THE KINETICS ON' THE DIAMOND _ OXYGEN R,EACTION
T. Ev¡Ns
Physics
Department,
(Manuscrilit
and C. Pne¡.r,
Uni,uersity
of Reading,
Engblul
received August 8, f 96f )
A method is described by which the oxidation rates of diamond single crystals can be determined.
The etch rates of lov¡ index surfaces have been measured under identical conditions.
Diamonds wero
etched in a fast, flowing stream of oxygen which had been previously purified.
The pressure dependence
of the diamond-oxygen
reaction at, a number of fixed temperatures was studied.
The reaction rates
were found to be linearly dependent, on oxygen preasure for the range of pressures from 5 X l0-2 to
0.5 mm Hg. Experiments for the reaction were carried out in the temperature range from 650.C to
1350'Cand relativeetchratesforthe
(llt), (lI0) and (100) surfaces were determineá throuEhout ühe
pressure and temperature ranges mentioned.
Below 1000'C úhe rate ofetching was slowest oo thu (tO¡)
surface. A graphite layer is formed on the diamond surface during oxidation.
A mschanism for the
diamond-oxygen reaction is proposed which úakes into account the role played by the graphite layer.
The effect of certain impurities on the reaction is also discussed.
I.
INTII,ODUCTION
fn the extensive work which has beendone
on carbon-gas reactions,l the carbon has
rangedin structure from essentiallyamorphous
to single crystals of graphite. A complicating
factor in the interpretation of results is the
porosity of the carbon specimen with the
associated difficulties of determining the real
surface area being attacked and the effect of
diffusion of the reacting gas into and the
reaction products out of the specimen. It
would be expected that the effects ofporosity
could be overcome by studying diamond-gas
reactions using single crystals of diamond.
As will become eyident, this expectation is
only partly fulfilled as a layer of free carbon
sometimes forms on the surface of diamond
during oxidation resulting again in diffusiorr
effects becoming important. Although the
etch features on diamond surfaces as a result
of oxidation have been investigated in some
detailz-s little rvork has previously been d.one
1 P. L. WalkerJr., F. Rusinko, Jr., andL. G. Austin,
'ín Catalgsis,\ol. XI, 133, 1959, for review.
Ad,uance.s
2 A. F. Williams, Genesis oJ the D'iamond (London:
Ernest Benn), 1932.
3 A. R. Patel and S. Tolansky, Proc. Roy. Soc.
4,243,4l (1957).
aM.OmarandM.Kenawi, Phil.Mag.2,859(f957).
5 T. Evans and D. H. Sauter, Phdl,. Mag. 6, 429
(1961')
on kirretics of the diamond-oxygen reaction.
In 1936 Lambert6 reported an actiyation
energy of 43 kcal mole-l for the oxidation
of diamond powder and Baner? investigated
the adsorption properties of diamond powders.
In the work reported here, the oxidation rates
for the (tll) (110)and (100)facesof diamond
single crystals have been measuredat temperatures between 650'C and 1350"C. It will be
shorvn that óhe diamond-oxygen reaction is
not simply oxidation of üamond to gaseous
COz or CO but involves an intermediate
process in which a carbon film forms on the
diamond surface.
II.
I]XPERIMENTAI]
A. Materials
Diamond octahedra with good surfaces
bounded by (lll)
faces are abundantly
found in nature and in the investigation of
this surface, flat triangular plates called
m&cles$'ere used. Crystals bounded by (100)
or (ll0) surfacesare comparatively rare and
rvhen they do occur the surfaces are deeply
pitted and not suitable for reaction rate
t47
6J. D. Lambert, ?rans. Faraday. Soc.32, 452,
1936.
7 A . M . B a n e r , J . C h e m . ¡ S o c . ,1 2 6 I ( 1 9 3 6 ) .
148
FrFTrr cA,RBoNCONFERENCE
(fff), (110) and (100) faces could be determined. As will be mentioned later the surfaces
were frequently covered with a carbon
coating which was removed by heating in a
mixture of perchloric, nitric and sulphuric
acids before the thickness measurements were
taken. At temperatures below 1000'C the
combustion chamber was constructed of silica
B. Apparatus and' Methoil
whilst for higher temperatures recrystallized
In order to minimize diffusion offects in the alumina was used. The temperature range
gas phase and reduce secondary reactions investigated rvas650'C to 1350'C at an oxygen
occurring on the diamond face asfar aspossible, pressure of 0.4 mm Hg. This pressure was
the apparatus was designed to expose the selected as it gave a measurable reaction rate
diamonds to a fast flowing oxygen stream at at the lower temperatures but not too rapid
a reaction at the higher temperatures. In a
reduced pressures.
