Diamond and non-diamond carbon synthesis in an oxygen-acetylene flame

Thin Solid Films. 196 (1991) 271-281 PREPARATION

271

AND CHARACTERIZATION

DIAMOND AND NON-DIAMOND OXYGEN-ACETYLENE FLAME L. M. HANSSEN*,

K. A. SNAIL?,

CARBON

W. A. CARRINGTONS,

SYNTHESIS

J. E. BUTLER/l,

IN AN

S. KELLOGG*

AND

D. B. OAKESB

Code 6522, Opiical Sciences Dioision, Naval Research Laboratory,

Washingron DC 20375-5000 I U.S.A.)

(Received January

8, 1990)

25, 1990; revised July 5, 1990: accepted

August

Diamond and non-diamond carbon deposition in an oxygen-acetylene combustion flame has been analyzed over a range of oxidizer:fuel ratios (R,) and substrate temperatures (T,). The effects on diamond deposition of substrate preparation and position in the oxygen-acetylene flame have been examined. Diagrams relating diamond, microcrystalline graphite and amorphous carbon growth to the oxygen:acetylene flow ratio and substrate temperature have been developed. In addition, the dependences of particle morphology and growth rate on T, and R, were examined. Micro-Raman spectroscopy, optical microscopy and scanning electron microscopy were used to analyze the growth.

1.

INTRODUCTION

The recent reports of diamond synthesis in an oxygen-acetylene flame in air have sparked a great deal of interest in the diamond chemical vapor deposition (CVD) community’~-6. This is due in part to the simplicity and low cost of the experimental apparatus as well as the high growth rates of diamond that can be achieved’**. Several groups have expanded on the initial work, detailing the appropriate experimental conditions for diamond growth’, growing diamond in a low pressure chamber5 as well as in a variety of hydrocarbon-oxidizer flames and on a variety of substrate materials 2,738. The diamond deposit has been characterized by various techniques, including Raman spectroscopy, scanning electron and optical microscopy, X-ray diffraction and electron channeling2,9. The combustion-flame-grown diamond has been found to be very similar to

*Present address: Bldg 220, Rm B-306, National Institute of Standards and Technology, Gaithersburg, MD 20899, U.S.A. t Author to whom correspondence should be sent. $ Present address: Crystalline Materials, 2411 Old Crow Canyon Road, Suite A, San Ramon, CA 94583, U.S.A. 11 Code 6174, Chemistry Division, Naval Research Laboratory, Washington, DC 20375-5000, U.S.A. 7 Geo-centers, 10903 Indian Head Hwy., Fort Washington, MD 20744, U.S.A. @Present address: Physical Sciences Inc., 20 New England Business Center, Andover, MA 01810. U.S.A. Elsevier Sequoia/Printed

in The Netherlands

272

L. M. HANSSEN et al.

CVD diamond and natural diamond. Other forms of carbon, such as amorphous carbon and microcrystalline graphite’, have also been detected in the flame deposits. The effects of critical parameters on the diamond growth have been partially investigated. However, a comprehensive study characterizing the deposit as a function of oxidizer:fuel ratio and substrate temperature has yet to be reported. In what follows, an examination of the flame deposit and its dependence on the conditions of growth is presented. 2. EXPERIMENTAL

SET-UP

The apparatus used for the carbon growth consisted of a commercial oxygenacetylene brazing torch, fed by a gas mass flow control system, and a watercooled copper substrate mount. A two-color IR pyrometer (which measures 2.2 and 2.4 urn IR radiation) was used to monitor the substrate temperature during growth. The arrangement of these components is shown in Fig. 1. MASS FLOW CONTROLLERS

PYROMETER

NOZZLE FLAME A

/a’

,

SUBSTRATE

COPPER

MOUNT

SCREWDRIVER

Fig. 1. Schematic diagram of the oxygen-acetylene flame deposition apparatus. The molybdenum threaded rod substrate temperature is controlled through height adjustment relative to the water-cooled mount.

