Influence of Steam on a Vanadyl Pyrophosphate Catalyst During

Subscriber access provided by UNIVERSITY OF THE SUNSHINE COAST

Article

Influence of Steam on a Vanadyl Pyrophosphate Catalyst During Propane Oxidation Maria Heenemann, Christian Heine, Michael Hävecker, Annette Trunschke, and Robert Schloegl J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.7b06314 • Publication Date (Web): 26 Sep 2017 Downloaded from http://pubs.acs.org on September 28, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry B is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Influence of Steam on a Vanadyl Pyrophosphate Catalyst During Propane Oxidation Maria Heenemann,1 Christian Heine,1 Michael Hävecker,2 Annette Trunschke,*,1 and Robert Schlögl1,2

1

Fritz-Haber-Institut der Max-Planck-Gesellschaft

Department of Inorganic Chemistry Faradayweg 4-6, 14195 Berlin (Germany) 2

Max-Planck-Institut für Chemische Energiekonversion

Department Heterogeneous Reactions Stiftstr. 34-36, 45470 Mülheim an der Ruhr (Germany)

*Corresponding author: Annette Trunschke Department of Inorganic Chemistry, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin (Germany) Phone: +49 30 8413 4457 E-mail: [email protected]

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 31

Abstract We have investigated electronic and catalytic modifications of the p-type semiconducting selective oxidation catalyst vanadyl pyrophosphate (VPP) for propane oxidation in the presence and absence of steam. Steam changes propane conversion only slightly, but increases the selectivity towards oxygenates (acrylic acid, acetic acid) and the olefin propylene resulting in reduced selectivity of the undesired total oxidation products CO and CO2. Contact-free in operando microwave conductivity measurements at 0.1 MPa revealed that the modified catalytic performance is accompanied by a reduced electrical conductivity. Surface sensitive NearAmbient-Pressure X-ray Photoelectron Spectroscopy (NAP-XPS), and X-ray Absorption Spectroscopy (XAS) measurements at 25 Pa showed that steam depletes the topmost surface of VPP in phosphorus and enhances the average vanadium oxidation state slightly. These findings are accompanied by a decreased work function, but no detectable shift of the valence band edge is observed. Thus, the chemical surface modification changes the surface dipole but leaves the barrier height of the surface induced space charge layer basically unaffected. Hence, we conclude that steam does not affect the electron hole concentration (majority charge carriers) and hence the oxygen vacancy concentration. Therefore, the reduced conductivity can be understood in terms of charge carrier mobility changes, which may affect the selectivity of VPP towards oxygenates with steam. In addition, the modification of local properties, such as the concentration of acid sites as well as the nature and number of adsorption sites may have an impact on the catalytic properties.

2


Page 3 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Upcoming raw material changes in the chemical industry call for innovations in heterogeneous catalysis technologies.1-2 Oxidative dehydrogenation and selective oxidation of short-chain alkanes are desirable reactions that often suffer from low selectivity. Successful oxidation catalysts, such as vanadium phosphorous oxides (VPO), industrially applied in the direct oxidation of n-butane to maleic anhydride,3-5 have been developed applying empirical strategies guided by phenomenological concepts.6-7 But the complexity of the resulting oxide materials in terms of structure and composition hampers detailed understanding and targeted catalyst design which is, however, necessary to develop the economic utilization of other alkane molecules. VPO embrace several phases with vanadium in different formal oxidation states including: +5 (α-, β-, γ-, and δ-VVOPO4), +4 ((VIVO)2P2O7), and +3 (VIIIPO4).8-10 The bulk phase of the working industrial catalyst that produces maleic anhydride in n-butane oxidation with a yield of 65 mol% is exclusively composed of crystalline vanadyl pyrophosphate (VIVO)2P2O7 (VPP, Figure 1).11-12 This phase is formed by transformation of a VOHPO4·0.5 H2O precursor in an activation procedure.9

Figure 1. Schematic illustration of the (VIVO)2P2O7 structure with oxygen atoms in red, [VO6] octahedrons (or [O4V=O] square pyramids) in blue and [PO4] tetrahedrons in grey.

