From molybdenum based model catalysts to technically applied
systems
vorgelegt von
Diplom-Chemiker
Stefan Knobl
aus Selb
Fakultät II - Mathematik und Naturwissenschaften
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Naturwissenschaften
Dr. rer. nat.
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr. rer. nat. R. Schomäcker
Berichter/Gutachter: Prof. Dr. rer. nat. M. Lerch
Berichter/Gutachter: Prof. Dr. rer. nat. R. Schlögl
Tag der wissenschaftlichen Aussprache: 09. März 2004
Berlin 2004
D 83
Acknowledgement
I would like to express my gratitude towards all my colleagues in the AC department
for a good working atmosphere and support.
Thanks to the team in Kuala Lumpur, especially Norli, Diana and Q. and Prof. Dr.
Sharifah B. Hamid.
Thanks to the team in Novosibirsk, especially Galina A. and Galina N.
I would like to thank especially Olaf for his guidance and his readiness to discuss all
kind of things. Further, I would like to thank Pablo and Gisela.
I would like to thank Prof. Dr. M. Lerch for the report, and Prof. Dr. R. Schomäcker
for being the chairman.
Further, I am in debt to Dr. G. Mestl for the first supervising part.
Thanks to Dr. D. Niemeyer for supervising the second part and for his guidance.
Finally I would like to thank Prof. Dr. R. Schlögl for putting up with me, for his
support, his ideas, his readiness to discuss and for the good working conditions.
Many thanks to my parents and to Bärbel.
Table of contents
From molybdenum based model catalysts to technically applied systems 1
1 Introduction 1
1.1 General Introduction 1
1.2 Mo-only model systems 3
1.3 Multi Metal Oxides 5
2 Controlled preparation of molybdenum oxide catalysts 10
2.1 Introduction 10
2.2 Experimental 12
Preparation 12
XRD 13
TEM 13
Raman DRIFTS 13
UV/Vis/NIR 13
BET 14
TG 14
TPRS 14
2.3 Results 14
2.3.1 Orthorhombic MoO315
pH-dependency 16
XRD 16
Raman 16
DRIFTS 16
UV/Vis NIR 17
Thermal analysis 17
TPRS 18
2.3.2 Trimolybdate 18
pH dependency 18
XRD 18
Raman 20
DRIFTS 20
UV/Vis/NIR 21
Thermal Analysis 21
2.3.3 Hexagonal MoO321
pH dependency 22
XRD 23
Raman 24
DRIFTS 24
UV/Vis/NIR 25
Thermal Analysis 26
SEM/TEM 28
TPRS 29
2.3.4 Mo36O1128- supramolecular molybdenum oxide 30
pH dependency 32
XRD 32
Raman 32
DRIFTS 33
UV/Vis NIR 33
Thermal Analysis 34
SEM/TEM 36
TPRS 37
2.3.5 Up scaling 38
Preparation and pH-dependency 38
Electron Microscopy and Electron Diffraction 38
TPRS 39
2.4 Discussion 41
2.4.1 Effect of counter cation 43
2.4.2 Effect of temperature and proton to molybdenum ratio 44
2.4.3 Effect of water 45
2.4.4 Structure determination by XRD and TEM 47
2.4.5 Raman and DRIFTS 50
2.4.6 TPRS 51
2.5 Conclusion 53
2.6 Tables 54
3 In situ Raman investigation of the decreasing pH preparation
method leading to various MoO3structures 74
3.1 Introduction 74
3.2 Experimental 75
Preparation 75
Raman spectroscopy 76
3.3 Results 76
3.3.1 Reaction at 30 °C 76
3.3.2 Reaction at 50 °C 77
3.3.2 Reaction at 70 °C 78
3.4 Discussion 79
3.5 Conclusion 83
4 Nanoclusters as Precursors to (MoVW)5O14: In situ and
chemical characterisation
of the systems of a single phase oxidation catalyst 87
4.1 Introduction 87
4.2 Experimental 88
4.3 Results 90
4.3.1 UV/Vis spectroscopy 90
4.3.2 Conductivity measurements 92
4.3.3 95Mo NMR spectroscopy 93
4.3.4 ESR spectroscopy 96
4.4 DISCUSSION 97
4.5 CONCLUSIONS 100
5 The Synthesis and Structure of a Single Phase, Nanocrystalline
MoVW Mixed
Oxide Catalyst of the Mo5O14-Type 105
5.1 Introduction 105
5.2 Experimental 107
5.3 Results and discussion 109
5.3.1 SEM 109
5.3.2 Particle size distribution and BET surface area 110
5.3.3 TG-MS 110
5.3.4 XRD 112
5.3.5 HRTEM 116
5.3.6 Raman Spectroscopy 119
5.3.7 Catalytic properties 123
5.4 Conclusion 126
6 Conclusion and Outlook 133
6.1 Structure - activity relation 133
6.2 Preparation and reactions in solution 134
6.3 Outlook 135
1
From molybdenum based model catalysts
to technically applied systems
1 Introduction
1.1 GENERAL INTRODUCTION
About one quarter of added value world wide is produced via catalytic reactions that
originate from partial oxidation reactions and contribute to the gross national product
of industrialised countries[1]. Many catalytic systems are based on molybdenum
oxides. Common reactions are the partial oxidation of propane, propene and acrolein.
The availability and price of the feed stock mentioned above depend largely on
economic conditions, caused by changes in refinery technique, resources (natural
gas) and political interests in oil producing countries. Propane however needs
different catalysts e.g. a molybdenum based catalyst with vanadium, tellurium and
niobium (MoVTe) for partial oxidation reactions[1;5;6], whereas the acrolein oxidation
is carried out on V and W promoted Mo suboxides[2-4]. Test reactions in this thesis
were the partial oxidation of propene (first part) and the partial oxidation of acrolein
(second part).
High performance catalysts have been developed for these reactions and optimised
by empiric methods. High throughput experimentation (precipitation and testing) is
very successfully applied for development and testing of high performance systems.
However many important details are not monitored by this approach. Effects caused
by minor amounts of additives cannot be precisely recorded. Further parameters
which are often considered to be minor but decisive for the preparation are the speed
of addition, concentrations of precursor solutions, concentration gradients, speed of
stirring, type and form of stirrer, temperature, size of the reaction vessel, solubility,
super-saturation, rate of precipitation, mode of operation and mixing sequence. These
2
influences are difficult to control and to elaborate in detail scientifically in such
experiments. The major disadvantage of high throughput experimentation (hthe) is
that it does not improve the understanding of the chemistry behind the preparation
process. Consequently high throughput experimentation is most effective with a
knowledge based support and knowledge based key experiments.
Only a knowledge based approach will make straightforward developments of new
processes possible and improve industrial catalysts. The ultimate goal of this work
was to contribute to the fundamental understanding of partial oxidation catalysis i.e.
to understand the function of a catalytically active material. One step towards this
goal was the identification of a structure reactivity relation. Suitable model catalysts
have been prepared because industrial systems are both structurally and chemically
complex[2;4;7-15].
In order to rationalise and explain catalytic reactions various theories are discussed in
the literature such as spill over phenomena, site isolation and phase cooperation
[1;16;17]. To test these theories and to generate more knowledge about catalytic
reaction mechanisms the focal point of this thesis was to provide well defined
materials with varying degrees of complexity.
Structural Inorganic Chemistry was a good starting point for the development of
preparation strategies for larger scale functional materials such as catalysts based on
the paradigm of a principle structure-reactivity relationship. In order to be a useful
candidate for amodel catalyst more than just the structural information is necessary.
Relevant for catalysis is the `real structure´ of the material. This includes knowledge
about nanostructuring, particle size, composites, surface phenomena and amorphous
parts. Only after having gained thorough information about this structurally complex
systems and the influence of the addition of vanadium and tungsten can be
understood (chapter 4 and 5).
This enables a knowledge based approach of catalyst preparation and the specific
improvement of catalytic properties. This can be seen as an iterative improvement
starting with well defined and documented preparation and activation. After a
thorough ex- and even more important in situ characterisation of the material the
second iterative cycle started with the preparation of the identified active phases in a
pure form. Having this well characterised material in hand a structure reactivity
relation was established which was refined in the following cycles.
3
This strategy enabled generation of information about industrially relevant processes
under realistic conditions (e.g. reaction temperature and pressure). The obtained
material was more similar to the industrial catalysts than single crystals applied in
UHV and surface science studies but still well defined enough for scientifically
valuable conclusions. Consequently the results are relevant for industrial multi
million ton processes.
1.2 MO-ONLY MODEL SYSTEMS
Structural changes of molybdenum species in solution were investigated as a
function of pH under preparative conditions. Precursors such as ammonium hepta
molybdate (AHM), sodium molybdate (Na2MoO4), lithium molybdate (Li2MoO4)
and potassium molybdate (K2MoO4) were dissolved in water. The pH was decreased
by adding HNO3. Precipitation led to solid catalyst precursor. Another method
applied to obtain a solid was spray drying[18; 19].
Depending on the preparation conditions the molybdenum oxygen systems
developed structures which were also identified in high performance catalysts used in
production plants. A hexagonal molybdenum oxide phase showed a structure similar
to the M 2 phase of the MoVTeNb system used industrially for the partial oxidation
of propane[1;16;17;21-24]. Other authors found the same structure and named it
Mo5O16[20].
Another example is Mo5O14 which is identified as active phase of an industrial
MoVW Oxide catalyst for the selective partial oxidation of acrolein or propene to
acrylic acid. The characteristic structural motif is a pentagonal bipyramid. This motif
is also found in Mo36O1128-. Therefore the model system allows studying structurally
highly complex materials with low chemical complexity.
The complex parameter space such as temperature, molybdenum concentration, and
acid concentration was investigated with an automated titration machine which is an
excellent tool for fast screening. The influences of up scaling were examined with a
reaction vessel one order of magnitude larger than the titrator.
One important aim of chapter two was further to bridge gaps between Structural
Inorganic Chemistry in aqueous solution and Solid State Chemistry and further
between Solid Phase Chemistry and functional materials.
4
Very early investigations on aqueous molybdenum systems were carried out[18; 19].
The main methods used at this time were determination of diffusion coefficients,
UV/Vis and conductometric-, potentiometric-, and thermometric–titrations. Saski
and Sillen presented the first comprehensive potentiometric investigation[25]. A
milestone was the identification of the hepta and octa molybdate structures by X-ray
single crystal analysis[26-28].
Significant contributions were made by Aveston and Anacker and also by Tytko and
Glemser[29-37]. They combined Raman spectroscopy and single crystal analysis,
which enabled them to compare solid state structures with compounds in solution.
Contributions to reveal details of the reaction mechanism were delivered by O-NMR
spectroscopy[38-40]. Recently Cruywagen published more potentiometric data and a
computer aided evaluation[41-43]. All experiments were carried out at room
temperature with diluted solutions.
With the help of these data four different structures were obtained on a preparative
scale. Using TEM it was shown that so called `minor species´ in solution obviously
play an important role in the solidification process. Therefore the role of integral
analytical techniques such as Raman spectroscopy and XRD were reflected critically.
One task of this work was to test whether this knowledge is applicable under
preparative conditions. This was achieved by carefully controlling parameters such
as pH, conductivity, temperature etc. in small and large scale reactions. To bridge the
above described gaps the solid phase chemistry of the obtained catalyst precursors
was investigated by TG/DSC-MS and the catalytic behaviour was monitored by
temperature programmed reaction spectroscopy (TPRS).
1.3 MULTI METAL OXIDES
The second part of this work was dedicated to the more complex system
(Mo0.68V0.23W0.09)5O14. Based on the knowledge generated by using the model
systems a single crystalline phase was obtained. This procedure was investigated by
a variety of analytical techniques such as in situ Raman spectroscopy, pH-
measurements, conductivity-measurements and UV/Vis spectroscopy. In order to
follow the preparation in detail a novel in situ method was developed for UV/Vis
measurements based on a reflection fibre optic.
5
The reaction mechanism that has been set up for the binary model system was in
principle transferred to the complex system. The role of vanadium was identified as
the linking element between isomerising octamolydate species.
From the catalytic point of view it was shown that the very ill defined phase mixture
in the industrial catalyst can be reduced to only one catalytically relevant crystalline
phase, namely Mo5O14. Additional phases are most likely just side products of the
preparation procedure.
However, it was shown that a single crystalline phase theory is too simple to explain
the whole catalytic process as differences in catalytic activity were detected with
compounds of identical stoichiometry and XRD pattern. Similar to the model case
amorphous overlayers were detected on the single phase material. Therefore the
approach to control the real-and nano structure is much more important than the
influence of additives.
Comparing this thesis to traditional catalyst research and optimisation it is striking
that almost only additives are changed. Here however a novel approach is presented.
Carefully chosen model catalysts which contain only molybdenum and oxygen were
described. Several reasons support this choice. Firstly, much catalytic data is
available on orthorhombic MoO3. This has been reviewed lately by Haber[44-46].
Secondly, as nanostructuring oxygen mobility and vacancies are considered to play a
major role in catalysis (Mars van Krevelen) molybdenum oxides are good candidates
for investigation. Related to this are substoichiometric oxides so called Magneli
phases[47-55]. Therefore on the one hand this work is the logical consequence of
preceding work carried out in this department on molybdenum oxides e.g. three
preceding Ph.D. theses by Dieterle[56], Wienold[57] and Blume[58] and numerous
publications[11;15;59;60]. On the other hand this preparative approach opens the way to
check recent theories on tailor like materials.
Additionally molybdenum oxides are very sensitive to changes of the preparation
conditions. Therefore an investigation of this system is absolutely needed. The role
of additives can only be understood after the effects of the matrix are known
completely.
6
Reference List
[1.] R. K. Grasselli, Catal.Today 1999, 49 141-153.
[2.] Hibst, H. and Unverricht, S. BASF AG. [DE 19815281 A 1]. 1981.
Ref Type: Patent
[3.] Hibst, H. and Unverricht, S. BASF AG. [DE 19815281 A 1]. 1981.
Ref Type: Patent
[4.] Tenten, A., Martin, F.-G., Hibst, H., Marosi, L., and Kohl, V. BASF AG. [EP 668104 B1].
1995.
Ref Type: Patent
[5.] R. K. Grasselli, J. D. Burrington, Advances in Catalysis 1981, 30 133.
[6.] R. K. Grasselli, Surface Reaktion Mechanisms of Selective Olefin Oxidation over Mixed
Metal Oxides, (Eds.: M. Che, G. C. Bond) Elsevier, Amsterdam 1985, p. 275.
[7.] Kurtz, A. N., Cunningdam, R. W., and Naumann, A. W. Union Carbide Co. [US 4111983].
1978.
Ref Type: Patent
[8.] Novak, V., Sokol, L., and Jelinek, J. CS1207807B. 1981.
Ref Type: Patent
[9.] Bertolini, N. and Ferlazzo, S. Euteco Impianti S.p.A. [US 4289654]. 1981.
Ref Type: Patent
[10.] Kawajiri, T., Uchida, S., and Hironaka, H. Nippon Shokubai Kagaku. [EP 427 508 A1].
1991.
Ref Type: Patent
[11.] M. Dieterle, G. Mestl, J. Jäger, Y. Uchida, R. Schlögl, J.Mol.Catal.A: Chemical 2001, 74
169.
[12.] M. Dieterle, G. Mestl, Phys.Chem.Chem.Phys. 2002, 4822.
[13.] S. Knobl, G. A. Zenkovets, G. N. Kryukova, O. Ovitser, M. Dieterle, R. Schlögl, G. Mestl,
J.Catal. 2003, 215 177-187.
[14.] G. Mestl, Ch. Linsmeier, R. Gottschall, M. Dieterle, J. Find, D. Herein, J. Jäger, Y. Uchida,
R. Schlögl, Journal of Molecular Catalysis 2000, 162 455-484.
7
[15.] T. Ressler, O. Timpe, T. Neisius, J. Find, G. Mestl, M. Dieterle, R. Schlögl, Journal of
Catalysis 2001, 191 75-85.
[16.] R. K. Grasselli, Topics in Catalysis 2001, 15 93.
[17.] R. K. Grasselli, Topics in Catalysis 2002, 21 79.
[18.] G. Jander, K. F. Jahr, W. Heukeshoven, Zeitschrift fur Anorganische und Allgemeine Chemie
1930, 194 383-428.
[19.] A. Rosenheim, Zeitschrift fur Anorganische und Allgemeine Chemie 1916, 96 139.
[20.] B. Krebs, I. Paulat-Böschen, Acta Cryst. 1976, B32 1697.
[21.] Hatano, M. and Kayou, K. [EP 318295]. 1988.
Ref Type: Patent
[22.] T. Ushikubo, K. Oshima, A. Kayou, M. Hatano, Study of Surface Science and Catalylis,
(Eds.: C. Li, Q. Xin) Elsevier, Amsterdam 1997, p. 473.
[23.] M. M. Lin, Applied Catalysis A-General 2003, 250 305-318.
[24.] M. M. Lin, Applied Catalysis A-General 2003, 250 287-303.
[25.] Y. Sasaki, L. Sillén, Arkiv För Kemi Mineralogi och Geologi 1968, 29 253.
[26.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 18 325-341.
[27.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 2349.
[28.] I. Lindquist, Acta Cryst. 1950, 3159.
[29.] J. Aveston, E. W. Anacker, J. S. Johnson, Inorganic Chemistry 1964, 3735-746.
[30.] I. Böschen, B. Buss, B. Krebs, O. Glemser, Angew.Chem. 1973, 85 409.
[31.] O. Glemser, W. Holznagel, A. Iftikhar, Z.Naturforsch. 1965, 20b 192.
[32.] K.-H. Tytko, B. Schönfeld, B. Buss, O. Glemser, Angewandte Chemie 1973, 85 305-307.
[33.] K.-H. Tytko, B. Schönfeld, Zeitschrift fur Naturforschung 1975, 30b 471-484.
[34.] K.-H. Tytko, O. Glemser, Adv.in Chem.Series 1976, 19 239-315.
[35.] K.-H. Tytko, O. Glemser, Adv.Inorg.Chem.Radiochem 1976, 19 239.
[36.] K.-H. Tytko, G. Baethe, E. R. Hirschfeld, K. Mehmke, D. Stellhorn, Z.Anorg.Allg.Chem.
1983, 503 43-66.
8
[37.] K.-H. Tytko, G. Baethe, Z.Anorg.Allg.Chem. 1987, 555 85-97.
[38.] V. W. Day, M. F. Fredrich, W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1977, 99 6146.
[39.] W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1976, 98 8291.
[40.] W. G. Klemperer, Angew.Chem. 1978, 17 246.
[41.] J. J. Cruywagen, J. B. B. Heyns, Inorganic Chemistry 1987, 26 2569.
[42.] J. J. Cruywagen, Advances in Inorganic Chemistry 2000, 49 127.
[43.] J. J. Cruywagen, A. G. Draaijer, J. B. B. Heyns, E. A. Rohwer, Inorganica Chimica Acta
2002, 331 322.
[44.] J. Haber, Catalysis by Transition Metal Oxides, (Eds.: R. Grasselli, J. Brazdil) American
Chemical Society, Washington D. C. 1985, p. 1.
[45.] J. Haber, 3 rd World Congress on Oxidation Catalysis, (R.K.Grasseli, S.T.Oyama,
A.F.Gaffney, J.E.Lyons) 1997, 1997 Elsevier, Amsterdam 1.
[46.] J. Haber, V. Turek, Journal of Catalysis 2000, 190 320-326.
[47.] G. Andersson, A. Magnéli, Acta Chemica Scandinavica 1950, xxx 793-797.
[48.] A. Magnéli, Arkiv För Kemi Mineralogi och Geologi 1946, 24 a1.
[49.] A. Magnéli, Acta Chemica Scandinavica 1948, 2501-517.
[50.] A. Magnéli, Acta Chemica Scandinavica 1948, 2861-871.
[51.] A. Magnéli, G. Andersson, B. Blomberg, L. Kihlborg, Analytical Chemistry 1952, 24 1998.
[52.] A. Magnéli, Research 1952, 5394.
[53.] A. Magnéli, Acta Cryst. 1953, 6495.
[54.] A. Magnéli, G. Andersson, Acta Chemica Scandinavica 1955, 91378-1381.
[55.] A. Magnéli, B. Blomberg-Hansson, L. Kihlborg, G. Sundkvist, Acta Chemica Scandinavica
1955, 91382.
[56.] Dieterle, M., TU Berlin, 2000.
[57.] Wienold, J., 2003.
[58.] Blume, A., TU Berlin, 2004.
10
2 Controlled preparation of molybdenum
oxide catalysts
2.1 INTRODUCTION
The production of high value petrochemicals such as acrolein or acrylic acid is carried
out by selective oxidation of lower alkanes over molybdenum oxide based catalysts[1-9].
Empirical optimisation of the catalyst performance through doping with foreign cations
has already reached a high level again stressing the huge potential of such catalysts.
There is a number of recent patents for a material with the approximate composition of
Mo1V0.33Te0.22Nb0.11Ox[10-15]. Characterisation of this material has led to the
identification of two major phases amongst other minor phases. The first one (M1) is
orthorhombic and isostructural of Cs0.7(NbW)5O14[16] and shows a great similarity to
Mo5O14 because of its pentagonal bipyramides as the most striking feature. The second
phase (M2) is pseudo hexagonal[11] and isostructural to Sb4Mo10O31.
Although additives are proven to enhance catalytic activity there is an ongoing debate
about whether they act as structural builders or linkers and therefore favour a certain
type of structure or they form active sites themselves. Grasselli[17-19] has outlined `seven
pillars of selective oxidation´, in which `phase cooperation and site isolation´ are two of
his major concepts. As testing these concepts on multi component and multi phase
industrial catalysts is rather difficult, there is a need for good model catalysts. A
breakthrough was achieved recently by the successful synthesis of single phase
(MoVW)5O14 material. This synthesis was done with a nanostructured precursor as
intermediate in which vanadyl linkers connect octamolybdate species, that were
distorted by tungsten atoms[20]. During the catalytic reaction at elevated temperature the
material undergoes structural and electronic changes and its activity and selectivity are
enhanced. The post mortem sample shows the Mo5O14-type structure, which was first
discovered by Kihlborg[21]. This work led to the conclusion that the function of V and
W atoms is only to form and stabilise the catalytically active site of Mo5O14. This
assumption would be further validated, if a solid containing such sites would be
successfully prepared containing only molybdenum and oxygen. This material should
then have a similar catalytic activity to the industrial (MoVW)5O14 catalyst.
11
Another matter of debate is whether the catalytic reaction happens on bulk-terminated
oxide, as can be deducted from Grassellis work, or on differently coordinated Mo oxide
surface layers. The latter statement can be supported by studies from Wachs et al. who
investigated methoxy chemisorption on monolayer supported Mo oxide prepared by
incipient wetness[22]. They found a correlation between the monolayer of methoxy
species of the surface and the number of active surface sites. These sites were formed by
transformation of crystalline (bulk) metal oxides into active surface metal oxide species
by reaction-induced spreading. This phenomenon clearly depends on the nature of the
support. Bell et al. identified the active surface species as two dimensional MoOx
polymers, who are favoured at low surface densities <4.5 nm per Mo. Formation of
Mo6+ centers and oxygen vacancies as active sites was observed for ODH of propane.
Increasing density resulted in a decline in catalytic activity per Mo atom, because of
transformation into three dimensional MoO3structures with inaccessible Mo species[23-
27]. Small additions of alkali (<0.2 per Mo) do not lead to any structural changes,
however, some electronic modifications are detected. Haber et al. performed kinetic
measurements of propene oxidation on bismuth molybdates and postulated the existence
of two different oxygen species at the surface that carry out different functions during
the catalytic process[28]. Whilst lattice oxygen is responsible for the partial oxidation of
propene into acrolein or acrylic acid, total oxidation is carried out by surface oxygen
species that are in equilibrium with the gas phase.
The results mentioned above generally imply that the active species consist of
nanostructured binary molybdenum oxides. In order to test this hypothesis any influence
of dopants or support on the catalytic reaction has to be ruled out, therefore it is highly
desirable to synthesise unsupported and undoped MoOxpolymers and test their catalytic
activity against the industrial catalysts and as well against bulk MoO3. A variety of such
polymolybdates has been successfully obtained by controlled precipitation. The first
prominent compound of this type was reported by Böschen et al. in K4Mo36O112 *8
H2O[29-33]. Like the Mo5O14 and M1 industrial catalysts, this species also contains the
pentagonal bipyramide as a structural motif, that is recognised as an important building
block[34] for large polymolybdates (keplerates, >500 atoms). Even in the photochemical
molybdenum blues reaction this motif is discussed[35].
The approach used in this paper was to obtain the pentagonal bipyramides directly from
precipitation of nano sized clusters. The aim was to directly influence the nature of the
nano clusters by careful adjustment of the reaction conditions. A variety of control
12
variables have been reported in earlier work, dealing with pH dependent reactions of
Mo compounds in aqueous solution[36-38] [39-46] [47-50] [51;52] [53] [34] [52;54-62] [63-66] [67-69].
These variables are extrapolated on a preparative scale in order to ensure the production
of different structural families. Clusters containing these pentagonal bipyramides should
show catalytic activity without any need for calcination or activation, as these
procedures cause uncontrolled reformation of the clusters, leading to pure ortho-MoO3
revealed in its defect-free form, if the temperature is raised above 500 °C. This material
shows unsuitable local chemical bonding for selective oxidation[9;70-72] and requires
tuning by the introduction of defects that can be achieved by heating the sample above
the limit of oxygen diffusion to the surface. Finally up scaling is an important factor to
ensure a reproducible production of catalytic material especially when diffusion plays
an important role in the reaction mechanism.
2.2 EXPERIMENTAL
Preparation
All solids were obtained by precipitation. Small-scale preparation (200 ml) to scan the
parameter range was carried out in a Mettler-Toledo DL 77 Titrator to which an
automated Rondo 60 sample changer was attached. The Rondo was modified in such a
way that a water bath maintained constant temperature conditions. As starting material
aqueous solutions of (NH4)6Mo7O24 * 4 H2O (AHM), Na2MoO4, Li2MoO4, K2MoO4(all
MERCK, p.a.) at a concentration range between 0.28 mol/l and 2 mol/l calculated on
Mo was used. As precipitation agent HNO3(1 mol/l – 5 mol/l) was applied. The
experiments were carried out at 30 °C, 50 °C and 70 °C.
In a separate set of experiments the effects of up scaling have been studied by using a
home-made 4l computerised semi-technical preparation set-up allowing to control the
temperature accurately and to measure pH and electrical conductivity. The same
concentration ranges have been used as before, the starting solution was 1l. The pH and
was recorded and analyzed digitally. After the addition of HNO3the samples were
allowed to age for 1 h at 35 °C. All samples were filtered and dried in a desiccator over
dry gel. It is crucial not to wash the samples, as this would cause partial re- dissolution.
13
XRD
A STOE STADI-P focusing monochromatic transmission diffractometer equipped with
a Ge(111) monochromator and a position sensitive detector. Cu-Karadiation was used.
The phase analysis was performed with the STOE Win XPOW software package
(version 1.06; Stoe Darmstadt, Germany) and with PowderCell (V 2.3; Bundesanstalt
für Materialforschung und -prüfung (BAM) Berlin, Germany).
TEM
The samples were prepared for transmission electron microscopy (TEM) by standard
preparation routines. The powder is suspended in ethanol and dispersed onto a standard
meshed copper grid coated with a holey carbon film, by dipping the grid into the slurry.
The samples are studied in a Philips CM 200 FEG electron microscope operated at
200~kV and equipped with a Gatan Image Filter and a CCD camera. Scanning electron
microscopy (SEM) images are acquired with an S 4000 FEG microscope (Hitachi).
