Structure and Functional Properties of Heteropolyoxomolybdates Supported on Silica SBA-15

Structure and functional properties of heteropolyoxomolybdates supported on silica SBA-15

vorgelegt von

Dipl.-Chem. Rafael Zubrzycki geb. in Berent

von der Fakultät II - Mathematik und Naturwissenschaften der Technischen Universität Berlin zur Erlangung des akademischem Grades

Doktor der Naturwissenschaften -Dr. rer. nat.-

genehmigte Dissertation

Promotionsausschuss

Vorsitzender: Prof. Dr. rer. nat Thomas Friedrich Berichter/Gutachter: Prof. Dr. rer. nat. Thorsten Ressler Berichter Gutachter: Prof. Dr. rer. nat. Malte Behrens

Tag der wissenschaftlichen Aussprache: 20. März 2015

Berlin 2015

Abstract

Heteropolyoxomolybdates with Keggin structure (HPOM) were supported on SBA-15 and introduced as model catalysts for investigating structure-property correlations during selective propene oxidation. The chemical composition of the HPOM was varied by substituting molybdenum with vanadium or tungsten. Subsequently, the various heteropolyoxomolybdates were supported on nanostructured silica SBA-15. Additionally, unsubstituted HPOM were deposited on SBA-15 with different pore radii. Unsupported and supported heteropolyoxomolybdates were characterized by ex situ techniques yielding a detailed knowledge about structure and chemical composition of the model catalysts. Afterwards, the unsupported and supported heteropolyoxomolybdates were characterized by in situ techniques and tested for their catalytic properties in the partial oxidation of propene. HPOM supported on SBA-15 were investigated to elucidate the influence of addenda atoms, the silanol groups of SBA-15, the pore radii of SBA-15, and the HPOM loading on the resulting structures forming during propene oxidation conditions. The initial Keggin structure was retained after supporting HPOM on SBA-15. The removal of adsorbed water and a following dehydroxylation of silanol groups of SBA-15 lead to a destabilizing effect on the Keggin ion during propene oxidation conditions. Subsequently, the HPOM supported on SBA-15 formed a mixture of [MoOx] and [(V,W)Ox] species on the support material under catalytic conditions. The [MoO6] units were influenced by the structural evolution of neighboring [VO6] and [WO6] units of the initial Keggin ion structure. The structural evolution of the [MoOx] and [(V,W)Ox] species lead to predominantly tetrahedral [MoO4] and [VO4] units in vanadium substituted HPOM and to predominantly octahedral [MoO6] and [WO6] units in tungsten substituted HPOM. The formation of [MoO4] units or [MoO6] depended on the degree of vanadium or tungsten substitution. The resulted [MOx] (M = V, W) units were in close vicinity to the [MoOx] species. The various structures resulting for supported HPOM exhibited an influence on the catalytic activity. The reaction rates at similar propene conversions for supported

HPOM decreased with higher [MoO4]/[MoO6] ratio. The higher reaction rate resulted in an increased formation of total oxidation products. Hence, samples with an increased

[MoO4]/[MoO6] ratio exhibited an increased selectivity towards C3 oxidation products.

Zusammenfassung

Heteropolyoxomolybdate mit Keggin Struktur (HPOM) geträgert auf SBA-15 wurden als Modellkatalysatoren für die selektive Propenoxidation verwendet und hinsichtlich ihrer Struktur-Eigenschafts-Beziehungen untersucht. Die chemische Zusammensetzung der HPOM wurde durch Substitution von Molybdän mit den sog. Addenda-Atomen Vanadium oder Wolfram variiert. Anschließend wurden die verschiedenen Heteropolyoxomolybdate auf SBA-15 geträgert. Zusätzlich wurden unsubstituierte HPOM auf SBA-15 mit unterschiedlichen Porenradien geträgert. Die ungeträgerten und geträgerten HPOM wurden charakterisiert, um detaillierte Informationen über die Struktur und die chemische Zusammensetzung der Modellkatalysatoren zu erhalten. Danach wurden die ungeträgerten und geträgerten HPOM unter Reaktionsbedingungen charakterisiert und auf ihre katalytischen Eigenschaften bei der partiellen Oxidation von Propen getestet. Die auf SBA- 15 geträgerten HPOM wurden untersucht, um den Einfluss der Addenda-Atome, der Silanolgruppen des SBA-15, der unterschiedlichen Porenradien des SBA- 15 und der HPOM-Beladung auf die sich unter Propenoxidationsbedingungen bildenden Strukturen aufzuklären. Die Kegginstruktur blieb nach der Trägerung der HPOM auf SBA-15 erhalten. Die Entfernung von adsorbiertem Wasser und eine folgende Dehydroxylierung der Silanolgruppen des SBA-15 führten zu einer Destabilisierung der Keggin-Ionen unter Propenoxidationsbedingungen. Anschließend bildeten die geträgerten HPOM unter katalytischen Bedingungen eine Mischung aus [MoOx]- und [(V,W)Ox]-Spezies auf dem

Trägermaterial. Die [MoO6]-Einheiten wurden durch die strukturelle Entwicklung der benachbarten [VO6]- und [WO6]-Einheiten aus der ursprünglichen Kegginstruktur beeinflusst. Die Strukturentwicklung der [MoOx]- und [(V,W)Ox]-Spezies führte zu

überwiegend tetraedrischen [MoO4]- und [VO4]-Einheiten in den vanadiumsubstituierten

HPOM und zu überwiegend oktaedrischen [MoO6]- und [WO6]-Einheiten in den wolframsubstitutierten HPOM. Die Bildung der [MoO4]- oder [MoO6]-Einheiten waren von der Anzahl der Addenda-Atome pro Keggin-Ion abhängig. Die [MOx]-Einheiten

(M = V, W) befanden sich in unmittelbarer Nähe zu den [MoOx]-Einheiten. Die verschiedenen Strukturen, die sich aus den geträgerten HPOM bildeten, zeigten einen Einfluss auf die katalytische Aktivität. Die Reaktionsrate bei ähnlichen Propenumsätzen nahm für die geträgerte HPOM mit höherem [MoO4]/[MoO6] Verhältnis zu. Die höhere Reaktionsgrate führten zu einer erhöhten Bildung von Totaloxidationsprodukten. Die

Proben mit einem erhöhten [MoO4]/[MoO6] Verhältnis zeigte eine erhöhte Selektivität gegen C3 Oxidationsprodukten.

Contents

Abstract ...... III Zusammenfassung ...... V Contents ...... VII Abbreviations ...... X 1 Introduction ...... 1 1.1 Motivation ...... 1 1.2 Heteropolyoxomolybdates in partial oxidation reactions ...... 3 1.3 Supported heteropolyoxomolybdates partial oxidation reactions ...... 5 1.4 Outline of the work ...... 7 2 Characterization Methods ...... 8 2.1 Structural Characterization ...... 8 2.1.1 Powder X-ray diffraction ...... 8 2.1.2 Vibrational spectroscopy ...... 9 2.1.3 Physisorption ...... 10 2.1.4 X-ray absorption spectroscopy ...... 11 2.1.5 Nuclear magnetic resonance spectroscopy ...... 13 2.2 Element Analysis ...... 14 2.2.1 X-ray fluorescence (XRF) spectroscopy...... 14 2.2.2 Atomic absorption spectroscopy (AAS) ...... 15 2.3 Thermal analysis ...... 15 2.4 Catalytic Characterization ...... 15

3 Charaterization of bulk P(V,W)xMo12-x (x = 0, 1 ,2) ...... 17 3.1 Sample Preparation ...... 17 3.2 Sample characterization ...... 19

3.3 Ex situ characterization of P(V,W)xMo12-x (x = 0, 1, 2) ...... 24 3.3.1 Quantification of metal loading by XRF ...... 24

3.3.2 Long-range structure of as-prepared P(V,W)xMo12-x (x = 0, 1, 2) ...... 24

3.4 Short-range order structural characterization of P(V,W)xMo12-x (x = 0, 1, 2) ...... 26 3.5 In situ Characterization of bulk heteropolyacids ...... 32

3.5.1 In situ XRD of PMo12-x(V,W)x x = 0, 1, 2 during oxidation conditions ...... 32 3.5.2 Functional characterization of bulk HPOM ...... 36

VII

3.6 Summary...... 41

4 Characterization of P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) (10 wt.% Mo) ...... 42 4.1 Sample Preparation ...... 42 4.2 Sample characterization ...... 43 4.3 Results of the Characterization ...... 45

4.3.1 Long-range structure of as-prepared P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) ...... 45

4.3.2 Short-range order structural characterization of as-prepared P(V,W)xMo12-x- ...... SBA-15 (x = 0, 1, 2) ...... 48 4.4 Conclusion ...... 55

5 Characterization of PVxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions ...... 56 5.1 Experimental...... 56 5.1.1 Sample Characterization ...... 56 5.1.2 Sample preparation ...... 59

5.2 Structural characterization of PVxMo12-x-SBA-15 (x = 1, 2) under ...... catalytic conditions ...... 59

5.2.1 Local structure in activated PVxMo12-x-SBA-15 (x = 0, 1, 2) and a ......

reference V2Mo10Ox-SBA-15 under catalytic conditions ...... 62

5.2.2 Local structure of P in activated PV2Mo10SBA-15 under catalytic conditions ...... 68 5.2.3 Structure directing effects of vanadium and the support material on the structure ....

of activated PV2Mo10-SBA-15 under catalytic conditions ...... 70

5.3 Functional characterization of PVxMo12-x-SBA-15 (x = 0, 1, 2) ...... 71 5.3.1 Reducibility ...... 71 5.3.2 Catalytic performance ...... 72 5.3.3 Influence of phosphorus species on catalytic activity ...... 74 5.4 Summary...... 76

6 Characterization of PWxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions ...... 77 6.1 Experimental...... 77 6.1.1 Sample Characterization ...... 77 6.1.2 Sample preparation ...... 80

6.2 Structural evolution of PWxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions . 80

6.2.1 Local structure in activated PWxMo12-x-SBA-15 (x = 0, 1, 2) and a ......

reference W2Mo10Ox-SBA-15 under catalytic conditions ...... 86

6.2.2 Comparison of the local structure around Mo centers in act. PW2Mo10-SBA-15 ......

and a reference act. W2Mo10Ox-SBA-15 under catalytic conditions ...... 90

VIII

6.3 Functional characterization of PWxMo12-x-SBA-15 (x= 1, 2) ...... 94 6.3.1 Reducibility ...... 94 6.3.2 Catalytic performance ...... 95 6.4 Summary ...... 98

7 Characterization of PMo12 supported on SBA-15 with tailored pore radii ...... 99 7.1 Experimental ...... 100 7.2 Structure of the support materials ...... 104

7.3 Characterization of PMo12-SBA-15 (10, 14, 19 nm) ...... 106

7.4 Structural evolution of PMo12- SBA-15 (10, 14, 19 nm) under catalytic ...... conditions ...... 108

7.5 Functional characterization of PVxMo12-x-SBA-15 (x= 1, 2) ...... 113 7.5.1 Influence of the resulting structures to catalytic activity ...... 113 7.6 Summary ...... 115

8 Characterization of PVMo11 supported on SBA-15 with different metal loading . 116 8.1 Experimental ...... 117

8.2 Characterization of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and ...... 1 wt.% Mo) ...... 120

8.3 Structural evolution of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and ...... 1 wt.% Mo) under catalytic conditions ...... 123

8.4 Functional characterization of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, ...... and 1 wt.% Mo) ...... 128 8.4.1 Reducibility ...... 128 8.4.2 Influence of the resulting structure on catalytic activity...... 129 8.5 Summary ...... 132 9 General discussion and Summary ...... 133 9.1 Structure directing effect of the support material ...... 133 9.2 Structure directing effects of the addenda atoms ...... 135 9.3 Structure activity relationships ...... 137 10 Conclusions ...... 142 11 References ...... 148 12 Appendix ...... 164 Danksagung ...... XII

Abbreviations act. activated AHM ammonium heptamolybdate BET Brunauer-Emmet-Teller BJH Barrett-Joyner-Halenda cf. compare (Latin "confer") DTG derivative thermogravimetric e.g. or example (Latin "exempli gratia") eq. equation et al. and others (Latin "et alii") EXAFS extended X-ray absorption fine structure exp. experimental FID flame ionization detector FT Fourier transformed HASYLAB Hamburg Synchrotron Radiation Laboratory HPOM heteropolyoxomolybdate with Keggin structure i.e. that is (Latin "id est") IR infrared IUPAC International Union of Pure and Applied Chemistry m/e mass-charge ratio

MoO3-SBA-15 molybdenum oxides supported on SBA-15 nom. nominal Norm. normalized

PMo12 H3[PMo12O40

PVMo11 H4[PVMo11O40]

PV2Mo10 H5[PV2Mo10O40]

PWMo11 H3[PWMo11O40]

PW2Mo10 H3[PW2Mo10O40] PZC point of zero charge RT room temperature SBA-15 mesoporous silica (Santa Barbara amorphous type material No. 15) SDA structure-directing agent TA thermal analysis

TCD thermal conductivity detector TG thermogravimetry

V2Mo10Ox-SBA-15 vanadium and molybdenum oxides supported on SBA-15

W2Mo10Ox-SBA-15 tungsten and molybdenum oxides supported on SBA-15 wt.% weight percent XAFS X-ray absorption fine structure XANES X-ray absorption near edge structure XAS X-ray absorption spectroscopy XRD X-ray diffraction XRF X-ray fluorescence

1 Introduction

1.1 Motivation

Heteregenous catalyzed reactions play a fundamental role in the production of industrial organic chemicals and intermediates. Selective catalytic oxidation processes generate approximately one quarter of the value produced world wide by catalytic processes.[1] The products include such intermediates as acrolein, acrylic acid, acrylonitrile, methacrylic acid, MTBE, maleic anhydride, phthalic anhydride, ethylene and propylene oxide.[2] One important industrial process is the selective oxidation of propene towards acrolein and acrylic acid.[3,4] Acrylic acid is an important raw material in the fine chemicals industry. The acid and its esters are important monomers for the preparation of polymers and are used in the manufacture of paints and adhesives, in the treatment of paper and textile, as well as superabsorbent.[5] The industrial production of acrylic acid is a two step process. In the first reaction, propene is oxidized using a bismuth molybdate based catalyst resulting in the production of acrolein. In the second reaction, acrolein is oxidized using a bismuth molybdate based catalyst mixed with additional metal oxides with transitions metals such as vanadium or tungsten[3]:

propene acrolein acrylic acid

The product yield of acrylic acid in this process is about 90%.[6] Increasing the product yield using new or improved catalyst is of particular interest, because the industrial processes can be made more economical and sustainable. Molybdenum based catalyst are of particular interest as they are often used in partial oxidation catalysts for industrial application.[7] The catalyst may be improved by varying the chemical complexity. Additional metals such as W, Nb, or V stabilize characteristic crystallographic structures which lead to oxidation catalysts with improved activity and selectivity.[8,9] However, the influence of structural variety and chemical complexity in the mixed oxide systems on catalytic performance is difficult to distinguish. Moreover, the functionality of individual metal centers or particular structural motifs of these highly active mixed oxide catalysts

can hardly be determined. Hence, model systems are required which combine structural invariance with compositional variety or vice versa.[10–12] Heteropolyoxomolybdates (HPOM) with Keggin structure exhibit a broad compositional range while maintaining their characteristic structural motif.[13–16] Therefore, substituting Mo atoms with addenda atoms (i.e. V, W, Nb) make Keggin type HPOM suitable model system to study structure activity relationships. A challenge within the study of structure-activity relationships of those catalysts is to distinguish between the bulk and surface structures of this materials. Catalytic reaction occurs on the crystalline surface of the bulk compounds. Therefore, the majority of the bulk compound is not involved in the catalytic reaction. This leads to an analytical problem, because the average of the bulk compound is measured with most analytical methods. Therefore, a study of structure-activity relationships of surface structures corresponding to the "real" catalyst is not feasible. An approach to solve this analytic problem is the use of supported metal oxide catalysts.[17,18] Supported catalytic species possesses high dispersions and an improved surface to bulk ratio. Hence, differentiating between bulk and surface structures is no longer necessary. Therefore, structure activity relationships can be readily deduced from the characteristic oxide species observed on the support material under catalytic reaction conditions. Suitable support materials for catalysts posseses a large surface area and a homogenous internal pore structure with sufficiently large pores. The precursors are ideally highly dispersed on the support material. Hence, all the centers are accessible on the surface and are involved in the catalytic reaction. Furthermore, the support may interact with the precursor to stabilize particular structural motifs without affecting the catalytic reaction.[19–21] Nanostructured

SiO2 materials such as SBA-15 represent suitable support systems for metal oxide catalysts.[22–25] Additionally, the mechanical, thermal, and hydrothermal stability improves SBA-15 as support material during catalytic conditions like high temperatures or in the presence of steam.[26] The deposition of vanadium or tungsten substituted HPOM on SBA-15 lead to well dispersed HPOM Keggin ions. HPOM supported on SBA-15 can be used to vary the chemical composition while maintaining good accessibility of the supported molybdenum based catalyst.[27] Therefore, a study of structure-activity relationships of surface structures corresponding to the "real" catalyst was possible.

1.2 Heteropolyoxomolybdates in partial oxidation reactions

The first characterization of the Keggin structure was performed by J.F- Keggin at 1934 using XRD.[28] Fig. 1-1 depicts the Keggin ion structure typically represented by the x-8 4+ 5+ formula [XM12O40] where X is the central atom (Si , P , etc.), x is the oxidation state and M is the metal ion (Mo6+or W6+). The smallest structural units of the Keggin structure are metal-oxygen octahedra (MO6) and are called primary structure. The octahedra are arrranged in four M3O13 (triad) groups surrounding the central tetrahedron XO4. The Keggin ion structure is called secondary structure and the arrangement of the Keggin ion in a crystal structure represents the tertiary structure.[29]

P Mo O

triad Keggin structure

Fig. 1-1: (left) Triad of the Keggin structure (Mo3O13); (right) Keggin structure (secondary structure).

Heteropolyoxomolybdates with Keggin structure are able to catalyze a variety of reactions as homogeneous or heterogeneous catalyst (Table 1-1). Thus, HPOM are part of current investigations in catalysis research.[30–33] Substituted HPOM were intensely investigated in the past with regard to partial oxidation reactions.[13–16,34,35] Li et al. showed, that introducing vanadium in the Keggin ion structure of HPOM lead to an enhanced

Table 1-1: Summary of reactions, which are catalyzed by Heteropolyoxomolybdates with Keggin structure. reactions catalysing by HPOM[34–38] isomerisation of alkanes polymerisation of THF MeOH to olefins Diels-Alder Reaction alkylation of paraffins oxidation of alkanes oligomerisation of alkenes oxidation of alkenes Friedel-Crafts Acylation hydrogenation of alkenes Beckmann rearrangement methacrolein to methacrylic acid

catalytic activity in partial oxidation of propane.[36] Bondavera et al. investigated the influence of vanadium on the catalytic activity in ammoxidation of methylpyrazine.[35] The vanadium substituted HPOM showed an enhanced catalytic activity depending on the degree of substitution. Comparable correlations between catalytic activity and the degree of vanadium substitution were found for the oxidation of acrolein, isobutylene, and isobutane.[37–39] Ressler et al. investigated in various studies the structural evolution of vanadium substituted HPOM during propene oxidation.[13–16] H4[PVMo11O40] (PVMo11) loses crystal water during treatment under propene oxidation conditions in the temperature range from 373 to 573 K.[15] The release of crystal water is followed by partial decomposition, reduction of the average Mo valence, and formation of cubic HPOM

(Mox[PVMo11-xO40]) at 573 K. The formation of cubic Mox[PVMo11-xO40] with Mo centers outside the Keggin ion structure and V centers remaining in a lacunary Keggin ion coincides with the onset of catalytic activity.[15] Niobium substituted HPOM

(H4[PNbMo11O40]) formed the characteristic cubic HPOM structure, similar to the structural evolution of H3[PMo12O40], H4[PVMo11O40], and H5[PV2Mo10O40].[14] In contrast to H3+x[PVxMo12-xO40] (x = 0, 1, 2) the lacunary Keggin ion decompose rapidly towards MoO3 at about 673 K. The decomposition process correlated with a decrease in catalytic activity.[14] Mestl et al. investigated the thermal induced decomposition of

PVMo11.[40] Upon loss of crystal water vanadyl and molybdenyl species are expelled from the Keggin ion structure forming the lacunary Keggin ion. This defective structure further disintegrated to triads and finally condensed to the thermodinamically stable MoO3.[40]

Therefore, it may be assumed, that the vanadium and niobium substitution in HPOM have a structure directing effect, stabilizing a structure active in oxidation of propene. However, the role of addenda atoms is still under discussion, because addenda atoms may have both a structure directing effect and/ or a functional effect during propene oxidation conditions.

1.3 Supported heteropolyoxomolybdates in partial oxidation reactions

Support material

Mesoporous material SBA-15 was fist synthesized in 1998.[22,23] The SBA-15 structure composed of hexagonal channels has a high surface area and a narrow pore size distribution. Supramolecular aggregates are used for the synthesis of mesoporous systems as structure-directing agents (SDAs).[42,43] In the synthesis of SBA-15 a block copolymer is used as SDA.[22,23] The self-organization of the block copolymer promotes the formation of a silica based inorganic network around the organic aggregates. Afterwards, the organic template is removed through a calcination process. Fig. 1-2 illustrates a schematic representation of the preparation of SBA-15.

Globe Rod-Shaped Liquid-Crystalline Organic-Inorganic Mesoporous Micelle Micelle Phase Composite Material

Calcination

Fig. 1-2: Schematic representation of the preparation of SBA-15 (adapted from [45]).

The surface area and pore width are tailored by the preparation procedure. The typical synthesis of SBA-15 leads to surface areas between 600 and 1000 m2/g and pore diameters between 5 and 10 nm.[22] The pore radius is tunable with swelling agents resulting in pore radii up to 50 nm.[46] The swelling agents are for example benzene, 1,3,5,-

trimethybenzene, decane, and gelatin.[47–50] The swelling agents enrich in the hydrophobic chains of the surfactants in the micelles and expand the micelles resulting in a larger diameter. The important conditions during the synthesis of SBA-15 with larger pores are the initial synthesis temperature, the amount of swelling agent, and the hydrothermal treatment time and temperature.[51] Generally, the pore diameter increases with lower initial synthesis temperature influencing the formation of the micelles. The expansion of the micelles is limited, because mesocellular foams with spherical mesopores are formed, when higher relative amounts of swelling agents are used.[49] Further parameters for adjusting the pore radius of SBA-15 are the hydrothermal treatment time and temperature.

The increase in the hydrothermal treatment temperature allows to achieve larger pore sizes in a shorter period of time.[22,46] Disadvantages in changing the hydrothermal conditions are long hydrothermal treatment times and high temperatures (e.g. 2 days at 130 C) that lead to merging of adjacent cylindrical mesopores.[51]

Supported heteropolyoxomolybdates in partial oxidation reactions

Industrial applications and investigations of supported tungstate or molybdate heteropoly acids with Keggin ion structure have been recently reviewed.[52,53] Various authors reported, that the Keggin ion structure of supported tungstate or molybdate heteropoly acids retained intact after deposition on silica, titania or zirconia based support materials.[54–58,27] For H3[PMo12O40] supported on ZrO2 (PMo12-ZrO2) Devassy et al. investigated the nature of the phosphorous species depending on Keggin loading and calcination temperature.[39] They, reported a decomposition of the HPOM to oxide species at temperatures above 723 K. The thermal stability of H3[PW12O40] supported on

ZrO2 (PW12-ZrO2) was investigated by López-Salinas et al.. The structural behaviour of

PW12-ZrO2 during calcination was comparable to that of PMo12-ZrO2. PW12-ZrO2 decomposed at temperatures above 773 K to form the corresponding supported oxides.[46]

Ressler et al. reported for H4[PVMo11O40] supported on SBA-15 (PVMo11-SBA-15) a decomposition under propene oxidation conditions above 573 K resulting in Mo oxide species.[27] The resulting molybdenum oxide species are comparable to that of molybdenum oxide species synthesized from an ammonium hepta molybdate (AHM) precursor. Both molybdenum oxide species on SBA-15 revealed comparable structural motifs during treatment in propene oxidation conditions.[59] Results of the Mo K edge

XANES analysis of activated PVMo11-SBA-15 indicated tetrahedrally coordinated MoOx species. A Comparison with references afforded about 50% of tetrahedrally coordinated

MoOx species on SBA-15.[27] Ressler et al. assumed that the [MoO6] units exhibited a connectivity similar to that of the building blocks of MoO3. The MoO4 units may be isolated or connected to other MoOx species on the surface of SBA-15.[27] Therefore, no stable HPOM supported on SBA-15 could be obtained on SBA-15 under reactions conditions. The role and structural evolution of V and P in PVMo11-SBA-15 under catalytic conditions remained largely unknown.

1.4 Outline of the work

The objective of this work is to elucidate the role and structural evolution of Mo, V, W and P in HPOM supported on SBA-15 under catalytic conditions. Additionally, the influence of the support material SBA-15 on the stability of the supported HPOM is investigated with respect to the role of dehyrdation processes of the SBA-15, the pore radii of SBA-15 and HPOM loading. Therefore, the chemical composition of HPOM is varied by substituting molybdenum with vanadium or tungsten. Subsequently, the various heteropolyoxomolybdates are supported with different loading on nanostructured silica SBA-15. Additionally, unsubstituted HPOM are deposited on SBA-15 with different pore radii. The unsupported and supported heteropolyoxomolybdates are characterized by ex situ techniques ensuring a detailed knowledge about structure and chemical composition of the model catalysts. Afterwards, the unsupported and supported heteropolyoxomolybdates are characterized by in situ techniques and tested for their catalytic properties in the partial oxidation of propene. HPOM supported on SBA-15 are intensively investigated to clarify the influence of the silanol groups of SBA-15, the pore radii of SBA-15, and the HPOM loading on SBA-15 on the structures forming during propene oxidation conditions. Additionally, the influence of the addenda atoms of the substituted HPOM supported on SBA-15 on the resulting structures is investigated during propene oxidation conditions. The characterization of the structural evolution especially of the addenda atoms (V, W) and the heteroatom (P) focus on identifying the metal oxide structure and correlating the structure with catalytic activity.

2 Characterization Methods

2.1 Structural Characterization

2.1.1 Powder X-ray diffraction

X-ray diffraction (XRD) is used for determining the long-range order structures of the synthesized bulk samples in this work. For that, powder samples are irradiated with monochromatic X-ray photons. The X-ray photons are inelastically and elastically scattered by the electrons of atoms arranged in a periodic structure.[60] The X-ray photons that are elastically scattered by the electrons of atoms are used for determining the crystal structure. The scattered X-ray wave interfere constructively or destructively depending on the distance between the lattice planes and the angle (θ), between the incident X-rays and the lattice planes. The Bragg equation described the detectable X-ray photons resulting from constructive interference as a relationship between the lattice spacing (d), and the angle (θ), between incident X-rays and lattice plane.

nλ = 2d sin θ n = 1, 2, ... (2.1) with: n diffraction order λ wavelength of the X-ray photons d lattice spacing θ angle between incident X-rays and lattice planes

Fig. 2-1 depicts a schematic representation of the scatted X-ray waves and interfere constructively. Constructive interference takes place only when sum of the path lengths of + is an integer factor of the wavelength, λ. Diffraction peaks can be characterized by the Miller indices, which describe the corresponding lattice planes. Detailed information about data analyzing and structure refinement can be found elsewhere.[61,62]

X-rays

A C d B

lattice planes

Fig. 2-1: X-ray photons strikes the ordered lattice at an angle . X-ray photons are scattered and interfere constructively in direction given by the Bragg equation (equation 2.6).

2.1.2 Vibrational spectroscopy

The advantage of Infrared (IR) spectroscopy is that the method can be used to analyze amorphous compounds. Amorphous compounds could not be analyzed by XRD. Therefore, IR spectroscopy is chosen for excluding additional amorphous compounds besides the crystalline compounds in the synthesized bulk samples. In solid states the atoms in a crystal structure vibrate internally by changing the interatomic distances. This vibrations lead to transitions that are within the range of the infrared radiation (0.7μm- 1000μm). The wavelength of the IR radiation is varied and the decrease of intensity is measured.[63,64] A prerequisite for the absorption of IR radiation is a change in the dipole moment in molecules or structural motifs in solid state compounds. Thus, the electric field of the electromagnetic radiation couples to the dipole moment in the structure, resulting in an absorption of the electromagnetic radiation. The observed vibration are called modes. Detailed information can be found elsewhere.[68,69] Similar to IR spectroscopy different vibrational modes are also observed in Raman spectroscopy. In contrast to IR spectroscopy that uses the absorption resulting by various vibrational modes at different irradiation energies, another method of excitation is selected in Raman spectroscopy. In Raman spectroscopy the sample is irradiated with monochromatic radiation, usually with a laser.[64,65] The wavelength of the radiation should be chosen outside the absorption range of electronic transitions of the analyte

excluding electronic absorption effects at the excitation wavelength. In Raman spectroscopy, the electromagnetic radiation induces a dipole moment in the molecule achieving higher vibration levels. Hence, linear and homoatomic structural motifs can be studied. Thus, Raman spectroscopy is complementary to IR spectroscopy. There are three possible types of interactions subdivided into two categories, the Raman and Rayleigh scattering. In the Rayleigh scattering, a photon with the energy causes a transition to a higher virtual energy state of the structural motif and relaxes to a vibrational level of the ground electronic state. The absorbed and emitted energy is equal in this process resulting in an elastic scattering interaction. The excitation occur from any vibrational states depending on the Boltzmann distribution. At room temperature, a excitation from the vibrational ground state and the first excited vibrational state of the electronic ground state is usual. In Raman scattering, the incident energy is different from the given energy . Thus, Raman scattering is an inelastic scattering interaction. The energy difference is located above (anti-Stokes radiation) and below (Stokes radiation) of the incident monochromatic electromagnetic radiation.[63,69]

2.1.3 Physisorption

Physisorption measurements are used for determining the surface area and pore structure of the used support material SBA-15. Physisorption (physical adsorption) denoted an attaching gas to a surface, which was bound by van der Waals interactions at this. The gas molecules are denoted as adsorbate and the surface as an adsorbent.[66,67] A dynamic equilibrium is established between the adsorbent and adsorbate from the gas phase. In this case, under isothermal conditions, the number of adsorbed molecules n depends on the equilibrium pressure p of the gas. This relationship can be expressed by the following equation.The physisorption isotherm can be obtained by plotting the relationship shown in equation 2.2. The isotherms are divided into six classes by IUPAC.[68] The relevant type of isotherms for this work is the type IV isotherm. The type IV isotherm has a specific hysteresis curve, which is characteristic for mesoporous materials. The adsorption of the monolayer and multi-layer on the walls of the mesopores is following by capillary condensation in the mesopores of the adsorbent.

with: number of adsorbed molecules equilibrium pressure

saturation pressure of the adsorptive

The specific surface area of the investigated materials can be determined by the Brunauer-Emmer-Teller (BET) method from the measurements of the isotherms.[69] In addition to the specific surface area, the pore size distribution can be determined in the mesoporous material from the capillary condensation. This method is the Barret-Joyner- Halenda (BJH) method and is derived from the Kelvin equation.[70] In this case, the emptying of pores is considered by stepwise lowering of the relative pressure , in which the thinning of the multilayer film is taken into account. Detailed informations can be found elsewhere.[66–70]

2.1.4 X-ray absorption spectroscopy

The X-ray absorption fine structure was first discovered in the 1920s. X-ray absorption fine structure spectrocopy was interesting for the analysis of atoms in biological molecules, catalysts, and amorphous materials since the presence of particle accelerators (synchrotron) in the 1970`s.[71,72] X-Ray Absorption Spectroscopy (XAS) is an analytical method for elucidating electronic properties and the local structure of the absorber atoms.[73,74] The particular advantage of XAS is the applicability of the method even in the absence of long- range order. These property is especially relevant for supported catalysts. XAS is used in this work to determine the structural evolution of the Mo, V, and W centers in supported HPOM during propene oxidation conditions. Absorption of X-rays at the core-level binding energy leads to excitation of an electron from a core level. Excitation of the electron at the core-level binding energy lead to a sharp rise in absorption, which is denoted as absorption edge. The energy for the excitation is specific for the absorber atom. XAS can be subdivided in two regions, X-ray near edge structure (XANES) and the extended X-ray absorption fine structure (EXAFS).

The XANES region is typically located in a 50 eV range around the absorption edge. This region contains information about the element specific oxidation states and coordination geometries around the absorber atom.[75,76] The XANES pre-edge region resulted through the excitation of an electron from a core level (K, L, M, depending on the observed absorption edge) to an unoccupied state. The resulting dipole transitions obey the following selection rules: the spin quantum number cannot be changed during the transition Δs = 0, the orbital quantum number l and the total spin j have to change (Δl, Δj) = ± 1. Therefore, electronic transitions from the s orbital of the K shell to a higher p orbital and from the p orbital of the L shell to higher s or d orbitals are typical. The final state of the transition can be also a hybridized orbital. Therfore, transitions from the s orbital of the K shell to pd hybridized orbitals are possible due to the partially different

] μd scattering atom

Absorption [ Absorption

0 absorber atom

19.9 20.0 20.1 20.2 20.3 20.4 20.5 20.6 dipol transitions E [keV] XANES EXAFS

Fig. 2-2: X-ray absorption spectrum with a schematic representation of the processes at the absorption edge,. XANES region: Absorption of an X-ray photon and excitation to a higher unoccupied level (dipole transition). EXAFS region: Representation of the fine structure of the absorption edge resulting from the interference of the outgoing photoelectron wave from the absorber atom with the incoming photoelectron wave resulting by backscaterring of neighbored atoms.

orbital character. Transitions, that are forbidden in principle in quantum mechanics, occur also in a non-centrosymmetric structure geometry, as in the tetrahedral geometry. Additionally, transitions take place in non centrosymmetric structures such as a distorted octahedral structure. Thus, the shape of the XANES region is caused by the local electronic and geometric structure of the absorbing atom. All quantum mechanically forbidden but occurring transitions are characterized by a pre-edge peak in the K edge/LI edge for the investigated elements in this work (Mo, V, W). In octahedrons no pre-edge peak is expected. The highest intensity of the pre-edge peak are observed for a tetrahedral geometry. Therefore, a mixed coordination geometry (tetrahedral and octahedral) can be quantified by a linear combination of reference spectra of tetrahedral and octahedral coordinated references. In the EXAFS region the generated photoelectron interact with the electron density of adjacent atoms. The outgoing electron wave from the absorber atom reaches the neighboring atoms and will be scattered back.[76] The incoming spherical electron waves interferes with the outgoing photoelectron wave resulting in an oscillation of the absorption coefficient and in the fine structure in the absorption spectrum (Fig. 2-2). The absorption coefficient can be extracted from the EXAFS region of the absorption spectrum. The oscillatory part can be separated from the atomic absorption of a free atom and is denoted as EXAFS function . The EXAFS function contains information about the coordination number, element specific backscattering amplitude, and the mean-squared displacement of the neighboring atoms around the absorber atom. The EXAFS function χ(k)can be transferred to a pseudo radial distribution function FT (χ(k)) through a Fourier transformation. Detailed explanations can be found in Ref. [76].

2.1.5 Nuclear magnetic resonance spectroscopy

Nuclear magnetic resonance (NMR) spectroscopy is an analytical method to elucidate structures in solid states and molecules. NMR spectroscopy is used in this work for determining the local structure of phosphorus in the activated supported HPOM after treatment under propene oxidation conditions. NMR spectroscopy could be used for atoms with a nuclear spin I ≠ 0. If such a nucleus is in an external magnetic field, the nuclear level split as function of the nuclear spin. The lower nuclear level can be excited with

electromagnetic radiation resulting in transition between the energy levels. The transition corresponds to the Larmor frequency and the chemical and electronic environment of the atom, resulting in the resonant frequency . Thereby, the resonant frequencies change depending on the local structure of the atom. The resulting NMR spectrum is normalized in the abscissa to a standard substance. The normalization results in a device-nonspecific scale, the chemical shift . NMR spectra of liquid or gas phase exhibit narrow signals with characteristic splitting signals caused by coupling to other nuclei in the environment. This results from the rapid rotation of the molecules in all directions and leads to averaging of all orientation-dependent spin interactions. In the solid-state NMR spectroscopy rather broad signals are obtained that show a characteristic shape. This broad signals result due to the different orientation of the structural motifs in space, caused by an orientation- dependent spin interactions. The MAS technology (Magic Angle Spinning) averages these anisotropic spin interactions. Therefore, the sample is placed at an angle of = 54.7° to the magnetic field and rotates around its own axis. With this construction, anisotropic interactions are averaged and discrete signals obtained. These signals are comparable to signals in liquid or gaseous phase.

2.2 Element Analysis

2.2.1 X-ray fluorescence (XRF) spectroscopy

XRF spectroscopy is particularly suitable for the quantitative analysis of metals in metal oxides such as used in this work. Irradiation of the sample with X-ray photons results in removal of an electron from a core shell. The resulted core hole is filled by an electron from a higher level, which results in the emission of an X-ray photon. The energy of the resulting fluorescence photons is described by Mosley`s law. The energy of the emitted fluorescence photons is characteristic for each element and can be used to identify the elemental composition of the sample. Detailed explanation can be found in Ref. [65].

2.2.2 Atomic absorption spectroscopy (AAS)

AAS is used in this work for the quantitative analysis of phosphorus. XRF spectroscopy is due to overlapping of the main P peak and a Mo peak of limited use. For the measurements, the samples are dissolved in a suitable solvent. Afterwards, the solution is transferred in the atomizer.[77] Flames or electrothermal (graphite tube) atomizer are typical atomizer using in AAS. In this work a pyrocoated graphite tube was used as atomizer. A matrix modifier and the sample were injected into the graphite tube. The graphite tube was heated up to 2873 K resulting in atomization of the phosphorus species. Phosphorus atoms are irradadiated with a Hollow Cathode Lamp (HCL) containg small amounts of phosphorus The HCL emmited the line spectrum of the element of interest (here phosphorus). This lead to an excitation of the valence electrons in the atoms of the element of interest resulting in an absorption of the line spectrum of the HCL. The absorption is proportional to the amount of the element of interest and could be used for quantitative analysis. Detailed information can be found elsewhere.[64]

2.3 Thermal analysis

Thermal analysis consist of analytical methods, which detect physical or chemical properties of a substance or substance mixture as a function of temperature or time, while the sample undergoes a controlled temperature program.[78] Thermogravimetry (TG) is used mainly for the investigation of decomposition processes, thermal stability, or dehydration processes.[79] The sample weight is monitored using a sensitive thermobalance as a function of time or temperature. The first derivative of the TG signal corresponds to the differentiated thermogravimetric curve, the DTG signal. The DTG signal facilitates the identification of mass decreases due to the resulting maxima.

2.4 Catalytic Characterization

Quantitative catalysis measurements were performed using a fixed bed laboratory reactor connected to an online gas chromatography system. Gas chromatography is a suitable

chromatographic method for the separation of analyte mixtures in gas phase.[80] A detailed description of various reactor types and the individual assets of each setup can be found elsewhere.[81,82] The used fixed-bed reactor consisted of a SiO2 tube. Reactants are passed through the catalyst bed. At the outlet of the reactor, reaction products are analyzed by gas chromatography. Therefore, the sample is injected in the inert gaseous mobile phase (carrier gas). The sample in the carrier gas stream are passed through the stationary phase. The interaction between the sample and the mobile or stationary phase strongly depends on their physical and chemical properties. The larger the interaction with the stationary phase, the slower the migration velocity of the sample components. A low interaction with the stationary phase shorts retention time of the sample components. Therefore, the adsorption on the stationary phase or dissolving in the mobile phase depends on whether and how a sample component interacts with the stationary phase. The different migration velocities of the current sample components resulted in various times at which the individual compounds passed through the stationary phase. The various components can finally be analyzed as discrete bands with suitable detectors. The retention time is the time each component requires to move from the point of injection to the detector. The resulting bands are detected as a function of retention time. Thus, individual sample components can be qualitatively and quantitatively analyzed in this way.[83,84]

3 Charaterization of bulk P(V,W)xMo12-x (x = 0, 1 ,2)

Heteropolyoxomolybdates (HPOM) of the Keggin type (e.g., H3[PMo12O40]) are active catalysts for the partial oxidation of alkanes and alkenes.[7,85–88] HPOM with Keggin structure exhibit a broad compositional range while maintaining their characteristic structural motifs.[14–16] Substituting Mo atoms with addenda atoms (i.e. V, W, Nb) make Keggin type HPOM suitable model systems to study structure-activity relationships. Thus, HPOM have been frequently studied as active catalysts for selective oxidation reactions.[89] Therefore, H3[PMo12O40] (PMo12), H4[PVMo11O40] (PVMo11),

H5[PV2Mo10O40] (PV2Mo10), H3[PWMo11O40] (PWMo11), and H3[PW2Mo10O40]

(PW2Mo10) were synthesized as model catalysts in selective propene oxidation. For elucidating structure activity correlations of model systems for catalytic investigations, a detailed knowledge about structure and chemical composition is indispensable. Therefore, various characterization methods, such as XRF, AAS, XRD, IR, Raman and XAS are necessary for a sufficient characterization of the used catalyst systems. In this chapter, in addition to ex situ characterization, in situ XRD at oxidizing conditions P(V,W)xMo12-x (x = 0, 1 ,2) was performed. Catalytic testing revealed the functional properties of

P(V,W)xMo12-x (x = 0, 1 ,2) under propene oxidation conditions.

3.1 Sample Preparation

Preparation of H3[PMo12O40]

19.72 g MoO3 (Sigma Aldrich) was dissolved in 650 ml water and heated under reflux. 95 ml of 0.12 M phosphoric acid were added dropwise to the reaction mixture. The resulting suspension was heated for 3 h and kept at 298 K for 24 h until a clear yellow solution was obtained. The remainder was filtered of and the volume of the resulting yellow solution was reduced to ~30 ml using an evaporator. H3[PMo12O40] crystallized during storage at 277 K for several days.

Preparation of H3+x[PVxMo12-x] (x=1,2)

H4[PVMo11O40] was prepared as follows. 17.85 g MoO3 (Sigma Aldrich) and 1.02 g V2O5 (Sigma Aldrich) were dissolved in 650 ml water and heated under reflux. 95 ml of 0.12 M phosphoric acid were added dropwise to the reaction mixture. The resulting suspension was heated for 3 h and kept at 298 K for 24 h until a clear orange solution was obtained. The remainder was filtered of and the volume of the resulting orange solution was reduced to ~30 ml using an evaporator. H4[PVMo11O40] crystallized during storage at 277 K for several days.

