The Impact of the Bulk Structure on Surface Dynamics of Complex Mo-V-based Oxide Catalysts

Article abstract of DOI:10.1021/acscatal.7b00130

Mixed oxides composed of the four metals Mo, V, Te, and Nb are known to be efficient catalysts in selective oxidation of lower alkanes. The outstanding catalytic performance of such mixed oxides is attributed to the presence of the so-called M1 crystal phase that contains all four elements in the metal positions of the structure. In the present work, an M1 phase composed only of Mo and V has been prepared by hydrothermal synthesis. High crystallinity was achieved by applying a synthesis temperature of 200 °C. The phase-pure mixed MoV oxide was studied as catalyst in the oxidation of propane. In contrast to previous reports, the desirable oxidation product acrylic acid is formed over the Te-free M1 structure in significant amounts, implying that Te is not necessarily required as a component of the active ensemble responsible for selective oxygen insertion. The MoV M1 oxide is, however, less selective compared to that of the M1 structure composed of the four metals Mo, V, Te, and Nb. The reason has been determined by applying a combination of synchrotron-based single crystal structure analysis and near-ambient-pressure X-ray photoelectron spectroscopy. Determination of the crystal structure of MoV M1 oxide reveals partial occupation of sites in the hexagonal channels of the M1 structure by V, which are occupied by Te in MoVTeNb M1 oxide. Hydrolysis of the M-O bonds (M = V, Te) under reaction conditions leads to migration of the metal in the hexagonal channels to the surface. Accumulation of more than 50 at % V on the surface of the MoV M1 oxide most likely causes postcombustion of formed acrylic acid, whereas enrichment of Te at the surface of MoVTeNb M1 oxide results in dilution of surface V⁵⁺ species and, consequently, high selectivity.

Full text of DOI:10.1021/acscatal.7b00130

Subscriber access provided by University of Newcastle, Australia
Article
The Impact of the Bulk Structure on Surface
Dynamics of Complex Mo-V-based Oxide Catalysts
Annette Trunschke, Johannes Noack, Sergej I. Trojanov, Frank Girgsdies, Thomas Lunkenbein, Verena
Pfeifer, Michael Hävecker, Pierre Kube, Christoph Sprung, Frank Rosowski, and Robert Schloegl
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.7b00130 • Publication Date (Web): 21 Mar 2017
Downloaded from http://pubs.acs.org on March 25, 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.
ACS Catalysis 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 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
The Impact of the Bulk Structure on Surface Dynamics of Complex MoꢀVꢀbased
Oxide Catalysts
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
Annette Trunschke,1,* Johannes Noack,1,2 Sergej Trojanov,3 Frank Girgsdies,1
Thomas Lunkenbein,1 Verena Pfeifer,1,4 Michael Hävecker,1,5 Pierre Kube,1 Christoph Sprung,1
Frank Rosowski,6 Robert Schlögl1
1Department of Inorganic Chemistry, FritzꢀHaberꢀInstitut der MaxꢀPlanckꢀGesellschaft,
Faradayweg 4ꢀ6, 14195, Berlin, Germany
2UniCatꢀBASF Joint Lab, Technische Universität Berlin, Sekr. EW K 01, Hardenbergstraße 36,
10623 Berlin, Germany
3HumboldtꢀUniversität zu Berlin, Institut für Chemie, BrookꢀTaylorꢀStr. 2, 12489 Berlin,
Germany
4Catalysis for Energy, Group EMꢀGKAT, HelmholtzꢀZentrum Berlin für Materialien und
Energie GmbH, Elektronenspeicherring BESSY II, AlbertꢀEinsteinꢀStr. 15, 12489 Berlin,
Germany
5Department of Heterogeneous Reactions, MaxꢀPlanckꢀInstitut für Chemische
Energiekonversion, Stiftstr. 34ꢀ36, 45470 Mülheim a. d. Ruhr, Germany
6BASF SE, Process Research and Chemical Engineering, Heterogeneous Catalysis, CarlꢀBoschꢀ
Straße 38, 67056, Ludwigshafen, Germany
Corresponding author: trunschke@fhiꢀberlin.mpg.de
ACS Paragon Plus Environment
1

ACS Catalysis
Page 2 of 34
1
2
3
4
5
ABSTRACT
6
7
8
9
Mixed oxides composed of the four metals Mo, V, Te, and Nb are known to be efficient catalysts
in selective oxidation of lower alkanes. The outstanding catalytic performance of such mixed
oxides is attributed to the presence of the soꢀcalled M1 crystal phase that contains all four
elements in the metal positions of the structure. In the present work, a M1 phase composed only
of Mo and V has been prepared by hydrothermal synthesis. High crystallinity was achieved by
applying a synthesis temperature of 200°C. The phaseꢀpure mixed MoV oxide was studied as
catalyst in the oxidation of propane. In contrast to previous reports, the desirable oxidation
product acrylic acid is formed over the Teꢀfree M1 structure in significant amounts implying that
Te is not necessarily required as a component of the active ensemble responsible for selective
oxygen insertion. The MoV M1 oxide is, however, less selective compared to the M1 structure
composed of the four metals Mo, V, Te, and Nb. The reason has been figured out by applying a
combination of synchrotronꢀbased single crystal structure analysis and nearꢀambientꢀpressure
XPS. Determination of the crystal structure of MoV M1 oxide reveals partial occupation of sites
in the hexagonal channels of the M1 structure by V, which are occupied by Te in MoVTeNb M1
oxide. Hydrolysis of the MꢀO bonds (M=V, Te) under reaction conditions leads to migration of
the metal in the hexagonal channels to the surface. Accumulation of more than 50 at% V on the
surface of the MoV M1 oxide most likely causes postꢀcombustion of formed acrylic acid, while
enrichment of Te at the surface of MoVTeNb M1 oxide results in dilution of surface V5+ species,
and, consequently, in high selectivity.
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
ACS Paragon Plus Environment
2

