Beruflich Dokumente
Kultur Dokumente
3, 1971
The kinetics and mechanism of copper oxidation have been measured over the
temperature range 900-1050~ and the pressure range 5 x 10 -3 to 8 • 10-1
arm. It has been shown that, at the pressures lower than the dissociation pressure
of CuO, the oxide scale formed on fiat fragments of the copper specimens is
compact and composed of a single layer, adhering closely to the metallic base.
Growth of the scale proceeds under these conditions by outward diffusion of
metal. The rate of the process under the conditions for which single-phase
scales are formed increases with increasing oxygen pressure according to the
equation :
kp = const p1~3.9
tt
INTRODUCTION
The mechanism of high-temperature oxidation of copper has been a subject
of extensive theoretical and experimental studies. 1-3 Classical works of
Wagner *-v published in the 1930's were concerned with the kinetics and
mechanism of copper oxidation and with physicochemical properties of
cuprous oxide, which was the main product of the oxidation. In the following
* Institute of Solid State Chemistry, School of Mining and Metallurgy, Cracow, Poland.
291
292 Mrowec and Stok]7osa
years studies on the copper-oxygen system were concerned with the elucida-
tion of the kinetics and mechanism of scale formation and the physico-
chemical properties of the two oxides of copper. 1-1~
It has been shown that the scale formed on fiat fragments of the copper
specimens is compact and its growth takes place by outward diffusion of the
metal. ~~ Hence it may be assumed that the defect structure in cuprous
oxide is limited only to the cation sublattice. This conclusion is in full agree-
ment with the results of studies on deviations from stoichiometry in this oxide,
conducted at various temperatures and at various oxygen pressures, and is
confirmed by the results of Hall effect measurements 13,14and by the observed
dependence of the electrical conductivity on the oxygen pressure, s;15'~6
Specifically, it has been proved that cuprous oxide shows deviations from
stoichiometry involving cation defects (Cu2_yO) and that the electrical
conductivity of this oxide increases with increasing oxygen pressure, electron
holes being the current carriers. Cuprous oxide is thus a p-type, metal-deficit
semiconductor, having cationic vacancies and electron holes. These con-
clusions are also confirmed by the results of studies on. diffusion of copper
and oxygen in Cu20, which show that the mobility of the metal in the
cuprous oxide is considerably higher than the mobility of oxygen. 11 17
Although much information has been collected, many problems related to
the mechanism of copper oxidation and physicochemical properties of
Cu20 await solution. In particulaar there exist marked discrepancies between
the results obtained by-~arious authors concerning the dependence of the
oxidation rate on the oxygen pressure 7'~8'~9 and the dependence of defect
concentration on the equilibrium pressure of oxygen. ~6'2~ Accurate data
on the mobility of point defects in Cu20 are also lacking. The only data
available reported by Moore and Seliks~Sn,1~ are limited to one pressure of
oxygen and were subjected to considerable experimental error because of the
very short half-life of 64Cu ('~1/2 ~--- 12 hr).
Because of the discrepancies mentioned above, one cannot formulate
unambiguous conclusions about the defect structure of cuprous oxide and in
particular about the degree of ionization of cation vacancies and their
mobility. Therefore, the mechanism of copper oxidation cannot be described.
The present work is an attempt to explain these problems on the basis of
precise measurements of parabolic rate constants of copper oxidation as a
function of oxygen pressure and temperature.
EXPERIMENTAL
Kinetics Measurement
The kinetics of copper oxidation were studied with the continuous
gravimetric method in an apparatus containing a spiral microbalance, a
Oxidation of Copper at High Temperatures 293
constant
~er
gas o1.1,
8
F
10
conducted in a constant stream of gas, the flow rate of which was varied
between 5 and 15 ml/min. The oxygen pressure ranged from 6.6 x 10 -3 to
0.8 atm and the temperature from 900 to 1050~ The lower limit of the pres-
sure range was set by evaporation of cuprous oxide which under these
conditions began to affect the results of the kinetics. The upper value of the
pressure was determined by the dissociation pressure of cupric oxide (CuO).
