Carbon H169. Vol. 7, I'p. I-tt p(~q(;un(.H1 Press.

Printed in Great Britain

GROWTH OF SINGLE-CRYSTAL GRAPHITE BY

PYROLYSIS OF ACETYLENE OVER METALS
A. E. B. PRESLAND
Department of Chemical Engineering and Chemical Technology, Imperial College,
London S.W.7, England
and
P. L. WALKER. JR.
Department of Materials Science, The Pennsylvania State University, University Park,
Penn. 16802, U.S.A.

(Received 4 October 1968)

Abstract- Carbon has been prepared by pyrolysis of acetylene over foils of platinum and
nickel heated to 1000°C. The deposits are shown to consist of three-dimensionally
ordered graphite, with crystal dimensions in the a-direction (parallel to the substrate
surface) as large as 5 u: Optical and electron microscopy has been used to show that the
graphite nucleates preferentially at substrate grain boundaries and kink sites. In the case
of nickel, surface steps also influence growth, and the graphite film produced by com-
plete coverage of the substrate is under compressive stress. Mechanisms for the forma-
tion of graphite by this means are discussed. and evidence is provided which indicates
that precipitation from solution in the substrate cannot be involved.

1. INTRODUCTION of the carbide, this carbon reappears as

The formation of crystalline graphite is graphite.
a process of considerable interest. The largest By anology with the growth of thin single
single crystals are found in geological crystal metal films, it might be supposed that
deposits (e.g. Ticonderoga, Madagascar) an alternative approach to. the production
where they may have crystallized from of crystalline graphite would be to deposit
solution in the mineral matrix. Many transi- carbon from the vapor phase on to a suitable
tion metals. too. dissolve significant quantities substrate, at a temperature high enough to
of carbon at high temperatures, which give adequate mobility: A. pyrolytic carbon
precipitates as graphite on cooling. The with almost theoretical density, three-
best-known example of this is the 'kish' dimensional ordering nuda c-spacirig vof
which forms during steel-making, Austerrnan 3-354 A has in fact been reported [3]. Such
et al.[l] have used melts of iron and nickel carbons usually have a turbostratic structure,
to produce large, relatively perfect graphite however, raising the substrate temperature
crystals in this way. does not significantly improve this, Most
Transition metals have also been used to pyrolytic carbons are deposited on an
lower the temperature of graphitization of existing graphite surface, whereas there is
commercial carbons [2]. It has been suggested some evidence to suggest that ametalsurface
that carbon either dissolves in, or forms a may be more effective in providing the
carbide with, these so-called graphitization required mobility.
catalysts; and that subsequently, either by Thus, studies of the oxidation of single
precipitation from solution. or by dissociation crystal graphite using metal catalysts[4,5]