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 31

The crystal structure of VPP consists of distorted [VO6] octahedrons, which are joined by edges. A pair of [VO6] octahedrons is equatorial linked to [PO4] tetrahedrons forming a layered structure in the (100) plane.8 The octahedral coordination sphere of vanadium is often considered as a [VO5] square pyramid with a short apical V=O bond (1.6 Å). The V=O bonds are in trans position and the layers are connected by pyrophosphate groups (P2O7).13-15 However, as will be discussed below, the ideal crystal structure of VPP in any type of termination has little to do with the real atomistic configuration on the surface of the catalyst under operation. Importantly, VPP is a p-type semiconductor with electron holes as major charge carriers.11, 16-17 VPP has been found to be also active in direct oxidation of propane to acrylic acid, but the selectivity to the desired product is low. Ai reported an acrylic acid yield of approximately 6 mol% over V2O5/P2O5 mixtures.18 The yield was doubled by using VPP as catalyst and optimized by adding 15% steam to the gas feed.19 The differences in selectivity to the corresponding acid/anhydride in C3 and C4 oxidation, respectively, over VPP and the impact of steam may be related to differences in bulk and surface electronic properties. These properties further depend on the chemical potential of the applied gas feed with implications on the activation of molecular oxygen.20 Indeed, gas-phase-dependent conductivity response of VPP was measured under conditions of n-butane oxidation by in operando microwave cavity perturbation technique (MCPT)21 showing a behavior of VPP like an oxide gas sensor.11-12, 21 Furthermore, the surface of crystalline VPP undergoes changes under reaction conditions of nbutane oxidation. Formation of microcrystalline and amorphous VPO surface domains has been reported in previous literature.22-24 The oxidation state of vanadium at the surface (4.0 – 4.3) differs from the bulk (formal 4.0), in particular in the presence of reactive gases.12, 20 The V4+/V5+ redox couple was identified as surface state exchanging electrons with the reaction partners alkane and oxygen. The surface concentration of V5+ changes with varying composition of the 4


Page 5 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


reaction feed and correlates well with the energy position of the XP valence band onset and secondary electron cut-off. These changes were interpreted as variation in surface induced band bending and electron affinity.20 Thus, band bending variation can explain the observed conductivity variation induced by the gas phase composition. The aim of the present work is the investigation of steam induced electronic structure modifications of VPP during propane oxidation. Microwave conductivity measurements in a contact-free manner at ambient pressure in combination with product analysis by gas chromatography will be used to monitor changes in the conductivity as a function of feed composition in operando. Near-Ambient-Pressure (NAP) X-ray Photoelectron Spectroscopy (XPS), and X-ray Absorption Spectroscopy (XAS, measured in total electron yield) are applied to investigate charge carrier and chemical dynamics on the surface of the working VPP catalyst. Operation will be proven by mass spectrometry. The impact of catalyst dynamics on catalytic performance will be discussed.

2. Experimental Section 2.1

Catalyst

The vanadyl pyrophosphate catalyst precursor was prepared via an alcohol route and activated in n-butane oxidation feed resulting in a polycrystalline powder with a specific surface area of 25 m2/g.11-12 Phase purity of the (VIVO)2P2O7 catalyst (Internal catalyst ID: 10449) and stability of the bulk phase under reaction conditions of propane oxidation was proven by X-ray powder diffraction (XRPD) (Figure S1).

5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2.2

Page 6 of 31

Operando microwave cavity perturbation technique

Electrical conductivity measurements were conducted in a contact-free manner using the microwave cavity perturbation technique (MCPT)21, 25 to determine the conductivity of VPP in propane oxidation feed and to monitor the influence of steam on the charge transport properties of VPP. The measurement principle is based on the fact that the electric and magnetic field distribution inside a microwave cavity is modified by a dielectric sample. The electromagnetic field inside the cavity is at resonance with a characteristic resonance frequency and a quality factor , = ⁄∆ where ∆ is the full width at half minimum of the power peak. The

unloaded quality factor and the resonance frequency for every data point were determined from the S11 parameter in reflection mode. We applied a Smith chart analysis of the complex reflection factor and the transmission line theory.26-27 The introduction of the dielectric catalyst in the maximum of the electromagnetic field shifts and decreases . This dielectric response is described by the complex permittivity ( = + ).21, 25 The real and imaginary part of

the complex permittivity can be computed according to − = − 1 1 1 − =

(1)

