Colorimetric measurements of carbon erosion and deposition rates on extended areas of plasma facing components in TEXTOR

m

ELSEVIER

.i

l

G~

i

I

I

Journal of Nuclear Materials 220-222 (1995) 452-456

Colorimetric measurements of carbon erosion and deposition rates on extended areas of plasma facing components in TEXTOR P. Wienhold, F. Weschenfelder, J. Winter Institut fiir Plasmaphysik, Forschungszentrum Jiilich GmbH, Ass. EURA TOM-KFA, P.O. Box 1913, D-52425 Jiilich, Germany

Abstract

Colorimetry is introduced at TEXTOR as a technique to observe the carbon erosion on extended areas. It quantifies the interference colours of thin ( ~ 200 nm) transparent co-deposits of carbon and hydrogen which are mainly depending on thickness. The paper describes the facility as a video and an image processing system and the thickness determination by means of the colour coordinate hue (hxr). Erosion rates up to - 22 n m / s are measured on a precarbonized spherical graphite test piece exposed in the SOL for neutral beam heated plasmas. Part of the carbon is redeposited promptly with + 4 n m / s , but the transition range between net erosion and deposition is not stationary. Erosion is observed around the tangency point. Beyond 2 cm into the SOL, deposition is found to be influenced by the limiter geometry.

1. Motivation

Erosion at wall components in future fusion devices is realized to be a serious concern. For ITER, unacceptable high erosion rates are estimated [1] for graphite if no control techniques can be developed. After its transport as a plasma impurity, the eroded material is co-deposited with hydrogen at other places inside the machine. The deposit may cause intolerable sinks for the radioactive tritium. Direct observation on extended wall areas is required in order to investigate the spatial distribution of the erosion rates or to control the deposition by means of newly developed in situ repair techniques [2]. Pulse integrating information will not be sufficient to understand the process dynamics. In machines like TEXTOR material deposition is easy to observe because mainly carbon (graphite limiters, carbonized liner) faces the plasma. It forms amorphous a-C : H films which are reformed after plasma erosion. As long as the films remain transparent over extended areas they show interference colours in illuminating light which mainly depend on thickness. The observation of the changing colour patterns and, hence thickness changes can yield carbon erosion/deposition rates

if the density of the deposit of 6.5 × 1022 C / c m 3 [3] is taken into account. Before interference colours could be quantified colorimetrically first attempts were made in TEXTOR in 1988 by means of simple fringe analysis [4] in order to estimate the carbon rates and to verify transition ranges from deposition to erosion. Colour fringes on a boronized spherical test piece were compared ex situ by visual inspection before and after exposure (53.4 s) in the SOL of ohmically heated plasmas [5]. Deposition of +3 n m / s was found 2.5 cm from the edge and confirmed later [3,6], but erosion of - 2 nrn/s at 1 cm distance (tangency point). The existence of a transition range suggests that the competing erosion and deposition processes can balance each other [7]. Fringe analysis after long term exposures is not adequate, however, to observe the time evolution before the steady state is reached, because of the limited thickness resolution (20-50 nm). Other techniques like ion beam analysis of exposed samples turned out to be time consuming or are - as reflectometry - under further development at TEXTOR. Colorimetry aims at the quantification of the interference colours with better resolution (1 nm) in order to investigate changes after each discharge

0022-3115/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0022-3115 (94)00504-4

P. Wienhold et al. /Journal of Nuclear Materials 220-222 (1995) 452-456

and to contour the rates over extended areas (about 20 cm2).

2. Colorimetric determination of erosion rates in TEXTOR 2.1. Experimental setup

In order to observe areas a CCD colour camera views via a mirror through a top window of TEXTOR, but declined by 7.5 ° against the vertical direction (Fig. 1). The test piece is at the bottom of the torus at a total distance of about 2 m from the camera. Sufficient illumination was provided vertically through a second window by means of halogen light, and the camera was adjusted for 3200 K colour temperature. The development of the fringes is monitored and recorded via a 60 m long video connection. Before introduction into TEXTOR, the cylindrical graphite test piece had been prepared for an erosion experiment by covering it with an a - C : D film in the T E X T O R vessel along with a carbonization of the machine. A thin Al-film of some 100 nm sputtered before as a marker onto the graphite surface made the colours well visible. The a - C : D deposit was not homogeneous and appeared in yellow (150 nm) and purple (170 nm) of second interference order at the spherical end (insert). The later colour measurement was restricted to this central area (34 mm in diameter) outside of which colours became shifted too much due to the inclining incidence angle mirror

