Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1)

Article abstract of DOI:10.1016/j.cattod.2015.03.033

Adsorption and reaction of acetic acid on a CeO₂(1 1 1) surface was studied by a combination of ultra-high vacuum based methods including temperature desorption spectroscopy (TPD), soft X-ray photoelectron spectroscopy (sXPS), near edge X-ray absorption spectroscopy (NEXAFS) and reflection absorption IR spectroscopy (RAIRS), together with density functional theory (DFT) calculations. TPD shows that the desorption products are strongly dependent upon the initial oxidation state of the CeO₂ surface, including selectivity between acetone and acetaldehyde products. The combination of sXPS and NEXAFS demonstrate that acetate forms upon adsorption at low temperature and is stable to above 500 K, above which point ketene, acetone and acetic acid desorb. DFT and RAIRS show that below 500 K, bridge bonded acetate coexists with a moiety formed by adsorption of an acetate at an oxygen vacancy, formed by water desorption.

Full text of DOI:10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
Catalysis Today xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Reactivity and reaction intermediates for acetic acid adsorbed on
CeO2(1 1 1)
Florencia C. Calazaa,1, Tsung-Liang Chena,2, David R. Mullinsa, Ye Xub,
Steven H. Overburya,∗
a Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA
b Department of Chemical Engineering, Louisiana State University, Baton Rouge, LA 70803, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Adsorption and reaction of acetic acid on a CeO2(1 1 1) surface was studied by a combination of ultra-high
vacuum based methods including temperature desorption spectroscopy (TPD), soft X-ray photoelectron
spectroscopy (sXPS), near edge X-ray absorption spectroscopy (NEXAFS) and reflection absorption IR
spectroscopy (RAIRS), together with density functional theory (DFT) calculations. TPD shows that the
desorption products are strongly dependent upon the initial oxidation state of the CeO2 surface, including
selectivity between acetone and acetaldehyde products. The combination of sXPS and NEXAFS demon-
strate that acetate forms upon adsorption at low temperature and is stable to above 500 K, above which
point ketene, acetone and acetic acid desorb. DFT and RAIRS show that below 500 K, bridge bonded
acetate coexists with a moiety formed by adsorption of an acetate at an oxygen vacancy, formed by
water desorption.
Received 16 December 2014
Received in revised form 28 February 2015
Accepted 9 March 2015
Available online xxx
Keywords:
CeO2(1 1 1)
Acetic acid adsorption
Ketonization
Surface chemistry
Oxide catalysis
DFT
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
processes occurring on the surface, and identity of adsorbed surface
species present on the surface during reactions.
Cerium oxide has been widely recognized as a catalytic mate-
rial and employed in catalytic systems primarily because of its
reducibility. This property makes it able to store and release oxygen,
and so is used during changes in catalytic cycles in an automotive
three-way catalysts. Working with a precious metal catalyst, CeO2
and especially mixed Ce-based oxides, can buffer oxygen availabil-
ity during oxidation of CO, reduction of NO, and water gas shift
reactions [1]. However, ceria is also interesting for its acid–base
properties that make it able to catalyze reactions such as dehydro-
genation, ketonization, and dehydration reactions even without the
necessity for a catalytic metal [2]. In these reactions that typically
involve organic oxygenates, the redox capability of the CeO2 may
enter into the reaction pathways, and this can be expected to lead
to a selectivity that varies depending upon the reduction of the
ceria surface. It is difficult to discern these changes for polycrys-
talline CeO2 under reactor conditions but studies of model surfaces
can help relate the reaction selectivity to surface structure, redox
With this goal in mind, we and others have explored reac-
tion pathways for C1 and C2 oxygenates reacting at single crystal
surfaces [3] and CeO2 nanoparticles with well-defined crystallo-
graphic terminations [4,5]. In studies of ethanol dehydrogenation
by these CeO2 nanoparticles, dehydrogenation to acetaldehyde
was favored over dehydration to ethylene but the selectivity was
found to depend upon surface structure [4]. Sensitivity to struc-
ture and to the extent of reduction was found in TPD studies from
CeO2(1 0 0) and CeO2(1 1 1) of ethanol and other alcohols [6]. Reac-
tions of acetaldehyde demonstrated weak interactions with a fully
oxidized CeO2(1 1 1) but introduction of vacancies by reduction
leads to formation of the enolate form [7]. This enolate is identi-
fied as an intermediate for coupling reaction to crotonaldehyde, the
primary reaction product observed over CeO2 nanoparticles. Dur-
ing in situ DRIFTS studies of CeO2 nanoparticles during reaction,
targeted to identify reaction intermediates and pathways, it was
found that the oxidized product of these reactions, acetate, was
frequently observed. This is of practical importance, since it has
been shown that carboxylates are the primary intermediate from
reactions of aldehydes that lead to long chain ketones via ketoniza-
tion [8]. Cerium oxide has been shown to ketonize acetic acid to
acetone, but depending upon conditions the ceria can undergo
transformation to form bulk cerium acetate [9]. Mechanisms for
ketonization of carboxylic acids and the acid enolization pathway
∗
Corresponding author. Tel.: +1 865 574 5040; fax: +1 865 241 5152.
E-mail address: overburysh@ornl.gov (S.H. Overbury).
Current address: Fritz Haber Institute of the Max Planck Society, 14195 Berlin,
1
Germany.
2
Current address: Applied Materials, Inc., Gloucester, MA 01930, USA.
http://dx.doi.org/10.1016/j.cattod.2015.03.033
0920-5861/© 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
2
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
leading to ketone have been of intense interest and are still a matter
of debate [10–12]. Clarification of acetate and its role as spectator,
intermediate or deactivator can be aided by model studies aimed
at identifying the temperature evolution of adsorbed acetic acid.
Identification of surface intermediates by IR methods during C2
reactions requires better identification of possible reaction inter-
mediates by their spectroscopic signature. First principles density
functional theory (DFT) calculations now make it possible to exam-
ine bonding configurations of surface intermediates in detail when
combined with surface science techniques. Comparisons of vibra-
tional spectra simulated using DFT calculations with experimental
FTIR spectra obtained for formate, carbonates and acetaldehyde
have clarified the interpretation of the experimental spectra, which
enables more reliable detection of the surface species [7,13,14].
To obtain a better understanding of the interactions between
acetic acid and cerium oxide catalysts, we have now combined UHV
based measurements on a model CeO2(1 1 1) surface with DFT com-
putations. Surface science measurements, performed as a function
of temperature on a surface with known structure, provide infor-
mation about the evolution of the acetic acid on the surface and
provide signatures of bonding configurations. DFT is used to pre-
dict models for the surface bonding configurations and to simulate
vibrational spectra that can be compared directly with experimen-
tal RAIRS spectra. From this we have identified the stages of acetic
acid deprotonation and decomposition. Measurements are per-
formed upon both oxidized and reduced surfaces to identify the
important role of oxygen vacancies upon the reaction products.
From these studies, we find that bridge bonded acetate coexists
with another type of acetate that is bonded by its interaction with
oxygen vacancies.
(m/z = 45 and 43), CO (m/z = 28) and CO2 (m/z = 44). In the TPD spec-
tra shown and in their peak integration, interferences from overlap-
ping co-products were subtracted. Various other possible products
were monitored but were not definitively detected, including
methane, formaldehyde, dimethyl ether and diethyl ether, vinyl
acetate, methyl acetate, ethyl acetate, and various alcohols.
RAIRS was performed as described previously [7]. Spectra were
collected using grazing angle (∼86◦) p-polarized light from the IR
beam of a Mattson Infinity infrared spectrometer, and the reflected
IR beam was detected with a liquid-nitrogen cooled MCT detec-
tor. All spectra shown are differences between the IR absorption
spectrum at a certain condition and a reference spectrum recorded
for the clean ceria surface. All spectra were collected at an instru-
mental resolution of 4 cm−1 and the signal averaged for 1024
scans.
Soft X-ray photoelectron spectroscopy (sXPS) was conducted
using synchrotron radiation on beamline U12a at the National Syn-
chrotron Light Source (NSLS). The XPS data were collected using
a VSW 125 hemispherical analyzer with the sample oriented such
that the photon incident angle was 35◦ and electron emission angle
was 30◦ relative to the surface normal. C 1s and O 1s spectra were
recorded using photon energies of 400 and 600 eV, respectively.