The stones were placed in a fast florving separate series of experiments the oxygen
(500-1500 cm/sec) stream of oxygen which presslrre was also varied at particular temperatures so as to determine the order of the
had previously been purified by passing it
reactions. When the rate of reaction was too
through a cleaning train containing heatei
palladized asbestos, soda lime, silica gel and small to measure conveniently by the method
magnesium perchlorate. The reduced pressure described, e.9., the rate for the (100) face
w¿s maintainedwith a mercury diffusion pump between 650"C and 700"C, the weight loss per
and the pressure and flow rate could be unit surface area was determined and the
changed by adjusting an irrlet valve and also decrease in thickness calculated.
the pumping speed of the system. The diamonds were mounted in such a, way that the
III.
B,ESULTS
low index faces under investigation were
Initially it was established that within the
parallel to the gas flow. Prior to every run the
range of flow rates used, the rate of reaction
specimens were cleaned by boiling in nitric
acid followed by refluxing in ethyl alcohol. was independent of the flow rate at a particuGlass tweezers were used in the handling of lar pressure. This was shown to occur for
the stones to reduce the possibility of surface oxygen flow rates of between 500 and 1550
cm/sec. No difficultv was experienced in
contamination.
The experimental procedure for a run was to obtaining reproducible results provided the
evacuate the system to 10-6 mm Hg and to reduction in specimen thickness masked any
heat the diamonds in vacuum to the required effect due to localized pitting at such surface
defects as microcracks which were sometimes
temperature. The oxygen was then admitted
for the required time at the required pressure present originally. Ilowever for a particular
and flow rate. At the completion of the run experiment it was found that the previous
the system was evacuated and allowed to cool. history of the stone was important. An
The rate of reaction was determined by untreated smooth stone initially gave a low
measuring the decrease in thickness of the rate of oxidation which rapidly increased to a
diamonds by means of a dial gauge which was constant value as the time of oxidation
accurate to I ¡^c.The mean of twenty readings increased. This constant value was taken as
was taken. As in every case the opposite the rate for the conditions. Microscopic
parallel faces of the specimens were the same examination showed that the initially smooth
low index faces, the reaction rates for the surface was progressively roughened until
experiments. Because of this the specimens
used for the reaction rates of (110) and (f00)
surfaces were polished cubes bounded by
either (100) faces or a mixture of (ll0) and
(100) faces. The surfaces were rvithin one
degree of the low index plane and almost free
of polishirrg marks.
TIIE
KINETICS
OF TIIE
DIAMOND
uniform roughening oocurred at the steady
reaction rate. On the other hand diamonds
which had been oxidized at high temperatures
gave high values when oxidized subsequently
at a lower temperature before again reaching
a steady reaction rate. To minimize such
effects the rates were only measured after the
diamond had been given a preliminary etch
under the particular condition of the experiment.
OXYC}EN IiEACTION
149
e
I
d
E
l¡J
F
É
f
u
F
U
A. Variation of Reaction Rate with Orygen
Pressure
The pressure dependence of the reaction
rates for the three low index faces at 700'C,
850'C and 950'C is given in X'igs. l, 2 and 3
respectively. At all these temperatures the
reaction on all three faces is first order in the
pressure range used. However there rvas a
'E
s
U
t
I
(J
F
lf,l
oto20304(.j-50
O X Y G E N P R E S S U R Em ml O 2H s .
Fro. 3. Etch rate vs. oxygen pressure at 950.C.
difference in the appearance of the faces as
a result of the reaction. At 700"C, the (lf l)
and (tl0) faces were covered wióh a $ey layer
of carbon whereas the (100) face was not
covered. At 850"C and above, all three faces
were coated with a carbon layer. The difference in surface covering is shown in X'ig. 4
which is a diamond cube v'ith four (lt0) faces
and 2 (100) faces after reaction at 830"C in
0.4 mm Hg of oxygen pressure. The four (110)
faces ar:ecovered with an opaque carbon layer
whereas the (100) faces are uncoated.
