Journal of Nuclear Materials 196-98 (1992) 569-572 North Holland

jnurnalof nuclear

materials

Erosion of tungsten by self-sputtering and light ion at oblique angles of incidence

C.H. Wu The NET Team, Max-Planck-lnstitut f~r Plasmaphysik, D-8046 Garching bei Miinchen, Germany

E. Hechtl Physik Department, Technische Universith't Miinchen, D-8046 Garching bei Miinehen, Germany

irradiation

Tungsten is being considered as plasma facing material in next step fusion devices. The advantage of this material is its high surface binding energy and consequently its low physical sputtering yield by light ions. The drawback is the low tolerable limit of this impurity in the fusion plasma. However, this material may be a favorable choice for operation at low plasma edge temperature. This paper provides values of sputtering yields for self-sputtering of tungsten as a function of the angle of ion incidence. Sputtering yields are measured in the range between 300 eV and 2 keV and between normal incidence and 60 ~ to normal. Tungsten shows a strong angular dependence of the yield. A self-sputtering yield of tungsten, y = 1, is reached at an angle of incidence 0 = 42 ~ at an impinging energy E = 350 eV. The results are compared with computer simulation calculations. The dependence of the sputtering yields on the structure of the target material was also investigated. In addition, the physical sputtering yields of tungsten by the light ions D, T and He as a function of angle of incidence and impinging particle energy were calculated. The results imply that a self-sputtering yield of tungsten y < 1 is expected at an angle of incidence 0 < 30 ~ and at an impinging particle energy E < 350 eV. On the basis of all knowledge available, the potential of tungsten for use as plasma facing material was discussed in detail.

I. Introduction

The objectives of the I T E R / N E T next step fusion devices are [1]:

a) the demonstrat ion of self-sustained burn of a D / T plasma and controlled long pulse operation;

b) the demonstrat ion of safe operat ion of a device that uses basic technology appropriate to a future fusion reactor and that can test essential components under conditions where the fusion reaction is self-sus- taining.

To meet these objectives, next step machines will operate at pulse lengths between 700 and 2000 s, which will require substantial extrapolation in the plasma physics data base from present tokamaks, e.g. JET, TFTR, JT60, A S D E X and T E X T O R . The fusion power of I T E R / N E T will probably exceed 1100 MW, and, consequently, the energy deposition on the divertor plates could be as high as 25 M W / m z, which means that the plasma flux density impinging the divertor plates at low divertor temperature T e < 20 eV may exceed 5 x 10~9/cm 2 s. At this high plasma flux den- sity, very high erosion yields of low-Z materials, e.g. carbon and beryllium, are expected.

This indicates that, when low-Z materials are used as divertor armour, the impurity production and the lifetime of the divertor plates are the critical issues.

Apart from the low impurity limit tolerable in fu- sion plasmas, high-Z materials possess high surface binding energies and consequently low physical sput- tering yields by light ions. In particular, the erosion of high-Z materials, e.g. W, Mo, by light ions may become very marginal, if the divertor temperature is kept low, T e < 30 eV. However, impurity production via self- sputtering may be critical depending on the impinging particle energy and angle of particle incidence.

In continuation of our erosion study on tungsten [2], we performed experimental measurements of tungsten self-sputtering and theoretical calculation on tungsten erosion via physical sputtering by light ions as a func- tion of the impinging energy and angle of incidence, the energy range for self-sputtering is 350 e V - 2 keV, the angle of incidence 0 ~ 0-75 ~ whilst the energy for light ions is 150-350 eV.

This paper presents experimental and calculated results of a study of tungsten erosion. On the basis of these results, the potential of tungsten as material for PFCs is discussed and tentative conclusions are drawn.

0022-3115/92/$05.00 �9 1992 - Elsevier Science Publishers B.V. All rights reserved

570 C.H. Wu, E. Hechtl /Erosion of tungsten

2. Experimental

The sputtering irradiation was performed with a Harwell-type isotope separator at Technische Univer- sit,it MiJnchen. In the isotope separator the beam is normally accelerated to 20 keV and then decelerated to the desired energy. With normal impact of the ion beam a single immersion lens is used for this beam retardation, see ref. [3]. With oblique angles of inci- dence it is necessary to have a beam retardation system that yields a field-free space in front of the target. Otherwise the impact conditions are not well defined. A special decelerat ion system was therefore designed for oblique angle irradiation; see ref. [4]. We use a Freeman-type ion source in the separator. A W + ion beam is produced with a tungsten fi lament as the feed supply. Carbon-tetrachloride is fed into the source via a gas inlet system.