The instrumental resolution was ca. 0.5 eV. The Ce 4d photoemi-
ssion was used for binding energy calibration using the 122.3 eV
satellite feature. The Ce oxidation state was determined using XPS
of the Ce 4d regions at Beamline U12a.
NEXAFS was performed at the C k-edge. The energy resolution
was less than 0.5 eV and the photon energy was calibrated using
the dip in the photon flux at 284.7 eV [18]. The X-ray absorption
was recorded using a partial yield electron detector. The high-pass
retarding grid was set at −230 V. Higher order X-ray excitation cre-
ated apparent absorption features from the ceria substrate due to
the O k-edge and the Ce MIV and MV edges. The absorption due to
only higher order radiation was determined by recording spectra
with a retarding grid voltage of −307 V, i.e. greater than the first-
order photon energy. The background resulting from the higher
order excitation was subtracted from the NEXAFS spectra.
2. Experimental and computational methods
Experiments were performed in three separate ultra-high
vacuum chambers described previously, one used for reflection
absorption infra-red spectroscopy (RAIRS) [7], one based at the
National Synchrotron Light Source (NSLS) for soft X-ray photo-
electron spectroscopy (sXPS) and near edge X-ray absorption fine
structure (NEXAFS), and a third for performing temperature pro-
grammed desorption (TPD). In each of the UHV systems the model
CeO2 catalyst were prepared as thin films deposited on Ru(0 0 0 1)
single crystals, leading to a flat CeO2 surface with (1 1 1) orienta-
tion [15]. Briefly, cerium metal is evaporated from a metal doser
onto the Ru crystal held at 700 K while under a controlled pressure
of oxygen gas, typically around 10−7 torr. By varying the oxygen
pressure, a range of catalyst with different ratios of Ce3+/Ce4+
could be obtained, as quantified from the XPS spectra (Supplemen-
tal information, Fig. S1). Reduced surfaces could also be obtained
by exposure to methanol at ∼700 K [16]. Glacial acetic acid was
degassed by several freeze–pump–thaw cycles and dosed onto the
surface through either a directed, effusive gas doser or a leak valve.
Temperature programmed desorption (TPD) experiments were
performed using a temperature ramp rate of 2 K/s while biasing the
sample at −70 V to prevent electrons generated by the mass spec-
trometer ionizer from stimulating reactions at the surface. For the
TPD spectra, the prepared surfaces were exposed to acetic acid from
a directed doser at 175–185 K, chosen to be slightly higher than the
multilayer desorption temperature. Following dosing, the sample
is positioned in front of the quadrupole mass spectrometer (Hiden
HAL/3F 301) and the QMS signal at multiple masses was monitored
during the temperature ramp. The intensities of the observed prod-
ucts were scaled according to the method described by Ko et al.
[17]. The masses used and the product that they most represent
are H2 (m/z = 2), water (m/z = 18), acetylene (m/z = 26), acetaldehyde
(m/z = 29), ketene (m/z = 42), acetone (m/z = 43 and 58), acetic acid
Periodic, spin-polarized DFT calculations were performed
within the generalized gradient approximation (GGA-PW91) [19]
using the Vienna Ab initio Simulation Package (VASP) [20–22]. DFT
at the GGA level fails to properly describe the localization of the Ce
4f electrons that occurs with the reduction of Ce4+ to Ce3+ due to
self-interaction errors [23]. To compensate for this deficiency, the
DFT+U formalism of Dudarev et al. [24] was used. It is an efficient,
ad hoc method that directly modifies the on-site Coulomb interac-
tion, which has been shown to satisfactorily reproduce localized f
states in bulk Ce2O3 and in oxygen vacancies in CeO2, although no
single value of U can simultaneously improve the predictions for
all properties of bulk CeO2 and Ce2O3 [25,26]. A U-value of 2 was
used in this study based on our previous results in modeling the
adsorption of formate and acetaldehyde on CeO2(1 1 1) [7,14].
The core electrons were described by the projector-augmented
wave method (PAW) [27], and the Kohn–Sham valence states
(including Ce(4f5d6s), O(2s2p), C(2s2p), H(1s)) were expanded in
a plane wave basis up to a kinetic energy of 400 eV. The CeO2(1 1 1)
surface was represented by a slab consisting of three O Ce O tri-
layers with a (2 × 2) surface unit cells, which corresponds to ¼
monolayer (ML) of coverage for each adsorbate per unit cell.
The periodic slabs were separated in the z direction by ca.
˚
12 A of vacuum. The top trilayer and all adsorbates were fully
relaxed and the remaining two trilayers were held fixed at the
bulk positions. Adsorption was allowed on one side of the slab
only, with dipole decoupling [28] in the z direction. The surface
Brillouin zone was sampled with a 2 × 2 × 1 Monkhorst–Pack k-
point grid [29]. Geometry optimization was converged to below
0.01 eV/Å in each degree of freedom for all relaxed atoms. Different
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
3
CO2
210 K
310 K
520 K
CO2
600 K
210 K
550 K
CO
CO
590 K
H2O
H2O
H2
C2H2 x5
H2
C2H2
OC=CH2
x5
x5
OC=CH2
CH3CHO
CH3CHO x5
CH3COOH x5
(CH3)2CO x5
CH COOH
x5
x5
3
(CH3)2CO
200
300
400
500
600
700
800
900
200
300
400
500
600
700
800
900
T / K
T / K
Fig. 1. TPD spectra obtained following exposure to acetic acid at 175 K on fully
oxidized CeO2. Spectra are offset for clarity.
Fig. 2. TPD spectrum following exposure to acetic acid at 175 K on a surface that
is 60% reduced, i.e. corresponding to a surface stoichiometry of CeO1.7. Spectra are
offset for clarity.
spin states were calculated and the lowest adsorption energy was
reported for each adsorbate. The adsorption energy was calculated
as ꢀE = Etotal − Eslab − Emol, where Etotal, Eslab, and Emol are the ener-
gies of the surface with an adsorbate, the clean surface without
adsorbate, and the adsorbate molecule isolated in the gas phase in
a neutral state (closed-shell or radical).
310 K. Ketene, the primary C2 product, desorbs at 590 K along with
H2O. These correlated products imply cleavage of a C O bond and
of a methyl C H bond, although it cannot be determined from these
data which is rate limiting. A trace amount of acetone is detected at
nearly the same temperature as the ketene. Formation of acetone
implies a coupling reaction such as ketonization between nearby
acetates. Acetone is the only coupling product observed. At higher
temperatures (above 600 K) CO and CO2 desorb in broad states
implying C C bond breaking occurs. Small amounts of acetylene
were detected at 700 K, but methane is not observed at any point
of the TPD experiments. To speciate the O in the water desorption, a
fully oxidized film of Ce18O2 was grown and used as a substrate for
TPD of CH3C16O2H. As shown in Fig. S3 (Supplemental Information)
the majority of H2O desorbing at 210 and 310 K extracts O from the
lattice, while the water desorbing near 590 K derives in equal parts
from lattice oxygen and from acetic acid.
The vibrational frequencies and simulated IR spectra of surface
adsorbates were calculated using the Atomic Simulation Environ-
ment [30]. The vibrational modes and frequencies were calculated
from a finite difference approximation of the dynamical matrix, and
the IR intensities were calculated from a finite difference approxi-
mation of the gradient of the dipole moment in the z direction [31].
˚
The magnitude of the displacement was 0.01 A in each degree of
freedom. The assignments of the calculated modes were based on
visual inspection. Additional details of the setup of our DFT calcu-
lations can be found in Refs. [7,14].
3. Results
Several TPD experiments were run on a ceria surface that was
reduced, either by sequential dosing and flashing of acetic acid
or by treatment with methanol at 700–900 K (for example, Fig.
