Two active sites for photocatalytic oxidation of formic acid on TiO2: Effects of H2O and temperature

Article abstract of DOI:10.1006/jcat.2002.3640

Photocatalytic oxidation (PCO) of formic acid on TiO₂ produces CO₂ without forming any long-lived intermediates. Formic acid PCO is not limited by surface diffusion of either formic acid or active oxygen species. The formic acid oxidation rate increases significantly with temperature up to 373 K, which was the highest temperature used in this study. At least two types of active sites exist for PCO of formic acid on TiO₂ and their activities differ by a factor of at least 20. The more-active sites comprise less than one-fourth of a room-temperature formic acid monolayer. Water, which is produced during PCO, redistributes adsorbed formic acid by displacement. Water readily displaces approximately one-third of a formic acid monolayer but it does not effectively compete for the remaining formic acid adsorption sites.

Full text of DOI:10.1006/jcat.2002.3640

Journal of Catalysis 209, 105–113 (2002)
doi:10.1006/jcat.2002.3640
Two Active Sites for Photocatalytic Oxidation of Formic Acid
on TiO2: Effects of H2O and Temperature
Darrin S. Muggli1 and Michael J. Backes
Department of Chemical Engineering, University of North Dakota, Grand Forks, North Dakota 58202-7107
Received November 29, 2001; revised April 10, 2002; accepted April 11, 2002
adsorbed intermediates and determine PCO mechanisms
(2, 4, 21–25), and design a photocatalyst with increased
selectivity to partial oxidation products (11). These tran-
sient studies provided information on the PCO surface pro-
cesses that would not be easily determined by steady-state
reaction.
Photocatalytic oxidation (PCO) of formic acid on TiO2 produces
CO2 without forming any long-lived intermediates. Formic acid
PCO is not limited by surface diffusion of either formic acid or
active oxygen species. The formic acid oxidation rate increases sig-
nificantly with temperature up to 373 K, which was the highest
temperature used in this study. At least two types of active sites
exist for PCO of formic acid on TiO2 and their activities differ by a
factor of at least 20. The more-active sites comprise less than one-
fourth of a room-temperature formic acid monolayer. Water, which
is produced during PCO, redistributes adsorbed formic acid by dis-
placement. Water readily displaces approximately one-third of a
formic acid monolayer but it does not effectively compete for the
Although transient reaction techniques have proven to
be useful in studying the surface processes during PCO, care
must be taken when interpreting transient reaction data.
Previous transient PCO studies (4, 12–14, 21, 24) measured
high PCO rates for carboxylic acids, indicating that they re-
act readily on TiO2. In contrast, steady-state studies (26–29)
using the same acids found that these species accumulate
on the surface, suggesting that they react slowly and may
poison the surface. This study analyzes transient PCO of
a model carboxylic acid, formic acid, to resolve the differ-
ences between transient and steady-state PCO studies. In
addition, studying formic acid PCO provides a better un-
derstandingofsurfaceprocessesandadsorptionsiteactivity
because formic acid oxidizes to CO2 without forming any
long-lived intermediates (1, 4), so that the surface reaction
rate can be measured directly. Specifically, this study will
show that multiple reaction sites exist for formic acid PCO
and formic acid adsorbed at less-active sites is displaced by
H2O to more-active sites during PCO.
c
remaining formic acid adsorption sites.
ꢀ 2002 Elsevier Science (USA)
INTRODUCTION
Volatile organic compounds (VOCs) are major contrib-
utors to air pollution. Many waste streams that contain
VOCs are dilute in air with the VOC concentration be-
low 1000 ppm (1–5). For low concentrations, an efficient
low-temperature remediation method is preferable to high-
temperature catalytic oxidation or thermal incineration.
Heterogeneous photocatalytic oxidation (PCO) effectively
removes low concentrations of organic contaminants at
room temperature from both gas and liquid phases by
oxidizing organics to environmentally benign compounds
(1–16). During PCO, near-UV irradiation of a semiconduc-
tor catalyst (usually TiO2) excites electrons from the va-
lence to the conduction band, leaving holes behind. These
electron–hole pairs migrate to the surface where they oxi-
dize adsorbed organics to CO2 and H2O.
EXPERIMENTAL METHODS
The transient and temperature-programmed reaction
system used in this study has been described previously
(13). An annular Pyrex reactor, which was identical to
those used in previous studies (13, 18), allowed for high
gas flow rates and uniform UV irradiation of the catalyst.
Approximately 30 mg of each catalyst was coated in a thin
layer (average thickness <0.4 µm) in the annular region
of the reactor. During PCO, 12 UV lamps (Johnlite, 8 W)
surrounded the reactor to uniformly irradiate the catalyst.
These lamps generate light in the 300 to 500 nm range with
a maximum intensity near 360 nm (30). A furnace made of
Ni–Cr wire wrapped around a quartz cylinder surrounded
the reactor. The end of a 0.5-mm chromel-alumel, shielded
Transient reaction is a powerful technique for probing
thesurfaceprocessesduringphotocatalyticoxidation. Tran-
sient reaction techniques have been employed to compare
activities of different catalysts (17, 18), correlate adsorption
sites with PCO behavior (3, 4, 11), determine the role of
lattice oxygen during PCO (12–14, 19, 20), identify strongly
1 To whom correspondence should be addressed. Fax: 701-777-3773.
E-mail: darrin muggli@und.nodak.edu.
