Dimethyl methylphosphonate decomposition on fully oxidized and partially reduced ceria thin films

Article abstract of DOI:10.1016/j.susc.2009.12.028

The thermal decomposition of dimethyl methylphosphonate (DMMP) on crystalline ceria thin films grown on Ru(0 0 0 1) was studied by temperature programmed desorption (TPD), X-ray photoelectron spectroscopy (XPS) and infrared absorption reflection spectroscopy (IRAS). TPD experiments show that methanol and formaldehyde desorb as the two main products at 575 K, while water, formaldehyde and CO are produced above 800 K. IRAS studies demonstrate that DMMP adsorbs via the phosphoryl oxygen at 200 K, but the P{double bond, long}O bond converts to a bridging O{single bond}P{single bond}O species at 300 K. DMMP decomposition initially occurs via P{single bond}OCH₃ bond scission to form methyl methylphosphonate (MMP) and methyl phosphonate (MP) between 300 and 500 K; XPS and IRAS data are consistent with a methoxy intermediate on the surface at these temperatures. The more stable P{single bond}CH₃ bonds remain intact up to 700 K, and the only surface intermediate at higher temperatures is believed to be PO_x. Although the presence of PO_x decreases activity for DMMP decomposition, some activity on the ceria surface remains even after 7 cycles of adsorption and reaction. The ceria films become reduced by multiple DMMP adsorption-reaction cycles, with the Ce⁺⁴ content dropping to 30% after seven cycles. Investigations of DMMP reaction on reduced ceria surfaces show that CO and H₂ are produced in addition to methanol and formaldehyde. Furthermore, DMMP decomposition activity on the reduced ceria films is almost completely inhibited after only 3 adsorption-reaction cycles. Similarities between DMMP and methanol chemistry on the ceria films suggest that methoxy is a key surface intermediate in both reactions.

Full text of DOI:10.1016/j.susc.2009.12.028

Surface Science 604 (2010) 574–587
Contents lists available at ScienceDirect
Surface Science
journal homepage: www.elsevier.com/locate/susc
Dimethyl methylphosphonate decomposition on fully oxidized
and partially reduced ceria thin films
a
a
b
Donna A. Chen a, , Jay S. Ratliff , Xiaofeng Hu , Wesley O. Gordon ,
*
Sanjaya D. Senanayake b, David R. Mullins b
a Department of Chemistry and Biochemistry, University of South Carolina, Columbia, SC 29208, United States
b Oak Ridge National Laboratory, Oak Ridge, TN 37831, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 25 October 2009
Accepted for publication 18 December 2009
Available online 29 December 2009
The thermal decomposition of dimethyl methylphosphonate (DMMP) on crystalline ceria thin films
grown on Ru(0 0 0 1) was studied by temperature programmed desorption (TPD), X-ray photoelectron
spectroscopy (XPS) and infrared absorption reflection spectroscopy (IRAS). TPD experiments show that
methanol and formaldehyde desorb as the two main products at 575 K, while water, formaldehyde
and CO are produced above 800 K. IRAS studies demonstrate that DMMP adsorbs via the phosphoryl oxy-
gen at 200 K, but the P@O bond converts to a bridging OAPAO species at 300 K. DMMP decomposition
initially occurs via PAOCH3 bond scission to form methyl methylphosphonate (MMP) and methyl phos-
phonate (MP) between 300 and 500 K; XPS and IRAS data are consistent with a methoxy intermediate on
the surface at these temperatures. The more stable PACH3 bonds remain intact up to 700 K, and the only
surface intermediate at higher temperatures is believed to be POx. Although the presence of POx decreases
activity for DMMP decomposition, some activity on the ceria surface remains even after 7 cycles of
adsorption and reaction. The ceria films become reduced by multiple DMMP adsorption-reaction cycles,
with the Ce+4 content dropping to 30% after seven cycles. Investigations of DMMP reaction on reduced
ceria surfaces show that CO and H2 are produced in addition to methanol and formaldehyde. Further-
more, DMMP decomposition activity on the reduced ceria films is almost completely inhibited after only
3 adsorption-reaction cycles. Similarities between DMMP and methanol chemistry on the ceria films sug-
gest that methoxy is a key surface intermediate in both reactions.
Keywords:
Soft X-ray photoelectron spectroscopy
Thermal desorption spectroscopy
Catalysis
Surface chemical reaction
Phosphorus
Ó 2009 Elsevier B.V. All rights reserved.
1. Introduction
stoichiometric rather than catalytic, ultimately limiting the capacity
for decomposition of the chemical agents [6–10]. The metal oxide
The development of materials for the catalytic decomposition of
organophosphorus nerve agents, such as VX, Sarin and Soman, is a
research area of critical importance for national security. While
carbon adsorbents have been employed in filtration devices for
decontamination in the field, the adsorption is reversible and
has limited capacity. Dimethyl methylphosphonate (DMMP,
(CH3O)2PO(CH3)) is frequently used as a simulant molecule for
understanding the activity of the organophosphorus nerve agents
because DMMP is nontoxic but has structural similarities to the
nerve agents. Despite excellent activity for breaking CAH, CAP
and PAO bonds, transition metal surfaces are poor catalysts for
organophosphorus decomposition due to poisoning of active
sites by phosphorus byproducts [1–5]. More recently, MgO and
CaO nanoparticles have been used as reactive adsorbents for
organophosphorus compounds, but the surface reactions are
surfaces are believed to be passivated by the formation of POx or me-
tal phosphates [9,11–13]. However, there is evidence that highly
reducible metal oxides like ceria and titania are less susceptible to
phosphorus poisoning from DMMP decomposition. For example,
ceria powders have been reported to be highly active for DMMP
decomposition, and this unusual activity was attributed to the abil-
ity of lattice oxygen to participate in the reaction [14]. In a previous
study, it was reported that the TiO2(1 1 0) surface maintains some
activity for DMMP decomposition after multiple cycles of adsorp-
tion and reaction [15]. Moreover, phosphorus from DMMP decom-
position could be removed from titania by heating in oxygen.
In this work, we are investigating the thermal chemistry of
DMMP on crystalline ceria thin films under ultrahigh vacuum con-
ditions in order to develop a fundamental understanding of the
decomposition of organophosphorus compounds on ceria. Ceria
is known as a good catalytic material, particularly for reactions
requiring a medium for oxygen storage [16]. Ordered ceria thin
films grown on Ru(0 0 0 1) are ꢀ2–5 nm thick [17–19] and exhibit
* Corresponding author. Tel.: +1 803 777 1050; fax: +1 803 777 9521.
E-mail address: chen@mail.chem.sc.edu (D.A. Chen).
0039-6028/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2009.12.028