Raman DRIFTS
Raman spectroscopy was performed on a Labram I (Dilor) instrument equipped with a
confocal microscope (Olympus). A notch filter (Kaiser Optical) was applied to cut off
the laser-line and the Rayleigh scattering up to 150 cm–1. The spectrometer is equipped
with a CCD camera (1024*298 diodes), which is Peltier cooled to –30 °C to reduce the
thermal noise. A He-laser (Melles Griot) was used to excite the Raman scattering at 632
nm with a laser power of 1,4 mW. The following spectrometer parameters were used:
microscope objective: 10; slit width: 200 µm (spectral resolution: 2.5 cm-1), integration
time: 30 s per spectrum and 5 averages.
DRIFTS were recorded on a Bruker ifs 66 FTIR-spectrometer equipped with OPUS
software. 100 spectra were averaged and the resolution was 4 cm-1. The spectra were
recorded in air. For background measurement KBr was used. The aperture was set to 5
mm, scan velocity 7; 20 kHz.
UV/Vis/NIR
A commercial UV-Vis-NIR spectrometer (Lambda 9, Perkin Elmer) equipped with an
integrating sphere was supplemented with a construction to measure in situ diffuse
reflectance spectra from room temperature to 450 °C. A full description of the complete
setup has been given elsewhere[73].
14
BET
BET was carried out on a Quantachrom Autosorb-1 apparatus. Degassing of the sample
was carried out at 120 °C for two hours. 0.3 g of sample was weighed in and nitrogen
was used as adsobate.
TG
Thermal analysis (TA) was performed with a STA 449 C Jupiter apparatus (Netzsch).
Flowing helium and air atmosphere were applied (the flow rate was set at 15 ml/min in
both cases). The heating rate was set at 5 °C/min. Mass spectrometric analysis (TG-MS)
of the evolved gases was performed with an Omnistar quadrupole mass spectrometer
(Pfeiffer Vacuum).
TPRS
The test reaction for TPRS was the partial oxidation of propene. The instrumentation for
the TPRS runs consists of a simple tubular reactor. The sample (50 mg) is diluted in a
matrix of BN and SiC (for amelioration of heat transfer).
The total gas flow is set to 100 ml/min and the resulting hsv ranges about 13500. The
feed is composed of 10 % propene and oxygen each in He. The temperature is ramped
twice to 500 °C by 5 °C/min in two successive cycles.
2.3 RESULTS
Depending on preparation conditions such as temperature, molybdenum concentration,
acid concentration and the choice of starting material and counter ion such as Li+, Na+,
K+, NH4+four different families of products were obtained by precipitation:
orthorhombic MoO3, hexagonal MoO3, trimolybdate and a supramolecular Mo36O1128-
like compound (Fig. 2.1).
15
2.3.1 Orthorhombic MoO3
Orthorhombic MoO3is only obtained with Li+as counter ion. Regardless of the
preparation temperatures further heating to the boiling point was necessary to force
precipitation.
pH-dependency
Lithium monomolybdate shows typically an initial pH at around 6.5 (Fig. 2.2).
Acidification with nitric acid leads first to a sudden pH drop followed by a buffering
region around pH = 5.5 and a further drastic pH drop down to pH = 2. This pH curve is
very temperature dependent as the sample treated at 50 °C needs the smallest amount of
acid in this series to reach its final pH = 1. The second largest acid volume is used for
the sample at 70 °C. The 30 °C sample needs the largest amount of acid. In Fig. 2.2 the
amount of acid added is normalised to the acid concentration and to the molybdenum
concentration. Therefore all pH data shown in this plot can be directly compared.
Figure 2.1: Lead stru
ctures of the obtained materials identified by XRD and
Raman; a) supramolecular compound; b) hexagonal MoO3
; c) trimolybdate;
d) orthorhombic MoO3
16
XRD
All three samples show the pattern of orthorhombic MoO3however poorly crystalline.
One typical pattern is shown in Fig. 2.4 B (222). The indexing was done according to
PDF 5-508.
Raman
Bands are detected at 996, 825, 667, 376, 335, 292 and 244 cm-1. This is in line with the
bands reported in the literature[74] for orthorhombic MoO3apart from the band at 825
cm-1, which should be at 820 cm-1. Further a shoulder is detected at 980 cm-1, which
does not belong to MoO3. Fig. 2.5 (222) shows a typical Raman spectrum, the other
samples are listed in Table 2.6.
DRIFTS
Bands are found at 840, 743, 1000, 1387, 1636, 1946, 2389, 3584 cm-1. The band at
3584 cm-1 is assigned to OH stretching and deformation frequencies of hydrated water
respectively. The band at 1636 cm-1 suggests that water is present as water molecules as
well as hydroxyl[75-77]. Representative spectra are shown in Fig. 2.9 and 2.10.
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4
1
2
3
4
5
6
7251 30 °C
222 50 °C
230 70 °C
pH
normalised [HNO3]:[Mo]
Figure 2.2: pH curve for samples (LiMoO40.28 mol/l HNO3
2 mol/l) 251,
30 °C; 222, 50 °C; 230, 70 °C. Data normalised to molybdenum and acid
concentration respectively
17
UV/Vis NIR
Bands in the NIR region are detected at 1430, 1790, 1930 and 2280 nm as seen in Fig.
2.25. The sample prepared at 50 °C shows the biggest band areas and a band gap energy
of 3.37 eV. The band gap energy of the other two samples is 3.48 eV (Fig. 2.7 I, II).
Results are listed in Table 2.4. A strong band at 970 nm is observed in the samples
prepared from 0.28 mol/l Li2MoO4with 2 mol/l HNO3at 30 oC and 50 oC. Such a band
cannot be observed for all other samples with hexagonal MoO3, supramolecular
Mo36O1128-, and trimolybdate structure as can be seen in Fig. 2.13.
Thermal analysis
Although precipitation was only achieved after boiling of the aqueous solutions, the
precipitation temperature seems to have an effect on the behaviour during calcination.
The sample (Fig. 2.3; 222) prepared at 50 °C shows the most characteristic events in the
TG-MS. The biggest overall mass loss is detected for the sample prepared at 50 °C (7
%) and the last mass loss at 320 °C is most pronounced and accompanied by a
significant endothermic signal. This sample shows a peak in the water signal (MS) at 75
°C. After that water is evolved constantly up to 240 °C. NOXat 320 °C is evolved. The
intensity of the water signal is nicely reflected by the DSC trace. The maximum of the
water MS trace coincides with a minimum in the DSC curve. As soon as the water MS
trace is declining the DSC curve is getting steeper.
Relative to the other two samples the NOXsignal is much bigger. The TG traces of the
samples prepared at 30 °C and at 70 °C are similar showing an almost constant mass
100 200 300 400 500
90
91
92
93
94
95
96
97
98
99
100 230
222
Temperature [oC]
TG [%]
-0.30
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
243
DSC [mV]
Figure 2.3: TG-DSC of 222, 230 and 243
18
loss up to 266 °C and hardly any mass loss up to appr. 310 °C. From 372 °C upwards no
more mass losses are detected. Some decisive differences however appear in the DSC
trace. The sample prepared at 30 °C shows an exothermic transition at 238 °C which
coincides with the declining water MS trace. As soon as no more water is present the
DSC trace is changing its curvature. An endothermicity is detected at 352 °C at the
same time NOXis evolved. In contrast to that the sample prepared at 70 °C shows a
plateau in the DSC curve at 311 °C. From this temperature on no more water is evolved.
At 330 °C water is leaving the sample.
TPRS
The obtained orthorhombic MoO3starts to convert propene to CO2at around 300 °C. At
350 °C a plateau is reached. Another rise in activity is observed at 370 °C and at 430 °C
the maximum CO2production is measured (Fig. 2.20 and 2.21).
2.3.2 Trimolybdate
Trimolybdate is obtained with K+ as counter ion and at a reaction temperature of 70 °C
and a minimum molybdenum and acid concentration of 2 mol/l. In this investigation
trimolybdate was obtained with K2MoO4 2 mol/l, HNO32 mol/l and 5 mol/l. The
compound precipitates spontaneously (sample 244 and 243).
pH dependency
The titration starts at around pH = 9, buffers at around pH = 7 and reaches the final pH
= 1 asymptotically (Fig. 2.18).
XRD
Using K2MoO42 mol/l and HNO32 mol/l the XRD pattern (Fig. 2.4, sample 243)
indicates a phase mixture containing hexagonal MoO3, trimolybdate and at least one
more phase. The interpretation is difficult because peaks can shift depending on the
preparation conditions.
19
It is reported[46;78] that cation spaces could be either occupied by H3O+or K+. Depending
on this crystal parameters will change. However, it cannot be ruled out that more phases
are present. Increasing the acid concentration increases the amount of trimolybdate.
Figure 2.4: A XRD pattern of represent
ative samples. 256, supramolecular
phase from AHM; 227, hexagonal MoO3from AHM; 229, hexagonal MoO
3
from K2MoO4; 243, supramolecular from K2MoO4;
B 222, orthorhombic MoO3from Li2MoO4
; 245, supramolecular compound
10 20 30 40 50 60 70
A
002
310
111
501
200
110
100
210
243
229
227
256
diffraction angle
intensity
10 20 30 40 50 60 70
B
200
060
111
021
040 110
002
020
245
222
intensity
diffraction angle
20
Raman
The main bands are observed at 949, 938, 909, 612, 372 and 217 cm-1. Taking the
intensity pattern into account it fits with the literature values reported for K2O * 3 MoO3
* x H2O (Fig. 2.5). However there is a slight, non-systematic shift in band positions. In
the literature a triad of very sharp bands is reported. This triad is not well resolved when
HNO32 mol/l is used as precipitation agent and a further shoulder is detected at 873 cm-
1,which is not reported in the literature. The band detected at 1051 cm-1 is assigned to
nitrate from the precipitation agent. Again an increase of the acid concentration leads to
a purer product and the band positions fit exactly with the ones reported in the literature.
DRIFTS
Bands are detected at 726, 804, 880, 955, 1337, 1447, 1633, 2170, 2383, 2754, 3298,
3532 cm-1. The assignment of the water bands corresponds to the one done above for
the orthorhombic MoO3. At lower wavenumbers different bands are detected. Therefore
IR can discriminate among these compounds (Fig. 2.9 and 2.10).
1000 800 600 400 200
0222
0243
0245
0229
0227
0256
intensity
Raman shift (cm-1)
Figure 2.5: Raman spectra of representative samples. 256, supramolec
ular
phase from AHM; 227, hexagonal MoO3from AHM; 229, hexagonal MoO3
from K2MoO4; 245, supramolecular from K2MoO4
; 243, trimolybdate; 222,
orthorhombic MoO
3
21
UV/Vis/NIR
NIR bands are detected at 1435 and at 1935 nm with quite low intensity. In addition two
weak bands at 1790/1810 nm and stronger bands at 2000, 2095 and 2250 nm were
detected. Besides a small LMCT centred at 284 nm and an Eg of 3.77 eV are observed
(Table 2.4). The band gap energy in the sample prepared from 2 mol/l K2MoO4 with 2
(5) mol/l HNO3at 70 oC is 3.77 eV. The sample prepared from the higher concentrated
acid (5 mol/l) shows a band gap energy of 3.77 eV.
Thermal Analysis
Trimolybdate (Fig. 2.3, sample 243) reaches mass constancy at 130 °C. At this
temperature a sharp endothermic signal is detected. A further endothermic signal
appears at 333 °C. At around 420 °C the samples start to melt. The two samples do not
show any decisive differences.
2.3.3 Hexagonal MoO3
Hexagonal MoO3[79-82] is obtained with Na+, K+and NH4+as counter ion. It is either
obtained by spontaneous precipitation at higher temperatures or by further heating of the
100 200 300 400 500
92
94
96
98
100
DSC (mV)
TG (%)
Temperature / oC
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
229
219
227
Figure 2.6: TG-DSC of 219, 229 and 227
22
solution at pH = 1 to the boiling point. Finally hexagonal MoO3is only obtained from
rather dilute molybdenum stem solutions.
Employing Na+as counter ion and a molybdenum and acid concentration of 2 mol/l
respectively no reaction temperature yielded hexagonal MoO3spontaneously. After
heating hexagonal MoO3 was obtained (samples 252, 226, 255 and 231).
Potassium as counter ion yielded hexagonal MoO3when the molybdenum concentration
was 0.28 mol/l and HNO32.0 mol/l for the reaction temperatures 50 °C and 70 °C only
after further heating (samples 219, 229, 247 and 233).
Ammonia as counter ion yielded hexagonal MoO3with an AHM concentration 0.7
mol/l (Mo) and HNO31 mol/l at 50 °C and at 70 °C after further heating (samples 227,
232 and 248).
pH dependency
Using Na+as counter ion the starting pH is 8.2 for the reaction at 30 °C (Fig. 2.8).
Adding 2 mol/l HNO3leads to a pH drop. At around pH = 7.5 a buffering sets in, which
is followed by a large pH drop again down to pH = 1. The other two reaction
temperatures show a similar behaviour, however the starting pH is lower and the first
300 400 500 600 700 800
0
2
4
6
8
10
2,5 3,0 3,5 4,0 4,5
0
2
4
6
8
10
296
230
288
251
I
222
306 nm
Kubelka-Munk units
wavelength (nm)
222
3.37
II
251
230
3.48 eV
d(Kubelka-Munk)/d(hv)
Photon energy (eV)
Figure 2.7: Spectroscopic characteristics of orthorhombic MoO3
, 251, 222
and 230. I. UV/Vis spectra; II. band gap energies Eg
23
buffering takes place at around pH = 6.8. It is noteworthy that the sample prepared at 50
°C needs considerably less acid to reach the final pH. The highest amount of acid is
needed for the sample prepared at 30 °C.
In the potassium case the reaction starts at around pH = 7. Buffering occurs at around
pH = 6.5. Apart from that the curve is very similar to the one described above for
Sodium. Again the reaction at 50 °C needs considerably less acid (Fig. 2.18).
The reaction is slightly changing with ammonia as counter ion because the starting
compound is now the heptamolybdate ion rather than the mono molybdate ion.
Therefore the reaction starts at around pH = 5.3. After that the pH is dropping fairly
constantly until pH = 2 is reached. At this point the two pH traces separate. The reaction
at 50 °C reaches pH = 1 fairly quickly whereas the solution at 70 °C shows a stronger
buffering and needs much more acid (Fig. 2.14 and 2.19).
XRD
One representative diffraction pattern of the samples prepared from Na2MoO4 is shown
in Fig. 2.4 A indexed according to PDF-39-35 and ICSD-38415. The pattern fits quite
good to the one reported in the literature apart from a decisive broadening of the signal
at 25.5 2 theta and some additional small signals in the region from 49.3 2 theta to 53.3
2 theta and at around 61 2 theta.
Potassium as counter ion produced clearly hexagonal MoO3. The signals however
possesses shoulders, consequently a second phase with different d-spacings might be
present. Ammonia as counter ion yielded a similar result (Fig. 2.4 A).
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40
2
4
6
8231 70 °C
226 50 °C
252 30 °C
pH
normalised [HNO3]:[Mo]
Figure 2.8: pH curve for samples (Na2MoO42 mol/l, HNO3
2 mol/l) 252, 30
°C; 226, 50 °C; 231, 70 °C. Data normalised to molybdenum and acid
concentration respectively
24
Raman
The Raman spectra recorded from the hexagonal MoO3obtained from Na2MoO4,
K2MoO4and AHM shows bands at 978, 904, 884, 692, 494, 399, 255 and 229 cm-1.
Only minor shifts and changes in band shape occur. Representative spectra are shown in
Fig. 2.5.
DRIFTS
Depending on the counter ion small changes are observed. The main bands are detected
at around 726, 804, 886, 955, 996 (sh), 1360, 1612, 3305, 3571 cm-1. Especially the
region below 1000 cm-1 does not fit exactly to literature values. However in the
literature it is shown that not only Mo-O bond stretching and bending vibrations are
detected but also H2O liberational motions[75]. Therefore even the band positions in this
region depend on the quantity of water and ammonia. However the main features match
(Fig. 2.9 and 2.10).
1000 900 800 700 600
0222
0243
0245
0229
0227
0256
intensity [a.u.]
wavenumber (cm-1)
Figure 2.9: Diffuse reflectance fourier transform spectroscopy (DRIFTS) of
representative samples. 256, supramolecular phase from AHM; 227,
hexagonal MoO3from AHM; 229 hexagonal MoO3from K2MoO4
; 245
supramoleculart from K2MoO4; 243, trimolybdate; 222, orthorhombic MoO3
25
UV/Vis/NIR
Table 2.4 shows the spectroscopic characteristics of samples prepared from 2 mol/l
Na2MoO4with 2 mol/l HNO3at 30 °C, 50 °C and 70 °C, which indicated different band
positions depending on temperature. The spectrum of the sample prepared at 30 °C
shows a LMCT band at about 320 nm. This band is red shifted by 8 nm in the spectrum
of the sample prepared at 70°C and by 12 nm in the spectrum of the sample prepared at
50 °C. All three samples have the same Eg values of 3.27 eV (Table 2.4). In addition
these samples exhibit a very similar behaviour in the NIR region. A very strong
absorption band at 1955 nm, a small band at 1430 nm and relatively weak absorption
bands at 1820 and 2280 nm appear, whereas the sample prepared at 70 °C leads to a
spectrum with slightly higher band intensities. The spectra in the NIR range recorded of
the compound prepared from Na2MoO4are very similar, the main bands are detected at
1430, 1820, 1955 and 2280 nm. Whereas the sample prepared at 70 °C shows the
biggest band areas. Band gap energies are very similar as well. For all three compounds
3.27 eV are found.
Switching to K+as counter ion the bands are located at 1435, 1820, 1935, 2090 and
2235 nm. The band gap energy amounts to 3.30 eV.
1000 2000 3000 4000 5000
0222
0243
0245
0229
0227
0256
intensity [a. u.]
wavenumber [cm-1]
Figure 2.10: DRIFTS of representative samples. 256, supramolecular
phase from AHM; 227, hexagonal MoO3
from AHM; 229, hexagonal
MoO3from K2MoO4; 245, supramolecular from K2MoO4
; 243,
trimolybdate; 222, orthorhombic MoO3
26
Finally the NIR bands for the ammonia-containing compound (0.7 mol/l AHM) are
situated at 1440, 1570, 1945, 2040 and 2150 nm. The band gap energy is 3.35 eV for
the hexagonal compound prepared at 50 °C and 3.36 eV for the one prepared at 70 °C.
Thermal Analysis
Figure 2.11: A) SEM of hexagonal MoO3
, 227; B) TEM, 227; C) HRTEM,
227; D) SEM, 130 large scale; E) HRTEM of 130; F) TEM 130
27
With the sodium-containing compound the 50 °C sample is showing the biggest mass
loss (8.5 %) accompanied by a sharp endothermic signal at 350 °C (Fig. 2.12). All three
samples show a sharp exothermic signal at 378 °C. A sharp endothermic signal is
detected at 307 °C for the sample prepared at 30 °C and at 50 °C. The samples prepared
at 30 °C and at 70 °C show a broad endothermic signal at 331 °C. The sample prepared
at 30 °C evolves NOXat 325 °C and at 380 °C. The 50 °C sample shows only one very
large peak at 350 °C. The sample prepared at 70 °C looses NOXat 340 °C and at 380
°C.
The potassium containing samples show a similar thermal behaviour. The overall mass
loss of the sample prepared at 70 °C is 5 % and it is 6 % of the sample prepared at 50
°C. The TG-trace shows a step at around 100 °C. In between 190 °C and 320 °C the
samples prepared at 50 °C and 70 °C are fairly mass constant and their TG traces run
almost parallel. At temperatures higher than 320 °C a rapid mass loss (2 %) sets in
which is related to an exothermic signal in the DSC curve in the case of the sample
prepared at 50 °C.
Simultaneously, a trace of water and NOXis detected in the MS. The sample prepared at
70 °C does not show a clear step and no characteristic signals in the DSC trace. At
around 350 °C the evolution of water and NOXis detected when the final phase
transition to orthorhombic MoO3sets in. The final phase transition is shifted to lower
temperatures compared to the hexagonal phase prepared with ammonia as counter ion.
100 200 300 400 500
91
92
93
94
95
96
97
98
99
100
101
232
DSC [mV]
0231
0226
Temperature / oC
TG / %
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
Figure 2.12: TG-DSC, sample 226, 231, 232
28
The compound obtained from AHM exhibit an overall mass loss of 7 %. At 400 °C an
exothermic event is recorded coinciding with the evolution of water and NOXand a
small final mass loss (less than 1 %). The mass loss does not show clear steps and the
MS traces water and NOxcontinuously (Fig. 2.6).
In a separate experiment the sample was heated to the temperature of the exothermic
event and left at this temperature for one hour. During this procedure the sample
transformed into orthorhombic MoO3.
SEM/TEM
The morphology of sample 227 is studied with SEM. A characteristic SEM of the
sample is shown in Fig. 2.11 A, B, C. The sample consists of regular hexagonal rods
with a length of several µm and a diameter ranging from 1-2 µm.
500 1000 1500 2000 2500
0,00
0,05
0,10 245
244
233
970
251
252
Kubelka-Munk units
wavelength (nm)
Figure 2.13: NIR bands of 252, 245, 233, 244, 251
29
The particles reveal well-defined facets. However, smaller irregular particles are
observed in the size range below 500 nm. The projected morphology of the hexagonal
rods is recognised in TEM images such as Fig. 2.11.
The small particles are observed here as well. As the well-defined hexagonal shaped
rods not are transparent for 200 keV electrons due to the thickness, lattice fringe
imaging is concentrated on the small irregular shaped particles.
In Fig. 2.11 a characteristic high-resolution image of the small, electron transparent
agglomerates is shown. Randomly oriented clusters of molybdenum oxide reveal lattice
fringes separated by characteristic distances of 0.34 nm and 0.37 nm. The whole area
depicted in Fig. 2.11 seems to be crystalline but it is clear that it is an agglomerate
rather than a single crystal.
TPRS
The most active material in this family is produced from the ammonium containing
precursor. CO2production sets in at 330 °C. A decisive rise in activity is monitored at
420 °C. At 450 °C another plateau is reached and at 480 °C a maximum in activity is
reached. The same material starts to produce acrolein at around 350 °C. The activity is
rising with temperature only a small bend in the trace is observed at 370 °C (Fig. 2.20,
Fig. 2.21, Fig. 2.22, Fig. 2.23).
0,00 0,05 0,10 0,15 0,20 0,25 0,30
0,5
1,0
1,5
2,0
2,5
3,0
3,5
4,0
4,5
5,0
5,5
pH
normalised [HNO3]:[Mo]
256 30 °C
227 50 °C
232 70 °C
Figure 2.14: pH trace of 256, 227, 232. Data normalised to
molybdenum and acid concentration respectively
30
The catalytic behaviour of the sodium containing hexagonal MoO3is much lower.
However the first activity is registered at a quite low temperature 250 °C. At 330 °C and
at 350 °C two local maxima are observed in a comparatively shallow trace. The
maximum activity is found at 480 °C.
The potassium containing compound shows a very similar behaviour however the first
local maximum is more pronounced and the second one is missing.
2.3.4 Mo36O1128- supramolecular molybdenum oxide
The supramolecular compound is obtained with ammonia or with potassium as counter
ion. Usually low temperatures and high molybdenum concentrations need to be
employed (sample 256, 257, 228, 258, 225, 223, 249, 250, 245, 253, 246).
Using potassium as counter ion the supramolecular compound is obtained with K2MoO4
0.28 mol/l and HNO32.0 mol/l at 30 °C, K2MoO4yields the supramolecular compound
at 30 °C and at 50 °C the same is true when the acid concentration is increased to 5
mol/l.
Figure 2.15: A) HRTEM, 74; B) HRTEM, 74; C) SEM, 74, D) TEM, 74
31
Ammonia as counter ion (Fig. 2.14) yields the supramolecular phase when AHM 0.7
mol/l and HNO31 mol/l is used at 30 °C. From a 1 mol/l AHM solution this phase is
precipitated at 50 °C. Employing a 0.7 mol/l AHM solution and HNO32 mol/l
Mo36O1128- is obtained at 50 °C.
pH dependency
Using the 0.28 mol/l K2MoO4at 30 °C the reaction starts at pH = 7. The titration curve
is similar to the one leading to hexagonal MoO3with this concentration. However, in
this case much more acid is needed to reach the final pH. The titration curves of the 2
mol/l solution titrated with the 2 mol/l HNO3start at pH = 7.5. The buffering regime at
around pH = 7 is quite extensive and roughly the same amount of acid is needed to
reach the final pH. The same shape is observed for the case 2 mol/l K2MoO45 mol/l
HNO3. The reaction at 30 °C needs less acid than the one at 50 °C to reach the final pH
(Fig. 2.18).
With ammonia as counter ion (AHM 0.7 mol/l, HNO31 mol/l) the reaction starts at pH
= 5.3, the first buffering is less extensive and a comparatively big amount of acid is
used. In principle the same applies for the other reactions (Fig. 2.14 and 2.19).
100 200 300 400 500
88
90
92
94
96
98
100
DSC (mV)
TG (%)
Temperature / oC
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
0.20
0.25
0.30
228
256
Figure 2.16: TG-DSC of 256 and 228
32
XRD
The supramolecular compounds obtained all showed a variety of signals in the early 2
theta range, all similar to the original single crystal pattern from Böschen et al. [32]. No
sample matched that pattern exactly; the differences in the powder patterns could arise
from different amounts of crystal water present in the precipitates.
In the potassium case one needs to distinguish among the products obtained from high
and from low concentrations. Whilst the low concentration product shows the 100 %
signal at 7 2 theta, the other products show various signals in this region and a 100 %
signal is difficult to determine (details Fig. 2.4).
Raman
The bands for the potassium-containing product are situated at 963, 882, 373 and 229
cm-1 for the low concentration product. Compared to that the bands for the high
concentration product are slightly shifted (961, 898, 372 and 240 cm-1). However the
differences in band shape and relative band intensities are even more decisive.
100 200 300 400 500
90
92
94
96
98
100
DSC (mV)
TG (%)
Temperature / oC
-0,10
-0,05
0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
0,40
245
250
249
Figure 2.17: TG-DSC trace of 250, 245 and 249
33
The ammonia-containing product shows bands at: 983, 959, 889, 372 and 234 cm-1.
This is in line with the bands reported in the literature. The samples show a large
variation in relative band intensities in the region assigned to the terminal Mo=O
stretching frequencies (Fig. 2.5).
DRIFTS
Again small shifts depending on the counter ion are observed: 596, 658, 715, 818, 942,
967, 987, 1305, 1412, 1619, 3579 cm-1 (Fig. 2.9 and 2.10).
UV/Vis NIR
Looking at the NIR spectrum (Fig. 2.13) of the experiment carried out with the 2 (0.28)
mol/l K2MoO4solution and the 2 (5) mol/l HNO3it is to be seen that the compound
prepared at 50 (30)°C shows in Table 2.5 the most intense water (NIR) bands at 1435
nm and at 1935 nm. The compound prepared at 30 °C exhibits slightly smaller bands.
Band gap energies are is 3.30 eV and 3.30 eV. Switching from a 2 (0.28) mol/l acid
(K2MoO4) to a 5 mol/l the water (NIR) bands do not change their behaviour and are still
almost located at the same wavelengths. The band gap energy of both samples is found
at 3.43 eV.
0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4
2
4
6
8
10 249 30 °C
219 50 °C
229 70 °C
250 30 °C
245 50 °C
244 70 °C
246 30 °C
246 50 °C
243 70 °C
pH
normalised [HNO3]:[Mo]
Figure 2.18: pH-trace 249, 219, 229; 253, 246, 243; 250, 245,
244.
