Journal of Catalysis 307 (2013) 132–139

Contents lists available at ScienceDirect

Journal of Catalysis

journal homepage: www.elsevier .com/locate / jcat

Near ambient pressure XPS investigation of the interaction of ethanolwith Co/CeO2(111)

0021-9517/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.jcat.2013.07.015

⇑ Corresponding author.E-mail address: christian.papp@chemie.uni-erlangen.de (C. Papp).

1 Shared first authors.

L. Óvári a,1, S. Krick Calderon b,1, Y. Lykhach b, J. Libuda b, A. Erd}ohelyi c, C. Papp b,⇑, J. Kiss a,c,H.-P. Steinrück b

a MTA-SZTE Reaction Kinetics and Surface Chemistry Research Group, H-6720 Szeged, Rerrich Béla tér 1, Hungaryb Lehrstuhl für Physikalische Chemie II, University of Erlangen-Nürnberg, Egerlandstr. 3, 91058 Erlangen, Germanyc Department of Physical Chemistry and Materials Science, University of Szeged, H-6720 Szeged, Aradi vértanúk tere 1, Hungary

a r t i c l e i n f o

Article history:Received 20 March 2013Revised 13 July 2013Accepted 16 July 2013

Keywords:Ethanol steam reformingNear ambient pressure XPSCobaltCeria

a b s t r a c t

Near ambient pressure X-ray photoelectron spectroscopy was applied to study the interaction of ethanol(CH3CH2OH) with a well-ordered CeO2(111) film on Cu(111) and with a Co/CeO2(111)/Cu(111) modelcatalyst. The oxidation state of the surface and the chemical nature of reaction intermediates were ana-lyzed. At 300 K, the oxidation state of ceria decreased gradually with increasing ethanol pressure. At aconstant pressure of 0.1 mbar, the reduction of Ce4+ to Ce3+ increased significantly between 320 and600 K due to a higher mobility of oxygen or Ce3+ centers at elevated temperatures. The main intermedi-ate, ethoxide, was formed by dissociative adsorption of ethanol at room temperature. No coke formationwas observed up to 600 K on CeO2. Upon deposition of metallic cobalt, partial reduction of ceria wasobserved, leading to the formation of Co2+ sites but still leaving metallic Co in the metal particles. Duringthe reaction of ethanol with the Co/CeO2(111) model catalyst, the amount of Co2+ decreased drasticallywith increasing temperature, and at 600 K, the majority of Co was metallic. This process was accompa-nied by the severe reduction of ceria and the formation of significant carbon deposits.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Ethanol obtained by the fermentation of agricultural wastes canbe an important carbon-neutral renewable energy source and alsoa renewable raw material for chemical industry. Great efforts arecurrently undertaken to produce hydrogen by heterogeneouslycatalyzed processes from these renewable sources. Hydrogen haspotential applications as energy carrier to be used in fuel cells orin large-scale processes like ammonia synthesis. This demand in-spired studies of the dehydrogenation of oxygenated hydrocar-bons, especially ethanol [1–4]. The steam reforming of ethanol(SRE), the partial oxidation of ethanol (POX), and the oxidativesteam reforming (OSR) are currently in the focus of catalytic re-search as potential candidates for H2 production [4]. One advan-tage of bio-derived ethanol is its low sulfur content, reducing thepoisoning of catalysts, and its low toxicity as compared to metha-nol. Furthermore, ethanol from bio sources contains water andthus is particularly well suited for steam reforming, since the dis-tillation step to produce pure ethanol can be omitted.

During the steam reforming of ethanol, acidic supports likeAl2O3 favor dehydration and thereby increase the tendency forcoke formation due to the polymerization of ethylene [5]. How-ever, on ceria (CeO2), which is considered to be a basic support,dehydration is limited and its redox properties hinder coke forma-tion. Additionally, ceria promotes the water–gas-shift (WGS) reac-tion [6,7]. Noble metals, especially Rh, proved to be excellentcatalysts for the reaction, but their price is prohibitively high. Asan alternative to expensive transition metals, Co is considered apromising catalyst for the reaction [7–9]. Co achieves a high etha-nol conversion and selectivities of over 90% for H2 and CO2 on CeO2

and also on other supports, even at relatively low temperatures(�723 K) [10,11]. Supported Co catalysts break the C–C bond in ad-sorbed ethanol [12]. It was found that addition of a CeO2 promoterto the unsupported Co powder catalyst stabilizes the more activehcp structure of Co and hinders sintering during SRE [13].

Surface science studies on adsorption of alcohols on group VIIIBand IB transition metal surfaces, performed in ultrahigh vacuum(UHV), were reviewed by Mavrikakis and Barteau [14]. Dependingon the particular metal, dehydrogenation and C–C bond scissionlead to the formation of alkoxide, oxametallacycle, aldehyde, acyl,and coke on the surface and mostly H2, CH4, CO, and aldehyde inthe gas phase. The rupture of the C–C bond generally requires high-er activation energy than the scission of C–H and O–H bonds. C–O

L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139 133

bond breakage was not observed in the majority of cases. In orderto elucidate the role of Co in SRE, the reactivity of metallic and oxi-dized Co foils toward ethanol was investigated by temperature-programmed desorption [15]. On metallic Co, decarbonylationwas dominant, possibly through an oxametallacycle intermediate,yielding CO, H2, traces of CH4, and surface C as products. However,CoOx surfaces, exposing Co2+ as majority cobalt sites, were selec-tive toward dehydrogenation to produce acetaldehyde, wherebyother products were CO2, CO, and H2. On the Co3O4 surface, totaloxidation prevailed. The authors suggested that Co2+ sites are theactive centers in SRE, and Co0 sites are responsible for coke forma-tion. Though the initial oxidation state of cobalt was analyzed byXPS, the surface was not monitored during or after ethanol decom-position in this study. Thus, the role of Co2+ and Co0 in SRE is still asubject of debate [8,15,16]. In TPD measurements on ethanoldecomposition on the single crystalline surface of Co(0001), themajor products were H2 and CO, and acetaldehyde was also de-tected in small amounts [17], in analogy with the results on metal-lic Co foil [15]. Nevertheless, the nature of some desorbingproducts is still under debate [15,17]. High-resolution photoemis-sion study performed during ethanol adsorption on Co(0001) [17]at 160 K showed that O–H bond scission led to the formation ofethoxide. In contrast to many other low index Group VIIIB surfaces,it was found that ethoxide on Co(0001) mostly decomposed intoethylene/acetylene and atomic O via C–O bond scission, presum-ably through an acetaldehyde intermediate.