High purity oxygen (99.99%) and acetylene (99.6%) were used as source gases. A calibrated mass flow control system was used to set the individual gas flow rates of approximately 1.5 slm (standard litres per minute) to within +O.Ol slm. For all growths performed the combined flow rate of the gases was kept constant at 3 slm, while the ratio (R,) of the flow rate of oxygen to the flow rate of acetylene was varied. The substrate temperature (T,) was adjusted and controlled through variation of the

DIAMOND

AND CARBON

SYNTHESIS

IN AN OXYGEN-ACETYLENE

FLAME

273

substrate surface position relative to the top of the copper mount (see substrate description below). The substrate temperature was monitored and controlled to within f 20 “C. The substrates used for this study were silicon-coated molybdenum. Each substrate was an approximately 1.5 cm long section of g-16 molybdenum threaded rod, slotted on one end and polished with Sic and diamond grit on the other. The polished side was chemically cleaned, argon ion beam sputtered and coated with approximately 2000 A of silicon. The silicon layer was applied to improve the adhesion of the flame deposit to the molybdenum substrate. As a last step prior to flame growth, the coated samples were lightly scratched (two passes over 2 cm of coated cloth) with diamond 6 pm grit. Without this final step the nucleation density during flame growth was found to be significantly reduced. On the other hand, heavy scratching led to high nucleation density and continuous films, which tended to lift off the substrate and sometimes break apart upon cooling. Threaded molybdenum rod was chosen as a substrate material to achieve substrate temperature control and uniformity, avoiding the severe temperature drift between and gradients associated with other substrates4T6. The thermal conductance the substrate surface and the cooled holder, and thereby the substrate temperature, was varied and set by adjustment of the extension of the growth surface above the holder. Using this method, substrate surface temperatures were selected and maintained over a range from about 650 to 1200 “C. The deposition time for all samples was 8 min. The primary tool used to analyze the deposit was a micro-Raman spectrometer. Raman spectroscopy can be used to identify diamond, amorphous carbon and microcrystalline graphite as well as to detect them in combination”,“. The microRaman spectrophotometer used for this work (described in detail in ref. 12) employed a 5145 A argon ion laser, a 1024-channel OMA detector array and a specially adapted optical microscope. The spectra presented in this paper were taken with approximately 1 pm laser spot size on individual particles of the flame deposit. The spectra were integrated for 100 s with 4.3 mW of laser excitation radiation. 3.

SPATIAL

VARIATION

OF GROWTH

The substrate growth surface was positioned perpendicular to the flame axis as shown in Fig. 1. The flame deposit was found to be non-uniform across the substrate surface and varied with position radially outward from the center of the growth and flame axis. The deposit was seen to vary in average size, nucleation density, growth rate, particle morphology and type of carbon deposit. Both scanning electron microscope (SEM) observations and Raman spectra demonstrate this radial dependence. The variation of deposit type and amount with radial position on the substrate has been studied by Oakes et al. for a silicon substrate in O,-C*H,-H, and O,-C,H, flames”. An example of the radial dependence of morphology of growth on a silicon-coated molybdenum substrate in an O,-C,H, flame can be seen in Fig. 2. SEM photos are shown for an 8 min deposit at T, = 700 “C and R, = 0.95. Figure 2(a) is taken from the center of the growth, while Figs. 2(b), 2(c) and 2(d) are taken

274

L. M. HANSSEN e*t al.

Fig. 2. SEM photographs of typical flame deposit (grown for 8 min at T, = 700 C and R, = 0.95) from four points across the circular region of the deposit on the substrate: (a) center, (b) 2 mm away from the center,(c) 3 mm and(d) 4 mm.

from positions 2,3, and 4 mm radially outward from the center respectively. In the center well-faceted particles are seen with minimal amounts of secondary growth. At 2 mm away from the center the amount of secondary growth increases significantly. At 3 mm away the particles retain only a hint of a faceted structure. At 4 mm away, near the edge of the growth the particles have become almost perfectly spherical (ball like). This behavior is dependent on the growth conditions. For some samples grown with other R, and T, values no significant change in particle morphology with position was observed, but the average particle size (indicative of the growth rate) and/or nucleation density varied radially and not necessarily monotonically (see ref. 12 for details).

DIAMOND

AND CARBON

SYNTHESIS

IN AN OXYGEN-ACETYLENE

275

FLAME

For all substrates the deposit was observed to depend on the deposit’s position relative to the central flame axis. Two parameters which have the same radial symmetry as the deposit variation are substrate temperature and flame species distribution. During growth the temperature variation across the molybdenum substrates was considerably less than that observed across silicon Nevertheless, a similar radial variation of deposit substrates reported earlier4.“. was observed for both the molybdenum and silicon substrates. This suggests the flame species distribution as the primary cause of the radial variations in growth. 4.