Experiments were also done at temperatures
between 600"C and 700"C at an oxygen
pressure of one atmosphere in a fast gas flow.
The reaction in this temperature range was
X're. l. Etch rato vs. oxygen pressure for the
three lov¡ index faces aü 700'C.
a
E
l¡J
r
É.
f
U
U
O X Y G E N P R E S S U R En m . H e r l O 2
Frc. 2. Etch rate vs. oxygen pressure at 850oC.
Frc. 4. Diamond block consisting of 4 (f f0)
faces and 2 (100) faces after oxidation at 830'C.
150
FIFTII
CAR,BON
found to be zero order at atmospheric pressure
and the (Ill) surface studied did not have a
carbon layer as it did when the same type of
experiment was done at 0.4 mm Hg pressure.
B. Temperature Depenilenceof the Reacti,on
The variation in the reaction rate with
temperature between 650"C and 1350"C was
investigated at an oxygen pressure of 0.4 mm
Hg. The Anhenius plot of the results is shown
in n'ig. 5, where the solid lines denote the
curves at 0.4 mm Hg. It can be seenthat the
;
o
E
€
t¡l
F
G.
03/r"K
Frc. 5.
Arrhenius plot for the three low index
faces between 650'C and 1350"C.
CONFERENCE
1000"C. Between 650"C and 750"C an activation energy 44 + 2 kcal mole-l was measured
(between C and D in X'ig. 5). Then followed
a decrease in measured activation energy
between D and E to 23 ! 2 kcal mole-1.
Microscopic examination showed that the
surfaces had been attacked uniformly, resulting in a matt type of surfaces.
D. (L00) Iace
Unlike in the caseof (111) and (110) faces,
the (100) face was not always covered by a
carbon layer. Between 650"C and 850'C the
(100) face was completely free of a visible
surface carbon layer. However, above 850'C
a carbon layer appeared on the surface and
becameofappreciablethicknessabove 1000"C.
An activation energy of 55 * 2 kcal mole-l
was measured between 650"C and 750'C
(A to B) for the (100) face, with no carbon
layer being present. The onset of a carbon
layer above 850oCresulüedin a decreaseofthe
measured activation energy to 37 t 2 kcal
mole-l between 850'C and 1000'C.
The type ofattack on the (100)face differed
from that ofthe other two low index surfaces
considered. Between 650'C and 750'C the
attack was primarily by the movement of
steps over the (I00) surface. This is illustrated
in n'igs. 6 and 7 which is a (100) face after
oxidation at 650"C at 0.4 mm Hg for 3f and
8$ hr respectively. It can be seen that the
(100) face is the slowest etching face at
temperatures less than I000"C and that above
1000'C the reaction rates for the three faces
are identical. The three curves also have a
maximum value between 1000"C and 1050"C
before decreasing in reactivity at, higher
temperatures. We shall consider the (f ll)
and (110) faces together and then consider
the (100) face.
C. (111) and,(Ll0) Xaces
At all the temperatures consideredthe (f lf )
and (110)faceshad a thin coating ofcarbon on
their surfaces and this layer became appreciably thicker at temperatures greater than
Fre. 6. (100) surface after oxidation at 650'C
at 0.4 mm Ifg pressure of oxygen for 3| hr.
TIIE KINETICS
ON'THE
DIAMOND
reaction is occurring primarily at steps which
a,ppear to be more kinetically favourable for
attack than the flat surface. For this reason it
would appear that the rate ofreaction for the
(100) face would be more structure sensitive
than the other two low index surfaces as it
would depend upon the density of the source
of steps such as microcracks, dislocations and
possible impurities in the lattice.
Frc. 7. (100) surface after oxidaüion at 6b0"C
aü 0.4 mm IIg pressure of oxygen for 8$ hrs.
E. Reactions at Atmospheric Pressure
The rates ofreaction for the (llt) face were
measured between 600'C and 700'C and are
shown as the dotted curve FG in X\g. b.
Unlike at 0.4 mm Hg pressure, the face was
now clear of any surface carbon layer and the
measured activation energy was 55 ! 2 kcamole-l. A similar dotted curve HI is shown of
the (100) face between 600"C and 750.C at
one atmosphere pressure which again gave an
activation energy of 55 f 2 kcal mole-r. The
surface was again clear of a carbon layer as
was shown to be the casefor the sametemperature range at 0.4 mm Hg.
