Kinetic study of heterogeneous oxygen-exchange reactions and bulk self-diffusion of oxygen

Article abstract of DOI:10.1021/jp981191j

Apparent activation energies are determined for oxygen-isotope exchange reactions between O₂, CO, or NO and preoxidized or prereduced CaO surfaces. Oxygen exchange between N¹⁶O or C¹⁶O and isotope labeled Ca¹⁸O surfaces produced an apparent activation energy of 2 kcal/mol. Similar values are obtained for single-isotope-exchange between ¹⁸O¹⁸O and prereduced Ca¹⁶O surfaces. Apparent activation energies of 15-18 kcal/mol are found for single and double exchange between ¹⁸O¹⁸O and preoxidized Ca¹⁶O surfaces, as well as for double exchange between ¹⁸O¹⁸O and prereduced Ca¹⁶O surfaces. The low apparent activation energies are believed to result from adsorbed intermediates, whereas the high values may involve the formation of a singlet O₂ transient. It is shown that eventually the self-diffusion of oxygen ions in the bulk becomes the rate-determining step in the isotope-exchange reaction. Apparent activation energies are determined, and the values are found to depend on surface treatment, (i.e., 44 kcal/mol under reducing and 78 kcal/mol under oxidizing conditions). The involvement of oxygen vacancies under reducing conditions is discussed.

Full text of DOI:10.1021/jp981191j

5158
J. Phys. Chem. B 1998, 102, 5158-5164
Kinetic Study of Heterogeneous Oxygen-Exchange Reactions and Bulk Self-Diffusion of
Oxygen
Filip Acke* and Itai Panas
Department of Inorganic Chemistry, Chalmers UniVersity of Technology and Go¨teborg UniVersity,
S-412 96 Go¨teborg, Sweden
ReceiVed: February 18, 1998; In Final Form: April 22, 1998
Apparent activation energies are determined for oxygen-isotope exchange reactions between O2, CO, or NO
and preoxidized or prereduced CaO surfaces. Oxygen exchange between N16O or C16O and isotope labeled
Ca18O surfaces produced an apparent activation energy of 2 kcal/mol. Similar values are obtained for single-
isotope-exchange between 18O18O and prereduced Ca16O surfaces. Apparent activation energies of 15-18
kcal/mol are found for single and double exchange between 18O18O and preoxidized Ca16O surfaces, as well
as for double exchange between 18O18O and prereduced Ca16O surfaces. The low apparent activation energies
are believed to result from adsorbed intermediates, whereas the high values may involve the formation of a
singlet O2 transient. It is shown that eventually the self-diffusion of oxygen ions in the bulk becomes the
rate-determining step in the isotope-exchange reaction. Apparent activation energies are determined, and the
values are found to depend on surface treatment, (i.e., 44 kcal/mol under reducing and 78 kcal/mol under
oxidizing conditions). The involvement of oxygen vacancies under reducing conditions is discussed.
1. Introduction
followed by the formation of a bridging bond with an 16O of
the CaO surface. No activation energy for the oxygen-isotope
exchange was reported here, but a value can be calculated from
the work of Cunningham and Healy.14 An Arrhenius plot based
on the initial reaction rates for heterophase oxygen-isotope
exchange on powders, outgassed at different temperatures,
predicted apparent activation energies of about 12 kcal/mol. The
authors proposed a mechanism on the basis of an exchange with
hydroxyl groups as the most likely pathway for heterophase
isotope exchange at 623 K. Mechanisms for oxygen-isotope
exchange based on adsorption have also been suggested for CO
and NO as probe molecules. Huzimura et al.8 investigated the
isotopic exchange of oxygen between a CO adsorbate and a
MgO surface at temperatures between 100 and 800 K. Proof
of the CO bond breakage was provided at temperatures as low
as 200 K. This is rather surprising considering the high binding
energy in a CO molecule (about 11 eV). A mechanism was
proposed on the basis of CO molecules being trapped on low
coordinated surface-oxygen sites. The oxygen atom of the CO
molecule was suggested to enter into an oxygen vacancy. This
understanding implies an adsorbed intermediate and the oxygen
exchange to be a concerted reaction driven by thermal heat.
Support for this understanding is found in several infrared
spectroscopy studies which report a significant number of CO
species adsorbed on alkaline earth oxides at temperatures
between 77 and 300 K.9-11 The results presented by Yanag-
isawa,15 who studied the NO interaction with thermally activated
CaO and SrO surfaces using a temperature-programmed
desorption technique, confirm the importance of adsorbed
species in the isotopic exchange of oxygen. Adsorbed NO
species were observed at temperatures as high as 773 K, and
three different sites for oxygen-isotope exchange were pro-
posed. Infrared spectroscopy was used to detect adsorbed
NO molecules at temperatures between 40 and 298 K,16
and adsorptions on both lattice oxygens and cations were
suggested.