On high surface area CeO2, the adsorption and decomposition ofethanol were investigated by TPD and infrared spectroscopy (IR)[18,19]. Literature agrees that there is ethanol, ethoxide, and ace-tate species on the surface, but concerning the formation of acetylno agreement exists. In TPD experiments, mostly ethanol and acet-aldehyde were desorbing and only small amounts of decomposi-tion products such as CH4, CO2, CO, and only trace amount ofacetone were detected. Reduced ceria was more reactive towardsethanol decomposition, resulting in less molecular ethanol desorp-tion, with acetaldehyde again being the product with the highestyield. CO formation was negligible and the CO2 formation shiftedto higher temperatures (T > 650 K). Ethoxide fragments were stableagainst annealing up to 623 K in vacuum. The changes in IR spectraobserved at 623–673 K were assigned to the transformation of ace-tate groups to carbonates. The importance of the metal-supportperimeter in SRE is frequently emphasized [4]. Surface OH groupscan facilitate the transformation of ethoxide to acetate [4,20]. Thepresence of steam contributes to the demethanation of acetategroups to produce carbonate leading to CO2 and CH4, which is par-ticipating in a later SR reaction [4].

Since real catalysts are often structurally very complex, for thebetter understanding of surface reactions, it is helpful to investi-gate simplified, but well-controlled model catalyst systems, likemetal nanoparticles on supported single crystalline oxide surfaces[21,22]. With relevance to ceria, the formation of CeO2(111) filmswas characterized on Cu(111) [23–30] and on other supports[31,32]. Evaporation of Ce onto Cu(111) at T = 520 K substratetemperature in an O2 atmosphere resulted in the formation ofnearly stoichiometric CeO2 films with a (1.5 � 1.5) unit cell ofCeO2(111)/Cu(111) [26]. From low energy electron diffraction(LEED), it was concluded that the films are indeed closed layers[27]. For further details on the growth mechanism and on theeffects of temperature and partial pressures, we would like to referto the literature cited above.

X-ray photoelectron spectroscopy (XPS) is a well-establishedUHV technique providing information on the elemental composi-tion and chemical state of the solid surfaces. Herein, we use a nearambient pressure system [33] to study interactions between etha-nol and a model Co/CeO2(111) catalyst system in situ under morerealistic conditions, i.e., up to pressures of 1 mbar. Our aim was to

get closer to the real catalytic conditions and to determine not onlythe transient intermediates but also the oxidation states of activecomponents during the reaction. The investigation of the modelcatalyst under ambient conditions allows us to bridge the pressuregap between the surface science and realistic conditions.

2. Experimental

The details of the near ambient pressure XPS setup can be foundin previous papers [33,34]. The XPS apparatus used for photoelec-tron and in situ photoelectron investigations of the Co/CeO2 systemallows conventional UHV experiments but also high-pressure XPS(with reactive gas background pressures of up to 1 mbar). Thekey feature of this equipment is the efficient combination of fourdifferential pumping stages between the sample (electron emis-sion) and analyzer (electron detection). The analyzer is a modifiedOmicron EA-125 unit, and the X-ray source is a modified dual an-ode source (Specs XR-50) whose vacuum environment is com-pletely isolated from that of the analysis chamber with the helpof a 3 lm thick Al foil and additional gaskets. Gas dosing wasachieved via flooding the main chamber with the required gas.Photoemission spectra of the Ce 3d, Co 2p, O 1s, and C 1s regionswere recorded using Al Ka radiation (1486.6 eV) at normal emis-sion. The pass energy was 50 eV for all regions, with the exceptionof the C 1s region, for which 100 eV pass energy was used, if notmentioned otherwise, to obtain higher intensities. The acquisitionof the spectra took�2 h, with the exception of the 1 mbar pressure,when �6 h were required for a complete data set of all relevant re-gions. For peak fitting, Tougaard (Ce 3d) or Shirley (O 1s, Co 2p)baselines and Gauss–Lorentz peak shapes were used.

The Cu(111) crystal was cleaned with cycles of Ar+ sputtering(1 keV, 300 K, 45 min) and annealing at 873 K for 5 min, leadingto a sharp hexagonal LEED pattern and a contamination level (typ-ically C and O) below 1%, verified by XPS. The CeO2(111) film wasprepared on Cu(111) in a similar way as reported in the literature[24,26–28]. Ce metal (Alfa Aesar, purity 99.9%) was evaporatedwith an e-beam evaporator from a Mo crucible at an evaporationrate of 0.15 Å/min in an O2 background of 1 � 10�6 mbar at a sub-strate temperature of 523 K for 100 min. O2 was pumped down atthe end of the growth, after the sample cooled down to 430 K. TheCe deposition rate was calibrated already in the absence of O2 witha quartz crystal microbalance and it was stable during the filmgrowth, as it was checked also at the end of Ce evaporation. Theceria film thickness obtained this way was 17.3 Å assuming CeO2

stoichiometry, corresponding to 5.5 ML of CeO2(111). Investiga-tions of the film with LEED showed the characteristic (1.5 � 1.5)pattern of CeO2 with the complete absence of the Cu(111) spots;the latter observation is attributed to the formation of a continuousceria film. Additional treatments in 1 � 10�6 mbar of O2 up to673 K did not result in any appreciable changes in LEED, in theCu/Ce XPS intensity ratios and in Ce 3d peak shapes. No dewettingof the ceria film was observed during our experiments, since no in-crease in the Cu 2p3/2 XPS intensity was observed. Cobalt was evap-orated from a wire (99.99% purity, MaTecK GmbH). The cobaltdeposition rate was calibrated by a quartz microbalance. Ethanol(cleanliness: 99.9%) was further purified with freeze-pump-thawcycles.