105
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

106
MUGGLI AND BACKES
thermocouple contacted the catalyst film to provide feed-
back to the temperature programmer. A Balzers QMS 200
quadrupole mass spectrometer monitored the reactor efflu-
ent directly downstream of the reactor. A 25-µm ID fused
silica capillary, which fed directly to the mass spectrom-
eter ionizer, sampled the reactor effluent. This sampling
system provided high sensitivity and detected gas-phase
species rapidly. A computer that was interfaced with the
mass spectrometer recorded multiple mass peaks simul-
taneously. Calibration was performed by injecting known
amounts of each species into the gas flow downstream of
the reactor. Integration of the areas under the calibration
curves provided a means to convert mass spectrometer sig-
nals to reaction rates.
Before each isothermal PCO, the reactor was held for
20 min at 723 K in 100 sccm flow of 20% O2 in He (Praxair,
UHP)tocreateareproduciblesurface. FortransientPCOat
room temperature, three pulses (850 µmol/g catalyst each)
of formic acid (Aldrich, 96%) were injected upstream of
the reactor so that formic acid adsorbed onto the catalyst.
Transient PCO was carried out after the carrier gas flushed
excess gas-phase formicacidfromthe reactor. Afterashield
was placed between the lights and the reactor, the lights
were turned on and allowed to reach a steady output for ap-
proximately10min. Removingtheshieldsexposedthecata-
lyst to UV irradiation and initiated transient PCO. After
transient PCO for a specified time, the UV lights were
switched off and either temperature-programmed desorp-
tion (TPD) or oxidation (TPO) identified surface species
that did not desorb during PCO. TPD was carried out in
pure He flow. The Ni–Cr furnace heated the catalyst at 1 K/s
to 723 K and maintained this temperature until no products
were detected in the gas phase. After cooling the catalyst to
room temperature, a subsequent TPO oxidized species that
did not desorb during TPD. The TPO procedure was identi-
cal to that of TPD, except the carrier gas was 20% O2 in He.
FIG. 1. The CO2 formation rate (top) and normalized CO2 formation
rate (bottom) during transient PCO of a formic acid monolayer on TiO2
in 20% O2 at room temperature. The UV lights were turned on at 60 s and
turned off at 1800 s.
diates. After PCO stopped, a subsequent TPD confirmed
that only formic acid and H2O were adsorbed. Formic acid
coverage was determined by summing the amounts of CO2
that formed during PCO with those of CO and CO2 that
desorbed during subsequent TPO. For Degussa P-25 TiO2,
a monolayer of formic acid contained 375 25 µmol/g cata-
lyst, which is in good agreement with the value of 345 re-
ported by Muggli et al. (4).
RESULTS
Transient PCO
Figure 1 (top panel) shows the CO2 formation rate dur-
The CO2 formation rate curve in Fig. 1 (top panel) was
ing transient PCO of a monolayer of formic acid. After further analyzed to explore the activities of the formic
UV irradiation of the catalyst at 60 s, oxidation of adsorbed acid adsorption sites on TiO2. The CO2 formation rate
formic acid to CO2 quickly reached a maximum rate. The was normalized by dividing it by the amount of formic
CO2 formation rate decreased slowly during the first 50 s of acid that remained adsorbed at each sample point. For ex-
PCO, after which it decreased more rapidly. Monitoring the ample, the CO2 formation rate after 100 s of PCO was
CO2 formation rate is a direct measurement of the surface divided by the formic acid coverage at 100 s. Figure 1
reaction since CO2 appearance in the gas phase is reaction (bottom panel) plots this normalized rate versus the frac-
limited and not desorption limited. This was shown in sev- tion of a formic acid monolayer that formed CO2 to
eral experiments in which the UV lights were turned off show the change in catalyst activity with formic acid cov-
after 60–300 s of PCO and the CO2 formation rate dropped erage. The plot shows that the normalized CO2 forma-
quickly to zero. The immediate maximum in the CO2 for- tion rate increased from an initial value of 6 × 10−3 to a
mation rate (Fig. 1, top panel) indicates that formic acid maximum of 7 × 10−3 s−1 after 35% of the formic acid
oxidized to CO2 without forming any long-lived interme- monolayer formed CO2 and then subsequently decreased.

PHOTOCATALYTIC OXIDATION OF FORMIC ACID ON TiO2
107
0.016
After 80% of a formic acid monolayer oxidized, the nor-
malized CO2 formation rate remained constant at 4.0 ×
10−4 −1, which is more than an order of magnitude less
373 K
333 K
s
0.012
0.008
0.004
0
than the initial catalyst activity. Although the experiment
shown in Fig. 1 was stopped after 85% of a formic acid
monolayer reacted, other PCOs that were carried out
for longer times confirmed that the normalized rate re-
mained constant after 80% of the formic acid monolayer
reacted.
