Selective oxidation of ethanol over vanadia-based catalysts: The influence of support material and reaction mechanism

Catalysis Today 279 (2017) 95–106

Contents lists available at ScienceDirect

Catalysis Today journal homepage: www.elsevier.com/locate/cattod

Selective oxidation of ethanol over vanadia-based catalysts: The influence of support material and reaction mechanism T.V. Andrushkevich a , V.V. Kaichev a,b,c,∗ , Yu.A. Chesalov a,b,c , A.A. Saraev a,b,c , V.I. Buktiyarov a,b a

Boreskov Institute of Catalysis, Lavrentiev Ave. 5, 630090 Novosibirsk, Russia Novosibirsk State University, Pirogova Str. 2, 630090 Novosibirsk, Russia c Research and Educational Center for Energy Efficient Catalysis in Novosibirsk State University, Pirogova Str. 2, 630090 Novosibirsk, Russia b

a r t i c l e

i n f o

Article history: Received 8 January 2016 Received in revised form 25 April 2016 Accepted 27 April 2016 Available online 28 May 2016 Keywords: Heterogeneous catalysis Ethanol oxidation Acetaldehyde Acetic acid Vanadia

a b s t r a c t The catalytic performance of vanadia supported on silica, alumina, zirconia, and titania was investigated in the selective oxidation of ethanol. It was shown that the activity and product distribution strongly depend on the support material, which determines the structure of supported vanadia species. On silica and alumina, low-active V2 O5 crystallites were mainly formed regardless of the vanadium content. These catalysts demonstrated high selectivity toward only acetaldehyde. In contrast, monomeric surface vanadia species and polymeric surface vanadia species were mainly formed over TiO2 when the vanadium content did not exceed what is necessary for the ideal monolayer. Over zirconia, both the surface vanadia species and the V2 O5 crystallites existed regardless of the vanadium content. It was found that the surface vanadia species are more active in the selective oxidation of ethanol than the V2 O5 crystallites. The highest activity was observed for the polymeric vanadia species and, correspondingly, the best catalytic performance was achieved on the monolayer V2 O5 /TiO2 catalyst. At low temperatures between 110 and 150 ◦ C, this catalyst demonstrated high activity in the oxidation of ethanol to acetaldehyde with the selectivity ranging between 80% and 100%. At temperature near 200 ◦ C, the same catalyst was active in the oxidation of ethanol to acetic acid with the selectivity of approximately 65%. The surface intermediates and the catalyst state were also studied in situ by X-ray photoelectron spectroscopy and Fourier transform infrared spectroscopy. It was shown that under reaction conditions near 100 ◦ C, non-dissociatively adsorbed molecules of ethanol, ethoxide species, and adsorbed acetaldehyde exist on the catalyst surface, while at higher temperatures, V2 O5 /TiO2 is mainly covered with acetate species. Titanium cations remained in the Ti4+ state, whereas V5+ cations underwent a reversible reduction under reaction conditions. On the basis of the in situ data complemented by the results of kinetic measurements, a reaction mechanism for the selective oxidation of ethanol to acetaldehyde and acetic acid over the monolayer catalysts was proposed. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Acetaldehyde and acetic acid are important reagents and industrial chemicals. The global demand for acetic acid, which is mainly used for the synthesis of vinyl acetate, acetic anhydride, and acetates, as well as a solvent for the production of purified terephthalic acid, is above 10 million tons per year [1]. Acetaldehyde is

∗ Corresponding author at: Boreskov Institute of Catalysis, Lavrentiev Ave. 5, 630090 Novosibirsk, Russia. E-mail address: [email protected] (V.V. Kaichev). http://dx.doi.org/10.1016/j.cattod.2016.04.042 0920-5861/© 2016 Elsevier B.V. All rights reserved.

also an industrially important solvent and an intermediate for the synthesis of a wide range of organic compounds, such as pentaerythritol, crotonaldehyde, and pyridine derivatives. The global market for acetaldehyde is forecast to reach 1.2 million tons by the year 2015. Currently, approximately 85% of acetaldehyde is produced from ethylene via the Wacker-Hoechst process, 75% of acetic acid for the chemical industry is produced by the catalytic carbonylation of methanol and 25% by classical fermentation [2]. All these processes are liquid-phase, and the development of more effective gas-phase technologies is one of the most important tasks of the large-scale chemical industry.

96

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

To solve this task, the catalytic gas-phase oxidation of ethanol by molecular oxygen (from air) is more attractive because it is an economic and environmentally friendly process. Indeed, acetaldehyde, which is more expensive than ethanol, can be produced with high effectiveness by the gas-phase oxidative dehydrogenation (ODH) over supported transitional metal catalysts or by oxidation with O2 over different vanadium and molybdenum based oxides [3–7]. A number of heterogeneous catalysts have been also reported for the formation of acetic acid by the oxidation of ethanol in the gas phase: Mo0.61 V0.31 Nb0.08 Ox /TiO2 [7], Ce-meso TiO2 [8], V2 O5 /TiO2 [9], MoCeOx /SnO2 [10], Vx M1-x SbO4 (M = Fe, Al, Ga) [11], and Pt/Al2 O3 [12]. Definitely, the development of new catalytic technologies requires the understanding of the mechanism for the selective oxidation of ethanol on the atomic level. The oxide catalysts demonstrate high activity in the selective oxidation of ethanol to both acetaldehyde and acetic acid under mild conditions. The high selectivity toward acetaldehyde is usually observed at low temperatures ranging between 100 and 200 ◦ C, while at 200–250 ◦ C, because of the further oxidation of acetaldehyde, the reaction shifts toward acetic acid. For example, over Ce-meso TiO2 catalysts, the selectivity toward acetaldehyde achieves 93% at 150 ◦ C at the conversion of ethanol of 77% [8]. Similar selectivity toward acetaldehyde was observed over MoOx /TiO2 at 200 ◦ C but at low conversions near 17% [7]. Over V0.7 M0.3 SbO4 (M = Fe, Al, Ga), the direct oxidation of ethanol to acetaldehyde with the selectivity above 80% was observed in a wide temperature range from 150 to 230 ◦ C [11]. Multi-component metal oxides (Mo-V-NbOx ) can catalyze the direct oxidation of ethanol to acetic acid at 240 ◦ C with the selectivity up to 95% at 100% ethanol conversion [7]. A special attention has been paid to supported vanadia catalysts, which demonstrate excellent catalytic performance in the selective oxidation of ethanol [9]. Depending on reaction conditions, ethanol can be transformed to acetaldehyde, acetic acid, diethyl ether, ethyl acetate, ethyl formate, crotonaldehyde, ethylene, or carbon oxides (CO and CO2 ). The product distribution is also dependent on the support material and vanadia surface density [10,13–17]; however, the reason for this effect is still a topic of debate. Recently, we have shown that the catalytic performance of supported vanadia catalysts in the selective oxidation of methanol to dimethoxymethane and methyl formate is mainly determined by the structure of vanadia species [18,19]. Herein, we demonstrate that this hypothesis is also applicable for the selective oxidation of ethanol. It is well known that vanadia can form different structures over the surface of oxide supports: monomeric and polymeric species as well as crystallites of vanadium oxide. As shown by Kilos et al. [14], the acetaldehyde formation rate during the oxidation of ethanol over VOx /Al2 O3 catalysts depends on the vanadia surface density. Polyvanadate surface species supported on alumina exhibit a somewhat higher rate of ethanol ODH than monovanadate surface species. In contrast, according to DFT calculations [20], vanadylterminated monomers on CeO2 (111), that is, VO2 , are the most active species. VOx components in the Mo0.61 V0.31 Nb0.08 Ox /TiO2 catalyst [7] are responsible for the high reactivity of this material in the selective oxidation of ethanol to acetaldehyde and acetic acid while V-O-Ti linkages are responsible for the unselective oxidation of acetaldehyde to COx . The contrary point of view was stated by Beck el al. [21] who suggested that the V O S linkages (where S is a cation of a support) take part in the selective oxidation of ethanol to acetaldehyde over vanadia supported on CeO2 , Al2 O3 , ZrO2 , and TiO2 . The authors suggested that the alcohol first adsorbs dissociatively, resulting in a breaking of the V O S bond to form ethoxide and S-OH species. In order to elucidate these conflicting data and to develop the mechanism for the selective oxidation of ethanol, we synthesized a series of V2 O5 /SiO2 , V2 O5 /Al2 O3 , V2 O5 /ZrO2 , and V2 O5 /TiO2 catalysts and tested their catalytic activity in a flow reactor in a wide