The experiments were carried out on spectrally pure copper (Matthey and
Johnson). The flat specimens, 12.5 x 8.00 x 0.45 ram, were cut from sheet
obtained by cold rolling of 5 mm diameter rods of copper. The surface of the
specimens was polished with emery paper through 4/0, and then washed
successively with water, acetone, and methanol. After fixing, the specimen in
the reaction chamber, purified argon was passed through the apparatus. The
reaction chamber was then evacuated down to the pressure of the order of
10-6 mm Hg, which, taking into account the preliminary purification of the
system with argon, enabled the partial pressure of oxygen to be reduced to
about 10- lo atm. Under these conditions the specimen was not oxidized in
the temperature range applied. After the pressure was reduced to 10-6 mm
Hg, the reaction chamber was heated to the desired temperature. After the
desired temperature was attained, the specimen was further heated for 20 min,
and then oxygen was admitted, its pressure and flow rate being controlled by
Alpert-type valves. The oxidation kinetics were followed by registering the
mass gains of the oxidized samples as a function of reaction time. The
required pressure of oxygen was established within 2 min, and this time was
regarded as the zero on the time scale.
I
!
20 30 5O 7O
tmm
que = qo -- P Am (2)
9~here qM~ is the instantaneous surface area, q0 is the initial surface area, p is
a constant. This equation was used to calculate gh. An example of kinetic
curves for copper oxidation at a pressure of 2.6 x 10 -3 a t m is shown in
Fig. 3. Values of the parabolic rate constants obtained are given in Table I.
9 The temperature dependence of the oxidation rate is plotted in Fig. 4 for
various pressures of oxygen. Curves 1-5 correspond to the oxygen pressures
below the dissociation pressure of CuO. These conditions are those for which
single-phase scales of cuprous oxide are formed. Curve 6 corresponds to the
oxygen pressure in air. This latter pressure exceeds the dissociation pressure
of C u O except at 1050~ This curve represents the temperature dependence
2 2
25 50 '75
t m/n
Table I. Parabolic Rate Constants of Copper Oxidation in the Temperature Range 900 1050~
(k~ x 108 g2 cm-4sec-1)
Temperature
(~
900 950 980 1000 1050
Pressure \ .
(Po2 X 103 atm)
r ~
f050 ~ 1000 o 950 ~ 900 *
~za
r "K
Fig, 4, Temperature dependence of the rate of copper oxidation for
several oxygen pressures. Curves i - 5 correspond, respectively, to
6.6 x 1 0 - 3 a t m , 1.3 x 1 0 - 2 a t m , 2.6 x 1 0 - 2 a t m , 3.9 x 1 0 - 2 a t m ,
and 7.9 x 10 -2 atm. Curve 6 corresponds to the oxygen pressure of air.
Oxidation of Copper at High Temperatures 299
//
-7,o ///z////
"t,
Ii ii
g / o o
-~'8
Metallographie Studies
The measurements of oxidation kinetics were supplemented by studies on
material transport through the scale using the marker method. A platinum
r ~
f050 9 fO00 9 950 9 900 9
E]
-0.5
.~ -/.0
-/.5
-/-/0 3 ,/(
T
Fig. 6. Dependence of the dissociation pressure of cupric oxide (CuO)
on temperature. (D, Kubashevski31; A, Landolt32; [], Charotta3;
Z~, Bidwell 3s ; 0 , present work.
Oxidation of Copper at High Temperatures 301
DISCUSSION
The kinetic data obtained allow us to formulate a number of important
conclusions concerning the mechanism of high-temperature oxidation of
copper, the mobility of point defects in cuprous oxide, and the degree of
ionization of cation vacancies in this oxide. It has been found, that under
suitably selected conditions the oxidation of copper produces a single-
layered compact scale adherent to the metal. Scale growth proceeds under
these conditions by outward diffusion of metal. These facts corroborated by
strictly following the parabolic rate law imply that the slowest process
determining the rate of scale growth is diffusion of cation vacancies according
to the theoretical model of Wagner. 37 Good contact of the scale with the
metallic core during the determination of the kinetics and strict parabolic
behavior suggest that, under the conditions described above, there are no
perturbations of local thermodynamic equilibria at either phase boundary.
The results of the kinetics may be thus employed in calculations of self-
diffusion coefficients of copper in cuprous oxide from the general equation of
Wagner 37 :
--.--(a'~/zx ) (3)
kr=ceqj~kl-~D~e + Dx d l n a x
where kr is a rational oxidation constant expressed in gram equivalents per
centimeter per second; Ceq is the concentration of metal ions in gram
equivalents per cm 2 of the scale; z 1 and z2 are the valences of metal and an
oxidant in the reaction product, respectively; a~ and a~ are the thermo-
dynamic activities of the oxidant at the metal scale and scale-oxidant phase
boundaries; and DM, and Dx are the self-diffusion coefficients of metal and
anions in the scale. Since in the case under consideration Dcu >> Do2,17 the
general equation of Wagner can be simplified to the following form :
~ Z1
kr = C~q
~f -- D~e d In a x
k Z2
(4)
302 Mrowec and Stok4"osa
The rational rate constant in Eqs. (3) and (4) may be replaced by the parabolic
rate constant kp expressed in g2 • cm-4 x sec i.
