Erosion behaviour and surface composition modifications of SiC under D+ ion bombardment

cm

Nuclear Instruments and Methodsin Physics Research B 111 (1996) 63-69

. . __

IiWMl

__

B

Beam Interactions with Materials&Atoms

!!!z ELSEVIER

Erosion behaviour and surface composition modifications of Sic under Df ion bombardment H. Plank

*,

R. Schwiirer, J. Roth

Max-Planck-Institutjiiir Plasmaphysik, Euratom Association, D-85748 Garching, Germany

Received 20 July 1995;revised form received 12 October 1995 Abstract The erosion behaviour of fibre reinforced Sic under deuterium ion bombardment is investigated in the ion energy range between 20 eV and 1 keV. The sputtering yields are compared with earlier results obtained from SiC samples with different

structure and fabrication. At D+ energies below 100 eV chemical erosion of C is found which results in a surface enrichment of Si. The surface composition modifications are investigated by Auger Electron Spectroscopy. The surface concentrations have been measured as a function of target temperature up to 1000 K. It is found that the Si surface enrichment reflects the ion energy and temperature dependence of the chemical erosion of C in Sic. At low Df energies a Si rich thin altered layer covering the Sic target is formed.

1. Introduction

In recent years carbon based fibre reinforced materials came into wide use as plasma limiting surfaces in fusion devices. In general, these materials show higher resistance against crack propagation and failure by fracture compared to monolithic grades. For fibre reinforced Sic the limited thermal stability, the low thermal conductivity and the neutron damage behaviour limits the application as plasma facing material (Ref. [ 11 and references therein). However, the use in receded areas of the blanket with softer neutron spectrum would allow operation to reasonable lifetimes of the material. Another requirement for the application in fusion devices is a low erosion yield of the plasma exposed material. Carbon based materials show chemical erosion under hydrogen bombardment, i.e. the release of gaseous hydrocarbons. Compared to pure graphite doped graphites and carbides show reduced chemical erosion [23]. In binary systems however, sputtering can cause a change of the surface composition due to threshold and/or chemical effects [2]. For Sic a depletion of the surface from carbon is expected, mainly due to chemical effects. These surface composition modifications may change the erosion behaviour of the material. We have investigated the erosion of SIC composite material (fibre reinforced SiC: Sic/Sic) under deuterium

bombardment at energies below 1 keV. A comparison to earlier results obtained from Sic with different structure and fabrication is given. The composition changes of the surface ,due to deuterium impact are investigated in situ by Auger Electron Spectroscopy @ES).

* Corresponding author. Tel. + 49 89 3299 2227, fax + 49 89 3299 1149, e-mail [email protected]. 0168-583X/%/$15.00 0 1996 Elsevier Science B.V. All rights reserved SSDI 0168-583X(95)01258-3

2. Experimental

2.1. Sample characterization SiC/SiC samples were provided by General Atomics (GA) [4]. They consist of Nicalon fibre reinforcement with a carbon coating of about 0.3 ym. The matrix is Sic deposited by chemical vapor infiltration at 1100°C. The fibre architecture is four layers of 2-D woven fabric. The residual porosity is estimated by GA to be = 10%. The material should be considered as a based-line material which has not been optimized anyway. Before performing the experiments the samples were mechanically cut and cleaned in an ultrasonic bath. One sample was analysed by X-Ray-Photoelectron Spectroscopy (XPS) combined with 3 keV Arf sputtering to determine the bulk composition and the chemical state of the material. No preferential sputtering of Si and C in Sic samples by Ar+ sputtering is reported in the literature [5,6]. The Sic composite consists of silicon and carbon in the stochiometric proportion and small oxygen contaminations (< 3 at.%). No further impurities could be detected. ‘Ihe energetic positions of the Si 2p and the C 1s photoelectron peaks at 100.3 eV and 283.2 eV binding energy

64

H. Plank et al./Nucl. Instr. and Meth. in Phys. Res. B III (1996) 63-69

Fig. 1. Scanning electron microscopy images of surface topogmphy of SiC/SiC after mechanically cutting and cleaning in an ultrasonic bath (before D+ bombardment). (a) Low magnikation, showing the fibre structure; (b) large magnification, showing the surface topography of a fibre.