temperature range. In addition, we carried out an in situ study of the oxidation of ethanol over the most active monolayer V2 O5 /TiO2 catalyst using Fourier transform infrared spectroscopy (FTIR) and near ambient-pressure X-ray photoelectron spectroscopy (XPS). 2. Experimental 2.1. Catalyst preparation Supported vanadia catalysts containing 2–22 wt% V2 O5 were prepared by the impregnation of supports (SiO2 , ␥-Al2 O3 , ZrO2 , and TiO2 ) with an aqueous solution of vanadyl oxalate synthesized from V2 O5 (>99.6%, Reachim, Russia) and oxalic acid (>97%, Reachim, Russia). The samples were dried in air at 110 ◦ C for 24 h and then calcined in an air flow (50 ml/min) at 400 ◦ C for 4 h. As support materials, we used commercial aerosil SiO2 (>99.6%, Reachim, Russia) with the specific surface area SBET of 200 m2 /g and TiO2 (anatase, AlfaAesar) with SBET of 350 m2 /g. ␥-Al2 O3 with SBET = 250 m2 /g was synthesized by the calcination of boehmite AlOOH·nH2 O (n = 0.3–1.0) in air at 550 ◦ C for 4 h. ZrO2 was prepared by the precipitation of Zr(OH)4 from a ZrOCl2 solution with aqueous ammonia at 50 ◦ C; final pH was 8.5. The resulting zirconium hydroxide was dried in air at 110 ◦ C for 12 h and then calcined at 400 ◦ C for 4 h. Synthesized zirconia was a mixture of the monoclinic (85%) and cubic (15%) phases. The specific surface area of ZrO2 , which was calculated by the Brunauer–Emmett–Teller (BET) method, was 120 m2 /g. 2.2. Catalyst characterization The catalysts were characterized by elemental analysis, Raman spectroscopy, N2 adsorption and X-ray diffraction (XRD) techniques. The elemental analysis was performed using an inductively coupled plasma atomic emission spectrometer (Baird). Powder XRD measurements were carried out using a Siemens D500 diffractometer using monochromatic CuK˛ radiation. The 2␪ scan covered a range of 10–70◦ . The specific surface area was calculated with the BET method using nitrogen adsorption isotherms measured at liquid nitrogen temperatures with an automatic Micromeritics ASAP 2400 sorptometer. Raman spectra were obtained on a RFS 100/S Raman spectrometer (Bruker) using a Nd:YAG laser as an excitation source (␭ = 1064 nm, 100 mW). The laser radiation was focused onto a spot with a diameter 50 ␮m. Before recording the spectra, the samples were calcined in air at 400 ◦ C for 30 min. 2.3. Catalytic testing The steady-state activity of the catalysts was tested at atmospheric pressure in a differential reactor with a flow-circulating configuration [22]. The reactor was constructed from a Pyrex glass tube with a 12-mm inner diameter and a 50-mm length. A coaxial thermocouple pocket with a 4-mm outer diameter was fitted in the catalyst bed to control the temperature. The reactor was placed inside an electric oven. The temperature was controlled within ±0.5 ◦ C by a K-type thermocouple. The feed consisted of ethanol, oxygen, and nitrogen in the molar ratios of 1:4:15 (5 vol.% C2 H5 OH in air). The catalyst fraction 0.25–0.50 mm was used in the experiments. Concentrations of the reactants and products were determined with an on-line gas chromatograph equipped with thermal conductivity and flame ionization detectors. Ethanol, acetaldehyde (CH3 CHO), acetic acid (CH3 COOH), diethyl ether ((C2 H5 )2 O), ethyl acetate (CH3 COO CH2 CH3 ), crotonaldehyde (CH3 CH CHCHO), ethylene, water, and CO2 were analyzed with a Porapak T column, while CO, oxygen, and nitrogen were analyzed with a NaA molecular sieve column. All gas lines from the reactor to

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

97

a sampling valve were maintained at 120 ◦ C to prevent the condensation of reactants and products. Ethanol (A.C.S. reagent grade, 99% purity) obtained from Aldrich was used in all the experiments. The selectivity toward each product was calculated as the amount of the detected product divided by the amount of converted ethanol using corresponding stoichiometric coefficients. The carbon balance was 97 ± 2%. 2.4. In situ XPS and FTIR measurements In situ XPS experiments were performed at the Innovative Station for In Situ Spectroscopy (ISISS) beamline at the synchrotron radiation facility BESSY II (Berlin, Germany) using the monolayer V2 O5 /TiO2 catalyst. The experimental station was described in detail elsewhere [23]. In short, the station was equipped with an electron energy analyzer PHOIBOS-150 (SPECS Surface Nano Analysis GmbH), a gas cell, and a system of differential pumping, which allowed us to obtain high-quality core-level spectra at pressures up to 1 mbar. A powder sample was pressed into a thin self-supporting pellet. The pellet was mounted on a sapphire sample holder between two stainless steel plates. The first plate had an aperture of 5 mm in diameter for measuring the core-level spectra of the catalyst surface. The second plate was used for heating by a NIR semiconductor laser (␭ = 808 nm). The sample temperature was measured with a K-type thermocouple pressed directly against the rear of the sample. The flows of ethanol and oxygen into the gas cell were regulated separately with calibrated massflow controllers (Bronkhorst). The flow rate of ethanol in all the experiments was approximately 2 sccm. The total pressure in the gas cell was of 0.25 and 0.5 mbar in the experiments with ethanol and with an equimolar C2 H5 OH/O2 mixture, respectively. The synchrotron worked in the multi bunch hybrid mode that provided a constant photon flux. The C1s, Ti2p3/2 , V2p3/2 , and O1s core-level spectra were recorded at the photon energy of 720 eV. The charge correction was performed by setting the Ti2p3/2 peak at 459.0 eV [24]. The curve-fitting was done with the CasaXPS software. The core-level spectra were resolved into their components after a Shirley-type background subtraction. The line-shape of each component was considered to be a product of Lorentzian and Gaussian functions. To identify reaction intermediates involved in the oxidation of ethanol, FTIR spectra were obtained in situ with a Cary 660 FTIR spectrometer (Agilent Technologies). The experiments were performed using vanadia-titania catalysts that contained 10% and 20% of V2 O5 . The spectrometer was operated in the transmission mode using a specially designed quartz cell-reactor with BaF2 windows. The volume of the cell-reactor was approximately 1.5 cm3 . The catalyst powder (35–50 mg) was pressed into a thin selfsupporting pellet (∼15 mg/cm2 , 1 × 3 cm in size) and placed into the cell-reactor. The FTIR experiments were performed at atmospheric pressure using a feed of 1.5 vol.% C2 H5 OH in air flowing at 50 sccm. Ethanol was dosed by bubbling air through a glass saturator filled with liquid ethanol at 0 ◦ C. As a result, the molar ratios C2 H5 OH:O2 :N2 were approximately 1:14:52. Before the exposure to the reactant mixture, the sample was treated in a flow of air at 250 ◦ C for 1 h. Subsequently, the cell-reactor and the catalyst sample were cooled to a desired temperature, and the air flow was replaced with the flow of the ethanol/air mixture. The FTIR spectra were recorded in a range of 1100–4000 cm−1 at a resolution of 4 cm−1 . The spectra of the gas phase were also recorded using a Cary 660 FTIR. In these experiments, a special gas-cell with the optical path length of approximately 70 mm was connected to the outlet of the cell-reactor. Bands at 1065 cm−1 (Q branch of (C O) band), 1760 cm−1 (P branch of (C O) band), 1790 cm−1 (P branch of (C O) band), 2115 cm−1 (R branch of (C O) band), and

Fig. 1. XRD patterns of bulk oxides SiO2 and Al2 O3 as well as V2 O5 /SiO2 and V2 O5 /Al2 O3 supported catalysts.

