The isotopic exchange reaction of oxygen on metal oxides

Article abstract of DOI:10.1006/jcat.1998.2377

Two methods have been used to determine the individual rate constants of the isotopic exchange reaction of oxygen on metal oxides. The best way to determine the three rate constants is to use the kinetic model of Klier and co-workers and fit the equations

Full text of DOI:10.1006/jcat.1998.2377

Journal of Catalysis 182, 390–399 (1999)
Article ID jcat.1998.2377, available online at http://www.idealibrary.com on
The Isotopic Exchange Reaction of Oxygen on Metal Oxides
1
C. Doornkamp, M. Clement, and V. Ponec
Leiden Institute of Chemistry, Leiden University, Gorlaeus Laboratories, P.O. Box 9502, 2300 RA Leiden, The Netherlands
E-mail: knegtel@chem.leidenuniv.nl
Received July 16, 1998; revised December 3, 1998; accepted December 10, 1998
1
the surface oxygen of the metal oxide (R -mechanism).
Two methods have been used to determine the individual rate
1
8
O2(g) + Os → O O(g) + 18Os
16
18 16
constants of the isotopic exchange reaction of oxygen on metal ox-
ides. The best way to determine the three rate constants is to use
the kinetic model of Klier and co-workers and fit the equations to
the experimental data. The method of Tsuchiya and co-workers can
be used to check the results obtained by the method of Klier and
—the oxygen molecule exchanges both its atoms with the
surface oxygen of the metal oxide (R -mechanism).
2
0
1
2
co-workers. The three rate constants, R , R , and R , of the three
different exchange mechanisms are determined for the period IV
metal oxides at various temperatures. The rate constants are cor-
related with parameters characterising the oxide, i.e. the position
of the metal element in the periodic table and the average metal–
18
O2(g) + 2 Os → O2(g) + 2 18Os
16
16
The isotopic exchange reaction was extensively studied
in the 1960s by Klier and co-workers (3–8), Winter (9, 10),
and Boreskov and co-workers (11–18). These authors used
different methods to determine the rate constants of the dif-
ferent mechanisms of the exchange reaction. Klier and co-
workers showed that it was possible to determine the three
different rate constants by fitting their kinetic model (inte-
gral equations) to the experimental data. However, since
the 18O concentration in these experiments was low (ap-
oxygen bond strength.
�c 1999 Academic Press
Key Words: oxygen isotopic exchange; metal oxides; activity pat-
tern; metal–oxygen bond strength; Tammann temperature.
INTRODUCTION
One of the essential functions of the catalyst in oxidation proximately 18% ), these fits were not very accurate (3–8).
reactions is the activation of oxygen. The way in which oxy- Winter, who also used kinetic equations to derive the rate
gen is activated is a crucial problem in oxidation catalysis. It of exchange, determined only the exchange rate of the main
is generally accepted that lattice oxygen plays an important mechanism (9, 10). Boreskov and co-workers determined
role in selective oxidation reactions. However, there is still the exchange rates by following the 16O O concentration
a debate about the role of various adsorbed oxygen species with time over an oxide whose oxygen had been enriched
18
to the same concentration of 18O as the gas phase. In this
in oxidation reactions (1, 2).
The interaction of molecular oxygen with a metal oxide case the overall rate constant K, being the sum of the three
0
1
2
can be investigated with the oxygen isotopic exchange re- individual rate constants R , R and R , was determined.
action. From the course of the oxygen exchange reaction, Boreskov and co-workers also determined the exchange
conclusions can be drawn about the way in which oxygen rate of 18O with the oxide’s oxygen starting with the nat-
takes part in the catalytic processes.
In the presence of an oxide there are three possible ways R is the sum of the individual rate constants R and R
for an oxygen molecule to exchange its atoms:
(11–18).
ural isotopic composition of the oxide. This rate constant
1
2
Another method to determine the rate constant of the R2
mechanism was shown by Tsuchiya et al. (19) in the hydro-
gen–deuterium exchange reaction, since with the hydro-
gen–deuterium exchange reaction on hydrides or hydrogen
containing catalysts in general, the same three types of
mechanisms as described above can take place. Therefore,
this method can also be used for determination of the R2
rate constant of the isotopic oxygen exchange reaction.
To summarise, the rate constants of the isotopic oxygen
exchange reaction published so far are either the overall
—
the oxygen molecule exchanges one of its atoms with
another oxygen molecule from the gas phase, without the
participation of the oxide’s oxygen (R -mechanism).
0
18
16
18 16
O2(g) + O2(g) → 2 O O(g)
the oxygen molecule exchanges one of its atoms with
—
1
To whom correspondence should be addressed.
390
0021-9517/99 $30.00
Copyright �c 1999 by Academic Press
All rights of reproduction in any form reserved.