The chlorine reacts with the hot fi lament and yields a total W + current of a few tens of micro amperes, which, though small, is sufficient for our investigations. To run the source, a support gas, normally argon, is used. The particle flux on the target was about 3 • 1013 W + cm -2 s -1. The base pressure in the irradiation chamber was 10 -6 Pa. During ion source operat ion the pressure rises to about 10 -5 Pa. The tungsten sheet targets with a purity of at least 99.95% were supplied by Plansee, Reut te , Austria.

The gross release yield was determined by the weight loss method. The integrated beam current was mea- sured with an accuracy of better than 10% and the mass change of the target with bet ter than 1 p.g by a Sartorius ultramicrobalance. The irradiation time was chosen to reach a mass change of about 100 ~g. In the vicinity of a gross yield of unity, the mass change approaches zero since gain and loss almost cancel out. In these cases the irradiation times were chosen such as to reach a total impinging ion mass of about 200 Ixg.

The experiment was carried out for two relevant energies namely 350 eV and 2 keV, and the angle of incidence was in the range of 0-75 ~ .

3. Results and discussion

3.1. Tungsten self-sputtering

The results are summarized in table 1. The gross yield Y + R (Y is the self-sputtering yield, R is the particle reflection coefficient) versus the angle of inci- dence for two energies is shown in fig. 1.

In the mass loss method we used, we cannot distin- guish between sputtered and reflected particles. With increasing impact angle measured against the normal, the reflection coefficient increases and cannot longer be neglected (as for normal impact). It is seen that two sets of experimental results were obtained for 350 eV.

Table 1 Gross self-sputtering yields of tungsten for two ion energies E and various angle of incidence 0

0 Y + R E = 2 keV E = 350 eV

With Without buildup buildup layer layer

0 ~ 0.30 1.76 15 ~ 0.40 2.07 30 ~ 0.86 0.70 2.96 45 ~ 1.25 1.07 3.72 50 ~ 1.16 60 ~ 1.16 4.44 75 ~ 3.74

The gross yield for 350 eV ions is below unity for a wide range of angle of incidence. In these cases a buildup of layer occurs.

Since the target was reused for a couple of mea- surements, the buildup of the layers accumulated, this making it possible to measure the yields of such buildup layers even if the value is above unity. The yield of such a buildup layer is shown as the dashed curve in fig. 1. The yield of the sheet metal with no buildup layer is shown as the bold curve. The thin curve repre- sents the calculation Y + R with the TRIM-SP code [5]. The agreement of the calculation with the gross yield with no buildup layers is very good up to an angle of 50 ~ , the yield of the buildup layer is higher. The surface binding energy is obviously lower in such a layer.

To obtain an idea of the quality of the surface, the SEM micrographs shown in fig. 2, were made. The left picture shows a tungsten sampled eroded with 2 keV W +, and the right one a sample bombarded with 350 eV W* at an angle of incidence of 30 ~ This dune-like surface is probably responsible for the higher yield

5 , i , f , I ' i , i , , , i , i ,

"6 350 eV

=+ ) . .

L O.lJ , i , i , ] i ~ i ] , i J i , ] ,

0 10 20 30 /,,0 50 60 70 80 90

ANGLE OF INCIDENCE, O[degree]

Fig. i. Gross yield of tungsten self-sputtering as a function of energy and angle of incidence.

C.H. Wu, E. Hechtl /Erosion of tungsten 571

Fig. 2. SEM micrographs of tungsten samples: (left) bombarded with 2 keV W+; (right) bombarded with 350 eV W +.

owing to a lower binding energy. The yield curve for 2 keV ion impact is in good agreement with the calcu- lated values (this curve). In this case the yield is always above unity and no buildup occurs.

3.2. Erosion of tungsten by light ions

Erosion of tungsten via physical sputtering by the light ions D +, T +, and He + has been calculated with the Yamamura formalism [6,7] for energies of 150, 200, 250, 300 and 350 eV as a function of the angle of incidence. The calculated results are given in figs. 3-5.

Figs. 3 and 4 show the ratio of the physical sputter- ing yield at incident angle (0) to that at normal inci- dence, Y(O)/Y(O), as a function of incident angle. It is seen that the physical sputtering yields reach a maxi- mum at incident angles between 75 and 80 ~ for D and T ions. The angularly dependent sputtering yields of 4He show behaviour similar to that of D and T ions. However, the maximum ratio, Y(O)/Y(O), decreases with increasing ion mass.