2360 cm−1 (P branch of (CO2 ) band) were used for the detection of ethanol, acetaldehyde, acetic acid, CO, and CO2 , respectively. The concentrations of surface intermediates and gas-phase products were measured in separate experiments carried out under the same conditions. 3. Results and discussion 3.1. Structure of catalysts In full agreement with previous studies (see Ref. [24] and refs therein), we have found that vanadia over SiO2 , Al2 O3 , ZrO2 , and TiO2 can form the monomeric vanadia species VOx , the polymeric vanadia species (VOx )n , and crystallites of vanadium oxide. The relative concentrations of these vanadia forms depend on the support material and the vanadium content. The main characteristics of the catalysts under study and their designations are summarized in Table 1. Because of a weak vanadia-support interaction, the surfaces of alumina and silica contain only V2 O5 crystallites, regardless of the vanadium content [18,25]. The XRD patterns of these catalysts exhibit well-defined reflections of the V2 O5 phase (Fig. 1), whereas the core-level spectra of these catalysts demonstrate low V/Si and V/Al atomic ratios even at the high vanadium content [18]. This finding clearly indicates the low concentration of the surface vanadia species. Moreover, in the Raman spectra, only two sharp peaks at 702 and 994 cm−1 were observed (spectra not shown), which are typical of crystalline V2 O5 [26]. The peaks were assigned to lattice vibrations of the orthorhombic polymorph of V2 O5 [18]. In contrast, in the V2 O5 /TiO2 catalysts, mainly the surface vanadia species were observed. According to the literature [27–29], due to strong vanadia-suport interaction, at the vanadia content under 10% of a monolayer (ML), unhydrous TiO2 contains only isolated monomeric species with a tetrahedral coordination. Polymeric structures such as chains and ribbons of VOx units with an octahedral coordination appear at the vanadia concentration above 20% of the monolayer [28]. When the vanadium content exceeds what is necessary for the ideal monolayer, V2 O5 crystallites are favorable. These crystallites are characterized by weak bonding to TiO2 [29]. The monolayer coverage of polymerized vanadia species on different oxides was measured by Raman spectroscopy, and for TiO2 , it was found to be approximately 7.9 vanadium atoms/nm2 [30]. In full agreement with this model, the V2 O5 phase was detected by XRD in the 20VTi catalyst, which contained approximately 1.5 ML of vanadia, and only reflections of TiO2 (anatase) were observed in the XRD patterns of the 2VTi and 10VTi catalysts in which the sur-

98

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

Table 1 Characteristics of catalysts under study. Sample

V2 O5 content, wt%

SBET , m2 /g

Surface concentration of vanadiuma , atom V/nm2

Phase composition

Relative concentration of vanadia formsb

15VSi 20VAl 20VTi 10VTi 2VTi 17VZr 7VZr 3.4VZrc

15 22 20 10 2.0 17 7.1 3.4

134 146 111 153 135 96 120 122

10.3 9.9 11.7 4.2 0.9 11.9 3.9 1.9

V2 O5 , SiO2 V2 O5 , ␥-Al2 O3 V2 O5 ,TiO2 (anatase) TiO2 (anatase) TiO2 (anatase) V2 O5 , ZrO2 V2 O5 , ZrO2 ZrO2

100% V2 O5 ∼100% V2 O5 ∼30% V2 O5 ∼70% (VOx )n 100% (VOx )n 100% VOx ∼80% V2 O5 ∼20% (VOx )n ∼30% V2 O5 ∼70% (VOx )n 100% VOx

a b c

Refers to the specific surface area. The data obtained from the results of Raman spectroscopy and XRD study. In this catalyst, the V2 O5 phase was removed selectively by washing in HNO3 [19].

Fig. 3. XRD patterns of bulk ZrO2 and V2 O5 /ZrO2 supported catalysts.

Fig. 2. Raman spectra of bulk oxides V2 O5 (1) and TiO2 (2) as well as supported catalysts 2VTi (3), 10VTi (4), and 20VTi (5).

face concentration of vanadium was 0.11 and 0.53 ML, respectively [18]. The presence of the surface vanadia species in the 2VTi and 10VTi catalysts was confirmed by Raman spectroscopy. The spectra of the 2VTi, 10VTi, and 20VTi catalysts, as well as the spectra of bulk oxides V2 O5 and TiO2 , are presented in Fig. 2. One can see that in the range between 550 and 1100 cm−1 , the spectrum of TiO2 contains only a single sharp peak at 640 cm−1 . The spectra of supported catalysts contain an extra broad peak near 830 cm−1 , which can be attributed to the surface vanadia species. Indeed, broad peaks near 840–940 cm−1 , which were assigned to the monomeric and polymeric vanadia species, have been observed in similar systems [31–34]. The spectrum of V2 O5 contains two sharp peaks at 700 and 994 cm−1 . The first peak corresponds to lattice vibrations localized within the V O V bridge in the V2 O5 structure, while the second peak corresponds to the stretching vibration of vanadyl V O bonds [35]. The spectrum of the 20VTi catalyst contains weak peaks at 700 and 994 cm−1 , indicating the formation of bulk vanadium oxide over the titania surface at the high vanadia content. Because of a specific interaction between vanadia and zirconia, all the vanadia species mentioned above are formed over ZrO2 regardless of the vanadium content. This question was considered in detail elsewhere [18,19]. Indeed, the V2 O5 phase was detected by XRD in both catalysts 7VZr and 17VZr in which the surface concentration of vanadium was approximately 0.6 and 1.8 ML, respectively (Fig. 3). These values were estimated on the basis of previous Raman studies [30], where the monolayer coverage of surface vanadia

species on ZrO2 was found to be 6.8 vanadium atoms/nm2 . The Raman spectra of the 7VZr and 17VZr catalysts contain two peaks at 617 and 632 cm−1 due to the lattice vibrations of the monoclinic polymorph of ZrO2 , and two peaks at 702 and 994 cm−1 due to crystalline V2 O5 . In addition, the Raman spectra of these catalysts exhibit two peaks at 930 and 1030 cm−1 due to a bridge vibration of V O M (M = V or Zr) and a stretching vibration of vanadyl V O bonds in the surface vanadia species, respectively [18,19]. The relative intensity of the peaks at 994 and 1030 cm−1 allowed us to estimate the relative concentration of different vanadia forms. Correspondingly, only 20% of vanadia exists in the form of surface vanadia species in 17VZr, whereas in 7VZr, the part of surface vanadia is approximately 70% (Table 1). In addition, the 3.4VZr catalyst, which contains only surface vanadia species, was prepared by washing the 7VZr catalyst in an aqua solution of HNO3 . Earlier, we have shown that such treatment removes V2 O5 crystallites, while the surfaces vanadia species remain unchanged [24]. Indeed, no reflections of the V2 O5 phase were observed in the XRD pattern after this treatment (Fig. 3). 3.2. Catalytic performance The catalytic properties of the catalysts were tested at atmospheric pressure in the temperature range between 130 and 250 ◦ C. It was found that, depending on the reaction temperature and the contact time, ethanol can be transformed with different selectivity to acetaldehyde (AA), acetic acid (AcA), diethyl ether (DEE), ethyl acetate (EA), crotonaldehyde (CA), ethylene (C2 H4 ), and carbon oxides (COx ). The results of the catalytic tests of the V2 O5 /SiO2 and V2 O5 /Al2 O3 catalysts are presented in Fig. 4. Both catalysts demonstrate high selectivity toward acetaldehyde even at low

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

99

Fig. 4. Effect of reaction temperature on the conversion of ethanol and the selectivity to acetaldehyde, acetic acid, ethyl acetate, and carbon oxides over V2 O5 /SiO2 and V2 O5 /Al2 O3 catalysts.