D.A. Chen et al. / Surface Science 604 (2010) 574–587
575
high reflectivity associated with the metal support, providing high
signal intensity in the infrared reflection absorption spectroscopy
(IRAS) experiments; these ultrathin films are also sufficiently con-
ductive for X-ray photoelectron spectroscopy (XPS) and scanning
tunneling microscopy (STM). The ceria thin films have been exten-
sively studied and characterized by the Mullins group [17,20,21].
Another advantage of working with the ceria films is that the oxi-
dation states of ceria can be controlled. For the fully oxidized CeO2
films, low energy ion scattering studies indicate that the surface is
predominantly oxygen-terminated, and the films exhibit sharp
LEED patterns characteristic of the (1 1 1) surface of the fluorite
structure [20]. STM studies show that the ceria films are atomically
flat with single-layer high steps and terraces of 50–100 nm, and
the films can be imaged with atomic scale resolution [17]. How-
ever, Lu et al. have demonstrated that the film growth conditions
can influence the quality of the film [22]; morphological differ-
ences in the film are evident in STM but not in LEED. For the re-
duced CeOx films grown in a lower pressure of O2, ion scattering
indicates that a significant amount of Ce is exposed due to numer-
ous oxygen vacancies [20].
the ceria film prior to DMMP exposure was used as the back-
ground. Dry air-purged bags enclosed the entire beam path, but
some minor water interference was removed from the spectra by
manually subtracting reference water spectra collected with the
experimental setup. The crystal was initially cooled to 100 K and
transferred into a turbomolecularly-pumped IR cell. Heating to
the various temperatures was performed at a rate of ꢀ1 K/s, and
the sample was re-cooled to 100 K during collection of IR spectra.
DMMP was introduced into the IR cell at ꢀ10À5 Torr above the base
pressure; this exposure was sufficient to produce condensed
DMMP multilayers on the surface.
Soft X-ray photoelectron spectroscopy (sXPS) was conducted
using synchrotron radiation on beamline U12a at the National Syn-
chrotron Light Source (NSLS), and the associated UHV system has
been described elsewhere [23,24]. The base pressure in this system
was <4 Â 10À10 Torr. The sXPS data were collected with a VSW 125
hemispherical analyzer using a photon incident angle of 35° and
electron emission angle of 30° with respect to the surface normal.
Binding energies were referenced to the position of a satellite peak
at 122.3 eV in the Ce(4d) spectrum [19]. C(1s), O(1s) and P(2p)
spectra were collected at beam energies of 400 eV, 600 eV and
340 eV, respectively, and the instrumental resolution was roughly
0.5 eV. The CeO2 films were reduced by exposure to acetone at
600 K for 20 min at a pressure of 2 Â 10À6 Torr above the base
pressure. Previous studies show acetone is an effective reductant
for CeO2 [25], and an analysis of the XPS Ce(4d) region indicated
2. Experimental
CeO2(1 1 1) and partially reduced CeOx(1 1 1) films were grown
in situ by vapor deposition of cerium onto a Ru(0 0 0 1) surface held
at 700 K under an oxygen atmosphere. This procedure has been de-
scribed in detail in several previous publications [19,20,23]. The re-
duced ceria films were prepared by decreasing the oxygen
background pressure from 10À7 Torr to 10À8 Torr unless otherwise
specified. Growth of labeled Ce18O2(1 1 1) films was achieved by
substituting 16O2 with labeled 18O2 obtained from Cambridge Iso-
topes (>95% isotopic purity). The Ce oxidation states were deter-
mined from XPS data for the Ce 3d and Ce 4d regions by fitting
the spectra with a linear combination of pure Ce+3 and Ce+4 spectra
[21,24].
that the reduced CeOx films were ꢀ50% Ce+4
.
3. Results
3.1. TPD and conventional XPS studies
3.1.1. On stoichiometeric ceria (CeO2)
The reaction of DMMP on fully oxidized ceria produces formal-
dehyde (30 amu, CH2O) and methanol (31 amu, CH3O) as the major
gaseous products (Fig. 1a). The main desorption peaks for both
products occur at 575 K, but small desorption features are also ob-
served at 440 K; trace amounts of H2O desorption are detected at
575 K. Furthermore, a smaller high temperature peak for formalde-
hyde appears at 800 K and is accompanied by H2O desorption
(18 amu) at the same temperature. The evolution of DMMP itself
is monitored at 79 amu (PO3) since this is one of the more intense
large fragments of DMMP observed in the mass spectrum. DMMP
evolution occurs at ꢀ200 K in a sharp peak that is attributed to
multilayer desorption, but there are also smaller desorption fea-
tures at 240 K and 420 K. Masses other than those corresponding
to the cracking fragments of the identified products are not ob-
served. Specifically, there is no H2 (2 amu), methane (16 amu),
CO2 (44 amu), PO (47 amu) and PO2 (63 amu) beyond the intensi-
ties expected from DMMP cracking. Dimethyl ether production
was monitored at mass 46 amu, as further discussed in the next
paragraph. All of the 28 amu signal at 575 K can be accounted for
by cracking of formaldehyde and methanol, demonstrating that
the production of CO does not occur; however, the 28 amu signal
at 800 K most likely has some contribution from CO. The assign-
ment of the 31 amu peak to methanol is confirmed by the
31:32 amu ratio, which is the same as for pure methanol. The
assignment of the 79 amu peak at 420 K to DMMP rather than
PO3 or some other fragment is based on a comparison of the
47:63:79:94:124 amu ratios for the desorbing product with those
of DMMP itself. Similarly, identification of the 30 amu desorption
species as formaldehyde is based on the 29:30 amu ratio after both
of these signals are corrected for methanol fragmentation. Given
that the 30:31 amu ratio for pure methanol leaked into the cham-
Before all experiments, DMMP was purified by multiple freeze–
pump-thaw cycles. In some cases, the DMMP sample was found to
be contaminated by methyl bromide, which was removed by
immersing the DMMP bottle in a water–ice bath and pumping on
it for 10 h. Molecular sieves were also placed in the DMMP bottle
to eliminate water, and IRAS data for DMMP multilayers did not
indicate any impurities.
Temperature programmed desorption (TPD), conventional XPS
and IRAS experiments were performed in two different chambers
at Oak Ridge National Laboratory (ORNL). The base pressure in
the TPD/XPS system was 2 Â 10À10 Torr. TPD spectra were re-
corded by a UTI 100C mass spectrometer, and the temperature
was ramped at a rate of 2 K/s using an RHK TM310 temperature
programmer. During heating in front of the mass spectrometer,
the sample was biased at À70 V to prevent surface damage due
to electrons from the mass spectrometer ionizer. DMMP was dosed
at 150 K through an effusive gas doser [25,26] to minimize desorp-
tion from the sample holder, and the crystal was flashed to >600 K
prior to DMMP exposure to remove ambient water adsorbates. XPS
experiments were carried out with a MgKa anode and a Physical
Electronics double pass cylindrical mirror analyzer [24]. The re-
duced ceria films in the TPD experiments were ꢀ65% Ce+4
.
For the IRAS experiments, the near-stoichiometric films were
>99.5% Ce+4 and 85% Ce+4 according to XPS, while the CeOx films
were highly reduced with ꢀ30% Ce+4, corresponding to CeO1.65
.
IRAS data were collected with a Mattson Infinity infrared spec-
trometer by passing the light through a ZnSe wire grid p-polarizer,
reflecting off of the sample at a ꢀ85° angle and focusing onto a li-
quid nitrogen-cooled MCT detector. All spectra were acquired at
4 cmÀ1 resolution and summed over 1000 scans. The spectrum of

576
D.A. Chen et al. / Surface Science 604 (2010) 574–587
46 amu
CeO2
CeOx
0.30
0.25
0.20
0.15
0.10
0.05
0.00
0.30
0.25
0.20
0.15
0.10
0.05
0.00
a
1.0
0.8
31 amu
0.6
30 amu
0.4
28 amu
0.2
a
0.0
18 amu
1
2
3
4
5
6
7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
CeO2
CeOx
79 amu
800
200
400
600
b
46 amu
31 amu
30 amu
b
1
2
3
4
5
6
7
28 amu
Cycle
Fig. 2. Product yields for: (a) formaldehyde (30 amu) and (b) methanol (31 amu)
formation at 575 K, based on integrated areas in the TPD spectrum, as a function of
adsorption-reaction cycle. Data for the stoichiometric and reduced surfaces are
shown as circle and squares, respectively. The 30 amu signals were not corrected for
the cracking contribution of methanol, which accounts for ꢀ10% of the total 30 amu
signal on all surfaces.
18 amu
79 amu
800
TPD data for the first and second cycles are qualitatively very sim-
ilar with the main difference being a decrease in DMMP desorption
at 420 K for the second cycle. Major changes in the desorption
spectra occur only after three cycles of adsorption and reaction.
Specifically, the formaldehyde and methanol peaks become signif-
icantly broader and shift to higher temperature in addition to los-
ing intensity, as shown in the data for the fourth cycle (Fig. 1b).
Furthermore, the high temperature water and formaldehyde
desorption peaks disappear, and the distinct DMMP desorption
peak at 420 K is replaced with a much broader feature extending
from 200 to 400 K. After multiple adsorption-reaction cycles, we
attribute the absence of the 420 K DMMP peak and the shift of
DMMP desorption to lower temperatures to the blocking of sites
for strong DMMP adsorption due to byproducts of DMMP decom-
position. The desorption of dimethyl ether (46 amu) is also ob-
served in the same temperature range as formaldehyde and
methanol, and the identity of this product is supported by the
45:46 amu ratio, which matches that for dimethyl ether. The di-
methyl ether formation is not detected until the third adsorp-
tion-reaction cycle; its yield reaches a maximum for the fourth
cycle and decreases to zero by the seventh cycle.
200
400
600
Temperature (K)
Fig. 1. Temperature programmed desorption data for a saturation dose of DMMP on
stoichiometric ceria (CeO2) during: (a) the first adsorption-reaction cycle; and (b)
the fourth adsorption-reaction cycle. The heating rate was 2 K/s. For the 28 amu
signals in both (a) and (b), the rising baselines were subtracted out in order to more
clearly show the high temperature 28 amu peaks.
ber is roughly 1:10, ꢀ90% of the 575 K peak at 30 amu is estimated
to be from formaldehyde.
To investigate the ability of the ceria surface to sustain activity
for DMMP decomposition, TPD experiments were carried out for
successive cycles of adsorption and reaction on the same surface.
The major reaction products, which are formaldehyde and metha-
nol at 575 K, were monitored over seven reaction cycles (Fig. 2, cir-
cles). Product yields do not decrease substantially over the first
three cycles, but there is a sharp drop off between cycles three
and six, with the formaldehyde and methanol yields decreasing
to 15–18% of the first cycle values. The drop in product yields is
more gradual between cycles six and seven, and the fact that the
yields do not decrease to zero after the seventh cycle shows that
the surface sustains some activity for DMMP decomposition. The
The ceria films were grown in isotopically labeled 18O2 to ad-
dress the role of lattice oxygen in DMMP reaction (Fig. 3). TPD of
DMMP on the labeled ceria surface demonstrates that the DMMP
desorbed at 420 K exchanges oxygen with the ceria lattice, given

