Creator:J. L. Margrave,M. P. D'Evelyn,R. H. Hange Date Created:October 23, 1991 Place Created: Keywords:diamond growth,carbon atoms Context:Mechanism of diamond growth by chemical vapor deposition: Carbon-13 studies ************************************************** Mechanism of diamond growth by chemical vapor deposition: Carbon-13 studies M. P. D'Evelyn, C. J. Chu,a) R. H. Hange,a) and J. L. Margrave3' (Received 23 September 1991; accepted for publication 23 October 1991) Previous 13CH4/12C2H2 isotopic competition experiments on the mechanism of diamond growth by chemical vapor deposition are reanalyzed in light of recent evidence for a nonlinear dependence of the first-order Raman shift frequency on 13C mole fraction. The new Raman data imply a l3C mole fraction for mixed-isotope diamond films several percent higher than that reported previously. The corrected carbon-13 mole fractions of polycrystalline diamond films and homoepitaxial films grown on (100), (111), and (110) natural diamond substrates were each equal, within experimental error, to that of the methane above the substrate but significantly different from that of gas-phase acetylene. As the 13C mole fractions of methyl radical and methane should be nearly identical, the methyl radical is concluded to be the predominant growth precursor regardless of the crystallographic orientation of the diamond substrate. The growth of diamond films at low pressure by chemical vapor deposition (CVD) has generated tremendous interest within the past five years,1 and much recent effort has been focused on determining fundamental aspects of the growth mechanism. The identity of the gas phase precursor primarily responsible for diamond growth has been highly controversial, and both methyl radical2,3(e> and acetylene3 mechanisms have been proposed. We recently obtained the first direct evidence that the methyl radical is the primary growth precursor in hot-filament CVD of polycrystalline diamond films4 and homoepitaxial films grown on (100), (111), and (110) natural diamond substrates5 through isotopic competition experiments involving growth from a mixture of 13CH4 and 12C2H2. Martin and co-workers6 have shown that diamond grows more readily when CH4 is injected into a plasma-generated stream of hydrogen atoms than when acetylene is injected, indicating that the methyl radical is a more effective growth precursor than acetylene. Johnson et al.1 have very recently grown polycrystalline diamond films from a mixture of l3CH4 and ^QHj by microwave plasma CVD, avoiding complete isotopic scrambling by operating at high flow rates. They found that the 13C mole fraction of diamond was only slightly below that of methane, indicating that ~90% of the carbon atoms in the diamond resulted from species derived from methane (presumably methyl radicals). Indirect evidence suggesting growth predominantly by the methyl radical has been obtained by several groups,8 while indirect evidence suggesting growth predominantly by acetylene or from a combination of both precursors has also been seen.9 Our experiments, carried out under conditions where complete isotopic scrambling between 13CH4 and 12C2H2 did not occur, involved comparing the isotopic composition of gases collected immediately adjacent to the substrate to that of the diamond films. The experimental ap- paratus and growth conditions have been described in detail.4 5 The isotopic composition of the diamond films was derived from the first-order Raman shift frequency, which we determined4 to be approximately linear in 13C mole fraction. However, recent experiments by Hass et a/.,10 Banholzer et al.11 and by Nisida and Kanda12 have shown that the dependence of Raman frequency on 13C mole fraction has a significant downward curvature. The new results are shown in Fig. 1, together with a cubic polynomial fit to the data of Banholzer et al. Earlier Raman data of Chrenko13 showed a similar effect but the l3C compositions were subsequently found to be slightly inaccurate.14 The cubic fit