DEPENDENCE

OF

GROWTH

QUANTITY,

MORPHOLOGY

AND

TYPE

ON

PROCESS

PARAMETERS

The morphologies of the deposited particles observed in this study can be grouped into three categories: “ball-like” growth, faceted growth and intermediate growth. The intermediate growth particles exhibit facets, but with varying amounts of secondary facet growth such as those seen in Figs. 2(at2(c). Typical ball and facet growths from four samples are shown in Fig. 3. The vibrational Raman spectra, discussed in detail below, were used to identify the nature of the carbon deposits. Faceted diamond crystal particles grown with T, = 1000°C and Rf = 1.05 are shown in Fig. 3(a); ball-like particles of microcrystalline graphite grown with T, = 1150 “C and R, = 0.80in Fig. 3(b); amorphous carbon grown with T, = 750 “C and R, = 0.80in Fig. 3(c); and ball-like particles containing all three carbon phases grown with T, = 900 “C and R, = 0.95 in Fig. 3(d). Samples were deposited with oxygen-to-acetylene flow ratios ranging from R, = 0.70 to 1.20 and with substrate temperatures ranging from T, = 660 to 1200 “C. The results of optical observations of the deposit type in the center of growth are summarized in Fig. 4. Particles exhibiting any amount of facetting are grouped together under the facet growth symbol. All spherical particles observed are grouped together under the ball-like symbol. No carbon particle growth could be detected with gas flow ratios equal to or greater than 1.1. However, at high substrate temperature (105&1200 “C) and a flow ratio of 1.1, SIC particles were formed from the flame reactants and the silicon coating of the substrate. The region of diamond growth as evidenced by the Raman spectra discussed below is shaded in. Micro-Raman spectra were taken from particles near the center of growth of each sample. The spectra show evidence of diamond (1332 cm- ‘), amorphous carbon (15 10 cm - ‘), microcrystalline graphite (1350 and 1590 cm- ‘) and a broad fluorescence background, possibly due to defects in the diamond lattice’j. A series of Raman spectra for samples grown with R, = 1.0 and for T, ranging from 700 to 1200 “C in 100 “C steps (curves labeled (a)-(f) respectively) is shown in Fig. 5. At the lower temperatures 700 “C (Fig. 5(a) and 800 “C (curve (a)) diamond and amorphous carbon lines comprise the spectrum. As the substrate temperature increases, the diamond peak increases in relative magnitude to 900 “C (Fig. 5(c)), then diminishes with increasing temperature. The broad amorphous carbon peak (from about 1700 to 1000 cm I, peaked near 1500 cm - ‘, with a shoulder near 1350 cm- ‘) is present at all temperatures, strongest at lower temperatures and reduced but still significant at the higher temperatures 1100 “C (Fig. 5(e)) and 1200 “C (Fig.

276

L.

M.

HANSSEN et al.

(b)

Cc)

(4

Fig. 3. SEM photographs of deposit (see text for growth conditions): faceted crystallites of (a) diamond and ball-like structures composed of(b) microcrystalline graphite, (c)amorphous carbon and (d) all three carbon forms.

5(f)). The microcrystalline graphite peaks (at 1350 and 1580 cm- ‘) begin to appear at 900 “C and increase monotonically with increasing temperature. Using the peak height information from the Raman spectra of all the samples, the regions in R,-T, space in which diamond, amorphous carbon and microcrystalline graphite grow can be examined. As is evident from Fig. 5, these regions are not completely exclusive, but rather, they overlap. The relative Raman peak heights of the various carbon phases do not give an absolute measure of relative volume or mass amounts of material deposited14; they can, however, paint an informative qualitative picture outlining the regions of growth. The existence of each phase of carbon identified through the relative magnitudes of their Raman peaks vs. R, and T, is

DIAMOND

c

c

z ;i; $

AND CARBON

SYNTHESIS

IN AN OXYGEN-ACETYLENE

FLAME

277

1200 i 1100 1000

E” f

900

d s

800

z d

700 600

--l q l

’

0

0

L 0.6

ball growth facet growth nogrowth Raman diamond

peak

I 0.7

0.8

0.9

Oxygen/Acetylene

1.0

1.1

1.2

1.3

Flow Ratio, Rt

Fig. 4. Diagram of carbon growth morphology in the range of R,-T, space studied. Note that the shaded region, where Raman analysis indicates the presence of diamond, contains all of the facet growth as well as some of the ball-like growth.