D.A. Chen et al. / Surface Science 604 (2010) 574–587
577
Ce+3
0.5
0.4
0.3
0.2
0.1
0.0
Ce+4
81 amu (18O-DMMP)
79 amu (DMMP)
Ce(3d)
e
d
33 amu (18O-methanol)
32 amu (methanol/
18O-formaldehyde)
c
31 amu (methanol/
18O-formaldehyde)
b
30 amu (16O/
18O-formaldehyde/C18O)
a
29 amu (methanol/
formaldehyde)
920
910
900
890
880
Binding Energy (eV)
Fig. 4. X-ray photoelectron spectroscopy data for the Ce(3d) region for the
stoichiometric ceria film: (a) before DMMP reaction; (b) after one adsorption-
reaction cycle; (c) after two adsorption-reaction cycles; (d) after three adsorption-
reaction cycles; and (e) after four adsorption-reaction cycles with DMMP.
200
300
400
500
600
700
800
Temperature (K)
tric ceria to monitor changes in the oxidation state. The
spectrum for the stoichiometric surface before DMMP reaction
(Fig. 4a) exhibits six peaks assigned to three pairs of spin–orbit
doublets. The multiple peaks associated with the 3d7/2 and 3d5/2
binding energies are due to different final state occupancies in
the 3d levels [24]. There are two features that serve as good mark-
ers for gauging the level of reduction in the ceria surface: the peak
at 916 eV is characteristic of Ce+4, and the peak at 884 eV is char-
acteristic of Ce+3 state [24]. After the first DMMP adsorption-reac-
tion cycle, the intensity of the 916 eV decreases while the valley
near 884 eV begins to fill in (Fig. 4b). The decrease in the 916 eV
is more pronounced after the second cycle, and a peak at 884 eV
clearly appears (Fig. 4c). Further evidence of Ce+4 reduction is
found after the third and fourth cycles (Fig. 4d and e). The extent
of reduction can be quantified by fitting the data with a linear
combination of the fully oxidized and fully reduced spectra. A plot
of % Ce+4 as a function of TPD adsorption-reaction cycle (Fig. 5,
circles) shows a large decrease in Ce+4 after the first three cycles,
with the Ce+4 content dropping from 100% to 45%. After cycles 4–
7, the decrease in Ce+4 is more gradual and diminishes to 30% after
the seventh cycle. Notably, the activity for DMMP decomposition
also decreases sharply after the third cycle as the Ce+4 content
drops below 50%, suggesting that Ce+4 sites may be important for
decomposition activity.
Fig. 3. Temperature programmed desorption data for a saturation exposure of
DMMP on Ce18O2. The heating rate was 2 K/s.
that both 18O-labeled (81 amu) and unlabeled DMMP (79 amu) are
detected with approximately equal intensities. No oxygen ex-
change with the lattice occurs in methanol since the 33 amu
(CH318O) signal from the main cracking fragment of 18O-labeled
methanol is not detected at any temperature, and therefore the
18O-methanol fragments do not contribute to the intensity ob-
served at other masses. Masses 29–32 amu represent a combina-
tion of cracking fragments from unlabeled methanol as well as
18O-formaldehyde and 16O-formaldehyde. There is no oxygen ex-
change in formaldehyde and methanol produced at 420 K because
the 29:30:31:32 amu ratios are identical on the labeled and unla-
beled surfaces. However, oxygen exchange occurs in formaldehyde
production at 580 K: the 31:29 and 32:29 amu ratios are higher on
the labeled surface compared to the unlabeled due to increased
contributions from masses 31 and 32 amu from the cracking frag-
ments of 18O-formaldehyde. Only 18O formaldehyde (32, 31,
30 amu) and no unlabeled formaldehyde (29 amu) is produced at
780 K, indicating that all of the formaldehyde formed at this tem-
perature undergoes oxygen exchange with the ceria. The formalde-
hyde produced at 780 K contains only lattice oxygen instead of
oxygen from DMMP because this product arises from PACH3 bond
breaking and combination of CH3 with lattice oxygen, rather than
from PAOCH3 bond breaking in DMMP. Since the 30:32 amu ratio
at 780 K is greater than the corresponding 28:30 amu ratio for
unlabeled formaldehyde, some of the desorption at 30 amu is
attributed to C18O [27].
3.1.2. On reduced ceria (CeOx)
DMMP decomposition on an intentionally reduced ceria surface
was studied by TPD in order to further elucidate the relationship
between surface reduction and activity. The Ce(3d) region in the
XPS spectrum (Fig. 6a) exhibits peaks at 884 eV and 905 eV charac-
teristic of Ce+3, and the surface is estimated to be ꢀ35% Ce+3. The
X-ray photoelectron spectra for the Ce(3d) region were col-
lected after each of the adsorption-reaction cycles on stoichiome-

578
D.A. Chen et al. / Surface Science 604 (2010) 574–587
100
80
60
40
20
0
0.4
a
CeO2
CeOx
79 amu
31 amu
0.3
0.2
30 amu
28 amu
0
1
2
3
4
5
6
7
18 amu
Cycle
Fig. 5. Ce+4 content as a function of DMMP adsorption-reaction cycle as determined
by XPS. The circles are for the initially stoichiometric film and the squares are for
the initially reduced CeOx film.
0.1
2 amu
0.0
0.4
300 400 500 600 700 800 900
Ce+3
Ce+4
Ce(3d)
d
b
79 amu
31 amu
0.3
0.2
0.1
0.0
c
30 amu
28 amu
b
18 amu
2 amu
a
920
910
900
890
880
300 400 500 600 700 800 900
Binding Energy (eV)
Temperature (K)
Fig. 6. X-ray photoelectron spectroscopy data for the Ce(3d) region for the reduced
ceria surface: (a) before DMMP reaction; (b) after one adsorption-reaction cycle; (c)
after two adsorption-reaction cycles; and (d) after three adsorption-reaction cycles
with DMMP.
Fig. 7. Temperature programmed desorption data for a saturation dose of DMMP on
reduced ceria for: (a) the first adsorption-reaction cycle; and (b) the third
adsorption-reaction cycle. The heating rate was 2 K/s.
TPD of DMMP on the reduced surface produces formaldehyde and
methanol as the major products (Fig. 7a), but the desorption tem-
perature is 630 K, which is ꢀ55 K higher than on stoichiometric
ceria. The desorption of H2 (2 amu) is also observed at 630 K. After
correcting the 28 amu signal at ꢀ600 K for mass fragmentation
from methanol and formaldehyde, little if any of the 28 amu signal
can be attributed to CO. Formaldehyde desorption is not observed
at high temperature, and instead, H2 and CO production are de-
tected in addition to H2O at 820–850 K. While the lower tempera-
ture methanol peak is more pronounced than on the stoichiometric
surface and is shifted to 535 K, the formaldehyde peak at this tem-
perature is absent. In contrast to activity on CeO2, DMMP desorp-
tion at 420 K is not observed. Methanol and formaldehyde
production decreases for the second adsorption-reaction cycle on
the reduced surface, and there is a pronounced drop in activity
for the third cycle (Figs. 7b and 2 squares). The formaldehyde
and methanol desorption peaks become broader, less intense and
shift to high temperature. Furthermore, the water and H2 peaks
at ꢀ630 K disappear completely although H2 desorption is still ob-
served at 800–900 K. A comparison of the formaldehyde and meth-
anol yields on stoichiometric and reduced ceria as a function of
adsorption-reaction cycle is shown in Fig. 2. While the initial activ-
ity CeOx is greater than that of CeO2, the activity for DMMP decom-
position on CeOx drops off more rapidly. XPS data for the Ce(3d)
region (Fig. 6) demonstrate that there is substantial reduction of
ceria after the first cycle, but spectral changes after subsequent cy-