ISOTOPIC EXCHANGE REACTION OF OXYGEN ON METAL OXIDES
391
rate constants as determined by Boreskov (18) or some in-
dividual rate constants which were determined with a lim- and co-workers to determine the individual rate constants
ited accuracy (9). in the isotopic exchange reaction, two different oxides were
To test the methodsof Klier and co-workers and Tsuchiya
Since it is very interesting to establish the exact contri- used, V2O5 and Fe2O3. Because of large differences in the
bution of the three exchange mechanisms, a good method activity of these oxides in the exchange reaction, the ex-
is desirable to determine all the individual rate constants. change reaction on Fe2O3 was investigated in a different
In the first part of this paper, we have used the methods of temperature range than on V2O5. For Fe2O3 the tempera-
Klier et al. (3) and Tsuchiya et al. (19) and made an attempt ture range under investigation was between 623 and 823 K
to determine all three rate constants with higher accuracy and for V2O5 between 743 and 823 K.
than achieved earlier. In the second part we determined the
Commercially available V2O5 (BDH) was used. Fe2O3
individual rate constants for the period IV metal oxides at was prepared by thermal decomposition of Fe(OH)2, ob-
various temperatures and correlated these values with pa- tained by precipitation of the corresponding metal nitrate
rameters characterising the oxide, i.e., the position of the at pH 9. The hydroxide was decomposed in air at 773 K.
metal element in the periodic table and the average metal– X-ray diffraction was used to check the structure of Fe2O3.
oxygen bond strength. This will be done having in mind the
potential impact of this correlation on the theory of activity Isotopic Exchange Reaction on the Period IV Metal Oxides
and selectivity of metal oxides in oxidation reactions.
The exchange reaction on the period IV metal oxides
was carried out at various temperatures. The temperature
range was dependent on the activity of the oxide under in-
vestigation. Since the difference in activities of the oxides in
EXPERIMENTAL
the exchange reaction is quite large, the temperature ranges
varied with the metal oxide. The amount of oxide which was
used for an experiment was also chosen according to the ac-
tivity of the metal oxide. The temperature and amount of
oxide were chosen in such a way that within 1–10 h, equilib-
rium between gas phase and oxide had been reached. The
total oxygen pressure was always 0.9 mbar.
The characteristics of the period IV metal oxides are
shown in Table 1. NiO was prepared by thermal decom-
position of Ni(NO3)2 at 773 K for 18 h. For the other period
IV metal oxides commercially available oxides have been
used as supplied.
The heat of formation, 1Hf, of the oxide has been used
as a measure for the average M–O bond strength (20). The
M–O bond strength is defined hereby as the heat of forma-
tion of the metal oxide divided by the number of oxygen
atoms contained in the oxide (1Hf/O atom).
Testing the Method of Klier et al. and Tsuchiya et al.
The isotopic exchange reactions were carried out in a
static apparatus at low pressures (in the range of mbar).
The concentration of all three isotopes in the gas phase was
followed in time by a mass spectrometer (VG instruments
8
–80 mm) connected with the apparatus via a continuous
leak. The pressure decrease was negligible during the exper-
iment. The reactor was made of quartz and had a volume
of 912 ml. All experiments have been performed at a to-
tal pressure lower than 1 mbar to minimise mass transport
problems.
Two types of exchange experiments were performed. Ei-
ther the exchange reaction was started with an equilibrium
mixture of all the dioxygen isotopes or with a nonequilib-
16
18
rium mixture of only O2 (Messer Griesheim) and O2
Isotec Inc. 99.1% 18O). The equilibrium mixture of all the
dioxygen isotopes was prepared by introducing a 1 : 1 mix-
(
16
18
ture of O2 and O2 into a vessel with a sealed-in platinum
wire, which was heated to approximately 1300 K. In both
TABLE 1
Characteristics of the Period IV Metal Oxides
M–O bond
types of experiments, the 16O2, O O, and O2 concentra-
tion was followed with time and the total dioxygen pressure
was in both cases 0.9 mbar.
16
18
18
Before each experiment the oxides were treated in a vac-
Surface area
strength
(kcal/mol) point (K)
Melting
�
3
2
uum (10 mbar) at the temperature of the exchange reac-
Oxide Precursor Supplier
(m /g)
tion for 3 h. After treatment in a vacuum, the oxides were
equilibrated for 3 h with 16O2 at a pressure of 0.9 mbar and
V2O5
BDH
BDH
Merck
Baker
BDH
Baker
Merck
AZL
5.1
2.6
2.9
22.9
7.3
7.0
5.3
5.1
74.1
90.8
62.2
65.7
53.3
57.3
36.6
83.2
963
Cr2O3
MnO2
Fe2O3
2539
at the temperature of the reaction.
Since water and CO2 can have an influence on the oxy-
gen isotopic exchange reaction, the water vapour and CO2 Co3O4
pressure have been monitored by the mass spectrometry
during the standard treatment of the oxides. At the end of
this treatment these pressures were at the noise level and
do not have any influence on the exchange reaction.