CAlION VcU. No.1-A

2 A. E. B. PRESLAND and P. L. WALKER, Jr.

have shown that some of these metals (e.g. acetone, and cleaned by dipping in cone.
Pt, Ag) are extremely mobile on the graphite HCI, followed by lengthy washing in distilled
basal plane, indicating a very low interfacial water.
energy. Strips of foil about 5 x I cm were clamped
Furthermore, Banerjee et al.[6] found that between the electrodes of an Elion vacuum
carbon films deposited on nickel by the unit, which was then pumped to a pressure of
pyrolysis of C302 were more graphitic than between 5 x 10-5 and 5 X 10-6 Torr. Current
would have been expected at the temperature was passed through the foil until the tempera-
used. Irving and Walker [7] evaporated ture at the center was approximately 1000°C,
carbon in vacuo on to foils of Mo, W, Pt and as measured with a Leeds and Northrup
Ta heated to 1000°C and obtained enhanced disappearing filament pyrometer. For each
graphitization in all cases. as compared with run. the foil was held in vacuo at this
carbon evaporated on to metallic substrates temperature for not less than 15min, since
at room temperature. Pt and Ta, in parti- Karu and Beer reported [8] that vacuum
cular, gave spot patterns in selected area annealing was a necessary prerequisite for
electron diffraction, indicating graphite the production of graphite films. The foil
crystal diameters (in the a-direction) in was then allowed to cool to room temperature
excess of 1000A. in vacuo, and cylinder acetylene admitted to
Karu and Beer pyrolyzed methane over the work chamber via an air admittance
foils of nickel heated to temperatures in the valve.
range 9OO-HOO°C and obtained thin film Karu and Beer claimed that all their
deposits, which they stripped and examined pyrolyses were carried out using methane at
by transmission electron diffraction [8]. a pressure of between 1-1O,u. as measured
These films gave selected area diffraction on a cold cathode ionization gauge. Initial
patterns corresponding to graphite single experiments by the present authors, using
crystals. both methane and acetylene at pressures of
In view of this evidence, it was decided to 1-50,u gave completely negative results. in
attempt the production of thin films of that no visible deposit could be obtained even
crystalIine graphite by the pyrolysis of after 1 hr pyrolysis at 1000°C. Clearly higher
acetylene over metal foils. Acetylene was used pressures were required, but the vacuum
for this purpose since it is thought to play apparatus could not be maintained at a
a key role as an intermediate in the formation constant pressure in excess of 50,u. Even-
of carbon from all gaseous organic corn- tually the following technique was used, and
pounds [9]. Nickel and platinum foils were proved successful. although interpretation of
used as substrates as they are good dehydro- the results was rendered more complex. With
genating agents. Optical and electron the' work chamber being pumped by the
microscopy were employed to characterize diffusion pump and the metal foil at room
the deposits and to obtain information about temperature, acetylene was admitted until
their nucleation and growth. the pressure rose to about 20,u. The foil was
then heated to 1000±25°C and the flap valve
2. EXPERIMENTAL above the diffusion pump closed. With the
The Ni foils used were 0·0025 in. thick. pump cut off and acetylene continuing to
polycrystalline, with a surface finish suffici- flow in, the pressure in the work chamber
ently rough to give non-specular reflection. rose rapidly; at no time, however, did it
The Pt foil was 0·010 in. thick. polycrystalline, exceed about I Torr. The foil temperature
with an extremely high polish. Both foils was continuously adjusted manually, to take
were reagent grade; they were degreased in account of the increasing gaseous thermal
 GROWTH OF SINGLE-CRYSTAL GRAPHITE 3

conduction. At the end of the desired a continuous, electron-transparent film,

pyrolysis time, the foil was rapidly cooled to supporting all the denser features which had
room temperature by cutting off the current been evident in the optical microscope.
and the work chamber pumped out again. Selected area diffraction from the continuous
The surface of each foil was first examined film' gave a pattern consisting of two weak
with a metallurgical microscope. The central diffuse rings at 2·13 and 1'23 A, corres-
region of the foil was then cut out and the ponding to a pyrolytic carbon with its c-axis
pyrolytic deposit stripped by dissolution in predominantly perpendicular to the film
an appropriate solvent. For Ni this was cone. plane and an La value of about 50 A. The
HCI, while Pt was left overnight in a Petri grain boundary deposit (Fig. 7), however,
dish of cone. HCI containing one drop of gave in addition to the amorphous carbon
cone. HN0 3 • After thorough washing in pattern a single hexagonal spot pattern
distilled water, small pieces of the deposit completely consistent with single crystal
were mounted on grids and examined by graphite (from a selected area - 2[) p}). The
transmission electron microscopy and diffrac- crystallographic nature of the deposit
tion. . boundaries and the occasional moire patterns
observed provided further evidence for this.
3. RESULTS Selected area diffraction, the tendency
3.1 Platinum towards regular morphology, and the
3.1.1 Optical microscopy. Figure 1 is a low- occasional presence of extinction contours
magnification optical micrograph of a suggested that the patch deposits (Fig. 8)
platinum foil after 20 min pyrolysis. Every were also of single crystal graphite. This was
grain boundary was decorated with carbon, confirmed by dark field microscopy.
and considerable deposition had also taken
place,at points within the grains. giving rise
to an overall nucleation density which varied 3.2 Nickel
from grain to grain. These nucleation points 3.2.1 Optical microscopy. The nickel foil
were usually randomly distributed but were surface appeared, under the optical micro-
sometimes arranged. in. sets. of parallel scope. to be very different from that of
straight lines (Fig. 2). In one or two instances. platinum. after both had received the pre-
low-angle boundaries also appeared to be liminary heat treatment. Whereas the effect
decorated (Fig. 3). Some of the random of the latter on platinum was merely to
deposits were large enough to be described produce grain boundary grooving. the
as patches (Fig. 3); when these were situated annealed Ni surface showed, in addition,
in a region ofhigh nucleation density, they extensive step' and terrace formation. This
were surrounded by a partially or completely surface structure had a marked effect On the
denuded zone (Fig. 4). Groups of short. deposit morphology. Thus. nucleation
straight lines were also observed, parallel occurred along single '. stcpsandargrain
within a single grain (Fig. 5). In some cases. boundaries (Fig. 9). At. a later stage. con-
considerable grain boundary migration had siderable lateral. intergrowth had· occurred,
clearly taken place (Fig. 6), one position of and. the tentacular appearance ofthe dep?~it
the boundary being heavily decorated with boundary.indicated thatgroWlhalongsteps
carbon while the other was visible by surface was favored (Fig. 10).
grooving (or, in some cases, by alight carbon As the deposit :thickness increased,lhe
deposit). underlying, step ~tructure. of the substrate
3.1.2 Electron microscopy. The deposit could no longer be seen; .theuppersurfa~e
stripped from the Pt surface was found to be of the deposit took on the 'polygonal· net-
4 A. E. B. PRESLAND and P. L. WALKER, Jr.