F. Eflect of Impurities
The effect of impurities on the diamondox¡'gen reaction will now be briefly mentioned.
It would be expected that any impurity which
reduced the oxidation rate of the carbon
layer would red.uce the measured diamond
reaction rate. This is because. as will be
OXYGEN REACTION
r5I
described in the discussion, the diamond.oxygen reaction rate is determined by the
oxidation rate of the overlying carbon layer
when it is present. Water vapot of 2.5o/o
was added to the oxygen and the rates determined at 0.4 mm Hg pressure. The water
vapor retarded the reaction and resulted in
a clear stone at temperatures less than 800.C
for the (lll) face with measured activation
energy of approx. 75 kcal mole-l. The fact
that the face was not coated indicates that
although the water vapor retarded the
carbon layer and oxygen reaction it inhibited
the diamond-oxygen reaction more strongly,
resulting in the rate of removal of the carbon
layer exceeding the production of surface
carbon. This was the reverse ofthe situation
for the (1I I ) surface when reacted with oxygen
alone. Chlorine was also found to have a
strong inhibiting effect on the diamondoxygen reaction.
IY.
DISCUSSION
The presence of a surface carbon layer
cannot be accounted for by a purely physical
phase transformation from the diamond to
graphite structure at the temperatures used
in the experiments. The transition temperature for the transformation at normal pressures
is uncertain, but is certainly higher than the
maximum temperature of l3b0'C used. This
can easily be shown by heating a üamond to
this temperature in a yacuum of better than
l0-o mm Hg when no surface graphite is
formed and no surface attack takes place.
Thus it appearsthat three primary reactions
must be considered in the diamond-oxygen
reaction:
(l) Direct oxidation of diamond by oxygen
to gaseous CO2 and CO. This reaction
has not been shown to occur but the
possibility cannot be ignored. Let this
have a reaction rate .Rr.
(2) Formation of a carbon layer on the
diamond surface. Oxygen is necessary
for the carbon film to form and thus the
T52
T'IFTI{
CABBON CONT'ERENCN
reaction must involve chemisorbed
oxygen, a rearrangement of the surface complexes resulting in a transference of carbon from the diamond
lattice to the fllm' Reflection electron
diffraction ofthe carbon layer gave very
diffuse graphite rings indicating ü
crystallite size of less than 50 A for
a layer formed at 900'C. Let the rate of
carbon layer formation by this reaction
be R2.
(3) Direct oxidation of the carbon layer to
CO2and CO. Let this rate be .83.
If ,82 is greater than R3 a carbon layer will
result as in the case of the (11f) and (ll0)
surfaces when reacted at 0.4 mm Hg pressure.
When -Rzis equal or less than -83, a clear surface will result as for the (100) surface reacted
between 650"C and 850"C at 0.4 mm Hg
pressure and the (1ll) surface reacted at'
atmospheric pressure between 600'C and
700'C. The surface carbon oxiclation has an
oxygen pressure dependence which differs
from that of the diamond-oxygen reaction.
This results in -82 becoming less than 'R3 as
the pressure is raised from 0.4 mm Hg to one
atmosphere causing the (lfl) face to change
from a covered surface to a clear one al
temperatures between 600"C to 700"C' At
the start of a run with a clear stone, when liz
is greater than -83, a carbon layer is formed
and increases in thickness. This makes .B3
larger by increasing the number of active sites
in the layer and also red'ucesR2 by reducing
the amount of oxygen which can diffuse
through the active carbon layer to the underlying diamond surface. \Yhen ,82 equals R3
the carbon layer attains an equilibrium
thickness which remains constant throughout
the run. The attainment of equilibrium
conditions must be extremely rapid as the
experiments have not been able to detect any
variation in reaction rates at different times
of reaction for otherwise identical conditions.
At equilibrium, the reactivity of the carbon
layer controls the diamond-oxygen reactions
as it determines the amount of oxygen which
can diffuse to the diamond surface which
gives the equality between R2 and Rs.
