Sb(III) as a surface site for water adsorption on Sn(Sb)O2, and its effect on catalytic activity and sensor behavior

Article abstract of DOI:10.1021/jp981391v

Surface segregation of Sb in polycrystalline Sn(Sb)O₂ is known to affect the rate of surface-catalyzed combustion of hydrocarbons and carbon monoxide over this material. This combustion rate is also known to be affected by the presence of water vapor. We show that Sb segregation to the surface as Sb(III) in Sn_1-xSb_xO₂ (x = 0.005, 0.05), controllable by thermal treatment, strongly alters the effect of water vapor both upon the surface-catalyzed combustion rate of carbon monoxide and upon the elevated-temperature electrical conductivity. We suggest that the surface defect states which mediate both the electrical behavior and the surface-catalyzed combustion are best formulated as an association complex of an oxygen vacancy with Sn(II) or Sb(III), and are then able to propose a simple model which unifies the interpretation of the behavior of SnO₂ as both a gas sensor and a combustion catalyst. We postulate: that a correct formulation of the "adsorbed oxygen" (O^2-) surface species mediating the electrical response is an oxygen molecule trapped in or on a surface oxygen vacancy; that the combustion reaction proceeds partly through these species and partly through lattice oxygen at the surface; that water competes with oxygen for the surface vacancies, blocking this route; and that the binding energy of water to the Sb(III)-V_o surface defect complex is less than that to the Sn(II)-V_o complex.

Full text of DOI:10.1021/jp981391v

6732
J. Phys. Chem. B 1998, 102, 6732-6737
Sb(III) as a Surface Site for Water Adsorption on Sn(Sb)O2, and Its Effect on Catalytic
Activity and Sensor Behavior
Vincent Dusastre and David E. Williams*
Chemistry Department, UniVersity College London, 20 Gordon Street, WC1H OAJ, UK
ReceiVed: March 5, 1998; In Final Form: June 3, 1998
Surface segregation of Sb in polycrystalline Sn(Sb)O2 is known to affect the rate of surface-catalyzed
combustion of hydrocarbons and carbon monoxide over this material. This combustion rate is also known to
be affected by the presence of water vapor. We show that Sb segregation to the surface as Sb(III) in Sn1-xSbxO2
(x ) 0.005, 0.05), controllable by thermal treatment, strongly alters the effect of water vapor both upon the
surface-catalyzed combustion rate of carbon monoxide and upon the elevated-temperature electrical
conductivity. We suggest that the surface defect states which mediate both the electrical behavior and the
surface-catalyzed combustion are best formulated as an association complex of an oxygen vacancy with Sn(II)
or Sb(III), and are then able to propose a simple model which unifies the interpretation of the behavior of
SnO2 as both a gas sensor and a combustion catalyst. We postulate: that a correct formulation of the “adsorbed
oxygen” (O2-) surface species mediating the electrical response is an oxygen molecule trapped in or on a
surface oxygen vacancy; that the combustion reaction proceeds partly through these species and partly through
lattice oxygen at the surface; that water competes with oxygen for the surface vacancies, blocking this route;
and that the binding energy of water to the Sb(III)‚VO surface defect complex is less than that to the Sn(II)‚
VO complex.
Introduction
lost from the hydrous oxide, the onset of antimony surface
enrichment at 400 °C corresponded exactly to a change in color
from white to deep blue, which the authors interpreted as being
due to antimony lattice diffusion and reconstruction. At 600
°C, by analogy with (NH4)2SbBr6,13 the color change is possibly
due to charge transfer between Sb3+ and Sb5+ within a random
array of disordered SnO6 and SbO6 octahedra.7 The results are
therefore consistent with the prediction that catalyst preparation
forces antimony into an environment in which it is not
particularly stable, so that at high temperatures segregation to
surfaces takes place where antimony could occupy an energeti-
Antimony-tin oxide catalysts are well-known to be active
and selective for olefin oxidation, oxidative dehydrogenation,
and ammoxidation of alkenes, notably propylene to acrolein and
acrylonitrile,1-3 and have previously been developed for the
selective oxidation of hydrocarbons (methane4). The behavior
contrasts with that of pure SnO2, which favors deep oxidation.5,6
Optimal catalytic activity is obtained with fairly large antimony
concentrations (∼30% Sb cations1). Although Sb substitution
appears to occur without strong modification of the SnO2 lattice,
the reported solubility limit of antimony cations in the rutile
phase of SnO2, from structural studies, varies from 3 7 to 40%.8
The catalytic performance of these compounds depends critically
upon the conditions of preparation and particularly upon the
calcination temperature adopted. These conditions have been
reported9,10 to have an effect on the properties of the materials,
such as the surface area, the particle size, and the surface
composition. The surface composition of these oxides varies
from the bulk composition. There is, especially, surface
enrichment in antimony which takes place to an extent depend-
ent upon bulk composition and calcination temperature. Pro-
nounced enrichment by antimony at the surface was observed
by X-ray photoelectron spectroscopy (XPS) with a heat of
segregation varying from roughly 10 (40 at. % Sb) to 40 (0.5
at. % Sb) kJ/mol.11 Comparison of the influence of calcination
temperature upon surface composition (by XPS) with the effect
upon bulk phase composition (X-ray diffraction study) has
shown9,12 a region which contains a Sb2O4 surface phase which
appears at high temperature and which exhibits decreasing
surface antimony composition with increasing temperature as
a consequence of vaporization of Sb2O4.
