Ozone decomposition over manganese oxide supported on ZrO2 and TiO2: A kinetic study using in situ laser Raman spectroscopy

Article abstract of DOI:10.1006/jcat.2001.3167

Ozone decomposition of ZrO₂- and TiO₂-supported manganese oxide was studied utilizing in situ laser Raman spectroscopy along with simultaneous steady-state kinetic measurements. The Raman spectra revealed the presence of an adsorbed peroxide species for both catalysts under reaction conditions. Experimental data were well represented by a nonuniform surface kinetic treatment which assumed that the activation energies of adsorption and desorption were linear functions of the surface coverage of the peroxide species. Steady-state reaction rates were monitored for varying ozone partial pressure (0.6-3 kPa) and temperature (293-343 K). The reaction order with respect to ozone partial pressure for the ZrO₂- and TiO₂-supported sample was 0.98 and 0.85, respectively. At zero surface coverage of the peroxide species, the activation energy for adsorption was 7 and 6 kJ/mole for the TiO₂- and ZrO₂-supported catalysts, respectively. The activation energies for desorption at zero coverage were 44 and 62 kJ/mole, respectively for the same catalysts. The difference in ozone decomposition activity for the catalysts was due to the difference in desorption kinetic parameters.

Full text of DOI:10.1006/jcat.2001.3167

Journal of Catalysis 199, 282–290 (2001)
doi:10.1006/jcat.2001.3167, available online at http://www.idealibrary.com on
Ozone Decomposition over Manganese Oxide Supported
on ZrO and TiO : A Kinetic Study Using in Situ
2
2
Laser Raman Spectroscopy
�
�
, ,1
Rakesh Radhakrishnan and S. Ted Oyama †
�
Environmental Catalysis and Materials Laboratory, Departments of Chemical Engineering and †Chemistry,
Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061-0211
Received September 14, 1999; revised January 14, 2001; accepted January 4, 2001; published online March 21, 2001
ates on the surface of the working catalysts. This follows
Ozone decomposition on ZrO2- and TiO2-supported manganese recent work done in our laboratory (1, 2) on the mech-
oxide was studied using in situ laser Raman spectroscopy along anism and reaction kinetics for ozone decomposition on
with simultaneous steady-state kinetic measurements. The Raman
spectra showed the presence of an adsorbed peroxide species for
both catalysts under reaction conditions. The experimental data
were well represented by a nonuniform surface kinetic treatment
that assumed that the activation energies of adsorption and des-
netic parameters that are compared to those obtained in
orption were linear functions of the surface coverage of the perox-
ide species. Steady-state reaction rates were monitored for varying
ozone partial pressure (0.6–3 kPa) and temperature (293–343 K).
The reaction order with respect to ozone partial pressure for the
ZrO2-supported sample was 0.98 and that for the TiO2-supported
manganese oxide supported on Al2O3. The present study
confirms the presence of an adsorbed peroxide species on
the surface of the two new catalysts and the applicability
of the earlier proposed kinetics. The result is a set of ki-
the previous work and which help explain ozone decompo-
sition activity differences for the various manganese oxide
catalysts.
The detection of an adsorbed peroxide species on the
sample was 0.85. At zero surface coverage of the peroxide species, catalyst surface by Raman spectroscopy was critical in ob-
the activation energy for adsorption was 6 and 7 kJ/mol for the taining coverage data which were used in the analysis.
ZrO2- and TiO2-supported catalysts, respectively. The activation
energies for desorption at zero coverage were 62 and 44 kJ/mol,
respectively, for the same catalysts. The difference in ozone decom-
position activity for the catalysts was primarily due to the difference
Raman spectroscopy is a powerful technique that has been
widely used in heterogeneous catalysis (15). In situ Raman
spectroscopy has emerged as a useful method for studying
catalyst structure changes and adsorbed reaction interme-
diates (16–18). The information provided by these stud-
ies combined with kinetic measurements can be utilized to
determine reaction mechanisms (19). Raman spectroscopy
has been used to study several transition metal oxide sys-
tems. Among the materials that commonly appear in the lit-
erature are supported oxides of Mo (20, 21), V (15, 22), Cr
in desorption kinetic parameters.
�c 2001 Academic Press
Key Words: ozone; manganese oxide; kinetics; nonuniform sur-
face; laser Raman.
INTRODUCTION
Ozone is a toxic environmental pollutant and its abate- (23), Nb (24), W (25), and Re (26). There have been veryfew
ment at ground level is important for protecting plant and Raman studies which present spectra for supported man-
animal life. Several academic studies and patents have re- ganese oxide (1, 5, 27, 28). These studies have concentrated
ported on the development and testing of ozone decom- on manganese oxide supported on alumina (1, 5, 27, 28) or
position catalysts (1–4) and many of these have identified silica (28). Adsorption of ozone at low temperatures (77 K)
manganese oxide as an excellent catalyst for ozone elimi- has been studied earlier for ZrO2 (29) and TiO2 (30) sub-
nation. Manganese oxides have also found use in catalyz- strates using infrared spectroscopy.
ing a variety of oxidation (5–10) and reduction reactions
(
11–13). The chemical and catalytic properties of ozone are
EXPERIMENTAL
described in a recent book (14).
