Inorganic Microporous Membranes for Gas Separation in Fossil Fuel ...

Mitglied der Helmholtz-Gemeinschaft
Inorganic Microporous Membranes for Gas Separation in
Fossil Fuel Power Plants
George van der Donk

Schriften des Forschungszentrums Jülich
Reihe Energie & Umwelt / Energy & Environment Band / Volume 8

Forschungszentrum Jülich GmbH
Institut für Energieforschung (IEF)
Werkstoffsynthese und Herstellungsverfahren (IEF-1)
Inorganic Microporous Membranes
for Gas Separation in
Fossil Fuel Power Plants
George van der Donk
Schriften des Forschungszentrums Jülich
Reihe Energie & Umwelt / Energy & Environment Band / Volume 8
ISSN 1866-1793 ISBN 978-3-89336-525-8

Bibliographic information published by the Deutsche Nationalbibliothek.
The Deutsche Nationalbibliothek lists this publication in the Deutsche
Nationalbibliografie; detailed bibliographic data are available in the
Internet at http://dnb.d-nb.de.
Publisher Forschungszentrum Jülich GmbH
and Distributor: Zentralbibliothek, Verlag
D-52425 Jülich
phone:+49 2461 61-5368 · fax:+49 2461 61-6103
e-mail: zb-publikation@fz-juelich.de
Internet: http://www.fz-juelich.de/zb
Cover Design: Grafische Medien, Forschungszentrum Jülich GmbH
Printer: Grafische Medien, Forschungszentrum Jülich GmbH
Copyright: Forschungszentrum Jülich 2008
Schriften des Forschungszentrums Jülich
Reihe Energie & Umwelt / Energy & Environment Band / Volume 8
D 294 (Diss., Bochum, Univ., 2007)
ISSN 1866-1793
ISBN 978-3-89336-525-8
The complete volume is freely available on the Internet on the Jülicher Open Access Server
(JUWEL) at http://www.fz-juelich.de/zb/juwel
Neither this book nor any part may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, microfilming, and recording, or by any
information storage and retrieval system, without permission in writing from the publisher.

List of Abbreviations
Ac Acetyl Acetonate
ADA 1-amino Adamantine
AKP30 Alumina powder from the Japanese company Sumitomo
ATSB Aluminum-Tri-Sec-Butoxide
BCY BaCe0.8Y0.2O3-δ
BET Brunauer, Emmet and Teller
BJH Barret, Joyner and Halenda
BM-1 Ball Milled sample 1
C10(-NH2) Decylamine
C16(-NH2) Cetylamine
Ca Caproic acid
CCS Carbon Capture and Storage
CTA+ Cetyltrimethylammonium
CVD Chemical Vapour Deposition
D1H Dodecasil 1h zeolite type
DDR Deca-dodecasil 3R
DEA Diethanolamine
DIPA Diisopropanolamine
dkin
Kinetic diameter
dlayer
Layer thickness
DLS Dynamic Light Scattering
DOH Dodecasil 1h code by International Zeolite Association
dpore
Pore diameter
DTA Differential Thermal Analysis
EDX Energy-dispersive X-ray analysis
EN Ethylenediamine
EO Ethylene oxide
ETS-4 Engelhard Titano-Silicates zeolite type
Exo Exothermic directing in Differential Thermal Analysis plot
F127 Pluronic F127
FAU Faujasite zeolite type
FER Ferrierite zeolite type
HIP(-1) Hot Isostatic Pressure (sample 1)
HK-SF Horvath-Kawazoe modified by Saito-Foley
HR-TEM High Resolution Transmission Electron Microscopy
ICP-OES Inductively Coupled Plasma - Atomic Emission Spectrometer
IPCC Intergovernmental Panel on Climate Change
ISO International Organization for Standardization
i

IUPAC International Union of Pure and Applied Chemistry
IZA International Zeolite Association
LTA Linde zeolite Type A
MEP Melanophlogite Zeolite type
MFI Mobile Five
MIEC Mixed ionic electronic conductor
MOR Mordenite zeolite type
MTN Dodecasil-3C Zeolite type
NON Zeolite type
PO Propylene oxide
ppm Parts per million
PTFE Polytetrafluoroethylene (Teflon)
PVA Poly(vinyl-alcohol)
RPM Rotations per minute
SAXS Small Angle X-ray Scattering
SBET
Specific surface area by Brunauer, Emmet and Teller
SDA’s Structure Directing Agents
SEM Scanning Electron Microscopy
SGT Sigma 2 zeolite type
SIMS Secondary Ion Mass Spectroscopy
SOD Sodalite zeolite type
TEOS Tetraethyloxysilane
TGA Thermogravimetric analysis
Ti-1 Titanium doped dodecasil 1H sample 1
TMOS Tetramethyloxysilane
XPS X-ray Photoelectron Spectroscopy
XRD X-ray Diffraction
YSZ Yttria Stabilised Zirconia
ZSM-5 Zeolite Socony Mobile no. 5 synthesised by (Exxon) Mobile
ii

Summary
CO2 capture and storage might have an important role in stabilising the global concentration
of CO2. Power plants are primary candidates for CO2 capture and storage because they have
great potential (up to 45% of CO2 emission reduction in the year 2050 compared to 2005).
Inorganic membranes (Zeolites or TiO2-ZrO2) are candidates for separating H2 from CO2
(precombustion) or CO2 from the flue gas (mainly N2) at the End-of-Pipe (postcombustion) in
fossil fuel power plants.
All-silica Dodecasil 1H (DOH) zeolite type was selected for the hydrothermal stability
and possible ability to separate H2 from other gasses under precombustion concept conditions.
The DOH crystal size can be synthesised to thin hexagonal plates with sizes in the order of 10
µm. This crystal size is too large for membrane formation or to act as seeds for the layer
formation, by means of secondary growth of DOH nuclei.
The removal of the complete structure directing agent (SDA) content from the pores in the
DOH structure could not be obtained through calcination in air for extended periods at
elevated temperatures. Quasi SDA-free DOH, with a high crystallinity, was obtained after
calcination at 900ºC for 5 hours when the atmospheric pressure was increased twice to 50
MPa for 30 min.
The prepared all-silica DOH, that is quasi SDA-free, might present hydrothermal stable
microporous material with pores that are inaccessible for CO2 and accessible to H2.
Polymeric Y2O3 or TiO2 mixed ZrO2 sols are synthesised for the preparation of
ultramicroporous powders and thin films on γ-Al2O3 intermediate layers supported by α-
Al2O3 disks as potential gas separation membranes. Two routes have been selected being the
Ketone- and the Amine-approach based on the precursor modifiers.
8 mol% yttria stabilised zirconia (8YSZ) calcined at 450ºC is microporous with a BET
specific surface area of ~50 m 2 /g. 8YSZ layers might form microporous layers with low
permeability. 30-50 nm thin cubic 8YSZ films, prepared by the Ketone-approach, show He
and N2 transport by Knudsen diffusion due to defects or to the too large pores in the final
membrane layer.
As expected from the linear polymeric Amine-Sols, the amorphous binary TiO2-ZrO2
materials are microporous between 400 and 500ºC. The highest BET specific surface area of
~200 m 2 /g with an estimated pore size of ~1.0 nm (gas physisorption) is obtained for the
Ti0.5Zr0.5O2 calcined at 500ºC using the Amine-approach. The crystallisation temperature of
orthorhombic Ti0.5Zr0.5O2 is between 550 and 600ºC which is ~250ºC higher than that of
single oxides.
20-60 nm thin, homogeneous films can be prepared from TiO2, ZrO2 and binary oxides on γ-
Al2O3 membranes using calcination temperatures in the range of 400 to 600ºC. H2/CO2
permselectivity higher than the Knudsen factor is observed for Ti0.5Zr0.5O2 films calcined at
500 and 600ºC. These films can contain a pore size distribution of ~0.3 to ~0.5 nm in
diameter which is comparable to state of the art SiO2 membranes. The maximal He permeance
of 1·10 -7 mol/m 2 sPa and a maximal permselectivity of He/N2 =14 or H2/CO2=6 values are
lower than state of the art SiO2 membranes but higher than TiO2-ZrO2 membranes published
iii

in literature. These results fulfil the primary aim of achieving gas separation membranes from
TiO2-ZrO2 material. However, the permeability and permselectivity of these membranes
should be increased for applications in power plants.
The micropores in Ti0.5Zr0.5O2 membranes are stable for at least 1500 hours. Steam test results
in reversible pore blocking without layer delamination or destruction while maintaining the
permselectivity. This evidence indicates that Ti0.5Zr0.5O2 membranes can be alternatives to
SiO2 membranes for precombustion applications.
iv

Zusammenfassung
Die Abtrennung und anschließende Speicherung von anthropogen erzeugtem CO2 wird im
Hinblick auf die Stabilisierung der globalen CO2-Konzentration künftig eine bedeutende
Rolle einnehmen. Fossile Kraftwerke bieten dafür ein enormes Potential. Im Vergleich zu
2005 könnten die CO2-Emissionen bis 2050 um 45% gesenkt werden. Anorganische
Membranen (z.B.: aus Zeolithen oder TiO2-ZrO2) eignen sich prinzipiell, H2 von CO2 vor der
Verbrennung (precombustion) oder CO2 vom stickstoffhaltigen Abgas nach der Verbrennung
(postcombustion) im fossilen Kraftwerk abzutrennen.
Der Silika Dodecasil 1H Zeolith (DOH) wurde aufgrund seiner hydrothermalen
Stabilität und seiner Fähigkeit, H2 von anderen Gasen unter den Bedingungen des
Precombustion-Verfahrens zu trennen, ausgewählt. Die DOH-Kristalle konnten als dünne
sechseckige Platten in der Größenordnung von 10 µm synthetisiert werden. Sie waren jedoch
zu groß für die Membranbildung oder um als Kristallkeime für die Schichtbildung durch das
sekundäre Wachstum von DOH-Keimen zu fungieren.
Die komplette Abspaltung des Templates von den Poren der DOH-Struktur konnte durch die
Kalzinierung an Luft bei erhöhten Temperaturen und über längere Zeiträume nicht erreicht
werden. Nahezu template-freie DOH-Kristalle mit einer Käfigbelegung von nur 1% und einer
hohen Kristallinität konnten nach einer Kalzinierung bei 900ºC mit einer Haltezeit von 5
Stunden und einer zweimaligen Druckbehandlung von 50 MPa für 30 Minuten hergestellt
werden.
Nahezu template-freie Silika DOH-Kristalle wurden hergestellt, um hydrothermal stabile und
mikroporöse Materialen zu erhalten, welche Poren aufweisen, die für CO2 unzugänglich sind,
die H2 jedoch passieren kann.
Polymere Y2O3/ZrO2-Sole oder TiO2/ZrO2-Sole wurden für die Herstellung von
ultramikroporösen Pulvern und dünnen Filmen auf γ-Al2O3-Zwischenschichten, die von α-
Al2O3-Membranscheiben getragen werden, als potenzielle Gastrennungsmembranen
synthetisiert. Dazu sind zwei Sol-Gel-Routen verwendet worden: abhängig von den
Additiven, die die Metallalkohole modifizieren, wurde entweder die Keton- oder die Amin-
Route für die Herstellung gewählt.
Aus Yttrium-stabilisiertem Zirkonoxid (8YSZ), das bei 450ºC kalziniert wurde, ist ein
mikroporöses Pulver mit einer spezifischen BET-Oberfläche von ~50 m 2 /g hergestellt
worden. 8YSZ-Schichten könnten daher mikroporös sein und eine niedrigere Durchlässigkeit
aufweisen. 30-50 nm dicke kubische 8YSZ-Schichten, die mit Hilfe der Keton-Route
hergestellt wurden, zeigten einen He- und N2- Transport gemäβ Knudsen-Diffusion, da
Defekte oder zu große Poren in der endgültigen Membranschicht vorhanden waren.
Wie erwartet, waren die aus linearen polymeren Amin-Solen hergestellten binären TiO2-ZrO2
Materialien zwischen 400 und 500ºC amorph und mikroporös. Die höchste spezifische BET-
Oberfläche von ~200 m 2 /g mit einer geschätzten Porengröße (gas physisorption) von ~1.0 nm
wurde beim Ti0.5Zr0.5O2-Pulver erreicht, das bei 500ºC mit Hilfe der Amin-Route hergestellt
wurde. Die Kristallisationstemperatur des orthorhombischen Ti0.5Zr0.5O2-Pulvers lag zwischen
v

550 und 600ºC, und damit ~250ºC höher als die Kristallisationstemperatur der einzelnen
Oxide.
Aus TiO2-, ZrO2- und binäre Oxidsolen konnten 20-60 nm dünne homogene Filme auf γ-
Al2O3-Membranen bei Kalzinierungstemperaturen zwischen 400 bis 600ºC synthetisiert
werden. Bei Ti0.5Zr0.5O2-Filmen, die bei 500 und 600ºC kalziniert wurden, lag die H2/CO2-
Permselektivität oberhalb des Knudsen Faktors. Es wurde eine Porengrößenverteilung mit
einem Durchmesser von ~0.3 bis ~0.5 nm bestimmt, die mit den weit entwickelten SiO2-
Membranen, die allerdings in Wasserdampfatmosphären degradieren, vergleichbar sind. Die
maximale He-Permeabilität betrug 1·10 -7 mol/m 2 sPa und die maximale Permselektivität
He/N2=14 und H2/CO2=6. Diese sind niedriger als die Selektivitäten von referierten SiO2-
Membranen, jedoch höher als die in der Literatur angegebenen Selektivitäten von TiO2-ZrO2-
Membranen. Diese Ergebnisse belegen, dass das Hauptziel, Membranen aus TiO2-ZrO2 zur
Gastrennung herzustellen, erreicht wurde. Allerdings ist für die Anwendung im Kraftwerk
noch eine deutliche Steigerung von Permselektivität und Permeabilität erforderlich.
Die Mikroporen in den Ti0.5Zr0.5O2-Membranen wiesen eine Stabilität von mindestens 1500
Stunden auf. Ergebnisse von Dampftests zeigten eine umkehrbare Porenblockierung ohne
Schichtabplatzung oder Zerstörung bei gleichbleibender Permselektivität. Aus den
Untersuchungen geht hervor, dass Ti0.5Zr0.5O2-Membranen vielversprechende Alternativen zu
SiO2-Membranen für Anwendungen zur Gastrennung im Precombustion-Verfahren sein
können.
vi

Content
Content
CONTENT......................................................................................................................... 1
1 INTRODUCTION AND AIMS .................................................................................... 3
1.1 GLOBAL WARMING AND CO2 EMISSIONS...................................................................................... 3
1.2 POWER PLANT AND GAS SEPARATION CONCEPTS.......................................................................... 4
1.3 OBJECT AND OUTLINE OF THIS THESIS .......................................................................................... 5
2 THEORETICAL BACKGROUND ............................................................................. 7
2.1 INORGANIC GAS SEPARATION MEMBRANES .................................................................................. 7
2.1.1 Dense metallic and ceramic membranes ................................................................................ 8
2.1.2 Microporous inorganic membranes...................................................................................... 13
2.2 ZEOLITE MEMBRANES ................................................................................................................ 14
2.2.1 Zeolite membrane material synthesis ................................................................................... 15
2.2.2 Zeolite membrane preparation methods ............................................................................... 16
2.2.3 State of the art zeolite membranes........................................................................................ 20
2.2.4 Selection of zeolite types....................................................................................................... 21
2.3 SOL-GEL DERIVED MEMBRANES ................................................................................................. 24
2.3.1 Sol-Gel chemistry ................................................................................................................. 24
2.3.2 Sol-Gel membrane preparation methods.............................................................................. 26
2.3.3 State of the art Sol-Gel derived membranes ......................................................................... 27
2.3.4 Selection of Sol-Gel derived membrane material ................................................................. 29
2.4 MICROPOROUS MEMBRANE CHARACTERISATION METHODS ....................................................... 31
2.4.1 Morphology .......................................................................................................................... 31
2.4.2 Mass transport...................................................................................................................... 32
3 EXPERIMENTAL....................................................................................................... 37
3.1 SUPPORTS AND INTERMEDIATE LAYERS ..................................................................................... 37
3.1.1 Support formation................................................................................................................. 37
3.1.2 Intermediate layer material characterisation and layer formation ...................................... 38
3.2 SYNTHESIS OF ZEOLITE MATERIAL AND MEMBRANE PREPARATION............................................ 39
3.2.1 Hydrothermal zeolite syntheses ............................................................................................ 39
3.2.2 SDA removal from dodecasil-1H by post synthesis treatments ............................................ 40
3.2.3 Dodecasil-1H layer formation.............................................................................................. 41
3.3 SOL-GEL DERIVED TIO2, ZRO2 AND BINARY OXIDE MEMBRANES............................................... 42
3.3.1 Polymeric TiO2, ZrO2 and binary oxide sol synthesis........................................................... 42
3.3.2 Drying and sintering of TiO2, ZrO2 and binary oxide bulk material .................................... 44
3.3.3 Microporous TiO2, ZrO2 and binary oxide membrane preparation ..................................... 44
3.4 CHARACTERISATION FOR MICROPOROUS MATERIALS AND MEMBRANES .................................... 45
3.4.1 Microporous material characterisation................................................................................ 45
3.4.2 Microporous membrane characterisation ............................................................................ 47
4 RESULTS AND DISCUSSION .................................................................................. 50
4.1 SUPPORTS AND INTERMEDIATE LAYERS ..................................................................................... 50
1

1 Introduction and aims
4.2 ZEOLITE MATERIALS AND LAYERS ............................................................................................. 53
4.2.1 Deca-dodecasil-3R synthesis ................................................................................................ 53
4.2.2 Structure directing agent free Dodecasil-1H synthesis ........................................................ 54
4.2.3 Dodecasil-1H layer characterisation ................................................................................... 66
4.3 SOL-GEL DERIVED MEMBRANES................................................................................................. 70
4.3.1 Polymeric sol characterisation............................................................................................. 70
4.3.2 Microstructure properties of membrane material ................................................................ 75
4.3.3 Sol-Gel derived membrane characterisation........................................................................ 96
4.3.4 Gas permeation................................................................................................................... 106
5 CONCLUSIONS AND RECOMMENDATIONS................................................... 110
6 REFERENCES........................................................................................................... 114
2

1 Introduction and aims
1 Introduction and aims
1.1 Global warming and CO2 emissions
Studies of Antarctica’s ice cores have concluded that the global temperature and
concentration of CO2 in the atmosphere are coupled over the last 650.000 years. This
indicates that the global temperature is influenced by the concentration of CO2. The
Intergovernmental Panel on Climate Change (IPCC) reported that the concentration of
CO2 in the atmosphere increased from approximately 280 ppm in the pre industrial era to
an average of 381 ppm in 2006: a 36% increase. This concentration has not been
exceeded in the last 420,000 years. 1
The concentration of CO2, which is measured at Mauna Loa (Hawaii), is increasing
approximately 1% per year. Simulation models predict a concentration of CO2 of 600
ppm by the year 2050. This concentration of CO2 increase is believed to be based on the
supply of primary energy dominated by fossil fuels. Stabilising the CO2 concentration
between 450-750 ppm requires persistent actions in order to reduce the emissions of CO2
such as conservation and energy efficiency, coal to gas substitution, renewable energy,
nuclear energy and CO2 capture and storage as indicated in the Figure 1 published by
IPCC. Clearly, CO2 capture and storage might play a dominant role in stabilising the
global concentration of CO2.
Emissions of CO2 originate from a number of different sources. It is believed that the
human activity forms a major part of these emissions of CO2. The human activity
contribution consists mainly of combustion of fossil fuels used in power plants to
generate energy, transportation, combustion of biomass, and residential and commercial
buildings. Significant CO2 emissions are also found in industrial processes, for instance
for the production of cement, iron, steel and hydrogen. The total CO2 emission
originating from the use of fossil fuel is estimated at 25-28 GtCO2/y in the year 2000
(maximal 84 GtCO2/y in 2050), of which more than 60% can be attributed to large scale
stationary emission sources. Power plants are primary candidates for CO2 capture and
storage due to their great potential: up to 45% emission reduction of CO2 can be achieved
in this sector alone in the year 2050. In addition, many of the coal based power plants in
Germany and elsewhere are depreciated and need to be renewed in the coming decades.
3

1 Introduction and aims
Year
Figure 1 Mitigation potentials as part of the CO2 emissions based on MiniCAM model. This is one of
many models and is based on single scenarios and does not convey the full scale uncertainties. 1
1.2 Power plant and gas separation concepts
The reduction and capture of the CO2 emissions poses great technical and economical
problems, especially for the so-called ‘end of pipe’ solutions. Among the biggest
challenges, and the general aim of this thesis, is the need to concentrate and purify the
CO2 present in the existing exhaust gasses. It is favourable to concentrate CO2 at high
pressures on the production side to reduce transportation and storage costs.
The vast majority of large scale emission sources such as power plants, generate diluted
CO2 flues, typically less than 15% (main components are H2O and N2). High
concentration of CO2, namely more than 90%, is required in order to be cost effective and
to be able to liquefy gaseous CO2. Up to now, there are no large (>500MW) power plants
equipped with methods to capture CO2. CO2 capture can be performed by scrubbing the
flue gas with an organic solvent. Membranes are good candidates for CO2 capture due to
their relatively low energy consumption. 2 Several CO2 capture concepts for fossil power
plants are discussed for future power supply. 1-4
1. Postcombustion capture
In postcombustion capture, CO2 is captured after the combustion process, conventionally
through the use of chemical solvents (e.g. amine scrubbing). This existing technology
operates at relatively low temperatures (~100ºC) and low exhaust gas CO2 concentration
(

1 Introduction and aims
Candidate materials include polymer (organic) membranes as well as ceramic (inorganic)
membrane systems with tailored properties and operational characteristics.
2. Precombustion capture
Gasification processes of fossil fuel with a subsequent CO-shift reaction offer high
potential for CO2-removal. The CO-shift reaction results in the formation of CO2 and H2
(180-550ºC 1 ). If H2 can be continuously removed from the gas mixture, CO2 can easily
be separated and stored. H2 can be used for generating electricity in gas turbines and fuel
cells, and for the production of chemicals and synthetic fuels based on fossils and
biomass. Physical absorption for H2/CO2 separation (net efficiency losses approx. 10 %
points) will be the major competing technology for molecular sieving (porous ceramic
membranes) or proton conducting membranes.
3. Oxyfuel combustion processes
Combustion of fossil fuels and biomass in pure O2 results in formation of CO2 and H2O
as combustion products. H2O can easily be separated from the combustion gas by
condensation at low temperatures. Conventional O2-production by air liquefaction
requires high investment costs and results in a significant efficiency drop of about 10 %
points for a power plant. Membrane systems offer a high potential for the supply of pure
oxygen for combustion processes while providing a drop in efficiency penalties. High
temperature (>800ºC) ceramic membrane systems with both ionic and electronic
conductivity are attractive materials due to the high selectivity of these systems for
oxygen separation.
1.3 Object and outline of this thesis
Object of this thesis
Inorganic membranes are promising candidates for the gas separation in the above
mentioned power plant concepts to capture CO2. Mixed oxygen ion and electronic
conductive membranes have potential for the oxyfuel process (O2/N2 separation) and
microporous inorganic membranes have high prospects for both pre- and postcombustion
concepts. Microporous inorganic membranes will be the topic of this thesis.
The general aim is to prepare inorganic microporous hydrothermal stable membranes for
separating power plant gasses such as H2/CO2 or CO2/N2. These aims can be subdivided
into:
5

1 Introduction and aims
Selection of membrane material
The selection of the membrane materials must fulfil the requirements of chemical,
mechanical and hydrothermal stability in combination with high permeation of one of
the species and a high separation factor.
o Zeolite material with narrow pore size distributions will be selected for the
separation of H2/CO2 or CO2/N2. All silica pore free dodecasil 1H and decadodecasil
3R zeolite types may fulfil this role. The focus on zeolites in this thesis
is mainly on the material research.
o Sol-gel derived TiO2 and ZrO2 materials can be alternatives to state of the art
SiO2 membranes for both pre- and postcombustion applications.
Preparation of sol-gel derived membranes
The primary aim of these sol-gel derived TiO2 and ZrO2 materials is to prepare
functional membrane layers with significant permselectivity for the above mentioned
applications.
Characterisation of membranes
Characterisation of the Sol-gel derived membranes will be studied through gas
permeance measurements. The membrane permeability setup must be designed and
build-up to study the single gas permeation and selectivity performance. The
hydrothermal stability must be investigated in order to evaluate the applicability of such
membranes for pre- and postcombustion CO2 capture.
Outline
In Chapter 2 “Theoretical background”, the inorganic gas separation membranes are
discussed. These inorganic membranes are subdivided in to dense and porous membranes
based on the transport mechanism. The selection of the membrane materials is based on
the properties of microporous membranes found in literature. This is followed by the
characterisation methods to study microporous materials and membranes. In Chapter 3
“Experimental”, the material synthesis, the membrane preparation and the
characterisation procedures of the equipment are described. Chapter 4 “Results and
discussion” contains chemical and physical properties of the selected zeolites and sol-gel
based materials, the microstructural properties of membrane layers and membrane
permeation performance. The aims will be correlated with the final results and the future
prospective is given in Chapter 5 “Conclusions and recommendations”.
6

2 Theoretical background
2 Theoretical background
Membranes are semi-permeable barriers which are able to selectively transport different
species. The selectivity is a measure of the different transport abilities of the species. Gas
separation membranes inhibit the transport of favour gas molecules where others are
either blocked or have a lower transport flux. The most common gas separation
membranes can be distinguished by the choice of membrane material:
o Organic polymer membranes
o Ceramic or metallic membranes
Organic polymer membranes can be applied in power plants for the capture of CO2. In
the future, highly efficient power plants in combination with high temperature separation
technology for CO2 capture will be preferred. Ceramic or metallic membranes may fulfil
this role.
This chapter contains a literature study on the state-of-the-art ceramic or metallic gas
separation membranes. Review of this literature is important for understanding the
restrictions and possibilities of promising gas separation membranes for the previously
mentioned power plant concepts. Crystalline (page 14) and sol-gel derived (page 24)
microporous ceramic membranes are the focus of this thesis. Finally, mass transport in
microporous membranes is outlined.
2.1 Inorganic gas separation membranes
Three types of inorganic membranes for gas separation applications can be distinguished:
o Dense ceramic membranes
o Dense metallic membranes
o Molecular sieving membranes
Dense ceramic membranes such as mixed ionic electronic conductive membranes are
candidates for oxygen production from air with high selectivity due to their ability to
transport oxygen ions and electrons at elevated temperatures in the range of 600-1000ºC.
Dense ceramic protonic membranes are able to separate H2 from other gasses at
temperatures between 500 and ~800ºC.
Dense metallic membranes are able to separate H2 from other gasses in a temperature
range of 300-600ºC. The operating temperatures of ceramic and metallic protonic
conductors match the typical reforming reaction temperatures in the precombustion
concept at 180-550ºC.
7