808a
Fe(NO3)
Ni(NO3)
1838
1223
2257
1599
2248
NiO
CuO
ZnO
a
Decomposition at 808 K.

3
92
DOORNKAMP, CLEMENT, AND PONEC
RESULTS
Two methods have been used for determining the rate
constants of the three mechanisms of the isotopic exchange
reaction. These methods depend on the initial composition
of the gas phase. The first method (the method of Tsuchiya
et al. (19)) can be used if the experiment is started with
an equilibrium composition of the gas phase. The second
method (the method of Klier et al. (3)) can be used with any
kind ofinitialcomposition ofthe gasphase (equilibrium and
nonequilibrium composition). With the first method only
2
one rate constant (R ) can be determined, while with the
second method all three rate constants can be determined.
Method of Tsuchiya et al.: Starting with an Equilibrium
Composition of the Gas Phase
16
18
2
16
18
If the quantity 9 is defined as 9 = [ O O] /[ O2][ O2]
and the gas phase composition is at equilibrium, the quan-
tity 9 is equal to the equilibrium constant 9eq, which is 4.
When the exchange reaction is started with this equilibrium
0
1
composition and onlythe R or R mechanismsoperate then
2
d9/dt = 0. However if the R mechanism operates as well,
then d9/dt
6= 0. In the case that the oxide only consists of
2 1 0
16
O and when R � R or R , the initial decrease of 9 is
given by
�
�
d ln 9
1
2
=
� R z
,
[1]
dt
t=0
FIG. 1. The dependence of 9 versus time on (a) V2O5 at 823 K and
(
b) Fe2O3 at 743 K.
with z the amount of 16O2 molecules in the system at t = 0
or
1
8
O2(g) + Os → O O(g) + 18Os
16
18 16
�
�
R1
d ln 9
2
16
18 16
16
16
18
R = � ( O2)
.
[2]
O O(g) + Os → O (g) + Os
2
dt
t=0
1
8
O2(g) + 2 Os → O2(g) + 2 18Os
16
16
18
16
2
Thus, if the R mechanism is large enough, it will always
18 16
O O(g) + 2 Os → O2(g) + Os + 16Os R2
18
18
change the equilibrium composition. This means that the
18
O O(g) + 2 Os → O (g) + O + 16Os
16
18
18
2
2
s
rate constant for the R mechanism can be determined from
the slope of ln 9 with time. An example of the decrease of
1
8
Klier and co-workers showed that the number of O atoms
(
9
versus time is shown in Fig. 1a for V2O5 at 823 K.
w) in the gas phase is given by
Figure 1b shows the dependence of 9 versus time for
Fe2O3 at 743 K. The quantity 9 decreases with time and
�
�
� �
2
a + m
2
1
reaches 4 again after 200 min. The return of 9 to the equi- w = w∞ + (w0 � w∞) exp � (2R + R )
t . [3]
2am
librium value is much faster for Fe2O3 than for V2O5.
The symbols used here are the same as used by Klier et al.
3), whereby m is the total number of exchangeable oxygen
atoms in the oxide, a the total number of oxygen molecules,
Method of Klier et al.: Starting with a Nonequilibrium
Composition of the Gas Phase
(
1
2
1
2
The second method to determine the individual rate con- and R and R the rate constant for the R and R mecha-
stants is by fitting the kinetic equations as described by Klier nism, respectively. By fitting this equation to the time de-
18
and co-workers to the experimental data. The model of pendence of the concentration of O in the gas phase, the
2
1
Klier is based on six different reactions in which the ex- total exchange rate (2R + R ) with the oxide can be de-
change leads to a change in the isotopic composition:
termined. The program used for fitting Eq. [3] to the ex-
perimental data performed a least square fitting using a
modified Powell algorithm.
18
16
18 16
R0
O2(g) + O2(g) → 2 O O(g)

ISOTOPIC EXCHANGE REACTION OF OXYGEN ON METAL OXIDES
393
FIG. 3. Best fit of Eqs. [3] and [4] to the time dependence of the partial
1
8
pressure of O on Fe2O3 at 743 K.
which describe the 16O2, 16O O, 18O2, and 18O concentra-
18
tion versus time were used to fit the time dependence of all
18
oxygen isotopes. Fitting only the time dependence of O2
and 18O gives more then one solution for the best-fit pa-
rameters, depending on the input parameters. Thus, in this
way only local minima were found with the least square
fitting. By using all kinetic equations the global minimum
0
1
2
could be found resulting in one solution for R , R , and R
independent of the input parameters.
In Tables 2 and 3 the rate constants of the exchange reac-
tion on Fe2O3 and V2O5 at various temperatures, determi-
FIG. 2. Best fit of Eq. [3] to the time dependence of the partial pres- ned by the two described methods, are shown. Both tables
1
8
sure of O on (a) V2O5 at 773 K and (b) Fe2O3 at 743 K.