Mechanistic studies of chemical and physical processes at
surfaces of alkaline earth oxides have led to an increased
understanding of the reduction of NO by CO, H2, and
hydrocarbons and the oxidative coupling of methane over these
materials.1-6 Involvement of lattice oxygens has been suggested
through a surface-oxygen abstraction step by the reducing
species. This was confirmed by a recent study which investi-
gated the reduction and reoxidation of CaO surfaces by CO and
H2 as reducing agents and NO as oxidant.7 The production of
a CO2 transient was observed and a correlation was shown
between the amount of CO2 formed and the amount of NO
reduced. These processes are understood to involve an inter-
mediate electron accepting capacity, a property enhanced by
the presence of alkali metals.2-6 It has been suggested that
semiconductor models2,3 and models based on F-type centers4,5
describe the fate of the excess electrons. In the former model,
alkali metal ions were introduced to provide impurity levels in
the band gap, while in the latter they were introduced to stabilize
the F-type centers.
The purpose of the present study is to clarify further the
essential properties of the chemically active substrate. This is
achieved by isotope-exchange experiments, a commonly em-
ployed technique to determine reaction pathways and identify
reactive sites. In this work, kinetic parameters are determined
for the oxygen-isotope exchange between 18O18O and preoxi-
dized or prereduced Ca16O surfaces, and between C16O or N16O
and preoxidized Ca18O surfaces. The results for the oxygen-
isotope exchange between O2 and preoxidized CaO surfaces are
consistent with the data available in the literature.
Isotope exchange between an O2 adsorbate and CaO surfaces
was studied by Yanagisawa et al.12,13 The authors suggested it
occurred by adsorption of an O2 molecule at an O-vacancy,
* Corresponding author: tel, + 46 (0)31 772 28 86; fax, + 46 (0)31
772 28 53; e-mail, filip@inoc.chalmers.se.
S1089-5647(98)01191-2 CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/11/1998

Oxygen Exchange Reactions
J. Phys. Chem. B, Vol. 102, No. 26, 1998 5159
Chang et al.17 studied the oxygen-isotope exchange between
gas-phase molecular oxygen and Li-doped MgO lattice oxygens
in an attempt to characterize the reactive sites for the oxidative
coupling of methane. The amount of oxygen exchanged
corresponded to approximately 20 atomic layers. Similar
amounts were observed by Peil et al.18 The exchange reaction
became dependent on the self-diffusion of oxygen in the bulk.
Measuring the isotope exchange at different temperatures
allowed them to determine the apparent activation energy for
this process (i.e., 63.5 kcal/mol for a MgO bulk).19 This
apparent activation energy falls in the range, 30-85 kcal/mol,
as reported in the literature for temperatures below 1300 °C.20,21
The present work reports the effects of surface pretreatment
on the apparent activation energy for single-oxygen-isotope
exchange between O2 and CaO. Surface preoxidation resulted
in a significantly higher apparent activation energy than for
prereduced CaO. Comparisons of the apparent activation
energies for oxygen-isotope exchange between C16O or N16O
and preoxidized Ca18O surfaces will be made. The asymptotic
features in the latter studies are exploited to determine the
apparent activation energy for self-diffusion of bulk oxygen.
3. Results
The initial rate of oxygen-isotope exchange will be shown
to be controlled by the actual isotope-exchange reaction. Once
the outermost oxygen layers of the CaO have been exchanged,
the rate becomes controlled by the self-diffusion of oxygen ions
in the CaO bulk. Results for the oxygen-isotope-exchange
reactions are presented for the complete investigated time
interval and subsequently used to determine apparent activation
energies for the isotope-exchange reactions and the self-diffusion
of oxygen.
3.1. Oxygen Isotope-Exchange Experiments. The follow-
ing reactions were studied:
18O18O + 16Os T 18O16O + 18Os
18O18O + 216Os T 16O16O + 218Os
(R1a)
(R1b)
N16O + 18Os T N18O + 16Os
C16O + 18Os T C18O + 16Os
(R2)
(R3)
2. Experimental Section
A fixed bed reactor connected to a quadropole mass spec-
trometer (QMG 421C-3 of Balzers) was used for the experi-
ments. The quartz reactor (i.d. 22 mm, length 500 mm) had an
asymmetric construction to avoid heating the upper metal fitting
and the vacuum tight Viton O-ring. Gas sampling was carried
out 2 mm under the bed-supporting sintered quartz filter, using
a quartz capillary. The temperature was measured 3 mm under
the capillary tip with a K-type thermocouple. This position of
the thermocouple avoids any interference in the results due to
its catalytic activity. The temperature was controlled with a
second K-type thermocouple in contact with the heating coil
and connected to a Eurotherm temperature controller.