2 Theoretical background
The development of molecular sieving membranes started about 50 years ago as a
separating agent for 235 UF6 and 238 UF6 for nuclear fuel enrichment. 5 Since then, a number
of other applications have emerged such as liquid-solid and liquid-liquid separation,
pervaporation and finally gas separation, generally determined by the pore size of the
selective layer. Microporous ceramic can be used as molecular sieve membranes and
might have high prospects in the post and precombustion concepts provided that both the
membrane permeability and permselectivity are substantial in order to minimise
efficiency penalties in power plants.
2.1.1 Dense metallic and ceramic membranes
Dense membranes are gastight layers. The first group of dense inorganic membranes
studied in the past decade for gas separation is the metallic membrane type; primarily
palladium alloy membranes for hydrogen (H2/CO2) separation. The latest development in
this field is the study of supported thin metallic membranes, as thin as a few tenths of
microns. 6 The most extensively studied group of dense membranes includes the oxygen
ionic conductive and mixed oxygen ionic and electronic conductive ceramic membranes. 6
In the early 1980s, a third group of dense membranes emerged from high temperature
hydrogen semi permeable dense ceramic membranes. These membranes are based on
proton-conducting ceramics.
2.1.1.1 Dense metallic membranes
The hydrogen transport properties of dense metallic membranes such as silver doped
palladium (Pd/Ag) membranes are determined by many factors. The factors can be
hydrogen transport to the metal membrane, adsorption of hydrogen on the surface,
splitting of hydrogen to be incorporated into the lattice of the metallic membrane
material, bulk diffusion of the protons and counter electrons, regeneration of the
hydrogen molecules, desorption of the hydrogen and finally diffusion from the membrane
which is extensively discussed in the article of Bredesen et al. 7
There are several methods used to coat thin metallic films on porous metallic or ceramic
supports: electroless plating, chemical and physical vapour deposition or sputtering. All
these methods can be used to obtain good quality, thin Pd/Ag membranes with hydrogen
to helium (or nitrogen) permselectivity.
Pex et al. 8 and Pan et al. 9 prepared palladium based membranes on alumina supports
(see Table 1). Others prepared thin palladium 10 or palladium nickel 11 layers on stainlesssteel
porous substrates, which offer high hydrogen permeance in combination with
excellent permselectivities. A sandwich structure based on a dense vanadium layer and a
thin porous palladium layer on both sides resulted in even higher permselectivitiy. 12
8

2 Theoretical background
Despite the good permeability and permselectivity of these membranes upscaling to a
plant is rarely accomplished. Generally, these membranes suffer from hydrogen
embrittlement, which occurs as a result of crystal transformation between α- and β-phases
in hydrogen atmosphere at temperatures below 295°C, 13,14 and poor stabilities in
atmospheres containing: CO, S, Cl and H2O. This can lead to complete deterioration of
the structure. Furthermore, the raw materials, namely palladium and vanadium, are
expensive.
Table 1 Gas permeance, permselectivity and thickness of some the state of the art metallic membranes
Permeance in
mol/m 2 ·s·Pa x10 -8
H2 N2 H2/ N2
Permselectivity Thickness (µm) Reference
100 [8]* 0.33 300 3-5
87 0.087 1000 3
1750 1.09 1600 1
1970 0.42 4700 0,5
1400 0.03 50000 0.5/40/0.5
* H2 permeance converted in m 3 /m 2 hbar
2.1.1.2 Oxygen ionic conductive membranes
Dense ionic conductive ceramic membranes derive their capability for oxygen separation
from the presence of oxygen vacancies in the material crystal lattice. The presence of
oxygen vacancies enables ionic oxygen to be selectively transported through the lattice
via a hopping mechanism in the presence of an oxygen chemical potential as the driving
force. This reaction is presented in equation (1), with the Kröger-Vink notation.
O 2V 2O 4h •
••
+ ↔ + (1)
2 O
x
O
In practical terms, dense membranes only transport oxygen ions at elevated temperatures,
typically above 800°C, with ideally, 100% selectivity when gas leakage is excluded.
Since the charged oxygen ions are transported, the charge must be compensated with
electron counter-transport to maintain the charge neutrality. The charge can be
transported externally as Figure 2-A suggests and one refers to an ion conducting
material or oxygen pump. The membrane is called a mixed ionic electronic conductor
(MIEC) when both oxygen ions and electrons are transported through the same
membrane material (see Figure 2-B). The oxygen flux can be presented by the simplified
Wagner equation (2). 15
9
8
9
10
11
12

2 Theoretical background
pO2′′
RT
j = − ∫ t σ d ln pO
(2)
O2 2 e ion
16F
L pO2′
2
Thus, the oxygen flux (jO2) through dense ceramic membranes is dependent on the
temperature (T), thickness (L), the electronic conductivity (σe; te= σe/[σe+σion]) and ionic
conductivity (σion) of the membrane material and the oxygen partial pressure gradient
(∆pO2) over the membrane. 16
pO’2
e
A
O 2-
A B
pO’’2
10
pO’2
O 2-
pO’’2
Figure 2 Dense membrane concepts. A): oxygen pump. b): Mixed ionic and electronic conductor 16 .
The most promising materials for oxygen separation and methane reforming membranes
are the acceptor-doped perovskite-type oxides with the general formula La1-xAxByO3-δ
(A = Sr, Ba and B = Fe, Cu, Ni, Cr, Co). These compounds show high electronic and
ionic conductivity. The highest oxygen flux has been obtained by Vente et al. 17,18 for a
strontium barium iron doped cobaltate at 900˚C, see Table 2. The highest oxygen flux
found in literature has been obtained with an oxygen partial pressure at the feed side of
1 bar and a membrane thickness of only 0.2 mm, whereas the other results are based on
membranes thicker than 1 mm. If the oxygen flux is normalised by the thickness than
Mertins et al. 19 measured the highest oxygen flux with atmospheric air at the feed side.
The flux of the membrane reported by Mertins et al. 19 is obtained by using methane as a
sweep gas in combination with a Ni-catalyst for methane reforming application.
Decreasing the thickness would improve the bulk diffusion. The oxygen transport will be
rate determined by the oxygen surface exchange 20 for membranes with a thickness less
than 500 µm. Thin La1-xSrxCo1-yFeyO3-δ films supported on tailored ceramic substrates
can be prepared 21 . Fluxes between 1 and 30 ml/cm 2 ·min can be expected. 7
2e -

2 Theoretical background
Table 2 Oxygen permeation (j(O2)), thickness (d) and temperature (T) of various oxygen conductive
membranes. The flux is normalised for membranes with a thickness of more 500 µm to a thickness of
1000 µm assuming that the bulk diffusion is rate determined.
Membrane material
j(O2)
normalised
j(O2) T D Feed-
Side
ml/(min·cm 2 ) °C mm
11
Sweep-
side
La0.2Sr0.8Fe0.8Co0.1Cr0.1O3-δ 5.47 5.47 800 1 Air He/CH4
(Ni cat.)
SrCo0.8Fe0.2O3 1.6 1.1 900 1.5 Air He
SrCo0.33Fe0.66O3 3.37 2.73 1000 1.2 Air He
SrCo0.66Fe0.33O3 2.79 2.79 1000 1 Air He/air
SrCo0.66Fe0.33O3 4.11 4.11 1000 1 Air He/air
Ba0.5Sr0.5Co0.8Fe0.2O3 3.1 900 0.2 p(O2) = 0.2 atm He
Ba0.5Sr0.5Co0.8Fe0.2O3 13.3 1000 0.2 p(O2)= 1.0 atm He
Ba0.5Sr0.5Co0.8Fe0.2O3 2 1.35 900 1.5 Air He
2.1.1.3 Dense protonic conductive membranes
The hydrogen transport through dense ceramic protonic membrane materials is similar to
the oxygen ion transport. The oxygen vacancies in mixed electronic and protonic
conductors are in equilibrium with water vapour, see equation (3) and (4) below:
•• 1
x •
VO + O2 ↔ OO + 2h
2
(3)
x
H O V O 2OH
•
••
+ + ↔ (4)
2
O O O
The OHO • is an interstitial proton that associates strongly with a neighbouring oxygen
ion. Protons can also be formed in the lattice of the membrane material when H2 gas is in
equilibrium (5) with the oxides:
2O + H ↔ 2OH + 2e′
(5)
x
•
O 2
O
The hydrogen transport through these membranes is dependent on many factors such as
the mobility of the interstitial proton as a function of the hydrogen partial pressures
gradient. 24
Lin et al. 6,25 studied the hydrogen permeance of strontium thulium doped ceria (see
Table 3). Balachandran et al. 26 and Meulenberg et al. 27 prepared BaCe0.8Y0.2O3-δ (BCY)
based membranes and found similar results as Lin et al. 6 Eltron U.S. 28 produced a
strontium iron cobaltate with a reasonable hydrogen permeation. Balachandran et al. 26
Ref
19
22
23
23
23
18
18
22

2 Theoretical background
proved that the electronic conductivity of BCY based membranes is insufficient for future
gas separation. The electronic conductivity can be increased, thereby increasing the
hydrogen flux by dispersing a metal powder (Pd/Ni) into the ceramic (BCY) matrix to
form a cermet. In a later stage, Balachandran et al. 29 claimed an extremely high hydrogen
permeance of 1.3·10 -5 mol/cm 2 ·s for a strontium iron cobaltate cermet with a thickness of
only 20 µm. Generally, state of the art proton conductors are the Sr- and Ba- based
perovkites. However, the stability towards CO2 and H2O at elevated temperatures must be
improved for industrial applications. 30-33
Table 3 Hydrogen permeance and thickness of the state of the art hydrogen conducting membranes.
Material H2 Permeation Temperature Thickness Ref.
mol/cm 2 ·s x10 -8 (ml/cm 2 ·min) [˚C]
SrCe0.95Tm0.05O3-δ 2.98 0.04 900 16mm
SrCe0.95Tm0.05O3-δ 9.37 0.13 900 150 µm
BaCe0.8Y0.2O3-δ 3.60 0.05 - -
SrFeCo0.5O3 7.44 0.1 - 2 mm
SrFeCo0.5O3 1300 17 900 20 µm
SrCe0.85Yb0.05O3 1100 15 1000 100 µm
12
24
29,34,35
26
28
29
31

2 Theoretical background
2.1.2 Microporous inorganic membranes
Porous inorganic materials are classified by pore size as stated by the International Union
of Pure and Applied Chemistry, 36 listed in Table 4. Microporous membranes can
function as molecular sieves due to size exclusion in order to separate the power plant
gasses such as H2/CO2 or N2/CO2 based on the kinetic diameter (see Table 5).
Table 4 IUPAC 36 pore size classification
Structure sublevel Pore size
Macroporous >50 nm
Mesoporous 2-50 nm
Microporous Supermicroporous 0.7-2 nm
Ultramicroporous

2 Theoretical background
2.2 Zeolite membranes
The Swedish mineralogist Cronstedt found a new type of mineral in 1756 and named it
zeolite. Zeo-lite is derived from a Greek word and means boiling stone. It seemed that the
new mineral desorbs water at high temperatures. Since then, many natural and industrial
zeolites have been found and synthesised. Microporous zeolites are crystalline
(alumino)silicates * and can be used for adsorption, catalysis and ion-exchange. In
general, zeolites have higher chemical and thermal stability than microporous amorphous
silica.
The zeolite structure is formed from a three dimensional network of SiO4 and AlO4
tetrahedra. These tetrahedra are connected to each other through shared oxygen atoms
and form channel or cage-like structures. These voids have well-defined rings (pore
opening) sizes and shapes. A selection of zeolite structures is listed in Table 6. 38 More
than 130 zeolite framework types are approved and published on the internet site of the
International Zeolite Association. 39 The zeolites can be subdivided based on their largest
pore aperture. The pore aperture is a window with a certain amount of neighbouring
tetrahedral centres that share an oxygen atom. A zeolite with 12 tetrahedra centres is
referred as 12-ring. The path of the lowest resistance for transported species is
determined by the pore aperture and interconnectivity of pores forming channels of
different dimensionality.
Table 6 Structural characteristics and properties of different zeolite types and zeolite related materials.
Code Zeolite type [Si]/[Al] Ring Channel system Pore size Å (x-y)
SGT Sigma 2 70-∞ 6 0D ~2.8
DOH Dodecasil 1H ∞ 6 0D ~2.8
SOD Sodalite 1-∞ 6 0D 2.2-2.8
DDR Deca-dodecasil 3R ∞ 8 2D 3.6-4.4
ETS-4 - 3-4
LTA Zeolite A 1-∞ 8 3D 3.8-4.3
FER Ferrierite - 10 2D 5.2·5.4
MFI Silicalite-1 ∞ 10 3D
5.2·5.7
ZSM-5 10-1000 10 3D
FAU Zeolite X 1-1.5 12 3D 7.4
Zeolite Y >1.5 12 3D 7.4
MOR Mordenite - 8/12 2D 4.0-7.0
Breck 40 outlined the properties of several zeolite structures including the individual pore
size, shape and interconnectivity. The pore sizes vary between 2-12 Å and the chemical
composition (e.g. the Si/Al ratio) can determine the hydrophilic, hydrophobic, acidic, and
basic properties of the zeolite. 38
* The largest number of zeolites is represented by aluminosilicates. Aluminogermanates, gallophosphates
or silicoaluminophosphates belong to this group as well and are documented at www.iza-structure.org.
14

2 Theoretical background
Poly-crystalline zeolite material can form a membrane layer useful in separating different
species via micropore diffusion. Poly-crystalline zeolite membranes are promising
candidates for separating gasses under the harsh conditions associated with power plants
due to their chemical and thermal stability. In general, zeolite membranes can act as
molecular sieves for gasses due to their well-defined pore size. At present, zeolite
membranes are applied in the field of pervaporation.
2.2.1 Zeolite membrane material synthesis
Zeolites are crystallised from nutrients at relatively moderate temperatures and mild
pressures. This process is hereafter referred as hydrothermal synthesis and is
schematically drawn in Figure 3. The resulting zeolite structure is dependent upon the
hydrothermal conditions, such as the reactant concentrations and nature, the pH, the
presence of structure directing agents or templates (hereafter referred as SDA),
temperature, reaction time etc. The most important factors affecting the zeolite synthesis
are the gel composition and the reaction conditions.
The precursor mixture or gel composition is generally a mixture of aluminium and silicon
sources, including a base and a solvent. SDA can be added in order to assist in crystal
formation. The most common aluminium sources are sodiumaluminate, aluminiumnitrate
or aluminiumphosphate.
15

2 Theoretical background
Stainless Steel
Autoclave
Gastight
Hydrothermal crystallisation

2 Theoretical background
membranes (Figure 4). The supported zeolite membranes are by far the most extensively
studied. The most investigated support material is α-alumina. Stainless Steel, γ-alumina
and titania are examples of other support materials suitable for zeolite membranes. 43,44
A B C
Figure 4 Schematic drawing of polymer – zeolite composite membrane (A), self supported zeolite membrane
(B) and a (α-Al2O3) supported zeolite membrane (C).
Ideal zeolite membranes are gastight directly after the hydrothermal crystallisation. A
common nitrogen permeation test is applied to test the gas tightness. Zeolite membranes
must be activated, or in more detail, the water and organics, such as structure directing
agents (SDA’s), must be removed in order to obtain an open and interconnected
microporous structure. Removing these SDA’s can be done via:
o Chemical leaching
o Ion exchange techniques which is applicable for the larger pore zeolites (e.g. 12ring)
o Thermal treatment at specified temperatures, heating rate, atmosphere and
duration. The heating rate and cooling rate can be key factors for removing SDA
without destroying the zeolite layers due to different expansion coefficient of
templated and template free zeolite structures
o Low temperature ozone treatment. 43
The thermal expansion coefficient difference of the membrane layer and support can
cause stress, undesired cracks and peeling off. Porous intermediate layers or thin zeolite
layers with randomly ordered crystals are recommended in order to reduce the thermal
stresses.
The preparation methods for different supported zeolite membranes are known and
include (a) in situ hydrothermal synthesis in the presence of a substrate, (b) “dry or wet
gel conversion method” and (c) the application of two stage (phase) secondary growth
methods. The seeds in the secondary growth methods can be colloidal zeolites that are
deposited by dip coating, spin coating or sputtering 38 .
17

2 Theoretical background
In situ hydrothermal synthesis
The most widely reported and successful route to prepare zeolite membranes is in situ
hydrothermal synthesis on a substrate. The substrate is immersed into the zeolite
precursor mixture which is contained in an autoclave, and then heated for a
predetermined time, see Figure 5. Both, nucleation and growth of the zeolite crystals,
occurs on the support surface.
Figure 5 Schematic arrangement for the in-situ hydrothermal synthesis of a zeolite membrane. The support
for the zeolite film is immersed in the precursor mixture within the autoclave during the hydrothermal
synthesis.
These synthesis routes are used for many zeolite types. However, there is little scope for
the control of the microstructure of the final films. Intercrystal defects can be difficult to
avoid when using in situ growth routes and this limits the applications in gas separation. 45
The gel compositions, which result in zeolite layers, usually are less concentrated than
the optimal composition for making powders. 43
Synthesis routes in which the zeolite is formed within the pores of the support are studied
in order to overcome the problems of the defects and to increase the strength.
The preferential nucleation and growth is dependent on many factors such as time and
temperature. Generally, the growth is slow at low temperatures and is observed mostly
within the pores. Fast growth and the formation of an outer layer occur at higher
temperatures (190 °C). Several processes have been studied to minimise the amount of
layer defects such as Chemical vapour deposition (CVD) and microwave heating. 45
18

2 Theoretical background
Dry or wet gel conversion method
This method is based on deposition of amorphous alumino-silicate gel on the support.
The crystallisation is initiated using the vapour of amines such as triethylamine,
ethylenediamine and water. The process is called ‘steam-assisted crystallisation’ if the
dry gel contains structure directing agents (SDA). Vapour Phase Transport is the
terminology used when the gel composition is SDA-free. However, many zeolite layers
present cracks, which can be explained by the large volume shrinkage from the gel state
to the zeolite layer.
Secondary growth
Seeding the membrane initially can be beneficial due to the decoupling of the nucleation
and the growth processes to obtain a controlled microstructure (Figure 6). Generally, the
seeds are synthesised at lower temperatures than the classical recipes in order to form
smaller crystals. 43 The seed size can vary between 50 to 2000 nm. Depending on the
temperature, induction period and gel composition, the seed synthesis time can be very
long (up to several months). 43 Commercial zeolite crystals can be grinded to acquire
small seeds 200 nm in size.
Figure 6 Schematic drawing of secondary grown zeolite membrane layer
There are several methods for attaching seeds to the surface of the support. By (i)
changing the pH of the solution for matching the zeta potentials of alumina support and
zeolitic seeds. Seeds (ii) can be rubbed into the support with help of cationic polymers.
As an alternative to seed rubbing, colloidal zeolite particles, which achieve a greater
control of the membrane microstructure, can be deposited on the supports. The
attachment can be improved by heating the seeded supports to 150-200ºC in order to
condensate the hydroxyl groups.
The use of sols with nanosized primary building units followed by a secondary growth
process were firstly studied by Tsapatsis et al. 46 The substrate is deposited (e.g. by dip
coating) with nanosized zeolitic nuclei and these zeolite sols are connected
hydrothermally. This process can result in a zeolite membrane with a thickness of about
200 nm.
Another (iii) recent zeolite seeds preparation method is the use of laser ablation. The
depositing of zeolite seeds is prepared by using pulsed laser deposition and followed by
hydrothermal growth of the layer. 38
19

2 Theoretical background
2.2.3 State of the art zeolite membranes
Despite their large pore size of 0.74 nm, the 12-ring Faujasite (FAU, see Figure 7)
zeolite type can be selective for small gasses due to competitive adsorption. Ion-exchange
with alkali metals such as cesium increased the CO2 permeability of the FAU membrane
while N2 permeance remained unchanged. A maximal CO2/N2 permselectivity of 149 was
obtained with a CO2 permeance of ~0.8·10 -6 mol/m 2 ·s·Pa using an equimolar CO2 and N2
feed, 47,48 see Table 7. This selectivity is predominantly determined by preferential CO2
adsorption. The definition of permeance and permselectivity is given at the end of this
chapter.
The most intensively studied zeolite membrane material is the 10-ring MFI (Mobile Five,
see Figure 7). So far, there are no MFI membranes, with H2/CO2 permselectivities
significantly higher than the Knudsen factor, were found in literature. The low
permselectivity can be explained by the large pore aperture (0.51x0.55 nm) compared to
the kinetic diameters of 0.289 and 0.33 nm for H2 and CO2 respectively.
Table 7 Gas permeance and permselectivities of MFI and FAU-type zeolite membranes
Permeation rate in mol/m 2 ·s·Pa x10 -8 Permselectivity Ref.
H2 CO2 O2 N2 H2/CO2 CO2/N2
MFI 260 19 13.7
27 15 1.8
330 121 121 2.73 1.0
74 23
32
700
56
12 7 4.0 2.67 3.0
10 1.0 10
65.6 2.6 25.5
FAU 80 0.5 149
* Based on a stainless steel support
# Permselectivity based on a binary equimolar CO2/N2 feed
Tsapatsis et al. 54 studied the 10-ring ETS-4 type zeolites (titaniasilicate as chemical
composition) for the use of molecular sieves. ETS-4 collapses near 200°C to an
amorphous phase due to the loss of structural water chains present along the channel
system. The thermal stability of ETS-4 can be improved via ion exchange of e.g. Na
against Sr. The amorphous phase of Sr-ETS-4 is obtained at temperatures above 350 °C.
This type of zeolite is stable upon exposure to moist air for at least two weeks. 55 The unit
cell dimension of ETS-4 becomes smaller as the temperature increases due to
dehydration. The contraction of the unit cell dimensions shows a clear correlation with
20
2*
2
49
50
46
51
52
12
53
47,48#

2 Theoretical background
the material’s adsorption capacity for various gas molecules. Kuznicki et al. 55 claimed
even that ETS-4 can be tuned to penetrate or absorb oxygen and to exclude nitrogen into
the crystal cages, resulting in an oxygen-selective adsorbent.
Known 8-ring zeolite type membranes are Zeolite Type A (LTA, Linde Type A, Figure
7) and deca-dodecasil 3R. The Si/Al ratio for zeolite (Type) A is 1-3.7. Recently,
hydrophobic all-silica zeolite A was prepared by using a supramolecular organic
structure-directing agent. 56 Guan et al. 57 prepared aluminosilicate LTA zeolite
membranes with rather low permselectivity of 7.5 for H2/N2. Hedlund et al. 58,59 prepared
ultra thin layers of 200 nm or less of LTA.
The low H2/CO2 permselectivity of LTA membranes can be explained by the possibility
that both H2 and CO2 can enter the pores and by the fact that intercrystalline pores may
be present as a result of poor poly-crystallisation. This drawback is less critical for
pervaporation applications. Nowadays, the Japanese company NGK prepares
commercially available LTA and DDR membranes.
2.2.4 Selection of zeolite types
A B C
Figure 7 Zeolite frameworks of MFI (A), LTA (B) and FAU (C).
Lab scale zeolite membranes offer good gas separation factors predominantly determined
by the sorption properties and less by molecular sieving. This is mainly due to the
presence of intercrystalline pores (larger than 2 nm) as a result of poor polycrystallisation.
Nevertheless, zeolite membranes are promising candidates for CO2
capture.
In this study, zeolite types are selected for the pre and postcombustion concept. Sorption
properties of H2, CO2 and N2 are less pronounced at elevated temperatures. Therefore the
zeolite types are selected primarily as molecular sieves. These requirements exclude
zeolite types with 10 or more rings (Figure 8).
21

2 Theoretical background
The additional requirement for membranes in the pre and postcombustion concept is
hydrothermal stability. Generally, hydrophobic zeolites can be stable up to elevated
temperatures in humid conditions due to their high Si/Al ratio.
The zeolite types fulfilling these requirements are the 6-rings MEP, NON, MTN, SOD,
SGT and DOH. The only known all silica 8-ring zeolite type membrane is deca-dodecasil
3R (DDR). Recently, all silica zeolite Type A have been prepared. 56 In this thesis, the
all-silica clathrasils DOH and DDR have been selected. DOH has not been studied for the
purpose of separating gasses. To the best of our knowledge, no all-silica 6-ring zeolite,
with significant permselectivity, membranes have been published up to now.
MFI
LTA
DDR
DOH
2 3 4 5 6
H 2 CO 2 N 2
Pore aperture in Å
Figure 8 Pore aperture of zeolite types compared to the kinetic diameters of H2, CO2 and N2.
2.2.4.1 Synthesis of small DDR crystal
Only a few authors have published on the synthesis of deca-dodecasil 3R (possessing the
zeolite framework type DDR-Figure 9). The name deca-dodecasil 3R originates from
dodecahedra and decahedra as structural building units forming three layers, which are
rhombohedrally stacked. Gies et al. 60 were the first to develop this zeolite structure. Den
Exter et al. 44 showed the difficulties of preparing this type of structure. The company
NGK 61,62 published and patented 63 this zeolite structure as membrane material. Even
Exxon Mobil is actively working on this material. 64 Recently, van den Berg 65 studied this
material for the use of hydrogen storage. 1-amino adamantane (hereafter referred as
ADA) is the most frequent structure directing agent (SDA) that results successfully in all
silica single phase DDR structure. DDR-type membranes with an aperture of 0.36·0.44
nm possess a relatively high CO2/CH4 selectivity (equimolar feed) due to preferential
absorption. The permeance of CO2 is higher than the permeance of He or H2 for this 5-10
µm thin DDR type membrane. 62 Significant CO2/H2 permselectivity might be obtained
from this zeolite structure.
22

2 Theoretical background
Figure 9 Deca-dodecasil 3R (DDR) structure 66 Figure 10 Dodecasil 1H (DOH) structure 66
2.2.4.2 Synthesis of small SDA-free DOH crystals
The all silica zeolite type Dodecasil 1H 67 (possessing the zeolite framework type DOH-
Figure 10) 68 with a pore aperture of ~2.8 Å 44 might be a promising membrane material
for separating hydrogen at elevated temperatures. The name dodecasil 1H originates from
dodecahedra as the structural building units forming layers which are hexagonally
stacked.
Only a few groups have studied DOH as a potential candidate for gas separation. The
research group of Gies 69,70 was among the first to study this zeolite type. Grebner et al.
71-74 and den Exter 66 studied this zeolite material for several applications. DOH zeolites
can be synthesised with SDA’s only. The most common are 1-amino adamantine (ADA),
piperidine and 1-adamantyl trimethylammonium bromide 71-73 in aqueous solutions or the
metal organic salts 75 [Co(C5H4Me)2]PF6 or [Fe(C6H6)(C5H5)2]PF6 in solvent free
systems. After synthesis the templates occupy pores in the zeolite structure and, hence,
have to be removed. DOH zeolite material is stable at elevated temperatures, for instance
up to temperature of 900ºC, while still trapping some of the SDA. 71-73 In contrast, ADA
from deca-dodecasil 3R is removed at 450°C, 44 but 700ºC is preferred. 63
DOH as membrane materials must be prepared template free and these crystals should be
prepared as small crystals in order to form thin layers. Small crystals might be prepared
from a homogeneous gel for short crystallisation times or by means of aging. The number
of crystals can be enhanced by decoupling the nucleation from crystal growth by means
of aging the gel composition at moderate temperatures before applying the common
crystallisation temperature.
Synthesising SDA-free DOH might be possible by using seeds and SDA free gel
compositions as Grebner et al. claimed. 71-73 Post synthesis treatment to remove the
template can be achieved via conventional calcination for extended periods or calcination
of smaller crystals to shorten the transport path length of oxygen or combusted organics.
23

2 Theoretical background
2.3 Sol-gel derived membranes
Microporous initial amorphous derived membranes, particularly microporous silica
membranes, have been studied intensively in the last decades. Silica membranes can be
prepared via:
o Chemical vapour infiltration or deposition
o Sol-gel techniques
Chemical vapour deposited silica membranes are prepared starting from gaseous silica
precursors that either infiltrate or deposit a silica layer in or on top of meso or
macroporous supports. This method can reach high permselectivities but is generally
coupled with a significant decrease of permeability due to the effect of pore blocking.
Sol-gel derived silica membranes offer high H2 permeance 1-2·10 -6 mol/m 2 sPa in
combination with reasonable permselectivities (terms are explained on page 36).
However, the stability of silica membranes in humid conditions at elevated temperatures
is poor.
First, the synthesis of sols is discussed, followed by the common preparation methods of
microporous sol-gel derived membranes. An overview of these membranes is given in
section 2.3.3 followed by the selection of the promising sol-gel derived microporous
materials in section 2.3.4.
2.3.1 Sol-Gel chemistry
The sol-gel process is one of the most appropriate methods for the preparation of
mesoporous and microporous layers. There are two sol-gel routes, polymeric and
colloidal (Figure 11). Different steps involved in the process determine the porous
structure, even at the very first stage of precursor synthesis. The first step in sol-gel
process is the preparation of a sol using molecular precursors. There are two types of
precursors, metal organics (preferentially metal alkoxides) and metal salts. Clusters or
colloids are formed at the sol stage by condensation reaction. The chemistry of the sol
formation is well documented by Livage et al. 76 Depending on conditions, nearly linear
polymeric structures (hereafter referred as polymeric sols) and dense colloidal particles
(colloidal sols) can be prepared. These clusters and colloids collide at the final stage to
form a gel.
24