D.A. Chen et al. / Surface Science 604 (2010) 574–587
579
cles are less extensive; after the first cycle, the Ce+4 content drops
from 65% to 45% and further decreases to ꢀ30% after the second
and third cycles (Fig. 5).
ms(PAO) [37]. Infrared spectra for Na2CH3PO3 in solution also
shows PAO modes at 1085, 1050 and 972 cmÀ1 [38]. Furthermore,
heating DMMP on magnesia, lanthana and alumina to 673–773 K
results in the appearance of modes at 1190–1132 cmÀ1 and
3.2. Infrared spectroscopy studies
1065–1051 cmÀ1 that are attributed to
m
(PO3) [32]. Although these
peaks overlap with the positions of the CAO stretching vibrations
in DMMP and methoxy as well as (OCH3), the disappearance of
the da(OCH3) and (OCH3) modes illustrates that DMMP and meth-
3.2.1. On stoichiometric and near-stoichiometric ceria
q
Infrared spectroscopy experiments were carried out to identify
surface intermediates at various stages of the thermal decomposi-
tion of DMMP on ceria. The data shown in Fig. 8a are from a surface
that was 85% Ce+4. Experiments were also carried out on a fully oxi-
dized ceria film; the data from the fully oxidized film was the same
q
oxy do not remain on the surface at these temperatures [32].
At 800 K (data not shown), the decrease in intensity for ds(PCH3)
at 1301 cmÀ1 demonstrates that all of the PACH3 bonds are broken.
The intense peak at 1045 cmÀ1 is lost, and a peak at 1025 cmÀ1 ap-
pears, in addition to peaks at 987 and 924 cmÀ1, which are also ob-
served in the spectrum at 700 K. These features are assigned to the
PAO stretches of POx, which is believed to be the main surface spe-
cies at 800 K. It is unclear why there is a huge decrease in intensity
and a shift to lower frequency for the PAO mode originally ob-
served at 1045 cmÀ1. However, changes in the infrared spectrum
have also been observed for CePO4 crystals heated to different tem-
peratures to induce structural transformations [39].
as that in Fig. 8a although the da(OCH3) and
q(PCH3) peaks were
not as clearly visible due to issues with baseline subtraction. At
200 K, vibrational modes characteristic of intact DMMP are ob-
served, and peak assignments are based on the spectra for gas
and liquid phase DMMP [28–31] (Table 1) as well as DMMP ad-
sorbed oÀn1 other oxide surfaces [12,30–33]. The
m(P@O) mode at
1225 cm is significantly red-shifted from its gas phase value of
1275 cmÀ1 while the other modes are close to their gas phase val-
ues, supporting the interaction between the phosphoryl oxygen
and the oxide surface. Other features observed at this temperature
The formation of MMP and MP via PAOCH3 bond scission sug-
gests that surface methoxy should also be an intermediate. Deter-
mining if surface methoxy appears in the infrared spectra is
complicated by the fact that the position of the CAO stretch in
methoxy intermediates is known to be sensitive to binding site.
Studies of methanol on ceria powders [28,40] and crystalline thin
films [41] report CAO stretches for singly, doubly and triply bound
methoxy at 1105–1108, 1058–1065 and 1015–1028 cmÀ1, respec-
tively [28,40,41], with the latter two species requiring the exis-
tence of oxygen vacancies [41]. At 300 and 500 K, a shoulder at
1058 cmÀ1 appears and could be assigned to methoxy at twofold
sites; the intensity of this feature is difficult to determine since it
overlaps with major peaks at 1076 and 1045 cmÀ1. The presence
of a bridging methoxy species at lower temperatures is consistent
with the mix of 16O- and 18O-formaldehyde produced at 575 K. At
400 K, the peak at 1020 cmÀ1 could be attributed to threefold coor-
dinate methoxy, and this feature disappears almost completely at
500 K. Furthermore, in the spectra collected between 300 and
include the
m
(CAO) in DMMP at 1070 cmÀ1, da(OCH3) and
q(OCH3)
at 1467 and 1188 cmÀ1, respectively, and ds(PCH3) and
q(PCH3) at
1311 and 922 cmÀ1, respectively. After heating to 300 K, the most
notable change in the spectrum is the disappearance of the P@O
stretch. This is accompanied by a rise in intensity at 1080 cmÀ1
,
which is attributed to the
[7,30,31,33–36]. The a(OAPAO) mode is expected to appear
around 1189 cmÀ1, which overlaps with the
(OCH3) mode from
adsorbed DMMP. However, at 300 K, the low intensity of this peak
compared to s(OAPAO) and the fact that the peak does not in-
ms(OPO) vibration in a bridging P species
m
q
m
crease between 200 and 300 K suggests that the ma(OAPAO) mode
does not contribute to the 1188 cmÀ1 peak. It is more likely that
the bridging OAPAO species is oriented perpendicular to the sur-
face so that the ma(OAPAO) vibration is forbidden by the surface di-
pole selection rule.
Between 200 and 400 K, the decrease in intensity for the
da(OCH3) and
q(OCH3) modes indicates PAOCH3 bond scission in
500 K, there is intensity at 1100 cmÀ1 from the
ms(OPO) vibration,
DMMP, which produces a methyl methlyphosphonate (MMP,
(CH3O)PO(CH3)) intermediate. In contrast, the PACH3 rock and
deformation modes do not lose intensity, demonstrating that the
PACH3 bonds are not broken at these temperatures. At 500 K, the
and it is impossible to rule out some contribution from methoxy
at top sites. Thus, the infrared data are consistent with the pres-
ence of surface methoxy between 300 and 500 K. Surface methoxy
is not expected at 600 K, which is above the desorption tempera-
ture for the main methanol and formaldehyde desorption peaks.
1088 cmÀ1 peak from s(OPO) and 1070 cmÀ1 peak from
m m(CAO)
in the POACH3 groups both lose intensity due to continued
PAOCH3 and OAPAO bond scission. At 700 K, the complete disap-
pearance of the OAPAO mode demonstrates that all of the bridging
phosphorus bonds are dissociated, and the loss of intensity at
1070 cmÀ1 is consistent with complete PAOCH3 bond scission.
The PCH3 modes at 1311 and 922 cmÀ1 are still present, indicating
intact PACH3 bonds. In fact, the da(OCH3) peak at 1311 cmÀ1 be-
comes sharper and more intense after heating to 500 K, and this
suggests a change in structure of the surface intermediate.
3.2.2. On reduced ceria (CeOx)
On the reduced ceria surface, molecular DMMP is observed at
200 K (Fig. 8b), and the spectrum closely resembles that of DMMP
on stoichiometric ceria. Adsorption of DMMP via the phosphoryl
oxygen is demonstrated by the shift of the P@O 1230 cmÀ1 to low-
er frequencies while other modes are unshifted compared to gas
phase DMMP. After heating to 300 K, the P@O stretch disappears,
and the
oxygen is converted to a OAPAO bridging species. The peak at
1063 cmÀ1 from
(CO) in DMMP becomes more pronounced, and
it is possible that both (CAO) in DMMP and surface methoxy con-
m
s(OPO) peak at 1078 cmÀ1 grows in as the phosphoryl
These spectra are consistent with a methyl phosphonate (MP,
CH3PO) species at 700 K and with MP being the majority interme-
diate at 500 K, given the loss in intensity for the da(OCH3) and
m
m
q
(OCH3) vibrations. Other changes in the spectrum after heating
tribute to the intensity of this peak. As on the stoichiometric sur-
face, heating from 200 to 400 K causes the intensity of the
da(OCH3) peak to decrease significantly, indicating PAOCH3 bond
breaking to form first MMP and later MP. At 400 K, there is very lit-
tle MMP on the surface, and at 580 K, the surface intermediate ap-
pears to be exclusively MP since the da(OCH3) peak disappears. The
ds(PCH3) mode is more broad on CeOx compared to CeO2, possibly
due to greater surface site heterogeneity. Furthermore, this peak
splits into two features at 300 K and is significantly red-shifted at
the higher temperatures, suggesting perturbation of the CAP
to 700 K include an increase in intensity at 1045 cmÀ1, and this
sharp peak is ascribed to the PAO stretches in an MP intermediate
based on previous studies of DMMP on metal oxides [12,32]. Work
by Tripp et al. suggests that on TiO2 and WO3 powders, MP bound
to three lattice oxygen atoms remains on the surface above 623 K;
this assignment is supported by infrared spectra that exhibit peaks
at 1150, 1050 and 983 cmÀ1 [12]. Based on the infrared spectra of
crystalline CePO4 orthophosphate, the 1150 and 1050 cmÀ1 modes
are assigned to m
a(PAO), and the 983 cmÀ1 mode is assigned to

580
D.A. Chen et al. / Surface Science 604 (2010) 574–587
a
700 K
500 K
400 K
300 K
200 K
1500
1400
1300
1200
1100
1000
900
Wavenumber (cm-1)
b
700 K
580 K
400 K
300 K
200 K
1500
1400
1300
1200
1100
1000
900
Wavenumber (cm-1)
Fig. 8. Infrared reflection absorption spectroscopy data for a saturation dose of DMMP adsorbed at 100 K and heated to various temperatures on: (a) near-stoichiometric ceria
(CeO1.8); and (b) reduced ceria (CeO1.65). Heating rates were approximately 1 K/s.
bonds, which remain intact up to 700 K. The
splits into two features at temperatures of 300 K and higher. The
q
(OCH3) peak also
OAPAO mode and CAO stretch from MMP disappear completely
at 580 K, and a new peak at 968 cmÀ1 becomes one of the domi-

D.A. Chen et al. / Surface Science 604 (2010) 574–587
581
Table 1
Assignments for vibrational modes of DMMP on the near stoichiometric (CeO1.9) and reduced (CeO1.65) surfaces at 200 K. For comparison, assignments for gas phase DMMP, [31]
liquid phase DMMP [29], DMMP on alumina at 200 K [30], DMMP on WO3 at 300 K [12], and DMMP on titania at 300 K [12] are also shown. All values are reported in cmÀ1, and
peak positions were determined from infrared spectroscopy experiments, except for the DMMP on alumina data, which is from inelastic tunneling spectroscopy measurements.
DMMP on
CeO1.9
DMMP on
CeO1.65
Gas phase
DMMP
Liquid phase
DMMP
DMMP on alumina
at 200 K
DMMP on
WO3 at 300 K
DMMP on
TiO2 at 300 K
da (OCH3)
ds(PCH3)
1467
1311
1225
1188
1068
922
1467
1317
1230
1189
1063
–
1467
1314
1276
1466
1314
1245
1452
1308
–
1177
1048
900
1449
1313
1223
1459
1314
1237
m
(P@O)
(OCH3)
(CO)
(PCH3)
q
1190
1186
1193
1186
m
1048–73
916
1038–65
915
1036–58
918
1037–61
919
q
nant spectral features. The 968 cmÀ1 peak is ascribed to the PAO
stretches in a MP intermediate coordinated to lattice oxygen, given
that solid cerium phosphate samples also have an infrared band in
the 925–996 cmÀ1 range [37,39]. It is unlikely that the 968 cmÀ1
peak is from the PACH3 rock because this mode normally appears
at a lower frequency of around 922 cmÀ1 and is already absent at
lower temperatures even though the ds(PACH3) peak is still visible.
MP is also proposed as an intermediate in DMMP decomposition
on the stoichiometric CeO2 films, but the PAO stretches are ob-
served at higher frequencies, perhaps suggesting that the struc-
tural form of POx is different on the oxidized and reduced
surfaces. A small peak at 1218 cmÀ1 appears at 580 K and remains
at 700 K. While this feature cannot be definitively assigned, it is in
gies are consistent with the vibrational modes of a POx species, as
discussed in the previous section.
3.3. Soft X-ray photoelectron spectroscopy studies
3.3.1. On stoichiometric ceria (CeO2)
The C(1s) region for DMMP multilayers (10 L exposure) at 100 K
shows two main peaks at 288.1 eV and 286.3 eV that correspond to
the methoxy and methyl carbons, respectively, in condensed
DMMP (Fig. 9) [43–45]. After heating to 200 K, the peaks at
287.8 eV and 286.1 eV are again attributed to the methoxy and
methyl carbons in molecular DMMP, and the 1.3:1 peak area ratio
compared to the expected 2:1 ratio implies that some decomposi-
tion occurs at this temperature. A smaller peak at ꢀ290 eV is also
observed and has a binding energy that could be assigned to sur-
face formate [19]. This may have been caused by X-ray induced
damage as has previously been observed for formaldehyde and
acetone on ceria [18,25]. The presence of molecular DMMP on
the surface at 200 K is consistent with the TPD data that show
DMMP desorption above this temperature.
the correct frequency range for
m(P@O) [28,42]. The intense peak at
1050 cmÀ1 is assigned to
ma(PAO), as on the stoichiometric surface.
At 700 K, the PACH3 bonds remain intact, and the dominant peaks
at 1050 and 968 cmÀ1 lose intensity although smaller features
around 1050, 1000 and 973 cmÀ1 can still be detected. These ener-
At 300 K, the main surface species still appear at ꢀ287.6 eV and
286.2 eV although the ꢀ1:1 peak area ratio indicates more exten-
sive DMMP decomposition. In addition, the valley between these
two peaks has filled in, suggesting the presence of a new feature
near 287 eV that persists after further annealing at 400 K. This
additional feature near 287 eV is assigned to methoxy adsorbed
on CeO2, given that methoxy resulting from the decomposition of
methanol on CeO2 has a binding energy of 287.2 eV [19]. PAOCH3
bond scission is typically observed as part of the decomposition
mechanism of DMMP on metal oxide surfaces and would produce
surface methoxy [6,30,32,46,47]. At 600 K, the main surface species
has a binding energy of ꢀ285.7 eV, and this is the only peak ob-
served after heating to 800 K. This binding energy is lower than
any of the previously reported OACHx species adsorbed on CeO2
[18,19,25,48] and is therefore likely related to C bound to an elec-
tropositive element such as PACHx or CeACHx. At 900 K, no carbon
is detected on the surface.
100
C(1s)
900 K
P-CH3
Ce-O-CH3
80
800 K
600 K
P-OCH3
60
400 K
300 K
40
P-OCH3
200 K
In terms of overall signal intensity, the integrated C(1s) signal
exhibits the sharpest decrease between 100 and 200 K as con-
densed DMMP desorbs from the surface. The carbon signal inten-
sity continues to drop between 400 and 600 K, and this can be
explained by the desorption of carbon-containing products such
as methanol, formaldehyde and molecular DMMP.
20
The O(1s) spectrum of the ceria film is dominated by the lattice
oxygen peak at 530.3 eV, which is the only feature observed in the
absence of DMMP. After DMMP adsorption at 100 K, the lattice
oxygen peak is significantly attenuated, and two new peaks appear
at 535 eV and 533 eV (Fig. 10). Based on their relative intensities
and the stoichiometry in DMMP, these peaks are assigned to the
methoxy oxygens (POCH3) and the phosphoryl oxygen (P@O) in
DMMP, respectively; their binding energies are similar to what
has been reported for multilayer DMMP on Rh(1 0 0) [45]. At
100 K
292
290
288
286
284
282
Binding Energy (eV)
Fig. 9. Soft X-ray photoelectron spectroscopy data for the C(1s) region for a 10 L
dose of DMMP adsorbed on stoichiometric ceria (CeO2) at 100 K and then heated as
indicated. Spectra were collected with a 400 eV photon energy.