Following Brooks analysis [8,9], at relevant plasma conditions: B = 5 T, t0=87 ~ eto 0 = - 3 k T ~ , T e = 3 0

5

o w r > . . .

3 (D

>" 2

, 1 1 1

T ~ W

l l l l l

10 20

, , i r 1350'eVl ~

ov 2so .v \',: 2oo ~v ~ " , - - - , \', 150 eV ~ / \,i,

I I I I I I I I I I I

30 40 SO 60 70 80

O

9O

Fig. 4. Ratio of physical sputtering yields, Y(O)/Y(O), of tungsten as a function of incident angle and incident energy

for tritium.

eV, and N~o = 102o m 3. The angle of incidence of tungsten will be around 18 ~ and the average charge state of tungsten is about 2.1. The impinging energy of tungsten is far below 350 eV, which implies, that the over all release, Y+ R, will not exceed unity. There-

CD > -

0 '

I i

13~W

1 i I i I

10 20 30

T T I I I I I I / ~

350 eV - - _

3 o o e v - _ //.~'~:~ 2 oov - _ _ /~. 7 / \ 'd 200 eV - ~ h!

.~' . :

I I I I I I I I L I

40 50 60 70 gO

@ 90

Fig. 3. Ratio of physical sputtering yields, Y(O)/Y(O), of tungsten as a function of incident angle and incident energy

for deuterium.

~ I i t I ~ f i l I I I t I I I I I t

30 He--,- W .......... �9 / / ~ , , .

2 5 . . / / ~ ~

20 ~ \

. . . . . . . . . _ . . . . - - "*~ 300 ev 1 0 250 eV /

20o ev J i. 0 5 150 eV / ~: ' k

0 I I I t I I I I I I I l I I I I " % ~

10 20 30 40 50 60 70 80 90

@

Fig. 5. Ratio of physical sputtering yields, Y(O)/Y(O), of tungsten as a function of incident angle and incident energy

for helium.

572 C.H. Wu, E. Hechtl / Erosion of tungsten

fore, runaway self-sputtering is not expected under these conditions.

The angle of incidence of the light ions D, T, and 4He will most probably be smaller than 0 = 60 ~ At Te = 30 eV, the erosion of tungsten via physical sput- tering by light ions is therefore negligibly small.

4. Conclusion

In the present study, an experimental investigation of the self-sputtering and a theoretical calculation of light ion erosion of tungsten as a function of incident angle and particle energy are performed. Furthermore, the influence of the charge state of the ions on the use of tungsten as for PFCs material has been discussed, and the following conclusions can be made:

a) It is shown by this study that the results of TRIM-SP calculations of tungsten self-sputtering of a virgin surface as a function of the energy and angle of incidence agrees well with those of experimental re- suits. However, the erosion yield of redeposited mate- rial is higher than the TRIM-SP calculated value. The surface topography will influence the rate of impurity release.

b) The erosion of tungsten via physical sputtering by light ions is negligibly small at 0 < 60 ~ and T e = 30 eV.

c) From this study a preliminary conclusion can be drawn, namely that under tokamak relevant conditions, T e = 30 eV, 0 = 18 ~ average charge state of tungsten

ions, 2.1, a runaway self-sputtering of tungsten is not expected. Therefore, it can be stated that the T e = 30 eV will be the upper temperature for use W as plasma facing material.

Acknowledgements

It is a pleasure to thank Dr. W. Eckstein for TRIM- SP code calculation on tungsten self-sputtering.

References

[1] D.E. Post et al., ITER Documentation Series no. 21 (IAEA, 1991).

[2] E. Hechtl, H.R. Yang, C.H. Wu and W. Eckstein, J. Nucl. Mater. 176 & 177 (1990) 874.

[3] E. Hechtl, Nucl. Instr. and Meth. 186 (1981) 453. [4] E. Hechtl, Proc. 12th Int. Conf. on Electromagnetic Iso-

tope Separators and Techniques related to their Applica- tions, ed. M. Fujioka, Sendai, Japan, 1991, Nucl. Instr. and Meth. B70 (1992) 441.

[5] J.P. Biersack and W. Eckstein, Appl. Phys. 34 (1984) 73. [6] Y. Yamamura, Y. Itakawa and N. Itoh, IPPJ-AM-26,

Nagoya University (1983). [7] C.H. Wu, J. Nucl. Ma~er. 160 (1988) 103. [8] J.N. Brooks, Phys. Fluids B2(8) (1990) 1858. [9] J.N. Brooks and D.N. Ruzic, J. Nucl. Mater. 176 & 177

(1990) 278.