2 Theoretical background
Figure 11 Schematic drawing of the colloidal and polymeric sol-gel technique 77
The particle size of polymeric sols may be in the range of 2-20 nm whereas colloidal sols
have particle sizes up to 1000 nm in diameter. Polymeric sols are prepared by adding
small amounts of water to keep the hydrolysis reaction slow, equation (6). The water can
be added in a controlled method by:
i) slowly adding a high ratio alcohol/water solution,
ii) in situ production of H2O by an esterification reaction,
iii) dissolving an alkaline base or
iv) adding a hydrated salt into the alkoxide solution in alcohol.
The catalyst can be applied to enhance the condensation reactions. Precursor modification
or stabilisation is another method used to modify the precursors in order to reduce the
hydrolysation rate. Ketones, amines and carboxylic acids can stabilise the precursor and
thereby decrease the hydrolysis reaction. Common precursor modifiers are acetyl acetone
78,79 and diethanolamine 78-83 for titania and zirconia polymeric sols.
25

2 Theoretical background
Colloidal sols are prepared with excessive amounts of water. Most applied sols are
synthesised by using a metal alkoxide. The metal alkoxide is hydrolysed. The formed
metal hydroxides are reacted in the alcohol condensation reaction which is usually
occurring in parallel to the water condensation reaction. These reactions (6)-(8) may
occur simultaneously.
hydrolysis
≡ Si − OR + H 2 O ←⎯⎯
⎯ → ≡ Si − OH + ROH
(6)
alcohol condensation
≡ Si − OR + HO − Si ≡ ←⎯⎯⎯
⎯⎯
→ ≡ Si − O − Si + ROH ≡
water condensation
≡ Si − OH + HO − Si ≡ ←⎯⎯⎯
⎯
→ ≡ Si − O − Si ≡ H O
(8)
2.3.2 Sol-Gel membrane preparation methods
Sol-gel derived microporous membranes are based on a graded structure. Macroporous
substrates can be alumina, porous stainless steel and titania. 84 In contrast to zeolite
membranes, see also page 16, an (mesoporous) intermediate layer is required to i) prepare
a low roughness surface for the microporous layer and ii) avoiding infiltration into the
supporting layer (Figure 12). Metastable γ-Al2O3 is the most widely used material for
intermediate layers.
Microporous dpore < 2 nm
Mesoporous 2< dpore > 50 nm
Macroporous dpore > 50 nm
Figure 12 Left, schematic build-up of an asymmetric inorganic porous membrane and right the schematic
overview of the sol-gel process. Multilayers are prepared by repeating the last 3 steps.
Micro and mesoporous membranes are formed at the sol stage. The most common
coating methods are dip or spin coating. Many factors have an influence on the
microstructural properties of the final layer. The most important factors are the gelation,
viscosity and the properties of the support. The support may have a significant influence
26
2
Sol
Coating
Drying
Fired
(7)

2 Theoretical background
due to its capillary pressure drop in combination with its hydrophobicity. These
properties are described by Bhave. 5
Finally, the membrane is formed by a thermal treatment. First, the gel layer is dried at
low temperatures (

2 Theoretical background
hydrophobic membranes. They reported generally low permeabilities in combination with
reasonable permselectivities. Similar study is performed by Yoshida et al. 96 who claimed
to have produced relatively hydrothermal stable silica/zirconia membranes. The highest
hydrogen permeance of silica based membranes is published by Oyama et al. 97 (Table 8).
However, the H2/CO2 permselectivity is rather low. This permselectivity can be increased
to 4300 [-] by a silica chemical vapour deposition layer in combination with reasonable
permeability.
Silica membranes, as published by de Vos, 87 showed a decrease of permselectivity after
exposure to mild humid conditions at elevated temperatures. The hydrothermal stability
could be increased by changing the hydrophobicity of the functional silica layer using
methyl groups calcined in oxygen-poor atmosphere. However, the increase of
hydrothermal stability was coupled with a decrease of permselectivity. The decrease of
permselectivity of these silica membranes might be based partly on the poor adhesion
between the α-Al2O3 support and the γ-Al2O3 intermediate layer.
Table 8 Permeance, permselectivity, pore size (dpore) and layer thickness (dlayer) of silica membranes
Permeation in mol/m 2 ·s·Pa x10 -8 [m 3 /m 2 ·h·bar] Permselectivity dpore (nm)/ Ref
H2 CO2 O2 N2 H2/CO2 CO2/N2 O2/N2 dlayer (µm)
200 [16] 50 10 1 4 50 10 -/0.03
87 *
60 1.0 1 70 -/0.03
87 #
60 10 10 10 7 1 -/0.03
87 &
- 100 20 5 4.3-5/0.05
93
10 0.3 0.06 7.0·10 -3 7 42.9 8.57 0.35/-
95 $
45 0.45 -/-
96
20 4.65·10 -3 4300 -/0.01-10 97 †
2000 628 4.7 -/0.01-10
130 90 17 1.44 5.29 0.35/
1 Silica membrane calcined at 400ºC.
# Silica membrane calcined at 600ºC.
& Methylated silica membrane calcined at 400ºC
$ Silica membrane doped with zirconia
† Silica layer applied by means of CVD
A delamination between the crystalline phases was observed by Nijmeijer et al. 99 after
exposure to pre combustion mimicked by using hydrothermal conditions. The
hydrothermal stability of the γ-Al2O3 intermediate layer could be improved by doping the
γ-Al2O3 with La2O3 in combination with phosphate bonding between the α-Al2O3 support
and the γ-Al2O3 intermediate layer. Replacing the γ-Al2O3 by other mesoporous materials
such as titania 82,100-102 or zirconia 82,103 is recommended for further studies.
28
97
98

2 Theoretical background
2.3.3.2 Microporous transition metal oxide membranes
In contrast to sol-gel derived non-transition silica membranes, the formation of
microporous transition titania or (yttria stabilized) zirconia is difficult due to the fast and
hard to control hydrolysis and condensation reactions of Ti or Zr-precursors. 81 The fast
reactivity can be a result of the larger charge density of Zr or Ti compared with Si 76 or
by the heterogeneousity of the commercially purchased alkoxide precursors. 80,81 Several
groups have reported the synthesis of microporous YSZ (yttria doped zirconia) 78,81,83,104-
111 or titania 78,82,83,100-102,104,112,113 materials, however they were not successful in
producing membranes with a significantly larger permselectivity than the Knudsen factor
for the gasses H2, N2, CO2 and CH4 as a result of the pore sizes of ~0.9 nm and larger. 114
2.3.4 Selection of Sol-Gel derived membrane material
Microporous sol-gel derived silica membranes are found nowadays industrial in
pervaporation applications. Lab scale sol-gel derived membranes offer good gas
separation factors which is mainly based on the molecular sieving compared with
sorption properties. Good prospects are therefore envisaged for applications such as
dehydrogenation of hydrocarbons, coal gasification, the postcombustion concept and the
water-gas shift process in the precombustion concept.
Modified sol-gel derived non-transition metal oxide silica membranes have not found
their way into the large scale gas separation. An alternative method to obtain
hydrothermal stable microporous membranes is to develop non-silica sol-gel derived
materials with the following requirements:
o Separative to H2 from CO2 or CO2 from N2 by means of molecular sieving
o Defect free microporous layers
o Comparable permeabilities to silica membranes
o Stable in hydrothermal conditions
2.3.4.1 Microporous transition metal oxide membranes
Sol-gel derived microporous crystalline titania membranes are commercially today
available with a pore size of 0.9 nm. 101 Sol-gel derived microporous crystalline (yttria
doped) zirconia material is prepared at lab scale. Both, titania and zirconia, sol-gel
derived transition metal oxide materials are being increasingly studied due to a more
controllable sol-gel synthesis using precursor modifiers. 78-80,82,83,110,115
Crystalline materials are desirable due to their expected thermal and chemical stability.
However, state of the art silica layers are amorphous and contain pore sizes in the range
of 0.5 nm, whereas the titania and zirconia materials tend to crystallise at lower
temperatures and might have larger particles than their silica counterparts, resulting in
larger pores. It is known that a mixture of titania and zirconia has a higher crystallisation
29

2 Theoretical background
temperature, thereby maintaining its amorphous phase rather than the corresponding
single oxides 83,112,113,116 . Structuring microporous material might be beneficial for
increasing the porosity and therefore the permeability.
2.3.4.2 Structuring microporous metal oxides
The structuring of metal oxides (yttria doped) zirconia or titania with long chain primary
amines as structure directing agents (SDA) at intermediate temperatures might form high
permeable and selective membranes. Knowles and Hudson 117 prepared high surface area
(~300 m 2 /g) calcined structured zirconia with the addition of alkyltrimethylammonium
halides to the zirconia sols. The distance between zirconia particles turns out to be
linearly dependent on the chain length of these SDA’s. The potential of these SDA’s as
pore formers on stabilized zirconia has been reported 118 and offers interesting prospects.
Recently, block copolymers have been increasingly used to organize structured
composite solids, because the architectures of the amphiphilic block copolymers can be
adjusted to control the interaction between the inorganic and organic species in the sols.
Stucky 119 and Hon 120 intensively studied these structured zirconia’s with the use of these
non-toxic SDA’s with interesting properties. BASF’s Pluronic P-123 (EO20PO70EO20) in
ZrO2 and F127 (EO106PO70EO106) in 4YSZ (where EO is ethylene oxide and PO is
propylene oxide) are used as SDA’s and resulted in surface areas of 90 and 150 m 2 /g,
respectively.
30

2 Theoretical background
2.4 Microporous membrane characterisation methods
Ultramicroporous membranes have pores with dimensions similar to those of small
molecules (pores

2 Theoretical background
The pore sizes in materials produce a certain shape of isotherm, which can be subdivided
into groups based on the pore size and adsorbate-adsorbent interaction 121 , see Figure 14.
Isotherm Type I presents microporous materials showing a sharp increase of adsorbed
volume at low relative pressures and is referred as micropore filling. At higher relative
pressures a plateau occurs. No monolayer formation is obtained due to the comparable
size of the adsorbed gas kinetic diameter and the pore diameter (dkin~dpore). A monolayer
is formed (point 1 in Figure 14) followed by multilayer formation in mesoporous
materials presented by isotherm type IV (point 2 in Figure 14). Adsorbent dense material
obtaining mono and multilayers is schematically drawn in isotherm type II.
Adsorped volume →
Relative pressure →
Figure 14 isotherms based on the pore size and adsorbate-adsorbent interaction. 121
2.4.2 Mass transport
Microporous membranes usually comprise a stack of layers of different materials, as is
depicted schematically in Figure 15. The thin top layer provides the membrane with its
selective features, whereas the other layers essentially provide mechanical strength.
Transporting of molecules through the stack of layers is influenced by all layers present,
which essentially form a series of resistances to the transport. Multicomponent mixtures
mass transport can be further influenced by mutual interactions between the different
components presented by Benes. 91,122 In this section, emphasis will be on the transport of
single gasses. The permeance is the flux divided by the applied pressure difference over
the layer. The permeance depends, amongst other things, on the nature of the gas, the
pore morphology and the applied conditions.
32

2 Theoretical background
Figure 15 Flux resistance in each layer including defects in the functional membrane layer and in the optional
intermediate layer
Sol-gel derived microporous membrane mass transport is predominantly micropore
diffusion. Mass transport in zeolite membranes is a combination of micropore diffusion
and intercrystalline diffusion. Microporous membranes exhibiting the Knudsen
permselectivity can be a result of mesoporous defects. Macroporous defects such as
defects in the intermediate layer or intercrystalline pores can be identified by observing
the contribution of viscous flow. A simplified overview of transport mechanisms is
outlined in Table 10 and will be discussed below.
Table 10 Transport mechanisms of gasses in different porous media. 38
Transport
mechanism
Pore size and classification 36 Selectivity
Viscous Flow > 20 nm Non-selective
Knudsen Diffusion 2-100 nm
33
=
M
M
2 α ; M1

2 Theoretical background
size, (viscous) properties of the transported species, activity of the pore or particle wall,
applied pressure and temperature, texture and microstructure of the supports such as the
connectivity of the pores, the porosity, the concentration and the connectivity of the
defects, and the thickness of the support and supporting layers.
Viscous flow
Viscous flow occurs when mutual gas molecule interactions are more frequent than
molecule-wall collisions. Convective flow of the gas molecules occurs when a pressure
gradient is applied over the membrane. This transport mechanism is referred as viscous
flow due to the dominant transport factor relating to the collisions between the gas
molecules as a function of their viscosity η. The flux can be described using the modified
Poiseuille type law, 45 as seen in equation (9).
2
ε r P dP
N = −
(9)
τ 8η
RT dz
The term N is the molar flux in mol/m 2 s, r is the pore radius in m, P is the pressure in Pa,
R is the gas constant, and T is the gas temperature in K. The geometric factors are the
porosity ε and the tortuosity τ, in which the latter can present a ratio of the effective
distance that the species travels per membrane thickness. No significant gas selectivity
can be obtained using viscous flow.
Knudsen diffusion
Knudsen diffusion occurs in the regime where the mean free path length of the gas is
largely compared to the pore size, which is typical in mesoporous materials. 45 The
molecule-wall collisions are more frequent than the gas molecule interactions. Knudsen
diffusion is dependent on the molar mass (M in kg/mol) of the gas species (Table 10) and
the thickness of the membrane layer (L in m), see equation (10) (valid for ideal gasses in
cylindrical pores).
ε 2 r 8
N = − ∆P
(10)
τ 3 L πMRT
Defects contribution
Intercrystalline defects in zeolite membranes and defects in sol-gel derived membrane
layers can be large. The combination of both transport mechanisms where the gas
molecule to molecule collisions are in the same order of magnitude as the gas molecule to
34

2 Theoretical background
wall interaction is often referred as transition flow and is presented in equation (11). The
first term in equation (11) originates from the viscous flow and the second from the
Knudsen diffusion equation. Both contributions can be identified from experimental data
(12).
N = −
2 ( Ar P + Br)
ε dP
τ dz
⎛ 1 ⎞ 2 8
A = ⎜ ⎟ B =
(12)
⎝ 8ηRT
⎠ 3 πMRT
The α-Al2O3 support with γ-Al2O3 intermediate layer show gas transport predominantly
by Knudsen diffusion as reported by Benes et al. 91
Micropore diffusion
Several research groups 45,91,122,124 have studied the gas transport in microporous media
such as the well-defined pore size MFI zeolite type membranes and amorphous sol-gel
derived silica membranes. The transport mechanisms in microporous membranes are a
function of the pore size, (condensable) properties of the transported species, activity of
the pore wall, applied temperature, texture and microstructure of the microporous layer
such as the connectivity of the pores, the porosity, the concentration and the connectivity
of the defects, and the thickness of the layers.
The single gas flux in microporous membranes is based on diffusion where the gas
molecule pore wall interaction is more frequent than the gas molecules interaction. The
diffusion flux (N in mol/m 2 s) is determined by the concentration of the gas and the
mobility of the gas to hop from one void (hereafter referred as the pore) to the
neighbouring pore. This theory of irreversible thermodynamics, where the flux of single
gas ‘i' is proportional to the gradient in its chemical potential for microporous
membranes, is thoroughly outlined by Benes. 122 The mass transport in microporous
membranes follows the Henry regime at high temperatures with relatively low pressure
and results in the following equation (13):
~
D
N = ∆
L
( T ) K(
T )
P
∆P represents the pressure difference over the membrane. L is the thickness of the
membrane layer. D ( T )
~
35
(11)
(13)
is the concentration independent single component chemical
diffusion coefficient and is related to the mobility of the gas component. K(T) is the
Henry coefficient. The diffusion flux is a thermally activated process described by an

2 Theoretical background
Arrhenius type function 122 presenting the activation energy to hop from one pore to the
neighbouring unoccupied pore and the isosteric heat of sorption, respectively.
Equation (14) is based on the following assumptions and is suitable for sol-gel derived
microporous membranes:
o The gas molecule position in the pore is balanced by the Van der Waals and
repulsive forces
o Only one gas molecule can occupy a pore
o Mass transport occurs only in the direction perpendicular to the membrane layer
o No external forces (like electromagnetic) are applied on the gas molecules
o Temperature in every pore is similar
o The microporous membrane material obeys the Henry isotherm.
The permeance is the ratio of the flux and the applied pressure difference (∆P), resulting
in equation (14). The permselectivity (Fα) is the ratio between the permeance F of gas ‘i’
and gas ‘j’ equation (15).
~
N D(
T ) K(
T )
F = =
∆P
L
(14)
Fi
F α =
F
(15)
j
36

3 Experimental
3 Experimental
The material synthesis and layer preparation methods are outlined in this chapter. First,
the support and the supporting membrane layer formation are discussed. Next, the
preparation of zeolite materials, post treatments and zeolite layer formations are
described. In section 3.3 the sol-gel synthesis and the coating steps are outlined. The last
section includes the characterisation techniques used to study the bulk material and the
microporous membranes.
3.1 Supports and intermediate layers
3.1.1 Support formation
Macroporous supports are the basis of graded porous membranes. Macroporous α-Al2O3
discs were prepared by colloidal filtration of ultra pure alumina powder (AKP30
Sumitomo Chemical Co., Tokyo, Japan) in HNO3. Alumina powder was mixed with
equal amounts of grams of 0.02 M HNO3 by ultrasonic vibration for 14 min at 200 Watt
(Model 450 Sonifier, Branson Ultrasonics Corp., Danbury, CT). This suspension was
sieved with a 200 µm metal filter. Next, the mixture was filtrated by a Schleicher &
Schuell 800 nm membrane sieve using a vacuum pump for 3 to 5 hours. After drying
overnight, a ‘green’ substrate was obtained (Figure 16).
AKP30
Suspension
HNO 3
Sinter oven
1100ºC 1100ºC 2h +2ºC/min
powder
mixture
Ultrasonic
Grind in 36Øx2.4d
in mm
37
Polish
filter
Vacuum filitration filitration
Vacuum >3h
Drying > 12h 12h
Figure 16 Schematic drawing of support formation by means of vacuum filtration

3 Experimental
Finally, the substrates were sintered at 1100°C for 2 hours with a heating and cooling rate
of 2°C/min. Fired substrates were then grinded into dimensions of 36 mm Ø and 2.0-2.4
mm thickness. The surface roughness was decreased by polishing to avoid pinhole in the
membrane layers. First, the α-Al2O3 supports were roughly polished with a 74 µm
diamond in a metal polishing plate (Ecomet 4 Buehler, Germany) with a pressure of 12.5-
25 kPa for 15 minutes. This step was repeated with a 30 µm diamond in a paraffin plate
with identical pressure and duration. The polished supports were ultrasonically cleaned in
ethanol and fired at 800ºC for 2 hours with a heating and cooling ramp of 2ºC/min. This
resulted in a vague mirror like disc with a thickness in the range of 1.90-2.35 mm.
3.1.2 Intermediate layer material characterisation and layer formation
0.5 M Boehmite sols were coated on the supports under clean-room conditions. Boehmite
sols were prepared by reacting 0.5 mol of aluminum-tri-sec-butoxide (ATSB; 97% purity,
Aldrich) with 70 mol H2O at >90ºC. The ATSB was added drop wise under argon-gas
flow to avoid premature hydrolysis. The temperature of the reaction mixture was kept
>80°C to avoid the formation of bayerite (Al(OH)3). After the ATSB was added, the
mixture was kept at 90°C for 1 h to evaporate the formed butanol. The mixture was
subsequently cooled to 60°C. The acidity was adjusted with 1 M HNO3 (E. Merck,
Darmstadt, Germany) to a pH of 2.8 to break down and prevent agglomeration. 5 During
the synthesis, the sol was stirred vigorously. The peptized mixture was refluxed for 20 h
at 90°C, resulting in a 0.5 M Boehmite sol with a clear white/blue appearance. These sols
remained stable for several weeks.
The Boehmite sol coating in clean-room conditions was performed by dip or spin coating.
The dip coating procedures were similar as those reported by Nijmeijer et al. A dipcoating
solution of 0.5 M Boehmite sol and 3 wt% poly(vinyl-alcohol) (hereafter referred
as PVA, Aldrich) in 0.05 M HNO3 with a 2/3 volume ratio was stirred carefully to
prevent air bubbles. PVA acted as a drying chemical controlling additive and was
dissolved in 0.05 M HNO3 to avoid agglomeration of the Boehmite particles in the
Boehmite sol/PVA mixture. Both solutions were filtered with 800 nm filters (Model
FP030/50, Green Rim, Schleicher and Schuell GmbH, Dassel, Germany) to avoid large
agglomerates.
The polished and cleaned surface of α-Al2O3 supports were coated by bringing it in
contact with the Boehmite sol for a certain time, using an automated dedicated dipcoating
apparatus that includes electronic height adjustment (Nima technologies UK)
under clean-room conditions, see Figure 17. The fixation of the support was performed
by mechanical clips or by a vacuum positioning system.
Spin coating procedures are performed similarly. An excessive amount of Boehmite sol
was applied on the polished surface of α-Al2O3 supports, which is fixed by an integrated
38

3 Experimental
vacuum sample holder of the spin-coater (Delta+80 7 equipment SÜSS Microtec).
Evaporation between zero and 60 seconds was performed before rotating at 500-5000
RPM.
After the membranes were dip or spin-coated, they were dried at 40ºC in a relative
humidity of 60% (climate chamber) and fired in air at temperatures of 600°C for 3 h with
a heating/cooling rate of 1°C/min to obtain the γ-phase of Al2O3. Unsupported γ-Al2O3
bulk membrane material was obtained for gas physisorption and XRD characterisation by
drying the dip-coating solutions and subsequently firing the dried gels at the abovementioned
conditions.
Figure 17 A) Nima technology rotational dip coater and B) SÜSS Microtec spin coater
3.2 Synthesis of zeolite material and membrane preparation
The synthesis of the selected zeolite materials of deca-dodecasil-3R (DDR) and
dodecasil-1H (DOH) are outlined in this section. The synthesis of DDR zeolite is
described as well as the synthesis and post-treatments of pure DOH zeolite. Finally, the
formation of DOH layers is presented.
3.2.1 Hydrothermal zeolite syntheses
A B
Deca dodecasil-3R synthesis
DDR small crystals were synthesised according to the published recipes. 44,60 The recipe
of Gies 60 was reproduced by dissolving the structure directing agent (SDA)
1-adamantaneamine (ADA, Aldrich) in a basic solution of ethylenediamine (EN, Aldrich)
in water. The tetramethyloxysilane (TMOS, Aldrich) was slowly hydrolysed by drop
wise addition into the aqueous solution. Another approach, described by Tomita et al. 62
and den Exter et al., 44 was used. Namely, dissolving ADA in EN followed by rapid
addition of deionised water. This mixture was heated for 1 hour at 95ºC after being
shaken for 1 hour at RT to dissolve the ADA completely. TMOS at 0ºC was added in the
39

3 Experimental
ice cooled solution to decrease the hydrolysis rate of the silica precursor. ADA was kept
dissolved by heating the final solution at 95ºC and pouring this hot solution into a Teflon
liner that was placed in a preheated autoclave (at 100ºC).
A modification of the synthesis route by Tomita and co workers 125 was used. A diluted
silica sol was added to the ADA/EN suspension. This mixture was poured into a Teflon
liner and DDR was synthesised at 140-180ºC for 10-31 days with or without rotation of
the autoclaves.
The 50 ml Teflon liners were filled up to 2/3 with the sol solution. Nucleation of the
Teflon liners was avoided by cleaning with a diluted HF (Aldrich) followed by washing
with deionised water or by diluted NH4F (Aldrich) and H2O. The Teflon liners were
placed in homemade stainless autoclaves and placed in a preheated oven at 150-160ºC for
21-60 days. No DDR layers were attempted to prepare, the focus was on the DDR
material synthesis.
Dodecasil-1H synthesis
The “small crystal DOH synthesis route” using ethylenediamine (EN) as base was
followed by dissolving 1-adamantaneamine (ADA) in an aqueous EN solution.
Tetramethyloxysilane (TMOS) was added drop wise. 50 ml Teflon liners were filled up
to 2/3 with this gel composition. Crystallisation was performed at 200ºC for 1 to 60 days
in autoclaves. The “small crystal DOH ammonia route“ was similar but the aqueous
solution of EN is replaced by 33% NH3 (Aldrich). The crystallisation was performed
statically or dynamically. Dynamic crystallisation was performed by rotating the
autoclaves or by stirring the gel by means of a combination of an oil bath and magnet
stirrer as agitation.
Small amounts of titanium isopropoxide (Alfa Aesar) or titanium n-propoxide (Aldrich)
were added to the silica source or to the final gel composition for the formation of Ti-
DOH.
Seeded DOH syntheses were performed to obtain DOH SDA-free. The DOH seeds were
prepared as described above followed by grinding in ethanol or in acetone. The DOH
seeds were added as nuclei to the gel composition.
3.2.2 SDA removal from dodecasil-1H by post synthesis treatments
The synthesised zeolitic products were washed with H2O and ethanol. Subsequently
several post synthesis treatments have been applied to remove the structure directing
agent (SDA) from the [5 12 6 8 ] cages of DOH.
A. Calcination at 700-900ºC in air at ambient pressure. The heating time change
between 5 hours and 21 days with a heating and cooling ramp of 5 to 10ºC/min
40

3 Experimental
o of as-made DOH crystals
o of as-made DOH crystals synthesised with traces of titanium (Si/Ti~100)
to assist the burnout by catalytic oxidation
o of DOH being ball milled to reduce the crystal size. The DOH crystals
were ball milled for 4 to 72 hours.
B. calcination in pressurised air: Hot Isostatic Pressure (HIP) calcination was
performed with a vacuum sinter oven of Engineered Pressure Systems
International N.V. Belgium. The samples were calcined between 700 and 900ºC
for 5 hours. During this time the air pressure changed between 1 and 200 MPa
with a frequency between 1 and 8 times.
3.2.3 Dodecasil-1H layer formation
The DOH Zeolite layers were prepared on commercially available alumina disks
(Inocermic GmbH Hermsdorf Germany). A 0.5 wt% suspension of DOH nuclei in H2O
was spincoated 1, 3 or 5 times on the substrates with a Delta+80 7 equipment of SÜSS
Microtec (Table 11). The seeded substrates were heated at 600ºC for 3 hours with a
heating and cooling rate of 1ºC/min. The seeded substrates were located horizontally at
the bottom of a Teflon lined steel autoclave. A clear solution with a molar ratio of
100SiO2:11ADA:267EN:7400H2O was added. Finally, the seeded substrates were
subjected to hydrothermal crystal growth at 200ºC for 24-48 hours.
The second attempt of DOH zeolite layer formation was carried out by depositing a
suspension of 0.25 wt% DOH nuclei in H2O on a substrate. The seeds were dried at
200ºC prior to the hydrothermal step. A clear sol solution with a molar ratio of
100SiO2:5.9ADA:69EN:3500H2O was added. The hydrothermal treatment was
performed at 200ºC for 24 hours. The zeolitic layers were washed with ethanol and H2O
and finally calcined at 700ºC for 5 hours with a heating and cooling rate of 1ºC/min.
Table 11 Seeding and coating properties of DOH membranes
Deposition Nuclei Gel composition Crystallisation
Technique # Deposition steps SiO2:ADA:EN:H2O Temperature Time
A 1 100:11:267:7400 200ºC 48h
Spin-coating, 600ºC B 3 100:11:267:7400 200ºC 48h
C 5 100:11:267:7400 200ºC 24h
D 1 100:5.9:69:3500 200ºC 24h
Depositing, 200ºC E 1 100:5.9:69:3500 200ºC 24h
F 1 100:5.9:69:3500 200ºC 24h
41