(2)

where and are cavity constants, and are the empty and sample-loaded frequencies and

and are quality factors of the empty and sample-loaded cavity. The obtained effective

permittivity values are the values for the polycrystalline catalyst in powder form, which can be

transformed to effective bulk values applying the Landau-Liftshitz-Looyenga formalism.28-30 Using the imaginary part of the bulk permittivity , , the microwave electrical conductivity σ can be calculated (Equation 3), where is the vacuum permittivity.

6


Page 7 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


= ,

(3)

For the in operando microwave conductivity measurements we used a custom-designed, cylindrical microwave cavity with a height of 20 mm and an inner diameter of 45 mm. The experiment was performed with the first transverse magnetic mode (TM010) at an empty resonance frequency of about 4.79 GHz. A network analyzer was used to generate microwaves (Agilent PNA-L N5230C) and was connected via a coaxial cable with the cavity. Here, a coupling loop inductively coupled the microwave into the cavity. For the catalytic studies, we used a plug-flow reactor, which was located in the electric field maximum and connected with a gas supply (Bronkhorst, El-Flow). The reactor tube was filled with the catalyst (100 – 200 µm particle sieve fraction) to a bed length of 9.8 mm corresponding to a weight of 52.3 mg. A total gas flow of 20 ml/min was applied (gas hourly space velocity of 17 323 h−1). The catalyst was heated to 400 °C with a heating rate of 10 °C/min. Preheated nitrogen (8 l/min) passed the reactor tube and heated the catalyst. After reaching the final temperature the experiment was performed subsequently under dry, wet, dry, wet, and dry feed conditions. The dry gas atmosphere contained 3 vol% propane, 6 vol% oxygen, and 91 vol% nitrogen. The wet gas atmosphere contained 3 vol% propane, 6 vol% oxygen, 5 vol% steam, and 86 vol% nitrogen. The gas purity was at least 99.95 vol%. The effluent gas flow was analyzed with an online gas chromatograph (Agilent 7890). A detailed description of the method and the setup was reported previously.21, 31

2.3

Near-ambient pressure X-ray photoelectron and soft X-ray absorption spectroscopy

In a subsequent step, we investigated the catalyst with near-ambient pressure X-ray photoelectron spectroscopy (NAP-XPS) and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy in the total electron yield (TEY) mode. The experiments were conducted at the Innovative 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 31

Station for In-Situ Spectroscopy beamline located at the BESSY II synchrotron in Berlin (Germany). The experimental setup is described in literature.32-33 The catalyst (10 mg) was pressed into a pellet (diameter of 8 mm, 1 ton pressing pressure). Spectroscopic measurements were performed at 25 Pa and 400 °C with a heating rate of 10 °C/min. A near-infrared laser (808 nm, continuous wave) was used for heating. The experiment was started in dry propane gas atmosphere (1 Nml/min C3H8, 2 Nml/min O2, 3 Nml/min He), changed to wet propane gas atmosphere (1 Nml/min C3H8, 2 Nml/min O2, and 3 Nml/min steam), and vice versa. The reaction products were analyzed by proton-transfer reaction mass spectrometry (IONICON). Core level spectra of C1s, O1s, V2p, and P2p were recorded at different kinetic energies with pass energy of 20 eV. V2O5 is used as a model substance to calculate the inelastic mean free path (IMFP) applying the Tanuma formalism.34-35 For our study, the computed IMFP was 0.55 nm and 1.4 nm at kinetic energies of 150 eV and 750 eV, respectively. In the following, we refer to 0.55 nm as surface sensitive and 1.4 nm as "bulk" sensitive. The binding energy (BE) calibration was done using the C1s peak (BE: 284.4 eV). The V2p3/2, P2p, and O1s peaks were fitted with Gaussian-Lorentzian functions after a Shirley background correction with CasaXPS software.36 Gaussian-Lorentzian functions are convolutions of Gaussian (considers instrumental response, thermal broadening, X-ray line shape) and Lorentzian (considers lifetime broadening) functions.37 The average V oxidation state for surface and "bulk" sensitive measurements was derived from the V4+ and V5+ peak area of the V2p3/2 core level peak while the P/V ratio was calculated from the photon intensity and cross section corrected peak areas.38 The secondary electron cutoffs and the valence band spectra were measured with constant photon energy of 100 eV. A voltage of -18.2 V was applied between detector and sample for the photoelectron cutoff measurements. A pass energy of 2 eV has been used. 8


