Professional Documents
Culture Documents
Spitsyn 1981
Spitsyn 1981
It is shown that diamond crystallization by chemical vapor deposition should preferably be carried out at reduced pressures.
Selective growth of diamond is ensured by introducing atomic hydrogen into the crystallization zone; this suppresses crystal-
lization of graphite. The growth rate of homoepitaxial diamond films reached 1 ~m/h at 1000°C; film properties were identical to
those of bulk crystals. The lattice parameter in boron-doped films (-0.I at.%) decreased by 0.0009 A; the film and substrate
parameters coincide at dopant concentrations of -1 at.%, and the semiconductor diamond film intergrows with the substrate
without stress. Diamond crystals up to several tens of microns in thickness were grown also on non-diamond substrates. At large
supersaturation, the crystal habit is octahedral and at low supersaturation, it is cubic. The linear growth rate is constant at the
early stages of crystal growth but then it diminishes to a level typical for the homoepitaxial growth of diamond films.
219
220 B. V. Spitsyn et al. / Vapor growth of diamond
transform them to the tetravalent state charac- (here and throughout the text ci are constants).
teristic of carbon atoms in the diamond struc- The rate of chemical crystallization depends both
ture. on the supersaturation and on the energy of
A number of recently suggested techniques of activation E of decomposition of a molecule
carbon condensation into a so-called "diamond- AB(gas) at the substrate surface. Consequently,
like carbon" grown from excited carbon atoms in the approximation of direct impact the rate of
and ions on cold diamond and non-diamond CC is
substrates [7-10] can be classified as physical
methods. Usually, dielectric layers grown by Vchem = c2(PAB-- P°m) e x p ( - E / R T )
these techniques contain only fragments of the
= C~(achem-- 1)P ° e x p ( - E / R T ) . (2)
diamond structure, with grain size of the order of
1 0 ~ ; in some cases diamond crystallites with
The dependence of p o and p O on crystallization
sizes up to - 1 / z m were observed [9].
temperature T can be found by using the equa-
Chemical vapor deposition opens new pos-
tions of chemical thermodynamics
sibilities for diamond crystallization in the region
of thermodynamic metastability. As indicated by
P°A = C3 e x p ( - M R T ) ,
the published data, only this technique makes it
possible to achieve acceptable growth rates at 0 0
PAB/PB = Keq = exp(AG°T/RT),
moderate supersaturations, and to grow perfect
homoepitaxial diamond films. where A is the heat of evaporation of the crystal
Let us compare the simplest processes of phy- A(solid), AG ° is the change of the Gibbs free
sical crystallization (PC) and chemical crystal- energy in the chemical reaction, and Keq is the
lization (CC) of a solid A. In the case of PC, equilibrium constant. By assuming
consider a crystal in equilibrium with vapor of
the same composition: P ° ~ p ° ~- px,
crystal will presumably be of greater perfection characteristics of this new process and of the
than crystals grown by PC-methods at the same pyrolysis of methane.
rate and at the same substrate temperature, since CTR synthesis of diamond was carried out
in this case O~chern"~ O~phys. For instance, hydrogen from hydrocarbons in a closed system, by carbon
reduction of tungsten fluoride at -850°C [11] transfer from a graphite source to a diamond
produced single-crystalline tungsten layers with substrate. The gas medium obtainable according
the theoretically predicted density. Physical to the mode of chemical transport reaction con-
sublimation of tungsten cannot lead to similar tains the following simplest hydrocarbons:
results. According to eq. (3), the advantages of methane, ethane, ethylene, acetylene supplying
CC compared to PC are enhanced as the crystal- carbon to the diamond surface. Selectivity of the
lization temperature is lowered and the pressure process was achieved by introducing an increased
Px increased. However, this increases 0~chem , and concentration of atomic hydrogen. The latter
may cause a deterioration of the quality of the interacts strongly with graphite but practically
deposits. The above analysis obviously suffers does not affect diamond [13]. Different methods
from idealizations, since it neglects mass transfer are available for obtaining a superequilibrium
in the gas phase, diffusion in adsorption layers, concentration of atomic hydrogen. The selective
and possible concurrent reactions in the bulk and etching effect with regard to graphite was nearly
on the crystal surface. the same when the thermal (several % H) and
Growth rates and crystal perfection are very the electric discharge (about 10% H) methods
different in different modifications of the chem- were used for obtaining a superequilibrium con-
ical crystallization techniques. For example, the centration of atomic hydrogen. But it is neces-
well-known method of growth on diamond seed sary to note that within the temperature range
crystals by methane pyrolysis [2, 6, 12] is charac- between 800 and 1200°C, we obtain virgin
terized by a low linear growth rate and by an diamond, not graphitized at 1500°C in vacuum,
unavoidable deposition of graphite. This makes in contrast to the so-called diamond-like carbon.