3 Experimental
3.3 Sol-gel derived TiO2, ZrO2 and binary oxide membranes
Sol-gel derived membranes are prepared using the supports with intermediate layers as
discussed previously. The final titania zirconia and binary metal oxide membrane
materials are prepared from polymeric sols, a process which will be outlined in the next
section. Finally, the formation of the sol-gel derived membrane is described and the
characterisations of these materials are discussed.
3.3.1 Polymeric TiO2, ZrO2 and binary oxide sol synthesis
The formation of a titania or zirconia polymeric sol can be controlled using a precursor
modifier to stabilise the alkoxide precursors. The ketone acetyl acetone and the amine
diethanolamine were selected as precursor modifiers. The sol syntheses in this work are
subdivided in to the ketone acetyl acetone and the amine diethanolamine route. A few
experiments were conducted with an alternative amine: diisopropanolamine.
3.3.1.1 Ketone route
The general procedure for synthesis of the acetyl acetone sols was carried out as follows
(Figure 18). A solution (1:1 molar) of acetyl acetone (Ac) and the corresponding
carboxylic acid was added under stirring to a solution of zirconium n-propoxide (Mateck,
70%) in 2-propanol (Aldrich), with a final molar ratio Ac/Zr ~1.
Figure 18 Procedure followed in the synthesis of different acetyl acetone 8YSZ sols
42

3 Experimental
Another solution was prepared comprising H2O and yttrium nitrate (Treibacher) in 2propanol
to prepare YSZ sols. This hydrolysis solution was added drop wise at 0ºC to the
Zr-solution, and the hydrolysis molar ratio (W) was water/Zr ~4. The final sol contained
variable contents of solids with a nominal molar ratio Zr/Y ~ 85/15. The carboxylic acids
used as catalysts include acetic, propionic, caproic (Ca) and nanonic acid, all of which
were supplied by Aldrich. The different solid contents were obtained by adjusting the
amount of 2-propanol used as solvent. Single oxide titania sols are synthesised using
titanium isopropoxide (Alfa Aesar) instead of zirconium n-propoxide. No binders are
used in order to avoid large pores in the sintering steps.
The structure directing agents (SDA’s) such as, decylamine (C10), cetylamine (C16),
cetyltrimethylammonium (CTA+) and Pluronic F127 are added to zirconia sols with
weight ratios of 0.32, 0.48, 0.73 and 1.2 for C10, C16, CTA+ and F127, respectively.
3.3.1.2 Amine route
Polymeric sols were prepared by mixing a precursor solution and a hydrolysis solution.
The alkoxide (zirconium n-propoxide or titanium n-propoxide Aldrich) was mixed with
with diethanolamine (hereafter referred as DEA) and n-propanol (ACS Reagents Aldrich)
with a molar ratio 1:2:72.6 under argon atmosphere, forming the precursor solution. TiO2
and ZrO2 mixture sols were prepared by mixing the alkoxide precursors, via mixing 10,
25, 50, 66, 75 or 90% of titanium n-propoxide to zirconium n-propoxide. The hydrolysis
solution, consisting of 7-10 mol equivalents (with respect to the alkoxide) of 1.0 M HNO3
(Aldrich) and n-propanol (volume ratio 1.8:50), was added slowly (1.5 mL/min) to the
precursor solution under vigorous stirring in an argon atmosphere. The molar ratio of the
final sol was 1:2:7:120 for respectively the alkoxide, diethanolamine, H2O and
n-propanol.
The effect of acidification on hydrolysis and condensation was investigated by selecting
1.0 M HNO3, 0.1 M HNO3 or H2O. The effect of DEA to stabilise the precursors is
compared to the longer chain diisopropanolamine (hereafter referred as DIPA).
n-Propanol (Reagent plus, 99.7%), DEA and DIPA were purchased from Aldrich and
used as received.
43

3 Experimental
Figure 19 Procedure followed in the synthesis of different amine sols
3.3.2 Drying and sintering of TiO2, ZrO2 and binary oxide bulk material
The polymeric sols were dried in a glass beaker at 80ºC for 4 hours to remove the
propanol and the dried further at 140ºC to the obtain bulk material. The dried powders
were characterised with DTA/TG and element analyses (carbon and nitrogen). The
powders were calcined at 400, 500, 550 and 600ºC for 2 hours with heating rates of
0.4ºC/min. The YSZ bulk materials were further calcined at 800 and 1080ºC for 2 hours.
Organic extractions were performed by mixing 1 wt% suspensions of the gelled sols in a
n-propanol/n-heptane (mol ratio of 2:1) solution. These mixtures were refluxed overnight
followed by conventional washing with H2O and drying.
3.3.3 Microporous TiO2, ZrO2 and binary oxide membrane preparation
The polymeric sols were applied on suitable substrates by spin coating with a rotating
speed ranging from 500 to 5000 RPM or by dip coating. The wet films were dried at
room temperature or in a climate chamber set at 40ºC with 60% humidity for 3 hours.
The films were fired at 400, 450, 500 or 600ºC for 2 hours with heating rates of
0.4ºC/min. These steps were repeated to form a second layer. The films were made on a
microscope glass slip (Chance Propper Ltd.) or on γ-Al2O3 membrane supports described
in the first section of this chapter.
44

3 Experimental
3.4 Characterisation for microporous materials and membranes
3.4.1 Microporous material characterisation
Particle size analysis of the sols was performed by using dynamic light scattering (DLS)
equipment (Horiba Nanoparticle Size Analyzer LB-550) and small angle X-ray
scattering. Prior to analysis, the sols were filtered with a Schleicher & Schuell 800 nm
membrane sieve.
Small angle X-ray scattering
SAXS studies 126 were performed using a NANOSTAR camera (Bruker AXS), fitted with
a rotating anode source, run at 40 kV and 40 mA and filtered to yield only CuKα with
wavelength λ=1.540 Å. The collimation path consisted of 3 pinholes and the primary
beam stop at about 1.05 m from the sample position was chosen at 2 mm in diameter to
allow the smallest scattering vector q~0.006 Å -1 to be obtained. Scattering patterns in
2-dimensionals were collected on a Hi-Star Xenon-filled 1000·1000 wired grid area
detector. Typical acquisition times for the dilute solutions, which were kept for 10 hours,
for statistical reasons, in sealed quartz capillaries of 1.5 mm diameter in the evacuated
sample chamber. Transmissions of the solutions were in the order of 40-50% from
absorbance measurements using a glassy carbon standard which was inserted in the
optical path between the sample and the detector. The spot size of the beam on the
sample was 500 µm. All data was corrected for background scattering of the quartz after
a radially averaging process of the raw data to yield a usable scattering vector range
0.008

3 Experimental
X-ray diffraction
Powder X-ray diffraction measurements were conducted using a Siemens D500
diffractometer with a CuKα radiation. The crystallinity of the sample is calculated based
on the ratio of the crystal area and the amorphous area as displayed by the XRD pattern
which was previously corrected by the background contribution. The total integrated area
(intensity*2Θ) of the XRD patterns is assumed to be an accumulation of background,
amorphous and crystalline intensity in a selected XRD angle range, in this case selected
from 10

3 Experimental
Element analysis
The carbon content of the DOH crystals and sol-gel derived sols and bulk materials was
obtained by combusting the sample with the use of radiofrequency heating (2300ºC) in an
oxygen flow with combusting additives. The quantity of combusted carbon was
determined by means of Infrared (Leco CS 600). Nitrogen content in the DOH crystals
and sol-gel derived bulk materials was determined by thermal conductivity detection, the
inorganic samples were heated in helium flow by means of resistance heating (Leco TCH
600). The metal ion impurities in the DOH zeolite crystals were studied with Inductively
Coupled Plasma with Optical Emission Spectroscopy (TJA-IRIS-INTREPID).
Static corrosion tests were performed in nitric acid and sodium hydroxide at a pH ranging
from 1 to 13 by stirring the suspension of the sol/gel derived bulk materials fired at 400,
500 and 600ºC. Corrosion tests were carried out at room temperature for 8 days.
3.4.2 Microporous membrane characterisation
Scanning and Transmission Electron Microscopy (SEM-TEM)
Surface and cross section SEM analysis were performed using a LEO 1530 (Gemini)
electron microscope. Electron energy dispersive X-ray analysis (EDX) was used for
qualitative analysis of the powder. TEM investigations were performed using a Philips
CM200 or a Tecnai F20 G 2 operating at 200 kV.
Secondary Ion Mass Spectroscopy
The infiltration depth of DEA sols was studied by means of Secondary Ion Mass
Spectroscopy (SIMS). By repeatedly sputtering the membrane surface with argon the
chemical composition could be determined at various depths. Profiles were obtained with
a resolution of ~0.4 nm. The chemical composition was determined to be in an area of
0.01 mm 2 inside a crater of 0.09-0.16 mm 2 excluding edge effects.
Raman spectroscopy
The Raman spectroscopy measurements were made using a Horiba Jobin Yvon
SpexT64000 system equipped with two scanning mechanisms, one for the
foremonochromator and another one for the spectrograph. The SpexT64000 is equipped
with 1800 grooves/mm gratings. The 488 nm line of an Ar + ion laser (COHERENT
Innova® 300C Series) was used for the probes excitation. The spectra were collected in a
backscattering geometry with a confocal Raman microscope equipped with an Olympus
MPlan 50x objective. The detection of the Raman signal was carried out with a CCD
detector operating at -196°C. The probes were also measured using a Raman micro
spectrometer (Horiba Jobin Yvon, model LabRam) using the 514.5 nm excitation line
47

3 Experimental
from an Ar + laser (model 2213) and the 632.8 nm excitation line from a He-Ne laser. The
spectra were collected in a backscattering geometry with a Raman microscope equipped
with an Olympus LMPlanFl 50x/0.50 objective. The acquisition time was set at 5
cycles/20 seconds.
X-Ray Photoelectron Spectroscopy
X-Ray Photoelectron Spectroscopy (XPS) were performed using a XPS/AES
spectrometer based on Physical Electronics components. In-depth profiling is obtained
with a 4kV Ar-sputter gun for 2 min. Quantification of the results was done using the
program MultiPak resulting in at% with 15% relative accuracy.
Mass transport
Gas permeation characteristics of the sol-gel derived membranes were determined in a
setup at the University of Twente and at the University of Eindhoven as schematically
outlined in Figure 20. This was done with pure gasses at pressures and temperatures
ranging from 2 to 5 bars and 100 to 200°C, respectively. In this setup the rate of
permeation is measured at atmospheric conditions using a thermal mass flow controller
(MFC). Feed pressure and temperature were controlled with a mechanical pressure
reducer (PR) and programmable electrical heater. Upon installation of the membranes in
a stainless steel, they were first degassed for at least 60 h at 200°C (heating rate 1 ºC/min)
under helium flush. During all measurements feed pressure, transmembrane pressure and
temperature were logged automatically. All gasses, i.e. hydrogen (H2), helium (He),
nitrogen (N2), argon (Ar) and carbon dioxide (CO2), were used with a minimum purity of
99.999%. Only sulphur hexafluoride (SF6) was used with a minimum purity of 99.8%.
Pressure reducer
PR
P
Membrane in
sample holder
48
Valve V
∆∆∆∆P ∆∆∆∆P
Mass Flow MFC Controller
Figure 20 Schematic outline of the setup for the determination of gas permeation characteristics. P
presents the feed pressure and ∆P the pressure difference over the membrane.

3 Experimental
Additionally, the hydrothermal stability of one of the membranes was evaluated by
measuring of the He and N2 permeation upon exposure to water vapour at 200°C. In a
setup as schematically outlined in Figure 21, the partial vapour pressure of the water in
this setup was controlled with a thermostated reservoir through which He was fed to the
membrane module. At the permeate side water can be collected in a cold-trap (CT) before
the He flow is measured with a thermal mass flow controller (MFC). The hydrothermal
treatment consisted of 3 consecutive 24 hour periods during which the vapour pressure
was kept at 1, 2 and 3 bar, respectively. Upon termination of this treatment the membrane
was degassed under a sweep of ‘dry’ He, while periodically measuring He and N2
permeation in time.
Pressure
PR
reducer
P
Membrane Membrane in
in
sample sample holder
holder
Boiler
Boiler
49
Valve V
∆∆∆∆P
Mass MFC Flow Controller
CT Cold trap
Figure 21 Schematic outline of the setup used for the evaluation of hydrothermal stability.

4 Results and discussion
4 Results and discussion
The first section deals with the support and supporting ceramic membrane layers. The
second section includes the properties and potentials of deca-dodecasil 3R and dodecasil
1H zeolite materials and of the first dodecasil 1H layers. The final section describes the
properties of sol-gel derived TiO2-(Y2O3)ZrO2 membranes including the sol, the
microstructure and the membrane layer characterisation and discussions.
4.1 Supports and intermediate layers
Supports were made of α-Al2O3 material (AKP30) and were prepared by colloidal
filtration as reported by Nijmeijer et al. 87 The supports have a diameter of 36 mm and a
thickness of 2-2.4 mm (Figure 22-A). AKP30 alumina bulk material has a surface area
7.1 m 2 /g measured by gas physisorption, which is comparable with the 6.9 m 2 /g reported
by the supplier. The particle size is 340 nm (D50) and the purity >99.99%. The purity
was analysed by means of X-ray fluorescence resulting in only 8 ppm Ni, 5 ppm Zn and 2
ppm Fe. The porosity of the colloidal filtrated and (1100ºC) fired substrate is 31% with
an average pore size of ~80 nm and is comparable with similar prepared substrate. 87,99
α-Al2O3 SEM surface image is presented in Figure 22-B.
36 mm
A B
Figure 22 A) 36 mm in diameter produced α-Al2O3 support and B) the surface SEM image.
The intermediate layers are prepared with γ-Al2O3 to form layers of good quality to
support the final membrane layers. However, the final membrane layers might be
supported in the future by mesoporous titania or zirconia materials to produce chemically
more stable membranes with matching thermal expansion coefficients with titania or
zirconia functional membrane layers.
The intermediate layers are prepared from Boehmite sols having particle sizes (d50)
typically ~30 nm (Figure 23) and are comparable with results reported in literature. 87,99
50
α-Al2O3

4 Results and discussion
Abundance %
20
15
10
5
0
1 10 100 1000
Particle size in nm
51
A 28nm
B 26nm
Figure 23 Particle size distribution of Boehmite sol measured twice by means of Dynamic Light Scattering
The γ-Al2O3 layer obtained by dip coating is ~1 µm in thickness and can be increased to
approximately 3 µm by repeating the coating and firing steps (Figure 24-A).
Homogeneous γ-Al2O3 intermediate was obtained by means of spin-coating. The
thickness of the twice spin-coated γ-Al2O3 intermediate is ~3 µm (using 1000 or 3000
RPM), indicating that the layer thickness is more dependent on the concentration of the
Boehmite sol and the evaporation time than the actual spinning speed (Figure 24-B and
C). Raman or XRD analysis could not distinguish the γ from the α-Al2O3 phase of the
membranes. The γ-Al2O3 bulk material was analysed with XRD and gas physisorption
after drying and firing treatments.
γ-Al2O3
α- Al2O3
A B
γ-Al2O3
α- Al2O3
γ-Al2O3
α- Al2O3
2-3 µm
Figure 24 Cross section SEM images of γ-Al2O3 intermediate layers by dipcoating (single layer, A) or
spincoating at 1000 RPM (B) and 3000 RPM (C).
The γ-Al2O3 intermediate layer material calcined at 600ºC for 3 hours showed a type IV
isotherm indicating a mesoporous material. XRD pattern results confirm the presence of
pure γ-Al2O3 and the absence of α-Al2O3 and δ-Al2O3. The specific surface area is
derived from the Brunauer, Emmet and Teller (SBET) theory. 130 The Barret, Joyner and
Halenda (BJH) algorithm 130 is selected for estimating the pore size of mesoporous
material that presents type IV isotherms. The BET specific surface areas are in the range
C

4 Results and discussion
of 205-225 m 2 /g with a median pore size of 4.1-4.6 nm (Figure 25). This is comparable
with literature results. 5
VPore / cm 3 g -1
0,3 8
0,2
0,1
102 10-1 102 100 102 101 102 0,0
0
r / nm
Figure 25 Pore size distribution of the γ-Al2O3 powder calcined at 600ºC for 3 hours using the BJH model.
The pore volume was calculated from the absorbed gas for certain mass of powder (cm 3 /g). The green line
presents the differential of the absorbed volume.
52
6
dV/dr / cm 3 nm -1 g -1
4
2

4 Results and discussion
4.2 Zeolite materials and layers
All silica clathrasil 8-ring deca-dodecasil 3R (DDR) zeolite type might be a potential
membrane material to separate CO2 from other gasses. 6-ring dodecasil 1H (DOH), also
an all silica clathrasil, could be an interesting membrane material for H2 to CO2
separation.
First, DDR syntheses results will be described briefly. Second, the successful syntheses
of DOH will be described including the results to obtain quasi SDA-free DOH structures,
followed by a discussion on the formation of secondary grown DOH layers on α-Al2O3
supports.
4.2.1 Deca-dodecasil-3R synthesis
Deca-dodecasil-3R (DDR) is not commercially available but can be synthesised from
silicon precursors which are hydrolysed in the presence of the structure directing agents
(SDA) at high pH. These reactions are performed in Teflon lined steel autoclaves at high
temperatures for several days. Aminoadamantane (ADA) is the most frequently used
known SDA for the formation of DDR.
Dodecasil-1H is one of the competing phases of DDR using the same gel composition. 60
DOH is synthesised at temperatures above 190ºC while DDR requires synthesising
temperatures in the range of 160-170ºC.
Table 12 presents the obtained phases as a function of the synthesis temperature. It shows
shown that DOH is synthesised at higher temperatures. Changing the gel composition did
not result in single phase DDR zeolite material. Different polymorphs of DOH, SGT and
DDR (SGT is zeolite type Sigma 2) are formed at temperatures below 200ºC, as seen in
Figure 26 A and B.
A)
Polymorphs
Figure 26 A) Optical microscope image of polymorph products obtained at 170ºC for 25 days. B) DOH-
DDR polymorphs found by den Exter et al. 67 .
53
B)

4 Results and discussion
Table 12 DDR synthesis conditions presenting the molar ratio and crystallisation time in days
SiO2:ADA:EN:H2O Crystallisation
conditions
Obtained phase
Days Temp. ºC
100:30.7:266:7300 60 160 DDR/SGT
100:30.7:266:7300 25 170 DDR/SGT/DOH
100:11:267:7405 27 200 DOH
The Sigma 2 (SGT) zeolite type coexists in the products synthesised at temperatures
below 200ºC. SGT can be synthesised in the presence of sodium and aluminium as well.
Inductively Coupled Plasma with Optical Emission Spectroscopy found traces of these
compounds. These impurities might act as structure directing agents. SGT in the products
could not be excluded by changing the gel composition (Table 13).
Single phase DDR synthesis seems to be complex. Only a few groups are able to prepare
pure DDR 65,125 who used a well studied synthesis conditions and raw materials.
Table 13 DDR synthesis conditions presenting the molar ratio of the gel composition and crystallisation
time in days synthesised at 160ºC
SiO2:ADA:EN:H2O Crystallisation conditions in days Obtained phase
100:11:267:7405 28 DDR/SGT
100:30.7:266:7300 60 DDR/SGT
100:47:266:3600 33 DDR/SGT/DOH
100:47:404:6800 25 DDR/SGT/DOH
4.2.2 Structure directing agent free Dodecasil-1H synthesis
4.2.2.1 Synthesis of small DOH nuclei
Dodecasil 1H (DOH) can be synthesised successfully using ammonia and
ethylenediamine as the base. The zeolite type SGT (Sigma 2) or a mixture of DOH and
deca-dodecasil 3R (DDR) instead of single phase DOH crystals was obtained when using
the gel composition as reported by Den Exter. 66 Pure DOH was obtained using a gel
composition with less ADA and EN (Table 14).
Pure DOH was synthesised with a 9-18 Si/ADA mol ratio. Decreasing the water content
in the gel may reduce the hydrolysis rate and thereby result in smaller crystals. Partial
crystalline DOH product was obtained when the mol ratio of H2O/Si was reduced from
74 to 10, most likely due to inhomogeneous components in the gel (Table 15).
54

4 Results and discussion
Table 14 DOH synthesis conditions presenting the molar ratio of the gel composition and the crystallisation
time in days.
SiO2:ADA:EN:H2O Crystallisation conditions Obtained phase
Days Temp. ºC
100:47:420:5500 11 200 SGT
100:47:420:5500 11 ~190 DOH/DDR
100:23:267:7405 24 200 DOH+Gel
100:11:267:7405 27 200 DOH
Table 15 Gel composition with molar ratio 100SiO2:xADA:267EN:yH2O syntheses crystallised at 200ºC
static or dynamic agitation with resulting products.
Gel Crystallisation Product
composition conditions
x y days Agitation
5.5 7405 10 Rotation DOH
5.5 7405 20 Rotation DOH
11 7405 24 Static DOH
23 7405 24 Static DOH+gel
11 3000 24 Static DOH
11 1000 21 Static DOH+gel
The time for complete DOH crystallisation can be reduced from 21 to 7 days by rotating
the autoclave. The crystallisation time was further reduced to only 4 days by stirring the
gel in situ (Table 16). Increasing the gel homogeneity clearly decreases the crystallisation
time. The DOH crystallisation under stirring was further studied for the gels, containing
ammonia instead of ethylenediamine. The DOH crystallinity, using the NH3 route, does
not increase after 5 days of stirring (Figure 27) at 200ºC. The smallest DOH crystals, in
the range of 5-13 µm in diameter, were obtained using the NH3 route in combination with
autoclave rotation for 18 days (Figure 28).
Table 16 Autoclave agitation crystallisation conditions using 100SiO2:11ADA:267EN:7405H2O gel
composition at 200ºC as function of crystallisation time
Autoclave agitation days Product
Static 21 DOH
14 DOH
7 DOH + Gel
Rotation 7 DOH
Stir 4 DOH
55

4 Results and discussion
Figure 27 Crystallinity of DOH, using the NH3 route, as function of the crystallisation time with a
crystallisation temperature of 200˚C.
Figure 28 Optical micrograph of synthesised DOH crystals
Aging the gel composition prior to the hydrothermal crystallisation of DOH
The average crystal size, obtained after 4 days of crystallisation at 200˚C, is a function of
the aging temperature of the gel. The smallest DOH crystals, 5-10 µm in size, are
obtained after 4 days of crystallisation of the aged gels. This gel was aged at 80˚C for 24
56
Crystallinity %
5 10 15 20 25 30 35
2Θ [º]
DOH crystals
100
75
50
25
0
12days
NH 3 route
2 4 6
Days
8 10 12
2days
3days
4days
5days

4 Results and discussion
hours using ethylenediamine as a base. Lower aging temperatures resulted in larger
crystals (Figure 29). The crystal size is estimated from several crystals observed with the
optical micrograph. This indicates that the nucleation of primary building units occurs at
moderate temperatures (50-80ºC), which is typical for zeolite formation. At higher
temperatures (200ºC), DOH crystal growth is stimulated. These results suggest that the
DOH nucleation and growth can be decoupled.
The crystal size in the range of 10 µm can be too large to act as nuclei for the formation
of DOH thin layers. It is believed that zeolite membrane layers should have a desired
thickness of less than 20 µm. 133
Crystal size [µm]
160
120
80
40
0
20 40 60 80
Aging temperature [ºC]
Figure 29 Crystal size as function of the aging temperature using the gel composition of
100SiO2:11ADA:267EN:7405H2O. Two samples at each aging temperature of 25, 50 and 80ºC were
prepared. The crystallisation was performed by rotating the autoclave at 200˚C for 4 days. The crystal
sizes were determined with an optical microscope.
4.2.2.2 Preparation of quasi template free DOH
Seeding DOH synthesis
The addition of DOH seeds enhances the DOH crystallisation. However, the
reproducibility of a synthesis with seeds in combination with low concentration of SDA
was proven to be difficult. Nevertheless, low yield of pure DOH was obtained using a
low amount of seeds (0.28-0.60 wt% compared to the total gel mass) in combination with
Si/ADA = 100 molar ratio instead of 9. These results confirm the possibility of DOH
synthesis by secondary growth as published by Grebner et al. 71-73 However, the
57

4 Results and discussion
successful reproduced seeded DOH synthesis still includes significantly amount of SDA.
This could have resulted in almost completely templated DOH. The chemical analysis
confirm this expectation with a 5.5 wt% (0.22 mol%) carbon content. This amount,
related to the theoretically determined SiO2/ADA ratio of 34 in DOH, results in a ~75%
cage occupancy. To summarise, it was not possible to synthesize DOH without the
addition of (small amounts of) SDA by using secondary growth of seeds.
Post treatments
Several post treatments are compared in order to remove the organic template from the
largest DOH [5 12 6 8 ] cages, namely 1) conventional calcinations, 2) conventional
calcinations of DOH synthesised with traces of titanium to assist the burnout by catalytic
oxidation, 3) reducing the crystal size to shorten the path length of the oxygen and
combusted gasses by means of mechanical ball milling and 4) calcination in combination
with pressurised air.
1) Conventional calcinations
The structure directing agent (SDA) occluded in DOH could not be removed completely
by means of conventional calcination in air under ambient pressure and at temperatures
between 700 and 900ºC. This is in line with other published results. 66,71-73 If all large
cages are occupied by an ADA molecule the organic compounds amount to a total of
6.9 wt.%. A complete SDA removal is observed at about 1400ºC, which is coupled with
DOH structure breakdown (Figure 30). The XRD patterns of the ‘as-made’ and the
calcined sample at 900ºC for 5 hours clearly show a broad diffraction around 22° 2Θ
indicating amorphous silica (Figure 31). This typical crystallinity loss after calcination is
coupled with an increase in the intensity for the low angle and is typical of emptying
zeolite cages. Several samples suffer significant crystallinity loss after calcination.
58

4 Results and discussion
Intensity (a.u.)
DTA in µV/mg
0,05
0,00
-0,05
-0,10
-0,15
-0,20
-0,25
exo
200 400 600 800 1000 1200
85
1400
Temp [ºC]
Figure 30 DTA/TG results of DOH synthesised by means of the NH3 route (5ºC/min)
10 15 20 25 30
2Theta [º]
59
As-made
900ºC /5 hours
Figure 31 XRD pattern of DOH sample as-made (black) and calcined at 900ºC for 5 hours (grey)
The cage occupancy is calculated from the experimentally determined Si/C molar ratio of
the DOH samples. 34SiO2:1ADA:xN2 69 is the chemical composition of DOH. C10NH17 is
the chemical composition of ADA. The cage occupancy is 100 % when the Si/C ratio
is 3.4.
105
100
95
90
TG weight %