H5[PV2Mo12O40] was prepared as follows. 16.89 g MoO3 (Sigma Aldrich) and 2.13 g V2O5 (Sigma Aldrich) were dissolved in 675 ml water and heated under reflux. 98 ml of 0.12 M phosphoric acid were added dropwise to the reaction mixture. The resulting suspension was heated for 3 h and kept at 298 K for 24 h until a red solution was obtained. The remainder was filtered of and the volume of the resulting red solution was reduced to ~30 ml using an evaporator. H5[PV5Mo12O40] crystallized during storage at 277 K for several days.

Preparation of H3[PWxMo12-x] (x=1,2)

H3[PWMo11O40] was prepared as follows. 3.12 g Na2WO4 · H2O (Merck) were dissolved in 50 ml water. The aqueous solution of Na2WO4 was passed through an ion-exchange resin (Ion exchanger I (Merck)).[90] The resulted sol of tungsten acid and 15 g MoO3 (Sigma Aldrich) were dissolved in 500 ml water and heated under reflux. 79 ml of 0.12 M phosphoric acid were added dropwise to the reaction mixture. The resulting suspension was heated for 3 h and kept at 298 K for 24 h until a clear yellow solution was obtained. The remainder was filtered of and the volume of the resulting yellow solution was reduced to ~30 ml using an evaporator. H3[PWMo11O40] crystallized during storage at 277 K for several days.

H3[PW2Mo10O40] was prepared as follows. 3.97 g Na2WO4 · H2O (Merck) were dissolved in 50 ml water. The aqueous solution of Na2WO4 was passed through an ion-exchange resin (Ion exchanger I (Merck)).[90] The resulted sol of tungsten acid and to 13 g MoO3 (Sigma Aldrich) were dissolved in 455 ml water and heated under reflux. 75 ml of 0.12 M phosphoric acid were added dropwise to the reaction mixture. The resulting suspension

was heated for 3 h and kept at 298 K for 24 h until a clear yellow solution was obtained. The remainder was filtered of and the volume of the resulting yellow solution was reduced to ~30 ml using an evaporator. H3[PW2Mo11O40] crystallized during storage at 277 K for several days.

3.2 Sample characterization

X-Ray Fluorescence Analysis

Elemental analysis by X-ray fluorescence spectroscopy was performed on an X-ray spectrometer (AXIOS, 2.4 kW model, PANalytical) equipped with a Rh K alpha source, a gas flow detector and a scintillation detector. 60-80 mg of the samples were diluted with wax (Hoechst wax C micropowder, Merck) at a ratio of 1:1 and pressed into 13 mm pellets. Quantification was performed by standardless analysis with the SuperQ 5 software package (PANalytical).

Atom Absorption Spectroscopy (AAS)

45 mg of P(V,W)xMo12-x (x = 0, 2) obtained before and after treatment under propene oxidation condition (5% propene + 5% O2 in He; 723 K; 0 h and 12 h time on stream) were diluted with aqueous ammonia to 10 ml. Measurements were performed with the Perkin Elmer 1100B AAS-spectrometer equipped with pyrocoated graphite tubes, including a platform, and an AS700 autosampler. Extinction was measured for 4 s (Table 3-1) at 213.6 nm using a super hollow cathode lamp (HCL) of Photron at 35 mA. Background correction was achieved by alteration of D2 lamp radiation and HCL radiation, respectively. Background was subtracted from peak area of extinction. Each sample was measured 3 times. Therefore, 10 µl of LaCl3 solution as matrix modifier (Roth ≥ 99.9 %, 100 mg/l) were injected, and subsequently 90 µl of sample solution were added.[77] The applied temperature program is presented in Table 3-1. The, syringe of autosampler was purged before and after each step with diluted nitric acid.

Table 3-1: AAS measurement program with applied temperatures [K], heating durations [s], dwelling times [s], and internal gas purges [ml min-1 (Ar)]. temperature time to reach dwelling internal gas purge step setpoint [K] setpoint [s] time [s] [ml min-1 (Ar]) drying 363 7 5 300 drying 393 3 30 300 pyrolysis 573 3 20 300 pyrolysis 1673 5 40 300 pyrolysis 1673 1 5 0 atomization/measurement 2873 0 4 0 cleaning 2873 1 8 200 cooling RT fast - 300

X-ray absorption spectroscopy (XAS)

Transmission XAS experiments were performed at the Mo K edge (19.999 keV) at beamline X, V K edge (5.465 keV) and W LIII edge (10.204 keV) at beamline C at the Hamburg Synchrotron Radiation Laboratory, HASYLAB, using a Si(311) double crystal monochromator at beamline X and a Si(111) double crystal monochromator at beamline C. X-ray absorption fine structure (XAFS) analysis was performed using the software package WinXAS v3.2..[91] Background subtraction and normalization were carried out by fitting linear polynomials and 3rd degree polynomials to the pre-edge and to the post- edge region of an absorption spectrum, respectively. The extended X-ray absorption fine structure (EXAFS) χ(k) was extracted by using cubic splines to obtain a smooth atomic 3 background μ0(k). The FT(χ(k)·k ), often referred to as pseudo radial distribution function, was calculated by Fourier transforming the k3-weighted experimental χ(k) function, multiplied by a Bessel window, into the R space. EXAFS data analysis was performed using theoretical backscattering phases and amplitudes calculated with the ab-initio multiple-scattering code FEFF7.[92] Structural data employed in the analyses were taken from the Inorganic Crystal Structure Database (ICSD).

with a lower limit of 4.0% in amplitude with respect to the strongest backscattering path. EXAFS refinements were performed in R space simultaneously to magnitude and imaginary part of a Fourier transformed k3-weighted and k1-weighted experimental χ(k) using the standard EXAFS formula.[94] This procedure reduces the correlation between the various XAFS fitting parameters. Structural parameters allowed to vary in the refinement were (i) disorder parameter σ2 of selected single-scattering paths assuming a symmetrical pair-distribution function and (ii) distances of selected single-scattering paths. The statistical significance of the fitting procedure employed was carefully evaluated in three steps as outlined in.[95] The procedures accounts for recommendations of the International X-ray Absorption Society on criteria and error reports.[96] First, the number of independent parameters (Nind) was calculated according to the Nyquist theorem Nind = 2/Π · ΔR· Δk + 2. In all cases, the number of free running parameters in the refinements was well below Nind. Second, confidence limits were calculated for each individual parameter. In the corresponding procedure, one parameter was successively varied by a certain percentage (i.e. 0.05% for R and 5% for σ2) and the refinement was restarted with this parameter kept invariant. The parameter was repeatedly increased or decreased until the fit residual exceed the original fit residual by more than 5%. Eventually, the confidence limit of the parameter was obtained from linear interpolation between the last and second last increment for an increase in fit residual of 5%. This procedure was consecutively performed for each fitting parameter. Third, a F test was performed to assess the significance of the effect of additional parameters on the fit residual.[97]

Powder X-ray diffraction (XRD)

Ex situ XRD measurements were conducted on an X’Pert PRO MPD diffractometer (Panalytical, θ-θ geometry), using Cu K alpha radiation and a solid-state multi-channel PIXcel detector. Wide-angle scans (5-90° 2θ, variable slits) were measured in reflection mode using a silicon sample holder. In situ XRD measurements were conducted on a STOE diffractometer (θ-θ Mode) using an

Anton Paar in situ cell. Thermal stability tests were conducted in 20 % O2 in He (total flow 100 ml/min) in a temperature range from 323 K to 723 K. The gas phase composition at the cell outlet was continuously monitored using a non-calibrated mass spectrometer in a

multiple ion detection mode (Pfeiffer Omnistar). Phase analysis was performed using the X´Pert Highscore Plus software package (Panalytical).

IR Spectroscopy

IR spectra were recorded on a Magna System 750 of Nicolet in a wavenumber range of 400 – 4000 cm-1. Samples were pressed into pellets of 13 mm in diameter after diluting with KBr.

Raman Spectroscopy

Raman spectra were recorded on a FT-RAMAN spectrometer (RFS 100, Bruker). A Nd:YAG laser with a wavelength of 1064 nm was used for excitation. Samples were measured in a glass sample holder with a resolution of 1 cm-1. The laser power at the sample position was adjusted to 150 mW. All measurement consisted of 200 scans for each sample. Scans were averaged for improvement of the signal-to-noise ratio.

Catalytic testing - selective propene oxidation

PMo12 (98 mg), PV2Mo10 (55 mg), and PW2Mo10 (55 mg) was placed in a SiO2 tube (30 cm length, 3 mm inner diameter) and fixed between two layers of quartz wool and treated under catalytic conditions For catalytic testing in selective propene oxidation a mixture of 5% propene (Linde Gas,

10% propene (3.5) in He (5.0)) and 5% oxygen (Linde Gas, 20% O2 (5.0) in He (5.0)) in helium (Air Liquide, 6.0) was used in a temperature range of 293-723 K Reactant gas flow rates of oxygen, propene, and helium were adjusted with separate mass flow controllers

(Bronhorst) to a total flow of 40 ml/min. All gas lines and valves were preheated to 473 K. Hydrocarbons and oxygenated reaction products were analyzed using a Carbowax capillary column connected to an AL2O3/MAPD column or a fused silica restriction (25 m x

0.32 mm) each connected to a flame ionization detector. O2, CO, and CO2 were separated using a Hayesep Q (2 m x 1/8``) and a Hayesep T packed column (0.5 m x 1/8``) as precolumns combined with a back flush. For separation, a Hayesep Q packed column (0.5 m x 1/8``) was connected via a molsieve (1.5 m x 1/8``) to a thermal conductivity detector (TCD). Conversion, product selectivity, and reaction rate were calculated by the following equations:

Conversion: (3.1)

Selectivity: (3.2)

reaction rate: (3.3)

with: volume fraction stoichiometric factor desired product volume flow

mass of molybdenum

3.3 Ex situ characterization of P(V,W)xMo12-x (x = 0, 1, 2)

3.3.1 Quantification of metal loading by XRF

Quantitative analysis of P(V,W)xMo12-x (x = 0, 1, 2) was performed to verify the synthesis process. Results of quantitative XRF measurements and nominal composition of

P(V,W)xMo12-x (x = 0, 1, 2) are summarized in Table 3-2. Experimental composition corresponded very well to the nominal composition and confirmed a successful synthesis.

Table 3-2: Results of quantitative XRF measurements and nominal composition of P(V,W)xMo12-x (x = 0, 1, 2). elements H P Mo V W O

PVMo11 nom.. wt.% 0.23 1.75 59.25 2.86 - 35.93

PVMo11 exp. wt.% - 0.75 62.07 2.24 - 34.48

PV2Mo10 nom. wt.% 0.29 1.78 55.22 5.86 - 36.84

PV2Mo10 exp.. wt.% - 0.89 58.62 4.91 - 35.14

PWMo11 nom. wt.% 0.16 1.62 55.16 - 9.61 33.45

PWMo11 exp.wt.% - 0.81 56.60 - 9.98 32.60

PW2Mo10 nom. wt.% 0.15 1.55 47.94 - 18.38 31.98

PW2Mo10 exp.wt.% - 0.66 50.73 - 17.35 31.27

3.3.2 Long-range structure of as-prepared P(V,W)xMo12-x (x = 0, 1, 2)

Long-range order structural analysis of P(V,W)xMo12-x (x = 0, 1, 2) was performed using X-ray powder diffraction (XRD). Fig. 3-1 shows the XRD powder pattern together with the theoretical pattern from structure refinement. H3[PMo12O40]·13H2O (ICSD 31128) was used as model for the refinement.[98] The samples P(V,W)xMo12-x (x = 0, 1, 2) were synthesized as 13 hydrate. Vanadyl or molybdenyl salts of the corresponding heteropolyacids typically crystallize in the cubic crystal systems.[99,100] Other Hydrates, for example 6 H2O, 30 H2O of the heteropolyacids, lead also to a cubic crystal system.[101,102]

PMo12

tensity

PVMo11

ntensity I

PV2Mo10

ntensity I

PWMo11

ntensity

PW2Mo10

ntensity I

10 20 30 40 50 60 70 80 Diffraction angle 2θ [°] experiment refinement difference

Fig. 3-1: XRD pattern of PMo12, PVMo11, PV2Mo10, PWMo11, and PW2Mo10 together with the XRD structure refinement of the 13 hydrate phase.

Therefore, the 13 hydrate, crystallizing in a triclinic crystal system, was chosen as desired compound to ensure the incorporation of the addenda atoms (V, W) in the Keggin ion and excluding the formation of the corresponding salt compounds. For the refinement, the molybdenum atoms were statistically replaced by the addenda atoms (V, W) depending on the degree of substitution in the model structure. Atomic coordinates were kept invariant. All samples showed the typical pattern of the 13 hydrate of HPOM.[98] The observed deviations are explained by different crystal water content due to grinding of the sample. Table 3-3 summarized the lattice parameters of the refinements. The volumes of the unit cells of the corresponding PVxMo12-x (x = 1, 2) samples indicated slightly decreased

volumes with higher degrees of vanadium substitution. The volumes of the unit cells of the corresponding PWxMo12-x (x = 1, 2) samples were nearly independent of the tungsten degree of substitution. A possible simplified explanation for the different behaviour of V and W centers substituting for Mo in Keggin type HPOM may be the considerably different ion radii of V (68 pm) and W (74 pm) in a six-fold coordination compared to Mo (74 pm).[103] The smaller ionic radius of V resulted in a decreased unit cell in contrast to W with identical ionic radius compared to Mo.

Table 3-3: Lattice parameters resulting from a refinement for PMo12, PVMo11, PV2Mo10, PWMo11, and PW2Mo10.

Sample PMo12 PVMo11 PV2Mo10 PWMo11 PW2Mo10 Space group P P P P P Lattice parameters a (Å) 13.546 13.551 13.566 13.531 13.577 b (Å) 14.088 14.058 14.022 14.047 14.051 c (Å) 14.134 14.145 14.126 14.109 14.135 alpha (°) 60.616 60.542 60.584 60.784 60.656 beta (°) 67.871 67.619 67.524 67.919 67.752 gamma (°) 70.213 70.143 70.189 70.346 70.210 V (Å3) 2137.213 2130.313 2124.298 2129.766 2136.110 GOF 2.52 2.98 2.79 2.97 3.38

3.4 Short-range order structural characterization of P(V,W)xMo12-x (x = 0, 1, 2)

IR-/RAMAN-Spectroscopy

IR and Raman spectra of P(V,W)xMo12-x (x = 0, 1, 2) are shown in Fig. 3-2. All samples exhibited identical IR and Raman bands and were comparable to bands known from the literature for H3[PMo12O40].[104–106] Therefore, it may be assumed that the Keggin ion structure existed for all samples without significant influence of the addenda atoms (V, W).

PMo12 PMo12

PVMo11

PV2Mo10

ntensity

ransmission T

PWMo11

PW2Mo10

1400 1200 1000 800 600 400 1000 800 600 400 200 Wavenumber [cm-1] Wavenumber [cm-1]

Fig. 3-2: IR (left) and Raman spectra (right) of PMo12, PVMo11, PV2Mo10, PWMo11, and PW2Mo10.

The IR and Raman signals are summarized in the Appendix (Table A 1; Table A 2). -1 H3+x[PVxMo12-xO40] (x = 1, 2) showed two shoulders in the IR spectra at ~1080 cm -1 (νas (P-Oi)) and ~980 cm (νas (V-Ot)) which could be assigned to V incorporated in the -1 Keggin ion.[107,108] Additionally, the IR signals for νas (P-Oi) (~1064 cm ) and νas (M- -1 Ot) (~977 cm ) shifted to lower wavenumbers for vanadium substituted and to higher wavenumbers for tungsten substituted HPOM. This confirmed the assumption of incorporated addenda atoms (V, W) in the HPOM.[108–110] The absence of a band in the Raman spectra at 1034 cm-1 related to a vanadyl cation suggested the V in the secondary structure.[111–113]. Therefore, it may be assumed that no further structural motifs except of the Keggin ion was formed independent of the degree of substitution.

X-ray Absorption Spectroscopy

Mo K edge analysis of P(V,W)xMo12-x (x = 0, 1, 2)

3 Fig. 3-3 depicts the theoretical and experimental Mo K edge FT(χ(k)·k ) of P(V,W)xMo12-x 3 (x = 0, 1, 2). The very similar shape of Mo K edge FT(χ(k)·k ) of P(V,W)xMo12-x

0.50

0.45 PMo12 0.40

0.35 PVMo11

0.30

) 3

k 0.25

)· k

( PV2Mo10

χ 0.20 FT( 0.15 PWMo 0.10 11

0.05 PW Mo 0.00 2 10

-0.05 0 1 2 3 4 5 6 R [Ǻ]

Fig. 3-3: Theoretical (dotted) and experimental (solid) Mo K edge FT(χ(k)·k3) of as prepared

PMo12, PVMo11, PV2Mo10, PWMo11, and PW2Mo10.

(x = 0, 1, 2) indicated a similar local structure of the Mo centers in P(V,W)xMo12-x (x = 0, 1, 2). For a more detailed analysis, theoretical phase and amplitudes were calculated for Mo-O and Mo-Mo distances and used for EXAFS refinement. The results of the 3 refinements of the Mo K edge FT(χ(k)·k ) of P(V,W)xMo12-x (x = 0, 1, 2) are summarized 3 in Table 3-4. The shapes of Mo K edge FT(χ(k)·k ) of P(V,W)xMo12-x (x = 0, 1, 2) were comparable to that of PVxMo12-x (x = 0, 1, 2) investigated by Ressler et al..[13,14,16] 3 Substituting PMo12 lead to decreasing amplitudes in the FT(χ(k)·k ) with higher amount of addenda atoms in the Keggin structures. The diminished amplitudes were accounted for by an increased disorder parameters σ2 depending on the degree of substitution. Probably the decreased amplitudes resulted from a distortion of the [(V,W)O]6 units in substituted

Keggin ions based on PMo12. The M-O and Mo-Mo distances were comparable for all

P(V,W)xMo12-x (x = 0, 1, 2) samples. This indicated an incorporation of addenda V and W atoms into the Mo based Keggin ion without influencing the structure. Hence, V and W were suitable elements for substituting Mo atoms in HPOM.

Table 3-4: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in as prepared P(V,W)xMo12-x (x = 0, 1, 2). Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental Mo K -1 edge XAFS χ(k) of P(V,W)xMo12-x (x = 0, 1, 2) (k range from 3.0-13.7.0 Å , R range from 0.9 to

4.0 Å, E0 = ~1.7, residuals ~11.3-20.0 Nind = 22, Nfree = 9). Subscript c indicates parameters that were correlated in the refinement.

Keggin model PMo12 PVMo11 PV2Mo10 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 1 1.68 1.64 0.0024 1.65 0.0039 1.66 0.0052

Mo-O 2 1.91 1.78 0.0030c 1.78 0.0043c 1.77 0.0059c

Mo-O 2 1.92 1.95 0.0030c 1.95 0.0043c 1.94 0.0059c Mo-O 1 2.43 2.40 0.0006 2.40 0.0011 2.40 0.0051

Mo-Mo 2 3.42 3.42 0.0054c 3.43 0.0068c 3.43 0.0076c

Mo-Mo 2 3.71 3.73 0.0054c 3.73 0.0068c 3.72 0.0076c

Keggin model PMo12 PWMo11 PW2Mo10 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 1 1.68 1.64 0.0024 1.63 0.0036 1.62 0.0026

Mo-O 2 1.91 1.78 0.0030c 1.77 0.0039c 1.77 0.0033c

Mo-O 2 1.92 1.95 0.0030c 1.94 0.0039c 1.94 0.0033c

Mo-O 1 2.43 2.40 0.0006 2.40 0.0020 2.41 0.0019

Mo-Mo 2 3.42 3.42 0.0054c 3.42 0.0082c 3.43 0.0094c

Mo-Mo 2 3.71 3.73 0.0054c 3.73 0.0082c 3.74 0.0094c

V K edge analysis of PVxMo12-x (x = 1, 2)

3 Fig. 3-4 depicts the theoretical and experimental V K edge FT(χ(k)·k ) of PVMo11 and 3 PV2Mo10. The very similar shape of the FT(χ(k)·k ) indicated similar local structures around the V centers in the unsupported HPOM Keggin structure. A detailed structure 3 analysis was performed by EXAFS. The shapes of the V K edge FT(χ(k)·k ) of PVMo11 3 and PV2Mo10 were comparable to the V K edge FT(χ(k)·k ) of PVMo11 and PV2Mo10

0.10 PVMo11

)

3 0.05

)·

( χ

PV2Mo10 FT(

0.00

-0.05 0 1 2 3 4 5 6 R [Ǻ]

Fig. 3-4: Theoretical (dotted) and experimental (solid) V K edge FT(χ(k)·k3) of as prepared

PVMo11 and PV2Mo10.x.

investigated by Ressler et al..[16,14] The results of the refinement of the V K edge 3 FT(χ(k)·k ) of PVMo11 and PV2Mo10 are shown in Table 3-5.. All V-O and V-Mo distances 2 were comparable for PVxMo12-x (x = 1, 2). The disorder parameters (σ ) for all V-O and

Table 3-5: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the V atoms in as prepared PVMo11 and PV2Mo10. Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental V K edge -1 XAFS χ(k) of PV2Mo12-SBA-15 (k range from 3.0-11.0 Å , R range from 0.9 to 4.0 Å, E0 = 0.6, residuals 9.1-11.7; Nind = 16, Nfree = 9). Subscript c indicates parameters that were correlated in the refinement.

Keggin model PVMo11 PV2Mo10 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) V-O 1 1.68 1.63 0.0078 1.62 0.0041

V-O 2 1.91 1.94 0.0154c 1.95 0.0120c

V-O 2 1.92 1.95 0.0154c 1.95 0.0120c

V-O 1 2.43 2.38 0.0178 2.38 0.0093

V-Mo 2 3.42 3.36 0.0173c 3.34 0.0151c

V-Mo 2 3.71 3.78 0.0173c 3.74 0.0151c

V-Mo distances were decreased for PV2Mo10 compared PVMo11. This decreased disorder 2 parameters (σ ) indicated a lower degree of distortion of the [VO6] units incorporated in the

Keggin ion for PVMo11.

W K edge analysis of PWxMo12-x (x = 1, 2)

EXAFS analysis of PWxMo12-x (x = 1, 2) rarely have been reported in the literature. 3 Therefore, W LIII edge FT(χ(k)·k ) of PWxMo12-x (x = 1, 2) were compared to H3[PW12O40]

(PW12) as reference to exclude the formation of PW12. The results confirmed the assumption, that tungsten was incorporated in the HPOM. Fig. 3-5 (left) depicts the 3 theoretical and experimental W LIII edge FT(χ(k)·k ) of PWxMo12-x (x = 1, 2) and the 3 experimental W LIII edge FT(χ(k)·k ) of PW12. All EXAFS spectra showed comparable 3 shapes, indicating a comparable structure around the W centers The W LIII edge χ(k)·k of 3 PW12 exhibited distinct differences compared to the W LIII edge χ(k)·k of PWxMo12-x (x = 1, 2) in the range above 9 Å-1. This indicated a slightly different structure or other 3 backscattering atoms around the W centers. W LIII edge FT(χ(k)·k ) of PWxMo12-x

14 0.25 PW12 12

0.20 10

) 8

k 0.15

3 )·

PWMo11 k

k 6

)·

(

k (

0.10 χ 4 FT( 2 0.05 0 PW2Mo10

0.00 -2 -4 0 1 2 3 4 5 6 4 6 8 10 R [Ǻ] k [Ǻ-1] Fig. 3-5: (left) Theoretical (dotted) and experimental (solid) W K edge FT(χ(k)·k3) and (right) 3 χ(k)·k of as prepared PW12, PWMo11, and PW2Mo10.

(x = 1, 2) could be sufficiently simulated using only W-Mo distances (Table 3-6). Hence, 2 the formation of predominantly PW12 was excluded. All disorder parameters (σ ) in the

EXAFS refinement for PWxMo12-x (x = 1, 2) were nearly identical indicating a comparable degree of distortion.

Table 3-6: Type and number (N), and XAFS disorder paramters (σ2) of atoms at distance R from the Mo atoms in as prepared PWMo11 and PW2Mo10. Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental W LIII edge -1 XAFS χ(k) of PWxMo12-x (x = 1, 2) (k range from 3.4-11.5 Å , R range from 0.9 Å to 3.8 Å, E0 =

3.2, residuals 9.1-13.2 Nind = 8, Nfree = 16)

Keggin model PWMo11-SBA-15 PW2Mo10-SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) W-O 1 1.68 1.69 0.0024 1.72 0.0025 W-O 2 1.91 1.82 0.0025 1.83 0.0026 W-O 2 1.92 1.94 0.0025 1.95 0.0026 W-O 1 2.43 2.32 0.0028 2.29 0.0028 W-Mo 2 3.42 3.46 0.0047 3.46 0.0045 W-Mo 2 3.71 3.69 0.0047 3.69 0.0045

3.5 In situ Characterization of bulk heteropolyacids

3.5.1 In situ XRD of PMo12-x(V,W)x x = 0, 1, 2 during oxidation conditions

For elucidating structure-activity relationships, an analysis of the structural evolution during oxidizing conditions was useful. Fig. 3-6 depicts selected in situ powder pattern of

PMo12 during treatment in 20% oxygen He (323 K to 723 K). At 323 K the typical pattern peaks for H3[PMo12O40]·3-8 H2O were identified.[114,115] Between 373 K and 473 K a dehydration of the HPOM lead to H3[PMo12O40] without constitutional water.[116,117]

The dehydration process continued until 673 K, where the anhydrous [PMo12O38.5] and β-

MoO3 were found.[117,34,118] Appendix Fig. A 1depicts the ion current m/e = 18 during temperature programmed oxidation representing the amount of water. The ion current increased between 498 K and 648 K confirming the loss of constitutional water. At

723 K α-MoO3 + β-MoO3

673 K [PMo12O38.5] + β-MoO3

623 K H3[PMo12O40]+ [PMo12O38.5]

573 K H3[PMo12O40]+ [PMo12O38.5]

523 K H3[PMo12O40]+ [PMo12O38.5] Intensitiy 473 K H3[PMo12O40]

423 K H3[PMo12O40]

373 K H3[PMo12O40]

323 K H3[PMo12O40]*3-8 H2O

10 15 20 25 30 35 40 45 Diffraction angle 2θ [°] Fig. 3-6: Selected in situ powder pattern during treatment in 20% oxygen in He (temperature range 323 K to 723 K) of H3[PMo12O40] · n H2O.

723 K a mixture of α-MoO3 + β-MoO3 with a higher concentration of the metastabile β-

MoO3 was identified.[118] Fig. 3-7 shows the in situ powder pattern of PVMo11 and

PV2Mo10 during treatment in 20% oxygen in He (323 K to 723 K). The powder pattern at

323 K shows for both PVMo11 and PV2Mo11 a hydrated structure (3-8 hydrate).[18,19]

Subsequently a dehydration of the water of crystallization resulted in PVxMo12-x (x = 1, 2) without water of crystallization. Between 473 K and 573 K for PVMo11 and 523 K for

PV2Mo10 a mixture of H3-x[PVxMo12-xO40] (x = 1, 2) and the anhydrous [PVxMo12-xO38.5-

0.5x] (x = 1, 2) was found. Above 623 K (for PVMo11) and 573 K (for PV2Mo10) only the

[PVxMo12-xO38.5-0.5x] (x = 1, 2) phase was identified.[117] At 673 K a mixture of

[PV2Mo10O37.5], α-MoO3, and β-MoO3 could be detected for PV2Mo10. At 723 K only

MoO3 was found for both PVMo11 and PV2Mo10. The phase composition of the two MoO3 phase was different to that of PMo12 at 723 K. The amount of β-MoO3 decreased with higher degree of vanadium substitution resulting in mainly α-MoO3 for PV2Mo10.

Fig. 3-8 shows the in situ powder pattern of PWMo11 and PW2Mo10 during treatment in

20% oxygen in He (323 K to 723 K). The powder pattern at 323 K shows for both PVMo11 and PV2Mo11 a hydrated structure (3-8 hydrate) comparable to H3-x[PVxMo12-xO40] (x = 0, 1, 2).[18,19] Subsequently dehydration of water of crystallization resulted in

PVMo11 PV2Mo10

723 K 723 K 673 K

673 K

623 K 623 K 573 K 573 K

523 K Intensitiy Intensitiy 523 K 473 K 473 K 423 K 423 K 373 K 373 K 323 K 323 K

10 15 20 25 30 35 40 45 10 15 20 25 30 35 40 45 Diffraction angle 2θ [°] Diffraction angle 2θ [°]

H3+x[PVxMo12-xO40] · 3-8 H2O (x = 1, 2) [PVxMo12-xO40-1.5·x] (x = 1, 2)

H3+x[PVxMo12-xO40] (x = 1, 2) [PV2Mo10O37.5] + MoO3

H3+x[PVxMo12-xO40]+ [PVxMo12-xO40-1.5·x] MoO3 (x = 1, 2)

Fig. 3-7: Selected in situ powder pattern during treatment in 20% oxygen in (temperature range

323 K to 723 K) of H4[PVMo11O40] · n H2O (left) and H5[PV2Mo10O40] · n H2O (right).

H3[PWxMo12-xO40] (x = 1, 2) without water of crystallization. Between 473 K and 623 K for PWMo11 and PW2Mo10 a mixture of H3[PWxMo12-xO40] (x = 1, 2) and the anhydrous

[PWxMo12-xO38.5] (x = 1, 2) was detectable. In contrast to H3-x[PVxMo12-xO40] (x = 0, 1, 2) no separated anhydrous [PWxMo12-xO38.5] (x = 1, 2) phase was found. At 673 K a mixture of [PWxMo12-xO38.5] (x = 1, 2) and at 623 K (for PVMo11) and β-MoO3 was found. At

723 K only MoO3 was found for both PWMo11 and PW2Mo10. This phase composition of the MoO3 phase were different to that of PMo12, PVMo11, and PV2Mo10. The amount of β-

MoO3 increased with higher degree of tungsten substitution resulting in mainly β-MoO3 for PWMo11 and PW2Mo10.

Both MoO3 modifications (α-MoO3, and β-MoO3) possess slight different structures. The smallest structural motif in α-MoO3, and β-MoO3 is a distorted MoO6 octahedron.[119]

α-MoO3 possesses a characteristic layer structure, with edges MoO6 octahedra sharing

PWMo11 PW2Mo10

723 K

673 K 673 K 623 K 623 K

573 K Intensitiy

573 K Intensitiy 523 K 523 K 473 K 473 K 423 K 423 K 373 K 373 K 323 K 323 K

10 15 20 25 30 35 40 45 10 15 20 25 30 35 40 45 Diffraction angle 2θ [°] Diffraction angle 2θ [°]

H3[PWxMo12-xO40] · 3-8 H2O (x = 1, 2) [PWxMo10O38.5] (x = 1, 2)+ MoO3

H3 [PWxMo12-xO40] (x = 1, 2) MoO3

H3[PWxMo12-xO40]+ [PWxMo12-xO38.5] (x = 1, 2)

Fig. 3-8: Selected in situ powder pattern during treatment in 20% oxygen in (temperature range

323 K to 723 K) of H3[PWMo11O40] · n H2O (left) and H3[PW2Mo10O40] · n H2O (right).

within the layer. The layers in α-MoO3 are connected via van der Waals interaction of corner-sharing MoO6 octahedra.[120] Similar to ReO3 the MoO6 octahedra in β-MoO3 are exclusively corner-shared.[121] An explanation for the favored formation of α-MoO3 depending on the degree of vanadium substitution in PVxMo12-x (x = 1, 2) could be Pauling`s 3rd rule.[122] This rule states, that the stability of an ionic structure decreases with higher amount of corner-sharing polyhedra. The distance between the neighboring

MoO6 octahedra in α-MoO3 is smaller than that of edge-sharing MoO6 octahedra in 5+ 6+ β-MoO3.[123] Therefore, the energy loss due to the incorporation of V compared to Mo 6+ 6+ was compensated by the formation of predominantly edge-sharing α-MoO3. W and Mo have an identical charge and the consideration of a lower charge was irrelevant. That is the reason why tungsten substituted HPOM lead to the formation of predominantly

β-MoO3. Additionally, many tertiary Mo/V or rather Mo/W as well as V/W/Mo mixed oxides are known in the literature.[124–126]The high miscibility of the mixed metal oxides

may be explained by the comparable ion radii in a six fold coordination and the preferred formation of MO6 (x = V, W, Mo) octahedra.[125,126] A further structure stabilizing effect of addenda atoms (V, W, Nb) in the synthesis of mixed molybdenum oxides with

Mo5O14-type structure has been shown.[127] Therefore, it may be assumed that the addenda atoms (V, W) possessed a structure directing effect during decomposition of HPOM under oxidizing conditions. The resulting structures depended on the addenda atom. Hence, vanadium substituted PVxMo12-x (x = 1, 2) favored the formation of α-MoO3 with edge-shared MO6 octahedra (Mo, V). Tungsten substituted PWxMo12-x (x = 1, 2) lead to the formation of β -MoO3 with corner-shared MO6 octahedra (Mo, W).

3.5.2 Functional characterization of bulk HPOM

Catalytic testing of P(V, W)xMo12-x (x = 0, 1, 2)

Reaction rates and selectivities of P(V,W)xMo12-x (x = 0, 1, 2) in propene oxidation at

723 K are shown in Fig. 3-9. Reaction rates for of P(V,W)xMo12-x (x = 0, 1, 2) were calculated for similar propene oxidation conditions (4-5 % propene conversion). Reaction rates for PVxMo12-x (x = 0, 1, 2) were different depending of the degree of vanadium and tungsten substitution. The reaction rates for PVxMo12-x (x = 0, 1, 2) increased with higher

acrylic acid acetone CO acetic acid propionaldehyde CO2

] 1 acrolein acetaldehyde - 100 s

(Mo) 1

80 12 -

10 60 8 40 6

Selectivity [%] Selectivity 4 20 2 0 0

a b c d e Reaction rate [µmol(proene)g rate Reaction -1 -1 Fig. 3-9: Reaction rate (µmol(propene)·g (Mo)·s ) and selectivity of (a) PV2Mo10, (b) PVMo11,

-1 -1 degree of vanadium substitution from 5.8 µmol(propene)g (Mo)s (PMo12) to 12.3 -1 -1 µmol(propene)g (Mo)s (PV2Mo10). The tungsten substituted samples showed an -1 -1 increased reaction rate to 7.4 µmol(propene)g (Mo)s too compared to PMo12. The influence on the catalytic activity was higher for the vanadium substituted samples

PVxMo12-x (x = 1, 2) in contrast to the tungsten substituted samples PWxMo12-x (x = 1, 2). Selectivities towards acrolein decreased with higher degree of vanadium substitution from

29% (PMo12) to 22% (PV2Mo10). In contrast to the selectivities of PVxMo12-x (x =' 1, 2), the selectivities towards acrolein for both tungsten substituted samples (27%) were comparable to that of unsubstituted PMo12 (29%). Selectivities towards propionaldehyde and acetaldehyde decreased with higher degree of substitution independent of the addenda atoms (V or W). Additionally, the selectivities towards acetic acid increased with higher degree of vanadium substitution. In contrast to that PWxMo12-x (x = 1, 2) showed not significant selectivities towards acetic acid (below 1%). The amount of total oxidation products (CO; CO2) increased with higher degree of substitution. The distribution of the total oxidation products was different depending to the nature and degree of substitution.

Compared to the product distribution of unsubstituted PMo12, PVxMo12-x (x = 1, 2) and

PWxMo12-x (x = 1, 2) showed an increased production of CO and CO2, respectively.

Influence of phosphorus and structure to the catalytic activity

In situ XRD characterization (chapter 3.5.1) showed an influence of the addenda atoms in HPOM on the resulting structures during thermal treatment under oxidizing conditions. A structural characterization during propene oxidation at temperatures above 698 K was not feasible due to coke formation in the in situ cell of the XRD. Therefore, the reactor for catalytic testing is used for catalytic treatment (5% propene and 5% O2 at 723 K; 0 h and

12h time on stream) of pure PMo12, PV2Mo12 and PW2Mo10. Subsequently, the samples were investigated with XRD after treatment (0 h and 12 h time on stream) to obtain structural information. Additional, the samples were investigated with AAS to quantify the content of phosphorus. Fig. 3-10 depicts the evolution of the propene conversion for

PMo12, PV2Mo10, and PW2Mo10 up to 12 h time on stream at 723 K. The sample weights of

PMo12, PV2Mo10, and PW2Mo10 were chosen to reach a comparable propene conversion (11.8-13.5) at 723 K with time on stream 0 h (c.f. chapter 3.2) All three samples exhibited a decreased activity with longer time on stream. The degree of deactivation was higher for

Time on stream 0 h 12 h

723 14 12 623

523 8 T [K] T 6 423

4 propene conversion [%] conversion propene 323 2 0 0 10 20 30 40 Cycles

Fig. 3-10: propene conversion for PMo12 ( ), PV2Mo10 ( ), and PW2Mo10 ( ) during catalytic testing

(5% propene and 5% O2 at 723 K; 12h time on stream).

PMo12 than for PV2Mo10 which again was higher than that of PW2Mo10. One explanation for the deactivation process is an enrichment of phosphorous on the surface of bulk HPOM. Phosphorus containing catalysts (i.e. VPO, FePO, MoPO) play a crucial role as oxidation catalysts.[128] Millet et al. showed that adding small amounts of phosphoric acid to the feed showed positive effects on long-term stability and catalytic performance of FePO catalysts during ODH of isobutyric acid into methacrylic acid.[129] The phosphorus source was needed to maintain a constant P/Fe ratio at the surface of the catalyst.[129] Another example for the relevance of P in oxidation reactions are VPO catalysts.[128,130] VPO catalysts showed migration of phosphorus species to the surface and a decreasing amount of phosphorus in the catalyst during water vapor treatment. The excess of phosphorus on the surface may suppress oxidation of VPO catalysts hindering formation of active sites for oxidation reactions. Subsequently, hydrolysis of P-O-P or P-O-V groups resulted in a removal of phosphate groups on the surface and an increasing activity.[128,130] The quantification of phosphorus in all samples (P(V,W)xMo12-x x = 0, 2) showed that the content of phosphorus was similar before and during treatment under catalytic conditions (0 h and 12 h time on stream) (Appendix Fig. A 4). Therefore, removal of phosphate groups from the HPOM was excluded. Hence, XRD structural analysis of the treated (time on stream 0 h and 12 h) samples was used to elucidate

structural differences (Fig. 3-11). Structural differences may be also responsible for the different deactivation processes and/or different reaction rates and selectivities (Fig. 3-9). XRD powder pattern for each sample measured at 0 h and at 12 h were nearly identical.

XRD pattern peaks of PMo12 (0 h and 12 h time on stream) corresponded to that of α-

MoO3 (c.f. Appendix Fig. A 2).[41] α-MoO3 is the thermodynamically stable modification of molybdenum oxides in their highest oxidation state (+6) and indicated an oxidizing character of the catalytic gas composition.[131,132] Kühn et al. showed that the treatment of α-MoO3 under propene oxidation conditions had no influence on the structure.[133]

Therefore, it may be assumed that PMo12 decomposed during propene oxidation conditions at 723 K to the thermodynamically stable α-MoO3 which lead in a poor catalytic activity (c.f. Fig. 3-9).[41]

The resulting structure for PV2Mo10 (0 h and 12 h time on stream) was probable a mixture of lacunary Keggin ions and Keggin ions. Ressler et al. described for PV2Mo10 a dynamic behaviour by isothermally switching from propene (reducing) to an oxidizing (propene and oxgen) and back to propene (reducing) conditions at 723 K. This in situ XAS experiments showed the formation of a short vanadium-molybdenum distance of about 2.8 Å for the

0h 12h

PMo12 PMo12

PV2Mo10 PV2Mo10

Intensitiy Intensitiy

PW2Mo10 PW2Mo10

10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 Diffraction angle 2Ɵ [°] Diffraction angle 2Ɵ [°]

Fig. 3-11: XRD powder pattern of PMo12, PV2Mo10, and PW2Mo10 after treatment in 5% propene and 5% oxygen in He at 723 K with time on stream 0 h (left) and 12 h (right).

reduced state. The oxidized state exhibits a longer distance of the vanadium center to an extra-Keggin molybdenum center at 3.2 Å.[16] The results suggested a mixture of at least two sites around two different V centers.[16] Therefore, it may be assumed, that the resulted structure for PV2Mo10 treated during catalytic conditions (5% propene + 5% oxygen in He at 723 K) corresponded to a mixture of lacunary Keggin ions and Keggin ions. Hence, a final structural solution was not feasible. Nevertheless, the mixture of the various structures may be responsible for the enhanced catalytic activity (c.f. Fig. 3-9).

XRD pattern of PW2Mo10 after catalytic treatment (0 h and 12 h time on stream) showed broad peaks indicating an amorphous or less crystalline compound. A mixture of α-MoO3 and Mo17O47 could be indentified from the XRD pattern (Appendix Fig. A 3).

Molybdenum centers in Mo17O47 have an average valence of ~ +5.5. This indicated, that the degree of reduction was higher for PW2Mo10 than for PMo12. Hence, tungsten lead to an increased reducibility of PW2Mo10 during propene oxidation conditions. The different structures formed during catalytic conditions (5% propene + 5% oxygen in He at 723 K) depended on the substituted element (V, Mo). Therefore a structure directing effect of the addenda atoms to structures formed during thermal treatment under catalytic may be assumed. The different structures resulting during catalytic conditions (5% propene + 5% oxygen in He at 723 K) were an explanation for the different catalytic behaviours of

P(V,W)xMo12-x ( x = 0, 1, 2). α-MoO3 resulting from PMo12 showed the lowest catalytic activity in propene oxidation. The mixture of lacunary Keggin ions and Keggin ions resulting for PV2Mo10 and the mixture of α-MoO3 and Mo17O47 resulting for PW2Mo10 showed an increased catalytic activity in propene oxidation.

3.6 Summary

P(V,W)xMo12-x (x = 0, 1, 2) were examined by a combination of various characterization techniques. The synthesis of P(V,W)xMo12-x (x = 0, 1, 2) lead to HPOM with the desired

Keggin type structure and chemical composition. P(V,W)xMo12-x (x = 0, 1, 2) crystallized as 13 hydrate in a triclinic crystal system. The 13 hydrate structure of the HPOM indicated the incorporation of the addenda atoms (V, W) in the Keggin ion. Hence, the formation of the corresponding salt compounds was excluded. The volume of the unit cell decreased with higher vanadium substitution because of the smaller ionic radius of V in a six-fold coordination in contrast to W with an identical ionic radius compared to Mo. The results of the IR and Raman measurements confirmed, that the addenda atoms (V, W) were incorporated in the Keggin ion. Peaks in the IR and Raman spectra indicating additional structure motifs except for the Keggin ion structure were not found. Results of the EXAFS refinements indicated, that the addenda atoms (V, W) were incorporated in the Keggin ion independent of the degree of substitution.