The bed material investigated consisted of 0.250 g CaO
(Fisher Scientific) mixed with 1.000 g of quartz sand, pro analysi
(Merck). The quartz sand was added to reduce the pressure
drop in the bed. The bed was calcined at 1073 K in an Ar
flow until adsorbed CO2 and H2O species were evacuated (after
about 1 h), and the bed was then alternatively exposed to 18O18O
and C16O or N16O. The reaction products, 16O18O, 16O16O,
C18O, C18O16O, C18O18O, and N18O, formed during the experi-
ments, were analyzed by means of a mass spectrometer. A
labeled oxide film, Ca18O, was formed during exposure of the
initial Ca16O material to 18O18O at 1073 K, enabling investiga-
tions of the temperature dependence on the oxygen exchange
between CO or NO and Ca18O. This initial step was repeated
after every exposure to CO or NO in order to obtain constant
surface and bulk properties. The same procedure was followed
after exposure of the Ca16O surface to 18O18O at different
temperatures. An intermediate exposure to CO or NO (reduced
or oxidized surfaces respectively) at 1073 K was done in order
to obtain a similar Ca16O material before each 18O18O exposure.
The kinetics of the respective isotope-exchange reactions were
determined by evaluating the initial exchange rate constants for
different temperatures in Arrhenius plots.
The subscript s indicates a surface oxygen. Reactions R1a and
1b describe the oxygen isotope exchange between 18O18O and
R
prereduced or preoxidized Ca16O surfaces (section 3.1.1), and
R2 and R3 describe the oxygen-isotope exchange reactions
between Ca18O surfaces and N16O or C16O (section 3.1.2).
3.1.1. Isotope Exchange between 18O18O and Preoxidized
or Prereduced Ca16O. The dependence of the surface pretreat-
ment on the isotope exchange between 18O18O and Ca16O
surfaces was investigated, taking into consideration both oxidiz-
ing and reducing conditions. Preoxidized surfaces were ob-
tained after repeated intermediate exposure of the surface to
N16O at 1073 K for 45 min. Figures 1a and b show, for
temperatures between 706 and 873 K, the oxygen-isotope
exchange between 18O18O and these surfaces. Figure 1a shows
the 16O16O production, corresponding to the double-exchanged
oxygen (reaction R1b). An initial increase in 16O16O concentra-
tion is observed for all three temperatures tested. The initial
slopes increase with increasing temperature. A distinct maxi-
mum can be observed for the double-exchange reaction,
becoming even more pronounced with increasing temperature.
The qualitative effects were similar for the single-oxygen
exchange (reaction R1a, Figure 1b). Characteristic differences
between single and double exchange are the significantly larger
amounts of single-exchanged oxygen and the systematically
higher initial slopes for the single-exchanged oxygen produc-
tion. Both figures include the feed signal for the 18O18O
concentration as a measure of the time delay due to the
experimental set up.
Oxygen isotope exchange between prereduced CaO surfaces
and O2 was also investigated. Reduced surfaces were obtained
by repeated intermediate exposure of the surface to C16O at 1073
K for 45 min. A transient CO2 formation, containing all three
isotopes (C18O18O, C16O18O, and C16O16O), was observed during
the initial CO exposure. Figures 2a and b show the results of
the oxygen-isotope exchange for temperatures between 668 and
817 K. The single-oxygen-exchange reaction (reaction R1a)
highlights the main difference between the exchange reactions
over preoxidized and prereduced surfaces (i.e., the initial slope
in the 16O18O concentration is independent of temperature). The
temperature effect on the double-oxygen exchange (reaction R1b)
over prereduced surfaces is similar to that for preoxidized
The gases used were 5000 ppm NO, 5000 ppm CO, both in
Ar, and 98 atom % 18O18O. The gases were mixed with pure
Ar, using mass flow controllers (series 5800 of Brooks), to a
flow of about 25 mL/min and concentrations of about 2500 ppm
for CO and NO, and 7200 ppm for 18O18O. The background
due to Ar on the m/e signal 36 of the mass spectrometer was
subtracted for all experiments.

5160 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
Figure 2. Oxygen-isotope exchange between 18O18O and prereduced
Ca16O surfaces at 668, 718, and 738 K. The mass spectrometer m/e
signals 32 (16O16O) and 34 (16O18O) are shown in a and b, respectively.
Figure 1. Oxygen-isotope exchange between 18O18O and preoxidized
Ca16O surfaces at 706, 773, and 873 K. The mass spectrometer m/e
signals 32 (16O16O) and 34 (16O18O) are shown in a and b, respectively.
surfaces (i.e., a temperature-dependent initial slope and an
increase in the maximum 16O16O formation with increasing
temperatures).