3. Results and discussion

3.1. XPS characterization of CeO2(111) on Cu(111)

After the preparation of the CeO2(111) film on the Cu(111) sur-face, it was characterized by XPS. The Ce 3d spectrum shown inFig. 1A (‘‘clean’’) is composed of three doublets, (u0 0 0, v0 0 0), (u00, v00)

540 535 530 525

1

534.

8

Cou

nt ra

te (a

rb. u

.)Binding energy (eV)

529.

2

535.

5

p(mbar)

1

10-1

10-4

10-6

clean

BO 1s

(gas)

300 290 280

287.

5

1

288.

728

9.4

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

285.

9

p(mbar)

1

10-1

10-4

10-6

clean

C C 1s

(gas)

920 910 900 890 880 870

u''' u'' u v''' v''

916.

6

907.

5

900.

789

8.2

888.

8

p(mbar)

1

10-1

10-4

10-6

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

clean

A Ce 3d

882.

3

v

x5

Fig. 1. XP spectra of the CeO2(111) surface collected at 300 K in the presence of ethanol at increasing pressures. For comparison, spectra of the clean ceria surface and spectraof gas-phase ethanol (1 mbar) are also displayed. Before exposure to ethanol at 300 K at each pressure, the surface was reoxidized in 10–6 mbar of O2 at 623 K.

134 L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139

and (u, v) corresponding to the emission from the spin-orbit split3d3/2 and 3d5/2 core levels [35–38]. The three doublets are assignedto different final states of tetravalent Ce (Ce4+ ions) in Ce com-pounds: u0 0 0 (916.6 eV) and v0 0 0 (898.2 eV) are due to a Ce 3d94f0 O2p6 final state, u00 (907.5 eV) and v00 (888.8 eV) to a Ce 3d94f1 O2p5 final state, and u (900.7 eV) and v (882.3 eV) to a Ce 3d94f2 O2p4 final state. For a more detailed picture, which might allow adeeper understanding of the core levels and the properties of Cein these spectra, we refer to [39]. Note that especially the well sep-arated u00 0 peak at 916.6 eV is characteristic for the presence of Ce4+

[36].Indeed, after evaporating a 0.3 nm Ce layer onto the CeO2(111)

film at 523 K, the intensity of u0 0 0 peak at 916.6 eV significantly de-creased (Fig. 2), and four new features developed in the Ce 3d re-gion: They are assigned to two spin-orbit split doublets, (u0, v0)and (u0, v0), resulting from two different final states of Ce3+. TheseCe3+ cations are formed by the reaction of Ce metal with CeO2.

920 910 900 890 880 870

880.

7

907.

5

v0

v'u0

899.

3

888.

8

916.

6

903.

8

898.

2

885.

2

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

clean+0.3nm Ce at 523K+600K 10min

882.

3

Ce 3d u'

Fig. 2. Ce 3d spectrum of the clean CeO2(111) surface, after deposition of (nominal)0.3 nm of metallic Ce at 523 K, and after subsequent annealing at 600 K.

According to literature, u0 (903.8 eV) and v0 (885.2 eV) are assignedto a Ce 3d94f1 O 2p6 final state, and u0 (899.3 eV) and v0

(�880.7 eV) to a Ce 3d94f2 O 2p5 final state [36]. Interestingly,upon heating to 600 K for 10 min, the six-peak structure of Ce4+,characteristic of the oxidized surface, was largely recovered(Fig. 2). This shows an increased exchange between sub-surfaceand surface oxygen vacancies on CeO2(111) upon heating [40].To summarize, in XPS the Ce3+ and Ce4+ species are differentiatedwith distinct line shapes corresponding to their various final states.Note that also the freshly prepared CeO2(111) film showed a smalldegree of reduction as demonstrated by the deconvolution of theCe 3d region (Fig. 3A, bottom spectrum).

Additional information on the quality of the CeO2(111) film isobtained from the O 1s spectrum in Fig. 1B (clean), which showsthe signal of lattice oxygen at 529.2 eV. The C 1s region in Fig. 1C(clean) exhibits a peak at 289.4 eV that is not resulting from carbonimpurities, but from the Ce 4s core level. Overall, one can state thespectra of our CeO2(111) film resemble literature spectra very well[36], confirming the quality of our preparation.

3.2. Adsorption and decomposition of ethanol on CeO2(111) atdifferent pressures at 300 K

As a reference, we first measured XP spectra of vapor-phaseethanol (CH3CH2OH) at 1 mbar. The corresponding C 1s and O 1sspectra are displayed in Fig. 1B and C as topmost spectra. The O1s region displays one peak at 535.5 eV, and in the C 1s region,two peaks at 287.5 and 288.7 eV are observed, in agreement withliterature data [41]. The two C 1s peaks are assigned to methyl car-bon and alkoxy carbon, respectively.

The adsorption of ethanol on CeO2(111) at 300 K was investi-gated by recording spectra at 10�6, 10�4, 10�2, 10�1, and 1 mbar.Prior to each experiment, the sample was oxidized at 623 K, to en-sure identical conditions. The evolution of Ce 3d spectra is shownin Fig. 1A. The increase in ethanol pressure resulted in attenuationof the intensity of Ce 3d spectrum. Simultaneously, a relative in-crease in the intensity in the valleys between v and v00, and u andu00 was noted. This indicates the emergence of v0 (�885.2 eV) andu0 (903.8 eV) peaks from Ce3+, due to mild reduction of ceria underincreasing ethanol pressure at 300 K. The peak fitting of selected Ce3d spectra shown in Fig. 3A confirms this trend. To quantify theamount of Ce3+, the ratio of the integrated intensities of Ce3+ spec-tral contribution to the total Ce 3d spectrum, i.e., Ce3+/(Ce3++ Ce4+),is plotted in Fig. 4A as a function of ethanol pressure. The plot

920 910 900 890 880

clean

10 -4

Cou

nt ra

te (c

ps)

Binding energy (eV)

p(mbar)

0.1

Ce 3d A

540 535 530 525

534.