Data normalised to molybdenum and acid concentration respectively
34
For the compound prepared from 0.7 and 1 mol/l AHM bands in the NIR region are
located at 1440, 1570, 1945, 2040 and 2150 nm. The first band, the third and the fourth
are assigned to a surface OH group the second to H2O and the last band is caused by
ammonia. The LMCT band is blue shifted from 316 to 308 nm with increasing AHM
concentration and the band gap energy is 3.46 eV (0.7 mo/l AHM) and 3.55 eV (1 mol/l
AHM) as shown in Table 2.5.
Thermal Analysis
The supramolecular phase prepared at 30 °C (0.28 mol/l K2MoO4, HNO32 mol/l) shows
only two sharp steps in the TG-trace (100 °C, 322 °C), the overall mass loss is around 8
% (Fig. 2.17). The DSC curve indicates three small exothermic peaks at 213 °C, 286 °C
and 355 °C. At the endothermic signal and the shoulder in the 100 °C region H2O and
OH is detected by MS. The occurrence of the first exothermic peak at 213 °C is linked
with the evolution of OH and H2O. Between the two signals at 286 °C and 355 °C (at
344 °C) a NOXsignal is detected by MS.
When the same experiment (Fig. 2.17) is performed with a solution containing 2 mol/l
K2MoO4the two supramolecular phases obtained at 30 °C and at 50 °C show almost
identical TG traces (overall mass loss 8 %). Water is removed at 100 °C and this
coincides exactly with the appearance of the endothermic signal including the shoulder,
between 200 and 300 °C a creeping mass loss is detected. Another peak assigned to
NOXin the MS is detected at around 340 °C. After 350 °C mass constancy was
0,0 0,1 0,2
0,5
1,0
1,5
2,0
2,5
3,0
3,5
4,0
4,5
5,0
5,5 227 50 °C
228 50 °C
225 50 °C
hex
pH
normalised [HNO3]:[Mo]
Figure 2.19: pH trace, 227, 228, 225.
Data normalised to
molybdenum and acid concentration respectively
35
achieved. The 50 °C sample shows two distinct exothermic signals at 281 °C and at 343
°C. The gases evolved are the same as in the previous case and at the same
temperatures. The 30 °C sample shows signals at the same positions but less distinct.
The same experiment is carried out only with a higher HNO3concentration of 5 mol/l.
The TG-traces of the two supramolecular compounds are very similar to the ones
described in the preceding paragraph. However the total mass loss of the sample
prepared at 30 °C is comparatively high: 12 %. The sample prepared at 50 °C looses 10
% of its mass. Both samples show a broad endothermic DSC signal at 100 °C where
water is released. At 130 °C both samples show a sharp endothermic signal, which is
unrelated to any mass loss. At 278 °C an exothermic mass loss is detected and only at
the endothermic signal at 340 °C a final mass loss is observed setting NOXfree.
The compound obtained from the AHM solution 0.7 mol/l shows an endothermic event
at around 100 °C, when a large amount of water is removed from the sample (ca. 7 %).
After that a sequence of three exothermic events is registered at 279 °C, 332 °C and at
400 °C. Each of them connected with a mass loss and at each of them water and NOx
are evolved. Total mass loss is 12 % (Fig. 2.16).
In a separate experiment samples were heated to the temperature of each event and then
the temperature was kept constant for one hour. After this the sample was investigated
200 250 300 350 400 450 500
temperature [°C]
signal m/e = 56
Li
Na
NH4
K
Figure 2.20: TPRS first cycle; showing the effect of
counter ions in acrolein production
36
by Raman spectroscopy. It turned out that the supramolecular phase is transformed into
hexagonal MoO3after the first exothermic event at 279 °C. No change in phase is
observed after 332 °C. At 400 °C it transformed into orthorhombic MoO3.
SEM/TEM
A representative SEM image is shown in Fig. 2.15 C. The particles observed in the
SEM images are typically between 1 and 5 µm and show an elongated shape with
facets. However, the facets are not perfect and some irregularities are observed. The
molybdenum to oxygen ratio is found to be close to constant over the sample by EDX
analysis (not shown here).
The SEM images reveal some smaller features (tenths of µm). These features are more
visible in TEM images. In Fig. 2.15 D such an image is shown. The large somewhat
irregular shaped particles are observed in projection in the images. Moreover is the
carbon film covered with small (<200 nm) agglomerates.
A typical high-resolution image of the small agglomerates is shown in Fig. 2.15 A and
B. The image shows the sample consisting of small randomly oriented clusters revealing
200 250 300 350 400 450 500
0
1
2
3
4
5
6
7
temperature [°C]
conversion in oxygen [%]
Li
Na
NH4
K
K amplified by 6.3
Figure 2.21: TPRS first cycle; showing the effect of
counter ions in oxygen conversion
37
lattice fringes encapsulated by amorphous/non-crystalline material. The lattice fringes
are separated with a characteristic distance of 0.37 nm.
TPRS
The most active material prepared for this investigation is the supramolecular
compound (Fig. 2.22 and Fig. 2.23). Its CO2trace is characterised by a rather sharp
local maximum at 230 °C. Catalytic activity rises steeply at 380 °C. Another
local maximum at quite high level is observed at 430 °C. The highest overall
activity towards CO2is seen at the end of the heating ramp.
Acrolein production starts at 380 °C and no preceding local maximum is observed. At
430 °C a turning point in the trace is observed. The maximum activity toward CO2is
registered at the end of the heating ramp.
Figure 2.22: TPRS run first cycle; showing compounds prepared from
AHM starting material; oxygen conversion
200 250 300 350 400 450 500
0
5
10
15
20
conversion oxygen [%]
temperature [°C]
256
257
258
227
223
228
225
232
38
2.3.5 Up scaling
Preparation and pH-dependency
The hexagonal phase and the supramolecular phase were also synthesised in a 4 l
reactor in order to test the effects of up scaling. The pH curves of two supramolecular
samples that were precipitated in the different reactors under otherwise identical
conditions display a very similar behaviour up to pH = 3, further down the `small scale´
sample showed an increased buffering region that can be associated with protonation
reactions. After the titration the `large scale´ sample was allowed to age, as mentioned
above. The Raman spectra showed identical positions for all major bands. XRD analysis
showed both the characteristic signal for the supramolecular phase at 7 degrees 2 theta,
but each sample showed a number of foreign signals.
Electron Microscopy and Electron Diffraction
Well-defined rods are observed in typical TEM images Fig. 2.11 of the hexagonal
material obtained by the up scaling experiments. The typical hexagonal rods are around
5 µm in length and 0.5 µm wide. Smaller agglomerates are observed as well in the SEM
images. However, the amount is much less than the amount observed in sample 227.
The mixture of big and small rods is clearly visible in TEM images such as Fig. 2.11.
The size of the big rod in Fig. 2.11 is 0.6 µm times 4.3 µm. The small rod located below
the big one in Fig. 2.11 has the dimension 50 nm times 400 nm, corresponding to a
volume approximately 1500 times less than the big rod. Lattice fringe imaging of the
suitable particles shows big (>50 nm) regular crystals. A typical example of such a
lattice fringe image is shown in Fig. 2.11. An FFT of the image is shown in Fig. 2.15.
The dominant lattice fringes are separated with distances of 0.61 nm and 0.36 nm,
respectively, in an angle of 70º. The outermost 0.5-1 nm do not show the same lattice
fringes as the inner parts of the crystals.
The supramolecular particles obtained during the up scaling experiments (sample 74)
are more regular shaped for the large precipitation batch sample compared to the small
scale samples. This is observed in both SEM and TEM images. A characteristic TEM
image of sample 74 is shown in Fig. 2.11. Both large (>1 µm) and small (<200 µm)
agglomerates are observed in the images. However, the amount of smaller agglomerates
39
seems to be smaller and the shape better defined compared to sample prepared on a
smaller scale.
HRTEM images of the smaller agglomerates found in sample 74 reveal randomly
oriented 3-5 nm clusters embedded in non-crystalline material as shown in Fig. 2.11.
The distance between the lattice fringes revealed in the clusters are found to 0.34 nm
and 0.37 nm. The clusters seem to be more well-developed in this sample than in the
small scale sample and the non-crystalline material less. Furthermore, larger (20-50 nm)
well-crystalline particles are observed in the sample Fig. 2.11. A Fourier Transform of
the crystal in shown in Fig. 2.15. The main features in the FFT correspond to lattice
fringe distances of 0.53 nm and 0.40 nm in an angle of 40º. The outermost 0.5 nm of the
crystal shown in Fig. 2.11 is observed to be a different structure than the rest of the
crystal.
TPRS
Two samples prepared under similar conditions are compared (Fig. 2.24). In both cases
a 1 mol/l AHM stem solution was used and HNO31 mol/l. The large scale sample was
produced under preparative conditions including a 1 h aging time.
300 350 400 450 500
temperature [°C]
signal m/e = 56
256
257
258
227
223
228
225
232
Figure 2.23: TPRS run first cycle; showing compounds prepared
from AHM starting material; acrolein production
40
Looking at the CO2production the preceding local maximum is slightly shifted. Apart
from that the two traces match exactly until 250 °C are reached. From that point on the
large scale’s trace is steeper and shows a marked plateau at 320 °C. The overall
maximum is higher but reached at the same temperature. After that the activity towards
CO2is declining. From 330 °C onwards the two traces match again. During the second
cycle the activity is in the same range as at the end of the first one. However the activity
of the small scale sample is higher.
In the acrolein case the two traces match until the first maximum is reached. During the
constant temperature period the large scale sample is declining faster. In the second
cycle the range of activity is comparable to the end of the first one and again the small
scale sample is more active than the large scale.
0 60 120 180 240 300 360
400
600
800
m/e 56 [a.u.]
time [min]
m/e 44 [a.u.]
large scale
small scale
T [K]
Figure 2.24: TPRS first and second cycle; showing the activity and
selectivity of the large scale sample and the small scale sample
41
2.4 DISCUSSION
Principally the herein described method allows producing four different families of
materials in a controlled manner. The assignment is based on integrative methods such
as Raman spectroscopy and XRD. TEM shows that every sample contains differently
sized and shaped particles with a high content of amorphous material. This is
particularly important to be considered for any efforts to derive structure activity
correlations. The presence of a non-uniform precipitate is easily understandable bearing
in mind the multitude of parallel reactions that occur during the titration. An earlier
Raman investigation[29] already identified two independent reactions, more are likely. A
closer look at the Raman bands show identical band positions for each member of the
same family but different intensities, again indicating the presence of more than one
species. Another indicator for this is the band gap energies determined from the UV/Vis
experiments. This is most likely one of the reasons of many contradictions in the
literature concerning these materials.
500 750 1000 1250 1500 1750 2000 2250 2500
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
1790
c
1930
1430
III
b
a
2280
970
Kubelka-Munk units
wavelength (nm)
Figure 2.25: Spectroscopic characteristics NIR of orthorhombic
MoO3. 251, 222 and 230
42
MoOx spectra show NIR bands with different intensities, distinguishable LMCT bands
and band gap energies (Eg) at RT. Based on the exact determination of such
spectroscopic characteristics the following LMCT bands (nm) (I) and Eg’s (eV) (II) are
attributed to the above mentioned MoOxfamilies: (I) 312 (NH4+), 319 (K+); (II) 3.50
(NH4+), 3.43 (K+) to supramolecular Mo36; (I) 313 (NH4+), 312 (K+), 327 (Na+); (II)
3.35 (NH4+); 3.30 (K+), 3.27 (Na+) to hexagonal MoO3; (I) 296 (Li+); (II) 3.44 (Li+) to
orthorhombic MoO3and (I) 284 (K+); (II) 3.77 (K+) to trimolybdate MoOx.
From a blue shift of the LMCT band in the series supramolecular/hexagonal
orthorhombic trimolybdate and a decreasing broadening of this band it may be
concluded that the cluster size decreases. All MoOxsamples evolved NIR bands at
1435, 1940, and 2040 nm. They are assignable to an overtone mode of the OH
stretching vibration and a combination mode of the OH stretching and bending
vibration, respectively. Other NIR bands, e.g. those detected in MoOxsamples prepared
from AHM at 1570 and 2150 are caused by ammonia.
1000 800 600 400 200
Supramolecular compound 257
Hexagonal MoO3280 °C
Intermediate 325 °C
Intermediate 340 °C
Orthorhombic MoO3
Intensity
Raman shift [cm-1]
Figure 2.26: Raman investigation of TG post mortem samples at
characteristic signals
43
2.4.1 Effect of counter cation
Despite the inhomogeneity mentioned above, this work has clearly shown the
considerable structure directing effects of counter cation. The influence of the counter
cation is based on its size and other properties such as activity and ionic strength. Whilst
only the big ions K+and NH4+caused spontaneous precipitation, the presence of Li+and
Na+required a heating period, leading to orthorhombic MoO3in case of Li+and to
hexagonal MoO3in case of Na+. The latter structure was initially reported by Krebs[31;83]
and labeled as Mo5O16. Heating of K+ and NH4+containing material yields either the
trimolybdate[43;78] or the hexagonal MoO3whereas ammonia yields only supramolecular
or hexagonal MoO3, as reported.
In order to assess the thermodynamic stability of the four different families it is
assumed that there is the same general trend in solution and in solid material. Wienold
et al.[84] carried out thermal decomposition of ammonium heptamolybdate. They
showed that in kinetically determined steps the system turned into hexagonal MoO3and
subsequently into orthorhombic MoO3at 350oC. The last step was identified as
thermodynamically favoured. If the formation of orthorhombic MoO3is also
thermodynamically favourable in solution, the Na+, K+and NH4+counter cations
prevent such formation by precipitation of other intermediate phases. Formation of such
other phases would also be kinetically controlled, as also indicated by the large
temperature dependence. As Li+is not able to force the precipitation of the hexagonal
phase because of its lower activity the orthorhombic phase is formed.
In order to address the question about the role of the counter cations during the
precipitation process Lehn´s definition of supramolecular chemistry as `chemistry
beyond the molecule´ intermolecular forces such as van der Waals forces, London
dispersion forces and hydrogen bonding becomes important. Further polarisability
introduced in the HSAB concept needs to be applied. It is likely that the comparatively
soft NH4+could serve as an endo-template [85;86]. Fragments of the Octamolybdate will
group around the ammonia and interlink. As soon as the particle size is big enough and
a certain amount of species is formed, the solubility product is exceeded and
precipitation of the hexagonal MoO3sets in[86]. A retro synthetic approach corroborates
to this idea. Taking the ammonia inside the channel as a centre, only a few Mo-O-Mo
bridges need to be broken to yield these nanocluster building blocks. A comparable
mechanism should be possible with K+, taking into account that this ion is very similar
44
in size and ionic strength to ammonia. However K+seems to be the limit for this
mechanism because under some experimental conditions (high Mo concentration, high
acid concentration, 70 °C and spontaneous precipitation) the layered trimolybdate
structure is obtained. This trend is further followed as the smaller and harder ions Na+
and Li+do not cause spontaneous precipitation, but rather hexagonal/ orthorhombic
MoO3is formed after heating, as reported above. The smaller the ion the more difficult
it becomes to build up a stable spherical wrapping, consequently layered structures are
obtained from smaller ions and channel structures from bigger ones.
2.4.2 Effect of temperature and proton to molybdenum ratio
The obtained results show that temperature and the proton to molybdenum ratio are also
important control variables. 50 °C is the most important boundary as the temperature
dependent investigations show. Therefore at this temperature most likely the reaction
mechanism is changing. High molybdenum concentrations and low temperature (30 °C)
lead to a spontaneous precipitation of a supramolecular compound, which is very
similar to the Mo36O112 reported by Krebs. Low concentration and high temperature
leads to the formation of a hexagonal MoO3[79-81] the same structure is reported by
Krebs[83;87] and labelled as Mo5O16, orthorhombic MoO3or trimolybdate depending on
the counter ion.
Up to pH = 3.5 the curves are identical and therefore the reaction mechanisms are alike.
This seems to be plausible because at this stage most likely protonation of the
heptamolybdate is taking place. This fits nicely to NMR data[67]. At around pH = 3
where the reorganization to the Mo36O112 sets in, the 30 °C curve is different to the other
curves. Octamolybdate is reported to be the dominant species [63;65;88] [44;68;89] in this pH
region. As transformation from octamolybdate into Mo36O112 requires many bonds to be
broken and reassembled by polycondensation reactions the slow down in the pH drop
seems plausible. At around pH = 2 another change in mechanism sets in. The 50 °C
sample is acidified easily without yielding a precipitate. The 70 °C moves parallel to the
30 °C sample however at lower pH. It is likely that 50 °C is not enough to overcome the
activation energy barrier to precipitate the hexagonal phase.
Using K2MoO4as starting compound the coordination of the Mo is tetrahedral and the
experiment starts at higher pH. As the curves show the starting pH is a function of the
molybdenum concentration. As the first pH drop is very rapid the tetrahedral species is
45
not to be protonated. The first buffering region probably coincides with the formation of
a heptamolybdate species. Again this event depends on molybdenum concentration. The
following reaction pathway will be similar to the one described with ammonia as
counter ion. It is noteworthy that trimolybdate is only obtained with high molybdenum
and acid concentrations at higher temperatures.
According to Tytko the Mo36O112 is the major compound in solution at low pH and
precipitates as soon as the solubility product is reached. It is remarkable that the
connectivity has changed compared to the starting material[42]. In the AHM precursor
only corner sharing octahedra are observed. In the Mo36O112 edge sharing connectivity
prevails and the pentagonal bipyramid as structural motif, which appears in catalytic
active material like Mo5O14, is formed[20;90;91]. Corner sharing also turns up in the
hexagonal MoO3.
2.4.3 Effect of water
Krebs points out that Mo36O112 is only stable in the mother liquor. Water is responsible
for the hydrogen bonding which is probably the dominating coherent force in this
molecule[32]. This phenomenon has been studied in detail in comparable
compounds[54;86]. As soon as water is removed the compound will decompose. This
explains the variation in XRD. Nevertheless the close similarity of the XRD to the
reference and the exact agreement of the Raman band positions corroborates to the idea
that the main structural motives prevail.
Investigating the reaction in the solid phase again the differences due to counter ions are
striking. Using ammonia as a counter ion, the final transformation is taking place at 50
°C higher temperatures than in the potassium case. Therefore solid-state
thermodynamics are clearly influenced by the nature of the counter ion.
Moreover the phase transformation using ammonia as counter ion will be very different
because of the combustion of the ion. The large amount of water released at the first
exothermic event is probably to some extend related to this combustion. The same holds
true for the exothermic DSC signal. A good deal of heat is produced because of the
combustion.
As NOxis released at 3 different temperatures showing three clearly distinguishable
DSC and MS peaks (water and NOxdata not shown) it is probably situated in two
46
different places. At the first event the ammonia serving as linker between
supramolecular building blocks will be removed. After that ammonia is situated in the
channels of the hexagonal MoO3 and will only be removed at higher temperatures. The
last NO signal is detected with the K+containing compounds as well. At this
temperature the NO3-that was brought into the solution by the nitric acid is removed.
As a mass loss is detected in this region as well it is very likely that the phase
transformation form the hexagonal phase to the orthorhombic MoO3is related to a
removal of oxygen from the lattice or a further condensation. This idea is further
supported by traces of water that are also detected.
Comparing the supramolecular compound obtained with K+as counter ion with the
ammonia containing compound it becomes clear that the first transformation is also a
condensation step. The water being released in this step by the K+containing compound
cannot originate from a combustion process but from polycondensation. In this respect
solid phase and aqueous chemistry run parallel because some activation barrier needs to
be overcome to induce this process.
The similarities between the supramolecular phase and the hexagonal MoO3become
very clear when the sample prepared from a 2 mol/l K2MoO4solution is compared to
the sample prepared from a 0.28 mol/l K2MoO4solution. Mass loss and DSC trace are
very similar. Therefore only the condensation process will vary slightly.
The water contents can be estimated from the NIR bands and from the TG-MS results
for the Li+, Na+and K+containing precursors. Compounds, which show a big mass loss,
always have marked water bands. This correlation is not possible for the ammonia
containing samples because the water and ammonia vibrations are combined.
The XRD pattern of the orthorhombic MoO3obtained from Li2MoO4shows broad
signals and a comparatively high background indicating low crystallinity. One reason
for this is that the degree of polymerisation due to a condensation in water at the chosen
conditions is too low to form big crystalline particles or the solubility product is reached
too early. DSC and TG-MS clearly shows an ongoing polycondensation.
The sample prepared at 50 °C shows the biggest overall mass loss concerning water.
However the first step where only associated water is removed is biggest as well. One
reason for this is probably a high degree of protonation at the outer shell of the
polymeric precursor in solution being a good docking place for hydrogen bonding to
water molecules. Assuming that the second step is the water from the polycondensation
47
process the 50 °C sample looses 2.9 % in this step. For the 30 °C sample the same
amount is calculated. Only for the 70 °C sample 3.2 % are calculated. Therefore in this
sample more OH groups suitable for further condensation were created. Again a change
in the reaction mechanism can be established.
Taking all these results together it can be concluded that at higher temperatures in
solution the sites for protonation are changing and protonation is much more efficient.
This idea is corroborated by the above described changes in the pH curves of the
ammonia containing samples as a function of temperature.
2.4.4 Structure determination by XRD and TEM
Phase characterisation of materials by XRD is widely used. However, such integral
techniques are insufficient to reveal the detailed local phase-and structure
determination, which might be relevant for the purposes of the studied material,
especially in catalysis. Lattice fringe images of sample 130, such as Fig. 2.11 show
well-developed crystals, from which the relevant lattice plane distances can be
measured. Several high-resolution images revealing lattice fringes acquired from sample
130 are analysed, and the measured lattice plane distances are compared with the
corresponding XRD pattern. This is shown in Fig. 2.15 C. The XRD shows narrow
distinct peaks coinciding with diffraction peaks calculated from the model structure of
hexagonal molybdenum trioxide. The signal to noise ratio is very high and no additional
features are observed between the well-defined peaks. The observed lattice plane
distances measured in the high-resolution images are marked under the XRD pattern in
Fig. 2.15 C. HRTEM images cannot and should not be used quantitatively, so the
comparison is done qualitatively. Several lattice plane distances observed in the
HRTEM images are not revealed in the XRD, indicating that the HRTEM images are
acquired of crystalline structures which are not resolved by the integral technique of
XRD. In Fig. 2.11 both a very big particle and a small particle are observed. The
volume of the bigger particle is approximately three orders of magnitude larger than the
volume of the smaller one. The XRD technique is volume sensitive and hence the
structure of the largest volume is favoured in the XRD. The diffraction pattern of the
structure of the smaller crystals is suppressed by the diffraction pattern of the large
volume of big well-defined crystals in the sample. The local structure of the smaller
particles in the sample might be highly relevant for the resulting structure of the active
48
catalysts, and should not be neglected. In order to validate the comparison between X-
ray diffraction and diffraction of electrons in the sample, diffraction simulations are
performed using X-rays and electrons. The simulated powder diffraction spectra of the
hexagonal molybdenum trioxide model reveal exactly the same features for electron
diffraction and X-ray diffraction, although slight changes in the intensities are observed.
The morphology of the biggest particles in the samples shows clear differences in the
two families (hexagonal and supramolecular). The regular hexagonal rods with well-
Figure 2.27: A FFT of 74, B FFT of 130, Comparison electron diffraction
versus X-ray diffraction
49
defined facets observed in Fig. 2.11 are a further indication of the single crystalline
hexagonal atomic structure of biggest particles in sample 227. The more irregular
morphology of the largest particles observed in Fig. 2.11 might indicate a more
complicated atomic structure or the fact that the biggest particles in the small scale
sample are agglomerates of smaller crystals.
In the following the crystallographic structure of the studied samples will be discussed
on the basis of the lattice fringe images acquired from area of the sample, which are
electron transparent, i.e. thin agglomerates.
The samples precipitated in the small-scale set-up consist of randomly oriented clusters
in the 3-5 nm scale, as shown in Fig. 2.11. The fact that lattice fringes are revealed
indicates that the clusters are crystalline. The characteristic separations between the
revealed lattice fringes are measured to 0.34 nm and 0.37 nm respectively. These
separations coincide with the distance found between Mo atoms in edge-sharing and
corner-sharing MoO6octahedra. The crystalline clusters in both sample 227 and the
small scale sample are embedded in non-crystalline material, although this is more
pronounced in the small scale. This non-crystalline material might very well contain the
nitrogen, which is released during the activation of the catalyst. The clusters will then
serve as building blocks for the final active structure of the catalyst.
Up scaling the precipitation process results in larger, more ordered crystalline
structures. The structure shown in Fig. 2.11 is characteristic to sample 130 and shows a
single crystal of more than 50 nm in size. Although the lattice fringe images acquired
from sample 130 are insufficient to determine the structure of the crystalline material, it
can be concluded that the resolved lattice fringes do not match the unit cell parameters
of hexagonal or orthorhombic molybdenum trioxide. The outermost 0.5-1 nm of the
single crystal do not reveal the well-defined lattice fringes found in the inner part of the
crystal. This might be due to ill-defined lattice plane termination or to reconstruction at
the surface due to oxygen vacancies caused by oxygen vaporisation into vacuum. The
more ordered and bigger crystals are also observed in Fig. 2.11 A and B acquired from
sample 74. The amount of non-crystalline material seems to be reduced in samples
precipitated in bigger scale. Both sample 130 and sample 74 undergo longer aging times
due to the up scaled precipitation, hence the system has more time to attempt the
thermodynamically equilibrium state, resulting in larger more ordered structures.
50
2.4.5 Raman and DRIFTS
As Me-O distances and connectivity play an important role in the discussion of catalytic
activity it is worthwhile having a closer look on Raman and DRIFTS. Starting with
hexagonal MoO3(Fig. 2.28) with the point symmetry C6h a band is reported at 976 cm-1.
This band is assigned to terminal Mo-O stretching (OMo); the bands at 904 cm-1 and at
889 cm-1 are assigned to bridging Mo-O-Mo (OMo2) of corner sharing octahedra; the
band at 691 cm-1 is assigned to a 3 oxygen stretching frequency (OMo3) of edge
sharing octahedra; the bands at 494 cm-1, 400 cm-1 and 317 cm-1 are assigned to the
respective bending vibration (OMo3); the band at 256 is assigned to the bending
vibration (OMo). Finally the band at 224 cm-1 is assigned to (OMo2).
Figure 2.28: Hexagonal MoO3, Assignment of Raman and IR modes
In orthorhombic MoO3with the point group D2h the band assignment is as follows (Fig.
2.29). The band at 997 cm-1 is assigned to (OMo); the band at 824 cm-1 is assigned to
(OMo2); the bands at 668 cm-1, 476 cm-1 and 379 cm-1 are assigned to (OMo3); the
band at 339 cm-1 is assigned to (OMo3); the band at 291 cm-1 is assigned to (OMo);
the band at 247 is assigned to (OMo2)[92-97].
51
IR bands were found at 743 cm-1 (not reported in the literature), at 840 cm-1 assigned to
(OMo2) and at 1000 cm-1 which is assigned to (OMo)[98] [99;100]. Remarkably the
Raman and IR-bands in orthorhombic MoO3are not equal. This could be explained by
employing group theory. The space group is D2h and in this case IR and Raman active
modes do not coincide. Consequently Raman and IR excite totally different modes.
Figure 2.29: Orthorhombic MoO3, Assignment of Raman and IR modes
2.4.6 TPRS
All tested molybdenum oxide samples show catalytic activity towards propene selective
oxidation into acrolein, the degree of activity is strongly structure dependent. The main
information of the TPRS can be extracted out of the traces emerging from the first
heating. In this sequence, the precursor samples undergo decomposition and structural
changes resulting from the thermal and atmospheric conditions they are exposed to.