582
D.A. Chen et al. / Surface Science 604 (2010) 574–587
200 K, the desorption of multilayer DMMP results in an increase in
the intensity of the lattice oxygen peak and a small shift to
524.8 eV for the binding energy of the methoxy oxygens. The de-
crease in intensity between 200 and 300 K in the peak assigned
to POCH3 is partly related to the desorption of DMMP but also
due to PAOCH3 bond scission. Some of the PAOCH3 groups persist
up to 400 K, which is consistent with the presence of POCH3 spe-
cies in the C(1s) spectra. The phosphoryl oxygen binding energy
in DMMP shifts from 533 eV and 532 eV; similar shifts to lower
binding energies are generally reported in the conversion of multi-
layer to chemisorbed species [25], and these shifts also are ob-
served in the C(1s) region. As PAOCH3 and PACH3 bonds are
cleaved, new bonds may form between P and surface oxygen.
POx species are presumably the only non-lattice O-containing spe-
cies at 800 K and 900 K. Therefore, the peak near 532 eV that is
present in all spectra from 200 to 900 K is primarily attributed to
POx. Methoxy and hydroxyl on oxidized CeO2 have binding ener-
gies of ꢀ531.8 eV [19], and these species may also contribute
intensity to the binding energy region between the lattice oxygen
and the POx peaks in Fig. 10.
(Fig. 11). Features in the P(2p) spectra can be most easily assigned
for the data collected at 100 K and after annealing at 900 K. Upon
adsorption at 100 K, the dominant feature in the spectrum is the
peak at 134.8 eV, which is attributed to molecular DMMP and is
in agreement with the binding energy of DMMP on other surfaces
[43–45,49]. At 900 K, the P(2p) peak at 133.6 eV is ascribed to POx
species, and this binding energy is consistent with the 133.4 eV re-
ported for CePO4 [50]. CePO4 forms when phosphorus-containing
molecules react on ceria surfaces and is known to be thermally sta-
ble [51–54]. The P(2p) doublet lineshape at 900 K is not as clear as
in the spectrum of DMMP at 100 K, suggesting heterogeneity due
to different species or adsorption sites.
At 200 K, the peak from molecular DMMP shifts to ꢀ134.5 eV.
The appearance of intensity near 133.5 eV reflects the decomposi-
tion of PAOCH3 bonds and the formation of POx bonds. At 300 K,
most of the intensity above 135 eV disappears, and the intensity
around 133.5 eV increases as DMMP decomposes and desorbs
upon further heating. After heating to 400 K and 600 K, the P(2p)
peaks shift to lower binding energy due to surface species that con-
tribute intensity in the 133 eV region. However, the peak maxi-
mum shifts back to higher binding energy after heating to 800 K
and 900 K, and this shift may be due to a change in adsorption site,
as was postulated for sulfur on CeOx [55].
The P(2p) region is the most difficult to interpret since there are
multiple phosphorus-containing species on the surface at all tem-
peratures, and it is not possible to uniquely identify all species.
However, assignments can be made that are consistent with the
infrared data and XPS data from the oxygen and carbon regions.
The P(2p) signal consists of a P(2p3/2) and P(2p1/2) peaks that are
separated by 1 eV and have a relative intensity ratio of 2:1. For
clarity, the position of only the P(2p3/2) peak is noted here
3.3.2. On reduced CeOx
The interpretation of the XPS spectra recorded for DMMP ad-
sorbed on CeOx is very similar that obtained on CeO2. As a general
rule, the binding energies for similar species adsorbed on a reduced
ceria surface are 0.5–1 eV greater than their counterparts on oxi-
dized ceria [18,19,25,48], and this behavior is observed for DMMP
in the C(1s), O(1s) and P(2p) regions. For example, the C(1s) peaks
associated with PAOCH3 and PACH3 in DMMP at 100 K on CeOx
lattice
oxygen
250
P=O/POx
O(1s)
POx
20
P(2p)
900 K
200
150
100
50
900 K
800 K
800 K
15
CeOCH3
600 K
600 K
P-OCH3
400 K
10
400 K
300 K
300 K
DMMP
200 K
5
200 K
P=O
100 K
528
100 K
538
536
534
532
530
138
136
134
132
130
Binding Energy (eV)
Binding Energy (eV)
Fig. 10. Soft X-ray photoelectron spectroscopy data for the O(1s) region for a 10 L
dose of DMMP adsorbed on stoichiometric ceria (CeO2) at 100 K and then heated as
indicated. Spectra were collected with a 600 eV photon energy.
Fig. 11. Soft X-ray photoelectron spectroscopy data for the P(2p) region for a 10 L
dose of DMMP adsorbed on stoichiometric ceria (CeO2) at 100 K and then heated as
indicated. Spectra were collected with a 340 eV photon energy.

D.A. Chen et al. / Surface Science 604 (2010) 574–587
583
(Fig. 12) have shifted from 288.1 eV to 288.7 eV, and from 286.3 eV
to 287.0 eV, respectively, compared to the positions on CeO2.
In the C(1s) spectrum of DMMP on CeOx, there is a small shift to
lower binding energy at 200 K as the multilayer desorbs, and the
decrease in the relative intensity of the peak assigned to the POCH3
feature implies partial decomposition of these moieties. The inten-
sities of the C(1s) peaks from PAOCH3 and PACH3 remain nearly
equal between 200 and 400 K, indicating that roughly half of the
PAOCH3 groups have decomposed. The intensity around 288 eV
between the PAOCH3 and PACH3 peaks can again be assigned to
surface methoxy, given that methoxy on CeOx has a C(1s) binding
energy of 288 eV [19]. After annealing at 600 K, the C(1s) spectra
is quite broad, and the PACH3 peak shifts to lower binding energy
as the remaining PAOCH3 fragments fully decompose. The persis-
tence of intensity around 288 eV suggests that methoxy remains
on surface. In addition, a new feature near 285 eV begins to appear,
and remnants of this peak are observed even after annealing to
900 K. A feature with a similar binding energy was assigned to a
CHx species bound to Ce following acetone decomposition on CeOx
[25].
oxy decomposes at higher temperature, suggesting that the POx
species has a slightly lower binding energy than CeAOCH3 on CeOx.
The P(2p) spectra show clearer changes at various temperatures
on reduced CeOx (Fig. 14) compared to the spectra on CeO2. At
300 K and 400 K, the spectra is dominated by a well-defined
P(2p) doublet with a 2p3/2 peak near 134.7 eV, and this feature is
ascribed to phosphorus in MMP. After annealing to 600 K, the
P(2p) peaks shifts to 134 eV as the second methoxy group is re-
moved to form MP. There is only a small binding energy shift
(<0.2 eV) between 600 and 800 K, indicating that MP and POx can-
not be easily distinguished in the P(2p) spectra on reduced ceria,
presumably because the phosphorus atom in MP is coordinated
to lattice oxygen. In contrast to behavior on the CeO2 surface, there
is also only a small shift back to higher binding energies after
annealing to 900 K on CeOx.
4. Discussion
4.1. Reaction mechanism
In the O(1s) spectra of DMMP on reduced ceria (Fig. 13), there is
greater separation between the adsorbate related features and the
lattice O peak since the adsorbate features are shifted to higher
binding energy compared to the values on CeO2. After annealing
to 300 K, the peak at ꢀ532.8 eV is much more intense on CeOx
compared to the peak at 532.2 eV on CeO2. This is because the
binding energies of POx and the CeAOCH3 are nearly coincident
on reduced ceria since methoxy on CeOx appears at 532.8 eV
[19]. There is a small shift to lower binding energy as surface meth-
Based on a combination of TPD, IRAS and XPS data, a reaction
mechanism for DMMP decomposition on stoichiometric ceria is
proposed (Fig. 15). At 200 K, DMMP is chemisorbed to the surface
via the phosphoryl oxygen, and XPS data indicate that some
decomposition occurs, most likely via PAOCH3 bond scission. The
P(2p) data also suggest that a bridging OAPAO species is formed
at this temperature. After heating to 300 K, all of the phosphoryl
groups are converted to the bridging OAPAO species, and contin-
ued PAOCH3 bond scission occurs. Above 400 K, molecular DMMP
desorbs as well as a small amount of methanol and formaldehyde.
DMMP undergoes oxygen exchange with the ceria, which is consis-
100
C(1s)
900 K
lattice
oxygen
P-CH3
P=O/POx
250
200
150
100
50
O(1s)
800 K
80
Ce-O-CH3
P-OCH3
900 K
600 K
60
800 K
600 K
400 K
Ce-O-CH3
P-OCH3
300 K
40
400 K
300 K
200 K
20
200 K
P=O
100 K
100 K
528
292
290
288
286
284
282
538
536
534
532
530
Binding Energy (eV)
Binding Energy (eV)
Fig. 12. Soft X-ray photoelectron spectroscopy data for the C(1s) region for a 10 L
dose of DMMP adsorbed on reduced ceria (CeOx) at 100 K and then heated as
indicated. Spectra were collected with a 400 eV photon energy.
Fig. 13. Soft X-ray photoelectron spectroscopy data for the O(1s) region for a 10 L
dose of DMMP adsorbed on reduced ceria (CeOx) at 100 K and then heated as
indicated. Spectra were collected with a 600 eV photon energy.