313 K
298 K
PCO with Heating
0
0.2
0.4
0.6
0.8
1
Several formic acid PCOs were carried out in which the
catalyst was heated to temperatures at or below 373 K
before or during PCO. TPD of formic acid showed that
formic acid does not react without UV at temperatures be-
low 400 K. Therefore, any change in PCO rate after heating
TiO2 to 373 K was not the result of formic acid reacting ther-
mally to form new surface species. To further determine
the effect of heating formic acid to 373 K, a monolayer
of adsorbed formic acid was heated to 373 K and cooled
to room temperature prior to performing PCO. Heating
to 373 K desorbed approximately 20% of the formic acid
monolayer, and during subsequent PCO (Fig. 2), the ini-
tial normalized CO2 formation rate was the same as a
formic acid monolayer. This indicates that heating to 373 K
did not cause adsorbed formic acid to react to a species
that oxidized more rapidly during PCO. Figure 2 shows
that the normalized CO2 formation rate started to de-
crease at the same coverage as did PCO of a formic acid
monolayer.
Fraction Reacted (mL)
FIG. 3. Normalized CO2 formation rates during transient PCO of
formic acid at various temperatures.
tures, the normalized rate started to decrease when formic
acid coverage was approximately equal to 65% of a mono-
layer.
In Figure 4, room-temperature PCO of formic acid was
stopped after 130 s of UV irradiation, the catalyst was
heated in the dark to 373 K, and PCO was resumed after
cooling to room temperature. The mass spectrometer de-
tected H2O but not formic acid desorption when the cata-
lyst was heated. This contrasts TPD of a formic acid mono-
layer in which weakly bound formic acid desorbed at 373 K.
Since formic acid coverage was less than a monolayer af-
ter 130 s of PCO, any weakly bound formic acid that des-
orbed during heating subsequently readsorbed on vacant
catalyst sites. Therefore, heating removed adsorbed H2O
and redistributed formic acid on the surface. When room-
temperature PCO resumed after heating and cooling in
the dark, catalyst activity more than doubled; the CO2
formation rate after the dark time was 2.6 times the rate
before the lights were turned off. The lights were turned
off and the catalyst was heated and cooled in the dark
two more times during the experiment; the CO2 forma-
tion rate after PCO resumed was 2.4 and 2.0 times greater
than before the lights were turned off for these two sub-
sequent heatings, respectively. Figure 4 shows that after
the first heating, the normalized CO2 formation rate in-
creased to a value that was 10% greater than the previous
maximum but it decreased quickly after reaching a maxi-
mum, in contrast to the slightly increasing normalized rate
initially. The increase in PCO rate after heating (Fig. 4)
may be due to either redistribution of formic acid on the
surface or removal of H2O which desorbed during heat-
ing. Additional experiments investigated the role of H2O
during PCO.
Figure 3 shows transient PCOs carried out at tempera-
tures ranging from 298 to 373 K. As temperature increased
from 298 to 373 K, the amount of adsorbed formic acid de-
creased from 375 to 315 µmol/g catalyst, but the maximum
CO2 formation rate doubled. Figure 3 shows that the initial
normalized CO2 formation rate at 373 K was approximately
2.3 times that at room temperature. At all PCO tempera-
0.008
Formic acid monolayer
0.006
0.004
0.002
Heated before PCO
0
0
0.2
0.4
0.6
0.8
Fraction Reacted
PCO of Formic Acid and H2O
FIG. 2. Normalized CO2 formation rates during room-temperature
transient PCO of formic acid with and without preheating the catalyst to
373 K.
Figure 5 shows PCO of formic acid with and without
coadsorbed H2O. Injecting 2 and 8 µL of H2O displaced 24

108
MUGGLI AND BACKES
respectively. Injecting 2 µL of H2O decreased formic acid
coverage by 90 µmol formic acid/g catalyst but the amount
of H2O that desorbed was the same as a formic acid mono-
layer. Since one molecule of adsorbed formic acid produces
one molecule of H2O during TPD, the amount of H2O ad-
sorbed equaled the amount of formic acid displaced. Simi-
larly, when 8 µL of H2O was injected, H2O displaced formic
acid in a 1 : 1.3 ratio. Another experiment repeated the ad-
sorption procedure of Fig. 5, but TPD rather than PCO
was carried out to determine if H2O changes formic acid
adsorption.Coadsorbing H2O decreased formic acid cover-
age due to displacement, but it did not otherwise change the
formic acid TPD spectra; all TPD products were detected
at the same temperatures and the desorption curves had
the same shapes with and without H2O. Changing the or-
der of adsorption (adsorbing H2O before formic acid) did
not change the resulting TPD spectra.
When 2 µL of H2O was injected prior to PCO (Fig. 5), the
normalized rate reached a maximum of 0.006 s−1, remained
fairly constant as PCO consumed approximately 10% of
a formic acid monolayer, and then decreased throughout
PCO. Quadrupling the amount of injected H2O only dis-
placed an additional 6% of a formic acid monolayer and the
normalized rate reached the same maximum of 0.006 s−1
.