In situ XRD pattern of P(V,W)xMo12-x (x = 0, 1, 2) showed that the addenda atoms have an structure directing effect during decomposition under oxidizing conditions. Vanadium substitution in HPOM (PVxMo12-x (x = 1, 2)) lead to an increased formation of predominantly α-MoO3 depending on the degree of vanadium substitution. Conversely, tungsten substituted PWxMo12-x (x = 1, 2) exhibited an increased formation of β-MoO3 depending on the degree of tungsten substitution.

Catalytic tests revealed an increased activity for the substituted HPOM (P(V,W)xMo12-x (x = 1, 2)) depending on the degree of substitution. The addenda atoms (V, W) had a structure directing effect to the structures forming during propene oxidation conditions.

Unsubstituted PMo12 formed α-MoO3 with poor catalytic activity in propene oxidation. Vanadium substitution probably lead to a mixture of lacunary Keggin ions and Keggin ions, resulting in an enhanced catalytic activity. Tungsten substituted HPOM (P(W)xMo12-x

(x = 1, 2)) formed a mixture of α-MoO3 and Mo17O47 with an enhanced catalytic activity. Hence, the different structures resulting during catalytic condition and the different chemical compositions (P(V,W)xMo12-x (x = 0, 1, 2)) seemed to be responsible for the different catalytic activities and selectivities during propene oxidation.

4 Characterization of P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) (10 wt.% Mo)

For elucidating structure activity correlations of model systems for catalytic investigations, a detailed knowledge about structure and chemical composition is indispensable. Therefore, various characterization methods, such as XRF, XRD, Nitrogen Physisorption, and XAS are necessary for a sufficient characterization of the used catalyst systems. In this chapter it will be shown that the initial Keggin structure was retained after supporting of

P(V,W)xMo12-x (x = 0, 1, 2) onto SBA-15. Furthermore it should be ensured, that the supporting process lead to the desired metal loadings of 10 wt.% Mo for the current model catalyst. Additionally, an analysis of the structure of the supported model catalyst was carry out to confirm well dispersed Keggin ion structure motifs on the support material.

4.1 Sample Preparation

Preparation of the support material silica SBA-15

Silica SBA-15 was prepared according to Ref. [23]. 16.2 g of triblock copolymer (Aldrich, P123) were dissolved in 294 g water and 8.8 g hydrochloric acid at 308 K and stirred for 24 h. After addition of 32 g tetraethyl orthosilicate for 24 h, the reaction mixture was stirred for 24 h at 373 K. The resulting gel was transferred to a glass bottle and the closed bottle was heated to 388 K for 24 h. Subsequently, the suspension was filtered and washed with a mixture of H2O/EtOH (100:5). The resulting white powder was dried at 378 K for 3 h and calcined at 453 K for 3 h and at 823 K for 5 h. Three batches of silica SBA-15 were used for P(V, W)xMo12-x-SBA-15 (10 wt. % Mo).

Preparation of HPOM supported on silica SBA-15

HPOM (Chapter 3.1) were supported on SBA-15 via incipient wetness. The amount of molybdenum was adjusted to 10 wt.%. Therefore an aqueous solution of HPOM was used.

4.2 Sample characterization

X-Ray Fluorescence Analysis

Physisorption measurements

Nitrogen physisorption isotherms were measured at 77 K on a BEL Mini II volumetric sorption analyzer (BEL Japan, Inc.). The silica SBA-15 sample was treated under vacuum at 368 K for about 20 min and at 448 K for about 17 h before starting the measurement. Data processing was performed using the BELMaster V.5.2.3.0 software package. The specific surface area was calculated using the Brunauer–Emmett–Teller (BET) method in 2 the relative pressure range of 0.03-0.24 assuming an area of 0.162 nm per N2 molecule.[69] The adsorption branch of the isotherm was used to calculate pore size distribution and cumulative pore area according to the method of Barrett, Joyner, and Halenda (BJH).[70]

Powder X-ray diffraction (XRD)

XRD measurements were conducted on an X’Pert PRO MPD diffractometer (Panalytical, θ-θ geometry), using Cu K alpha radiation and a solid-state multi-channel PIXcel detector. Wide-angle scans (5-90° 2θ, variable slits) were measured in reflection mode using a silicon sample holder. Small-angle scans (0.4-6.0° 2θ, fixed slits) were collected in transmission mode with the sample spread between two layers of Kapton foil.

X-ray absorption spectroscopy (XAS)

The modified H3[PMo12O40] (ICSD 209 [14,93]) model structure was modified by substituting a Mo for either V or W (V for PVxMo12-x; W for PWxMo12 (x = 1, 2)) .Single scattering and multiple scattering paths of the model structure were calculated up to 6.0 Å with a lower limit of 4.0% in amplitude with respect to the strongest backscattering path. EXAFS refinements were performed in R space simultaneously to magnitude and imaginary part of a Fourier transformed k3-weighted and k1-weighted experimental χ(k) using the standard EXAFS formula.[94] This procedure reduces the correlation between the various XAFS fitting parameters. Structural parameters allowed to vary in the refinement were (i) disorder parameter σ2 of selected single-scattering paths assuming a symmetrical pair-distribution function and (ii) distances of selected single-scattering paths. Detailed information about the fitting procedure are described in chapter 3.2.

4.3 Results of the Characterization

Quantification of metal loading XRF

Quantitative analysis of P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) was performed to verify the supporting process. Results of quantitative XRF measurements and nominal composition of P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) was summarized in Table 4-1. Experimental composition corresponded very well to the nominal composition and confirmed a successful supporting process. For simplification of the nomenclature the samples were still denoted as P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) (10 wt.% Mo).

Table 4-1: Results of quantitative XRF measurements and nominal composition of P(V,W)xMo12-x- SBA-15 (x = 0, 1, 2). Elements H P Mo V W O Si

PMo12-SBA-15 nom. wt.% 0.03 0.27 10.01 - - 50.37 39.32

PMo12-SBA-15 exp. wt.% - 0.33 9.89 - - 50.30 39.41

PVMo11-SBA-15 nom. wt.% 0.04 0.29 10.00 0.48 - 50.33 38.85

PVMo11-SBA-15 exp. wt.% - 0.42 9.15 0.42 - 50.44 39.46

PV2Mo10-SBA-15 nom. wt.% 0.05 0.32 9.99 1.06 - 48.99 37.15

PV2Mo10-SBA-15 exp. wt.% - 0.50 9.86 1.1 - 50.13 38.13

PWMo11-SBA-15 nom. wt.% 0.03 0.29 9.99 - 1.74 49.67 38.28

PWMo11-SBA-15 exp. wt.% - 0.43 9.92 - 1.83 49.57 38.24

PW2Mo10-SBA-15 nom. wt.% 0.03 0.32 10.02 - 3.84 48.81 36.97

PW2Mo10-SBA-15 exp. wt.% - 0.43 9.80 - 3.94 48.74 37.09

4.3.1 Long-range structure of as-prepared P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2)

The long-range structure of as-prepared SBA-15 and P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) were investigated by low-angle and wide-angle X-ray diffraction (Fig. 4-1). At small angles SBA-15 and P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) exhibited the characteristic patterns ((10l, (11l) and (20l)) representing the hexagonal pore structure of nanostructured

SBA-15. The lattice constant of the hexagonal unit cell of a = 12.1 nm of SBA-15 was determined from the (10l) peak (Fig. 4-1; left). Lattice constants of the hexagonal unit cell resulting for P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) were nearly identical to unsupported

SBA-15. Supporting P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) on SBA-15 showed no influence to the pore structure of the support material. Wide-angle X-ray diffraction patterns of a

SBA-15 and P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) with a loading of 10 wt.% Mo, showed no long-ranged ordered molybdenum oxide species (Fig. 4-1; right). This indicated sufficiently dispersed Keggin ions without formation of extended crystalline HPOM structures.

(10l) (11l) (20l)

SBA-15 SBA-15

PMo12-SBA-15

PVMo11-SBA-15 PVMo -SBA-15

ntensity 11

ntensity I I

PV2Mo10-SBA-15 PV2Mo10-SBA-15

PWMo -SBA-15 PWMo11-SBA-15 11

PW2Mo10-SBA-15 PW2Mo10-SBA-15

1 2 3 10 20 30 40 50 60 70 80 Diffraction angle 2Ɵ [°] Diffraction angle 2Ɵ [°]

Fig. 4-1: Low-angle (left) and wide-angle (right) X-ray diffraction patterns of SBA-15, PMo12-

SBA-15, PVMo11-SBA-15, PV2Mo10-SBA-15, PWMo11-SBA-15, PW2Mo10-SBA-15. (10l), (11l), and (20l) reflections are indicated.

Nitrogen physisorption

Nitrogen adsorption/desorption isotherms and pore distributions (BJH) of SBA-15 and

PMo12-SBA-15 are shown in Fig. 4-2. The isotherms of SBA-15 and PMo12-SBA-15 were of type IV indicative of mesoporous materials. Adsorption and desorption branches were nearly parallel in SBA-15 and PMo12-SBA-15 as expected for regularly shaped pores. N2 physisorption isotherms of PMo12-SBA-15 resembled that of the original SBA-15.

Constrictions of the pores due to deposition of PMo12 on SBA-15 could be excluded. Pore radius of the supported material decreased slightly. Therefore, the results of nitrogen physisorption measurements indicated, that PMo12 was sufficient dispersed on the support material without influence on the pore structure. SBA-15 possessed a BET surface area of 843 m2/g. BJH calculations of pore size distributions resulted in a pore diameter of dBJH = 10.6 nm. Given the pore diameter and the lattice constant (a = 12.1 nm), a wall thickness of the SBA-15 material used amounted to ~2 nm. PMo12-SBA-15 was chosen exemplary to elucidated an influence on pore diameter and surface area of the support material after deposition. PVMo11-SBA-15,

PV2Mo10-SBA-15, PWMo11-SBA-15 and, PW2Mo10-SBA-15 were not analyzed by

] ] 1

- 0.4

g 1

800 -

0.3

] [ml nm [ml

1 0.2

600 p

/d 0.1

V d 0 400 5 7 9 11 13 15

Pore Diameter [nm] Volume [ml g [ml Volume

200

0 0.0 0.2 0.4 0.6 0.8 1.0

Relative Pressure p/p0

Fig. 4-2: Nitrogen physisorption isotherms of silica SBA-15 (square) and PMo12-SBA-15 (circle), and pore distributions of silica SBA-15 (square) and PMo12-SBA-15 (circle), (inset).

nitrogen physisorption because of the required pretreatment (0.5 h at 450°C in He) which leads to a dehydration and dehydroxylation process concomitant with structural changes of the orginate Keggin structure. Therefore the result was used as indicator to exclude a significant influence on the pore structure.

Table 4-2 summarizes specific surface area aBET, external surface area aEXT, area corresponding to the mesopores aMeso, pore diameter dpore, and mesopore volume VMeso of

SBA-15 and PMo12-SBA-15

Table 4-2: Specific surface area aBET (calculated by BET method), external surface area aEXT

(calculated as the difference between aBET and aMeso), area corresponding to the mesopores aMeso, pore diameter dBJH (calculated by BJH method), mesopore volume VMeso, of SBA-15 and PMo12- SBA-15. 2 2 2 3 aBET (m /g) aExt (m /g) aMeso (m /g) dBJH (nm) VMeso (cm /g) SBA-15 843 145 698 10.6 1.233

PMo12-SBA-15 603 67 536 9.2 0.940

4.3.2 Short-range order structural characterization of as-prepared P(V,W)xMo12-x-SBA- 15 (x = 0, 1, 2)

X-Ray Absorption Spectroscopy

XAS was chosen as suitable spectroscopic method to analyze the structure of the Keggin ion motif on the support material. XRD was unsuitable for a structural analysis, because no long-ranged ordered molybdenum oxide species (Fig. 4-1; right) were expected. IR- and RAMAN-spectroscopy were not chosen as analyzing method as well. The IR- and

RAMAN spectra of P(V,W)xMo12-x (x = 0, 1, 2) were superimposed with the excess of SBA-15. Additionally, XAS is an element specific method and suitable for analyzing substituted compounds.

Mo K edge analysis of P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2)

Fig. 4-3 (left) depicts the Mo K edge XANES spectra of P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) and (right) the theoretical and experimental Mo K edge FT(χ(k)·k3) of

P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2). A comparison of the Mo K edge XANES spectra confirmed identical structural motifs in P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2). Analysis of the Mo K edge position (Fig. 4-3, left; broken line) yielded an average valence of ~6 (cf. Appendix Fig. A 6) of the substituted supported HPOM.[134] Therefore, substitution and the supporting process did not have an influence on the average valence of Mo in 3 P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2). The very similar shape of Mo K edge FT(χ(k)·k ) of

P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) indicated a similar local structure around the Mo centers in the unsupported and supported HPOM Keggin structure. P(V,W)xMo12-x-SBA- 15 (x = 0, 1, 2). The results of the refinements of the Mo K edge FT(χ(k)·k3) of

0.50 PMo12-SBA-15 PMo -SBA-15 0.45 12

PVMo11-SBA-15 0.40

0.35 PVMo11-SBA-15

PV2Mo10-SBA-15

0.30

) 3 PWMo11-SBA-15 0.25 PV2Mo10-SBA-15

χ(k)·k 0.20 PW Mo -SBA-15 2 10 FT(

0.15 PWMo11-SBA-15 Normalized absorption Normalized 0.10

0.05 PW2Mo10-SBA-15 0.00

-0.05 20.0 20.1 20.2 0 1 2 3 4 5 6 Photon energy [keV] R [Ǻ]

Fig. 4-3: (left) Mo K edge XANES with indicated Mo K position (broken line) and (right) 3 theoretical (dotted) and experimental (solid) Mo K edge FT(χ(k)·k ) of as prepared PMo12-SBA-15,

PVMo11-SBA-15, PV2Mo10-SBA-15, PWMo11-SBA-15, PW2Mo10-SBA-15.

P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) are shown in Table 4-3. Substituting PMo12 lead to decreasing amplitudes in the FT(χ(k)·k3) with higher amount of addenda atoms in the Keggin structures. The diminished amplitudes were discernible in all increased disorder parameters σ2 with substitution degree. Probably the decreased amplitudes resulted from a distortion of the [(V,W)O]6 units in substituted Keggin ions based on PMo12. Comparable results were found in the unsupported P(V,W)xMo12-x (x = 0, 1, 2) (cf. chapter 3)

Table 4-3: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in as prepared P(V, W)xMo12-x-SBA-15 (x= 0, 1, 2). Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental Mo K edge XAFS χ(k) of P(W,V)xMo12-x-SBA-15 (x = 0, 1, 2) (k range from 3.0-13.7 -1 Å , R range from 0.9 to 4.0 Å, E0 = ~1.7, residuals ~11.3-20.0 Nind = 22, Nfree = 9). Subscript c indicates parameters that were correlated in the refinement.

Keggin model PMo12-SBA-15 PVMo11-SBA-15 PV2Mo10-SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 1 1.68 1.64 0.0025 1.64 0.0035 1.66 0.0053

Mo-O 2 1.91 1.78 0.0033c 1.78 0.0035c 1.77 0.0067c

Mo-O 2 1.92 1.95 0.0033c 1.95 0.0035c 1.94 0.0067c Mo-O 1 2.43 2.40 0.0008 2.40 0.0006 2.39 0.0008

Mo-Mo 2 3.42 3.42 0.0052c 3.43 0.0057c 3.43 0.0076c

Mo-Mo 2 3.71 3.74 0.0052c 3.73 0.0057c 3.72 0.0076c

Keggin model PMo12-SBA-15 PWMo11-SBA-15 PW2Mo10-SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 1 1.68 1.64 0.0025 1.64 0.0028 1.62 0.0026

Mo-O 2 1.91 1.78 0.0033c 1.77 0.0038c 1.77 0.0033c

Mo-O 2 1.92 1.95 0.0033c 1.94 0.0038c 1.94 0.0033c

Mo-O 1 2.43 2.40 0.0008 2.40 0.0013 2.41 0.0019

Mo-Mo 2 3.42 3.42 0.0052c 3.42 0.0073c 3.43 0.0094c

Mo-Mo 2 3.71 3.74 0.0052c 3.74 0.0073c 3.73 0.0094c

V K edge analysis of PV2Mo10-SBA-15 (x = 1, 2)

Fig. 4-4 (left) depicts the V K edge XANES spectra of PV2Mo10 and PV2Mo10-SBA-15 3 and (right) the theoretical and experimental V K edge FT(χ(k)·k ) of PV2Mo10 and

PV2Mo10-SBA-15. XAS analysis at the V K edge for PVMo11-SBA-15 was hardly feasible due to the low content of V (~0.5 wt.%) beside Mo (~10 wt.%). Hence only an XAS analysis at V K edge for PV2Mo10-SBA-15 was performed and compared to unsupported

PV2Mo10. V K edge XANES spectra of PV2Mo10 -SBA-15 were identical with V K edge

XANES spectra of PV2Mo10 (Fig. 4-4, left). Comparing the pre-edge peak at the V K edge of PV2Mo10SBA-15 and vanadium oxide as reference compound indicated an average V valence between 4 and 5 (Appendix Fig. A 5). Results of the Mo and V K edge XANES spectra indicate a successful deposition of heteropolyoxomolybdates on the support material. The very similar shape of the FT(χ(k)·k3) indicated similar local structure around the V and Mo centers in the unsupported and supported HPOM Keggin structure. A detailed structure analysis was performed by EXAFS. The results of the refinement of the 3 V K edge FT(χ(k)·k ) of PV2Mo10-SBA-15 are shown in Table 4-4. Supporting PV2Mo10

1.5 PV Mo 0.10 2 10

PV2Mo10

1.0 ) 3 PV Mo -SBA-15

2 10 0.05

χ(k)·k FT( PV2Mo10-SBA-15 0.5

Normalized absorption Normalized 0.00

0.0 5.45 5.50 5.55 5.60 0 1 2 3 4 5 6 Photon energy [keV] R [Ǻ]

Fig. 4-4: (left) V K edge XANES of PV2Mo10-SBA-15 and PV2Mo10; (right) Theoretical (dotted) 3 and experimental (solid) V K edge FT(χ(k)·k ) of as prepared PV2Mo10-SBA-15 and PV2Mo10.

on SBA-15 resulted in a decreased amplitude of the corresponding V K edge FT(χ(k)·k3). 3 Conversely, the Mo K edge FT(χ(k)·k ) of PVxMo12-x-SBA-15 (x = 0, 1, 2) showed a minor increasing amplitude in the range between 2.5-3.8 Å. This increase in amplitude of the 3 FT(χ(k)·k ) was previously reported for PVMo11 and PVMo11-SBA-15.[27] Eventually, the Mo and V K edge FT(χ(k)·k3) confirmed that the Keggin ion structure motifs were maintained upon supporting PVxMo12-x on SBA-15.

Table 4-4: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the V atoms in as prepared PV2Mo12-SBA-15. Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental V K edge -1 XAFS χ(k) of PV2Mo12-SBA-15 (k range from 3.0-11.0 Å , R range from 0.9 to 4.0 Å, E0 = 0.6, residuals 13.8; Nind = 16, Nfree = 9). Subscript c indicates parameters that were correlated in the refinement.

Keggin model PV2Mo10-SBA-15 N R(Å) R(Å) σ2(Å2) V-O 1 1.68 1.60 0.0053

V-O 2 1.91 1.94 0.0017c

V-O 2 1.92 1.95 0.0017c V-O 1 2.43 2.43 0.0156

V-Mo 2 3.42 3.32 0.0178c

V-Mo 2 3.71 3.71 0.0178c

W LIII edge analysis of PWxMo12-x-SBA-15 (x = 1, 2)

Fig. 4-5 (left) depicts the W LIII edge XANES spectra of PWMo11-SBA-15 and PW2Mo10- 3 SBA-15 and (right) the theoretical and experimental W LIII K edge FT(χ(k)·k ) of

PWMo11-SBA-15 and PW2Mo10-SBA-15. XANES spectra of PWxMo12-x-SBA-15 (x= 1,

2) exhibited increased white lines comparing to bulk PWxMo12-x (x= 1, 2) (cf. chapter 0) The change of white line intensities indicated a change of the absorption properties. Therefore, it could be assumed, that the supported Keggin ions were well dispersed on the

SBA-15 causing an increased whiteline. The very similar shape of W LIII edge EXAFS 3 spectra of PWxMo12-x-SBA-15 (x = 1, 2) FT(χ(k)·k ) indicated a similar local structure of

0.15 4.0 PWMo11-SBA-15

3.5

3.0 0.10

)

2.5 3

2.0 (k)·k χ 0.05 PWMo11-SBA-15 1.5 FT( PW2Mo10-SBA-15 PW2Mo10-SBA-15

Normalized absorption Normalized 1.0 0.00 0.5

0.0 10.2 10.3 10.4 0 1 2 3 4 5 6 Photon energy [keV] R [Ǻ]

Fig. 4-5: (left) W LIII edge XANES of PW2Mo10-SBA-15 and PWMo11-SBA-15; (right) 3 Theoretical (dotted) and experimental (solid) W K edge FT(χ(k)·k ) of as prepared PW2Mo10-SBA-

15 and PWMo11-SBA-15. the W centers in the unsupported and supported HPOM Keggin structure. Comparable to the white line height, the amplitudes representing the W-Mo distances were increased. Thus, a dispersion effect could be determined from the amplitudes or rather from decreased disorder parameters σ2 representing the heights of the amplitudes compared to bulk

PWxMo12-x (x = 1, 2). The results confirmed the results of the vanadium substituted

PVxMo12-x (x = 1, 2). The supporting process did not have an influence on the Keggin ion structure on the support material.

Table 4-5: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in as prepared PWxMo11-x-SBA-15 (x=1, 2). Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental W LIII -1 edge XAFS χ(k) of PWxMo12-x-SBA-15 (x = 1, 2) (k range from 3.4-11.5 Å , R range from 0.9 Å to

3.8 Å, E0= ~ 3.2, residuals ~9.1-13.5 Nind = 8, Nfree = 16). Subscript c indicates parameters that were correlated in the refinement.

Keggin model PWMo11-SBA-15 PW2Mo10-SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) W-O 1 1.68 1.71 0.0018 1.72 0.0017

W-O 2 1.91 1.82 0.0019 1.83 0.0018 W-O 2 1.92 1.95 0.0019 1.95 0.0018 W-O 1 2.43 2.31 0.0020 2.29 0.0020

W-Mo 2 3.42 3.46 0.0049 3.47 0.0035

W-Mo 2 3.71 3.70 0.0049 3.69 0.0036

4.4 Conclusion

P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) were examined by a combination of various characterization techniques. The supporting process of P(V,W)xMo12-x (x = 0, 1, 2) on SBA-15 via incipient wetness lead to the desired metal loadings of 10 wt.% Mo for the current model catalyst. P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) were sufficiently dispersed on the support material without any influence on the pore structure of the support material.

Supporting P(V,W)xMo12-x on SBA-15 resulted in regular and well dispersed Keggin ions on the support material. The formation of extended crystalline HPOM structures could be excluded. Therefore, substituting Mo atoms with addenda atoms (i.e. V, W) make Keggin type heteropolyoxomolybdates suitable model systems to study structure activity relationships. Supported catalytic species posses high dispersions and an improved surface to bulk ratio. Therefore, structure activity relationships can be readily deduced from the characteristic oxide species observed on the support material under catalytic conditions.

Hence, in the following chapters (5-6) P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) with a loading of 10 wt.% Mo were investigated under catalytic conditions and structure activity relationships presented.

5 Characterization of PVxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions

Keggin type H4[PVMo11O40] has been reported to exhibit a pronounced interaction effect with SBA-15 as support material.[27] This effect resulted in a further decreased thermal stability of the supported Keggin ions compared to the bulk materials. Under catalytic conditions PVMo11-SBA-15 formed a mixture of tetrahedrally and octahedrally coordinated [MoO4] and [MoO6] units.[27] However, the role and structural evolution of V and P in PVxMo12-x-SBA-15 (x = 0, 1, 2) under catalytic conditions remained largely unknown. In this chapter in situ X-ray absorption spectroscopy investigations at the Mo K edge of PVxMo12-x-SBA-15 (x = 0, 1, 2) and at the V K edge of PV2Mo10-SBA-15 during propene oxidation conditions were performed. Moreover, 31P MAS NMR measurement of

PV2Mo10-SBA-15 and the reference H3PO4 supported on SBA-15 (denoted as H3PO4- SBA-15) after catalytic oxidation with propene are described. Correlations between structural evolution of [MoOx] and [VOx] units and performance under catalytic conditions will be described. Additionally, the obtained structures and catalytic performances were compared to a suitable supported reference material.

5.1 Experimental

5.1.1 Sample Characterization

31P-NMR measurement

31P MAS NMR spectra were recorded on a Bruker Avance 400 spectrometer (31P: 161.92 MHz) using a 4 mm double resonance HX MAS probe. Data collection used a 90° pulse with a relaxation delay of 60 s under a MAS rotation of 12 kHz. Spectra were referenced to

85% H3PO4 in aqueous solution using solid NH4H2PO4 (δ = 0.81 ppm) as a secondary reference.

X-ray absorption spectroscopy (XAS)

Transmission XAS experiments were performed at the Mo K edge (19.999 keV) at beamline X at the Hamburg Synchrotron Radiation Laboratory, HASYLAB, using a Si(311) double crystal monochromator. Transmission XAS experiments at the V K edge (5.465 keV) were also performed at HASYLAB, using a Si(111) double crystal monochromator at beamline C. In situ experiments were conducted in a flow reactor at atmospheric pressure (5 vol% oxygen in He, total flow ~30 ml/min, temperature range from 303 to 723 K, heating rate 4 K/min). The gas phase composition at the cell outlet was continuously monitored using a non-calibrated mass spectrometer in a multiple ion detection mode (Omnistar from Pfeiffer). X-ray absorption fine structure (XAFS) analysis was performed using the software package WinXAS v3.2..[91] Background subtraction and normalization were carried out by fitting linear polynomials and 3rd degree polynomials to the pre-edge and to the post- edge region of an absorption spectrum, respectively. The extended X-ray absorption fine structure (EXAFS) χ(k) was extracted by using cubic splines to obtain a smooth atomic 3 background μ0(k). The FT(χ(k)·k ), often referred to as pseudo radial distribution function, was calculated by Fourier transforming the k3-weighted experimental χ(k) function, multiplied by a Bessel window, into the R space. EXAFS data analysis was performed using theoretical backscattering phases and amplitudes calculated with the ab-initio multiple-scattering code FEFF7.[92] Structural data employed in the analyses were taken from the Inorganic Crystal Structure Database (ICSD).

Single scattering paths in the hexagonal MoO3 model structure (ICSD 75417 [135]) and a modified Na2MoO4 structure (ICSD 24312 [136]) were calculated up to 6.0 Å with a lower limit of 4.0% in amplitude with respect to the strongest backscattering path. EXAFS refinements were performed in R space simultaneously to magnitude and imaginary part of a Fourier transformed k3-weighted and k1-weighted experimental χ(k) using the standard EXAFS formula.[94] This procedure reduces the correlation between the various XAFS fitting parameters. Structural parameters allowed to vary in the refinement were (i) disorder parameter σ2 of selected single-scattering paths assuming a symmetrical pair- distribution function and (ii) distances of selected single-scattering paths. Detailed information about the fitting procedure are described in chapter 3.2.

Powder X-ray diffraction (XRD)

Temperature programmed reduction

Temperature programmed reduction (TPR) was performed with a catalysts analyzer from BEL Japan Inc. equipped with a silica glass tube reactor. Samples were placed on silica wool inside the reactor next to a thermocouple. A gas flow (5 % H2 in Ar) of 60 ml/min was adjusted during reaction. A heating rate of 8 K / min to 973 K was used while H2 consumption was measured with a TCD unit. All samples were pretreated with a gas flow of 60 ml/min Ar at 393 K for about 45 min before starting the measurement. For measurements 37.2 mg PMo12-SBA-15, 40.4 mg PVMo11-SBA-15, and 39.8 PV2Mo10- SBA-15 were used.

Catalytic testing - selective propene oxidation

Quantitative catalysis measurements were performed using a fixed bed laboratory reactor connected to an online gas chromatography system (Varian CP-3800) and a non-calibrated mass spectrometer (Pfeiffer Omnistar). The fixed-bed reactor consisted of a SiO2 tube (30 cm length, 9 mm inner diameter) placed vertically in a tube furnace. In order to achieve a constant volume and to exclude thermal effects, catalysts samples (~ 38 mg) were diluted with boron nitride (Alfa Aesar, 99.5%) to result in an overall sample mass of 375 mg. For catalytic testing in selective propene oxidation a mixture of 5% propene (Linde Gas, 10% propene (3.5) in He (5.0)) and 5% oxygen (Linde Gas, 20% O2 (5.0) in He (5.0)) in helium (Air Liquide, 6.0) was used in a temperature range of 293-723 K Reactant gas flow rates of oxygen, propene, and helium were adjusted with separate mass flow controllers (Bronhorst) to a total flow of 40 ml/min. All gas lines and valves were preheated to 473 K. Hydrocarbons and oxygenated reaction products were analyzed using a Carbowax capillary column connected to an AL2O3/MAPD column or a fused silica restriction (25 m·0.32 mm

each) connected to a flame ionization detector. O2, CO, and CO2 were separated using a Hayesep Q (2 m x 1/8``) and a Hayesep T packed column (0.5 m x 1/8``) as precolumns combined with a back flush. For separation, a Hayesep Q packed column (0.5 m x 1/8``) was connected via a molsieve (1.5 m x 1/8``) to a thermal conductivity detector (TCD). Details about the calculation of conversion, selectivity, and reaction rate are described in chapter 3.2.

5.1.2 Sample preparation

PVxMo12-x (x = 0, 1, 2) supported SBA-15 were prepared as described in chapter 4.1.

A reference material (denoted as V2Mo10Ox-SBA-15) was prepared as follows.

232.1 mg (NH4)6Mo7O24·4H2O and 29.3 mg (NH4)6V10O28·6H2O were dissolved in water and were deposited via incipient wetness on 1 g SBA-15 to obtain metal loading of 10 wt.% Mo and 1 wt.% V. The sample was dried for 18 h at room temperature and calcined for 3 h at 773 K. H3PO4-SBA-15 was prepared by depositing 1.1 ml of 0.12 M phosphoric acid on 1 g of silica SBA-15.

5.2 Structural characterization of PVxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions

In situ XANES analysis

PVxMo12-x-SBA-15 (x = 1, 2) was investigated by in situ XAS in propene oxidation conditions. Fig. 5-1 depicts the evolution of molybdenum XANES spectra of PVMo11- SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen. XAS analysis at V K edge for PVMo11-SBA-15 was hardly feasible due to the low content of V (~0.5 wt.%) beside Mo (~10 wt.%). Fig. 5-2 shows the evolution of vanadium (a) and molybdenum (b) XANES spectra of

PV2Mo10-SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen. The pre-edge peak features can be employed to elucidate the local structure around the metal centers. Using the pre-edge peak height sufficed to quantify the contribution of tetrahedral [MO4] and distorted [MO6] (M = V, Mo) units present during

absorption Normalized Normalized

673 20.20 573 20.15 T [K] 473 20.10 373 20.05 20.00 Photon energy [keV]

Fig. 5-1: in situ Mo K edge XANES spectra of PVMo11-SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen in helium in a temperature range between 300 K and 723 K. thermal treatment of the catalysts. The pre-edge peak heights of in situ V and Mo K edge

XANES spectra at 298 K (Fig. 5-2) were attributed to the distorted [MO6] (M = V, Mo) building units of the Keggin ion with the metal centers in their highest oxidation states.[14,133] Mo K edge XANES spectra of PVxMo12-x-SBA-15 (x = 1, 2) and V K edge

XANES spectra of PV2Mo10-SBA-15 remained unchanged within the temperature range from 298 K through 473 K. Hence, the Keggin structure appeared to be stable up to 473 K. Between 473-648 K the pre-edge peak height increased with temperature. This indicated

V K edge Mo K edge

absorption

Normalized Normalized

absorption Normalized Normalized 20.20 673 5.54 673 20.15 573 5.52 573 20.10 473 5.5 473 T [K] T [K] 20.05 5.48 373 373 20.00 5.46 Photon energy Photon energy 00 [keV] [keV] Fig. 5-2: (left) in situ V K edge XANES spectra and (right) in situ Mo K edge XANES spectra of

PV2Mo10-SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen in helium in a temperature range between 300 K and 723 K.

structural changes from octahedral to tetrahedral [MOx] (M = V, Mo) units during thermal treatment under catalytic conditions comparable to unsubstituted PMo12-SBA-15 (chapter 7.4).[95,59] Evolution of the normalized pre-edge peak height together with the ion currents of H2O, CO, CO2, acrolein, and acetone during oxidation of propene are shown in Fig. 5-3, (right). In situ Mo K edge FT(χ(k)·k3) (Fig. 5-3, left) indicated that the Keggin structure was intact up to 473 K. Structural changes between 473 K and 648 K as observed in the Mo K edge FT(χ(k)·k3) coincided with the evolution of pre-edge height of V and Mo 3 K edge of PV2Mo10-SBA-15. No structural changes in the FT(χ(k)·k ) of Mo could be determined above 648 K. Moreover, changes in pre-edge heights of V and Mo occurred on the same time scale thereby confirming the incorporation of V centers in the Keggin structure. The onset of catalytic activity and the formation of various selective oxidation

m/e=18 (H2O)

m/e=28 (CO)

m/e=44 (CO2)

m/e=56 (acroleine) )

3 0.06 m/e=58 (acetone) (k)·k

χ 0.04 673

height height height orm. ion current orm. ion FT( 573

0.02 N 473 peak edge orm. pre

T [K] N

373

0 2 4 6 373 473 573 673 R [Ǻ] T [K] pre edge peak height V

pre edge peak height Mo 3 Fig. 5-3: (left) Evolution of Mo K FT(χ(k)·k ) of PV2Mo10-SBA-15 during thermal treatment in 5% propene and 5% oxygen in helium in the temperature range from 303 to 723 K (4 K/min); (right) evolution of normalized ion current of H2O (m/e 18), CO (m/e 28), CO2 (m/e 44), acroleine (m/e

56), and acetone (m/e 58), and normalized pre-edge height of V and Mo K edge of PV2Mo10-SBA- 15 during thermal treatment in 5% propene and 5% oxygen in helium in the temperature range from 303 to 723 K (4 K/min). products coincided with the detected structural changes. Apparently, the catalytically active species formed during thermal activation from the original Keggin structure under reaction conditions. These mainly consisted of V and Mo species centers in a particular tetrahedral coordination. The evolution of the [MoO4]/[MoO6] ratio of PVxMo12-x-SBA-15

(x = 0, 1, 2) based on a linear combination of bulk MoO3 and bulk Na2MoO4 (cf. chapter

7.4) during propene oxidation conditions was shown in Fig. 5-4. A comparison of the structural changes of PVxMo12-x-SBA-15 (x = 0, 1, 2) indicated identical onset temperatures. An increase of tetrahedral [MoO4] units with V substitution was determined.

60 PMo 12-SBA-15

PVMo 11-SBA-15 50 PV2Mo10-SBA-15

] ratio [%] ratio ] 6

30 ]/[MoO

4 20

[MoO 10

0 300 400 500 600 700 T [K]

Fig. 5-4: Quantification of the [MoO4]/[MoO6] ratio of PMo12-SBA-15, PVMo11-SBA-15, and

PV2Mo10-SBA-15 during thermal treatment under propene oxidation conditions.

5.2.1 Local structure in activated PVxMo12-x-SBA-15 (x = 0, 1, 2) and a reference

V2Mo10Ox-SBA-15 under catalytic conditions

Local structure of Mo centers in act. PVxMo12-x-SBA-15 (x = 0, 1, 2)

3 Fig. 5-5 (left) shows the Mo K edge FT(χ(k)·k ) of act. PMo12-SBA-15, act. PVMo11-SBA-

15, and act. PV2Mo10-SBA-15 after thermal treatment under catalytic conditions at 723 K. 3 The Mo K edge FT(χ(k)·k ) were nearly similar for all three PVxMo12-x-SBA-15 (x = 0, 1, 2) and exhibited features similar to that of previously reported dehydrated molybdenum oxides and HPOM supported on SBA-15.[27,59] Minor differences between 3 the three PVxMo12-x-SBA-15 (x = 0, 1, 2) are marked in the Mo K edge FT(χ(k)·k ) and the 3 Mo K edge χ(k)·k . For a more detailed analysis hexagonal MoO3 was used as structural model. Theoretical XAFS phases and amplitudes were calculated for Mo-O and Mo-Mo distances and used for EXAFS refinement. The results of the refinement are shown in Table 5-1. The first coordination sphere of the Mo K edge FT(χ(k)·k3) of as prepared

0.30 PMo12-SBA-15

0.25

0.20 12

PVMo10-SBA-15

)

0.15 3 8

(k)·k

(k)·k χ χ

0.10 4 FT(

0.05 PV2Mo10-SBA-15 0

0.00 -4 4 6 8 10 12 14 16 -1 -0.05 k [Ǻ]

0 1 2 3 4 5 6 R [Ǻ]

3 3 Fig. 5-5: (left) Mo K edge FT(χ(k)·k ) and (right) Mo K edge χ(k)·k of act. PMo12-SBA-15, act.

PVMo11-SBA-15, and act. PV2Mo10-SBA-15 after thermal treatment under propene oxidation conditions at 723 K.

PVxMo12-x-SBA-15 (x = 1, 2) (cf. chapter 4.3) exhibited differences compared to act. 3 PVxMo12-x-SBA-15 (x = 1, 2). The first peak in the FT(χ(k)·k ) originated mainly from the tetrahedral species on the SBA-15 support and could be sufficiently simulated using four Mo-O distances. These four distances sufficiently accounted for the minor amount octahedral [MoO6] species. Confirming the results of the XANES analysis 1st and 2nd 2 2 disorder parameters (1st-σ , 2nd-σ ) were higher for act. PMo12-SBA-15 and indicated a decreasing amount of tetrahedral structural motifs compared to act. PVxMo12-x-SBA-15 (x = 1, 2). In addition, the 4th disorder parameter (4th-σ2) was smaller than the disorder parameters for act. PMo12-SBA-15. This disorder parameter mainly represented the fraction of octahedral [MoO6] species. Hence, the reduced disorder parameter indicated an increasing amount of octahedral structural motifs in act. PMo12-SBA-15 compared to act.

PVxMo12-x-SBA-15 (x = 1, 2). Therefore a structure directing effect of addenda vanadium resulting in an increased concentration of tetrahedral [MoO4] units under propene oxidation conditions was determined. A distinct peak at ~3 Å in the FT(χ(k)·k3) indicated

the formation of dimeric or oligomeric [MoxOy] units on SBA-15. Hence, only isolated tetrahedral [MoO4] units can be excluded as major molybdenum oxide species.[137] The 2 disorder parameters σ of the Mo-Mo distances for act. PVxMo12-x-SBA-15 (x = 0, 1, 2) were nearly identical indicating a comparable degree of oligomerization of Mo species on silica SBA-15 independent of the V substitution in contrast to PMo12 supported on SBA-15 with larger pore radii (chapter 7.4).

Table 5-1: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in act. PVxMo12-x -SBA-15 (x = 0, 1, 2). Experimental parameters were obtained from a refinement of a hexagonal MoO3 model structure (ICSD 75417 [135]) to the experimental -1 Mo K edge XAFS χ(k) of act PVxMo12-x -SBA-15 (x = 0, 1, 2) (k range from 3.4-16.0 Å , R range from 0.9 to 4.0 Å, E0 = -5.2, residuals ~12.3-12.8 Nind = 26, Nfree = 12). Subscript c indicates parameters that were correlated in the refinement.

hex-MoO3 act. PMo12- act. PVMo11- act. PV2Mo11- model SBA-15 SBA-15 SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 2 1.67 1.67 0.0015 1.67 0.0012 1.67 0.0013

Mo-O 2 1.96 1.89 0.0038c 1.89 0.0034c 1.88 0.0034c

Mo-O 1 2.20 2.19 0.0038c 2.18 0.0034c 2.18 0.0034c Mo-O 1 2.38 2.35 0.0011 2.36 0.0014 2.34 0.0017

Mo-Mo 2 3.31 3.49 0.0068c 3.50 0.0066c 3.49 0.0061c

Mo-Mo 2 3.73 3.63 0.0068c 3.63 0.0066c 3.63 0.0061c Mo-Mo 2 4.03 3.73 0.0100 3.75 0.0100 3.75 0.0100

Comparison of the local structure around the Mo centers in act. PV2Mo10-SBA-15

and a reference act. V2Mo10Ox-SBA-15 under catalytic conditions

3 Fig. 5-6 shows the Mo K edge FT(χ(k)·k ) of act. PV2Mo10-SBA-15 and activated

V2Mo10Ox-SBA-15 after thermal treatment under catalytic conditions. Linear combinations of the XANES spectra of Na2MoO4 and MoO3 references were used to determine the amount of tetrahedral [MoO4] and octahedral [MoO6] units in act. PV2Mo10-SBA-15 and act. V2Mo10Ox-SBA-15. Apparently, act. PV2Mo10-SBA-15 consisted of a mixture of tetrahedral [MoO4] and octahedral [MoO6] units in a ratio of 1:1. For act. V2Mo10Ox-SBA-

15 a ratio of 1:3 was found. A comparison of the pseudo radial distribution function of act.