is seen to fit all the data rather well. Hass et al.10 have shown that the curvature results from isotopic disorder in diamond as a consequence of the nonsymmetric way that 12C and 13C couple to lattice pho-nons. It is not entirely clear why the curvature in the dependence of Raman shift on mole fraction was not evident in our CVD-grown polycrystalline films.4 The apparent linear dependence was discussed4 in a framework equivalent to the virtual crystal approximation described by Hass et al.10 The relative flow rates of 13CH4 and 12CH4 were determined in our initial work using factory-calibrated mass flowmeters but the isotopic composition of the gas phase was not independently verified. As the l3CH4/'2CH4 studies were the first growth experiments we performed, the error bars on either the Raman frequency or mole fraction may have been larger than was thought, or perhaps impurities or stress were present in the films. As discussed previously,5 our analysis presumes that the incorporation rates of 12CH3 and 13CH3 (or the carbon isotopes of C2H2) into the growing diamond film are identical. Derjaguin and Fedoseev15 have claimed that carbon-13 is preferentially incorporated into diamond, but recent preliminary results of Banholzer16 indicate no significant 13C/12C kinetic isotope effect. We have reanalyzed the Raman data for our mixed-isotope diamond films4'5 using the cubic polynomial fit shown in Fig. 1. The results are summarized in Table I. 1 February 1992 FIG. 1. Dependence of first-order Raman shift of diamond on carbon-13 mole fraction. Data are shown from Banholzer el at. (see Ref. 11) and Nisida and Kanda (see Ref. 12). Also included in Table I are the 13C mole fraction of methane and acetylene determined previously4'5 by matrix isolation infrared spectroscopy after collection via a probe located immediately adjacent to the substrate. The 13C mole fraction of the diamond films, polycrystalline and homoepitaxial films grown on (100), (111), and (110) substrates, were each equal, within experiment error, to that of methane and significantly different from that of acetylene. As discussed previously,4 the 13C mole fraction of CH3 is approximately equal to in the limit of quasiequilibrium between 13CH3, 12CH3 and l3l3C2H2, 12,,3C2H2. However, as pointed out by Frenklach,17 interconversion of CH3 and C2H2 is well out of quasiequilibrium away from the filament. It is readily seen that Eq. (1) is also approximately valid in the oppo- site limit, irreversible conversion of l3CH3 and 12CH3 to 13,13C2H2 and 12,13C2H2. In the latter limit the concentration of 13,13C2H2 would be proportional to a cross section times a collision number multiplied by [13CH4]2, while [12,13C2H2] would be proportional to twice (the factor of two results from the inversion symmetry present in 13,13C2H2 but lacking in I2,I3C2H2) the same cross section and collision number multiplied by [12CH4] [I3CH4], resulting in Eq. (1). However, irreversible acetylene formation from methyl radicals is contradicted by the observation4,5 of partial conversion of 12C2H2 to ' C2H2, presumably via 12CH3. Equation (1) is likely to be reasonably accurate given its validity in either limit but is clearly subject to uncertainty. On the other hand, interconversion of CH3 and CH4, via CH4 + H^CH3 + H2, has been calculated by Goodwin and Gavillet18 to remain in quasiequilibrium to within a few mm of the substrate. The quasiequilibrium will insure that CH3 and CH4 remain in isotopic equilibrium. Near the substrate, however, quasiequilibrium fails due to surface-catalyzed H recombination (via abstraction of surface hydrogen), as predicted by Goodwin and Gavillet,8(b) and observed by Hsu.19 CH3 and CH4 will nonetheless remain in isotopic equilibrium unless reactions intercon-verting C, and C2 species occur at a significant rate within the boundary layer or on the diamond surface. The preponderance of evidence, including detailed modeling calculations,18 indicates that Cj^±C2 interconversion is much slower than CH3^±CH4 