kr = ~Mox(z2/Ax) 2 k~ (5)
Substituting the oxidant activity by its partial pressure, one obtains the
following relation between the parabolic rate constant of oxidation and the
cation self-diffusion coefficient in the scale :
,, A~zl fpk
k p - -2 3 Duedlnpx (6)
V~MeXZ2 p,.
where Pr denotes the oxidant pressure at the metal-scale phase boundary
and Px is the oxidant pressure at the outer phase boundary. To integrate
Eq. (6) we should know a priori the dependence of self-diffusion coefficient of
metal in the scale on the oxidant pressure. Fueki and Wagner 39 have proposed
the solution of this problem which does not require the knowledge of the
dependence of DMe on the oxidant pressure. They drew attention to the fact
that Pr is, at a given temperature, independent of the oxidant pressure.
According to the assumptions of the Wagner theory,37 a local thermo-
dynamic equilibrium is attained at the metal-scale phase boundary, and
hence the partial pressure at this phase boundary is very close to the dissocia-
tion pressure of the scale. The above-mentioned assumption of the Wagner
theory was proved in a number of experiments.4~ Taking this fact into
account we may integrate Eq. (6) to give:
zt~ Ax ] ~dtnpx]p.
In the case of copper oxidation to Cu20, Eq. (7) assumes the form :
I dk"
Dc~ = 1.818| ---'~p | (8)
ld log Po2]v,
To calculate the self-diffusion coefficient of metal in the scale, it is
necessary to determine the dependence of the oxidation rate on oxygen
pressure. Having determined this function, one can calculate the value of
DM~by finding the differential of the function for a given pressure. Fueki and
Wagner 39 performed these calculations for the Ni-NiO-Oz system and
Mrowec, Walec, and Werber 42 for the Co-CoO-O 2 system. The results
reported were in agreement with the data obtained with the use of radio-
isotopes of nickel and cobalt. Analogous agreement was also found by
Mrowec 43'4~ for the Fe-FeS-S2 system. Equation (8) has been employed in
the present work for calculations of the self-diffusion coefficients of copper
in Cu20 at selected oxygen pressures and at a temperature of 1000~
Oxidation of Copper at High Temperatures 303
A2xzl D n ~-I/n
kp- iv~__3 o ~ex - p ~ l , ) (11)
ghieXZ2
r162
The above equation may be simplified, if the measurements of kp were
performed at pressures considerably above the oxidant pressure at the
Me-MeX boundary (Px >> Pr) :
t! --
A~-zl
kp ~2 3 Donpx/" (12)
VMeXZ2
An analogous equation was reported much earlier by Wagner 37 in a
discussion of the possibilities of simplifying the solution of his general
equation. However the simplification used by Wagner has a form identical to
Eq. (9), and hence it does not define the proportionality coefficient of this
equation. This excludes the possibility of employing this relation for the
calculation of other physicochemical parameters, in particular in determin-
ing D o and 1In.
Because the kinetics of copper oxidation were determined at pressures
considerably higher than the oxygen pressure at the C u - C u 2 0 phase
boundary, the function
log kp = f(log Po2) (13)
was obtained by the least-squares method, and then the values of D o and
1/n were calculated from the constants obtained. Substituting these data
304 Mrowec and Stokt'osa
Self-diffusion coefficient
Pressure Dcu x lOs cm z sec-I
Poz • 103
(atm) Values derived with equations
may thus conclude that the Fueki-Wagner method is more general, and the
results may be equally accurate when the measurements are carried out over a
suitably wide pressure range. If this is not possible, however, the method
proposed by the present authors appears to yield more accurate results.
To compare the results of calculations of Dcu with the values obtained
by Moore and Selikson 11 by radioisotope methods, the results of the present
work were extrapolated to an oxygen pressure of 1.3 x 10 . 4 atm making
use of the equations given in Table III. Moore and Selikson determined the
diffusion coefficients of copper in C u 2 0 in the temperature range 800-
1050~ at only one value of oxygen pressure (1.3 • 10 -4 atm). In Fig. 7 the
results of the present work and the data obtained by Moore and Selikson are
presented graphically. As already mentioned, the results obtained with the
radioisotope method show considerable scatter of the experimental points
which may be due to serious experimental difficulties resulting from the
short half-life of radiocopper. As seen from the plot, the results of the present
r'c
i
0
-7.6
8 o .
g o o
8 ~
O~
176 ~gO I
O,84 I
O,00
~. . 10 3 oK
Fig. 7. The temperature dependence of the self-diffusion coefficient of copper at Po2 = 1.3 •
10 -'~ atm. Open circles denote the experimental values obtained from radioisotope measure-
ments; open triangles are the values extrapolated from the results of the present work.