4 Results and discussion
The coupled loss of SDA and crystallinity during calcination might be explained by the
collapse of the DOH surface structure. While the inner part of the DOH crystals
maintains their crystal structure including the trapping of some of the template and the
outer part of the DOH crystals collapses. In principle, O2 with a kinetic diameter of 3.46
Å, CO of 3.70 Å and CO2 of 3.30 Å are inaccessible to the DOH structure due to the
smaller window opening compared to the kinetic diameter of these gasses. However, the
DOH structure might ‘breathe’ (elastic vibration) at elevated temperatures allowing the
transport of these gasses. This behaviour has been observed for other zeolites. 134
The remains of the burned out ADA might react with the oxygen from the silica
framework at elevated temperatures. This could explain the decrease in cage occupancy
coupled with the DOH structure loss. This might occur preferentially at the outer surface
of the DOH crystals. SDA combustion is studied by DTA/TG measurements in air and in
nitrogen. DTA/TG plots show similar behaviour for both gasses indicating that the
weight loss is independent of the atmosphere and could explain the internal combustion.
In a nut shell, conventional calcination at ambient pressure removes only a part of the
organic SDAs.
2) Catalytic oxidation by titanium incorporation in the DOH framework
Titanium was incorporated into the DOH framework in order to enhance the conventional
calcination by means of catalytic oxidation. Small amounts of titanium isopropoxide or
titanium n-propoxide were added to the silica source or to the final sol composition.
Similar attempts have been performed to incorporate iron or aluminium and have resulted
in multi phase zeolite products.
Highly crystalline Ti-DOH products are obtained after ~27 days of crystallisation at
200ºC in the presence of a small amount of titanium in the gel (molar ratio Si/Ti 10-∞,
see Table 17). Electron energy dispersive X-ray analysis and Inductively Coupled Plasma
analysis confirm the presence of titanium in the samples. Raman spectroscopy presents
bands at 960 cm -1 indicating the presence of Ti-O-Si bonds.
Table 17 Cage occupancy of as-made and after calcination for samples with different Si/Ti molar ratios
Sample # Crystal size Si/Ti Cage occupancy
As-made Calcined at 900ºC
Ti-1 Ca. 100 µm ∞ 91% 31%
Ti-2 100 76% 24%
Ti-3 Ca. 40 µm ∞ - 11%
Ti-4 100 - 4%
60

4 Results and discussion
The carbon content (≡ cage occupancy) after calcination is lower if titanium had been
present in the gel composition. In addition, more cages are emptied after the calcination
at 900ºC when the crystallisation is performed for smaller DOH crystals. This is indicated
by comparing the samples Ti-2 with sample Ti-4 (Table 17). The reduced path length
(smaller crystals) the combusted species have to travel until they are expelled from the
crystal show a significant contribution to the final cage occupancy.
A broad exothermic peak in the temperature range of 300 to 600ºC is observed for the all
silica DOH sample (Figure 32). The addition of titanium to the gel resulted in an even
broader exothermic peak extending to about 900ºC. The TG results are in agreement with
the DTA results. The addition of titanium results in weight loss of ~5 wt % in the range
of 100-900ºC which is a factor 2 higher compared to the undoped DOH.
DTA in µV/mg
0,0
-0,1
-0,2
-0,3
-0,4
-0,5
exo
D1H
Ti-D1H
100 200 300 400 500 600 700 800 900
92
1000
Temp [ºC]
Figure 32 DTA/TG plots of DOH as function of titanium incorporation.
The decrease in remaining carbon and the larger weight loss as observed by TG suggest
that the small amount of titanium which was added to the synthesis gel has an influence
on the calcination product. Whether this can be explained by catalytic oxidation is still
unclear. This hypothesis should be studied in more detail.
3) Calcination of Ball milled DOH
The smallest synthesised DOH crystal size is in the order of 10 µm. Reducing the crystal
size is attractive for SDA removal and beneficial for the formation of thin DOH layers.
The expulsion of the organic remains from the D1H crystal during the calcination could
61
108
104
100
96
TG weight %

4 Results and discussion
be enhanced by a shorter path length. The path length can be reduced by grinding the
as-made crystals into smaller particles. Ball milling was used as the grinding method.
The crystal size obtained by optical microscopy reduces two orders of magnitude to
1-10 µm by ball milling for 0 to 26 hours (Figure 33).
Crystal size [µm]
120
100
80
60
40
20
0
0 5 10 15 20 25 30
Ball mill in hours
Figure 33 Crystal size of large synthesised DOH as function of ball milling time.
~1 µm DOH crystal can be observed after ball milling for 72 hours (Figure 34-A and B).
Decreasing the crystal size by ball milling is coupled with a significant crystallinity
decrease during the 72 hours of ball milling (Table 18). Attempts to hydrothermally
recrystallise the ball milled products were unsuccessful. Treating the ball milled DOH
products with aqueous NH4F did not separate the crystalline and amorphous phases.
Calcination of the 72 hours ball milled sample at 900ºC for 5 hours did not result in
further crystallinity loss. The ~35% crystalline DOH product did not contain any carbon.
Once again, the peak intensity at the low diffraction angle increased after calcination
indicating the presence of empty cages (XRD pattern not shown). The few DOH crystals
left are SDA free. This indicates that the template removal is strongly dependent on the
crystal size. The quantity of expelling of the combusted SDA is dependent on the path
length of the transported species should travel.
62

4 Results and discussion
DOH crystals
Figure 34 A) Optical microscope image of as-made DOH crystals and B) SEM image of the 72 hours ball
milled DOH crystals
Table 18 Crystallinity loss and cage occupancy of as-made and calcined DOH as function of ball milling
times.
Sample # Ball milled Crystallinity Cage filled
Before
calcination
63
Grinded
DOH crystals
A B
Calcined
at 900ºC
Asmade
Calcined
at 900ºC
BM-1 0h 72% -
72h 32% 35% - 0%
BM-3 0h 93% 82% 83%
16h 97% - 49%
22h 71% - 7%
Figure 35 presents the XRD patterns and its crystallinity (samples BM-2 in Table 18) of
DOH crystals ball milled for 0, 16 and 22 hours. Significant crystallinity is maintained
after 16 to 22 hours of ball milling. The cage occupancy decreases sharply after 16 hours
of ball milling (Inlay Figure 35). Again, reducing the crystal size seems to be coupled
with crystallinity and SDA loss. Partly SDA-free ~71% crystalline DOH could be
obtained after ball milling for 22 hours and calcination of 900ºC for 5 hours. A broad
particle size distribution is observed (5-125µm) by means of optical microscopy. The
largest remained DOH crystals could contain some of the SDA and could explain the
small carbon concentration.

4 Results and discussion
10 15 20 25 30 35 40
2Theta [º]
Figure 35 XRD pattern of DOH as-made, ball milled for 16 and 22 hours. Inlay presents the cage
occupancy after calcination at 900ºC for 5 hours.
4) DOH calcination at high pressure
The SDA removal from D1H crystals might be easier if free oxygen molecules were
present in the cages to support the destruction of the SDA. At higher air pressures during
calcination it might be possible to force oxygen molecules into the cages which then may
react with the remains of the combusted SDA by forming e.g. CO or CO2. A pressure
swing process could in particular enhance the expulsion of the SDA by enhanced oxygen
penetration into the cages at high pressure levels followed by the outgassing of the
reaction products - CO and CO2 at low pressure level (1 MPa). It is expected that the gas
transport might be a function of the environmental pressure.
Calcination at 700ºC for 5 hours in combination with 10 times 10 MPa air (Figure 36) on
D1H crystals resulted in a strong reduction in cage occupancy while maintaining high
crystallinity (Table 19). In contrast, only a slight cage occupancy loss was observed when
the sample was exposed to 700ºC for 5 hours with atmospheric pressure. The Hot
Isostatic Pressure (HIP) cycling enhances the SDA removal. However, the pressure,
temperature or time might not have been enough to obtain the low cage occupancy of
11 % as has been observed after calcination at 900ºC for 21 days (Table 19).
64
Cage occupancy [%]
80
60
40
20
0
0 5 10 15 20 25
ball milling in hours
As made
16h Ball milled
22h Ball milled

4 Results and discussion
900 [ºC]
700 [ºC]
Calcination temp.
Time
5 hours
65
200 MPa
Pressure
50 MPa
10 MPa
Figure 36 The HIP calcination program. Solid lines present the temperature profile and the dashed lines the
applied pressure of technical air
Table 19 Crystallinity loss and cage occupancy of Hot Isostatic Pressure calcined D1H samples. All the
HIP calcinations were performed for 5 hours with heating and cooling ramps of 10ºC/min.
Temperature Pressure Crystallinity loss Cage occupancy
As made - 75 - 78%
700ºC/5h - Ca. 14% 67%
700ºC/5h 10x10 MPa Ca. 1% 15%
900ºC-21 Days - Ca. 13% 11%
900ºC/5h - Ca. 9% 64%
900ºC/5h 2x50 MPa Ca. 13% 1%
900ºC/5h 200 MPa Ca. 23% -
Increasing the temperature and pressure to 900ºC and 200 MPa for 5 hours resulted in
significant crystallinity loss (Table 19). The air pressure of 200 MPa might have been
sufficient to destroy the D1H crystals partly.
Decreasing the pressure to 50 MPa and applying this pressure twice at 900ºC for 5 hours
resulted in very low carbon content and maintaining the high crystallinity (Figure 37).
HIP calcination using a pressure cycle of 50 MPa on templated D1H crystals results in
quasi-SDA free D1H. Again, a peak intensity increase is observed in the low angle scale
indicating free (or nearly free) zeolitic pores (Table 19). All the samples used in this HIP
calcination are in the range of 5-40 µm.
The results suggest that the gas transport of penetrating oxygen and combusted gasses is
enhanced by the pressure swing process at elevated temperatures. This quasi SDA-free
D1H is a potential membrane material with a well defined pore size of 2.8 Å and might

4 Results and discussion
be able to selectively transport H2 (or He) from gasses with kinetic diameter larger than
3 Å at moderate temperature.
Cage occupancy [%]
80
60
40
20
0
As made 900ºC 21 days 900ºC 2x50MPa 5h
Figure 37 the cage occupancy of small crystal D1H samples that are as-made, calcined at 900ºC for 21 days
and calcined at 900ºC for 5 hours including 2 times 50 MPa pressurised air.
4.2.3 Dodecasil-1H layer characterisation
Thin zeolite films can be prepared on α-Al2O3 substrates 58 by using direct coating or by
the so-called secondary growth method to form zeolitic membranes. Secondary growth is
preferred to direct growth for DOH layers due to the fact that direct growing can result in
a low crystalline DOH or in large DOH crystals. Besides, nucleation and growth are
decoupled in secondary growth which could result in homogenous thin layers. No zeolitic
layer was observed after 24 hours by direct coating using a standard DOH synthesis sol
as described in the first section of DOH growth.
Spin coating a substrate with a suspension of DOH nuclei resulted in random oriented
DOH crystals. Small DOH crystals that were calcined at 900ºC for 5 hours were selected
to act as DOH nuclei. DOH nuclei were grown hydrothermally to form a layer in an
autoclave at 200ºC for 48 hours. The autoclave contained a gel composition
100SiO2:11ADA:267EN:7405H2O (in molar ratio). This gel composition, including
SDA, was chosen to get pure DOH. Though, SDA free gel solution is advisable to avoid
calcination at elevated temperature. SDA calcination of DOH layers on top of α-Al2O3
substrates could result in cracks due to thermal expansion coefficient mismatches.
The hydrothermal secondary growth resulted in island growth (Figure 38-A) or in DOH
layers which are too thick for membrane applications when the support was spin-coated 5
times. The remaining product obtained during the hydrothermal secondary growth was
66

4 Results and discussion
characterised with XRD which showed highly crystalline DOH (XRD pattern not shown).
Microscopic images clearly show hexagonal crystals on the surface (Photos not shown).
A B
Figure 38 DOH Zeolite layer by spin coating (A) followed by hydrothermal growth. B is deposited wet
followed by hydrothermal growth and finally calcined.
A similar suspension of DOH nuclei in H2O was deposited and subsequently dried at
200ºC for 2 hours for seed attachment. The DOH nuclei on the substrates were grown
hydrothermally in a 100SiO2:5.9ADA:69EN:3506H2O (in molar ratio) gel composition at
200ºC for 24 hours. This gel composition was selected as described in 125 for preparation
of thin DDR membrane layers. DDR layers are grown with this gel composition at 160-
170ºC, DOH layer crystallisation temperature is selected at 200ºC. Finally, the washed
membranes were calcined forming a brown layer like Figure 38-B.
Figure 39 presents the surface of membrane B. 10-20 µm hexagonal DOH crystals are
clearly present.
Figure 39 Optical microscope image of DOH membrane A’s surface
Figure 40-A presents a cross section of DOH membrane B consisting of an α-Al2O3
support with the delaminated DOH layer on top. Some areas indicate attachment between
67
Hexagonal
DOH crystals
50 µm

4 Results and discussion
the α-Al2O3 support and the DOH layer (Figure 40-B). The delamination is likely a result
of the stresses occurring during the heat-treatment of 700ºC.
The thickness of the DOH layer is in the range of 50 to 80 µm. No DOH nuclei can be
distinguished which could have grown into crystals of 10-20 µm in size (Figure 40-C).
The thickness/width ratio is about 10, indicating thin plates. Figure 40-C presents a
highly porous zeolitic area close to the α-Al2O3 surface and on the membrane surface. A
more dense area is observed between these porous layers and is magnified (Figure 40-D).
The intercrystalline channels transport liquids despite the dense areas in the zeolitic layer.
A B
DOH
α-Al2O3
C D
DOH
Figure 40 Cross section SEM images of DOH layer on α-Al2O3 support. A) presents the full thickness of the
membrane. The zeolitic layer is delaminated. B) higher magnification resulting. C) orientation of the zeolite
crystals on the α-Al2O3 surface and D high density area of the zeolitic layer.
The DOH crystals are present on the α-Al2O3 surface. The alumina layer did not contain
any silicon and the zeolitic layer was free of any aluminium. This indicates that all-silica
DOH can be prepared on alumina supports without interfering.
Striking is the crystallisation time if compared to other synthesis attempts. While a
crystallisation time with a minimum of 4 days is required to form highly crystalline DOH
68
DOH
α-Al2O3
DOH

4 Results and discussion
without a substrate, D1H layers are obtained after only 24 hours. This offers prospects for
making thin homogeneous DOH layers in relatively short time. Nevertheless, the DOH
crystals are too large thereby forming thick layers. The recommendation is to optimise
the DOH crystal growth before further studies on DOH layers and membranes.
69

4 Results and discussion
4.3 Sol-Gel derived membranes
TiO2, ZrO2 and the binary oxides TiO2-ZrO2 and Y2O3-ZrO2 (hereafter referred as YSZ)
can be prepared by means of the sol-gel technique. Generally, the hydrolysis of highly
reactive Ti and Zr-precursors is controlled with precursor modifiers such as ketones and
amines as described in the chapter ‘Theory’.
Y2O3 or TiO2 doped ZrO2 sol-gel derived materials and membranes were studied and
prepared. A thorough study has been performed on polymeric YSZ sols using ketone
acetyl acetone as precursor modifier to prevent a full hydrolysis of the alkoxides in the
presence of pore formers. This approach has been concentrated on the sol and bulk
material properties. The first results obtained for YSZ layers using this approach
(hereafter referred as ketone approach) are discussed.
The second approach is based on the promising diethanolamine precursor modifier
outlined by Van Gestel et al., 82,104,135 Spijksma et al. 80,81 and Aust et al 83 (hereafter
referred as amine approach). This work is concentrated on the study of (TiO2-doped)
ZrO2 powder and membranes using diethanolamine as the precursor modifier.
Diethanolamine in the sol synthesis combines the prevention of fast hydrolysis of the
alkoxides and the higher viscosity (than acetyl acetone), which is beneficial to the coating
procedures. Diisopropanolamine can be an alternative to diethanolamine precursor
modifier.
Acetyl acetone Diethanolamine
The selection of the membrane layer material is mainly based on the microstructure
property results of these bulk materials. Several (TiO2 doped) ZrO2 membranes have
been prepared and characterised. Finally, the permeability, permselectivity and
permeability as function of hydrothermal stability will be discussed.
4.3.1 Polymeric sol characterisation
The 8 mol % Y2O3 doped ZrO2 (8YSZ) sol particle size, viscosity and stability have been
studied for the ketone approach. The sol composition of 25, 50, 66, 75, 90, 100% molar
ratio TiO2 in ZrO2 have been prepared by the amine approach.
70

4 Results and discussion
4.3.1.1 Ketone approach
A summary of the stability of the final 8YSZ sols obtained while simultaneously varying
the type of carboxylic acid and the solid content is depicted in Table 20. A sol is
considered stable when a sol is clear for several months at room temperature. It is clear
that the higher the aliphatic chain of the carboxylic acid, the higher the sol stability. In
fact, when nanonic acid was used, the obtained sols were stable in the whole range of
solid content. Since the carboxylic acid acts as hydrolysis/polymerization catalyst, this
behaviour can be explained considering that the lower acid strength of long-chain acid
reduces the reaction (growth) rate, decreasing in turn the formation of bigger particles
and agglomerates. Bigger particles and agglomerates can destabilize the sol due to
sedimentation. These bigger particles are avoided using long-chain acids. Moreover,
another effect to be taken into account is when the acid chain length is increased, the
8YSZ nano-particles are somehow better protected or capped, 136 reducing the possibility
of agglomeration (Figure 41-A and B).
Table 20 Summary of the screened synthesis compositions showing influence of solid content wt% and the
type of carboxylic acid on the final sol stability. Shadowed cells represent unstable sols while white ones
represent stable clear sols (stable for several months). Fixed molar ratios of the sols include Ac/Zrpropoxide
~1, water/Zr-propoxide ~4 and acid/Ac ~1. § Maximal 8YSZ wt% content was limited by the
carboxylic acid solubility, being the maximum values: 20% (acetic), 19% (propionic), 18% (caproic) and
17% (nanonic). Unstable sols start precipitating after a few days or weeks.
Solid Content,
8YSZ wt%
2.5
5.0
10
15
Max.
§
71
Carboxylic Acid
Acetic Propanoic Caproic Nanonic
Unstable Sols
Stable Sols
On the other hand, the sols with a higher solid content show a larger average particle size
(Figure 41-C) and are more stabile. The sols have been analysed with the dynamic light
scattering technique immediate after the synthesis was finalised. This effect is attributed
to the higher concentration of protecting/complexing agents (acetyl acetone and the
corresponding carboxylic acid), although the concentration of particles is higher and the
collision probability too. 5wt% structure directing agent added to stable sols showed a
similar particles size distribution with an average particle diameter of 4.5 nm and a
standard deviation (δ) of 1.9 nm (Figure 41-D). Ketone sols with 0, 50, 75, 100 mol %

4 Results and discussion
TiO2 in ZrO2 showed similar particle size distributions. The particle size does not seem to
be influenced by the type of metal organic precursor.
For the whole set of sols, the viscosity was measured in a wide range of shear rates (see
Figure 42). The nature of the carboxylic sols, apparently, does not influence the viscosity
while the solid content does in an exponential manner. For highly concentrated sols (~ 20
wt% 8YSZ), the viscosity reaches a maximum value of 12.5 mPa•s. Likely, the drying
rate is a function of the viscosity which increases when the solid content is decreased. In
fact, the difference among the different sols given an acid type is only the amount of 2propanol
used as solvent. The drying rate is decreased when the chain length of the
carboxylic acid increases. This rheological and drying behaviour has a direct effect on the
coating quality obtained, done by manual dipping of glass slips, since apparently the best
layers (based on the sol stability and viscosity) are obtained with concentrated sols and
those synthesised with caproic or nanonic acid. These pictures are not shown. However,
low-concentrated sols are needed to form crack-free layers consisting of small particles.
Generally, high concentrated sols results in cracked layers, possible due to too thick
coatings. Thus, low-concentration caproic- or nanonic (pelargonic)-catalysed sols are the
most promising coating materials.
Solid Content YSZ (% YSZ)
20
15
10
5
A
NH 2
O
OH
OH
O
OH
O
OH
O
OH
O
O
OH
0 4 8 12 16
Particle Diameter (nm)
C
Acetic Acid
72
Carboxylic Acid
Aminocaproic
Oleic
Linoleic
Pelargonic
Caproic
Acetic
8YSZ+F127
8YSZ
YSZ 5%wt
0 4 8 12 16
Particle Diameter (nm)
0 4 8 12
Particle diameter (nm)
Figure 41 (A) carboxylic chains. Particles size of the sols as function of the carboxylic groups (B), (C) Particle
size of the sols as function of the sol content and (D) the influence of the 5wt% structure directing agent on the
particle size.
D
B

4 Results and discussion
Solid Content,
8YSZ wt%
0
2.5
5.0
10
15
Max §
73
Carboxylic Acid
Acetic Propanoic Caproic Nanonic
12 6.6 3.2 2
Viscosity (mPa·s)
Figure 42 Viscosity of the obtained sols as a function of the type of carboxylic acid and the solid
concentration.
4.3.1.2 Amine approach
The synthesis of sols from diethanolamine (DEA) modified precursors results in 0, 25,
50, 66, 75, 90, 100 mol% TiO2 in ZrO2 sols that are stable at least for several weeks. The
particle sizes of these DEA sols as measured by means of dynamic light scattering (DLS)
and small angle X-ray scattering (SAXS). These results indicate that all sol particles are
similar in size (2-5 nm) and that their size does not depend significantly on the Ti to Zr
ratio (Table 21-Figure 43).
Replacing the precursor modifier DEA with diisopropanolamine (DIPA) resulted in
similar average particle sizes (Figure 44).
Table 21 Particle sizes with errors between brackets and fractal dimensions of mol % TiO2 in ZrO2 DEAsols
Sample Concentration Particle size by SAXS in nm Fractal dimension
TiO2 2 wt% 3.8 (±0.5) 1.37 (±0.01)
Ti0.5Zr0.5O2 2 wt% 2.8 (±1.0) 1.04 (±0.01)
Ti0.5Zr0.5O2 4 wt% 3.6 (±1.0) 1.16 (±0.01)
ZrO2 2 wt% 2.0 (±0.5) -

4 Results and discussion
Particle size in nm
8
7
6
5
4
3
2
1
0
8
7
6
5
4
3
2
1
0
DLS
0 25 50 75 100
SAXS
0 25 50 75 100
mol % TiO 2 in ZrO 2
Figure 43 Particle sizes measured by DLS (dark grey) and SAXS (light grey) as function mol% TiO2 in
ZrO2 DEA-sols.
Abundance %
35
30
25
20
15
10
5
50 mol% TiO 2 in ZrO 2
DEA
DIPA
0
1 10 100
Particle size [nm]
Figure 44 Particle size distribution by DLS of 50 mol% TiO2 in ZrO2 with DEA (solid line) and DIPA
(dashed line) as precursor modifier
The particle size measured by DLS or SAXS increased approximately 25% if the solid
content of the 50 mol% TiO2 in ZrO2 DEA-sol was doubled (Table 21). This increase in
the particle size diameter as a function of the concentration is within the error. A
significant difference in particle size is observed between ZrO2 and TiO2 DEA-sols when
only the SAXS results are considered. More measurements are required to explain this
74

4 Results and discussion
difference. The charge density difference between zirconium (0.65) and titanium (0.63)
might explain these results. The higher charge density of zirconium can result in a higher
reactivity of zirconium alkoxides compared to titanium alkoxides.
The fractal dimensions of pure titania sol (1.37) and 50 mol% TiO2 in ZrO2 (1.04) present
linear polymeric sols that are weakly branched, like silica sols (1.1-1.8). 132 The fractal
dimension of pure ZrO2 could not be characterised with SAXS because these particles
were too small (Table 21).
These stable DEA or DIPA sols have similar particle sizes as silica sols. The fractal
dimensions of the DEA sols are comparable with the silica sols that form pores in the
range of 3Å. Therefore, these TiO2-ZrO2 sols with matching silica sols properties might
form ultramicroporous membrane layers.
4.3.2 Microstructure properties of membrane material
The porous properties of dried and extracted or calcined sols are studied with N2-sorption
measurements. The pore size distributions obtained by gas physisorption does not have to
correlate quantitatively with the actual pore size in the supported membranes. This is due
to the fact that supported membrane materials are exposed to capillary forces of the
porous support during the coating and the drying steps.
Gas physisorption results can give trends in changing the pore structure. Moreover, the
pore structure is generally determined by N2-sorption having a kinetic diameter of 0.365
nm and a sorption area of 0.166 nm 2 , which can only be applicable for pores that are
accessible for the penetrating species (dpore>dkin). Pores might be H2 open and
inaccessible for N2 or CO2. This makes this characterization method only partially
suitable for ultra-microporous membrane materials.
4.3.2.1 Ketone approach – 8YSZ
The gas physisorption study on powders obtained from ketone 8 mol% yttria doped ZrO2
(8YSZ) sols is focussed for the sols that include the addition of all mentioned structure
directing agents (SDA); decylamine, cetylamine, cetyltrimethylammonium and Pluronic
F127 (hereafter referred as C10, C16, CTA+ and F127). The specific surface area is
derived from the Brunauer, Emmet and Teller (SBET) theory. 130 The sols with the addition
SDA F127 are studied in combination with acetyl acetone modified precursor with
caproic acid.
The microstructure properties of the 8YSZ powders are listed in Table 22. The pore
diameter was calculated with Horvath-Kawazoe model 131 adapted for cylindrical pores
75

4 Results and discussion
(Saito-Foley 137 ) for the microporous materials (HK-SF). A material is considered to be
microporous when the relative microporous adsorbed volume Vmicro/Vtotal is large in
combination with type I N2-sorption isotherm. The Barret, Joyner and Halenda (BJH)
algorithm 130 is selected for estimating the pore size in diameter of mesoporous material
that presents type IV isotherms.
Table 22 Microstructure properties of 8YSZ powders prepared from sols that are extracted or sols that are
dried and calcined at 450 and 500ºC.
Surfactant Temp. in ºC Crystal size
nm
76
SBET
m 2 /g
Vtotal *
cm 3 /g
Vmicro *
cm 3 /g
BJH Ø
nm
HK-SF
Ø nm
C10 450 - 77 0.096 0.021 3.4
C16 450 - 35 0.410 0.009 2.6
CTAB 450 - 67 0.114 0.017 3.5
F127 450 6.5 106 0.106 0.027 3.6
F127 Extracted Amorphous 106 0.260 0.034 7.4
F127 500 8.5 70
Acetyl acetone and caproic acid added to the 8YSZ sol
F127 500 7.0 49 0.037 0.013 3.5
- 500 5.7 57 0.028 0.015 1.0
* Vtotal = Absorbed pore volume at P/P0 of 0.95 128 and the Vmicro= Absorbed pore volume at P/P0 of 0.02
Sols that contained F127 as SDA result in powders with the largest specific surface area.
No clear trend could be observed on the microstructural properties as function of the
SDA type. The addition of the SDA C16 to the 8YSZ sols results in powders with the
lowest specific surface area and the lowest relative absorped microporous volume (ratio
between Vmicro and Vtotal). The microstructural properties of the C16 sample indicate a
more dense material compared to the others. The N2 sorption isotherms of the samples
with C10, C16 and CTA and F127 added to the 8YSZ sols are outlined in Figure 45-B.
8YSZ sol with F127 added as SDA were extracted or calcined at 450 and 500ºC after
drying. A lower specific surface area (106 to 70 m 2 /g) was observed for 8YSZ sols with
F127 when the calcination temperature was increased from 450 to 500ºC. This indicates
that the material is densifying above 450ºC.