This paper presents a study of ozone decomposition on
manganese oxide supported on TiO2 and ZrO2 using in situ
laser Raman spectroscopy to quantify reaction intermedi-
Catalyst preparation and characterization. Supported
manganese oxide catalysts (3 wt% ) were prepared by im-
2
� 1
pregnating ZrO2 (Degussa, VP ZrO2, Sg = 50 m g ) and
TiO2 (Degussa, Titanoxid P25, Sg = 50 m g ) with aqueous
1
2
� 1
To whom correspondence should be addressed.
282
0021-9517/01 $35.00
Copyright �c 2001 by Academic Press
All rights of reproduction in any form reserved.

OZONE DECOMPOSITION ON SUPPORTED MANGANESE OXIDE
283
solutions of manganese acetate (Mn(CH3COO)2 � 4H2O, by passing filtered oxygen through a high-voltage silent-
Aldrich, >99.99% ) to the point of incipient wetness. For discharge ozone generator (OREC, V5–0). The total pres-
example, 1.69 g of manganese acetate was dissolved in 4 cm3 sure in the system was maintained at 101 kPa and the gen-
of distilled water and this solution was added dropwise to erated ozone/oxygen mixture was introduced into the cell
20 g of ZrO2. After impregnation, the sample was heated with ozone partial pressures varying from 0.4 to 3 kPa. The
at 393 K for 6 h and calcined at 773 K for 6 h to produce reaction temperature was varied from 293 to 340 K and the
3
3
wt% MnOx/ZrO2. A similar method was used to prepare flow rate of the ozone/oxygen mixture was maintained at
wt% MnOx/TiO2. 744 �mol s (1000 cm min ). The inlet and outlet ozone
The total number of metal oxide sites on the surface concentrationswere monitored usinga UV absorption-type
� � 1
1
3
of the catalysts was estimated through the temperature- ozone analyzer (IN-USA, AFX-H1). Steady-state turnover
programmed desorption (TPD) of oxygen. A detailed de- rates were based on ozone conversions and the number of
scription of the experiment can be found elsewhere (31). Mn active sites calculated from the ozone desorption ex-
Briefly, the procedure involved pretreatment of the cata- periments, using the expression, rate = (conversion � flow
lyst at 773 K in oxygen flow followed by the introduction rate) � (ozone concentration)/number of sites.
of a 2 mol% ozone/oxygen mixture at 233 K for 2 h. Subse-
quently, the catalyst was heated in helium flow at 0.17 K/s
to 1173 K while monitoring the m/e signals at 16, 32, and
RESULTS
4
8 corresponding to the possibly desorbing molecular oxy- TPD Traces
gen species. Desorption/decomposition peaks for molecu-
lar oxygen were detected and these were quantified through
calibration pulses of to determine the amount of oxygen
adsorbed per gram of catalyst. This value was used as the
corresponding site density (sites per gram of catalyst) for
turnover rate calculations. X-ray diffraction (XRD) pat-
terns were acquired on a powder diffractometer (Scintag,
Model ASC0007) with a Cu K� monochromatized radiation
source (45 kV and 40 mA).
The TPD traces for molecular oxygen (m/e = 32) from
the ZrO2-supported manganese oxide are shown in Fig. 1.
In no case was ozone (m/e = 48) observed. The traces for
the support ZrO2 without ozone adsorption reveals no
desorption/decomposition peak (Fig. 1a). However, when
In situ Raman spectroscopy. Details about the experi-
mental setup for the in situ Raman spectroscopy system can
be found elsewhere (1). Briefly, the system used an argon
ion laser (514.5 nm, Spex Lexel 95) as a light source, a holo-
graphicnotch filter (Kaiser, Super Notch Plus) for removing
Rayleigh scattering, a single-stage monochromator (Spex,
5
00 M) for light dispersion, and a CCD detector (Spex,
Spectrum One) for spectral acquisition. The laser was oper-
ated at 200mW and the detector slit width wasset at 100 �m.
�
1
The resolution of the Raman spectrometer was 6 cm . The
samples were in the form of thin wafers held at the end of a
rotating cylinder (1500 rpm) in a Suprasil quartz in situ cell.
Heating was carried out using heating tape around the sam-
ples. Cooling of the samples was achieved by passing chilled
water through Teflon tubing wrapped around the cell.