interconversion. We observed essentially identical 13C mole fractions in methane and acetylene with and without a diamond substrate, suggesting that surface-catalyzed interconversion is insignificant. However, both Harris et a 1.20 and Wu and co-workers9(a) observed, using a quartz probe, a CH4/C2H2 ratio that continually increased with increasing distance from the hot filament, which might be interpreted as implying non-negligible C,^C2 interconversion chemistry. These spatial variations in the CH4/C2H2 ratio are probably due primarily to thermal diffusion,20 as the ap- Polycrystalline Diamond homoepitaxial Gas phase reactants (100) (111) (110) CH4" C2H2 CH^CjHJ 57.7 57.7 32.4 4.1 61.0 64.8 27.7 3.9 58.4 64.2 33.5 3.3 55.6 59.8 36.1 4.0 Mean: 58.2 ± 3.6 61.6 ± 5.5 32.4 ± 5.6 3.8 ±0.6 57.6 53.9 34.8 2.8 55.5 61.1 36.0 1.8 55.3 58.5 30.5 2.6 57.0 57.6 57.8 61.1 38.4 2.9 Mean: 56.8 ± 1.2 58.6 ± 5.4 34.9 ± 5.3 2.5 ± 0.8 parent carbon balance was low near the filament,9(a)'20 and perturbation of the gas phase temperature profile and chemistry by the probe is probably significant.20 In any case, our sampling probes were located at a distance from the filament within ~ 1 mm of the filament-substrate distance, and therefore sampled a CH^/C2H2 ratio representative of that at the substrate surface. Any further reactions of stable hydrocarbon species within the water-cooled probe would be quenched. We thus conclude that the l3C mole fractions of CH3 and CH4 at the substrate should be very nearly equal, and the proximity of the gas-sampling probes to the substrate should assure a reasonably accurate determination of the isotopic compositions of methane and acetylene. Clearly, a true in situ determination of the 13C mole fractions of the gas phase species above the surface, by molecular beam sampling mass spectrometry, for example, would be preferable to our ex situ measurements, as noted previously.4 We think it is very unlikely, however, that the 13C mole fraction of CH3 at the substrate surface differed from our measured value for CH4 (59-62%) by more than 5-10%. Given the magnitude of the difference in 13C mole fractions between diamond (57-58%) and acetylene (32-35%), it seems quite clear that methyl radicals were the predominant growth precursor. We cannot exclude contributions to growth by acetylene, however. If the differences between the 13C mole fractions of diamond and methane (methyl radical) are real, then sr90% of the diamond originated from methyl radicals and the remainder from acetylene. Acetylene-based growth mechanisms,3 while not disproved by our experiments, are shown not to be dominant, at least under the growth conditions employed here. As discussed elsewhere,5,21 existing methyl radical mechanisms have shortcomings as well. Considerably more experimental and theoretical work, particularly on surface intermediates and surface rate processes, will be necessary in order to elucidate the details of the adsorption, decomposition, and incorporation of methyl radicals into the diamond lattice. The authors acknowledge the National Science Foundation for support of this work by Grant CHE-8807546 and the Office of Naval Research for additional support. The authors thank Dr. Mike Tamor for bringing the new 13C Raman data to our attention and Dr. Bill Banholzer, Dr. Curtis Johnson, Dr. Barbara Garrison, Dr. Michael Frenklach, and Dr. Steve Harris for providing us with manuscripts prior to publication. MPD also thanks Dr. Bill Banholzer, Dr. Steve Harris, Dr. Dave Goodwin, Dr. Ching-Hsong (George) Wu, and Dr. Mike Tamor for useful discussions. '(a) R. C. DeVries, Ann. Rev. Mater. Sci. 17, 161 (1987); (b) A. R. Badzian and R. C. DeVries, Mater. Res. Bull. 23, 385 (1988); (c) J. C. Angus and C. C. Hayman, Science 241, 913 (1988); (d) K. E. Spear, J. Am. Ceram. Soc. 72, 171 (1989); (e) J. C. Angus, in Diamond and Diamondlike Films, edited by J. P. Dismukes, A. J. Purdes, B. S. Mey-erson, T. D. Moustakas, K. E. Spear, K. V. Ravi, and M. Yoder (The Electrochemical Society, Pennington, NJ, 1989), p. 1. 