Fig. 5. Effect of reaction temperature on the conversion of ethanol and the selectivity to acetaldehyde, acetic acid, ethyl acetate, and carbon oxides over V2 O5 /TiO2 and V2 O5 /ZrO2 catalysts with high vanadia content (17VZr and 20VTi).

temperatures. Over the V2 O5 /SiO2 catalyst, the selectivity toward acetaldehyde achieves 98% at 150 ◦ C. An increase in the reaction temperature leads to an increase in the rate of the oxidation of ethanol over both catalysts, which is accompanied with a decrease in the selectivity toward acetaldehyde and results in the appearance of the other products mentioned above (Table S1 in the Supplementary data). The conversion of ethanol also increases with the reaction temperature even at the same contact time. It should be noted that the V2 O5 /SiO2 and V2 O5 /Al2 O3 catalysts demonstrate unsatisfactory activity in the selective oxidation of ethanol to acetic acid. For example, in both cases, the selectivity toward acetic acid does not exceed 5% at 200 ◦ C even at the conversion of ethanol above 70%. The maximal selectivity toward acetic acid achieves 22.6% over V2 O5 /SiO2 at 250 ◦ C, which is even lower than the selectivity toward acetaldehyde. At higher temperatures, the major products are carbon oxides with the total selectivity above 50%. The V2 O5 /Al2 O3 catalyst demonstrates similar catalytic performance apart of higher selectivity toward ethyl acetate, which achieves approximately 18% at 250 ◦ C. The results of the catalytic tests of the V2 O5 /TiO2 and V2 O5 /ZrO2 catalysts with high vanadia content are presented in Fig. 5. These catalysts also demonstrate high selectivity toward acetaldehyde at low temperatures, which achieves approximately 90% at 130 ◦ C. An increase in the reaction temperature leads to an increase in the conversion of ethanol and to changes in the product distribution.

Near 200 ◦ C, the main product is acetic acid. The maximal selectivity toward acetic acid is approximately 62% over the V2 O5 /TiO2 catalysts at 220 ◦ C. At higher temperatures, the reaction shifts to the nonselective oxidation of ethanol to carbon oxides. The selectivity toward ethylene, diethyl ether, and crotonaldehyde does not exceed 0.7%, 1.3%, and 7.2%, respectively (Table S2). The maximal selectivity toward ethyl acetate achieves 14.5% over V2 O5 /TiO2 at 180 ◦ C. The V2 O5 /TiO2 and V2 O5 /ZrO2 catalysts are more active in the selective oxidation of ethanol to acetaldehyde than V2 O5 /SiO2 and V2 O5 /Al2 O3 . At 150 ◦ C, the rate of oxidation of ethanol normalized to the number of supported vanadium atoms is 0.34 and 1.24 mmol/s over V2 O5 /ZrO2 and V2 O5 /TiO2 (Table S2), respectively, while over V2 O5 /SiO2 and V2 O5 /Al2 O3 the rate is distinctly lower (Table S1). In comparison with the rate over the V2 O5 /TiO2 catalyst, the rate over V2 O5 /SiO2 and V2 O5 /Al2 O3 is 10–20 times lower. The catalytic activity increases with the reaction temperature. Over the more active V2 O5 /TiO2 catalyst, the normalized rate of oxidation of ethanol increases from 0.34 to 4.1 mmol/s when the reaction temperature increases from 130 to 230 ◦ C. It should be noted that the rate of the selective oxidation of ethanol over bare supports is lower by 2 orders of magnitude than that over vanadia supported catalysts. Moreover, acetic acid is not detected among the products of the oxidation of ethanol over bare supports in the entire temperature range. It means that the selective oxidation of

100

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

Fig. 7. in situ FTIR spectra of the 10VTi catalyst obtained at different temperatures under the flow of the following mixtures: 1.5 vol.% C2 H5 OH in air (a) and 1.5 vol.% C2 H5 OH in He (b).

Fig. 6. Rate of the oxidation of ethanol vs. the surface concentration of vanadium atoms over V2 O5 /ZrO2 (1) and V2 O5 /TiO2 (2) catalysts at the reaction temperature of 150 ◦ C and conversion of ethanol of 40% (a) and at 200 ◦ C and conversion of ethanol of 70% (b).

ethanol to both acetaldehyde and acetic acid proceeds mainly on supported vanadia species. To compare the activity of different vanadia forms, we tested the V2 O5 /TiO2 and V2 O5 /ZrO2 catalysts with different vanadia contents in the selective oxidation of ethanol. The experiments were performed at 150 and 200 ◦ C, when the main products were acetaldehyde and acetic acid, respectively. The results are presented in Fig. 6. In these experiments, we used three types of the catalysts with the coverage of vanadia species near 1–2, 4, and 12 vanadium atoms per nm2 . In accordance with the XRD and Raman spectroscopy results (Table 1), these catalysts contain mainly the isolated monomeric species, the polymerized vanadia species, and a mixture of the polymerized vanadia species and the V2 O5 crystallites, respectively. One can see that the catalysts with the coverage of vanadia species near 4 V atom/nm2 demonstrate the maximal activity in the oxidation of ethanol to both acetaldehyde and acetic acid (Fig. 6). It indicates that the surface polymeric vanadia species have higher activity in the selective oxidation of ethanol to acetaldehyde and acetic acid than the isolated monomeric species and the V2 O5 crystallites. This finding agrees well with a conclusion by Kilos et al. [14], who showed that polyvanadate domains over ␥-Al2 O3 are more reactive in the oxidative dehydrogenation of ethanol to acetaldehyde than monovanadate structures. Summarizing the catalytic data, we can conclude that the catalytic performance of the vanadia-based catalysts in the selective oxidation of ethanol is determined by the vanadia-support interaction. For alumina and silica, the vanadia-support interaction is weak and, as a result, mainly the low-active V2 O5 crystallites are formed on the catalyst surface regardless of the vanadium content. Definitely, the effective surface of vanadia in this case is also low. In contrast, the chemical properties of the titania surface provide the strong vanadia-support interaction and, as a result, the V2 O5 crystallites start to form only after completion of the monolayer coverage. Zirconium oxide is characterized by intermediate strength of the vanadia-support interaction, and thus its surface contains a mixture of the surface vanadia species and the V2 O5 crystallites regardless of the vanadium content.

Table 2 shows the results of the catalytic tests of more active sub-monolayer 2VTi and 10VTi catalysts, containing 0.11 and 0.53 ML of vanadia, respectively. On the one hand, these data illustrate in more detail the data presented in Fig. 6 to confirm our conclusion about the higher activity of the polymeric vanadia species in the selective oxidation of ethanol in comparison with the isolated monomeric species. On the other hand, one can see that the selectivity toward acetaldehyde can reach 100% over the catalyst with the sub-monolayer vanadia coverage at 110 ◦ C. An increase in the reaction temperature leads to a decrease in the selectivity toward acetaldehyde, which is accompanied with an increase in the conversion of ethanol. Simultaneously, the selectivity toward acetic acid also increases, achieving 65% at 200 ◦ C. It means that the oxidation of ethanol proceeds successively: first, ethanol oxidizes to acetaldehyde and then acetaldehyde oxidizes to acetic acid. This question is considered in detail elsewhere [13]. Hence, the monolayer vanadia-titania catalysts are perspective for the selective oxidation of ethanol to acetaldehyde at low temperatures in the range of 100–150 ◦ C, and these catalysts can be used for the selective oxidation of ethanol to acetic acid at higher temperatures near 200 ◦ C. 3.3. Mechanistic study To understand clearly why different vanadia species have different reactivity in the selective oxidation of ethanol, the elucidation of the reaction mechanism is needed. The application of in situ FTIR spectroscopy has allowed the identification of the key intermediates formed on the catalyst surface during the reaction. Then, we studied the catalyst state directly in pure ethanol and in the ethanol/O2 mixture using in situ X-ray photoelectron spectroscopy. First, we performed an in situ FTIR study using the 10VTi and 20VTi catalysts. Fig. 7 shows the FTIR spectra for the 10VTi catalyst collected at 100, 150, and 230 ◦ C under the flow of two reactant mixtures. The reactant mixtures contained 1.5 vol.% C2 H5 OH in air and 1.5 vol.% C2 H5 OH in He. The spectrum from the catalyst before the exposure to the reaction mixture and the spectrum from gasphase ethanol were subtracted from the raw spectra to identify the contributions of the adsorbed surface species. In the spectrum obtained at 100 ◦ C in the C2 H5 OH/air flow (Fig. 7a), positive bands appear at 2976 (as (CH3 )), 2933 (as (CH2 )), 2876 (s (CH2 )), 1730 ((C O)), 1678 ((C O)), 1532 (as (COO)), 1444 (s (COO)), 1383 (␦s (CH3 )), 1267 (␦(OH)), 1144 (␳(CH3 )), and 1090 ((CCO)) cm−1 . The assignment of the bands is based on the

