Ethylene Glycol Adsorption and Reaction over CeOX(111) Thin Films

ARTICLE pubs.acs.org/JPCC

Ethylene Glycol Adsorption and Reaction over CeOX(111) Thin Films T.-L. Chen and D. R. Mullins* Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6201, United States ABSTRACT: This study reports the interaction of ethylene glycol with well-ordered CeOX(111) thin film surfaces. Ethylene glycol initially adsorbs on fully oxidized CeO2(111) and reduced CeO2
X(111) through the formation of one C
O
Ce bond and then forms a second alkoxy bond after annealing. On fully oxidized CeO2(111) both recombination of ethylene glycol and water desorption occur at low temperature leaving stable
OCH2CH2O
(ethylenedioxy) intermediates and oxygen vacancies on the surface. This ethylenedioxy intermediate goes through C
C bond scission to produce formate species which then react to produce CO and CO2. The formation of water results in the reduction of the ceria. On a reduced CeO2
X(111) surface the reaction selectivity shifts toward a dehydration process. The ethylenedioxy intermediate decomposes by breaking a C
O bond and converts into an enolate species. Similar to the reaction of acetaldehyde on reduced CeO2
X(111), the enolate reacts to produce acetaldehyde, acetylene, and ethylene. The loss of O from ethylene glycol leads to a small amount of oxidation of the reduced ceria.

1. INTRODUCTION Cerium oxide is known for its ability to store, release, and transport oxygen ions. The unique ability for releasing or absorbing oxygen has made ceria widely used as an oxygen storage material in automotive catalytic converters. Ceria-supported or doped catalysts are also used extensively in applications including oxidative dehydrogenation processes and steam reforming of alcohols such as methanol and ethanol.1
6 Because of the participation of surface oxygen and oxygen vacancies in catalytic reactions on ceria, surface chemistry studies on initially oxidized or reduced surfaces are good models for understanding reactions under oxidizing or reducing atmospheres, respectively. The adsorption and reaction of organic oxygenates such as alcohols,7
11 aldehydes,12
14 ketones,15 and carboxylic acids16,17 have been studied on CeO2(111)/CeO2
X(111) surfaces. These investigations have shown that the ceria stoichiometry, CeO1.5(111)
CeO2(111), dramatically influences the surface reaction. Primary alcohols, such as 1-propanol and ethanol, were oxidized by CeO2(111) forming aldehydes through a dehydrogenation process, while on a reduced surface a dehydration process was more favorable leading to alkene products.8 Carbonyl functionalized oxygenates, such as acetaldehyde14 and acetone,15 have demonstrated little reactivity on fully oxidized ceria while both deoxygenation and dehydration occurred on reduced CeO2
X(111). C
O bond scission was never evident in the C1 oxygenates and C
C bond cleavage did not occur in the longer-chain alcohols and aldehydes. All of the alcohols were shown to bond to the surface through the dissociation of the O
H forming a surface-bound alkoxy,
CHX
O
Ce.7,8 In this work, we use ethylene glycol as a model to probe how ceria will interact with more than one alcohol group on a single molecule. As the simplest diol, ethylene glycol has two symmetric hydroxyl groups which are available to interact with the surface. Two of the key questions in this study are whether only one hydroxyl reacts with the surface r 2011 American Chemical Society

or whether both hydroxyl groups react forming ethylenedioxy, Ce
OCH2CH2O
Ce, and whether ethylene glycol’s reactivity differs significantly from the analogous monoalcoholic molecule, ethanol. The adsorption and reaction of ethylene glycol have previously been studied over several metal surfaces including Cu, Ag, Ni, Rh, Mo, Pd, and Pt, but we have not found any studies on well-defined single crystal metal oxide surfaces.18
26 On the metals various intermediate species and reaction products have been reported. On Cu(110), both alcohol end groups dehydrogenated and yielded glyoxal, OCHCHO, as the product.18 The intermediate was suggested to be bound with one hydroxyl dissociated and the other still intact but interacting with surface. Ag(110), however, did not react with ethylene glycol unless oxygen was preabsorbed.19 The ethylene glycol reacted with surface oxygen to form ethylenedioxy and H2O below 170 K. The ethylenedioxy intermediate further evolved into glyoxal and hydrogen at higher temperature. When there was excess O on the Ag(110) the C
C bond cleaved in the ethylenedioxy yielding water, formaldehyde, and surface formate.19 On Mo(110), intermediates bound through either one or both oxygens were observed on the surface, and it was suggested that the species bound at only one end reacted through a short-lived intermediate bound at both ends leading to C
O bond scission and ethylene desorption.20 CO desorption, also the result of C
C bond cleavage, was seen when ethylene glycol reacted with Rh(111),21 Rh(100),22 Pd(111),23 Ni(111),24,25 and Pt(111).24,25 Ethylene glycol decomposed to CO and H2 via an ethylenedioxy species on both the Rh(111) and Rh(100) surfaces.21,22 The intermediate could either begin the decomposition via C
C or C
H bond scission. Received: February 18, 2011 Revised: May 2, 2011 Published: June 27, 2011 13725

dx.doi.org/10.1021/jp201639n | J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C On the Ni(111) surface the initial decomposition proceeded through O
H bond cleavage, forming the ethylenedioxy intermediate. This intermediate reacted through further dehydrogenation and C
C bond scission to form CO on the surface.24,25 The same intermediate (ethylenedioxy) was also seen in the reaction on the Pd(111) surface.23 It converted, however, into glyoxal which either desorbed molecularly or decomposed through C
C bond scission, dehydration, and decarbonylation to form CO and hydrogen. On Pt(111), ethylene glycol primarily desorbed reversibly but a small amount of CO and H2 was also observed.24,25 In the present work, we will show that initially ethylene glycol partially deprotonates on the ceria surface producing a mixture of species bound by either one or both O atoms. As the temperature increases water desorbs and all of the intermediates are bound through both O atoms. This ethylenedioxy then proceeds along different reaction pathways depending on whether the ceria is oxidized or reduced. CO and CO2 are the principal products on oxidized CeO2(111). These result from breaking the C
C bond of the ethylenedioxy producing a formate intermediate during decomposition. On reduced ceria, C
O bond cleavage creates an enolate species which later produces dehydration products such as acetaldehyde, ethylene, and acetylene.