However, the CO2 formation rate started to decrease al-
most immediately after reaching a maximum. For all three
experiments, the normalized rate was the same initially and
it started to decrease at the same formic acid coverage. At
longer reaction times, the normalized CO2 formation rates
were identical for PCO with and without coadsorbed H2O.
FIG. 4. CO2 formation rates (top) and normalized CO2 formation
rates (bottom) during transient PCO of formic acid. The lights were turned
off (black arrows) and on (white arrows) during the experiment. During
the dark periods, the catalyst was heated to 373 K and then cooled to room
temperature.
PCO of 12C- and 13C-Formic Acid
Figure6showsPCOofamonolayerof 12C-formicacidfor
75 s, followed by an injection of 0.2 µL of 13C-formic acid
in the dark before the resumption of PCO. To minimize
displacement of 12C-formic acid, the amount of injected
and 30% of the formic acid monolayer, respectively. Dur-
ing subsequent TPOs, the amounts of H2O that desorbed
were 410, 410, and 440 µmol H2O/g catalyst after PCO of
a formic acid monolayer with 0-, 2-, and 8-µL H2O pulses,
FIG. 6. CO2 formation rates during transient PCO of formic acid. The
UV lights were turned off (black arrow) at 100 s and 13C-formic acid was
injected in the dark before PCO resumed at 430 s (white arrow).
FIG. 5. Normalized CO2 formation rates during room-temperature
transient PCO of formic acid and coadsorbed H2O/formic acid.

PHOTOCATALYTIC OXIDATION OF FORMIC ACID ON TiO2
109
13C-formic acid was an order of magnitude less than other to the surface did not limit transient PCO; when the UV
pulses used in this study. Similar to Fig. 1, the 12CO2 for- lights were turned off and back on during PCO, the CO2
mation rate reached an immediate maximum, decreased formation rate after the dark time was the same as before
slowly for approximately 30 s, and then dropped more the UV lights were turned off. If O2 diffusion to the sur-
quickly as PCO continued. After 75 s of UV irradiation, face limited PCO, O2 would have diffused to the surface in
approximately 180 µmol/g catalyst of 12C-formic acid ox- the dark and the CO2 formation rate would have increased
idized to CO2, leaving approximately 170 µmol/g catalyst when the lights were turned back on. This experiment also
adsorbed. AftertheUVlightswereturnedoff, injecting13C- indicates that surface diffusion of formic acid to active sites
formic acid in the dark resupplied the sites left vacant by did not limit PCO rate because formic acid also would have
PCO;massbalancesshowedthatapproximately170 µmol/g diffused to active sites in the dark.
catalyst of 13C-formic acid adsorbed. When the UV lights
During transient PCO of a formic acid monolayer (Fig. 1,
were turned back on at 430 s, the sum of the 12CO2 and bottom panel), the normalized CO2 formation rate de-
13CO2 formation rates was equal to the initial 12CO2 for- creased dramatically as formic acid coverage decreased.
mation rate, which indicates that catalyst activity after the The sharp drop in the normalized CO2 formation rate dur-
dark time was the same as that initially. Figure 6 shows that ing PCO might be explained by surface diffusion of active
the 12CO2 formation rate was the same before and after surface oxygen species limiting the PCO rate at low formic
the lights were turned off but that after the dark time it acid coverage. Formic acid that is adsorbed near oxygen ad-
decreased more slowly than before for approximately 30 s sorption sites would react initially and then oxygen would
and then decreased rapidly. The 13CO2 formation rate also surface diffuse to formic acid adsorbed at greater distances
quickly reached a maximum, decreased slowly for 30 s, and from the oxygen adsorption sites as PCO proceeds. How-
subsequently dropped rapidly. Note that after adsorbing ever, when PCO was carried out at one-half the typical UV
13C-formic acid, the 13CO2 formation rate was 1.35 times intensity, the CO2 formation rate was 59% of that at full
that of the 12CO2 rate, even though 12C- and 13C-formic intensity, and the shapes of the normalized CO2 formation
acid coverages were nearly the same.
rate curves were exactly the same. That is, after multiplying
the normalized CO2 formation rate at half UV intensity by
1.7, the curves were coincident at all coverages. If oxygen
surface diffusion limits PCO at low formic acid coverages,
the normalized rate should begin to decrease at a lower
coverage during PCO at one-half UV intensity because
decreasing UV intensity would slow the surface reaction
rate but not the rate of oxygen surface diffusion. Because
the normalized CO2 formation rate curves have exactly the
same shape for different UV intensities, surface diffusion
of active oxygen species does not limit PCO rate. In ad-
dition, this experiment confirms that diffusion of reactants
and products does not limit PCO rate.