S2, see SI). Acetic acid TPD performed from a surface that is 60%
3.1. Desorption products
TPD experiments were performed on ceria samples with differ-
ent levels of reduction. Fig. 1 shows the TPD obtained from the fully
oxidized surface following adsorption at 175 K. Acetic acid (mass
45) desorbs in two states near 210 K and 520 K. Acetic acid fragmen-
tation pattern presents two intense fragments at 43 and 45 amu
and a less intense parent peak at 60 amu. All three masses were
followed to discriminate against acetone which would contribute
at mass 43. The lower temperature state corresponds to desorp-
tion of undissociated, chemisorbed acetic acid molecules and the
high temperature peak presumably corresponds to recombination
of acetate with its proton bonded on nearby O anions (Ce OH).
Water also desorbs in multiple peaks, including a sharp peak at
reduced, i.e. corresponding to a surface stoichiometry of CeO1.7,
is shown in Fig. 2. The product distribution is substantially differ-
ent than for the fully oxidized surface. Desorption of undissociated
acetic acid is still visible below 300 K, but there is no desorption of
H2O at or below 300 K as seen from the oxidized surface. Recom-
binative desorption of acetic acid, seen at 510 K for the oxidized
surface, does not occur from this highly reduced surface at any tem-
perature. There is also less ketene and acetone formed compared
to the oxidized surface, although there is increased acetaldehyde
that desorbs at 550 K just below the H2 desorption temperature.
Desorption of CO and CO2 at 600–700 K is decreased compared to
the oxidized surface and little or no CO2 is observed up to 900 K.
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
4
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
dehydrogenation
dehydration uni and bimolecular
decarbonylation
bimolecular only
acetone
acetaldehyde
acetic acid
ketene
H2
water
0.0
30
40
50
60
70
80
90
100
Ce+4(%)
Fig. 3. Integrated TPD products are shown as a function of the extent of oxidation of the surface. The intensities shown for acetone, acetaldehyde and acetic acid are scaled
by a factor of 5.
Desorption of H2 increases substantially compared to the oxidized
surface, peaking strongly at 600 K. It has been previously reported
[32] that hydroxyls (from water) are stabilized on a reduced sur-
faces of ceria, leading to desorption of H2 (rather than H2O) above
500 K as seen prominently in this case. The amount of acetylene
detected at ∼700 K is increased somewhat compared to the fully
oxidized CeO2. A portion of the remaining C containing species are
consumed by reacting with lattice oxygen at high temperatures
(above 800 K) liberating mainly CO and not CO2, probably because
of the decrease of available O on the reduced surface.
The variation in product distribution as a function of the extent
of reduction of the initial CeO2 surface is shown in Fig. 3. The
oxidation state of the Ce is determined from the Ce 3d XPS (see Sup-
plemental Information). To avoid inclusion of physisorbed acetic
acid, the acetic acid product is integrated over only the high tem-
perature desorption peak.
as the sample is annealed to 500 K and above, but a small new fea-
ture appears at 285 eV that can be assigned to a reduced form of
carbon. By 600 K the carboxylate and methyl peaks are essentially
gone and only small amounts of the reduced carbon remain. All of
the C is gone after annealing to 900 K.
From the peak fitting of the two components in the methyl (at
286.4 and 285.8 eV) and carboxylate (at 290.0 and 289.1 eV), the
amounts of each type of carbon can speciated and are summarized
in Fig. 5 as a function of the annealing temperature. The transfer of
intensity from the lower binding energy components to the higher
energy components occurs over the temperature range 200–400 K,
the range where water desorbs from the surface (Fig. 1). Desorption
of water creates O-vacancies on the surface possibly allowing the
acetate to shift to a different adsorption site or to change configu-
ration.
The corresponding O 1s spectra are shown in Fig. 4 (upper
panel). The peak at 530.4 eV, due to ceria lattice O, is attenuated
by the adsorbate at 200 K, and an adsorbate related peak appears at
532.2 eV with a tail on the higher binding energy side. Upon anneal-
ing to 300 K the tail disappears while there is a 0.4 eV shift in the
adsorbate peak to 532.6 eV. This peak further shifts and sharpens
at 400 K. The loss of intensity on the low binding energy side likely
reflects the loss of surface OH through water desorption that occurs
in this temperature range (see Fig. 1) while the loss of the high bind-
ing energy tail is due to desorption of the molecular (undissociated)
acetic acid. The temperature-dependent loss of all adsorbate oxy-
gen (OH and other O-containing molecules) is shown in Fig. 5. As
the sample is annealed above 500 K the intensity drops dramat-
ically as O-containing products desorb from the surface or the O
incorporates into lattice. The O 1s intensity approximately tracks
the total C 1s intensity above 300 K. However, at 600 K the adsor-
bate O 1s peak is gone whereas there is still some residual, reduced
carbon present.
3.2. Adsorption states
3.2.1. sXPS
sXPS spectra were recorded after the fully oxidized CeO2(1 1 1)
was exposed to 10 L (1 L = 10−6 torr s) of acetic acid at 100 K and then
annealed to sequentially higher temperatures, shown in Fig. 4. At
the lowest annealing temperatures, the C 1s spectrum (Fig. 4 lower
panel) shows two peaks near 290.2 and 286.1 eV. A shoulder is evi-
dent at 289 eV and upon annealing to 200 K this shoulder becomes
the dominant feature with a loss of the peak at 290.2 eV. It may
be assumed that the transition upon annealing to 200 K is associ-
ated with loss of multi-layer, physisorbed acetic acid, and possible
deprotonation of the remaining chemisorbed acetic acid. Most of
the physisorbed acetic acid desorbs leaving primarily two features
peaked at 289.1 and 285.7 eV that are assigned to the carboxyl and
the methyl carbons. However, peak fitting suggests that shoulders
are evident on the high binding energy side of both peaks. As the
sample is annealed to 400 K, both features appear to sharpen and
shift to higher binding energy. The intensity of both peaks decreases
Fig. 6 shows sXPS spectra following adsorption of acetic acid on
partially reduced CeO2(1 1 1). In this case the sample was approx-
imately 60% reduced (CeO1.7) and was exposed to a 10 L dose
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
5
A) O 1s
Lattice O
200 K
8000
Adsorbate O
6000
4000
2000
0
300 K
400 K
500 K
Fig. 5. Speciation of surface carbon and total adsorbate O by sXPS following heat-
ing adsorbed acetic acid adsorbed on a fully oxidized CeO2(1 1 1) surface. Values
obtained from integration of C 1s and O 1s features in Fig. 4.
550 K
600 K
respectively, in acetate. As the sample is annealed the nominal peak
positions shift to higher binding energy and the peaks become nar-
rower and more symmetric. At 500 K new features are evident on
both higher and lower side of the methyl peak. Upon annealing the
sample to 550 K the intensity of the carboxylate feature decreases
dramatically, the peak positions of the methyl and carboxylate shift
to lower binding energy, and the new features at 288.6 eV and
286.3 eV grow. The peak near 288.6 is poorly resolved and may
consist of a dioxy species or an enolate. The peak at 286.3 eV is
assigned to a CHX species possibly bound to Ce [33,34]. As the sam-
ple is annealed to 900 K all of the C 1s peaks disappear except an
intense peak at 286 eV. This peak could only be removed by oxygen
treatment at elevated temperature.
540
535
530
525
Binding Energy (eV)
2000
1500
1000
500
0
B) C 1s
-CCH3
150 K
In the O 1s spectra (Fig. 6 upper panel), adsorption and anneal-
ing to 200 K leads to a broad adsorbate peak centered at 533.3 eV.
As on the oxidized surface, there are shifts and sharpening of the
adsorbate peak as the sample is annealed to higher temperatures,
but the overall O 1s intensity (acetate and hydroxyl) remains con-
stant from 200 K to 500 K. After annealing to 550 K the adsorbate
intensity drops and the peak shifts to 533 eV. These changes in the
adsorbate O 1s feature correspond to loss of the carboxylate and
growth of the alkoxy in the C 1s spectrum (Fig. 6A). Only the lattice
O feature remains at 900 K indicating that the C that remains on the
surface is not bonded to O.