Page 3 of 34
ACS Catalysis
1
2
3
4
KEYWORDS
5
6
7
8
9
selective oxidation, M1 structure, NAPꢀXPS, single crystal analysis, MoV, Te, propane, acrylic
acid
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
INTRODUCTION
Mixed oxides are widely applied as catalysts in selective oxidation reactions and
electrocatalysis.1ꢀ8 Conceptual understanding of bulk mixed oxides on the molecular level is,
however, limited due to the often very complex structural and surface chemistry of this materials
class. Activity and selectivity are frequently attributed to the crystal structure and the exposure of
specific surface facets. One example is the ab plane of the soꢀcalled “M1 phase” (ICSD 55097)9ꢀ
10
of MoVTeNb oxide.11ꢀ13 Crystalline MoVTeNb oxide exclusively composed of the M1 phase
catalyzes the selective oxidation of propane to acrylic acid and shows high activity and
selectivity, whereas isomorphous MoV oxide is unselective and transforms propane mainly into
carbon oxides.14
Selectivity to acrylic acid has been attributed to the function of tellurium in catalytic ensembles
at the catalyst surface.14 It has been postulated that Te is responsible for αꢀH abstraction of the
chemisorbed propene molecule, once formed by oxidative dehydrogenation of propane at
cooperating vanadium and molybdenum oxide sites.12 The activation of the first CꢀH bond in the
methylene group of the propane molecule is attributed to V5+=O surface species, while
O=Mo6+=O sites contribute to propene formation by abstraction of the second hydrogen atom.
The intermediate product propene may desorb and reꢀadsorb or migrate to Te4+ꢀO sites
responsible for the formation of a πꢀallylic intermediate by αꢀH abstraction. Subsequent oxygen
insertion is again attributed to molybdenum dioxo sites. The entire mechanistic concept has been
ACS Paragon Plus Environment
3

ACS Catalysis
Page 4 of 34
1
2
3
4
5
6
7
8
9
postulated taking into account the chemical complexity of the MoVTeNb mixed oxide assuming
structural definition of the surface in terms of an ideal cut of the bulk crystal structure.12
Differences in metal site occupancy hence will have an impact on the chemical composition of
the active ensemble,15 and on catalytic activity.16 Clear experimental proof of this concept is still
missing. Alternatively an active function of Te=O in propane activation has been proposed.17
According to the soꢀcalled reductionꢀcoupled oxo activation mechanism, the Te=O bond accepts
the proton from the propane molecule while the electron goes to adjacent reducible V or Mo
sites. The latter concept is in agreement with the experimentally observed enrichment of the
MoVTeNb oxide surface in tellurium and vanadium under reaction conditions of propane
oxidation.18ꢀ20
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 current work is aimed at probing the surface of MoV M1 oxide under working conditions
by synchrotronꢀbased nearꢀambientꢀpressure Xꢀray photoelectron spectroscopy (NAPꢀXPS) to
shed light on the differences in the catalytic properties of MoV and MoVTeNb oxide in
oxidation of propane to acrylic acid. For this purpose, phaseꢀpure MoV oxide composed of the
M1 structure has been prepared. The high crystallinity of the material permitted a synchrotronꢀ
based singleꢀcrystal structure analysis. Our results clearly show an impact of the crystalline bulk
structure on surface composition and reactivity of Moꢀbased mixed oxides in selective oxidation
of propane. The experiments illustrate critical relations between solidꢀstate and surface dynamics
in oxidation catalysis.
EXPERIMENTAL DETAILS
Highly crystalline MoVOx catalyst composed of the M1 structure was synthesized by a
hydrothermal method and subsequent washing procedure. A specially built reactor (Premex
ACS Paragon Plus Environment
4

Page 5 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
Reactor AG, Lengnau, Switzerland) from corrosion resistant Hastelloy Cꢀ22 (2.4602) was
utilized, which enables precise control of the reaction parameters. A mixture of 9.18 g
(NH4)6Mo7O24·4H2O (AHM, Merck, 52 mmol Mo) dissolved in 230 g water (MilliꢀQ, Merck)
and 3.30 g VOSO4 (Acros Organics, 12.9 mmol V) dissolved in 30 g water was loaded into a
hydrothermal reactor at room temperature. The vessel was heated to 200°C at a rate of 1°C/min
and hold at this temperature for 17 h. The reaction mixture is stirred during the whole experiment
at a rate of 100 rpm. The recorded reaction parameters are presented in Fig. S1.
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
After cooling to room temperature, a black solid (3.4 g) was isolated by filtration (pore 5 glass
frit), washed twice with distilled water and dried at 80°C for 16 hours. To remove amorphous
components, 1 g of the powder was washed with 25 ml oxalic acid solution (0.25 M, Acros
Organics) at 60°C for 30 min under stirring. The solid was centrifuged (5000 rpm, 30 min),
washed twice with 25 ml water and dried at 80°C over night. About 0.66 g of the washed sample
was recovered. Thermal treatment before analysis and catalytic testing was done in flowing
argon (flow rate 100 ml/min) at 400°C (heating rate 10°C/min) for 2 h in a rotary tube furnace
(Xerion, Freiberg, Germany), (catalyst ID 18075).
Powder XRD measurements were performed on a STOE STADI P transmission diffractometer
equipped with a primary focusing germanium monochromator (Cu Kα1 radiation) and DECTRIS
MYTHEN 1K position sensitive solidꢀstate detector. The samples were mounted in the form of
small amounts of powder sandwiched between two layers of polyacetate film and fixed with a
small amount of Xꢀray amorphous grease. The powder pattern was analyzed using the TOPAS
software (version 5, Bruker AXS 2014) by whole powder pattern fitting. Initially, a Pawley fit
based on the M1 unit cell was used to judge the phase purity. After obtaining a crystal structure
ACS Paragon Plus Environment
5

ACS Catalysis
Page 6 of 34
1
2
3
4
5
6
7
8
9
model from single crystal synchrotron data, this model was used for a Rietveld fit of the powder
data. In both cases no residual reflections not attributable to the M1 structure were observed.
The crystal structure of MoV M1 oxide was analyzed by single crystal analysis applying
synchrotron radiation at the MX 14.2 beamꢀline at BESSY II of the HZB (HelmholtzꢀZentrum,
Berlin. The crystal structure was refined with SHELXLꢀ2014 (G. M. Sheldrick, 2014), using the
coordinates of Sadakane et al. as starting model.21 However, as the z value of the origin in space
group Pba2 can be freely chosen, we shifted the average plane of equatorial oxygen atoms to z =
0.5. Initially, all framework metal atoms were refined as mixed Mo/V sites. Several sites, which
converged sufficiently close to full Mo occupancy, were subsequently treated as Mo only sites.
Before the finalization of the structure model, all metal sites were refined anisotropically, while
oxygen atoms had to be treated isotopically. Since the anisotropic displacement parameters for
M2 and M12 indicated possible site splitting, these sites were subsequently split and refined
isotropically. Some residual electron density peaks which were observed in the hexagonal and
heptagonal channels (apart from M12) were, judging from their distances to other atoms,
considered to represent either oxygen atoms associated with M12, water molecules, or
ammonium ions. In all cases, these were treated as oxygen atoms. Since the low electron density
and overlapping contact radii indicated only partial occupancy, the occupancy was refined while
the isotropic thermal parameter was arbitrarily fixed at 0.02 Å2 to avoid strong parameter
correlations.
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
Scanning electron microscopy (SEM) images were recorded on Hitachi Sꢀ4800 scanning
electron microscope equipped with field emission gun (FEG) with an acceleration voltage of 1.5
kV. Prior to imaging, the samples were dispersed on a conductive carbon tape (plano). Elemental
ACS Paragon Plus Environment
6