2. EXPERIMENTAL SECTION Experiments were performed in two separate UHV chambers. Soft X-ray photoelectron spectroscopy (sXPS) and near edge X-ray absorption fine structure (NEXAFS) were conducted using synchrotron radiation on beamline U12a at the National Synchrotron Light Source (NSLS). CeO2(111) films were grown in situ by Ce vapor deposition onto a Ru(0001) surface at 700 K under an oxygen atmosphere. The procedure has been reported previously.27 Reduced CeO2
X(111) films were produced by exposing a CeO2(111) surface to the equivalent of ca. 100 L of methanol at 700 K.7 A directed gas doser was used to minimize the total pressure rise in the vacuum chamber.28 This method resulted in highly reproducible reduced oxidation states. The Ce oxidation state was determined using XPS of the Ce 3d and Ce 4d regions at ORNL and Beamline U12a, respectively. We estimate that the reduced surface is 65((5)% Ce3+ (CeO1.67(111)). Ethylene glycol (99.8%) was obtained from Sigma Aldrich. At ORNL MgSO4 was added to the glycol to remove any latent water. On both systems the diol was pumped at room temperature for 24 h before use to minimize the water contamination. The manifold system was gently heated to prevent condensation during the gas transport. At beamline U12a the adsorbate was introduced to the UHV chamber by way of a variable leak valve using a “backfilling” method. At ORNL the adsorbate exposure was achieved using a effusive gas doser28 to minimize the rise in the background pressure and desorption from other parts of the sample holder during TPD. Because of its low vapor pressure and tendency to stick to the dosing tube the ethylene glycol exposure was difficult to control reproducibly. The doses were always sufficiently large so that the multilayer peak at 205 K was observed. C 1s and O 1s spectra were recorded at photon energies of 400 and 600 eV, respectively. The instrumental resolution was ca. 0.5 eV. NEXAFS was performed at the C k-edge using a partial yield electron detector. 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.29 The high-pass retarding grid was set at
230 V.

ARTICLE

Figure 1. TPD of ethylene glycol (Mass 31) following the adsorption of ethylene glycol at 185 K from fully oxidized CeO2(111) after the first exposure (solid black line), the third exposure (gray line), and from 65% reduced CeO1.67(111) (dotted black line).

Higher order X-ray excitation created 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. The temperature programmed desorption (TPD) experiments were performed in a chamber at ORNL. The temperature was ramped at 2 K/s, and the sample was biased at
70 V to prevent electrons generated by the mass spectrometer ionizer from stimulating reactions at the surface. A Hiden HAL/3F 301 mass spectrometer was used in these experiments. The TPDs occurred in a “line-of-sight” geometry with the sample face ca. 2 cm from the mass spectrometer aperture. Sequential TPD cycles were conducted to determine whether the ethylene glycol reactions modified the surface chemistry through reduction or oxidation of the surface. Three cycles were conducted on the fully oxidized surface, then the surface was reduced in methanol and then three cycles were conducted on the reduced surface.

3. RESULTS 3.1. Ethylene Glycol on Fully Oxidized CeO2(111). Ethylene glycol was adsorbed on CeO2
X(111) at 185 K as described above. The desorption products during TPD from the first, third and fourth cycles are shown in Figures 1
3. The desorption from the fully oxidized surface is indicated by the solid black lines in each figure. Ethylene glycol desorption was monitored at Mass 31 which is its most intense cracking fragment (Figure 1).30 Identical TPD spectra were seen at Mass 33 which is another major fragment of ethylene glycol. The physisorbed ethylene glycol desorbs near 205 K.20
22 More strongly absorbed ethylene glycol desorbs between 240 and 500 K. It is significant that there is no desorption above 500 K under any conditions at Mass 31. Mass 31 is a major fragment for possible products such as ethanol 13726

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C

ARTICLE

Figure 3. TPD of H2 and H2O following the adsorption of ethylene glycol adsrobed at 185 K from fully oxidized CeO2(111) after the first exposure (solid black line), the third exposure (gray line), and from 65% reduced CeO1.67(111) (dotted black line).

Figure 2. TPD spectra following the adsorption of ethylene glycol adsorbed at 185 K from fully oxidized CeO2(111) after the first exposure (solid black line), the third exposure (gray line), and from 65% reduced CeO1.67(111) (dotted black line).

and glyoxal therefore the formation of these products can be ruled out. C-containing decomposition products are shown in Figure 2. All of the desorption below 400 K can be attributed to ethylene glycol desorption. The ethylene glycol desorption dominated the spectra below 300 K and could not be reliably subtracted from the masses shown in Figure 2. For clarity the data is only shown above 250 K. The CO desorption at Mass 28 and the acetylene desorption at Mass 26 have been isolated by subtracting the ethylene contribution based on the ethylene desorption at Mass 27. Similarly the CO2 desorption at Mass 44 was isolated by subtracting the acetaldehyde contribution based on the desorption at Mass 29. The spectra in Figure 2 have been scaled using a method suggested by Ko et al.31 in order to produce a more accurate depiction of the relative intensities of the products. These scale factors are based on the fragmentation and ionization probabilities of the molecules and the mass-dependent response of the mass spectrometer. On the fully oxidized surface (black lines) CO is the most intense product. There are three distinct CO desorption states at 500, 600, and 650 K. CO2 is also observed at 500, 590, and 700 K. In addition to CO and CO2 the most intense C-containing desorption product was acetaldehyde monitored at Mass 29. Weaker fragments at Masses 15, 42, and 43 qualitatively match the Mass 29 desorption and the relative intensities match those seen in the mass spectrum for acetaldehyde in our system.