cally more favorable situation, probably as Sb3+ 14,15
.
SnO2 is a semiconductor with a direct band gap of 3.6 eV
between the full oxygen 2p valence band and the tin states at
the bottom of the conduction band. The electronic conduction
of stannic oxide has been attributed to oxygen vacancy donor
levels and can be enhanced by reduction.16 Introducing carriers
in SnO2 by antimony doping results in a dramatic increase in
the electrical conductivity because of substitution of Sb(V) into
the SnO2 lattice.17,18 Model calculations compared with infrared
reflectance measurements for various Sb doping levels sug-
gested, however, that for doping levels > 1 at. %, the sample
might be covered with a reduced carrier concentration surface
layer, consistent with Sb segregation to the surface as Sb(III).19
Catalytic reaction between reducing species and surface-
adsorbed oxygen at elevated temperatures releases carriers in
the tin dioxide conduction band. Changes in conductivity of
tin dioxide associated with the interaction with reducing gases
provide the basis for application of these systems as gas
sensors.20 The electrical conductivity is, in fact, a very sensitive
probe of the surface state of the oxide. Semiconductor gas
detectors are sensitive to water vapor and their response to
combustible gases is affected by the ambient humidity.21,22 Thus,
it has been observed that water adsorbs in greater quantities
Cross and Pyke11 studied Sb segregation upon calcination of
oxides prepared by a coprecipitation method. As water was
S1089-5647(98)01391-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 08/06/1998

Sb(III) for Water Adsorption on Sb(Sb)O2
J. Phys. Chem. B, Vol. 102, No. 35, 1998 6733
than oxygen on the surface of tin dioxide.23 Water is strongly
bound to an SnO2 surface, possibly as surface Sn-OH groups.
On heating, minor water loss is observed at 373 K, the major
loss taking place in the range 520-720 K. The effect of
moisture on gas sensitivity has been discussed24 in terms of the
displacement of chemisorbed oxygen by H2O and OH-. On
SnO2 the chemisorption of water seems to introduce a surface
electronic state such as a surface hydroxyl species.24 Recently,
the presence of dissociatively adsorbed water on SnO2(110) has
been observed using a combination of ultraviolet photoelectron
spectroscopy (UPS) and temperature-programmed desorption
spectroscopy (TPDS).25 In these studies, H2O acts as an electron
donor, bending the bands down at the surface and increasing
the surface carrier concentration. The increase in surface
conductivity caused by H2O is greater on slightly reduced
surfaces than on heavily reduced, disordered ones.26
Figure 1. Electrical resistivity of Sn1-ySbyO2 porous pellets exposed
to dry air at 400 °C, as a function of calcination temperature.
The behavior of Sb-doped tin dioxide as a gas sensor material
has been extensively reported. However, there has been little
or no investigation of the effect of antimony enrichment at the
surface as Sb(III) on either adsorption, catalysis, or gas sensor
behavior, despite the speculation by Egdell et al,27 that there
should be an interaction of Sb(III) surface species with
molecules which can act as electron acceptors. In the present
paper, we show that the surface segregation of Sb in polycrys-
talline Sb-doped SnO2 can be controlled by an appropriate
preparation regime. We confirm earlier XPS observations
implying the presence of Sb(III) at the surface and correlate
this with the measurements of surface segregation. By measur-
ing the electrical behavior in response to exposure to water vapor
in air at elevated temperature we then deduce the existence of
an electrically active surface state associated with preferential
water adsorption on Sb(III), thereby supporting earlier specula-
tion concerning the significance of this state.27,28 Finally, we
demonstrate an effect of Sb(III) segregation on oxidation
catalysis. The effect is found only in the presence of water
vapor. We propose a connection between the electrical proper-
ties and the catalytic activity which rationalizes the results.