CCD camera

video connection

3Omm

- -

48ram

LI

453

(0 > 45°). Because of the curvature of the probe these points are 7.1 mm more apart from the plasma edge than the tangency point. Since it has been outlined earlier how interference colours develop on a system which consists of a thin a - C : H film onto an opaque substrate [4,8], brief remarks about the colorimetric formalism [9] and monitoring [10] will be given only. Colorimetry makes use of the fact that any visible spectral distribution creates a unique colour which can be described according to the CIE standards [11] in terms of three coordinates. On the other hand the identical colour impression can be admixed from three proper set visible wavelengths, such as emitted from the red, green and blue phosphors of a video screen. They are quantifiable and measurable, e.g. as voltages which are related to the colour coordinates [12]. Due to technical limitations of the video system the colour coordinate hue hxr (colour tone) only can be trasmitted with its full natural range. This is no restriction, however, because the colour tone of the reflected spectral distribution mainly depends on the thickness of the a-C : H film. The hue hxy is defined as a polar angle in the circle of colours [9] and ranges between 0° and 360°. Most pronounced for example are the colour impressions yellow (0-750), green (75-145°), blue (145-245 °) and purple (245-360 °) used for simple fringe analysis. Note, that a new interference order usually begins with yellow and that the hue decreases with increasing film thickness. The colorimetric system set up at T E X T O R measures hxy with a resolution of few degrees. Its calibration with samples is described elsewhere [5]. The a - C : D covered test piece was positioned for the first exposure with its tip at r = 47.3 cm by means of a vacuum lock system. The toroidal belt limiter ALT II defined the plasma radius a = 46 cm, other limiters were withdrawn. Plasma discharges in deuterium (n~ --2.8 X 1013/cm 3, I o = 357 kA, B T = 2.25 T) lasting 3.6 s were heated by co-injected hydrogen neutral beams with 1.3 MW over 2 s. The liner temperature was 150°C. After the first exposure the probe was moved closer to the plasma by 5 mm (r = 46.8 cm) and exposed during the following discharges until the a - C : D deposit had partly been eroded down to the Al-marker. Between shots, camera and illumination system were activated and short sequences recorded. 2.2. Results and discussion

i

PC + image processing- recorder -Fig. 1. Colorimetric setup at TEXTOR. The insert shows in top view the initial colour pattern of the a-C: D deposit on the spherical end of the carbonized test piece.

One single video image is sufficient for the colorimetric analysis and has to be read digitally by means of a frame grabber board into the computer where the hue values are calculated. Since we observed the test piece from top the images looked similar to the insert in Fig. 1 but with a changing fringe pattern. Not all of the data can be shown here, but the results given are

454

P. Wienhold et al. / Journal of Nuclear Materials 220-222 (1995) 452-456 o o

?~0 3 6 0

'k

t

before e x p o s u r e ~ 210

ion drift side electron drift side ..,.., .,..,..~..,..~..,..,..,..,.. 18-15-12-9-6-3 0 3 6 9 12 15 18

TOROIDAL DIRECTION ( r n m ) Fig. 2. Hue profiles measured across the test piece in toroidal direction before (full squares) and after the first (open squares) and second (triangles) exposure. The poloidal location is - 1.4 mm (out). characteristic. Hue values were measured along lines parallel to the toroidal field direction at different poloidal positions. The line scans connect two locations on ion and electron drift side (i and e) of the probe which are each 17 mm apart from the mid point (abscissa 0). Although still in the scrape-off layer, the spherical end of the test piece had approached an erosion dominated zone. This is indicated in Fig. 2 by an increase of almost all hue angles (open squares) compared to the ones measured over the virgin surface (full squares). The remarkable hue change at the probe tip (from 230° to 290°) continued during exposure 2 (full triangles) resulting into a change from purple to yellow colour. On the ion drift side, the hue increases even more while on the electron drift side a small decrease seems to indicate some deposition. Fig. 3 shows the thickness profiles at the same poloidal distance ( - 1 . 4 mm outside the midplane) as evaluated by means of a calibration curve for the 190 170-180-

before e x p o s u r ~

160v 150_ % dli O'} 140- ~

1202 11090

,dr sid . . 18 - 1 5 - 1 2 - 9 - 6

, -3

, 0

, 3

e,leetron drift, side 6 9 12 15 18

TOROIDAL DIRECTION ( m m )

Fig. 3. Thickness profiles as derived from hue in Fig. 2 and for two following exposures #54476-79) taking into account the hue-thickness relation as calibrated for the system a-C: H/AI.