DISCUSSION
Mass Transfer and Other Experimental Artifacts
The behavior of P-25 (Fig. 1) during transient PCO does
not appear to be due to nonuniform UV irradiation of the
P-25 surface. Repeat experiments for catalysts with aver-
age film thicknesses ranging from 1.5 to 5 µm produced the
same results as in Fig. 1. In addition, a previous study (31)
of acetaldehyde PCO on several TiO2 catalysts showed that
the normalized CO2 formation rates on some catalysts were
approximately constant with coverage, indicating that UV
irradiation was essentially uniform. In this study, the same
coating procedure was used, the TiO2 films were approxi-
mately the same thickness, and the UV lights and reactor
were fixed in the same positions as in the previous study.
Active Sites
If all formic acid adsorption sites were equally active, the
normalized rate in Fig. 1 (bottom panel) would be constant
Several experiments determined that the PCO rate data throughout PCO. If at least two adsorption sites of differ-
were free of mass transfer limitations. Adsorbing formic ing activity were present, the normalized rate would de-
acid in the dark ensured that the rate of formic acid diffu- crease initially, as formic acid was consumed on the more
sion and adsorption to the surface did not affect the PCO active sites, and eventually reach a constant value when
rate. Since CO2 adsorbs weakly to TiO2 at room temper- only formic acid adsorbed on less-active sites remained.
ature, its appearance in the gas phase is reaction limited; However, Fig. 1 (bottom panel) shows that the normal-
when the UV lights were turned off during PCO, the CO2 ized rate increased slightly during the initial stages of PCO
formation rate dropped quickly to zero, indicating that CO2 and reached a maximum after 35% of a formic acid mono-
desorption and diffusion are fast compared to the surface layer formed CO2. The initial increase in the normalized
reaction rate. Since CO2 formation is reaction limited and rate is attributed to H2O, which formed during PCO and
formic acid produces CO2 without forming any long-lived redistributed adsorbed formic acid. The normalized rate
intermediates, monitoring the CO2 formation rate during increased initially because formic acid displacement resup-
transient PCO is a direct measurement of the surface reac- plied the more-active sites and the CO2 formation rate was
tion rate. Another experiment determined that O2 diffusion normalized by dividing it by the total formic acid coverage

110
MUGGLI AND BACKES
at each point in time. Further evidence of H2O displacing on the more-active sites and produce high reaction rates
adsorbed formic acid will be discussed in the following. in transient studies. After steady-state PCO, however, car-
When nearly all of the formic acid that is easily displaced boxylic acids adsorbed on less-active sites are expected to
by H2O was removed, the normalized rate decreased and be present on the surface since they react slowly on these
eventually reached a constant value, which is consistent sites and are more strongly bound than the original reac-
with two types of active sites. At low formic acid coverage, tants (which were aromatics and alcohols in these studies).
the normalized rate was more than an order of magnitude Therefore, carboxylic acids are expected to be on the sur-
less than the initial rate. At long reaction times, only formic face during PCO regardless as to whether or not they poison
acid on the less-active sites remained and therefore the nor- the catalyst.
malized rate decreased at the same rate as coverage, pro-
ducing a constant normalized rate (Fig. 1, bottom panel).
The variation in site activity may be due to differences in
Effect of H2O during Transient PCO
Understanding theroleofH2Oisimportant since it forms
and remains adsorbed during transient PCO of formic acid.
Photocatalytic oxidations with coadsorbed H2O and formic
acid (Fig. 5) clearly showed that H2O displaced adsorbed
formic acid. Muggli et al. (3) determined that approximately
750 µmol H2O/g catalyst adsorbed on P-25 at room temper-
ature. After an injection of 2 µL of H2O, which corresponds
to approximately 5 monolayer equivalents of H2O, only
90 µmol/g catalyst H2O adsorbed, which is 12% of an H2O
monolayer. Each adsorbed molecule of H2O displaced one
molecule of formic acid so that approximately three-fourths
of a formic acid monolayer remained adsorbed. When
the amount of injected H2O quadrupled to approximately
20 monolayer equivalents, H2O displaced formic acid in a
1 : 1.3 ratio. Quadrupling the amount of pulsed H2O only
displaced 20% more formic acid. That is, injecting 20 mono-
layer equivalents of H2O displaced only 110 µmol formic
acid/g catalyst, which is approximately 30% of the adsorbed
formic acid monolayer. This suggests that H2O readily dis-
places approximately one-third of a formic acid monolayer
but competes less effectively for the sites to which the re-
maining formic acid adsorbed. Weakly bound formic acid
that is easily displaced by H2O is most likely molecularly ad-
sorbed, as indicated by Liao et al. (33). They studied formic
acid adsorption with FTIR and reported that 30% of the
adsorbed formic acid was molecularly adsorbed and the re-
maining 70% adsorbed dissociatively as formate.
As discussed previously, the increasing normalized CO2
formation rate during the initial stages of PCO (Fig. 1, bot-
tom panel) was attributed to H2O displacing formic acid,
which replenished the more-active sites. When most of the
formic acid that is easily displaced by H2O was removed,
the normalized CO2 formation rate decreased quickly since
the more-active PCO sites were not being resupplied with
formic acid as readily as before. Coadsorbed H2O/formic
acid PCOs (Fig. 5) also show that injecting H2O removed
formic acid that is easily displaced so that the normalized
rate started to decrease at the same coverage as did a formic
acid monolayer. These results are also similar to those of
Figs. 2 and 3, where heating removed a portion of adsorbed
formic acid. That is, when formic acid coverage was approx-
imately 65% of a monolayer, the normalized rate started to
decrease regardless of whether the means used to reduce
formic acid adsorption, but TPD was not sensitive enough
to determine different forms of adsorbed formic acid.