The acetaldehyde desorbs in a broad doublet at 550 and 600 K with a small feature at 650 K. There was no desorption at Mass 30 which precludes formaldehyde formation. The only other C-containing products were small amount of acetylene and ethylene which desorb at 650 K. The formation of CO and CO2 liberates hydrogen from the ethylene glycol while acetaldehyde and acetylene will liberate hydrogen and oxygen. These elements should produce H2 and H2O. Note that the ceria substrate may be involved in any reaction mechanism so that lattice O may also be incorporated in H2O formation. The H2 and H2O desorption is shown in Figure 3. Water desorption starts at a low temperature. Much of the intense initial peak at 205 K is associated with ethylene glycol desorption however the broad desorption between 250 and 400 K is more intense relative to the 205 K feature than the corresponding region for ethylene glycol (Figure 1). Therefore the desorption at Mass 18 between 250 and 400 K is attributed primarily to water. Water also desorbs in a series of high temperature states between 470 and 700 K. This is the same region where CO, CO2, and acetaldehyde are observed. Only a small amount of H2 was observed near 650 K from the oxidized surface. C 1s sXPS spectra were recorded after fully oxidized CeO2(111) was exposed to 20 L (1 L = 10
6 Torr s) of ethylene glycol at 190 K to ensure a full coverage and then annealed to higher temperatures. The resulting spectra after each annealing step are shown in Figure 4. Upon heating to 250 K most of the physisorbed species have desorbed (Figure 1), and a single peak is observed at 287.1 eV, which is assigned to C in an alkoxy moiety.7 This feature could be associated with either
C
OH or
C
O
Ce. After annealing to 350 K, this peak shifts to 286.8 eV and becomes narrower with a loss of intensity. The loss of intensity is consistent with the desorption of ethylene glycol that occurs between 250 and 350 K (Figure 1). There is little change in the C 1s spectra until 550 K when the peak at 286.8 eV decreases in intensity and a new peak appears at 290 eV. 13727

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C

ARTICLE

Figure 4. C 1s sXPS spectra from ethylene glycol adsorbed on CeO2(111) at 190 K and then annealed as indicated.

Figure 5. O 1s sXPS spectra from ethylene glycol adsorbed on CeO2(111) at 190 K and then annealed as indicated.

The 290 eV peak is assigned to a carboxylate species (COO).17 In some cases the formation of carboxylate species has been associated with X-ray damage in adsorbed oxygenates although this appears to be a bigger problem with carbonyls than alkoxys.7,15 We tested for X-ray damage by comparing spots on the sample that had been exposed to the X-ray beam for an extended period to areas that had only limited X-ray exposure. We found no evidence of X-ray induced chemistry in this system. The corresponding O 1s spectra are shown in Figure 5. The peak at 530.4 eV is due to the lattice O in the ceria substrate. This peak loses intensity after the adsorption of ethylene glycol. This is due to attenuation by the absorbate and presumably conversion of some of the lattice O into Ce
OH.7 At 250 K two additional peaks are apparent at 531.8 and 533.6 eV. The feature at 531.8 is associated with both Ce
O
C
and Ce
OH groups.7,32 Previous studies have indicated that
OH in molecular water and methanol have O 1s binding energies between 533 and 534 eV.7,32 Therefore the feature at 533.6 eV eV is assigned to a
C
OH group. When the sample is annealed to 350 K there is an increase in the intensity of the lattice oxygen peak. This is consistent with desorption of ethylene glycol and water (Figures 1 and 3). There is also a decrease in the intensity of the
C
OH feature at 533.6 eV but little change in the Ce
O
C-/Ce
OH feature at 531.8 eV. This indicates that
C
OH groups were converted into Ce
O
C-/Ce
OH. The new Ce
O
C-/Ce
OH groups are balanced by the loss of these groups through ethylene glycol and water desorption. Note that there was also a 0.3 eV shift in the C 1s feature between 250 and 350 K (Figure 4). This may be associated with a C 1s binding energy difference between
C
OH and
C
O
Ce. When the sample was annealed to 450 K the intensity of the O 1s feature at 531.8 eV increases while the feature at 533.6 eV disappears. Since there is no significant desorption in this temperature range these changes in the O 1s spectra are associated with the total conversion of
C
OH to
C
O
Ce. After annealing to 550 K the intensity of the adsorbate related feature decreases in intensity and shifts to 532.5 eV with a new shoulder appearing on the higher binding energy side. The shift