The comparison of the samples under investigation elucidates the different activation
(and deactivation) behaviour depending on structure and composition of the precursor.
Most striking is the dependency on the cation linked to the starting molybdate material;
the ammonium-derived samples show the highest activity.
As described above ammonia is the only counter ion which decomposes during heating.
One reason for the comparatively low activity of the materials still containing counter
ions could be the different phase changes as a function of temperature as shown in the
TG-DSC experiment. Another reason might be a change in the defect structure of the
material. During the ammonia combustion NOxis detected. This can lead to a partial
auto reduction of the material and hence a different defect structure which is
catalytically inactive. This is corroborated by Mc Carron Ill´s findings which show that
52
the position of ammonia in the channels of the hexagonal MoO3is shifted compared to
Na+and K+. Consequently the different interactions leading to different channel sizes
and different defect structures[82]. The different band gap energies determined for
materials produced from different counter ions also show that different defects prevail.
However, there is a general feature non-depending on the chemical composition
stressing out the principal role of structural properties for heterogeneous catalysis. The
samples classified supramolecular give rise to a first oxidation maximum centred at 230
°C arising from a very active and non selective oxygen species released from these
samples while their structure is rearranged. The role of transient and metastable states in
catalysis is further spotted by the observation, that all samples perform quite the same
during the second heating.
It is further noteworthy that the acrolein trace of the supramolecular material (223) and
the hexagonal MoO3(232) run almost parallel at temperatures higher than 400 °C.
Consequently the reaction mechanism after a different activation behaviour is very
similar. This is in line with the TG-DSC experiment the two traces are clearly different
between 250 °C and 370 °C. Moreover the catalytic process is strongly dependent on
the final phase transition because the catalytic activity is rising decisively in this region.
Compared to the above described material the hexagonal MoO3prepared at 50 °C
shows a much higher onset temperature. As shown above the pH curve is very different
as well. This proves that the details of the preparation process do matter. Differently
protonated precursors will lead to a different nanostructure in the material and this
results in a different catalytic behaviour.
The careful investigation of the precursors revealed that the integrative methods such as
XRD, vibrational spectroscopy and thermal methods do not detect the smaller
unidentified particles determined by TEM. The comparison of the large scale
preparation to the smaller scale clearly showed different catalytic behaviour. The TEM
investigation showed that the small scale preparation produced a larger extent of small
particles. TPRS showed that this material deactivates more slowly and shows higher
activity in the second cycle. Therefore these small particles could be the reason for
catalytic activity. This could only be revealed in this comprehensive investigation
taking the effect of up scaling into account.
Another reason –also detected by TEM – could be that catalytic activity is connected to
a nano structuring of the material. Consequently a structural rearrangement or even
53
worse crystallisation in the thermodynamically most stable form or sintering needs to be
avoided. This would be in line with earlier investigations showing that highly crystalline
material usually possesses low activity[8].
It is important to note that in this investigation the catalytic activity of a system
consisting only of molybdenum and oxygen can be altered over a wide scale only by
controlling the defect -and nano -structure of the material. Haber[101] postulates that the
introduction of Bi3+ in MoO3as a promoter generates a new active site responsible for
hydrogen abstraction from hydrocarbons. Taking the results of this investigation into
account such a change could also be achieved in a system without additives. The same
holds true for Grasselli´s investigation[17-19].
Considering Haber´s data and the herein presented results it seems highly likely that
catalyst optimization could be possible in a chemical simple system by careful control
of the structure development during preparation and by controlling thermodynamic
parameters such as reaction temperature and partial pressure of reactants and oxidizing
reagents during activation and catalysis.
2.5 CONCLUSION
A number of nanostructured molybdenum oxides with high structural complexity has
been successfully prepared. Characterization of the compounds has shown that material
with the same structural features as the industrial material (e.g. pentagonal bipyramides
or hexagonal channels) can be prepared without the need for any additive or promoter.
All of the obtained material is active in selective oxidation of propene, same
supramolecular samples even show comparable activities to the industrial material. No
material is active as precipitated, but calcination transforms them into metastable active
phase at about 300 °C, as seen from the TPRS experiments. Above 450 °C phase
transformation into orthorhombic MoO3significantly reduces the catalytic activity.
Whilst it is often reported that additives increase the catalytic activity, it is more likely
that they merely stabilise the metastable Mo-O phase.
Many attempts have been carried out in the literature in order to develop structure
activity correlations. This work has demonstrated that such correlations by integrative
methods as XRD or Raman have to be treated with extreme caution, as the existence of
many differently shaped and structured particles could only be observed with HRTEM.
54
The TPRS experiments show clearly that even a chemically simple system can be tuned
by controlling its thermodynamic parameters. This opens two new pathways. First,
catalytic reactions can be studied and understood on comparatively simple model
catalysts. Second, catalysts can be optimised by varying and controlling thermodynamic
parameters and not only by additives whose role often is obscure.
2.6 TABLES
Table 2.1: Precipitation conditions, (* S= supramolecular, H= hexagonal, T=
trimolybdate, O= orthorhombic)
Sample
Number
Conditions
Family* counter-
ion
Mo-
concentration
[mol/l]
HNO3-
concentration
[mol/l]
Temperature
/ °C
0249 S K +0.28 2 30
0219 H 0.28 2 50
0229 H 0.28 2 70
0250 S 2 2 30
0245 S 2 2 50
0244 T 2 2 70
0253 S 2 5 30
0246 S 2 5 50
0243 T 2 5 70
0286 S 0.28 5 30
0247 H 0.28 5 50
0233 H 0.28 5 70
0251 O Li +0.28 2 30
0222 O 0.28 2 50
0230 O 0.28 2 70
0252 H Na +2 2 30
0226 H 2 2 50
0231 H 2 2 70
0255 H 2 5 30
0256 S NH4+0.7 1 30
0227 H 0.7 1 50
0232 H 0.7 1 70
55
0257 S 1 1 30
0228 S 1 1 50
0258 S 1 5 30
0248 H 1 5 50
0225 S 0,7 2 50
0223 S 1 2 50
Table 2.2: Scale up
Sample
Number
Conditions
Family* counter-
ion
Mo-concentration
[mol/l]
HNO3-concentration
[mol/l]
Temperature /
°C
0072 S NH4+1 1 35
0073 S NH4+1 1 35
0074 S NH4+0.5 1 35
0075 S NH4+0.5 1 35
0130 H NH4+0.2 135/80
0131 H NH4+0.2 135/80
0132 H NH4+0.1 135/80
0133 H NH4+0.1 135/80
0134 S NH4+2 1 35
0135 S NH4+2 1 35
0136 S NH4+1 1 35
0137 S NH4+2 1 35
0138 S NH4+2 1 35
0140 S NH4+1 1 35
Table 2.3: BET surface areas
Sample Nr. cation phase identification from Raman Surface Area (m2/g)
0219 K+Hexagonal MoO31
0229 K+Hexagonal MoO30.8
0250 K+Supermolecular compound 0.8
0245 K+Supermolecular compound 1.2
0244 K+Trimolybdate 3.7
56
0253 K+Supermolecular compound 0.7
0246 K+Supermolecular compound 1.5
0243 K+Trimolybdate 3.5
0254 (286) K+Supermolecular compound 0.8
0247 K+Hexagonal MoO31.1
0233 K+Hexagonal MoO30.8
0230 Li+orthorhombic MoO316.2
0226 Na+Hexagonal MoO33.6
0255 Na+Hexagonal MoO31.1
0231 Na+Hexagonal MoO30.5
0228 NH4+Supramolecular Compound 1.7
0131 NH4+Hexagonal MoO31.0
0137 NH4+Supramolecular Compound 1.6
0139 NH4+Supramolecular Compound 1.5
0140 NH4+Supramolecular Compound 1.4
0142 NH4+Supramolecular Compound 0.6
Table 2.4: UV/Vis data
molybdenum oxides
counter ion structure band position/nm Eg/eV
295 (a) 3.48 (a)
304 (b) 3.37 (b)Li+orthorhombic
289 (c) 3.48 (c)
K+trimolybdate 284 (d) 3.77 (d)
332 (e) 3.27 (e)
328 (f) 3.27 (f)Na+hexagonal
320 (g) 3.27 (g)
319 (h) 3.30 (h)
K+hexagonal 305 (i) 3.30 (i)
316 (k) 3.35 (k)
57
NH4+hexagonal 316 (k) 3.35 (k)
NH4+hexagonal 310 (l) 3.36 (l)K+supramolecular 319 (m) 3.30 (m)
284sh (n) 329b(n) 3.43 (n)
K+
308 (o) 3.43 (o)
NH4+supramolecular 316 (p) 3.46 (p)
NH4+supramolecular 277sh (r) 308b(r) 3.55 (r)
Table 2.5: NIR data
molybdenum
oxides
counter
ion structure
Li+orthorombic 970vw 1430 1790 1930 2280
K+ trimolybdate 1435 1790/
1810 1930 2000 2095 2250
Na+hexagonal 1430 1820 1955 2280
K+hexagonal 1435 1820 1935 2090 2235
NH4+hexagonal 1440 1570 1810 1945 2040 2150
K+supramolecular 1435 1810 1935 2090 2235
K+supramolecular 1435 1790/
1810 1930 2000 2095 2250
NH4+supramolecular 1440 1570 1945 2040 2150
NH4+supramolecular 1440 1570 1945 2040 2150
Table 2.6: Major Raman band positions and intensities
Sample
Nr
Fa
mil
y
Raman bands wavenumber cm-1
249 SPos. 963 882 373 229
Int. 7674 5573 2247 3529
219 H Pos. 978 904 692 496 401 255 224
Int. 3478 4875 2112 961 1333 4464 767
229 HPos. 978 904 692 493 401 254 224
Int. 3355 4619 1913 719 1103 4237 881
250 SPos. 961 898 372 240
Int. 2871 2550 960 1519
245 SPos. 961 894 371 237
Int. 3945 3867 1301 1950
58
244 TPos. 1052 949 938 911 612 372 217
Int. 808 3579 2673 1311 843 1519 2473
253 SPos. 960 899 372 238
Int. 2460 1587 594 970
246 SPos. 961 897 372 238
Int. 4073 3385 1243 1829
243 TPos. 1052 949 930 911 603 378 219
Int. 2498 10972 3181 2397 1448 2113 3541
254 SPos. 962 898 371 241
Int. 3271 2748 1462 1805
247 HPos. 980 904 693 401 254 224
Int. 4326 5929 2814 1429 6062 3085
233 HPos. 977 904 692 494 400 254 224
Int. 3923 5324 2304 982 1388 4830 2514
251 OPos. 997 825 668 381 339 292 247
Int. 8702 20442 3375 2189 3279 8264 3773
222 OPos. 997 824 668 382 340 291 247
Int. 9647 23411 3651 2489 3630 9262 4065
230 OPos. 997 825 668 382 339 291 249
Int. 3651 9694 1545 998 1587 3759 1835
252 OPos. 977 902 692 492 398 256 227
Int. 3772 6110 2365 902 1322 4667 3022
226 HPos. 979 902 692 493 398 256 228
Int. 5033 7759 3205 1726 2150 5753 3970
231 HPos. 977 902 692 495 399 255 226
Int. 1543 2382 914 370 534 1832 1178
255 HPos. 976 901 692 493 398 255 229
Int. 3353 5739 2027 823 1220 4118 2838
256 SPos. 959 893 374 233
Int. 6085 3866 1535 2473
227 HPos. 976 904 691 494 400 256 224
Int. 4803 6823 2744 997 1593 6029 1597
232 HPos.976 904 692 494 400 254 224
Int. 5776 7834 3208 1270 1889 6716 1186
257 SPos. 960 884 372 230
Int. 931 661 265 417
228 SPos. 959 893 375 237
Int. 3304 2196 845 1301
258 SPos. 957 895 374 237
Int. 2030 1467 530 865
248 HPos. 976 904 692 494 400 224
59
Int. 2126 2974 1285 456 703 1760
225 SPos. 960 893 373 237
Int. 4802 3098 1175 1865
Table 2.7: XRD data
Cation Sample Name Phase Peak Position Intensity
K249 Supramolecular 6.88 3498
10.25 1639
11.57 1684
16.06 768
22.16 765
25.16 775
27.11 940
219 Hexagonal 9.67 10279
16.77 1549
19.4 4439
25.78 15626
29.31 8269
35.38 4794
43.08 1501
45.39 2385
48.84 3014
56 3055
57.87 1447
61.67 873
68.87 1465
229 Hexagonal 9.66 12215
16.77 1950
19.4 4908
25.79 16943
29.32 9341
30.95 1011
35.39 5188
43.07 1797
45.4 2532
48.84 4128
56.03 3195
57.87 1759
60
67.13 989
250 Supramolecular 3.87 907
7.24 1379
8.59 2237
11.24 2956
13.03 1068
14.2 870
17.64 596
19.51 613
23.18 694
25.16 596
27.3 789
29.43 661
34.54 391
37.47 346
46.88 316
56.02 261
245 Supramolecular 5.56 1313
8.63 1926
11.04 1248
13.66 1004
16.93 930
20.57 766
22.57 911
25.01 930
27.29 1037
28.72 758
32.53 519
37.1 364
42.84 299
44.81 376
47.94 313
49.32 431
54.71 306
57.12 266
59.31 272
63.28 255
65.46 239
61
244 Trimolybdate 9.77 6569
11.77 1717
15.17 1227
16.56 903
18.72 1077
23.84 1580
25.28 1450
27.61 2117
29.56 1553
31.39 1406
33.86 715
39.78 867
41.16 847
44.12 470
47.7 953
48.8 683
57 515
253 Supramolecular 5.94 1613
8.44 2646
11.19 2197
13.29 878
18.08 685
19.92 661
23.18 738
23.55 877
26.21 837
29.45 746
30.75 654
34.07 483
36.82 351
41.87 283
46.74 342
49.71 355
57.72 262
246 Supramolecular 5.54 1134
6.89 1481
8.61 1545
12.23 1105
12.86 873
62
16.96 897
20.48 702
22.61 746
26.75 799
27.3 990
28.78 684
33.28 471
33.81 406
38.33 341
38.82 319
41.17 296
44.83 332
48.14 301
49.26 329
51.59 302
54.65 259
60.34 228
61.19 208
64.34 209
66.43 190
69.72 172
243 Trimolybdate 9.77 13269
15.17 2213
18.71 1934
20.77 1115
23.83 2170
25.25 2469
27.58 4512
29.54 2811
31.37 3081
35.13 888
39.74 1851
47.51 1894
48.61 1335
56.94 1074
63.61 759
254 Supramolecular 6.51 1677
8.6 2507
11.23 5825
63
13.24 1294
17.65 855
19.51 864
23.19 1037
28.19 930
29.49 834
34.54 520
37.48 432
45.65 397
247 Hexagonal 9.65 7016
16.77 1292
19.38 2755
25.81 9273
29.31 6272
35.38 2933
43.02 925
45.32 1217
48.88 3236
56.01 2149
58.77 628
68.84 782
79.22 545
82.5 531
233 Hexagonal 9.67 13527
16.78 2122
19.4 5611
25.78 19242
29.32 10308
30.94 1153
35.39 6077
43.06 2132
45.39 3059
48.85 3853
56.01 3646
57.87 2085
61.64 1113
67.11 1183
Li 251 Orthorombic 12.73 712
64
16.92 408
19.31 399
23.29 1519
27.32 2310
33.65 847
39.11 443
45.67 397
49.25 874
52.37 284
55.11 452
57.78 245
64.71 263
69.32 191
78.83 200
222 Orthorombic 12.79 808
23.31 1679
27.34 2285
33.63 905
39.14 421
45.55 416
49.2 886
52.33 318
55.11 466
230 Orthorhombic 12.79 905
23.26 1647
27.34 2379
33.58 884
39.08 478
45.52 427
49.18 954
52.19 305
55.08 505
64.63 297
Na 252 Hexagonal 9.61 9803
16.69 1500
19.29 3476
25.61 8615
29.31 6632
65
35.31 3330
42.83 1230
45.13 1685
48.96 2171
56.04 1783
57.6 1217
68.72 787
226 Hexagonal 3.37 1686
9.61 12939
16.67 1954
19.27 5226
25.58 11394
29.27 9605
30.87 1003
33.69 1397
35.27 4824
42.77 1591
45.09 2754
48.89 3587
53.03 795
55.96 3011
57.54 1791
66.95 844
68.69 1562
231 Hexagonal 9.61 8617
16.68 1319
19.27 3172
25.59 7620
29.28 5842
35.28 2989
42.78 1157
45.11 1717
48.91 2125
53.03 582
55.99 2060
57.56 1462
61.21 544
68.7 998
86.55 528
66
255 Hexagonal 9.61 7672
16.7 1155
19.28 3086
25.59 6950
29.28 5160
35.28 2740
42.79 942
45.09 1371
48.91 1808
53.02 473
55.98 1426
57.57 920
68.69 623
NH4+256 Supramolecular 4.54 1727
7.02 15128
11.54 3377
227 Hexagonal 9.66 9954
16.78 1692
19.39 3057
25.78 11123
29.31 6866
35.38 3597
43.05 1222
45.37 1725
48.85 2758
56 2401
57.86 1195
61.67 627
68.81 1122
232 Hexagonal 25.78 12037
29.32 7525
35.38 4002
43.03 1500
45.34 1909
48.87 2772
56.03 2518
57.87 1306
67
61.59 730
68.89 1109
257 Supramolecular 4.35 1678
6.94 13840
11.27 2690
26.38 1154
228 Supramolecular 9.66 9954
16.78 1692
19.39 3057
25.78 11123
29.31 6866
35.38 3597
43.05 1222
45.37 1725
48.85 2758
56 2401
57.86 1195
61.67 627
68.81 1122
258 Supramolecular 6.94 4962
9.33 1057
11.54 1785
26.44 743
248 Hexagonal 9.66 9367
16.76 1713
19.38 2712
25.8 9555
29.31 6768
35.37 3708
43.01 1364
45.33 1783
48.87 3856
53.03 607
56.02 2396
57.87 912
61.54 635
68.87 990
68
82.5 545
225 Supramolecular 4.35 2174
6.94 13852
10.81 3894
26.09 1062
69
Reference List
[1.] Kurtz, A. N., Cunningdam, R. W., and Naumann, A. W. Union Carbide Co. [US 4111983].
1978.
Ref Type: Patent
[2.] Novak, V., Sokol, L., and Jelinek, J. CS1207807B. 1981.
Ref Type: Patent
[3.] Bertolini, N. and Ferlazzo, S. Euteco Impianti S.p.A. [US 4289654]. 1981.
Ref Type: Patent
[4.] Kawajiri, T., Uchida, S., and Hironaka, H. Nippon Shokubai Kagaku. [EP 427 508 A1]. 1991.
Ref Type: Patent
[5.] Tenten, A., Martin, F.-G., Hibst, H., Marosi, L., and Kohl, V. BASF AG. [EP 668104 B1]. 1995.
Ref Type: Patent
[6.] Hibst, H. and Unverricht, S. BASF AG. [DE 19815281 A 1]. 1981.
Ref Type: Patent
[7.] G. Mestl, Ch. Linsmeier, R. Gottschall, M. Dieterle, J. Find, D. Herein, J. Jäger, Y. Uchida, R.
Schlögl, Journal of Molecular Catalysis 2000, 162 455-484.
[8.] O. Ovsitser, Y. Uchida, G. Mestl, G. Weinberg, A. Blume, J. Jager, M. Dieterle, H. Hibst, R.
Schlögl, J.Mol.Catal.A: Chemical 2002, 185 291-303.
[9.] M. Dieterle, G. Mestl, J. Jäger, Y. Uchida, R. Schlögl, J.Mol.Catal.A: Chemical 2001, 74 169.
[10.] Hatano, M. and Kayou, K. [EP 318295]. 1988.
Ref Type: Patent
[11.] T. Ushikubo, K. Oshima, A. Kayou, M. Hatano, Study of Surface Science and Catalylis, (Eds.:
C. Li, Q. Xin) Elsevier, Amsterdam 1997, p. 473.
[12.] H. Watanabe, Y. Koyasu, Applied Catalysis A-General 2000, 194 479-485.
[13.] M. M. Lin, Applied Catalysis A-General 2003, 250 305-318.
[14.] M. M. Lin, Applied Catalysis A-General 2003, 250 305-318.
[15.] M. M. Lin, Applied Catalysis A-General 2003, 250 287-303.
[16.] M. Lundberg, M. Sundberg, Ultramicroscopy 2003, 52 429-435.
70
[17.] R. K. Grasselli, Catal.Today 1999, 49 141-153.
[18.] R. K. Grasselli, Topics in Catalysis 2001, 15 93.
[19.] R. K. Grasselli, Topics in Catalysis 2002, 21 79.
[20.] S. Knobl, G. A. Zenkovets, G. N. Kryukova, O. Ovitser, M. Dieterle, R. Schlögl, G. Mestl,
J.Catal. 2003, 215 177-187.
[21.] L. Kihlborg, Arkiv För Kemi Mineralogi och Geologi 1963, 21 427-437.
[22.] L. E. Briand, W. E. Farneth, I. E. Wachs, Catalysis Today 2000, 62 219-229.
[23.] K. Chen, S. Xie, E. Iglesia, A. T. Bell, Journal of Catalysis 2000, 189 421-430.
[24.] K. Chen, S. Xie, A. T. Bell, E. Iglesia, Journal of Catalysis 2000, 244-252.
[25.] K. Chen, A. T. Bell, E. Inlesia, J.Phys.Chem.B 2000, 104 1292-1299.
[26.] K. Chen, S. Xie, E. Iglesia, A. T. Bell, Journal of Catalysis 2000, 189 421-430.
[27.] K. Chen, S. Xie, A. T. Bell, E. Iglesia, Journal of Catalysis 2001, 198 232-242.
[28.] J. Haber, V. Turek, Journal of Catalysis 2000, 190 320-326.
[29.] K.-H. Tytko, B. Schönfeld, B. Buss, O. Glemser, Angewandte Chemie 1973, 85 305-307.
[30.] B. Krebs, I. Paulat-Böschen, Acta Cryst. 1976, B32 1697.
[31.] B. Krebs, I. Paulat-Böschen, Acta Cryst. 1982, B 38 1710.
[32.] I. Paulat-Böschen, J.C.S.Chem.Comm. 1979, 780.
[33.] H. Peters, L. Till, K. H. Radeke, Zeitschrift fur Anorganische und Allgemeine Chemie 1969, 365
14-21.
[34.] L. Cronin, P. Kögerler, A. Müller, Journal of Solid State Chemistry -JSSCB 2000, 152 57.
[35.] T. Yamase, P. V. Prokop, Angew.Chem.Int.Ed. 2002, 114 484.
[36.] J. J. Cruywagen, J. B. B. Heyns, Inorganic Chemistry 1987, 26 2569.
[37.] J. J. Cruywagen, Advances in Inorganic Chemistry 2000, 49 127.
[38.] J. J. Cruywagen, A. G. Draaijer, J. B. B. Heyns, E. A. Rohwer, Inorganica Chimica Acta 2002,
331 322.
[39.] I. Böschen, B. Buss, B. Krebs, O. Glemser, Angew.Chem. 1973, 85 409.
71
[40.] O. Glemser, W. Holznagel, A. Iftikhar, Z.Naturforsch. 1965, 20b 192.
[41.] O. Glemser, G. Wagner, B. Krebs, Angew.Chem. 1970, 82 639.
[42.] K.-H. Tytko, B. Schönfeld, B. Buss, O. Glemser, Angew.Chem.Int.Ed. 1973, 12 330.
[43.] K.-H. Tytko, B. Schönfeld, Zeitschrift fur Naturforschung 1975, 30b 471-484.
[44.] K.-H. Tytko, O. Glemser, Adv.Inorg.Chem.Radiochem 1976, 19 239.
[45.] K.-H. Tytko, G. Baethe, E. R. Hirschfeld, K. Mehmke, D. Stellhorn, Z.Anorg.Allg.Chem. 1983,
503 43-66.
[46.] K.-H. Tytko, G. Baethe, Z.Anorg.Allg.Chem.1987, 555 85-97.
[47.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 2349.
[48.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 18 325-341.
[49.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 2349.
[50.] I. Lindquist, Acta Cryst. 1950, 3159.
[51.] M. T. Pope, Progress in Inorganic Chemistry 1991, 39 181.
[52.] M. T. Pope, A. Müller, Angewandte Chemie 1991, 103 56-70.
[53.] J. Aveston, E. W. Anacker, J. S. Johnson, Inorganic Chemistry 1964, 3735-746.
[54.] H. D. Lutz, R. Nagel, S. A. Mason, A. Müller, H. Bögge, E. Krickemeyer, J.Solid State
Chemistry 2002, 165 199.
[55.] A. Müller, E. Krickemeyer, M. Penk, V. Wittneben, J. Döring, Angewandte Chemie 1990, 102
85-87.
[56.] A. Müller, W. Plass, E. Krickemeyer, S. Dillinger, H. Bögge, A. Armatage, A. Proust, C.
Beugholt, U. Bergmann, Angewandte Chemie 1994, 106 897-899.
[57.] A. Müller, F. Peters, M. T. Pope, D. Gatteschi, Chem.Rev. 1998, 98 239-271.
[58.] A. Müller, S. Sarkar, S. Q. N. Shah, H. Bögge, M. Schmidtmann, Sh. Sarkar, P. Kögerler, B.
Hauptfleisch, A. X. Trautwein, V. Schünemann, Angewandte Chemie 1999, 111 3435-3443.
[59.] A. Müller, E. Krickemeyer, S. K. Das, P. Kögerler, S. Sarkar, H. Bögge, M. Schmidtmann, S.
Sarkar, Angew.Chem. 2000, 112 1674-1676.
72
[60.] A. Müller, S. Q. N. Shah, H. Bögge, M. Schmidtmann, P. Kögerler, B. Hauptfleisch, S. Leiding,
K. Wittler, Angew.Chem. 2000, 112 1677-1679.
[61.] A. Müller, S. K. Das, P. Kögerler, H. Bögge, M. Schmidtmann, A. X. Trautwein, V.
Schünemann, E. Krickemeyer, W. Preetz, Angewandte Chemie 2000, 112 3556-3559.
[62.] D. Volkmer, A. Du Chesne, D. G. Kurth, H. Schnablegger, P. Lehmann, M. J. Koop, A. Müller,
J.Am.Chem.Soc. 2000, 122 1995-1998.
[63.] V. W. Day, W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1977, 99 952.
[64.] V. W. Day, M. F. Fredrich, W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1977, 99 6146.
[65.] W. G. Klemperer, Angew.Chem. 1978, 17 246.
[66.] W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1976, 98 8291.
[67.] O. W. Howarth, P. Kelly, L. Pettersson, J.Chem.Dalton Trans. 1990, 81.
[68.] G. Johanson, L. Pettersson, N. Ingri, Acta Chemica Scandinavica 1979, A 33 305-312.
[69.] G. Johanson, L. Teterson, N. Ingri, Acta Chemica Scandinavica 1979, A 33 305-312.
[70.] M. Dieterle, G. Weinberg, G. Mestl, Phys.Chem.Chem.Phys. 2002, 4812.
[71.] G. Mestl, Ramanspektroskopische Charakterisierung von Oxidkatalysatoren, Universität
München, München 1994, p. 1.
[72.] T. Ressler, O. Timpe, T. Neisius, J. Find, G. Mestl, M. Dieterle, R. Schlögl, Journal of Catalysis
2001, 191 75-85.
[73.] J. Melsheimer, M. Thiede, R. Ahamad, Phys.Chem.Chem.Phys. 2003, 54366-4370.
[74.] G. Mestl, T. K. K. Srinivasan, Cat.Rev.Sci.Eng. 1998, 40 451-570.