2
show that the R rate constants determined by the method
of Klier et al. and Tsuchiya et al. agree reasonably well.
In Figs. 2a and 2b the best fit of Eq. [3] for the 18O con-
centration over, respectively, V2O5 and Fe2O3 is shown. The
best fit is represented by the continuous line. In the case of
V2O5 Eq. [3] fits the data very well, but this is not the case
for Fe2O3.
Oxygen Exchange Reaction on the Period IV Metal Oxides
The three rate constants of the oxygen exchange reac-
tion on the period IV metal oxides were determined by
using the modified method of Klier et al. The results are
shown in Fig. 4. In Fig. 4a an Arrhenius plot of the exchange
According to the model of Klier et al. the concentration
18
of O in the gas phase (w) should be constant after equilib-
rium between the gas phase and the oxide has been reached.
With Fe2O3 the 18O concentration still decreases, also when
equilibrium (9 = 4) has been reached. The apparent equi-
librium concentration of 18O (w∞) is dependent on time.
Therefore, a time dependence of w∞ was introduced into
the model of Klier et al. in the form of
TABLE 2
Rate Constant for the Oxygen Exchange Reaction on V2O5 De-
termined by the Method of Klier et al. and Tsuchiya et al. at
Various Temperatures
V2O5
w∞ = w∞0 exp(� kwt).
[4]
Equilibrium
Nonequilibrium
Temp.
K)
Tsuchiya
R (mol/min/m )
Klier
R (mol/min/m )
Klier
R (mol/min/m )
In Fig. 3 the best fit is shown by combining Eqs. [3] and
[
kinetic equations fit the data very well.
2
2
2
2
2
2
(
4] for the 18O concentration over Fe2O3. In this case the
�
�
�
8
8
9
� 8
� 8
� 9
� 8
� 8
� 9
823
03
743
5.6 10
2.6 10
2.5 10
6.8 10
2.2 10
2.5 10
5.9 10
2.2 10
1.6 10
8
With the kinetic equations as described by Klier and co-
1
8
18
workers, fitting the time dependence of O2 and O should
0
1
be sufficient to determine the three rate constantsR , R and
Note. The use of initial equilibrated or nonequilibrated dioxygen mix-
R ofthe exchange reaction. However, the kinetic equations tures is indicated.
2

3
94
DOORNKAMP, CLEMENT, AND PONEC
TABLE 3
Rate Constants for the Oxygen Exchange Reaction on Fe2O3
Determined by the Method of Klier et al. and Tsuchiya et al. at
Various Temperatures
mechanisms. The oxides which show activity in all three ex-
change mechanisms are MnO2, Fe2O3, Co3O4, CuO, NiO,
and ZnO. Their activities as functions of temperature are
shown in Figs. 4c to 4h, respectively.
Figure 4 also shows that with increasing temperature the
1
Fe2O3
increase in activity is steeper for the R exchange mecha-
2
nism than for the R mechanism. In other words, the acti-
Equilibrium
1
vation energy for the R exchange reaction is higher than
Nonequilibrium
2
for the R reaction. The only exception to this rule is ZnO.
Tsuchiya
mol/min/m
Klier
mol/min/m
Klier
mol/min/m
2
2
2
The activation energies for the exchange reaction for the
period IV metal oxides are shown in Table 4.
Temp.
K)
(
R1
R2
R1
R2
R1
R2
In the case of Co3O4, NiO, and CuO a decrease in ex-
�
7
7
7
8
� 6
� 7
� 7
� 8
� 7
� 7
� 7
� 8
1
2
7
7
7
6
73
43
43
73
—
—
—
—
5.9 10
2.0 10
4.3 10
7.3 10
1.2 10
2.0 10
2.9 10
2.8 10
7.9 10
2.0 10
3.6 10
9.7 10
—
—
change rate for the R an R mechanisms is observed from a
�
�
�
� 7
� 7
� 8
� 7
� 7
� 7
2.3 10
1.4 10
1.7 10
1.8 10
2.9 10
1.1 10
certain temperature onwards. Thisdecrease in activitystarts
1
for the R exchange reaction at a higher temperature than
2
for the R exchange reaction. In the temperature region
Note. The use of initial equilibrated or nonequilibrated dioxygen mix- where a decrease of exchange with the oxygen of the oxide
0
tures is indicated.
is observed, the R mechanism starts to become the main
mechanism.
reaction on V2O5 is shown. In the case of V2O5 the exchange
reaction only proceeds via the R mechanism. In Fig. 4b the change activity of MnO is 673 K. Above 673 K MnO starts
The maximum temperature possible to measure the ex-
2
2
2
isotopic exchange reaction on Cr2O3, is shown. On this ox- to decompose and an increase ofdioxygen pressure isobser-
1
2
ide, oxygen is exchanged via the R and the R exchange ved during the measurement.