3.1.2. Oxygen Isotope Exchange between Ca18O and N16O
or C16O. CaO surfaces were alternately exposed to 18O18O at
1073 K for 15 min and to N16O at temperatures between 573
and 1073 K. This treatment resulted in a fully oxidized material
at all times. The oxygen-18 treatment was carried out in order
to label the oxygens in the CaO matrix. An 18O concentration
gradient from the surface into the bulk can be expected.
Oxygen-isotope exchange between the labeled Ca18O surface
and N16O (reaction R2) was observed for all tested temperatures.
Figure 3 shows both the isotope exchanged N18O and the N16O
feed concentrations. The initial slope in N18O concentration
was independent of temperature, whereas the maximum in N18O
production increased with increasing temperature. The N16O
feed is also included in Figure 3 to give an indication of the
response time of the experimental set up.
Figure 3. Oxygen-isotope exchange between N16O and a Ca18O surface
investigated in the temperature range 573-1073 K.
with the relations
[N16Of] - [N16O] ) [N18O]
The above experiment with NO was repeated, replacing NO
with CO. Figure 4 shows the oxygen isotope exchange between
C16O and Ca18O (reaction R3). Similar results as for the N16O/
Ca18O system were obtained (i.e., the amount of oxygens
exchanged are a function of temperature), whereas the initial
slopes were insensitive to temperature. The decreasing part of
the C18O and N18O concentration curves (Figures 3 and 4) are
discussed in detail under section 3.2.2, where modeling of
oxygen self-diffusion was attempted.
3.2. Apparent Activation Energies. 3.2.1. Apparent Ac-
tiVation Energies for Oxygen Isotope Exchange. The mass
balances for reactions R1a, R1b, and R2 over preoxidized CaO
surfaces are straightforward. The loss of reactant corresponded
to the amount of isotope exchanged molecules formed in accord
and
[18O18Of] - [18O18O] ) [18O16O] + [16O16O]
where f stands for feed conditions. The single (R1a) and double
(R1b) oxygen isotope exchange between 18O18O and prereduced
surfaces are, however, affected by an initial reoxidation. This
loss of 18O18O was compensated for in the analysis by taking
the sum of all O2 species as the feed (i.e., [18O18O] + [18O16O]
+ [16O16O]). Similarly, an initial loss was also observed for
CO, since CO reduced the surface by forming CO2. This effect
was again accounted for in the mass balance for reaction R3 by
taking the sum of [C16O] and [C18O] as the C16O feed
concentration, [C16Of].

Oxygen Exchange Reactions
J. Phys. Chem. B, Vol. 102, No. 26, 1998 5161
Figure 4. Oxygen-isotope exchange between C16O and a Ca18O surface
at 573, 773, and 1073 K.
The apparent activation energies for the actual isotope
exchange can be determined from the initial slopes in a
concentration-time diagram. Rate constants were computed
assuming a second-order reaction proportional to [NO/CO/O2]‚
[Os] where [Os] is the concentration of 16Os for reactions R1a
and R1b and 18Os for reactions R2 and R3. The surface-oxygen
concentration was treated as a constant as long as the slope
remained constant. The above assumption of a pseudo-first-
order reaction is similar to the one proposed by Lizuka et al.22
and was shown to be efficient in the interpretation of the oxygen-
isotope exchange between CO2 and MoO3. The rate constant
was evaluated by solving the material balance, treating the
reactor as an integral reactor
Figure 5. Arrhenius plots for the oxygen-isotope exchange for (a)
single- and double-oxygen-isotope exchange between 18O18O and both
prereduced and preoxidized Ca16O surfaces and (b) oxygen exchange
between N16O or C16O and Ca18O surfaces.
TABLE 1: Apparent Activation Energies (Ea) and
Preexponential Factors (A) for the Examined
Isotope-Exchange Reactions
Ea (kcal/mol)
A
k ) (F/(Ctot COsW)) ln(yf/y)
(1)
O2
single exchange
preoxidized
prereduced
double exchange
preoxidized
prereduced
where F is the molecular flow, Ctot the total concentration, COs
the surface-oxygen concentration, W the amount of catalyst, and
y the molar ratios of the respective species tested. The surface-
oxygen concentration was estimated by treating the CaO grains
as cubes with a total surface area of 0.9 m2 (determined by
BET). Evaluations of the rate constants in Arrhenius plots for
oxygen-isotope exchange between 18O18O and Ca16O and
between N16O or C16O and Ca18O are made in Figure 5a and b,
respectively. Apparent activation energies and preexponential
factors for the respective reactions are collected in Table 1.