[75.] N. Sotani, Bull.Chem.Soc.Japan 1975, 48 1820-1825.
[76.] L. J. Bellamy, The Infraredspectra of Complex Molecules, Mathuen, London 1966, p. 346.
[77.] B. Herzberg, Molecular Spectra and Molecular Structure, D. van Norstrand, New Jersey 1968,
p. 167.
[78.] B. M. Gatehouse, P. Leverett, J.Chem.Soc.(A) 1968, 1398-1405.
[79.] N. A. Caiger, S. Crouch-Baker, P. G. Dickens, J.Solid State Chemistry 1987, 67 369-373.
[80.] J. D. Guo, P. Zavalij, M. S. Whittingham, Eur.J.Solid State Inorg.Chem. 1994, 31 833-842.
73
[81.] E. M. McCarron III, J.Chem:Soc., Chem.Commun. 1986, 336.
[82.] E. M. McCarron III, D. M. Thomas, J. C. Calabrese, Inorganic Chemistry 1987, 26 370-373.
[83.] B. Krebs, I. Paulat-Böschen, Acta Cryst. 1976, B32 1697.
[84.] Wienold, J., 2003.
[85.] F. Schüth, Angew.Chem.Int.Ed. 2003, 42 3604-3622.
[86.] A. Müller, F. Peters, M. T. Pope, D. Gatteschi, Chem.Rev. 1998, 98 239-271.
[87.] J. L. Garin, J. M. Blanc, Journal of Solid State Chemistry -JSSCB 1985, 58 98-102.
[88.] V. W. Day, M. F. Fredrich, W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1977, 99 6146.
[89.] K.-H. Tytko, O. Glemser, Adv.in Chem.Series 1976, 19 239-315.
[90.] L. Kihlborg, Arkiv För Kemi Mineralogi och Geologi 1963, 21 427-437.
[91.] N. Yamazoe, L. Kihlborg, Acta Cryst. 1975, B31 1666.
[92.] B. Chae, Y. M. Jung, X. Wu, S. B. Kim, J.of Raman Spectroscopy 2003, 34 451.
[93.] M. A. Py, K. A. Maschke, Physica B1981, 105 376.
[94.] C. Julien, G. A. Nazri, Solid State Ionics 1994, 68 111.
[95.] G. A. Nazri, C. Julien, Solid State Ionics 1992, 376 53-56.
[96.] F. A. Cotton, R. M. Wing, Inorganic Chemistry 1965, 4867.
[97.] L. Seguin, M. Figlarz, R. Cavagnat, J. C. Lassegues, Spectrochimica Acta Part A-Molecular and
Biomolecular Spectroscopy 2222, 51 1323-1344, 1995.
[98.] I. R. Beattie, T. R. Gilson, J.Chem.Soc. 1969, A2322.
[99.] M. Fournier, C. Rocchicioli-Deltcheff, L. P. Kazansky, Chemical Physics Letters 1994, 223 297-
300.
[100.] H. Jeziorowski, H. Knözinger, Journal of Phys.Chem. 1979, 83 1166.
[101.] J. Haber, E. Lalik, Catalysis Today 1997, 33 119-137.
74
3 In situ Raman investigation of the
decreasing pH preparation method leading
to various MoO3structures
3.1 INTRODUCTION
Precipitation is one of the major synthesis routines for molybdenum oxide based
catalysts, because it is inexpensive and can be carried out industrially on a large
scale. The most common precursor is ammonium heptamolybdate tetra hydrate, and
the precipitating agent is often nitric acid, because all counter cations are nitrogen
based and can be easily removed by subsequent calcination. However, in many cases
not enough attention is drawn to the fact that the process of precipitation is crucially
affected by a multitude of parameters, such as temperature, precursor and acid
concentration or the nature of the counter cation. Reproducible results are only
obtained if all these parameters are fully controlled. A large study recently showed
that four different families of products can be obtained by just altering the parameters
mentioned[1]. During the reaction many isopolymolybdates are formed as
intermediates which are essentially pH-dependent and exist in equilibrium[2-14]. At
low temperatures a Mo36O1128- like compound was obtained with ammonia
heptamolybdate as precursor. At 70 °C hexagonal MoO3was obtained. At 50 °C
depending on the concentration of the starting solution either Mo36O1128- or
hexagonal MoO3could be obtained. The pH curves of the reactions at the different
temperatures were different and therefore different formation mechanisms most
likely prevail.
In order to achieve a rational catalyst preparation it is vital to know the reaction
mechanism in solution. Very early investigations on aqueous molybdenum systems
were carried out[15-17] in which determination of diffusion coefficients, UV/Vis and
conductometric-, potentiometric-, and thermometric–titrations were used as main
methods. Saski and Sillen[18] presented the first comprehensive potentiometric
investigation. A milestone was the identification of the hepta and octa molybdate
structures by X-ray single crystal analysis[19-21].
75
Another breakthrough was achieved by Aveston and Anacker[22] and Tytko and
Glemser[23-25] by combining Raman spectroscopy and single crystal analysis which
enabled them to correlate solid state structures with compounds in solution. As a
result a distinction was made between the heptamolybdate that dominates between
pH six and four, and the octamolybdate, which dominates, between pH four and two.
Contributions to reveal details of the reaction mechanism were delivered by O-NMR
spectroscopy[26-29]. Recently Cruywagen[30-32] published more potentiometric data
and an computer aided evaluation. Further, a combination of X-Ray Scattering and
Raman spectroscopy was applied[33].
Whilst the investigations mentioned above have brought many insights into the
solution chemistry of the molybdate system, they have only limited significance for
large scale catalyst preparation as almost all of these reactions were carried out at
room temperature (whatever temperature this is), in low dilutions and further
electrolytes such as NaClO4or NaCl were added. Catalyst synthesis is commonly
done at large scale, with high precursor concentrations, and the presence of chlorine
is avoided. In addition there is a lack of in situ analysis that shows exactly the
formation of precipitates out of octamolybdates. Therefore the study aims at
bridging the gap between structural inorganic chemistry and catalyst preparation.
3.2 EXPERIMENTAL
Preparation
A starting solution was prepared by dissolving ammonium heptamolybdate tetra
hydrate (AHM, Merck) in bidestilled water in order to obtain a Mo concentration of
0.7 mol/l. As precipitation agent HNO31 mol/l was applied. Whilst the molybdate
solution was stirred with a magnetic stirrer, constant temperature conditions were
maintained by using a water bath. Acid was added in 5 ml portions with an
automated titrator (Metrohm) and the pH measurement was performed with a pH
electrode (Mettler Toledo) carefully calibrated to the respective reaction
temperatures. Standards applied: pH = 2, 4, 7 (Merck).
76
Raman spectroscopy
Raman spectroscopy was performed on a Labram I (Dilor) instrument equipped with
a confocal microscope (Olympus). A notch filter (Kaiser Optical) was applied to cut
off the laser-line and the Rayleigh scattering up to 150 cm–1. The spectrometer is
equipped with a CCD camera (1024*298 diodes), which is Peltier cooled to –30 °C
to reduce the thermal noise. A He-laser (Melles Griot) was used to excite the Raman
scattering at 632 nm with a laser power of 1,4 mW. The following spectrometer
parameters were used: microscope objective: 10; slit width: 400 µm (spectral
resolution: 2.5 cm-1), integration time: 30 s per spectrum and 10 averages. To get
spectra from the solution the laser beam was redirected on the beaker with an optical
angle. The laser beam was used as well to check for any Tyndall effect, which
indicates particle formation in the solution.
3.3 RESULTS
The most decisive spectral changes in this investigation were observed between 740
and 1000 cm-1. Whilst the region between 740 and 800cm-1 is associated with Mo-O-
Mo bridging modes the region between 800 and 1000 cm-1 is assigned to terminal
Mo-O stretching frequencies[3].
3.3.1 Reaction at 30 °C
The starting pH of the reaction is 5.5 and the respective Raman spectrum (Fig. 3.1)
shows bands at 943, 898, 560, 362 and 219 cm-1. At pH 5.0 the band positions are
still the same however the intensity of the bands at 943 and 898 cm-1 has decreased
considerably.
At pH 4.5 the former band at 898 cm-1 has turned into a shoulder and its position is
slightly shifted to higher wavenumbers. The band at 943 cm-1 has now shifted to 948
cm-1. At pH 4.0 the band has a lower maximum height but it is broader. Its centre is
now situated at 950 cm-1. This tendency is continued at pH 3.5. However it is
noteworthy that a shoulder develops at 970 cm-1 pointing to the existence of
octamolybdate[34].
At pH 3.0 a band is detected at 959 cm-1, commonly assigned to octamolybdate[34].
This band shows a weak shoulder at 953 cm-1 and a more pronounced one at 921 cm-
1. Another band appears at 970 cm-1. At pH 2.5 the band at 959 and 970 cm-1 still
prevails the other spectral features underwent an up shift of 2 cm-1. These two bands
77
however are not very clearly separated. Therefore it could be interpreted as a single
band at 964 cm-1 which is split. This situation prevails until pH 1.5 is reached at this
pH the Tyndall effect starts to appear, which indicates formation of solid particles in
the solution.
At pH 1.4 bands are detected at 983 (sh), 970, 955 902 and 854 cm-1; the reaction
mixture is slightly cloudy. At pH 1.3 precipitation occurred coinciding with a total
decline in resolution. A similar effect was observed by Mestl[35]. Looking at the
bands at lower wavenumbers the band at 362 cm-1 underwent an up shift and
decreased in intensity. The band at 219 cm-1 decreased as well in intensity but shifted
to lower wavenumbers.
3.3.2 Reaction at 50 °C
The reaction at 50 °C starts with pH 5.17 and bands are detected (Fig. 3.2) at 943,
898, 560, 455, 362, 219 cm-1. Apart from the band at 455 cm-1 this matches with the
experiment carried out at 30 °C. At pH 4.5 the main band has shifted to 945 cm-1 its
height is decreased and it has broadened and at 897 cm-1 only a shoulder is detected.
980 960 940 920 900 880
30 °C
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.4
pH
wavenumber [cm-1]
relative intensity
Figure 3.1: In situ Raman investigation at 30 °C
78
980 960 940 920 900 880
50 °C
5.2
4.5
4.0
3.5
3.0
2.5
2.0
1.5
0.81
Wavenumber [cm-1]
pH
relative intensity
Figure 3.2: In situ Raman investigation at 50 °C
At pH 4.04 a shoulder appears at 970 cm-1 the main band starts to split and shows
two signals at 956 and at 949 cm-1, the shoulder at 896 cm-1 has lost intensity again.
At pH 3.51 a marked band has developed at 970 cm-1. Further bands a registered at
961, 954 (sh), 925 (sh) and 853 cm-1. At pH 2.99 the band positions are still the
same. The band at 970 cm-1 however has gained more intensity. Until pH = 2.06 is
reached the situation stays the same. At this pH the first Tyndall effect is observed.
The spectrum is changing at pH = 1.51. Bands are found at 981 cm-1 (assigned to the
presence of [Mo36O112]8-), 970 (sh), 955 and 900 cm-1. At pH 1.09 the same spectrum
is recorded but the intensities are lower. At pH = 0.81 precipitation sets in.
3.3.2 Reaction at 70 °C
At 70 °C the reaction starts at pH 5.0 bands are (Fig. 3.3) detected at 953 (sh), 943,
898, 560, 362, 254 (sh) and 220 cm-1. At pH 4.5 the situation has changed. Bands are
recorded at 955, 947 and at 896 cm-1. At pH 4.0 the intensity is increasing and bands
are situated at 969, 961 (sh), 956, 950 (sh) 946, 939 (sh) and 908 cm-1. At pH 3.5 the
band at 969 cm-1 is gaining intensity the one at 956 does not considerably change and
the band at 946 cm-1 diminishes.
79
980 960 940 920 900 880
70 °C
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
pH
wavenumber [cm-1]
relative intensity
Figure 3.3: In situ Raman investigation at 70 °C
At pH = 3.0 the above described trend prevails. The spectra at pH = 2.5, 2.0 and 1.5
are almost identical. They are characterised by bands at 970, 948, 920 (sh) and 847
cm-1. At pH = 1 precipitation sets in.
3.4 DISCUSSION
Although the solution chemistry of molybdates has been very well studied in the
past, this work is the first to present an in situ study of the system at different
temperatures. A summary of the findings is displayed in Fig. 3.5.
Mo7O246- + H+↔HMo7O245- (1)
MoO42- + Mo7O246- + 2H+↔ H3Mo8O285- (2)
H3Mo8O285-+ H+↔ Mo8O264- + H2O (3)
4 Mo8O264- + 4MoO42- + 16H+↔ Mo36O1128- + 8 H2O (4)
5Mo8O264- + 4H+↔8Mo5O162- + 2 H2O (5)
Mo8O264- + MoO42- ↔3Mo3O102- (6)
Mo8O264- + 4H+↔8MoO3+ 2 H2O (7)
80
The equations show the net reactions that occur during titration of the molybdate
solutions. For a closer evaluation of the pH curves the following assumptions have to
be taken into account. The solution consists of many different entities like
monomolybdates, which are present over the whole range of pH. The species
mentioned above are dominant at relevant pHs, transformation from one species into
the other occurs via three steps. The mechanism includes protonation and
condensation reactions, further it can be assumed that MoO6entities are always
detached and reattached.
The first three steps describe the transformation from hepta to octamolybdate. The
first step involves protonation of the heptamolybdate ion (1), which causes
instabilities of the system. Detachment and reattachment of MoO4units lead to (2) as
net reaction, a subsequent condensation reaction leads to octamolybdate (3). It has
been mentioned earlier, that at least three different octamolybdate isomers are
present in solution. Whilst the octamolybdate is present a pH = 2.7, the
octamolybdate is present at lower pH, and the octamolybdate is formed as
intermediate step between the and form (Fig. 3.4).
As the pH curves are more or less similar up to pH = 2.5 it can be assumed that every
reaction path has the octamolybdate as intermediate step. The further reaction path is
controlled by the external factors mentioned above like temperature, reactant
concentrations or nature of counter cation.
The pathway with the lowest activation barrier might be polycondensation, as this
allows distribution of the negative charge across many MoO6units. The equation (4)
will be shifted to the right if either the amount of molybdate or acid is large, further
the presence of a sufficiently large counter cation is required. Higher temperature
causes polycondensation of a lower extent, and reaction (5) is pursued, leading to the
hex phase.
It is well established that the dominating molybdate species in a solution between pH
= 6 and 3.5 is the heptamolybdate ion. The Raman spectra presented in this work
clearly support this assignment, as the major band at 943 cm-1 was assigned to the
terminal terminal Mo-O (MoO) frequency of heptamolybdate (C2v) in many
publications[22][36]. Only Ueda observed this band at 939 cm-1, but this small shift
may just arise from different pH or temperature conditions.
81
It has been reported earlier, that increasing amounts of protons lead to protonation
and condensation reactions[37]. In the Raman spectrum protonation will cause a
decrease in intensity of the terminal Mo-O vibration, and condensation will cause a
shift in band position. Indeed both phenomena have been observed. Further, a
correlation between the effects observed in the Raman spectra and the pH curve, that
was published earlier by the above authors can be made.
Protonation corresponds to a large buffering section in the pH curve and
condensation to a drop in pH. The first major condensation product is the
octamolybdate, a mechanism for that has been reported in the literature[31;38;39]. Many
octamolybdate isomers are reported to exist in equilibrium, whilst the -
octamolybdate is reported to be the dominating species at around pH = 2.7,
acidification is reported to shift the equilibrium to the molybdate[34]. There are
many publications about the mechanism of isomerization between the and the
form[26;40-42] although there is no convincing explanation about the factors that favour
isomerisation. Bridgeman assessed the thermodynamic stability of the two isomers
by carrying out DFT calculations. He reported that the octamolybdate is more
thermodynamically stable than the form, because of unfavourable steric
interactions of the form. Therefore it can be concluded that the isomerisation
process is kinetically controlled.
A mechanism for the isomerisation process has been postulated by Klemperer et
al.[26] with an intermediate octamolybdate. Whilst this isomer has been found in the
solid state[34] it has been not observed in solution in any large quantities, therefore it
can be considered as rather unstable. In any way, transformation of the into the
0,00 0,05 0,10 0,15 0,20 0,25
1,0
1,5
2,0
2,5
3,0
3,5
4,0
4,5
5,0
5,5
6,0
(2)
(3)
(1)
(4)
256 (0.7/1)(1/1.5)
257 (1/1)
258 (1/5)
pH
[HNO3]:[Mo]
Figure 3.4: pH-trace; interpretation of equivalence points
82
form requires breaking of at least two bonds. It is postulated by a number of authors
[43], that is process is enhanced by creation of hydrogen bonds of the type N-H...O-
Mo. A decrease in pH would drastically enhance the strength of this bond, therefore
the process of transformation into the form would be enhanced. In addition this
transformation via bond breaking is expected to have a considerable activation
barrier, hence a raise in temperature should ease the isomerisation process. Both
effects are observed in the Raman spectra presented.
The drastic temperature effect on the precipitation products has been reported
recently [44]. Whilst at 30 °C a smooth transformation into the supramolecular Mo36
species is observed, at 70 °C the octamolybdate is still the dominant species at pH
= 1. The octamolybdate is assigned as precursor for formation of the hexagonal
phase, but as this process requires massive bond breaking and reassembling, as the
form has only edge sharing connections, whilst the hexagonal precipitate has many
corner sharing connection. Again, the activation barrier has to be overcome by high
temperature. The fact that no precipitation of the supramolecular compound was
observed might be explained by the non existence of the octamolybdate, which is
largely present at lower temperatures. The form might either be a direct precursor
or at least a necessary compound for the Mo36 formation. If that is to be the case,
then the ratio of to octamolybdate, which is temperature dependent, determines
also the nature of the final precipitate.
A more quantitative interpretation of the Raman spectra is difficult for as there may
be many more molybdate species present in solution. The fact that no isosbestic
points were found supports this assumption[23]. Even more, neither the hexagonal
phase nor the supramolecular phase can be formed just out of octamolybdates,
therefore other intermediates need to appear.
Taking all this findings into account the variation of reaction temperature shifts the
equilibrium between the -Mo8O264- and the -Mo8O264- species. The location of this
equilibrium seems to determine the product obtained. All this can be very clearly
seen by the in situ Raman experimentation.
83
3.5 CONCLUSION
The in situ Raman data presented in this work has clearly demonstrated the severe
temperature effect on the nature of molybdate species in solution. Although the
temperature steps chosen in this work are rather large (20 °C), it must be assumed
that even a couple of degrees can significantly alter the ratio of species in solution.
The numerous preceding work in this field has to be re-evaluated in the light of the
results presented. Earlier pH studies, that showed different reaction mechanisms,
which occurred at different temperatures are confirmed by the results presented.
Figure 3.5: Dominant species in solution, dependent on temperature and pH.
Step 1: Protonation of the heptamolybdate ion. Step 2: T dependent
isomerisation. Step 3: Precipitation of a supramolecular Mo36 compound. Step
4: Kinetically controlled Precipitation of hex. MoO3
84
Reference List
[1.] S. Knobl, O. Timpe, Niemeyer D., S. B. A. Hamid, R. Schlögl, J.Catal. 2004.
[2.] A. Müller, F. Peters, M. T. Pope, D. Gatteschi, Chem.Rev. 1998, 98 239-271.
[3.] G. Mestl, T. K. K. Srinivasan, Cat.Rev.Sci.Eng. 1998, 40 451-570.
[4.] M. T. Pope, Progress in Inorganic Chemistry 1991, 39 181.
[5.] M. T. Pope, A. Müller, Angewandte Chemie 1991, 103 56-70.
[6.] L. Cronin, P. Kögerler, A. Müller, Journal of Solid State Chemistry -JSSCB 2000, 152 57.
[7.] M. I. Khan, A. Müller, S. Dillinger, H. Bögge, Q. Chen, J. Zubieta, Angewandte Chemie
1993, 105 1811-1814.
[8.] H. D. Lutz, R. Nagel, S. A. Mason, A. Müller, H. Bögge, E. Krickemeyer, J.Solid State
Chemistry 2002, 165 199.
[9.] A. Müller, E. Krickemeyer, M. Penk, V. Wittneben, J. Döring, Angewandte Chemie 1990,
102 85-87.
[10.] A. Müller, W. Plass, E. Krickemeyer, S. Dillinger, H. Bögge, A. Armatage, A. Proust, C.
Beugholt, U. Bergmann, Angewandte Chemie 1994, 106 897-899.
[11.] A. Müller, S. Q. N. Shah, H. Bögge, M. Schmidtmann, P. Kögerler, B. Hauptfleisch, S.
Leiding, K. Wittler, Angew.Chem. 2000, 112 1677-1679.
[12.] A. Müller, S. K. Das, P. Kögerler, H. Bögge, M. Schmidtmann, A. X. Trautwein, V.
Schünemann, E. Krickemeyer, W. Preetz, Angewandte Chemie 2000, 112 3556-3559.
[13.] A. Müller, S. Sarkar, S. Q. N. Shah, H. Bögge, M. Schmidtmann, Sh. Sarkar, P. Kögerler, B.
Hauptfleisch, A. X. Trautwein, V. Schünemann, Angewandte Chemie 1999, 111 3435-3443.
[14.] A. Müller, R. Rohlfing, J. Döring, M. Penk, Angewandte Chemie 1991, 103 575-577.
[15.] G. Jander, K. F. Jahr, W. Heukeshoven, Zeitschrift fur Anorganische und Allgemeine Chemie
1930, 194 383-428.
[16.] A. Rosenheim, Zeitschrift fur Anorganische und Allgemeine Chemie 1916, 96 139.
[17.] O. Glemser, W. Holznagel, A. Iftikhar, Z.Naturforsch. 1965, 20b 192.
85
[18.] Y. Sasaki, L. Sillén, Arkiv För Kemi Mineralogi och Geologi 1968, 29 253.
[19.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 18 325-341.
[20.] I. Lindquist, Arkiv För Kemi Mineralogi och Geologi 1950, 2349.
[21.] I. Lindquist, Acta Cryst. 1950, 3159.
[22.] J. Aveston, E. W. Anacker, J. S. Johnson, Inorganic Chemistry 1964, 3735-746.
[23.] K.-H. Tytko, B. Schönfeld, B. Buss, O. Glemser, Angewandte Chemie 1973, 85 305-307.
[24.] K.-H. Tytko, B. Schönfeld, Zeitschrift fur Naturforschung 1975, 30b 471-484.
[25.] K.-H. Tykto, G. Petridis, B. Schönfeld, Z.Naturforsch. 1980, 35b 45-56.
[26.] V. W. Day, W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1977, 99 952.
[27.] V. W. Day, M. F. Fredrich, W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1977, 99 6146.
[28.] W. G. Klemperer, Angew.Chem. 1978, 17 246.
[29.] O. W. Howarth, P. Kelly, L. Pettersson, J.Chem.Dalton Trans. 1990, 81.
[30.] J. J. Cruywagen, J. B. B. Heyns, Inorganic Chemistry 1987, 26 2569.
[31.] J. J. Cruywagen, Advances in Inorganic Chemistry 2000, 49 127.
[32.] J. J. Cruywagen, A. G. Draaijer, J. B. B. Heyns, E. A. Rohwer, Inorganica Chimica Acta
2002, 331 322.
[33.] G. Johanson, L. Pettersson, N. Ingri, Acta Chemica Scandinavica 1979, A 33 305-312.
[34.] S. Himeno, H. Niiya, T. Ueda, Bulletin of the Chemical Society of Japan 1997, 70(3) 631-
637.
[35.] G. Mestl, J.of Raman Spectroscopy 2002, 33 333-347.
[36.] K.-H- Tytko, G. Petridis, B. Schönfeld, Zeitschrift fur Naturforschung 1980, 35b 45-56.
[37.] S. B. A. Hamid, D. Othman, N. Abdullah, O. Timpe, S. Knobl, Niemeyer D., R. Schlögl,
Topics in Catalysis 2003, 24 87-95.
[38.] K.-H. Tytko, O. Glemser, Adv.in Chem.Series 1976, 19 239-315.
[39.] K.-H. Tytko, O. Glemser, Adv.Inorg.Chem.Radiochem 1976, 19 239.
[40.] W. G. Klemperer, W. Shum, J.Am.Chem.Soc. 1976, 98 8291.
86
[41.] A. Bridgeman, G. Cavigliasso, Inorganic Chemistry 2002, 41 3500-3507.
[42.] A. Bridgeman, J.Phys.Chem. 2002, 106 12151-12160.
[43.] X.-J. Wang, B.-S. Kang, C.-Y. Su, K.-B. Yu, H.-X. Zhang, Z.-N. Chen, Polyhedron 1999, 18
3371-3375.
[44.] S. Knobl, O. Timpe, N. Abdullah, D. Othman, Niemeyer D., S. B. A. Hamid, R. Schlögl,
Journal of Catalysis 2004, submitted .
87
4 Nanoclusters as Precursors to (MoVW)5O14:
In situ and chemical characterisation of the
systems of a single phase oxidation catalyst
4.1 INTRODUCTION
Selective partial oxidation of hydrocarbons is an important synthesis step for about 25%
of all organic products that are produced worldwide[1]. Particularly important reactions are
the oxidation of propane/propene to acrolein and acrylic acid. The common catalyst
material used for these reactions involves vanadium and tungsten doped molybdenum
oxides[2-8]. Therefore considerable efforts have been made to resolve the structure of the
catalytically active phase, as this would be of enormous economic and scientific benefit.
Mestl et al.[9-11] observed a Mo5O14 type single phase emerging during oxidation of
ethanol, acrolein and propene over mixed VWMo oxide and this structure was assigned to
the active phase of the catalyst.
The structure of the individual Mo5O14 was discovered first by Kihlborg[12]. The main
structural motif of Mo5O14 is a pentagonal bipyramidal coordination. The same motif has
also been observed by Müller[13] and identified as an important building block[14] for
formation of large polymolybdates (keplerates, >500 atoms), in which these pentagonal
bipyramidal units are linked and placed at the 12 corners of an icosahedron, and also by
Böschen et al. in K4Mo36O112 *8 H2O[15;16]. Even in the photochemical molybdenum
blues reaction this motif is discussed[17]. It is also known that binary Mo based oxides
doped with V, Nb, W or Ta[18;19] possess a structure similar to Mo5O14. Knobl et al.[20]
described the preparation of single phase (MoVW)5O14, which was achieved by spray
drying of a mixture of ammonium heptamolybdate (AHM), ammonium metatungstate
(AMT) and vanadyl oxalate solutions followed by thermal treatment. Raman spectroscopy
of the solution of the (MoVW)5O14 precursor showed bands that were assigned to terminal
M=O and bridging M-O-M vibrations characteristic of polyoxo metallates. No reference
was made to the element and oxidation state of these bridges. Structure and composition
of the pure Mo, W and V complexes as well as those of the mixed V-Mo, W-Mo species
in the solutions have been studied at the wide ranges of pH and component ratio[13;21-26].
88
However, the nature of the species of ternary VWMo obtained from the mixed solution of
AHM, AMT and vanadyl oxalate is still obscure.
The aim of the present study is to analyse the species in the precursor solution of
(MoVW)5O14 and to investigate the role of vanadium and tungsten concerning structure
development. UV/Vis, ESR and 95Mo NMR spectroscopy have been applied to get more
information about complexes developed in the precursor solution and hence about the
reaction scheme leading to the formation of the complex oxide phase.
4.2 EXPERIMENTAL
The precursor for the single phase Mo5O14 type catalyst was prepared according to Knobl
et al.[20] by mixing aqueous solutions of ammonium heptamolybdate (AHM with Mo
concentration of 0.963 mol/l ), ammonium metatungstate (AMT with W concentration of
0.271 mol/l), and vanadyl oxalate (V concentration of 0.84 mol/l), so that an atomic ratio
between Mo, Vand W will be equal to 0.68:0.23:0.09.