showed the presence of a single well-crystallized rutile phase,
whose lattice parameters were the same as those of undoped
SnO2. The average crystallite size, determined from the
diffraction line width and cross-checked by scanning electron
microscopy and surface area measurement, increased smoothly
with calcination temperature from roughly 25 (600 °C; consistent
with ref 7) to 150 nm (1200 °C) for 0.5% Sb-doped materials.
XPS spectra were recorded (VG ESCALAB 220iXL instrument)
using focused (300 µm spot size) monochromatized Al KR
radiation. The binding energies were referenced to the hydro-
carbon C 1s peak at 284.80 eV and the sample charging was
controlled with a 3 eV flood gun. Spectrum quantification and
curve fitting was performed using a Shirley background and
sensitivity factors were obtained from Wagner et al.30
Gas combustion measurements were performed with a quartz
reactor tube (15 mm in diameter) and quartz frit upon which 1
g of the powder was placed. The reacting gases (1000 ppm in
21% O2, 79% N2, or Ar) were supplied at 50 cm3min-1 through
mass-flow controllers (Tylan) to the reactor tube mounted
vertically in a tube furnace. The temperature was ramped from
25 to 600 °C at 10 °C‚min-1. A quadrupole mass spectrometer
was used to measure the concentrations in the gas stream after
passage through the reaction tube. For CO oxidation the gas
mixtures contained Ar instead of N2 to remove the mass 28
interference. Carbon dioxide was the only product detected.
The mass spectrometer signal was linear in concentration of
CO, CH4, and CO2. The pseudo-first-order rate constant for
the surface-catalyzed combustion was calculated as previously
described.31 Specific surface area was determined by krypton
adsorption at 77 K using a volumetric procedure. Values very
similar to those noted by Brown and Patterson4 were obtained.
For comparison of materials, the rate constant was normalized
by the surface area of the sample.
Experimental Section
The oxides Sn1-ySbyO2 (y ) 0, 0.005, 0.05) were prepared
by coprecipitation7 using the following procedure. Weighed
amounts of antimony (Ventron, purity > 99.999%) and tin
(Fluka, purity > 99.999%) were dissolved in aqua regia. An
excess of NH3 was added to the solution to make it alkaline.
After the mixture boiled for a few hours, the resulting precipitate
was washed and collected. After the precipitate dried overnight
at 120 °C, the powdered materials were fired for 12 h at different
temperatures (600, 800, 1000, and 1200 °C) in air in recrystal-
lized alumina crucibles. The coprecipitated materials were white
in color after drying at 120 °C, and became blue on heating to
600 °C. The blue color darkened with increasing firing
temperature. The materials were then ground, pressed at 1 ton
into 13 mm diameter 1.5 mm thick pellets and fired at the same
temperature as the initial firing for 12 h.
The electrical behavior of the pellets under controlled
humidity and temperature conditions was monitored in a
computer-controlled rig (described in ref 29). Two-terminal dc
resistance measurements were employed, the accuracy of these
measurements having been independently cross-checked against
ac measurements. Water vapor pressure was controlled by
mixing dry gas with gas that had been saturated with water vapor
by passing through a bubbler.
Results
Sb Surface Segregation. The variation of electrical con-
ductivity with calcination temperature is shown in Figure 1. The
material calcined at 600 °C had properties essentially the same
as those of the undoped SnO2 calcined at the same temperature,
and it must be concluded that this material was unreacted. Such
a material has been described as an unordered assembly of Sb-
and Sn-centered octahedra.4,7 The variation of conductivity with
reaction temperature could, in absence of other information, be
taken simply to illustrate the variation in degree of reaction with
calcination temperature. However, XPS demonstrates that the
effect is related to the temperature-dependent surface segregation
of Sb. Al KR excited photoelectron spectra in the region O1s
and Sb3d are shown in Figure 2. The Sb 3d5/2 peak overlaps
X-ray powder diffractrometry (Siemens D5000, transmission
mode with incident beam monochromator and Cu KR radiation)

6734 J. Phys. Chem. B, Vol. 102, No. 35, 1998
Dusastre and Williams
Figure 2. Photoelectron spectra of the O 1s and Sb 3d region for Sn0.995
-
Sb0.005O2 materials calcined at different temperature.