584
D.A. Chen et al. / Surface Science 604 (2010) 574–587
tent with a bridging OAPAO intermediate, whereas no lattice oxy-
gen exchange is found in methanol or formaldehyde at this tem-
perature. At 500 K, OAPAO bonds are still observed, most likely
from a bridging MP species since the majority of PAOCH3 bonds
are cleaved. There is also evidence that PAOCH3 bond scission be-
low 600 K produces surface methoxy. At 575 K, formaldehyde and
methanol desorb from the surface, and exchange with lattice oxy-
gen occurs in formaldehyde but not in methanol. After heating to
700 K, the PACH3 bonds remain intact in what is assumed to be
a MP intermediate, and no intact OAPAO bonds are detected.
Formaldehyde and water desorption occur in a broad peak cen-
tered around 800 K. The oxygen in formaldehyde is solely from
the ceria lattice, and the same is true for water. A proposed mech-
anism for formaldehyde production is PACH3 bond breaking to
produce a surface CH3 species that binds to lattice oxygen on the
ceria surface. Subsequent CAH bond scission produces formalde-
hyde and releases surface hydrogen, which combines with lattice
oxygen to desorb water. No carbon remains on the surface at
900 K, and the only surface species observed is POx.
On reduced CeOx surfaces, the mechanism of DMMP reaction is
the same as on CeO2 although the product distribution is slightly
different. For example, more CAH bond breaking on CeOx gives rise
to CO rather than formaldehyde production at high temperatures.
Moreover, hydrogen leaves as H2 at the expense of H2O on the re-
duced, oxygen-deficient surface. The activity for DMMP decompo-
sition is initially higher on CeOx, but activity decreases more
rapidly after only a few cycles of adsorption and reaction.
as MMP, MP, methoxy and POx have been proposed. On ceria
[56], titania [46,57] alumina [32], MgO [32,46], La2O3 [32], Fe2O3
[32], WO3 [46], and Y2O3 [33] powders, vibrational spectroscopy
has shown that DMMP binds to the surface via the phosphoryl oxy-
gen. The frequency of the P@O stretch is red-shifted from the
1275 cmÀ1 gas phase value [9,30,31,33,35] and in some cases it is
proposed that adsorption occurs on the metal cation via the lone
pairs on the electron-rich phosphoryl oxygen [9,31,32,35,58]. Com-
putational studies of DMMP on alumina confirm surface adsorp-
tion via AlAO@P bond formation [59] and predict the 60–
85 cmÀ1 shift to lower energies observed in experiments [28]. Fur-
thermore, on titania [31,36], yttria [33], and magnesia [7,35], the
P@O group is converted to a bridging OAPAO species upon heat-
ing, as is observed on the ceria films in this work. This is demon-
strated by the loss of the P@O stretch at 1234–1214 cmÀ1 and
the appearance of PAO modes for POx species at 1170–
1090 cmÀ1, such as
ms(OAPAO) and ma(OAPAO) modes at 1085
and 1189 cmÀ1, respectively [7,30,31,33–36].
On most oxide surfaces studied, DMMP decomposition initially
occurs via step-wise scission of the PAOCH3 bonds, as illustrated
by the loss of intensity in the
m(CAO) mode in DMMP and the
appearance of (CAO) stretches associated with methoxy
m
[12,30,32,33,36,57]. The more stable PACH3 bonds remain intact
until higher temperatures ranging from 573 to 673 K [12,30,32].
The exception to these trends is the reducible Fe2O3 surface, on
which PACH3 and PAOCH3 bond scission are proposed to occur
simultaneously via a Mars van Krevelen mechanism in the 373–
473 K temperature range [32,60]. The Fe+2/Fe+3 redox couple par-
ticipates in oxidative PACH3 bond scission to form H3COÀ on the
surface, and gas phase oxygen is reduced by the surface to com-
plete the redox cycle [32]. Although the highly reducible ceria sur-
face would also be capable of forming a Ce+3/Ce+4 redox couple for
oxidative PACH3 bond scission, it is clear that on ceria, the PAOCH3
bond is broken at lower temperature than PACH3.
4.2. Comparison with reaction mechanisms on other surfaces
The mechanism for DMMP decomposition on the crystalline
ceria films is similar to what has been observed on a number of
other powdered metal oxide surfaces, where surface species such
The presence of surface hydroxyls is believed to play a key role
in these reaction mechanisms. Specifically, the mechanism for
PAOCH3 bond scission involves surface hydroxyl groups, which
typically adsorb on the metal oxides under ambient conditions
[12,30,32,33,36,57]. Although the PAOCH3 bond is thermodynam-
ically stronger than PACH3, methoxy is a better leaving group than
methyl in nucleophilic displacement reactions [30]. One of the first
mechanistic studies of DMMP decomposition of metal oxides was
the inelastic tunneling spectroscopy studies of DMMP on alumina
from the Weinberg group [30,61]. It was proposed that the MMP
intermediate is formed via attack on phosphorus by a nucleophilic
surface oxygen. Further decomposition of MMP occurs via attack at
the methoxy carbon by nucleophilic surface oxygen or hydroxyls to
induce CAO bond scission rather than continued PAOCH3 bond
scission [30]; this mechanism was based on the appearance of
the OAH stretch at 2440 cmÀ1 associated with hydroxy methyl-
30
POx
P(2p)
DMMP
900 K
25
20
15
10
5
800 K
600 K
400 K
MP
MMP
300 K
200 K
phosphonate, which receives
a proton from surface hydoxyl
groups. The hydroxy methyl phosphonate then decomposes to
methylphosphonate [30,61]. Work by the Mitchell group on pow-
dered metal oxide surfaces such as Al2O3, MgO, and La2O3 [32] also
reports that surface hydroxyls act as nucleophiles, displacing the
methoxy groups on the phosphorus atom in DMMP and resulting
in methanol production from combination of methoxy and surface
hydrogen [32]. Formation of PAO bonds with lattice oxygen make
the phosphorus atoms more electrophilic and susceptible to attack
by surface hydroxyls. On many metal oxide surfaces, DMMP is also
weakly adsorbed on oxide surfaces via H-bonding between surface
hydroxyls and the phosphoryl oxygen [12,30,57] although stronger
chemisorption occurs via adsorption of P@O to the Lewis acid sites
of the metal cation [30,57].
100 K
138
136
134
132
130
Binding Energy (eV)
Fig. 14. Soft X-ray photoelectron spectroscopy data for the P(2p) region for a 10 L
dose of DMMP adsorbed on reduced ceria (CeOx) at 100 K and then heated as
indicated. Spectra were collected with a 340 eV photon energy.
On surfaces that have been intentionally kept free of most
hydroxyls, by either heating or preparing the oxides in vacuum,