Page 7 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
analysis was obtained by using Xꢀray dispersive spectroscopy (EDX). The measurement was
carried out at 15 kV and sampled at an EDAX detector connected to the Hitachi Sꢀ4800.
Scanning transmission electron microscopy (STEM) was conducted on a double corrected
JEOL JEMꢀARM200F equipped with a CEOS CESCOR and a CEOS CETCOR hexapole
aberration correctors for probe and image forming lenses, respectively. A cold FEG (CFEG) is
used for electron extraction. Images were recorded with an acceleration voltage of 200 kV
inserting a high angle annular dark field (HAADF) or annular bright field (ABF) detector after
the sample. The sample was prepared via cross sectioning the MoV M1 oxide crystals.
The surface area determination was carried out in a volumetric N2 physisorption apparatus
(Autosorbꢀ6ꢀB, Quantachrome) at 77 K. The sample was degassed in dynamic vacuum at a
temperature of 150°C for 2 h prior to adsorption. The adsorption isotherm was measured in the
range p/p0 = 0.0–0.35. The specific surface area was calculated according to the BET method in
the pressure range p/p0 = 0.05–0.35 using 11 data points. Pore size distribution in the micropore
range was analyzed applying the method of Horvath and Kawazoe based on adsorption in the
pressure range p/p0 = 0.0ꢀ0.001.22
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 gasꢀphase propane oxidation was carried out in fixedꢀbed tubular reactors in continuous
flow mode in a 10ꢀfold parallel reactor unit (ILS, Germany) that has been described in detail
before.23 Typically, 0.5 g of powdered catalyst was first binderꢀfree pressed under 1.25 MPa,
crushed and sieved to a particle size of 150ꢀ200 ꢁm. Then, 40 mg of the catalyst were loaded into
a tubular reactor with an inner diameter of 2 mm. The feed was composed of C3H8, O2, H2O, and
N2 in a molar ratio of 3 : 6 : 20 : 71, and 3 : 6 : 40 : 5 vol%, respectively. The reaction was
carried out at 250 – 300°C. Gas analysis was performed by an online gas chromatograph (7890A,
Agilent). A combination of two capillary columns (Poraplot Q and Plot Molesieve) in connection
ACS Paragon Plus Environment
7

ACS Catalysis
Page 8 of 34
1
2
3
4
5
6
7
8
9
with a thermal conductivity detector is used to analyze the permanent gases. A combination of
two capillary columns (HPꢀFFAP and Poraplot Q) connected to a flame ionization detector is
used to analyze alkanes, olefins, and oxygenates. A carbon balance of 100 ± 5% was calculated.
Inꢀsitu XPS was performed at the synchrotron facility BESSY II of the HZB (Helmholtzꢀ
Zentrum Berlin) using tunable, monochromatic radiation of the ISISS (Innovative Station for Inꢀ
Situ Spectroscopy) beam line. The spectra were collected in the presence of reactive gases at
elevated temperature using a nearꢀambientꢀpressure XPS (NAPꢀXPS) system described in detail
elsewhere.24 Briefly, 10.7 mg of MoV oxide powder was pressed (5 tons, disk of diameter 8 mm)
into a selfꢀsupporting pellet. Mounted onto a sapphire sample holder, the pellet was inserted in
the reaction and XPS measurement cell, in which mixtures of O2/C3H8/He and O2/C3H8/H2O(g)
with volume flows of 2 sccm/0.5 sccm/1.7 sccm were introduced via calibrated Bronkhorst mass
flow controllers. A constant total pressure of 25 Pa was established by a pressure regulating
valve. The overall gas phase composition was monitored by online mass spectrometry and,
furthermore, a proton transfer reaction mass spectrometer (PTRꢀMS, IONICON) was used to
determine the abundance of acrylic acid. Heating of the pellet was realized from the sample back
using an NIR laser, whose power was adjusted through a PID feedback loop controlled by a Kꢀ
type thermocouple in contact with the sample surface. The heating rate was 5 K/min up to a final
temperature of 270°C. Core level spectra of O 1s, V 2p, Mo 3d, and C 1s were collected at
constant kinetic energies of the photoelectrons of 150 eV and 750 eV, resulting in inelastic mean
free paths (IMFP) of ~0.6 nm and ~1.6 nm, respectively, according to the model of Tanuma et
al.25 The XPS spectra were normalized by the storage ring current and the photon energy
dependent incident photon flux. The binding energy (BE) scale was calibrated with respect to the
valence band onset. V 2p3/2 peak deconvolution and quantitative element abundance calculations
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
ACS Paragon Plus Environment
8

Page 9 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
were performed with the commercially available CASA XPS data analysis software (Neil Farley,
www.casaxps.com) after subtracting a Shirleyꢀtype background and by taking into account the
photon energy dependent photoꢀionization cross sections.26 The error bar of the absolute
elemental composition can be estimated up to 20 % due to uncertainties in the monochromatic
photon flux, cross sections, and peak areas. However, only the uncertainty in peak area
determination contributes to relative uncertainties in an experimental series. Therefore, the
relative error bar in the XPS figures can be estimated to be ~2 %.
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
RESULTS AND DISCUSSION
1. Structural Characterization of the Catalyst
Xꢀray powder diffraction (Fig. S2a) and SEM images (Fig. S3) of (Mo,V)Ox reveal phase
purity and high crystallinity of the synthesized material. The particle size distribution is broad
and the shape of the particles not uniform. Very small particles are also observed (Fig. S3).
However, highꢀresolution electron microscopy of the small particles selected by transmission
electron microscopy reveal that also these particles are composed of the crystalline M1 phase
(Fig. S4). The bulk metal composition determined by EDX corresponds to 29.5 at% V and 70.5
at% Mo.
The bulk structure of individual oxide particles of (Mo,V)Ox was investigated by ABFꢀSTEM
(Fig. 1). As opposed to the HAADF detector, the ABF detector allows to image light elements in
the vicinity of heavy atoms. The image resembles the complex nanoꢀstructural array of metal
cations typical for orthorhombic (Mo,V)Ox. In addition, the images display a partial occupation
of hexagonal and heptagonal channels for these complex metal oxides (indicated by the red
arrow), which cannot be observed using a HAADF detector (Fig. S5).
ACS Paragon Plus Environment
9