353

4.8

531.

5

1

clean

10-4Cou

nt ra

te (c

ps)

Binding energy (eV)

p(mbar)

0.1

O 1s B

529.

2

Fig. 3. Peak fitting for selected curves of Fig. 1. For the assignment of the Peaks in A and C please refer to the text.

L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139 135

shows that the reduction of CeO2(111) increases up to 10�4 mbarand levels off thereafter. This behavior suggests that at pressures of10�4 mbar or higher, the further reduction is hindered, due to a re-duced mobility of either oxygen or Ce3+ centers at 300 K.

To reveal, if the reduction is due to the high exposure dose(time � pressure) or due to the high pressure, we investigatedthe effect of time in this reaction, see Fig. 4. The CeO2(111) surfacewas in contact with 1 � 10�5 mbar of ethanol for 69 h at 300 K(exposure: �1.9 � 106 L). Thereafter, the spectra were taken in10�5 mbar ethanol (b) and after 15 min of pump-down (c), withthese two spectra showing only minor differences. In a separateexperiment, a spectrum was taken in 10�1 mbar ethanol (d) after�1 h at this pressure, yielding a much higher exposure of�2.8 � 108 L (see also Fig. 1A): The analysis shows that the extentof reduction is higher for the lower total exposure, i.e., at lowerpressure but higher exposure time, with a Ce3+/(Ce3++ Ce4+) ratioof 0.22, as compared to the value of 0.10 at 10�1 mbar. Note thatthis effect is not attributed to the exposure of ceria to vacuum,

10 -8 10 -6 10 -4 10 -2 1000.00

0.02

0.04

0.06

0.08

0.10

0.12

Ce

(III)/

(Ce(

III)+

Ce(

IV))

p (mbar)

A

Fig. 4. (A) Reduction of ceria at 300 K, calculated from peak fitting of the Ce 3d region; (69 h (b), followed by 15-min evacuation at 300 K (c), and after exposure to ethanol at 1

since for CeO2 none or only slight reduction is observed [25,42].It is rather attributed to diffusion of oxygen or Ce3+ centers throughthe reduced surface layer as the rate-limiting step, as also sug-gested at the end of last paragraph.

During ethanol adsorption at 300 K, the O 1s peak at 529.2 eV oflattice oxygen develops a shoulder at the high binding energy sideat 531.5 eV (see Fig. 3B). This value would be in line with the for-mation of OH groups, which were observed after exposure to H2O[43,44] or CH3OH [45]. This assignment is however equivocal, asthe O 1s peaks of ethoxide and acetaldehyde surface species arealso expected at this binding energy. Especially, ethoxide forma-tion is likely as proposed in literature [4,18,20]. At higher pres-sures, an additional contribution from weakly held, molecularlyadsorbed ethanol should also be considered. Matolín et al. foundthe O 1s peak of a methanol multilayer on CeO2(111) at 534 eV[45], and the corresponding O 1s peak for physisorbed ethanol isexpected to occur at a similar position. Indeed, the inspection ofthe O 1s spectra in Fig. 1B and the fits in Fig. 3B reveal a weak peak

920 910 900 890 880 870

d

c

b

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

d: in 10 -1 mbar 1 h

c: after 15 min pump down

b: in 10 -5mbar 69 h a: clean

Ce 3dB

a

B) Ce 3d spectra of clean CeO2(111) (a), after exposure to ethanol at 10�5 mbar for0�1 mbar for 1 h (d).

920 910 900 890 880 870

600

500

400

320

916.

6

907.

5 900.

790

3.8

898.

2

888.

888

5.2

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

882.

3

Ce 3d

T(K)

clean

A

535 530 525

600

500

400

320

529.

452

9.2C

ount

rate

(arb

. u.)

Binding energy (eV)

O 1s

T(K)

clean

B

300 295 290 285 280

289.

428

6.7

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

285.

9

600

500

400

320

T(K)

clean

C 1sCe 4s

C

Fig. 5. XPS spectra collected from CeO2 in 0.1 mbar of ethanol at temperatures of up to 600 K. The ‘‘clean’’ spectrum was recorded before ethanol exposure. The pass energywas 50 eV for all regions.

136 L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139

at 534.3 eV, which becomes observable at 10�1 mbar (note that thecomparably large peak at 534.8 eV at 1 mbar is due to the ethanolgas phase). The negligible intensity of the peak due to molecularethanol adsorption below 10�1 mbar suggests that dissociativeadsorption of ethanol prevails at 300 K. At 0.1 and 1 mbar, how-ever, the presence of molecularly adsorbed ethanol cannot be ex-cluded. The gas-phase spectrum of ethanol with the sample notpresent (topmost spectrum in Fig. 1B) displays a somewhat higherbinding energy (535.5 eV) than with the sample present, due to thedifferences in the work functions of the sample and the spectrom-eter [33].