Apparent activation energies in two significantly different
ranges were found. Low values are seen for the oxygen-isotope
exchange between C16O or N16O and the Ca18O surfaces, as
well as for single-oxygen-isotope exchange between 18O18O and
prereduced Ca16O surfaces. This is contrasted by double-
oxygen-isotope exchange over prereduced Ca16O surfaces and
both single- and double-oxygen-isotope exchange over pre-
oxidized surfaces. Apparent activation energies of about 17
kcal/mol are observed for these. Note also the effects on the
preexponential factor. The value for single-oxygen-isotope
exchange over preoxidized surfaces was about twice as large
as that for the double exchange. Prereduction of the surface
resulted in an increased preexponential factor for the double-
oxygen-isotope exchange reaction. This trend can also be
observed from the relative positions of the straight lines in the
Arrhenius plots (Figure 5a). No significant difference was
observed between NO and CO (Figure 5b).
16.6
1.8
15.1
2.8 × 10-3
17.7
16.1
9.5
30.4
NO
CO
2.2
3.0
6.1 × 10-3
1.1 × 10-2
second law for a semi-infinite body and constant partial
pressures of 16O/18O at the gas-solid interphase produces23
C(x,t) ) Cs erfc[x(4Dt)-1/2
]
(2)
where Cs is the surface-oxygen concentration in mol/m3 and D
is the intrinsic diffusivity expressed in m2/s. After a certain
time, depending on probe molecule and temperature, the oxygen-
isotope exchange rate is determined by the rate of self-diffusion
of 18O/16O from the bulk to the surface, or equivalently, of 16O/
18O transport from the surface into the bulk. At a certain time
and for a certain position, the atomic flux in the bulk becomes
a function of the concentration gradient as described by Fick’s
first law:
J(x,t) ) -D dC(x,t)/dx
(3)
where J is the flux of oxygen ions (mol/m2 s), D the diffusivity,
and C the concentration as defined in eq 2. Differentiation of
eq 2, and calculating the flux J according to eq 3 at the surface,
for t sufficiently large gives
3.2.2. Apparent ActiVation Energies for Self-Diffusion of
Oxygen: A Model. The concentration profile of 16O in the bulk
of Ca18O, as well as of 18O in the bulk of Ca16O, are functions
of the distance to the surface (x) and time. Solving Fick’s
J(0,t) ) Cs (2/π1/2)1/(4Dt)1/2
(4)
where Cs (2/π1/2) has a value of 0.112 mol/m3.

5162 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
TABLE 2: Values for the Parameters r, â, r1, â1, and γ for
the Functions Describing the Oxygen Isotope Exchange for
NO (Eq 5) and CO (Eq 6) as Probe Molecules
NO and CO
COinitially
â1
temperature
(K)
R
â
R1
γ
573
706
773
873
1073
Ea (kcal/mol)
10 129 -227.35 16 610 -437.53 1.9357 × 10-6
14 161 -287.71
17 030 -253.25 25 183 -574.51 1.891 × 10-7
22 308 -238.96
33 175 -197.91 33 489 -380.13 1.2236 × 10-9
74.3
44.9
Figure 7. The effect of temperature on the shape of the normalized
weight function.
Contrary to the case of NO, no acceptable fit to the simple
model could be obtained for CO as the probe molecule. This
is not surprising as exposure of the preoxidized surface to CO
has been shown to result in surface-oxygen abstraction and
formation of vacancies. Initially, the self-diffusion of oxygen
should be controlled by a vacancy-induced mechanism, which
is expected to have a substantially lower activation energy than
in fully oxidized substrates. Even for a diffusion mechanism
involving vacancies, Fick’s laws should be obeyed and propor-
tionality of J to t-1/2 should prevail. Eventually, the vacancies
are expected to have diffused into the bulk and the self-diffusion
of oxygen will be similar to that found for NO as the probe
molecule. Therefore, the tails of the oxygen-isotope exchange
curves should be modeled by a sum of two functions, each of
the form 5, and weighted according to
J(0,t) ) [exp(-γ t2)J1(0,t)] + [(1 - exp(-γt2))J2(0,t)] (6)
where
J1(0,t) ≈ R1/t1/2 + â1
J2(0,t) ≈ R/t1/2 + â
Figure 6. Comparison between the employed model for the self-
diffusion of oxygen and the experimental points for (a) isotope exchange
between N16O and a Ca18O surface and (b) isotope exchange between
C16O and a Ca18O surface in the temperature range 573 and 1073 K.
where R and â are the parameters obtained for the NO case at
the corresponding temperatures and γ determines weight factors
which are optimized together with the parameters R1 and â1 by
the least-squares method. See Table 2 for the fitted values of
the parameters and Figure 6b for the result of the modeling.
The values of R increase with increasing temperature.
