Chemical Vapour Deposition Of Silicon Carbide By Pyrolysis Of Methylchlorosilanes

JOURNAL OF MATERIALS SCIENCE LETTERS 16 (1997) 33—36

Chemical vapour deposition of silicon carbide by pyrolysis of methylchlorosilanes BYUNG JIN CHOI Department of Welding Engineering, Taegu Junior Health College, Taejeun-dong san 7, Pook-gu, Taegu 702-260, Korea

DONG WON PARK, DAI RYONG KIM

Department of Metallurgical Engineering, Kyungpook National University, Sankyuk-dong 1370, Pook-gu, Taegu 702-701, Korea

Silicon carbide (SiC) prepared by chemical vapour deposition (CVD) remains of importance for both structural [1, 2] and electronic [3,4] applications. A variety of gaseous precursors has been used for SiC deposition under widely varying conditions of input gas composition, temperature and pressure [5]. For

DDS, TCS or TS as a reactant and to investigate

CVD of SiC, in general, a carrier gas is bubbled through a silicon containing liquid mixed with

imate the temperature of substrate in accordance with a previous calibration with a thermocouple.

another stream of carrier gas. The necessary carbon is either contained in the chlorosilane or supplied by introducing a hydrocarbon. Methylchlorosilane (MTS: CH3SiCI3), as a pre-

cursor, is one of the most useful of the available chlorosilanes because it contains silicon and carbon in stoichiometric proportions [6]. Accordingly, it has been expected to give stoichiometric SiC deposition. Recently, from thermodynamic studies on the CVD

of SiC in the MTS + H2 gas system [7, 8], it has been shown that SiC is the only stable solid phase present in a wide range of temperature, pressure and concentration of the reactant. In experimental works,

however, there have been many reports in the literature [9, 10] showing that excess Si is codeposited with SiC at lower temperature. Furthermore, in

our previous work [11] conducted with the MIS + H2 gas system in the temperature range l000—1500 °C, we always found excess Si deposition

at temperatures below 1400 °C. In order to obtain stoichiometric SiC deposition, we recently reported [12] the addition of propane (C3H8) as an excess carbon source to the MTS + H2 gas system, which resulted in the deposition of stoichiometric SiC. Therefore, it is conceivable that silanes are more reactive with the substrate than hydrocarbon and supplying excess carbon is needed for stoichiometric

SiC deposition. There are some precursors that supply excess carbon, such as dimethyldichlorosilane (DDS: (CH3 )2 SiC12), trimethylchlorosilane (TCS: (CH3)3SiCl) and tetramethylsilane (TS: (CH3)4Si). These chlrorosilanes contain both silicon and carbon,

and the ratios of C/Si in the molecules are 2, 3 and

4, respectively. In addition DDS, TCS and TS molecules decompose more easily than MIS and supply sufficient hydrocarbon above the .substrate. The purpose of the work reported here was thus to

prepare stoichiometric SiC deposit by supplying 0261-8028

1997 Chapman & Hall

the change of the microstructure. The SiC coatings were deposited in a horizontal quartz reactor at 1000— 1500 °C and under atmospheric pressure. The temperatures were measured with an optical pyrometer and corrected to approxGraphite plates and SiC-coated graphite were used as substrate and susceptor, respectively. The three types of methylchlorosilane (DDS, TCS and TS) were used

and they were carried by hydrogen from the evaporator maintained at 0 °C. The concentration of methylchlorosilane was controlled with hydrogen. The total flow rate of X (X = DDS, TCS or TS) + H2 gas mixture was kept constant at 1600 standard cubic centimetres per minute (sccm). The growth rate was estimated by measuring the weight increase during deposition periods. The crystal structure was

analysed by X-ray diffractometry (XRD) and the silicon content of the coating layer was determined by energy dispersive spectrometry (EDS) and Auger electron spectroscopy (AES). The surface morphol-

ogy of the coating layer was investigated by scanning electron microscopy (SEM). Details of the

experimental procedure were similar to those reported previously [11, 12]. The molar fraction dependence of the deposition rate is shown in Fig. 1. The results of previous work [11], carried out in the MTS + H2 system, are also

shown for comparison. The total flow rate of methylsilane was 1600 sccm and the temperature of

the graphite substrate was 1300 °C. As shown in Fig. 1, the deposition rate increased linearly with increasing reactant concentration. Hunt and Stirl [13] have shown that the growth rate r is determined by the relationship expressed as r = 7E Po TF, where 11E, Po and TF are deposition yields in experimental work, partial pressure of reactants and total system

pressure, respectively. Thus, increase of the molar fraction (X/(X + H2), X = MTS, DDS, TCS or TS) leads to linear increase in the growth rate. By using DDS (C/Si = 2), a higher growth rate was achieved, while it decreased with TCS or TS. The temperature dependence of the deposition rate

is shown in Fig. 2. In the MTS + H2 system, as is

33

250

always found the formation of SiC powder in front

of the susceptor while using TCS and TS as reactants. This can be rationalized on the basis of