The C 1s region at different ethanol pressures at 300 K is shownin Fig. 1C. As mentioned above, the broad peak centered at289.4 eV originates from the Ce 4s level. Upon exposure to ethanolat 300 K, a very broad C 1s feature is observed at �285.9 eV. Itsassignment is challenging because of the possible presence of var-ious surface species (mostly ethoxide, acetaldehyde, and acetategroups), giving multiple overlapping peaks in that energy range.The adsorption of ethanol [46] and acetaldehyde [47] was previ-ously studied by Mullins and coworkers on CeO2(111) and by

920 910 900 890 880

v0v'u 0u'vv''v'''uu''

400

500

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

T(K)

Ce 3d

600

u'''

A

Fig. 6. (A) Peak fitting for selected curves of Fig. 5; (B) Ce3+/(Ce3+ + Ce4+) ratio as a mea0.1 mbar of ethanol; calculated from the peak fitting of the Ce 3d region shown in (A).

other authors on polycrystalline samples [48,49]. Formation of alk-oxy species was detected by XPS on oxidized CeO2(111) afteradsorption of alcohols at 300 K [46]. The two non-equivalent C 1slevels of the ethoxide species were located at �285.9 eV and at�286.9 eV [46]. Acetaldehyde did not decompose on the stoichi-ometric CeO2(111) surface and desorbed below room temperature.On reduced CeOx(111), chemisorbed acetaldehyde was more sta-ble, characterized at room temperature by C 1s peak positions at286.9 and 289.5 eV and assigned to non-dissociated acetaldehydeas majority species. The C 1s binding energy of the carboxylic car-bon atom in acetate was observed at 290 eV [46]. Since for bothacetate and acetaldehyde one of the C 1s peaks was found in theliterature at �289.5–290.0 eV and we observed only a slight in-crease in that region upon ethanol adsorption, we conclude thatthese species can only be minorities in our case and the dominat-ing surface fragment is ethoxide up to 10�1 mbar, formed by reac-tion (1):

CH3CH2OHþ Os ! CH3CH2Oa þ OHa ð1ÞOHa þ OHa ! H2Og þ Os þ Vo ð2Þ

300 400 500 6000.0

0.1

0.2

0.3

0.4

0.5

in 0

.1 m

bar o

f eth

anol

Ce

(III)/

(Ce(

III)+

Ce(

IV))

Temperature (K)

clea

n

B

sure for the reduction of ceria at temperatures of up to 600 K during exposure to

L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139 137

Reaction (2) may lead to the formation of water, lattice oxygen andoxygen vacancies (Vo) and finally to the reduction of ceria. Indeed,water production was observed during adsorption of ethanol onCeO2 at low temperatures (�200 K) [46]. At 1 mbar, the C 1s regionis dominated by gas-phase ethanol (Fig. 1C). A clear distinction ofsurface species was not possible due to the small intensity of thesurface species.

3.3. Reactions of 0.1 mbar C2H5OH on CeO2(111) film at differenttemperatures

Next, the interaction of 0.1 mbar of ethanol with CeO2(111) wasinvestigated at different temperatures up to 600 K. The correspond-ing spectra are shown in Fig. 5. In the Ce 3d region, pronouncedchanges are observed, which become most striking starting at500 K. While the u0-peak (903.8 eV) and the v0-peak (885.2 eV)due to Ce3+ markedly increase in intensity, the u0 0 0-peak(916.6 eV) characteristic for Ce4+ significantly decreases inFig. 5A; the corresponding peak fitting for selected Ce 3d spectrais shown in Fig. 6A. These changes indicate a strong reduction inthe surface layer. To quantify the changes, we plotted the ratio ofCe3+/(Ce3+ + Ce4+) in Fig. 6B for increasing temperatures; this plotshows an almost linear dependence on temperature to �0.5 at600 K. From this large value, it is concluded that an in-depth reduc-tion of the ceria film occurred at this temperature. This is notsurprising in light of Fig. 2, which suggests that diffusion of ionswithin the ceria film is more pronounced at elevated temperatures.

920 910 900 890 880 870

600

500

400

320

916.

6

907.

5 900.

790

3.8

898.

2

888.

888

5.2

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

882.

3

Ce 3d

T(K)

clean

A

800 790

786.

1

Cou

nt ra

te (a

rb. u

.)

Binding ener

Co 2p

300 295 290 285 280

284.

3

600

500

400

320

289.

4

285.

4

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

C 1s

Ce 4s

T(K)

D

290 28

Nor

mal

ized

XP

inte

nsity

Binding ener

285.

9

C

Fig. 7. (A–D) XP spectra recorded after deposition of 0.7 ML of Co on the CeO2(111) surannealing the surface in ethanol at higher temperatures. Spectra were collected in 0.1 mregion of the CeO2(111) and 0.7 ML Co/CeO2(111) surfaces collected in 0.1 mbar of ethaReduction of ceria in the presence cobalt during exposure to ethanol at temperatures of

With increasing temperature, the O 1s peak in Fig. 5B decreasessignificantly, in accordance with the reduction of CeO2 observed inthe Ce 3d region. In the C 1s region, shown in Fig. 5C, we again ob-serve a broad feature with maxima at 285.9, 286.7 and 289.4 eV,with their relative intensities changing with temperature. For thequalitative interpretation of the spectral changes during heat treat-ment, we consider the literature results for CeOx(111) thin films[46–48,50] discussed in the introduction. The peak assigned to eth-oxide (285.7 eV) is found at all temperatures, in agreement withliterature; its shift to higher binding energies at 600 K is attributedto the different bonding situation of ethoxide on the reduced ceriasurface [46]. Please note that identification of the surface species israther challenging, due to the overlap of the signals and the lowintensity in the C 1s region.

3.4. Interaction of C2H5OH with Co/CeO2(111)

The growth of Co on CeO2(111) was briefly studied previouslyusing XPS [50]. Based on the linear dependence of the Ce 3d/Co2p intensity ratio vs. cobalt coverage, two dimensional growth ofCo was proposed up to 1 ML. From the Co 2p peak shape, theresulting surface species were identified as Co0 and Co2+. Further-more, reduction of Ce was observed upon Co deposition, due tothe reaction of Co with CeO2. In the literature data, for metallic co-balt (Co0), an asymmetric Co 2p3/2 peak is observed at 778.0–778.5 eV. Co2+ is characterized by a Co 2p3/2 peak at 780–781 eV,

780 770

600

500

400

320

779.