2 (a) M. Tsuda, M. Nakajima, and S. Oikawa, J. Am. Chem. Soc. 108, 5780 (1986); (b) M. Tsuda, M. Nakajima, and S. Oikawa, Jpn. J. Appl. Phys. 26, L527 (1987); (c) S. J. Harris, Appl. Phys. Lett. 56, 2298 (1990); (d) W. A. Yarbrough, "Diamond growth on the (110) surface," in Diamond Optics IV, edited by S. Holly and A. Feldman (SPIE, Bellingham, WA), (in press); (e) S. J. Harris and D. N. Belton, Thin Solid Films (unpublished); (f) B. J. Garrison and D. W. Brenner, Science (in press). 3 (a) M. Frenklach and K. E. Spear, J. Mater. Res. 3, 133 (1988); (b) D. Huang, M. Frenklach, and M. Maroncelli, J. Phys. Chem. 92, 6379 (1988); (c) M. Frenklach and H. Wang, Phys. Rev. B 43, 1520 (1991); (d) D. N. Belton and S. J. Harris, in Diamond Materials, edited by A. J. Purdes, J. C. Angus, R. F. Davis, B. M. Meyerson, K. E. Spear, and M. Yoder (The Electrochemical Society, Pennington, NJ, 1991), p. 170; J. Chem. Phys. (in press); (e) D. Huang and M. Frenklach, J. Phys. Chem. (in press). 4C. J. Chu, M. P. D'Evelyn, R. H. Hauge, and J. L. Margrave, J. Mater. Res. 5, 2405 (1990). 5C. J. Chu, M. P. D'Evelyn, R. H. Hauge, and J. L. Margrave, J. Appl. Phys. 70, 1695 (1991). 6 (a) L. R. Martin and M. W. Hill, J. Mater. Sci. Lett. 9, 621 (1990); (b) S. J. Harris and L. R. Martin, J. Mater. Res. 5, 2313 (1990). 7C. E. Johnson, W. A. Weimer, and F. M. Cerio (unpublished). 8 (a) L. S. Piano, D. A. Stevenson, and J. R. Carruthers, in New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 257; (b) W. A. Yarbrough, K. Tankala, and T. DebRoy, in New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 341; (c) S. J. Harris and A. M. Weiner, Thin Solid Films (unpublished). '(a) C.-H. Wu, M. A. Tamor, T. J. Potter, and E. W. Kaiser, J. Appl. Phys. 68, 4825 (1990); (b) T. Yasuda, K. Miyamoto, M. Ihara, and H. Komiyama, in New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 353. I0K. C. Hass, M. A. Tamor, T. R. Anthony, and W. F. Banholzer, Phys. Rev. B (in press). "(a) W. Banholzer, T. Anthony, and R. Gilmore, in New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 857; (b) W. Banholzer and T. Anthony, Thin Solid Films (unpublished). 12 Y. Nisida and H. Kanda, "Isotope Effect of Raman Scattering of Diamond (to be published). ,JR. M. Chrenko, J. Appl. Phys. 63, 5873 (1988). I4M. A. Tamor (private communication to MPD). 15B. V. Derjaguin and D. V. Fedoseev, Diamond Wrought by Man (MIR, Moscow, 1985), pp. 97-98. I6W. F. Banholzer (private communication to MPD). Diamond grown from commercial methane (presumably 1.08% 13C) was found to have a 13C mole fraction of 1.037%. He indicated that further studies are underway. 17 M. Frenklach, in Diamond Materials, edited by A. J. Purdes, J. C. Angus, R. F. Davis, B. M. Meyerson, K. E. Spear, and M. Yoder (The Electrochemical Society, Pennington, NJ, 1991), p. 142. Frenklach suggests, as we also argue, that CH, and CH4 may be in isotopic equilibrium. He goes on to propose yet another explanation of our isotopic labeling experiments involving interconversion of CH3 and C2H2 via C3 species, arguing that no inferences about the relative importance of methyl radicals and acetylene in diamond growth can be drawn from our data. The latter argument is specious, however, as the C3 interconversion model is inconsistent with the data which it purports to explain. 18 (a) D. G. Goodwin and G. G. Gavillet, J. Appl. Phys. 68, 6393 (1990); (b) D. G. Goodwin and G. G. Gavillet, in New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 335. "(a) W. L. Hsu, in Diamond Materials, edited by A. J. Purdes, J. C. Angus, R. F. Davis, B. M. Meyerson, K E. Spear, and M. Yoder (The Electrochemical Society, Pennington, NJ, 1991), p. 217; (b) W. L. Hsu, Appl. Phys. Lett. 59, 1427 (1991). 20S. J. Harris, D. N. Belton, A. M. Weiner, and S. J. Schmieg, J. Appl. Phys. 66, 5353 (1989). 21 (a) Y. L. Yang and M. P. D'Evelyn, J. Am. Chem. Soc. (in press); (b) L. F. Sutcu, M. S. Thompson, C. J. Chu, R. H. Hauge, J. L. Margrave, and M. P. D'Evelyn (unpublished).