-COO
200 K
300 K
400 K
500 K
550 K
3.2.2. RAIRS
RAIRS was performed to get additional information about the
nature of the bonding and the bonding configurations of the
acetates. In Fig. 7 (lower panel), the RAIRS spectra in the region
of the C O and OCO stretches are shown following a sequence
of exposures at 175 K. The surface was held at constant temper-
ature, so the coverage increases to saturation (Fig. 7) at the highest
dose. Prior to exposure, the surface was fully oxidized by heating
to 700 K in 10−7 torr O2. A feature at 1430 cm−1 is most promi-
nent at low coverage, but features at 1530 and 1695 cm−1 become
more obvious with increasing coverage and a feature at 1450 cm−1
dominates at the highest coverage. Following the exposure the sat-
urated surface was subsequently annealed to sequentially higher
temperatures following the adsorption as shown in Fig. 7 (upper
panel). After annealing to 226 K, the spectrum is dominated by a
strong feature peaking at 1450 cm−1, although it appears to have
shoulders and it is accompanied by a peak at 1695 cm−1. Upon
annealing to 326 K the peak at 1695 cm−1 disappears, the strong
600 K
900 K
293 292 291 290 289 288 287 286 285 284 283
Binding Energy (eV)
Fig. 4. sXPS of the O 1s region (A) and the C 1s region (B) obtained following
adsorption of acetic acid on oxidized ceria at 200 K and subsequent annealing to
sequentially higher temperatures.
of acetic acid at 100 K and subsequently heated to sequentially
higher temperatures. Upon heating to 200 K there are two promi-
nent peaks in the C 1s spectra (Fig. 6 lower panel) at 287.4 eV and
291.0 eV, with shoulders clearly evident on the high binding energy
side of both peaks. As on the fully oxidized surface the low and
high binding energy peaks are assigned to the methyl and carboxyl,
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
6
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
Acetic acid on CeO2(111)
1450
A) O 1s
Lattice O
1430
8000
6000
4000
2000
0
Adsorbate O
1390
1330
200 K
1695
226 K
300 K
326 K
527 K
400 K
500 K
1000
1100
1200
1300
1400
1500
1600
1700
1800
Wavenumber (cm-1)
550 K
Acetic acid adsorption on CeO2
1430
1450
700 K
900 K
1695
increasing
dose
540
535
530
525
1330
Binding Energy (eV)
1600
1400
1200
1000
800
600
400
200
0
B) C 1s
-COO
-CCH3
1000
1100
1200
1300
1400
1500
-1
1600
1700
1800
200 K
Wavenumber / cm
300 K
400 K
500 K
Fig. 7. Lower: evolution of RAIRS spectra with coverage obtained by exposing a
fully oxidized CeO2(1 1 1) surface to acetic acid at 175 K with sequential doses to
saturation. Upper: evolution of RAIRS spectra from the saturated surface during
subsequent annealing to temperatures indicated in the plot. Spectra are difference
spectra (relative to clean surface), background subtracted and offset vertically.
The nature of the four features remaining at 527 K can be fur-
ther elucidated by observing acetic acid on the reduced surface as
shown in Fig. 8. This surface was 60% reduced (to CeO1.7). Sequential
exposures to acetic acid at 175 K were used to increase the cover-
age while monitoring RAIRS. At the initial adsorption two peaks are
obvious at 1330 and 1396 cm−1, while a weaker shoulder is appar-
ent at about 1430 cm−1 (Fig. 8 lower panel). These three features
align nearly perfectly with features observed after annealing the
acetic acid adsorbed on the fully oxidized surface to 527 K (Fig. 7).
As exposure increases, these peaks grow in intensity but remain
at roughly the same position (or red shift less than 3 cm−1). At the
highest coverage the peak at 1423 cm−1 grows to dominate the
spectrum, and it is accompanied by a peak at 1705 cm−1. Growth
of these two peaks at highest coverage is approximately corre-
lated, and may be associated with an additional adsorbate state
that differs from that observed at the lowest coverage.
-CO-Ce
550 K
700 K
C-Ce
900 K
293 292 291 290 289 288 287 286 285 284 283
Binding Energy (eV)
Fig. 6. sXPS of the O 1s region (A) and the C 1s region (B) obtained following adsorp-
tion of acetic acid on reduced ceria at 200 K and subsequent annealing to sequentially
higher temperatures.
3.2.3. NEXAFS
peak at 1450 cm−1 decreases and its shoulders resolve into two
peaks at higher and lower wavenumbers. In addition, two sharp
features at 1330 and 1390 cm−1 appear. Further annealing to 527 K
removes all traces of the 1450 cm−1 peak, and four prominent peaks
remain in the RAIRS spectrum. A further anneal to 627 K removes
all traces of these peaks (not shown).
C k-edge NEXAFS spectra were recorded after oxidized
CeO2(1 1 1) and reduced CeO1.7(1 1 1) were exposed to 10 L of acetic
acid at 90 K and then annealed to higher temperature. The spectra
from fully oxidized CeO2(1 1 1) are shown in Fig. 9A and the spectra
from reduced CeO1.7(1 1 1) are shown in Fig. 9B. The upper spec-
trum in Fig. 9A was recorded immediately after adsorption at 90 K
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
7
30
1423
1465
Acetic acid on CeO1.7
1390
1330
C k-edge
A)
π* CO
1705
25
20
15
10
5
175 K
244 K
4p CO
σ* C-C
4p CH3
3s CH3
90 K
σ* C-O
300 K
444 K
(∗)
1000
1100
1200
1300
1400
1500
1600
1700
1800
550 K
305
Wavenumber (cm-1)
0
285
290
295
300
Acetic acid adsorption on CeO1.7
Photon Energy
1423
1390
1330
14
12
10
8
B)
π* CO
1705
increasing
dose
300 K
550 K
700 K
π
* C=C (CH2)
6
1000
1100
1200
1300
1400
1500
1600
1700
1800
4
π* C=C (CO)
Wavenumber (cm-1)
2
Fig. 8. Lower: evolution of RAIRS spectra with coverage obtained by exposing a
partially reduced CeO1.7(1 1 1) surface to acetic acid at 175 K with sequential doses
to saturation. Upper: evolution of RAIRS spectra from the saturated surface during
subsequent annealing to temperatures indicated in the plot. Spectra are difference
spectra (relative to clean surface), background subtracted and offset vertically.
0
285
290
295
300
305
Photon Energy
Fig. 9. NEXAFS at the C 1s edge, A) on fully oxidized, and B) on highly reduced
ceria. In B, the ␲*(C C) resonance at 284.6 eV is associated with a C at the alkyl end
and is from a multilayer of acetic acid. A number of adsorption reso-
nances are evident and have been labeled in accordance with the
absorption spectrum reported for gas phase acetic acid by Robin
et al. [35]. When the sample was annealed to 300 K the multilayer
acetic acid desorbs and many of the minor features are no longer
evident but the ␲* resonance from the carboxylate is still dominant
at 288.2 eV. The peak position has shifted to a slightly lower photon
energy compared to the multilayer, consistent with the lower C 1s
sXPS binding energy of acetate relative to acetic acid.
After annealing at 550 K the ␲* resonance from the carboxyl-
ate is weaker due to the loss of carboxylate as observed in the
C 1s sXPS spectra (Fig. 4). This temperature corresponds to the
high temperature desorption of acetic acid and the onset of ketene
desorption (Fig. 1). A new feature emerges in the NEXAFS near
286.5 eV, labeled (*) in Fig. 9A. Peaks near this position are gener-
ally associated with a ␲* resonance from a carbonyl [18]. However,
carbonyl is not indicated at 550 K in the C 1s sXPS spectrum (Fig. 4).
Previous studies with acetaldehyde, formaldehyde and acetone on
oxidized CeO2(1 1 1) placed the carbonyl binding energy between
288.0 eV and 288.5 eV [33,36,37]. Alternatively, the NEXAFS peak
may be associated with a ␲* C C resonance such as found in eno-
late or a ketene-like intermediate that is more clearly evident on
the reduced surface (see below).
(
(
C
CH2) and the ␲*(C C) resonance at 286.4 eV is assigned to the oxygenated C
OOC CH2 or OCH CH2).
the fully oxidized surface (Fig. 9A). After annealing to 550 K this res-
onance decreases and new peaks appear at 286.4 eV and 284.6 eV,
but the total C absorption intensity, indicated by the magnitude
of the “jump” between 283 eV and 305 eV, is roughly unchanged.
The NEXAFS indicates that the acetate has converted to a new
species. Two likely candidates are (CH2 COOad) and (CH2 CHOad).