4 Results and discussion
V [cm
ads 3
/g]
150
120
90
60
30
V ads / cm 3 g -1
dV/d2r (cm 3 /g·nm)
60
50
40
30
20
10
0.05
0.04
0.03
0.02
0.01
8YSZAcCa+F
8YSZAcCA
0,0 0,2 0,4 0,6 0,8 1,0
p/p0 0
C16
CTA
C10
F127
0.00
0.1 1 10 100
diameter in [nm]
B1 B2
0
0.0 0.2 0.4 0.6 0.8 1.0
P/P 0
77
F127
C10
CTA
C16
Figure 45 N2-sorption isotherms of different calcined sols A) isotherms of the 8YSZ powders calcined at 500ºC
that contained acetyl acetone and caproic acid in the sols (grey dots) and with the additional SDA F127 (black
dots) 500ºC. B1) Pore size distribution and B2) isotherms of 8YSZ powders calcined at 450ºC with different
SDA’s C10, C16, CTA and F127 added to the 8YSZ sols.
A

4 Results and discussion
The addition of the combination acetyl acetone as precursor modifier and caproic acid as
catalyst to the 8YSZ sol resulted in a decrease of specific surface area (70 to 49 m 2 /g) of
the powders but increased the relative absorbed microporous volume. One could state
that the material is becoming more microporous due to the addition of this combination
of acetyl acetone and caproic acid. Similar powder that were synthesised without the
addition of the SDA F127 resulted in even higher relative absorbed microporous volume.
Figure 45-A clearly indicates the effect of the SDA F127 to 8YSZ sols. The pore
diameter of this powder was estimated at 1.0 nm using the HK-SF method. The 8YSZ
powder that was synthesised with the combination of acetyl acetone and caproic acid is
the most microporous material found in this test. In spite its microporosity, it is believed
that microporous material with 57 m 2 /g will likely result in layers with low permeation.
The uncalcined 8YSZ powder was totally XRD-amorphous, as expected from the
polymeric nature of the particles and the remaining organics embedded and bonded
within the Zr-O / Y-O branches. The 8YSZ sol with the addition of the SDA F127 that
was dried and followed by leaching procedure with n-propanol/n-heptane (see chapter
3.3.2.) is referred as the extracted sample. The extracted sample is almost amorphous but
showed some weak order, deduced from the broad peak at 30º (main peak of the fluorite
structure, see Figure 46). The samples calcined at 450ºC and 500ºC show already the
pattern corresponding to the fluorite (cubic) structure without any other phase impurity
(such as tetragonal or monoclinic zirconia phases), as expected from the stabilization
effect of yttria (8 mol%).
Intensity (a.u.)
25 30 35 40 45 50 55 60 65 70 75
2Theta [º]
78
CubicYSZ
500ºC Acetyl acetone+Caproic acid
500ºC
450ºC
Extracted
Figure 46 XRD diffraction patterns of the 8YSZ powders that were extracted or calcined at 450 and 500ºC.
The upper line presents the 8YSZ powder calcined at 500º with the addition of the carboxylic caproic acid
and acetyl acetone to the sol. All peaks are assigned to cubic ZrO2 structure (JCPDS card #30-1468).

4 Results and discussion
The average crystal size is calculated from XRD diffraction pattern using the Scherrer’s
equation at the peak at ~30º. The average crystal sizes of the 8YSZ powders are listed in
Table 22. The average crystal size increases from 5.7 to 7.0 nm for the 8YSZ powders
that were calcined at 500ºC when the SDA F127 was added to the 8YSZ sol that
contained the combination of acetyl acetone and caproic acid. The larger particle might
have been formed due to the structuring effect of the long polymeric chains of F127. This
effect seems to be enhanced when the combination of acetyl acetone and caproic acid is
not present, and result in an increase of average crystal size from 7.0 to 8.5 nm. This
might be explained by the absence of the hindering of crystal growth due to the capping
effect of the acetyl acetone or caproic acid groups.
The pristine polymeric 8YSZ sols maintain their fluorite structure at any tested
temperature with increasing crystal size from 5.5, 7.0, 13.5 to 72 nm at respectively, 500,
600, 800 and 1080ºC (Figure 47). The increase of average crystal size is likely attributed
to the sintering behaviour of the nano-crystalline particles.
Intensity (a.u.)
20 30 40 50 60 70
2 Theta [º]
Figure 47 XRD diffraction patterns of dried 8YSZ sols calcined at 500-1080ºC with 2h dwell time (heating
and cooling ramps 2 K/min). The inlay presents the crystal size as a function of the calcination temperature.
The XRD data suggest the formation of cubic 8YSZ material at temperature below 500ºC
(Figure 47) and even below 450ºC (Figure 46). All the calcined 8YSZ powders listed in
Table 22 present the cubic phase. So the cubic phase formation is independent on the
organic additives. Likely, the cubic phase formation is determined by the 8YSZ sol
properties.
79
Crystal size [nm]
80
60
40
20
0
500 600 700 800 900 1000 1100
T [ºC]
1080ºC
800ºC
600ºC
500ºC

4 Results and discussion
DTA/TG analyses were performed to determine the crystallisation temperatures and the
temperature where the organics are removed. The organics can be the propoxide groups
attached to the zirconium precursor, nitrate-groups from the yttrium precursor, bonded
2-propanol groups, caproic acid, acetyl acetone and the structure directing agents. Figure
48-A shows the DTA curves from the powders that were prepared from 8YSZ sols, 8YSZ
sols with the SDA 127 and the 8YSZ sols with the SDA 127 plus the combination acetyl
acetone and caproic acid.
Figure 48-A presents the DTA curve of 8YSZ powder and shows a broad exothermic
peak between 150 and 400ºC. Approximately 55 wt% is lost in this temperature range
from the 8YSZ powder when the TG curve is considered (Figure 48-B). This can be
assigned to the gradual burn out of the (bonded) n-propanol, nitrate and propoxide
groups. Exothermic peaks are observed at approximately 160, 200, 270 and 360ºC.
Further analysis with e.g. DTA in combination with a mass spectrometer is required to
indentify these exothermic peaks. Clearly, no crystallisation temperature can be extracted
from this data. 8YSZ powders with F127 added to the sol show a less complex DTA
curve. Less weight loss is observed as a result of drying at 140 ºC for 8 hours instead of
standard 4 hours. The dehydration of the 8YSZ powder that contains F127 is observed at
100ºC by show an endothermic peak. The exothermic peak at 435ºC can not be assigned
to the crystallisation temperature due to the fact that the material is show cubic phase
formation at 400ºC (results not shown). Therefore, both exothermic peaks at 235 and
435ºC are likely related to the burnout of the organic species mentioned for the 8YSZ in
combination with the combustion of the F127. F127 seem to organise the gradual burn
out of the organic species when the 8YSZ powder and 8YSZ powder with F127 are
compared. This effect is no longer visible when the combination acetyl acetone and
caproic acid are added to the sol. Again, a broad exothermic peak between 175 and 410ºC
with a weight loss of 35% is observed. One is unable to distinguish the crystallisation
temperature and the temperature when each of the organic compounds is combusted. The
combination of acetyl acetone and caproic acid seem to be bonded to the inorganic solids
in a broad temperature range.
80

4 Results and discussion
DTA µV/mg
0.1
0.0
-0.1
-0.2
-0.3
-0.4
Exo
8YSZ
8YSZ +F127
8YSZ +F127
+Acetyl acetone
+caproic acid
100 200 300 400 500 600
T [ºC]
A
81
Weigth [%]
100
90
80
70
60
50
40
30
8YSZ
8YSZ +F127
+Acetyl acetone
+Caproic acid
100 200 300 400 500 600
Figure 48 DTA (A) and TG (B) curves 8YSZ dried sols of blank, +F127 and AcCa+F127
T [ºC]
8YSZ +F127
In summary, first 8YSZ powder can be prepared microporously with the addition of
acetyl acetone as the precursor modifier and caproic acid as the condensation catalyst to
the sol. However, the specific surface area of only ~50 m 2 /g of this microporous material
might form membrane layers with too low porosity. This could lead to low
permeabilities, see page 96.
Second, the 8YSZ sol is, in combination with block copolymer Pluronic #F127
mesostructuring the nanocrystals and offers the most interesting mesoporous membrane
material due to its high specific surface area. The block copolymer F127 seems to prevent
densification during the firing steps. The F127 mesostructering of 8YSZ can be
interesting for several applications such as catalytic reactions or membranes that may be
more suited for separating bulkier molecules.
Third, the formation of porous 8YSZ is less effective for extraction compared to regular
calcination. Extraction of the dried 8YSZ sols results in amorphous material with relative
large pore sizes.
Forth, the organics in the sols are gradually released during the calcination. As a
consequence, these ZrO2 coatings should be calcined carefully in the first temperature
period (up to 450ºC) in order to maintain the film morphology unchanged.
4.3.2.2 Amine approach-TiO2
A combination of type I and IV N2-sorption isotherms is obtained when the dried TiO2
DEA sols are calcined at 400, 500 and 600ºC (Figure 49). Similar trends are found in
literature. 83 The BET specific surface area (SBET) of the TiO2 samples decreases with
increasing calcination temperature (Table 23). This trend is coupled with an increase in
the pore size (BJH algorithm 130 ) from 3.8, 6.1 to 8 nm and an increase of crystal size
B

4 Results and discussion
from 11.3, 15.1 to 22.1 nm determined with the Scherer equation for respectively 400,
500 and 600 ºC. The relative micropore volume (Vmicro/Vtotal) decreases from 25 at 400ºC
to 8% at 600ºC. TiO2 microstructure properties trends clearly indicate a densification of
the material with increasing calcination temperature.
Table 23 Microstructure properties of dried TiO2 sols
Calcination
Temp. in [ºC]
Equivalent
1 M HNO3
Crystal
size nm
SBET
m 2 /g
Vtotal
cm 3 /g
Vmicro
cm 3 /g
BJH
nm
140 7 amorphous - - - -
400 7 11.3 216 0.24 0.06 3.8
400 10 - 214 0.26 0.06 3.8
500 7 15.1 90 0.18 0.03 6.1
500 10 - 75 0.16 0.02 6.0
600 7 22.1 36 0.09 0.01 8.0
The effect on the TiO2 microstructural properties was studied by changing the quantity of
the hydrolysis solution. The hydrolysis solution (1 M HNO3) was increased from 7
equivalent to 10 equivalent to the titanium precursor. The microstructure properties are
listed in Table 23. The specific surface area and the relative microporous absorbed
volume of the TiO2 powders calcined at 400 and 500ºC seem be independent on the
increase of equivalent 7 to 10 of the hydrolysis solution. This indicates that the sol
modification is mild as expected from similar average particle size in the sol. An average
particle size of 4.4 nm was found with 7 equivalent 1 M HNO3and 4.1 nm with 10
equivalent 1 M HNO3.
The dried TiO2 sols listed in Table 23 transform from XRD-amorphous to anatase TiO2 in
the temperature range of 140-400ºC. The TiO2-TG plot (Figure 50) shows a weight loss
with three decomposition steps in the interval of 30-600ºC. The weight loss below
~225ºC is assigned to the water desorption and organic removal. The decomposition step
between 225 and 475ºC can be related to the gradual decomposition of the amine ligands
on the outer and inner surface of the material which is believed to be bonded more strong
than e.g. n-propanol. The third step in weight loss is small.
The DTA pattern is complementary with the TG observations. The exothermic peak in
the DTA plot at ~225ºC suggests the removal of loosely bound organic residues. The
second broad exothermic peak (320-360ºC) might be related to the gradual
decomposition of the amine groups in combination with the gradual anatase TiO2
crystallisation.
82

4 Results and discussion
Vads / cm 3 g -1
200
150
100
50
0,0 0,2 0,4 0,6 0,8 1,0
p/p0 0
Figure 49 N2 isotherms TiO2 calcined at 400ºC (◊), 500ºC (○) and 600ºC (□). Open symbols present the
adsorption and closed symbols the desorption points.
Exo
uV/mg
0,2
0,1
0,0
-0,1
-0,2
-0,3
400ºC
500ºC
0
100 200 300 400 500 600
Temperature [ºC]
83
DTA
TG
Figure 50 DTA-TG curves of dried TiO2 sol.
600ºC
A gradual anatase to rutile TiO2 transformation in combination with removal of the
remaining organics might explain the weight stabilisation above ~450ºC. Similar results
have been found by Zou et al. 116 A mixture of anatase and rutile TiO2 is formed when the
120
100
80
60
40
20
weight loss %

4 Results and discussion
dried TiO2 sol is calcined at 600ºC. Xu et al. 112 found rutile phase transformation
temperature higher than 600ºC. This might be explained by the difference in sol synthesis
or 30 minutes dwell time instead of 2 hours.
A combination of rutile and anatase TiO2 was observed at 500ºC when the ketone
approach was applied. DEA seems to prevent the anatase rutile transformation compared
to the ketone approach and might be explained by the stronger bonded precursor
modifier.
TiO2 layers prepared similar as the powder discussed above might have the same
microstructural properties and form crystalline mesoporous TiO2 membranes. However,
the microstructural properties of the layers can differ from the powders due to i) the
formation of thin layers instead of agglomerated particles and ii) capillary forces when a
thin layer is applied on a porous substrate.
10 20 30 40 50 60
2Theta [º]
Figure 51 XRD diffraction patterns of dried TiO2 sols calcined at 140, 400, 500, 600ºC. The crystal size
determined with the Scherrer relationship is presented in the inlay.
4.3.2.3 Amine approach - ZrO2
Type I N2-sorption isotherms are obtained when the dried ZrO2 DEA sols were calcined
at 400, 500 and 600ºC (Figure 52). The specific surface area seems to be stable in the
temperature range of 400-500ºC and decreases at higher calcination temperatures (Table
24). The ZrO2 type I N2 sorption isotherm indicates lower specific surface area values
than TiO2 but are similar to the results published in literature. 116
84
600ºC
Crystal size [nm]
500ºC
30
25
20
15
10
5
0
400ºC
140ºC
400 500 600
T [ºC]

4 Results and discussion
The increase in pore size in combination with a decrease of the relative microporous
adsorbed volume Vmicro/Vtotal (56-6%) with increasing temperature is in agreement with
the results found by Xu et al. 112 The pore size increase seems to be coupled with the
increase in crystal size, which increases from 6.1 at 400ºC to 19 nm at 600ºC (Figure 54).
Similar trend have been observer for the TiO2 powders.
Table 24 Microstructure properties of dried ZrO2 sols
Calcination
Temp. in [ºC]
Crystal
size nm
SBET
m 2 /g
Vtotal
cm 3 /g
Vmicro
cm 3 /g
SF nm BJH
nm
140 amorphous - - - -
400 6.1 74 0.034 0.019 1.5 -
500 8.3 77 0.054 0.019 2.2 -
600 19 6 0.011 0.001 4.0
Vads / cm 3 g -1
60
50
40
30
20
10
500ºC
400ºC
600ºC
85
ZrO 2
0,0 0,2 0,4 0,6 0,8 1,0
p/p0 0
Figure 52 N2 isotherms ZrO2 calcined at 400ºC (◊), 500ºC (○) and 600ºC (□).
Open symbols present the adsorption and closed symbols the desorption points.
The dried ZrO2 sols crystallize to tetragonal ZrO2 in the temperature range of 140-400ºC.
The TG plot (Figure 53) of dried ZrO2 can be subdivided into three steps. The first step
presents the water and loosely bounded organic removal in the temperature range of 50 to
300ºC. A sharp weight loss is obtained in the second step up to 400ºC where the amine
groups leave the inorganic-organic material in combination with crystallisation. No

4 Results and discussion
significant weight loss is observed in the third step where the remaining organics are
combusted. The carbon content measurements at 140 to 600ºC are complementary to
these results (Figure 72).
Exo
uV/mg
0,3
0,2
0,1
0,0
-0,1
-0,2
-0,3
100 200 300 400 500 600 700
0
800
Temperature [ºC]
86
DTA
TG
Figure 53 DTA-TG curves of dried ZrO2 sol.
The DTA results (Figure 53) are complementary to the TG results. The exothermic peak
in the DTA plot at 308ºC suggests the removal of organic residues (Figure 53). More
organics remain in the ZrO2 microporous material in the temperature range of 140-600ºC
compared to TiO2 (Figure 72). The second exothermic peak at 358ºC might be related to
the formation of tetragonal ZrO2.
A combination of tetragonal and monoclinic ZrO2 (tetra/mono >>1) is formed when the
dried ZrO2 sol is calcined at 600ºC (Figure 54). No cubic ZrO2 could be observed as
expected from the mild temperature conditions and absence of a cubic phase stabilisator
such as Y2O3. Raman spectroscopy did observe only monoclinic and tetragonal ZrO2 on
the ZrO2 layers and will be discussed in the next section. The phase transformation from
tetragonal to monoclinic ZrO2 may form gradually in the temperature range of 450-
650ºC.
The gas physisorption, DTA and XRD results on ZrO2 DEA 2 wt% sols that are dried and
calcined at 140, 400, and 500ºC for 3 hours with a heating rate of 25ºC/hours suggest that
these materials might form microporous membranes with rather low permeability due to
the small specific surface areas.
120
100
80
60
40
20
weight loss %

4 Results and discussion
20 30 40 50 60
2Theta [º]
Figure 54 XRD diffraction patterns of dried ZrO2 sols calcined at 140, 400, 500, 600ºC. The crystal size
determined with the Scherer equation is presented in the inlay.
4.3.2.4 Amine approach - Binary oxides Ti1-xZrxO2
The TiO2 addition into the ZrO2 was conducted in the first stage of the powder formation
by mixing the alkoxides so it can be assumed that, after the sol-gel formation, a
homogeneously distributed TiO2 in the ZrO2 network. Transmission Electron Microscopy
- Energy Dispersive using X-ray analysis and Inductively Coupled Plasma
measurements 116 were performed and found similar metal oxide ratios as starting
mixtures supporting this assumption.
Type I N2-sorption isotherms are observed for the Zr-rich binary oxides calcined at 500ºC
(Figure 55). The specific surface area increases and the pore size decrease with increasing
the mol% TiO2 in ZrO2 (0 and 25% in Table 25 and Figure 56-A).
Doping TiO2 with 10 and 25 mol% ZrO2 calcined at 500ºC resulted in combination of
type I and IV isotherms. The relative microporous adsorbed volume Vmicro/Vtotal increases
and pore size decreases with increasing ZrO2 content (Figure 56-B).
All the binary oxides contain significant relative microporous absorbed volume at 500ºC.
This was expected from the linear polymeric sols. The N2-sorption isotherm of
Ti0.5Zr0.5O2 sample calcined at 400ºC is comparable to what has been reported in
literature of Ti0.66Zr0.34O2 80 and Ti0.8Zr0.2O2. 83 The highest specific surface area, the
highest relative microporous absorbed volume and the smallest estimated pore size in this
study is obtained for Ti0.5Zr0.5O2 calcined at 500ºC (Table 25 and Figure 56). XRD
87
600ºC
500ºC
400ºC
Crystal size [nm]
140ºC
25
20
15
10
5
0
400 500
T [ºC]
600

4 Results and discussion
amorphous 50 mol % TiO2 in ZrO2 calcined at 500ºC prepared by the ketone approach
contained significantly less specific surface area, results not shown.
Table 25 Microstructure properties of calcined binary oxide TiO2-ZrO2 DEA sols
Calcination
temperature
mol% Ti Crystal size
nm
SBET
m 2 /g
Vtotal
cm 3 /g
Vmicro
cm 3 /g
SF nm BJH
nm
400ºC 0 6.1 74 0.034 0.019 1.5 -
25 Amorphous 180 0.086 0.052 - -
50 Amorphous 124 0.080 0.055 - -
100 11.3 216 0.24 0.06 - 3.8
500ºC 0 8.3 77 0.054 0.019 2.2 -
25 Amorphous 166 0.086 0.044 1.2 -
50 Amorphous 190 0.109 0.057 1.0 -
75 Amorphous 121 0.158 0.037 3.8
100 15.1 90 0.18 0.03 6.1
600ºC 0 19 6 0.011 0.001 4.0
25 18.6 - - - -
50 20.1 44 0.037 0.01 3.5
90 19.2 - - - - -
100 22.1 36 0.09 0.01 8.0
Vads / cm 3 g -1
100
80
60
40
20
50 mol %
500ºC 0, 25, 50 mol% TiO2
25 mol %
0,0 0,2 0,4 0,6 0,8 1,0
p/p0 0
Figure 55 N2 isotherms Zr-rich binary oxides calcined at 500ºC. 0 mol% TiO2 in ZrO2(□), 25 mol%
TiO2 in ZrO2 (○) 50 mol% TiO2 in ZrO2 (◊).Open symbols present the adsorption and closed symbols
the desorption points.
88
0 mol %
TiO2

4 Results and discussion
V micro /V total [%]
Spec. Surface area [m 2 /g]
50
40
30
20
10
0
200
160
120
80
40
0
0 25 50 75 100
mol % TiO 2 in ZrO 2
Figure 56 Relative absorbed microporous volume of powders calcined at 500ºC with varying mol% TiO2
in ZrO2. B) Specific surface areas of powders calcined at 500ºC with varying mol% TiO2 in ZrO2.
Vads / cm 3 g -1
100
80
60
40
20
500ºC
400ºC
50 mol% TiO2
0,0 0,2 0,4 0,6 0,8 1,0
p/p0 0
Figure 57 N2 isotherms 50 mol% TiO2 in ZrO2 calcined at (◊) 400, (○) 500 and (□)
600ºC. Open symbols present the adsorption and closed symbols the desorption points.
The sol and microstructure properties may be influenced by the acidity of the hydrolysis
solution as Spijksma et al. 80 reported for the Ti0.66Zr0.34O2 calcined at 400ºC. The effect
89
600ºC
A
B

4 Results and discussion
of hydrolysis solution acidity on the microstructure of the Ti0.5Zr0.5O2 bulk material was
studied with gas physisorption.
The fractal dimension of the 50 mol% TiO2 in ZrO2 sol is larger (1.29) with similar
particle size if H2O was used instead of 1 M HNO3 as hydrolysis solution.
Figure 58-A and B show a simultaneous decrease in the specific surface area (190 to
57 m 2 /g) and the relative microporous adsorbed volume (52 to 6%) as a result of a higher
pH of the hydrolysis solution. It should be noted that Spijksma et al. 80 reported results
for Ti0.66Zr0.34O2 calcined at 400ºC that show an inverse relation. In that particular work
an acidic hydrolysis solution resulted in a significant increase of both the average pore
size of the powder and the particle size of the sol.
V micro /V total [%]
Spec. Surface area [m 2 /g]
50
40
30
20
10
0
200
160
120
80
40
0
1 M HNO3 0.1 M HNO3 H2O
Figure 58 A) Relative micropore volume and B) Specific surface areas of powders calcined at 500ºC as a
function of the hydrolysis agent acidity.
DTA and XRD analysis are used to study the crystallisation temperature of the binary
oxides. All binary oxides containing 25 to 75 mol % TiO2 in ZrO2 are still XRD
amorphous at 500ºC and become nanocrystalline at 600ºC. 25 mol% TiO2 doped ZrO2
calcined at 600ºC possess a minor phase of highly disturbed monoclinic ZrO2 and a
higher ratio of tetragonality compared to the undoped ZrO2. The higher tetragonality can
be explained by the incorporation of titanium into the ZrO2 lattice (Figure 59-C). Clearly,
the monoclinic ZrO2 phase is suppressed by the addition of 25 mol% TiO2.
Ti0.5Zr0.5O2 crystallises between 550 and 600ºC (Figure 59-A) to a highly distorted and
disturbed orthorhombic phase similar to the orthorhombic TiZrO4 published in. 112
90
A
B

4 Results and discussion
In pure TiO2, the anatase phase transforms partly to the rutile phase if the calcination
temperature is increased from 500 to 600ºC (Figure 59-B). Apparently, this phase
transformation is already suppressed by the presence of 10 mol % ZrO2. These results
confirm the incorporation of ZrO2 into the TiO2 framework found in literature, 83,112 up to
25 mol % ZrO2 might be dissolvable into anatase TiO2. The hydraulic radius decrease 112
and the broadening of the anatase (110) diffraction peak from 0 to 10 mol% ZrO2 doped
TiO2 clearly indicate retardation of TiO2 densification.
The TG and DTA results of the binary oxides are complex due to gradual organic loss in
a broad temperature range with many exothermic peaks, results not shown. A large
weight loss 50% is observed for these binary oxides in the temperature range of 100-
500ºC and might have masked the endothermic peak of phase transformation. A more
detailed DTA curve can be prepared by applying a long dwell time at temperatures below
the crystallisation temperature.
91

4 Results and discussion
Intensity (a.u.)
Intensity (a.u.)
Intensity (a.u.)
20 30 40 50 60 70
100 mol% Ti
90 mol% Ti
Orthorhombic TiZrO 4
2 Theta [º]
15 20 25 30 35 40 45 50 55
2Theta [º]
92
600ºC
550ºC
500ºC
400ºC
600ºC
Anatase TiO 2
Rutile TiO 2
0 mol% Ti
25 mol% Ti
20 30 40 50 60 70
2Theta [º]
600ºC
Monoclinic ZrO 2
Tetragonal ZrO 2
Figure 59 XRD patterns of A) 50 mol % TiO2 in ZrO2 dried sols calcined at 400, 500, 550 and 600ºC for 2
hours with reference of orthorhombic TiZrO4. B) XRD patterns of 90 and 100 mol % TiO2 in ZrO2 calcined
at 600ºC with references of anatase and rutile TiO2 and C) 0 and 25 mol % TiO2 in ZrO2 calcined at 600ºC
with references of tetragonal and monoclinic ZrO2
A
B
C

4 Results and discussion
In summary, anatase TiO2 contains mesopores in the temperature range of 400 to 600ºC.
The average crystal size and the pore size seem to be coupled and increase with
increasing calcination temperature. Tetragonal ZrO2 is mainly microporous below 600ºC.
However, the ZrO2 powder with a specific surface area of approximately 75 m 2 /g might
result in low permeable membrane layers. Besides, both TiO2 and ZrO2 show phase
transformations in the temperature range of 500-600ºC. This could lead to cracks during
the calcination as a result of thermal expansion mismatches of the phases.
The crystallisation temperature of Ti0.5Zr0.5O2 is between 550 and 600ºC which is
approximately 250ºC higher than single oxides 320-360ºC. The higher crystallisation
temperature could suppress the crystal growth while maintaining the microporosity. The
crystallisation temperature, high porosity, high microporosity and small pore size bulk
material properties of Ti0.5Zr0.5O2 calcined at 500ºC are similar to state of the art SiO2.
This indicates that Ti0.5Zr0.5O2 might form microporous membrane layers as well.
However, amorphous microporous materials might have a lower chemical stability
compared to their single oxide relatives.
4.3.2.6 Amine approach – Diethanolamine / Diisopropanolamine
The precursor modifier diethanolamine (DEA) is replaced in the sol synthesis by
equimolar diisopropanolamine (DIPA) to i) increase the viscosity, ii) better compatibility
with the propanol solvent and iii) prevention of densification by forming higher porosity.
No literature was found on this potential precursor modifier.
No significant difference in the DTA/TG analyses and carbon content results are obtained
when DIPA was used instead of DEA as precursor modifier for the formation of binary
oxides with 25 and 50 mol % TiO2 in ZrO2 up to 500ºC.
Figure 60-A presents type I sorption isotherms of 50 mol % TiO2 in ZrO2 using DIPA,
which have specific surface area trends comparable with 50 mol % TiO2 in ZrO2 using
DEA (Table 26). However, the specific surface areas of 25 and 50 mol % TiO2 in ZrO2 at
400ºC using DIPA are a factor 2 higher than when DEA was used (Table 26 and Figure
60-B). The higher specific surfaces are of dried and calcined DIPA sols may be explained
by the longer organic chain of DIPA compared to DEA.
93

4 Results and discussion
Vads / cm 3 g -1
150
100
50
Table 26 Microstructure properties of dried binary oxide TiO2-ZrO2 DIPA sols
Calcination Temp. Mol% Ti SBET m 2 /g
400ºC 25 334
50 224
500ºC 25 219
50 115
600ºC 50 48
50 mol% TiO2 in ZrO2 DIPA
400ºC
500ºC
600ºC
0,0 0,2 0,4 0,6 0,8 1,0
p/p 0
0
25 mol% TiO2 in ZrO2 A DIPA-DEA 400ºC B
200
94
Vads / cm 3 g -1
150
100
50
DIPA
DEA
0,0 0,2 0,4 0,6 0,8 1,0
p/p 0
0
Figure 60 A) N2 isotherms 50 mol% TiO2 in ZrO2 calcined at (□) 400, (○) 500 and (◊) 600ºC using DIPA. B)
N2 isotherms 25 mol% TiO2 in ZrO2 calcined at 400ºC using DIPA (○) and DEA (□). Open symbols present
the adsorption and closed symbols the desorption points.
Ti0.5Zr0.5O2 calcined at 600ºC is XRD-amorphous when diethanolamine is replaced by
diisopropanolamine in the sol synthesis (Figure 61). The crystallisation temperature of
TiZrO4 prepared by DIPA is delayed and could be explained by a more porous structure
(larger specific surface area).
Acidification of the hydrolysis solution in the DIPA sol synthesis is less effective
compared to the DEA sol synthesis. The specific surface area and relative microporous
adsorbed volume are comparable when the hydrolysis solution is prepared with H2O
instead of 1 M HNO3. These results suggest that the DIPA is better bonded to alkoxide
precursors. Therefore, DIPA might be an interesting alternative to DEA for the formation
of microporous membrane materials.