In the first chamber the sputtering yield was determined from the weight loss of the target due to Df ion bombardment at normal incidence. The chamber was equipped with a Mettler 22 vacuum-microbalance having an absolute accuracy of + 1 pg. Additionally, reaction products during D+ bombardment were analysed with a quadrupole mass spectrometer which was separated from the target chamber by a liquid nitrogen cooled tube aperture directed onto the beam spot of the target. The target was heated by electron bombardment from the rear of the sample. The temperature was measured with an infrared pyrometer. Before performing the erosion experiments the samples were annealed in situ up to 1100°C to avoid weight loss due to outgassing during the irradiation. The second chamber was used to determine the surface composition of Sic targets during D+ ion bombardment by means of Auger Electron Spectroscopy (AES) with a four grid retarding field analyser (RFA) [9]. Ion beams which were extracted from the ion source and deflected into chamber II passed through the centre of the RFA and struck the target at normal incidence. 3 keV primary electrons were directed to the samples at an angle of incidence of 73” to the surface normal. The electron beam current was 10 ALA. The C KLL and the Si LVV Auger signals were monitored using Lock-In-technique with a modulation voltage of 2 V peak-to-peak. The energy resolution AE/E was about 1% in the energy range between 60 and 1000 eV which could be recorded with the RFA. The alternating performances of ion bombardment and AES analysis were automatically controlled. In order to investigate temperature dependent effects changing the surface composition the target holder was heated ohmitally. Target temperatures ranging from 300 to 1000 K were measured with a thermocouple.

3. Results and discussion show that the material exists predominantly in the form of Sic [6]. For energy calibration the Au 4f,,, peak was used as a reference with a binding energy of 84.0 eV [7]. Fig. 1 shotis scanning electron microscopy images clearly demonstrating the fibre structure of the material (low magnification, Fig. la) as well as the initial surface shape and roughness (large magnification, Fig. lb). The images are taken’before bombardment.

3.1. Erosion behaviour The sputtering yields Y were determined from the weight loss Am of SiC/SiC targets after D+ ion bombardment at normal incidence according to the equation AmNo Y=-.

2.2. Experimental

setup

The erosion experiments were performed on the high current ion source described elsewhere [8]. Here only some important features will be described. A 3 keV Dl ion beam was extracted from the ion source, then magnetically mass analysed and decelerated to the required energy by biasing the target. Typical beam currents for 3 keV Dl were 120 IJ.A equivalent to a flux of 7 X 1015 D+/cm*s. Two different chambers were used for the measurements.

WN, Here N,, is Avogadro’s number, N, the number of incoming ions and M2 the average atomic target mass. Fig. 2 shows the results for the investigated Sic composite material (filled symbols). Sputtering yields from different SIC samples (single crystalline, plasma sprayed) obtained from earlier measurements by Bohdansky et al. [lo] are plotted for comparison (open symbols). There is no significant difference between the old data for Sic and the present ones for Sic/Sic. The different structure and the

H. Plank et al./Nucl.

Instr. and Meth. in Phys. Res. B 111 (1996) 63-69

0.15 -

Bohdansky fit

I

D*

-

65

Sic/Sic

-30eV D’ _______50 eV D+ 150 eV D+ ----

1 keV D+

I

o.oo-’ 400

600

800

TEMPERATURE

1000

(K)

Fig. 3. Temperature dependence of the CD, production yield

ENERGY (keV)

Fig. 2. Sputtering yields of SiC for D+ bombardment as a function of ion energy (0 [IO], n Sic/Sic, this work). The yields above the threshold energy of 29.4 eV are fitted with the revised Bohdansky formula (solid line) [ 111.

during bombardment of SiC/SiC with D+ ions of various energies. The data are normalized to the CD, yield of graphite bombarded with 1 keV D+ at 850 K ( = 1).