to higher binding energy and the high energy shoulder are consistent with the conversion of some of the Ce
O
C
groups into carboxylate species on the surface. Formate has an O 1s peak near 533 eV at 500 K on CeO2(111).17 3.2. Ethylene Glycol on Reduced CeO2
X Surface. The CeO2(111) surface was reduced to CeO1.67(111) by annealing in methanol at 700 K. Again ethylene glycol was adsorbed at 185 K, and the resulting TPD spectra are indicated by the dotted lines in Figures 1
3. The ethylene glycol exposure appears to have been lower on the reduced surface compared to the oxidized surface, but ethylene glycol multilayer desorption is evident from 200 to 220 K in Figure 1. The ethylene glycol desorption that occurred from 240 to 500 K on the oxidized surface (black line) collapses into a shoulder on the multilayer peak and decays by 280 K (dotted line). Significant amounts of acetaldehyde, ethylene, and acetylene are produced from the reduced surface (Figure 2). The acetaldehyde and acetylene have relatively sharp desorption peaks at 600 and 630 K, respectively. The ethylene desorbs in two peaks at 600 and 630 K. The CO desorption is reduced significantly compared to the oxidized surface. The low temperature peak near 500 K completely disappears, and the asymmetric feature between 560 and 660 K is greatly attenuated although essentially in the same temperature region as on the oxidized surface. CO2 desorption is completely eliminated. The water desorption between 500 and 700 K virtually disappeared on the CeO1.67(111) surface (Figure 3, dotted line). However water is still evident below 400 K with features at 280 and 380 K. The high-temperature water desorption is replaced by H2 desorption between 580 and 700 K. Figure 6 shows the results of C 1s sXPS spectra of ethylene glycol on reduced CeO1.67(111). The sample was exposed to a 20 L dose of ethylene glycol at 190 K and then heated to sequentially higher temperatures. After annealing to 250 K the prominent peak at 287.8 eV again represents a mixture of
C
OH and
C
O
Ce. As the sample was annealed to 450 K, this peak shifts to 287.5 eV and becomes narrower; however, the 13728

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C

ARTICLE

Figure 6. C 1s sXPS spectra from ethylene glycol adsorbed on CeO1.67(111) at 190 K and then annealed as indicated.

Figure 7. O 1s sXPS spectra from ethylene glycol adsorbed on CeO1.67(111) at 190 K and then annealed as indicated.

integrated intensity is unchanged. The peak position for the alkoxy is 0.7 eV higher on the reduced surface compared to the oxidized surface. This is consistent with what was observed for various alcohols on oxidized and reduced CeO2
X(111).7 Upon annealing to 550 K the intensity of the alkoxy peak decreases and a new feature appears at 285.8 eV. The feature at 285.8 eV is assigned to an alkyl moiety,
CHX.14,15 Upon annealing to 650 K, virtually all of the C 1s intensity has disappeared. No carboxylate species at ca. 290 eV was detected at any temperature. The O 1s spectra from ethylene glycol on CeO1.67 at 190 K are shown in Figure 7. The lattice oxygen peak from the adsorbatefree surface is at 530.4 eV. As on the oxidized surface the adsorption of ethylene glycol attenuates the lattice O peak. When the adsorbate covered surface was annealed to 250 K two adsorbate-related features are evident at 532.6 and ca. 534.4 eV. The feature at 534.4 eV is again assigned to an alcohol group,
C
OH, while the feature at 532.4 is associated with
C
O
Ce and Ce
OH groups. As in the C 1s spectra, the peaks on the reduced surface are shifted to higher binding energy compared to the oxidized surface.7 The relative intensity of the 532.4 eV feature is greater on the reduced surface compared to the oxidized surface because hydroxyls are more stable on the reduced surface and the additional O vacancies enables ethylenedioxy to form even in the presence of surface hydroxyl.7,32 The feature at 534.4 eV diminishes after annealing to 350 K and disappears at 450 K indicating the complete conversion of
C
OH into
C
O
Ce. The intensity of the peak at 532.4 eV remains largely unaffected up to 450 K because as alcohol is converted into alkoxy, Ce
OH is lost through the desorption of water. At 550 K the intensity of the lattice O peak starts to recover with an intensity drop for the feature at 532.4 eV, which is consistent with the desorption of products and a decrease in the C
O bonding that was indicated in the C 1s spectrum at 600 K (Figure 6). At 650 K the adsorbate related features are mostly gone. The C 1s spectrum of the species formed by ethylene glycol on reduced CeO2
X(111) at elevated temperatures is reminiscent of the spectrum produced by acetaldehyde at elevated temperatures

on CeO2
X(111). In Figure 8a the C 1s spectrum of 20 L of ethylene glycol on CeO1.67(111) annealed to 575 K (bottom) is plotted along with spectrum of acetaldehyde on reduced CeO1.7(111) that was annealed to 550 K (top). The peak positions and separation in the two adsorbates’ features is nearly identical. In acetaldehyde the peak near 288 eV is associated with a modified alkoxy moiety while the peak at 285.8 eV is related to an alkyl group.14 The surface species formed by acetaldehyde and ethylene glycol were further examined by the C k-edge NEXAFS spectroscopy to help identify the molecular species more precisely. Figure 8b shows the C k-edge NEXAFS spectra from ethylene glycol adsorbed on CeO1.67(111) and then annealed to 550 K (bottom) and acetaldehyde adsorbed on CeO1.7(111) annealed to 450 K (top). Two distinct pre-edge features appear in both spectra. The feature at 284.5 eV is an indication of an alkene bond (π* (CdC)). The 286.3 eV feature is most commonly assigned to a π* (CdO) resonance; however, as shown for phenol and dihydroquinone,33 it could also be associated with an alkene bond adjacent to an alcohol. The species that is consistent with the C 1s and C k-edge NEXAFS spectra is a carbanion/enolate, [OCHCH2]1
.14 On the basis of the C 1s position at 287.5 eV, which is closer to an alkoxy than a carbonyl, this surface species is assign to an enolate, Ce—O—CHdCH2. Ethylene glycol forms a mixture of the enolate and ethylenedioxy at 575 K as indicated by the greater intensity of the CO peak relative to the CHX peak in the C 1s spectrum (Figure 8a) and the smaller intensity of the pre-edge features relative to the C k-edge jump in the NEXAFS spectrum (Figure 8b). 3.3. Reduction and Oxidation of Ceria by Ethylene Glycol. Ethylene glycol can either reduce CeO2(111) by producing water and other O-containing products or it can oxidize reduced CeO2
X(111) by producing H2 and O-depleted products thus leaving the O on the substrate. Ethylene glycol appears to be an unlikely reductant because of the amount of O contained in the molecule. However the formation of oxygen rich products such as water, CO and CO2 (Figures 2 and 3), and only small amounts of oxygen-deficient products such as H2 and acetylene, suggests 13729