After mixing all parent solutions, concentrations of Mo, W and V became equal to 0.52,
0.07 and 0.175 mol/l, respectively. The pH of the mixed precursor solution became equal
to 3.5. For comparison some single and binary mixed aqueous solutions have been also
prepared with the same element concentrations as in mixed solution. Binary AHM and
AMT aqueous solutions with ratio of W/(W+Mo) equal to 9% and with concentration of
0.52 mol/l Mo and 0.07 mol/l W were prepared. The pH of the single and binary solutions
was close to 5.1. For some UV/Vis and NMR experiments the pH was lowered to 3.5 by
adding oxalic acid. Another set of the model binary solutions between AHM and vanadyl
oxalate with different ratios of V/(V+Mo) equal to 4.6, 9.2, 13.8, 18.4, 23% and
concentrations of Mo and V in the binary solution equal to 0.52 mol/l and 0.03, 0.07, 0.10,
0.14, 0.17, respectively, was synthesised. The pH of these binary solutions depends
strongly on the V content.
89
Further experiments concerning pH and conductivity measurements as well as in situ
UV/Vis spectroscopy were carried out in a computerised preparation machine. The
addition of the vanadyl oxalate solution into the AHM solution was carried out by Ismatec
micro pumps, the data recording is done by a computer at a sampling rate of 1 Hz. The pH
electrodes and conductivity probes are calibrated to respective temperatures. After mixing
was finished the prepared solution underwent a heating to 353K for one hour.
UV/Vis spectroscopy was carried out on a UV-2501 PC Shimadzu instrument. Whilst a
0.001 mm quartz cuvette was used for the region 3.7 eV to 6.2 eV, the spectral range from
1.8 eV to 5 eV was investigated by using a 0.006 mm quartz cuvette. Fibre reflection
UV/Vis spectroscopy was performed on a Ocean Optics SD 2000 spectrometer equipped
with a 600 lines/mm grating. As reflectant boron nitride (BN) [Alfa Aesar, hexagonal, 325
mesh, ~78 µm] was chosen.
The fibre has got a 8 around 1 configuration and a halogen lamp was used as light source.
The spectrometer was calibrated to a solution containing 10 g BN per one litre of water.
The calibration was done in the reaction vessel under reaction conditions. Outside light
was excluded by foil coverage. The reaction mixture (composition see above) contained
the same amount of BN. Comparison between the traditional transmission spectra and the
reflectance in-situ spectra shows identical band positions. Some intensities are inverted
3,0 3,5 4,0 4,5 5,0 5,5 6,0
0
10
20
30
40
50
60
70 A
B
C
D
Transmission [%]
eV
Figure 4.1:UV/Vis spectra of AHM and AHM+AMT solutions in the ligand
-
metal charge transfer region at different pH values: A. AHM at pH =
5.1, B.
AHM+AMT at pH = 5.1, C. AHM+AMT at pH = 3.5, D. AHM at pH = 3.5
90
due to crystallisation onto the BN particles but this should not derange the reaction
pathway because without BN an inhomogeneous crystallisation prevails.
The ESR spectra were obtained with a Bruker ESP-500 spectrometer with 100 kHz field
modulation. The spectra were recorder at 77K. A g-factors were determined using DPPH
as a standard. The accuracy of the g-factors measurements was not higher than 0.01. The
line width was measured with the accuracy 10 G (10 G=1 mT).
The number of VO2+ ions was determined by a method of double integration of the
experimental first derivative spectrum previously described by Blumenfeld et.al.[27] and
by comparison of the spectral intensity of each sample with a calibration curve obtained
by recording the spectral intensity of several solutions of VOC2O4. In the range of
concentration adopted the calibration was linear.
95Mo NMR spectra have been recorded on an MSL-400 Bruker spectrometer at a
frequency 26.08 MHz, with 20 KHz sweep width, 20 s pulse width and 1 s interpulse
delay using 1M Na2MoO4as an external standard.
4.3 RESULTS
4.3.1 UV/Vis spectroscopy
Fig. 4.1 shows UV/Vis spectra obtained in the 3.7-6.2 eV region of the ligand-metal
charge transfer for the pure AHM and mixed binary (AHM+AMT) solutions prepared at
different pH values. Acidification of the AHM solution to 3.5 pH by adding oxalic acid
led to a significant increase in absorption in the ligand-metal charge transfer region. This
effect can be assigned to protonation of the heptamolybdate species as well as to the
structural rearrangements leading to the formation of octamolybdates.
Indeed the species (HMo7O24)5- has been reported at pH = 2.5[22], also a similar effect was
observed on calculated spectra[29]. Favourable sites for protonation of the [Mo7O24]6- ions
are the twelve terminal oxygen atoms of the complex[37;38]. In this pH range the reaction to
[Mo8O26]4- occurs[24;25;47;48;41-45;50]. Addition of AMT to AHM to a ratio of 9%
(W/(W+Mo)) also increased the intensity of absorption; this effect was small at pH = 1
and became larger at pH = 3.5. The doubly protonated metatungstate ion (H2W12O42)10- is
reported to be the major species at pH = 3.5[24].
91
Addition of acidic vanadyl oxalate to AHM led to the decrease of pH and an increase in
intensity of the band (Fig. 4.2): again an effect largely due to protonation of the Mo-O
species and/or transformation of the heptamolybdate into octamolybdate. The maximum
intensity was achieved at 4.6 % of V/(V+Mo), further addition resulted in a slight
decrease in intensity. Addition of AMT re-increased the intensity (Fig. 4.2).
For all binary V-Mo solutions the d-d transitions at 2 eV and 1.6 eVthat are characteristic
for vanadyl species[28-29] are observed though their positions are slightly shifted into lower
eV values (see curves in Fig. 4.3 and 4.4). This fact may indicate the interaction of
vanadyl ion with Mo complex. After adding vanadyl oxalate to the ratio of 4.6 % in
(V/V+Mo) a broad band lying between 2.2 and 2.4 eV appeared. At 18.4 % vanadyl
oxalate (V/V+Mo) this band was gaining intensity.
It can be assigned towards heteronuclear ligand-metal charge transfer transitions between
V and W or Mo[30], meaning the creation of a Mo-O-V type bridge. Such a bridge is
4,0 4,5 5,0 5,5 6,0
0
10
20
30
40
50 A
B
C
E
D
F
I
G
H
Transmission [%]
eV
Figure 4.2: UV
/Vis spectra of the pure AHM (G), AMT (H), Vanadyl oxalate
(I), mixed V/V+Mo at different vanadium concentration (A: 4.6, B: 9.2, C:
13.8,
D: 18.4, E: 23%) and ternary (Mo0.68V0.23W0.09) (F) solutions in the ligand-
metal
charge transfer region
92
formed by condensation reactions between the protonated W-doped heptamolybdate
[HMo7O24]5- or octamolybdate [HMo8O26]3- and the vanadyl oxalate.
Addition of AMT did not change the shape of the spectral features, whereas the intensity
of the signal has diminished. Addition of tungstate to the mixed AHM/vanadyl oxalate
solution enhances both the band assigned to protonation and the band assigned to
condensation, these processes are increased by tungstate addition.
4.3.2 Conductivity measurements
This model is further supported by pH and conductivity measurements. Fig. 4.5 shows the
conductivity displayed against the pH. Acidification of the heptamolybdate caused by
addition of the vanadyl oxalate solution leads at first to an increase in conductivity. Such a
behaviour is expected because of additional fast charge carriers such as protons. The
conductivity reaches a maximum at pH=4.5, further vanadyl oxalate addition causes a
1,5 2,0 2,5 3,0 3,5
0
20
40
60
80
100
120
140
160
180
200 A
B
C
D
E
F
Vanadyl oxalate
AHM
AMT
Transmission [%]
eV
Figure 4.3: UV/Vis spectra of pure AMT, mixed V/V+Mo at their dif
ferent ratios
(A: 4.6, B: 9.2, C: 13.8, D: 18.4, E: 23%) and ternary (Mo0.68V0.23W0.09
) (F)
solutions in d-d transition region
93
decrease in pH: This effect can only be attributed to a substantial decrease in the number
of charge carriers by condensation reactions.
Addition of AMT solution causes a further substantial decrease in conductivity and
therefore enhanced condensation. However, changes in concentration and therefore
different activity of the respective species and different hydratisation leading to altered
conductivities needs to be taken into account. Different speeds of vanadyl oxalate and
AMT addition show different pH conductivity curves. This is a strong indicator, that the
protonation and condensation/polymerization processes are kinetically controlled.
4.3.3 95Mo NMR spectroscopy
It is well established that polynuclear anionic molybdenum oxocomplexes are formed
upon acidification of the aqueous solutions of MoO4tetraoxoanions[31;32].
Polycondensation reaction is described as:
Figure 4.4: In situ UV/Vis characterization; Addition of Va
nadyl oxalate (black)
and AMT (grey) to AHM; speed of addition 92 ml/min,
speed of stirring 200 t/min,
spectrum gaining intensity with increasing amount of Vanadyl oxalate
400 500 600 700 800 900
AHM
Start
5
10 min
15 min
20 min
25 min
30 min
35 min
40 min
45 min
R/R
wavelength [nm]
94
[MoO4]2- [Mo7O24]6-
[Mo8O26]4-.
The composition of the polyanions in the solution depends on the acidification degree[24-
26;46;47-49]. NMR spectroscopy gives a direct evidence for the type and composition of the
polynuclear species. Thus, the 95Mo NMR spectrum of [Mo7O24]6- consists of two peaks at
200 and 32 ppm with the intensity ratio of 1:6[23;33;38;39]. The spectrum of [Mo8O26]4- in the
aqueous solution represented by two peaks situated at about 100 and 10 ppm[32;38].
Fig. 4.6 shows the 95Mo NMR spectrum of the AHM precursor solution. By the position
of two peaks (32.8 and 210 ppm) visible in the spectrum (curve 1) species in this solution
are identified as those of [Mo7O24]6- that are in line with the literature data. After addition
of AMT to AHM an additional peak with 24.5 ppm appears in the NMR spectrum, at the
same time, the position of the peak at about 32 ppm does practically not change (see curve
B in Fig. 4.6).
Figure 4.5: Conductivity against pH. Addition of Vanadyl ox
alate and
AMT to AHM; speed of addition 6.5 ml/min and 92 ml/min, speed of
stirring 200 t/min
2,5 3,0 3,5 4,0 4,5 5,0 5,5
32
34
36
38
40
42
44
slow addition, 6,5 ml/min
fast addition, 92 ml/min
(conductivity [mS/cm])
(pH)
95
Earlier data[23;32-33] indicates that Mo6+ could be substituted by W6+ and form mixed
heptametalates of [Mo7O24]6-- type in the weak acidic solutions with a pH between 6 and
4. This complex is stable at the wide range of the ratios between the components (Mo7-
xWxO24 with x=1-6). In the binary solution under study the ratio between Mo and W is
equal to 7.5 that is why two peaks are observed in the spectra: the bigger one corresponds
to the pure [Mo7O24]6- complexes, and the smaller one arises due to the appearance of the
same species promoted by tungsten.
After addition of the vanadyl oxalate to the binary solution the shape of the NMR
spectrum has drastically changed. The peak at 210 ppm disappeared from the spectrum;
the broadened band at about 99 and peak at 4.3 ppm became visible (curve D). The band
position fits more closely to [Mo8O26]4- rather than [Mo7O24]6- [32;38].
The broadening of the signals in the NMR spectrum of the ternary mixed solutions in
comparison to the peaks obtained from the pure octamolybdates can be attributed to the
effects caused by tungsten doping and by the interaction with VO2+ ions. This suggestion
is further supported by the fact that no NMR spectrum of the ternary mixed solution peaks
responsible for the metatungstates which form in this pH range is observed.
(ppm)
-150-100-50
050
100150
200250
4
1
2
3
Figure 4.6: 95Mo NMR spectra of AHM (1), binary AHM+AMT (2), [Mo8O26]4
—
species (3) and ternary (Mo0.68V0.23W0.09) (4) solutions
96
4.3.4 ESR spectroscopy
ESR spectroscopy performed at 77 K and a room temperature did not reveal the presence
of Mo5+ and W5+ in the parent AHM and mixed binary and ternary solutions which
indicates that they are all in a diamagnetic oxidation state (6+). Fig. 4.7 shows ESR
spectra of the V-Mo solutions with different vanadium concentration and the final ternary
mixed solution.
This spectrum is characterised by hyperfine splitting and parameters of gII = 1.93 AII =
200 Gs and g= 1.975 A= 77 G which has been assigned to the isolated vanadyl ions in
the distorted octahedral environment[34;35]. However, additional peaks of the low intensity
closely situated at the parallel components of the hyperfine splitting are observed in the
ESR spectrum (see Fig. 4.7).
These lines could be assigned to the vanadyl ions (VO2+) but due to the low intensity of
additional maxima exact spectral parameters of this type could not be evaluated properly.
2500 3000 3500 4000 4500
A
B
C
D
E
F
(B [Gauss])
Figure 4.7
: ESR spectra of the binary AHM and vanadyl oxalate solutions
containing the following ratios of V/(V+Mo): 4.6% (A) 9.2% (B)
13.8% (C)
18.4% (D) and 23 % (E) and mixed VMoW solution (F).
DPPH was taken as
standard
97
After increasing the vanadium content in the solution (over 13.8 %) the appearance of the
slightly anisotropic and broad line with geff = 1.96 and gII < g, along with the earlier
observed spectrum, are registered by ESR. Similar spectra are observed for the MoVW
mixed solution (see Fig. 4.7) and the concentration of the isolated vanadyl ions in the
MoVW solution does not exceed 10 %.
Broadening of ESR signals may be caused by the formation of associated V4+ ions.
According to[40] in the case of the formation of associated species geff 1/3 [gII+2g], and
this ratios is valid for ESR spectrum of complex solution. Therefore, we suggested that a
broad and anisotropic ESR signal observed for the ternary solution could be assigned to
the associated vanadyl ions in the distorted octahedral coordination.
From the other side, it is well known[61]that magnetic exchange in –VO2+-O-VO2+-
assosiates is so strong that ESR spectra from these species are practically not observed,
e.g. in V2O4. Due to this fact one of the explanations of the broadening of ESR spectrum
given in Fig. 4.7 is associated with a possible presence of species of VO2+-O-Mo6+-O-
VO2+ -type with magnetic exchange through diamagnetic Mo6+ ions in the solutions.
This interaction of VO2+ species through Mo6+ ions is weaker than that between pure
vanadyl ions but is strong enough to broaden the ESR spectrum. The interpretation is in
line with UV/Vis spectroscopic data and conductivity measurements indicating the
formation of Mo-O-V type bridges.
4.4 DISCUSSION
This work shows the highly complex interaction between the initial compounds AHM,
AMT and vanadyl oxalate. During the mixing process several reactions are happening
simultaneously and lead to an important precursor for the final Mo5O14 structure.
Although the current literature describes either only the binary systems Mo-V and Mo-W
[51;24-26;52-57], a reaction mechanism can be set up that is based on the current literature as
well as on the UV/Vis, conductivity, ESR and 95Mo NMR measurements shown in this
work.
UV/Vis spectroscopy and the conductivity experiments have shown that addition of
vanadyl oxalate to ammonium heptamolybdate leads to kinetically controlled protonation-
and condensation reactions.
98
5
247
6
247 1OHMoHOMo k(Protonation)
OHOMoHOMo k2
4
268
6
247 107208 2
(Octamolybdate formation)
OnHMoOVOOMonVOOHMonn
k2
32
5
247 )(2
(Condensation)
OnHMoOVOOMonVOOHMonn
k2
42
5
268 )(2
(Condensation)
n
Figure 4.8: Connection between the octamolybdate units. The arrows denote further
terminal oxygen atoms, where additional vanadyl groups can connect
Addition of the vanadyl oxalate is the rate-determining step. Whilst quick addition leads
to higher polymerisation/condensation products, slower addition leads to a higher proton
concentration (lower pH), as observed (Fig. 4.5). In parallel protonated AHM is
transferred into octamolybdate. As bond distances in the octa-and heptamolybdate are
completely different this is one reason for the decisive changes in the UV/Vis spectra.
Further support is drawn from the 95Mo NMR study presented above that formation of
[Mo8O26]4- -type species is dominant in the MoVW mixed solutions (see also [24-28;41-50]).
These findings are in line with a long series of pH dependent investigations of
polyoxometallate chemistry. This complex depicted in Fig. 4.8 has 14 terminal and 12
bridging oxygen bonds. Terminal bonded oxygen atoms can be easily protonated and react
with the vanadyl cations[37;38]. Earlier Raman spectroscopy on this system[20] revealed the
formation of the molybdate species that are linked via Me-O-Me bridges, and the UV/Vis
and ESR measurements specified these bonds as VO2+-O-Mo6+-O-VO2+ bridges.
The recent findings in this work have been summarised in Fig. 4.8-4.10 The vanadyl
group links discrete [Mo8O26 ]4- and [Mo7O24 ]6- species in solution and form a polymer
consisting of [Mo8O26 ]4- and [Mo7O24 ]6- units where Mo atoms are partially substituted
99
by W. In this polymer the vanadyl ion is interconnected via the four oxygen atoms of the
equatorial plane of the vanadyl group with the molybdenum atoms possessing terminal
oxygen bonds from isolated [Mo8O26]4- complexes. An octahedral coordination can be
completed by water (Fig. 4.9). As [Mo8O26]4- has seven of such coordination sites
formation of a three dimensional oligomeric network is likely (Fig. 4.10).
According to[24] tungsten enhances the oligomerisation because tungsten-oxygen bonds
are much more favourable sites for protonation; hence the tendency of tungsten to form
chains is stronger. This is in line with various literature findings. Krebs et al. report
polymers consisting of Mo8O276- units interlinked by an europium species. In
hydrothermal conditions these polymers are found and octamolybdate can polymerise as
well[42;45;51;54-57].
In parallel to the oligomerisation process there is also internal isomerisation of the
octamolybdate units. Klemperer used 17O-NMR spectroscopy as a structural probe to
investigate the polymetallate chemistry in solution[58]. Especially -[Mo8O26]4- shows a
very dynamic behaviour in solution including the breaking of Mo-oxygen bonds[59]. A
detailed mechanism is suggested by Klemperer et al. how this rearrangement and
isomerisation reaction between and -[Mo8O26 ]4- occur. In systems containing only
Mo a pentagonal bipyramide is achieved at low pH (e.g. in [Mo36O112]8- )[60]. It is highly
likely that during these reactions the mainly edge sharing connections in the
heptamolydates and octamolybdates are broken and corner-sharing types are built up
which dominate the Mo5O14-type structure (Fig. 4.10).
Müller et al. point out the structural similarities between the Mo5O14 type structure and the
Mo36(H2O)24O96[14-17]. Therefore, the reactions leading to these compounds should be very
similar. Krebs et al. suggest from the analysis of the single crystal x-ray data that
[Mo7O24]6- acts as nucleus for the formation of [Mo36O112]8- and identifies water as an
important bridging ligand[48;49]. The removal of water could therefore lead to the
breakdown of the K8Mo36O112(H2O)16 unit and lead to the formation of the Mo5O14 type
solid. Knobl et al.[20] observed several releases of water by thermo gravimetric
experiments.
At least one more element apart from Mo seems to be necessary for that. It is therefore
likely that the presence of tungsten in the octamolybdate units favours the formation of the
pentagonal bipyramides.
Molecular structures close to those observed in the solutions can be preserved by spray
drying[20]; calcination of the spray dried material leads to the final (MoVW)5O14. Thus, the
100
main steps of the formation of the precursor solution of the complex MoVW catalyst are
described as follows. Upon mixing of AHM and AMT solutions pure [Mo7O24]6-
complexes along with W doped [Mo7O24]6- species appear. Addition of the Vanadyl
oxalate to the binary mixture leads to the protonation of the molybdate isopolyanions and
their transformation into the [Mo8O26]4- -type structure. The formed species become
interconnected with each other via vanadyl groups such providing a kind of the oligomeric
network. Incorporation of tungsten into the octamolybdate units leads to formation of the
pentagonal bipyramidal motive.
4.5 CONCLUSIONS
This work has substantially increased the understanding of the detailed synthesis routine
towards a single Mo(VW)5O14 oxidation catalyst. By using UV/Vis, ESR and 95Mo NMR
spectroscopy the nature of molybdate species in the AHM, AMT and mixed precursor
solutions of the ternary mixed oxide was studied. Results obtained show that structure
Figure 4.9: Three dimensional connections of octam
olybdate units via
vanadyl ions (Mo = light grey, O = red, V = blue)
101
formation starts as early as during the mixing process of the initial precursor solutions.
Addition of acidic vanadyl oxalate causes protonation and condensation of the
heptamolybdate ions into octamolybdate ions that are linked by vanadyl species and form
a polymeric network. Addition of tungsten atoms clearly enhance this process. This
polymeric network is an essential precursor for the final Mo5O14 –type structure.
Figure 4.10: Reaction steps. 1: Protonation of the heptamolybdate, 2:
Formation of octamolybdate (3 modifications), 3: Formation of the pentagonal
bipyramidal motif following a mechanism set up by Klemperer [58]
102
Reference List
[1.] N. Kurtz, R. W. Cunningdam, A. W. Naumann (Union Carbide Co.) [US 4111983] (1978).
[2.] V. Novak, L. Sokol, J. Jelinek, [CS 1207807] B(1981).
[3.] N. Bertolini, S. Ferlazzo (Euteco Impianti S.p.A.) [US 4289654] (1981).
[4.] T. Kawajiri, S. Uchida, H. Hironaka (Nippon Shokubai Kagaku) [EP 427 508 A1] (1991).
[5.] F.-G.Tenten, H. Martin, H. Hibst, L. Marosi, V. Kohl (BASF AG), [EP 668104 B1] (1995).
[6.] H. Hibst, S. Unverricht (BASF), [DE 19815281 A 1].
[7.] M. Tanimoto, H. Himeji-shi, I. Mihara, H. H. Aboshi-ku, T. Kawajiri, H. Himeji-shi, (Nippon
Shokubai), [EP 0 711 745 B1].
[8.] F.-G. Tenten, H. Hibst, H. Martin, L. Marosi, V. Kohl (BASF), [DE 4405514 A1].
[9.] G. Mestl, Ch. Linsmeier, R. Gottschall, M. Dieterle, J. Find, D. Herein, J. Jäger, Y. Uchida, and R.
Schlögl, J. Mol. Catal. A, 2000, 162, 455.
[10.] O. Ovsitser, Y. Uchida, G. Mestl, G. Weinberg, A. Blume, M. Dieterle, H. Hibst, and R. Schlögl, J.
Mol. Catal. A, 2002, 185, 291.
[11.] M. Dieterle, G. Mestl, J. Jäger, Y. Uchida and R. Schlögl, J. Mol. Catal. A, 2001, 74, 169.
[12.] L. Kihlborg, Ark Kemi, 1963, 21, 427.
[13.] A. Müller, P. Kögerler, C. Kuhlmann, Chem.Comm., 1999, 1347-1358.
[14.] L. Cronin, P. Kögerler, A. Müller, Journal of Solid State Chemistry, 2000, 152, 57.
[15.] I. Paulat-Böschen, J.C.S. Chem. Comm., 1979, 780.
[16.] B. Krebs, I. Paulat-Böschen, Acta Cryst., 1982, B 38, 1710.
[17.] T. Yamase, P.V. Prokop, Angew. Chem. Internat. Ed., 2002, 114, 3, 484.
[18.]T. Ekström, M. Nygren, Acta Chem. Scand., 1972, 26, 1827.
[19.]T. Ekström, M. Nygren, Acta Chem. Scand., 1972, 26, 1836.
[20.] S. Knobl, G. A. Zenkovets, G. N. Kryukova, O. Ovsitser, D. Niemeyer, R. Schlögl, and G. Mestl, J.
Catal., 2003, 215, 177.
[21.] M.T. Pope, Heteropoly and Isopoly Oxomatallate, Springer-Verlag, Berlin, 1983.
[22.] J.J. Cruywagen, B. Krebs, I. Paulat-Böschen, Acta, Cryst. Sect. B., 1976, B32, 1697.
[23.] R.I. Maksimovskaya, K.G. Burtseva, Polyhedron, 1985, 4, 1559.
[24.] J.J. Cruywagen, Advances in Inorganic Chemistry, 2000, 49, 127.
[25.] J.J. Cruywagen, B.B. Heyns, Inorg. Chem., 1987, 26, 2569.
103
[26.] J.J. Cruywagen, A.G. Draaijer, J.B.B. Heyns, E.A. Rohwer, Inorganica Chimica Acta, 2002, 331,
322.
[27.] A.A. Blumenfeld, V.V. Voevodskii, A.G. Semenov, Application of Electron Paramagnetic
Resonance in Chemistry, Nauka, Novosibirsk, 1962 [in Russian].
[28.] J. Ballhausen, H. B. Grey, Inorg. Chem., 1962, 1, 111.
[29.] A.B.P. Lever, «Inorganic electronic spectroscopy », 1984, Elsevier (Amsterdam, Oxford, New-
York, Tokyo).
[30.] L. David, C. Craciun, M. Rusu, O. Cozar, P. Ilea, D. Rusu, Polyhedron, 2000, 19, 1917-1923.
[31.] S.F. Gheller, M. Didney, A.F. Masters, Austr. J. Chem., 1984, 37, 1825.
[32.] P. Sarrazin, B. Mouchel, S. Kazstelan, J. Phys. Chem, 1989, 93, 904.
[33.] I. Andersson, J. J. Hastings, O. W. Howarth, L. Pettersson, J. Chem. Soc. Dalton Trans., 1994, 1061.
[34.] M.C. Paganini, L. Dall’Acqua, E. Giamello, L. Lietti, P. Forzatti, G. Busca, J. Catal., 1997, 166,
195.
[35.] F. Kubec, Z. Sroubek, J. Chem. Phys., 1972, 57, 1660.
[36.] Y.Y.H. Chao, D.R. Kearns, J. Phys. Chem., 1977, 81, 666.
[37.] J.P. Jolivet, M. Henry, J. Livage, E. Bescher, Metal Oxide Chemistry and Synthesis: From Solution
to Solid State, John Wiley & Sons, Chichester, 2000, p. 20.
[38.] O.W. Howarth, P.J. Kelly, J. Chem. Soc. Dalton Trans. 1990, 81-84.
[39.] O.W. Howarth, P.J. Kelly, Chem. Soc. Chem. Comm., 1988, 1236.
[40.] S.A. Al’tshuler, B.M. Kozyrev, «Electronic paramagnetic resonance of the compounds of elements
of transition metals», Nauka, Moscow, 1972 [in Russian].
[41.] J. Fuchs, H. Hartl, Angew. Chem. 1967, 88, 385.
[42.] J. Fuchs, H. Hartl, W.D. Hunnius, S. Mahjour, Angew. Chem. 1975, 87, 634.
[43.] I. Paulat -Böschen, B. Buss, B. Krebs, Acta Cryst. 1974, B30, 48.
[44.] O. Glemser, G. Wagner, B. Krebs, Angew. Chem. 1970, 82, 639.
[45.] I. Paulat -Böschen, B. Buss, B. Krebs, O. Glemser, Angew. Chem. 1973, 85, 409.
[46.] J. Aveston, E.W. Anacker, J.S. Johnson, Inorg. Chem. 1964, 3, 735.
[47.] H. Tytko, G. Baethe, E.R. Hirschfeld, K. Mehmke, D. Stellhorn, Z. anorg. allgem. Chem. 1983 503,
43.
[48.] K. H. Tytko, O. Glemser, Adv. Inorg. Chem. Radiochem, 1976, 19, 239.