system a - C : H / A l . Refraction and absorption constants n =2.22 and k = 0 . 1 3 , respectively, were assumed as measured ellipsometrically for carbonization layers [5]. n can be lower in redeposited material ( ~ 1.8) or under presence of other constituents [6] leading to an according underestimate of the thicknesses, which will, however, not influence the rates figured out from the differences of two succeeding profiles. The mismatch due to the varying angle of incidence does not exceed 10 nm. Up to 45 nm are lost at the ion drift side after the first exposure (# 54 476) and correspond to an erosion rate of about - 1 2 n m / s at r = 48.0 cm. Assuming a yield of 3% [13] for chemical carbon erosion this would require a hydrogen flux of at least 2.6 X 1018 cm -2 s-1 if self erosion and redeposition are neglected. The erosion rate is much higher than observed for ohmic discharges ( - 2 n m / s ) and most probably caused by higher fluxes due to the neutral injected beam. Fluxes of some 1019 D cm 2 s-1 are estimated near the edge of a test limiter under similar conditions [14]. Temperature excursions higher than 200°C can be excluded [15]. The erosion rates are higher ( ~ - 2 2 n m / s ) during the second (and the following) exposures with the probe 5 mm closer to the edge which results into an almost full erosion of the deposit beyond - 1 5 mm toroidal distance. This boundary line progressively moves towards the mid point. The appearance of newly deposited material between - 1 0 and 0 mm at places which showed net erosion before is hardly plausible without assuming the direct redeposition of carbon. Prompt deposition of Mo marker material has recently been reported [16]. After its erosion the carbon can travel radially until its ionization in the SOL plasma and follows the field lines until deposition of part of it on the spherically curved surface. About 2 mm ionization length would have to be assumed if the peak shift measures the radial travel distance here. This is likely if erosion is due to hydrocarbon formation with thermal energies. Profiles are shown only for the first four exposures (until # 54479), but the inspection of the following seven reveals that the peak shifts further and reaches stationarity at about + 3 mm. Physical sputtering does not seem to play an important role, because the aluminium substrate layer was still present after the total exposure time of 77.7 s. It protected the graphite bulk from further chemical erosion and prevented eventually the regeneration of deposits in the naked zones. Erosion is also observed around the tangency point (0 mm) despite vanishing of the projected surface. This can better be seen from the two-dimensional rate contour plots in Figs. 4a, 4b which were created for the first two exposures from scans made at different poloidal locations. A loss of - 4 n m / s would correspond to a flux of 0.9 × 1018 D cm -z s -1 if erosion by

P. Wienhold et al. /Journal of Nuclear Materials 220-222 (1995) 452-456

to the short (2 m) connection to limiter blade 4 which shadows the hydrogen flux and may act as a source of carbon. On the ion drift side, the connection length is about 35 m.

tO.

Z ~ r.)

0

3. Conclusions and summary

,-.-1 .,C ~

455

-10

O -15 -20

-15

r=47.3cm

-10

-5

0

5

10

15

20

15

20

TOROIDAL DIRECTION [mm]

10

~ Z

5

~

0

0 -15 -20

r=46.Scm

-15

-10

-5

0

5

10

TOROIDAL DIRECTION [mm]

Fig. 4 a, b Contour plots of rates in nm/s as evaluated from line scans at different poloidal locations after the first two exposures. Deposition zones (positive numbers) are shadowed.

hydrogen is assumed only. Direct cross field diffusion of hydrogen ions to limiter and probe surfaces (funnel effects) have recently been discussed [17] in terms of a new model in order to explain observations made in TFTR. The poloidal asymmetry of the erosion/deposition transition range is possibly due to E x B forces or poloidal plasma rotation which would cause preferential ion transport into the positive poloidal direction [18,2], but as mentioned above the deposition dominated zone (shadowed) moves in positive toroidai direction as exposure time proceeds. On the electron drift side net deposition develops between 10-15 mm toroidal distance (corresponding to r = 47.2 cm) while on the ion drift side deposition was found beyond r = 48.0 cm (outside information range). It continued into the cylindrical parts of the test piece, which were oriented perpendicular to the toroidal field lines, and levelled off at r = 52 cm. This carbon is transported via the SOL from other sources to the sample. Peak deposition rates of ~ + 3 n m / s could be estimated from the fringe patterns after the dismounting of the sample, but 5 mm more close to the plasma edge on the electron drift side. This is presumably due