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C

ARTICLE

Figure 8. (A) C 1s sXPS and (B) C k-edge NEXAFS spectra from ethylene glycol adsorbed on CeO1.67 and acetaldehyde adsorbed on CeO1.7(111). The ethylene glycol was annealed to 575 K in (A) and 550 K in (B). The acetaldehyde was annealed to 550 K in (A) and 450 K in (B). NEXAFS spectra were recorded at a normal angle of incidence.

Figure 9. Ce 3d XPS spectra from (A) fully oxidized CeO2(111), (B) after three exposure/TPD cycles of ethylene glycol on oxidized CeO2(111), (C) reduced CeO1.67(111), and (D) after reaction of ethylene glycol on CeO1.67(111).

that the decomposition of ethylene glycol may lead to the reduction of oxidized ceria. Figure 9 shows the Ce 3d XPS spectra of the CeOX(111) substrate at various stages during the ethylene glycol TPD cycles and methanol reduction. The initial, fully oxidized CeO2(111) surface is shown in Figure 9a. The labels above the peaks in spectrum A indicate features that are characteristic of Ce4+.34,35 After three ethylene glycol TPD cycles (Figure 9b) the peak U000 has decreased and there is an increase in intensity at V0, V0 , and U0 that are indicative of reduction from Ce4+ to Ce3+. Analysis of this spectrum indicates that the ceria is reduced to 70% Ce4+ after three ethylene glycol TPD cycles.35 Figure 9c shows that the surface was further reduced to 40% Ce4+ after exposure to methanol at 700 K. Ethylene glycol reactions on the reduced surface resulted in a slow reoxidation of the reduced surface (Figure 9d). After three ethylene glycol TPD cycles on the reduced surface the degree of oxidation increased to 45% Ce4+.

The change in the Ce oxidation state is also evident in the successive TPD spectra. The gray lines in Figures 1
3 show the ethylene glycol TPD during the third exposure/TPD cycle on an initially fully oxidized CeO2(111) surface. In Figure 1 it can be seen that the chemisorbed ethylene glycol desorption from 230 to 400 K resembles that from the reduced surface (dotted line) rather than that from the oxidized surface (black line). The acetaldehyde production rises dramatically during the third cycle (Figure 2, black and gray lines). The absolute amount of acetaldehyde is relatively unchanged between the lightly reduced and the highly reduced surfaces (Figure 2, gray and dotted lines); however, the production appears to shift between a lower temperature channel near 560 K into a higher temperature channel near 610 K as the degree of reduction increases. As the acetaldehyde formation increases the CO and CO2 formation decrease significantly. The shift in selectivity from CO and CO2 to acetaldehyde is also reflected in the water production at elevated temperatures (Figure 3). Because acetaldehyde retains more H than CO or CO2 less water is produced. The amounts of ethylene and acetylene (Figure 2) are relatively unchanged during the third cycle compared to the fully oxidized surface although their desorption temperatures shift. Formation of totally dehydrated products therefore appears to be most favorable at higher degrees of reduction. Reoxidation of reduced ceria by ethylene glycol is also evident in the TPD spectra (data not shown). Compared to the most reduced surface that was created after the methanol treatment (Figures 1
3, dotted lines) the amounts of ethylene, acetylene, and H2 decrease in each subsequent TPD cycle.

4. DISCUSSION The adsorption and reaction of ethylene glycol with oxidized CeO2(111) is shown in Scheme 1. Ethylene glycol initially adsorbs on the CeO2(111) surface at low temperature through deprotonation of one or both of the OH groups forming 13730

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C

ARTICLE

Scheme 1. Adsorption and Reaction of Ethylene Glycol on Oxidized CeO2(111)