the method suitable only for increasing the mass Because of the multicomponent nature of the
of highly dispersed diamond powders in a cyclic system and the large number of reactions taking
process involving a periodically repeated remo- place in the gas phase and on the surface of the
val of non-diamond carbon from the surface of substrate (hydrogenation of hydrocarbons, recom-
fine powder diamond. High-perfection single- bination of atomic hydrogen, and the crystal-
crystalline layers with thickness of the order of lization process itself), the absolute value of the
several microns can be grown only in a selective supersaturation could not be determined. The
process (suppressing the deposition of graphite) growth rate depends on the character and tem-
which provides a substantially higher linear perature of the substrate, on the partial pressures
growth rate. We have achieved this goal by using of the components, and on the geometry of the
a chemical transport reaction (CTR) for the system. According to ref. [14], in the CTR mode
growth of diamond. Table 1 gives the main the supersaturation at the substrate increases
Table 1
Synthesis of diamond by pyrolysis of CH4 and by the chemical transport reaction
techniques (CTR)
Table 2
Some properties of synthetic diamond
hydrogen which has a concentration in homoe- Boron is mostly incorporated into the {111}
pitaxial films about twice that in natural faces and the incorporation on {100} faces is very
diamond. An increased hydrogen content in low. The electric conductivity of {100} diamond
CTR-grown diamond is caused by the active role films is lower by approximately a factor of 104
of atomic hydrogen in this process. Within the than that of films grown on {111} faces. Boron
experimental errors, the properties of the syn- differs in its electronic structure from the carbon
thetic and natural diamond are identical (see atom which it replaces, and the diamond lattice
table 2). near a dopant atom undergoes a distorition.
Also, boron has a valence equal to 3 and can be
incorporated more easily into a {111} face having
three-fold bonding. This, in the authors' opinion,
4. Semiconducting diamond is one of the causes of a preferred incorporation
of boron during the growth of {111} diamond
Introduction of gaseous boron compounds in faces.
the process of diamond deposition made it pos-
sible to grow and investigate [19] single-crystal-
line semiconducting diamond films. As the 5. Diamond growth on foreign surfaces
doping level is increased from -1017 to
~10Wcm -3, the activation energy of electrical One of the reasons permitting the nucleation
conduction falls from 0.37 to - 0 . 1 e V . of diamond on foreign substrates is the sup-
Measurements of thermo-emf and Hall effect pression of the graphite deposition in the
demonstrated that the semiconducting diamond presence of atomic hydrogen. A n o t h e r reason is
layers have p-type conductivity. As shown by related to the weak effect of the substrate surface
X-ray topography data, boron doping to a con- forces on the nucleation under the conditions of
centration - 0 . 1 a t % decreases the lattice comparatively high supersaturation realized in
parameter of synthetic diamond by 0 . 0 0 0 9 ~ the C T R technique.
[16]. The lattice contraction and hole conduc- Diamond crystallization on foreign surfaces
tivity of diamond indicate that, at low doping was conducted under synthesis conditions typical
levels, boron atoms are mostly found in lattice for homoepitaxial growth but poly- or single-
sites. The lattice parameter of the highly doped crystalline Cu, Si, W, or other substrates were
(up to - 1 a t % boron) semiconducting diamond used instead of diamond seed crystals. Spon-
is practically equal to that of the substrate. It taneous nucleation of diamond crystals was
appears that in this case the film contraction due observed mostly on defects like scratches, grain
to boron incorporation in lattice sites is com- boundaries, dislocation outcrops etc.; this sup-
pensated by lattice expansion due to interstitial ports the assumption that nucleation takes place
boron atoms. on the substrate and not in the gas phase.
224 B. V. Spitsyn et al. / Vapor growth of diamond
grow with the same growth and faceting mode as [2] W.G. Eversole, US Patent No. 3030188, 23.07.1958.
the diamond seed crystals. The crystal habit, [3] J.A. Brinkman, C.J. Meechan and H.M. Dieckamp, US
morphology, and growth rate become identical Patent No. 3175885, 01.06.1960.
[4] J. Lander and J.J. Morrison, Surface Sci. 4 (1966) 241.
for all types of diamond seed crystals (spon-
[5] R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser,
taneously nucleated, natural, and synthesized at K.K. Kelley and D.D. Wagman, Selected Values of the
high pressures) after approximately 30min of Thermodynamic Properties of the Elements (US Atomic
growth. The transient process is caused by the Energy Commission with the Assistance of the Ameri-
irregular shape of the rounded crystals and by a can Society for Metals, Menlo Park, Ohio, 1973).