Calculating the diffusivities at the temperatures tested and
plotting these in an Arrhenius plot produces an apparent
activation energy of 74.3 kcal/mol when NO is used as the probe
molecule. The diffusivity as determined with CO is initially
enhanced by the presence of vacancies, consistent with a lower
apparent activation energy, 44.9 kcal/mol. This enhanced
oxygen transport affects â since it is determined by the residual
activity of 16O. In Figure 7 the normalization factor is plotted
as a function of time for the three temperatures investigated.
The importance of this factor decreases with increasing tem-
peratures. At 1073 K, R and R1 are approximately equal,
whereas significantly more pronounced differences were found
at the lower temperatures.
Modeling the decline in the oxygen-isotope-exchange curves
for NO as the probe molecule was made by optimizing the
constants R and â in relation 5 by the least-squares method
J(0,t) ≈ (R/t1/2) + â1/2
(5)
The parameter R is proportional to the inverse of the square
root of the diffusivity as seen by comparing eq 4 and 5. The
parameter â was introduced to account for deviations from the
idealized model (i.e., a semi-infinite substrate is assumed rather
than grains) as well as the fact that not all oxygens in the bulk
are isotopically labeled. The applicability of eq 5 rests on the
understanding of the oxygen distribution in the substrate being
100% 18O at the surface, 100% 16O in the core, and a
concentration gradient in between. The model applies as long
as this gradient is negligible, meaning that a thin layer in the
vicinity of the surface is probed. Particularly, â is understood
to be a sensitive function of the residual 16O activity. Table 2
gives the values for the parameters R and â for different
temperatures. The temperature dependence of the diffusivity
is described by an Arrhenius equation and the relation between
R and the diffusivity is used to evaluate the activation energy
for self-diffusion of oxygen. Figure 6a shows the result of the
modeling: good agreement with the experimental data points
is obtained.
4. Discussion
The CaO catalyst can be divided into three regions comprising
the physical surface where the exchange between the gas phase
and the solid occurs, seVeral subsurface layers readily available
for exchange, and the bulk oxide.19 This partitioning is used in
the present study, where kinetic parameters for surface-oxygen
exchange and self-diffusion of oxygen in the bulk are investi-
gated by means of isotopic labeling. Oxygen-exchange rates

Oxygen Exchange Reactions
J. Phys. Chem. B, Vol. 102, No. 26, 1998 5163
between the physical surface and probe molecules revealed
information concerning the actual isotope-exchange reaction.
Apparent activation energies for single- and double-oxygen-
isotope exchange between 18O18O and preoxidized or prereduced
Ca16O surfaces and oxygen exchange between C16O or N16O
and preoxidized Ca18O surfaces were determined and are
discussed in section 4.1. An increase in temperature resulted
in an increased involvement of the subsurface oxygens in the
isotope-exchange reaction. The bulk oxygens gradually become
more important, resulting in an asymptotic exchange rate
controlled by the self-diffusion of oxygen through the bulk.
Effects of reduction and oxidation on apparent activation
energies of self-diffusion of oxygen were demonstrated and are
discussed in section 4.2.
Figure 8. Relation between the γ-parameter in the normalized weight
function and the rate constant k for the surface-oxygen abstraction.
4.1. Oxygen Isotope Exchange. The apparent activation
energies for oxygen isotope exchange (Table 1) are consistent
with the values reported in the literature. Observations of two
distinct classes of apparent activation energies for isotope
exchange have been discussed previously for exchange between
an ionic solid MX and the corresponding diatomic gas X2.24
Winter25 suggested the involvement of F-type centers in the
isotope exchange reactions that correspond to the lower apparent
activation energies. However, the higher values are not well
understood. No correlation with electronic properties of the
materials was observed.25 The higher apparent activation
energies obtained in this work for the single- and double-oxygen
exchange between O2 and a preoxidized CaO surface, as well
as the double-oxygen exchange between O2 and a prereduced
CaO surface, were similar to estimations from the initial rates
of oxygen-isotope exchange between O2 and CaO as calculated
from the results of Cunningham and Healy.14 These values were
contrasted by the low apparent activation energy observed in
the present study for single-oxygen-isotope exchange between
O2 and prereduced CaO surfaces. The magnitude of the latter
is comparable to those obtained for oxygen exchange between
CO or NO and CaO surfaces. Exposure of a CaO surface to a
reducing agent at elevated temperatures has been shown to result
in the formation of reactive sites, as a direct consequence of
the surface-oxygen abstraction step.7 The promoting effect of
F-type centers has been demonstrated for the oxygen exchange
over zinc oxide and other metal oxides26,27 and can be expected
to cause the drastic reduction in the apparent activation energy
for the single-oxygen-isotope exchange over prereduced surfaces
observed in the present study. The high apparent activation
energy for double-oxygen exchange over prereduced surfaces
suggested that the reoxidation of the surface is a fast process.
reduced sites unlikely in the exchange reactions with these
molecules. Previous results from thermal desorption gas
analysis are consistent with this understanding. It was shown
that CO, NO, and O2 desorb at temperatures up to about 1000
K.8,12,15 Thus, the first factor controlling the isotope exchange
is the availability of adsorption sites. The physical origin of
the 17 kcal/mol apparent activation energy remains unsolved.