Steady-state kinetic studies. A low volume sample cell
3
(
volume <25 cm ) allowed effective investigation of sur-
face reactions under varying pressure and temperature con-
ditions. The temperature of the sample was monitored
through a thermocouple placed 3 mm from the sample
in a thermocouple well. The sample (0.3 g) was pressed
into a thin cylindrical wafer with a diameter of about
1
5 mm and a thickness of 1 mm, which was rotated at
FIG. 1. Oxygen (m/e = 32) TPD trace for ZrO2-supported man-
ganese oxide: (a) ZrO2 without ozone adsorption; (b) ZrO2 after ozone
adsorption at 233 K; (c) Mn/ZrO2 without ozone adsorption; (d) Mn/ZrO2
approximately 1500 rpm to prevent local overheating due
to the laser. The wafer was pretreated in flowing oxygen
(
Airco, >99.6% ) at 773 K for 2 h. Ozone was generated with ozone adsorption at 233 K.

2
84
RADHAKRISHNAN AND OYAMA
FIG. 3. In situ Raman spectra of ZrO2-supported manganese oxide:
(
1
(
a) in oxygen; (b) in 2 mol% ozone/oxygen mixture. Reactant flow =
3
� 1
000 cm min at room temperature. Spectrum acquisitions conditions
�
1
Figs. 1 and 2): laser power = 200 mW, resolution = 6 cm , exposure
time = 60 s, n = 60 scans
�
1
� 1
feature at 852 cm and a broad band at 930 cm . This peak
�
1
at 880cm wasaccompanied bytwo overtone peaksat 1754
FIG. 2. Oxygen (m/e = 32) TPD trace for TiO2-supported manganese
oxide: (a) TiO2 without ozone adsorption; (b) TiO2 after ozone adsorption
at 233 K; (c) Mn/TiO2 without ozone adsorption; (d) Mn/TiO2 with ozone peared once ozone flow was cut off. Similar measurements
�
1
and 2615 cm (not shown). All the new peaks slowly disap-
adsorption at 233 K.
were carried out at different temperatures (273, 313, 328,
and 343 K). It was found that the maximum peak intensity
for the adsorbed species did not change for measurements
at 273 and 293 K .
Figure 4 presents the Raman spectra for TiO2-supported
manganese oxide under similar conditions, again at room
ozone is adsorbed on the surface of ZrO2, a desorp-
tion/decomposition peak at 532 K is observed (Fig. 1b).
In the case of ZrO2-supported manganese oxide, the trace
without ozone adsorption (Fig. 1c) reveals a strong feature
at 775 K and weaker features at 560, 790, and 1050 K. Ad-
sorption of ozone on the catalyst surface at 233 K results in
a TPD trace (Fig. 1d) which shows strong features at 370,
4
10, 549, and 780 K and weak features at 790 and 1060 K.
The corresponding oxygen (m/e = 32) TPD traces for
the TiO2-supported manganese oxide are shown in Fig. 2.
The traces for the TiO2 support (Figs. 2a, 2b) with and
without ozone adsorption contain no desorption/decom-
position peak. The TiO2-supported manganese oxide with-
out any ozone adsorption (Fig. 2c) produces desorption/
decomposition peaks at 770, 836, and 1005 K. The catalyst
with ozone adsorption at 233 K (Fig. 2d) shows desorption/
decomposition features at 378, 830, and 995 K.
In Situ Laser Raman Spectra
Figure 3 shows the Raman spectra at room temperature
(
293 K) for manganese oxide on ZrO2, under both oxygen
flow and ozone/oxygen flow. Under oxygen flow (Fig. 3a),
�
1
there were peaks at 336, 383, 470, 552, and 626 cm . The
introduction of ozone (Fig. 3b) resulted in the appearance
FIG. 4. In situ Raman spectra of TiO2-supported manganese oxide:
(
a) in oxygen; (b) in 2 mol% ozone/oxygen mixture. Reactant flow =
�
1
3
� 1
of a very intense peak at 880 cm along with a shoulder 1000 cm min at room temperature.

OZONE DECOMPOSITION ON SUPPORTED MANGANESE OXIDE
285
FIG. 5. Effect of temperature and ozone partial pressure on ozone
decomposition rates over ZrO2-supported manganese oxide. Reactant
flow = 1000 cm min
FIG. 7. Reaction isotherms of ozone decomposition rates over ZrO2-
supported manganese oxide measured using in situ Raman spectroscopy.
3
� 1
.
3
� 1
Reactant flow = 1000 cm min . Spectrum acquisitions conditions (Figs. 5
and 6): laser power = 200 mW, resolution = 6 cm , exposure time = 60 s,
�
1
temperature (293 K). In oxygen flow (Fig. 4a), there were n = 15 scans.
�
1
peaks at 390, 509, and 630 cm . Under continuous ozone
�
1
flow (Fig. 4b), a peak at 878 cm appeared along with a Steady-State Kinetics
�
1
very weak feature at 928 cm . The first and second over-
Steady-state ozone decomposition rates were measured
under varying temperature and ozone partial pressure con-
ditions. For both catalysts, the decomposition rate increased
with increasing ozone partial pressure and temperature
�
1
tone peaks in this case were at 1750 and 2614 cm , respec-
tively. Again, similar measurements were made at different
temperatures (273, 313, 328, and 343 K).