D.A. Chen et al. / Surface Science 604 (2010) 574–587
585
MMP
DMMP
CH3O
CH3
CH3
P
CH3
O
P
OCH3
CH3O
O
O
O
300 - 400 K
200 K
DMMP +
CH2O +
CH3OH, 420 K
CH2O +
CH3OH,
575 K
MP
CH3
P
MP
CH3
P
CH3
O
O
O
O
O
O
O
500 K
700 K
CH2O + H2O +
CO, 700-900 K
POx
900 K
Fig. 15. A proposed reaction scheme for DMMP decomposition on CeO2 showing the main surface species at various temperatures. Although the structures of all of the
intermediates have not been conclusively identified in this work, reasonable structures have been proposed based on literature studies of DMMP decomposition on other
oxide surfaces [12,30,31,36,61,68].
surface chemistry is still believed to be dictated by residual hydro-
xyl groups. Titania surfaces that have been intentionally dehydr-
oxylated by heating in vacuum [36] or in an oxygen environment
[57] exhibit isolated hydroxyl groups remaining on the surface;
these surface hydroxyls undergo hydrogen bonding to physisorb
DMMP via the phosphoryl oxygen and also hydrolyze DMMP to
form surface methoxy [36,57]. For yttria powders grown in vac-
uum by laser-heated gas phase condensation, the OAH stretch of
isolated surface hydroxyls at 3705 and 3676 cmÀ1 in the infrared
spectrum signifies trace hydroxyls on remaining on the surface
[33]. Hydrogen bonding interactions between DMMP and surface
hydroxyls broadens the associated OAH peak and shifts it to
3200 cmÀ1 [33]. Furthermore, nucleophilic attack by hydroxyls
on DMMP resulted in PAOCH3 bond scission and formation of sur-
On metal surfaces, the decomposition of DMMP is also consis-
tent with a reaction mechanism of PAOCH3 bond breaking to form
surface methoxy. Thus, the chemistry of DMMP resembles that of
methoxy on both metals and metal oxide surfaces. For example,
on Ni(1 1 1) [1], Pt(1 1 1) [5], Ni [43] and Pt [15] clusters supported
on TiO2, and Pd(1 1 1) [1], DMMP reaction produces CO and H2 as
the main products, and these are also formed from methoxy
decomposition on the respective single-crystal surfaces [62–65].
Likewise, formaldehyde is formed from DMMP reaction on TiO2-
supported Cu clusters when methoxy on Cu is known to produce
formaldehyde. On Rh(1 0 0) [45], the H2, H2O, CO, CO2, and CH3OH
production from DMMP decomposition is also similar to methanol
reaction on Rh(1 0 0), which forms H2, CO and CH3OH as the main
gaseous species [66].
face methoxy groups, which were identified by the
m
(CO) at
On many oxide surfaces, a POx species remains on the surface
after decomposition, and this species generally poisons the surface
to further DMMP decomposition [9,11–13]. The stable phospho-
nate is believed to be responsible for passivating the surface to-
ward DMMP adsorption and reaction [12]. However, the ceria
surface is able to sustain activity after seven cycles of DMMP
adsorption and desorption, unlike on metal surfaces and metal
clusters [1,4,15]. This suggests that ceria is more resistant than me-
tal surfaces to poisoning by phosphorous-containing decomposi-
tion byproducts. Similar behavior has been observed for DMMP
adsorption and reaction on TiO2(1 1 0) surfaces, on which activity
is not completely eliminated after five TPD cycles [15]. The ceria
surface appears to have greater activity than TiO2 because DMMP
decomposition on CeO2(1 1 1) does not require oxygen vacancies
whereas decomposition on TiO2(1 1 0) is believed to largely occur
at oxygen vacancy defects [43,44]. Unlike TiO2(1 1 0), the ceria sur-
face could not be reoxidized to regain activity for DMMP decompo-
sition after multiple cycles of DMMP adsorption and reaction.
When the ceria surface was oxidized at 600 and 900 K and in O2
(30 L) following seven cycles DMMP adsorption-reaction cycles,
the Ce+4 remained at ꢀ40%, and DMMP TPD experiments on this
1065 cmÀ1 [33].
On the ceria films in our studies, there is no clear evidence of
hydroxyls on the surface. The O(1s) region of freshly prepared ceria
films did not exhibit signal at ꢀ532 eV, which is the binding energy
attributed to surface hydroxyls [18]. Moreover, the
m(OAH) at
3600 cmÀ1 was not observed in the infrared spectrum of the ceria
films although this vibrational mode would have reduced intensity
for OAH bonds parallel to the surface due to the surface dipole
selection rule. Furthermore, surface hydroxyls would be expected
to desorb as water upon heating the surface, but no water desorp-
tion was observed from the clean ceria films in TPD experiments.
Thus, it is unlikely that the presence of surface hydroxyls is the
driving force behind DMMP chemistry on the ceria films even
though it is impossible to rule out the presence of a small fraction
of isolated OH species. The mechanism for DMMP decomposition
on the ceria thin films is also believed to be step-wise scission of
the PAOCH3 bonds followed by PACH3 bond breaking despite the
lack of surface hydroxyls. The infrared spectroscopy experiments
clearly show that the PACH3 bond remains on the surface until
700 K while the PAOCH3 bonds are broken by 500 K.

586
D.A. Chen et al. / Surface Science 604 (2010) 574–587
surface showed no activity for DMMP decomposition beyond what
was observed before oxidation. Apparently it is more difficult to re-
move phosphorus by oxidation from the ceria surface than from a
titania surface.
methoxys and decreased in yield with increasing DMMP exposure.
Interestingly, no dimethyl ether was formed in the reaction of
methanol itself, which also decomposes on ceria via a methoxy
intermediate. Diffuse reflectance infrared spectroscopy studies
showed that DMMP decomposition at room temperature produced
surface methoxy, and two different binding sites were identified
based on the CAO stretch for methoxy: the on top site at
1098 cmÀ1 and 2-fold site at 1047 cmÀ1 [56,67]. However, it should
4.3. Comparison with methanol reaction on ceria: common methoxy
intermediate
The similarities in the chemistry of methanol and DMMP on cer-
ia thin films provides evidence that DMMP decomposition occurs
via a methoxy intermediate. Methoxy was identified by high reso-
lution XPS studies as the only surface intermediate in methanol
reaction on CeO2 films grown on Ru(0 0 0 1) [19]. For both DMMP
and methanol reaction on the ceria thin films, formaldehyde and
methanol desorb at ꢀ560–575 K as the major products, and meth-
oxy was proposed to disproportionate to form methanol and form-
aldehyde. Moreover, multiple adsorption-reaction cycles for
DMMP or methanol resulted in reduction of the ceria. For reaction
on Ce18O2, both 18O- and 16O-formaldehyde were observed at 560–
575 K.
In the case of methanol reaction, it is proposed that the meth-
oxy intermediate can undergo methyl hopping so that only some
of the formaldehyde contains lattice oxygen. This may also occur
when methoxy is formed from DMMP reaction although it is pos-
sible that lattice oxygen incorporation occurs due to bidentate
adsorption of methoxy to form a dioxymethylene intermediate
via coordination with lattice oxygen. On crystalline ceria thin films
grown in UHV on Cu(1 1 1), methanol reaction on the ceria film
produced methoxy at 550 K, which was identified by infrared spec-
troscopy and X-ray photoelectron spectroscopy [41]; formalde-
hyde was produced from methanol reaction, but CO and H2 were
observed as the primary products, unlike on the CeO2 films on
Ru(0 0 0 1). However, since CO and H2 are major products formed
from methanol reaction on reduced ceria films [19], the difference
in chemistry may be related to the oxidation state of the ceria films
on Cu(1 1 1).
be noted that these two modes overlap with the ms(OAPAO) stretch
expected around 1058–1085 cmÀ1 [30,31,33]. The lower frequency
for CAO stretch for the bridging species compared to the value on
pure ceria was explained by methoxy at a bridge site between cer-
ium and aluminum ions [56].
5. Summary
DMMP readily decomposes on crystalline ceria thin films upon
heating the surface. The main gaseous products are formaldehyde,
methanol, water and CO, with POx remaining on the surface after
heating to 900 K. The ceria films are capable of sustaining decom-
position activity after multiple cycles of adsorption and reaction,
but the activity significantly diminishes with each cycle. This loss
of activity is attributed to passivation of active sites by POx and also
corresponds to reduction of Ce+4 to Ce+3; even after reoxidation of
the ceria, the activity is not restored, indicating that DMMP decom-
position on ceria is not catalytic. At 100 K, DMMP initially adsorbs
to the surface via the phosphoryl oxygen, but the P@O bond con-
verts to a bridging OAPAO species at 200 K. Upon further heating,
DMMP decomposes via PAOCH3 bond scission, forming MMP and
MP surface intermediates as well as surface methoxy. The more
stable PACH3 bond is not cleaved until temperatures above
700 K. DMMP reaction on reduced ceria films is similar to that on
the stoichiometric films although more H2 and less high tempera-
ture formaldehyde are produced on the reduced surface.
Acknowledgements
The main difference between methanol and DMMP reaction is
the presence of surface hydroxyls from OAH bond scission in
methanol. These hydroxyls give rise to water desorption at 200 K
from methanol reaction, whereas water desorption in DMMP reac-
tion is not observed below 800 K because CAH bonds are not bro-
ken below this temperature. Furthermore, formaldehyde
production from methanol reaction is not observed above 800 K;
this is in agreement with the high temperature formaldehyde orig-
inating from PACH3 bond scission during DMMP decomposition to
form a transient methoxy intermediate with lattice oxygen.
Methanol [19] and DMMP reactions are also similar on the re-
duced ceria films. Both reactions produce CO and H2, in addition
to the formaldehyde and water observed on the stoichiometric sur-
face. For methanol reaction, the H2 yield increases while the H2O
yield decreases for more highly reduced surfaces, as expected
due to deficiency of oxygen in the reduced films. Furthermore,
the temperature of formaldehyde and methanol desorption shifts
from 560 to 575 K to ꢀ630 K for both DMMP and methanol reac-
tion on the reduced surface. It is proposed that the greater number
of oxygen vacancies on the reduced surface stabilize methoxy and
OH since it becomes harder to remove lattice oxygen as water [19].
An earlier study of DMMP decomposition on powdered ceria
further supports methoxy as a surface intermediate although the
product distribution is not identical to that on ceria thin films.
Work from the Mitchell group involving DMMP decomposition
on ceria supported on alumina reports that methanol and dimethyl
ether are the primary products formed via a methoxy intermediate
[56]. In these studies, it was established that most of the observed
activity is from reaction on ceria rather than alumina. The dimethyl
ether was proposed to be formed from combination of two surface
We gratefully acknowledge financial support from the US Army
Research Office (W911NF-05-1-0184). D.R.M., W.O.G. and S.D.S. are
supported by the Division of Chemical Sciences, Geosciences, and
Biosciences, Office of Basic Energy Sciences, US Department of En-
ergy, under Contract DE-AC05-00OR22725 with Oak Ridge Na-
tional Laboratory, managed and operated by UT-Battelle, LLC. Use
of the National Synchrotron Light Source, Brookhaven National
Laboratory, was supported by the US Department of Energy, Office
of Science, Office of Basic Energy Sciences, under Contract No. DE-
AC02-98CH10886.
References
[1] X. Guo, J. Yoshinobu, J.T. Yates, J. Phys. Chem. 94 (1990) 6839.
[2] T.Z. Tzou, S.W. Weller, J. Catal. 146 (1994) 370.
[3] J.G. Ekerdt, K.J. Klabunde, J.R. Shapley, J.M. White, J.T. Yates, J. Phys. Chem. 92
(1988) 6182.
[4] V.S. Smentkowski, P. Hagans, J.T. Yates, J. Phys. Chem. 92 (1988) 6351.
[5] M.A. Henderson, J.M. White, J. Am. Chem. Soc. 110 (1988) 6939.
[6] Y.X. Li, O. Koper, M. Atteya, K.J. Klabunde, Chem. Mat. 4 (1992) 323.
[7] Y.X. Li, K.J. Klabunde, Langmuir 7 (1991) 1388.
[8] G.W. Wagner, P.W. Bartram, O. Koper, K.J. Klabunde, J. Phys. Chem. B 103
(1999) 3225.
[9] S.R. Segal, L.X. Cao, S.L. Suib, X. Tang, S. Satyapal, J. Catal. 198 (2001) 66.
[10] O. Koper, E. Lucas, K.J. Klabunde, J. Appl. Toxicol. 19 (1999) S59.
[11] M.A. Henderson, T. Jin, J.M. White, J. Phys. Chem. 90 (1986) 4607.
[12] C.S. Kim, R.J. Lad, C.P. Tripp, Sensors Actuat. 76 (2001) 442.
[13] L. Cao, S.R. Segal, S.L. Suib, X. Tang, S. Satyapal, J. Catal. 194 (2000) 61.
[14] M.B. Mitchell, V.N. Sheinker, A.B. Tesfamichael, E.N. Gatimu, M. Nunley, J. Phys.
Chem. B 107 (2003) 580.
[15] J.S. Ratliff, S.A. Tenney, X. Hu, S.F. Conner, S. Ma, D.A. Chen, Langmuir 25 (2009)
216.
[16] A. Trovarelli, Catal. Rev.– Sci. Eng. 38 (1996) 439.