This value is only obtained for oxygen exchange including an
O2 molecule and may be ascribed to an inherent property of
this molecule upon adsorption. The singlet-triplet splitting in
O2 is such a feature.29 It is reasonable to expect such a property
to be effective in the entrance channel of the exchange process
(i.e., an 3O2 to 1O2 conversion is necessary for oxygen
exchange). This constraint is not present for NO and CO and
consequently different apparent activation energies for oxygen
exchange were observed. The presence of excess electrons,
resulting from the prereduction treatment, makes the 2O2
adsorption channel accessible for single-oxygen exchange
between O2 and prereduced CaO surfaces, and consequently
the lower apparent activation energy was observed.
4.2. Self-Diffusion of Oxygen. A straightforward under-
standing of the oxygen self-diffusion emerges from the NO
study, and an apparent activation energy of 74.3 kcal/mol is
obtained. A similar value, 63.5 kcal/mol, was determined for
MgO.19 Determination of the apparent activation energy for
oxygen self-diffusion using CO is more complicated. A surface-
oxygen abstraction step upon CO exposure introduces vacancies
in the surface layer of the material. Despite the necessity of a
temperature of about 773 K to obtain reasonable CO2 produc-
tion, even at 573 K some surface oxygens are expected to be
removed. In previous work, the temperature dependence of the
rate constant for surface-oxygen abstraction was determined,7
and this temperature dependence is shown in Figure 8. The
temperature dependence of the γ parameter, which determines
the weight function, is also included in Figure 8 (see section
3.2.2). For 573 K, a low rate constant and a high γ value were
observed. This implies an initial dominating mechanism based
on vacancies introduced by the surface-oxygen abstraction. With
time, the vacancies diffused into the bulk and a similar
mechanism as for NO as the probe molecule is expected to
become dominant. Increased temperature results in an increased
rate constant for surface-oxygen abstraction and the importance
of the vacancy-controlled mechanism diminishes. The experi-
ment at 1073 K confirms this trend. Note the very small effect
of the weighing at this temperature, as seen by comparing the
parameters R and R1. This is expected due to the small number
of vacancies in the vicinity of the surface, and consequently
the bulk property prevails at these elevated temperatures.
The oxygen-isotope exchange between C16O and Ca18O
surfaces could involve similar sites as the 18O18O single-oxygen
exchange with a prereduced Ca16O surface, since exposure of
CaO to CO resulted in a surface-oxygen abstraction.7 However,
the amount of surface oxygens abstracted is low at temperatures
below 773 K, which makes a mechanism based on adsorbates
such as CO2-, CO22-, and CO32- more likely. The former was
observed using electron spin resonance,28 whereas the latter two
are among the many species suggested by IR spectroscopy
data.9-11 Further evidence for a mechanism based on adsorption
for CO is found in the similarity with NO at elevated
temperatures, since for the latter a fully oxidized surface is
obtained at all times. Yanagisawa suggested that NO was
- 15
adsorbed as NO2- or NO3
.
Necessary for these intermediates
is the presence of O- sites, and the existence of such finds
support in the pretreatment of the oxide (i.e., 7500 ppm O2 at
1073 K). The similarity between CO and NO as probe
molecules, seen in the present study, makes the involvement of

5164 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
(5) Wu, M.-C.; Truong, C. M.; Coulter, K.; Goodman, D. W. J. Catal.
1993, 140, 344.
5. Concluding Remarks
Discriminating molecular properties that control adsorption
are of interests in order to achieve high selectivity in the NOx
reduction activity of catalysts. The present study has focused
on some relevant molecules, choosing CaO as a model substrate.
The rates of oxygen-isotope exchange between CO, NO or O2,
and CaO surfaces were investigated in a comparative study and
were shown to be initially controlled by the actual isotope-
exchange reaction. After this first stage, the rate was found to
depend increasingly on the rate of self-diffusion of oxygen in
the CaO bulk.
(6) Zhang, X.; Walters, A. B.; Vannic, M. A. J. Catal..; 1994, 146,
568.
(7) Acke, F.; Panas, I.; Stro¨mberg, D. J. Phys. Chem. 1997, 101, 6484.
(8) Huzimura, R.; Yanagisawa, Y.; Matsumura, K.; Yamabe, S. Phys.