[49.] O. Glemser, W. Holznagel, S. Iftikhar Ali, Z. Naturforschung, 1965, 20b, 192.
[50.] J. Bridgeman, J. Phys. Chem. A., 2002, 106, 12151.
[51.] B. Krebs, I. Loose, M. Bösing, A. Nöh, E. Droste, C. R. Acad. Sci. Paris, t. 1, Serie II c, 1998, 351.
104
[52.] T. Pope, Progress in Inorganic Chemistry, 1991, 39, 181.
[53.] A. Müller, F. Peters, M. T. Pope, D. Gatteschi, Chem. Rev., 1998, 98, 239.
[54.] G. Liu, Y-G. Wei, Q. Yu, Q. Liu, S-W. Zhang, Inorganic Chemistry Communications, 1999, 2, 434.
[55.] R. Villanneau, A. Proust, F. Robert, F. Villain, M. Verdaguer, P. Gouzerh, Polyhedron, , 2003, 22,
1157.
[56.] C-L. Pan, J-F. Song, J-Q. Xu, G-H. Li, L. Ye, T-G. Wang, Inorganic Chemistry Communications,
2003, 6, 535.
[57.] R. S. Raring, L. Bewley, J. Zubieta, Inorganic Chemistry Communications, 2003, 6, 539.
[58.] W.G. Klemperer, Angew. Chem. Int. Ed., 1978, 17, 246.
[59.] V.W. Day, M. F. Frederich, W. G. Klemperer, W. Shum, J. Am. Chem. Soc., 1977, 99, 952.
[60.] R. Schlögl, D. Niemeyer, S. B. A. Hamid, O. Timpe, S. Knobl, D. Othman, N. Abdullah, Topics in
Catalysis, 2003, submitted.
[61.] C.P. Slichter, in Principles of Magnetic Resonance, Harper-Row, New York, 1963.
5 The Synthesis and Structure of a Single
Phase, Nanocrystalline MoVW Mixed Oxide
Catalyst of the Mo5O14-Type
5.1 INTRODUCTION
About one quarter of all organic products produced world-wide are synthesised via
catalytic selective partial oxidation reactions. The industrial and economic relevance
of research in this field, hence is self evident. Generally, these industrial processes
are highly developed. Improvements of such processes can only be achieved if
fundamental understanding is reached about the active catalyst structures and their
relation to the catalytic performance.
Currently, MoVW supported catalysts are used in industry for the synthesis of acrylic
acid[1-5]. Despite this industrial importance, fundamental information is not available
neither on the structure formation during synthesis nor on the peculiarities of the
atomic arrangements in these systems depending on different preparation routes and
element ratios.
Previously, the Mo5O14-type phase was found to be highly important for selective
partial oxidation catalysis for example methanol oxidation[6], acrolein oxidation[7],
and propene oxidation[8]. Proceeding work[6-9] showed that the selectivity for partial
oxidation products could be considerably augmented when the amount of Mo5O14
was increased. The structure of this oxide is built up by pentagonal bipyramids and
octahedrally co-ordinated metal centres[10]. Coordination sphere changes can easily
be imagined in this structure which might explain its activity in partial oxidation
catalysis. This oxide forms at a definite transition metal ratio of Mo:V:W equal to
0.68:0.23:0.09. At the same time binary molybdenum based oxides doped with
different elements such as Nb, W and Ta have been synthesised and their structure
was identified as that of the Mo5O14-type[11;12]. These phases were found to be stable
at a wide temperature range and a broad variation of the element ratios. However, the
ternary system has not yet been synthesised as a single phase material without traces
106
of other molybdenum oxides, e.g. thermal treatment in inert atmosphere results in the
appearance of MoO2, while treatment in air leads to MoO3.
Its identification as an important phase for partial oxidation catalysis, makes this
(MoVW)5O14 oxide also important as a model substance for fundamental surface
science studies, provided single crystals could be grown. Attempts to grow single
crystals via gas phase transport reactions or sintering[8;13;14] showed that this phase
forms from mixtures of molybdenum and tungsten, from mixtures of molybdenum,
tungsten and vanadium, but not in case of binary Mo-V mixtures with high V
concentrations. Both tungsten and vanadium play important roles as structural
promoters in the formation and stabilization of this oxide and hence for catalytic
activity.
It seems plausible that different thermal treatments of the precursor solutions effect
a) the composition of the usually mixed phase catalysts and b) the crystal sises of the
different constituting phases. Thus, the understanding of the aqueous precursor
chemistry is required to control the preparation of such mixed oxide catalysts.
Furthermore, subsequent drying and activation procedures from the liquid precursor
to the active and selective catalyst are of paramount importance for the development
of the optimum catalytic performance. Accordingly, knowledge has to be generated
about the detailed processes which occur during these synthesis steps. Only then, it
might be possible to fully control not only the phase composition of the mixed oxide
catalyst, but also the crystal size, the crystallinity, and the morphology of the active
phase. A developed synthesis routine thus could lead to defined crystal sizes or even
nano-crystalline (MoVW)5O14 mixed oxide catalysts. Moreover, it offers a versatile
path to control its elementary composition. Effects of crystallite size / morphology
and elemental composition could be studied separately on the catalytic performance.
To this end, some steps of the developed aqueous preparation procedure are
characterised by in situ micro Raman spectroscopy. The important, subsequent
drying process as well as further activation and formation procedures are investigated
by in situ Raman spectroscopy, HREM and XRD. Comparison with Raman spectra
of well defined, single-crystalline reference oxides[13] was used to assign the obtained
spectra during these catalyst preparation routes to certain oxides, such as MoO2,
Mo4O11, Mo8O23, MoO3, or Mo5O14.
107
This paper was aimed at the preparation of a single phase Mo5O14-type material and
the study of its molecular architecture at every step of the synthesis. Possible changes
of the particle morphologies of the ternary oxide and its precursors during thermal
treatment were also in the focus of this study. Last but not least, the detected physical
alterations during synthesis should be related to the changing catalytic activity and
selectivity in the partial oxidation of acrolein.
This approach of a knowledge-based development of all synthesis steps combined
with the use of complementary physicochemical characterization techniques led to
the defined preparation of a single phase (MoVW)5O14 oxide catalyst for acrolein
partial oxidation.
5.2 EXPERIMENTAL
The mixed oxide catalyst with the Mo0.68V0.23W0.09Oxcomposition was prepared by
spray-drying of mixed solutions of ammonium heptamolybdate (AHM, Merck, p.a.),
ammonium metatungstate (AMT, Fluka, purum, > 85 % WO3gravimetric), and
vanadyl oxalate of the respective transition metal concentrations and a pH value of 2.
The aqueous solution of AHM with the concentration of 0.963 mol/l MoO3was
prepared by dissolving AHM in bi-distilled water at 353 K. The aqueous solution of
AMT with the concentration of 0.271 mol/l WO3was prepared by dissolving AMT
in bi-distilled water at 353 K. The aqueous solution of vanadyl oxalate with the
concentration of 0.379 mol/l was prepared by dissolving V2O5(Merck, extra pure) in
an aqueous solution of oxalic acid (EGA-Chemie, >99%) with the concentration of
1.93 mol/l at 353 K. The different transition metal precursor solutions were allowed
to cool to room temperature and then mixed by adding the corresponding amounts
with a metering pipette. The mixed solutions were heated at 353 K for 1h.
The Mo0.68V0.23W0.09 oxide solid catalyst precursor was prepared by spray-drying the
mixtures of the aqueous solutions of AHM, AMT and vanadyl oxalate with a spray-
dryer of the `Anhydro´-type. Subsequently, the obtained product was calcined at 623
K for 120 min in flowing air (flow rate: 3.6 l/h) and at 723 K in flowing helium (flow
rate: 3.6 l/h) for 120 min or only at 713 K in flowing air for 240 min. The heating
rate to these end temperatures was 6 K/min. The calcination was done in a gas flow
reactor placed in a tube furnace.
108
The Mo, V, and W concentrations in the precursor solutions, the solid precursor and
the catalysts were determined by Atom Absorption Spectroscopy (AAS), Perkin
Elmer (PE 4100).
Raman spectroscopy was performed on a Labram I (Dilor) instrument equipped with
a confocal microscope (Olympus). A notch filter (Kaiser Optical) was applied to cut
off the laser-line and the Rayleigh scattering up to 150 cm–1. The spectrometer is
equipped with a CCD camera (1024*298 diodes), which is Peltier cooled to 243 K to
reduce the thermal noise. A He-laser (Melles Griot) was used to excite the Raman
scattering at 632 nm with a laser power of 1,4 mW. The following spectrometer
parameters were used: microscope objective: 100; slit width: 200 µm (spectral
resolution: 2.5 cm-1), integration time: 240 s per spectrum and 20 averages.
X-ray diffraction (XRD) analysis was carried out using an URD-63 spectrometer
with Cu Kradiation in the 5 - 70 2Θ range by continuous scanning (2° Θ/min). The
integration time was set at 20 s. The POLYCRYSTAL software package[15] was used
to refine the structure of the sample calcined in air at 623 K and He at 713 K.
Additional X-ray diffraction (XRD) measurements were done at room temperature
on a STOE STADI-P focusing monochromatic transmission diffractometer equipped
with a Ge(111) monochromator and a position sensitive detector. Cu-Karadiation
was used. The phase analysis was performed with the STOE Win XPOW software
package (version 1.06; Stoe Darmstadt, Germany) and with PowderCell (V 2.3;
Bundesanstalt für Materialforschung und -prüfung (BAM) Berlin, Germany).
The particle size distribution was measured on a Coulter Counter TA- 2 from
Coultronics. The specific surface areas were determined by the BET method
(Micromeritics).
The morphology and the size of the catalyst particles after spray-drying and during
calcination was determined by SEM analysis. SEM was conducted on a S 4000 FEG
microscope (Hitachi). The acceleration voltage was set at 10 kV, the objective
aperture was 30 mm, and the working distance was 10 mm.
High resolution transmission electron microscopy (HRTEM) analysis was carried out
on a CM 200 electron microscope (Philips) (point resolution: 0.2 nm, acceleration
voltage: 200 kV) equipped with an EDX system (EDAX). For HRTEM analysis, the
samples were prepared from suspensions of the powders in methanol. A drop of the
methanol slurry was put on holey-carbon / copper grids.
109
Thermal analysis (TA) was performed with a STA 449 C Jupiter apparatus
(Netzsch). Flowing helium and air atmosphere were applied (the flow rate was set at
15 ml/min in both cases). The heating rate was set at 10 K/min and 20 K/min. Mass
spectrometric analysis (TG-MS) of the evolved gases was performed with an
Omnistar quadrupole mass spectrometer (Pfeiffer Vacuum).
The sieve fraction between 0.4 and 0.6 mm of crushed pellets was used for the
catalytic measurements. The catalytic tests were performed in a quartz tubular flow
reactor (i.d. 4mm). The catalyst (0.025-0.1 g) was diluted with quartz (1:10 –1:4 by
weight) to achieve a better temperature control. Reaction mixtures of 4% C3H4O, 8%
O2, 20% H2O, and the balance He with total flow rates between 0.2 to 2 ml/s were
used for the catalytic measurements. The reactants and products were analyzed by an
on-line gas chromatograph (Varian 3800) equipped with TCD and FID detectors. A
Porapak-QS column (2 m×1/8 IN, s.s.) and a 60/80 CarboxenTM 1000 column (15
FTs×1/8 IN, s.s.) were employed for the analysis of permanent gases and organic
substrates. The carbon mass balance was 100±5 %.
5.3 RESULTS AND DISCUSSION
5.3.1 SEM
Table 5.1 EDX Result
Elements (At %) T = 383 KT = 623 KT = 713 K
V 18.9 21.2 24.4
W10.1 11.1 7.3
Mo 70.9 67.7 68.3
Fig. 5.1 A shows a characteristic SEM image of catalyst particles after spray-drying.
All particles are spherically shaped and have smooth surfaces. Their sizes range from
2 – 50 m with main values lying between 5 – 10 m. The shape and the size of the
spherical particles did not change after the calcination in air of the spray-dried
material at 623 K. But the observed morphology of the sample has changed after
110
heating at 713 K in helium or in air. These results are given in Fig. 5.1 B, C. One can
see that now the spherical particles consist of numerous smaller particles of regular
shape and size ranging from 0.2 to 0.5 m. The EDX results given in Tab. 5.1 show
that the concentrations of the different elements slightly change with thermal
treatment.
Figure 5.1: Typical SEM images of the spray-dried precursor A), the MoVW
oxide catalyst heated at 623 K in air and at 713 K in He B, C) at different
magnifications
5.3.2 Particle size distribution and BET surface area
Analysis of particle size distribution shows that most particles have sizes between 5
and 10 m (59.5 %) and this data is in line with the SEM analysis. About 21 % of the
particles are smaller than 5 m, and a few particles are bigger than 10 m (19 %).
After calcination at 623 K and 713 K, the particle size distribution did not decisively
change.
The specific BET surface area of the dried precursor was very low with 0.8 m2/g, the
one of the calcined oxide was 4.1 m2/g. These values are close to those reported for
the industrial MoVW mixed oxide catalyst[6-9].
5.3.3 TG-MS
The TA data of the spray-dried Mo0.68V0.23W0.09Oxsample, measured in flowing
synthetic air, is shown in Fig. 5.2. Four main mass losses are observed at 385, 545,
621, and 675 K. DSC performed simultaneously with TA shows endothermic events
at 385 K, 525 K and 675 K, and two exothermic effects at 563 K and 621 K, which
are similar to the mass losses as evident from the DTG trace.
111
The first endothermic effect is associated with a mass loss of 3.5 % in the TA. As
seen from the MS traces, 33.4 % of the total amount of water in the catalyst is
released from the sample. Water, carbon dioxide, ammonia and nitrogen oxides are
released at the endothermic and exothermic effects at 525 and 563 K, respectively.
The release of carbon dioxide is due to the decomposition of vanadyl oxalate.
Nitrogen oxides are formed from the oxidation of ammonia and the reduction of the
vanadium and molybdenum precursors. An exothermic effect is detected at 621 K,
which is related to only a slight mass loss of 3.5 %. Water and traces of ammonia,
carbon dioxide and nitrogen dioxide are observed by MS. This exothermic effect is
associated with a crystallisation process as revealed by SEM and XRD.
Fig. 5.3 displays the TA data obtained in flowing helium of the sample which was
pre-treated in flowing air at 623 K for 2h. Two major mass losses are observed of 4.6
60
70
80
90
100
30
40
50
60
70
80
90
100
norm. MS-Signal
b
exo
endo
621
545
675
1.6 %
1.9 %
18.7 %
3,5 %
c
a
DSC Signal [µV]
300 400 500 600
385
NO
NO2
CO2
NH3
H2O
Mass Loss [%]
Tem perature [K]
Figure 5.2: TA in flowing synthetic air of the spray-dried Mo0.68V0.23W0.09O
x
catalyst. a) TG data, b) DTG data, c)
DSC data (upper panel); MS traces of
evolved H2O, NH3, CO2NO2, and NO
112
and 1.4 %, respectively. The DSC curve shows endothermic events at 378 K and 555
K, which correlate with the mass losses as also confirmed by the DTG trace. The first
effect is accompanied by a mass loss of 4.6 % as shown in the TG curve. The MS-
analysis of the evolved gases shows that only water is released from the sample. The
second endothermic effect at 555 K is also accompanied by loss of water. Other
thermal effects or mass losses were not observed up to 743 K. In this experiment, the
release of carbon dioxide or nitrogen oxides was not observed. These compounds
obviously were completely released during the pre-treatment of the precursor in air at
623 K for 2h, in line with the TA data shown in Fig. 5.2.
5.3.4 XRD
Fig. 5.4 displays the results of the XRD analysis of the Mo0.68V0.23W0.09 mixed oxide
samples after the different treatment steps. The XRD pattern shown in Fig. 5.4 of the
initial and calcined samples differ considerably. The XRD patterns reveal that the
mixed Mo0.68V0.23W0.09 oxide precursor was poorly crystallised after spray-drying
(Fig. 5.4 a) and after calcination in air at 623 K (Fig. 5.4 b).
92
94
96
98
100
300 400 500 600
555
378
1.4%
4.6%
Mass Loss [%]
-0,1
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
c
b
a
DSC Signal [µV]
H2O
MS-Intensity [a.u.]
Temperature [K]
Figure 5.3: TA in helium of the spray-dried Mo0.68V0.23W0.09Ox
catalyst
pre-treated in air at 623 K for 2h. a) TG-data, b) DTG data, c)
DSC data
(upper panel); lower panel: MS trace of evolved water
113
The broad halo lying between 5 and 14o2 is characteristic for the XRD pattern
of the spray-dried sample as evident from Fig. 5.4 a. It should be noted that a
set of reflections of the Mo5O14 structure, such as (210), (310), appear in
this 2 region (pattern c and d of Fig. 5.4). After thermal treatment of the
spray-dried sample at 623 K in air, the halo disappeared and a broad XRD peak
is observed at 22° 2in the XRD pattern of this sample (see diffractogram b of
Fig. 5.4).
In this case, the peak location is very close to that of the (001) reflection of Mo5O14.
The spray-dried material shows a second broad signal at 27.6° 2 The broad
reflection, observed in this range (pattern a of Fig. 5.4) transforms into a set of sharp
10 20 30 40 50 60 70 80
821
741
731
721
551
631 541
601 740
730640
550
630 540
610 600 001
520
510
420
330
400
310
210
200
e
d
c
b
a
Intensity
2
25 30 35
111
021
040
110
Figure 5.4: X-ray pattern of the MoVW oxide precursor/catalysts: a) spray-
dried
precursor , b) catalyst calcined in air at 623 K for 2h, c) catalyst calcined in air
at
723 K for 2h, d) catalyst calcined in air at 623 K and in helium at 713 K, and e)
catalyst calcined in air at 623 K for 2h after operation in acrolein oxidation for
80h. The
inset shows the 2Θ region of diffractogram c which shows reflections due
to MoO3
114
peaks, including the (550) reflection of Mo5O14, after calcination in air at 723 K (Fig.
5.4 c), and in air at 623 K plus in helium at 713 K (Fig. 5.4 d).
These observations indicate that a Mo5O14-type structure is pre-formed in the spray-
dried precursor but of course with a very low degree of structural ordering within the
basal plane. This observation is in agreement with published results of the structural
studies of the nanocrystalline, industrial catalyst[6,16].
The XRD pattern of the MoVW sample calcined in air at the higher temperature of
723 K is shown in Fig. 5.4 c. This temperature results in the crystallisation of the
complex Mo-V-W oxide of the Mo5O14-type from the nanocrystalline, spray-dried
material as confirmed by the appearance of the sharp and characteristic peaks of this
phase in the XRD diffractogram (pattern c of Fig. 5.4).
However, traces of a MoO3phase are also detected for this sample. The inset of Fig.
5.4 shows the 2Θ range of the diffraction pattern c, in which the reflections of MoO3
occur as marked by their respective indices. The result of the assignment of the
reflections are given in Tab. 5.2. These XRD data are corroborated by Raman
spectroscopy (vide infra). Hence, the single phase Mo0.68V0.23W0.09 O compound of
the Mo5O14–type structure could not be identified when the sample was heated in air
at 723 K. MoO3, known for its total oxidation activity, is definitely present in the
catalyst.
The single phase Mo0.68V0.23W0.09O compound of the Mo5O14–type structure is only
formed after subsequently heating the catalyst, pre-calcined in air at the lower
temperature of 623 K, in helium at 713 K as evident from the XRD pattern d of Fig.
5.4, which additionally shows the indices of the reflections of Mo0.68V0.23W0.09O.
It has to be noted that the XRD pattern recorded for the catalyst, which was
thermally treated in air at 623 K and then operated in the acrolein oxidation for 80 h
(pattern e of Fig. 5.4) is similar to pattern d of Fig. 5.4. This observation confirms
recent results on the enhanced formation of the Mo5O14-type structure during
acrolein oxidation[7].
The XRD pattern d of Fig. 5.4. was used for the structural refinement of the ternary
oxide sample. According to the results obtained, the structure is well described by the
space group P4/nmm with following lattice parameters: a = 4.54063, b = 4.54063, c
= 0.39979 nm, respectively. A slight decrease of the unit cell parameters of the
115
thermo-activated ternary oxide is observed as compared to the reference Mo5O14
Kihlborg[10] (P4/mbm space group, a = 4.5990, b = 4.5990, c = 0.3936 nm).
The incorporation of V and W ions into the structure of this molybdenum oxide
could be the reason for this. It was shown above that crystalline particles of this
sample have W and V contents as high as 9 and 23 at.%, respectively.
Data given in Fig. 5.4 may provide the supposition that the structure of the spray-
dried catalyst precursor only exhibits close-range order rather than nanocrystalline
3D periodicity, because there is only one single halo in XRD pattern, which
disappears upon calcination with the simultaneous growth of the broad (001)
reflection of nanocrystalline Mo5O14 (Fig. 5.4 b).
In case of nanocrystallinity, there should be several very broad peaks, instead of one
single halo, whose intensities grow with raising the calcination temperature due to
the increasing degree of sample crystallinity. Hence, it seems likely that the structure
of the spray-dried sample is characterised by some structural ordering in the layer
plane but does not possess any periodicity in the perpendicular direction, e.g. along
[001].
Moreover, the gradual increase of the intensity of the (001) reflection during the
crystallization of the Mo5O14-type structure from an amorphous, spray-dried
precursor indicates a lamellar, or pseudolamellar character of this mixed oxide
(pattern b, c, d of Fig. 5.4). When the layers begin to pack into slabs, the 3D
periodicity of the MoVW oxide becomes evident. That is also demonstrated by the
increasing intensities of the basal reflections. The proposed structural model fits well
with Raman spectroscopic data, which also indicates the presence of nuclei of the
Mo5O14 phase in the spray-dried material, and with TEM observations, which
confirm the pseudolamellar character of the Mo5O14 structure (vide infra).
Attention should also be focused on the fact that the crystallization process of the
Mo-V-W oxide from the amorphous precursor is governed by a proper maintenance
of the cation oxidation states during the thermal treatment steps to the final catalyst.
According to the XRD analysis, calcination of the spray-dried precursor in an
oxidizing medium, i.e. air, leads to the partial disintegration of the complex oxide
phase of the Mo5O14-type as proven by the reflections of MoO3in the XRD pattern
(Fig. 5.4, inset). Hence, the treatment of the spray-dried precursor under well chosen
116
tempering or reaction conditions is a critical requirement for the synthesis of well-
crystalline, single phase Mo-V-W oxide (Fig. 5.4 c, d, e).
Table 5.2: XRD data
2DInt. (%) Phase
7.77 11.4 32 (MoVW)5O14
8.69 10.17 29.9 (MoVW)5O14
12.30 7.19 10.5 MoO3 (r)
15.6 5.68 8.4 (MoVW)5O14
22.29 4.0 100 (MoVW)5O14
23.13 3.84 55.2 (MoVW)5O14
23.44 3.79 39.0 (MoVW)5O14
24.82 3.58 39.4 (MoVW)5O14
27.28 3.266 36 MoO3 (r)
31.66 2.824 17.1 (MoVW)5O14
33.74 2.66 32.8 MoO3 (r)
38.98 2.31 12.2 (MoVW)5O14
5.3.5 HRTEM
According to TEM, the particles of the spray-dried, and the calcined sample (air, 623
K; helium, 713 K) are of platelet-like shape with sizes of about 200 nm in length and
10 nm in width. With the help of selected area electron diffraction (SAED) patterns,
the structure of the sample was identified as a Mo5O14-type. SAED pattern oriented
along [001] zone axis and given in the inset of Fig. 5.5. reflects the closely packed
character of the Mo5O14 structure in the basal plane. SEAD pattern shown in the inset
of Fig. 5.5 displays a set of strips similar to those characteristic of lamellar materials
Thus, this ternary oxide can be considered as pseudolamellar. These layers are
117
alternating along the [010] direction. It is evident from the TEM observations that the
(010) plane is the most developed plane of the crystalline particle.
At higher magnification, the structure of the sample is rather complicated. A cubic
motif of the dark spots is clearly visible in the HRTEM micrograph of the sample
indicating the Mo5O14-type structure viewed along the [001] orientation (see Fig.
5.5). The distance between two neighbouring spots is equal to 2.25 nm, which fits
well with half of the unit cell parameter of Mo5O14. This value also coincides with
the distance between two channels formed by the pentagonal bipyramides in the
structure of this complex ternary oxide[17,18]. Therefore, it can be suggested that the
micrograph in Fig. 5.5 shows the channels in the basal plane. On the other hand, the
pseudolamellar nature of the crystal structure is evident from the HRTEM image of
the crystal viewed down the [010] projection (Fig. 5.6) from which an interlayer
parameter as long as 1 nm can be measured.
EDX measurements indicate similar elemental compositions for all investigated
particles of the ternary oxide sample. As evident from EDX, the concentrations of
Mo, V, and W are equal to 68.13, 22, and 9.88 at%, respectively. These values are in
a good agreement with the EDX data obtained by SEM analysis.
Figure 5.5: HRTEM image and SAED patterns of Mo5O14
structure viewed
along the [001] direction
118
The XRD and TEM data presented in this work are in a good agreement with earlier
observations on the structural arrangement of the MoVW oxide with an admixture of
MoO3[6,9]. The structural evolution of the main ternary phase during thermal
treatment seems to be similar to the structural transformation of the catalyst in the
present study. The XRD pattern of the precursor described in[6,9,16] shows the same
diffuse peaks, which became sharp and strong due to crystallisation of the initial
structure upon thermal treatment. Note that the (210) and (200) reflections, although
very broad, already exist in the XRD pattern of the precursor, whereas the (001)
reflection is almost absent in the XRD diffractogram. After the first calcination step
of the precursor, the reverse situation was observed. The former reflections
disappear, and the latter start to grow until the sharp and intense peaks appear after
further heating. From the structural point of view, this implies that there is a kind of
short-order and structural periodicity in the [210] direction, but it is very poor along
the [001] direction. This finding is in line with recent TEM data[7,9], which indicated
diffusion reflections in the SAED pattern. Irregular packing of thin layers within the
ternary oxide particles in the direction perpendicular to the (001) basal plane leads to
the so-called bundle structure. The TEM results obtained in this study also show this
pseudolamellar character of the ternary MoVW phase.
Figure 5.6: HRTEM image and SAEDpattern of Mo5O14
structure viewed
along the [010] direction
119
5.3.6 Raman Spectroscopy
Figure 5.7 shows the Raman spectrum of an aqueous mixed solution of the three
transition metal compounds and the spectra of individual AHM, AMT and vanadyl
oxalate solutions for comparison. The spectrum (a of Fig. 5.7) of the pure, colourless
AHM (c=0.96 mol/l Mo) solution displays bands at 940, 894, 818, and 700 cm-1 in
agreement with literature[19].
The Raman spectrum of the pure, blue vanadyl oxalate solution (c = 0.76 mol/l V)
displays bands at 974, 908 and 678 cm-1 (b of Fig. 5.7), while a band is detected at
971 cm-1 for the pure, colorless AMT (c = 0.27 mol/l W) (c of Fig. 5.7). The
spectrum d of Fig. 5.7 was recorded for the freshly mixed solution of the precursor
solutions.
1100 1000 900 800 700 600
0,0
0,5
1,0
1,5
C
A
B
D
E
F
Norm. Intensity
Raman Shift [cm-1]
Figure 5.7: Raman spectrum normalised to the intensity of the band at 944
cm-1 of A the ammonium heptamolybdate solution; B
vanadyl oxalalte
solution, Cthe ammonium meta tungstate solution, D
the mixed solution (0,53
mol/l Mo; 0,18 mol/l V; 0.07 mol/l W) E
the sum as calculated from the
respective weighted spectra of the pure starting compounds and fitt
ed to the
experimental spectrum, F t
he mixed solution (of D) after heat treatment at
353 K for 1h
120
Raman bands of the fresh mixed solutions are observed at 964, 944, 910, 821, 790,
711 and 682 cm-1, which are in the typical regime for Raman bands of polyoxo
metallates. These bands tentatively could be assigned to terminal M=O vibrations in
the regime between 1000 and 890 cm-1 and to M-O-M bridges in the regime between
890 and 600 cm-1.