4 Results and discussion
Intensity (a.u.)
Ti 0,5 Zr 0,5 O 2 at 600ºC
10 20 30 40 50 60 70
2Theta [º]
95
Diethanolamine
Diisopropanolamine
Orthorhombic TiZrO 4
Figure 61 XRD diffraction patterns of 50 mol % TiO2 in ZrO2 dried sols calcined at 600ºC when
diethanolamine and diisopropanolamine are compared.
4.3.2.7 Static corrosion tests
A static corrosion test is preformed as a simplified test to study the chemical stability of
these powders. The results of these tests can be informative but not conclusive on the
hydrothermal stability of membrane layers.
The TiO2, ZrO2 and Ti0.5Zr0.5O2 bulk materials that are calcined at 400, 500 and 600ºC
powders showed to be stable at a pH of 13 for 8 days. The chemical stability of
Ti0.5Zr0.5O2 bulk materials is slightly lower than that of the single oxides at a pH of 1 for
8 days. However, only a few Ti and Zr ppms are dissolved in the acid. The chemical
stability of these microporous inorganic materials in acids indicates that they are
promising candidates for the formation of hydrothermally stable membranes.
4.3.2.8 Microstructure properties comparison by the Ketone and Amine approach
The sol syntheses of the Ketone and Amine-approach are different in the concentration,
pH, raw material and hydrolysing agents. Despite that, the crystallisation temperature and
the microstructure properties of calcined and dried ketone and amine sols are compared.
The amine-approach offered membrane materials with generally higher crystallisation
temperatures, higher specific surface areas and smaller pore sizes. This makes the amineapproach
used in this study more interesting for membrane materials than the ketoneapproach.
However, both approaches can be applied for the formation of microporous
TiO2, ZrO2 or binary oxides. Mesoporous materials were made by templating 8YSZ or by
synthesising TiO2 rich powders. TiO2 – (Y2O3)ZrO2 membranes and membrane layers are
characterised in the next section.

4 Results and discussion
4.3.3 Sol-Gel derived membrane characterisation
Supported 8YSZ layers were prepared from the ketone sols and characterised with
scanning electron microscopy (SEM), transmission electron microscopy (TEM) and
single gas permeance measurement. Similar analyses are performed on TiO2, ZrO2 and
binary oxide layers that were prepared by amine sols. Additional, X-ray photoelectron
spectroscopy (XPS), Raman spectroscopy and Secondary Ion Mass Spectroscopy (SIMS)
measurements are prepared on the latter.
4.3.3.1 Ketone approach - 8YSZ layers
8YSZ was prepared on glass substrates and on graded porous alumina supports (γ-Al2O3).
Figure 62-B shows a coating applied by simple drying of a film composed of 8YSZ sol
using acetyl acetone, caproic acid and block copolymer F127 (YSZ/F127 ~ 1/4 wt/wt). In
this case, the substrate was a microscope glass slip and the calcination temperature was
500ºC for 2 hours. The obtained film has a thickness of about 200 nm, and a fine porosity
can be observed, although the primary particles cannot be recognized due to the
resolution of SEM technique.
A B
YSZ
Glass Substrate
Figure 62 Cross section SEM image of a coated YSZ sol blended with a block copolymer on a glass slip and
calcined at 500ºC.
Two kinds of 8YSZ layers have been prepared on graded porous γ-Al2O3 membranes
being i) 8YSZ sol using acetyl acetone in combination with nanonic acid and ii) 8YSZ
sol using acetyl acetone, caproic acid and F127. A ~50 nm thin 8YSZ layer is obtained on
a γ-Al2O3 membrane after calcining at 450ºC an 8YSZ sol that included nanonic acid.
The surface and cross section of the membrane indicate a homogeneous layer (Figure 63).
96

4 Results and discussion
nanonic YSZ
γ-Al2O3
α-Al2O3
Figure 63 SEM photograph of a coated nanonic YSZ sol α-and γ-Al2O3 substrate and calcined at 450ºC.
Figure 64 presents the TEM study of the supported 8YSZ layer that included acetyl
acetone, caproic acid and F127. The substrate was prepared from α-Al2O3 and is highly
crystalline. The γ-Al2O3 intermediate layer has a thickness of 6 µm (Figure 64-A). The
YSZ functional layer presents a nanocrystalline phase and has a thickness of only 20-30
nm (Figure 64-B). Nanostructuring due to the addition of F127 could not be observed in
this analysis. Clearly thin cubic YSZ layers can be prepared using the ketone approach.
The cubic phase was identified by Raman spectroscopy (Figure 65). The band at 1155
cm -1 can be assigned to the inorganic bonding of ZrO2 with the Al2O3 supporting layer.
A B
6µm γ-Al2O3
Figure 64 TEM analysis of a membrane: electron diffraction pattern of the (A) α-Al2O3 and (B) γ-Al2O3,
and TEM images of (C) a 6 µm γ-Al2O3 intermediate layer and (D) detail of the top 8YSZ layer that
contained F127, acetyl acetone and caproic acid in the sol (20-30 nm in thickness) on the intermediate
layer.
97
γ-Al2O3
8YSZ+F

4 Results and discussion
A) 8YSZ membrane
Excitation: 514.5 nm; 20.6 mW
Raman Intensity
Al 2 O 3
380
cubic YSZ
262
400 380 360 340 320 300 280 260 240 220 200
Wavenumber/cm -1
cubic YSZ
225
98
Raman Intensity
Al-O-Zr
1155
B) 8YSZ membrane
Excitation: 514.5 nm; 20.6 mW
cubic YSZ
1050
cubic YSZ
1033
1200 1180 1160 1140 1120 1100 1080 1060 1040 1020
Wavenumber/cm -1
Figure 65 A and B Raman spectra of the 8YSZ layer that contained F127, acetyl acetone and caproic acid
in the sol.
4.3.3.2 Amine approach - TiO2-ZrO2 and binary oxide layers
TiO2, ZrO2 and 50 mol % TiO2 in ZrO2 layers are prepared on γ-Al2O3 membranes. The
films are prepared by spin or dip coating using the as-made DEA sols. The amount of
mesoporous defects in the final membrane layer was minimised by consequently coating
each γ-Al2O3 membrane twice with the DEA sol. The first DEA layer was calcined
before applying the second DEA layer. The layer thickness of the intermediate layer and
the final membrane layer is observed with SEM and TEM. 20-40 nm homogeneous thin
films have been observed for the membrane layers prepared from TiO2, ZrO2 and
Ti0.5Zr0.5O2 calcined at 400, 500 and 600ºC (Table 27 and Figure 67 A-C).
The infiltration depth of DEA sols was studied by means of Secondary Ion Mass
Spectroscopy (SIMS). A penetration depth into the γ-Al2O3 was approximately half the
final membrane layer thickness, considering the TiO2, ZrO2 and Ti0.5Zr0.5O2 layers
calcined at 500ºC. Titanium and zirconium atoms were found at a distance of 10-15 nm
from the Ti0.5Zr0.5O2 – γ-Al2O3 boundary for the Ti0.5Zr0.5O2 membrane with a thickness
of 25 nm.
Table 27 Intermediate and final membrane layer thickness
Calcination mol% Ti Thickness γ-Al2O3 in µm Thickness Final layer in
temperature
nm by TEM
400ºC 50 2.5 20-30 (SEM)
0 5.8 36
500ºC 50 2.6 25
100 2.4 22
600ºC 50 4.5 23

4 Results and discussion
~50nm TiO2
γ-Al2O3
~50nm Ti0.5Zr0.5O2
~25nm ZrO2
γ-Al2O3
γ-Al2O3
Figure 66 SEM cross section images of A) TiO2, B) Ti0.5Zr0.5O2 and C) ZrO2 calcined at 600ºC
The surfaces of the DEA-layers were studied with XRD to identify the crystalline phase
of the membrane layer. However, only the crystalline support material could be observed.
Focussed Raman spectroscopy studies observed crystalline phase and are compared with
the electron diffraction patterns obtained by means of high resolution transmission
electron microscopy (HR-TEM).
Figure 67-A present the HR-TEM image of the 25 nm TiO2 layer calcined at 500ºC.
Rough estimation of the particle size diameter would be 5-10 nm. The particle sizes
observed with TEM does not seem to change significantly from the particle size in the
99
A
B
C

4 Results and discussion
sol. The average crystal size is 15 nm obtained from the XRD analysis on powders which
is more than half of the final membrane layer thickness. Similar particle size range has
been found for commercial TiO2 membranes. Defect free 25 nm thin TiO2 layers based
on the stacking of 2-5 particles is expected to be difficult.
γ-Al2O3
Tungsten
A B
γ-Al2O3
Figure 67 A) TEM image of a TiO2 membrane calcined at 500ºC supported by γ-Al2O3. B) ordered atomic
structure is observed presenting anatase TiO2
Figure 67-B shows a higher magnification to show the atomic ordering presenting anatase
TiO2. No rutile TiO2 was observed as in agreement with XRD analysis on bulk
membrane material. Focussed Raman spectroscopy was applied on the TiO2 and ZrO2
membranes obtaining mainly bands assigned to the supporting alumina (Table 28 and
Figure 68). Raman on the TiO2 membrane confirms the presence of anatase TiO2 by the
bands at (520, 148 cm -1 ). Rutile TiO2 can not be excluded due to a very weak band at 450
and 453 cm -1 , respectively 514 and 632 nm laser excitation.
100
TiO2

4 Results and discussion
Raman Intensity
Table 28 Raman bands of the TiO2 and ZrO2 membrane assigned to the microstructural phases. 138
(s: strong, w: weak, vw: very weak).
Laser excitation TiO2 membrane ZrO2 membrane
514.5 nm, 40.1 mW 632.8 nm, 4.8 mW
Wavelenth in cm -1
520
520w 522w Anatase TiO2
468vw Monoclinic ZrO2
450vw 453vw
Monoclinic ZrO2
or Rutile TiO2
432w 433w α-Al2O3 α-Al2O3
419s 419s α-Al2O3 α-Al2O3
380s 381s α-Al2O3 α-Al2O3
148s 148s Anatase TiO2
468
450
432
419
A -- TiO2
-- ZrO2
B
381
148
500 400 300 200 100
Wavenumber/cm -1
101
Raman Intensity
522
453
433
419
374
148
500 400 300 200 100
Wavenumber/cm -1
Figure 68 Raman spectra of TiO2 (red) and ZrO2 (brown) membranes calcined at 500ºC. Excitation: A) 514.5
nm, 40.1 mW and B) 632.8, 4.8mW.
Figure 69-B presents a homogeneous thin 25-30 nm ZrO2 layer containing ordered
zirconium atoms (Figure 69-A). No particles and pores could be observed. Ordered ZrO2
was observed in the scale of the membrane layer thickness indicating fully crystalline
layer. Fully crystalline 25 nm thin layers with the absence of small particles might form a
gastight layer. The presence of these ZrO2 nanocrystals might explain the rather small
BET specific surface area due to the inaccessibility of N2.
The electron diffraction pattern obtained from the HR-TEM analysis on the ZrO2
membrane calcined at 500ºC showed crystalline and amorphous intensity. The

4 Results and discussion
amorphous intensity originates from the supporting γ-Al2O3 layer and the nanocrystalline
intensity to tetragonal ZrO2 phase as could expected from the XRD results on powders.
Raman spectroscopy presents mainly alumina (Table 28 and Figure 68). Monoclinic
ZrO2 enrichments on the surface might be identified by the very weak band at 468 and
450 cm -1 . Tetragonal ZrO2 bands (e.g. 370 cm -1 ) are hard to distinguish because this band
is in the same wave number region where the α-Al2O3 bands are observed. Raman studies
with 488 nm excitation on both TiO2 and ZrO2 membranes could not reveal more
contrast.
A
ZrO2
γ-Al2O3
Figure 69 A and B) TEM images of a ZrO2 membrane calcined at 500ºC supported by γ-Al2O3.
Cross section TEM image show 20-30 nm Ti0.5Zr0.5O2 membranes that are calcined at
500 and 600ºC. Both membranes obtain amorphous structures. The electron diffraction
patterns obtained by TEM do not indicate the presence of crystalline domains in the
Ti0.5Zr0.5O2 layer calcined at 500 and 600ºC. The absence of orthorhombic Ti0.5Zr0.5O2
which can be observed by XRD at 600ºC in the bulk material can be the result of effects
that result from interactions between the layer and the support material.
Raman spectroscopy demonstrates mainly alumina. No clear trends are observed on the
difference in calcination temperature of the membranes. The weak band at 450 cm -1 , at
both the laser excitations, can either be assigned to rutile TiO2 or to monoclinic ZrO2
(Table 29 and Figure 71). Raman spectroscopy could not show any signs for tetragonal
ZrO2, anatase TiO2 or orthorhombic TiZrO4. Ti-O-Al, Zr-O-Al or Ti-O-Zr 139 bands were
not visible.
102
Tungsten
ZrO2
γ-Al2O3
B

4 Results and discussion
Tungsten 500ºC A 600ºC
B
Ti0.5Zr0.5O2
γ-Al2O3
Figure 70 TEM images of a Ti0.5Zr0.5O2 membrane calcined at 500 (A) and 600ºC (B) supported by γ-Al2O3.
Table 29 Raman bands on Ti0.5Zr0.5O2 membranes assigned to the microstructural phases. 138
(vs: very strong, s: strong, w: weak)
Laser excitation Ti0.5Zr0.5O2 membranes
488 nm, 250 mW 514.5 nm, 40.1 mW 500ºC 600ºC
Wavelenth in cm -1
753s 752s α-Al2O3
647s 647s
α-Al2O3
579w 579w
α-Al2O3
450w 450w Monoclinic ZrO2 or Rutile TiO2
433w 432w
α-Al2O3
419vs 419vs
α-Al2O3
380s 381s
α-Al2O3
103
Tungsten
γ-Al2O3
Ti0.5Zr0.5O2

4 Results and discussion
Raman Intensity
752
647
579
800 600 400
Wavenumber/cm -1
104
419
432
450
380
600ºC
500ºC
Figure 71 Raman spectra of the Ti0.5Zr0.5O2 membranes calcined at 500ºC (green) and at 600ºC (blue)
using the excitation of 514.5 nm, 20.6 mW.
4.3.3.3 Amine approach - Organic content in membrane layers
The effective pore size in the layers can be determined by an amorphous film around the
nanocrystals or in combination with the presence of organic residues originating from the
amines. The latter hypothesis was put forward by Duke et al. 94 for carbon remainings
and by Xomeritakis et al. 140 for amine groups in microporous silica membranes.
The carbon content at the outer surface of the membrane layers as determined by XPS
decreases with calcination temperature and is slightly higher than the carbon content of
the bulk material as determined by ICP (Figure 72). However, sputtering the surface with
argon for 2 minutes reduces the carbon content to comparable values. This suggests that
the observed high carbon concentrations originate significantly from atmospheric
contamination. Striking is the agreement between the carbon content trend in the powders
and the layer with increasing ZrO2 in the sample. Both bulk material and membrane layer
show an increase in carbon content with an increasing ZrO2 content. This can be
explained by the slightly higher reactivity of the titanium precursor compared to that of
the zirconium precursor resulting in stronger organic-inorganic bonds.

4 Results and discussion
Mol% Carbon
0,8
0,6
0,4
0,2
0
140
400
Temperature [ºC]
105
500
600
100%
50%
0%
Mol % TiO 2
in ZrO 2
Figure 72 mol% carbon in the dried DEA sols calcined at 140, 400, 500 and 600ºC for 0, 50 and 100 mol% TiO2
in ZrO2.

4 Results and discussion
4.3.4 Gas permeation
The results of all He and SF6 gas permeation measurements are summarized in Figure 73.
The membrane numbers are listed in Table 30. The permselectivity of the membranes
that have been calcined at 400ºC and of the γ-Al2O3 membrane are all similar to what can
be expected from the Knudsen diffusion (6.0). An increase in the calcination temperature,
to 500ºC, results in a strong reduction of He permeation from 4⋅10 -8 mol/s⋅m 2 ⋅Pa to
approximately 2⋅10 -9 mol/s⋅m 2 ⋅Pa. A calcination temperature of 600ºC causes a
significant increase in the He permeance to at least 2⋅10 -8 mol/s⋅m 2 ⋅Pa and a He/SF6
permselectivity of up to 30. It should be noted that both the He and the SF6 permeability
of membrane #204 significantly exceed those of all other membranes, which suggests
that this particular membrane contains substantially more large pores or defects. All other
results indicate that mass transport through the Ti0.5Zr0.5O2 layers calcined above 400ºC is
no longer dominated by the Knudsen diffusion, but that at least for the He transport there
is a substantial contribution from micropore diffusion.
He Permeance (10 -8 mol/s m 2 Pa)
109 15
10
5
0
Permeance
Permselectivity
Double Coating
Support 400ºC 400ºC 500ºC 500ºC 600ºC 600ºC
#109 #206 #208 #132 #202 #205 #204
Figure 73: He permeance and He/SF6 permselectivity of Ti0.5Zr0.5O2 membranes calcined at 400, 500 and
600ºC measured at 200ºC.
Table 30 Overview of characterised membranes with gas permeation
Final layer Calcination temperature Membrane #
γ-Al2O3 600ºC 109
Ti0.5Zr0.5O2 400ºC 206, 209
500ºC 132, 143, 202
600ºC 204, 205
106
35
30
25
20
15
10
5
0
He/SF 6 Permselectivity

4 Results and discussion
Permeance (10 -9 mol/s m2Pa)
Permeance (10 -9 mol/s m 2 Pa)
1.E-06
10 -6
1.E-07
10 -7
1.E-08
10 -8
1.E-09
10 -9
1.E-10 1E-08
10 -10
10 -8
10 -9
1E-09
10 -10
107
#204; 600ºC
#205; 600ºC
1E-10
0.2 0.3 0.4 0.5 0.6 0.7 0.8
Kinetic diameter (nm)
#143; 500ºC
#132; 500ºC
#202; 500ºC
Figure 74: Permeability (measured at 200ºC) of Ti0.5Zr0.5O2 membranes calcined at 500 and 600ºC for
different gasses as a function of their kinetic diameter. The lines serve only as a guide to the eye.
In Figure 74 the measured permeances of a series of gasses with kinetic diameters
ranging from 0.26 to 0.55 nm are presented for all membranes calcined at 500 and 600ºC.
All membranes, demonstrate a significant increase in permeability for gasses with a
kinetic diameter below 0.3 nm. These data are indicative of the existence of pores in the
Ti0.5Zr0.5O2 layers having dimensions which make them primarily accessible to relatively
small molecules such as He and H2. This is especially manifest in membrane #205, which
achieves a measured permselectivity for He/H2 and He/N2 of 2.0 and 14, respectively.
In Figure 75 the effect of temperature on H2 and CO2 permeation are presented for
membrane #205 in Arrhenius-type plots. The linearity of the relation between logarithm
of permeance and reciprocal temperature confirms a thermally activated transport
mechanism. The activation energy Ea can have any sign; in this study the sign is positive
when the permeation increases with temperature. The measured activation energy for H2
and CO2 transport amount to 8.4 and 4.9 kJ/mol, respectively. As a result of these

4 Results and discussion
differences the H2/CO2 permselectivity of this membrane increases from 4.2 to 5.3 as the
temperature increases from 115 to 190ºC. This is comparable to values observed for ZrO2
membranes. 141,142 The activation energies for He transport through membranes #202,
#204 and #205 are 3.9, 7.9 and 15 kJ/mol, respectively, which is indicative of an
increasing degree of micropore diffusion in these membranes.
Ln[Permeance s m 2 Pa/mol)]
0.24 228 0.25 0.26 189 0.27 0.28 156 0.29 0.30 103 0.31 103 0.32
-12.0
7
-12.5
-13.0
-13.5
-14.0
T (ºC)
-14.5
3
0.24 0.25 0.26 0.27 0.28 0.29 0.30 0.31 0.32
1000/RT (mol/kJ)
108
Hydrogen
Carbon dioxide
Permselectivity
Figure 75: H2/CO2 permselectivity as a function of temperature and Arrhenius plots for H2 and CO2
permeation of a Ti0.5Zr0.5O2 membrane (#205) calcined at 600ºC.
Because of its outstanding performance, the hydrothermal stability of this particular
membrane (#205) has been tested. The effect of exposure to water vapour on He and N2
permeation is presented in Figure 76. During the exposure, a strong decrease in both the
He and N2 permeance is immediately observed. The amount of permeating water was too
small to be determined gravimetrically. After termination of the steam exposure,
degassing results in a steady increase of both He and N2 permeation. Only after a period
of approximately 750 hours the He permeation starts to decrease. Both permeation
measurements and visual inspection of the membrane surface after the experiment
exclude the occurrence of layer delamination which can result from hydrothermal
treatment. In fact, a He/N2 permselectivity above 10 is observed again after exposure to
steam after flushing the membrane for 500 h which demonstrates a high degree of
stability of the micropore structure. Besides, the γ-Al2O3 intermediate layer was not
delaminated as expected from the hydrothermal testing on SiO2 membranes of the
recently reported work of Duke et al. 94
6
5
4
H 2/CO 2 Permselectivty

4 Results and discussion
He Permeance (mol/s m 2 Pa)
2.5E-08
2.5�10 8
2.0E-08
2.0·10 8
1.5E-08
1.5·10 8
1.0·10 8
1.0E-08
0.5·10 8
5.0E-09
0.0·10 8
Helium Nitrogen
0.0E+00
0.0E+00
-250 0 250 500 750 1000 1250
Time after Steam (h)
109
5.0E-09
5�10 -9
4.0E-09
4·10 -9
3.0E-09
3·10 -9
2·10 -9
2.0E-09
1·10 -9
1.0E-09
0·10 -9
Figure 76: He (♦) and N2 (■) permeance of a Ti0.5Zr0.5O2 membrane (#205) calcined at 600ºC before and
after exposure to 3 bar of steam at 200ºC.
N 2 Permeance (mol/s m 2 Pa)

5 Conclusions and recommendations
5 Conclusions and recommendations
First, this chapter includes the conclusions of the studied zeolite materials. The basic aim
was to prepare zeolite material that could separate gasses by molecular sieving.
Recommendations on the zeolite materials are discussed to achieve this goal.
Second, the results on sol-gel derived TiO2-ZrO2 sols, bulk materials, layers and
membranes are summarised and related to the preparation of the TiO2-ZrO2 gas
separation membrane goal. Additionally, suggestions for further research are outlined to
achieve industrial goals.
8-ring Deca-dodecasil 3R and 6-ring Dodecasil 1H zeolite materials
Deca-dodecasil 3R (DDR) hydrothermal synthesis is complex and formed zeolite
mixtures of DDR with DOH or SGT (Dodecasil 1H or Sigma 2). The synthesis of DDR
as a potential CO2/N2 separation membrane material was unsuccessful. Further syntheses
with varying gel compositions of nutrients from several suppliers and reaction conditions
should be performed in order to understand how to prepare single phase DDR.
Dodecasil 1H (DOH) crystal synthesis, structure directing agent (SDA) free synthesis and
SDA-free DOH by means of post synthesis treatments were performed to prepare allsilica
zeolite material with pores inaccessible to gasses with the exception of gasses with
a kinetic diameter of H2 or smaller.
DOH can be synthesised under standard conditions (Teflon lined autoclave heated
under static conditions, 200°C, gel composition: 100SiO2:9-11ADA:70-270EN:3000-
7400H2O forming crystals >100 µm). Changing the synthesis conditions from static to
dynamic by means of autoclave rotation or stirring the gel in situ to obtain a more
homogeneous gel mixture reduces the DOH crystallisation time from 21 days to 4 days
coupled with a decrease in crystal size. The DOH crystal size can also be reduced by
aging the gel at 80ºC for several hours before hydrothermal crystallisation at 200ºC. The
DOH nucleation and growth can be decoupled.
The DOH crystal size can be reduced to thin hexagonal plates with sizes in the order of
10 µm being too large for membrane formation or to act as seeds for the layer formation
by means of the secondary growth of DOH nuclei. The DOH crystal size might be
synthesised smaller by means of further crystallisation optimisation. This might be
accomplished by changing the gel composition, aging of the gel and reaction condition
such as the heating ramp and stirring/rotating conditions.
Secondary growth of DOH crystals in gel composition with the absence of SDA
was unsuccessful. Low DOH yield was obtained using a small amount of SDA in the gel
composition. These synthesised DOH crystals have a ~75% SDA cage occupancy. SDAfree
DOH synthesis by means of seeded synthesis was not obtained. This proposed
110

5 Conclusions and recommendations
method by Grebner et al. 71,72 was not confirmed which might be explained by the
absence of SDAs or guest molecules to form DOH cages.
DOH Calcination in air for extended periods (21 days) at elevated temperatures
(900ºC) could not result in complete SDA removal. A minor calcination enhancement is
observed for the DOH samples that are doped with traces of titanium. It is not certain if
this can be explained by catalytic oxidation of titanium. Significant DOH cages were
emptied by calcination when the DOH crystal size was reduced mechanically. However,
crystal structure loss is inevitable during ball milling DOH crystals. Quasi SDA-free
DOH, with an cage occupancy of only 1% and high crystallinity, was obtained after
calcination at 900ºC for 5 hours when the atmospheric pressure was increased twice to 50
MPa for 30 min (Hot isostatic pressure calcination).
All silica DOH that is quasi SDA-free was prepared and might present hydrothermal
stable microporous material with pores that are inaccessible for CO2 and possibly open to
H2. These DOH crystals should be studied with gas sorption to prove the H2 accessibility
and diffusion. First DOH layers on α-Al2O3 supports were prepared with crystallisation
times of only 24 hours.
Sol-Gel derived membranes
Polymeric Y2O3 or TiO2 mixed ZrO2 sols are synthesised for the preparation of
ultramicroporous powders and thin films on γ-Al2O3 supporting membranes as potential
gas separation membranes. Two routes have been selected, the Ketone and the Amine
approach based on the precursor modifiers.
Acetyl acetone modified Zr-precursor, carboxylic catalysed and yttrium-hydrated
sol syntheses resulted in ~5 nm stable yttrium stabilised zirconia (YSZ) sols. Structure
directing agents (SDA) such as Pluronic F127 did not change the particle size in the sol.
Caproic and nanonic acids catalysed sols resulted in the highest viscous stable sols.
Fractal dimensions of these Ketone-sols would be informative to study.
8 mol % Y2O3 addition to the zirconia sols resulted in cubic 8YSZ material at
temperatures above 400ºC. 8YSZ calcined dried sols are microporous with BET specific
surface areas of ~50 m 2 /g. SDA addition to the 8YSZ sols, to mesostructure the cubic
8YSZ nanoparticles, transformed the material to a more mesoporous nature. TEM
investigations on powders and on the surface of mesostructured 8YSZ membranes are
advised. 8YSZ layers with the absence of SDAs might form microporous layers with low
permeability. Optimal calcination conditions such as drying, heating ramp, dwell time
calcination time and cooling ramp should be studied in detail.
30-50 nm cubic 8YSZ films prepared by the Ketone-approach show He and N2
transport by Knudsen diffusion due to defects or by too large pores in the final membrane
layer. Defect free 8YSZ layers using the Ketone-approach should be prepared in order to
111