The presenceofthe carbon layer reducesthe
measured activation energy as would be
expected rvhen the diamond surface reaction
is controlled by diffusion of oxygen through
the layer. With the (100)surfaceat 0.4 mm Hg
pressure as the temperature was raised above
850"C the stone becamecoated. Corresponding to the appearanceofthe layer was a reduction in the measured activation energy from
55 to 37 kcal mole-1. Similarly in the 600'C
to 700'C region for the (1ll) surface disappearance of the surface layer as the oxygen
pressurewas increasedfrom 0.4 mm Hg to one
atmosphere resulted in an increaseof measured
activation energy ftom 44 to 55 kcal mole-l.
From the results it is concluded that 55 t2
kcal mole-l is the true activation energy for
the diamond-oxvgen reaction as diffusion
effects would be expected to be negligible due
to the absenceofthe carbon layer and also the
independence of the reaction upon the flow
rate. The other measured"activation energies
with the covered surfaces give an apparent
activation energy which is less than the true
activation energy for the diamond-oxygen
reaction sinceit is affected by diíI'usion through
the over-lying carbon layer.
At first sight it may appear that the activation energy of 44 kcal mole-l measured on the
(111)anct(110)surfacesat 0.4 mm Hg between
650 and 750'C would be the activation energy
for the carbon layer-oxygen reaction itself,
since at equilibrium .B2equals -R3and diffusion
effects may be expected to be small at these
low temperatures. However the following
assumptions would have to be made:
(f ) That the carbon layer thickrressremains
of constant, thickness between 650 and
750'C so as to ensure a constant
quantity of material with which the
oxygen can react. There is no reason
why the thickness should remain constant as the temperature is raised.
(2) That there is no variation in the oxygen
pressure gradient through the layer as
TIIE KINSTICS
OX'THE
DIAMOND
the temperature is changed. This is not
so as the presenceofthe layer has been
seen to reduce the measured activation
energy by limiting the amount of
oxygen which can reach the surface.
(3) That the possible direct oxidation of
diamond to CO and COz does not vary
with temperature.
As these assumptions are not tenable, no
special significance can be placed upon the
measured activation energies when a surface
coating of carbon is present on the diamond
surface.
The maximum in the Arrhenius plot
between 1000and 1050"Cand the subsequent
decreasein the reaction rate at higher temperatures is interesting in that similar effects have
beenobservedinwork on graphite oxidation.8'e
It was also noted that the thickness of the
surface carbon became markedly greater at
1000'C and above. It has been mentioned
that, the diamond reaction is controlled by the
carbonlayer oxidation and ifthere is a decrease
in reactivity of the surface carbon the layer
gets thicker as a consequence and further
inhibits the arrival of oxygen at the diamond
surface. The measured maximum and subsequent decreasein the reaction rate can therefore be accounted for by the reduction in
reactivity at these temperatures of the overlying carbon layer.
It hasbeenseenthatthe (100)faceis a slower
8 X. Duval, J. Chi,m. Phgs. 3 (196f ).
e G. Blyholder, J. S. Binford and H.
J. Phgs. Chem.62,263, (f 958).
Eyring,
OXYGEN REACTION
153
etching face than the (lff) below 1000"C
despite the fact that the measured activation
energiesin the region of 600'C and 700'C are
the same. 'Ihe reason for this can be found in
the type of attack on the two surfaces. In the
case of the (f lf ) face a matt tlpe of equilibrium surface is formed whereas the (f00)
reaction occurs primarily by step movement.
The relative oxidatiorr rates are determined
therefore by the difference in the number of
active sites on the (llt) and (100) faccs. All
surface atoms on the (11f) face are reacting
whereas the reactive sites on the (100) face
are situated primarily at steps on the surface.
The relevance of this work to more usual
carbon-oxygen rate studies is that diffusion
effects cannot be ignored even at as low a
temperature as 650'C. The carbon layer
formed on the (tll) diamond surface at this
temperature is extremely thin and cannot be
measured by v'eight difference. Despite this,
the layer has a marked effect at the diamond
surface due to the oxygen gradient through
the thin layer. This makes one suspect thaü
diffusiorr still has an important effect even
in the so called "chemical region" where, in the
literature, it is assumed that the reactivity of
the carbon controls the reaction rate.
The authors wish to thank Professor R. W. Ditchburn for his interest throughout this work.
Thanks
are also due to Dr. J, F. I{. Custers, Director of the
Diamond Research Laboratory,
Johannesburg for
the supply of stones and also for his interest.
Tho
work has been financed by a grant from Industrial
Distributors (1946) Ltd. which is gratefully acknowIedged.