work' appearance shown at C in Fig. 10. This 4. DISCUSSION

polygonal network is clearly in evidence in 4.1 Effect ofhydrocarbon pressure
Fig. II, which shows the appearance of the Whatever the detailed mechanisms in-
foil when completely covered with deposit. volved in the dehydrogenation of acetylene
The features at D are 'blisters', i.e, raised to carbon, it is clear that the rate of carbon
areas of deposit film which have parted com- production must increase with pressure and
pany with the underlying substrate. must, therefore, in the experiments described
3.2.2 Electron microscopy. After 3! min increase with pyrolysis time.
pyrolysis, the stripped deposit had the The requirements for single crystal forma-
appearance of a lacework of individual tion are (i) low rate of nucleation, (ii) low
crystals (Fig. 12). The original grain boun- rate of growth, (iii) high mobility. In these
daries of the substrate can be seen, decorated experiments the surface temperature, and
by enhanced deposition. The inset electron hence the mobility, were kept constant. The
diffraction pattern was obtained from an carbon deposition rate was steadily increasing,
area of the deposit corresponding to one however; and it must be assumed that for
such substrate grain. The graphite is clearly some time after the start of pyrolysis the
microcrystalline, albeit with a tendency to combination of deposition rate and mobility
preferred a-axis orientation. Figure 13 was such as to permit the formation of cry-
shows a higher magnification micrograph of stalline graphite. This graphite was nucleated
one of the larger graphite crystals, in which heterogeneously at favored sites on the
can be seen a number of low-angle boun- substrate surface. Eventually, carbon must
daries. The equivalent diffraction pattern have been produced so rapidly that an
(inset) confirms this. amorphous deposit was formed homogen-
After 10 min pyrolysis, the diffraction eously over the surface. On stripping, this
pattern had become more spotty, indicating acted as an extraction replica, bearing the
an increase in crystal size (Fig. 14). The localized graphitic deposits formed earlier
graphite crystals were now becoming in the pyrolysis.
increasingly acicular in shape, with the long
axes lying parallel to one another (Fig. 15). 4.2 Nucleation
The continuous film formed after 33 min 4.2.1 Grain boundary nucleation. The fact
pyrolysis proved to be rather thick for that both platinum and nickel grain boun-
examination in transmission; in view of the daries act as favorable sites for the nucleation
large extinction distance for graphite, the of graphite could be due to (a) the local
film thickness was probably of the order of change in topography at the intersection of
2000-5000 A. Figure 16 shows that the film the grain boundary with the surface, (b) the
contained an extensive tilt-boundary network. action of the grain boundary itself as a pre-
Some of these tilt boundaries may correspond ferred route for the diffusion of carbon, or
to the polygonal structure observed in the of point defects, or (c) the effect of localized
optical microscope. Moire patterns were impurity concentrations as a result of segrega-
also observed (e.g. at M, Fig. 17), indicating tion to the boundary. Some light is thrown
the presence of twist boundaries between on this by consideration of the effects shown
overlapping crystals. The selected area in Fig. 6 where the grain boundary has
diffraction pattern (Fig. 18) corresponded to clearly migrated during pyrolysis.
single crystal graphite, with add~ional we~ This configuration could arise in two ways;
continuous rings through the 1010 and 1120 either site A or site B (Fig. 6) could represent
reflections, caused by the presence of some the final position of the grain boundary.
amorphous carbon. On the whole, migration from A to B seems
 Fig. I. Opt ical micrograph of surface of Pt foil
after 20 min pyrolysis, show ing decorated grain
boundaries and variable nucleation density (154 x ).