ACS Catalysis
Page 10 of 34
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 1. Imaging the nanostructure of the M1 phase viewed along [001]. (A) Atomic resolution
ABFꢀSTEM image of (Mo,V)Ox;. With the help of the color the various polyhedrons and channel
sites can be distinguished; The color corresponds to the color of the polyhedrons used in Figure 2
(pentagonal biꢀpyramidal unit in yellow, octahedrons surrounding the pentagonal unit in dark
blue, linking octahedrons in light blue, and channel sites in red); In order to highlight the
occupation of the channel sites the inverted ABFꢀimage is shown in (B) (red arrows indicates
occupation); The original image is presented in Fig. S6.
The size of some crystallites in the polycrystalline (Mo,V)Ox allowed a synchrotron Xꢀray
single crystal analysis. A summary of the crystallographic data is given in Table 1, and Tabs. S1ꢀ
S2. The structural details of the orthorhombic M1 phase were first elucidated for the
(Mo,V,Te,Nb)Ox system by combined Rietveld refinement of synchrotron Xꢀray and neutron
powder diffraction data.27 Later, (Mo,V,Sb)Ox was the first M1 type structure to be characterized
by synchrotronꢀbased single crystal analysis.21 Generally speaking, M1 belongs to the family of
molybdenum bronzes, which have a ca. 4 Å crystallographic axis in common. This distance
represents the length of two M–O bonds forming the backbone of linear chains of cornerꢀsharing
polyhedra in this direction.
ACS Paragon Plus Environment
10

Page 11 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
Table 1. Crystallographic data for (Mo,V)Ox
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
Formula
M (g molꢀ1)
T (K)
Mo7.88V2.61O29.25
1356.76
100(2)
λ (Å)
0.85507
crystal system
space group
a (Å)
Orthorhombic
Pba2 (No. 32)
21.0590(7)
26.5310(8)
3.9810(10)
2224.2(6)
4
b (Å)
c (Å)
V (Å3)
Z
calcd (g cmꢀ3)
(mmꢀ1)
4.052
ρ
8.435
ꢀ
F000
2500
measured reflections
independent reflections
data / restraints / parameters
goodness of fit
31034
5381 (Rint = 0.0485)
5381 / 1 / 232
1.061
0.0474
R1 (I > 2σ(I))
wR2 (all data)
0.1206
Flack parameter
extinction coefficient
largest diff. peak / hole (e Åꢀ3) 4.617 / ꢀ2.091
0.045(16)
0.00182(19)
ACS Paragon Plus Environment
11

ACS Catalysis
Page 12 of 34
1
2
3
4
5
B
A
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 2. Polyhedral representation of the M1 structure (A), and asymmetric unit of the
(Mo,V)Ox M1 crystal structure with atom numbering scheme (B); Axial oxygen atoms are
numbered like the corresponding metal atoms and have been omitted for clarity.
One prominent feature of the M1 structure is the presence of a pentagonal building block
(colored in yellow in Fig. 2 A). This unit consists of a MO7 pentagonal bipyramid, which shares
its equatorial edges with five MO6 octahedra (colored in dark blue in Fig. 2 A). This structural
motif is not only shared with a family of supramolecular polyoxometallate compounds,28 but has
also been shown to play a role as an actual building block during the M1 synthesis.21, 29ꢀ30
Furthermore, the same motif can be found in several related molybdenum bronze structures,
most notably the Mo5O14 type.31 In both structures, the pentagonal motifs are linked by
additional MO6 octahedra (colored in light blue in Fig. 2 A) via corner sharing in equatorial
plane, i.e. the plane perpendicular to the 4 Å c axis. In contrast to Mo5O14, which is more densely
packed, the linkage in the M1 structure involves 6ꢀ and 7ꢀmembered rings, forming channels
along the c axis (Fig. 2). These channels, of which the 7ꢀmembered ones can act as micropores,32
may contain a variety of statistically occupied species, such as water molecules or further metal
atoms (indicated in red in Fig. 2 A).
ACS Paragon Plus Environment
12

Page 13 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
For the crystal structure analysis presented here, the atom numbering scheme used by DeSanto
et al.9 and most subsequent authors was adopted (Fig. 2B). The general structural characteristics
agree well with the previously published studies. Inside the oxygen polyhedra, the metal atoms
exhibit a pronounced trend for offꢀcenter distortion, which is especially prominent along the c
axis of the M1 structure, resulting in alternating longer and shorter M–O distances in this
direction. In conjunction with the tendency of neighboring metal centers to avoid too short
metalꢀmetal distances, this leads to a characteristic alternating “up and down” pattern
(puckering) of metal atoms with respect to the average metal plane (Fig. 3). It should be noted
that in the (Mo,V,Sb)Ox system, extensive disorder with respect to this outꢀofꢀplane
displacement has been reported.21 For all framework metal sites, there were two alternative
positions above and below the average (“pseudoꢀmirror”) plane. This site splitting yielded
unequal occupation, i.e. there were majority and minority sites found in this disorder.
Furthermore, the refined distribution between the two alternative positions was generally similar
for all sites, hinting a possible correlation. When regarded separately, the majority sites exhibit
the same alternating displacement pattern, which is seen in all other studies. Thus, we tend to
believe that this effect could be better explained as being caused by unresolved twinning, rather
than a correlated disorder across all metal sites.
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
ACS Paragon Plus Environment
13

ACS Catalysis
Page 14 of 34
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)
(B)
Figure 3. Outꢀofꢀplane displacement (puckering pattern) of metal sites in the M1 structure.
(A) “Top” view onto the cꢀaxis; only sites around the equatorial plane are shown, i.e. axial
oxygen atoms have been omitted for clarity. (B) “Side” view along the bꢀaxis. Atoms close to the
z = 0 and z = 0.5 planes are shown in gray, while significant displacements of metal centers from
the average plane are indicated in white (z = 0.5+δ) and black (z = 0.5–δ), respectively.
Concerning the distribution of Mo and V on the various metal sites (for atomic numbering see
Fig. 2B), our results are also consistent with the trends seen in previous reports: the sites
belonging to the pentagonal unit (M9 and M5, M6, M8, M10, M11) are exclusively occupied by
Mo, while most linker sites show a mixed Mo/V occupancy. Table 2 compares the metal
distribution found in several studies.
ACS Paragon Plus Environment
14