The former, (CH2 COOad), has been described previously as a “sur-
face ketene” formed from enolization of acetate [38], a surface
acetic acid dianion intermediate [11], or a dioxyketene [39]. The
latter, (CH2 CHOad), corresponds to enolate as has been previously
observed from adsorption of acetaldehyde [7,40] and ethylene gly-
col [41] on reduced CeOx(1 1 1). The two new peaks in the NEXAFS
are assigned to ␲* (C C) resonances, in either the surface ketene
or the enolate, with the peak at 284.6 eV associated with a C at the
alkyl end (
C CH2) and the peak at 286.4 eV assigned to the oxy-
genated C ( OOC CH2 or OCH CH2) [42]. Similar NEXAFS spectra
have been observed from enolate formed by acetaldehyde adsorp-
tion on CeO2(1 1 1) [40], and on CeO2(1 0 0) [39]. On the CeO2(100)
both enolate and dioxyketene (surface ketene) formation was pro-
posed based upon the sXPS.
When the acetic acid is adsorbed on reduced CeO1.7(1 1 1) and
annealed to 300 K the NEXAFS spectrum (Fig. 9B) is dominated by
the ␲* CO resonance associated with carboxylate, as for the case of
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
8
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
Fig. 10. DFT-calculated minimum-energy structures for various states of acetate on CeO2(1 1 1): (A) ␮-acetate; (B) ␮-acetate co-adsorbed with H; (C) acetate/Vo; (D) acetate/Vo
co-adsorbed with H; (E) ␮-acetate co-adsorbed with acetate/Vo; (F) acetate/VVo co-adsorbed with H; (G) CH2 COO/Vo; (H) CH2 COO/VVo). In each panel, side view is on left
and top view is on right. Large brown; medium green, red, and black; and small white spheres represent lattice O, lattice Ce, O, C, and H atoms, respectively. Top-layer lattice
O atoms are colored more lightly to be distinguished from lower-layer lattice O atoms. Medium orange spheres in panels (B), (D), and (F) represent coadsorbed H atoms that
form surface hydroxyl groups. Bonds between O in acetates and lattice Ce are for illustrative purposes only. Molecular images are made using VESTA [60].
After the sample is annealed to 700 K the absorption peak at
285 eV increases while the other peaks decrease. At this temper-
temperature and extent of reduction of the ceria, as it affects oxygen
availability, the acetic acid can desorb intact or decompose all the
way to H2, CO, and carbonaceous surface deposits. More interesting
are the pathways and surface intermediates that lead to other C1,
C2 and C3 desorption products.
ature, the C 1s spectrum (Fig. 6A) indicates that there are no C
O
bonds remaining. This implies that the NEXAFS peaks have changed
and are no longer due to carboxylate or enolate as assigned at 550 K.
Instead the peaks at 285 eV, 286.7 eV and 288.2 eV are consistent
with a carbonaceous deposit containing unsaturated C C bonds.
Similar spectra have been observed for C60 fullerenes [43] and
amorphous C films [44]. The spectrum has some similarities with
NEXAFS spectra from metal carbides [45] however metal carbides
generally have C 1s XPS binding energies near 282–283 eV [46].
Therefore cerium carbide is ruled out based on the C 1s spectrum
in Fig. 6A.
Adsorption at 175 K on a fully oxidized CeO2 surface leads to
accumulation of molecularly adsorbed, undissociated acetic acid.
This conclusion is clear from the RAIRS spectrum (Fig. 7) which
shows characteristic peaks that can be assigned as ꢁ(C O) at
1695 cm−1, ı(CH3) at 1430 cm−1, ꢁ(C O) at 1190 cm−1, and CH3
rock at 1020 cm−1. These positions approximately match with gas
phase (or liquid phase) acetic acid [47–49] although the presence
of OH bending mode is not obvious and this mode may be most
affected by adsorption and interactions with the surface and neigh-
boring molecules. Annealing the oxidized surface to 226 K removes
most indications of the acetic acid, consistent with its desorption
seen in TPD at 200 K (Fig. 1). Not all adsorbate is acetic acid at 175 K
and the sXPS implies that deprotonated acetate is present as early
as 150 K. Coexistence of acetic acid and acetate explains the peaks
in the C 1s sXPS spectrum and their changes between 150 and 200 K
(Fig. 4).
After desorption of weakly bonded molecular acetic acid, it is
generally assumed that dissociated acetic acid remains on the sur-
face in the form of acetate [50,51], but the RAIRS data suggest that
more than a single state remains. Following anneal to 326 K, the
RAIRS features sharpen up and two new modes appear at 1330 and
1390 cm−1, separated by 60 cm−1. These two lower wavenumber
modes grow in intensity even as the modes at higher wavenumber
(1430 and 1465 cm−1) persist, suggesting that at least two types
3.2.4. DFT
A number of different acetate states on both the fully oxidized
and the partially reduced CeO2(1 1 1) surfaces were examined. The
latter was represented by a point oxygen vacancy (denoted Vo) or
a dimer of oxygen vacancies (denoted by VVo) in the surface oxy-
gen layer of the CeO2(1 1 1) slab, as we have done previously [7,14].
Several of these states possess IR-active modes that are consistent
with the various observed RAIRS spectra. The energy-minimized
structures of these states are shown in Fig. 10, including bridging
(␮) acetate (located atop a pair of adjacent Ce4+ sites; Fig. 10A);
␮-acetate co-adsorbed with H (Fig. 10B); acetate adsorbed in an
oxygen vacancy (acetate/Vo; Fig. 10C); acetate/Vo co-adsorbed with
H (H bonded to a lattice O; Fig. 10D); acetate/Vo co-adsorbed with
␮-acetate (Fig. 10E); acetate/VVo co-adsorbed with H (Fig. 10F);
CH2 COO adsorbed in an oxygen vacancy (CH2 COO/Vo; Fig. 10G)
and in a vacancy dimer (CH2 COO/VVo; Fig. 10H) respectively. The
adsorption energies of these states are reported in Table 1. The
simulated IR spectra for these species are shown in Fig. 11A and
B.
Table 1
DFT-calculated adsorption energies (in eV) of the various acetate states included for
comparison with experimental IR spectra.
State
ꢀE (eV)
4. Discussion
␮-Acetatea
−0.77
−0.95
−4.18
−2.50
−3.28
−2.71
−1.09
−1.34
␮-Acetate + Hb
a
The combination of desorption measurements and surface spec-
troscopies allows us to put together a detailed description of
the transformations that acetic acid undergoes on a CeO2 surface
as shown in Scheme 1. This scheme summarizes the desorption
products that are observed and shows in an approximate way
the adsorbate species and their evolution with temperature. It is
clear that lattice oxygen participates in the temperature-dependent
transformations and that oxygen vacancies play an important role
both in stabilizing and in altering surface adsorbates. Depending on
Acetate/Vo
Acetate/Vo + Hb
Acetate/Vo + ␮-acetatea,*
Acetate/VVo + Hb
c
CH2 COO/Vo
CH2 COO/VVo
c
a
With respect to gas-phase acetate.
With respect to gas-phase acetic acid.
With respect to gas-phase ketene.
Averaged over two acetate groups.
b
c
*
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
9
conclude that these modes are associated with acetate adsorption
at a vacancy. However, there is another sharp mode at 1465 cm−1
that remains after annealing to 527 K (Fig. 7). This feature is not
observed during the initial stages of adsorption on the reduced sur-
face, and so it must be distinguished from those associated with
the adsorbate trapped at the vacancies. Although the sXPS indi-
cates two types of C present at 500 K, consistent with carboxyl and
methyl of acetate, sXPS alone cannot differentiate two structurally
distinct acetate-like adsorbates. However, the NEXAFS at the C 1s
edge indicates a new resonance at 286.5 eV that grows between
300 and 500 K. It is associated with another type of ␲ hybridized
bonding that is distinct from the ␲* CO of acetate.
A
exp. oxidized 527 K
1
acetate/Vo + μ-acetate
acetate/Vo
The challenge is to identify more definitively the two types of
adsorbates, and the modes observed in the RAIRS provide insights.