1600

1400

Raman Shift

1200

1000

(cm-l)

Fig. 5. Micro-Raman spectra taken from the center of flame deposit, grown at R, = 1.00 and T, ranging from (a) 700 “C to (f) 1200 “C in 100 “C intervals. The sharp diamond peak is at 1332 cm-‘. The broad (I 700-1000 cm- ‘) amorphous carbon line has a peak near 1500 cm ’ and a shoulder near 1350 cm- ‘. The two peaks at 1350 and 1590 cm ’ are associated with microcrystalline and crystalline graphite. Each spectrum has been scaled to unity at 1332 cm- ‘.

shown in Figs. 6(at6(c). The values of the individual Raman peak heights are shown normalized to the sum of the peak heights of the four Raman lines described below. The normalization removes spurious effects due to fluctuations in the micro-Raman system from run to run and day to day. Empty squares, which must be distinguished

278

L. M. HANSSEN et a/.

Fig. 6. Relative strength of Raman peaks corresponding to (a) diamond,(b) microcrystalline graphite and (c)amorphous carbon for all samples exhibiting growth, US.flow ratio and substrate temperature. Empty sites indicate that either no sample was grown or no growth was seen.

from squares with a “0” value, indicate that either no growth was seen on the sample or no sample was made. The plot of the diamond Raman line (1332 cm-‘) peak height data shown in Fig. 6(a) exhibits a trend toward higher quality of diamond at higher flow ratios and lower substrate temperatures. The sum of the crystalline graphite peaks (1350 and 1590 cm- ‘) shown in Fig. 6(b) exhibits the opposite trend, becoming stronger with decreasing flow ratio and increasing temperaure. Evidence of amorphous carbon (1500 cm ‘) in Fig. 6(c) is found on almost all samples; the Raman peak is, however, strongest at the lowest temperatures and diminishes with increasing temperature.

DIAMOND

AND

CARBON

SYNTHESIS

IN AN OXYGEN-ACETYLENE

FLAME

279

The flame-grown samples exhibited a wide range of growth rates, from approximately 8 urn hh’ for R, = 1.05 and K = 700 “C to approximately 100 urn h- ’ for R, = 0.90 and T, = 1150 “C. The fraction of the substrate area covered by growth (after 8 min) vs. R, and K is shown in Fig. 7. Each sample was categorized (high, med, low, none) through optical microscope observation. This represents a measure of combined growth rate and nucleation density. Note that the coverage is generally a monotonic function of temperature, ranging from the growth (low) of small particles covering a few per cent or less of the substrate for T, < 700 “C to nearly continuous film growth (high) for T, 2 950 “C.

1300

-

1200

-

G e ;

1100

5

1000

5 $ Q) E :: s fs

.

0

e...

.

-

.

. .

9000

E”

f

.

800

-

700

-

600

* 0.6

0

.

.

+

I

.

+

.

.

0

.

.

..t

.

9

+ .

x

0

.*xx x

0

x

x

0 x

0.8

0.9

00

0

+ +

l

17

000

high med low

none

ox*+ x

0

+

f 0.7

1.0

1.1

1.2

1.3

Oxygen/Acetylene Flow Ratio, Fit Fig. 7. Quantity of growth as a function of Rf and T, for the same samples described in Fig. 4. The area1 coverage C, of the substrate in the center of growth (total growth time 8 min) is shown: high, CA > 85%; medium, 10% < C, < 85%: low, 0.01% < C, < 10%.

The data presented above were taken from substrates which were scratched to enhance nucleation. A few samples were produced without diamond scratching. Microscope observations and Raman spectra of these samples indicate general agreement with the scratched substrate results. However, in T,-R, space (Fig. 4) the facet-ball boundary and the Raman diamond line boundary are shifted to lower R, and higher T,, expanding the region of diamond growth. In addition, the observed crystallite quality (degree of secondary nucleation) and the relative strength of the Raman diamond line were both improved for the unscratched (OS. scratched) substrates. 5.

CONCLUSIONS

AND DISCUSSION

Several phases of carbon, including diamond,.amorphous carbon and microcrystalline graphite, are deposited by an oxygen-acetylene flame within the substrate

280

L. M. HANSSEN

et d.