Blank experiments conducted with the supports indi-
cated that the presence of ozone did not result in a decrease
in the intensity of the support peaks. The intense support
peaks for the ZrO2- and TiO2-supported manganese oxide
prevented the observation of any manganese oxide peak or
changes to it during the course of the reaction.
(
Figs. 5 and 6). The reaction order in ozone partial pres-
sure was found to be 0.98 and 0.85 for the ZrO2- and
TiO2-supported catalysts, respectively. These reaction or-
ders were found to be temperature independent.
� 1
The peak at 880cm for the ZrO2-supported catalyst and
� 1
the peak at 878 cm for the TiO2-supported catalyst were
used to determine the surface coverage of the adsorbed
species, �, under reaction conditions. This was done by in-
tegrating the peak intensities and normalizing the values to
the peak area at saturation, at the highest ozone partialpres-
sures and the lowest temperatures. Figures 7 and 8 present
reaction isotherms of the surface coverage for ZrO2- and
TiO2-supported manganese oxide catalysts. In both cases,
the surface coverage increased with increasingozone partial
pressure, eventually reaching saturation. Surface coverage
values decreased with increasing temperature as expected.
DISCUSSION
TPD Spectra
The absence of desorption/decomposition peaks for the
TPD trace of ZrO2 without ozone adsorption shows that
no bulk reduction of the support occurred by heating
to elevated temperatures. The TPD trace for the pure
ZrO2 support after ozone adsorption showed a broad
desorption/decomposition peak at 532 K (Fig. 1b). This
FIG. 6. Effect of temperature and ozone partial pressure on ozone
decomposition rates over TiO2-supported manganese oxide. Reactant
3
� 1
.
flow = 1000 cm min

2
86
RADHAKRISHNAN AND OYAMA
When ozone was adsorbed on the surface of the catalyst,
broad reduction peaksat 830and 995K appeared alongwith
the adsorbed peroxide species peak at 378 K (Fig. 2d). An
interesting observation here is the absence of a desortion
peak from atomic oxygen. This suggests that the recombina-
tion reaction for adsorbed atomic oxygen species does not
occur for the catalyst under the measurement conditions.
Integration of the desorption/decomposition peaks for
oxygen from the catalysts can be used to calculate the
amount of adsorbed oxygen on the surface. Taking val-
ues at the lowest temperatures (233 K) and highest pres-
sures (3 kPa) gives coverages at saturation, and allows de-
duction of the active center site density on the surface of
the catalyst (moles per gram of catalyst). Measurements at
slightly higher temperatures showed insignificant changes,
confirming that saturation had been reached. The corre-
�
1
sponding values were 163 and 31 �mol g for the Mn/ZrO2
and Mn/TiO2 catalysts, respectively.
FIG. 8. Reaction isotherms of ozone decomposition rates over TiO2-
supported manganese oxide measured using in situ Raman spectroscopy.
3
� 1
.
Reactant flow = 1000 cm min
Laser Raman Spectra
The Raman spectrum obtained for each of the supported
peak is indicative of an adsorbed species on the surface samples is indicative of dispersed species at the surface.
of ZrO2 upon exposure to ozone. Peaks at 775 and 1050 K In the absence of ozone for both the ZrO2- and TiO2-
(
Fig. 1c) for zirconia-supported manganese oxide (without supported samples, the only peaks visible are those of the
ozone adsorption) suggest reduction of manganese oxide support. For the ZrO2 supported sample, the peaks at 336,
�
1
at these high temperatures. These reduction peaks shifted 383, 470, 552, and 626cm belongto ZrO2. Similarly, for the
�
1
slightly to 780 and 1060 K when ozone was adsorbed on the TiO2-supported sample, the peaks at 390, 509, and 630 cm
surface of the catalyst. The peaks at 370 and 410 K (Fig. can be assigned to TiO2. In addition, no XRD peaks besides
d) can be attributed to the desorption/decomposition of those of the support are observed in these samples. Never-
1
adsorbed oxygen species. The oxygen species are identi- theless, the supported samples do undergo reduction, and
fied as peroxide groups on the surface of the catalysts by chemisorb oxygen. This evidence suggests that the man-
their band positions as obtained from their Raman spectra. ganese oxide is in a dispersed state in the near surface of
The range of vibrational frequencies expected for neutral the support. The degree of dispersion calculated from the
�
1
�
� 1
O2 is 1460–1700 cm , that for O2 is 1015–1180 cm , and active site density assuming one oxygen molecule per site
that for O2 is 640–970 cm (32) and the observed bands was 47% for the Mn/ZrO2 catalyst and 9% for the Mn/TiO2
�
� 1
2
clearly fall in the peroxide range. Also the bands are close catalyst.
to those of O2 on MnOx/Al2O3 where isotopic substitu-
�
These conclusions are confirmed by studies conducted by
2
tion has unambiguously identified the peroxide intermedi- Ma et al. (7, 8) for ZrO2- and Al2O3-supported manganese
ate (1). When the MnOx/ZrO2 sample was heated to 500 K, oxide samples also using manganese acetate as a precursor.