Previous compilations of IR vibrational data for metal carboxylates,
including acetate complexes, have described them by character-
istic symmetric and asymmetric OCO modes and their splitting
ꢀ = ꢁa − ꢁs [52,53]. Although there are difficulties in postulat-
ing structure from the positions of the modes, values of ꢀ that
are significantly less than ionic values are indicative of chelating
or bridging bidentate bonding [52]. Sodium acetate in solution
exhibits values of ꢀ around 140 cm−1 between the ꢁa mode at
1560 cm−1 and the ꢁs manifold that peaks at 1420 cm−1 [54,55].
The splitting between the strongest observed modes is 60 cm−1 on
the reduced surface (at 1330 and 1390 cm−1) either before anneal-
ing at low coverage or after annealing to 444 K (Fig. 8). The two
peaks are also dominant features on the oxidized surface (Fig. 7).
This splitting is considerably smaller than the ionic values for ꢀ.
Based upon the compilation of Deacon [52], the asymmetric mode,
ꢁa(OCO), is expected to be found at wavenumbers ranging from
about 1525 to 1575 cm−1 for chelating acetate and roughly 1530 to
1630 cm−1 for bridging acetates, a range that includes the solution
value of 1560 cm−1. Unidentate acetate groups are typically higher
still. Similarly, solid Ce acetate itself shows a peak at 1564 cm−1 [9]
and acetic acid adsorbed on polycrystalline CeO2 leads to a peak
1540 assigned to ꢁa(OCO) consistent with other tabulated acetates
[56]. All of these values are higher than the most prominent modes
seen on the oxidized surface at 1450–1465 cm−1 (Fig. 7) or on the
pre-reduced surface at 1425 cm−1 (Fig. 8), suggesting that these
adsorbates are more strongly interacting with the CeO2 than typical
acetate.
μ-acetate
μ-acetate + H
0
900
1000 1100 1200 1300 1400 1500 1600 1700
Wavenumber (cm-1)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
B
exp. reduced 444 K
acetate/VVo + H
acetate/Vo + H
μ-acetate +H
900
1000
1100
1200
1300
1400
1500
1600
1700
Wavenumber (cm-1)
Based on the evidence from RAIRS we postulate that there
are at least two types of acetate on the oxidized surface in the
temperature from around 300 K up to the onset of ketene and
acetone desorption above 500 K. The apparently low wavenum-
Fig. 11. Comparison of experimental and computed spectra for structures proposed
for acetate on the oxidized (A) and reduced (B) surfaces of CeO2. Computed spectra
are stable structures found by DFT and described in Fig. 10 (see text: “+” indicates co-
adsorption of two species). Experimental spectrum in A is for acetic acid adsorbed
on fully oxidized CeO2 followed by annealing to 527 K; experimental spectrum in
(B) is for acetic acid adsorbed on CeO1.7 followed by annealing to 444 K. Spectra are
offset vertically and the experimental spectra are arbitrarily scaled.
bers of the O
C O modes suggest a strongly bonded ␮-acetate
remaining on oxidized parts of the surface, where it is preferen-
tially located above a pair of Ce4+ sites (Fig. 10A). H2O removal
below 300 K suggests acetate adsorbed in an oxygen vacancy
(acetate/Vo) (Fig. 10C). Another possibility is co-adsorbed ␮-acetate
and acetate/Vo (Fig. 10E) because the surface coverage of acetate
is likely to be high before decomposition begins substantially.
The simulated IR spectra for these three configurations are pre-
sented in Fig. 11A. The computed IR-active modes in the range of
1260–1390 cm−1 correspond to ꢁs(OCO) and ꢁa(OCO) with consid-
erable variation in position, while the modes from near 1430 cm−1
correspond to CH3 scissor. Qualitatively these features are fully con-
sistent with the three broad features observed at 1330, 1390, and
1430 cm−1 at 527 K (Fig. 7). Note that the computed spectrum for
acetate/Vo exhibits two strong peaks split by 49 cm−1 (Fig. 11A)
consistent with the observed ꢀ (60 cm−1) for the vacancy stabilized
acetate described above.
of adsorbates are coexisting. Annealing to 326 K must also lead
to desorption of water, as demonstrated by the TPD that features
water desorption at 310 K (Fig. 1). It is apparent that much of this
water derives from reaction of the dissociated acid proton with lat-
tice oxygen, a conclusion supported by the desorption experiments
performed on the Ce18O2 film that indicate that the desorbing water
contains lattice oxygen (Fig. S3). Therefore during the TPD, water
desorption leads to reductive oxygen vacancy formation on the sur-
face. Reduction of the surface by the loss of water explains the
shift in the sXPS positions to higher binding energies when the
oxidized surface is annealed from 300 to 400 K [40]. The resulting
vacancies become traps for surface acetates, which may migrate a
short distance to adsorb there in a more stable configuration. Fur-
ther evidence for this interpretation derives from the adsorption
on the pre-reduced surface, which exhibits the same two modes
(1330 and 1390 cm−1) already at the lowest exposures (Fig. 8). We
On the reduced surface, TPD indicates that there is no exit chan-
nel for hydrogen either as H2 or H2O below ∼500 K. Therefore the
surface is H- and defect-rich until the onset of H2 desorption above
500 K. Two acetate states co-adsorbed with H were considered
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
10
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
Scheme 1. Reaction pathways and surface species during temperature increase following acetic acid adsorption on a highly reduced (lower) and a fully oxidized (upper)
CeO2 surface.
using DFT, including acetate/Vo with H (Fig. 10D) and acetate/VVo
with H (Fig. 10F). For these structures, the computed modes corre-
spond to the methyl rock near 1010 cm−1, the ꢁs(OCO) and ꢁa(OCO)
fall in the range of 1310–1390 cm−1 while the CH3 scissor mode falls
in the range 1410–1440 cm−1. Qualitatively these features are also
consistent with the main features observed at 1050, 1330, 1396, and
1430 cm−1 at 444 K (Fig. 7). Two other partly dehydrogenated states
described as surface ketene intermediate (CH2 COO) adsorbed at
a single oxygen vacancy (Fig. 10G) or an oxygen vacancy dimer
(Fig. 10H) were also considered. The simulated IR spectra for these
structures do not contain any strong IR mode and are omitted from
Fig. 11.
Two factors should be noted. First, the DFT spectra suggest
that the peak positions are affected by neighboring acetate and
H, for example compare ␮-acetate and ␮-acetate + H in Fig. 11A.
There may be many other similar local arrangements involving
␮-acetate, acetate in vacancies, and H that are frozen at low tem-
peratures but evolve as temperature is raised, and these can lead to
shifting and broadening of the peaks. Second, the DFT-calculated
peaks are generally red-shifted compared to experiments, as has
been reported RAIRS studies of formic acid adsorption on ceria
[13,14]. Taking these two factors into consideration, the RAIRS
spectra and calculated IR spectra suggest that ␮-acetate and
acetate/Vo coexist on the oxidized surface and that acetate is
predominantly adsorbed in oxygen vacancies on the reduced sur-
face.
is increased above 500 K. They are unimolecular dehydration to
form ketene:
CH3COOH → CH2
C
O + H2O
(1)
and the bimolecular coupling reaction to form acetone:
2CH3COOH → (CH3)2C O + CO2 + H2O
(2)
Except for acetic acid, ketene and acetone are the main multiple
C containing products observed for the initially oxidized surface
and ketene is favored over acetone. These balanced reactions do
not describe the pathways that occur during the TPD, since the
low-temperature water evolution and the formation of acetate are
indications of stepwise processes of deprotonation to an intermedi-
ate. Acetone formation implies reaction between two deprotonated
acetates with methyl transfer from one to the other, thereby form-
ing COO that leads to subsequent release of CO2:
CH3COO + CH3COO/Vo → (CH3)2C O + COO
(3)
The CO2 TPD product correlates reasonably with the acetone as
expected from Eq. (3). Specifically, on the initially oxidized surface
the acetone slightly precedes the (larger) CO2 production (Fig. 1),
while on the highly reduced surface both acetone formation and
CO2 formation are eliminated (Fig. 2). Although acetone production
yields CO2 it is possible that CO2 may also result from unimolecular
decomposition of acetate. It is of mechanistic importance that for
the oxidized surface ketene and acetone are both evolved at nearly
the same temperature (600 K) with the same TPD peak signature,
although acetone is smaller. This coincidence suggests formation
of surface ketene may be rate-limiting the formation of acetone.