V2Mo10Ox-SBA-15 and act. PV2Mo10-SBA-15 confirmed the results of the XANES 3 analysis. The first peak of the Mo K edge FT(χ(k)·k ) of act. V2Mo10Ox-SBA-15 exhibited 3 differences compared to that of act. PV2Mo10-SBA-15. The first peak in both FT(χ(k)·k ) originated from the tetrahedral and octahedral species on the SBA-15 support and could be sufficiently simulated using four Mo-O distances accounting for the amount of octahedral 2 2 [MoO6] species (Table 5-2). The 1st and 2nd disorder parameters (1st-σ , 2nd-σ ) were

0.08

V2Mo10Ox-SBA-15

)

0.04

3 (k)·k

χ 0.00 PV2Mo10-SBA-15

Intensity FT(

-0.04 MoO3

0 1 2 3 4 5 6 10 20 30 40 50 60 70 R [Ǻ] Diffraction angle 2Ɵ [°]

act. PV2Mo12-SBA-15 act. V2Mo10Ox -SBA-15

3 Fig. 5-6: (left) Mo K edge FT(χ(k)·k ) of activated PV2Mo10-SBA-15 and activated V2Mo10Ox- SBA-15 after thermal treatment in 5% propene and 5% oxygen in helium at 723 K; (right) XRD of as prepared PV2Mo10-SBA-15, as prepared V2Mo10Ox-SBA-15, and simulated MoO3.

higher for act. V2Mo10Ox-SBA-15 and indicated a decreasing concentration of tetrahedral

[ MoO4] units. The third Mo-O distance is considerably shorter than the distance in act. 2 PV2Mo10-SBA-15. In addition, the 4th disorder parameter (4th-σ ) is smaller than the disorder parameter for act. PV2Mo10-SBA-15. This disorder parameter mainly represented the fraction of octahedral MoO6 units. Hence, the reduced disorder parameter indicated an increasing amount of octahedral structural motifs in act. V2Mo10Ox-SBA-15 compared to 3 act. PV2Mo10-SBA-15. Furthermore, the amplitude in the Mo K edge FT(χ(k)·k ) of act.

V2Mo10Ox-SBA-15 at higher Mo-Mo shells resembled the shape of α-MoO3. The Mo-Mo distance at ~3.3 Ǻ is characteristic for α-MoO3. These results confirmed the existence of crystalline α-MoO3 which was also identified by XRD before thermal treatment under catalytic conditions (Fig. 5-6, right). Estimating the amount of α-MoO3 from the amplitude

at ~3.3 Ǻ (not phase corrected) in the pseudo radial distribution function yielded an amount of about ~20%.

Table 5-2: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in act. PV2Mo10-SBA-15 and act. V2Mo10Ox-SBA-15. Experimental parameters were obtained from a refinement of a hexagonal MoO3 model structure (ICSD 75417 [135]) to the -1 experimental Mo K edge XAFS χ(k) of act. PV2Mo10-SBA-15 (k range from 3.6-16.0 Å , R range from 0.9 to 4.0 Å, E0v= ~ -5.2, residual ~12.8 Nind = 27, Nfree =12) and act. V2Mo10Ox-SBA-15 (k -1 range from 3.6-16.0 Å , R range from 0.9 to 4.0 Å, E0 = ~ -0.4, residual ~ 10.9, Nind = 26, Nfree =14). Subscript c indicates parameters that were correlated in the refinement.

Type act. PV2Mo11-SBA-15 Type act. V2Mo10Ox-SBA-15 N R(Ǻ) σ2(Ǻ2) N R(Ǻ) σ2(Ǻ2) Mo-O 2 1.67 0.0013 Mo-O 2 1.68 0.0023

Mo-O 2 1.88 0.0034c Mo-O 2 1.90c 0.0042c

Mo-O 1 2.18 0.0034c Mo-O 1 2.11c 0.0042c Mo-O 1 2.34 0.0017 Mo-O 1 2.34 0.0001

Mo-Mo - - - Mo-Mo 1 3.14 0.0061c

Mo-Mo 2 3.49 0.0061c Mo-Mo 1 3.28 0.0061c

Mo-Mo 2 3.63 0.0061c Mo-Mo 2 3.71 0.0057 Mo-Mo 2 3.75 0.0100 Mo-Mo 2 3.92 0.0113

Comparison of the local structure around V centers in act. PVxMo12-x-SBA-15

(x = 0, 1, 2) and a reference act. V2Mo10Ox-SBA-15 under catalytic conditions

The evolution of the local structure around V centers in the supported catalysts and reference materials differed from that of the Mo centers. Fig. 5-7 shows the V K edge 3 FT(χ(k)·k ) of act. PV2Mo10-SBA-15 (left) and act. V2Mo10Ox-SBA-15 (right). The amplitudes at distances between 3-4 Ǻ indicated different scattering atoms. Single scattering paths of a Na2MoO4 structure (ICSD 24312 [136]) were used for the EXAFS refinement of act. V2Mo10Ox-SBA-15 and act. PV2Mo10-SBA-15.The theoretical model for act. PV2Mo10-SBA-15 based on Na2Mo2O7 with replaced Mo atoms with V atoms per formula unit. The theoretical model structure for act. PV2Mo10-SBA-15 based on the

Na2Mo2O7 structure with two replaced Mo atoms by V atoms per formula unit. Results of the refinements for the V K edge FT(χ(k)·k3) are given in Table 5-3. The distances between 2 1-2 Ǻ corresponded to a tetrahedral [VO4] unit. Distances R and disorder parameters σ were nearly identical for act. PV2Mo10-SBA-15 and act. V2Mo10Ox-SBA-15 (Fig. 5-7). The first Mo coordination sphere corresponded to a mixture of octahedral and

Experiment 0.04 Experiment

Theory Theory

)

0.02 3

) 0.02

(k)·k

(k)·k χ

χ 0.00 0.00

FT( FT( -0.02 -0.02 act. PV2Mo10-SBA-15 act. V2Mo10Ox-SBA-15 0 1 2 3 4 5 6 0 1 2 3 4 5 6 R [Ǻ] R [Ǻ]

3 Fig. 5-7: V K edge FT(χ(k)·k ) of (left) act. PV2Mo10-SBA-15 and (right) act. V2Mo10Ox-SBA-15 after thermal treatment in 5% propene and 5% oxygen in helium at 723 K.

tetrahedral [MoOx] species. In contrast to the first Mo-O peak with six individual Mo-O distances, the first V-O peak could be sufficiently simulated using two V-O distances. The two V-O distances sufficiently accounted for the tetrahedral [VO4] species. Silica atoms from the support were found at a distance of ~2.55 Ǻ. Additionally, a V-Mo distance was 3 identified in the V K edge FT(χ(k)·k ) of act. PV2Mo10-SBA-15. V-O and V-V distances in act. V2Mo10Ox-SBA-15 were very similar to those in dehydrated VxOy-SBA-15 synthesized with a butylammonium decavanadate precursor.[95] Assuming only V-V distances resulted in a sufficient agreement between experimental and theoretical spectra in contrast to act. PV2Mo10-SBA-15. This indicated that [VOx] and

[MoOx] species were not present in close vicinity to each other. The results of Mo K and V

K edge analysis of act. PV2Mo10-SBA-15 and act. V2Mo10Ox-SBA-15 confirmed that only act. PV2Mo10-SBA-15 contained supported V-O-Mo mixed oxides structural motifs forming under catalytic conditions. Active sites of selective oxidation catalysts often consist of multiple metal atoms.[137] Synthesis routes of supported ternary oxides with different metal oxide precursors rarely have been reported. Vanadium substituted Keggin ions enabled the synthesis of connected

[VOx] and [MoOx] species not readily available from physically mixed precursors. Apparently, the proximity of vanadium and molybdenum in the Keggin precursors is a prerequisite for obtaining connected metal oxide species on a support material.

Table 5-3: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the V atoms in act. PV2Mo10-SBA-15 and act. V2Mo10Ox-SBA-15. Experimental paramters were 2- obtained from a refinement of a modified [Mo2O7] model system (ICSD 24312 [136]) were Mo is replaced by V and Si additional added compared to the experimental V K edge XAFS χ(k) of act. -1 PV2Mo10-SBA-15 (k range from 3.0-10.0 Å , R range from 0.9 to 3.8 Å, E0 = ~8.8, residual ~ 6.0 -1 Nind = 13, Nfree =8) and act. V2Mo10Ox -SBA-15 (k range from 3.0-10.0 Å , R range from 0.9 to 3.8

Å, E0 = ~8.8, residual ~ 11.3, Nind = 13, Nfree =8). Subscript c indicates parameters that were correlated and f that were fixed in the refinement.

Type act. PV2Mo11-SBA-15 Type act. V2Mo10Ox-SBA-15 N R(Ǻ) σ2(Ǻ2) N R(Ǻ) σ2(Ǻ2) V-O 2 1.83 0.0183 V-O 2 1.82 0.0160

V-O 2 1.83c 0.0183c V-O 2 1.82c 0.0160c

V-Si 1 2.54 0.0172 V-Si 1 2.55 0.0094

V-O 1 2.91 0.005f V-O 1 2.95 0.0046f

V-O 1 3.10 0.0246 V-V 1 3.30 0.0089

V-Mo 1 3.60 0.0221 V-V 1 3.58 0.0089c

5.2.2 Local structure of P in activated PV2Mo10SBA-15 under catalytic conditions

31 Fig. 5-8 shows P-MAS-NMR measurements of as-prepared and act. PV2Mo10-SBA-15 in 31 comparison to as-prepared and activated reference H3PO4-SBA-15. The P MAS NMR spectrum of PV2Mo10-SBA-15 resembled that of bulk PVMo11.[138] This confirmed that the majority of P centers was located in Keggin ions supported on SBA-15.[139,140] The peak in the spectrum of H3PO4-SBA-15 at 0.8 ppm could be assigned to molecular H3PO4.

In the spectrum of act. H3PO4-SBA-15 four pronounced peaks can be seen at chemical shifts of 0.8, -10.8, 22.8, and -35.9 ppm. Zhi-qiang Zhang et al. reported similar results for

SiO2 impregnated with H3PO4.[140,141] Accordingly, the peak at 0.1 ppm is characteristic for H3PO4, while the peaks at -10.8 and -22.8 ppm were attributed to terminal and internal

act. H3PO4-SBA-15 intensity

H3PO4-SBA-15

orm. n

100 80 60 40 20 0 -20 -40 -60 -80 -100

 (ppm)

act. PV2Mo10-SBA-15 tensity

PV2Mo10-SBA-15 norm. norm. in

100 80 60 40 20 0 -20 -40 -60 -80 -100  (ppm)

31 Fig. 5-8: P MAS NMR spectra of asprepared H3PO4-SBA-15, PV2Mo10-SBA-15, and thermal treated under catalytic conditions (5% propene and 5% oxygen in He) at 723 K act. H3PO4-SBA-

15 and act. PV2Mo10-SBA-15.

phosphate groups of condensed phosphates, respectively.[141] Krawietz et al. assigned the peak at -35.9 ppm to silicon hydrogen tripolyposphate (SiHP3O10, -35 ppm).[142] For

H3[PMo12O40] supported on ZrO2 (PMo12-ZrO2) Devassy et al. investigated the nature of phosphorous species depending on Keggin loading and calcination temperature.[54] 31 PMo12-ZrO2 exhibited a comparable broadening of the peaks in the P MAS NMR spectrum with increasing calcination temperature. Decomposition of the HPOM to the oxide species was observed at temperatures above 723 K. Thermal stability of

H3[PW12O40] supported on ZrO2 (PW12-ZrO2) was investigated by López-Salinas et al..

The structural behaviour of PW12-ZrO2 during calcination was comparable to that of

PMo12-ZrO2. PW12-ZrO2 decomposed at temperatures above 773 K to form the corresponding supported oxides.[56] The authors assigned an additional peak at -30 ppm to phosphorous oxides exhibiting P-O-P motifs. In the 31P MAS NMR spectra of act.

PV2Mo10-SBA-15 studied here, a broad resonance indicated structural rearrangement and a

partial decomposition of the Keggin ions during thermal treatment under catalytic 31 conditions. Moreover, the P MAS NMR spectra of act. PV2Mo10-SBA-15 resembled that of a VPO-SBA-15 sample treated under oxidative and reductive conditions.[143] A broad resonance observed for the VPO sample between -12 and -38 ppm was attributed to various vanadyl orthophosphates phases (-8.4 to 21.2 ppm) and phosphorus bound to the SBA-15 support (ca. -38 ppm). Comparable structural motifs can be assumed for act.

PV2Mo10-SBA-15. However, formation of phosphorous oxide or SiHP3O10 exhibiting linked P-O-P structures could be excluded. In total, the 31P MAS NMR results indicated a variety of structural motifs in the activated samples studied here. Apparently, phosphorus remained connected to the molybdate- and/or vanadate-species of the [(Mo,V)Ox] units during propene oxidation conditions.

5.2.3 Structure directing effects of vanadium and the support material on the structure of

activated PV2Mo10-SBA-15 under catalytic conditions

Characteristic differences were revealed by comparing the structural evolution of bulk HPOM during thermal treatment to that of supported HPOM. In bulk HPOM the Keggin ion partially decomposes under catalytic conditions to form a lacunary Keggin anion.[13] In this process Mo cations migrate on extra Keggin sites while remaining coordinated to the resulting lacunary Keggin anion.[13] Driving force for the formation of lacunary Keggin anions may be the relaxation of the Keggin structure at elevated temperature upon removal of structural water. Eventually, this leads to the formation of more extended oxide structures. These structural changes at temperatures above 573 K are accompanied by reduction of the metal centers.[16,13] Vanadium incorprated in bulk HPOM acts as a structural promoter facilitating the formation of the active (Mo, V) oxide phase under catalytic conditions. The incorporated V centers result in a pronounced destabilization and accelerated decomposition of the Keggin ion at elevated temperatures.[14,117] The structural characteristics of model systems like MoOx-SBA-15 and VOx-SBA-15 depend on their hydration states.[144,95] A comparable effect could be responsible for the structural evolution of HPOM supported on SBA-15. Adsorbed water and silanol groups from the support may possess a structure stabilizing effect on the Keggin ion. This effect would be comparable to that of water of crystallization and constitutional water in bulk

HPOM under ambient conditions.[117] Vansant et al. reported that amorphous silica showed dehydroxylation of silanol groups between 473-673 K resulting in a decrease of a silanol density from 4.6 OH/nm2 (473 K) to 2.3 OH/nm2 (673 K).[145] TG measurement of PVMo11-SBA-15 showed a mass loss of about 2 wt.% between 473 and 673 K which correlated with the temperature range of structural rearrangement of PVMo11-SBA-15 under catalytic conditions (c.f. Chapter 8.2; Fig. 8-1). Thus, desorption of water and dehydroxylation of silanol groups may be responsible for the formation of act. PVxMo12-x-SBA-15 (x = 0, 1, 2). SBA-15 seemed to possess a directing effect on the formation of activated (Mo, V, P)Ox structures depending on the treatment conditions (i.e. temperature, gas composition). Apparently, the thermal stability of Keggin ions supported on SBA-15 was significantly decreased. While vanadium had a minor influence on the thermal stability, the interaction with the support material appeared to be more important. Nevertheless, vanadium still had a distinct structure directing effect to form V-O-Mo mixed structures under catalytic conditions. The presence of tetrahedral

[VO4] species lead to an increasing ratio of tetrahedral [MoO4] to octahedral [MoO6] species. Compared to SBA-15 other support materials exhibit different structure directing effects depending on the acidity of the surface.[25,146,147] For instance, mainly isolated

[MoO4] units existed on an alkaline MgO support in agreement with the behaviour of Mo oxides in alkaline solution.[137] Here, the acidic surface of silica SBA-15 resulted in mainly linked M-O-M (M = Mo, V) species again corresponding to the behaviour of vanadates and molybdates in acidic solutions.[37,148]

5.3 Functional characterization of PVxMo12-x-SBA-15 (x = 0, 1, 2)

5.3.1 Reducibility

Fig. 5-9 shows the H2 TPR profiles of PVxMo12-x-SBA-15 (x = 0, 1, 2). The resulted

H2 TPR profiles revealed one sharp (~800 K) and a very broad (873- 973 K) reduction peak. The shapes of the H2 TPR profiles were nearly identical, just the reduction temperatures slightly increased with the vanadium content from 790 K for PMo12-SBA-15 to 808 K for PV2Mo10-SBA-15. The H2 TPR profiles at ~800 K were comparable to molybdenum oxide supported on SBA-15 with Mo loadings between 9.5 wt.% and 13.3

PMo12-SBA-15 800 K PVMo -SBA-15 11 790 K 808 K

PV2Mo10-SBA-15

umption

cons

2 2 H

373 473 573 673 773 873 973 T [K]

Fig. 5-9: Temperature.programmed reduction (H2-TPR) of PMo12-SBA-15, PVMo11-SBA-15, and -1 PV2Mo10-SBA-15 measured at a heating rate of 8 Kmin 5% H2 in Ar.

wt.%.[149,150] Lou et al. assigned the sharp reduction peak from oligomeric MoOx species or small MoOx clusters and the broaded signal above ~800 K to the reduction of monomeric MoOx species.[150] Vanadium oxide supported on SBA-15 with a loading between 1.0 wt.% V and 4.5 wt.% V reduced between 769 K and 799 K depended on the dispersion degree.[151] Hence, the reducibility were comparable to molybdenum oxides supported on SBA-15. A comparison of vanadium and molybdenum oxides supported on

Al2O3 with similar metal surface density showed that the temperature of the maximum of

H2-consumption were slightly higher (~10 K) for the supported vanadium oxides.[152] A comparable intrinsic effect of the metals could be responsible for the shift to higher reduction temperatures with higher vanadium loading. However, no significant changes in the reducibility depending on the vanadium substitution degree were detected.

5.3.2 Catalytic performance

Reaction rates and selectivities of PMo12-SBA-15, PVMo11-SBA-15, PV2Mo10-SBA-15,

V2Mo10Ox-SBA-15, and bulk PV2Mo10 in propene oxidation at 723 K are shown in Fig.

5-10. Reaction rates for PVxMo12-x-SBA-15 (x = 0, 1, 2) were calculated for similar propene oxidation conditions (~ 14-16% propene conversion). The propene conversion for bulk PV2Mo10 (~ 3%) was lower due to the strongly decreased catalytic activity. Adjusting

to similar propene oxidation conditions for the low active sample would lead to an large volume and thermal effects. Hence, comparing of the catalytic performance of PVxMo12-x- SBA-15 (x = 0, 1, 2) to that of the low active sample needs to be done carefully.

Reaction rates for PVxMo12-x-SBA-15 (x = 0, 1, 2) were similar and independent of the degree of vanadium substitution. Selectivities for CO increased at the expense of acetaldehyd with higher degree of vanadium substitution. The results of the catalytic performance of bulk PVxMo12-x (x = 0, 1, 2) showed strong increased reaction

acrylic acid acetone CO

acetic acid propionaldehyde CO

2 ] 1 acrolein acetaldehyde - s

100 70

(Mo) 1 60 - 80 g 50 60 40

40 30

Selectivity [%] Selectivity 20 20

10 [µmol(propene) rate n

0 0

a b c d e Reactio

-1 -1 Fig. 5-10: Reaction rate (µmol(propene)g (Mo)s ) and selectivity of (a) PMo12-SBA-15, (b)

PVMo11-SBA-15, (c) PV2Mo10-SBA-15, (d) V2Mo10Ox-SBA-15, and (e) bulk PV2Mo10 in 5% propene and 5% oxygen in He at 723 K. rates for vanadium substituted bulk HPOM. Selectivities towards oxygenates, especially acetaldehyd decreased at the expense of CO with increased vanadium substitution degree.

PV 2Mo10 was chosen as bulk HPOM for comparing the product distribution (i.e. acrylic acid, acetic acid, acrolein, acetone, propionaldehyde, acetaldehyde, CO, and CO2) to

PV2Mo10-SBA-15. While, PV2Mo10 showed a slightly increased selectivity towards acrolein, PV2Mo10-SBA-15 exhibited an increased selectivity towards acetic acid. Total oxidation products CO and CO2 amounted to about ~55% in the resulting oxidation products. Conversely, the reaction rates of PV2Mo10 and PV2Mo10-SBA-15 exhibited considerable differences. The catalytic activity of PV2Mo10-SBA-15 was four times higher than that of PV2Mo10. Apparently, higher dispersion and an improved surface to bulk ratio of Keggin ions resulted in a much increased activity at comparable selectivity. Structural

analysis of act. PVxMo12-x-SBA-15 (x = 0, 1, 2) revealed an increased concentration of tetrahedral [MoO4] units at comparable degree of oligomerization. Apparently, the additional [VOx] species in act. PVxMo12-x-SBA-15 (x = 0, 1, 2) lead to new multifunctional active sites, resulting in a different product distribution without influence to the reaction rates.

In contrast to the HPOM samples, V2Mo10Ox-SBA-15 showed a decreasing activity and a different product distribution compared to PV2Mo10-SBA-15. While, the amount of total oxidation products in the gas phase was considerably lower, an increasing selectivity to acetaldehyde was determined. The structural analysis indicated that act. V2Mo10Ox-SBA- 15 possessed an increased amount of higher oligomerized Mo species and a decreased content of tetrahedral [MoO4] units. It was shown earlier, that higher oligomerized V and Mo species showed an increased selectivity towards oxidations products.[137,153]

Additionally, the majority of [VOx] and [MoOx] species in act. V2Mo10Ox-SBA-15 did not seem to be directly connected to each other. Local separation of the [VOx] and [MoOx] species may be responsible for the increased concentration of acetaldehyde, which is mainly formed by vanadium based catalysts in contrast to molybdenum based catalysts.[1,31,50] Apparently, the new multifunctional active site consisting of connected

[VO4] and [MoOx] units lead to an increased amount of total oxidation products for act.

PV2Mo10-SBA-15 in contrast to not connected [VO4] and [MoOx] units in act. V2Mo10Ox-

SBA-15. Furthermore, availability of dimeric or oligomeric [(V,Mo)Ox] units increased the selectivity towards oxygenates in contrast to isolated [MoO4] units [47,50]. Hence, connected [VO4] and [MoOx] units and the general degree of oligomerization of

[(V,Mo)Ox] units influenced the catalytic activity and selectivity towards propen oxidation.

5.3.3 Influence of phosphorus species on catalytic activity

Phosphorus containing catalysts (i.e. VPO, FePO, MoPO) play a crucial role as oxidation catalysts.[128] Adding small amounts of phosphoric acid to the feed showed positive effects on long-term stability and catalytic performance of FePO catalysts during ODH of isobutyric acid into methacrylic acid. The phosphorus source was needed to maintain a constant P/Fe ratio at the surface of the catalyst.[129] VPO catalysts showed migration of phosphorus species to the surface and a decreasing amount of phosphorus in the catalyst

during water vapor treatment. The excess of phosphorus on the surface suppressed oxidation of VPO catalysts and hindered formation of active sites for oxidation reactions. Subsequently, hydrolysis of P-O-P or P-O-V groups resulted in a removal of phosphate groups on the surface and an increasing activity.[128,130] Moreover, adding V and P to

MoOx based catalysts for ODH of ethane afforded an increasing selectivity and conversion towards ethane. Haddad et al. suggested synergistic effects between structurally related oxides like (V,Mo)5O14 and (V, Mo)PO phases to be responsible for the enhancened catalytic performance.[154] Here, the formation of water as byproduct during oxidation of propene may have favored the migration of phosphorus species under catalytic conditions.

The different surface to bulk ratios of PV2Mo10 and PV2Mo10-SBA-15 could lead to a different migration and hydrolysis of phosphate groups in the materials. The available

Keggins in PV2Mo10-SBA-15 were located on the surface of the support material.

Therefore, an enrichment of phosphate groups was not possible for PV2Mo10-SBA-15. An enrichment of phosphate groups on the surface of bulk PV2Mo10 would results in a higher P/M (M = Mo, V) with a possible influence on catalytic activity and selectivity. However, the comparable selectivity of bulk PV2Mo10 and PV2Mo10-SBA-15 (Fig. 5-10) was indicative of similar active centers despite different P/M (M = Mo, V) ratios. Therefore, the increased catalytic activity of PV2Mo10-SBA-15 was attributed to a higher dispersion and an improved surface to bulk ratio of supported Keggin ions.

5.4 Summary

The structural evolution of PVxMo12-x-SBA-15 (x = 0, 1, 2) and a mixture of V and Mo oxides supported on SBA-15 during propene oxidation conditions was examined by in situ X-ray absorption spectroscopy at the Mo K and V K edge. Additionally, 31P MAS NMR measurements of supported PV2Mo10-SBA-15 and H3PO4-SBA-15 after catalytic reaction were performed. During thermal treatment under propene oxidation conditions PVxMo12-x-

SBA-15 (x = 0, 1, 2) formed a mixture of mainly tetrahedral [MoOx] and [VOx] units. Changes in the local structure around the V centers coincided with structural changes of the Mo centers and the onset of catalytic activity. The concentration of tetrahedral [MoO4] units correlated with the degree of vanadium substitution without affecting to the degree of oligomerization of the [MoxOy] and [VxOy] species. Apparently, the mainly tetrahedral

[MoOx] and [VOx] units were in close vicinity and able to interacted under catalytic conditions. The new multifunctional active site consisting of connected [VO4] and [MoOx] units lead to an increased amount of total oxidation products without influencing the reaction rate. Conversely, structural analysis of activated reference V2Mo10Ox-SBA-15 synthesized with individual V and Mo precursors indicated that [VOx] and [MoOx] species were mostly separated from each other on the surface of SBA-15. Moreover, activated

V2Mo10Ox-SBA-15 possessed an increased amount of higher oligomerized [MoxOy] species and a decreased content of tetrahedral [MoO4] units. This may explain the observed increased selectivity towards partial oxidations products. The structural environment of phosphorus in PV2Mo10-SBA-15 under catalytic conditions corresponded to a mixture of various species. Phosphorus was linked to both the SBA-15 support via P-O-Si bonds and to the Mo or V centers of the [MoOx] or [VOx] units. In total, supported vanadium substituted Keggin ions are suitable precursors to synthesize connected [VOx] and [MoOx] species on SBA-15. Apparently, the proximity of vanadium and molybdenum in the Keggin precursors a prerequisite for obtaining connected metal oxide species.

6 Characterization of PWxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions

Keggin type H4[PVMo11O40] has been reported to exhibit a pronounced interaction effect with SBA-15 as support material.[14] This effect resulted in a further decreased thermal stability of the supported Keggin ions compared to the bulk materials. PVMo11-SBA-15 formed a mixture of tetrahedrally and octahedrally coordinated [MoO4] and [MoO6] units during propene oxidation.[14] The structural evolution and role of tungsten in PWxMo12-x- SBA-15 (x = 1, 2) during propene oxidation were not part of previous investigations. Therefore, a first structural and functional characterization were necessary to elucidated structure activity of supported molybdenum oxide based model catalysts. In this chapter in situ X-ray absorption spectroscopy investigations at the LIII-LI edges of PWxMo12-x-SBA- 15 (x = 1, 2) during propene oxidation conditions were performed. Correlations between structural evolution of [MoOx] and [WOx] units and performance under catalytic conditions will be described. Additionally, the obtained structures and catalytic performances were compared to a suitable supported reference material.

6.1 Experimental

6.1.1 Sample Characterization

X-ray absorption spectroscopy (XAS)

Transmission XAS experiments were performed at the Mo K edge (19.999 keV) at beamline at X and W LIII-LI edges (10.204-12.098 keV) at beamline C at the Hamburg Synchrotron Radiation Laboratory, HASYLAB. Using a Si(311) double crystal monochromator at Beamline X for the Mo K edge and a Si(111) double crystal monochromator at Beamline C for the W LIII-LI edges. In situ experiments were conducted in a flow reactor at atmospheric pressure (5 vol% oxygen in He, total flow ~30 ml/min, temperature range from 303 to 723 K, heating rate 4 K/min). The gas phase composition at the cell outlet was continuously monitored using a non-calibrated mass spectrometer in a multiple ion detection mode (Omnistar from Pfeiffer).

X-ray absorption fine structure (XAFS) analysis was performed using the software package WinXAS v3.2..[91] Background subtraction and normalization were carried out by fitting linear polynomials and 3rd degree polynomials to the pre-edge and to the post- edge region of an absorption spectrum, respectively. The extended X-ray absorption fine structure (EXAFS) χ(k) was extracted by using cubic splines to obtain a smooth atomic 3 background μ0(k). The FT(χ(k)·k ), often referred to as pseudo radial distribution function, was calculated by Fourier transforming the k3-weighted experimental χ(k) function, multiplied by a Bessel window, into the R space. EXAFS data analysis was performed using theoretical backscattering phases and amplitudes calculated with the ab-initio multiple-scattering code FEFF7.[92] Structural data employed in the analyses were taken from the Inorganic Crystal Structure Database (ICSD).

Single scattering paths in the hexagonal MoO3 model structure (ICSD 75417 [135]) and a modified H3[PMo12O40] structure (ICSD 209 [14,93]) were calculated up to 6.0 Å with a lower limit of 4.0% in amplitude with respect to the strongest backscattering path. EXAFS refinements were performed in R space simultaneously to magnitude and imaginary part of a Fourier transformed k3-weighted and k1-weighted experimental χ(k) using the standard EXAFS formula.[94] This procedure reduces the correlation between the various XAFS fitting parameters. Structural parameters allowed to vary in the refinement were (i) disorder parameter σ2 of selected single-scattering paths assuming a symmetrical pair- distribution function and (ii) distances of selected single-scattering paths. Detailed information about the fitting procedure are described in chapter 3.2.

Powder X-ray diffraction (XRD)

XRD measurements were conducted on an X’Pert PRO MPD diffractometer (Panalytical, θ-θ geometry), using Cu K alpha radiation and a solid-state multi-channel PIXcel detector. Wide-angle scans (5-90° 2θ, variable slits) were collected in reflection mode using a silicon sample holder. Small-angle scans (0.4-6.0° 2θ, fixed slits) were measured in transmission mode with the sample spread between two layers of Kapton foil.

Temperature programmed reduction

PW2Mo10-SBA-15 were used.

Catalytic testing - selective propene oxidation

6.1.2 Sample preparation

PWxMo12-x (x=1,2) supported SBA-15 were prepared as described in chapter 4.1.

A reference material (denoted as W2Mo10Ox-SBA-15) was prepared as follows. 241.0 mg

(NH4)6Mo7O24·4H2O and 29.3 mg (NH4)6W12O39·xH2O were dissolved in water and were deposited via incipient wetness on 1 g SBA-15 to obtain metal loading of 10 wt.% Mo and 3.8 wt.% W. The sample was dried for 18 h at room temperature and calcined for 3 h at 773 K.

6.2 Structural evolution of PWxMo12-x-SBA-15 (x = 1, 2) under catalytic conditions

In situ XANES analysis

PWxMo12-x-SBA-15 (x = 1, 2) was investigated by in situ XAS in propene oxidation conditions. Fig. 6-1 shows the evolution of molybdenum Mo K edge XANES spectra of

PW2Mo10-SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen. Mo K edge XANES spectra of PWxMo12-x-SBA-15 (x = 1, 2) remained unchanged

within the temperature range from 298 K through 473 K comparable to PVxMo12-x-SBA-15

absorption Normalized Normalized

673 20.20 573 20.15 473 20.10 T [K] 373 20.05 20.00 Photon energy [keV] Fig. 6-1: in situ Mo K edge XANES spectra of PW Mo -SBA-15 during temperature-programmed 2 10 treatment in 5% propene and 5% oxygen in helium in a temperature range between 300 K and 723 K.

(x = 0, 1, 2) (cf. chapter 5.2). Hence, the Keggin ion appeared to be stable up to 473 K independent of the degree of substitution. Between 473-648 K the pre-edge peak height increased during thermal treatment under catalytic conditions.[15,23] This increasing pre- edge peak corresponded to a structural rearrangement to tetrahedral [MoO4] species. Linear combinations of the XANES spectra of MoO3 and bulk Na2MoO4 references were used to determine the amount of tetrahedral [MoO4] and octahedral [MoO6] units of PWxMo12-x- SBA-15 (x = 0, 1, 2) during propene oxidation conditions (cf. chapter 7.4). The evolution of the tetrahedral [MoO4] to octahedral [MoO6] ratio of PWxMo12-x-SBA-15 (x = 0, 1, 2) during propene oxidation conditions was shown in Fig. 6-2. In contrast to PVxMo12-x- SBA-15 (x = 1, 2) the W substitution lead to decreased concentration of tetrahedral

[MoO4]

60 PMo12-SBA-15

PWMo 11-SBA-15 50 PW2Mo10-SBA-15

] ratio [%] ratio ] 6

30 ]/[MoO 4 20

[MoO 10

0 300 400 500 600 700 T [K]

Fig. 6-2: Quantification of the [MoO4]/[MoO6] ratio of PMo12-SBA-15, PWMo11-SBA-15, and

PW2Mo10-SBA-15 during thermal treatment under propene oxidation conditions.

species. Hence, tungsten substitution lead to mostly octahedral [MoO6] species resulting during propene oxidation conditions. Fig. 6-3 depicts the W LIII and LI edge XANES spectra of PWxMo12-x-SBA-15 (x = 1, 2) during temperature-programmed treatment in 5% propene and 5% oxygen. Compared to the onset temperature of 473 K of the structural rearrangement of the [VOx] and [MoOx] units for PVxMo12-x-SBA-15 (x = 1, 2) (c.f. 5.2), a delayed structural change of the initial octahedral [WO6] units was observed. The onset of structural changes increased to 550 K for PW2Mo12-SBA-15 and to 623 K for PWMo11-

SBA-15. The X-ray absorption W LIII edge corresponds to electron transitions from 2p3/2

absorption

Normalized Normalized

absorption Normalized Normalized 673 10.35 673 573 10.30 573 12.35 473 10.25 473 12.25 T [K] 373 10.20 T [K] 373 12.15 10.15 Photon energy 12.05 Photon energy [keV] [keV]

Fig. 6-3: in situ W LIII edge (left) and W LI edge (right) XANES spectra of PW2Mo10-SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen in helium in a temperature range between 300 K and 723 K. orbital to vacant 5d orbitals and to the vacuum level.[155] The contribution of the possible p-s transitions is ca. 50 times weaker.[156] Hence, especially the white line W LIII edge reflects the electronic state of the vacant states of the absorbing atoms. The white line in the W LIII edge is assigned to the 5d orbital split by the ligand field. The orbital split by an

(a)

absorption d-orbital Normalized Normalized splitting

(b) ives

eg 5d

t2g 2nd derivat 2nd 2p3/2

10.18 10.20 10.22 10.24 Photon energy [keV]

Fig. 6-4: (a) W LIII edge XANES spectra of PW2Mo10-SBA-15 and (b) 2nd derivates of W LIII edge XANES spectra of PW2Mo10-SBA-15; ( ) experiment, ( ) fitting function, and ( )fitting peaks.

octahedral ligand field is stronger than in a tetrahedral ligand field.[157,158] Therefore, an analysis of the white line was suitable to elucidate the electronic structure of the possible structural motifs. Fig. 6-4 shows an example for a white line analysis. The 2nd derivates of the W LIII edge spectra resulted in a splitted peak representing the electron transitions from

2p3/2 to split 5d states (t2g and eg orbitals). The difference in energy position of the splitted peaks corresponds to the energy difference between eg and t2g. For elucidating the energy difference, the 2nd derivates of two Lorentz functions were used to simulated the 2nd derivate of the W LIII edge spectra. The resulted positions of the two minima were used to calculate the energy difference.

W LI edge XANES spectra represent the transition from the 2s orbital and have "K edge character". Hence, the W LI edge XANES spectra may be interpreted comparable to a K edge spectrum. A change in the pre edge peak height may correspond to a structural rearrangement. Fig. 6-5 shows the results of the analysis of W LIII and LI edge XANES spectra meausred during propene oxidation conditions. XAS analysis at W LI edge for

PWMo11-SBA-15 was hardly feasible due to the low content of W (~1.8 wt.%) beside Mo

(~10 wt.%). The structural changes of the [WO6] in PVxMo12-x-SBA-15 (x = 1, 2) were delayed compared to [VOx] units in PVxMo12-x-SBA-15 (x = 1, 2) during propene oxidation conditions (cf. chapter 5.2). [WO6] species in PWMo11-SBA-15 and in

PW2Mo10-SBA-15 changed their local structure above 620 K and 560 K, respectively. This

0.400 3.2 PWMo11-SBA-15 PW2Mo10-SBA-15 0.395

3.1 0.390 3.0 0.385 2.9 0.380 2.8 0.375

2.7

pre edge peak height peak edge pre

0.370 I

( ) Energy Gap [eV] Gap ( Energy ) 2.6 0.365 2.5 0.360 L ( W ) 300 400 500 600 700 T [K]

Fig. 6-5: (square) Evolution of the energy gap of PWMo11-SBA-15 and PW2Mo10-SBA-15 during thermal treatment under propene oxidation conditions; (cycle) evolution of the pre edge height of W

LI edge XANES spectra of PW2Mo10-SBA-15 during thermal treatment under propene oxidation conditions.

temperatures corresponded to the temperatures where the major structural rearrangement of the [MoOx] species were finished. For elucidating, the type of structural rearrangement, a detailed analysis of the 2nd derivates of the W LIII edge XANES spectra was necessary.

Fig. 6-6 shows the 2nd derivates of the W LIII edge XANES spectra of PW2Mo10-SBA-15 before and after propene oxidation conditions and during propene oxidation at 723 K.

Additionally, the 2nd derivates of the W LIII edge XANES spectra of WO3 and Na2WO4 were used as references. Unexpectedly the 2nd derivates of the W LIII edge XANES spectra of PW2Mo10-SBA-15 after propene oxidation conditions exhibited an increased energy gap in contrast to the derivates of the W LIII edge XANES spectra of PW2Mo10- SBA-15 during propene oxidation conditions at 723 K. An analysis of the ratio of the areas of the Lorentz functions used for the refinement revealed that the ligand field under propene oxidation conditions was octahedral. Comparable result was obtained for

PWMo11-SBA-15. An identification of the type of ligand field of the unknown structure motif was possible, because the X-ray absorption intensity is t2g:eg = 3:2 for an octahedral

[WO6] and eg:t2g = 2:3 for a tetrahedral [WO4] unit.[157] Table 6-1 summarizes the results of the detailed W LIII edge analysis. Hence, the initial octahedral [WO6] units persisted

PW2Mo10SBA-15

WO3 before

at 723 K

Na2WO4

Derivatives

nd after

2 2nd

10.19 10.20 10.21 10.22 10.23 10.24 10.19 10.20 10.21 10.22 10.23 10.24 Photon energy [keV] Photon energy [keV]

Fig. 6-6: Second derivates of W LIII edge XANES spectra of (left) WO3 and Na2WO4 and (right)

PW2Mo10-SBA-15 before, at 723 K, and after propene oxidation; experiment ( ), fitting function ( ), and fitting peaks ( ).

during propene oxidation conditions in contrast to the [MoO4] and [MoO6] units in

PWxMo12-x-SBA-15 (x = 0, 1, 2) and vanadium substituted PVxMo12-x-SBA-15 (x = 1, 2)

(cf. chapter 5.2). Nevertheless structural rearrangements for PWMo11-SBA-15 (620-723 K) and PW2Mo10-SBA-15 (560-723K) were identified and could be assigned to a reversible distortion of the octahedral [WO6] units under propene oxidation conditions. The reversible distortion of the [WO6] units explained the delayed onset temperature of the structural rearrangement as well. The octahedral [WO6] seemed to influenced the structural rearrangements of the octahedral [MoO6] to tetrahedral [MoO4] units under catalytic conditions. Therefore, a shift to increased onset temperatures of the distortion process depending on the decreased degree of W substitution was detectable.

Table 6-1: Peak positions of the fitting Lorentz peaks, splitted peak energy (difference of the peak positions of the fitting Lorentz peaks), quotient of the fitting peak areas (peak 1area/ peak 2area), and the resulting ligand field of PW2Mo10-SBA-15 before and after propene oxidation conditions and at

723 K at propene oxidation conditions and the references WO3 and Na2WO4..

splitted peak peak 1area / Resulted Peak 1 [keV] Peak 2 [keV] energy [eV] peak 2area ligand field

PW2Mo10-SBA-15 10.2151 10.2183 3.2 1.38 Oh (before)

PW2Mo10-SBA-15 10.2153 10.2179 2.6 1.52 Oh (723 K)

PW2Mo10-SBA-15 10.2152 10.2183 3.1 1.37 Oh (after)

WO3 10.2147 10.2187 4.0 1.29 Oh

Na2WO4 10.2143 10.2167 2.4 0.68 Td

The interaction between the [MoOx] and [WO6] units resulted in a structure directing effect to mostly octahedral [MoO6] units under propene oxidation conditions depending on the degree of W substitution. The structure directing effect of the addenda tungsten atoms differed from the structure directing of addenda vanadium in supported HPOM. In

PVxMo12-x-SBA-15 (x = 1, 2), the [MoO6] units were influenced by the neighboring [VO6]

units of the initial Keggin ion structure resulting in mostly tetrahedral [MoO4] and [VO4] units depending on the degree of V substitution during thermal treatment under propene oxidation conditions (cf. chapter 5.2).

6.2.1 Local structure in activated PWxMo12-x-SBA-15 (x = 0, 1, 2) and a reference

W2Mo10Ox-SBA-15 under catalytic conditions

Local structure around the Mo centers in act. PWxMo12-x-SBA-15 (x = 0, 1, 2)

3 Fig. 5-5 (left) shows the Mo K edge FT(χ(k)·k ) of act. PMo12-SBA-15, act. PWMo11-

SBA-15, and act. PW2Mo10-SBA-15 after thermal treatment under propene oxidation 3 conditions at 723 K. The resulted Mo K edge FT(χ(k)·k ) of PWxMo12-x-SBA-15 ( x = 0, 1, 2) were similar. Minor differences are marked in the Mo K edge FT(χ(k)·k3) and the Mo K 3 edge χ(k)·k . For a more detailed analysis hexagonal MoO3 was used as structural model.

0.30 PMo12-SBA-15

0.25

0.20 12

PWMo10-SBA-15

)

0.15 3 8

(k)·k

(k)·k χ χ

0.10 4 FT(

0.05 PW2Mo10-SBA-15 0

0.00 -4 4 6 8 10 12 14 16 -1 -0.05 k [Ǻ]

0 1 2 3 4 5 6 R [Ǻ]

3 3 Fig. 6-7: (left) Mo K edge FT(χ(k)·k ) and (right) Mo K edge χ(k)·k of act. PMo12-SBA-15, act.

PWMo11-SBA-15, and act. PW2Mo10-SBA-15 after thermal treatment under propene oxidation conditions at 723 K.