Page 15 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
Table 2. Comparison of metal distribution and resulting (i.e. diffraction derived)
composition from several published M1 single crystal (SC) and Rietveld (R) structure
analyses. For the framework sites M1-M11, the Mo occupancy is given, and the reminder
of the site is filled with V unless noted otherwise. In contrast, the channel sites M12 and
M13 are only partially occupied. For convenient comparison, the alternative site
numbering scheme used by Ishikawa et al.33 is added in parentheses
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
SCthis work
Mo,V
SC33
Mo,V
76:24
76:24
SC21
R34
R9
R35
Composition
framework
total
Mo,V,Sb Mo,V,Nb,Te Mo,V,Nb,Te Mo,V,Ta,Te
79:21
81:19:0
77:18:5
73:17:10:0 78:12:10:0
82:14:4:0
75:13:4:8
75:25
68:16:9:6
71:11:9:9
M1 (M11)
M2 (M7)
M3 (M10)
M4 (M9)
M7 (M8)
M5 (M5)
M6 (M1)
M8 (M4)
M10 (M3)
M11 (M2)
M9 (M6)
M12 [hex.]
M13 [hept.]
0.26
0.11
0.32
0.46
0.74
0.38
0.58
1
1
0.37
0.34
1
1
0.75
0.84
0.63
0.24
0.24
0.39
0.46
0.55
0.62
0.53
0.88
0.45
0.33
0.60
0.62
0.68
1
0.59
1
1
1
1
1
1
1
1
ꢀ
1
0.8
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0.49 V
ꢀ
1
0.52 Sb
ꢀ
0 / 1 Nb
0.69 Te
ꢀ
0 / 1 Nb 0.56 / 0.44 Ta
0.74 Te
0.20 Te
0.67 Te
0.18 Te
ꢀ
The sites M1 and M2 are special positions located on 2ꢀfold rotational axes. M1 resides
between two 7ꢀmembered rings and seems to have a tendency to be rich in V, with the exception
of the (Mo,V,Te,Nb)Ox and (Mo,V,Te,Ta)Ox Rietveld studies of DeSanto et al.35 Position M2 is
located between two 6ꢀmembered rings and is mostly found to be rich in Mo. This M2 site shows
some interesting characteristics, which, to our knowledge, have not been discussed in detail
before. In our study, we observed a strongly anisotropic thermal displacement parameter, with its
long axis pointing approximately to the centers of the 6ꢀmembered rings. In the recent single
ACS Paragon Plus Environment
15

ACS Catalysis
Page 16 of 34
1
2
crystal study of (Mo,V)Ox by Ishikawa et al.,33 the same effect can be seen in the supporting CIF
data. For the final refinement, we decided to use an alternative model, describing M2 with an
isotropic thermal parameter, but liberating the atom position from the special position
restrictions, effectively splitting the site symmetrically around the twofold axis. The refined
position is displaced slightly towards the channel centers. While not explicitly discussed by
Sadakane et al., an analogous splitting is depicted in the supporting material for the majority
component of the M2 site in MoVSbOx.21 Thus, we believe that this outꢀofꢀaxis displacement is a
general characteristic of the M1 structure. There are two factors, which can be made responsible
for this disorder around the symmetry element. First, molybdenum and vanadium in their d0 state
show different preferences concerning the mode of offꢀcenter distortion in an octahedral
environment. V5+ (and also V4+ to a slightly lesser extent) prefers a displacement towards one
ligand (the vanadyl oxygen atom), i.e. along an axis of the octahedron. Mo6+, on the other hand,
is usually displaced towards an octahedron edge between two neighboring ligand atoms.36 For
the sites M1 and M2, the twofold rotational symmetry axis coincides with one axis of the
coordination octahedron. Thus, at the Vꢀrich M1 site, the site symmetry and the preferred
vanadium distortion along the axis are compatible with each other. In contrast, this symmetry
cannot accommodate the preferred distortion geometry for the Moꢀdominated M2 site, thus
resulting in statistical disorder of the metal atom around the rotation axis. The second factor
comes into play when the hexagonal channels contain additional metal atoms on the M12 site,
like, e.g., Te or Sb. In our crystal structure analysis, we found significant electron density at the
M12 site, which was too close (around 2.0 ± 0.2 Å) to the nearest oxygen sites (O16, O20) to
represent a water oxygen or ammonium nitrogen atom. Since Mo and V are the only metals in
our sample, we interpret this electron density as a statistically occupied vanadium site. As partial
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
ACS Paragon Plus Environment
16

Page 17 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
occupation seems to be a general feature of the M12 site, it is quite conceivable that this kind of
disorder will also affect the M2 site, to which is tightly connected via two common oxygen
atoms. In other words, we assume that the statistical disorder of the M12 site in the channels
enhances the already present outꢀofꢀaxis distortion at the mostly Moꢀbased M2 site by metalꢀ
metal repulsion. It should be noted that the published Rietveld analyses assign less Mo (more V)
to M2 than the single crystal studies. In the light of the evidence for M2 site splitting around the
twofold axis, one may argue that the Rietveld results could be biased in this respect. The splitting
would cause a dilution of electron density in the center, which may then be mistaken for a lower
average atomic number in an isotropic, symmetry restricted refinement model.
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 origin of the vanadium atoms in the hexagonal channels is not completely clear. For
comparison, we also investigated a crystal of the washed sample before thermal activation, and
found no significant electron density at the M12 site. Furthermore, the anisotropy (or splitting,
depending on the model used) of the associated M2 site was clearly less pronounced than in the
activated sample. Thus, the vanadium atoms must have migrated onto the M12 site during the
thermal activation treatment. There are two possibilities for the origin of this extraꢀframework
vanadium: either, it came from the surrounding framework sites, leaving some cation vacancies
there, or it migrated into and along the channels from the outside (surface/amorphous material).
Considering the behavior of M1ꢀtype catalysts under changing environmental conditions,20 the
migration hypothesis seems more likely. A direct crystallographic test by including framework
vacancies in the refinement model is, however, not possible. Dropping the boundary condition
that the Mo and V occupations of regular sites must add up to unity would leave the refinement
model mathematically underdetermined.
ACS Paragon Plus Environment
17