FIG. 4. Arrhenius plot of the isotopic oxygen exchange reaction on (a) V2O5, (b) Cr2O3, (c) MnO2, (d) Fe2O3, (e) Co3O4, (f) NiO, (g) CuO, and
(
h) ZnO.

ISOTOPIC EXCHANGE REACTION OF OXYGEN ON METAL OXIDES
395
FIG. 4—Continued
In the previous paragraph it was shown that the model of peratures are shown in Fig. 5. From these Arrhenius plots
Klier et al. had to be completed by an additional term, which an apparent activation energy for diffusion in the oxide can
takes into account diffusion of labelled oxygen from the sur- be calculated. The results are shown in Table 5. However,
face into the bulk of the oxide, since this took place during in the case of V2O5 no diffusion rate could be detected.
the exchange measurements. By fitting the corrected model The apparent diffusion parameters can only be determined
to the experimental data not only the rate constants for the
three exchange mechanisms could be determined, but also
an apparent diffusion parameter k. The apparent diffusion
parameters for the period IV metal oxides at various tem-
TABLE 4
Activation Energies of the Isotopic Exchange Reaction
0
1
2
R mechanism
R mechanism
R mechanism
Oxide
(kJ/mol)
(kJ/mol)
(kJ/mol)
V2O5
Cr2O3
MnO2
Fe2O3
Co3O4
NiO
—
—
150
—
170
389
307
145
—
211
111
83
130
36
149
148
151
160
117
190
103
210
258
108
FIG. 5. Apparent diffusion rates as a function of temperature deter-
mined by fitting the corrected model of Klier et al. to the experimental
data.
CuO
ZnO

3
96
DOORNKAMP, CLEMENT, AND PONEC
TABLE 5
TABLE 6
Apparent Activation Energies for Diffusion
Oxygen Exchange Activtiy of Some Period IV Oxides after Two
Different Pretreatments
Apparent activation energy
Pretreatment Aa
Pretreatment Bb
Oxide
for diffusion (kJ/mol)
Oxide
T (K)
R0
R1
R2
R0
R1
R2
Cr2O3
MnO2
Fe2O3
Co3O4
NiO
66
65
95
V2O5
Cr2O3
Fe2O3
Co3O4
CuO
773
773
713
548
598
0
0
0
21
0
0
0.78
2.6
9.9
0
0
0.39
0.32
74
2.3
0
2.3
16
33
16
0.84
3.0
15
3.6
14
1.1
6.3
3.5
5.1
66
107
116
114
CuO
ZnO
8.2
a
Evacuating for 3 h and equilibrating for 3 h at Treaction.
Evacuating for 3 h at Treaction (rate constants in 10 mol/m min).
b
16
2
after equilibrium between the gas phase and the first two
layers of the oxide has been reached. In the case of V2O5 change activity after standard pretreatment are shown in
no equilibrium was reached during the time of the experi- Table 6.
If the oxides are not equilibrated with 16O , a higher ex-
ment, since with this oxide it takes a very long time before
2
the equilibration is accomplished.
change activity is observed for all three exchange mecha-
In Fig. 6 the exchange activities of the period IV metal nisms. When only pretreatment in a vacuum was carried
0
oxides at one temperature (573 K) are shown. Almost all out, oxides which had no exchange activity in the R mech-
1
0
oxides of the period IV metal oxides show activity in the R
anism at a certain temperature showed some R exchange
2
and the R exchange reaction at 573 K. The only exception is activity at this temperature.
V2O5 as was already shown by Fig. 4a. At 573 K most oxides
2
1
show a higher activity in the R mechanism than in the R
mechanism. The only oxides which show a higher activity in
DISCUSSION
1
2
the R compared to the R mechanism are Co3O4 and ZnO. Method to Determine the Rate Constants of the Isotopic
The oxides which show the highest activity in the exchange
reaction are MnO2, Co3O4, and CuO. Figure 6 also shows
Exchange Reaction
The two methods suggested by Klier et al. (3) and
Tsuchiya et al. (19) were used to determine the rate con-
stants in the oxygen exchange reaction on metal oxides.
With the method of Tsuchiya et al. only the rate constant
1
that the activity pattern for the R exchange mechanism and
2
for the R exchange mechanism are quite similar.
To test the influence of the pretreatment of the oxides
on the exchange activity of the three mechanisms, the ex-
change reaction was performed on V2O5, Fe2O3, CuO, and
NiO after evacuating these oxides only at the reaction tem-
perature for 3 h. Thus, before the exchange experiment was
started, these oxides were not equilibrated with 16O2. The
exchange activity of these oxides compared with the ex-
2
for the R mechanism can be determined, while with the
0
1
2
method of Klier et al. three rate constants (R , R , and R )
are accessible. In Tables 2 and 3 the rate constants of the ex-
change reaction over Fe2O3 and V2O5 determined by both
methods are shown and compared, and the values for the
2
R mechanism agree reasonably well. Since both methods
differ substantially, the method of Klier et al. is an integral
technique, while the method of Tsuchiya et al. is a differen-
tial technique; this agreement shows that the theory of Klier
et al. is a good method to determine the three different rate
constants in the exchange reaction. An additional advan-
tage of the method of Klier et al. is that one gets all three rate
constants in one experiment. The method of Tsuchiya et al.
can be used to check the results obtained by the method of
Klier et al.