the peroxide Raman signal disappeared. Thus, the desorp- The catalysts were characterized using XPS, TPR, and
tion/decomposition peak observed at 549 K is likely a re- XRD, and it was deduced that the low loading catalysts pre-
sult of the recombination and desorption of atomic oxygen pared through incipient wetness resulted in well-dispersed
on the surface of the ZrO2 support. This process, though, manganese oxide systems with oxidation states between +2
occurs at much higher temperatures than those at which and +3. Kapteijn et al. have also studied the effect of pre-
reaction occurs (<350 K). Consequently, it probably does cursor and metal loading on Al2O3-supported manganese
not contribute to the reaction in the lower range. Isotopic oxide (5) using XPS, TPR, Raman, and IR spectroscopy.
tracer studies had earlier indicated that atomic oxygen did They observed that at low loadings (<10% ), manganese
not recombine at low temperatures on MnOx/Al2O3 (1, 2). oxide prepared from manganese acetate was a mixture of
The desorption traces for the pure TiO2 support with well-dispersed manganese oxide phases, MnO and Mn2O3.
and without ozone adsorption showed no signs of adsorbed Earlier work done in our laboratory has identified a peak at
�
1
species or bulk reduction (Figs. 2a, 2b). In the case of the 654 cm which was assigned to the mixed-valency species
Mn/TiO2 catalyst with no adsorption, manganese oxide re- Mn3O4 species (1). The presence of very intense sup-
duction features appeared at 770, 836, and 1005 K (Fig. 2c). port peaks in this region for ZrO2 and TiO2 precludes the

OZONE DECOMPOSITION ON SUPPORTED MANGANESE OXIDE
287
fied an adsorbed peroxide species at 884 cm 1 on the cata-
lyst surface through isotopic work and ab initio calculations.
According to these studies, the vibrational mode for the ad-
�
�
�
O3 + → O2 + O
(i)
�
�
O3 + O → O + O2
(ii)
2
�
�
O → O2 +
(iii) sorbed species is highly localized. This could be the reason
for the very small change in the vibrational frequency for
the peroxide species on ZrO2- and TiO2-supported cata-
lysts as compared to that for the Al2O3-supported
2
SCHEME 1. Proposed Ozone Decomposition Mechanism.
catalyst.
For the Mn/ZrO2 catalyst, the peak at 880 cm had a
observation of this feature in the present studies. Addition-
ally, the presence of MnO, Mn2O3, and MnO2 phases, which
are weakly Raman active, cannot be ruled out. These phases
are usually identified through XRD, but only for cata-
lysts with high metal loading (>10% ). Recently, Buciuman
et al. (28) have also studied manganese oxide supported on
Al2O3 and SiO2 (0.05–18 mol% ) using XRD, FTIR, and
Raman spectroscopy. These studies have confirmed the
presence of bulk phases at high loading (>6 wt% ) by the
observation of Raman features corresponding to � -Mn2O3,
�
1
�
1
shoulder on the lowwavenumber side at 852cm indicative
of two different adsorbed peroxide species on the catalyst
surface. The presence of these species was also indicated in
the oxygen TPD results (Fig. 1). It is likely that in the case of
the Mn/ZrO2 catalyst, there are two peroxide species. The
structure of these peroxide species is uncertain. They could
2
be bidentate � -O2 complexes as likely on the monoatomic
Mn center on Mn/Al2O3 (31), or they could be bridging
dioxygen species as possible on the multinuclear Mn cen-
ter on the Mn/TiO2 and Mn/SiO2 catalysts (31). The bands
on the Mn/ZrO2 samples could also arise from different
interactions of the peroxide species with the ZrO2 support.
Adsorbed ozone on ZrO2 and TiO2 at low temperatures
�
-Mn2O3, MnO, and Mn3O4 and XRD featuresfor �-MnO2.
In conclusion, based on comparisons to the literature, our
results indicate that the supported manganese species are in
a dispersed state with likely varied structures and oxidation
states. Crystallites of some bulk oxide species are possible,
but if present these would be smaller than 4 nm.
(
77 K) has been observed earlier (29, 30) using infrared
�
1
spectroscopy. On ZrO2 bands were observed at 2122 cm
(
As discussed in the previous section, the new signals
�
1
� 1
�
1
�1 + �3), 1138 cm (�1), and 1001 cm (�3) while on
that appear upon ozone exposure at 880 and 878 cm for
the ZrO2- and TiO2-supported manganese catalysts can be
attributed to an adsorbed peroxide species derived from
� 1
TiO2 (anatase) bands were seen at 2137 cm (�1 + �3),
� � 1
1
1
147 cm (�1), and 989 cm (�3). These correspond to
�
1
distorted molecular ozone vibrations with interaction of
O3 with Lewis sites on the oxides. No adsorbed peroxide
species were observed at these low temperatures.