In spite of the preliminary character of the interpretation, these first results show that carbon erosion and deposition in TEXTOR can directly be measured on extended areas ( ~ 10 cm 2) after a single discharge by colorimetry of the interference colours as long as the carbon exists as transparent film (a-C : D). This implies thicknesses less than 1 ixm and temperatures below 700°C. The rate resolution is about 1 n m / s . Observations can be made after each discharge and reveal transient phases before stationarity is reached. Modeling of the material transport in dependence on plasma parameters will not be possible without additional information from other diagnostics, but the great number of data which are yielded by colorimetry from extended areas should ease a solution. So far, the major results are: in neutral beam heated (2.6 MW) discharges, carbon erosion 2 cm into the SOL achieved rates of ~ - 2 2 n m / s corresponding to losses of 2.6 x 1017 C / c m 2 s. Part of it is promptly deposited about 2 mm more close to the plasma edge and with rates of + 4 n m / s . The short ionization length suggests chemical erosion rather than sputter -elease of the carbon. The deposited carbon undergoes ~rosion in the following discharges and causes a toroidal shift of the erosion/deposition transition range until the carbon source at the erosion dominated area is fully depleted. Erosion ( - 4 n m / s ) at vanishing projected areas with respect to the field direction (tangency point) may indicate newly predicted erosion channels (funnel effect). First rate contour plots reveal the poloidal asymmetry of the erosion pattern and point to a possible influence of E x B forces or plasma rotation. Stationarity is characterized by the transition to a deposition zone 1.2 and 2.0 cm apart from the plasma edge on electron and ion drift side, respectively, depending on connection length. This carbon is eroded from other sources and deposited with peak rates of + 3 nm/s. The results are little influenced by surface geometry or roughness. The data evaluation has still to be made interactively, in order to e.g. recognize interference orders, and is far from any automatic run. Fringe analysis must not be excluded. Optical constants of both, substrate and film enter, but mainly the refractive constant of the deposit influences the hue-thickness relation. At present, the error limits are about _+15%, but with much smaller statistical error.

456

P. Wienhold et al. /Journal of Nuclear Materials 220-222 (1995) 452-456

Acknowledgements The authors gratefully acknowledge the advice of R. M611er and H. Hailing, Zentralinstitut fiir Elektronik, K F A JiJlich, Germany, for the definition of the video and computer system and the adjustments of the electronic components. They also thank H. Bergs~tker and M. Rubel, K T H Stockholm, who prepared the graphite test piece with aluminum, B. Schweer and S. Musso helped us to run their lock system at T E X T O R .

References [1] M.F.A. Harrison, E.S. Hotston, G.P. Maddison, NET Report EUR-FU/80/90-97, 1990. [2] H.G. Esser et al., these Proceedings (PSI-11), J. Nucl. Mater. 220-222 (1995) 457. [3] P. Wienhold, J. von Seggern, H.G. Esser et al., J. Nucl. Mater. 176/177 (1990) 150. [4] P. Wienhold, U. Littmark, E-MRS Symp. Proc. Vol. XVII, p. 441, 1987, Strasbourg, France. [5] P. Wienhold, F. Weschenfelder and J. Winter, Nucl. Instr. and Meth. B 94 (1994) 503.

[6] F. Weschenfelder, Thesis, 1993, Univ. Diisseldorf, Germany. [7] G.M. McCracken, D.H.J. Goodall, P.C. Stangeby et al., J. Nucl. Mater. 162-164 (1989) 356. [8] P. Wienhold, F. Weschenfelder, R. M611er, 20th EPS Conf. on Controlled Fusion and Plasma Physics, Lisboa, 1993, Vol. 17C, part III, p. 1119. [9] D.L. MacAdam, Color Measurement, (Springer, Berlin, 1981). [10] G.R. Jones, P.C. Russell, Pure Appl. Opt. 2 (1993) 87. [11] Commission Internationale de l'Eclairage (CIE), Publication no. 13.2, Paris, 1974. [12] EBU Standard for Chromaticity Tolerances for Studio Monitors, Tech. 3213-E, Brussels, 1975. [13] E. Vietzke, V. Philipps, Fusion. Techn. 15 (1989) 108. [14] V. Philipps, A. Pospieszczyk, U. Samm et al., J. Nucl. Mater, 196-198 (1992) 1106. [15] V. Philipps, T. Tanabe, Y. Ueda et al., Nucl. Fusion 34 (1994) 1417. [16] D. Naujoks, R. Behrisch, V. Philipps, B. Schweer, 20th EPS Conf. on Controlled Fusion and Plasma Physics, Lisboa, 1993, vol. 17C, part II, p. 651. [17] P.C. Stangeby, C.S. Pitcher, J.D. Elder, Nucl. Fusion 32 (1992) 2079. [18] J. Winter, P. Wienhold, Contrib. to High-Temp. Plasma Phys. part II, eds. K.H. Spatschek and J. Uhlenbusch (Akademie Verlag, Berlin, 1994) p. 431.