XAFS phases and amplitudes were calculated for Mo-O and Mo-Mo distances and used for EXAFS refinement. The results of the refinement are shown in Table 6-2. The first peak of 3 Mo K edge FT(χ(k)·k ) of as prepared PWxMo12-x-SBA-15 (x = 1, 2) (cf. chapter 4.3) exhibited differences compared to act. PWxMo12-x-SBA-15 (x = 1, 2). The first peak in the 3 FT(χ(k)·k ) originated from tetrahedral and octahedral [MoOx] species on the SBA-15 support and could be sufficiently simulated using four Mo-O distances. These four distances sufficiently accounted for the minor amount of octahedral [MoO6] species. Confirming the results of the XANES analysis 1st and 2nd disorder parameters (1st-σ2, 2 2nd-σ ) were higher for act. PWMo11-SBA-15 and PW2Mo10-SBA-15 indicating a decreasing amount of tetrahedral structural motifs compared to act. PVxMo12-x-SBA-15 (x = 0, 1, 2). In addition, the 4th disorder parameters were smaller than the disorder parameter for act. PMo12-SBA-15. This disorder parameter represented mainly the fraction of octahedral [MoO6] species. Hence, the reduced disorder parameter indicated an increasing amount of octahedral structural motifs in act. PWxMo12-x-SBA-15 (x = 1, 2) compared to act. PMo12 -SBA-15. Therefore a structure directing effect of addenda tungsten resulting in a decreased ratio of tetrahedral [MoO4] to octahedral [MoO6] under

Table 6-2: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in act. PWxMo12-x -SBA-15 (x = 0, 1, 2). Experimental parameters were obtained from a refinement of a hexagonal MoO3 model structure (ICSD 75417 [135]) to the experimental -1 Mo K edge XAFS χ(k) of act. PWxMo12-x -SBA-15 (x = 0, 1, 2) (k range from 3.4-16.0 Å , R range from 0.9 to 4.0 Å, E0= ~ -5.2, residuals ~12.1-14.7 Nind = 26, Nfree =12). Subscript c indicates parameters that were correlated and f fixed in the refinement.

hex-MoO3 act. PMo12- act. PWMo11- act. PW2Mo10- model SBA-15 SBA-15 SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 2 1.67 1.67 0.0015 1.67 0.0019 1.68 0.0024

Mo-O 2 1.96 1.89 0.0038c 1.90 0.0042c 1.90 0.0045c

Mo-O 1 2.20 2.19 0.0038c 2.18 0.0042c 2.16 0.0045c Mo-O 1 2.38 2.35 0.0011 2.35 0.0009 2.35 0.0008

Mo-Mo 2 3.31 3.49 0.0068c 3.49 0.0077c 3.49 0.0072c

Mo-Mo 2 3.73 3.63 0.0068c 3.63 0.0077c 3.63f 0.0072c Mo-Mo 2 4.03 3.73 0.0100 3.72 0.0103 3.74 0.0095

propene oxidation conditions was determined. The distinct peak at ~3 Å (not phase 3 corrected) in the FT(χ(k)·k ) indicated the formation of dimeric or oligomeric [MoxOy] units on SBA-15. Therefore, isolated octahedral [MoO6] and tetrahedral [MoO4] units can be excluded as major molybdenum oxide species.

Local structure around the W centers in act. PWxMo12-x-SBA-15 (x = 1, 2)

3 Fig. 6-8 (left) shows the W LIII edge FT(χ(k)·k ) of act. PW2Mo10-SBA-15 at 723 K during 3 propene oxidation conditions. Fig. 6-8 (right) depicts the W LIII edge FT(χ(k)·k ) of act.

PWMo11-SBA-15, and act. PW2Mo10-SBA-15 after thermal treatment under propene 3 oxidation conditions at 723 K. Comparing the W LIII edge FT(χ(k)·k ) of act. PW2Mo10- SBA-15 at 723 K and after propene oxidation conditions resulted in significant differences. For a detailed structure analysis theoretical XAFS phases and amplitudes were calculated for W-O, W-Si and W-Mo distances and used for EXAFS refinement. The used theoretical

0.10 PWMo11-SBA-15

)

0.05 (k)·k

0.04 χ

)

FT( PW2Mo10-SBA-15

3 0.02 (k)·k

0.00 0.00 FT(

-0.02

0 1 2 3 4 5 6 0 1 2 3 4 5 6 R [Ǻ] R [Ǻ]

3 Fig. 6-8: Experimental (solid) and theoretical (dashed) W LIII edge FT(χ(k)·k ) of act. (left)

PW2Mo10-SBA-15 during thermal treatment in 5% propene and 5% oxygen in helium at 723 K;

(right) act. PWMo11-SBA-15 and act. PW2Mo10-SBA-15 after thermal treatment in 5% propene and 5% oxygen in helium at 723 K.

model system based on modified H3[PMo12O40] (ICSD 209 [14,93]) where P was replaced by Si. Thus, the model structure corresponded to a former triad of the Keggin ion with a Si bond. The result of the refinements are summarized in Table 6-3. The shapes of W LIII 3 edge FT(χ(k)·k ) of as prepared PWxMo12-x-SBA-15 (x = 1, 2) (cf. chapter 4.3) were different to that of act. PWxMo12-x-SBA-15 (x = 1, 2). This results confirmed the assumption of structural changes of [WOx] units in PWxMo12-x-SBA-15 (x = 1, 2) during 3 propene oxidation conditions. The first peak of W LIII edge FT(χ(k)·k ) of act. PWxMo12-x- SBA-15 (x = 1, 2) could be sufficiently simulated using three W-O distances. These distances sufficiently accounted for the amount of octahedral [WO6] species. The 3 refinement of the W LIII edge FT(χ(k)·k ) of PW2Mo10-SBA-15 during propene oxidation at 723 K indicated two decreased disorder parameters (1st-σ2, 2nd-σ2) and one increased 2 disorder parameter (3rd-σ ) compared to PW2Mo10-SBA-15 after catalytic conditions. The 3 resulting distances of the first shell of both W LIII edge FT(χ(k)·k ) were nearly identical. Hence, the different disorder parameters corroborated a distorted arrangement of the octahedral [WO6] units. Generally, disorder parameters will increase linear with temperature, if a structural rearrangement can be excluded.[159,13] Hence the identified

Table 6-3: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in act. PWMo11-SBA-15, act. PW2Mo10-SBA-15 after and act. PW2Mo10-SBA-15 during thermal treatment in 5% propene and 5% oxygen in helium at 723 K. Experimental paramters were obtained from a refinement of modified H3[PMo12O40] (ICSD 209 [14,93]) model structure to the experimental W LIII edge XAFS χ(k) of act. PWxMo12-x-SBA-15 (x = 1, 2) (k range -1 from 3.0-13.6 Å , R range from 1.0 to 3.8 Å, E0 = ~ -4.5 residual ~12.9 Nind = 10, Nfree = 20). Subscript c indicates parameters that were correlated in the refinement.

act. PWMo11- act. PW2Mo11- act. PW2Mo11-

SBA-15 SBA-15 (723 K) SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2)

W-O 2 1.68 1.79 0.0048c 1.76 0.0055c 1.77 0.0065c

W-O 2 1.91 1.69 0.0048c 1.69 0.0055c 1.71 0.0065c W-O 2 1.92 1.89 0.0021 1.89 0.0032 1.89 0.0026

W-Mo 2 3.42 3.56 0.0117c 3.53 0.0145c 3.53 0.0087c W-Si 1 3.10 3.08 0.0057 3.11 0.0063 3.06 0.0034

W-Mo 2 3.71 3.78 0.0117c 3.80 0.0145c 3.74 0.0087c

changes in the first disorder parameters (1st σ2, 2nd σ2 and 3rd σ2) were due to various distorted arrangement of the octahedral [WO6] units. The 2nd W-Mo distance was increased for act. PW2Mo10-SBA-15 at 723 K compared to act. PW2Mo10-SBA-15 after thermal treatment. Thus, it may be assumed that a different binding state of the W-Mo bonds existed at 723 K and after thermal treatment. A similar feature could be indentified in the W-Si bond. The resulted structural motifs especially the shorter 3rd W-O distance, indicated that the connection to phosphorus was broken as well. Nevertheless, the triad of the original Keggin ion may persisted resulting in connected edge- and corner-sharing octahedral [(W, Mo)O6] units. The [(W, Mo)O6] units were additionally connected to the support material. Typical distances for edge-sharing tungsten oxide compounds and corner-sharing H3[PW12O40], (NH4)10H2W12O42·4H2O, and WO3 were 3.4-3.6 Å and 3.7- 3.9 Å, respectively.[28,160,161] Therefore, the resulting W-Mo distances between 3.53-

3.80 Å in act. PW2Mo10-SBA-15 corresponded to both edge- and corner-sharing units.

Ross-Medgarden et al. found for WO3 supported on SiO2 a comparable structure motif. The resulted structure after dehydration conditions corresponded to a Si containing Keggin type cluster with corner- and edge-shared [WO6] units on the support material with an interacting bond to Si.[162] Therefore, a comparable structural motif of W substituted heteropolyoxomolybdates on SBA-15 resulting under propene oxidation conditions was expected.

6.2.2 Comparison of the local structure around Mo centers in act. PW2Mo10-SBA-15 and

a reference act. W2Mo10Ox-SBA-15 under catalytic conditions

3 Fig. 6-9 shows the Mo K edge FT(χ(k)·k ) of act. PW2Mo10-SBA-15 and act. W2Mo10Ox- SBA-15 after thermal treatment under catalytic conditions. The first peak of the Mo K 3 edge FT(χ(k)·k ) of act. W2Mo10Ox-SBA-15 resembled that of act. PW2Mo10-SBA-15. In contrast to the reference V2Mo10Ox-SBA-15 no Mo-Mo distance at ~3.3 Ǻ (not phase corrected) indicating crystalline α-MoO3 could be detected. Hence the resulted [MoxOy] species seemed to be very well dispersed on the support material SBA-15. The first peak in 3 both FT(χ(k)·k ) originated from tetrahedral [MoO4] and octahedral [MoO6] species on the SBA-15 support and could be sufficiently simulated using four Mo-O distances. These four distances sufficiently accounted for the amount of octahedral [MoO6] species. Linear

0.20

act. PW2Mo10-SBA-15 0.15

0.10

) 3

(k)·k act. W2Mo10Ox -SBA-15

χ 0.05 FT(

0.00

-0.05

0 1 2 3 4 5 6 R [Ǻ]

3 Fig. 6-9: Mo K edge FT(χ(k)·k ) of act. PWMo11-SBA-15 and act. W2Mo10Ox-SBA-15 after thermal treatment in 5% propene and 5% oxygen in helium at 723 K.

combinations of the XANES spectra of Na2MoO4 and MoO3 references were used to determine the amount of tetrahedral [MoO4] and octahedral [MoO6] units in act.

PW2Mo10-SBA-15 and act. W2Mo10Ox-SBA-15. Apparently, act. W2Mo10Ox-SBA-15 consisted of a mixture of increased tetrahedral [MoO4] and decreased octahedral [MoO6] units. For act. W2Mo10Ox-SBA-15 and for act. PW2Mo10-SBA-15 a [MoO4]:[MoO6] ratio of 3:2 and 1:4. were found, respectively. A comparison of the pseudo radial distribution function of act. W2Mo10Ox-SBA-15 and act. PW2Mo10-SBA-15 confirmed the results of the XANES analysis (Table 6-4). The 1st and 2nd disorder parameters (1st-σ2, 2nd-σ2) were higher for act. PW2Mo10-SBA-15 and indicated a decreasing amount of tetrahedral 2 MoO4 units. In addition, the 4th disorder parameter (4th-σ ) of is smaller than the disorder parameter for act. W2Mo10Ox-SBA-15. This disorder parameter mainly represented the fraction of octahedral MoO6 units. Therefore, the reduced disorder parameter indicated an increasing amount of octahedral structural motifs in act. PW2Mo10-SBA-15 compared to act. W2Mo10Ox-SBA-15. Furthermore, the Mo-Mo distances and disorder parameters were comparable for act. PW2Mo10-SBA-15 and act. W2Mo10Ox-SBA-15 indicating well dispersed [MoOx] units. A significant amount of crystalline MoO3 compared to the reference act. V2Mo10Ox-SBA-15 (cf. chapter 5.2.1) could be excluded. The ratio of

[MoO4]/[MoO4] units was increased indicating a favored interaction of the [MoxOy] species with the support material SBA-15 in contrast to act. PW2Mo10-SBA-15.

Table 6-4: Type and number (N), and XAFS disorder paramters (σ2) of atoms at distance R from the Mo atoms in act. PWxMo12-x -SBA-15 (x = 0, 1, 2). Experimental parameters were obtained from a refinement of a hexagonal MoO3 model structure (ICSD 75417 [135]) to the experimental -1 Mo K edge XAFS χ(k) of act. PWxMo12-x -SBA-15 (x = 0, 1, 2) (k range from 3.4-16.0 Å , R range from 0.9 to 4.0 Å, E0= ~ -5.2, residuals ~12.1-14.7 Nind = 26, Nfree = 12). Subscript c indicates parameters that were correlated and f fixed in the refinement.

hex-MoO3 act. PW2Mo12- act. W2Mo10Ox- model SBA-15 SBA-15 N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 2 1.67 1.68 0.0024 1.67 0.0017

Mo-O 2 1.96 1.90 0.0045c 1.88 0.0041c

Mo-O 1 2.20 2.16 0.0045c 2.18 0.0041c Mo-O 1 2.38 2.35 0.0008 2.35 0.0013

Mo-Mo 2 3.31 3.49 0.0072c 3.47 0.0072c

Mo-Mo 2 3.73 3.63f 0.0072c 3.61 0.0072c Mo-Mo 2 4.03 3.74 0.0095 3.70 0.0099

Comparison of the local structure around the W centers in act. PW2Mo10-SBA-15 and a reference act. W2Mo10Ox-SBA-15 under catalytic conditions

3 Fig. 6-10 (left) depicts the W LIII edge FT(χ(k)·k ) of act. PW2Mo10-SBA-15, act.

W2Mo10Ox -SBA-15 after thermal treatment under propene oxidation conditions at 723 K, and monoclinic WO3. Different structural motifs could be assumed comparing the shapes 3 of the W LIII edge FT(χ(k)·k ) of act. PW2Mo10-SBA-15 and act. W2Mo10Ox -SBA-15. W 3 3 LIII edge FT(χ(k)·k ) and χ(k)·k of act.W2Mo10Ox -SBA-15 were nearly identical to monoclinic WO3. Thus, the predominant [WOx] species seemed to be crystalline monoclinic WO3. XRD (Fig. 6-11) of the reference W2Mo10Ox-SBA-15 before thermal treatment confirmed this assumption. Comparable to the reference V2Mo10Ox-SBA-15 (cf. chapter 5.2.1) and in contrast to act. PW2Mo10-SBA-15 the [WOx] and [MoOx] units were not in a close vicinity. Active sites of selective oxidation catalysts are often multifunctional

0.5

act. PW2Mo10-SBA-15

0.4 30 25 0.3 act. W2Mo10Ox -SBA-15

)

3 15

(k)·k 0.2

(k)·k χ

χ 10 FT( 5 0.1 0 WO 3 -5 0.0 4 6 8 10 12 k [Ǻ]-1

-0.1 0 1 2 3 4 5 6 R [Ǻ] 3 3 Fig. 6-10: (left) W LIII edge FT(χ(k)·k ) and (right) W LIII edge χ(k)·k of act. PW2Mo10-SBA-15

(green), act. W2Mo10Ox-SBA-15 (red) after thermal treatment under propene oxidation conditions at 723 K and monoclinic WO3 (blue) as reference.

and consist of multiple metal centers.[28] Synthesis routes of supported ternary oxides with different metal oxide precursors rarely have been reported. Hence, the synthesis of supported tungsten substituted Keggin ions enabled the synthesis of connected [WOx] and

[MoOx] species on SBA-15 comparable to the vanadium substituted PVxMo12-x-SBA-15

PW2Mo10-SBA-15

ntensity W2Mo10Ox -SBA-15 I

WO3

10 20 30 40 50 60 70 80 Diffraction angle 2Ɵ[°]

Fig. 6-11: XRD of as prepared PW2Mo10-SBA-15, as prepared W2Mo10Ox-SBA-15, and monoclinic WO3.

(x = 1, 2). Therefore the results of the structural characterization showed that the proximity of tungsten and molybdenum in the Keggin precursors was necessary to obtain connected metal oxide species on a support material.

6.3 Functional characterization of PWxMo12-x-SBA-15 (x= 1, 2)

6.3.1 Reducibility

Fig. 6-12 shows the H2TPR profiles of PWxMo12-x-SBA-15 (x = 0, 1, 2). The resulted H2 TPR profiles revealed one sharp (~800 K) and a very broad (873- 973 K) reduction peak comparable to PVxMo12-x-SBA-15 (x = 0, 1, 2) (cf. chapter 5.3.1). The shapes of the H2 TPR profiles were nearly identical. The reduction temperatures increased slightly

PMo12-SBA-15 808 K PWMo -SBA-15 11 790 K 815 K

PW2Mo10-SBA-15

consumption

2 2 H

373 473 573 673 773 873 973 T [K]

Fig. 6-12: Temperature programmed reduction (H2 TPR) of PMo12-SBA-15, PWMo11-SBA-15, -1 and PW2Mo10-SBA-15 measured at a heating rate of 8 Kmin 5% H2 in Ar.

with the degree of tungsten substitution from 790 K for PMo12-SBA-15 to 815 K for

PW2Mo10-SBA-15. The H2 TPR profiles at ~800 K were comparable to molybdenum oxides supported on SBA-15 with Mo loadings between 9.5 wt.% and 13.3 wt.%.[149,150]

Lou et al. assigned the sharp reduction peak to oligomeric [MoxOy] species or small MoOx clusters and the broaded signal above ~800 K to the reduction of monomeric [MoOx] species (cf. chapter 5.3.1).[150] Tungsten oxide reduced at temperatures between 1063-

1273 K. Tungsten oxide supported on SiO2 with a loading of 8.0 wt.% W had a reduction

temperature of 804 K and 1073 K .[163,164] The reduction peak at the higher temperature (1073 K) was ascribed to the reduction of well dispersed tungsten species.[165] Comparing the typical reduction temperatures of supported tungsten oxides to molybdenum oxides, slightly higher reduction temperatures for supported tungsten oxides were determined. Therefore, the slight increase of reduction temperature with the degree of tungsten substitution degree may be interpreted as intrinsic effect of the addenda tungsten, comparable to vanadium substituted PVxMo12-x-SBA-15 (x = 0, 1, 2) (cf. chapter 5.3.1). However, slight changes in the reducibility depending on the degree of tungsten substitution were detected.

6.3.2 Catalytic performance

Reaction rate and selectivity of PMo12-SBA-15, PWMo11-SBA-15, PW2Mo10-SBA-15,

W2Mo10Ox-SBA-15, and bulk PW2Mo10 in propene oxidation at 723 K are shown in Fig.

6-13. Reaction rates for PWxMo12-x-SBA-15 (x = 0, 1, 2) were calculated for similar propene oxidation conditions (~ 14-17% propene conversion). The propene conversion for bulk PW2Mo10 (~ 3%) and W2Mo10Ox-SBA-15 (~ 2%) were lower due to the strong decreased catalytic activity. Adjusting to similar propene oxidation conditions for the low active samples would lead to a large volume of the samples and thermal effects. Hence, the comparison of the catalytic performance between PWxMo12-x-SBA-15 (x = 0, 1, 2) and the low active samples has to be done carefully.

Reaction rates for PWxMo12-x-SBA-15 (x = 0, 1, 2) slightly increased with the degree of tungsten substitution in contrast to constant reaction rates for vanadium substituted

PVxMo12-x-SBA-15 (x = 1, 2) (cf. chapter 5.3.2). Selectivities for CO increased at the expense of those of acetaldehyd with higher degree of tungsten substitution. Structural analysis of act. PWxMo12-x-SBA-15 (x = 0, 1, 2) revealed a decreased concentration of tetrahedral [MoO4] units as a function of tungsten substitution. The degree of oligomerization of [MoxOy] seemed to be comparable in all act. PWxMo12-x-SBA-15 (x =

0, 1, 2) samples. Therefore, the additional [WO6] species in act. PWxMo12-x-SBA-15 (x = 0, 1, 2) may lead to new multifunctional active sites, resulting in a slightly different product distribution and increased reaction rates. This [WO6] species in act. PWxMo12-x- SBA-15 (x = 0, 1, 2) showed a structure directing effect towards formation of octahedral

acrylic acid acetone CO

] 1 acetic acid propionaldehyd CO2 - s acrolein acetaldehyd

100 e (Mo)

e 70 1

- g

80 60

50 60

40 (propene)

40 30

µmol [

Selectivity [%] 20 20 10 0

a b c d e 0 reaction rate

-1 -1 Fig. 6-13: Reaction rate (µmol(propene)·g (Mo)·s ) and selectivity of (a) PMo12-SBA-15, (b)

PWMo11-SBA-15, (c) PW2Mo10-SBA-15, (d) W2Mo10Ox-SBA-15, and (e) bulk PW2Mo10 in 5% propene and 5% oxygen in He at 723 K.

[MoO6] units. Hence, due to the compositional and structural variety structure-activity correlation remained vague.

The reaction rate of PW2Mo10-SBA-15 strongly increased due to the improved surface to bulk ratio compared to bulk PW2Mo10. While, PW2Mo10 showed a slightly increased selectivity towards acrolein, PW2Mo10-SBA-15 exhibited an increased selectivity towards acetic acid. Total oxidation products CO and CO2 amounted to about ~55% in the resulting oxidation products. Apparently, higher dispersion and an improved surface to bulk ratio of Keggin ions resulted in a much increased activity at comparable selectivity similarly to

PVxMo12-x-SBA-15 (x = 1, 2) (cf. chapter 5.3.2).

In contrast to the supported HPOM samples, W2Mo10Ox-SBA-15 showed a strongly decreased activity. The product distribution of W2Mo10Ox-SBA-15 was comparable to unsubstituted act. PMo12-SBA-15. The amount of total oxidation products in the gas phase was slightly lower and an increasing selectivity to acetaldehyde and acrolein was determined compared to act. PWxMo12-x-SBA-15 (x = 1, 2). The structural analysis indicated that act. W2Mo10Ox-SBA-15 possessed an increased amount of oligomerized

[MoxOy] species and an increased content of tetrahedral [MoO4] units. Additionally, the predominant tungsten oxide species seemed to be crystalline monoclinic WO3 which may not significantly participate in the catalytic reaction. Apparently, the new multifunctional

active site resulting from connected [WO6] and [MoxOy] units lead to an increasing amount of total oxidation products for act. PW2Mo10-SBA-15 in contrast to not connected [WO6] and [MoxOy] units in act. W2Mo10Ox-SBA-15. Combining the results of the structural and functional characterization, the increase of octahedral [MoO6] units resulting from tungsten substitution lead to an enhanced catalytic activity and slightly different selectivity. The reference act. W2Mo10Ox-SBA-15 with decreased octahedral [MoO6] units and crystalline WO3 exhibited a decreased catalytic activity at comparable selectivity compared to unsubstituted act. PMo12-SBA-15.

Apparently, oligomerized octahedral [MoO6] species were the catalytically active sites in propene oxidation. The various selectivities as a function of tungsten substitution in act.

PWxMo12-x-SBA-15 (x = 0, 1, 2) may be caused by new multifunctional active sites consisting of connected [WO6] and [MoxOy] species. Apparently, tetrahedral [MoO4] species were not involved in selective propene oxidation which confirmed the results of previous studies.[137]

6.4 Summary

Structural evolution of PWxMo12-x-SBA-15 (x = 0, 1, 2) and a reference W2Mo10Ox-SBA- 15 during propene oxidation conditions were examined by in situ X-ray absorption spectroscopy at the Mo K and W K edges. During thermal treatment under propene oxidation conditions PWxMo12-x-SBA-15 (x = 1, 2) formed a mixture of mainly octahedral

[MoOx] and [WO6] units. Changes in the local structure around the W centers were delayed compared to the structural changes of the Mo centers depending on the degree of tungsten substitution. The delayed structural rearrangement corresponded to the temperatures where the major structural rearrangement of the [MoxOy] species was finished. The octahedral [WO6] units distorted during propene oxidation conditions.This influenced the structural changes of the [MoxOy] species resulting in mainly octahedral

[MoO6] units depending on the degree of tungsten substitution. The degree of oligomerization of the [MoxOy] species for all act. PWxMo12-x-SBA-15 (x = 0, 1, 2) was comparable and independent of the degree of tungsten substitution. Apparently, the mainly octahedral [MoOx] and [WO6] units were in close vicinity and able to interacted under catalytic conditions. The new multifunctional active site resulting due to connected [WO6] and [MoOx] units lead to an increased reaction rate and increased amount of total oxidation products. Conversely, structural analysis of activated reference W2Mo10Ox-SBA-15 synthesized with individual W and Mo precursors indicated that [WO6] and [MoOx] species were mostly separated from each other on the surface of SBA-15. Moreover, activated W2Mo10Ox-SBA-15 possessed a decreased amount of oligomerized [MoxOy] species and an increased content of tetrahedral [MoO4] units. Additionally, the predominantly tungsten oxide species seemed to be crystalline monoclinic WO3 and may not be involved in the catalytic reaction. This may explain the strongly decreased reaction rates and similar selectivities towards partial oxidations products compared to unsubstituted PMo12-SBA-15. In total, supported tungsten substituted Keggin ions are suitable precursors to synthesize connected [WO6] and [MoOx] species on SBA-15. Apparently, the proximity of tungsten and molybdenum in the Keggin precursors is a prerequisite for obtaining connected metal oxide species.

7 Characterization of PMo12 supported on SBA-15 with tailored pore radii

Nanostructured SiO2 materials such as SBA-15 represent suitable support systems for oxide catalysts.[22-25] Studies on H4[PVMo11O40] supported on SBA-15 revealed a structure directing effect of the silica support on the stability of the resulting Mo oxide species.[27] H4[PVMo11O40] supported on SBA-15 formed a mixture of tetrahedrally and octahedrally coordinated and linked [MoO4] and [MoO6] units under catalytic conditions.[27] Previous studies have shown that catalytic activity and selectivity scales with both the concentration and the degree of oligomerization of tetrahedral [MoO4] and octahedral [MoO6] units at the surface.[137] Isolated [MoO4] units supported on MgO were nearly inactive for propene oxidation. The catalytic activity and selectivity towards oxygenates increased with increasing amount of [MoxOy] species.[137] Previously, the degree of oligomerization was adjusted by either varying the metal loading or altering the surface acidity of the support material.[25,153,166,167] In addition, only few other characteristics of supported model systems are conceivable to alter the connectivity of supported MoOx species. A complimentary approach may be varying the pore radii of the support material. This could lead to modified structure directing effects on supported HPOM at constant metal oxide loading and identical surface acidity. Subsequently, the resulting [MoxOy] structures on tailored SBA-15 may be used to further elucidated structure-activity relationships. A study with tailored SBA-15 as support material was necessary elucidating the catalytically active structural motifs. Therefore, H3[PMo12O40] was supported on SBA-15 with modified pore radii (10, 14, 19 nm). The Samples were prepared with a surface 2 coverage of 1 Keggin ion per 13 nm independent of the pore radii. PMo12-SBA-15 (10, 14, 19 nm) were treated under propene oxidation conditions. In situ X-ray absorption spectroscopy investigations at the Mo K edge of PMo12 supported on SBA-15 with different pore radii (10, 14, 19 nm) during catalytic conditions are presented. A detailed analysis of the structures resulting under catalytic conditions is performed and correlated with the catalytic activity and product distribution towards propene oxidation. Additionall,y a comparison of the thermal stability of the supported samples was established.

100

7.1 Experimental

Sample Characterization

X-Ray Fluorescence Analysis

Physisorption measurements

Nitrogen physisorption isotherms were measured at 77 K on a BEL Mini II volumetric sorption analyzer (BEL Japan, Inc.). Silica SBA-15 samples were treated under vacuum at 368 K for about 20 min and at 448 K for about 17 h before starting the measurement. Data processing was performed using the BELMaster V.5.2.3.0 software package. The specific surface area was calculated using the Brunauer–Emmett–Teller (BET) method in the 2 relative pressure range of 0.03-0.24 assuming an area of 0.162 nm per N2 molecule.[69] The adsorption branch of the isotherm was used to calculate pore size distribution and cumulative pore area according to the method of Barrett, Joyner, and Halenda (BJH).[70]

Powder X-ray diffraction (XRD)

XRD measurements were conducted on an X’Pert PRO MPD diffractometer (Panalytical, θ-θ geometry), using Cu K alpha radiation and a solid-state multi-channel PIXcel detector. Wide-angle scans (5-90° 2θ, variable slits) were collected in reflection mode using a silicon sample holder. Small-angle scans (0.4-6.0° 2θ, fixed slits) were collected in transmission mode with the sample spread between two layers of Kapton foil.

101

Thermal analysis

Thermogravimetric (TG) measurements were conducted using a SSC 5200 from Seiko Instruments. The gas flow through the sample compartment was adjusted to 100 ml/min

(20% O2 and 80% He). Samples were measured with a rate of 2 K/min in the range from 298 K to 823 K.

X-ray absorption spectroscopy (XAS)

Transmission XAS experiments were performed at the Mo K edge (19.999 keV) at beamline X at the Hamburg Synchrotron Radiation Laboratory, HASYLAB, using a Si(311) double crystal monochromator. In situ experiments were conducted in a flow reactor at atmospheric pressure (5 vol% oxygen in He, total flow ~30 ml/min, temperature range from 303 to 723 K, heating rate 4 K/min). The gas phase composition at the cell outlet was continuously monitored using a non-calibrated mass spectrometer in a multiple ion detection mode (Omnistar from Pfeiffer). X-ray absorption fine structure (XAFS) analysis was performed using the software package WinXAS v3.2..[91] Background subtraction and normalization were carried out by fitting linear polynomials and 3rd degree polynomials to the pre-edge and post-edge region of an absorption spectrum, respectively. The extended X-ray absorption fine structure (EXAFS) χ(k) was extracted by using cubic splines to obtain a smooth atomic 3 background μ0(k) The FT(χ(k)·k ), often referred to as pseudo radial distribution function, was calculated by Fourier transforming the k3-weighted experimental χ(k) function, multiplied by a Bessel window, into the R space. EXAFS data analysis was performed using theoretical backscattering phases and amplitudes calculated with the ab-initio multiple-scattering code FEFF7.[92] Structural data employed in the analyses were taken from the Inorganic Crystal Structure Database (ICSD).

Single scattering and multiple scattering paths in the H3[PMo12O40] (ICSD 209 [14,93]) and hexagonal MoO3 (ICSD 75417 [135]) model structure was calculated up to 6.0 Å with a lower limit of 4.0% in amplitude with respect to the strongest backscattering path. EXAFS refinements were performed in R space simultaneously to magnitude and imaginary part of a Fourier transformed k3-weighted and k1-weighted experimental χ(k) using the standard EXAFS formula.[94] This procedure reduces the correlation between

102

the various XAFS fitting parameters. Structural parameters allowed to vary in the refinement were (i) disorder parameter σ2 of selected single-scattering paths assuming a symmetrical pair-distribution function and (ii) distances of selected single-scattering paths. Detailed information about the fitting procedure are described in chapter 3.2.

Catalytic testing - selective propene oxidation

Gas, 10% propene (3.5) in He (5.0)) and 5% oxygen (Linde Gas, 20% O2 (5.0) in He (5.0)) in helium (Air Liquide, 6.0) was used in a temperature range of 293-723 K Reactant gas flow rates of oxygen, propene, and helium were adjusted with separate mass flow controllers (Bronhorst) to a total flow of 40 ml/min. All gas lines and valves were preheated to 473 K. Hydrocarbons and oxygenated reaction products were analyzed using a Carbowax capillary column connected to an AL2O3/MAPD column or a fused silica restriction (25 m·0.32 mm each) connected to a flame ionization detector. O2, CO, and CO2 were separated using a Hayesep Q (2 m x 1/8``) and a Hayesep T packed column (0.5 m x 1/8``) as precolumns combined with a back flush. For separation, a Hayesep Q packed column (0.5 m x 1/8``) was connected via a molsieve (1.5 m x 1/8``) to a thermal conductivity detector (TCD). Details about the calculation of conversion, selectivity, and reaction rate are described in chapter 3.2.

Sample preparation

Silica SBA-15 samples with a pore diameter of ~10 nm was prepared according to Ref. [22]. 16.2 g of triblock copolymer (Aldrich, P123) were dissolved in 294 g water and 8.8 g hydrochloric acid at 308 K and stirred for 24 h. After addition of 32 g tetraethyl orthosilicate, the reaction mixture was stirred for 24 h at 373 K. The resulting gel was

103

transferred to a glass bottle and the closed bottle was heated to 388 K for 24 h. Subsequently, the suspension was filtered by vacuum filtration and washed with a mixture of H2O/EtOH (100:5). Silica SBA-15 with large pores were prepared according to Refs. [46],[51]. Silica SBA-15 with a pore diameter ~14 nm (or ~19 nm) was prepared as follows. 9.6 g of triblock copolymer (Aldrich, P123) were dissolved in 336 ml hydrochloric acid (1.3 M) at 288 K (or 290 K). After addition of 0.108 g ammonium fluoride the solution was stirred for 16 h (or 24 h). 20.7 g tetraethyl orthosilicate and 8.08 g 1,3,5-triisopropylbenzene were added to the solution. The resulting gel was transferred to a glass bottle and the closed bottle was heated to 393 K (or 373 K) for 29 h (or 48 h). Subsequently, the suspension was filtered by vacuum filtration and washed with a mixture of EtOH/HCl/H2O (100:10:100). The resulting white powders were dried at 378 K for 3 h and calcined at 453 K for 3 h and at 823 K for 5 h.

H3[PMo12O40] was prepared as described in chapter 3.1. H3[PMo12O40] was supported on SBA-15 via incipient wetness. The amount of molybdenum was adjusted to 10 wt.%, 6.7 wt.%, and 5.2 wt.% on SBA-15 (10, 14, 19 nm).

104

7.2 Structure of the support materials

Pore size distributions and specific surface areas of the synthesized support materials were calculated from N2 adsorption/desorption isotherms. SBA-15 (10 nm), SBA-15 (14 nm), and SBA-15 (19 nm) showed typical type IV isotherms indicative of mesoporous materials. Adsorption and desorption branches in hysteresis range were nearly parallel for

SBA-15 (10 nm) and SBA-15 (14 nm) indicating regular shaped pores Fig. 7-1. N2

isotherm for SBA-15 (19 nm) showed a slight broadening of the hysteresis loop, indicating

] ]

800 -

] 1

- 600

[ml nm [ml

5 p 10 15 20 25 d 400 V

d dp [nm] rrr Volume [ml g [ml Volume 200

0 0.0 0.2 0.4 0.6 0.8 1.0

Relative Pressure p/p0 Fig. 7-1: Nitrogen physisorption isotherms of silica SBA-15 (10 nm) (square), SBA-15 (14 nm) (circle), and SBA-15 (19 nm) (triangle) and pore distributions of of silica SBA-15 (10 nm) (square), SBA-15 (14 nm) (circle), and SBA-15 (19 nm) (triangle)(inset). the development of minor constrictions. BET surface areas were calculated from physisorption data. The tailored SBA-15 samples exhibited areas between 400 and 2 850 m /g. Specific surface area aBET (calculated by BET method), external surface area aEXT

(calculated as the difference between aBET and aMeso), and area corresponding to the mesopores aMeso are summarized in Table 7-1. Fig. 7-1 (inset) shows the pore size distribution derived from BJH analysis resulting in three different pore diameters of ~10, ~14, and ~19 nm. Small-angle X-ray diffraction patterns of the tailored SBA-15 are presented in Fig. 7-2 (left). The second derivates of small-angle X-ray diffraction patterns are shown for clarity in Fig. 7-2 (right). SBA-15 (10 nm) and SBA-15 (14 nm) exhibited

105

ntensity

2nd derivate 2nd

norm. norm. i Norm. intensity Norm.

-2.0 -1.5 -1.0 -0.5 0.5 1.0 1.5 2.0 -1.0 -0.5 0.5 1.0 Diffraction angle 2Ɵ [°] Diffraction angle 2Ɵ [°] SBA-15 (10nm) SBA-15 (14nm) SBA-15 (19nm)

Fig. 7-2: (left) Low-angle X-ray diffraction patterns and (right) 2nd derivates of the low-angle X- ray diffraction patterns SBA-15 (10 nm), SBA-15 (14 nm), and SBA-15 (19 nm). the typical patterns with low-angle 10l, 11l, and 20l peaks corresponding to the two- dimensional hexagonal symmetry. Small-angle X-ray diffraction pattern of SBA-15 (19 nm) showed one peak at low values of 2Ɵ. The lattice spacings d10l, derived from the Bragg equation (eq. 2.1.1), and unit cell constants a0, corresponding to the hexagonal pore arrangement, are given in Table 7-1. The received structural parameter of the tailored SBA-15 confirmed a successful synthesis of mesoporous SiO2 materials with different pore size distributions, and high surface areas.

Table 7-1: Specific surface area aBET (calculated by BET method), external surface area aEXT

(calculated as the difference between aBET and aMeso), area corresponding to the mesopores aMeso, pore diameter dpore (calculated by BJH method), mesopore volume VMeso, d10l-values (derived from low-angle XRD), unit cell constants a0 (corresponding to the hexagonal pore arrangement) of SBA-

15 (10 nm), SBA-15 (14 nm), and SBA-15 (19 nm).

aBET aExt aMeso dBJH 3 d10l a0 2 2 2 VMeso (cm /g) (m /g) m /g) (m /g) (nm) ( nm) (nm) SBA-15 (10 nm) 843 145 698 10.3 1.233 10.52 12.14 SBA-15 (14 nm) 525 83 442 13.8 1.344 12.52 14.46 SBA-15 (19 nm) 395 50 345 18.5 0.957 14.77 17.05

106

7.3 Characterization of PMo12-SBA-15 (10, 14, 19 nm)

Local structure around theMo centers in PMo12-SBA-15 (10, 14, 19 nm)

3 Fig. 7-3 shows the theoretical and experimental Mo K edge FT(χ(k)·k ) of PMo12-SBA-15 3 (10, 14, 19 nm). The shapes of the FT(χ(k)·k ) resembled that of bulk PMo12 indicating a similar local structure around the Mo centers in supported and unsupported HPOM Keggin ions. For a more detailed structural analysis H3[PMo12O40] Keggin ions (ICSD 209 [14,93]) was chosen as model structure. Comparing the distances R and disorder 2 parameters σ of PMo12-SBA-15 (10, 14, 19 nm) supported on SBA-15 with different pore radii exhibited no significant differences between the initial Keggin ion structure and Keggin ions supported on SBA-15 (chapter 3) structure. The good agreement between theory and experiment for PMo12-SBA-15 (10, 14, 19 nm) confirmed the Keggin ion structure upon supporting PMo12 on SBA-15.[27]

0.25

0.20 PMo12-SBA-15 (19 nm)

0.15

) 3

χ(k)·k 0.10 PMo12-SBA-15 FT( (14 nm) 0.05

0.00 PMo12-SBA-15 (10 nm) -0.05 0 1 2 3 4 5 6 R [Å]

3 Fig. 7-3: Theoretical (dotted) and experimental (solid) Mo K edge FT(χ(k)·k ) of PMo12 supported on SBA-15 (10 nm), SBA-15 (14 nm), and SBA-15 (19 nm).

107

Table 7-2: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in as prepared PMo12-SBA-15 (10, 14, 19 nm). Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the -1 experimental Mo K edge XAFS χ(k) of PMo12-SBA-15 (10, 14, 19 nm) (k range from 3.0-13.7 Å ,

R range from 0.9 to 4.0 Å, E0= ~ 2.3, residuals ~11.3-12.5 Nind = 22, Nfree = 9). Subscript c indicates parameters that were correlated in the refinement.

PMo12-SBA-15 PMo12-SBA-15 PMo12-SBA-15 Keggin model (10nm) (14nm) (20nm) N R(Å) R(Å) σ 2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 1 1.68 1.65 0.0025 1.65 0.0029 1.65 0.0021

Mo-O 2 1.91 1.78c 0.0033c 1.78c 0.0040c 1.79 c 0.0032c

Mo-O 2 1.92 1.95c 0.0033c 1.95c 0.0040c 1.94c 0.0032c Mo-O 1 2.43 2.40 0.0008 2.39 0.0010 2.40 0.0007

Mo-Mo 2 3.42 3.42 0.0052c 3.41 0.0056c 3.42 0.0054c

Mo-Mo 2 3.71 3.74 0.0052c 3.74 0.0056c 3.74 0.0054c

Thermal stability of PMo12 supported on SBA-15 with different pore radii

Fig. 7-4 depicts the measured thermogravimetric data of PMo12-SBA-15 (10, 14, 19 nm) in

20% O2 in He. The mass loss between 303 K and 373 K was ascribed to desorption of physically adsorbed water on the surface of the materials. Relative mass loss decreased for samples with large pore diameter. Afterwards a nearly constant mass in between 373 K and 448 K could be detected. The temperature range 448-523 K showed a mass loss of ~1%

(PMo12-SBA-15 (11 nm)), ~0.7% (PMo12-SBA-15 (14 nm)), and ~0.5% (PMo12-SBA-15 (19 nm)). Sample mass between 523 K and 823 K did not change for all samples. Comparable behaviour was shown for pure silica samples in vacuum.[168] Silica dehydrated between room temperature and 453 K followed by the dehydroxylation process of silanol groups between 453 K and 673 K. This resulted in the formation of siloxane groups and a decrease of silanol density from 4.6 OH/nm2 (473 K) to 2.3 OH/nm2 (673 K).[145,168] Comparing dehydration and dehydroxylation processes of supported HPOM and bulk HPOM revealed a correlation between dehydration and thermal stability of the Keggin ion.

Bulk PMo12 loses 1.5 molecules of constitutional water between 673-713 K

108

100 PMo12-SBA-15 (10 nm)

PMo12-SBA-15 (14 nm) 98 PMo12-SBA-15 (19 nm)

92 Normalized Mass [%] Mass Normalized Normalized Mass [%] Mass Normalized 90 373 473 573 673 773 T [K]

Fig. 7-4: Thermograms of PMo12-SBA-15 (10 nm), PMo12-SBA-15 (14 nm), and PMo12-SBA-15

(19 nm) at 20% O2 in He. under dry air conditions. This decomposition is accompanied by the formation of

MoO3.[117] Structures resulting for supported HPOM after thermal treatment under oxidizing conditions were comparable to stabilized two dimensional hexagonal MoO3 on SBA-15.[27,59] It has been shown that structural characteristics of supported model systems like MoOx-SBA-15 and VOx-SBA-15 depended mainly on their hydration states and previous calcination processes.[95,144] A comparable effect may be responsible for the structural evolution of HPOM supported on SBA-15. Adsorbed water and silanol groups from the support material may possess a structure stabilizing effect on the Keggin ion. This effect would be comparable to that of water of crystallization and constitutional water in bulk HPOM under ambient conditions.[117] Therefore, dehydroxylation of SBA- 15 may be the driving force for the structural decomposition of the Keggin ion resulting in the formation of Mo oxide species on SBA-15.