Ce
OCH2CH2OH, Ce
OCH2CH2O
Ce (ethylenedioxy), and Ce
OH. This is evident in the C 1s spectrum at 250 K (Figure 4) where alkoxy was the only feature in the spectrum indicating only C
O
Ce or C
OH bonds. The dissociation of the alcohol is consistent with C1
C3 alcohols that form alkoxy species and hydroxyl groups on oxidized CeO2(111).7,8 The O 1s data further indicate the coexistence at 250 K of unreacted
C
OH and surface bound
C
O
Ce species (Figure 5). Since Ce
OH has been shown to recombine and desorb as water on oxidized CeO2(111) below 250 K,7,8 most of the intensity in the 531.8 eV peak is due to
C
O
Ce. The feature at 531.8 eV is significantly bigger than the feature at 533.6 indicating that
C
O
Ce outnumber
C
OH and therefore ethylenedioxy is also present at 250 K. The decrease in the intensity of the O 1s
C
OH feature at 533.6 eV between 250 and 450 K suggests two possibilities: (a) the undissociated C
OH group reacts to form
C
O
Ce and thus creates a stable ethylenedioxy species or (b) recombination of Ce
OCH2CH2OH + Ce
OH into ethylene glycol which then desorbs. Both of these processes appear to occur. The TPD clearly shows ethylene glycol desorption above 250 K. However at 450 K the O 1s spectrum indicates that no
C
OH remains while the
C
O
Ce peak does not decrease in intensity and the C 1s indicates that alkoxy is the predominant species thus supporting the transformation of Ce-OCH2CH2OH into ethylenedioxy. Several studies over clean metal or oxygen covered metal surfaces have shown an adsorption mechanism similar to that on CeO2(111). On Cu(110) ethylene glycol shows a single peak in the C 1s spectrum at low temperature and a broad feature which could be resolved into two components similar to adsorbed methoxy and adsorbed methanol peaks in the O 1s spectrum.18 It was suggested that a strongly adsorbed intermediate was bound with the hydroxyl at one end completely dissociated and the other still intact but interacting with the surface. High-resolution electron energy loss spectroscopy (HREELS) studies of ethylene glycol adsorption on Rh(111) also indicated formation of ethylenedioxy intermediates.21 Another example that demonstrated the interaction between ethylene glycol and an O-covered surface has been reported by Capote et al.19 Ethylene glycol adsorbed on oxygen covered Ag(110) through the deprotonation of both OH groups in the ethylene glycol producing adsorbed
OCH2CH2O
and
OH on the surface. The formation of Ce
OCH2CH2O
Ce from 200 to 400 K coincides with the desorption of water throughout this same temperature range. As with the monoalcohols, dissociative adsorption of ethylene glycol at low temperature leads to adsorbed hydroxyl on CeO2(111).7,8 Below 250 K these hydroxyls disproportionate to produce desorbed water, surface O, and O vacancies. The vacancies allow the OH group at the other end of the adsorbate to react with the surface forming absorbed hydroxyl and a new
C
O
Ce bond (third stage of Scheme 1). The decomposition of the ethylenedioxy species, Ce
OCH2CH2O
Ce, formed by ethylene glycol is different from the

decomposition of the ethoxy species formed by ethanol adsorption on CeO2(111).8 While some acetaldehyde is produced from ethylenedioxy, the predominant products from ethylene glycol above 450 K are CO, CO2, and water (Figures 2 and 3, black lines). The formation of CO and CO2 requires C
C bond cleavage accompanied by dehydrogenation. A relatively small amount of hydrogen and a large amount of water are produced. The formation of O-rich products, CO2 and H2O, indicates that O from the ceria is being incorporated into these products. Ethoxy, on the other hand, shows no evidence of C
C bond scission. Ethoxy decomposes in approximately the same temperature range as ethylenedioxy, 500
600 K, but acetaldehyde and ethylene are produced. A pathway for the decomposition of ethylenedioxy to CO and H2O is suggested by the C 1s spectra (Figure 4). This pathway is depicted on the right side of Scheme 1. At 550 K carboxylate is evident at 290 eV. Although the nature of this carboxylate is not clear, it is difficult to envision a C2 species with one end bound as a carboxylate and the other bound as an alkoxy. Therefore we propose that the C
C bond cleaves producing formate, HCOO, on the surface. It has previously been shown that formic acid adsorbs as formate on CeO2(111) and produces CO and CO2.17 Therefore the formation of formate from ethylene glycol will also lead to these products. As with formic acid, disproportionation of the formates to produce formic acid was not observed from ethylene glycol. In Figure 4 both ethylenedioxy (286.8 eV) and formate (290 eV) are evident at 550 K. Water formation is very evident above 450 K for ethylene glycol but not for formic acid.17 A possible explanation is that the hydrogen content is greater for ethylene glycol compared to formic acid leading to more water formation. Brown et al. have previously shown that the decomposition products from the reaction of ethylene glycol on Rh(111) were CO and H2, and no carbon was detected on the surface after the TPD experiments.21 It was suggested that unstable formyl and formaldehyde intermediates would be produced if C—C scission occurred before complete dehydrogenation of the intermediate. On another metal system, ethylene glycol reacts with Ni(111) to form H2 and CO.24,25 HREELS results have shown C—H bond scission occurred at 250 K and by heating to 400 K the surface intermediate decomposes through C—C bond breaking to form CO on the surface. Incorporation of surface oxygen in the reaction process has been shown in the example of ethylene glycol on excessively oxygen covered Ag(110), where the OCH2CH2O intermediate reacted with the surface oxygen via C—H bond breaking to yield water, formalydehyde, and adsorbed formate species at 300 K.19 Unreacted OCH2CH2O and possibly OdCHCH2O then decomposed to yield ethylene glycol and glyoxal at 350
390 K, followed by the decomposition of formate at 415 K. On fully oxidized CeO2(111) the TPD results show that no other C1 products besides CO and CO2 are produced. However, they indicate that acetaldehyde and small amount of ethylene and 13731

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C

ARTICLE

Scheme 2. Adsorption and Reaction of Ethylene Glycol on Reduced CeO2
X(111)