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

101

Table 2 Conversion of ethanol, selectivity toward main products, and rate of oxidation of ethanol over sub-monolayer V2 O5 /TiO2 catalysts. T, ◦ C

t,g s mL−1

X, %

r, C2 Н5 OН mmol N−1 s−1

Selectivity, % AA

2VTi 150 180 200

1.02 1.02 1.02

35.7 65.1 73.3

73.3 55.7 31.6

10VTi 110 130 150 150 180 180 200

0.38 0.38 0.12 0.38 0.12 0.38 0.05

8.3 21.5 40.0 66.9 70.5 90.0 72.0

100 89.6 82.0 67.9 39.9 33.5 20.5

COx

EA

C2 Н4

CA

DEE

0 4.06 5.72

9.47 18.8 56.3

8.17 10.4 4.02

0.21 0.28 0.51

8.51 10.5 1.74

0.37 0.26 0.18

3.30 5.48 8.23

0 0 3.25 6.89 35.5 44.3 65.0

0 0 5.11 7.26 9.0 10.2 10.7

0 3.02 5.02 6.64 11.5 9.37 7.2

0 0.11 0.2 0.04 0.10 0.19 0.25

0 7.31 7.10 10.8 3.5 2.12 2.6

0 0 0.3 0.48 0.50 0.44 0.55

0.44 1.15 6.58 3.6 14.6 4.8 32.3

AcA

Fig. 8. Infrared-absorption intensities of ethanol and the main detected products − acetaldehyde, acetic acid, CO, and CO2 (a, b) as well as the main reaction intermediates adsorbed on the catalyst surface measured during the oxidation of ethanol on the 10VTi catalyst as a function of temperature. The data (a, c) are obtained in the ethanol/air mixture; the data (b, d) are obtained in the ethanol/He mixture.

literature data [36–40]. The ethoxide species (C2 Н5 O− V+n ) are the main adsorbed form of ethanol at 100 ◦ C. They are generated as a result of the dissociative adsorption of ethanol. The bands at 2976, 2933, 2876, 1383, 1144, and 1090 cm−1 are assigned to vibrational modes of these species. The weak band at 1267 cm−1 is assigned to ı(OH) vibration of molecularly adsorbed ethanol. The strong bands at 1730 and 1678 cm−1 can be assigned to adsorbed

acetaldehyde [36,37]. Both these bands decrease in intensity with an increase in the reaction temperature. Other bands developed with heating the catalyst could be attributed to adsorbed acetate complexes and acetic acid. Bands at 1532 and 1444 cm−1 could be assigned respectively to as (COO) and s (COO) modes of surface acetate complexes [36–38]. The bands of acetate species progressively increase in intensity with heating to around 200 ◦ C and

102

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

then decrease upon heating between 200 and 250 ◦ C. Molecularly adsorbed acetic acid is represented by the band at 1662 cm−1 due to the carbonyl stretching mode [37]. This band appears in the spectrum at 130 ◦ C and its intensity decreases with the temperature. The temperature dependence of the main bands is shown in Fig. 8c. Fig. 7b presents in situ FTIR spectra obtained at the same manner in the C2 H5 OH/He flow. The spectrum obtained at 100 ◦ C is similar to the spectrum collected at the same temperature in the presence of dioxygen (see Fig. 7a, spectrum 1). The only difference is the appearance of a rather narrow band with the maximum at approximately 1642 cm−1 . The band could be assigned to the C C stretching mode of adsorbed crotonaldehyde [41]. It is well known that the position of the (C O) band in the infrared spectrum of this complex is likely to be very close to the band of adsorbed acetic aldehyde [41]. Crotonaldehyde is the product of acetic aldehyde condensation. Acetate complexes are the main surface species formed under interaction of ethanol with the catalyst at 130–250 ◦ C. The surface concentration of acetates is much higher in the absence of O2 than in its presence (Fig. 8a). Adsorbed acetic acid is not formed in the absence of O2 . A strong band at 1776 cm−1 and a weak band at 1850 cm−1 are also observed in the spectra at 130–250 ◦ C. We suppose that these bands could be correspondingly assigned to asymmetric and symmetric (C O) modes of adsorbed maleic anhydride, which is the product of crotonaldehyde oxidation [41]. The temperature dependence of the intensities of main bands is shown in Fig. 8d. Fig. 8a and b demonstrate the temperature dependence of the concentration of gaseous components (ethanol and products) at the outlet of the IR cell with the 10VTi catalyst. Ethanol concentration decreases after the heating above 100 ◦ C, indicating that ethanol is oxidized under these conditions. The intensity of the acetaldehyde signal is under the detection threshold below 90 ◦ C. At higher temperatures, it strongly increases, achieving a maximum at 150 ◦ C. At higher temperature, acetic acid, CO, and CO2 appear among the products. The yield of acetic acid achieves a maximum near 200 ◦ C, while the yields of CO and CO2 increase with the reaction temperature. Thus, the dynamics of transformation of surface complexes and the changes in the concentration of gas-phase products indicate the consecutive transformation of ethanol into acetaldehyde and acetic acid. In the absence of O2 , acetaldehyde is the only detectable gaseous product of ethanol transformation (see Fig. 8b). It appears at 90 ◦ C, and its concentration increases with the temperature elevation. Acetic acid and carbon oxides are not formed in the absence of O2 in the reactant mixture. The transformation of ethanol on the 20VTi catalyst, which contains the V2 O5 crystallites, was also investigated by in situ FTIR spectroscopy. Fig. 9 demonstrates the FTIR spectra of this catalyst collected at 100, 150, and 230 ◦ C under the flow of the mixture containing 1.5 vol.% C2 H5 OH in air. Fig. 10 demonstrates the temperature dependence of the coverage of the catalyst by the main surface species (a) and the concentration of the main gaseous components at the outlet of the IR cell (b). The ethoxide species and molecularly adsorbed ethanol are detected only in the spectrum recorded at 100 ◦ C. The formation of the ethoxide species is accompanied with a decrease in the intensity of the band assigned to the first overtone of (V O) at 2036 cm−1 (Fig. 9) and with the appearance of a broad band at 3500–2900 cm−1 due to the O H stretching mode of H-bonded hydroxyl groups. Thus, we believe that in this process, the proton from the alcohol hydroxyl group is transferred to the vanadyl oxygen atom. When the temperature rises to 150 ◦ C, the concentration of ethoxide species decreases, and the bands due to vibrational modes of adsorbed acetaldehyde and crotonaldehyde appear in the spectra. With the further increase in the temperature to 230 ◦ C, the bands of acetate complexes dominate in the spectra. The bands of adsorbed maleic anhydride are also observed in the spectrum collected at 230 ◦ C. Acetaldehyde appears in the

Fig. 9. in situ FTIR spectra of the 20VTi catalyst obtained at different temperatures under the flow of the mixture of 1.5 vol.% C2 H5 OH in air. The spectra in the region of vanadyl groups are presented in the inset.