temperature range 700 S T, I 1200 “C and gas flow ratio range 0.80 i R, i 1.10. Even for a fixed substrate temperature and a fixed gas flow ratio the dominant carbon phase of the flame deposit can vary spatially. In addition, on a local scale of 1 urn or less the deposit can contain all three phases of carbon. Morphological evidence of diamond growth in the center of the sample (faceted crystallite particles as observed by microscopy) is limited to the temperature region near T, = 800 “C at R, = 0.90, which expands to 660 I T, I 1200 “C as R, is increased to 1.05. Either ball-like or no growth is observed outside this region. Micro-Raman analysis shows that the diamond growth region is expanded beyond that described above owing to the evidence of the diamond phonon in the ball-like particles near the facet-ball boundary in ‘T-R, space. The microcrystalline graphite content of the growth is found to increase as T, increases to 1200 “C and R, decreases to 0.70. The amorphous carbon content is largest at T, = 700 “C and decreases with increasing 7; for all flow ratios. The amount of carbon deposited is found to be strongly dependent on substrate temperature, ranging from less than 0.1% coverage at T, I 700 “C to loo”/, coverage at T, = 1200 “C. 660 “C is the lower temperature limit of these experiments; diamond growth may occur at lower temperatures. The limits of substrate temperature and oxygen:acetylene flow ratio on diamond growth in a flame, as determined in this study, are important clues to an understanding of diamond synthesis. Atomic hydrogen has been recognized to be a critical gas phase constituent in the CVD of diamond. It has been shown to selectively etch graphite in preference to diamond’ 5 and to improve the quality of flame-grown diamond 4-12. Hence its presence or absence would seem likely to be strongly linked to the limits of diamond growth. As the flow ratio of oxygen to acetylene is varied in a flame, upper and lower limits to diamond growth are found. As the flow ratio is decreased, the carbon supersaturation and hence the condensation rate of various forms of solid carbon should increase, leading to the gradual change in deposit from diamond to other forms of carbon. As the flow ratio is increased beyond 1.0, the fraction of hydrocarbon species in the flame should be drastically reduced, leading to a sharp drop-off in carbon deposition independent of substrate temperature. As the substrate temperature is varied, the gas phase species in most of the flame may vary only slightly for a fixed oxygen-to-acetylene flow ratio; however, the composition of species penetrating the boundary layer may vary significantly. The hydrogen termination of the growth surface will be reduced at temperatures above about 900-1000 “C as thermal desorption of hydrogen takes place’“. Hence the diamondnon-diamond boundary in Figs. 4 and 6 may be affected by a dynamic balance of atomic hydrogen adsorption and thermal desorption from the growth surface, as well as hydrogen abstraction processes. In addition, the variation in stability of diamond and non-diamond carbon nuclei on the substrate surface with flow ratio and temperature will affect the nature of the deposit and the location of the diamond-non-diamond boundary. At substrate temperatures below 660 ‘C the reduced surface mobility of carbonaceous species as well as the reduced thermal agitation of the lattice may also prevent diamond formation and lead to amorphous carbon growth.

DIAMOND

AND CARBON

SYNTHESIS

IN AN OXYGEN-ACETYLENE

FLAME

281

ACKNOWLEDGMENTS

The authors wish to acknowledge the support of the Office of Naval and SDIO/IST for a portion of the work described herein.

Research

REFERENCES

1 2 3 4

5 6 7 8 9 10 II I2 13 14 15 16

Y. Hirose and N. Kondo, Program and Book qf Abstrac1.v. Japan Applied Physics I988 Spring Meeting. March 29.1988, p. 95. L. M. Hanssen. W. A. Carrington, J. E. Butler and K. A. Snail, Mater. Lett.. 7 (7.8) (1988) 289. Y. Hirose, Proc. 1st Int. Conf: on the New Diamond Science and Technology, Tokyo, October 24-26, 1989. Japan New Diamond Forum (Tokyo) in the press. K. A. Snail, L. M. Hanssen, W. A. Carrington, D. B. Oakes and J. E. Butler, Proc. 1st Int. Cot$ on the New, Diamond Science und Technology, Tok!,o, October 24-26, 1989. Japan New Diamond Forum (Tokyo) in the press. W. Yarborough, Surf: Coot. Technol.. 39140 (1989) 241. P. Kosky and D. S. McAtee, Mater. Lett., 8 (9) (1989) 369. W. A. Carrington, L. M. Hanssen, K. A. Snail, D. B. Oakes and J. E. Butler, Metal/. Trans. A, 20 ( 1989) 1282. J. E. Butler, F. G. Celii, D. B. Oakes. L. M. Hanssen, W. A. Carrington and K. A. Snail, High Temp. Sci..27(1990) 183. R. Vardiman, C. Void. K. A. Snail, J. E. Butler and C. Pdnde, Mater Lett., 8 (1989) 468. D. S. Knight and W. B. White, J. Mater. Res., 4 (2) (1989) 385. R. J. Nemanich and S. A. Solin, Phys. Rev. B. 20 (1979) 392. D. B. Oakes. J. E. Butler. K. A. Snail. W. A. Carrington and L. M. Hanssen. Appl. PhJ.v.. in the press. D. S. Knight and W. B. White, Pro<. SPIE Co@ on Raman Scattering, Luminescence, and Spectroscopic Instrumentation in Technology, vol. 1055. SPIE, New York, 1989. 144. N. Wada and S. A. Solin, P/zysica B, IO5 (1981) 353. R. Yamada, J. Vat. Sci. Technol. A, 5 (4) (1987) 2222. B. B. Pate, SwjI Sri., 16s (1985)83.