[6] B.V. Derjaguin and D.V. Fedoseyev, Rost Aimaza i
sharp increase in the number of crystallization Grafita iz Gazovoj Fazy (Nauka, Moscow, 1977).
centers as a result of spontaneous nucleation. [7] S. Aisenberg and R. Chabot, J. Appl. Phys., 42 (1971)
After the crystals placed on the substrate achieve 2953.
a steady-state growth form, they differ from [8] V.M. Golyanov, L.A. Yelesin and M.I. Mikheyeva,
spontaneously nucleated crystals only in a larger Pis'ma v ZhETF 18 (1973) 569.
[9] E.G. Spencer, P.H. Schmidt, D.C. Joy and F,J. San-
size and in the absence of growth twins.
salone, Appl. Phys. Letters, 29 (1976) 118.
During the course of growing seed crystals on [10] N.N. Matyushenko, V.E. Strelnitsky, A.A. Romanov
a copper substrate, the authors have not noticed and V.T. Tolok, Dokl. Akad. Nauk. Ukr. SSR A5 (1976)
any formation of twins on their surface during 459.
about one hour. No further growth was carried [11] R.K. Chuzhko, I.V. Kirillov, Yu.I. Golovanov, V.M.
Lukyanovich, V.K. Petrov, E.I. Evko, A.P. Zakharov,
out because the surface of the copper substrates R.I. Nazarova and Z.G. Mendeleyeva, USSR Inv. Cer-
was overgrown by spontaneously nucleated crys- tif. No. 289147, 21.06.1968 (in Russian).
tals and because the linear growth rate has been [12] P.A. Tesner, A.E. Gorodetsky, E.V. Denisevich, A.P.
diminished on that account. Zakharov and T.V. Tekunova, Dokl. Akad. Nauk SSSR,
The linear growth rate of diamond crystals 222 (1975) 1384.
[13] J.C. Angus, N.C. Gardner, D.J. Poferl, S.P. Chauhan, T.
spontaneously nucleated on the foreign surface is Dyble and Pai Sung, in: Sinteticheskiye Almazy v
constant during the initial stages of growth Promyshlennosti, Ed. V.N. Bakul (Naukova Dumka,
(-5/.~m/h in the above example) but then falls as Kiev, 1974).
the substrate coverage by diamond overgrowth [14] L.N. Aleksandrov, Kinetika Obrazovanija i Struktury
increases. After the surface is completely Twerdich Slojev (Nauka, Novosibirsk, 1972).
[15] A.E. Aleksenko, B.V. Derjaguin, A.E. Gorodetsky,
covered with a polycrystalline layer, the linear A.P. Zakharov, M.O. Kliya, B.V. Spitsyn and L.L.
growth rate becomes nearly that characteristic of Bouilov, Abstracts of the 5th USSR Conf. on Crystal
the homoepitaxial diamond film under compar- Growth, Vol. 1, Mechanism and Kinetics of Crystal-
able synthesis conditions. lization (Tbilisi, 1977) pp. 122-133, 258-259 (in Rus-
sian).
[16] N.A. Bulyenkov, B.V. Derjaguin, V.P. Martovitsky and
Acknowledgements B.V. Spitsyn, 3rd USSR Conf. on Structural Defects in
Semiconductors, Abstract of Reports, Part 2 (Novosi-
birsk, 1978) p. 228 (in Russian).
The authors are grateful to A.E. Aleksenko [17] V.S. Vavilov, A.A. Gippius, A.M. Zaitsev, B.V. Der-
and A.A. Botev for participation in the experi- jaguin, B.V. Spitsyn and A.E. Aleksenko, 2nd USSR
ments, to A.E. Gorodetsky for electron diffrac- Conf. on Wide-Band Semiconductors, Abstracts of
tion studies, and to Z.E. Sheshenina for SEM Reports (Leningrad, 1979) (in Russian).
[18] B.V. Derjaguin, B.V. Spitsyn, A.E. Aleksenko, A.E.
photographs of the specimens. Gorodetsky, A.P. Zakharov and R.I. Nazarova, Dokl.
Akad. Nauk SSSR 213 (1973) 1059.
[19] A.E. Aleksenko, B.V. Spitsyn, S.D. Tkachenko, V.S.
References Vavilov, B.V. Derjaguin, M.A. Gukosyan, T.A. Kara-
tigina, E.A. Konorova and V.F. Serigiyenko, Dokl.
[1] B.V. Spitsyn and B.V. Derjaguin, USSR Inv. Certif. No. Akad. Nauk SSSR 233 (1977) 331.
339134, 10.07.1956 (in Russian). Official Bulletin of In- [20] N.A. Pangarov, in: Rost Kristallov, Vol. 10, Ed. N.N.
ventions of USSR, 1980, No. 17, p. 323. Sheftal' (Nauka, Moscow, 1974).