gaseous phase at 120 ◦ C. Its concentration achieves a maximum at 200 ◦ C, while acetic acid and carbon oxides appear in the spectra at 180–220 ◦ C (Fig. 10b). Hence, the same surface intermediates are formed during the oxidation of ethanol on both 10VTi and 20VTi catalysts. However, these surface complexes exhibit a significantly different reactivity. Indeed, on the surface of the monolayer 10VTi catalyst, acetate complexes are generated at considerably lower temperatures. Accordingly, this catalyst shows a significantly higher activity in the formation of acetic acid as compared with a catalyst comprising crystalline V2 O5 . The in situ XPS study with the monolayer V2 O5 /TiO2 catalyst was performed under similar conditions: in pure ethanol, i.e. without oxygen in the gas phase, and in the equimolar C2 H5 OH/O2 mixture. The results are presented in Figs. 11 and 12, respectively. In both cases, before the exposure to the reactant mixture or ethanol, the catalyst was pretreated in 0.25 mbar of flowing O2 at 350 ◦ C for 30 min directly inside the XPS reaction gas cell. This pretreatment led to the complete oxidation of vanadium. As a result, only a narrow single peak at 517.7 eV which can be attributed to V5+ cations was observed in the V2p3/2 spectra. According to the literature data [24,42–47], bulk and supported V2 O5 are characterized by the V2p3/2 binding energy in the range of 517.0–517.7 eV, whereas the V2p3/2 binding energies of V2 O4 and V2 O3 are within the ranges of 516.0–516.5 and 515.8–515.9 eV, respectively. The treatment in O2 led the complete burning of surface contaminations, and the C1s spectrum contains no peaks. This finding indicates high ability of the surface polymeric vanadia species to activate dioxygen from the gas phase. The treatment of the pretreated V2 O5 /TiO2 catalyst in the ethanol flow led to the complete reduction of V5+ to V4+ and V3+ even at low temperatures. Indeed, the V2p3/2 spectra consist of two peaks at 516.5–516.6 and 515.4–515.5 eV, which can be attributed to V4+ and V3+ cations, respectively. The amount of the V3+ cations grows slightly with temperature. The subsequent treatment in oxygen at 350 ◦ C led again to the complete oxidation of vanadium to V5+ . In contrast, no changes were detected in the Ti2p3/2 spectra under all used conditions: the spectra consist of only a narrow peak at 459.0 eV, which is typical of bulk TiO2 [24]. It means that titanium in the catalyst support remains in the Ti4+ state under reaction conditions. The C1s spectra obtained in situ during the heating of the monolayer V2 O5 /TiO2 catalyst in ethanol are well described with four peaks at 284.5, 285.2, 286.4, and 289.1 eV (Fig. 11). Two strong

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

103

Fig. 10. Infrared-absorption intensities of ethanol and the main detected products − acetaldehyde, acetic acid, and CO (a) as well as the main reaction intermediates adsorbed on the catalyst surface (b) measured during the oxidation of ethanol on the 20VTi catalyst as a function of temperature.

Fig. 11. Normalized V2p3/2 , Ti2p3/2 , and C1s core-level spectra of the monolayer V2 O5 /TiO2 catalyst: spectra 1 are obtained in 0.25 mbar flowing O2 at 350 ◦ C; spectra 2–5 are obtained in 0.25 mbar flowing ethanol at 110, 150, 200, and 250 ◦ C, respectively.

peaks at 285.2 and 286.4 eV could be assigned to two chemically distinct carbon atoms of molecularly adsorbed ethanol and of the surface ethoxide species. At least at low temperatures, both these species were detected by FTIR (Fig. 8c). The first peak corresponds to carbon atoms in the methyl groups, while the second peak corresponds to carbon atoms bonded with oxygen atoms. This assignment is based on an XPS study by Holroyd et al. [48], who observed two similar C1s peaks at 285.0 and 286.0 eV for ethanol adsorbed molecularly on Pd(110). The surface ethoxide species on TiO2 is characterized by similar C1s peaks at 285.5 and 286.8 eV [49]. Definitely, other species, such as adsorbed CO, which are characterized by the C1s binding energy of approximately 285.2 eV, may exist on the surface at low temperatures. The peak at 289.1 eV could be assigned to surface acetate, which is characterized on TiO2 , for example, by the C1s peak near 290 eV [49]. The presence of carbonate species, which are characterized by similar C1s binging energies cannot be excluded, as well [50]. The intensity of this

peak decreases with temperature because of the decomposition of acetate or carbonate species. The C1s peak at 284.5 eV is attributed to different adsorbed CHx species (x = 0–3) produced by the C C bond breaking in the ethoxide species and by the dehydrogenation of methyl groups [51–53]. At low temperatures near 100 ◦ C, when the rate of the C C bond breaking is small, the intensity of the C1s peak at 284.5 eV is negligible. An increase in the reaction temperature leads to an increase in the rate of this process and, as a result, the intensity of this C1s peak also increases. The V2p3/2 spectra obtained in situ under the C2 H5 OH/O2 mixture contain two peaks at 517.6–517.7 and 516.4–516.5 eV that can be attributed to V5+ and V4+ , respectively (Fig. 12). An exception is the spectrum obtained at 50 ◦ C (spectrum not shown), where an additional weak peak due to V3+ is observed at 515.7 eV. These data indicate that ethanol reduces V5+ to V3+ and V4+ ; however, in the presence of O2 in the gas phase, the fast re-oxidation of V3+ and V4+

104

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

Fig. 12. Normalized V2p3/2 and C1s core-level spectra of the monolayer V2 O5 /TiO2 catalyst: spectra 1 are obtained in 0.25 mbar flowing O2 at 350 ◦ C; spectra 2–5 are obtained under a flow of the equimolar C2 H5 OH/O2 mixture at 0.5 mbar during stepwise heating at 110, 150, 200, and 250 ◦ C, respectively.

cations occurs. Again, the Ti2p3/2 spectra consist of the single sharp peak at 459.0 eV, which corresponds to Ti4+ (spectra not shown). The C1s spectra consist of five peaks at 284.5, 285.2, 286.1, 286.4, and 289.1 eV (Fig. 12). Two peaks at 285.2 and 286.4 eV observed at low temperatures could be assigned to the ethoxide species adsorbed on the partially reduced vanadia surface. The peaks decrease in intensity with the temperature rise, which is in good agreement with the FTIR data (Fig. 8d). The origin of the peaks at 284.5 and 286.1 eV is not evident. Taking into account the FTIR data, we can speculate that this doublet originates from molecularly adsorbed ethanol. The slight shift of the C1s peaks to lower binding energy in comparison with the peaks observed under a flow of ethanol (Fig. 11) may be determined by the different oxidation state of vanadium on the catalyst surface. The peak at 284.5 eV also can originate from the adsorbed CHx species at least at low temperatures. At higher temperatures, these carbonaceous species burn to CO and CO2 , which leads to a decrease in the intensity of this peak [52,53]. The peak at 286.1 eV may originate from C O groups in the adsorbed products of acetic aldehyde condensation as well. In good agreement with the FTIR data (Fig. 7), the acetate C1s peak at 289.1 eV is observed only at high temperatures within the range between 150 and 250 ◦ C. The presented data agree well with the assumption that the selective oxidation of ethanol over vanadium oxide catalysts proceeds via the redox mechanism involving oxygen species from the surface lattice. It should be noted that no detailed mechanism for the selective oxidation of ethanol over oxide catalysts has been published elsewhere. Moreover, authors usually propose a sequence of elementary steps without any details of the reductive-oxidative process [7]. Here we develop a mechanism for the selective oxidation of ethanol to acetaldehyde and acetic acid over monolayer vanadia catalysts that includes the reduction and oxidation of vanadium cations. A proposed sequence of elementary steps for the oxidation of ethanol to acetaldehyde is shown in Scheme 1. We suggest that ethanol adsorbs intact on the acid-base sites of vanadia catalysts and further can dissociate to form adsorbed ethoxide species and