The observed reaction products that desorb above 500 K are the
result of the transformations occurring on the CeO2 surface. Two
major reaction pathways open up and compete as the temperature
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
11
Under UHV conditions the unimolecular formation of ketene may
be favored over acetone simply because the coupling reaction is
bimolecular and less favored as coverage decreases during TPD.
Although acetate reacts by these routes during TPD in UHV leav-
ing a clean surface (for the initially oxidized surface) or a surface
with carbon residue (for the initially reduced surface), the acetates
are stable up to almost 550 K. These results are therefore consistent
with the observation that bulk cerium acetate forms during steady
state reaction of acetic acid below 300 ◦C (573 K).[9]
Evidence for both of the net reactions (1) and (2) have been
reported previously on CeO2(1 1 1) and CeO2(1 0 0) surfaces, where
low temperature acetic acid adsorption led to the evolution of
ketene and a lesser amount of acetone during TPD [51]. However,
the effect of surface reduction upon the acetone:ketene selectivity
that is evident here was not previously reported. Similarly Barteau
and Kim [50] examined room temperature acetic acid adsorption on
TiO2 powders (anatase) and found acetone and ketene production
consistent with these reactions. However, they found that acetone
was favored over ketene, indicating altered selectivity compared
to CeO2 where acetone production is smaller. On TiO2 the CO2 pro-
duction lagged the acetone production slightly in TPD and was in
comparable yields to the acetone production.
also be due to enolate or carbonaceous residue. Enolate formed
from acetaldehyde has been observed on CeO2(1 1 1) by RAIRS and
is characterized by peaks near 1194 and 1600 cm−1 [7] and such
features are not seen clearly in the RAIRS data for either oxidized
or reduced ceria (Figs. 7 and 8). Enolate would be a natural precur-
sor to the small amount of acetaldehyde desorption that is observed
from the reduced surface. Therefore, we conclude that acetate/Vo is
the precursor to ketene, which is rapidly desorbed upon formation
(and so not observed spectroscopically) and that enolate in small
amounts is likely leading not to ketene but to acetaldehyde.
The present studies demonstrate that the TPD product distri-
bution is quite different when the initial surface is highly reduced.
Differences are shown from comparison of Figs. 1 and 2 and sum-
marized in Fig. 3 and Scheme 1. Briefly, highly reduced surfaces lead
to much more reduced products, a natural result of reoxidation of
the surface by the acetic acid. Consistent with this expectation is
that the ratio of H2:H2O is greatly increased on the reduced surface
compared to the oxidized surface, as is the ratio of CO:CO2. There is
a decrease of ketene but an increase in acetaldehyde and acetylene
resulting from unidentified, partially dehydrogenated intermedi-
ates. Reduction of the surface shuts down the coupling reaction
probably because of rapid deoxygenation of the acetate intermedi-
ates. This variation in the TPD product distribution is accompanied
by a variation in extent of reduction of the oxide surface. Adsorption
of acetic acid reduces a fully oxidized surface in part by providing
free H that reacts with the surface to evolve as water at low tem-
perature, but acetic acid also oxidizes a highly reduced surface by
leaving O on the surface. This behavior is similar to that reported
previously for formic acid [59] Another effect of a highly reduced
surface is that carbonaceous residue remains at high temperature
on the reduced surface for lack of available O. Finally, the fact that
the yield of acetone decreases on the more reduced surface suggests
that acetate/Vo is not sufficient for the formation of acetone or even
ketene, since both products decrease as the surface becomes highly
reduced. Formation of acetylene outcompetes ketene desorption on
the reduced surface, while the decrease in ␮-acetate on reduced
surface inhibits acetone formation. A practical implication is that
catalytic selectivity from a reactor may be adjusted by controlling
the reducibility of the reactant stream, for example by co-feeding
H2 or water.
Pestman et al. [57,58] studied the temperature-programmed
reaction of several acids on various oxide catalysts as a route to
ketone formation. For reactions of acetic acid, acetone was the pri-
mary product although they also found ketene and they examined
how its formation depends upon oxide and reaction conditions.
Interestingly, they proposed a flat-bonded “ketene-like” interme-
diate for the formation of acetone from acetic acid. Their evidence
was based upon comparison of aldehyde:ketone selectivity of sev-
eral acids with or without available ␣-H (i.e. H covalently bonded
to the C adjacent to the carboxylate carbon). Our TPD indicates that
ketonization to acetone is observed in small amounts from the oxi-
dized surface although it decreases on the more highly reduced
surfaces (Fig. 3). The acetate/Vo state observed at low tempera-
tures when vacancies are present, which is equivalent to a stably
bonded acetyl, could be an intermediate that is thermally activated
to yield both the observed desorption of ketene and the acetone
ketonization product. Evolution of ketene first requires the ther-
mally activated loss of the ␣-H to form a ketene-like intermediate
CH2 COO/Vo where one of the oxygens is strongly bonded at a
vacancy site (Fig. 10G), followed by the cleavage of the C O bond.
This intermediate can also be viewed as a bent ketene bonded to a
5. Conclusions
˚
surface O via the carboxyl C. Its C C bond has a length of 1.382 A,
which falls in the C C double bond range [7], and is −1.09 eV with
respect to gas-phase ketene. Based upon the DFT adsorption energy,
it is likely that upon the loss of the ␣-H in the acetate/Vo state at ele-
vated temperatures, the CH2 COO intermediate rapidly undergoes
Adsorption and transformations of acetic acid were studied
upon a model CeO2(1 1 1) surface. Near 175 K acetic acid adsorbs
to form a mix of molecular acetic acid and deprotonated acetate.
Acetic acid desorbs at low temperature along with water that leads
to reduction of the CeO2 surface. The resulting oxygen vacancies
form strong adsorption sites that trap acetate. Adsorbate states are
best described by a mixture of bridge bonded ␮-acetate and acetate
bonded at an oxygen vacancy (acetate/Vo), and a small amount
of enolate observed primarily on the more reduced surfaces. At
temperatures near 500–600 K products competitively desorb in a
mix of acetic acid, ketene, acetylene and acetone. Acetate/Vo is
the precursor to ketene and to acetylene, and its reaction with
␮-acetate leads to acetone. UHV conditions favor the unimolec-
ular product (ketene) over the coupling product (acetone) during
TPD. At higher temperature further decomposition to form CO and
CO2 is observed. The distribution of desorption products is strongly
dependent upon the extent of reduction of the surface prior to
adsorption. Increased reduction increases the ratio of H2:H2O and
the ratio of CO:CO2, and favors formation of acetaldehyde com-
pared to acetone or ketene. It is suggested that in a catalytic reactor,
the selectivity for acetone by ketonization of acetic acid should be
sensitive to co-fed reducing or oxidizing gases in the reactor stream.
C
O bond cleavage to release ketene. Linear ketene is calculated not
to adsorb on CeO2(1 1 1).
Experimentally there is faint spectroscopic evidence for stably
adsorbed ketene. In sXPS, a carbonyl bond of ketene should be
apparent at a binding energy between the carboxyl and the methyl
peaks. No such peak is evident in this range on the initially oxidized
surface (Fig. 4B) although on the initially reduced surface a weak
broad shoulder is observed at the higher annealing temperatures
(Fig. 6B). However, the reduced surface exhibits less ketene desorp-
tion than the oxidized surface, contradicting the hypothesis that
the peak is due to ketene. Also, the binding energy of the shoulder
feature is more closely positioned to be enolate than ketene. DFT
suggests that the ketene vibrational modes, being very weak, would
be difficult to see in RAIRS so failure to see them is inconclusive. The
NEXAFS resonance assigned to ␲* (C C) is compelling evidence for
a carbon–carbon double bond. But the resonance is also seen more
intensely on the initially reduced surface than the oxidized surface,
opposite to the trend of ketene desorption. These resonances could
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033

G Model
CATTOD-9550; No. of Pages12
ARTICLE IN PRESS
12
F.C. Calaza et al. / Catalysis Today xxx (2015) xxx–xxx
Acknowledgements
[18] J. Stohr, NEXAFS Spectroscopy, Springer-Verlag, New York, 2003.