Figure 4. Photoemission difference spectra (undoped SnO2 - 0.5%
Sb doped SnO2) for materials calcined at different temperature. The
spectra were normalized to the intensity of the valence band maximum.
removed by Sb implantation. Photoemission difference spectra
(Sb-implanted - pure SnO2) were used to display this effect
clearly. Our XPS measurements in the valence band region
confirmed this effect for the polycrystalline powder material
and showed its correlation with Sb segregation (see Figure 4).
Materials fired at 800 °C which correspond to the highest Sb
segregation showed a significant difference peak at 3 eV binding
energy. This difference peak did not appear for materials
calcined at 600 °C, confirming the interpretation that this
material comprised a random array of disordered SnO6 and SbO6
octahedra,7 rather than a solid solution of Sb in SnO2. For
materials calcined at higher temperature the difference peak
intensity correlated with the surface segregation, being greatest
for largest surface Sb concentration. As noted in ref 33, the
difference spectrum at higher binding energy is dominated by
a shift to higher binding energy of the main O 2p valence band
structure (higher binding energy in the doped material). For
these polycrystalline, untextured materials, the difference spec-
trum implies that Sb doping also causes a broadening of this
peak as well as a shift.
Figure 3. Variation of surface composition with calcination temper-
ature at constant bulk concentration and calcination period (12 h).
with the O1s peak, but from the intensity and assuming the
Sb3d3/2/Sb3d5/2 intensity ratio, the atomic percent of Sb segrega-
tion could be calculated. Variation of the calcination temper-
ature produced systematic changes in surface composition
determined from XPS data (Figure 3). The surface enrichment
factor was observed to vary dramatically with bulk composition
(by a factor of 10 between 0.5 and 5% Sb-doped compounds at
800 °C) at constant calcination temperature. Both series of
doped materials (0.5 and 5% Sb) exhibited a firing temperature
of maximum Sb enrichment, the maximum shifting to higher
temperature (1000 °C instead of 800 °C) at higher antimony
concentration. If the effect were simply due to a variation in
degree of reaction, then a monotonic variation of Sb signal with
calcination temperature would be expected, rather than the
observed maximum. The enhancement in segregation with
decrease of temperature corresponds to a heat of segregation
∆Hseg ≈ -20 kJ mole-1, consistent with the results in ref 11.
XPS and UPS study of conduction electrons in Sb-implanted
SnO2 films showed32 that not all of the implanted Sb ions on
the surface acted as donor centers but that some behaved as
trapping centers. The mechanism for carrier trapping was
explained by consideration of photoemission structure in the
band gap region between the valence band (VB) and the
conduction band (CB).33 The significant structure was a small
shoulder at the valence band edge of the pure SnO2, which was
Elevated-Temperature Electrical Response to Changing
Water Vapor Pressure. The sensitivity of the electrical
conductivity to variations of water vapor pressure (SH O) is
2
defined as SH O ) (Rdry - Rwet)/Rwet. The sensitivity to water
2
as a function of partial pressure of the gas at 25 °C, determined
at 400 °C increased roughly linearly with the partial pressure
and decreased significantly with the antimony doping. Whereas
for undoped materials the sensitivity decreased monotonically
with the increase in crystallite size as the calcination temperature
increased, the Sb-doped materials behaved differently, exhibiting
an enhanced sensitivity (by a factor of 2) to water at a calcination
temperature of 800 °C. These critical measurements were
repeated for confirmation. Comparing SH O (50% relative
2
humidity (RH) at 20 °C ) 1244 Pa) at 400 °C (Figure 5) for
undoped and 0.5% Sb-doped materials shows the clear correla-
tion of SH O with surface antimony segregation.
2
Catalyzed Combustion. The reaction product of both
methane and carbon monoxide (dry and wet) over Sn1-ySbyO2

Sb(III) for Water Adsorption on Sb(Sb)O2
J. Phys. Chem. B, Vol. 102, No. 35, 1998 6735
Figure 5. Sensitivity, Swater ) (Rdry-Rwet)/Rwet of electrical conductivity
to water vapor (1244 Pa ≡ 50% relative humidity at 25 °C) measured
at 400 °C, as a function of calcination temperature of Sn1-ySbyO2 porous
pellets.