The threshold energy of Si sputtering by deuterium bombardment is 20.9 eV while for carbon sputtering there is no threshold due to chemical effects [8]. In the D+ energy range below 100 eV chemical sputtering becomes the dominant erosion mechanism which means the release of volatile products due to chemical reactions between the incident deuterium ions and the Sic target material. In our experiments only CD, could be detected, no SiD, (x = 1,2,3,4) was seen. The temperature dependence of the methane production during deuterium bombardment with a flux of 3 X lOi* D+/cm*s is shown in Fig. 3 for D+ energies of 30 eV, 50 eV, 150 eV and 1 keV. The CD, signal is normalized to the signal as obtained from 1 keV D+ bombardment onto pyrolytic graphite at 850 K which corresponds to a ratio of CD,/D+= 0.1. For lower D+ ion energies (30 eV, 50 eV) the chemical erosion of C from SiC/SiC shows a maximum around 500 K. The CD, yield is lower by one order of magnitude with respect to pure graphite [12]. With increasing D+ ion energy the CD, production from the Sic composite decreases drastically. A small maximum can be seen around 900 K as it is known for graphite [3,13]. We cannot exclude that there are some small graphitic grains embedded in the Sic matrix. The penetration depth of Df into Sic was calculated for various energies up to 150 eV with the TRIDYN ([ 141) tering.

degree of surface roughness of SiC/SiC have only minor influence to the sputtering yields. The yields above a threshold of Efh = 29.4 eV are fitted using the revised Bohdansky formula [l l] (solid line in Fig. 2) Y(Eo)=Qsn~(E)

[ l-

(2r3][l-$r.

(2)

with E=-

Eo

and

% E,,(eV)

Ml +M*

= 30.74-----

z,z,(

2:‘s + z;/‘)“*,

(3)

M2

and

Eth =

Es y(l-Y)

4M,M, wi*

7 =

(M,

+

M2)2

*

(4)

Here E, is the projectile energy, s,“~(E) the nuclear stopping cross section based on the Kr-C potential and E the reduced energy. M, and Z, are the mass and the nuclear charge of the projectile, M, and Z, are the average atomic target mass and nuclear charge, respectively. Q is a fitting parameter and determines the maximum of the yield curve. Eth is calculated according to Eq. (4) using the heat of sublimation Es for Sic. Eq. (2) considers only physical sputtering which means the erosion by knocking off surface atoms via momentum transfer from the incoming ions to surface atoms. There is a marked difference at deuterium energies below 100 eV between the measured sputtering yields of SiC/SiC and the values obtained due to physical sputtering (Bohdansky fit). The measured data do not decrease towards a threshold energy as expected for physical sput-

Table 1 Mean penetration depth (x) and &andard deviation cr of D+ ions into SIC for various energies. The values are calculated with the computer program TRIDYN [14] D+ energy [eV]

(x>

[Al

u [Al

10

20

5.1

8.1

2.8

4.5

30

50

150

11

15

36

6.0

8.8

20

H. Plank et al./Nucl.

66

Instr. and Meth. in Phys.Res. B 111 (1996) 63-69

program described in Ref. [15]. The values are listed in Table 1. Due to the small penetration depth of low energy D+ ions and the energy dependence of methane release during deuterium bombardment the methane formation mainly occurs in the topmost atomic layers. Therefore and because of the threshold energies for Si and C it is expected that at low deuterium energies there is a depletion of carbon at the surface or in the subsurface of the material as observed previously for boron doped graphites [91. 3.2. Surface composition modi$cations The surface concentrations of carbon and silicon were determined by AES. Quantitative analysis was done by normalizing the Auger signals of the sample to signals of pure elements thus determining the relative elemental sensitivity factor. The surface concentration was calculated according to [ 161

and c,+c,=l.

(6)

Here’ c, is the surface concentration of element A, IA and c are the Auger signals of element A in an alloy AB and in pure A, respectively. The relative elemental sensitivity factor SR was calculated to be 8.5 from the Auger signals of pure silicon and graphite which are shown in Fig. 4. Measurements on pure Sic after cleaning by Hebombardment (shown also in Fig. 4) yield 48 at.% Si and 52 at.% C. Therefore, a further correction factor considerjng matrix effects was not taken into account.