OH groups. Because the formation of the ethoxide species is accompanied with a decrease in the bands due to (V O) and because the IR spectra exhibit the strong band due to H-bonded hydroxyl groups (Fig. 9), we believe that the chemisorption of ethanol is a heterolytic process, during which the proton from the alcohol hydroxyl group is transferred to the vanadyl oxygen atom and the oxidation state of vanadium changes from V5+ to V4+ (step 1). Acetaldehyde is formed in a subsequent step via a transfer of a proton from the CH2 group to the catalyst, which is accompanied with the partial reduction of an adjacent vanadium atom (step 2). According to deuterium isotopic substitution experiments [14], ␣-C H-bond cleavage is the rate-determining step in the oxidation of ethanol to acetaldehyde. From this point of view, we believe that during this process, the effect of a support must be significant and, by analogy with the oxidation of methanol [19], the nascent hydroxyl group may be bonded with titanium cations. The OH group ultimately recombines with another OH to form H2 O and vanadyl oxygen species (step 3); adsorbed acetaldehyde can desorb as a product. The catalytic cycle is completed by a reoxidation step, the details of which have been studied computationally [54]. Hence, both the terminal V O bond and the bridge V-O-Ti bond are involved in the oxidative dehydrogenation of ethanol through the transfer of two electrons. This conclusion is confirmed by the in situ data obtained during the oxidation of ethanol at low temperatures: adsorbed ethanol, ethoxide and acetaldehyde species were detected by FTIR (Fig. 7a) and the partial reduction of V5+ to V4+ was detected by XPS (Fig. 12). It is more probable that the selective oxidation of ethanol to acetaldehyde over the polymeric vanadia species supported on zirconia proceeds via a similar mechanism. At the same time, the support may also provide sites for binding substrate molecules, and, for instance, reducible supports such as CeO2 may be involved in the redox process as well [55]. However, such processes are improbable on the monolayer vanadia-titania catalysts. The oxidation of ethanol to acetic acid proceeds via acetaldehyde intermediates that are converted further to acetate species. Under certain conditions, the acetate species are

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

105

Scheme 1. Mechanism for the selective oxidation of ethanol to acetaldehyde on the monolayer vanadia catalyst.

Scheme 2. Mechanism for the selective oxidation of ethanol to acetic acid on the monolayer vanadia catalyst.

decomposed to form acetic acid. The proposed mechanism for this reaction is depicted in Scheme 2. We suppose that adsorbed acetaldehyde reacts with lattice oxygen atoms to form adsorbed acetate species, which were detected by FTIR during the oxidation of ethanol (Fig. 7). This process is accompanied with the further reduction of vanadium to V3+ and with the formation of oxygen vacancy. This hypothesis is confirmed by the XPS data (Fig. 11), when the high concentrations of adsorbed acetate species and V3+ cations were observed simultaneously in ethanol at 110 ◦ C. It is very important that no acetic acid and CO2 were detected among the products in the absence of O2 , whereas the surface concentration of acetate species was high (Figs. 7 and 8). It means that the formation and desorption of acetic acid and CO2 does not occur on the reduced catalyst due to a high stability of acetate species. At least the partial oxidation of the catalyst is needed for the acetic acid formation. Indeed, in the presence of O2 in the gas phase, a weak signal of the acetate species is observed at 200–250 ◦ C, despite no V3+ cations were detected by XPS (Fig. 12). Finally, we can conclude that the support effect in the selective oxidation of ethanol is a complex phenomenon. At least two factors that affect the catalytic performance can be distinguished. First, the support material determines the structure of supported vanadia species, which have different activities in the target reaction. Second, the support material affects the redox properties of

supported vanadium cations, which in turn determine the reactivity and the product distribution. Indeed, when the vanadia-support interaction is weak, low-active V2 O5 crystallites are mainly formed on the support surface. Such effect is observed, for example, in the V2 O5 /SiO2 and V2 O5 /Al2 O3 catalysts. Increasing the vanadia-support interaction initiates the preferred formation of surface vanadia species that are more active in the selective oxidation of alcohols. At the same time, the polymeric vanadia species on the titania surface and on the zirconia surface demonstrated different activity in the selective oxidation of ethanol to both acetaldehyde and acetic acid (Fig. 6). In this case, the role of the support goes far beyond a mere electronic polarization effect, as captured by the electronegativity scale [55]. Definitely, the electronic polarization effect changes the ionicity of the V O bonds to change the redox properties of the surface vanadia species. As a result, the material support influences the reactivity. According to a previous study by Wachs and Weckhuysen [56], the relative extent of reduction of the V5+ surface species during butane oxidation follows the row of supports: TiO2 > CeO2 > ZrO2 > Al2 O3 > SiO2 . The same trend in the activity of supported vanadia catalysts was observed for the selective oxidation of methanol to formaldehyde [57]. This dependence directly reflects the redox properties of the surface vanadia species and our catalytic data correlate well with this row.

106

T.V. Andrushkevich et al. / Catalysis Today 279 (2017) 95–106

4. Conclusions The influence of support material and the reaction mechanism for the selective oxidation of ethanol were studied. It was shown that the catalytic performance of vanadia supported on silica, alumina, zirconia, and titania differs and depends mainly on the structure of supported vanadia species. The catalysts characterized by a weak vanadia-support interaction, such as V2 O5 /SiO2 and V2 O5 /Al2 O3 , contain mainly low-active V2 O5 crystallites. When the vanadia-support interaction increases, surface vanadia species are predominantly formed. The best catalytic performance in the selective oxidation to both acetaldehyde and acetic acid has been shown by the monolayer V2 O5 /TiO2 catalyst. The selective oxidation of ethanol proceeds via the redox mechanism where the oxidized catalyst surface oxidizes the reactant and is reoxidized by gas phase oxygen. During the reaction on the V2 O5 /TiO2 catalyst, titanium cations remain in the Ti4+ state, whereas V5+ cations undergo a reversible reduction under reaction conditions to V4+ and V3+ . Ethanol dehydrates to acetaldehyde and then acetaldehyde transforms to acetic acid through the adsorbed acetate species. The formation and desorption of acetic acid occurs only in the presence of O2 in the gas phase because the reduction of the catalyst stabilizes the surface acetate complexes. It was found that the activities of the polymeric vanadia species on the titania surface and on the zirconia surface are different, which is caused by the vanadia-support interaction as well. Acknowledgments This work was partially supported by the Skolkovo Foundation (Grant Agreement for Russian educational organizations no. 5 of 30.12.2015). The authors are grateful to G.Ya. Popova and E.V. Danilevich for performing the catalytic tests and fruitful discussions. The authors thank A.Yu. Kluyshin, M. Hävecker, and A. Knop-Gericke for their assistance in carrying out the XPS experiments and L.M. Plyasova for performing the XRD measurements. The authors are also grateful to the staff of BESSY-II for their support during the beamtime. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cattod.2016.04. 042. References [1] K. Sano, H. Uchida, S. Wakabayashi, Catal. Surv. Jpn. 3 (1999) 55–60. [2] N. Yoneda, S. Kusano, M. Yasui, P. Pujado, S. Wilcher, Appl. Catal. A 221 (2001) 253–265. [3] T. Takey, N. Iguchi, M. Haruta, Catal. Surv. Asia 15 (2011) 80–88. [4] W. Song, P. Liu, E.J.M. Hensen, Catal. Sci. Technol. 4 (2011) 2997–3003. [5] J.C. Bauer, G.M. Veith, L.F. Allard, Y. Oyola, S.H. Overbury, S. Dai, ACS Catal. 2 (2012) 2537–2546. [6] T. Takei, N. Iguchi, M. Haruta, N. J. Chem. 35 (2011) 2227–2233. [7] X. Li, E. Iglesia, Chem. Eur. J. 13 (2007) 9324–9330. [8] Y. Eguchi, D. Abe, H. Yoshitake, Microporous Mesoporous Mater. 116 (2008) 44–50. [9] B. Jorgensen, S.B. Kristensen, A.J. Kunov-Kruse, R. Fehrmann, C.H. Christensen, A. Riisager, Top. Catal. 52 (2009) 253–257. [10] F.M. Goncalves, P.R.S. Medeiros, L.G. Appel, Appl. Catal. A 208 (2001) 265–270. [11] B. Mehlomakulu, T.T.N. Nguyen, P. Delichere, E. van Steen, J.M.M. Millet, J. Catal. 289 (2012) 1–10.