However, the kinetic equations of Klier et al. did not fit
our data exactly. While according to their model the con-
centration of 18O established after equilibrium had been
reached should be constant, our experimental data showed
a continuous decrease in time. This is because equilibrium
has been reached first only between the gas phase and the
first one or two layers of the oxides, and thereafter a slow
FIG. 6. Activitiesfor the period IV metaloxidesin the oxygen isotopic
exchange reaction at 573 K.

ISOTOPIC EXCHANGE REACTION OF OXYGEN ON METAL OXIDES
397
diffusion of 18O into the bulk of the oxide takes place. In is not more then one order of magnitude. At low tempera-
2
the literature a more complete description of both oxygen tures the exchange activity of the R mechanism is higher
1
exchange and diffusion can be found (21). Using the equa- and at higher temperatures the R mechanism shows the
tions of Klier and Kucera, the exchange rates and diffusion highest activity. An exception to this rule is V2O5, since no
1
coefficient can be determined simultaneously. However, we R exchange activity is observed. This lack of activity can be
have chosen a more empirical approach and added an extra addressed to a difference in the degree of participation of
term to Eq. [3]. The reader should notice that this is only a lattice oxygen. In the case of V2O5 all oxygen layers of the
small correction as can be seen from Figs. 2b and 3. When oxide take part in the exchange reaction while in the case
higher oxygen pressures are used in the oxygen exchange of the other oxides only one to two layers do so. All the
reaction, such as by Klier et al., a contribution of this size is layers can take part because diffusion of oxygen in V2O5
not visible.
is much faster compared to other oxides. This was already
18
The diffusion of O into the bulk of the oxide is tempera- mentioned by J ı´ ru and Boreskov (4, 18).
ture dependent and increases with increasing temperature.
The literature suggests that there are two kinds of R2
This temperature dependence suggests a real relationship exchange mechanisms: an exchange with oxygen from the
between the real diffusion coefficient and this apparent dif- lattice and the so-called place exchange (23). The place ex-
fusion constant. The rate constants of Eq. [4], characteris- change is just a displacement of a preadsorbed O2 molecule
ing the diffusion of 18O into the bulk of oxide at different by another gas molecule. This mechanism does not involve
temperatures, are shown in Fig. 5. In the case of V2O5 no dif- scission of any O-O bond. With place exchange an increase
18
fusion of O into the bulk could be monitored, because no in exchange rate is expected with a higher concentration
equilibrium between the gas phase and the first one or two of adsorbed oxygen molecules. However, Table 6 shows a
2
layers of the oxide has been reached, which has been the higher activity in the R exchange reaction if the oxides are
case with the other oxides. In the case of V2O5 all oxygen pretreated in a vacuum and not in oxygen. Since after pre-
16
layers take part in the exchange reaction simultaneously, treatment in a vacuum less O2 isadsorbed on the surface of
as was already mentioned by J ´ı ru and Nov a´ kov a´ (4). This the metal oxide compared to the situation after a pretreat-
oxide behaves as a well-mixed liquid.
ment in oxygen, such an increase in exchange rate suggests
that the prevailing mechanism for the R exchange cannot
be a place exchange but is a real exchange with oxygen of
the lattice.
2
1
2
R and R Mechanisms
The Arrhenius plots of the exchange reaction on the pe-
riod IV metal oxides show that at low temperatures the ex-
change reaction basically proceeds via the R mechanism.
0
R Mechanism
2
This means that at low temperatures it is easier for an ox-
At elevated temperatures almost all oxides show a high
2
0
ide to activate molecular oxygen via the R mechanism. At exchange rate by the R mechanism. In this mechanism oxy-
0
high temperatures on the other hand, the R mechanism is gen is probably activated by metal ions at the surface of the
the most important mechanism. This is in contrast to our oxides. At elevated temperatures the steady state of the
expectations since it is generally accepted that with oxida- surface of the oxides after pretreatment in oxygen is more
tion reactions a Langmuir–Hinshelwood-type mechanism reduced than at lower temperature. Apart from a higher
(
i.e., a mechanism without participation of lattice oxygen) number of vacancies, which leads to a higher activity in the
1
2
proceeds at low temperatures, while at higher temperatures R and R exchange reaction, reduced metal atoms could
the Mars–van Krevelen mechanism should prevail. Look- also appear at the surface. If the exchange reaction is per-
ingat oxidation reactionsover metaloxides, different mech- formed on a metal surface, as on Pt(18), then only the R0
anisms or forms of O2 activation are distinguished, depend- mechanism can be observed. From the literature it is also
ing on the temperature of the reaction. At low temperature known that CuO and NiO can be reduced or decomposed
the oxidation reaction can takes place by the Langmuir– to Cu(0) and Ni(0) at conditions used here (24). This is
Hinshelwood mechanism. At a temperature between about in agreement with Table 6. When the oxides are only pre-
4
73 and 873 K almost all oxidation reactions take place treated in a vacuum the surface of the oxide is on average
via the Mars–van Krevelen mechanism (also called a redox more deeply reduced compared to pretreatment in oxygen.