Fig. 2. Striated nucleation on Pt (630 ,< ).

Fig. 3. Sub-grain boundary decoration of Pt (630 x ),

[Facin.1J page 4]
 Fig. 4. Denuded zones around
grain boundaries and patch
deposits on Pt (630 x ),

Fig. 5. Linear deposits on Pt (630 x ).

Fig. 6. Effect of grain boundary

migrat ion of Pt (630 x ).
Fig. 7. Electron micrograph of grain boundary deposit stripped
from Pt (8000 x ), (lnset: Equivalent selected area electron
dilfraction pattern.
 Fig. 8. Grain boundary, patch and linear
deposits on Pt (3500 x ),

Fig. 9. Optical micrograph showing grain

boundary and step nucleation on Ni (630 x),

Fig. 10. Step growth on Ni. Note polygonal

network formation at C (630 x),
Fig. 11. Continuous graphite film on Ni, showing polygonal network and
raised 'blisters' at D (630 x ),

Fig. 12. Electro n micrograph of deposit str ipp ed from Ni a fter 3-1/2 min
pyro lysis (2500 x ). (II/Jet : Electron diffraction pattern from area co rres-
po nding to one substra te gra in of Ni .)
 Fig. 13. Low-angle boundaries in a single graphite crystal formed on Ni (21,000 x ), (Inset:
Equivalent selected area electron diffraction pattern of graphite crystal formed on Ni.)

Fig, 14. Electron diffraction pattern of graphite Fig. 15. Acicular crystals of graphite formed on Ni (conditions as
deposit stripped from Ni after 10 min pyrolysis. for Fig. 14) (3500 x ).
 Fig. 16. Transmission electron micrograph of continuous
graphite film stripped from Ni, showing tilt boundary
network (4800 x ).

Fig. 17. Moire patterns (e.g, at M) indicating rela-

tive rotation between overlapping graphite crystals
formed on Ni (17,600 X).

Fig. 18. Selected area electron diffraction pattern of IlTlIphlte

. crystals (corresponding m Fig. 17) formed on.Nl.
 GROWrJ-I OF SINGLE·CRYS'T'AL GRAPHITE