Popova et al. (32) used FTIR spectroscopy to study formic
acid adsorption on anatase TiO2. They determined that
formic acid adsorbs both molecularly and dissociatively,
in agreement with Liao et al. (33). In addition, they iden-
tified two types of dissociative adsorption: unsymmetrical
and bidentate formate. These spectroscopy studies indicate
that TiO2 adsorbs formic acid molecularly and dissocia-
tively at two formate adsorption sites. The primary role of
weakly bound, molecularly adsorbed formic acid appears
to be to resupply the formate adsorption sites, which are
active for PCO. The two forms of adsorbed formate react
at significantly different rates. These two formate adsorp-
tion sites will be hereafter referred to as the active sites and
their activities will be discussed in the following.
Figure 6 also shows that at least two types of active sites
exist for formic acid PCO. When 13C-formic acid was ad-
sorbed after 130 s of PCO of 12C-formic acid, the 13CO2
formation rate was approximately 35% greater than that of
12CO2, even though the coverages of 12C- and 13C-formic
acid were approximately the same. This indicates that 13C-
formic acid preferentially adsorbed on more-active sites
that were left vacant after the 130 s of PCO prior to 13C-
formic acid injection. In addition, the total CO2 formation
rate after 13C-formic acid injection was the same as that ini-
tially, indicating that the drop in CO2 formation rate during
PCO is not due to deactivation.
The identification of two active sites for carboxylic acid
PCO explains differences between steady-state and tran-
sient PCO studies. For example, previous steady-state stud-
ies (26, 27) of benzene and toluene PCO proposed that
benzoic acid accumulates on the surface and poisons sites,
whereas transient PCO studies (21, 24) found benzoic acid
to react readily. Similarly, accumulation of formic and acetic
acids has been proposed to block adsorption sites at steady
state (28, 29), whereas transient PCO showed that both
acids react quickly to CO2 (4, 12–14). The initial rates dur-
ing transient PCO primarily measure the activity of the
sites that are more active during PCO. In contrast, steady-
state studies identified carboxylic acids on the surface after
PCO, which is presumably due to carboxylic acids adsorbed
on less-active sites. That is, carboxylic acids react readily

PHOTOCATALYTIC OXIDATION OF FORMIC ACID ON TiO2
111
coverage was PCO (Fig. 1, bottom panel), H2O injections
(Fig. 5), or heating (Figs. 2 and 3).
In contrast to the coadsorbed H2O/formic acid experi-
ments, when H2O was pulsed into the reactor after 300 s
of transient PCO of formic acid, the CO2 formation rate
increased fourfold. This also appears to be due to H2O dis-
placing formic acid. When H2O was pulsed into the reactor
during PCO, formic acid that remained on the surface was
adsorbed primarily on less-active sites since formic acid on
more-active sites already reacted. The H2O pulse displaced
formic acid, which presumably readsorbed on the vacant
more-active sites and thus increased the CO2 formation
rate. A similar experiment was performed in which formic
acid PCO was stopped after 80 s and H2O was injected
in the dark; when PCO resumed, the CO2 formation rate
had approximately doubled. In both experiments, inject-
ing H2O apparently redistributed formic acid. At the time
of H2O injection, formic acid coverage on the more-active
sites was low whereas coverage on less-active sites was high
due to the large differences in activity. Therefore, since H2O
redistributed formic acid on the surface, the net effect of
H2O was to transport formic acid from less-active to more-
active sites. Theseresults are similartothose shown inFig. 4,
where heating the catalyst to 373 K in the dark redistributed
adsorbed formic acid. That is, plots of normalized rates ver-
sus fraction of formic acid that reacted were nearly identical
for PCO interrupted by either heating or injecting H2O in
the dark. This indicates that injecting H2O produced the
same effect as heating to 373 K; formic acid was redis-
tributed on the surface. Another experiment repeated the
procedure of that shown in Fig. 4 but a 0.2-µL pulse of H2O,
injected after heating the catalyst to 373 K and cooling to
room temperature in the dark, replenished the surface with
H2O after heating removed some of the adsorbed H2O.
When PCO resumed, the CO2 formation rate was nearly
the same as that after the first dark period in Fig. 4, indicat-
ing that the increase in PCO rate in Fig. 4 was not due to
the removal of adsorbed H2O.
FIG. 7. Decay curve for CO2 formation with two-site model fit.
creased initially. As discussed previously, this increase is
most likely due to displacement of weakly bound formic
acid by H2O. After 35% of a formic acid monolayer formed
CO2, the normalized rate decreased quickly, apparently be-
cause most of the weakly bound formic acid had been dis-
placed by H2O. This portion of the normalized CO2 for-
mation rate was fit to a two-site model (Fig. 7) to estimate
the number and activity of less- and more-active sites. The
following two-site model assumes constant oxygen concen-
tration and therefore first-order surface reactions:
k1
HCOOH(site 1) ⇒ CO2 + H2O
k2
HCOOH(site 2) ⇒ CO2 + H2O.