Figure 7. First-order rate constant, k1, for catalyzed oxidation of CO,
against temperature, T. The labels (20%, 50%, 80%) refer to the degree
of conversion of CO to CO2 on passage through the reactor bed.
Figure 6. Catalytic oxidation of CO and CH4 in a packed-bed, flow-
through reactor: mass 44 (CO2+) intensity against reactor temperature
for oxidation of 1000 ppm CO and 10000 ppm CH4 in 21% O2 + 79%
Ar over SnO2 calcined at 800 °C.
was carbon dioxide. The reaction of these gases was monitored
by looking at the mass 44 intensity as a function of temperature.
The onset temperature of methane combustion was around 450
°C for SnO2 (Figure 6) and around 500 °C for the antimony-
doped samples, and carbon monoxide oxidation occurred at
around 300 °C for all of our materials. At 400 °C the mass
spectrometry data indicated that 60% of the carbon monoxide
and virtually none of the methane had reacted upon passage
through the packed bed of the solid. The first-order rate constant
for methane combustion, calculated from these data, was
approximately 100 times greater than that given by Brown and
Patterson,4 consistent with the 100-times increase in oxygen
partial pressure.
Figure 8. Activation energy for CO oxidation in both wet (2500 Pa
≡ 100% relative humidity at room temperature) and dry gas against
calcination temperature of the Sn1-ySbyO2 catalyst.
with dry and wet CO is plotted against calcination temperature
in Figure 8. A clear maximum, correlating with the temperature-
dependent surface segregation of Sb, is shown. The effect of
Sb segregation is perhaps better described as causing, over the
temperature range of observation, an increase in rate of
combustion in wet gas from that characteristic of wet gas on
undoped SnO2 to that characteristic of dry gas on SnO2.
Figure 7 represents the variation of the pseudo-first-order rate
constant with temperature for surface-catalyzed combustion of
dry and wet CO for materials fired at different temperature. The
variation with temperature follows an Arrhenius form. The
consistency of results for the undoped SnO2 indicates a relative
error of no more than (10% in the rate constant normalized to
the surface area. On the undoped SnO2, the effect of introduc-
tion of water vapor was to decrease the rate but leave the
activation energy unchanged. In dry gas, the rate constant for
the Sb-doped materials was identical within experimental error
to that of the undoped SnO2. However, in wet gas, the activation
energy for combustion catalyzed on the Sb-doped materials was
increased to a degree dependent on the calcination temperature.
These critical measurements were repeated for confirmation.
The activation energy for combustion over all of the materials
Discussion
We have demonstrated that Sb(III) segregated to the surface
of SnO2 forms an electrically active surface state which interacts
with water. We have demonstrated an effect of this surface
interaction upon the rate of the surface-catalyzed combustion
of carbon monoxide.
Dramatic antimony segregation in SnO2 has previously been
demonstrated by XPS11 and it has been suggested32 that, because
not all the implanted Sb ions on the surface act as donor centers,
Sb could also act as a trapping center. The mechanism for
carrier trapping was supported by consideration of photoemis-
sion structure in the band gap region between valence and
conduction bands. SnO2 thin films have shown33 pronounced
photoemission intensity above the main valence band structure

6736 J. Phys. Chem. B, Vol. 102, No. 35, 1998
Dusastre and Williams
responsible for a shoulder that tails into the band gap region.
Experiments performed on SnO2(110)34 confirmed earlier sug-
gestions28 that this structure is to be associated with Sn ions at
surface sites that trap a pair of electrons to become Sn(II) rather
than Sn(IV). After implantation, the Sn(II) band gap structure
was significantly attenuated (compare with Figure 4.). The
evident changes in difference spectra suggest that antimony ions
substitute for surface Sn ions, and electrons are then trapped
into Sb3+-like surface states which lie at higher binding energy
(BE) than the Sn2+ states and overlap the main O 2p valence
band intensity. Hence, the energy of the Sn(II) surface state is
represented by a peak at 3 eV in the difference spectrum but it
is not possible to observe the Sb(III) surface state energy directly
in difference spectra.