200

the fact that the thermal decomposition temperatures

of the methylchlorosilanes decrease in the order

E

(2

C)

MTS > DDS > TCS > TS. Typical XRD patterns of the layer prepared with

150

DDS are shown in Fig. 3. The samples were

E

a)

C

0

deposited for 20 mm and the molar ratio of DDS was 0.01. As shown in Fig. 3, (111), (220) and

100

(3 11) peaks of fl-SiC, belonging to the cubic

0 0 a)

a

system, were observed. The diffraction pattern of the layer deposited at 1100°C is relatively broad,

50

0—

0.000

indicating micro-crystalline /3-SiC. With increasing substrate temperature, the diffraction pattern became 0.005

0.010

0.015

0.020

0025

Molar fraction (X/(X + H2)) Figure / Deposition rate as a function of molar fraction of reactants: substrate temperature 1300°C. (---), MTS [111; (•), DDS; (A), TS;

(.),

100

are somewhat different to the results of DDS (Fig. 3). As shown in Fig. 4, the diffraction pattern of the deposit prepared at 1100 °C using TCS shows

sharp and narrow peaks compared to Fig. 3. This means that the deposits were well crystallized at lower temperature. Relatively weak carbon and

7 6

5

E

II)

8 C

oa)

(a

the deposits obtained at 1500 °C, showing the

existence of free carbon in the deposits. Figs 4 and 5 show the diffraction patterns of the deposits obtained with TCS and TS. These patterns

8

0 (I) 0 a.

increased. There was no evidence of excess silicon deposit. Graphite peaks of (0 02) were detected in

TCS.

9

0)

sharp and the intensity of the diffraction lines

3

5.5

6.0

6.5

7.0

7.5

8.0

Reciprocal temperature (1 04/T, K1)

(b (c 80

I

I

70

60

50

I

I

I

40

30

20

20 (degrees)

Figure 2 Temperature dependence of deposition rate; molar fraction of reactants 0.01. (- --), MTS [11]; (I), DDS; (A), TS; (.), TCS.

Figure 3 XRD pattems of the deposits obtained in the DDS + H2

system at (a) 1100, (b) 1300 and (c) 1500 °C. DDS molar fraction:

the case with CVD, the temperature dependence was

divided into two linear regions, following the Arrhenius-type relationship. With DDS, TCS and TS, however, the deposition rate reached a maximum and further increase of the temperature decreased the

deposition rate. The maximum temperatures were 1400°C (DDS), 1100 °C (TCS) and 1000 °C (TS). Decrease of the deposition rate in the high temperature region is often reported. Cheng et al. [14], with the MTS + H2 system, showed that the maximum deposition rate occurred at 1300—1400 °C, and they explained that the decrease in the deposition rate at

high temperature was due to etching effects on the substrate surface. In our present investigation, the decrease of the deposition rate at higher temperature can be explained by etching effects in the case of DDS, while the decrease of the growth rate above 1100 °C in TCS or 1000 °C in TS is thought to be caused by gas phase reaction above the substrate. We

34

0.01; deposition time: 20 mm. SiC: (V), (Ill); (s), (220); (0), (311). C: (V), (002); (•), (102); (•), (110); (•), (004).

(a)——-—--————_—-————---——--'-—-—-—"

I

I

I

80

70

60

I

50 40 20 (degrees)

30

20

Figure 4 XRD patterns of the deposits obtained in the TCS + H2 systcm at (a) 1100, (b) 1300 and (c) 1500 °C. TCS molar fraction: 0.01;

deposition time: 20 mm. SiC: (7), (111); (LI), (220); (0), (311). C:

(V) (002); (•), (I 02); (•), (004).

(a a(\J

Co

here. TCS and TS also gave near stoichiometric SiC

deposition at ll00—1300°C. A large amount of excess carbon was produced above 1400 °C by using TCS and TS as a precursors.

As can be seen in the XRD patterns (Figs 3—5)

and EDS analysis, it is evident that chemical composition is dependent on temperature and the

of the deposits obtained in the TS + H system at (a) 1200. )b) 1300 and (C) i500 C. IS molar

C/Si ratio of the reactant i.e. the tendency of carbon codeposition is stronger when the substrate temperature and the C/Si ratio of reactants are higher. By controlling these two factors properly, stoichiometric SiC deposition can be obtained. The surface appearance was investigated by SEM.

fraction: 0.01: deposition time: 20 miii.

