter partially made of carbon led to different conclusions[¹⁶]. This could be due either to the fact that the carbon species were different either to the different working conditions.

REFERENCES

1. .H. Rebut et al., Proc. Int. Conf. on Surface effects in Cont. Therm. Fusion dev. and react., Argonne 10–12 Jan 1974.

2. TFR Group, Proc. Int. Conf. on surface effects in Conf. Fusion devices, San Francisco 16–20 Feb. 1976.

3. TFR Group, 3rd Int. Meet, on heating 28 June-2 July Grenoble 1976 EUR-CEA FC 834 July 1976.

4. TFR Group EUR-CEA FC 852 Oct. 1976.

5. M. Yvars, B.E. 76068 53425/Obj7623 MY/NL/Aout 1976. CEN-SACLAY/DGI/SEPCP/SSDP.

6. TFR Group, Nuclear Fusion 15 (1975). 1053–66.

7. G. Standenmaier, Proc. Int. Conf. on Surface effects in Cont. Fusion devices, San Francisco 16–20 Feb. 1976.

8. TFR Group, with collab. P. Staib, G. Staudenmaier, this conference.

9. TFR Group, with collab. K. J. Dietz, this conference.

10. TFR Group, EUR CEA FC 851 Sept. 1976.

11. J. Lelegard A. Schram, this conference.

12. M. H. Achard, A. G. Mathewson, CERN ISR VA/76−34. A. G. Mathewson, CERN ISR VA 176-41 and this conference.

13. R. Behrisch, (private communication).

14. S. K. Erents and al, Proc. Int. Conf. on Surface effects in Cont. Fusion devices, San Francisco 16–20 Feb. 1976 CLM P 451 1976.

15. S. Veprek, and al. Int. Conf. on Surface effects in Cont. Fusion devices, San Francisco 16–20 Feb. 1976.

16. Equipe Petula EUR-CEA-FC 810 p : 38 Petula Group IAEA-CN-35/A15

*With the collaboration of K.J. Dietz (Institut für Plasma Physik, KFA Jülich), P. Staib, G. Staudenmaier (Institut für Plasma Physik, Garching), V. Verchkov, Kurchatov Institute, Moscow.

DIVERTORS FOR IMPURITY CONTROL AND REVIEW OF IMPURITY EFFECTS IN PPL TOKAMAKS

Dale M. Meade,     Plasma Physics Laboratory, Princeton University, Princeton, New Jersey 08540, USA

ABSTRACT

was reduced to ≅1 for ne = 2 × 10¹³ cm−3, Te = 800 eV and Ti = 300 eV by gettering the interior of the vacuum chamber with titanium. In addition the interior vacuum wall surface has been monitored during discharge cleaning and plasma operation in an attempt to correlate impurities with wall conditions. Analysis of surface conditions in PLT indicate a decrease in carbon impurities in the plasma as hydrocarbons are removed from the surface test sample.The principles of divertor operation are reviewed. Requirements and methods of obtaining a shielding divertor are described.

1 INTRODUCTION

5%) can significantly increase the volt-seconds required to start a tokamak discharge; this is particularly troublesome for large tokamaks such as TFTR [1]. Later in the discharge impurities not only reduce τE but may be responsible for the density limit observed in tokamaks [2].

thereby decreasing the neutral-beam penetration and causing the plasma surface to be heated. Increased sputtering leads to more impurities and eventually a termination of beam heating. This "beam deposition instability [3]" will be of importance in future tokamaks and will limit the operating density of the device.

For low-Q (Q ≤ 1) reactors such as TFTR, the main effect of impurities is to deplete the fuel density leading to modest reduction in reactor performance while an ignition reactor is quite sensitive to small amounts of impurities [4].

2 IMPURITY CONTENT

The impurity concentrations in the PPL tokamaks were measured for several operating conditions. For standard PPL discharge cleaning (5–10 kA, 10 msec pulse length, 10 pulses per second for several days), the impurity content of ST during the steady-state part of the discharge approached the typical values [5] given in Column A of Table I. These impurity concentrations determined from UV spectroscopy (Fig. 2) and from x-ray spectroscopy [6] (Fig. 1), can be grouped into low-Z (oxygen) impurities that are fully stripped at the plasma center and high-Z metallic impurities that are never fully stripped and hence emit line radiation. The charge state of iron impurities was measured directly [7] by resolving the Kα iron lines (Fig. 3), the shift in the Kα line results from a reduction in shielding of the nuclear charge as electrons are removed the the L-shell.

TABLE I

Impurity Concentrations in ST

Figure 1 Time variation of the emission rate of various vacuum UV lines for a standard ST discharge. (PPPL 743644)

Figure 2 Typical x-ray spectrum measured in ST. (•) Lithium drifted silicon detector, 0.002 inch Be filter; (x) same detector, 0.005 inch Al filter; (o) NaI crystal detector. (PPPL 743677)

Figure 3 Spectrum of the iron Kα line structure for a ST plasma with a central temperature of 1200 eV. The coronal equilibrium is calculated using the theory of C. Jordan, Mon. Not. R. Astr. Soc. 148 (1970) 17. (PPPL 743749)

η may be due to the undetected tungsten.

in ST by removing questionable vacuum components and employing a more thorough discharge cleaning in hydrogen which was limited by the temperature rise in the vacuum vessel and limiter. In this case [η increased by ≅30%. Again the improved confinement for low-Z discharges was argued to be due to an improved current distribution rather then radiation, although this point may be debatable in the absence of total radiation