7.4 Structural evolution of PMo12- SBA-15 (10, 14, 19 nm) under catalytic conditions

PMo12-SBA-15 (10, 14, 19 nm) samples were investigated by in situ XAS under catalytic conditions. Fig. 7-5 shows the evolution of molybdenum XANES spectra of PMo12-SBA- 15 during temperature-programmed treatment in 5% propene and 5% oxygen. The

109

absorption Normalized Normalized

20.2 673 20.15 573 20.10 T [K] 473 20.05 373 20.00 Photon energy

Fig. 7-5: in situ Mo K edge XANES spectra of PMo12-SBA-15 during temperature-programmed treatment in 5% propene and 5% oxygen in helium in a temperature range between 300 K and 723 K. resulting structures exhibited an increased concentration of tetrahedral [MoO4] units. The pre-edge peak features in the Mo K edge XANES spectra can be employed to elucidate the local structure around the Mo center. Using the pre-edge peak height sufficed to quantify the contribution of tetrahedral [MoO4] and distorted [MoO6] units present under catalytic conditions. Fig. 7-6 showed the Mo K edge XANES spectra of PMo12-SBA-15 (14 nm)

and a spectrum calculated from a linear combination of bulk MoO3 and bulk

1.00

0.75 Na2MoO4

0.50

0.25 Normalized absorption Normalized

MoO3 0.00 20.0 20.1 20.2 Photon energy [keV]

Fig. 7-6: Refinement of the sum (dotted) of XANES spectra of references MoO3 and Na2MoO4

(dashed) to Mo K edge XANES spectrum of activated PMo12-SBA-15 (14 nm) after thermal treatment under propene oxidation conditions at 723 K.

110

Na2MoO4. The linear combination represented the amount of distorted [MoO6] and tetrahedral [MoO4] units. Quantitative evolution of the structural units was used to visualize changes in the structure of PMo12-SBA-15 (10, 14, 19 nm) during temperature programmed treatment in 5% propene and 5% oxygen. Fig. 7-7 depicts the calculated concentration of tetrahedral [MoO4] units during thermal treatment under catalytic conditions. No significant structural changes of PMo12-SBA-15 (10, 14, 19 nm) could be detected in the temperature range between 303 K and 448 K. Apparently, the Keggin structure was stable on silica SBA-15 in the temperature range 303-448 K. Subsequently, concentration of tetrahedral [MoO4] units considerably increased in the temperature range between 448 K and 598 K for PMo12-SBA-15 (10, 14, 19 nm). The onset of structural rearrangement was identical for all PMo12-SBA-15 (10, 14, 19 nm) samples and independent of the pore radii of the support materials. The stability of the Keggin ion seemed to depend only on the nature of the support material. The structural evolution in the temperature range (448-598 K) correlated to the dehydration and dehydroxylation process of SiO2 under oxidizing conditions (20% O2 in He). Apparently, dehydroxylation process was also the driving force for the structural rearrangement of the PMo12-SBA-15 (10, 14, 19 nm) samples under catalytic conditions. At a temperature of about 598 K a higher amount of the tetrahedral [MoO4] units for PMo12-SBA-15 (14, 19 nm) could be detected compared to conventional PMo12-SBA-15 (10 nm) (Fig. 7-7). The concentration of

80 PMo12-SBA-15 (10nm)

PMo12-SBA-15 (14nm) %] PMo12-SBA-15 (19nm)

] ratio [ ratio ] 6

]/[MoO 4

20 [MoO

300 400 500 600 700 T [K]

Fig. 7-7: Evolution of MoO4/MoO6 ratio of PMo12-SBA-15 (10 nm), PMo12-SBA-15 (14 nm), and

PMo12-SBA-15 (19 nm) during thermal treatment under propene oxidation conditions.

111

tetrahedral [MoO4] units for PMo12-SBA-15 (14, 19 nm) amounted to ~65% compared to

~45% for PMo12.SBA-15 (10 nm). Hence, the concentration of tetrahedral [MoO4] units increased with the larger pore radii of the support material. Quantification of tetrahedral

[MoO4] and octahedral [MoO6] units in the temperature range between 598 K and 723 K confirmed this assumption. The [MoO4] concentration of PMo12-SBA-15 (14 nm) and

PMo12-SBA-15 (19 nm) were comparable and reached the highest concentration with 75%

[MoO4] units at 723 K. PMo12-SBA-15 (10 nm) reach a maximum of tetrahedral [MoO4] units (~50%) at 657 K. Subsequently, a decreasing concentration of tetrahedral [MoO4] units to 40% at 723 K was determined.

Influence of the pore radii to the resulting structure of [MoxOy] species

Fig. 7-8 shows the Mo K edge FT(χ(k)·k3) of activated PMo12-SBA-15 (10, 19 nm) after 3 thermal treatment under propene oxidation conditions. The FT(χ(k)·k ) of act. PMo12-SBA- 15 exhibited features similar to that of previously reported dehydrated molybdenum oxides and HPOM supported on SBA-15.[27,59] For a more detailed structural analysis hexagonal MoO3 was chosen as model structure. Theoretical XAFS phases and amplitudes were calculated for Mo-O and Mo-Mo distances and used for EXAFS refinement. The results of the refinement are given in Table 7-3. The first peak of Mo K edge FT(χ(k)·k3) of act. PMo12-SBA-15 (10 nm) exhibited differences compared to act. PMo12-SBA-15 (14, 19 nm). The first peak in the FT(χ(k)·k3) originated mainly from the tetrahedral species on

0.08 PMo12-SBA-15 (10 nm) PMo12-SBA-15 (19 nm)

0.04

3 (k)·k

χ 0.00 FT(

-0.04

-0.08 0 1 2 3 4 5 6 R [Å] 3 Fig. 7-8: Mo K edge FT(χ(k)·k ) of activated PMo12-SBA-15 (10 nm) and activated PMo12-SBA- 15 (19 nm) after thermal treatment under propene oxidation conditions at 723 K.

112

the SBA-15 support and could be sufficiently simulated using four Mo-O distances. These four distances sufficiently accounted for the minor amount of octahedral [MoO6] species. 2 2 The 1st and 2nd disorder parameters (1st-σ , 2nd-σ ) were higher for act. PMo12-SBA-15 (10 nm) and indicated a lower amount of tetrahedral structural units. Additionally, the 4th 2 disorder parameter (4th-σ ) was smaller than that of act. PMo12-SBA-15 (14, 19 nm) with larger pores. This disorder parameter mainly represented the fraction of octahedral [MoO6] species, and corresponded to increasing amount of octahedral structural motifs in act.

PMo12-SBA-15 (10 nm) compared to act. PMo12-SBA-15 (14, 19 nm). A distinct peak at ~3 Å in the FT(χ(k)·k3) indicated a significant amount of dimeric or oligomeric [MoxOy] units on SBA-15 (10, 14, 19 nm) independent of the pore radii. Hence, isolated tetrahedral [MoO4] units can be excluded as major molybdenum oxide species.[137] The obtained Mo-Mo distances were identical for act. PMo12-SBA-15 (10, 14, 19 nm) samples and, thus, were independent of the pore radii. The disorder parameters 2 σ of the Mo-Mo distances for act. PMo12-SBA-15 (10 nm) were slightly increased compared to act. PMo12-SBA-15 (14, 19 nm). This indicated a decreased oligomerization degree of Mo silica SBA-15 with larger pore diameters. Apparently, the

Table 7-3: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in act. PMo12-SBA-15 (10, 14, 19 nm). Experimental parameters were obtained from a refinement of a hexagonal MoO3 model structure (ICSD 75417 [135]) to the experimental Mo K -1 edge XAFS χ(k) of act. PMo12-SBA-15 (10, 14, 19 nm) (k range from 3.4-16.0 Å , R range from

0.9 to 4.0 Å, E0= ~ -5.2, residuals ~12.5 Nind = 26, Nfree =12). Subscript c indicates parameters that were correlated in the refinement.

hex-MoO3 act. PMo12- act. PMo12-SBA- act. PMo12-SBA- model SBA-15 (10nm) 15 (14nm) 15 (20nm) N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 2 1.67 1.67 0.0015 1.67 0.0009 1.67 0.0009

Mo-O 2 1.96 1.89 0.0038c 1.88 0.0024c 1.88 0.0024c

Mo-O 1 2.20 2.19 0.0038c 2.17 0.0024c 2.17 0.0024c Mo-O 1 2.38 2.35 0.0011 2.34 0.0030 2.34 0.0029

Mo-Mo 2 3.31 3.49 0.0068c 3.49 0.0054c 3.49 0.0058c

Mo-Mo 2 3.73 3.63 0.0068c 3.62 0.0054c 3.62 0.0058c Mo-Mo 2 4.03 3.73 0.0100 3.73 0.0089 3.72 0.0096

113

concentration of the [MoxOy] species reached a minimum on the large pore samples

PMo12-SBA-15 (14, 19 nm). These large pore SBA-15 (dMeso = 14-19 nm) materials possessed a larger angle of inclination because of the decreased curvature of the pores in these samples. This may reduce the contact area between the decomposing Keggin ions eventually resulting in a higher concentration of dispersed and isolated oxide species. Conversely, thermal decomposition of Keggin ions on small pore support materials was prone to result in connected [MoxOy] species. The Mo-O, Mo-Mo distances and disorder parameters for act. PMo12-SBA-15 (14 nm) and act. PMo12-SBA-15 (19 nm) were nearly identical and different from those of PMo12-SBA-15 (10 nm). This confirmed the results of the quantification of tetrahedral [MoO4] (~75%) and distorted [MoO6] (~25%) units, present on large pore materials under catalytic conditions. Apparently, the formation of oligomeric [MoxOy] units mostly consisting of tetrahedral [MoO4] units depended on the pore radius of the silica SBA-15. Hence, act. PMo12-SBA-15 (10 nm) with smaller pores favored the formation of more extended structures on the support material.

7.5 Functional characterization of PVxMo12-x-SBA-15 (x= 1, 2)

7.5.1 Influence of the resulting structures to catalytic activity

PMo12-SBA-15 (10, 14, 19 nm) samples were tested under catalytic conditions for selective propene oxidation. Fig. 7-9 shows a comparison of the selectivities towards the oxidation products and reaction rates. The oxidation product distributions were comparable for all three PMo12-SBA-15 (10, 14, 19 nm) samples. The oxidation product distributions were comparable for all three PMo12-SBA-15 (10, 14, 19 nm) samples. The reaction rates for PMo12-SBA-15 (14 nm) and PMo12-SBA-15 (19 nm) were also nearly identical. In contrast to the samples with larger pores, PMo12-SBA-15 (10 nm) showed a ~14% higher reaction rate during propene oxidation. The theoretical Mo coverage of 0.9 Mo/nm2 was similar for all PMo12-SBA-15 (10, 14, 19 nm) samples. However, a significant difference between all SBA-15 (10, 14, 19 nm) materials was the curvature of the surface in the pores. The pore structure of mesoporous SBA-15 corresponds to that of hollow cylinders. Thus, the curvature of the walls of these cylinders decreases with higher pore radius. Therefore, arrangement of spherical Keggin ions on an area along the inner surface of pores with different pore radii leads to a decreasing distance between the spheres at smaller

114

acrylic acid acetone CO acetic acid propionaldehyde CO2

1 acrolein acetaldehyde -

100 70

(Mo)

1 -

80 g

60 60

50 (propene)

40 Selectivity [%] Selectivity 20 40

0 30

10 nm 14 nm 20 nm µmol rate reaction

-1 -1 Fig. 7-9: Reaction rate (µmol(propene)g (Mo)s ) and selectivity of PMo12-SBA-15 (10, 14, 20 nm) in 5% propene and 5% oxygen in He at 723 K. pore radius. Hence, assuming a volume of 1 nm3 per Keggin ion and considerating the various curvatures for SBA-15 (10, 14, 19 nm) lead to an increased effective coverage at smaller pore radius. Therefore, the increased effective distance of the Keggin ion on

PMo12-SBA-15 (14 nm) and PMo12-SBA-15 (19 nm) resulted in a lower concentration of

[MoxOy] units compared to PMo12-SBA-15 (10 nm). Hence, the catalytic activity in propene oxidation increased with higher concentration of [MoxOy] units under catalytic conditions. [MoxOy] units orginating from PMo12-SBA-15 (10 nm) resulted in an enhanced catalytic activity without significant influence on the product distribution. Therefore, a higher concentration of [MoxOy] units at similar loadings improved the catalytic activity towards propene oxidation. Comparable results have been shown for PVMo11 supported on 2 -1 SiO2 (Aerosil 300: 295 m g , Nippon Aerosil Co., Ltd.) with different loadings during oxidation of methacrolein.[167] Selective propene oxidation requires the transfer of more than two electrons. Therefore, [MoxOy] sites are necessary to selectively oxidize the propene molecule.[37,169,170] Catalytic activity of supported vanadium oxide based catalysts depended also on the concentration of [VxOy] units comparable to

[MoxOy].[153,171] Therefore, [MoxOy] units seemed to be necessary for catalytic activity while their concentration increased with the effective coverage.

115

7.6 Summary

Structural evolution of H3[PMo12O40] supported on SBA-15 (PMo12-SBA-15) with different pore radii (10, 14, 19 nm) was examined by in situ X-ray absorption spectroscopy investigations at the Mo K edge during propene oxidation conditions. Large pore SBA-15 was successfully used as support material for molybdenum based oxidation catalysts. Supporting heteropolyoxo molybdates on large pore SBA-15 resulted in regular Keggin ions on the support material. During thermal treatment in propene oxidation conditions

PMo12-SBA-15 (10, 14, 19 nm) formed a mixture of mostly tetrahedral [MoO4] and octahedral [MoO6] units. The onset temperature of structural changes of PMo12-SBA-15 (10, 14, 19 nm) during thermal treatment in propene oxidation conditions was largely independent of the pore size of SBA-15. The stability of the Keggin ions depended mostly on the nature of the support. Apparently, the dehydroxylation of silanol groups of the support material was the driving force for the structural instability of the Keggin ion. The resulting [MoxOy] structures present under catalysis conditions depended on the pore size of the support material. A higher concentration of octahedral [MoO6] units and higher oligomerized [MoxOy] units was detected for act. PMo12-SBA-15 (10 nm) compared to act.

PMo12-SBA-15 (14, 19 nm). The higher concentration of [MoxOy] units present in act.

PMo12-SBA-15 (10 nm) resulted in an increased catalytic activity compared to to activated

PMo12-SBA-15 (14, 19 nm) with a lower concetration of [MoxOy] units. Selectivities towards oxidation products during propene oxidation were comparable and largely independent of the pore radii of act. PMo12-SBA-15 (10, 14, 19 nm). Apparently, tailoring the pore radius of silica SBA-15 permitted to prepare Mo oxide model systems to investigate correlations between activity and structure of characteristic oxide species at similar loadings.

116

8 Characterization of PVMo11 supported on SBA-15 with different metal loading

H4[PVMo11O40] supported on SBA-15 forms a mixture of tetrahedrally and octahedrally coordinated and linked [MoO4] and [MoO6] units under catalytic conditions.[27] Supposingly, the catalytic activity and selectivity towards oxygenates increases with increasing amount of linked [MoxOy] species.[27] Accordingly, isolated [MoO4] units supported on MgO were nearly inactive for propene oxidation.[137] The degree of oligomerization may be varied by increasing or decreasing the metal loading or by altering the surface acidity of the support material.[25,148,153] Additionally, the variation of pore radii of the support material showed various structure directing effects of supported HPOM at constant metal oxide loading and identical surface acidity (cf. chapter 7). A higher concentration of octahedral [MoO6] and [MoxOy] units for PMo12-SBA-15 with smaller pore radius could be detected compared to PMo12-SBA-15 with larger pores. While, the catalytic activity increased with the amount of [MoxOy], the selectivities towards oxidation products during propene oxidation conditions were comparable and independent of the pore radius. This indicated similar active sites in act. PMo12-SBA-15 with various pore radii. For elucidating structure activity correlations a study with varied metal loading was necessary to established the catalytic active structure motifs. Variation of metal loading of

PVMo11-SBA-15 could lead to different structure directing effects during propene oxidation conditions. Therefore, H4[PVMo11O40] was supported on SBA-15 with different Mo loading (1 wt.% Mo, 5 wt.% Mo, and 10 wt.% Mo) to elucidate the resulting structure under propene oxidation conditions and the influence on the catalytic activity. Hence,

PVMo11-SBA-15 was treated under propene oxidation conditions. In situ X-ray absorption spectroscopy investigations at the Mo K edge of PVMo11 supported on SBA-15 with different Mo loading) under catalytic conditions were conducted. A detailed analysis of the structures present under catalytic conditions was performed and correlated with the catalytic activity and product distribution towards propene oxidation.

117

8.1 Experimental

Sample Characterization

X-Ray Fluorescence Analysis

X-ray absorption spectroscopy (XAS)

Transmission XAS experiments were performed at the Mo K edge (19.999 keV) at beamline X at the Hamburg Synchrotron Radiation Laboratory, HASYLAB, using a Si(311) double crystal monochromator. In situ experiments were conducted in a flow reactor at atmospheric pressure (5 vol% oxygen in He, total flow ~30 ml/min, temperature range from 303 to 723 K, heating rate 4 K/min). The gas phase composition at the cell outlet was continuously monitored using a non-calibrated mass spectrometer in a multiple ion detection mode (Omnistar from Pfeiffer). X-ray absorption fine structure (XAFS) analysis was performed using the software package WinXAS v3.2..[91] Background subtraction and normalization were carried out by fitting linear polynomials and 3rd degree polynomials to the pre-edge and post-edge region of an absorption spectrum, respectively. The extended X-ray absorption fine structure (EXAFS) χ(k) was extracted by using cubic splines to obtain a smooth atomic 3 background μ0(k) The FT(χ(k)·k ), often referred to as pseudo radial distribution function, was calculated by Fourier transforming the k3-weighted experimental χ(k) function, multiplied by a Bessel window, into the R space. EXAFS data analysis was performed using theoretical backscattering phases and amplitudes calculated with the ab-initio multiple-scattering code FEFF7.[92] Structural data employed in the analyses were taken from the Inorganic Crystal Structure Database (ICSD).

118

Single scattering and multiple scattering paths in the H3[PMo12O40] (ICSD 209 [14,93]) and hexagonal MoO3 (ICSD 75417 [135]) model structures were calculated up to 6.0 Å with a lower limit of 4.0% in amplitude with respect to the strongest backscattering path. EXAFS refinements were performed in R space simultaneously to magnitude and imaginary part of a Fourier transformed k3-weighted and k1-weighted experimental χ(k) using the standard EXAFS formula.[94] This procedure reduces the correlation between the various XAFS fitting parameters. Structural parameters allowed to vary in the refinement were (i) disorder parameter σ2 of selected single-scattering paths assuming a symmetrical pair-distribution function and (ii) distances of selected single-scattering paths. Detailed information about the fitting procedure are described in chapter 3.2.

Temperature programmed reduction

Temperature programmed reduction (TPR) was performed with a catalysts analyzer from BEL Japan Inc. equipped with a silica glass tube reactor. Samples were placed on silica wool inside the reactor next to a thermocouple. A gas flow (5 % H2 in Ar) of 60 ml/min was adjusted during reaction. A heating rate of 8 K / min to 973 K was used while H2 consumption was measured with a TCD unit. All samples were treated with a gas flow of 60 ml/min Ar at 393 K for about 45 min before starting the measurement. For measurements 37.2 mg PVMo11-SBA-15 (10 wt. % Mo), 35.0 mg PVMo11-SBA-15 (5 wt.

% Mo), and 34.5 mg PVMo11-SBA-15 (1 wt. % Mo), were used.

Catalytic testing - selective propene oxidation

119

were adjusted with separate mass flow controllers (Bronhorst) to a total flow of 40 ml/min. All gas lines and valves were preheated to 473 K. Hydrocarbons and oxygenated reaction products were analyzed using a Carbowax capillary column connected to an AL2O3/MAPD column or a fused silica restriction (25 m·0.32 mm each) connected to a flame ionization detector. O2, CO, and CO2 were separated using a Hayesep Q (2 m x 1/8``) and a Hayesep T packed column (0.5 m x 1/8``) as precolumns combined with a back flush. For separation, a Hayesep Q packed column (0.5 m x 1/8``) was connected via a molsieve (1.5 m x 1/8``) to a thermal conductivity detector (TCD).

Sample preparation

Silica SBA-15 was prepared according to [22,23] as described in chapter 4.1. PVMo11- SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) was prepared via incipient wetness. The amount of molybdenum was adjusted to 10 wt.%, 5 wt.% Mo, and 1 wt.% Mo. Therefore an aqueous solution of HPOM was used.

120

8.2 Characterization of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo)

Quantification of metal loading XRF

Quantitative analysis of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) were performed to verify the supporting process. Results of quantitative XRF measurements and nominal compositions of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) were summarized in Table 8-1.

Table 8-1: Results of quantitative XRF measurements and nominal composition of PVMo11-SBA- 15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo). elements H P Mo V O Si

PVMo11-SBA-15 nom. wt.% 0.04 0.29 10.00 0.48 50.33 38.85

PVMo11-SBA-15 exp. wt.% - 0.42 9.15 0.42 50.44 39.46

PVMo11-SBA-15 nom. wt.% 0.02 0.15 4.99 0.24 51.79 42.80

PVMo11-SBA-15 exp. wt.% - 0.19 4.37 0.24 51.93 43.28

PVMo11-SBA-15 nom. wt.% 0.00 0.03 1.00 0.05 52.96 45.96

PVMo11-SBA-15 exp.. wt.% - 0.03 0.83 0.04 53.00 46.10

Experimental composition corresponded well to the nominal composition and confirmed a successful supporting process. For simplification of the nomenclature the samples were still denoted as PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo).

Thermal stability of PVMo11 supported on SBA-15 with various metal loading (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo)

Fig. 8-1 depicts the measured thermogravimetric data of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) in 20% O2 in He. The mass loss between 303 K and 373 K was ascribed to desorption of physically adsorbed water on the surface of the supported materials. The loss of adsorb water delayed with higher Mo loading. The absolute mass

121

loss in this temperature range was between 8-12 wt.% for all PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) samples. The samples were not pretreated. Thus, the amount of physically adsorbed water on the surface of the samples may depend on the storage conditions. Afterwards a nearly constant mass between 373 K and 448 K could be detected. The mass showed a loss for PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo) in temperature range 448-523 K. The mass of PVMo11-SBA-15 (1 wt.% Mo) was nearly

100 10 wt.% Mo

98 5 wt.% Mo 1 wt.% Mo 96 94 92 90

NormalizedNormalizedMassMass [%][%] 86 373 473 573 673 773 T [K]

Fig. 8-1: Thermograms of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) at 20% O2 in He. const ant in this temperature range. Subsequently, between 523 K and 823 K the mass slightly decreased for all samples. Pure silica showed a comparable behaviour in vacuum. Silica dehydrated between room temperature and 453 K followed by the dehydroxylation process of silanol groups. This resulted in the formation of siloxane groups.[145,168] Adsorbed water and silanol groups from the support material may possess a structure stabilizing effect on the Keggion ion. This effect would be comparable to that of water of crystallization and constitutional water in bulk HPOM.[117] It has been shown that structural characteristics of supported model systems like MoOx-SBA-15 and VOx-SBA-15 depended mainly on their hydration states and previous calcination processes.[59,95] Therefore, the delayed desorption of physically adsorbed water at higher Mo loadings, may indicative a variable stability of the Keggin ion.

122

Structure of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo)

3 Fig. 8-2 shows the theoretical and experimental Mo K edge FT(χ(k)·k ) of PVMo11-SBA- 15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo). The shapes of the FT(χ(k)·k3) resembled that of bulk PVMo11 indicating similar local structure around the Mo centers in supported and unsupported HPOM Keggin structure. For a more detailed structural analysis, the

H3[PMo12O40] Keggin structure was chosen as model structure. A comparison of the

0.25

0.20 PVMo11-SBA-15 (10wt.% Mo)

0.15

) 3

χ(k)·k 0.10 PVMo11-SBA-15 FT( (5wt.% Mo) 0.05

0.00 PVMo11-SBA-15 (1wt.% Mo) -0.05 0 1 2 3 4 5 6 R [Å]

3 Fig. 8-2: Theoretical (dotted) and experimental (solid) Mo K edge FT(χ(k)·k ) of PVMo11 supported on SBA-15 (10wt.% Mo, 5wt.% Mo, 1wt.% Mo).

2 distances R and disorder parameters σ of PVMo11 supported on SBA-15 with different Mo loadings (10 wt.% Mo, 5 wt.% Mo) exhibited no significant differences between the initial

Keggin ion structure and Keggin ions supported on SBA-15. PVMo11-SBA-15 (1 wt.%

Mo) exhibited increased 2nd and 3rd Mo-O distances compared to PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo).This distances corresponded to the edge- and corner-sharing octahedral [MoO6] units. Additionally, the 2nd Mo-Mo distance was slightly increased confirming the increased Mo-O distances. The various Mo-O and Mo-Mo distances may indicate a slightly different binding state compared to PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo) possibly orginated from a stronger interaction of the Keggin ions with the

123

support material SBA-15. However, the good agreement between theory and experiment for PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo) confirmed the maintained Keggin ion structure upon supporting PVMo11 on SBA-15.[27]

Table 8-2: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in as prepared PVMo11-SBA-15 (10wt.% Mo, 5wt.% Mo, 1wt.% Mo). Experimental parameters were obtained from a refinement of H3[PMo12O40] model structure (ICSD 209 [14,93]) to the experimental Mo K edge XAFS χ(k) of PVMo11-SBA-15 (10wt.% Mo, 5wt.% Mo, 1wt.% -1 Mo) (k range from 3.0-13.7 Å , R range from 0.9 to 4.0 Å, E0 = ~ 1.7, residuals ~13.3-18.0 Nind =

22, Nfree = 9). Subscript c indicates parameters that were correlated in the refinement.

PVMo11-SBA- PVMo11-SBA-15 PVMo11-SBA-15 Keggin model 15 (10wt.% (5wt.% Mo) (1wt.%Mo) N R(Å) R(Å)Mo) Mo)σ2(Å 2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 1 1.68 1.64 0.0035 1.65 0.0036 1.66 0.0004 Mo-O 2 1.91 1.78 0.0035 c 1.78 0.0051c 1.82 c 0.0030c

Mo-O 2 1.92 1.95 0.0035c 1.96 0.0051c 1.99c 0.0030c Mo-O 1 2.43 2.40 0.0006 2.39 0.0010 2.38 0.0014

Mo-Mo 2 3.42 3.43 0.0057c 3.43 0.0063c 3.42 0.0066c

Mo-Mo 2 3.71 3.73 0.0057c 3.72 0.0063c 3.74 0.0066c

8.3 Structural evolution of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) under catalytic conditions

PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) samples were investigated by in situ XAS under catalytic conditions. Fig. 8-3 shows the evolution of molybdenum

XANES spectra of PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo) samples during temperature-programmed treatment in 5% propene and 5% oxygen. The resulting spectra exhibited an increasing pre-edge peak with higher temperatures. The pre-edge peak features in the Mo K edge XANES spectra can be employed to elucidate the local structure around the Mo center. This increasing pre-edge peak correspond to structural changes from octahedral [MoO6] species to partial tetrahedral [MoO4] species. The evolution of the ratio of tetrahedral [MoO4] to octahedral [MoO6] units of PVMo11-SBA-15 (10 wt.% Mo, 5

124

absorption absorption

Normalized Normalized Normalized 673 673 573 20.15 573 20.15 473 20.10 473 20.10 20.05 20.05 T [K] 373 20 T [K] 373 20 Photon energy Photon energy [keV] [keV]

Fig. 8-3: in situ Mo K edge XANES spectra of (left) PVMo11-SBA-15 (5 wt.% Mo) and (right)

PVMo11-SBA-15 (1 wt.% Mo) during temperature-programmed treatment in 5% propene and 5% oxygen in helium in a temperature range between 300 K and 723 K.

wt.% Mo, and 1 wt.% Mo) based on a linear combination of bulk MoO3 and bulk

Na2MoO4 (cf. chapter 7.4) during propene oxidation conditions was shown in Fig. 8-4. In contrast to P(V,W)xMo12-x-SBA-15 (x = 1, 2) (cf. chapter 5.2, 6.2) with identical Mo loading and to PMo12-SBA-15 (10, 14, 19 nm) (cf. chapter 7.4) with tailored pore radii, the onset temperatures the structural changes decreased with lower Mo loading. Structural changes of PVMo11-SBA-15 (1 wt.% Mo) could be detected above 400 K and of PVMo11-

SBA-15 (5 wt.% Mo) above 425 K. Apparently, the Keggin ions in PVMo11-SBA-15 (10

] ratio [%] ratio ] 6

]/[MoO 4

20 [MoO

300 400 500 600 700 T [K]

Fig. 8-4: Quantification of the tetrahedral MoO4 ratio of ( )PVMo11-SBA-15 (10 wt.% Mo),

( )PVMo11-SBA-15 (5 wt.% Mo), and ( ) PMo11-SBA-15 (1 wt.% Mo) during thermal treatment under propene oxidation conditions.

125

wt.% Mo) were stable on silica SBA-15 in the temperature range 303-448 K. Hence, the stability of the Keggin ion increased with loadings from 1 wt. Mo % to 10 wt. Mo %.

Subsequently, concentration of tetrahedral [MoO4] units for all PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) considerably increased with higher temperature.

However, the temperatures of the structural evolution to mainly tetrahedral [MoO4] units correlated to the dehydration and dehydroxylation process of SiO2 under oxidizing conditions (20% O2 in He). Therefore, dehydroxylation was the driving force for the structural rearrangement of the PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo) samples under catalytic conditions. This stability seemed to depend only on the nature of the support material comparable to PMo12-SBA-15 (10, 14, 19 nm) (cf. chapter 7.4).

The structural rearrangement finished at ~ 550 K (PVMo11-SBA-15 (1 wt.% Mo)) and 598

K (PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo). Hence, higher metal loadings resulted in increased temperatures, where the structural change were finished. Subsequently, the concentration of tetrahedral [MoO4] units for PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo,

1 wt.% Mo) was constant and reached the highest concentration with ~75% [MoO4] units at 723 K. PVMo11-SBA-15 (10 wt.% Mo) reach the highest concentration with ~50%

[MoO4] units at 723 K. Therefore, an increasing metal loading may lead to a decreased amount of tetrahedral [MoO4] units.

Influence of metal loading to the resulting structure of [MoxOy] species

3 Fig. 8-5 shows the Mo K edge FT(χ(k)·k ) of activated PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo) after thermal treatment under propene oxidation conditions. The 3 FT(χ(k)·k ) of act. PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo) exhibited features similar to that of dehydrated molybdenum oxides and HPOM supported on SBA-

15.[27,59] For a more detailed structural analysis hexagonal MoO3 was chosen as model structure. Theoretical XAFS phases and amplitudes were calculated for Mo-O and Mo-Mo distances and used for EXAFS refinement. The results of the refinement are shown in Fig. 3 8-5. The first peak in the Mo K edge FT(χ(k)·k ) of act. activated PVMo11-SBA-15 (10 wt.% Mo) exhibited differences compared to that of act. activated PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo). The first peak in the FT(χ(k)·k3) originated mainly from the tetrahedral [MoO4] species on the SBA-15 support and could be sufficiently simulated

126

0.35 act. PVMo12-SBA-15 (10wt.% Mo) 0.30 0.25

act. PVMo -SBA-15

) 12 3 0.20 (5wt.% Mo)

0.15 χ(k)·k

FT( 0.10 act. PVMo12-SBA-15 0.05 (1wt.% Mo) 0.00 -0.05

0 1 2 3 4 5 6 R [Å] 3 Fig. 8-5: Theoretical (dotted) and experimental (solid) Mo K edge FT(χ(k)·k ) of activated PVMo11 supported on SBA-15 (10wt.% Mo, 5wt.% Mo, 1wt.% Mo). using four Mo-O distances. These four distances sufficiently accounted for the minor 2 2 amount of octahedral [MoO6] species. The 1st and 2nd disorder parameters (1st-σ , 2nd-σ ) were higher for act. PVMo11-SBA-15 (10 wt.% Mo) and indicated a decreasing amount of 2 tetrahedral [MoO4] units. Additionally, the 4th disorder parameter (4th-σ ) was smaller than the disorder parameters for act. activated PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo) with lower metal loading. This disorder parameter mainly represented the fraction of octahedral [MoO6] species. Hence, the reduced disorder parameter indicated an increasing amount of octahedral structural motifs in act. activated PVMo11-SBA-15 (10 wt.% Mo) compared to act. PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo) and comparable to act. 3 PMo12-SBA-15 (14, 19 nm) (cf. chapter 7.3). The distinct peak at ~3 Å in the FT(χ(k)·k ) indicated the formation of dimeric or oligomeric [MoOx] units on SBA-15 independent of metal loading. The obtained Mo-Mo distances were comparable for act. PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo) samples and largely independent of the metal loading.

Additionaly, the disorder parameters of the Mo-Mo distances were lower for act. PVMo11-

SBA-15 (5 wt.% Mo) compared to act. PVMo11-SBA-15 (10 wt.% Mo) indicating a lower oligomerization degree for the [MoxOy] species in act. PVMo11-SBA-15 (5 wt.% Mo). The third Mo-Mo distance for act. PVMo11-SBA-15 (1 wt.% Mo) was increased compared to

127

Table 8-3: Type and number (N), and XAFS disorder parameters (σ2) of atoms at distance R from the Mo atoms in act. PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo). Experimental parameters were obtained from a refinement of a hexagonal MoO3 model structure (ICSD 75417

[135]) to the experimental Mo K edge XAFS χ(k) of act. PMo12-SBA-15 (10, 14, 19 nm) (k range -1 from 3.6-16.0 Å , R range from 0.9 to 4.0 Å, E0 = ~ -5.2, residuals ~12 Nind = 26, Nfree = 12). Subscript c indicates parameters that were correlated in the refinement.

hex-MoO3 act. PVMo11-SBA- act. PVMo11-SBA- act. PVMo11-SBA- model 15 (10wt.% Mo) 15 (5wt.% Mo) 15 (1wt.%Mo) N R(Å) R(Å) σ2(Å2) R(Å) σ2(Å2) R(Å) σ2(Å2) Mo-O 2 1.67 1.67 0.0012 1.67 0.0007 1.66 0.0007

Mo-O 2 1.96 1.89 0.0034c 1.87 0.0024c 1.86 0.0029c

Mo-O 1 2.20 2.18 0.0034c 2.18 0.0024c 2.17 0.0029c Mo-O 1 2.38 2.36 0.0014 2.34 0.0024 2.35 0.0026

Mo-Mo 2 3.31 3.50 0.0066c 3.49 0.0049c 3.49 0.0051c

Mo-Mo 2 3.73 3.63 0.0066c 3.62 0.0049c 3.64 0.0051c Mo-Mo 2 4.03 3.75 0.0100 3.75 0.0087 3.79 0.0118

act. PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo). This indicated a slightly different binding state of Mo species on silica SBA-15 with lower metal loading. Apparently, the degree of oligomerization of Mo of the [MoxOy] species reached a minimum on the samples with lower metal loadings. This was comparable to PMo12-SBA-15 (14, 19 nm) samples with large pores (cf. chapter 7.3). Additionally, the act. PVMo11-SBA-15 (1 wt.% Mo) showed a slight different binding state compared to samples with higher metal loading. Therefore, the Mo-O distances and disorder parameters for act. PVMo11-SBA-15 (5 wt.% Mo) and act. PVMo11-SBA-15 (1 wt.% Mo) were nearly identical and different from those of act.

PVMo11-SBA-15 (10 wt.% Mo). The comparable Mo-O distances and disorder parameters for act. PVMo11-SBA-15 (5 wt.% Mo) and act. PVMo11-SBA-15 (1 wt.% Mo) confirmed the results of the quantification of tetrahedral [MoO4] and distorted [MoO6] units. The quantification resulted in ~70% tetrahedral [MoO4] for both act. PVMo11-SBA-15 (5 wt.%

Mo) and act. PVMo11-SBA-15 (1 wt.% Mo) present under catalytic conditions. A comparable quantification was found for molybdenum oxide supported on SBA-15 with a

Mo loading of 5.5wt.%.[59] The slightly increased Mo-Mo distances in act. PVMo11-SBA- 15 (1 wt.% Mo) may indicate a further decreased degree of oligomerization. Therefore, the

128

formation of dimeric or oligomeric [MoxOy] units mostly consisting of tetrahedral [MoO4] units depended on metal loading on silica SBA-15. Hence, act. PVMo11-SBA-15 (10 wt.% Mo) favored the formation of more extended structures on the support material due to higher metal loading.

8.4 Functional characterization of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, and 1 wt.% Mo)

8.4.1 Reducibility

Fig. 8-6 shows the H2 TPR profiles of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.%

Mo). The resulted H2 TPR profiles of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo) revealed one sharp reduction peak (~800 K) and a very broad signal after the sharp peak comparable to PVxMo12-x-SBA-15 (x = 0, 1, 2) (cf. chapter 5.3.1). The H2 consumption for

PVMo11-SBA-15 (5 wt.% Mo) was due to lower metal loading compared to PVMo11-SBA-

15 (10 wt.% Mo). Additionally, the peak height of the sharp peak decreased for PVMo11-

SBA-15 (5 wt.% Mo) compared to the broaded signal above ~800 K. The H2 TPR profile of PVMo11-SBA-15 (1 wt.% Mo) exhibited only a broad signal above 748 K. The H2 TPR profiles of PVMo11-SBA-15 (10 wt.% Mo) were comparable to molybdenum oxides supported on SBA-15 with Mo loadings between 9.5 wt.% and 13.3 wt.%.[149,150] Lou et al. assigned the sharp reduction peak to oligomeric MoOx species or small MoOx clusters.

H2 TPR of PVMo11-SBA-15 (10 wt.% Mo) with the reduced sharp peak compared to the broaded signal above ~800 K indicated a decreased degree of oligomerization comparable to molybdenum oxide supported on SBA-15 with a Mo loading of 6.6 wt.%.[149] H2 TPR profile of PVMo11-SBA-15 (1 wt.% Mo) was different from H2 TPR profiles of supported molybdenum oxides with low Mo loadings (~2 wt.%). Typically, molybdenum oxides with Mo loadings below 5 wt.% exhibited a shift of the reduction temperature to higher temperatures (> 1000 K) indicating the reduction of predominantly monomeric MoOx species.[149,150] Therefore, the broad signal above 748 K may corresponded dimeric or oligomeric [(V,Mo)xOy] species. This species were lower oligomerized than the PVMo11-

SBA-15 (10 wt.% Mo, 5 wt.% Mo) samples. Apparently, mostly of the triads ([(V,Mo)xOy] species) of the initial Keggin ion persisted during temperature programmed reduction.

129

10 wt.% Mo 5 wt.% Mo

1 wt.% Mo

consumption

2 2 H

373 473 573 673 773 873 973 T [K]

Fig. 8-6: Temperature programmed reduction (H2 TPR) of PVMo11-SBA-15 (10 wt.% Mo),

PVMo11-SBA-15 (5 wt.% Mo), and PVMo11-SBA-15 (1 wt.% Mo) measured at a heating rate of 8 -1 Kmin 5% H2 in Ar.

Comparing the typical reduction temperatures of supported molybdenum oxides and

PVMo11-SBA-15 with different metal loading, a slightly higher degree of oligomerization for PVMo11-SBA-15 with higher metal loading was determined.[150] Therefore, the decreased height of the sharp peak compared to the broad signal above ~800 K PVMo11-

SBA-15 (5 wt.% Mo) indicated a decreased oligomeric [(V,Mo)xOy] species. Apparently, the degree of oligomerization reached a minimum for PVMo11-SBA-15 (1 wt.% Mo), indicated oligomerized [(V,Mo)xOy] species. In contrast to that, supported molybdenum oxide synthesized from a AHM precursor lead to predominantly monomeric MoOx species with Mo loadings below 2wt.%.[149,150]

8.4.2 Influence of the resulting structure on catalytic activity

Reaction rates and selectivities of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo) in propene oxidation at 723 K are shown in Fig. 8-7. The reaction rates of PVMo11-SBA- 15 (5 wt.% Mo, 1 wt.%) were measured under various propene conversion conditions. The propene conversion was varied to achieve a constant sample volume and to exclude thermal effects. Therefore, PVMo11-SBA-15 (10 wt.% Mo) was mechanically diluted with

130

SBA-15 to concentrations of 5 wt.% Mo and 1 wt.% Mo to achieve sample volumes comparable to PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo).

The reaction rate of PVMo11-SBA-15 (10 wt.% Mo) (35.6 µmol/g(Mo)s) at isoconversional conditions was slightly increased compared to that of PVMo11-SBA-15 (5 wt.% Mo) (34.1 µmol/g(Mo)s). The oxidation product distributions were comparable for

PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo) samples. The reaction rate for PVMo11-SBA- 15 (10 wt.% Mo) (29.9 µmol/g(Mo)s) at isoconversional conditions was also slightly increased compared to that of PVMo11-SBA-15 (1 wt.% Mo) (28.8 µmol/g(Mo)s). The

acrylic acid acetone CO

] 1 acetic acid propionaldehyd CO2 - s acrolein acetaldehyd 100 e 55 (Mo)

e 1 -

50 g

80 45 40

60 35 ty ty [%]

30 (propene)

40 25 µmol

20 [

Selectivi 15 20 10 5 0 0

a b c d e reaction rate propene conversion 14.1% 8.3% 7.7% 5.2% 3.7%

Fig. 8-7: Reaction rate (µmol(propene)g-1(Mo)s-1) and selectivity at different propene conversions of (a) PVMo11-SBA-15 (10 wt.% Mo), (b) PVMo11-SBA-15 (10 wt.% Mo), (c) PVMo11-SBA-15

(5 wt.% Mo), (d) PVMo11-SBA-15 (10 wt.% Mo), and (e) PVMo11-SBA-15 (10 wt.% Mo) in 5% propene and 5% oxygen in He at 723 K.

product distribution of PVMo11-SBA-15 (10 wt.% Mo) was different from that of PVMo11- SBA-15 (1 wt.% Mo). Selectivities towards acrolein and propionaldehyd increased and the amount of total oxidation products decreased for PVMo11-SBA-15 (1 wt.% Mo) compared to PVMo11-SBA-15 (10 wt.% Mo). Hence, an influence of the Mo loading on the catalytic properties may be assumed. Thus, samples with lower Mo loading exhibited a decreased catalytic activity comparable to PMo12-SBA-15 (10 nm, 14 nm, 20 nm) (cf. chapter 7.5.1). This indicated a lower degree of oligomerization for samples with lower Mo loadings

131

confirming the results of the XAS analysis and TPR of act. PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo). Nevertheless the catalytic activity in propene oxidation increased with higher degree of oligomerization of the [(V,Mo)xOy] units under catalytic conditions.