somewhat more likely, for the following (homogeneous) regime, only a thermal
reasons: groove will be formed at the intersection
with the surface. In either case, this now
(i) Grain boundaries migrate towards corresponds to site B. While all this is taking
their center of curvature. place, extensive further deposition of
(ii) If the boundary had moved from B to graphite is occurring at the original nuclei
A, it would be necessary to assume that
formed at site A, followed eventually by an
the boundary could move faster than overall coat.ing of amorphous carbon.
the thermal groove at the surface 4.2.2 Surface nucleation. If the grain boun-
could be eliminated by smoothing.
dary/surface intersection is a preferred
In view of the fact that surface diffu- nucleation site, it might be expected that the
sion coefficients are normally at least point of emergence of a dislocation would
as high, if not higher than those for be similarly favored. The surface nucleation
grain boundary diffusion, this appears density over platinum was found to vary
unlikely. between lOa and 108 em:", which seems too
(iii) The nucleation density in the region high to correspond to the dislocation density
between A and B corresponds to that in a fully-annealed f.c.c, metal. In any case,
outside A (i.e, on grain a) rather than
thermal annealing would have led to poly-
that inside B (on grain 13). Since this gonization, at least; and although occasional
density is a function of grain orienta-
examples interpretable as decoration of low-
tion (perhaps through the kink site angle boundaries were found (Fig. 3), the
density, see (4.2.2), the regionAB must
distribution of nuclei was usually either
have changed either from orientation
random, or heavily striated, as in Fig. 2. Thus
13 to orientation a very early in the some other site must be operative.
pyrolysis, before significant amounts
Those substrate grains whose outer surface
of graphite had been deposited on
corresponded to a high-index plane would.
area AB, or from orientation a to 13
on thermal annealing, be resolved into .01
at a much later time, when amorphous
multiple-faceted surface, which would
carbon was being deposited homo-
providea high density of kink sites suitable
geneously over the entire surface. In
for nucleation [10]. Further, these sites would
the latter Case, however, no decoration
often lie in parallel. rows, thus accounting
of the boundary when in the final A
for the observed striations. The density. of
position could occur.
these kink sites would depend Onthe orien...
The most probable interpretation of areas tation of the initial substrate grain and would,
such as Fig. 6 is, therefore, as follows. In the therefore, vary widely.
very early stages of pyrolysis, .graphite
nucleates along the substrate grain boundary 4.3 .Growth morphology
at A before there is any significant deposition The growth of graphite on beth-platinum
on the surface of grain 13. The boundary then and nickel takes place by the lateralextc:n-
migrates inward; and while it is moving, sion of island nuclei. In. the case of nickell
further deposition on it is inhibited (possibly however. this growth is strongly. influenced
because it moves rapidly). Its motion, of by the surface topography, ill. particular by
course, converts f3 into a, changing the local the step and. terrace structure .of.thesllb-
kink site density. ··1 f the boundary comes to a strate. Thus Fig. 10 shows that gwwth takes
halt •. while ..graphite is still being (hetero- place mainly along the. steps, and the .lath-
geneously) deposited, it willbe lightlydecora- shaped graphite crystals. in Fig. 15 were
ted; if it stops during the amorphouscarbon presumably formed in thill way.
6 A. E. 8. PRESLAND and P. L. WALKER, Jr.