[19] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J. Singh, C.
Fiolhais, Phys. Rev. B 46 (1992) 6671.
This research was sponsored by the Division of Chemical Sci-
ences, Geosciences, and Biosciences, Office of Basic Energy Sciences,
U.S. Department of Energy, under contract DE-AC05-00OR22725
with Oak Ridge National Laboratory, managed and operated by
UT-Battelle, LLC. F.C. Calaza and T.-L. Chen were sponsored by an
appointment to the Oak Ridge National Laboratory Postdoctoral
Research Associates Program administered jointly by Oak Ridge
Institute for Science and Education and Oak Ridge National Labora-
tory. Computational work performed at Louisiana State University
was in part supported by the Louisiana Experimental Program to
Stimulate Competitive Research (EPSCor) Program, funded by the
National Science Foundation and the Louisiana Board of Regents
Support Fund, and used high performance computational resources
provided by Louisiana State University. F.C. Calaza would like to
acknowledge Prof. Daniel Resasco for fruitful discussions regarding
the reaction mechanism for ketonization.
[20] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15.
[21] G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558.
[22] G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251.
[23] G. Kresse, P. Blaha, J.L.F. Da Silva, M.V. Ganduglia-Pirovano, Phys. Rev. B 72
(2005) 237101.
[24] S.L. Dudarev, G.A. Botton, S.Y. Savrasov, C.J. Humphreys, A.P. Sutton, Phys. Rev.
B 57 (1998) 1505.
[25] C. Loschen, J. Carrasco, K.M. Neyman, F. Illas, Phys. Rev. B 75 (2007) 035115.
[26] D.A. Andersson, S.I. Simak, B. Johansson, I.A. Abrikosov, N.V. Skorodumova,
Phys. Rev. B 75 (2007) 035109.
[27] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758.
[28] J. Neugebauer, M. Scheffler, Phys. Rev. B 46 (1992) 16967.
[29] H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188–5192.
[30] S.R. Bahn, K.W. Jacobsen, Comput. Sci. Eng. 4 (2002) 56–66.
[31] D. Porezag, M.R. Pederson, Phys. Rev. B 54 (1996) 7830.
[32] L. Kundakovic, D.R. Mullins, S.H. Overbury, Surf. Sci. 457 (2000) 51–62.
[33] S.D. Senanayake, W.O. Gordon, S.H. Overbury, D.R. Mullins, J. Phys. Chem. C 113
(2009) 6208–6214.
[34] D.R. Mullins, S.D. Senanayake, T.-L. Chen, J. Phys. Chem.
17112.
[35] M.B. Robin, I. Ishii, R. McLaren, A.P. Hitchcock, J. Electron Spectrosc. Relat. Phe-
nom. 47 (1988) 53–92.
C 114 (2010)
Appendix A. Supplementary data
[36] J. Zhou, D.R. Mullins, Surf. Sci. 600 (2006) 1540–1546.
[37] T.L. Chen, D.R. Mullins, J. Phys. Chem. C 115 (2011) 3385–3392.
[38] A.V. Ignatchenko, J. Phys. Chem. C 115 (2011) 16012–16018.
[39] D.R. Mullins, P.M. Albrecht, J. Phys. Chem. C 117 (2013) 14692–14700.
[40] T.-L. Chen, D.R. Mullins, J. Phys. Chem. C 115 (2011) 3385.
[41] T.-L. Chen, D.R. Mullins, J. Phys. Chem. C 115 (2011) 13725–13733.
[42] J.T. Francis, A.P. Hitchcock, J. Phys. Chem. 96 (1992) 6598.
[43] L.J. Terminello, D.K. Duh, F.J. Himpsel, D.A. Lapiano-Smith, J. Stohr, D.S. Bethune,
G. Meijer, Chem. Phys. Lett. 182 (1991) 491.
[44] G. Comelli, J. Stohr, C.J. Robinson, W. Jark, Phys. Rev. B 38 (1988) 7511–7519.
[45] J.G. Chen, Surf. Sci. Rep. 30 (1997) 1–152.
[46] L. Ramqvist, J. Appl. Phys. 42 (1971) 2113.
[47] N. C. WebBook, NIST Chemistry WebBook.
[48] T. Shimanouchi, NSRDS-NBS 39, U.S. Government Printing Office, Washington,
DC, 1972.
[49] J. Chao, B.J. Zwolinski, J. Phys. Chem. Ref. Data 7 (1978) 363–377.
[50] K.S. Kim, M.A. Barteau, Langmuir 4 (1988) 945–953.
[51] J. Stubenrauch, E. Brosha, J.M. Vohs, Catal. Today 28 (1996) 431–441.
[52] G.B. Deacon, R.J. Phillips, Coord. Chem. Rev. 33 (1980) 227–250.
[53] Z.-F. Pei, V. Ponec, Appl. Surf. Sci. 103 (1996) 171–182.
[54] N.J. Nicholas, G.V. Franks, W.A. Ducker, Langmuir 28 (2012) 7189–7196.
[55] C.C.R. Sutton, G.V. Franks, G. da Silva, Spectrosc. Acta A: Mol. Biomol. Spectrosc.
134 (2015) 535–542.
[56] M.A. Hasan, M.I. Zaki, L. Pasupulety, Appl. Catal. A: Gen. 243 (2003)
81–92.
[57] R. Pestman, A.v. Duijne, J.A.Z. Pieterse, V. Ponec, J. Mol. Catal. A: Chem. 103
(1995) 175–180.
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.cattod.2015.03.
033
References
[1] A. Trovarelli, P. Fornasiero, Imperial College Press; 2013.
[2] L. Vivier, D. Duprez, Chemsuschem 3 (2010) 654–678.
[3] D. Mullins, Surf. Sci. Rep. 70 (2015) 42–85.
[4] M. Li, Z. Wu, S.H. Overbury, J. Catal. 306 (2013) 164–176.
[5] A.K.P. Mann, Z. Wu, F.C. Calaza, S.H. Overbury, ACS Catal. 4 (2014) 2437–2448.
[6] D.R. Mullins, P.M. Albrecht, F. Calaza, Top. Catal. 56 (2013) 1345–1362.
[7] F.C. Calaza, Y. Xu, D.R. Mullins, S.H. Overbury, J. Am. Chem. Soc. 134 (2012)
18034–18045.
[8] A. Gangadharan, M. Shen, T. Sooknoi, D.E. Resasco, R.G. Mallinson, Appl. Catal.
A: Gen. 385 (2010) 80–91.
[9] R.W. Snell, B.H. Shanks, ACS Catal. 3 (2013) 783–789.
[10] A.V. Ignatchenko, E.I. Kozliak, ACS Catal. 2 (2012) 1555–1562.
[11] G. Pacchioni, ACS Catal. 4 (2014) 2874–2888.
[12] T.N. Pham, T. Sooknoi, S.P. Crossley, D.E. Resasco, ACS Catal.
2456–2473.
[13] G.N. Vayssilov, M. Mihaylov, P.S. Petkov, K.I. Hadjiivanov, K.M. Neyman, J. Phys.
Chem. C 115 (2011) 23435–23454.
[14] W.O. Gordon, Y. Xu, D.R. Mullins, S.H. Overbury, Phys. Chem. Chem. Phys. 11
(2009) 11171–11183.
[15] D.R. Mullins, P.V. Radulovic, S.H. Overbury, Surf. Sci. 429 (1999) 186–198.
[16] D.R. Mullins, M.D. Robbins, J. Zhou, Surf. Sci. 600 (2006) 1547–1558.
[17] E.I. Ko, J.B. Benziger, R.J. Madix, J. Catal. 62 (1980) 264–274.
3 (2013)
[58] R. Pestman, R.M. Koster, A. vanDuijne, J.A.Z. Pieterse, V. Ponec, J. Catal. 168
(1997) 265–272.
[59] S.D. Senanayake, D.R. Mullins, J. Phys. Chem. C 112 (2008) 9744–9752.
[60] K. Momma, F. Izumi, J. Appl. Crystallogr. 44 (2011) 1272–1276.
Please cite this article in press as: F.C. Calaza, et al., Reactivity and reaction intermediates for acetic acid adsorbed on CeO2(1 1 1), Catal.
Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.03.033