Page 9 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The NEXAFS spectra were taken in the TEY mode. The vanadium (V) L3-edge was recorded with photon energies of 505 - 560 eV and normalized to their maxima for a better comparison. The ∗ -resonance of gas-phase oxygen (530.8 eV) was used for energy calibration.39

3. Results and Discussion 3.1

Operando microwave conductivity

In order to probe the semiconducting and catalytic properties of the VPP catalyst the microwave electrical conductivity was measured in propane oxidation under dry and wet feed conditions. The propane to oxygen ratio was adjusted to the stoichiometric ratio of acrylic acid synthesis.40 Figure 2 shows the microwave conductivity (a) and the corresponding catalytic data (b). At the temperature of 400 °C, the conductivity decreased asymptotically towards 0.017 Sm−1 in dry feed with time on stream (Figure 2a). The first measurement of the gas phase composition was only possible after the conductivity values were largely stabilized. Accordingly, almost stable catalytic performance was observed henceforward (Figure 2b and Figure S2). Steam was introduced after 156 min. Then, the conductivity rapidly decreased reaching a constant value of 0.015 Sm−1 after 36 min. The conductivity increased again to 0.016 Sm−1 when the steam was switched off. Subsequent wet-dry cycles result in analogous changes, i.e. a conductivity decrease ( = 0.0146 Sm−1) under wet feed conditions and a conductivity increase under dry feed conditions ( = 0.0155 Sm−1).

9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 31

Figure 2. Microwave conductivity σ of VPP (a), and simultaneously measured catalytic performance under dry (3 vol% C3, 6 vol% O2, inert) and wet (3 vol% C3, 6 vol% O2, 5 vol% steam, inert) feed conditions as indicated in the graph at 400 °C (b); S describes the sum of the selectivity to desired products acrylic acid (AA) and propylene (PP) in red as well to unselective oxidation products CO, CO2, and acetic acid (AcA) in black; The selectivity for each individual product is presented in Figure S2 in the Supporting Information.

Overall, VPP ( ≈ 0.017 Sm−1) is less conductive than the efficient propane oxidation catalyst MoVTeNbOx (M1-phase) ( ≈ 2.3 Sm−1) under similar experimental conditions.40 Single-phase crystalline MoVTeNb M1 oxide is the most selective catalyst in oxidation of propane to acrylic acid that is known so far showing up to 80% selectivity at high propane conversion (>50%) depending on catalyst preparation and reaction conditions.41 It is well known that VPP acts as a p-type semiconductor11-12, phase).28,

40

17, 42-43

in contrast to the n-type semiconducting MoVTeNbOx (M1-

The major charge carriers are electron holes in VPP.11,

42

The electron hole 10


Page 11 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


concentration ℎ• increases when double ionized oxygen vacancies •• are filled by oxygen ! from the reactive gas feed, which is shown in Equation 4 (Kröger-Vink notation). This means that the conductivity increases for VPP (d /dp(!) > 0) and decreases for MoVTeNbOx (M1phase) (d /dp(!) < 0) with increasing oxygen partial pressure in the gas feed.11, 28, 42 ! + 2 •• ⇌ 2! % + 4ℎ•

(4)