D.A. Chen et al. / Surface Science 604 (2010) 574–587
587
[17] J. Zhou, A. Baddorf, D.R. Mullins, S.H. Overbury, J. Phys. Chem. C 112 (2008)
9336.
[18] J. Zhou, D.R. Mullins, Surf. Sci. 600 (2006) 1540.
[42] P. Hernandez-Fernandez, S. Rojas, P. Ocon, J.L.G. d la Fuente, J.S. Fabian, J.
Sanza, M.A. Pena, F.J. Garcia-Garcia, P. Terreros, J.L.G. Fierro, J. Phys. Chem. C
111 (2007) 2913.
[19] D.R. Mullins, M.D. Robbins, J. Zhou, Surf. Sci. 600 (2006) 1547.
[20] D.R. Mullins, P.V. Radulovic, S.H. Overbury, Surf. Sci. 429 (1999) 186.
[21] S.H. Overbury, D.R. Mullins, D.R. Huntley, L. Kundakovic, J. Catal. 186 (1999)
296.
[22] J.L. Lu, H.J. Gao, S. Shaikhutdinov, H.J. Freund, Surf. Sci. 600 (2006) 5004.
[23] D.R. Mullins, K.Z. Zhang, Surf. Sci. 513 (2002) 163.
[24] D.R. Mullins, S.H. Overbury, D.R. Huntley, Surf. Sci. 409 (1998) 307.
[25] S.D. Senanayake, W.O. Gordon, S.H. Overbury, D.R. Mullins, J. Phys. Chem. C
113 (2009) 6208.
[26] L. Kundakovic, D.R. Mullins, S.H. Overbury, Surf. Sci. 457 (2000) 51.
[27] In identical experiments, the peak temperature for formaldehyde and
methanol production shifted between 560 and 575 K for reasons that are
not well understood; these same temperature shifts were also observed for
formaldehyde and methanol production from methanol reaction on CeO2 films
[19].
[43] J. Zhou, S. Ma, Y.C. Kang, D.A. Chen, J. Phys. Chem. B 108 (2004) 11633.
[44] S. Ma, J. Zhou, Y.C. Kang, J.E. Reddic, D.A. Chen, Langmuir 20 (2004) 9686.
[45] R.I. Hegde, C.M. Greenlief, J.M. White, J. Phys. Chem. 89 (1985) 2886.
[46] B. Aurian-Blajeni, M.M. Boucher, Langmuir 5 (1989) 170.
[47] S.M. Kanan, Z. Lu, C.P. Tripp, J. Phys. Chem. B 106 (2002) 9576.
[48] S.D. Senanayake, D.R. Mullins, J. Phys. Chem. C 112 (2008) 9744.
[49] J. Zhou, K. Varazo, J.E. Reddic, M.L. Myrick, D.A. Chen, Anal. Chim. Acta 496
(2003) 289.
[50] B. Glorieux, R. Berjoan, M. Matecki, A. Kammouni, D. Perarnau, Appl. Surf. Sci.
253 (2007) 3349.
[51] M.L. Granados, F.C. Galisteo, P.S. Lambrou, R. Manscal, J. Sanz, I. Sobrados, J.L.G.
Fierro, A.M. Efstathiou, J. Catal. 239 (2006) 410.
[52] C. Larese, M.L. Granados, R. Mariscal, J.L.G. Fierro, P.S. Lambrou, A.M.
Efstathiou, Appl. Catal. B 59 (2005) 13.
[53] C. Larese, F.C. Galisteo, M.L. Granados, R. Mariscal, J.L.G. Fierro, P.S. Lambrou,
A.M. Efstathiou, J. Catal. 226 (2004) 443.
[54] C. Larese, F.C. Galisteo, M.L. Granados, R. Mariscal, J.L.G. Fierro, M. Furio, R.F.
Ruiz, Appl. Catal. B 40 (2003) 305.
[55] D.R. Mullins, T.S. McDonald, Surf. Sci. 601 (2007) 4931.
[56] M.B. Mitchell, V.N. Sheinker, W.W. Cox, E.N. Gatimu, A.B. Tesfamichael, J. Phys.
Chem. B 108 (2004) 1634.
[28] J. Lamotte, V. Moravek, M. Bensitel, J.C. Lavalley, React. Kinet. Catal. Lett. 36
(1988) 113.
[29] L. Bertilsson, I. Engquist, B. Liedberg, J. Phys. Chem. B 101 (1997) 6021.
[30] M.K. Templeton, W.H. Weinberg, J. Am. Chem. Soc. 107 (1985) 97.
[31] C.N. Rusu, J.T. Yates, J. Phys. Chem. B 104 (2000) 12292.
[32] M.B. Mitchell, V.N. Sheinker, E.A. Mitnz, J. Phys. Chem. B 101 (1997) 11192.
[33] W.O. Gordon, B.M. Tissue, J.R. Morris, J. Phys. Chem. C 111 (2007) 3233.
[34] A.E.T. Kuiper, J. Vanbokhoven, J. Medema, J. Catal. 43 (1976) 154.
[35] Y.X. Li, J.R. Schlup, K.J. Klabunde, Langmuir 7 (1991) 1394.
[36] D.A. Panayotov, J.R. Morris, Langmuir 25 (2009) 3652.
[37] T. Masui, H. Hirai, N. Imanaka, G. Adachi, Phys. Status Solidi A – Appl. Res. 198
(2003) 364.
[57] S. Gan, Y. Liang, D.R. Baer, A.W. Grant, Surf. Sci. 475 (2001) 159.
[58] S.R. Segal, S.L. Suib, X. Tang, S. Satyapal, Chem. Mater. 11 (1999) 1687.
[59] V.M. Bermudez, J. Phys. Chem. C 111 (2007) 3719.
[60] T.M. Tesfai, V.N. Sheinker, M.B. Mitchell, J. Phys. Chem. B 102 (1998) 7299.
[61] M.K. Templeton, W.H. Weinberg, J. Am. Chem. Soc. 107 (1985) 7745.
[62] B.A. Sexton, Surf. Sci. 102 (1981) 271.
[38] L.J. Bellamy, The Infrared Spectra of Complex Molecules, Chapman and Hall,
Thetford, UK, 1975.
[63] J.J. Chen, Z.C. Jiang, Y. Zhou, B.R. Chakraborty, N. Winograd, Surf. Sci. 328
(1995) 248.
[39] L. Xu, G. Guo, D. Uy, A.E. O’Neill, W.H. Weber, M.J. Rokosz, R.W. McCabe, Appl.
Catal. B 50 (2004) 113.
[40] S. Freni, S. Cavallaro, N. Mondello, L. Spadaro, F. Frusteri, Catal. Commun. 4
(2003) 259.
[64] S.M. Gates, J.N. Russell, J.T. Yates, Surf. Sci. 146 (1984) 199.
[65] S.M. Gates, J.N.J. Russell, J.T.J. Yates, J. Catal. 95 (1985) 23.
[66] J.E. Parmeter, X. Jiang, D.W. Goodman, J. Vac. Sci. Technol. A 9 (1991) 1810.
[67] A. Badri, C. Binet, J.C. Lavalley, J. Chem, Soc. – Faraday Trans. 93 (1997) 1159.
[68] D.A. Panayotov, J.R. Morris, J. Phys. Chem. C 112 (2008) 7496.
[41] A. Siokou, R.M. Nix, J. Phys. Chem. B 103 (1999) 6984.