The calculation of the theoretical spectrum of a hypothetical, mixed solution, which
was fitted to the experimental data (spectrum e of Fig. 5.7) shows bands at 973, 940,
894, 817, 701, and 678 cm-1. Note that the band positions and observed relative
intensities of the different signals vary from those of the freshly mixed solution. This
hypothetical sum spectrum is clearly different from that of the freshly mixed
solution. Hence, it can be assumed that already mixing the three starting solutions
induces a chemical reaction towards a polyoxo compound on Mo basis with
incorporated V and W.
Figure 5.7 F shows the spectra of a freshly mixed solution, after additional heat
treatment at 353 K for 1 hour. The Raman spectrum of this solution (spectrum b of
Fig. 5.7) also differs decisively from that of the freshly mixed solution (spectrum e of
Fig. 5.8). A new intense band appears at 878 cm-1 together with bands at 697 and 683
cm-1. These bands appear in the typical regime of frequencies of Me-O-Me bridge
stretching modes. The observation of these bands, thus, points to the formation of
polymeric aggregates after this additional heating presumably containing W and V
due to the absence of their respective Raman bands in the spectrum.
The Raman bands, detected at 938 and 872 cm-1 in the solution Raman spectrum
(Fig. 5.7, F), are also observed in the Raman spectrum of the spray-dried solid
(spectrum a of Fig. 5.8) together with a broad wing extending to lower wavenumbers
and a very weak band at 682 cm-1. Simultaneously, all other bands disappear which
were recorded for the solution.
Spectral alterations are evident as compared to the solution Raman spectra (Fig. 5.7,
5.8). By comparison with Raman spectra reported for Mo0.68V0.23W0.09 mixed metal
oxide catalysts[6,9], the band observed at 872 cm-1 after this treatment step is
tentatively attributed to a mixed oligomeric species with a molecular structure, which
might already contain the structural motif of Mo5O14, i.e. the central pentagonal
bipyramid surrounded by five octahedra[14]. In this context, is has to be mentioned
121
that similar Raman spectral bands were reported for the Keplerates also containing
the pentagonal bipyramids as a structural motif[20,21].
Thus, the combined information from XRD and Raman spectroscopy indicate the
presence of a molecular MoVW precursor species and the absence of long-range
order in the spray-dried solid precursor. After further drying at 383 K overnight, the
Raman spectrum (spectrum b of Fig. 5.7) of the sample has not changed.
The relative intensity of the band at 866 cm-1 attributed to the polyoxo (MoVW)5O14-
precursor however has increased somewhat as compared to the band at the same
wavenumber of the spectrum recorded in solution. A comparison with the XRD
results (Fig. 5.4) clarifies that the X-ray amorphous spray-dried precursor consists of
these polyoxo clusters.
Calcination of the sample in air at 623 K for 2h resulted in a Raman spectrum
(spectrum c of Fig. 5.8) which has considerably altered as compared to the spectra of
the dried materials (spectra a of Fig. 5.8).
1000 800 600
a
b
c
d
e
690
994
985
818
682
939
872
935
866
820
662
928
866
818
682
Intensity [a.u.]
Raman shift
Figure 5.8: Raman spectra of the MoVWO oxide precursor/catalysts: a) spray-
dried, b) dried in air at 383 K overnight, c) calcined in air at 623 K for 2h, d
)
calcined in air at 623 K for 2h and in helium at 713 K for 2h and e
) heated in air at
723 K for 2h
122
The formerly prominent bands between 800 and 1000 cm-1 have changed in shape
and have lost their structure. The band at 662 cm-1 assigned to Me-O-Me bridges
shows a higher relative intensity as compared to the bands between 800 and 1000
cm-1.
E
MoO
3
, V
2
O
5
WO
3
Spray drying
Calcination under low p (O
2
)
Calcination under high p (O
2
)
a
d
c
b
Figure 5.9: a) Oligo anions in solution b) Material after spray–
drying (with
oxygen defects) c) Material calcined
under low oxygen partial pressure where
a variety of defects can be obtained d) Binary oxide materials calcined under
higher oxygen partial pressure (i.e. air)
123
This increased intensity might indicate an increased degree of cross linking in the
solid. However, the Raman spectrum is still poorly resolved and is additionally
characterised by a high background extending to lower frequencies, which points to
an ill-defined, poorly crystallised material, in line with the XRD results (see Fig.
5.4).
The Raman spectrum d of Fig. 5.8 was recorded of the catalyst additionally tempered
in He at 713 K for 2h. The spectrum is more resolved compared to the spectrum of
the preceding step and shows bands and shoulders at 985, 949, 866, 818, 682, and
584 cm-1.
The Raman bands are more resolved and the background is reduced compared to the
mixed oxide after thermal treatment at 623 K in air. The observed band positions are
in agreement with those reported by Dieterle at al.[9] and Blume[13] for the catalyst
prepared by other methods and a different transition metal ratio. These spectral
alterations indicate an increase in crystallinity of the mixed oxide after this treatment
step that fits well with the XRD results, which proved the crystallisation of the
Mo5O14-type mixed oxide.
When the spray-dried sample was additionally calcined at 723 K for 2 h, Raman
shows bands or shoulders at 995, 875, 845, 818, and 667 cm-1 (spectrum f of Fig.
5.8). The Raman band at 891 cm-1 can be attributed to monoclinic Mo4O11[22],
whereas the band at 845 cm-1 may be assigned to orthorhombic Mo5O14, which is
characterised by Raman bands at 983, 909, 845, 790 and 728 cm-1 [13]. But XRD did
not reveal the presence of Mo4O11 in the sample [see Table 5.2] that is likely to be
caused by its XRD amorphous state. Obviously, calcination in air at temperatures
below 673 K (data not shown) leads to the formation of (MoVW)5O14 in phase
mixtures with traces of MoO3. Temperatures higher than 673 K result in the
decomposition the majority (MoVW)5O14 phase into a broad phase mixture.
5.3.7 Catalytic properties
The catalytic performances of the different, above described materials were studied
in the acrolein oxidation under identical reaction conditions. Fig. 5.10 shows the
acrolein conversions at 563 K and at = 0.12 g×s/ml as a function of time on stream.
The selectivities to acrylic acid and CO2+ CO with time on stream are displayed in
Fig. 5.11. It is evident from Fig. 5.10, 5.11 that the precursor pre-treatment in air, or
in air plus He considerably affected the final catalytic properties.
124
The catalyst, only calcined in air at 623 K for 2 hours, shows an activation period
similar to that observed for an industrial MoVWO catalyst[7]. It is important to note
that a significant change in catalytic performance was not observed during the first
minutes of operation, which could be related to surface processes, like desorption of
water or surface reduction.
Such processes can be excluded to be responsible for the improving catalyst
activities. Moreover, the BET surface areas of all materials were very low with about
4 m2/g and did not change upon the different calcination treatments or catalysis.
Textural changes, thus, can be excluded to be responsible for the changing catalytic
activity. As suggested in the previous work[7], such an activation period is related to
the formation of an active crystalline phase. In full agreement and confirming
previous results, XRD of the catalyst after operation in the acrolein oxidation for 80
h revealed the exclusive presence of crystalline (MoVW)5O14 in this sample
(diffractogram e of Fig. 5.4).
The selectivity to acrylic acid over the catalyst calcined in air also slightly increased
Figure 5.10: Acrolein conversion over the Mo9V3W1.2O14
catalyst at 563 K and
as a function of operation time on stream. Spray dried sample afte
r
calcination in air at 623 K for 2 hours (black), spray dried sample after
calcination in air at 623 K for 2 hours and subsequent treatment in He at 713
K for 2 hours (red). Reaction mixture composition: 4 % C3H4O, 8 % O2
, 20
% H2O, balance – He
0 5 10 15 60 70 80
0
5
10
15
20
25
30
35
Acrolein Conversion, %
Time on Stream, hours
125
with time on stream and reached 97% at an acrolein conversion of 25-26%, while
that to CO + CO2decreased (Fig. 5.8). This slight change in the selectivities over the
first 1.5 hours might be related to red ox processes within the catalyst material[6].
However, it has to be stated that the time scale of the selectivity changes does not
correlate with that of the increasing catalyst activity. Hence, this result once more
proves that this increasing catalyst activity is not directly related to red ox processes
in the material, which should be reflected by selectivity alterations.
The steady state selectivity to acrylic acid of 97 % finally reached is close to that of
the industrial catalyst[7]. The catalyst, which was additionally treated in He at 713 K
for 2 hours, only reached a 6-7 times lower acrolein conversion under identical
reaction conditions. Its selectivity to acrylic acid was also lower with 86%.
It is important to emphasise that treatments, which lead to the decomposition of the
Mo5O14–type structure, were not applied in the present catalytic tests. Previous
investigations have proven[9] that the Mo5O14-type structure is stable up to 818 K in
Figure 5.11: Selectivities to acrylic acid (triangles) and CO + CO2
(squares):
spray dried sample after calcinatio
n in air at 623 K for 2 hours (open
triangles and squares), spray dried sample after calcination in air at 623 K for
2 hours and subsequent treatment in He at 713 K for 2 hours (filled triangles
and squares).
0 1 2 3 4 70 75 80 85 90
0
20
40
60
80
100
Selectivities, %
Time on Stream, hours
126
He. Thermal treatments at temperatures above 829 K led to the decomposition of the
metastable Mo5O14 phase to stable MoO3and MoO2[9]. Additionally, strong oxidative
or strong reductive conditions were not used in the present study, which also result in
the formation of phases with a higher or with a lower degree of reduction, such as
MoO3or Mo2O5-x[6,9,23].
XRD analysis on the sample with the high catalytic activity (sample heated in air at
623 K for 2 h) revealed that a crystalline Mo5O14–structure appears after operation in
the acrolein oxidation reaction (Fig. 5.4 e). As was shown previously the structure of
the sample before reaction represents poorly crystalline (almost X-ray amorphous)
material.
It seems likely that the induction period, which was observed for this sample is
caused by crystallisation of the Mo5O14 phase during the reaction. Note that the
catalyst heated in air and in helium and composed of single crystalline Mo5O14–type
structure shows low activity and less selectivity for acrolein in spite of identical XRD
patterns of the activated material.
This phenomenon might be associated with different `real structures´ of the
compound under reaction conditions e.g. sise of the Mo5O14 particles, degree of
nanocrystallinity, different oxidation states of molybdenum, vanadium and tungsten.
Further, the character of the structural disordering during the reaction might also be a
reason. This result shows the full complexity of the system by demonstrating that the
long range structure of a catalyst is necessary but by no means sufficient information
to obtain a structure–function relationship. New ways in catalysis research are
therefore recommended that investigate in detail the real structure of the material in
situ under reactive conditions.
5.4 CONCLUSION
A synthesis procedure has been developed in this work, which allowed the
preparation of the single phase, well-crystallised Mo5O14-type oxide. The whole
process of structure formation starting from the precursor solutions up to the final
product has been closely monitored. A strong interaction of AHM, AMT and vanadyl
ions involving the formation of Me-O-Me bridges and a mixed Mo, V, W compound
was observed with a structure already being closely related to the Mo5O14-type. This
127
structural motif was maintained during the spray-drying process but the material
showed a low crystallinity. Further thermal treatment removed ammonia, water and
oxalate and increased the state of crystallinity. Whilst calcinations in air led to over
oxidation with a phase mixture amongst a well crystallised Mo5O14-type structure,
the pure single phase was only obtained by additional treatment in helium.
Particles of this Mo5O14-type compound are oriented along the [001] and the [010]
zone axes. Micro diffraction patterns show the closely packed character of the
Mo5O14-type structure in the basal plane. This ternary oxide is pseudolamellar with
alternating layers along the [010] direction and a strongly developed (010) face.
Structure refinement showed decreased a and blattice parameters with respect to
Kihlborg et al.[10], but an increased cparameter. This effect is probably due to
incorporation of V and W into the lattice. This statement is also supported by Raman
spectral features and shows the importance of V and W as structural promoters to
maintain the Mo5O14-type phase presumably by stabilizing the pentagonal
bipyramides.
Whilst the catalyst that formed the Mo5O14 structure after an induction period of
about 2.5 h achieved the highest catalytic activity for acrolein oxidation, the
synthesised single-phase material showed less activity but its selectivity was almost
as high as the industrial material. Some important conclusions can be drawn from
this. Apparently the active material is single-phase but distinctively different from
the perfect structure of Mo5O14. The Mo5O14 structure is an idealised endpoint that is
formed under reduced oxygen partial pressure by the organization process of a
mixture of oligo anions, which are generated in solution. The active phase is
metastable until crystallisation and oxidative decomposition into binary oxide phases
occurs under high oxygen partial pressure (air and above). This is illustrated in Fig.
5.9.
These experimental results are highly significant for future catalysis research,
because they confirm that a single-phase material can be catalytically active, and that
no phase cooperation or spill over phenomena are functionally essential.
Additionally, this work has highlighted the full complexity of these systems, because
x-ray powder diffractometry indicates the same crystallographic phase for two
differently active materials. This difference can only be explained by different `real´
structures of the two materials. It must be strongly emphasised that precise analytical
128
methods need to be developed, that look into the details of a real structure of a given
phase in order to determine structure–function relationships. The synthesis procedure
reported in this paper is an example for this approach.
129
Reference List
[1.] A. N. Kurtz, R. W. Cunningdam, A. W. Naumann (Union Carbide Co.) US
4111983 (1978).
[2.] V. Novak, L. Sokol, J. Jelinek, CS 1207807 B(1981)
[3.] N. Bertolini, S. Ferlazzo (Euteco Impianti S.p.A.) US 4289654 (1981)
[4.] T. Kawajiri, S. Uchida, H. Hironaka (Nippon Shokubai Kagaku) EP 427 508
A1 (1991)
[5.] A. Tenten, F.-G. Martin, H. Hibst, L. Marosi, V. Kohl (BASF AG), EP
668104 B1 (1995)
[6.] G. Mestl, Ch. Linsmeier, R. Gottschall, M. Dieterle, J. Find, D. Herein, J.
Jäger, Y. Uchida, and R. Schlögl, J. Mol. Catal. A, 162 (2000) 455-484.
[7.] O. Ovsitser, Y. Uchida, G. Mestl, G. Weinberg, A. Blume, M. Dieterle, H.
Hibst, and R. Schlögl, J. Mol. Catal. A, 185, 291-303, (2002).
[8.] M. Dieterle, PhD thesis, TU Berlin, 2001.
[9.] M. Dieterle, G. Mestl, J. Jäger, Y. Uchida and R. Schlögl, J. Mol. Catal. A,
174 (2001) 169-185.
[10.] L. Kihlborg, Ark Kemi, 21 (1963) 427.
[11.] T. Ekström, M. Nygren, Acta Chem. Scand., 26, (1972), 5, 1827-1835.
[12.] T. Ekström, M. Nygren, Acta Chem. Scand., 26, (1972), 5, 1836-1842.
[13.] A. Blume, PhD thesis, Berlin, 2002.
[14.] G. Mestl, J. Raman Spectrosc., 33, 335-347, (2002).
[15.] L.P. Solovyeva, S.V. Tsybulya, G.N. Kryukova, E.M. Moroz, JSTCAM 32
(1991) 18.
[16.] H. Werner, O. Timpe, D. Herein, Y. Uchida, N. Pfaender, U. Wild, R.
Schlögl, and H. Hibst, Catal. Lett., 44 (1997) 153.
130
[17.] Kihlborg, L., Acta Chem. Scand., 26, (1972), 3381.
[18.] W. Ueda, K. Oshihiara, Appl. Catal. A, 200 (2000) 135-143.
[19.] G. Mestl, T.K.K. Srinivasan, Catal. Rev.-Sci. Eng., 40 (4), 451-570 (1998).
[20.] A. Müller, S. Polarz, S.K. Das, E. Krickemeyer, H. Bögge, M. Schmidtmann,
B. Hauptfleisch, Angew. Chem., 111, 3493-3443 (1999).
[21.] A. Müller, S. Sarkar, S.Q.N. Shah, H. Bögge, M. Schmidtmann, S. Sarkar, P.
Kögerler, B. Hauptfleisch, A.X. Trautwein, V. Schünemann, Angew. Chem.,
111, 3435-3439 (1999).
[22.] A.A. Bolznan, B.J. Kenned, C.J. Howard, Austr. J. Chem., 48, 683 (1995).
[23.] Y. Uchida, G. Mestl, O. Ovsitser, J. Jäger, A. Blume and R. Schlögl, J. Mol.
Catal. A., 187, 247-257, (2002)
131
6 Conclusion and Outlook
6.1 STRUCTURE -ACTIVITY RELATION
Mo Oxide model catalysts with a variety of structural and chemical complexity have
been presented. A tremendous structural diversity was shown on binary molybdenum
and oxygen containing systems. Additionally various degrees of crystallinity and
particles sizes occurred as smaller particles do not necessarily show the same
crystalline structure as the bigger particles.
A number of nanostructured molybdenum oxides with high structural complexity has
been successfully prepared. Depending on the preparation conditions trimolybdate,
hexagonal MoO3, orthorhombic MoO3and a supramolecular compound was
prepared. Compounds showing a very similar crystal structure but a different degree
of protonation showed a decisively different catalytic behaviour.
No material is active as precipitated. Transformation into metastable active phases
occurred at about 300 °C, as seen from the TPRS experiments. Above 450 °C phase
transformation into orthorhombic MoO3significantly reduced the catalytic activity,
as shown in the second cycle of the TPRS run. Whilst it is often reported that
additives increase the catalytic activity, it is more likely that they merely stabilize the
metastable Mo-O phase.
Many attempts have been carried out in the literature to develop structure activity
relations. This work has demonstrated that such relations by integrative methods as
XRD or Raman have to be treated with extreme caution, as the existence of many
differently shaped and structured particles were only observed with HRTEM. Only a
minor amount of the mass balance exposed to the surface is responsible for catalysis.
The relevance of this effect for catalysis becomes obvious by comparing the small
scale samples with the large scale samples. The latter ones consist of larger, more
homogeneous particles. Their catalytic behaviour especially the deactivation is much
more rapid as with the small scale samples.
132
The TPRS experiments show clearly that even a chemically simple system can be
tuned by controlling its thermodynamic parameters. This opens two new pathways.
Firstly, catalytic reactions can be studied and understood on comparatively simple
model catalysts. Secondly, catalysts can be optimised by varying and controlling
thermodynamic parameters during the preparation process and not only by additives
whose role often is obscure.
A higher level of complexity is reached by addition of vanadium and tungsten into
the system. However, the model character still prevails because in contrast to real
multi element industrial systems only one crystalline phase (Mo5O14) is observed.
The importance of the real structure of a catalyst including its amorphous
components is nicely demonstrated in the work concerning the Mo5O14 system. The
catalyst which was calcined in air at 350 °C showed the highest activity for the
acrolein partial oxidation after an induction period of about 2.5 h. The post mortem
sample showed the Mo5O14 structure. The synthesised single phase material is
behaving differently. The activity is much lower but selectivity is close to the
industrial material.
Some important conclusions can be drawn from this. Apparently the active material
is a single crystalline phase but distinctively different from the perfect structure of
Mo5O14. The Mo5O14 structure is an idealised endpoint that is formed under reduced
oxygen partial pressure by the organization process of a mixture of oligo anions,
which are generated in solution. The active phase is metastable until crystallisation
and oxidative decomposition into binary oxide phases occurs under high oxygen
partial pressure.
6.2 PREPARATION AND REACTIONS IN SOLUTION
Often in the catalysis literature the precipitation process carried out in water is
neglected. In this thesis the aqueous chemistry and structure formation was studied
carefully. In addition to the potentiometric measurements in situ Raman spectroscopy
was performed.
The in situ Raman data on the molybdenum-only model systems presented in this
work has clearly demonstrated the severe temperature effect on the nature of
molybdate species in solution. Also the pH traces are very sensitive to a change in
133
reaction mechanism in solution. Especially a shift in the equilibrium between and
octamolybdate is the important factor influencing the nature of the final precipitate
and therefore of the catalyst precursor. Further in this context different activities and
consequently solubility products need to be mentioned. This is the reason why also
the normalised pH traces are different, especially when cations are changed.
The first model systems are vital for understanding the genesis of the second system
(multicomponent). The molybdenum chemistry as a function of pH and temperature
is very similar. At the octamolybdate equilibrium the added vanadyl is able to link
the prevailing elements and a different structure namely Mo5O14 is generated. It is
noteworthy that pentagonal bipyramids appear in the supramolecular compound as a
structural motif as well as in Mo5O14.
The whole process of structure formation starting from the precursor solutions up to
the final product has been closely monitored by using UV/Vis, ESR, Raman and
95Mo NMR spectroscopy. The results obtained show that structure formation starts as
early as during the mixing process of the initial precursor solutions. Addition of
acidic vanadyl oxalate causes protonation and condensation of the heptamolybdate
ions into octamolybdate ions that are linked by vanadyl species and form a polymeric
network. Addition of tungsten atoms clearly enhance this process. This polymeric
network is an essential precursor for the final Mo5O14–type structure. In this case
precipitation is inconvenient to obtain the solid catalyst precursor. Therefore spray-
drying was used to conserve the structural motif.
6.3 OUTLOOK
In the future the interplay between model development and complex real system can
be applied to even more challenging systems containing more additives such as
MoVTeNbOx. The key step identified in the model system, namely the isomerisation
reaction between and octamolybdate will be decisive in the MoVTeNbOx
preparation procedure or any other complex catalytic system containing
molybdenum.
During this isomerisation process the system is most reactive as shown by in situ
Raman spectroscopy in the model case. Therefore the structure formation will
depend strongly on the treatment during this stage e.g. on the temperature or on the
134
element added at the time. Very likely one step will determine the following.
Consequently the mixing sequence will play a role.
In this context another function of the additives can be discussed. One or more
components might act as a transporting reagent for gas phase vapour deposition. This
might be the key step for achieving high activity. As discussed in the Mo5O14-case
the crystalline phase might just be the support for an either ill defined amorphous or
highly dispersed material. Again, this hypothesis can only be confirmed by a well
investigated model system, because only in this case effects from the additives can
be separated from the effects from the support.
Table of Figures:
Figure 2.1: Lead structures of the obtained materials 15
Figure 2.2: pH curve for LiMoO4samples 16
Figure 2.3: TG-DSC of 222, 230 and 243 17
Figure 2.4: A XRD pattern of 256, 227, 229, 243 19
Figure 2.4: B XRD pattern of 222, 245 19
Figure 2.5: Raman spectra of 256, 227, 229, 245, 243, 222 20
Figure 2.6: TG-DSC of 219, 229 and 227 21
Figure 2.7: UV/Vis of 251, 222, 230 22
Figure 2.8: pH curve for 252, 226, 231 23
Figure 2.9: DRIFTS of 256, 227, 229, 245, 243, 222 24
Figure 2.10: DRIFTS of 256, 227, 229, 245, 243, 222 25
Figure 2.11: Electron Microscopy 26
Figure 2. 12: TG-DSC of 226, 231, 232 27
Figure 2.13: NIR bands of 252, 245, 233, 244, 251 28
Figure 2.14: pH trace of 256, 227, 232 29
Figure 2.15: Electron micrographs, 74 30
Figure 2.16: TG-DSC of 256, 228 31
Figure 2.17: TG-DSC of 250, 245, 249 32
Figure 2.18: pH-trace 249, 219, 229; 253, 246, 243; 250, 245, 244 33
Figure 2.19: pH trace, 227, 228, 225 34
Figure 2.20: TPRS first cycle; effect of counter ions, acrolein 35
Figure 2.21: TPRS first cycle; effect of counter ions, O2-conversion 36
Figure 2.22: TPRS run first cycle; AHM starting material, O2-conversion 37
Figure 2.23: TPRS run first cycle AHM starting material, acrolein 39
Figure 2.24: TPRS large scale sample versus small scale sample 40
Figure 2.25: NIR of 251, 222, 230 41
Figure 2.26: Raman of TG post mortem samples at characteristic signals 42
Figure 2.27: Comparison electron diffraction versus X-ray diffraction 48
Figure 2.28: Hexagonal MoO3, Assignment of Raman and IR modes 50
Figure 2.29: Orthorhombic MoO3, Assignment of Raman and IR modes 51
Figure 3.1: In situ Raman investigation at 30 °C 77
Figure 3.2: In situ Raman investigation at 50 °C 78
Figure 3.3: In situ Raman investigation at 70 °C 79
Figure 3.4: pH-trace; interpretation of equivalence points 81
Figure 3.5: Dominant species in solution 83
Figure 4.1: UV/Vis spectra of AHM and AHM+AMT solution 89
Figure 4.2: UV/Vis spectra of AHM, AMT, Vanadyl oxalate 91
Figure 4.3: UV/Vis spectra d-d transition region 92
Figure 4.4: In situ UV/Vis characterization 93
Figure 4.5: Conductivity against pH 94
Figure 4.6: Mo NMR spectra 95
Figure 4.7: ESR spectra of the binary solutions 96
Figure 4.8: Connection between the octamolybdate units 98
Figure 4.9: Three dimensional connections of octamolybdate units 100
Figure 4.10: Reaction steps 101
Figure 5.1: Typical SEM images 110
Figure 5.2: TA in flowing synthetic air 111
Figure 5.3: TA in helium 112
Figure 5.4: X-ray pattern of the MoVW oxide precursor/catalysts 113
Figure 5.5: TEM image and SAED pattern of Mo5O14 [001] 117
Figure 5.6: TEM image and SAEDpattern of Mo5O14 [010] 118
Figure 5.7: Raman spectrum normalised to the intensity 119
Figure 5.8: Raman spectra of the MoVWO oxide precursor/catalysts 121
Figure 5.9: Oligo anions in solution 122
Figure 5.10: Acrolein conversion over the Mo9V3W1.2O14 catalyst 124
Figure 5.11: Selectivities to acrylic acid 125
Table of Tables:
Table 2.1: Precipitation conditions 54
Table 2.2 Scale up 55
Table 2.3 BET surface areas 55
Table 2.4 UV/Vis data 56
Table 2.5 NIR data 57
Table 2.6 Major Raman band positions and intensities 57
Table 2.7 XRD data 59
Table 5.1 EDX Result 109
Table 5.2 XRD data 116
Stefan Knobl
Persönliche
Information Eltern: Katharina Knobl, geb. Bäuerlein,
Wilhelm Knobl
Familienstand: ledig
Staatsangehörigkeit: deutsch
Geburtsdatum: 23.09.1973
Geburtsort: Selb
Ausbildung 1980-1984: Dr. Franz-Bogner-Schule, Selb.
1985-1993: Abitur, Gymnasium Selb.
1993-1994: Zivildienst, Zentrale Diakonie-Station, Selb.
1996: Diplomvorprüfung, Universität Tübingen.
1996-1997: University Of Wales, Bangor.
2001: Diplom-Chemiker, Universität Tübingen.
2001-2004: Doktorand, Fritz-Haber-Institut der Max-Planck-
Gesellschaft, Berlin.
Praktika 1997: Bayer AG Leverkusen.
Tätigkeiten neben dem
Studium
Mitglied der Vertreterversammlung des Studentenwerks
Tübingen.
Mitglied des Verwaltungsrates des Studentenwerks Tübingen.
Mitglied der Studienkommission des Fachbereichs Chemie.