Compared to PVMo11-SBA-15 (5 wt.% Mo) [(V,Mo)xOy] units orginating from PVMo11- SBA-15 (10 wt.% Mo) resulted in an enhanced catalytic activity without significant influence on the product distribution. The different product distribution of PVMo11-SBA-

15 (1 wt.% Mo) may result from slightly different binding states of the [(V,Mo)xOy] units.

The [(V,Mo)xOy] species resulting from PVMo11-SBA-15 (1 wt.% Mo) lead to an increased selectivity towards acrolein and propionaldehyd. Grasseli et al. discussed the role of site isolation i.e. the spatial separation of active sites on the surface of a heterogenous catalyst.[170,172] A comparable effect could be responsible for the enhanced selectivity towards partial oxidation products in act. PVMo11-SBA-15 (1 wt.% Mo).

Hence, may be assumed, that the [(V,Mo)xOy] species resulting in act. PVMo11-SBA-15 (1 wt.% Mo) improved the favorable number of active oxygens. Apparently, this active oxygen species seemed to be bridged M-O-M (M = V, Mo) oxygen. Isolated [VOx] and

[MoOx] species would result mostly in total combustion (~ 90%). [137,153]

132

8.5 Summary

Structural evolution of H4[PVMo11O40] supported on SBA-15 (PVMo11-SBA-15) with various Mo loadings (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo) was examined by in situ X-ray absorption spectroscopy investigations at the Mo K edge during propene oxidation conditions. Supporting heteropolyoxo molybdates on SBA-15 with different Mo loadings resulted in regular Keggin ions on the support material. During thermal treatment in propene oxidation conditions the molybdenum oxide species on SBA-15 formed a mixture of octahedral [MoO6] and mostly tetrahedral [MoO4] units. The ratio of the tetrahedral

[MoO4] to octahedral [MoO6] units increased with lower Mo loading. The onset temperature of structural changes of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo, 1 wt.% Mo) during thermal treatment in propene oxidation conditions increased with higher Mo loading. A delayed desorption of physically adsorbed water in PVMo11-SBA-15 with higher Mo loading was indicative of a varied stability of the Keggin ion. Hence, desorption of physically adsorbed water and dehydroxylation of silanol groups of the support material were the driving forces for the structural instability of the Keggin ion. The instability lead to the formation of act.PVMo11-SBA-15. The resulting [(V,Mo)xOy] structures present under catalysis conditions depended on Mo loading. A higher concentration of octahedral

[MoO6] units together with higher oligomerized [(V,Mo)xOy] species could be detected for higher Mo loading. The higher oligomerized [(V,Mo)xOy] species present in act. PVMo11- SBA-15 (10 wt.% Mo) showed a slight increased catalytic activity compared to lower oligomerized [(V,Mo)xOy] species in act. PVMo11-SBA-15 (5 wt.% Mo, 1 wt.% Mo). The selectivities towards oxidation products during propene oxidation conditions were comparable and independent of the Mo loading (10 wt.% Mo, 5 wt.% Mo) indicating the same active sites in act. PVMo11-SBA-15. The product distribution for PVMo11-SBA-15

(1 wt.% Mo) was different from those of PVMo11-SBA-15 (10 wt.% Mo, 5 wt.% Mo). Selectivities towards acrolein and propionaldehyd increased and total oxidation products decreased for PVMo11-SBA-15 (1 wt.% Mo) compared to PVMo11-SBA-15 (10 wt.% Mo,

5 wt.% Mo). The different product distribution of PVMo11-SBA-15 (1 wt.% Mo) may result from a slightly different binding state of the [(V,Mo)xOy] species. This different binding state of the [(V,Mo)xOy] species may lead to an increased selectivity towards partial oxidation products.

133

9 General discussion and Summary

9.1 Structure directing effect of the support material

SBA-15 as support material had a significant influence on the structures of supported species that formed during thermal treatment. Generally, the structures of model systems such as MoOx-SBA-15, VOx-SBA-15, and WOx depend on their hydration states.[17,59,95,144,149,173,174] SBA-15 or rather SiO2 adsorbed water at ambient conditions. This water is removed at temperatures above 423 K.[168,175] In this work HPOM were supported via incipient wetness with an aqueous solution on the support material (cf. Chapter 4.1). Therefore, it may be assumed that the HPOM were incorporated in a matrix of adsorbed water comparable to higher hydrates of bulk HPOM or to an aqueous solution. The removal of adsorbed water and the following dehydroxylation of silanol groups during thermal treatment lead to a destabilizing effect on the Keggin ion.[145] This effect would be comparable to that of the removal of water of crystallization and constitutional water in bulk HPOM. The removal of constitutional water in bulk

HPOM leads to a decomposition resulting in MoO3 during thermal treatment.[116] TG measurement of PVMo11-SBA-15 showed a mass loss of about 2 wt.% between 473 and 673 K during oxidizing conditions indicating the dehydroxylation of silanol groups (c.f. Chapter 8.2; Fig. 8-1). The destabilizing effect on the Keggin ion was independent of the addenda atoms (V,W) in P(V,W)x-SBA-15 (x = 0, 1, 2), the pore radii of the SBA-15, and HPOM loading.

The structures forming during thermal treatment in propene oxidation conditions depended on the nature of SiO2, pore radii of the SBA-15, and HPOM loading. According to Wachs et al. the structure of supported metal oxides is correlated to the net pH at the point of zero charge (pzc) of the oxide support.[166] SiO2 has a pH of 3.9 at pzc. Therefore, the low pH at pzc of SiO2 leads to mainly linked M-O-M (M = Mo, V, W) species corresponding to the behaviour of molybdates, vanadates, and wolframates in acidic solutions.[37,148,176] Compared to SBA-15, other support materials exhibit different structure directing effects depending on the acidity of the surface.[69–71] For instance, mainly isolated [MoO4] and

[VO4] units existed on an alkaline MgO support in agreement with the behaviour of molybdates and vanadates in alkaline solution.[137,153]

134

The pore radius also had a significant influence on the structures formed during thermal treatment under propene oxidation conditions. Fig. 9-1 depicts a two dimensional schematic representation of pores with different radii. Squares in the pores represent the

Keggin ions. The radius (r1) of the smaller pore is half of the pore radius of the second

D1 < D2

r1 2r1 = r2

Fig. 9-1: Two dimensional schematic representation of pores with squares representing the Keggin ions.

pore (r2 = 2r1). Doubling of the pore radius lead to a doubling of the circumference of the pore and the number of squares representing the Keggin ions. Nevertheless, the effective distance D between the squares is clearly smaller in the smaller pore (D1) than in the larger pore (D2). Therefore, thermal treatment under propene oxidation conditions lead to lower oligomerized [MoOx] species and an increased [MoO4]/[MoO6] ratio at smaller pore radii as shown in Chapter 7.

The HPOM loading has a further influence on the structures formed during thermal treatment under propene oxidation conditions (Chapter 8). This effect was comparable to the effect of SBA-15 with different pore radii. The surface coverage for samples with lower HPOM loading was decreased resulting in less extended species on the support material. Hence, the degree of oligomerization of the [MoOx] species was decreased and the [MoO4]/[MoO6] ratio was increased at lower HPOM loading. This decreasing degree of oligomerization at lower metal loading has been shown for various supported metal oxides.[59,95,137,149,153,177,178]

135

9.2 Structure directing effects of the addenda atoms

Both bulk HPOM and supported HPOM exhibited a structure directing effect of addenda atoms. The structures forming during thermal treatment under oxidizing and propene oxidation conditions for bulk HPOM depended on the degree of substitution and type of addenda atoms (V, W). Bulk HPOM decompose during thermal treatment under oxidizing conditions resulting in MoO3. The loss of constitutional water depended also on the degree of substitution and type of addenda atoms (V, W). Vanadium substitution lead to decreased decomposition temperatures of 573 K for PV2Mo10, 623 K for PVMo11, and 673 K for unsubstituted PMo12 K according to the literature.[35,179] Tungsten substitution had no influence on the decomposition temperatures resulting in the loss of constitutional water.

Subsequently, the HPOM without constitutional water decomposed to MoO3. The resulting modifications were α-MoO3 and β-MoO3. Vanadium substituted HPOM (PVxMo12-x x = 1,

2) favored the formation of α-MoO3 and tungsten substituted HPOM (PWxMo12-x (x = 1,

2)) favored the formation of β-MoO3 (c.f. Chapter 3.5). An explanation of the structure directing effect are the different charges and ion radii of V5+, W6+, and Mo6+ resulting in the edge-shared structure of α-MoO3 for (PVxMo12-x (x = 1, 2)) and corner-shared structure of β-MoO3 for PWxMo12-x x = 1, 2 according to Pauling`s rules.[122]

During thermal treatment under propene oxidation conditions, bulk HPOM (PVxMo12-x x = 0, 1, 2) decomposed to various structures depending on the type of addenda atoms (V, W).

PMo12 decomposed to α-MoO3 as the thermodynamically stable modification of molybdenum oxides in their highest oxidation state (+ 6). PV2Mo10 decomposed during thermal treatment in propene oxidation conditions at 723 K into a mixture of various structures. Ressler et al. performed in situ XAS measurements of PVMo11 during propene oxidation conditions.[14] In this process Mo cations migrate on extra Keggin sites while remaining coordinated to the resulting lacunary Keggin anion.[13] Driving force for the formation of lacunary Keggin anions may be the relaxation of the Keggin structure at elevated temperature upon removal of structural water. These structural changes at temperatures above 573 K are accompanied by reduction of the molybdenum centers.[13,14] Subsequently, the reduced Mo centers reoxidized upon 723 K.[13,14] In another study on PV2Mo10 Ressler et al. described a dynamic behaviour by isothermally switching from propene (reducing) to oxidizing (propene and oxygen) and back to propene (reducing) conditions at 723 K. The results of the in situ XAS experiment revealed for the

136

reduced state the formation of a short vanadium-molybdenum distance of about 2.8 Å. The oxidized state exhibited a longer distance of the vanadium center to an extra-Keggin molybdenum center at 3.2 Å.[16] The results suggested a mixture of at least two sites around two different V centers.[16] Therefore, it may be assumed, that the structures formed during treatment of PV2Mo10 under propene oxidation conditions corresponded to a mixture of various structures comparable to lacunary Keggin ions or Keggin ions.

PW2Mo10 decomposed during thermal treatment in propene oxidation conditions at 723 K to a mixture of α-MoO3 and Mo17O47.[41,44] Molybdenum in Mo17O47 has an average valence of ~ +5.5 which indicated, that the degree of reduction was higher for PW2Mo10 than for PMo12. Hence, tungsten lead to an increased reducibility of PW2Mo10 during propene oxidation conditions.

The addenda atoms (V, W) in substituted HPOM supported on SBA-15 (P(V,W)xMo12-x- SBA-15 (x = 0, 1, 2)) exhibited also a structure directing effect on the structures forming during thermal treatment in propene oxidation conditions at 723 K. Vanadium lead to an increasing [MoO4]/[MoO6] ratio in contrast to tungsten substituted HPOM supported on SBA-15 (Fig. 9-2). The typical structure resulting for dehydrated molybdenum oxides supported on SiO2 was a mixture of tetrahedral [MoO4] and octahedral [MoO6] units.[27,59,149] Dehydrated vanadium oxides supported on SiO2 lead to the formation of predominantly [VO4] units.[17,25,95] Therefore, it may be assumed, that during the decomposition process, neighboring [MoO6] units were influenced by [VO6] units of the

PV2Mo10-SBA-15

PVMo11-SBA-15 PMo12-SBA-15 60 PWMo11-SBA-15

] ratio [%] ratio ] PW2Mo10-SBA-15

6 ]/[MoO

4 40 [MoO

Fig. 9-2: [MoO4]/[MoO6] ratio of P(V,W)xMo12-SBA-15 (x = 0, 1, 2) during thermal treatment under propene oxidation condition (5% propene + 5% O2 in He) at 723 K.

137

initial Keggin ion structure. This influence lead to tetrahedral [MoO4] and [VO4] units during thermal treatment under propene oxidation conditions (c.f. Chapter 5.2). Hence, the formation of [MoO4] units depended on the degree of vanadium substitution.

Dehydrated tungsten oxides or H3[PW12O40] supported on SiO2 corresponded to a Si containing Keggin type cluster with corner- and edge-shared [WO6] units on the support material.[58,162,180] Therefore, additional [WO6] species in PWxMo12-x-SBA-15 (x = 1, 2) influenced the Mo species of the initial Keggin ion structure resulting in predominantly

[MoO6] and [WO6] units during thermal treatment under propene oxidation conditions (c.f. Chapter 6.2). Both in vanadium and in tungsten substituted supported HPOM

(P(V,W)xMo12-x-SBA-15 (x = 1, 2)) the resulting [MOx] (M = V, W) units were in close vicinity to the [MoOx] species.

9.3 Structure activity relationships

The different structures forming under catalytic conditions (5% propene + 5% oxygen in

He at 723 K) for bulk P(V,W)xMo12-x (x = 0, 1, 2) depended on the substituted element (V, W). The different structures forming during catalytic conditions correlated with the different catalytic behaviours of P(V,W)xMo12-x ( x = 0, 1, 2). Fig. 9-3 depicts the resulting reaction rates and selectivities towards C3 oxidation products and CO, CO2. Both vanadium substituted PVxMo12-x (x = 1, 2) and tungsten substituted PWMo12-x (x = 1, 2) showed increased reaction rates compared to unsubstituted PMo12. The significant differences between the samples were the resulting structures. α-MoO3 resulting from

PMo12 showed the lowest catalytic activity in propene oxidation. The lacunary Keggin and

Keggin ion resulting from PV2Mo10 and the mixture of α-MoO3 and Mo17O47 resulting from PW2Mo10 showed an increased catalytic activity during propene oxidation. Therefore, it may be assumed that the different catalytic activities of the various structures depended on the type of addenda atom (V, W) and the degree of substitution. The resulting structures for the substituted HPOM indicated also a lower average valence of molybdenum for the lacunary Keggin and Keggin ion resulting from PV2Mo10 and the mixture of α-MoO3 and

Mo17O47 resulting from PW2Mo10 (c.f. Chapter 3.5.2).

138

13 80

1 12 - 70 11

(Mo)s 60

1 10 CO+CO2 - g 9 55 50 8 45

7 40

reaction rate rate reaction Selectivity [%] Selectivity (propene) 6 35 30 C3 oxidation

µmol 5 25 products 4 20 PV2Mo10 PMo12 PW2Mo10 PV2Mo10 PMo12 PW2Mo10

PVMo11 PWMo11 PVMo11 PWMo11

Fig. 9-3: (left) Reaction rates (µmol(propene)/g(Mo)) and (right) selectivities towards C3 oxidation products and CO+CO2 of bulk P(V,W)xMo12-x (x = 0, 1, 2) in 5% propene and 5% oxygen in He at 723 K.

Vanadium centers in substituted HPOM (PVxMo12-x (x = 1, 2)) can reversibly change their oxidation state from V5+ to V4+ without significant destabilization of the lacunary Keggin or intact Keggin ion.[15,16] Additionally, reduced V4+ has an ion radius of 72 pm which is comparable to that of Mo6+ (74 pm). The similar ion radius of V4+ may stabilize partially reduced intermediates.[103,181] Ressler et al. described in various XANES studies, that the Mo centers in bulk HPOM reduced between 573 K and 723 K to an average valence of ~ 5.85 during propene oxidation conditions. Subsequently, the Mo centers reoxidized above 723 K.[13–16] The ion radii of reduced Mo5+ centers (75 pm) and W5+ centers (76 pm) were larger than the ion radius of V4+ (72 pm) [103,181]. In particular at elevated temperatures the larger ion radii of reduced Mo5+ and W5+ centers destabilized the Keggin ion structure resulting in α-MoO3 or the corresponding partially reduced Mo17O47. Therefore, the reduction process during propene oxidation conditions above 573 K may lead to a destabilization of the lacunary Keggin ions or Keggin ions. Hence, the increased catalytic activity may result from the significantly different structures of PVxMo12-x

(x = 1, 2) in contrast to PMo12 (Fig. 9-3). The partial reduced Mo17O47 phase resulting from

PWxMo12-x (x = 1, 2) was stabilized by tungsten. Tungsten centers are able to occupy up to

30% of the sites of molybdenum atoms on Mo17O47.[182] Therefore, the formation of a mixture of α-MoO3 and Mo17O47 exhibited more reduced Mo centers than

139

α-MoO3 resulting from unsubstituted PMo12. This partially reduced mixture of α-MoO3 and Mo17O47 may lead to the enhanced catalytic activity.

The different structures and average valences of Mo resulting from P(V,W)xMo12-x (x = 0, 1, 2) may also lead to different selectivities (Fig. 9-3). Both structure and average valence of the resulting compounds have an influence on the product distribution.[2,4,183] Therefore, explaining the various selectivities was hindered by simultaneously varying composition and structure. Reduced metal centers may be particularly effective in activating gas-phase oxygen, resulting in an oversupply of surface-bond electrophilic oxygen. Generally, oxidation reactions proceed in two steps. The first step is the reduction of the catalyst with propene at the expense of lattice oxygen. The second step is the reoxidation of the catalyst with gas phase oxygen.[4] The reoxidation process is described by a series of reactions starting with adsorption of gas phase oxygen and eventually leading to the formation of lattice oxygen. If the reoxidation process is slow, propene may be attacked by nonselective oxygen species.[4] The reduced metal centers indicated, that the reaction rate of the reduction process was higher compared to that of the oxidation process. Hence, surface-bond electrophilic oxygen is prone to further oxidize propene or acrolein to

CO2.[4,16,184] Therefore, it may be assumed, that reduced metal centers were responsible for the decreased selectivity towards C3 oxidation products and increased formation of total oxidation products.

Fig. 9-4 (left) depicts reaction rates as a function of the [MoO4]/[MoO6] ratio resulting for act. P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) and PMo12-SBA-15 (14 nm, 19 nm) (coverage of 2 1 Keggin ion per 13 nm ) during propene oxidation (5% propene + 5% O2 in He; 723 K). The reaction rates were determined for similar propene conversions. The reaction rate decreased with higher [MoO4]/[MoO6] ratio independent of the addenda atoms (V,W) or pore radii of SBA-15. Additionally, the [MoO4]/[MoO6] ratio was an indicator for the degree of oligomerization. In all samples with higher [MoO4]/[MoO6] ratio, a decreased degree of oligomerization was assumed. This result confirmed the assumption, that selective oxidation takes place only in the presence of bridging M-O-M bonds.[4] The increased degree of oligomerization leads to a higher concentration of Mo-O-Mo bonds resulting in an increased reaction rate. The increased reaction rate resulted in increased total combustion (Fig. 9-4). The samples with a higher [MoO4]/[MoO6] ratio and a decreased degree of oligomerization exhibited an increased selectivity towards C3

140

70 2W 1 - 55 2V 2W W CO+CO 2

65 50

(Mo)s 1

- V W 45 g 60 40 55 2V

V 35 reaction rate rate reaction

(propene) 50 30 Selectivity [%] V 2V 25 C3 oxidation

µmol 45 2W W products 20

40 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 90 [MoO4]/[MoO6] ratio [%] [MoO4]/[MoO6] ratio [%]

Fig. 9-4: (left) Reaction rates and (right) selectivities towards C3 oxidation products and CO+CO2 as a function of the [MoO4]/[MoO6] ratio of act. P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2) in 5% propene and 5% oxygen in He at 723 K. The type of addenda atoms (V,W) and degree of substitution (x = 1, 2) are marked. oxidation products. Hence, reaction rates and selectivity towards the desired C3 oxidation products competed with each other. Therefore, it may be assumed that the excess of bridging M-O-M (M = Mo, V, W) lead to overoxidation according to the literature.[2,4,170] Addenda atoms had an additional influence on the reaction rates and selectivities for act.

P(V,W)xMo12-x-SBA-15 (x = 1, 2). It was shown for both act PVxMo12-x-SBA-15 (x = 1, 2) and act. PWxMo12-x-SBA-15 (x = 1, 2), that the reaction rates and selectivities differed from that of references synthesized with individual metal (Mo, V, W) precursors (c.f.

Chapter 5.3.2, 6.3.2). Structural analysis revealed that the resulting [MOx] (M = V, W) species in act. (P(V,W)xMo12-x-SBA-15 (x = 1, 2)) were in close vicinity to the [MoOx] species. Conversely, the [MOx] (M = V, W) species in act. V2Mo10Ox-SBA-15 and act.

W2Mo12Ox-SBA-15 were mostly separated from the [MoOx] species Apparently, the neighboring [MOx] (M = V, W) units and [MoOx] units in act. (P(V,W)xMo12-x-SBA-15 (x

= 1, 2)) resulted in an increased formation of CO and CO2. Various mixed metal oxides exhibited an increased formation of CO and CO2 with higher chemical complexity [185–

187]. Probably, the neighboring [MOx] (M = V, W) units and [MoOx] units were able to activate gas-phase oxygen, resulting in an oversupply of surface-bond electrophilic oxygen. This electrophilic oxygen is prone to further oxidize propene or acrolein to CO2

141

comparable to the effect of particularly reduced metal centers in bulk oxides.[8,118,121]

The references act. V2Mo10Ox-SBA-15 and act. W2Mo12Ox-SBA-15 with mostly separated

[MOx] (M = V, W) and [MoOx] species exhibited an increased selectivity towards C3 oxidation products and a decreased formation of CO and CO2. The neighboring [MOx]

(M = V, W) units and [MoOx] units in act. P(V,W)xMo12-x-SBA-15 (x = 1, 2) with higher chemical complexity exhibit an increased formation of CO2 and CO according to the literature.[185–187]

142

10 Conclusions

Introduction

Understanding structure-activity correlations of functional materials is an important issue in catalysis and materials science. Often model systems are investigated. For elucidating structure activity correlations of model systems for catalytic investigations, a detailed knowledge about structure and chemical composition is indispensable. Thus, various characterization methods are necessary for a sufficient characterization of the catalyst systems. Heteropolyoxomolybdates (HPOM) of the Keggin type exhibit a broad compositional range while maintaining their characteristic structural motifs. Therefore,

H3[PMo12O40] (PMo12), H4[PVMo11O40] (PVMo11), H5[PV2Mo10O40] (PV2Mo10),

H3[PWMo11O40] (PWMo11), and H3[PW2Mo10O40] (PW2Mo10) were synthesized as model catalysts in selective propene oxidation. P(V,W)xMo12-x (x = 0, 1 ,2) were supported on

SBA-15. The initial Keggin structure of P(V,W)xMo12-x (x = 0, 1 ,2) was retained after the supporting process. In situ XAS investigations under propene oxidation conditions of

P(V,W)xMo12-x-SBA-15 (x = 0, 1 ,2) were conducted. Catalytic testing elucidated the functional properties of P(V,W)xMo12-x-SBA-15 (x = 0, 1 ,2) during propene oxidation conditions. A detailed analysis of the structures formed under catalytic conditions was conducted and correlated with the catalytic activity and product distribution towards propene oxidation.

Synthesis of bulk HPOM and supported HPOM

The synthesis of P(V,W)xMo12-x (x = 0, 1, 2) lead to HPOM with the desired Keggin type structure and chemical composition. P(V,W)xMo12-x (x = 0, 1, 2) crystallized as 13 hydrate in a triclinic crystal system. The 13 hydrate structure of the HPOM ensured the incorporation of the addenda atoms (V, W) in the Keggin ion. The volume of the unit cell decreased with higher vanadium substitution because of the smaller ionic radius of V in a six-fold coordination. W with an identical ionic radius compared to Mo had no influence on the volume of the unit cell in contrast to the vanadium substitution. The results of IR and Raman measurements confirmed, that the addenda atoms (V, W) were incorporated in

143

the Keggin ion. The IR and Raman spectra exhibited the typical peaks of the Keggin ion structure. The EXAFS refinements indicated, that the addenda atoms (V, W) were incorporated in the Keggin ion independent of the degree of substitution. XRD and physisorption measurements of the tailored SBA-15 (10 to 19 nm) confirmed a successful synthesis of mesoporous SiO2 materials with different pore size distributions, high specific areas, and the typical pore structure of SBA-15. Supporting P(V,W)xMo12-x (x = 0, 1, 2) on SBA-15 (10 to 20 nm pore radius) via incipient wetness lead to the desired 2 metal loadings (1 to 10 wt. Mo or rather 1 Keggin ion per 130 to 13 nm ). P(V,W)xMo12-x- SBA-15 (x = 0, 1, 2) were sufficiently dispersed on the support material without affecting the pore structure of the support material. The formation of extended crystalline HPOM structures could be excluded.

Structure directing effect of SBA-15

SBA-15 as support material had a significant influence on the structures formed during thermal treatment. SBA-15 or rather SiO2 adsorbed water at ambient conditions. This water is removed at temperatures above 423 K. The removal of adsorbed water and the following dehydroxylation of silanol group lead to a destabilizing effect on the Keggin ion. Therefore, it may be assumed that the HPOM were incorporated in a matrix of physisorbed water comparable to higher hydrates of bulk HPOM or to an aqueous solution. This effect would be comparable to that of the removal of water of crystallization and constitutional water in bulk HPOM. The removal of constitutional water in bulk HPOM results in decomposition and formation of MoO3 during thermal treatment. The destabilizing effect on the Keggin ion was independent of the addenda atoms (V,W) in P(V,W)x-SBA-15 (x = 0, 1, 2), the pore radii of the SBA-15, and HPOM loading. Hence, the stability of Keggin ions supported on SBA-15 was significantly decreased compared to bulk HPOM, with decomposition temperatures between 623 K and 713 K. The resulting structures forming during thermal treatment in propene oxidation conditions depended on the nature of SiO2, pore radii of the SBA-15, and HPOM loading. The structure of supported metal oxides is correlated to the net pH at the point of zero charge

(pzc) of the oxide support. SiO2 have a pH of 3.9 at pzc. Therefore, the low pH at pzc of

SiO2 lead to mainly linked M-O-M (M = Mo, V, W) species corresponding to the behaviour of molybdates, vanadates and wolframates in acidic solutions.

144

The pore radius had also a significant effect on the structures formed during thermal treatment under propene oxidation conditions. A higher concentration of octahedral

[MoO6] units and higher oligomerized [MoxOy] units resulted for act. PMo12-SBA-15

(10 nm) compared to act. PMo12-SBA-15 (14, 19 nm). Enlarging the pore radii lead to an increased effective distances between the Keggin ions. The effective distances were clearly smaller in the smaller pores than in the larger pores. Therefore, the structures forming during thermal treatment under propene oxidation conditions consisted of lower oligomerized [MoOx] species and an increased [MoO4]/[MoO6] ratio with smaller pore radius. Apparently, tailoring the pore radius of silica SBA-15 permitted to prepare Mo oxide model systems to investigate correlations between activity and structure of characteristic oxide species at similar loadings. The HPOM loading had a further influence on the structures forming during thermal treatment under propene oxidation conditions. The surface coverage for samples with lower HPOM loading was decreased resulting in less extended species on the support material. Hence, the degree of oligomerization of the [MoOx] species was decreased and the [MoO4]/[MoO6] ratio was increased with lower HPOM loading. This effect was comparable to the effect resulting for SBA-15 with different pore radii.

Structure directing effect of the addenda atoms

The addenda atoms exhibited structure directing effect in both bulk HPOM and supported HPOM. The structures forming during thermal treatment under oxidizing and propene oxidation conditions for bulk HPOM depended on the degree of substitution and type of addenda atoms (V, W). Bulk HPOM decomposed during thermal treatment under oxidizing conditions resulting in MoO3. The release of constitutional water of the HPOM depended also on the degree of substitution and type of addenda atoms (V, W). Vanadium substitution lead to decreased decomposition temperatures of 573 K for PV2Mo10, of 623 K for PVMo11, and of 673 K for unsubstituted PMo12. Tungsten substitution had no influence on the release of constitutional water of the HPOM. Subsequently, the HPOM without constitutional water decomposed to MoO3. Vanadium substitution in HPOM (PVxMo12-x (x

= 1, 2)) lead to an increased formation of predominantly α-MoO3 depending on the degree of vanadium substitution. Conversely, tungsten substituted PWxMo12-x (x = 1, 2) resulted in an increased formation of β-MoO3 depending on the degree of tungsten substitution. The

145

different charges and ion radii of V5+, W6+, and Mo6+ were responsible for the structure directing effect resulting in the rather edge-shared structure α-MoO3 for (PVxMo12-x (x = 1,

2)) and corner-shared structure β-MoO3 for PWxMo12-x x = 1, 2 according to Pauling`s rules.

Bulk HPOM (PVxMo12-x x = 0, 1, 2) decomposed during thermal treatment in propene oxidation conditions to various structures depending on the type of addenda atoms (V, W). The Mo centers in bulk HPOM partially reduced between 573 K and 723 K to an average valence of ~ 5.85 and reoxidized above 723 K during propene oxidation conditions. The ion radii of reduced Mo5+ centers (75 pm) and W5+ centers (76 pm) were larger than the ion radius of V4+ (72 pm). In particular at elevated temperatures the larger ion radii of reduced 5+ 5+ Mo centers and W centers destabilized the Keggin ion structure. Therefore, PMo12 decomposed to the thermodynamically stable modification of molybdenum oxides, α-

MoO3. PW2Mo10 decomposed during thermal treatment in propene oxidation conditions to a mixture of α-MoO3 and Mo17O47. In contrast to PWxMo12-x (x = 0, 1, 2), PV2Mo10 decomposed during thermal treatment in propene oxidation conditions at 723 K to a mixture of various structures. The ion radius of reduced V4+ (72 pm) was comparable to the ion radius of Mo6+ (74 pm) stabilizing partially reduced intermediates of the initial Keggin ion structure. Vanadium substitution lead probably to a mixture of lacunary Keggin ion and Keggin ions.

HPOM lead to an increasing [MoO4]/[MoO6] ratio whereas tungsten substituted HPOM to a decreasing [MoO4]/[MoO6] ratio of the resulting [MOx] (M = Mo, V, W) species supported on SBA-15. The [MoO6] units were influenced by the neighboring [VO6] units and [WO6] units of the initial Keggin ion structure resulting in tetrahedral [MoO4] and

[VO4] units and octahedral [MoO6] and [WO6] units during thermal treatment under propene oxidation conditions. Hence, the formation of [MoO4] or [MoO6] units depended on the degree of vanadium or tungsten substitution. Both in vanadium and in tungsten substituted supported HPOM (P(V,W)xMo12-x-SBA-15 (x = 1, 2)) the resulting [MOx] (M

= V, W) units were in close vicinity to the [MoOx] species.

146

Structure-activity correlations

The different structures forming under catalytic conditions (5% propene + 5% oxygen in

He at 723 K) for bulk P(V,W)xMo12-x (x = 0, 1, 2) depended on the addenda atoms (V, W).

Both vanadium substituted PVxMo12-x (x = 1, 2) and tungsten substituted PWMo12-x

(x = 1, 2) showed increased reaction rates compared to unsubstituted PMo12. α-MoO3 resulting from PMo12 showed the lowest catalytic activity in propene oxidation. The mixture of lacunary Keggin ions and Keggin ions resulting from PV2Mo10 and the mixture of α-MoO3 and Mo17O47 resulting from PW2Mo10 showed an increased catalytic activity during propene oxidation. The resulting structures for the substituted HPOM indicated also a lower average valence of molybdenum compared to unsubstituted HPOM. Hence, both the structure and the average valence of the metal centers were responsible for the different catalytic behaviour. The various structures and average valences of Mo resulting from

P(V,W)xMo12-x (x = 0, 1, 2) may also lead to various selectivities. Both PVxMo12-x (x = 1,

2) and PWxMo12-x (x = 1, 2) lead to an increased formation of total oxidation products and a decreased selectivity towards C3 oxidation products depending on the degree of substitution. The reduced metal centers in the mixture of lacunary Keggin ions and Keggin ions resulting from PV2Mo10 and the mixture of α-MoO3 and Mo17O47 resulting from

PW2Mo10 may be particularly effective in activating gas-phase oxygen, resulting in an oversupply of surface-bond electrophilic oxygen. This electrophilic oxygen is prone to further oxidize propene or acrolein to CO2. Therefore, it may be assumed, that the reduced average valence in the structures resulting from P(V,W)xMo12-x (x = 1, 2) was responsible for the decreased selectivity towards C3 oxidation products and higher formation of total oxidation products.

The various structures resulting for supported HPOM exhibited also an influence on the catalytic activity. The reaction rates at similar propene conversion for supported HPOM decreased with higher [MoO4]/[MoO6] ratio for act. P(V,W)xMo12-x-SBA-15 (x = 0, 1, 2).

Therefore, the concentration of [MoO4] and [MoO6] correlated with the catalytic activity.

Additionally, the [MoO4]/[MoO6] ratio was an indicator for the degree of oligomerization.

In samples with higher [MoO4]/[MoO6] ratios a decreased degree of oligomerization is assumed. These results confirmed the assumption, that selective oxidation takes place only in the presence of bridging M-O-M bonds. The increased degree of oligomerization lead to

147

a higher concentration of Mo-O-Mo bonds resulting in an increased catalytic activity. However, the higher reaction rate resulted in an increased formation of total oxidation products. Samples with an increased [MoO4]/[MoO6] ratio exhibited in an increased selectivity towards C3 oxidation products. Hence, reaction rate and selectivity towards the desired C3 oxidation products competed with each other. The addenda atoms had an additional influence on the reaction rates and selectivities for act. P(V,W)xMo12-x-SBA-15 (x = 1, 2). The reaction rates and selectivities of both act

PVxMo12-x-SBA-15 (x = 1, 2) and act. PWxMo12-x-SBA-15 (x = 1, 2) were different from the reaction rates and selectivities of the references synthesized with individual metal (Mo,

V, W) precursors. Structural analysis revealed that the resulting [MOx] (M = V, W) species in act. (P(V,W)xMo12-x-SBA-15 (x = 1, 2)) were in close vicinity to the [MoOx] species.

Conversely, the [MOx] (M = V, W) species in the references act. V2Mo10Ox-SBA-15 and act. W2Mo12Ox-SBA-15 were mostly separated from the [MoOx] species Apparently, the neighboring [MOx] (M = V, W) and [MoOx] species resulted in an increased formation of

CO and CO2. Various mixed metal oxides exhibited an increased formation of CO and CO2 with higher chemical complexity. Probably, the new multifunctional active site was able to activate gas-phase oxygen, resulting in an oversupply of surface-bond electrophilic oxygen. This electrophilic oxygen is prone to further oxidize propene or acrolein to CO2 .

The references act. V2Mo10Ox-SBA-15 and act. W2Mo12Ox-SBA-15 with mostly separated

[MOx] (M = V, W) and [MoOx] species exhibited an increased selectivity towards C3 oxidation products and a decreased formation of CO and CO2.

148

11 References [1] R.K. Grasselli, Advances and future trends in selective oxidation and ammoxidation catalysis, Catal. Today 49 (1999) 141–153. [2] R.K. Grasselli, Fundamental Principles of Selective Heterogeneous Oxidation Catalysis, Top. Catal. 21 (2002) 79–88. [3] H.-J. Arpe, K. Weissermel, Industrielle organische Chemie: Bedeutende Vor- und Zwischenprodukte, 6th ed., Wiley-VCH, Weinheim, 2007. [4] B. Grzybowska-Świerkosz, Thirty years in selective oxidation on oxides: what have we learned?, Top. Catal. 11/12 (2000) 23–42. [5] M. Baerns, Technische Chemie, 2nd ed., Wiley-VCH, Weinheim, 2013. [6] T. Ohara, M. Ueshima, I. Yanagisawa, JP Patent 47-42241B (1972). [7] M.M. Lin, Selective oxidation of propane to acrylic acid with molecular oxygen, Appl. Catal. A: Gen. 207 (2001) 1–16. [8] R.K. Grasselli, D.J. Buttrey, P. DeSanto, J.D. Burrington, C.G. Lugmair, A.F. Volpe, T. Weingand, Active centers in Mo–V–Nb–Te–Ox (amm)oxidation catalysts, Catal.Today 91-92 (2004) 251–258. [9] T. Ushikubo, K. Oshima, A. Kayou, M. Hatano, Ammoxidation of propane over Mo-V-Nb-Te mixed oxide catalysts, in: Can Li and Qin Xin (Ed.), Studies in Surface Science and Catalysis Spillover and Migration of Surface Species on Catalysts Proceedings of the 4th International Conference on Spillover, Elsevier, 1997, pp. 473–480. [10] B.L. Kniep, T. Ressler, A. Rabis, F. Girgsdies, M. Baenitz, F. Steglich, R. Schlögl, Rational Design of Nanostructured Copper–Zinc Oxide Catalysts for the Steam Reforming of Methanol, Angew. Chem. Int. Ed. 43 (2004) 112–115. [11] U.S. Ozkan, R.B. Watson, The structure–function relationships in selective oxidation reactions over metal oxides, Catal. Today 100 (2005) 101–114. [12] T. Ressler, Solid-state kinetics and catalytic behavior of selective oxidation catalysts from time-resolved XAFS investigations, Catal. Today 145 (2009) 258– 266. [13] J. Wienold, O. Timpe, T. Ressler, In Situ Investigations of Structure–Activity Relationships in Heteropolyoxomolybdates as Partial Oxidation Catalysts, Chem. Eur. J. 9 (2003) 6007–6017.

149

[14] T. Ressler, O. Timpe, F. Girgsdies, J. Wienold, T. Neisius, In situ investigations of the bulk structural evolution of vanadium-containing heteropolyoxomolybdate catalysts during thermal activation, J. Catal. 231 (2005) 279–291. [15] T. Ressler, O. Timpe, F. Girgsdies, In situ bulk structural study on solid-state

dynamics and catalytic activity correlations of a H4[PNbMo11O40] partial oxidation catalyst, Z. Kristallogr. 220 (2005) 295–305. [16] T. Ressler, O. Timpe, Time-resolved studies on correlations between dynamic

electronic structure and selectivity of a H5[PV2Mo10O40] partial oxidation catalyst, J. Catal. 247 (2007) 231–237. [17] C. Hess, Direct correlation of the dispersion and structure in vanadium oxide supported on silica SBA-15, J. Catal. 248 (2007) 120–123. [18] I.E. Wachs, Recent conceptual advances in the catalysis science of mixed metal oxide catalytic materials, Catal. Today 100 (2005) 79–94. [19] R.A. Rajadhyaksha, H. Knözinger, Ammonia adsorption on vanadia supported on titania-silica catalyst: An infrared spectroscopic investigation, Appl. Catal. 51 (1989) 81–92. [20] J. Leyrer, D. Mey, H. Knözinger, Spreading behavior of molybdenum trioxide on alumina and silica: A Raman microscopy study, J. Catal. 124 (1990) 349–356. [21] J. Leyrer, R. Margraf, E. Taglauer, H. Knözinger, Solid-solid wetting and formation of monolayers in supported oxide systems, Surface Science 201 (1988) 603–623. [22] D. Zhao, J. Feng, Q. Huo, N. Melosh, G.H. Friedrickson, B.F. Chmelka, G.D. Stucky, Triblock Copolymer Syntheses of Mesoporous Silica with Periodic 50 to 300 Angstrom Pores, Science 279 (1998) 548–552. [23] D. Zhao, Q. Huo, J. Feng, B.F. Chmelka, G.D. Stucky, Nonionic Triblock and Star Diblock Copolymer and Oligomeric Surfactant Syntheses of Highly Ordered, Hydrothermally Stable, Mesoporous Silica Structures, J. Am. Chem. Soc. 120 (1998) 6024–6036. [24] G.M. Dhar, G.M. Kumaran, M. Kumar, K.S. Rawat, L.D. Sharma, B.D. Raju, K.R. Rao, Physico-chemical characterization and catalysis on SBA-15 supported molybdenum hydrotreating catalysts, Catal. Today 99 (2005) 309–314. [25] D.E. Keller, D.C. Koningsberger, B.M. Weckhuysen, Molecular Structure of a

Supported VO4 Cluster and Its Interfacial Geometry as a Function of the SiO2

Nb2O5 and ZrO2 Support, J. Phys. Chem. B 110 (2006) 14313–14325.

150

[26] K. Cassiers, T. Linssen, M. Mathieu, M. Benjelloun, K. Schrijnemakers, P. Van Der Voort, P. Cool, E.F. Vansant, A Detailed Study of Thermal, Hydrothermal, and Mechanical Stabilities of a Wide Range of Surfactant Assembled Mesoporous Silicas, Chem. Mater. 14 (2002) 2317–2324. [27] T. Ressler, U. Dorn, A. Walter, S. Schwarz, A. Hahn, Structure and properties of

PVMo11O40 heteropolyoxomolybdate supported on silica SBA-15 as selective oxidation catalyst, J. Catal. 275 (2010) 1–10. [28] J.F. Keggin, The Structure and Formula of 12-Phosphotungstic Acid, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 144 (1934) 75–100. [29] I. Kozhevnikov, Catalysis by polyoxometalates, J. Wiley, Chichester, West Sussex, New York, 2002. [30] D. Sawant-Dhuri, V.V. Balasubramanian, K. Ariga, D.-H. Park, J.-H. Choy, W.S. Cha, S.S. Al-deyab, S.B. Halligudi, A. Vinu, Titania Nanoparticles Stabilized HPA in SBA-15 for the Intermolecular Hydroamination of Activated Olefins, ChemCatChem 6 (2014) 3347–3354. [31] W. Alharbi, E. Brown, E.F. Kozhevnikova, I.V. Kozhevnikov, Dehydration of ethanol over heteropoly acid catalysts in the gas phase, J.Catal. 319 (2014) 174– 181. [32] M. Kuzminska, T.V. Kovalchuk, R. Backov, E.M. Gaigneaux, Immobilizing heteropolyacids on zirconia-modified silica as catalysts for oleochemistry transesterification and esterification reactions, J. Catal. 320 (2014) 1–8. [33] H.-J. Freund, Model Systems in Heterogeneous Catalysis: Selectivity Studies at the Atomic Level, Top. Catal. 48 (2008) 137–144. [34] L. Marosi, G. Cox, A. Tenten, H. Hibst, In Situ XRD Investigations of Heteropolyacid Catalysts in the Methacrolein to Methacrylic Acid Oxidation Reaction: Structural Changes during the Activation/Deactivation Process, J. Catal. 194 (2000) 140–145. [35] V.M. Bondareva, T.V. Andrushkevich, L.G. Detusheva, G.S. Litvak, Heteropoly compounds in ammoxidation of methylpyrazine, Catal. Lett. 42 (1996) 113–118. [36] X. Li, J. Zhao, W. Ji, Z. Zhang, Y. Chen, C. Au, S. Han, H. Hibst, Effect of vanadium substitution in the cesium salts of Keggin-type heteropolyacids on propane partial oxidation, J. Catal. 237 (2006) 58–66.