ACS Catalysis
Page 18 of 34
1
2
3
4
5
6
7
8
9
Inside the heptagonal channels, only low electron density was found. Judging from the ca. 3 Å
distance to the nearest oxygen atoms, these species can be interpreted as partially occupied water
molecules or residual ammonium ions. Specifically, the M13 site of DeSanto et al.9 was found to
be empty. This result is in line with a certain microporosity of the MoV M1 oxide (Fig. 4). The
external surface area of 9.8 m²/g can be differentiated from micropore surface area (18.1 m²/g)
by the tꢀplot method.22 The total surface area as determined by the BET method corresponds to
27.9 m²/g. The calculated pore diameter of 5 Å perfectly reflects the structural heptagonal pores
of the M1 phase. Accordingly, these pores are at least accessible for nitrogen. Adsorbed propane
is characterized by crossꢀsectional area of > 30 Å in average meaning that the molecule may be
adsorbed only at the pore mouth.37
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. Adsorption isotherm of nitrogen at ꢀ196°C at low relative pressures used for micropore
and surface area analysis (A), and pore size analysis of (Mo,V)Ox according to the Horvathꢀ
Kawazoe (HK) method (B).22
2. Oxidation of Propane
Due to its low thermal stability, MoV M1 oxide cannot be measured in propane oxidation
under the same reaction conditions as MoVTeNb M1 oxide. The typical working temperature of
MoVTeNb M1 oxide in selective oxidation of propane to acrylic acid in an oxygenꢀrich feed
ACS Paragon Plus Environment
18

Page 19 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
(O2:C3H8=2) that contains 20ꢀ40 vol% steam is 350ꢀ420°C. Under such conditions the yield of
the most desired reaction product acrylic acid approaches 30ꢀ50% depending on the preparation
method of the solid that influences chemical composition, particle size, and morphology.18 The
lower thermal stability of MoV M1 oxide is reflected in catalyst deactivation at T>300°C due to
decomposition of the mixed oxide into its binary oxides. Therefore, MoV M1 oxide has been
studied in the temperature range between 250 and 300°C. At these temperatures the MoV M1
oxide bulk structure is stable (Fig. S2). No phase separation and only slight changes in the lattice
parameters have been observed after catalysis (see caption of Fig. S2 for lattice parameters). The
contact time has been chosen to adjust the propane conversion between 5 and 15 % (Fig. 5).
Undesired molecules CO, CO2, and acetic acid are the major reaction products under these
conditions over MoV M1 oxide. Acetone is formed in lower amounts. Products of selective
reactions comprise propene (17ꢀ10%), and acrylic acid (15ꢀ10%). Most interestingly, the notable
amount of acrylic acid formed over MoV M1 oxide reveals that the presence of tellurium in the
catalyst is not necessarily required for oxygen insertion reactions. The experiment does,
however, not exclude that Te may facilitate acrylic acid formation.17, 38 Water has an impact on
the selectivity to propene, which increases slightly by decreasing the steam content from 40% to
20% (compare Fig. 5 and Fig. S7) at the expense of acetic acid and carbon oxides. The
selectivity to acrylic acid is not very much affected in this range of steam concentrations (Fig. 5
and Fig. S7). The selectivity to acetone and propene decrease with increasing conversion in favor
of acrylic acid formation (Fig. 5). For comparison, conversion and product distribution in
propane oxidation over MoVTeNb M1 oxide at the usually applied reaction temperatures (380ꢀ
420°C) and low reaction temperatures (250ꢀ300°C) are shown in Figs. S7ꢀ8. The temperature
may have an impact on the solidꢀstate kinetics of the catalyst with implications in terms of
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
ACS Paragon Plus Environment
19

ACS Catalysis
Page 20 of 34
1
2
3
4
5
6
7
8
9
oxygen activation and the abundance of differently reactive oxygen species on the catalyst
surface. An increased concentration of electrophilic oxygen species at lower temperatures should
be reflected in the product distribution. MoVTeNb M1 oxide is less active than MoV M1 oxide
at low temperature (Fig. S7), but the products seem to follow roughly the same SꢀX curves at
low and at high temperatures. This indicates that the reaction mechanism is not significantly
changed due to the difference in the reaction temperature. The diversity of oxygenates is higher
over MoV M1 oxide (Fig. S8). The trends in selectivity with increasing conversion (Fig. 5)
suggest that lower selectivity towards acrylic acid over MoV M1 oxide compared to MoVTeNb
M1 oxide is a result of postꢀcombustion of formed selective oxidation products.
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 5. Conversion of propane and product distribution over (Mo,V)Ox as a function of
temperature; Feed composition: 3% C3H8, 6% O2, 71% N2, and 40% H2O; Contact time: 0.36 g s
mlꢀ1.
This is also confirmed by an experiment in which the stability of acrylic acid over the catalyst at
reaction temperature was tested. The acrylic acid is largely decomposed to acetic acid, CO, and
CO2 (Fig. S9).
ACS Paragon Plus Environment
20

Page 21 of 34
ACS Catalysis
1
2
3
4
5
6
3. Surface Dynamics of (Mo,V)Ox
7
8
9
The surface composition of vanadium and molybdenum has been monitored by NAPꢀXPS
under operation of the catalyst at a reaction temperature of 270°C and a total pressure of 25 Pa.
Fig. 6 A shows the normalized vanadium content with time on stream in alternating dry and wet
feed (as indicated in the figure by steam ON and OFF) in surface sensitive measurement mode
that comprises an inelastic mean free path of 0.6 nm according to the model of Tanuma et al.25
The initial surface composition of MoV M1 oxide is in good agreement with the bulk
composition of V:Mo=29.5:70.5 at%. The majority of V atoms are present in oxidation state 4+,
but almost one quarter of the surface vanadium atoms occurs in the highest oxidation state.
Alternative wet and dry feed causes the appearance and disappearance of acrylic acid,
respectively, as monitored by proton transfer reaction mass spectrometry (black curve in Fig. 6
A). The surface composition does not follow these changes. Instead, a substantial increase in the
total surface vanadium concentration up to more than 50 at% is noticed (brown curve in Fig. 6
A) with increasing time on stream that goes along with a progressive deactivation of the catalyst
in terms of acrylic acid formation. The catalytic results presented for the plug flow reactor in Fig.
5 have been obtained under steady conditions, but changes in the selectivity to acrylic acid can
also be observed under normal conditions by consulting the time on stream behavior. The
selectivity to acrylic acid decreases within the first 30 minutes after reaching the reaction
temperature, however only in wet feed (Figs. S10ꢀ11). At the same time, the conversion of
propane increases slightly. The observations are in agreement with an enrichment of V5+ on the
surface of the catalyst in wet feed as observed by NAPꢀXPS. The deceleration of the acrylic acid
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
ACS Paragon Plus Environment
21