977

8.2

gy (eV)

T(K)

clean

B

535 530 525

600

500

400

320

529.

2

534.

8

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

O 1s

T(K)

clean

C

5 280

gy (eV)

285.

4

E0.7ML Co/CeO2(111)

eO2(111)

300 400 500 6000.0

0.1

0.2

0.3

0.4

0.5

0.6

in 0

.1 m

bar o

f eth

anol

Ce

(III)/

(Ce(

III)+

Ce(

IV))

Temperature (K)

afte

r Co

depo

sitio

n in

UH

V

F

face at 320 K (‘‘clean’’), followed by admission of 0.1 mbar of ethanol at 320 K andbar ethanol at the given temperatures. (E) Normalized XP spectra of the C 1s/Ce 4snol at 320 K. Spectra are displayed also after heavy smoothing to guide the eye. (F)up to 600 K.

790 785 780 775

600K

320K

Co 2+ sat.786.1

Co2+

780.3

Cou

nt ra

te (a

rb. u

.)

Binding energy (eV)

Co 0

778.1

Co 2p 3/2

clean

Fig. 8. Peak fitting of the Co 2p3/2 region for selected curves of Fig. 7B: after thedeposition of 0.7 ML of Co in UHV (‘‘clean’’), followed by exposure to ethanol at320 K, and stepwise annealing up to 600 K.

138 L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139

with a strong satellite at 786–787 eV. The signature of Co3+ is a Co2p3/2 peak at 780–781 eV with no satellite [15,50–52].

To study the reaction of ethanol on Co-covered ceria, we depos-ited 0.7 ML of Co at 300 K. The Co coverage was kept below 1 ML toprevent a closed Co layer. Upon Co deposition, we find, similar tothe previous findings [50], an increase in Ce3+; this is evident fromthe comparison of the bottom curves of Figs. 1A and 7A, and fromthe increase in the Ce3+/(Ce3+ + Ce4+) ratio from 0.04 ± 0.01 in Fig. 4to 0.11 ± 0.02 in Fig. 7F. The Co 2p3/2 region, shown in Fig. 7B, isdominated by a rather broad peak at 779.9 eV with a weaker satel-lite at 786.1 eV, indicative of Co2+; for details see also Fig. 8. Thelarge width of the main peak suggests the coexistence of severaloxidation states, predominantly Co2+ and some Co0. Since the reac-tion of Co and CeO2 and the resulting Co2+ formation is expected toproceed mainly at the ceria support/metal cluster interface, thesmall amounts of unreacted Co0 species are presumably locatedmainly on top of the cobalt clusters (i.e., away from the interface)and hence should be available for interaction with ethanol.

The O 1s peak after Co deposition was found to be slightly moreasymmetric toward higher binding energies compared to the cleanCeO2(111) surface (cf. Figs. 1, 3, 5 and 7). This is most likely due tothe O 1s contribution from CoO [15,52] with a binding energy of529.2–530.4 eV, but might partly also result from some ethanol,CO, or H2O adsorption from the background.

Next, the Co/CeO2(111) was exposed to 0.1 mbar of ethanol attemperatures of up to 600 K; at each temperature, a set of XP spec-tra was collected under the ethanol pressure. At 320 K, 0.1 mbar ofethanol led to no change in the shape of the Ce 3d spectra inFig. 7A, i.e., to no further reduction of the CeO2 surface, in contrastto the situation for Co-free ceria; this is evident from Fig. 7F, wherethe first two data points corresponding to the situation prior andafter exposure to ethanol are identical within the margin of error.As 0.7 ML CoO/Co is not sufficient to fully cover the substrate, weconclude that by Co deposition of this amount, the ceria surfaceis reduced to the approximately same level, i.e., a Ce3+/(Ce3+ + Ce4+)ratio of 0.11, which was achieved by ethanol on the Co-free ceriasurface (see Fig. 4A), and thus a further reduction does not occur.Apart from that, we find a decrease in the Co0 intensity at778.1 eV in Figs. 7B and 8. This is probably due to a C–O bond scis-sion, or the presence of ethoxide, the latter coordinating to the co-balt surface via the electronegative O atom. Indications forethoxide and/or OH are also seen in the O 1s region, where the highbinding energy shoulder (531.5 eV) has increased relative to themain ceria O 1s peak. Please note that the presence of CO cannotbe ruled out as was proposed in earlier studies [14,17]. The C 1sfeature in Fig. 7D is rather broad, containing a peak at 289.4 eV(mostly Ce 4s contribution) in addition to the dominant broad peakat 285.4 eV. In Fig. 7E, we compare the C 1s spectra obtained at320 K in 0.1 mbar of ethanol on Co/CeO2(111) and on CeO2(111),after intensity normalization. The comparison clearly shows thatthe low binding energy edge of the C 1s region is shifted downby ca. 0.5 eV in the presence of cobalt. Due to the overlap of the sig-nals of different species, one can only speculate on possible reasonsfor this difference, e.g., C–O or C–C bond breaking leading to CxHy

fragments, or different surface potentials for Co/CoO and ceria.The presence of acetyl groups cannot be excluded either, as suchspecies were proposed on Pt(111) [53] and on Pd(111) [54], butis unlikely, as it was not observed on Co(0001) [17].

Heating of Co/CeO2(111) in ethanol at 400 K induced onlyminor changes in the spectra in Fig. 7. However, annealing at500 K led to a significant reduction of Co, accompanied by a shiftof the C 1s feature toward smaller binding energies, and somefurther reduction of ceria (Ce3+/(Ce3+ + Ce4+) ratio is �0.3). At600 K, significant further reduction of Ce and Co is found leadingto a Ce3+/(Ce3+ + Ce4+) ratio of �0.5 ± 0.05. In the Co 2p region inFig. 7B, a narrow peak is observed at 778.2 eV, i.e., the position of

metallic Co known from literature [15,50–52]. By fitting the peaks(Fig. 8), we deduce that �55–65% of the total Co area originatesfrom Co0. In the C 1s region (Fig. 7D), the overall intensity stronglyincreased with the peak maximum at 284.3 eV, which is veryprobably due to the formation of carbonaceous species. Surfacecarbon is a known reaction product in the decomposition ofethanol on metal surfaces [14].