4.4 Epitaxy random at first and steadily improves

The formation of a single crystal over- as surface coverage increases. This, in
growth by deposition on to a crystalline sub- turn, implies recrystallization of the gra-
strate is usually the result of an epitaxial phite.
relationship between overgrowth and Clearly more experimental evidence is
substrate. Thus, it is natural to enquire required to determine whether this type of
whether this is the case here. The two metals graphite overgrowth is epitaxial or not.
used for the substrate were both f.c.c. and
both in the form of cold-rolled foil so that
although polycrystalline they may have had 4.5 Microstructure ofthe continuous graphite film
a similar preferred orientation, which would Complete coverage of the substrate sur-
have been (110)/[112][11]. Thus, most grains face was achieved only for nickel. The
would have had a surface plane close to polygonal network and the 'blisters' shown
(110), the symmetry of which is not parti- in Fig. 11 can be accounted for in the fol-
cularly conducive to epitaxial growth of a lowing way. The graphite overgrowth which
hexagonal structure such as graphite. How- forms at 1000°C is firmly keyed into the
ever, thermal etching may have led to the heavily terraced Ni surface; on cooling to
formation of surface steps involving lower room temperature, stresses are set up in
energy planes such as (111) which, with their the graphite due to differential contraction.
hexagonal symmetry, would have been much Since the graphite c-axis is everywhere
more likely to provide a suitable site for normal to the surface, the relevant expan-
epitaxial growth. sion coefficient is that for the a-axis, which
In the case of platinum, however, the grain is negative up to about gOODe. Extrapolating
boundaries were the prime source of single the results of Nelson and Riley[12], a value
crystal graphite, extending on either side of lil/l at lOOO°C of about 7 X 10-5 would
of the true boundary. Even if the substrate be obtained. This should be compared with
grain on one side were in the appropriate a lilli for Ni at l0000 e of 1·55 X 10-2[13]. The
orientation for epitaxy, it is unlikely that the thickness of the Ni substrate was -100x
correct relationship would be maintained on that of the graphite overgrowth; assuming
the other side. Since most of the grain boun- no deformation of the former, a compressive
daries were decorated with crystalline strain of about 1·5 X 10-2 would be imposed
graphite in this way, it appears either that on the latter, which would be more than
some of the latter grew non-epitaxially or, sufficient to give rise to plastic deformation.
possibly, that the density of (111) facets was The only mode of deformation of graphite
higher in the vicinity of the grain boun- single crystals at room temperature is by
daries. The presence of patches of graphite the motion of basal dislocations, giving rise
of more or less crystallographic shape (see to twinning and tilt boundary formation [14].
Fig. 8) may point to an epitaxial effect, Since the compressive stresses would not be
although this particular morphology could uniaxial but randomly distributed in the
be due to the step structure of the substrate plane of the film, the twins and tilt boun-
surface. daries would form just the sort of polygonal
The graphite deposits on nickel provide network observed.. In regions where the
even less evidence for an epitaxial relation- adhesion of the overgrowth to the substrate
ship. Comparison of Figs. 12,14 and 18shows was poor, the film could accommodate. the
that while the c-axis is normal to the sub- compressive stresses by buckling away from
strate surface from the early stages of the substrate, giving rise to the 'blisters'
pyrolysis, the a-axis orientation is almost shown in Fig. 11.
 GROWTH OF SINGLE-CRYSTAL GRAPHITE 7

4.6 Graphiteformation mechanisms vacuum evaporated on to foils of a number

Three possible mechanisms may be of metals (including Pt) held at 1000°C. The
suggested to account for the growth of resulting films, especially on Pt, were con-
crystalline graphite in the experiments siderably more graphitic than those deposited
which have been described. For convenience at room temperature, even though the
these mechanisms will be described as deposition rate (between I and 10A/sec)
'chemical', 'thin film', and 'metallurgical' was somewhat higher than that used here
respectively. (between 0·5 and 2·5 A/sec.).
4.6.1 Chemical model. Thomas[l5] has 4.6.3 Metallurgical model. Since carbon is
suggested that the formation of carbon being produced at the metal surface, the
during the combustion of aliphatic hydro- concentration gradient set up will cause it
carbons occurs through the free radical to diffuse normally to the surface. Nickel
polymerization of C1 and C2 species, such as dissolves 0'23 wt, % (1,1 at. %) of carbon at
CH 3 , CH 4 , C2H2 etc., to form, first, con- 1000°C[18]. Solubility values are not avail-
jugated polyene radicals, followed by cycli- able for the Pt-C system at this temperature,
zation to polybenzenoid radicals. The C:H but Collier et al.[19] estimate about 0·25 wt, %
ratio of the resulting polynuclear aromatic (4 at. %) at the (reduced) melting point of
hydrocarbons should increase with crystal- 1734°C. Thus significant amounts of carbon
lite size, as pointed out by Digonskii and are taken into solution in Ni and Pt and are
Krylov[16] and confirmed by Cullis et al for known to be precipitated as graphite on
carbon obtained by the pyrolysis of acetylene cooling.
over silica[17]. The efficiency of Pt and Ni It should be born in mind, however, that
substrates as dehydrogenation catalysts may, the thermal inertia of a thin metal foil is
therefore, be an important factor in the very small; and cooling from lOOO°C to room
formation of highly graphitic carbon by temperature would take a few seconds at
pyrolysis of hydrocarbons. The high reaction most. Even if back diffusion of the carbon to
velocities attainable during free radical the surface could take place as quickly as
polymerization make this model very this, the conditions would hardly be ideal for
attractive when considering the formation the formation of well-ordered crystalline
of flame carbon, which occurs typically graphite, which requires slow precipitation
within milliseconds. The production of of carbon from solution. Back diffusion to
crystalline graphite, however, requires the surface would, of course, take place
highly mobile (i.e. low molecular weight) most rapidly along grain boundaries;' which
species with a long lifetime. could account for their decoration at the
4.6.2 Thin film model. If the final product surface. A serious difficulty arises, however,
of the dehydrogenation of acetylene were a in those cases where grain boundarymigra-
single carbon atom (or pair of atoms) and tion has taken place; and both positions of
if such atomic carbon were mobile, then the grain boundary are decorated..lnview
nucleation (either homogeneous or at of what has been said earlier, it is difficult
preferred sites) followed by growth could to say with certainty which site corresponds
occur, as in the formation of thin metal to the final position of the boundary, but it
films by vacuum deposition. One would is obvious that only one can do so. Yct a
expect atomic carbon of this type to be back diffusion mechanism could only operate
extremely active chemically, resulting in a at the very end of pyrolysis, during the
low surface mobility. That this is not neces- cooling period, leading to decoration at
sarily so is shown by the experiments of only one position of the boundary, vi~ the
Irving and Walker[7], in which carbon was final one.
8 A. E. B. PRESLAND and P. L. WALKER, Jr.