ACS Catalysis
Page 22 of 34
1
2
3
4
5
6
decay in the XPS cell is attributed to the low total pressure, which may result in different solidꢀ
states kinetics, and differences in the flow conditions.
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
B
60
50
40
30
20
10
60
50
40
30
20
10
steam
OFF
steam
ON
ON
OFF ON
OFF
ON
OFF
ONOFF
ON
V5+ II
V5+ I
V5+ II
V5+ I
V4+
V4+
0
0
1,0
1,0
0,5
0,0
0,5
0,0
0
200
400
600
800
1000
1200
0
200 400 600 800 1000 1200 1400 1600 1800
time on stream / min
time on stream / min
Figure 6. Normalized abundance of different V species at the surface (0.6 nm IMFP) of
(Mo,V)Ox (A), and reꢀanalyzed data of (Mo,V,Te,Nb)Ox published previously20 (B) measured by
XPS in presence of flowing mixtures of O2/C3H8/He (steam OFF) and O2/C3H8/H2O(g) (steam
ON), respectively, (volume flows of 2 sccm/0.5 sccm/1.7 sccm) at T=270°C (A) and T=400°C
(B), respectively, and a total pressure of 25 Pa; The V content is given in at% with respect to the
sum of metals at the surface; The values are represented in an accumulative way: The blue curve
corresponds to the surface concentration of V4+, the red curve corresponds to the sum of the
surface concentration of V4+ and V5+(I), and the brown curve accounts for the overall surface
concentration of V composed of the sum of V4+, V5+(I) and V5+(II) (for assignment see text); The
black curve shows the acrylic acid abundance.
The changes in the peak profile of the V 2p3/2 peak shape with time after switching for the first
time from dry to wet feed are depicted in Fig. 7 (left). The V 2p3/2 peak is initially centered at
approximately 516 eV in dry feed with a small shoulder at higher binding energy (BE) around
517 eV. The shoulder got immediately more pronounced when switching to wet feed and
steadily increased in its intensity with time in wet feed, which indicates progressive oxidation of
vanadium from 4+ to 5+.39ꢀ40 The Mo 3d5/2 and Mo 3d3/2 peaks at 232.4 and 235.6 eV,
ACS Paragon Plus Environment
22

Page 23 of 34
ACS Catalysis
1
2
respectively, (Fig. 7, right) indicate that Mo exists essentially in oxidation state 6+,41 but small
contributions of molybdenum 5+ cannot be excluded.18 The oxidation state of molybdenum at
the surface remains constant under all reaction conditions. The asymmetry in the V 2p3/2 peak
develops progressively with time on stream in alternating wet and dry feed (Fig. 8, left). Again,
no change is observed concerning the state of molybdenum (Fig. 8, right).
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 7. Changes in the profile of V 2p3/2 (left), and Mo 3d (right) signals of (Mo,V)Ox with
time after first time switching from dry (red spectra) to wet feed (blue spectra) (140 min – 330
min time on stream in Fig. 6) ; For reaction conditions see caption of Fig. 6.
ACS Paragon Plus Environment
23

ACS Catalysis
Page 24 of 34
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 8. Long term changes (0 min – 1200 min time on stream in the experiment shown in Fig.
6) in the profile of the V 2p3/2 (left) and the Mo 3d signal (right) for (Mo,V)Ox during switching
from dry (red spectra) to wet (blue spectra) feed; For reaction conditions see caption of Fig. 6.
Xꢀray absorption spectroscopy (XAS) at the V L3,2 edges has been applied to get additional
information about the vanadium oxidation state and vanadium content at the surface. The overall
energy position of the V Lꢀedges at around 518eV (V L3ꢀedge) and 524.6eV (V L2ꢀedge) is a
probe of the vanadium oxidation state.42 The shift to higher photon energies as it is observed in
the course of the experiment with the (Mo,V)Ox catalyst indicates a steadily increasing vanadium
valence in agreement with the XPS V2p core level measurements (Fig. 7). While boundꢀbound
V2p ꢀ> V3d transitions result in resonance structures at the V L absorption edges, continuum
transitions become visible as a stepꢀlike intensity increase (“edge jump”). The edge jump is
proportional to the number of absorbers, i.e. to the V content in the material. The edge jump
intensity at the V L–edges at about 521 eV and 528 eV increases while the catalyst is exposed to
different conditions with the most striking effect observed when dosing water vapor to the feed,
ACS Paragon Plus Environment
24

Page 25 of 34
ACS Catalysis
1
2
3
4
5
6
as it becomes obvious in Fig. 9. Hence, XAS confirms both the increase of vanadium oxidation
state and the enrichment of vanadium on the surface of MoV M1 oxide during the experiment.
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
V L-edges
MoVOx-M1
270 °C
with steam
(final catalyst state)
no steam
(initial catalyst state)
512 514 516 518 520 522 524 526 528
excitation energy / eV
Figure 9. V L3,2ꢀnear edge Xꢀray absorption fine structure of (Mo,V)Ox measured in presence of
flowing mixtures of O2/C3H8/He and O2/C3H8/H2O(g), respectively, (volume flows of
2 sccm/0.5 sccm/1.7 sccm) at T=250°C.
The peak profile of the V 2p3/2 signal has been analyzed for all data points applying a
consistent constrained fit model to describe the variation in the data with high accuracy. An
example of a fit of the V 2p3/2 core level of (Mo,VO)x is shown in Fig. 10 and a summary of the
fit parameters can be found in Tab. S3. A fully constrained fit requires introduction of 3 fit
profiles with binding energies of 516.1, 517.1, and 517.3 eV, respectively. The peak at 516.1 eV
binding energy can readily be assigned to V4+ species while a BE of about 517 eV and higher
suggests the presence of V5+ components, in good agreement with the literature40 and references
therein. A smaller FWHM is characteristic for V5+ components, e.g. in binary vanadium oxides.40
ACS Paragon Plus Environment
25

ACS Catalysis
Page 26 of 34
1
2
3
4
5
6
7
8
9
Consequently, the slightly smaller FWHM of the fit profiles at BE 517 eV of about 1.25e Vꢀ1.35
eV compared to the FWHM of 1.4 eVꢀ1.45 eV found for the species at a binding energy of
516.1eV supports the assignment to V5+ and V4+ species, respectively.
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 10. Least square fit of V 2p3/2 XP core level spectra of (Mo,V)Ox without and with steam
in the feed. Black dots represent the experimental data, blue, red, and brown dots vanadium 4+,
5+ (I), and 5+ (II) species, respectively. Grey dots are the sum of fit profiles and a Shirley
background.
Literature values for BE of vanadium 2p core level lines vary considerably and make an
assignment based just on the binding energy challenging.43 Typically, no pronounced impact of
the local geometric arrangement around the vanadium atom on the V2p BE has been found.
Mekki et al. studied vanadium tellurite glasses and report only a very weak trend in BE shifts
with Te/V ratio.44 However, final state effects can significantly alter the BE of dispersed V5+
species.45ꢀ46 BE values of up to 518.7 eV have been reported for V5+ species. The inclusion of a
2nd V5+ species into the fit of the spectrum in steam (Fig. 10, bottom) is justified not only because
of the improvement of the fit quality but also because the variation in the data can be described
much more accurately. Furthermore, a function can be assigned to this additional V5+(II) species
ACS Paragon Plus Environment
26