However, the impact of steam on conductivity seems to have a different reason, because the same conductivity response was observed for the p-type semiconductor VPP and the n-type semiconductor MoVTeNbOx (M1-phase) in propane oxidation feed by changing from dry to wet feed (conductivity decrease under wet feed conditions and conductivity increase under dry feed conditions).40 Therefore, the observed behavior is independent of the charge carrier type. A reference experiment without propane (Figure S3) shows no conductivity response, which excludes also measurement artifacts due to the presence of water vapor in the gas phase. Referring to Equation 4, we conclude that steam does not influence the double ionized oxygen vacancies •• and, therefore, the overall electron hole concentration ℎ• . The catalytic activity decreases marginally by adding steam to the propane oxidation feed (Figure 2b). The propane conversion alternated between 6 % (dry feed) and 4 % (wet feed) with a constant carbon balance close to 100 %. In agreement with the literature, propane is oxidized to propylene, CO, CO2, acrylic acid, and acetic acid over VPP.18,

44-45

The selectivity towards

acrylic acid, acetic acid, and interestingly also propylene was increased in presence of steam, while formation of CO and CO2 were suppressed (Figure S2) resulting in a higher overall selectivity to the desired oxidation products propylene and acrylic acid in steam (Figure 2b). This could mean that for example a direct pathway of propane to COx is suppressed by steam addition. The selectivity changes are roughly reversible under alternating operation conditions. The 11



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 31

increase in acrylic acid selectivity under the influence of auxiliary steam was attributed to different reasons including the formation of new active sites, competitive adsorption of oxygen and water, or the suppression of consecutive reactions of acrylic acid.19,

46-48

In any case, the

product distribution is changed in steam, which has an impact on the chemical potential of the abundant gas phase. The gas phase potential may affect again the electronic and chemical state of the catalyst, which is reflected in the conductivity (Figure 2a). In the present case, the change in the electronic structure seems to be induced preferentially by the products rather than by the feed. The absence of a steam induced conductivity response without propane in the reaction feed supports this hypothesis. However, it should be discussed at this point that in absence of propane, i.e. in presence of a mixture of steam and inert gas, the reactivity of the VPP surface might be different compared to the reactivity in propane/oxygen/steam mixtures due to a different concentration of surface defects leading to differences in the reaction of the gas-phase water molecules with the surface of VPP. DFT calculations show that dissociation of water molecules hardly occurs on the stoichiometric V2O5 (010) surface due to the significant Coulombic repulsion of the lattice oxygen atoms around the exposed vanadium center to the approaching oxygen of the hydroxyl species.49 On the other hand, hydroxyl species are formed on the surface of an oxidation catalyst by reaction with propane. Furthermore, vacancy formation on V2O5 (010) by elimination of a hydroxyl group requires less energy than abstraction of an oxygen atom.50 The design of an appropriate reference experiment is, therefore, not straightforward.

The p-type conductivity of a semiconducting VPP catalyst is determined approximately by its hole concentration '( , carrier mobility μ( , and the elementary charge *; + ≈ |*|'( +μ( .

Furthermore, '( can be expressed in terms of the effective hole density of states in the valance

/ band ' .. , the difference between the valence band edge 0/ and the Fermi level 01 and the

thermal energy 23 according to51


12

Page 13 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


/ '( + = ' .. *+4 5−

0/ + − 01 6 23

(5)

Eichelbaum et al. showed that surface induced band bending of VPP is triggered by formation of the V4+/V5+ redox couple. The concentration of V5+ is changed with the abundant reaction gas feed in absence of auxiliary steam.11-12, 20 A schematic band diagram for a p-type semiconductor is shown in Figure S4 to further illustrate flatband, surface dipole, and band bending situations. Thus, a different chemical environment induced by steam in the reaction feed could possibly change the band bending (modifying the gap between 0/ + and 01 ) and, therefore, the hole concentration '( . To confirm this statement, we have performed NAP-XPS core level and

NEXAFS spectroscopy to identify changes in the surface chemical composition associated with the V4+/V5+ redox couple. We accompanied these studies by valence band and work function measurements.