Table 2 Inelastic mean free pathlength A of electrons [17,18], density p and mean atomic weight M in Si and Sic Matrix

h [A]

Si Sic

Si LVV (90 eV)

C KLL (272 eV)

5.1 4.5

9.5 7.5

M

2.3 3.2

28 20

Changes of the sample composition due to ion bombardment may form an altered layer with a thickness about two times the mean penetration depth of D+ ions at energies below 50 eV [14]. In this energy range the penetration depth of Dt ions (see Table 1) becomes comparable with or less than the escape depth of C KLL and Si LVV Auger electrons in Sic which are given in Table 2 according to Tanuma [17,1 S]. Therefore, a correction of measured surface concentrations is necessary. The additional use of the Si KLL Auger transition at 1619 eV kinetic energy was not possible in our experiment because of the limited energy range of the RFA (see Section 2). After bombardment the Sic samples are considered as a SIC substrate fully covered by a homogenous overlayer. The Auger signal of a substrate AB covered by a thin film of thickness d containing A and B is given by the sum of unattenuated signal from the thin film and an attenuated part of the substrate [ 161: Z A,(

E,)cos

0

dz

d h,( E,)cos

0

(7)

’

Here IA and ZL denote the measured signal of element A with and without overlayer (only substrate), respectively. A, is the inelastic mean free path of an electron with kinetic energy E, in the overlayer, 6 is the angle of emission to the surface normal. NT is the atom density of A in matrix m (m = t means the overlayer, m = s the substrate), given by Np = ( a,“)-3cz

4--

p&cm31

and p,N(

aAm)3= Mm,

with a2 the atomic volume and CA”the concentration of A in matrix m, respectively. P,,, is the density and M, the mean atomic weight of matrix m. N is Avogadro’s number. The geometry of a retarding field analyser requires a integration over the solid angle 0 of the detector. Thus from Eqs. (_5), (7) and (8) the measured ratio of c, and ca is given by 240

260

280

300

KINETIC ENERGY (&)

Fig. 4. Si LVV and C KLL Auger signals in SiC/SiC, pure silicon and graphite, respectively. The spectra are taken after cleaning the samples by He bombardment. The C KLL signal is magnified by a factor of 3.3.

‘A -=‘B

‘i

1 - L’A)cosBda+ “AAA/a(

”

..AB~~(l-~a~cos~d~+~~,cossdu.

1 L;cos8dR (9

67

H. Plank et al./Nucl. Instr. and Meth. in Phys. Res. B 111 (1996) 63-69 1 keV. This may be caused segregation of Si atoms [9].

at

by radiation

enhanced

3.3. Component sputtering yields

iT, I

0

20

40 FLUENCE

60 [ IO”

80 D+/cm’

1

t

100

]

Fig. 5. Si surface concentrations of Sic/Sic during 20 eV D+ ion bombardment as a function of fluence at various target temperatures. The concentrations then corrected by Eq. (9).

are calculated

by Eq. (5) and

with

From the surface concentration the sputtering yield ratio of the two components in a binary alloy can be calculated. The partial sputtering yield Y,P of component A in an AB compound is the average number of sputtered atoms A per incident ion while the component sputtering yield YL is the quotient of partial sputtering yield and surface concentration cJ, of element A, YL = YAP/c:. The surface concentration is used in calculating the component sputtering yields since at low ion energy the sputtered atoms originate from the topmost atomic layers. As in steady state the partial sputtering yields must reflect the bulk concentrations, the surface concentration can be calculated from the ratio of component sputtering yields by

WI YA_-- ‘k

and

‘A

ysc-

(10) The density pr, the mean atomic weight M, and the inelastic mean free path A, in the overlayer depend on the unknown overlayer composition CA. We assume a linear dependence of pt. Mr and A, from the Si concentration in the overlayer with the boundary values of Sic and pure Si (see Table 2). Then ca can be calculated solving Eq. (9) numerically. For the lowest ion energies, the correction of the surface concentrations amounts to about 10 at.%. In Fig. 5 corrected Si surface concentrations of Sic, cdlculated by Eqs. (51, (6) and (9), during bombardment with 20 eV D+ ions at different temperatures are shown as a ~function of fluence. Obviously the enrichment of silicon at the surface depends strongly on target temperature. At 300 K about 90 at.% silicon were observed, increasing to 190 at.% at 500 K and then decreasing to 75 at.% at 800 K and to 55 at.% - nearly the bulk concentration - at 1000

(11)

. 4

%

Here c, denotes the bulk concentration of element A. We assume in a first approximation that the component sputtering yield of element A in AB is equal to the sputtering yield of pure element A. Furthermore, it is assumed that the concentrations are constant over the depths which contribute to the sputtered flux. Then Eq. (11) can be used to calculate the expected “theoretical” surface concentrations in an alloy and to compare them with actually measured values. The surface concentrations determined by AES are corrected by the numerical procedure described in Section 3.2 taking into account the weighing of different layers for the Auger electrons and