A maximum Sb segregation was obtained at 800 °C for the
0.05% Sb-doped compounds (see Figure 3.). The enhancement
in segregation with decrease of temperature corresponds to a
heat of segregation ∆Hseg ≈ -20 kJ mol-1, consistent with the
results in ref 11. The maximum is consistent with the assertion
that, at the lowest reaction temperature, the product comprised
a random array of disordered SnO6 and SbO6 octahedra,7 rather
than a solid solution of Sb in SnO2, so there would be no
segregation. This assertion that at 600 °C firing the reaction to
form a solid solution was incomplete, is sound because the
valence band structure observed by XPS was almost indistin-
guishable from that of pure SnO2, and because the electrical
conductivity of this material was almost identical to that of pure
SnO2 calcined at the same temperature. At higher calcination
temperatures (1000 and 1200 °C) a decrease in Sb(III) concen-
tration was observed, from the maximum obtained at 800 °C,
which could be due to the diffusion of Sb3+ into the bulk of
the material. Our results on Sb surface segregation agree
quantitatively with those of Cross and Pyke,11 who showed that
for calcination temperatures greater than 400 °C the surfaces
became increasingly enriched in antimony, reaching a maximum
for calcination temperature close to 1000 °C, with a surface
enrichment factor of 7 with 4% atom of Sb and of 35 for 0.5%
atom of Sb.
variation in particle size with calcination temperature (paper in
preparation). Therefore, we have unequivocally demonstrated
that the sensitivity to one chemical species (water) is specifically
enhanced as a consequence of surface segregation changing the
surface composition.
Sb(III) is a surface trap state for electrons.28 The increase in
conductivity caused by water adsorption could therefore be
interpreted as water acting as an electron donor into this state.
This is a molecular adsorption of H2O, rather than the dissocia-
tive adsorption postulated by eq 1. Sn(II)surf is also a surface
trap for electrons, and the effect of water on undoped SnO2
could, therefore, also be due to molecular adsorption onto the
Sn(II) state. Later in this discussion we develop these ideas in
more detail, proposing that the surface adsorption sites are best
formulated as an association complex of an oxygen vacancy
with Sn(II) or Sb(III).
Water vapor had a very strong effect on the surface-catalyzed
oxidation of CO over doped and undoped SnO2. At a given
temperature (e.g., 400 °C) the rate of combustion normalized
to the surface area of all the materials for CO was greater under
dry conditions than wet. However, the effect of Sb segregation
was to cause a transition, over the temperature range of
observation, from behavior characteristic of “wet” SnO2 to that
characteristic of “dry” SnO2. We can develop an interpretation
which unifies the treatment of the electrical and the catalytic
behavior by following the arguments advanced by Brown and
Patterson.4 These authors showed that the exchange of oxygen
between the lattice and gaseous carbon dioxide or water was
very rapid, even at room temperature, and that, in the oxidation
of methane, the carbon dioxide formed incorporated equally
oxygen derived from the lattice and oxygen adsorbed onto the
surface from the gas. They showed that oxygen exchange
between gas and lattice required much higher temperature, and
that the reversible dissociative adsorption of oxygen contributed
negligibly to this exchange. They postulated that surface oxygen
vacancies mediated the oxygen exchange and that the catalyti-
cally active surface species was molecular oxygen adsorbed at
reduced Sn sites. These ideas contrast with those put forward
for the electrical behavior, which emphasize the importance of
dissociative ionosorption of oxygen. Thus, to interpret the
electrical behavior in the case of an n-type semiconducting
oxide, such as SnO2, the chemistry taking place at the surface
has been modeled by two main reactions. In the first reaction,
atmospheric oxygen becomes chemisorbed to the surface
consuming electrons
In the presence of gaseous oxygen, the elevated-temperature
electrical conductivity of SnO2 is determined by chemisorbed
oxygen, trapping conduction electrons in surface states formu-
lated as O2-, O-, and O2-. Gas sensitivity of the electrical
conductivity arises from changes in the surface concentration
of these states.24 Previous discussions starting from the observed
square-root dependence of conductivity on concentration of
reactive gas have particularly identified changes in the surface
concentration of the O2- state as being responsible for the gas
effects. Given this background, previous discussions24 of the
effect of water have postulated a transformation such as
1/2O2(g) + n′ { k1 } O-(ads)
(2)
k-1
Reducing gases such as CO, present in the ambient air,
produce a counter reaction
-
Oads2- + H2O f 2OHads
(1)
The trap state represented by OHads- is supposed to lie higher
in the band gap than that represented by O2-, to account for
the transformation causing an increase in conductivity. The
expected effect of substitution of an electron donor into the
lattice, giving an increase in conductivity, is to cause a decrease
in the sensitivity to gases. Thus, the expected trend in the
sensitivity to H2O, taking into account that the material calcined
at 600 °C was essentially unreacted, is shown by the dashed
line in Figure 5. Clearly, for a calcination temperature of 800
°C there was a significant enhancement of the electrical response
to H2O, correlating with the segregation of Sb and the formation
of the Sb(III) surface species. On these materials, the response
to CO and CH4, dry, showed only effects attributable to the
k2
CO + O-(ads)
98 CO2 + n′
(3)
where n′ is a conductance electron. The reducing gas removes
the chemisorbed oxygen, frees up an electronic carrier, n′, and
raises the conductivity of the sensor. Analysis of the conduc-
tance variation with gas concentration implies that the active
oxygen species is best formulated as O2- 24
As noted before,
.