ozone. There are also smaller features at 930 and 928 cm
,
respectively, that accompany the main peaks. These are
likely due to Mn == O vibrations formed from atomically ad-
sorbed oxygen (Scheme 1, vide infra). These are consistent
with the expected vibrational frequencies for such M == O as
Steady-State Kinetics
�
1
summarized by Che et al. (32) (900–1100 cm ). Atomically
adsorbed oxygen species are known to be more thermally
stable than adsorbed peroxide species(33). Table 1presents
the coverage of the two species as a function of increasing
temperature at 3000 Pa and supports our assignment for the
vibrations. The ratio of the two coverages clearly indicate
the thermal stability of the atomic oxygen species compared
to the peroxide species. These signals were not observed
during blank experiments when the supports alone were
used. Ozone decomposition work done on Al2O3 supported
manganese oxide using Raman spectroscopy (1) has identi-
The proposed ozone decomposition mechanism from the
studies done on an Al2O3-supported manganese oxide cata-
lyst (1) is presented in Scheme 1. The symbol is used
�
to denote surface sites. Briefly, experiments involving iso-
18
tope ( O3) (1, 2) substitution ruled out the possibility of
oxygen atom recombination and the finding that the ad-
sorbed ozone never desorbed established the irreversibility
of steps (i) and (ii). Furthermore, the finding that the per-
oxide species could not be formed from molecular oxygen
at any conditions showed the irreversibility of step (iii).
The acquired steady-state data were analyzed with a uni-
form surface kinetic analysis using a Langmuirian expres-
sion for the coverage � with an equilibrium constant K and
a reaction rate constant k:
TABLE 1
Effect of Temperature on Adsorbed Species Coverages
K(O3)
�O�a
� =
ss
[1]
[2]
Temperature/K
�O2
Ratio (�O� /�O2�
)
�
1
+ K(O3)
2
3
3
93
13
28
0.80
0.69
0.58
0.05
0.06
0.12
0.06
0.09
0.21
r = k� .
ss
ss
The results of the analysis for the MnOx/ZrO2 and
MnO /TiO coverage fits are presented in Figs. 9 and 10.
It is clear that the quality of the fit is not good and the
a
These coverages are qualitative. They were calculated from peak
areas assuming the same calibration factor as for the O2�
x
2
species.

2
88
RADHAKRISHNAN AND OYAMA
FIG. 11. Rate vs coverage: MnOx/ZrO2. A linear relationship is ex-
pected for a uniform surface.
FIG. 9. Surface coverage vs ozone partial pressure: MnOx/ZrO2. Fits
based on the Langmuir coverage expression.
likely presence of multiple manganese oxidation states on
average �2 value for these fits was found to be 10 . the surface of the catalyst which give rise to manganese
Figures 11 and 12 present the rate versus coverage plots oxide species with slightly different activities for ozone
for the two catalysts. From Eq. [2], a linear dependency of decomposition. The nonuniform analysis uses an adsorp-
steady-state rates on coverage is expected. The plots clearly tion activation energy which increases linearly with cover-
� 2
0
a
show a nonlinear dependency of rates on coverage. It can be age (E = E + gRT �) and a desorption/decomposition
a
ascertained from these results that the experimental data activation energy that decreases linearly with coverage
0
d
are inadequately predicted by a uniform surface kinetic (E = E � hRT �). Equations [3] and [4] are the result-
d
analysis.
ing rate expressions obtained for adsorption and desorp-
As a result, a nonuniform surface kinetic treatment (2) tion/decomposition, respectively.
was used to fit the acquired steady-state experimental data.
The interaction of the dispersed Mn oxide species with
different sites on the support surface probably gives rise
to a heterogeneous surface. This is compounded by the
0
� g�
ra = ka (O3)e
[3]
[4]
0
� h�
.
rd = k e
d
0
0
a
In these expressions k and k are the adsorption and
a
desorption/decomposition rate constants at zero coverage.
FIG. 10. Surface coverage vs ozone partial pressure: MnOx/TiO2. Fits
FIG. 12. Rate vs coverage: MnOx/TiO2. A linear relationship is ex-
based on the Langmuir coverage expression.
pected for a uniform surface.

OZONE DECOMPOSITION ON SUPPORTED MANGANESE OXIDE
289
Setting the adsorption and desorption rate expressions
equal provides equations for both the steady-state rate and
the steady-state coverage, Eqs. [5] and [6].
�
�
0
a
0
d
h/g+h
k
k
0
rss = kd
(O3)
[5]
[6]
�
�
0
1
k
a
�ss =
ln
(O3) .
0
g + h
k
d
The steady-state coverage and rates were monitored as
a function of ozone partial pressure and temperature. The
values for the various kinetic parameters in Eqs. [5] and [6]
were obtained through a least-squares fitting of the exper-
imental data. The linear fits of the steady-state rate versus
the ozone partial pressure (Figs. 5 and 6) and those of the
steady-state coverage of the peroxide species against the
logarithm of the ozone partial pressure (Figs. 13 and 14)
were used to simultaneously solve for the parameters in
Eqs. [5] and [6]. Clearly, the data fit the equations very well
throughout the entire range of variation of experimental
FIG. 14. Plots for the surface coverage vs logarithm of ozone partial
pressure: TiO2-supported manganese oxide.
position on the Mn/ZrO2 and Mn/TiO2 catalyst are com-
pared to those obtained for Mn/Al2O3 in the earlier work
2) in Table 2.