Page 27 of 34
ACS Catalysis
1
2
3
4
5
6
7
8
9
by comparison with the simultaneously obtained catalytic performance over MoV (Fig. 6 A) and
MoVTeNb (Fig. 6B) M1 oxide. For the purpose of comparison of the selective MoVTeNb M1
oxide with the rather unselective MoV M1 oxide, the data of MoVTeNb M1 oxide published
previously20 have been reꢀanalyzed applying the same fit procedure (Fig. 6B). The results
elucidate the full power of inꢀsitu spectroscopies. Both catalysts show a rather constant amount
of V4+ that is not affected by dosing of steam while the V5+(II) species with a BE of 517.3 eV is
created in the presence of steam in the feed. On (Mo,V)Ox the V5+(II) species remains on the
surface also in dry feed to grow even more intense when steam is added to the feed in the next
cycle. In contrast, this vanadium 5+ (II) species is removed from the surface when MoVTeNbOx
is treated in the same way, though at higher temperature. Also, the amount of V5+(I) species
while steam is dosed remains rather constant during built up of V5+(II) species in the case of the
(Mo,V,Te,Nb)Ox mixed oxide while the abundance of V5+(I) species decreases steadily with the
continuous increase of V5+(II) species for (Mo,V)Ox, but not by the same amount, though,
resulting in a slight increase in average vanadium oxidation state with time from 4.3 to 4.6 (Fig.
S12). The sum of V4+ and V5+(I) based on the metal content is about 30 at% for MoV M1 oxide
and 17 at% for MoVTeNb M1 oxide, respectively. These numbers are in agreement with the
vanadium concentration in the bulk of MoV M1 oxide (29.5 at% as determined by EDX), and
MoVTeNb M1 oxide (18.8 at% as determined by Xꢀray fluorescence20), respectively. This
suggests that V4+ and V5+(I) species belong to the lattice of the oxide, while V5+(II) is related to
extraꢀframework vanadium species.
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 comparison of acrylic acid abundance and V5+(II) intensity reveals that the (Mo,V)Ox
catalyst seems to deactivate with respect to acrylic acid formation due to the enrichment of
V5+(II) species at the surface with time on stream (Fig. 6 A). In contrast, a correlation between
ACS Paragon Plus Environment
27

ACS Catalysis
Page 28 of 34
1
2
acrylic acid production and the overall V5+ abundance20 was observed for MoVTeNb M1 oxide
(Fig. 6 B). In parallel to the accumulation of vanadium on the surface in wet feed, enrichment by
tellurium has been observed on MoVTeNb M1 oxide.20 These surface dynamics have been
explained by hydrolysis of TeꢀO bonds within the channels of the M1 structure followed by
migration of Te hydroxide species to the outer surface. Since no deactivation of the MoVTeNb
oxide has been observed over very long times under varying feed conditions,20 the formation of
vanadium tellurates at the surface has been postulated. Single crystal analysis of the (Mo,V)Ox
under study revealed partial occupation of metal position M12 in the hexagonal channels by
vanadium. In analogy to the MoVTeNb M1 oxide, the V5+(II) species might be associated with
vanadium oxide species that migrated from position M12 to the surface of the MoV M1 oxide
under reaction conditions due to hydrolysis of VꢀO bonds in the channels of the bulk oxide.
Increasing amount of such vanadium oxide species with V in the highest oxidation state leads to
decreasing selectivity towards acrylic acid as observed in the NAPꢀXPS experiment (Fig. 6 A).
Hydrolysis and migration to the surface seem to occur faster in case of MoVTeNb M1 oxide
perhaps due to the difference in temperature. In the experiment reported previously,20 the
standard reaction temperature of 400°C has been applied for the thermally more stable
MoVTeNb M1 oxide. Furthermore, improved reversibility of this redox process is observed for
MoVTeNb M1 oxide (Fig. 6 B) compared to MoV M1 oxide (Fig. 6 A).
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
SUMMARY AND CONCLUSIONS
Highly crystalline mixed molybdenumꢀvanadium oxide composed of the M1 crystal structure
has been prepared by hydrothermal synthesis. Single crystal analysis of one particle of the
polycrystalline powder revealed that
ACS Paragon Plus Environment
28

Page 29 of 34
ACS Catalysis
1
2
3
4
5
6
• The central atoms of the pentagonal unit M{M}5 are exclusively occupied by Mo;
• Most linker sites exhibit mixed Mo/V occupancy;
7
8
9
• The M12 site located in the hexagonal channels is statistically occupied by vanadium;
• The incorporation of V at the M12 sites occurs during thermal pretreatment of the
catalyst at 400°C in Ar.
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 MoV M1 oxide has been studied in oxidation of propane in a temperature range where the
crystal structure is stable (250ꢀ300°C). Acrylic acid is also formed over the Teꢀfree M1 structure
in significant amounts implying that Te is not necessarily required as a component of the active
ensemble responsible for selective oxygen insertion. As outlined above, tellurium is generally
considered to be an indispensable constituent of the active ensemble on the catalyst surface
involved in either consecutive selective oxidation of the intermediate propene to acrylic acid,12 or
in the activation of propane.17
Under working conditions, the (Mo,V)Ox surface is reconstructed and differs from the bulk in
terms of composition and oxidation state of V. A substantial enrichment of V5+ species at the
surface has been observed. Two electronically different V5+ species have been detected. The
particular enrichment in one of these species (V5+(II), BE=517.3 eV) is attributed to the
migration of vanadium from position M12 within the channels of the crystal structure to the
surface under reaction conditions. The process is facilitated by hydrothermal conditions when the
oxidation is carried out in presence of steam. Accumulation of vanadium in the highest oxidation
state is detrimental in terms of acrylic acid formation. The very high surface concentration of
more than 50 at% V might cause consecutive combustion of the desired reaction product acrylic
acid. This assumption is based on the measured conversionꢀselectivity behavior of the catalyst.
Detailed kinetic studies are in progress. The difference in selectivity of MoV and MoVTeNb M1
ACS Paragon Plus Environment
29