Fig. 7 compares the SEM micrographs of the as-

strong /3-SiC peaks were detected in the deposits

deposited SiC surfaces obtained by the pyrolysis of DDS, TCS and TS. The substrate temperature was 1300 °C, the molar fraction of the reactants was 0.01 and the deposition time was 20 mm. The SiC surface deposited by the pyrolysis of DDS showed cellular

80

70

60

50

40

30

20

20 (degrees) Figui'e 5 X—ray ditilaction patterns

obtained at 1300 °C. The 13-SiC peaks were not seen at 1500 Of it is thought that the carbon peaks in Fig.

4c correspond to graphite substrate. We could find neither SiC nor carbon deposition at 1500 °C in the TCS + H2 system. As can be seen in Fig. 5, carbon peaks were always detected with /3-SiC peaks in deposits obtained with TS regardless of the substrate temperature, and the carbon peaks become stronger with increasing substrate temperature. Therefire, these results indicate that the deposition of carbon is dependent on the substrate temperature and the C/ Si ratio of the reactants. The chemical compositions of the deposits are

structure (Fig. 7a). It is supposed that a large

shown in Fig. 6. The chemical composition was investigated by AES and EDS. In EDS analysis, only the silicon contents of the samples were determined. From the previous work, MTS yielded a considerable

amount of excess silicon at temperatures below 1400 °C; with increasing temperature, the amount of

excess silicon decreased gradually, and stoichiometric /3-SiC could be obtained only at 1500 0C. DDS, however, produced almost stoichiometric SiC at temperatures below 1400 °C in the work reported 100

80

in

60

C 5)

C

0 C)

C

0 40 0 Cl)

20

0— 900

1000 1100 1200 1300 1400 1500 1600 Deposition temperature (°C)

Fl gui-c 6 Si icon content as a function of substrate tenipelatuic: molar fraction of reactant: 0.01. )———). MIS [Ii]: )•), DDS: (A). TS: (.).

Figws 7 SEM micrographs of surface deposited at 300 C by using (a) DDS. )b( T(S and (c) TS. Molar fraction of reactants: 0.1)1:

'CS.

deposition time: 20 miii.

35

percentage of the cell boundary is composed of carbon, which hindered the further growth of SiC

2. K. MIYOSHI, D. H. BUCKLEY and M. SRINIVASAN,

crystal. The surface obtained with TCS (Fig. 7b), on the other hand, showed small worm-like structure.

3. W. VON MUENCH and E. PETTENPAUL, .1. App!. Phys. 48 (1977) 4823. 4. J. KENDALL, .1 Che,n. Phys. 21(1953) 821. 5. J. 5CHLICHTING, Powder Metal!. tnt. 12 (1980) 141.

The surface obtained by TS showed needle-like structure and the crystals were smaller than those of

TCS, as shown in Fig. 7c. These changes in microstructure are thought to be related to the carbon content in the deposits, i.e. the crystal size decreases with increasing C/Si ratio of the reactants because the excess carbon blocks the growth of the SiC crystal.

Ceram.

Bull. 62

(1983) 495.

6. R. J. PRICE, Nod. Technol. 35 (1977) 320. 7. A. I. KINGON, L. J. LUTZ and R. F. DAVIS, I. Amer Cera,n.

Soc. 66

(1983) 558.

8. G. S. FISCHMAN and W. 1. PETUSKEY, ibid. 68 (1985) 185.

9. K. MINATO and K. FUKUDA, I. Mater Sci. 23 (1988) 699.

10. J. CHIN and T. OHKAWA, Nuci. Technol. 32 (1977) 115. 11. B. J. CHOI and D. R. KIM, .1 Mate,: Sci. Lett. 10(1991) 860.

Acknowledgement This

paper was supported by the Non-Directed

Research Fund, Korea Research Foundation.

12. B. J. CHOI, S. H. JEUN and D. R. KIM, I. Em: Ceram. Soc.

9 (1992) 357.

13. L. P. HUNT and E. STIRL, .1 Electrochem. Soc. 135 (1988) 206.

14. D. J. CHENG, W. J. SHYY, D. H. KUO and M. H. HON, ibid. 134 (1987) 3145.

References 1.

36

M. K. BRUN and M. P. BOROM, I Ame,: Ceram. Soc. 72 (1989) 1993.

Received 23 April and accepted 1 August 1996