ReV. B, 1990, 41, 3786.
(9) Coluccia, S.; Garrone, E.; Guglielminotti, E.; Zecchina, A. J. Chem.
Soc., Faraday Trans. 1 1981, 77, 1063.
(10) Garrone, E.; Zecchina, A.; Stone F. S. J. Chem. Soc., Faraday
Trans. 1 1988, 84, 2843.
(11) Babaeva, M. A.; Bystrov, D. S.; Kovalgin, A. Y.; Tsyganeneko,
A. A. J. Catal. 1990, 123, 396.
(12) Yanagisawa, Y.; Yamabe, S.; Matsumura, K.; Huzimura, R. Phys.
ReV. B, 1993, 48, 4925.
The higher apparent activation energies for the actual
exchange process were speculated to be consequences of the
singlet-triplet spectroscopy of molecular oxygen. A compari-
son of the preexponential factors for the single- and double-
oxygen exchange between O2 and preoxidized CaO surfaces
suggested that the double-exchange reaction consists of two
successive single-exchange reactions. The apparent activation
energy for self-diffusion of oxygen was determined with NO
as the probe molecule. The temperature dependence of the
diffusivity was exploited and 74 kcal/mol was obtained for fully
oxidized CaO at all times. To understand the results for CO as
the probe molecule, the diffusion-controlled time domain was
divided into early and late subdomains. The early part was
controlled by the transient dynamics of oxygen vacancies, the
presence of which lowered the apparent activation energy, and
an estimated value of about 45 kcal/mol was obtained.
(13) Yanagisawa, Y.; Huzimura, R.; Matsumura, K.; Yamabe, S. Surf.
Sci. 1991, 242, 513.
(14) Cunningham, J.; Healy, C. P. J. Chem. Soc., Faraday Trans. 1,
1987, 83, 2973.
(15) Yanagisawa, Y. Appl. Surf. Sci. 1996, 100/101, 256.
(16) Platero, E. E.; Spoto, G.; Zecchina, A. J. Chem. Soc., Faraday
Trans. 1, 1985, 81, 1283.
(17) Chang, Y.-F.; Somorjai, G. A.; Heineman, H. J. Catal. 1993, 142,
697.
(18) K. P.; Goodwin, J. G., Jr.; Marcelin, G. J. Phys. Chem. 1989, 93,
7977.
(19) Peil, K. P.; Goodwin, J. G., Jr.; Marcelin, G. J. Catal. 1991, 131,
143.
(20) Viera, J. M.; Brook, R. J. In: Structure and Properties of MgO
and Al2O3 Ceramics; Kingery, W. D., Ed.; (American Ceramic Society
Press: Columbus, OH, 1984; p 438.
(21) Oishi, Y.; Ando, K. In: Structure and Properties of MgO and Al2O3
Ceramics; Kingery, W. D., Ed.; American Ceramic Society Press: Colum-
bus, OH, 1984; p 463.
(22) Lizuka, Y.; Tanigaki, H.; Sanada, M.; Tsunetoshi, J.; Yamauchi,
N.; Arai, S. J. Chem. Soc., Faraday Trans. 1994, 90, 1307.
(23) Wuensch, B. J.; Steele, W. C.; Vasilos, T. J. Chem. Phys. 1973,
58, 5258.
(24) Cunningham, J.; Goold, E. L.; Leahy, E. M. J. Chem. Soc., Faraday
Trans. 1 1979, 75, 305.
Acknowledgment. The authors are in debt to the Swedish
National Board for Industrial and Technical Development for
financial support (NUTEK).
(25) Winter, E. R. S. J. Chem. Soc. A 1968, 2889.
(26) Shvets, V. A.; Kuznetsov, A. V.; Fenin, V. A.; Kazansky, J. Chem.
Soc., Faraday Trans. 1 1985, 81, 2913.
(27) Cunningham, J.; Goold, E. L.; Leahy, E. M. J. Chem. Soc., Faraday
Trans. 1 1979, 75, 301.
References and Notes
(1) Ito, T.; Lunsford, J. H. Nature 1985, 314, 721.
(2) Shi, C.; Xu, M.; Rosynek, P.; Lunsford, J. H. J. Phys. Chem. 1993,
97, 216.
(28) Huzimura, R.; Kurisu, H.; Okuda, T. Surf. Sci. 1988, 197, 444.
(29) Huber, Herzberg Construction of Diatomic Molecules; van Nos-
trand: New York, 1979.
(3) Lin, C.-H.; Wang, J.-X.; Lunsford, J. H. J. Catal. 1988, 111, 302.
(4) Wu, M.-C.; Truong, C. M.; Goodman, D. W. Phys. ReV. B 1992,
46, 12, 688.