306 Mrowec and Stok]'osa
T ~
1060 9 tO00 * 950 ~ 900 o
-7,0
-7,~
-76
-7,8
i -- i __ i
~TG 0,80 O.B/~
-lr.lOJ oE
Fig. 8. The temperature dependence of the self-diffusion coefficient of
copper in Cu;O. Curves 1-5 correspond, respectively,to 6.6 x 10-3 atm,
1.3 x 10 2atm, 2.6 x 10-2atm, 3.9 x 10-2atm, and 7.9 x 10-Zatm.
Oxidation of Copper at High Temperatures 307
The exponent in Eq. (14) is very close to 1/4, implying that the cation vacancies
are practically nonionized. The latter conclusion follows from considerations
on the defect equilibria in the crystal lattice of cuprous oxide. 45'46
This conclusion is in agreement with the results of investigations on the
rate of copper oxidation as a function of oxygen pressure, shown in Fig. 5,
which is the basis for calculations of self-diffusion coefficients of copper in
cuprous oxide. It has been found that the pressure dependence of the para-
bolic rate constant is described by
kp = const p ~ (15)
where the exponent 1/n varies, depending on the temperature, between
1/3.9 and 1/4. An analogous dependence was obtained for the self-diffusion
coefficient of copper in C u 2 0 (Table IV and Fig. 8).
Bearing in mind the fact that the exponent of Eq. (10) has a value close
to 1/4, the general relationship between the coefficient of self-diffusion of
copper in C u 2 0 and temperature and the oxygen pressure may be described
as
t
kp = (p + 1)Die (19)
I
f
I
/
I
/I
/
I
/
I"
/
//
/
111 "+
x
xo
CONCLUSIONS
The results of studies on the kinetics and mechanism of copper oxidation,
under conditions for which the formation of a single-phase scale of cuprous
oxide forms, allow us to formulate the following conclusions.
1. The oxide scale on copper, formed over a range of oxygen pressures
below that of the dissociation pressure of CuO, is compact and adheres
310 Mrowec and Stokt~osa
closely to the metal on the flat portions of the sample. Growth of the scale,
consisting solely of Cu20, occurs by the outward diffusion of the metal. The
rate-determining step of oxidation is the diffusion of cation vacancies in the
scale.
2. When interpreting kinetic data, one should take into account the
changes in surface area of the metal during oxidation. Neglecting this factor
may give rise to erroneous conclusions, with respect to the rate law followed
and the absolute values of the rate constants of this process.
3. Under the conditions for which CuzO is the sole product of oxidation,
the rate of copper oxidation increases with an increase in oxygen pressure
according to the assumptions of t h e Wagner theory. It follows from this
dependence that in the temperature range 900-1050~ and at pressures
close to 1 atm, cuprous oxide contains mainly un-ionized cation vacancies.
This conclusion is confirmed by the results of calculations of Dco performed
with the use of Eqs. (8), (12), and (19).
4. The activation energy of copper oxidation when a single-phase
scale forms is different than that observed when a duplex scale forms (Table
II). This difference is due to the dependence of the rate of growth of C u 2 0
phase on oxygen pressure.
5. The self-diffusion coefficient of copper in cuprous oxide as a function
of oxygen pressure and temperature was derived from kinetic data by the
Fueki-Wagner method as well as by the new method proposed in the present
work. The agreement between the results obtained with these two methods is
good. This agreement is observed also between calculated results and data
obtained by M o o r e and Selikson who used radioisotopes.
6. The defect concentration gradient in the growing scale was analyzed
by the Pettit equations. This distribution was linear (Fig. 9). This enabled
us to employ the modified M o t t - G u r n e y equation [Eq. (19)] to calculate the
self-diffusion coefficients of copper in C u 2 0 as a function of temperature and
oxygen pressure. The agreement between values obtained in these calcula-
tions and those derived by other methods (Table IV) enabled the degree of
ionization to cation vacancies in C u 2 0 to be determined.
REFERENCES