151

[37] K.F. Jahr, J. Fuchs, Neue Wege und Ergebnisse der Polysäureforschung, Angew. Chem. 78 (1966) 725–735. [38] G. Lischke, R. Eckelt, G. Öhlmann, On the relationship between thermal stability and catalytic properties of supported molyboo-vanado-phosphoric heteropoly acids, React. Kinet. Catal. Lett. 31 (1986) 267–272. [39] G. Centi, Nieto, J. Lopez, C. Iapalucci, K. Brückman, E.M. Serwicka, Selective oxidation of n-pentane on 12-molybdovanadophosphoric acids, Appl. Catal. 46 (1989) 197–212. [40] G. Mestl, T. Ilkenhans, D. Spielbauer, M. Dieterle, O. Timpe, J. Kröhnert, F. Jentoft, H. Knözinger, R. Schlögl, Thermally and chemically induced structural transformations of Keggin-type heteropoly acid catalysts, Appl. Catal. A: Gen 210 (2001) 13–34. [41] G. Andersson, A. Magnéli, L.G. Sillén, M. Rottenberg, On the Crystal Structure of Molybdenum Trioxide, Acta Chem. Scand. 4 (1950) 793–797. [42] H. Tüysüz, C.W. Lehmann, H. Bongard, B. Tesche, R. Schmidt, F. Schüth, Direct Imaging of Surface Topology and Pore System of Ordered Mesoporous Silica (MCM-41, SBA-15, and KIT-6) and Nanocast Metal Oxides by High Resolution Scanning Electron Microscopy, J. Am. Chem. Soc. 130 (2008) 11510–11517. [43] A. Corma, From Microporous to Mesoporous Molecular Sieve Materials and Their Use in Catalysis, Chem. Rev. 97 (1997) 2373–2420. [44] L. Kihlborg, H. Hasselquist, L. Larsson, R.M. Dodson, The Crystal Structure of

Mo17O47 Acta Chem. Scand. 14 (1960) 1612–1622. [45] J.S. Beck, J.C. Vartuli, W.J. Roth, M.E. Leonowicz, C.T. Kresge, K.D. Schmitt, Chu, C. T. W., D.H. Olson, E.W. Sheppard, A new family of mesoporous molecular sieves prepared with liquid crystal templates, J. Am. Chem. Soc. 114 (1992) 10834–10843. [46] M. Kruk, L. Cao, Pore Size Tailoring in Large-Pore SBA-15 Silica Synthesized in the Presence of Hexane, Langmuir 23 (2007) 7247–7254. [47] F. Zhang, Y. Meng, D. Gu, Yan, Z. Chen, B. Tu, D. Zhao, An Aqueous Cooperative Assembly Route To Synthesize Ordered Mesoporous Carbons with Controlled Structures and Morphology, Chem. Mater. 18 (2006) 5279–5288. [48] L. Wang, X. Liu, X. Wang, X. Yang, L. Lu, Gelatin-assisted porous expansion of mesoporous silica, J Mater Sci 46 (2011) 634–640.

152

[49] J.S. Lettow, Y.J. Han, P. Schmidt-Winkel, P. Yang, D. Zhao, G.D. Stucky, J.Y. Ying, Hexagonal to Mesocellular Foam Phase Transition in Polymer-Templated Mesoporous Silicas, Langmuir 16 (2000) 8291–8295. [50] M. Luechinger, G.D. Pirngruber, B. Lindlar, P. Laggner, R. Prins, The effect of the hydrophobicity of aromatic swelling agents on pore size and shape of mesoporous silicas, Microporous and Mesoporous Mater. 79 (2005) 41–52. [51] L. Cao, T. Man, M. Kruk, Synthesis of Ultra-Large-Pore SBA-15 Silica with Two- Dimensional Hexagonal Structure Using Triisopropylbenzene As Micelle Expander, Chem. Mater. 21 (2009) 1144–1153. [52] M. Misono, Recent progress in the practical applications of heteropolyacid and perovskite catalysts: Catalytic technology for the sustainable society, Catal. Today 144 (2009) 285–291. [53] Y. Kamiya, T. Okuhara, M. Misono, A. Miyaji, K. Tsuji, T. Nakajo, Catalytic Chemistry of Supported Heteropolyacids and Their Applications as Solid Acids to Industrial Processes, Catal. Surv. Asia. 12 (2008) 101–113. [54] B.M. Devassy, F. Lefebvre, W. Böhringer, J. Fletcher, S.B. Halligudi, Synthesis of linear alkyl benzenes over zirconia-supported 12-molybdophosphoric acid catalysts, J. Mol. Catal. A 236 (2005) 162–167. [55] S. Li, J. Zheng, W. Yang, Y. Zhao, Preparation and Characterization of 3DOM

H3PMo12O40–SiO2 with Keggin Structure, Chem. Lett. 36 (2007) 758–759. [56] E. pez-Salinas, . ern ndez- ort z, I. Schifter, E. Torres- arc a, J. Navarrete, A. Gutiérrez-Carrillo, T. López, P. Lottici, D. Bersani, Thermal stability of 12- tungstophosphoric acid supported on zirconia, Appl. Catal. A: Gen. 193 (2000) 215–225. [57] G. Marcì, E. García-López, L. Palmisano, Photo-assisted degradation of 2-propanol

in gas–solid regime by using TiO2 impregnated with heteropolyacid H3PW12O40, Catal. Today 144 (2009) 42–47. [58] A.D. Newman, D.R. Brown, P. Siril, A.F. Lee, K. Wilson, Structural studies of

high dispersion H3PW12O40/SiO2 solid acid catalysts, Phys. Chem. Chem. Phys. 8 (2006) 2893. [59] J.P. Thielemann, T. Ressler, A. Walter, G. Tzolova-Müller, C. Hess, Structure of molybdenum oxide supported on silica SBA-15 studied by Raman, UV–Vis and X- ray absorption spectroscopy, Appl. Catal. A: Gen. 399 (2011) 28–34.

153

[60] V.K. Pecharsky, P.Y. Zavalij, Fundamentals of powder diffraction and structural characterization of materials, 2nd ed., Springer, New York, 2009. [61] R. Allmann, A. Kern, Röntgen-Pulverdiffraktometrie: Rechnergestützte Auswertung, Phasenanalyse und Strukturbestimmung, 2nd ed., Springer Berlin Heidelberg, Berlin, Heidelberg, 2002. [62] R.E. Dinnebier, Billinge, S. J. L, Powder diffraction: Theory and practice, Royal Society of Chemistry, Cambridge, 2008. [63] M. Hesse, H. Meier, B. Zeeh, Spektroskopische Methoden in der organischen Chemie, 7th ed., Thieme, Stuttgart, New York, 2005. [64] S. Wartewig, IR and Raman spectroscopy: Fundamental processing, Wiley-VCH, Weinheim, 2003. [65] D.A. Skoog, S.R. Crouch, F.J. Holler, Instrumentelle Analytik: Grundlagen, Geräte, Anwendungen, 6th ed., Springer, Berlin [u.a.], 2013. [66] P.W. Atkins, J. de Paula, Physikalische Chemie, 5th ed., Wiley-VCH, Weinheim, 2013. [67] C. Czeslik, H. Seemann, R. Winter, Basiswissen physikalische Chemie, 4th ed., Vieweg + Teubner, Wiesbaden, 2010. [68] Sing, K. S. W., Reporting physisorption data for gas/solid systems with special reference to the determination of surface area and porosity (Recommendations 1984), Pure & Appl. Chem. 57 (1985). [69] S. Brunauer, P.H. Emmett, E. Teller, Adsorption of Gases in Multimolecular Layers, J. Am. Chem. Soc. 60 (1938) 309–319. [70] E.P. Barrett, L.G. Joyner, P.P. Halenda, The Determination of Pore Volume and Area Distributions in Porous Substances. I. Computations from Nitrogen Isotherms, J. Am. Chem. Soc. 73 (1951) 373–380. [71] J. Wong, Extended x-ray absorption fine structure: A modern structural tool in materials science, Mater. Sci. Eng. 80 (1986) 107–128. [72] H. Fricke., The K-Characteristic Absorption Frequencies for the Chemical Elements Magnesium to Chromium, Phys. Rev. 16 (1920) 202–215. [73] J.W. Niemantsverdriet, Spectroscopy in catalysis: An introduction, 3rd ed., Wiley- VCH; [John Wiley, distributor], Weinheim, [Chichester], 2007. [74] B.M. Weckhuysen, In-situ spectroscopy of catalysts, American Scientific Publishers, Stevenson Ranch, Calif., 2004.

154

[75] J. Stöhr, NEXAFS spectroscopy, Springer-Verlag, Berlin, New York, 1992. [76] D.C. Koningsberger, R. Prins, X-ray absorption: Principles, applications, techniques of EXAFS, SEXAFS, and XANES, Wiley, New York, 1988. [77] B. Welz, Atomic absorption spectrometry, 3rd ed., Wiley-VCH, Weinheim, New York, 1999. [78] DIN 51005: Thermische Analyse (TA) - Begriffe. 2005 [cited 2014 10.08.2014]; Available from: http://www.beuth.de/langanzeige/DIN+51005/81449381.html. [79] W. Hemminger, H.K. Cammenga, Methoden der thermischen Analyse, Springer- Verlag, Berlin, New York, 1989. [80] Gottwald, GC für Anwender,Wiley-VCH, Weinheim, New York, Basel, Cambridge, Tokyo, 1995. [81] J. Haber, J.H. Block, B. Delmon, in: Ertl, Knözinger et al. (Ed.) 2008 – Handbook of Heterogeneous Catalysis, pp. 1230–1258. [82] J. Hagen, Industrial catalysis: A practical approach, 2nd ed., Wiley-VCH, Weinheim, 2006. [83] H.-J. Hübschmann, Handbuch der GC MS: Grundlagen und Anwendungen, VCH, Weinheim, New York, Basel, Cambridge, Tokyo, 1996. [84] R.L. Grob, E.F. Barry, Modern practice of gas chromatography, Wiley- Interscience, Hoboken, N.J., 2004. [85] T. Okuhara, N. Mizuno, M. Misono, Catalytic Chemistry of Heteropoly Compounds, in: D.D. Eley, Werner O. Haag and Bruce Gates (Ed.), Advances in Catalysis, Academic Press, 1996, pp. 113–252. [86] M.E. Davis, C.J. Dillon, J.H. Holles, J. Labinger, A New Catalyst for the Selective Oxidation of Butane and Propane This work was funded by BP., Angew. Chem. Int. Ed. 41 (2002) 858–860. [87] C.J. Dillon, J. Holles, J.A. Labinger, M.E. Davis, A substrate-versatile catalyst for the selective oxidation of light alkanes II. Catalyst characterization, J. Catal. 218 (2003) 54–66. [89] J. Holles, C.J. Dillon, J.A. Labinger, M.E. Davis, A substrate-versatile catalyst for the selective oxidation of light alkanes I. Reactivity, J. Catal. 218 (2003) 42–53. [89] N. Mizuno, M. Misono, Heterogeneous Catalysis, Chem. Rev. 98 (1998) 199–218.

[90] A. Chemseddine, F. Babonneau, J. Livage, Anisotropic WO3·nH2O layers deposited from gels, J. Non-Cryst. Solids 91 (1987) 271–278.

155

[91] T. Ressler, WinXAS: a Program for X-ray Absorption Spectroscopy Data Analysis under MS-Windows, J. Synch. Rad. 5 (1998) 118–122. [92] J. J. Rehr, C. H. Booth, F. Bridges, and S. I. Zabinsky, X-ray-absorption fine structure in embedded atoms, Phys. Rev. B 49 (1994) 12347–12350. [93] J. Boeyens, G. McDougal, J. Van R. Smit, Crystallographic study of the ammonium/potassium 12-molybdophosphate ion-exchange system, J. Solid State Chem. 18 (1976) 191–199. [94] T. Ressler, S.L. Brock, J. Wong, S.L. Suib, Multiple-Scattering EXAFS Analysis of Tetraalkylammonium Manganese Oxide Colloids, J. Phys. Chem. B 103 (1999) 6407–6420. [95] A. Walter, R. Herbert, C. Hess, T. Ressler, Structural characterization of vanadium oxide catalysts supported on nanostructured silica SBA-15 using X-ray absorption spectroscopy, Chem. Central J. 4 (2010) 3–23. [96] StandardsCriteria_July25_2000, available at http://www.ixasportal.net/ixas/images/ixas_mat/StandardsCriteria_July25_2000.pdf [97] W.H. Press, Numerical recipes: The art of scientific computing, 3rd ed., Cambridge University Press, Cambridge, 2007. [98] H. d´Amour, R. Allmann, Ein Kegginkomplex mit erniedrigter Pseudosymmetrie in

der Struktur des H3[PMo12O40]·(13-14)H2O, Zeitschrift für Kristallographie- Crystalline Materials 143 (1976) 1–13. [99] C. Marchal-Roch, C. Julien, J.F. Moisan, N. Leclerc-Laronze, F.X. Liu, G. Hervé, 3− Study of molybdenyl, vanadyl and mixed ammonium vanadyl salts of [PMo12O40] as oxidation catalysts, Appl. Catal. A: Gen.l 278 (2004) 123–131. [100] C. Marchal-Roch, N. Laronze, N. Guillou, A. Tézé, G. Hervé, Study of ammonium, IV mixed ammonium–cesium and cesium salts derived from (NH4)5[PMo11V O40] as isobutyric acid oxidation catalysts, Appl. Catal. A: Gen. 199 (2000) 33–44.

[101] A.J. Bradley, J.W. Illingworth, The Crystal Structure of H3PW12O40 29H2O, Proc. R. Soc. London, Ser. A 157 (1936) 113–131. [102] C.J. Clark, D. Hall, Dodecamolybdophosphoric acid circa 30-hydrate, Acta Crystallogr., Sect. B: Struct. Sci 32 (1976) 1545–1547. [103] M. Binnewies, M. Jäckel, H. Willner, G. Rayner-Canham, Allgemeine und Anorganische Chemie, 2nd ed., Spektrum, Akad. Verl, Heidelberg, 2011.

156

[104] A. . Bridgeman, omputational Study of the Vibrational Spectra of α- and β- Keggin Polyoxometalates, Chem. Eur. J. 10 (2004) 2935–2941. [105] C. Rocchiccioli-Deltcheff, M. Fournier, R. Franck, R. Thouvenot, Vibrational investigations of polyoxometalates. 2. Evidence for anion-anion interactions in molybdenum(VI) and tungsten(VI) compounds related to the Keggin structure, Inorg. Chem. 22 (1983) 207–216. [106] R. Thouvenot, M. Fournier, R. Franck, C. Rocchiccioli-Deltcheff, Vibrational investigations of polyoxometalates. 3. Isomerism in molybdenum(VI) and tungsten(VI) compounds related to the Keggin structure, Inorg. Chem. 23 (1984) 598–605. [107] A. Popa, V. Sasca, E. Kis, R. Marinkovic-Neducin, M. Bokorov, J. Halasz, Structure and Texture of some Keggin type heteropolyacids supported on silica and titania, J. Optoelectron. Adv. M. 7 (2005) 3169–3177. [108] A. Bielański, A. Małecka, . Kybelkova, Infrared study of the thermal

decomposition of heteropolyacids of the series H3+xPMo12–xVxO40, J. Chem. Soc., Faraday Trans. 1 85 (1989) 2847. [109] P. Villabrille, G. Romanelli, P. Vázquez, C. Cáceres, Vanadium-substituted Keggin heteropolycompounds as catalysts for ecofriendly liquid phase oxidation of 2,6- dimethylphenol to 2,6-dimethyl-1,4-benzoquinone, Appl. Catal. A: Gen. 270 (2004) 101–111. [110] P. Villabrille, G. Romanelli, L. Gassa, P. Vázquez, C. Cáceres, Synthesis and characterization of Fe- and Cu-doped molybdovanadophosphoric acids and their application in catalytic oxidation, Appl. Catal. A: Gen. 324 (2007) 69–76. [111] X. Gao, In Situ UV–vis–NIR Diffuse Reflectance and Raman Spectroscopic

Studies of Propane Oxidation over ZrO2-Supported Vanadium Oxide Catalysts, J. Catal. 209 (2002) 43–50. [112] K.V. Rao, P. Rao, P. Nagaraju, P.S. Prasad, N. Lingaiah, Room temperature selective oxidation of toluene over vanadium substituted polyoxometalate catalysts, J. Mol. Catal. A: Chem. 303 (2009) 84–89. [113] J.E. Molinari, L. Nakka, T. Kim, I.E. Wachs, Dynamic Surface Structures and

Reactivity of Vanadium-Containing Molybdophosphoric Acid (H3+xPMo12- xVxO40) Keggin Catalysts during Methanol Oxidation and Dehydration, ACS Catal. 1 (2011) 1536–1548.

157

[114] J. Black, Acrolein oxidation over 12-molybdophosphates I. Characterization of the catalyst, J. Catal.. 106 (1987) 1–15. [115] M. Langpape, J. Millet, U. Ozkan, M. Bourdeulle, Study of Cesium or Cesium- Transition Metal-Substituted Keggin-Type Phosphomolybdic Acid as Isobutane Oxidation Catalysts, J. Catal. 181 (1999) 80–90. [116] M. Misono, Heterogeneous Catalysis by Heteropoly Compounds of Molybdenum and Tungsten, Catal. Rev. 29 (1987) 269–321. [117] L. Marosi, E. Escalona Platero, J. Cifre, C. Otero Areán, Thermal dehydration of

H3+xPVxM12-xO40·yH2O Keggin type heteropolyacids; formation, thermal stability

and structure of the anhydrous acids H3PM12O40, of the corresponding anhydrides

PM12O38.5 and of a novel trihydrate H3PW12O40·3H2O, J. Mater. Chem. 10 (2000) 1949–1955. [118] E.M. McCarron, J.C. Calabrese, The growth and single crystal structure of a high

pressure phase of molybdenum trioxide: MoO3-II, Journal of Solid State Chemistry 91 (1991) 121–125. [119] J. Guo, P. Zavalij, M. Whittingham, Metastable Hexagonal Molybdates: Hydrothermal Preparation, Structure, and Reactivity, J. Solid State Chem. 117 (1995) 323–332. [120] D.P. Woodruff, The chemical physics of solid surfaces, Elsevier Science, Amsterdam, Oxford, 2001.

[121] J.B. Parise, E.M. McCarron, R. von Dreele, .A. oldstone, β-MoO3 produced from a novel freeze drying route, J. Solid State Chem. 93 (1991) 193–201. [122] L. Pauling, The Principles determinig the structure of complex ionic crystals, J. Am. Chem. Soc. 51 (1929) 1010–1026. [123] U. Müller, Anorganische Strukturchemie, 6th ed., Vieweg+Teubner Verlag / GWV Fachverlage, Wiesbaden, Wiesbaden, 2008.

[124] Eva Rödel, In situ bulk structural investigation of Mo5O14-type mixed metal oxide catalysts for partial oxidation reactions, Berlin, 2006. [125] E. Salje, R. Gehlig, K. Viswanathan, Structural phase transition in mixed crystals

WxMO1−xO3, J. Solid State Chem. 25 (1978) 239–250. [126] G.A. Zenkovets, G.N. Kryukova, V. Gavrilov, S.V. Tsybulya, V.A. Anufrienko, T.A. Larina, D.F. Khabibulin, O.B. Lapina, E. Rödel, A. Trunschke, T. Ressler, R.

Schlögl, The structural genesis of a complex (Mo,V,W)5O14 oxide during thermal

158

treatments and its redox behavior at elevated temperatures, Mater. Chem.Phys. 103 (2007) 295–304. [127] E. Rödel, O. Timpe, A. Trunschke, G.A. Zenkovets, G.N. Kryukova, R. Schlögl, T. Ressler, Structure stabilizing effect of tungsten in mixed molybdenum oxides with

Mo5O14-type structure, Catal. Today 126 (2007) 112–118. [128] A. Maiti, N. Govind, P. Kung, D. King-Smith, J.E. Miller, C. Zhang, G. Whitwell, Effect of surface phosphorus on the oxidative dehydrogenation of ethane: A first- principles investigation, J. Chem. Phys. 117 (2002) 8080–8088. [129] J.-M.M. Millet, FePO Catalysts for the Selective Oxidative Dehydrogenation of Isobutyric Acid into Methacrylic Acid, Catal. Rev. Sci. Eng. 40 (1998) 1–38. [130] F. Richter, H. Papp, T. Götze, G. Wolf, B. Kubias, Investigation of the surface of vanadyl pyrophosphate catalysts, Surf. Interface Anal. 26 (1998) 736–741. [131] P. Wongkrua, T. Thongtem, S. Thongtem, Synthesis of h- and α-MoO3 by Refluxing and Calcination Combination: Phase and Morphology Transformation, Photocatalysis, and Photosensitization, Journal Nanomaterials 2013 (2013) 1–8.

[132] Y. Wang, Y. Zhu, Z. Xing, Y. Qian, ydrothermal Synthesis of α-MoO3 and the Influence of Later Heat Treatment on its Electrochemical Properties, International Journal of Electrochemical Science (2013) 9851–9857. [133] S. Kühn, P. Schmidt-Zhang, A.H.P. Hahn, M. Huber, M. Lerch, T. Ressler, Structure and properties of molybdenum oxide nitrides as model systems for selective oxidation catalysts, Chem. Central J. 5 (2011) 42–52. [134] T. Ressler, J. Wienold, R. Jentoft, T. Neisius, Bulk Structural Investigation of the

Reduction of MoO3 with Propene and the Oxidation of MoO2 with Oxygen, J. Catal. 210 (2002) 67–83. [135] J.-D. Guo, P.Yu. Zavalij, M.S. Whittingham, Preparation and characterization of a-

MoO3 with hexagonal structure, Eur. J. Solid State Inorg. Chem. 31 (1994) 833– 842. [136] M. Seleborg, R.G. Hazell, Å. Nilsson, J. Sandström, H. Theorell, R. Blinc, S. Paušak, . Ehrenberg, . Dumanović, A Refinement of the rystal Structure of Disodium Dimolybdate, Acta Chem. Scand. 21 (1967) 499–504. [137] J. Scholz, A. Walter, A. Hahn, T. Ressler, Molybdenum oxide supported on nanostructured MgO: Influence of the alkaline support properties on MoOx

159

structure and catalytic behavior in selective oxidation, Microporous Mesoporous Mater. 180 (2013) 130–140. [138] K. Inumaru, A. Ono, H. Kubo, M. Misono, Catalysis by heteropoly compounds Part 39 The structure and redox behaviour of vanadium species in molybdovanadophosphoric acid catalysts during partial oxidation of isobutane, Faraday Trans. 94 (1998) 1765–1770. [139] R.-Y. Wang, D.-Z. Jia, L. Zhang, L. Liu, Z.-P. Guo, B.-Q. Li, J.-X. Wang, Rapid Synthesis of Amino Acid Polyoxometalate Nanotubes by One-Step Solid-State Chemical Reaction at Room Temperature, Adv. Funct. Mater. 16 (2006) 687–692.

[140] Z. Zhang, Y. Qu, S. Wang, J. Wang, The Initial Reactions of H3PO4 and NaH2PO4 Supported on Silica: A Joint Experimental and Theoretical Study, Chin. J. Chem. Phys. 22 (2009) 315–321. [141] M.M. Crutchfield, C.V. Callis, R.R. Irani, G.C. Roth, Phosphorus Nuclear Magnetic Resonance Studies of Ortho and Condensed Phosphates, Inorg. Chem. 1 (1962) 813–817. [142] T.R. Krawietz, P. Lin, K.E. Lotterhos, P.D. Torres, D.H. Barich, A. Clearfield, J.F. Haw, Solid Phosphoric Acid Catalyst: A Multinuclear NMR and Theoretical Study, J. Am. Chem. Soc. 120 (1998) 8502–8511. [143] J. Frey, C. Lieder, T. Schölkopf, T. Schleid, U. Nieken, E. Klemm, M. Hunger, Quantitative solid-state NMR investigation of V5+ species in VPO catalysts upon sequential selective oxidation of n-butane, J. Catal. 272 (2010) 131–139. [144] T. Ressler, A. Walter, Z. Huang, W. Bensch, Structure and properties of a

supported MoO3–SBA-15 catalyst for selective oxidation of propene, J. Catal. 254 (2008) 170–179. [145] E.F. Vansant, Voort, P. van der, K.C. Vrancken, Characterization and chemical modification of the silica surface, Elsevier, Amsterdam, New York, 1995. [146] R. Radhakrishnan, C. Reed, S.T. Oyama, M. Seman, J.N. Kondo, K. Domen, Y.

Ohminami, K. Asakura, Variability in the Structure of Supported MoO3 Catalysts: Studies Using Raman and X-ray Absorption Spectroscopy with ab Initio Calculations, J. Phys. Chem. B 105 (2001) 8519–8530. [147] . andzlik, P. Sautet, Structure of Dimeric Molybdenum(VI) Oxide Species on γ- Alumina: A Periodic Density Functional Theory Study, J. Phys. Chem. C 114 (2010) 19406–19414.

160

[148] G. Deo, I.E. Wachs, J. Haber, Crit, Rev. Surf. Chem 4 (1994) 141–187. [149] K. Amakawa, L. Sun, C. Guo, M. Hävecker, P. Kube, I.E. Wachs, S. Lwin, A.I. Frenkel, A. Patlolla, K. Hermann, R. Schlögl, A. Trunschke, How Strain Affects the Reactivity of Surface Metal Oxide Catalysts, Angew. Chem. Int. Ed. (2013) n/a. [150] Y. Lou, H. Wang, Q. Zhang, Y. Wang, SBA-15-supported molybdenum oxides as efficient catalysts for selective oxidation of ethane to formaldehyde and acetaldehyde by oxygen, J. Catal 247 (2007) 245–255. [151] Y.-M. Liu, Y. Cao, N. Yi, W.-L. Feng, W.-L. Dai, S.-R. Yan, H.-Y. He, K.-N. Fan, Vanadium oxide supported on mesoporous SBA-15 as highly selective catalysts in the oxidative dehydrogenation of propane, J. Catal. 224 (2004) 417–428. [152] K. Chen, The Relationship between the Electronic and Redox Properties of Dispersed Metal Oxides and Their Turnover Rates in Oxidative Dehydrogenation Reactions, J. Catal. 209 (2002) 35–42. [153] J. Scholz, A. Walter, T. Ressler, Influence of MgO-modified SBA-15 on the structure and catalytic activity of supported vanadium oxide catalysts, J. Catal. 309 (2014) 105–114. [154] N. Haddad, E. Bordes-Richard, A. Barama, MoOx-based catalysts for the oxidative dehydrogenation (ODH) of ethane to ethylene, Catal. Today 142 (2009) 215–219. [155] T. Kashiwabara, Y. Takahashi, M.A. Marcus, T. Uruga, H. Tanida, Y. Terada, A. Usui, Tungsten species in natural ferromanganese oxides related to its different behavior from molybdenum in oxic ocean, Geochim.Cosmochim. Acta 106 (2013) 364–378. [156] B.-K. Teo, P.A. Lee, Ab initio calculations of amplitude and phase functions for extended x-ray absorption fine structure spectroscopy, J. Am. Chem. Soc. 101 (1979) 2815–2832. [157] S. Yamazoe, Y. Hitomi, T. Shishido, T. Tanaka, XAFS Study of Tungsten L1- and

L3-Edges: Structural Analysis of WO3 Species Loaded on TiO2 as a Catalyst for

Photo-oxidation of NH3, J. Phys. Chem. C 112 (2008) 6869–6879. [158] S.R. Bare, G.E. Mitchell, J.J. Maj, G.E. Vrieland, J.L. Gland, Local site symmetry of dispersed molybdenum oxide catalysts: XANES at the Mo L2,3-edges, J. Phys. Chem. 97 (1993) 6048–6053.

161

[159] A. Kuzmin, J. Purans, Dehydration of the molybdenum trioxide hydrates

MoO3·nH2O: in situ x-ray absorption spectroscopy study at the Mo K edge, J. Phys.: Condens. Matter 12 (2000) 1959–1970. [160] O. Kirilenko, F. Girgsdies, R.E. Jentoft, T. Ressler, In Situ XAS and XRD Studies on the Structural Evolution of Ammonium Paratungstate During Thermal Decomposition, Eur. J. Inorg. Chem. 2005 (2005) 2124–2133. [161] X. Carrier, E. Marceau, H. Carabineiro, V. Rodríguez-González, M. Che, EXAFS spectroscopy as a tool to probe metal–support interaction and surface molecular

structures in oxide-supported catalysts: application to Al2O3-supported Ni(II)

complexes and ZrO2-supported tungstates, Phys. Chem. Chem. Phys. 11 (2009) 7527. [162] E.I. Ross-Medgaarden, I.E. Wachs, Structural Determination of Bulk and Surface Tungsten Oxides with UV−vis Diffuse Reflectance Spectroscopy and Raman Spectroscopy, J. Phys. Chem. C 111 (2007) 15089–15099. [163] S. Chaemchuen, W. Limsangkass, B. Netiworaraksa, S, Phatanasri, N. Sae-Ma, K.

Suriya, Novel catalyst of mixed SiO2-TiO2 supported tungsten for metathesis of ethene and 2- butene, Bulgarian Chemical Communications 44 (2012) 87–91. [164] D. Barton, S. Soled, G. Meitzner, G. Fuentes, E. Iglesia, Structural and Catalytic Characterization of Solid Acids Based on Zirconia Modified by Tungsten Oxide, J. Catal. 181 (1999) 57–72. [165] J. Ramírez, Characterization and Hydrodesulfurization Activity of W-Based

Catalysts Supported on Al2O3–TiO2 Mixed Oxides, J. Catal. 170 (1997) 108–122. [166] G. Deo, I.E. Wachs, Predicting molecular structures of surface metal oxide species on oxide supports under ambient conditions, J. Phys. Chem. 95 (1991) 5889–5895. [167] M. Kanno, T. Yasukawa, W. Ninomiya, K. Ooyachi, Y. Kamiya, Catalytic oxidation of methacrolein to methacrylic acid over silica-supported 11-molybdo-1- vanadophosphoric acid with different heteropolyacid loadings, J. Catal. 273 (2010) 1–8. [168] L.T. Zhuravlev, The surface chemistry of amorphous silica. Zhuravlev model, Colloids Surf., A 173 (2000) 1–38. [169] G. Jander, A. Winkel, Über amphotere Oxydhydrate, deren wäßrige Lösungen und Kristallisierende Verbindungen. XII Mitteilung. Hydrolysierende Systeme und ihre

162

Aggregationsprodukte mit besonderer Berücksichtigung der Erscheinungen in wäßrigen Aluminiumsalzlösungen, Z. anorg. allg. Chem. 200 (1931) 257–278. [170] R.K. Grasselli, Genesis of site isolation and phase cooperation in selective oxidation catalysis, Top. Catal. 15 (2001) 93–101.

[171] B. Solsona, A. Dejoz, M. Vázquez, F. Márquez, J. López Nieto, SiO2-supported vanadium magnesium mixed oxides as selective catalysts for the oxydehydrogenation of short chain alkanes, Appl. Catal. A: Gen. 208 (2001) 99– 110. [172] J.L. Callahan, R.K. Grasselli, A selectivity factor in vapor-phase hydrocarbon oxidation catalysis, AIChE J. 9 (1963) 755–760.

[173] X. Li, S. Huang, Q. Xu, Y. Yang, Preparation of WO3-SBA-15 mesoporous molecular sieve and its performance as an oxidative desulfurization catalyst, Transition Met. Chem. 34 (2009) 943–947. [174] X.-L. Yang, R. Gao, W.-L. Dai, K. Fan, Influence of Tungsten Precursors on the

Structure and Catalytic Properties of WO3/SBA-15 in the Selective Oxidation of Cyclopentene to Glutaraldehyde, J. Phys. Chem. C 112 (2008) 3819–3826. [175] R.K. Iler, The chemistry of silica: Solubility, polymerization, colloid and surface properties, and biochemistry, Wiley, New York, 1979. [176] Y. Sasaki, M. Hennichs, P.M. Jørgensen, S. Refn, V.K. Andersen, A. Jart, Equilibrium Studies of Polyanions. 7. The First Step in the Acidification of WO2-;

Equilibria in 3 M NaClO4 at 25 degrees C, Acta Chem. Scand. 15 (1961) 175–189. [177] Z. Miao, H. Zhao, H. Song, L. Chou, Ordered mesoporous zirconium oxophosphate supported tungsten oxide solid acid catalysts: the improved Brønsted acidity for benzylation of anisole, RSC Adv. 4 (2014) 22509. [178] N. Liu, S. Ding, Y. Cui, N. Xue, L. Peng, X. Guo, W. Ding, Optimizing activity of tungsten oxides for 1-butene metathesis by depositing silica on γ-alumina support, Chemical Engineering Research and Design 91 (2013) 573–580. [179] H.-G. Jerschkewitz, E. Alsdorf, H. Fichtner, W. Hanke, K. Jancke, G. Öhlmann, Über die thermischen Eigenschaften von Heteropolysäuren des Typs

H3+n[PVnMo12-nO40] x H2O (n = 0, 1, 2, 3) I. Thermogravimetrische, UV-VIS- und röntgenographische Untersuchungen, Z. Anorg. Allg. Chem. 526 (1985) 73–85. [180] S. Du Kim, M. Ostromecki, I.E. Wachs, S.D. Kohler, J.G. Ekerdt, Preparation and

characterization of WO3/SiO2 catalysts, Catal. Lett. 33 (1995) 209–215.

163

[181] A.F. Holleman, E. und Nils Wiberg, G. Fischer (Eds.), Lehrbuch der Anorganischen Chemie, Walter de Gruyter, Berlin, New York, 2007. [182] H. Gruber, E. Krautz, H.P. Fritzer, Magnetic susceptibility and electrical properties

of the X-phase Mo17O47 with substitutions of molybdenum, Phys. Stat. Sol. (a) 65 (1981) 589–594. [183] G. Mestl, MoVW mixed metal oxides catalysts for acrylic acid production: from industrial catalysts to model studies, Top. Catal. 38 (2006) 69–82. [184] G. Centi, F. Cavani, F. Trifiro, Selective Oxidation by Heterogeneous Catalysis, Springer US, Boston, MA, 2001. [185] M. Bettahar, G. Costentin, L. Savary, J. Lavalley, On the partial oxidation of propane and propylene on mixed metal oxide catalysts, Appl. Catal. A: Gen. 145 (1996) 1–48. [186] S. Carrazán, C. Martín, R. Mateos, V. Rives, Influence of the active phase structure Bi-Mo-Ti-O in the selective oxidation of propene, Catal. Today 112 (2006) 121– 125. [187] P. Concepción, P. Botella, J.L. Nieto, Catalytic and FT-IR study on the reaction pathway for oxidation of propane and propylene on V- or Mo–V-based catalysts, Appl. Catal. A: Gen. 278 (2004) 45–56.

164

12 Appendix

-1 Table A 1: Wave numbers [cm ] of the meausured IR vibration bands for P(V,W)xMo12-x (x = 0, 1, 2)(Chapter 3.4).

PMo12 PVMo11 PV2Mo10 PWMo11 PW2Mo10 Assignment

νas (P-O), νas (Mo-Ot) 1064 (s) 1062 (s) 1060 (s) 1066 (s) 1068 (s) (asymmetric coupling)

νas (P-O), νas (Mo-Ot) 961 (vs) 960 (vs) 959 (vs) 964 (vs) 964 (vs) (symmetric coupling)

869 (m) 865 (m) 862 (m) 870 (m) 874 (m) νas (Mo-Oe-Mo)

781 (vs) 779 (vs) 779 (vs) 784 (vs) 783 (vs) νas (Mo-Oe-Mo)

-1 Table A 2: Wave numbers [cm ] of the meausured Raman vibration bands for P(V,W)xMo12-x (x = 0, 1, 2) )(Chapter 3.4).

PMo12 PVMo11 PV2Mo10 PWMo11 PW2Mo10

999 (vs) 997 (vs) 1001 (vs) 996 (vs) 1001 (vs) νas (Mo-Ot)

972 (sh) 973 (sh) 974 (sh) 982 (sh) 982 (sh) νas (Mo-Ot)

909 (vw) 904 (vw) 904 (vw) 888 (vw) 888 (vw) νas (Mo-Oe-Mo)

νas (Mo-Oc-Mo), 611 (w) 609 (w) 617 (w) 608 (w) 608 (w) δ (Mo-Oc-Mo)

- 497 (vw) 507 (vw) 497 (vw) 498 (vw) δ (Mo-Oc-Mo)

- 460 (vw) 454 (vw) 460 (vw) 463 (vw) δ (Mo-Oc-Mo)

368 (vw) 370 (vw) 372 (vw) 370 (vw) 366 (vw) δ (Mo-Oe-Mo)

δ (Oc-Mo-Oc), 250 (m) 252 (m) 256 (m) 252 (m) 248 (m) δ (Oe-Mo-Oe')

227 (m) 224 (m) 224 (m) 224 (m) 231 (m) δ (Mo-Oe-Mo) δ (Mo-O-Mo), 156 (m) 160 (m) 161 (m) 160 (m) 159 (m) δ (O-Mo-O)

δ (Oc-Mo-Ot) 111 (m) 112 (m) 113 (m) 112 (m) 108 (m)

δ (Mo-O-Mo), 84 (w) 85 (w) 85 (w) 85 (w) 83 (w) δ (O-Mo-O)

165

m/e = 18 (H2O)

temperature 400

300

200

temperature [K] temperature Normalized ion current ion Normalized 100

0 0 1000 2000 3000 4000 5000 6000 Cycle Fig. A 1: Ion current m/e = 18 corresponding to water and temperature steps of in situ XRD meausurements of PMo12 during oxidation conditions (20% O2 in He; RT-723 K).

α-MoO3

Intensity

10 20 30 40 50 60 70 80 Diffraction angle [2Ɵ]

Fig. A 2: XRD powder pattern of PMo12 after thermal treatment during catalytic conditions :conditions (5% propene + 5% oxygen in He at 723 K) and diffraction peaks of bulk reference material α-MoO3 (ICSD 76365 [129]).

166

α-MoO3

Mo17O47

Intensity

10 20 30 40 50 60 70 80 Diffraction angle [2Ɵ]

Fig. A 3: XRD powder pattern of PW2Mo10 after thermal treatment during catalytic conditions :conditions (5% propene + 5% oxygen in He at 723 K) and diffraction peaks for bulk reference materials α-MoO3 (ICSD 76365 [41]) and Mo17O47 (ICSD 36098 [44]).

0.40

0.35

0.30

0h 12h

Absorption 0.25

0h 12h 0.20 0h 12h

0.15 PMo12 PV2Mo10 PW2Mo10

Fig. A 4: AAS absorption of phosphorus in PMo 12, PVMo11, and PV2Mo10 before and after treatment during propene oxidation conditions (5% propene + 5% O2 in He; 723 K; 0h and 12h time on stream).

167

2.5

2.0 VO2

V2O5 1.5

PV2Mo10

1.0

PV2Mo10-SBA-15

Normalized absorption Normalized 0.5

0.0

5.45 5.50 5.55 5.60 Photon energy [ke'V]

Fig. A 5: V K edge XANES of PV2Mo10-SBA-15, PV2Mo10 and the references V2O5, VO2.

6.0 Mo oxide references MoO3 PMo 12-SBA-15 PVMo -SBA-15 5.0 11 Mo4O11

PV 2Mo10-SBA-15

PWMo 11-SBA-15 MoO2 4.0 PW 2Mo10-SBA-15 Fit Curve 3.0

2.0 Mo average valence Moaverage 1.0

0.0 Mo

0 5 10 15 20 25 30 35 40 Mo K edge, (eV)

Fig. A 6: Mo average valence of the Mo oxide references and P(V,W)xMo12-x-SBA-15 ( x = 0, 1, 2) as a function of the Mo K edge position.

XII

Danksagung

Ich bedanke mich bei Herrn Prof. Dr. Thorsten Ressler für die interessante wissenschaftliche Fragestellung. Insbesondere danke ich ihm auch für die exzellente fachliche Betreuung während der gesamten Zeit meiner Forschungstätigkeit in seinem Arbeitskreis. Ich danke außerdem Herrn Prof. Dr. Malte Behrens für die Anfertigung des Zweitgutachtens. Mein besonderer Dank gilt der gesamten Arbeitsgruppe Ressler für die angenehme und freundschaftliche Arbeitsatmosphäre. Ich danke vor allem Gregor Koch, Alexander Müller, Sven Kühn und Dr. Juliane Scholz für ihre stete Diskussionsbereitschaft. Bei Alexander Hahn und Dr. Thomas Christoph Rödel bedanke ich mich für die wissenschaftliche und technische Unterstützung bei der Durchführung zahlreicher Experimente. Ich danke auch Semiha Schwarz für die technische Hilfestellung bei der Durchführung von TG- Messungen und die zahlreichen Ratschläge innerhalb- und außerhalb des Forschungsthemas. Besonders will ich mich an dieser Stelle bei Dr. Anke Walter bedanken, die mich als Forschungspraktikant und Diplomand in zahlreiche Methoden der analytischen Chemie eingeführt und mich in dieser Zeit für die Arbeit in der instrumentellen Analytik und Katalyseforschung exzellent vorbereitet hat. Darüber hinaus bedanke ich mich bei Lars Eggers, Mario Willoweit, Tina Somnitz und Larissa Braun, die mich im Rahmen ihrer Bachelorarbeiten unterstützt haben. Dem gesamten Arbeitskreis Lerch danke ich für die Aufnahmen der Weitwinkelbeugungsdaten. Ich bedanke mich bei den Arbeitskreisen Grohmann und Lerch für die freundliche Atmosphäre und tatkräftige Unterstützung im Syntheselabor. Ich bedanke mich bei allen Mitgliedern des Instituts für Chemie der TU Berlin, die mich bei meiner Arbeit unterstützt haben. Dem DESY und dem HASYLAB in Hamburg sei für die Bereitstellung zahlreicher Messzeiten gedankt. Bei der Deutschen Forschungsgemeinschaft (DFG) bedanke ich mich für die finanzielle Unterstützung. Ich danke meiner Frau Hanna und meiner kleinen Tochter Nell für die familäre Ablenkung nebem dem Forschungsalltag und die uneingeschränkte Unterstützung und Rücksichtnahme während der Anfertigung dieser Arbeit.