Because mass balances determine the total formic acid cov-
erage, the model contains three adjustable parameters: k1,
k2, and the ratio of formic acid coverage on sites 1 and 2. The
Figure 6 also suggests that H2O, produced during PCO
of formic acid, displaced adsorbed formic acid to more-
active sites. After PCO of 12C-formic acid and injection of
13C-formic acid in the dark, the 12CO2 formation rate was
the same before and after the dark period, indicating that
injecting 13C-formic acid did not displace 12C-formic acid.
Even though the 12C-formic acid was adsorbed on the same
sites, the 12CO2 formation rate decreased more slowly when
PCO resumed than before the dark period. Adsorbing 13C-
formic acid increased formic acid coverage and therefore
increasedtheratethatPCOproducedH2O, whichdisplaced
12C-formic acid from less-active to more-active sites.
value determined for k1 (2 × 10−2
s
−1) was approximately
20 times that of k2 (9 × 10−4 s−1), indicating a substantial dif-
ference in site activities. This model predicts that the more-
active sites comprise approximately one-fourth of the total
number of formic acid adsorption sites. The predicted cov-
erage assumes that when the normalized CO2 formation
rate started to decrease in Fig. 1 (bottom panel), the formic
acid coverage on the more-active sites was approximately
90%. This estimate was based on the large difference in site
activities predicted by this model and a CO2 formation rate
that was 90% of the initial rate.
As mentioned previously, H2O displaces formic acid,
which mainly occurs during the initial stages of reaction.
To obtain the simplest model that adequately fits the data,
the model used here does not include this displacement,
which is less important at lower coverages. Since H2O
Two-Site Model
Figure 1 (bottom panel) showed that during transient
PCO of a formic acid monolayer, the normalized rate in-

112
MUGGLI AND BACKES
displacementredistributesformicacidonthesurface, which
in effect transports formic acid from high coverage regions
(less-active sites) to low coverage regions (more-active
sites), this model overestimates site 1 coverage and under-
estimates both site 2 coverage and the difference in site
activities. Thus the values reported here represent lower
limits for site 1 activity and site 2 coverage and upper limits
on site 1 coverage and site 2 activity.
CONCLUSIONS
DuringPCOonTiO2, formicacidoxidizestoCO2 inasin-
glestepwithoutforminglong-livedintermediates. Lessthan
one-fourth of the formic acid adsorption sites are highly
active for PCO; these sites are at least 20 times more ac-
tive than other formic acid adsorption sites. Water readily
displaces approximately one-third of a formic acid mono-
layer whereas it does not effectively compete for adsorption
sites with the remaining formic acid. Mass balances showed
that, on average, each adsorbed H2O molecule displaces
one formic acid molecule. Water that is produced during
PCO redistributes adsorbed formic acid on the surface by
displacement. During transient PCO, redistributing surface
coverage transports formic acid from less-active to more-
active sites, which enhances the formic acid oxidation rate.
Adsorbed H2O does not poison the PCO activity of TiO2.
Heating TiO2 to 373 K also redistributes adsorbed formic
acid and the maximum CO2 formation rate during PCO at
373 K was twice that at room temperature.
Effect of Heating
Heating TiO2 from room temperature to 373 K (Fig. 3)
dramatically increased PCO activity; although formic acid
coverage at 373 K was 85% of that at room temperature,
the initial CO2 formation rate doubled. At longer reac-
tion times, the effect of heating was more dramatic. For
example, when one-fourth of a formic acid monolayer re-
mained adsorbed, the normalized CO2 formation rate at
373 K was eight times that at room temperature. This may
indicate that formic acid oxidation is activated and the ac-
tivation energy of the less-active sites is greater than that
of the more-active sites, or formic acid adsorbed at less-
active sites is more readily transported to more-active sites
at 373 K. Onishi et al. (34) studied formic acid decompo-
sition on TiO2. They observed that when part of a formic
acid monolayer was removed at 350 K, an ordered (2 × 1)
formic acid adsorption structure could not be maintained
due to surface diffusion. They proposed that formic acid
may surface diffuse above 350 K and noted that the melt-
ing point of TiHCOO was only 374 K. However, in an
experiment in which the UV lights were turned off for
1800 s and then back on after 180 s of transient PCO at
373 K (data not shown), the CO2 formation rate was the
same before and after the dark period. If the increase in
the PCO rate at 373 K was due to faster formic acid sur-
face diffusion, formic acid would have diffused to more
active sites in the dark and therefore the CO2 formation
rate would have increased when PCO resumed. Therefore,
formic acid surface diffusion does not appear to affect the
formic acid PCO rate at temperatures between 273 and
373 K.
ACKNOWLEDGMENT
Acknowledgment is made to the Donors of the Petroleum Research
Fund, administered by the American Chemical Society, for support of this
research.