2
conditions. The average � value for the coverage fits was
�
4
2
found to be 10 while the rate fits produced an average �
(
�
5
value of 10 . The pre-exponential and activation energy
values calculated for adsorption and desorption from the
analysis are presented in Table 1 and are compared with
the corresponding values obtained in the previous study
for the Al2O3-supported catalyst.
The valuesfor the pre-exponentialfor adsorption on each
� 18
� 3 � 1
of the catalysts is in the range of 10 cm
s and the ac-
tivation energy for ozone adsorption on each of the cata-
lysts is low (<10 kJ/mol). The activation energy for
desorption/decomposition is significantly larger than the
activation energy of adsorption for each catalyst. This could
be the reason why a stable reaction intermediate was ob-
served for the catalytic process. The activation energies and
pre-exponentials for the desorption/decomposition process
follow the trend Al2O3 > ZrO2 > TiO2. This suggests that
for the ozone decomposition process on supported man-
ganese oxide catalysts, the desorption/decomposition step
is rate determining.
The steady-state rates for both catalysts were found to
have a fractional order dependency on ozone partial pres-
sure which was independent of temperature. This is consis-
tent with the fractional order h/(g + h) predicted by Eq. [3]
with a value <1. The kinetic parameters for ozone decom-
There has been considerable work done in examining the
effect of support on the reactivity of transition metal ox-
ides. Work done in our laboratory with ethanol oxidation
over MoO3 catalysts has shown that metal oxide–support
TABLE 2
Kinetic Parameters for Ozone Decomposition on Supported
Manganese Oxide Catalysts
Al2O3
ZrO2
TiO2
Parameter
support
support
support
h
g
11.8
0.75
11.1
0.2
23.8
4.2
0
E
a
6 kJ/mol
6 kJ/mol
7 kJ/mol
0
d
E
69 kJ/mol
62 kJ/mol
44 kJ/mol
0
A
a
0
A
d
3.1 � 10� 18 cm s
3
� 1
1.1 � 10� 18 cm3 s� 1 1.2 � 10� 18 cm s
3
� 1
FIG. 13. Plots for the surface coverage vs logarithm of ozone partial
pressure: ZrO2-supported manganese oxide. A linear relationship is ex-
pected for a nonuniform surface.
7
� 1
5
� 1
3
� 1
1.6 � 10 s
8 � 10 s
1.9 � 10 s

2
90
RADHAKRISHNAN AND OYAMA
interactions are responsible for changing activities and se-
lectivities (18). The difference in the activities of the var-
ious support MoO3 catalysts for ethanol oxidation stud-
ies is reported to have resulted from a difference in the
pre-exponential term in the reaction rate constant (be-
cause there was little activation energy variation among
all the catalysts). The reactivity trend in this case followed
the order TiO2 > Al2O3 > SiO2. Similar trends in activity
were observed for methanol oxidation on supported MoO3
REFERENCES
1
. Li, W., and Oyama, S. T., J. Am. Chem. Soc. 120, 9041
(
1998).
2
3
. Li, W., and Oyama, S. T., J. Am. Chem. Soc. 120, 9047 (1998).
. Dhandapani, B., and Oyama, S. T., J. Appl. Catal. B: Environ. 11, 129
(1997).
. Oyama, S. T., Catal. Rev.-Sci. Eng. 42, 279 (2000).
. Kapteijn, F., van Langeveld, A. D., Moulijn, J. A., Andreini, A.,
Vuurman, M. A., Turek, M. A., Jehng, J. M., and Wachs, I. E., J. Catal.
4
5
1
50, 94 (1994).
(
34) and V2O5 (35) catalysts. The greater lability of the
6
7
8
9
. Kapteijn, F., Singoredjo, L., van Driel, M., Andreini, A., Moulijn, J. A.,
Ramis, G., and Busca, G., J. Catal. 150, 105 (1994).
. Ma, J., Chuah, G. K., Jaenicke, S., Gopalakrishnan, R., and Tan, K. L.,
Ber. Bunsenges. Phys. Chem. 99, 184 (1995).
. Ma, J., Chuah, G. K., Jaenicke, S., Gopalakrishnan, R., and Tan, K. L.,
Ber. Bunsenges. Phys. Chem. 100, 585 (1995).
. Baldi, M., Finochhio, E., Pistarino, C., and Busca, G., J. Appl. Catal.
A: Gen. 173, 61 (1998).
0. Kapteijn, F., Singoredjo, L., Andreini, A., and Moulijn, J. A., J. Appl.
Catal. B: Environ 3, 173 (1994).