acetylene are produced (Figure 2). For ethylene glycol these products represent a dehydration and deoxygenation reactions. These reactions require C—O bond cleavage during the decomposition of the OCH2CH2O intermediate but no C—C bond cleavage. After breaking one of the C—O bonds, the remaining OCH2CH2 group can follow the pathway that is similar to that for ethanol decomposition on CeO2(111) that produces acetaldehyde and ethylene.8 CO and CO2 formation appears to be somewhat more favorable than the formation of the C2 products (primarily acetaldehyde) indicating that cleavage of the C—C bond is favored over C—O bond cleavage when both ends of the molecule are bound to the oxidized ceria surface. However, either C—O or C—C bond scission always occurs during the decomposition of the OCH2CH2O intermediate as indicated by the absence of the C2 dehydrogenation products, glyoxal (OdCHCHdO) and hydroxyacetaldehyde (HOCH2CHdO). The adsorption and reaction of ethylene glycol on reduced CeO2
X(111) is depicted in Scheme 2. On reduced CeO2
X(111) the formation of the OCH2CH2OH and OCH2CH2O intermediates from deprotonation of one or two hydroxyls on ethylene glycol is again evident in the C 1s and O 1s sXPS spectra (Figures 6 and 7, respectively). As on the oxidized surface
C
OH is evident in the O 1s peak at 534.4 eV indicating the presence of OCH2CH2OH. However its relative intensity is weaker compared to the oxidized surface (Figure 5) because the presence of vacancies on the reduced surface provides more active sites for the adsorption of ethylenedioxy. The disappearance of the 534.4 eV feature upon annealing to 450 K is entirely due to the conversion of OCH2CH2OH to OCH2CH2O since no ethylene glycol desorbs in this temperature range (Figure 1). The concentration of C-containing species, as indicated by the C 1s intensities, is roughly the same at 250 K on the oxidized and reduced surfaces. However, whereas the C 1s intensity decreased on the oxidized surface as the sample was heated to 450 K due to desorption of ethylene glycol, the intensity was unchanged on the reduced surface. Previous studies on C1
C3 alcohols have shown adsorption on reduced ceria results in less alkoxy recombination and water desorption at low temperature with substantially more decomposition products produced at high temperature.7,8 The methoxy coverage from methanol at 450 K on reduced CeO2
X(111) was more than twice the methoxy coverage on oxidized CeO2(111).7 This is presumably due to an increase in the number of surface vacancies and the greater stability of OH on CeO 2
X. Recombination on the oxidized surface is less pronounced for ethylene glycol. The ethylenedioxy coverage at 450 K is only 30% greater on reduced ceria compared to oxidized ceria. Ethylene glycol also differs from the monoalcohols in that water is produced on a reduced surface. Ethylene glycol is unique among the oxygenates studied thus far in that a significant amount of water desorbs below 450 K on a highly reduced surface. This can be explained by the transition from Ce
OCH2 CH2OH to Ce
OCH2CH2O
Ce. The formation of Ce
OCH2CH2O
Ce is partially blocked by Ce
OH

on the surface. At low temperature
C
OH reacts with Ce
OH liberating water and forming a C
O
Ce bond. At higher temperatures, i.e., above 500 K, the Ce
OH species react to produce H2 leaving O with the substrate and the H and O in the enolate desorb in the acetaldehyde product. Therefore little water is seen above 500 K from ethylene glycol on the reduced surface. Ethylene glycol, however, is similar to the monoalcohols in that the selectivity toward different reaction products changes when the ceria substrate becomes reduced. The decomposition products for the primary alcohols such as ethanol and 1-propanol shifted toward the alkene dehydration products at the expense of the aldehyde dehydrogenation products as the substrate became reduced. This reflects the competition between the product and the substrate for the oxygen in the adsorbate. For ethylene glycol, a similar trend is seen where O-rich products such as CO and CO2 are produced on the oxidized surface but largely disappear on the reduced surface (Figure 2, black and dotted lines, respectively). Acetaldehyde production grows rapidly at the expense of CO and CO2 in response to a small amount of reduction (Figure 2, gray lines), but then the amount of acetaldehyde remains constant while the ethylene and acetylene production increase on a highly reduced surface (Figure 2, dotted lines). The formation of the formate intermediate which results from C
C bond scission is most favored on the most highly oxidized surface and the selectivity shifts toward the C2 products after a small amount of O is removed from the substrate. Additional insight into the selectivity for ethylene glycol decomposition on reduced CeO2
X(111) is obtained from the C 1s data (Figure 6) where an enolate intermediate is identified at high temperature. From previous studies in C1
C3 alcohols an alkoxy was the only C-containing intermediate observed.7,8 Carbanion/enolate intermediates were seen, however, in the study of acetone and acetaldehyde on the reduced CeO2
X surface.14,15 The carbanion/enolate was created by the deprotonation of the methyl group and possibly a bonding of the C to a surface Ce cation. The same intermediate is created by breaking the C
O bond at one end of the ethylenedioxy intermediate and a C
H bond at the other end (Scheme 2). The C 1s and C k-edge spectra produced by acetaldehyde and ethylene glycol at elevated temperatures display distinct similarities suggesting that the same intermediate is formed by both molecules (Figure 8). The TPD spectra also display similarities with acetaldehyde, H2, ethylene, and acetylene desorption occurring in a similar temperature range between 550 and 650 K. The desorption temperatures for these products also follow the same sequence with acetaldehyde < ethylene < acetylene. This progression can be explained by the stability of Ce
OH on the surface and the transport of O from the surface into the bulk. As the temperature increases H2 desorption competes with the formation of C2 products so that the organic molecules have less H as the temperature increases. In addition, as the temperature increases O transport from the surface to the bulk increases increasing the likelihood of cleaving the C
O bond to replenish the O on the 13732

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733

The Journal of Physical Chemistry C surface. We note that the similarities between the acetaldehyde and ethylene glycol TPDs are not perfect; in particular much more acetaldehyde is produced above 500 K from ethylene glycol than from acetaldehyde. This is likely due to the reaction of the ethylenedioxy intermediate directly to acetaldehyde as was observed on the fully oxidized surface (Scheme 1).