3.2

Near-ambient pressure X-ray photoelectron spectroscopy and near-edge X-ray absorption fine structure spectroscopy

The V L3-edge NEXAFS edge was analyzed together with the surface sensitive (IMFP: 0.55 nm) V2p3/2 core level in dry and wet propane feed condition. XP survey spectra (0( of 860 eV, see Figure S5 in the Supporting Information) were taken to check for impurities before performing high resolution XP core level scans. Impurities can promote or poison a catalyst and cause misleading interpretations of the XP spectra. We found V, P, O, and minor amounts of C and Si suggesting minor graphite and silica impurities, respectively. Neither graphite nor silica is known to influence selective oxidation reactions over VPP. Thus, these impurities do not interfere the analysis of the V2p3/2, P2p3/2 and V L3-edge NEXAFS edge. The analysis of the O1s core level is complicated by satellite peaks from the V2p1/2,3/2 in the oxidation state +5 and +4 reported by 13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 31

Zimmermann et al.52 Therefore, we did not analyze the O1s core level. The V2p1/2 core level was not evaluated, since it overlaps with photoemission satellites of the V2p3/2 core level. The V2p1/2 core level is also affected by Coster-Kronig Auger-Meitner decay channels.53 The proton-transfer reaction mass spectrometry (PTR-MS) signal at m/z = 73 (mass of protonated acrylic acid) and m/z = 43 (mass of protonated propylene), recorded during all applied gas feeds, proves that the catalyst was under catalytic operation (Figure S6). The general trend showed that the concentration of acrylic acid increases in wet feed. In contrast to the observation at normal pressure in the operando microwave conductivity measurements the abundance of propylene decreases in presence of steam. The trend for both products is reproducible.

The edge energy position in V L3-edge NEXAFS, which can be determined by a first momentum analysis, increases with increasing vanadium oxidation state.14, 54-55

Figure 3. V L3-edge NEXAFS edge (TEY mode normalized to its maxima) of VPP (in dry C3/O2/He (1/2/3 sccm) and wet C3/O2/H2O (1/2/3 sccm)) at 400 °C, 25 Pa (top); The difference spectra (TEY(wet feed)-TEY(dry feed)) is shown in red (bottom).

Figure 3 illustrates the normalized V L3-edge NEXAFS edge together with the difference spectrum between wet and dry feed conditions (TEY (wet) - TEY (dry)). A shift of the intensity 14


Page 15 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


towards higher photon energy is found when introducing steam, which can be clearly seen in the difference spectrum. Accordingly, the first momentum of the V L3-edge NEXAFS increases from 517.71 eV (dry feed) to 517.75 eV (wet feed) indicating an increase in the average V oxidation state. We identified two features in the difference spectrum in Figure 3. Feature I (518 eV) indicates a shift as well as an increase in relative intensity and feature II (516 eV) indicates a slight increase in intensity in wet feed compared to dry feed. The assignment is difficult since the changes are small. Maganas et al. reported a detailed assignment of several contributions to the V L3-edge spectrum of V2O5 using a systematic theoretical (DFT-ROCIS) approach.56 According to this calculation a component at higher energy (519 - 520 eV) can be assigned to 2p→3dz2 and a contribution at lower energy (516 - 519 eV) is characteristic for 2p→3dxz,yz single electron excitations. We conclude that the shift to higher energy that becomes apparent in the difference spectrum shown in Figure 3 is an indication for a partial change of the V oxidation state from V4+ to V5+ when changing from dry to wet gas feed. We interpret our observations in terms of a transformation of the VPP surface into a more “V2O5-like” character under wet conditions.

The surface sensitive V2p3/2 spectra (IMFP: 0.55 nm) are shown in Figure 4a. In the first dry propane feed, the best fit was achieved with three Gaussian-Lorentzian peaks with binding energies of 516.9, 517.8, and 518.0 eV. More details about the fit parameters are given in in the Supporting Information (Table S1). The peak at 516.9 eV binding energy in surface mode (Figure 4a) and 517.0 eV binding energy in "bulk" mode (IMFP: 1.4 nm)