mechanism). This is the case not only of total oxidation re- A higher degree of reduction results in more vacancies and
actions, but also of selective oxidation reactions. At higher a higher exchange rate.
temperatures a combination of surface reaction with rad-
ical formation at the surface occurs, with radicals reacting exchange and hetero-exchange (12). It turned out that for
further in the gas phase (22). almost all oxides which had been heated in oxygen before
Boreskov compared the activities of the so-called homo-
1
Figures 4b to 4h show a relationship between the R and the exchange experiment, the rates of homo- and hetero-
2
the R mechanisms. For all the period IV metal oxides the exchange were very similar. The results represented here
0
difference in exchange rate between those two mechanisms show that most oxides show the R activity only at elevated

3
98
DOORNKAMP, CLEMENT, AND PONEC
temperatures. Since the homo-exchange rate is the sum of
all three exchange mechanisms and the hetero-exchange
1
2
rate the sum of R and R activity, this agreement in ex-
change rates is not very surprising.
Correlation with Parameters Characterising the Oxide
1
2
In Figs. 6 to 9 the rate constants of the R and R mech-
anisms are correlated with the position of the metal ele-
ment in the periodic table, the average metal–oxygen bond
strength, and the Tammann temperature.
2
1
Figure 6 shows the same activity pattern for the R and R
mechanism. Thus, the individual rates have the same saw-
tooth-like pattern as the total exchange activity, determined
1
earlier by Boreskov. Although in the case of the R mech-
FIG. 8. Logarithm of the total activity of the R1 and R2 exchange
anism one oxygen vacancy is needed and in the case of the reaction at 573 K as a function of the Tammann temperature.
2
R mechanism two oxygen vacancies are required there is
no difference between those two mechanisms in the sense
mechanisms strongly and similarly depend on the average
suggested by Maltha and Ponec (25). As previously men-
metal–oxygen bond strength.
1
2
tioned, the mechanisms R and R are related with each
other, whereby the difference in activity is not much more
than one or two orders of magnitude. This explains proba-
bly why both mechanisms depend in the same way on the
average metal–oxygen bond strength as shown in Fig. 7.
In this graph the right y-axis represents the reciprocal of
the mean M–O bond strength (notice the higher the M–O
bond strength the lower the activity). A similar pattern for
the reciprocal M–O bond strength is observed as for the
Since diffusion is an additional factor in the exchange
reaction, as can be concluded from Fig. 5, a correlation be-
tween activity in the exchange reaction and the Tammann
temperature can be expected. The Tammann temperature
is defined as the temperature of reaction divided by the
melting temperature of the oxide. The temperature of re-
action is 573 K and the melting temperatures of the oxides
are shown in Table 1. Figure 8 shows the logarithm of the
sum of the exchange rates of the R and R mechanisms
plotted versus the Tammann temperature of the reaction.
A Tammann temperature between 0.2–0.5 means that
surface diffusion already takes place. Above 0.5, bulk dif-
fusion occurs as well. This is the case for vanadium oxide.
When surface diffusion takes place, which is the case of the
other oxides, the plot as in Fig. 8 shows a higher exchange
rate for a higher Tammann temperature of the reaction.
A similar relationship is observed if the apparent diffusion
rates at 573 K is compared with Tammann temperature,
which is shown in Fig. 9.
1
2
1
2
R and R mechanisms, which suggests that both exchange
1
2
FIG. 7. Oxygen exchange rates of the R and R mechanism compared
FIG. 9. Apparent diffusion rate k at 573 K as a function of the
to the M–O bond strength for the period IV metal oxides.
Tammann temperature.

ISOTOPIC EXCHANGE REACTION OF OXYGEN ON METAL OXIDES
399
This study has been performed to correlate the results
A linear relationship between the total exchange activ-
1
2
of the oxygen isotopic exchange reaction with those on the ity (R + R ) with the oxide and the Tammann tempera-
oxidation reactions in a more convincing way than in the ture shows that the surface diffusion of oxygen is somehow
paper of Maltha and Ponec (25). A detailed comparison of related to the exchange activities by these two exchange
this kind will be presented in the next paper (26). This paper mechanisms. The exchange rate and surface diffusion de-
will show that the same activity pattern as in Figs. 6 and pend on the same features of the solid state.
7
is observed with a great number of oxidation reactions.