5. CONCLUSIONS Ethylene glycol adsorbs on both fully oxidized and reduced CeOX(111) surfaces through its two hydroxyl groups. It is initially bound to the surface by formation of one C
O
Ce bond followed by another after annealing. Surface hydroxyls are created during the adsorption and removed as water leaving a stable
OCH2CH2O
(ethylenedioxy) intermediate on the surface. On the fully oxidized surface ethylenedioxy decomposes into formate groups by breaking its C
C bond. The formate decomposition leads to CO, CO2, and H2O production and ceria reduction. Ethylenedioxy can also react by breaking C
O bonds leading to acetaldehyde and a small amount of ethylene and acetylene. Reduction of the ceria substrate shifts the selectivity toward dehydration with ethylene glycol producing primarily acetaldehyde, ethylene, and acetylene. The ethylenedioxy intermediate converts into an enolate species that follows a reaction pathway that is similar to that of acetaldehyde on the same surface. The competition for the O atoms between the reduced surface and the absorbates has made the C
O bond easier to break than the C
C bond leading to more C2 dehydration products. ’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Phone: 865-574-2796. Fax: 865576-5235.

’ ACKNOWLEDGMENT Research sponsored by the Division of Chemical Sciences, 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. Use of the National Synchrotron Light Source, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-98CH10886.

ARTICLE

(10) Ferrizz, R. M.; Wong, G. S.; Egami, T.; Vohs, J. M. Langmuir 2001, 17, 2464. (11) Matolin, V.; Libra, J.; Skoda, M.; Tsud, N.; Prince, K. C.; Skala, T. Surf. Sci. 2009, 603, 1087. (12) Zhou, J.; Mullins, D. R. Surf. Sci. 2006, 600, 1540. (13) Mei, D. H.; Deskins, N. A.; Dupuis, M. Surf. Sci. 2007, 601, 4993. (14) Chen, T. L.; Mullins, D. R. J. Phys. Chem. C 2011, 115, 3385. (15) Senanayake, S. D.; Gordon, W. O.; Overbury, S. H.; Mullins, D. R. J. Phys. Chem. C 2009, 113, 6208. (16) Stubenrauch, J.; Brosha, E.; Vohs, J. M. Catal. Today 1996, 28, 431. (17) Senanayake, S. D.; Mullins, D. R. J. Phys. Chem. C 2008, 112, 9744. (18) Bowker, M.; Madix, R. J. Surf. Sci. 1982, 116, 549. (19) Capote, A. J.; Madix, R. J. J. Am. Chem. Soc. 1989, 111, 3570. (20) Queeney, K. T.; Arumainayagam, C. R.; Weldon, M. K.; Friend, C. M.; Blumberg, M. Q. J. Am. Chem. Soc. 1996, 118, 3896. (21) Brown, N. F.; Barteau, M. A. J. Phys. Chem. 1994, 98, 12737. (22) Jansen, M. M. M.; Nieuwenhuys, B. E.; Niemantsverdriet, H. Chemsuschem 2009, 2, 883. (23) Griffin, M. B.; Jorgensen, E. L.; Medlin, J. W. Surf. Sci. 2010, 604, 1558. (24) Skoplyak, O.; Barteau, M. A.; Chen, J. G. G. J. Phys. Chem. B 2006, 110, 1686. (25) Skoplyak, O.; Barteau, M. A.; Chen, J. G. G. Surf. Sci. 2008, 602, 3578. (26) Madix, R. J.; Yamada, T.; Johnson, S. W. Appl. Surf. Sci. 1984, 19, 43. (27) Mullins, D. R.; Radulovic, P. V.; Overbury, S. H. Surf. Sci. 1999, 429, 186. (28) Hagans, P. L.; Dekoven, B. M.; Womack, J. L. J. Vac. Sci. Technol. A 1989, 7, 3375. (29) Stohr, J. NEXAFS Spectroscopy; Springer-Verlag: New York, 2003; Vol. 25. (30) NIST Chemistry WebBook. (31) Ko, E. I.; Benziger, J. B.; Madix, R. J. J. Catal. 1980, 62, 264. (32) Kundakovic, L.; Mullins, D. R.; Overbury, S. H. Surf. Sci. 2000, 457, 51. (33) Francis, J. T.; Hitchcock, A. P. J. Phys. Chem. 1992, 96, 6598. (34) Burroughs, P.; Hamnett, A.; Orchard, A. F.; Thornton, G. J. Chem. Soc. Dalton 1976, 1686. (35) Mullins, D. R.; Overbury, S. H.; Huntley, D. R. Surf. Sci. 1998, 409, 307.

’ REFERENCES (1) Trovarelli, A. Catal. Rev. Sci. Eng. 1996, 38, 439. (2) Kennedy, E. M.; Cant, N. W. Appl. Catal. 1991, 75, 321. (3) Breen, J. P.; Burch, R.; Coleman, H. M. Appl. Catal., B 2002, 39, 65. (4) Diagne, C.; Idriss, H.; Kiennemann, A. Catal. Commun. 2002, 3, 565. (5) Kugai, J.; Velu, S.; Song, C. S. Catal. Lett. 2005, 101, 255. (6) Liu, Y. Y.; Hayakawa, T.; Suzuki, K.; Hamakawa, S. Appl. Catal. A: Gen. 2002, 223, 137. (7) Mullins, D. R.; Robbins, M. D.; Zhou, J. Surf. Sci. 2006, 600, 1547. (8) Mullins, D. R.; Senanayake, S. D.; Chen, T.-L. J. Phys. Chem. C 2010, 114, 17112. (9) Siokou, A.; Nix, R. M. J. Phys. Chem. B 1999, 103, 6984. 13733

dx.doi.org/10.1021/jp201639n |J. Phys. Chem. C 2011, 115, 13725–13733