(Figure 4b),

respectively, is assigned to V4+ species,57-59 while the two peaks at 518.0 eV (V5+(I)), and 517.8 eV (V5+(II)) suggest V5+ species in different coordination environment.60-63 Differentiation becomes possible due to the high resolution of the synchrotron experiment. The peak V5+(I) at 518.0 eV is assigned to a "bulk"-like phosphorous-containing vanadium oxide species since the peak is the only observable V5+ species in the "bulk" measurement (Figure 4b), and the BE 15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 31

position is in good agreement with literature values of vanadium phosphorous oxides (BE: 518 eV).57-59 The BE value is much higher compared to pentavalent vanadium in V2O5 (BE: 517.2 eV).64 The inclusion of an additional species with a BE of 517.8 eV (V5+(II)) is suggested by analyses of the V2p3/2 core level surface spectra in dry gas. The V5+(II) species at 517.8 eV represents the minor V5+ species in the first surface spectrum under dry conditions and may be assigned to a specific surface termination of VPP or the termination of a V2O5 (multi-)layer on VPP. After introduction of steam into the chamber, the peak profile of the V2p3/2 surface core level became narrower in comparison with the spectrum recorded first in dry feed conditions. The phosphorous-containing vanadium oxide V5+(I) species (BE: 518.0 eV) disappears from the surface (Figure 4a), while the peak area of the V5+(II) species (BE: 517.8 eV) increased significantly. This suggests a changed coordination of V5+ on the surface of the catalyst that occurs upon first contact with steam (see discussion concerning the surface phosphorous concentration below). The BE of V4+ remained constant. We observed no significant changes in the subsequent variation between wet and dry feed. Furthermore, no significant changes of the V5+(I) BE nor peak area in the "bulk" sensitive mode (IMFP: 1.4 nm) were detected (Figure 4b).

16


Page 17 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 4. V2p3/2 core level spectra of VPP (in dry C3/O2/He (1/2/3 sccm) and wet C3/O2/H2O (1/2/3 sccm)) feed in surface sensitive mode (a) and "Bulk" sensitive mode (b) at 400 °C, 25 Pa; Shirley BG describes the Shirley background correction.

The average V oxidation state derived from V4+ and V5+ peak areas of the V2p3/2 core level is summarized in Figure 5a for all applied gas feed conditions. In the first dry feed, an average V oxidation state of 4.28 for the surface sensitive mode and 4.13 for the "bulk" sensitive mode was found. The latter is close to the expected VPP "bulk" value of 4. These findings indicate that more V5+ species are present on the surface (higher degree of oxidation). When steam was added to the propane feed the oxidation state increased to 4.42 (surface sensitive) and 4.16 ("bulk" sensitive), respectively, which is in agreement with the shift observed in V L3-edge NEXAFS (Figure 3). No significant changes were observed in the subsequent variation between wet and

17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 31

dry feed. The overall oxidation state in the range of 4.28 to 4.42 for vanadium is in agreement with laboratory XPS studies.65

Figure 5. The average vanadium oxidation state of VPP as determined from V2p3/2 core levels (a), and the P/V ratio of VPP for surface and the "bulk" sensitive mode in dry C3/O2/He (1/2/3 sccm) and wet C3/O2/H2O (1/2/3 sccm) feed (b).

The catalyst loses phosphorous irreversibly, but in the present experiment only after the first switch of dry to wet feed which is visible in surface as well as "bulk" mode (Figure 5b). A ratio of 1/1/4.5 for V/P/O is proposed for VPP according to the Daltonid formula. The P/V atomic ratio on the surface of VPP has been studied by XPS before.65-67 Volta et al. found a surface enrichment of phosphorus where the P/V ratio ranged from 1.5 to 3.065-66 which is in agreement with our "bulk" measurements.

In subsequent dry/wet cycles the phosphorous concentration was stabilized at a lower level revealing always a lower concentration at the topmost surface compared to the "bulk" with a surface P/V ratio even below one. Hence, the hydrolysis of phosphate to PxOy followed by release of PxOy to the gas atmosphere24,

68

seems to occur under the applied conditions to a

considerable extent leading to surface depletion in phosphorous. The phenomenon is well known. 18


Page 19 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


To compensate loss of phosphorous under real operation conditions, phosphorus organic reagents, such as trimethyl phosphate,24 are added in the industrial n-butane process.9, 15

The stable P content on the surface in all dry/wet cycles except the first one supports the assignment of the V5+(I) species with a BE of 518.0 eV to phosphorous-containing V5+ species, because this species is only observed in the first surface spectrum under dry conditions (Figure 4a), and the disappearance of the species by switching to wet feed may be related to the irreversible loss of phosphorous. In all subsequent cycles the surface contains a constant amount of only one V5+ species (V5+(II), BE: 517.8 eV). The binding energy of the latter species is shifted to lower energy compared to the “bulk” V5+ species (V5+(I), BE: 518.0 eV), but it is still very high. Therefore, the species is attributed to a molecular vanadium oxide entity, such as a single VOx site or a VxOy cluster with a limited degree of oligomerization.