This activity pattern is not only observed for total oxidation
reactions, but also for the selective oxidation of allyl iodide
to acrolein.
ACKNOWLEDGMENTS
The authors are thankful to Dr. F. Kapteijn for supplying the software
and to Dr. J. Verouden for editorial advice.
CONCLUSION
REFERENCES
The methods of Klier et al. and Tsuchiya et al. have
been used to determine the individual rate constants of the
isotopic exchange reaction of oxygen on metal oxides. The
best way to determine the three rate constants is by using
the kinetic model of Klier et al. and fit all the equations to
the experimental data. The method of Tsuchiya et al. can be
used to check the results obtained by the method of Klier
et al.
Since diffusion of labelled oxygen into the bulk of the
oxide also takes place during the exchange reaction, the
kinetic model of Klier et al. had to be adjusted. An empir-
ical term was added to this model. By fitting the modified
model to the experimental data, not only the three different
rate constants could be determined, but also the apparent
diffusion parameter.
1
2
. Bielanski, A., and Haber, J., “Oxygen in Catalysis,” Dekker, New
York, 1991.
. Satterfield, C. N., “Heterogeneous Catalysis in Industrial Practice,”
McGraw-Hill, New York, 1991.
. Klier, K., Nov a´ kov a´ , J., and J ´ı ru, P., J. Catal. 2, 479 (1963).
. J ´ı ru, P., and Nov a´ kov a´ , J., Coll. Czech. Chem. Commun. 28, 1 (1963).
. Nov a´ kov a´ , J., and J ´ı ru, P., Coll. Czech. Chem. Commun. 29, 1114
(1964).
3
4
5
6
7
8
. Nov a´ kov a´ , J., J ı´ ru, P., and Klier, K., Coll. Czech. Chem. Commun. 33,
1
338 (1968).
. Muzykantov, V. S., J ´ı ru, P., Klier, K., and Nov a´ kov a´ , J., Coll. Czech.
Chem. Commun. 33, 829 (1968).
. Nov a´ kov a´ , J., Catal. Rev. 4, 77 (1970).
9. Winter, E. R. S., J. Chem. Soc., 2889 (1968).
0. Winter, E. R. S., Adv. Catal. 10, 196 (1958).
1
1
1. Dzizyak, A. P., Boreskov, G. K., and Kasatkina, L. A., Kinet. i Katal.
3, 81 (1962).
The rate constants for the oxygen exchange reaction were
determined for the period IV metal oxides at different
temperatures. The results show that with the exception of
1
1
2. Boreskov, G. K., Adv. Catal. 15, 285 (1964).
3. Muzykantov, V. S., Popovskii, V. V., and Boreskov, G. K., Kinet. Katal.
5, 624 (1964).
4. Muzykantov, V. S., Panov, G. I., and Boreskov, G. K., Kinet. Katal. 14,
948 (1973).
1
1
1
1
V2O5 the exchange reaction always proceeds with the R
2
and R mechanisms operating or with all three exchange
5. Muzykantov, V. S., Panov, G. I., and Boreskov, G. K., Kinet. Katal. 10,
2
mechanisms. On V2O5 only the R mechanism occurs. At
1047 (1969).
2
low temperatures the most important mechanism is the R
6. Muzykantov, V. S., Kinet. Katal. 6, 952 (1965).
mechanism, while at elevated temperatures the R mech- 17. Boreskov, G. K., and Muzykantov, V. S., Ann. N.Y. Acad. Sci. 213, 137
0
(
1973).
anism becomes the prevailing one. On V2O5 and Cr2O3,
0
18. Boreskov, G. K., Catal. Sci. Tech. 3, 39 (1982).
no exchange activity by the R exchange mechanism is ob-
1
2
9. Tsuchiya, S., Ponec, V., and Sachtler, W. M. H., J. Catal. 22, 280 (1971).
0. Weast, R. C. (Ed.), “Handbook of Chemistry and Physics,” 63rd ed.,
CRC Press, Boca Raton.
served.
1
2
It seems that the R and R exchange mechanisms are
closely related to each other. The difference in activity be- 21. Klier, K., and Kucera, E., J. Phys. Chem. Solids 27, 1087 (1966).
2
2
2
2. Golodets, G. I., Stud. Surf. Sci. Catal. 55, 693 (1990).
3. Che, M., and Tench, A. J., Adv. Catal. 32, 1 (1983).
4. Moulijn, J. A., van Leeuwen, P. W. N. M., and van Santen, R. A., Stud.
Surf. Sci. Catal. 79 (1993).
tween those two mechanisms at a certain temperature is not
more than one or two orders of magnitude. They both show
the same activity pattern for the period IV metal oxides and
are both in the same way dependent on the average M–O
bond strength.
2
5. Maltha, A., and Ponec, V., Catal. Today 17, 419 (1993).
26. Doornkamp, C., Clement, M., and Ponec, V., unpublished results.