the effects of water vapor have been discussed21,22,24 in terms
of the displacement of chemisorbed oxygen by H2O and OH-,
producing, on SnO2, a surface electronic state such as a surface
hydroxyl species, which lies higher in energy than the oxygen
species which is displaced.

Sb(III) for Water Adsorption on Sb(Sb)O2
J. Phys. Chem. B, Vol. 102, No. 35, 1998 6737
First, we postulate that a correct formulation of “adsorbed
oxygen” is an oxygen molecule trapped in or on a surface
oxygen vacancy. Following Brown and Patterson,4 we propose
that the combustion reaction proceeds partly through these
species and partly through lattice oxygen at the surface. Then
we propose that water competes with oxygen for the surface
vacancies, blocking this route. This accounts for the diminution
in the rate of the surface-catalyzed reaction in the presence of
water. To account for the effect of antimony segregation on
the reaction, we presume that the surface vacancy is correctly
formulated as an association of Sn(II) with the vacancy in the
absence of antimony, and of Sb(III) with the vacancy in the
presence of surface-segregated antimony. Then we propose that
the binding energy of water to the Sb(III)‚VO surface defect
complex is less than that to the Sn(II)‚VO complex. The
temperature variation of combustion rate would then correspond
simply to the desorption of water from the defect complex. The
apparent activation energy for combustion on the Sb-segregated
sample would measure the heat of adsorption of water onto the
surface defect.
identified consequences linked to the adsorption of water.
Discussion of these has suggested formulations for the surface
defect states which mediate both the electrical behavior and the
surface-catalyzed combustion.
Acknowledgment. We would like to thank the Engineering
and Physical Sciences Research Council of the UK, Capteur
Sensors and Analyzers Ltd., and the Ford Motor Company for
financial support of this work.
References and Notes
(1) Berry, F. J.; AdV. Catal. 1981, 30, 97.
(2) Weng, L. T.; Delmon, B. Appl. Catal. A 1992, 81, 141.
(3) McAteer, J.; J. Chem. Soc., Faraday Trans. 1 1979, 75, 2768.
(4) Brown, I.; Patterson, W. R., J. Chem. Soc., Faraday Trans. 1 1983,
79, 1431.
(5) Caldararu, M.; Popa, V. T.; Sprinceana, D.; Ionescu, N. I. Appl.
Catal. A 1995, 125, 247.
(6) Ono, T.; Yamanaka, T.; Kubokawa, Y.; Komiyama, M. J. Catal.
1988, 109, 423.
(7) Pyke, D. R.; Reid, R.; Tilley, R. J. D. J. Chem. Soc., Faraday Trans.
1 1980, 76, 1174.
(8) Volta, J. C.; Bussiere, P.; Coudurier, A.; Hermann, J. M.; Vedrine,
J. C. Appl. Catal. 1985, 16, 315.
Now we can make a connection between this interpretation
and the electrical behavior. First, an elementary reaction
between oxygen and a surface defect complex can be written:
(9) Wakabayashi, K.; Kamyia, Y.; Ohta, H. Bull. Chem. Soc. Jpn. 1967,
40, 2172.
(10) Trimm, D. L.; Gabbay, D. S. Trans. Faraday Soc. 1971, 67, 2782.
(11) Cross, Y. M.; Pyke, D. R. J. Catal. 1979, 58, 61.
(12) Crozat, M.; Germain, J. E. Bull. Soc. Chim. France 1973, 1, 1125.
(13) Atkinson, L.; Day, P. J. Chem. Soc. A 1969, 2423.
(14) Pyke, D. R.; Reid, R.; Tilley, R. J. D. J. Solid State Chem. 1978,
25, 1636.