Summarizing the XPS core level analysis, the surface species are strongly influenced when adding steam to the propane feed. We found that steam irreversibly depletes phosphorus in the topmost surface layer of VPP, but only in the first treatment with steam. Simultaneously, a V5+ species related to vanadium phosphorous oxide (V5+(I), BE: 518.0 eV) disappears from the surface. After the first treatment in steam the concentration of V5+ surface species due to VxOy clusters characterized by lower binding energy (V5+(II), BE: 517.8 eV) was increased compared to the dry starting point, but remains constant in further dry/wet cycles. The surface concentration of phosphorous in these cycles remains constant in the same way. These changes are detectable only in the surface sensitive mode, whereas the "bulk" sensitive mode just shows tiny changes. Consequently, steam leads to a rearrangement of the VPP surface chemical composition (formation of VxOy clusters) due to the loss of phosphorous. At this point it should be noted that

19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 31

the average V oxidation state in the MoVTeNb M1 oxide catalyst, which is more active and selective in propane oxidation compared to VPP, is always above 4.5.28

3.3

Valence band spectra, secondary electron cutoff and work function

Photoelectron spectroscopy allows the determination of the valence band position and, therefore, band bending variations, which affect the electron hole concentration of VPP (Equation 5). To understand the decrease in conductivity of VPP measured by MCPT in wet feed (Figure 2), we investigated the valence band and the secondary electron cutoff 0789:.. in detail. In this context, one has to keep in mind that photoelectron spectroscopy monitors the excited state of a solid.69 In particular, photoelectron excitations into the conduction band affect the barrier height of the space charge layer in semiconductors.70

Figure 6. Valence band spectra (normalized to their maximum) recorded at Ehv of 100 eV with constant photon flux (a), and secondary electron cutoff of VPP in different gas mixtures (b).

We recorded the valence band spectra at 0(; of 100 eV with constant photon flux for a comparison between valence band edge positions in VPP for different reaction feeds. The valence band spectra in the first dry, wet, and reducing C3/He feed are depicted in Figure 6a. For 20


Page 21 of 31

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


a better comparison, the valence band spectra were normalized to their maximum. It can be clearly seen that the valence band edge position does not shift in a detectable manner by adding steam. The unaffected valence band edge position was also reported for the n-type semiconducting MoVTeNbOx (M1-phase) by changing from dry to wet feed.40 In addition, similar ultra-high vacuum studies of V2O371 and α-Fe2O372 did not exhibit any valence band edge variation upon steam adsorption. However, in reducing C3/He feed a valence band shift to higher binding energies was observed. This shift is accompanied by a relative intensity increase of the valence band pre-edge feature at 1.9 eV BE (dry/wet feed) and 2.4 eV BE (C3/He). This pre-edge shift to higher binding energy suggests a reduction of the surface induced accumulation layer as described by Eichelbaum et al.20 The corresponding states are assigned to occupied vanadium 3d states for VPP.20 The increased pre-edge peak, therefore, indicates a lower V5+ concentration consistent with the oxidation state determined by analysis of V2p3/2 core level peaks. Thus, the barrier height reduction is caused by a change of the surface V4+/V5+ redox couple. The change in the oxidation state induced by switching from dry to wet feed (Figure 5a), however, is apparently too small to be reflected in the valence band spectra. Changes at the detection limit have been observed by variation of the alkane in the feed from ethane to propane to n-butane over MoVTeNb oxide.28 To confirm our results, we have measured the secondary electron cut-off 0

Influence of Steam on a Vanadyl Pyrophosphate Catalyst During

Recommend Documents