The reduction of Ce and Co can proceed in a variety of ways, e.g.,by the recombination of OH groups possibly formed as ethanol orthe CxHyOz surface fragments lose H, or by the reaction of surfaceC with O producing CO, etc. Oxygen-containing gas-phase productsoriginating from the substrate might be CO, CO2, acetaldehyde [18],and H2O. After the partially contrasting results related to ethanoldecomposition on a Co foil [15] and on Co(0001) [17], we cannotcompletely exclude the role of a transient oxametallacycle onmetallic Co sites. Presumably, the ceria support, acting as an oxy-gen ‘‘reservoir’’ at low temperatures, hinders reduction of Co andalso the formation of coke by O (reverse) spillover, but after thereduction of ceria at 500–600 K, this oxygen buffer is exhausted.Note that coke formation was not observed on pure ceria(Fig. 5C). The accumulation of carbon certainly contributed to thesevere intensity loss in the Co 2p region (Figs. 7B and 8), togetherwith a sintering of Co nanoparticles. Sintering of Co particles couldalso contribute to the downward shift of Co 2p3/2 binding energy,as was previously observed for metal clusters on oxide supports[21,55,56]. In our case, however, the presence of the satellite fea-ture of Co2+ at 786.6 eV at low temperatures and its strong atten-uation upon annealing in ethanol is a strong indication of thechange in the oxidation state itself. Moreover, the binding energyshifts due to cluster size effects are typically in the range of 0.5–1 eV, while we observed a shift of 1.7 eV (Fig. 7B). The sinteringof cobalt and the increased oxygen mobility facilitates the accessof ethanol to surface CeO2, which is subsequently reduced. Pleasenote that the use of pure ethanol gives different results than etha-nol/water mixtures over oxidized Co/CeO2 polycrystalline catalyst,where the reduction of Co, but not that of ceria, was observed[8,16].

L. Óvári et al. / Journal of Catalysis 307 (2013) 132–139 139

4. Conclusions

Near ambient pressure X-ray photoelectron spectroscopy wasapplied to study the interaction of ethanol with a well-orderedCeO2(111) film prepared on Cu(111) and with a Co/CeO2(111)model catalyst, with respect to the oxidation state of the surfaceand to the chemical nature of reaction intermediates. At 300 K,the oxidation state of ceria decreased gradually with increasingethanol pressure, presumably through H2O desorption involvinglattice O. The transformation of Ce4+ to Ce3+ increased significantlyupon increasing the temperature from 320 to 600 K at 0.1 mbar.The primary intermediate, ethoxide, was formed by dissociativeadsorption of ethanol at room temperature. No coke formationwas observed up to 600 K. On the Co/CeO2 model catalyst, we findthat a significant part of metallic Co reacted with ceria upon depo-sition, leading to the formation of Co2+ sites and to some reductionof ceria. During the interaction of 0.1 mbar ethanol with a 0.7 MLCo/CeO2(111) model catalyst, the amount of Co2+ decreased withincreasing temperature, and at 600 K, the majority of Co wasmetallic, accompanied by a severe reduction of Ce. At 600 K, a sig-nificant amount of carbon formed on Co/CeO2(111).

Acknowledgment

The financial support by the Cluster of Excellence ‘‘Engineeringof Advanced Materials’’ and by the Alexander von Humboldt Foun-dation within the Research Group Linkage Programme isacknowledged.

References

[1] L.F. Brown, Int. J. Hydrogen Energy 26 (2001) 381.[2] A. Harynato, S. Fernando, N. Murali, S. Adhikari, Energy Fuels 19 (2005) 2098.[3] G. Halasi, I. Ugrai, F. Solymosi, J. Catal. 281 (2010) 309.[4] L.V. Mattos, B.H. Jacobs, F.B. Noronha, Chem. Rev. 112 (2012) 4094.[5] A. Erdöhelyi, J. Rasko, T. Kecskes, M. Toth, M. Dömök, Catal. Today 116 (2006)

367.[6] P.D. Vaidya, A.E. Rodigues, Chem. Eng. J. 117 (2006) 39.[7] H. Song, U.S. Ozkan, J. Catal. 261 (2009) 66.[8] B. Bayram, I.I. Soykal, D. Deak, J.T. Miller, U.S. Ozkan, J. Catal. 284 (2011) 77.[9] H. Song, B. Mirkelamoglu, U.S. Ozkan, Appl. Catal. A 382 (2010) 58.

[10] S.S.-Y. Lin, D.H. Kim, S.Y. Ha, Catal. Lett. 122 (2008) 295.[11] J. Llorca, N. Homs, J. Sales, P.R. de la Piscina, J. Catal. 209 (2002) 306.[12] J. Llorca, N. Homs, P.R. de la Piscina, J. Catal. 227 (2004) 556.[13] S.S.-Y. Lin, D.H. Kim, S.Y. Ha, Appl. Catal. A 355 (2009) 69.[14] M. Mavrikakis, M.A. Barteau, J. Mol. Catal. A: Chem. 131 (1998) 135.[15] M.P. Hyman, J.M. Vohs, Surf. Sci. 605 (2011) 383.[16] J. Llorca, J.-A. Dalmon, P.R. de la Piscina, Appl. Catal. A 243 (2003) 261.[17] C.J. Westrate, H.J. Gericke, M.W.G.M. Verhoeven, I.M. Ciobica, A.M. Saib, J.W.

Niemantsverdriet, J. Phys. Chem. Lett. 1 (2011) 1767.[18] A. Yee, S.J. Morrison, H. Idriss, J. Catal. 186 (1999) 279.