REFERENCES
1. Nimlos, M. R., Wolfrum, E. J., Brewer, M. L., Fennell, J. A., and
Bintner, G., Environ. Sci. Technol. 30, 3102 (1996).
2. Muggli, D. S., Larson, S. A., and Falconer, J. L., J. Phys. Chem. 100,
15886 (1996).
3. Muggli, D. S., Lowery, K. H., and Falconer, J. L., J. Catal. 180, 111
(1998).
4. Muggli, D. S., McCue, J. T., and Falconer, J. L., J. Catal. 173, 470 (1998).
5. Jacoby, W. A., Blake, D. M., Fennell, J. A., Boulter, J. E., Vargo, L. M.,
George, M. C., and Dolberg, S. K., J. Air Waste Manage. Assoc. 46, 891
(1996).
6. Sauer, M.-L., and Ollis, D.-F., J. Catal. 149, 81 (1994).
7. Vorontsov, A. V., Barannik, G. B., Snegurenko, O. I., Savinov, E. N.,
and Parmon, V. N., Kinet. Catal. 38, 84 (1997).
8. Peral, J., and Ollis, D.-F., J. Catal. 136, 554 (1992).
9. Cunningham, J., Hodnett, B. K., and Walker, A., Proc. R. Irish Acad.
77B, 411 (1977).
10. Kennedy, J. C., and Datye, A. K., J. Catal. 179, 375 (1998).
11. Muggli, D. S., and Falconer, J. L., J. Catal. 175, 213 (1998).
12. Muggli, D. S., and Falconer, J. L., J. Catal. 187, 230 (1999).
Although Fig. 3 shows that increasing temperature de-
creased the initial formic acid coverage, the normalized
rate started to decrease at about the same coverage for all
reaction temperatures. As mentioned previously, the nor- 13. Muggli, D. S., and Falconer, J. L., J. Catal. 191, 318 (2000).
14. Muggli, D. S., Keyser, S. A., and Falconer, J. L., Catal. Lett. 55, 129
malized CO2 formation rate decreased due to depletion of
formic acid that is easily displaced by H2O. Because the
normalized rates decreased at the same coverages for each
PCO temperature, the formic acid that was removed by
(1998).
15. Muggli, D. S., and Falconer, J. L., J. Catal. 181, 155 (1999).
16. Jacoby, W. A., Nimlos, M. R., and Blake, D. M., Environ. Sci. Technol.
28, 1661 (1994).
heating in Fig. 3 appears to be the same formic acid that is 17. Blount, M. C., Kim, D. H., and Falconer, J. L., Environ. Sci. Technol.
35, 2988 (2001).
easily displaced by H2O. Similarly, Fig. 2 shows that heat-
ing to 373 K prior to room-temperature PCO removed a
portion of the weakly bound formic acid and the normal-
ized CO2 formation rate started to decreased at the same
coverage as that of PCO of a formic acid monolayer.
18. Muggli, D. S., and Ding, L. F., Appl. Catal. B: Environ. 32, 181 (2001).
19. Lee, G. D., and Falconer, J. L., Catal. Lett. 70, 145 (2000).
20. Lee, G. D., Tuan, V. A., and Falconer, J. L., Environ. Sci. Technol. 35,
1252 (2001).
21. Blount, M. C., and Falconer, J. L., J. Catal. 200, 21 (2001).

PHOTOCATALYTIC OXIDATION OF FORMIC ACID ON TiO2
113
22. Blount, M. C., Buchholz, J. A., and Falconer, J. L., J. Catal. 197, 303
28. Sauer, M. L., and Ollis, D. F., J. Catal. 158, 570 (1996).
(2001).
29. Obuchi, E., Sakamoto, T., Nakano, K., and Shiraishi, F., Chem. Eng.
Sci. 54, 1525 (1999).
23. Larson, S. A., and Falconer, J. L., Appl. Catal. B: Environ. 4, 325
(1994).
30. Kozlov, D. V., Paukshtis, E. A., and Savinov, E. N., Appl. Catal. B:
Environ. 24, L7 (2000).
31. Muggli, D. S., Ding, L., and Odland, M. J., Catal. Lett. 78, 23 (2002).
32. Popova, G. Y., Andrushkevich, T. V., Chesalov, Y. A., and Stoyanov,
E. S., Kinet. Catal. 41, 805 (2000).
33. Liao, L. F., Wu, W. C., Chen, C. Y., and Lin, J. L., J. Phys. Chem. B 105,
7678 (2001).
34. Onishi, H., Aruga, T., and Iwasawa, Y., J. Catal. 146, 557 (1994).
24. Larson, S. A., and Falconer, J. L., Catal. Lett. 44, 57 (1997).
25. Larson, S. A., Widegren, J. A., and Falconer, J. L., J. Catal. 157, 611
(1995).
26. Cao, L. X., Gao, Z., Suib, S. L., Obee, T. N., Hay, S. O., and Freihaut,
J. D., J. Catal. 196, 253 (2000).
27. Mendez-Roman, R., and Cardona-Martinez, N., Catal. Today 40, 353
(1998).