-A--

150&D’

K. The steady state surface concentration also depends on D” ion energy. In Fig. 6 Si concentrations are plotted vs. temperature for various ion energies. At 20 eV a pronounced Si concentration maximum around 500 K is observed. Increasing the ion energy the Si surface concentration decreases and approaches the bulk value. The measured surface concentrations shown in Fig. 6 reflect the energy and temperature dependence of the CD, production yield as seen in Fig. 3. Increasing the ion energy and/or the sample temperature the chemical erosion, i.e. the sputtering yield of carbon, decreases. Hence the Si surface concentration decreases. Si concentrations slightly enhanced with respect to the bulk composition of Sic are observed for D+ ion energies

Df -

SiClSiC

TEMPERATURE

(K)

Fig. 6. Steady state Si surface concentrations of Sic/Sic during D+ ion bombardment as a function of temperature for various ion energies. The concentrations are calculated by Eq. (5) and for Df ion energies below 50 eV corrected by Eq. (9).

68

H. Plank et al./ Nucl. Instr. and Meth. in Phys. Res. B 111 (1996) 63-69

D+

8 ii5

-

sic/sic

... ...

500 K

-.------

IOOOK calculated.

them.

tabulated,

Phys.

0,4_ “---“--“--‘-------_________..___~______ physical erOSiOn

0.2

I:. 0

I 200

I 400 ENERGY

I 600

1 800

I 1000

(eV)

Fig. 7. Steady state Si surface concentrations of SiC/SiC as a function of D+ ion energy obtained by experimental measured values at 500 and 1000 K (square and circle symbols) and by calculated component sputtering yields according to Eq. (11). The solid line is based on measured yields, the dotted line on calculated yields using Eq. (2).

for sputtering. The results for Sic/Sic are plotted in Fig. 7. Filled circles indicate measured Si concentrations at 500 K target temperature and filled squares at 1000 K. Also, Si surface concentrations calculated by using Eq. (11) are shown obtained from measured sputtering yields of pure graphite and pure silicon, which are taken from [S] (straight line - “chemical erosion”), and obtained from sputtering yields of pure elements calculated using Eq. (2) (Bohdansky’s formula, dotted line - “physical erosion”). The former includes chemical erosion, while Bohdansky’s formula represents only the case of physical sputtering, i.e. any chemical effects between the incoming ions and target atoms which could enhance the sputtering yield are not taken into account. In this case the threshold energy of graphite (24.3 eV) is higher than for silicon (20.9 eV). Thus the sputtering ratio Yc/Ysi and with it the Si surface concentration cf is zero for ion energies below the threshold energy of graphite. However, ratios obtained experimentally for pure elemental samples head to infinity due to chemical erosion of graphite with no threshold energy and no sputtering of silicon below 20.9 eV. This results in a surface depletion of carbon arriving at a Si concentration of c& = 1. At 500 K target temperature the measured surface concentration approaches the values calculated from the experimental sputtering yield of pure elemental samples (“chemical erosion”), while for 1000 K sputtering at low ion energies the measured values are more like results obtained from Bohdansky’s formula (“physical erosion”). Thus, at 500 K chemical erosion of carbon seems to be responsible for the high silicon concentration at the surface.

The depth of the maximum energy deposition for ion energies below 100 eV is similar to the mean free path of the penetrating ions. The deposited energy is concentrated in the topmost atomic layers. An amorphised surface develops with many broken bonds which represent binding the chemical sites for D+ ions. At high temperatures erosion of carbon is suppressed and no carbon depletion occurs. Two processes may be responsible for lowering the chemical erosion of carbon atoms in Sic. Firstly, broken bonds will heal more effectively at high temperatures reducing the number of binding sites for D atoms. Secondly, as reported for C:H films [20], the rate of thermal decomposition of radical hydrocarbon centres is large at high temperatures. The radical centres acting as erosion relevant groups are destroyed by H-split-off reactions. This reaction scheme may be enhanced in SIC due to the presence of Si. Thus, chemical erosion of carbon is reduced and the component sputtering yield ratio approaches the Bohdansky-fit based on physical sputtering only.