[12] L.V. Mattos, F.B. Noronha, J. Power Sources 145 (2005) 10–15. [13] V.V. Kaichev, Yu.A. Chesalov, A.A. Saraev, A. Yu. Klyushin, A. Knop-Gericke, T.V. Andrushkevich, V.I. Bukhtiyarov, J. Catal. 338 (2016) 82–93. [14] B. Kilos, A.T. Bell, E. Iglesia, J. Phys. Chem. C 113 (2009) 2830–2836. [15] Yu-Chuan Lin, Chih-Hung Chang, Chin-Chen Chen, Jih-Mirn Jehng, Shin-Guang Shyu, Catal. Commun. 9 (2008) 675–679. [16] R. Tesser, V. Maradei, M. Di Serio, E. Santacesaria, Ind. Eng. Chem. Res. 43 (2004) 1623–1633. [17] N.E. Quaranta, J. Soria, V. Cortes Corberan, J.L.G. Fierro, J. Catal. 171 (1997) 1–13. [18] E.V. Danilevich, G.Ya. Popova, T.V. Andrushkevich, V.V. Kaichev, I.G. Danilova, Yu.A. Chesalov, V.A. Rogov, V.I. Bukhtiyarov, V.N. Parmon, Appl. Catal. A 475 (2014) 98–108. [19] V.V. Kaichev, G.Ya. Popova, Yu.A. Chesalov, A.A. Saraev, T.V. Andrushkevich, V.I. Bukhtiyarov, Kinet. Catal. 57 (2016) 82–94. [20] C. Penschke, J. Parier, J. Sauer, J. Phys. Chem. C 117 (2013) 5274–5285. [21] B. Beck, M. Harth, N.G. Hamilton, C. Carrero, J.J. Uhlrich, A. Trunschke, S. Shaikhutdinov, H. Schubert, H.-J. Freund, R. Schlögl, J. Sauer, R. Schomäcker, J. Catal. 296 (2012) 120–131. [22] E.V. Ovchinnikova, T.V. Andrushkevich, G.Y. Popova, V.D. Meshcheryakov, V.A. Chumachenko, Chem. Eng. J. 154 (2009) 60–68. [23] A. Knop-Gericke, E. Kleimenov, M. Hävecker, R. Blume, D. Teschner, S. Zafeiratos, R. Schlögl, V.I. Bukhtiyarov, V.V. Kaichev, I.P. Prosvirin, A.I. Nizovskii, H. Bluhm, A. Barinov, P. Dudin, M. Kiskinova, Adv. Catal. 52 (2009) 213–272. [24] V.V. Kaichev, G. Ya. Popova, Yu.A. Chesalov, A.A. Saraev, D.Y. Zemlyanov, S.A. Beloshapkin, A. Knop-Gericke, R. Schlögl, T.V. Andrushkevich, V.I. Bukhtiyarov, J. Catal. 311 (2014) 59–70. [25] E.V. Danilevich, G.Ya. Popova, T.V. Andrushkevich, Yu.A. Chesalov, V.V. Kaichev, A.A. Saraev, L.M. Plyasova, Stud. Surf. Sci. Catal. 175 (2010) 463–466. [26] C. Sanchez, J. Livage, G. Lucazeau, J. Raman. Spectrosc. 12 (1982) 68–72. [27] G.C. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1–31. [28] B. Grzybowska-Swierkosz, Appl. Catal. A 157 (1997) 263–310. [29] G. Ya. Popova, T.V. Andrushkevich, E.V. Semionova, Yu.A. Chesalov, L.S. Dovlitova, V.A. Rogov, V.N. Parmon, J. Mol. Catal. A 283 (2008) 146–152. [30] I.E. Wachs, Catal. Today 27 (1996) 437–455. [31] X. Gao, J.L.G. Fierro, I.E. Wachs, Langmuir 15 (1999) 3169–3178. [32] W.C. Vining, J. Strunk, A.T. Bell, J. Catal. 281 (2011) 222–230. [33] Z. Wu, H.-S. Kim, P.C. Stair, S. Rugmini, S.D. Jackson, J. Phys. Chem. B 109 (2005) 2793–2800. [34] D.A. Bulushev, L. Kiwi-Minsker, F. Rainone, A. Renken, J. Catal. 205 (2002) 115–122. [35] R. Baddour-Hadjean, M.B. Smirnov, K.S. Smirnov, V. Yu Kazimirov, J.M. Gallardo-Amores, U. Amador, M.E. Arryo-de Dompablo, J.P. Pereira-Ramos, Inorg. Chem. 51 (2012) 3194–3201. [36] J.M. Coronado, K. Kataoka, I. Tejedor-Tejedor, M.A. Anderson, J. Catal. 219 (2003) 219–230. [37] Z. Yu, S.S.C. Chuang, J. Catal. 246 (2007) 118–126. [38] M. Li, Z. Wu, S.H. Overbury, J. Catal. 306 (2013) 164–176. [39] R.G. Greenler, J. Chem. Phys. 37 (1962) 2094–2100. [40] A.M. Nadeem, G.I.N. Waterhouse, H. Idriss, Catal. Today 182 (2012) 16–24. [41] V. Lochar, H. Drobná, Appl. Catal. A 269 (2004) 27–31. [42] M. Sambi, G. Sangiovanni, G. Granozzi, F. Parmigiani, Phys. Rev. B 55 (1997) 7850–7858. [43] M.E. Harlin, V.M. Niemi, A.O.I. Krause, J. Catal. 195 (2000) 67–78. [44] K.V.R. Chary, G. Kishan, C.P. Kumar, G.V. Sagar, J.W. Niemantsverdriet, Appl. Catal. A 245 (2003) 303–316. [45] L.K. Boudali, A. Ghorbel, P. Grage, F. Figueras, Appl. Catal. B 59 (2005) 105–111. [46] Z. Wu, F. Dong, Y. Liu, H. Wang, Catal. Commun. 11 (2009) 82–86. [47] H. Zhao, S. Bennici, J. Cai, J. Shen, A. Auroux, Catal. Today 152 (2010) 70–77. [48] R.P. Holroyd, R.A. Bennett, I.Z. Jones, M. Bowker, J. Chem. Phys. 110 (1999) 8703–8713. [49] P.M. Jayaweera, E.L. Quah, H. Idriss, J. Phys. Chem. C 111 (2007) 1764–1769. [50] A.A. Khassin, T.M. Yurieva, V.V. Kaichev, V.I. Bukhtiyarov, A.A. Budneva, E.A. Paukshtis, V.N. Parmon, J. Mol. Catal. A 175 (2001) 189–204. [51] V.V. Kaichev, V.I. Bukhtiyarov, G. Rupprechter, H.-J. Freund, Kinet. Catal. 46 (2005) 269–281. [52] V.V. Kaichev, A.V. Miller, I.P. Prosvirin, V.I. Bukhtiyarov, Surf. Sci. 606 (2012) 420–425. [53] A.V. Miller, V.V. Kaichev, I.P. Prosvirin, V.I. Bukhtiyarov, J. Phys. Chem. C 117 (2013) 8189–8197. [54] X. Rozanska, E.V. Kondratenko, J. Sauer, J. Catal. 256 (2008) 84–94. [55] T. Kropp, J. Paier, J. Sauer, J. Am. Chem. Soc. 136 (2014) 14616–14625. [56] I.E. Wachs, B.M. Weckhuysen, Appl. Catal. A 157 (1997) 67–90. [57] G. Geo, I.E. Wachs, J. Catal. 146 (1994) 323–334.