5. CONCLUSIONS 3. Diefendorf R. J.,]. Chim. Phys. 57, 815 (1960).

Single crystal graphite is formed when 4. Pres land A. E. B. and Hedley J. A.,J. Nucl.
Mater. 10, 99 (1963).
acetylene is pyrolyzed over Pt and Ni at 5. Thomas.J. M. and Walker P. L. Jr.,]. Chem.
lOOO°C. Phys. 41, 587 (1964).
Prior thermal annealing of the substrate 6. Banerjee B. C., Hirt T.J. and Walker P. L.Jr.,
influences the nucleation and growth of the Nature 192,450 (1961).
deposit through kink site and step formation. 7. Irving S. M. and Walker P. L. Jr., Carbon 5,
399 (1967).
It is not yet clear whether there is an epi- 8. Karu A. E. and Beer M.,]. Appl. Phys. 37, 2179
taxial relation between overgrowth and (1966).
substrate. 9. Porter G., Combustion Researches and Reviews,
A solution and reprecipitation model for p. 108. Agard (1955).
the formation of graphite in this system is 10. Allpress J. G. and Sanders J. V., Phil. Mag.
9,645 (1964).
untenable. Two possible mechanisms 11. Barrett C. S., Structure of Metals, p. 457.
remain - surface diffusion of atomic carbon McGraw-Hill, New York (1952).
species or polymerization and aromatization 12. Nelson J. P. and Riley D. P., Proc. Phys. Soc.
of short chain hydrocarbons. 57,477 (1945).
13. Handbook oj Thermophysical Properties of Solid
Acknowledgements- We should like to thank the Materials, Vol. I, p. 461. Pergamon Press,
U.S. Atomic Energy Commission for their support Oxford (1961).
on Contract No. AT(30-1)-1710 and also acknow- 14. Freise E. J. and Kelly A., Phil. Mag. 8, 1519
ledge the considerable help we have received (1963).
from Dr. D. Gibbon and the Mineral Constitution 15. Thomas A., Combust.. Flame 6, 46 (1962).
Laboratory, Pennsylvania State University. 16. Digonskii V. V. and Krylov V. N.,]. Appl.
Chem., USSR 33, 725 (1960).
17. Cullis C. F., Presland A. E. B., Read I. A. and
Trimm D. L., 2nd Conf. on Industrial Carbon
and Graphite, p. 195. Society of Chemical
REFERENCES Industry, London (1965).
1. Austerman S. B., Myron S. M. and Wagner 18. Hansen M., Constitution ofBinary Alloys, p. 374.
J. W., CarbonS, 549 (1967). McGraw-Hill, New York (1958).
2. Acheson R. G., U.S. Pat. Nos. 568328,617979, 19. Hansen M., Constitution ofBinary Alloys, p. 377.
645285,711031. McGraw-Hill, New York (1958).