4. Conclusion Different mechanisms depending on ion energy and target temperature yield different surface modifications of Sic during bombardment with D+ ions. For low ion energies below 100 eV the CD, release during sputtering shows a maximum at 500 K leading to a silicon rich surface layer. At high temperatures above 850 K physical sputtering dominates the erosion mechanism. The CD, release is strongly reduced and the silicon surface concentration approaches the bulk concentration independent on ion energy. Preferential sputtering, i.e. preferred sputtering of one component due to mass or surface binding energy differences may be responsible for slightly enhanced silicon surface concentration. For ion energies above 100 eV the CD, release is very small regardless of target temperature. Only a modestly enhanced silicon surface concentration at room temperature is observed decreasing towards the bulk level at temperatures of about 1000 K. No silicon concentration maximum can be observed at 500 K. Using appropriate multi-component materials as first wall armors in fusion devices the outermost atomic layers may differ from bulk composition under bombardment with hydrogen ions. There is a potential in selecting and designing first wall materials in such a way that the surface has favourable properties in view of sputtering, while the bulk has favourable properties in view of thermomechanical and electrical quality.

Acknowledgements It is a pleasure to thank W. Ottenberger cal support and assistance.

for his techni-

H. Plank et al./Nucl.

Instr. and Meth. in Phys. Res. B III (1996) 63-69

References [I] H. Bolt, Fusion Eng. Des. 22 (1993). 85. [2] J. Roth, J. Nucl. Mater. 176&177 (1990) 132. [3] C. Garcia-Rosales, J. Roth and R. Behrisch, J. Nucl. Mater. 212-215 (1994) 1211. [4] H. Strcckert, private communication; General Atomics, General Atomics Court, PO Box 85608, San Diego, CA 921869784. [5] F. Bozso, L. Muehlhoff, M. Trenary, W.J. Choyke and J.T. Yates Jr., J. Vat. Sci. Technol. A 2 (1984) 1271. [6] L. Muehlhoff, W.J. Choyke, M.J. Bozack and J.T. Yates Jr., J. Appl. Phys. 60 (8) 2842 (1986). [7] J.F. Moulder, W.F. Stickle, P.E. Sobol and K.D. Bomben, Handbook of X-Ray Photoelectron Spectroscopy, Perkin Elmer Corporation, Physical Electronics Division ( 1992). [8] W. Eckstein, C. Garcia-Rosales, J. Roth and W. Ottenberger, Sputtering Data, Max-Planck Institut I% Plasmaphysik, IPP Report 9/83 (1993). [9] R. SchwGrer and J. Roth, J. Appl. Phys 77(8) 3812 (1995). [lo] J. Bohdansky, H.L. Bay and W. Ottenberger, J. Nucl. Mater. 76&77 (1978) 163. [l l] C. Garcia-Rosales, W. E&stein and J. Roth, J. Nucl. Mater. 218 (1994) 8.

69

[ 121 J. Roth, in: Sputtering by Particle Bombardment II, Topics in Applied Physics, Vol. 52, ed. R. Behrisch (Springer, Berlin 1983) p. 91. [ 131 C. Garcia-Rosales and J. Roth, J. Nucl. Mater. 196- 198 (1992) 573. [14] W. Eckstein, unpublished results. [15] W. MBller, W. Eckstein and J.P. Biersack, Comp. Phys. Comm. 51 (1988) 355. [16] M.P. Seah, in: Practical Surface Analysis, 2nd Ed., Vol. 1, eds. J. Briggs and M.P. Seah (Wiley, Chichester, New York, Brisbane, Toronto, Singapore, 1990; Salle + Sauerlander, Aarau, Frankfurt am Main, Salzburg, 1990) p. 210. [ 171 S. Tanuma, C.J. Powell and D.R. Penn, Surf. Interface Anal. 17 (1991) 911. [ 181 S. Tanuma, C.J. Powell and D.R. Penn, Surf. Interface Anal. 17 (1991) 927. [19] G. Betz and G.K. Wehner, in: Sputtering by Particle Bombardment II, Topics in Applied Physics, Vol. 52, ed. R. Behrisch (Springer, Berlin, 1983) p. 11. [20] A. Horn, A. Schenk, J. Biener, B. Winter, C. Lutterloh, M. Wittmamr and J. Kiippers, Chem. Phys. Lett. 231 (1994) 193.