2n/
9
[SnSn//‚VO2•] + O2 f [SnSn‚OO]-O
8 [SnSn‚OO]-O2-
(1)
(15) Orchard, A. F.; Thornton, G. J. Chem. Soc., Dalton Trans. 1977,
13, 1238.
The overall stoichiometry of this reaction can be represented
(16) Marley, J. A.; Dockerty, R. K. Phys. ReV. 1965, 140, A304.
(17) Cox, P. A.; Egdell, R. G.; Orchard, A. F.; Harding, C.; Patterson,
W. R.; Tavener, P. J. Solid State Commun. 1982, 44, 837.
(18) Hermann, J. M.; Portefaix, J. L.; Forissier, M.; Figueras, F.; Pichat,
P. J. Chem. Soc., Faraday Trans. 1 1979, 75, 1346.
(19) Cox, P. A.; Egdell, R. G.; Flavell, W. R.; Kemp, J. P.; Potter, F.
H.; Rastomjee, C. S. J. Electron. Spectrosc. 1990, 54/55, 1173.
(20) McAleer, J. F.; Moseley, P. T.; Norris, J. O. W.; Williams, D. E.;
Tofield, B. C. J. Chem. Soc., Faraday Trans. 1 1987, 83, 1323.
(21) Morrison, S. R. Sens. Actuators 1982, 2, 329.
(22) Heiland, G. Sens. Actuators 1982, 2, 343.
(23) Yamazoe, N.; Fuchigami, J.; Kishikawa, M.; Seiyama, T. Surf. Sci.
1979, 86, 335.
(24) Williams, D. E. In Solid State Gas Sensors; Moseley, P. T., Tofield,
B. C., Eds.; Adam Hilger: Bristol, 1987.
(25) Gercher, V. A.; Cox, D. F. Surf. Sci. 1995, 322, 177.
(26) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides;
Cambridge University Press: Cambridge, 1994; p 316.
(27) Egdell, R. G.; Flavell, W. R.; Tavener, P. J. J. Solid State Chem.
1984, 51, 345.
as
4n/ + VO•• + O2 f OO + O2-
ads
in which two electrons are trapped for each surface oxygen atom,
as the interpretation of the electrical behavior requires. Interest-
ingly, the mechanism as formulated here does not require
dissociation of molecular oxygen.
A dissociative adsorption of water onto the surface defect
complex is proposed:
[SnSn//‚VO••] + OO + HOH f [OH•O‚SnSn//‚OH•O] (2)
In comparison with eq 1, the surface electron trap in eq 2
may lie at higher energy in the band gap, so replacement with
water of oxygen adsorbed at the defect results in an increase of
conductivity. Where Sb is surface segregated, the surface defect
complex could be written [SbSn/‚VO••‚SbSn/]. A similar formula-
tion for the surface reactions can then be made. If the energy
difference between OH state and oxygen state is bigger on this
site, then the effect of water vapor on the conductivity would
be larger. Such a statement is consistent with water being less
strongly bound at this state than at the tin defect state represented
in eq 2, and is thus consistent with the interpretation of the
effect on combustion kinetics advanced above.
(28) Cox, P. A.; Egdell, R. G.; Harding, C.; Patterson, W. R.; Tavener,
P. J. Surf. Sci. 1982, 123, 179-203.
(29) Henshaw, G. S.; Gellman, L. J.; Williams, D. E. J. Mater. Chem.
1994, 4, 1427-1431.
(30) Wagner, C. D. In Practical Surface Analysis, 2nd ed.; Briggs, D.;
Seah, M. P., eds.; Auger and X-ray Photelectron Spectroscopy, Vol. 1;
Wiley: Chichester, 1990; p 595.
(31) Williams, D. E.; Henshaw, G. S.; Pratt, K. F. E.; Peat, R. J. Chem.
Soc., Faraday Trans. 1995, 91(23), 4299-4307.
(32) Rastomjee, C. S.; Egdell, R. G.; Lee, M. J.; Tate, T. J. Surf. Sci.
Lett. 1991, 259, L769.
(33) Rastomjee, C. S.; Egdell, R. G.; Georgiadis, G. C.; Lee, M. J.;
Tate, T. J. J. Mater. Chem. 1992, 2(5), 511.
In summary, the study of the effect on electrical response
and on combustion kinetics of Sb segregation in SnO2 has
(34) Themlin, J. M.; Sporken, R.; Darville, J.; Caudano, R.; Gilles, J.
M.; Johnson, R. L. Phys. ReV. B 1990, 42, 11914.