1. Yamashita, T., and Vannice, A., J. Catal. 161, 254 (1996).
2. Maltha, A., Favre, L. F. T., Kist, H. F., Zuur, A. P., and Ponec, V.,
J. Catal. 149, 364 (1994).
metal–oxygen–support bridging bond for ZrO2- and TiO2-
supported catalysts was given as a reason for their higher
activities.
The desorption/decomposition step in the ozone decom-
position process can be viewed as a reductive elimination
involvingthe transfer ofelectronsfrom the peroxide species
to the manganese active center to form molecular oxygen.
NEXAFS (near-edge X-ray absorption fine structure) mea-
surements on the catalysts (31) revealed that there was a
chemical shift to lower energies in the Mn L-edge features
1
1
1
which followed the trend Al2O3 > ZrO2 > TiO2. The shift 13. Yongnian, Y., Ruili, H., Lin, C., and Jiayu Z., J. Appl. Catal. A: Gen.
1
01, 233 (1993).
toward lower energy with respect to bulk MnO2 indicates
more reduced species and larger number of unoccupied d
levels. Since it is the unoccupied d levels which accept elec-
trons, the NEXAFS results support the likely importance
of electron transfer in the reduction of the active center
during ozone decomposition.
1
4. Hunter, P., and Oyama, S. T., “Control of Volatile Organic Compound
Emissions.” Wiley, New York, 2000.
5. Wachs, I. E., Catal. Today. 27, 437 (1996).
6. Jehng, J. M., Hu, H., Gao, X., and Wachs, I. E., Catal. Today. 28, 335
(1996).
1
1
1
1
7. Wachs, I. E., Top. Catal. 1 (1998).
8. Zhang, W., Desikan, A., and Oyama, S. T., J. Phys. Chem. 99, 14468
(
1995).
CONCLUSIONS
19. Oyama, S. T., and Li, W., Top. Catal. 75 (1999).
2
2
2
0. Hu, H., Wachs, I. E., and Bare, S., J. Phys. Chem. 99, 10903
(1995).
1. Oyama, S. T., and Zhang, W., J. Am. Chem. Soc. 118, 7173
The presence of an adsorbed peroxide species on the sur-
face of ZrO2 and TiO2 during ozone decomposition was
confirmed using in situ laser Raman spectroscopy. Simulta-
neous steady-state kinetic measurements revealed a frac-
(
1996).
2. Went, G. T., and Oyama, S. T., J. Phys. Chem. 94, 4240
1990).
(
tional order dependency of the turnover rates with respect 23. Weckhuysen, B. M., and Wachs, I. E., J. Phys. Chem. 100, 14437
to ozone partial pressure of 0.98 and 0.85 for the ZrO2- and
TiO2-supported catalysts. The nonuniform surface kinet-
ics analysis used in an earlier study on Al2O3-supported
manganese oxide was found to work very well for the
(1996).
2
2
4. Hardcastle, F. D., and Wachs, I. E., Solid State Ionics 45, 201
(
1991).
5. Ostromecki, M. M., Burcham, L. J., Wachs, I. E., Ramani, N., and
Ekerdt, J. G., J. Mol. Catal. (A): Chem. 132, 43 (1998).
catalysts in this study. There was substantial variation in 26. Deo, G., and Wachs, I. E., J. Phys. Chem. 95, 5889 (1991).
the activation energy and pre-exponential values for the 27. Stroheimer, B. R., and Hercules, D. M., J. Phys. Chem. 95, 5889
(
1991).
desorption/decomposition step among the catalysts. Dif-
ferences in the structure of the manganese active center
seems to be the origin of the support effect on the desorp-
tion/decomposition kinetic parameters. The trend in these
2
2
8. Buciuman, F., Patcas, F., Cracium, R., and Zhan, R. T. D., Phys. Chem.
Chem. Phys. 1, 185 (1999).
9. Bulanin, K. M., Lavalley, J. C., and Tsyganenko, A. A., Colloids
Surf. A: Phys. Eng. Aspects 101, 153 (1995).
parameters, Al2O3 > ZrO2 > TiO2, seems to follow the re- 30. Bulanin, K. M., Lavalley, J. C., and Tsyganenko, A. A., J. Phys.Chem.
9
9, 10294 (1995).
ducibility of the manganese active center.
3
1. Radhakrishnan, R., Oyama, S. T., Chen, J. G., and Asakura, K., J. Phys.
Chem., in press.
ACKNOWLEDGMENTS
32. Che, M., and Tench, A. J., Adv. Catal. 32, 1 (1983).
33. Onishi, T., in “Dynamic Porcesses on Solid Surfaces” (K. Tamaru Ed.),
We gratefully acknowledge the financial support for this work by the
Director, Division of Chemical and Thermal Systems of the National Sci-
ence Foundation, under Grant CTS-9712047.
p. 237. Plenum, New York, 1994.
34. Hu, H., and Wachs, I. E., J. Phys. Chem. 99, 10911 (1995).
35. Deo, G., and Wachs, I. E., J. Catal. 146, 335 (1994).