[19] S.M. de Lina, L.O.O. da Silva, U.M. da Costa, G. Graham, B.H. Jacobs, L.V. Mattos,F.B. Noronha, J. Catal. 268 (2009) 268.

[20] J. Rasko, M. Dömök, K. Baan, A. Erdöhelyi, Appl. Catal. A 299 (2006) 202.[21] Q. Fu, T. Wagner, Surf. Sci. Rep. 62 (2007) 431.[22] J. Libuda, H.-J. Freund, Surf. Sci. Rep. 57 (2005) 157.[23] J.A. Rodiguez, J. Graciana, J. Evans, J.B. Park, F. Yang, D. Stacchiola, S.D.

Senanayake, S. Ma, M. Perez, P. Liu, J.F. Sanz, J. Hrbek, Angew. Chem. Int. Ed. 48(2009) 8047.

[24] F. Dvorak, O. Stesovych, M. Steger, E. Cherradi, I. Matolinova, N. Tsud, M. Skoda,T. Skala, J. Myslivecek, V. Matolin, J. Phys. Chem. C 115 (2011) 7496.

[25] V. Matolin, J. Libra, I. Matolinova, V. Nehasil, L. Sedlacek, F. Sutara, Appl. Surf.Sci. 254 (2007) 153.

[26] V. Matolin, L. Sedlacek, I. Matolinova, F. Sutara, T. Skala, B. Smid, J. Libra, V.Nehasil, K. Prince, J. Phys. Chem. C 112 (2008) 3751.

[27] F. Sutara, M. Cabala, L. Sedlacek, T. Skala, M. Skoda, V. Matolin, K. Prince, V.Chab, Thin Solid Films 516 (2008) 6120.

[28] T. Staudt, Y. Lykhach, L. Hammer, M.A. Schneider, V. Matolin, J. Libuda, Surf.Sci. 603 (2009) 3382.

[29] R. Wrobel, Y. Suchorski, S. BEcker, H. Weiss, Surf. Sci. 602 (2008) 436.[30] A. Siokou, R.M. Nix, J. Phys. Chem. B 103 (1999) 6984.[31] M. Alexandrou, R.M. Nix, Surf. Sci. 3321 (1994) 47.[32] D.C. Grinter, R. Ithnin, C.L. Pang, G. Thornton, J. Phys. Chem. C 114 (2010)

17036.[33] J. Pantförder, J.F. Zhu, D. Borgmann, R. Denecke, H.-P. Steinrück, Rev. Sci.

Instrum. 76 (2005) 14102.[34] K. Dumbuya, G. Cabailh, R. Lazzari, J. Jupille, L. Ringle, M. Pistor, O. Lytken, H.-P.

Steinrück, J.M. Gottfried, Catal. Today 181 (2012) 20.[35] P. Burroughs, A. Hamnett, A.F. Orchard, G. Thorton, J. Chem. Soc., Dalton Trans.

17 (1976) 1686.[36] A. Pfau, K.D. Schierbaum, Surf. Sci. 321 (1994) 71.[37] A. Fujimori, Phys. Rev. B 28 (1983) 4489.[38] D.R. Mullins, S.H. Overbury, D.R. Huntley, Surf. Sci. 409 (1998) 307.[39] C.J. Nelin, P.S. Bagus, E.S. Ilton, S.A. Chambers, H. Kuhlenbeck, H.-J. Freund, Int.

J. Quantum Chem. 110 (2010) 2752.[40] H.-Y. Li, H.-F. Wang, Y.-L. Guo, G.-Z. Lu, P. Hu, Chem. Commun. 47 (2011) 6105.[41] M. Abu-samha, K.J. Borve, L.J. Seathre, T.D. Thomas, Phys. Rev. Lett. 91 (2005)

103002.[42] M.A. Henderson, C.L. Perkins, M.H. Engelhard, S. Thevuthasan, C.H.F. Peden,

Surf. Sci. 526 (2003) 1.[43] V. Matolin, I. Matolinova, F. Dvorak, V. Johanek, J. Myslivecek, K. Prince, T.

Skala, O. Stetsovich, N. Tsud, M. Vaclavu, B. Smid, Catal. Today 181 (2012) 124.[44] Y. Lykhach, V. Johanek, H.A. Aleksandrov, S.M. Kozlov, M. Happel, T. Skala, P.S.

Petkov, N. Tsud, G.N. Vayssilov, K. Prince, K.M. Neymann, V. Matolin, J. Libuda,J. Phys. Chem. C 116 (2012) 12103.

[45] V. Matolin, J. Libra, M. Skoda, N. Tsud, K. Prince, T. Skala, Surf. Sci. 603 (2009)1087.

[46] D.R. Mullins, S.D. Senanayake, T.-L. Chen, J. Phys. Chem. C 114 (2010) 17112.[47] T.-L. Chen, D.R. Mullins, J. Phys. Chem. C 115 (2011) 3385.[48] H. Idriss, C. Diagne, J.P. Hindermann, A. Kiennemann, M.A. Barteau, J. Catal. 155

(1995) 219.[49] J. Rasko, J. Kiss, Appl. Catal. A 287 (2005) 252.[50] E. Martono, J.M. Vohs, J. Catal. 291 (2012) 79.[51] S.S.-Y. Lin, D.H. Kim, M.H. Engelhard, S.Y. Ha, J. Catal. 273 (2010) 229.[52] C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E. Muilenberg, Handbook

of X-Ray Photoelectron Spectroscopy, Perkin-Elmer Corp., Eden Prairie,Minnesota, USA, 1979.

[53] A.F. Lee, D.E. Gawthorpe, N.J. Hart, K. Wilson, Surf. Sci. 548 (2004) 200.[54] J.L. Davis, M.A. Barteau, Surf. Sci. 235 (1990) 235.[55] H.-P. Steinrück, F. Pesty, L. Zhang, T. Madey, Phys. Rev. B 51 (1995) 5427.[56] L. Ovari, J. Kiss, Appl. Surf. Sci. 252 (2006) 8624.