5 Conclusions and recommendations
study the permeability of these films as a function of pressure difference over the
membrane.
TiO2-ZrO2 and mixtures of 25, 50, 66, 75 and 90% mol TiO2 in ZrO2 were
prepared using the Amine-approach. Diethanolamine or diisopropanolamine precursor
modified Ti/Zr precursors can result in ~5 nm stable linear polymeric sols similar to the
state of the art SiO2 sols. It is believed that sol-gel derived ultramicroporous materials can
be synthesised with linear polymeric sols alone. The solid concentration and the molar
composition of these Amine sols should be studied by means of SAXS and viscosity
measurements.
The single oxides TiO2 and ZrO2 crystallise at temperatures below 400ºC
resulting in mesoporous and microporous material, respectively. As expected from the
linear Amine-sols, the amorphous binary TiO2/ZrO2 materials are microporous between
400 and 500ºC. The highest BET specific surface area of ~200 m 2 /g with ~1.0 nm
estimated pore size (gas physisorption) is obtained for the Ti0.5Zr0.5O2 calcined at 500ºC.
The crystallisation temperature of orthorhombic Ti0.5Zr0.5O2 is between 550 and 600ºC
which is ~250ºC higher than the single oxides. Diisopropanolamine instead of
diethanolamine postponed the crystallisation temperature of the TiO2-ZrO2 mixture.
These microporous materials exhibit even higher specific surface areas at 400ºC. This
might be explained by the higher viscosity of the Amine sols. The crystallisation
temperature and the specific surface areas of the binary TiO2/ZrO2 material prepared by
the Amine-approach are higher than the Ketone-approach. However, the solid
concentration and the sol composition of both approaches are different which make them
difficult to compare.
State of the art silica membranes have pores in the range of 0.3 nm mainly due to their
small particles of amorphous nature. It might not be excluded that ultramicroporous
inorganic membranes can be prepared from sol-gel derived amorphous ceramics alone.
Amorphous Ti0.5Zr0.5O2 layers might be ultramicroporous. Nevertheless, the physical
properties such as the results form XRD and gas sorption (H2, CO2, N2 at low and high
pressures and temperatures) on bulk material should be studied in more detail on these
TiO2-ZrO2 microporous materials. Changing the sol composition, drying and calcination
procedures might be beneficial.
20-60 nm thin homogeneous films can be prepared from TiO2, ZrO2 and binary
oxides on γ-Al2O3 membranes using calcination temperatures in the range of 400 to
600ºC.
Knudsen mass transport is observed in anatase TiO2 or tetragonal ZrO2 films prepared by
the Amine approach. This indicates the presence of pores larger than the kinetic diameter
of the transported gasses (~0.5 nm). The larger pores in TiO2 films are in agreement with
the mesoporous nature of the TiO2 bulk material. Larger pores in ZrO2 films can be
112

5 Conclusions and recommendations
mesoporous defects or intercrystalline channels between ZrO2 crystals that are in the
same order as the layer thickness. It is expected that crystalline TiO2 or ZrO2 may not
form ultramicroporous membranes with crystallisation temperatures of 500ºC and higher
due to the relative large crystallinity of these layers. Defect free, TiO2 or ZrO2
membranes should be prepared and studied with gas permeability to confirm the absence
of ultramicropores.
H2/CO2 permselectivity higher than Knudsen factor is observed for Ti0.5Zr0.5O2 films
calcined at 500 and 600ºC. He and H2 gas transport is thermally activated indicating the
presence of micropore diffusion. Knudsen mass transport is obtained for gasses with
kinetic diameter of CO2 and larger. These results show that Ti0.5Zr0.5O2 films calcined at
500 and 600ºC contain a pore size distribution of ~0.3 to ~0.5 nm in diameter.
The He permeance of 1·10 -7 mol/m 2 sPa with a He/N2 permselectivity of 5.9 or the He
permeance 2·10 -8 mol/m 2 sPa with a He/N2 permselectivity of 14 are lower than state of
the art SiO2 membranes but are higher than TiO2-ZrO2 membranes found in literature. 80
These results fulfil the primary aim of achieving gas separation membranes from TiO2-
ZrO2 material. The permeance might be increased by the formation of thinner or more
porous γ-Al2O3 intermediate layers. The permselectivity might be improved using
structure directing agents (SDAs) in the sols, avoiding larger pores or metal loading in
the final membrane layer.
The micropores in these membranes are stable for at least 1500 hours. Steam test results
in reversible pore blocking without layer delamination or destruction while maintaining
the He/N2 permselectivity of ~10. These gas separation membranes show the chemical
stability at high temperatures in humid conditions fulfilling the aim of hydrothermally
stable membrane preparation. This evidence indicates that binary oxide 50 mol % TiO2 in
ZrO2 microporous thin layers could be alternatives to SiO2 membranes for precombustion
applications. However, the permselectivity and permeability values of the prepared
Ti0.5Zr0.5O2 membranes do not reach industrial targets and these membranes might not
have sufficient hydrothermal stability. The hydrothermal stability might be improved by
using SDAs used in hydrothermal stable SiO2 membranes.
113

6 References
6 References
1. Metz, B., Davidson, O., de Coninck, H., Loos, M. & Meyer, L. IPCC special report on
carbon dioxide capture and storage (Cambridge university press, Cambridge, 2005).
2. Bernal, M. P., Coronas, J., Menendez, M. & Santamaria, J. Separation of CO2/N2
mixtures using MFI-type zeolite membranes. AICHE 50, 127-135 (2004).
3. Goettlicher, G. Energetik der Kohlendioxidrückhaltung in Kraftwerken. Fortschritt-
Berichte VDI, Energietechnik (1999).
4. Bohnes, S., Van der Donk, G. J. W., Meulenberg, W. A., Scherer, V. & Stoever, D. in
Proceeding of Porous Ceramic Materials (ed. Luyten, J.) (Brugge, Belgium, 2005).
5. Bhave, R. Inorganic Membranes, Synthesis, Characterization and Properties (Van
Nostrand Reinhold, New York, 1991).
6. Lin, Y. S. Microporous and dense inorganic membranes: current status and prospective.
Separation and Purification Technology 25, 39-55 (2001).
7. Bredesen, R., Jordal, K. & Bolland, O. High-temperature membranes in power generation
with CO2 capture. Chemical Engineering and Processing 43, 1129 (2004).
8. Pex, P. et al. in Eight international conference on inorganic membranes (eds. Akin, F. T.
& Lin, Y. S.) 524 (Cincinnati, Ohio, 2004).
9. Pan, X. L., Stroh, N., Brunner, H., Xiong, G. X. & Sheng, S. S. Pd/ceramic hollow fibers
for H2 separation. Separation and Purification Technology 32, 265-270 (2003).
10. Jun, C. S. & Lee, K. H. Preparation of palladium membranes from the reaction of
Pd(C3H3)(C5H5) with H2: wet-impregnated deposition. Journal of Membrane Science
157, 107-115 (1999).
11. Yeung, K. L. & Varma, A. Novel preparation techniques for thin metal-ceramic
composite membranes. AICHE 41, 2131-2139 (1995).
12. Moss, T. S., Peachey, N. M., Wilson, M. S. & Dye, R. C. in Fourth international
conference on inorganic membranes (ed. Fain, D. E.) 343 (Gatlinburg, Tennessee, 1996).
13. Edlund, D. J. & McCarthy, J. The Relationship between Intermetallic Diffusion and Flux
Decline in Composite-Metal Membranes - Implications for Achieving Long Membrane
Lifetime. Journal of Membrane Science 107, 147-153 (1995).
14. Morooka, S. & Kusakabe, K. Inorganic membranes for gas separation at elevated
temperatures. Industrial Ceramics 20, 15-18 (2000).
15. Wagner, C. Beitrag zur Theorie des Anlaufvorganges. Zeitschrift fuer Physikalische
Chemie B11, 25 (1930).
16. Van der Haar, L. M. Mixed-conducting perovskite membranes for oxygen separation
(PhD Thesis, University of Twente, Enschede, 2001).
17. Vente, J., McIntosh, S., Haije, W. G. & Bouwmeester, H. J. M. Properties and
performance of BaxSr1-xCo0.8Fe0.2O3 materials for oxygen transport membranes. Journal
of Solid State electrochemistry 10, 581 (2006).
18. Vente, J., Haije, H. & Rak, Z. in Eight international conference on inorganic membranes
(eds. Akin, F. T. & Lin, Y. S.) 595 (Cincinnati, Ohio, 2004).
19. Mertins, F. H. B. Perovskite-type ceramic membranes; Partial oxidation of methane in a
catalytic membrane reactor (PhD Thesis, University of Twente, Enschede, 2005).
20. Bouwmeester, H. J. M., Kruidhof, H. & Burggraaf, A. J. Importance of the Surface
Exchange Kinetics as Rate Limiting Step in Oxygen Permeation through Mixed-
Conducting Oxides. Solid State Ionics 72, 185-195 (1994).
21. Buechler, O., Serra, J. M., Meulenberg, W. A. & Sebold, D. Preparatin of thin La1xSrxCo1-yFeyO3-delta
membranes supported on tailored ceramic substrates. Solid State
Ionics 178, 91-99 (2007).
114

6 References
22. Shao, Z., Dong, H., Xiong, G., Cong, Y. & Yang, W. Performance of a mixed-conducting
ceramic membrane reactor with high oxygen permeability for methane conversion.
Journal of Membrane Science 183, 181-192 (2001).
23. Aasland, S., Tangen, I. L., Wiik, K. & Odegard, R. Oxygen permeation of
SrFe0.67Co0.33O3-δ. Solid State Ionics 135, 713-717 (2000).
24. Qi, X. W. & Lin, Y. S. Electrical conduction and hydrogen permeation through mixed
proton-electron conducting strontium cerate membranes. Solid State Ionics 130, 149-156
(2000).
25. Cheng, S., Gupta, V. K. & Lin, Y. S. in Eight international conference on inorganic
membranes (eds. Akin, F. T. & Lin, Y. S.) 50 (Cincinnati, Ohio, 2004).
26. Balachandran, U., Guan, J., Dorris, S. E., Bose, A. C. & Stiegel, G. J. in Fifth
international conference on inorganic membranes (ed. S.I., N.) 652 (Nagoya, 1998).
27. Meulenberg, W. A., Serra, J. M. & Schober, T. Preparation of proton conducting
BaCe0.8Gd0.2O3 thin films. Solid State Ionics 177, 2851 (2006).
28. White, J. H., Schwartz, M. & Sammells, A. F. 10 (Patent Eltron Research, Inc., Boulder,
Colo, U.S., 2000).
29. Balachandran, U. et al. in Eight international conference on inorganic membranes (eds.
Akin, F. T. & Lin, Y. S.) 163 (Cincinnati, Ohio, 2004).
30. Readman, J. E., Olafsen, A., Larring, Y. & Blom, R. La0.8Sr0.2Co0.2Fe0.8O3-delta as a
potential oxygen carrier in a chemical looping type reactor, an in-situ powder X-ray
diffraction study. Journal of Materials Chemistry 15, 1931-1937 (2005).
31. Norby, T. & Larring, Y. Mixed hydrogen ion-electronic conductors for hydrogen
permeable membranes. Solid State Ionics 136, 139-148 (2000).
32. Norby, T., Wideroe, M., Glockner, R. & Larring, Y. Hydrogen in oxides. Dalton
Transactions, 3012-3018 (2004).
33. Rothenberger, K. S. et al. Performance testing of hydrogen transport membranes at
elevated temperatures and pressures. Abstracts of Papers of the American Chemical
Society 218, U634-U635 (1999).
34. Balachandran, U., Lee, T. H., Wang, S. & Dorris, S. E. Use of mixed conducting
membranes to produce hydrogen by water dissociation. International Journal of
Hydrogen Energy 29, 291-296 (2004).
35. Song, S. J., Wachsman, E. D., Rhodes, J., Dorris, S. E. & Balachandran, U. Hydrogen
permeability of SrCe1-XMxO3-delta (x=0.05, M=Eu, Sm). Solid State Ionics 167, 99-105
(2004).
36. Rouquerol, J. et al. Recommendation for the characterization of porous solids. Pure &
applied chemistry 66, 1739 (1984).
37. Duke, M. C., da Costa, J. C. D., Do, D. D., Gray, P. G. & Lu, G. Q. Hydrothermally
robust molecular sieve silica for wet gas separation. Advanced Functional Materials 16,
1215 (2006).
38. Ramsay, J. D. F. & Kallus, S. in Recent Advances in Gas Separation by Microporous
Ceramic Membranes (ed. Kanellopoulos, N. K.) 373-395 (Elsevier, Attikis, 2000).
39. www.iza-structure.org.
40. Breck, D. W. Zeolite molecular sieves: Structure chemistry and use (Wiley, New York,
1974).
41. Barrer, R. M. Hydrothermal chemistry of zeolites (Acadamic Press, New York, 1982).
42. Julbe, A., Motuzas, J., Cazevielle, F., Volle, G. & Guizard, C. Synthesis of sodalite/alpha
Al2O3 composite membranes by microwave heating. Separation and Purification
Technology 32, 139-149 (2003).
43. Julbe, A. in Studies in Surface Science and Catalysis (eds. Cejka, J. & Bekkum van, H.)
p135 (Elsevier B.V., Amsterdam, 2005).
115

6 References
44. den Exter, M. J., Jansen, J. C. & Bekkum van, H. in Studies in Surface Science and
Catalysis (eds. Weitkamp, J., Karge, H. G., Pfeifer, H. & Hölderich, W.) 1159 (Elsevier
science B.V., Amsterdam, 1994).
45. Bruggraaf, A. J. in Fundamentals of inorganic membrane science and technology (eds.
Bruggraaf, A. J. & Cot, L.) 331-435 (Elsevier, Amsterdam, 1996).
46. Lovallo, M. C., Gouzinis, A. & Tsapatsis, M. Synthesis and characterization of oriented
MFI membranes prepared by secondary growth. AICHE 44, 1903-1913 (1998).
47. Hasegawa, Y. et al. Separation of CO2-CH4 and CO2-N2 systems using ion-exchanged
FAU-type zeolite membranes with different Si/Al ratios. Korean Journal of Chemical
Engineering 19, 309-313 (2002).
48. Hasegawa, Y., Kusakabe, K. & Morooka, S. Effect of temperature on the gas permeation
properties of NaY-type zeolite formed on the inner surface of a porous support tube.
Chemical Engineering Science 56, 4273-4281 (2001).
49. Coronas, J., Falconer, J. L. & Noble, R. D. Characterization and permeation properties of
ZSM-5 tubular membranes. AICHE 43, 1797-1812 (1997).
50. Tuan, V. A., Falconer, J. L. & Noble, R. D. Alkali-free ZSM-5 membranes: Preparation
conditions and separation performance. Industrial & Engineering Chemistry Research
38, 3635-3646 (1999).
51. Burggraaf, A. J., Vroon, Z., Keizer, K. & Verweij, H. Permeation of single gases in thin
zeolite MFI membranes. Journal of Membrane Science 144, 77-86 (1998).
52. Richter, H., Voigt, I., Fischer, G. & Puhlfurss, P. Preparation of zeolite membranes on the
inner surface of ceramic tubes and capillaries. Separation and Purification Technology
32, 133-138 (2003).
53. Ando, Y., Hirano, Y., Mase, S. & Taguchi, H. in Fifth international conference on
inorganic membranes (ed. S.I., N.) 124 (Nagoya, 1998).
54. Rocha, J. & Anderson, M. W. Microporous titanosilicates and other novel mixed
octahedral-tetrahedral framework oxides. European Journal of Inorganic Chemistry, 801-
818 (2000).
55. Kuznicki, S. M. et al. A titanosilicate molecular sieve with adjustable pores for sizeselective
adsorption of molecules. Nature 412, 720-724 (2001).
56. Corma, A., Rey, F., Rius, J., Sabater, M. J. & Valencia, S. Supramolecular selfassembled
molecules as organic directing agent for synthesis of zeolites. Nature 431,
287-290 (2004).
57. Guan, G. Q., Kusakabe, K. & Morooka, S. Gas permeation properties of ion-exchanged
LTA-type zeolite membranes. Separation Science and Technology 36, 2233 (2001).
58. Hedlund, J., Schoeman, B. & Sterte, J. Ultrathin oriented zeolite LTA films. Chemical
Communications, 1193-1194 (1997).
59. Hedlund, J., Schoeman, B. J. & Sterte, J. in Progress in Zeolite and Microporous
Materials, Pts a-C 2203-2210 (1997).
60. Gies, H. Studies on Clathrasils .9. Crystal-Structure of Deca-Dodecasil 3R, the Missing
Link between Zeolites and Clathrasils. Zeitschrift Fur Kristallographie 175, 93-104
(1986).
61. Yaijma, K., Nakayama, K., Niino, M., Tomita, T. & Yoshida, S. in Proceedings of the
9th international conference on inorganic membranes (eds. Bredesen, R. & Raeder, H.)
(Tapir Uttrykk, Lillehammer, Norway, 2006).
62. Tomita, T., Nakayama, K. & Sakai, H. Gas separation characteristics of DDR type zeolite
membrane. Microporous and Mesoporous Materials 68, 71-75 (2004).
63. Nakayama, K., Suzuki, K., Yoshida, M., Yajima, K. & Tomita, T. 33 (NGK insulators,
Ltd, Japan, 2005).
64. Clark, T. et al. in 14th Annual Meeting of the North American Membrane Society
(Jackson Hole, WY, 2003).
116

6 References
65. Van den Berg, A. W. C. in Opportunities and limitations of hydrogen storage in zeolitic
clathrates (University of Delft, 2006).
66. den Exter, M. J. Exploratory study of the synthesis and properties of 6-, 8- and 10-Ring
tectosilicates and their potential application in zeolite membranes (PhD thesis,
University of Delft, 1996).
67. Gies, H., Gerke, H. & Liebau, F. Dodecasile - eine neue Reihe polytyper
Einschlussverbindungen von SiO2. Angewandte Chemie 94, 215 (1982).
68. Baerlocher, C., Meier, W. M. & Olson, D. H. Atlas of Zeolite Framework Types
(Elsevier, Amsterdam, 2001).
69. Gies, H. 1-Aminoadamantane as guest molecule in dodecasil 1H: an X-ray
crystallographic study. Journal of Inclusion Phenomena 4, 85-91 (1986).
70. Gerke, H. & Gies, H. Studies on clathrasils. IV. Zeitschrift Für Kristallographie 166, 11-
22 (1984).
71. Grebner, M., Reich, A., Schüth, F. & Unger, K. (Merck Patent Gesellschaft 024962,
Germany, 1993).
72. Grebner, M., Reichert, H., Schüth, F. & Unger, K. in Characterization of Porous Solids
Iii 545-550 (1987).
73. Grebner, M. D., Reich, A., Schüth, F., Unger, K. K. & Franz, K. D. Influence of
Synthesis Conditions on the Morphology of Dodecasil 1h. Zeolites 13, 139-144 (1993).
74. Müller, U. et al. high-resolution sorption studies of argon and nitrogen on large crystals
of microporous zeolite ZSM-5. Fresenius Zeitung Analytic Chemie 333, 433-436 (1989).
75. Van de Goor, G., Lindlar, B., Felsche, J. & Behrens, P. Solvent-free synthesis of
clathrasils using metal-organic complexes as structure-directing agents. Journal of
chemical society chemical communications 24, 2559 (1995).
76. Livage, J. & Sanchez, C. Sol-Gel Chemistry. Journal of Non-Crystalline Solids 145, 11-
19 (1992).
77. Brinker, C. J. & Scherer, G. W. Sol-Gel science (Acadamic Press, 1990).
78. Benfer, S., Arki, P. & Tomandl, G. Ceramic membranes for filtration applications -
Preparation and characterization. Advanced Functional Materials 6, 495 (2004).
79. Richter, H. Herstellung keramischer nanofiltrationsmembranen aus ZrO2 und TiO2
(Technical University Bergacadamie Freiberg, 1999).
80. Spijksma, G. I. et al. Microporous zirconia-titania composite membranes derived from
diethanolamine-modified precursors. Advanced Materials 18, 2165 (2006).
81. Spijksma, G. (PhD Thesis, University of Twente, Enschede, 2006).
82. Van Gestel, T., Kruidhof, H., Blank, D. H. A. & Bouwmeester, H. J. M. ZrO2 and TiO2
membranes for nanofiltration and pervaporation - Part 1. Preparation and characterization
of a corrosion-resistant ZrO2 nanofiltration membrane with a MWCO < 300. Journal of
Membrane Science 284, 128-136 (2006).
83. Aust, U., Benfer, S., Dietze, M., Rost, A. & Tomandl, G. Development of microporous
ceramic membranes in the system TiO2/ZrO2. Journal of Membrane Science 281, 463-
471 (2006).
84. Meulenberg, W. A., Mertens, J., Bram, M., Buchkremer, H. & Stöver, D. Graded porous
TiO2 membranes for microfiltration. Journal of the European Ceramic Society, In press
(2005).
85. Cot, L. et al. Inorganic membranes and solid state sciences. Solid State Sciences 2, 313-
334 (2000).
86. Uhlhorn, R. J. R., Veld, M., Keizer, K. & Burggraaf, A. J. High Permselectivities of
Microporous Silica-Modified Gamma-Alumina Membranes. Journal of Materials
Science Letters 8, 1135-1138 (1989).
87. de Vos, R. M. High-selectivity, high-flux silica membranes for gas separation (University
of Twente, Enschede, 1998).
117

6 References
88. de Vos, R. M., Maier, W. F. & Verweij, H. Hydrophobic silica membranes for gas
separation. Journal of Membrane Science 158, 277-288 (1999).
89. de Vos, R. M. & Verweij, H. Improved performance of silica membranes for gas
separation. Journal of Membrane Science 143, 37-51 (1998).
90. de Vos, R. M. & Verweij, H. High-selectivity, high-flux silica membranes for gas
separation. Science 279, 1710-1711 (1998).
91. Benes, N. E., Nijmeijer, A. & Verweij, H. in Recent Advances in Gas Separation by
Microporous Ceramic Membranes (ed. Kanellopoulos, N. K.) 373-395 (Elsevier, Attikis,
2000).
92. Tsai, C. Y., Tam, S. Y., Lu, Y. F. & Brinker, C. J. Dual-layer asymmetric microporous
silica membranes. Journal of Membrane Science 169, 255-268 (2000).
93. Xomeritakis, G. et al. Organic-templated silica membranes - I. Gas and vapor transport
properties. Journal of Membrane Science 215, 225-233 (2003).
94. Duke, M. C., da Costa, J. C. D., Do, D. D., Gray, P. G. & Lu, G. Q. Hydrothermally
robust molecular sieve silica for wet gas separation. Advanced Functional Materials 16,
1215-1220 (2006).
95. Mase, S., Ando, Y., Hirano, Y. & Taguchi, H. in Fifth international conference on
inorganic membranes (ed. S.I., N.) 652 (Nagoya, 1998).
96. Yoshida, K., Hirano, Y., Fujii, H., Tsuru, T. & Asaeda, M. Hydrothermal stability and
performance of silica-zirconia membranes for hydrogen separation in hydrothermal
conditions. Journal of Chemical Engineering of Japan 34, 523-530 (2001).
97. Oyama, S. T., Zang, L., Lee, D. & Jack, D. S. 11 (Patent, Conoco Inc., Houston, TX;
Virginia Polytechnic Institute and State University, Blacksburg, VA, US, 2003).
98. Asaeda, M. & Yamasaki, S. Separation of inorganic/organic gas mixtures by porous silica
membranes. Separation and Purification Technology 25, 151-159 (2001).
99. Nijmeijer, A., Kruidhof, H., Bredesen, R. & Verweij, H. Preparation and properties of
hydrothermally stable gamma-alumina membranes. Journal of the American Ceramic
Society 84, 136-140 (2001).
100. Sekulic-Kuzmanovic, J. Mesoporous and microporous titania membranes (PhD thesis,
University of Twente, Enschede, 2004).
101. Voigt, I., Fischer, G., Puhlfurss, P., Schleifenheimer, M. & Stahn, M. TiO2-NFmembranes
on capillary supports. Separation and Purification Technology 32, 87-91
(2003).
102. Voigt, I., Mitreuter, G. & Futing, M. The pore structure of meso- and microporous TiO2membranes.
Ceramic Forum International 79, E39-E44 (2002).
103. Lin, Y. S., Chang, C. H. & Gopalan, R. Improvement of Thermal-Stability of Porous
Nanostructured Ceramic Membranes. Industrial & Engineering Chemistry Research 33,
860-870 (1994).
104. Van Gestel, T., Van der Donk, G. J. W., Meulenberg, W. A., Bouwmeester, H. J. M. &
Stoever, D. in International conference inorganic membranes (eds. Bredesen, R. &
Raeder, H.) 658 (Lillehammer, Norway, 2006).
105. Kueper, T. W., Visco, S. J. & Dejonghe, L. C. Thin-Film Ceramic Electrolytes Deposited
on Porous and Nonporous Substrates by Sol-Gel Techniques. Solid State Ionics 52, 251-
259 (1992).
106. Hebert, V., His, C., Guille, J., Vilminot, S. & Wen, T. L. Preparation and
Characterization of Precursors of Y2O3 Stabilized ZrO2 by Metal-Organic Compounds.
Journal of Materials Science 26, 5184-5188 (1991).
107. Okubo, T. & Nagamoto, H. Low-Temperature Preparation of Nanostructured Zirconia
and YSZ by Sol-Gel Processing. Journal of Materials Science 30, 749-757 (1995).
118

6 References
108. Wen, T. L., Hebert, V., Vilminot, S. & Bernier, J. C. Preparation of Nanosized Yttria-
Stabilized Zirconia Powders and Their Characterization. Journal of Materials Science 26,
3787-3791 (1991).
109. Pacheco, G. & Fripiat, J. J. Physical chemistry of the thermal transformation of
mesoporous and microporous zirconia. Journal of Physical Chemistry B 104, 11906-
11911 (2000).
110. Serra, J. M., Hansch, R., Meulenberg, W. A., Menzler, N. H. & Buchkremer, H. in
International conference on porous ceramic materials (eds. Luyten, J. & Snijkers, F.)
(Fransaer, D., Brugge (Belgium), 2005).
111. Van Gestel, T., Meulenberg, W. A., Buchkremer, H., Van der Donk, G. J. W. &
Bouwmeester, H. J. M. in International conference inorganic membranes (eds. Bredesen,
R. & Raeder, H.) 104 (Lillehammer, Norway, 2006).
112. Xu, Q. Y. & Anderson, M. A. Sol-Gel Route to Synthesis of Microporous Ceramic
Membranes - Thermal-Stability of TiO2-ZrO2 Mixed Oxides. Journal of the American
Ceramic Society 76, 2093-2097 (1993).
113. Xu, Q. Y. & Anderson, M. A. Sol-Gel Route to Synthesis of Microporous Ceramic
Membranes - Preparation and Characterization of Microporous TiO2 and ZrO2 Xerogels.
Journal of the American Ceramic Society 77, 1939-1945 (1994).
114. Pacheco, G., Zhao, E., Valdes, E. D., Garcia, A. & Fripiat, J. J. Microporous zirconia
from anionic and neutral surfactants. Microporous and Mesoporous Materials 32, 175-
188 (1999).
115. Sekulic, J., ten Elshof, J. E. & Blank, D. H. A. A microporous titania membrane for
nanofiltration and pervaporation. Advanced Materials 16, 1546 (2004).
116. Zou, H. & Lin, Y. S. Structural and surface chemical properties of sol-gel derived TiO2-
ZrO2 oxides. Applied Catalysis A-General 265, 35-42 (2004).
117. Knowles, J. A. & Hudson, M. J. Preparation and Characterization of Mesoporous, High-
Surface-Area Zirconium(IV) Oxides. Journal of the Chemical Society-Chemical
Communications, 2083-2084 (1995).
118. Mamak, M., Coombs, N. & Ozin, G. Mesoporous yttria-zirconia and metal-yttria-zirconia
solid solutions for fuel cells. Advanced Materials 12, 198 (2000).
119. Yang, P. D., Zhao, D. Y., Margolese, D. I., Chmelka, B. F. & Stucky, G. D. Block
copolymer templating syntheses of mesoporous metal oxides with large ordering lengths
and semicrystalline framework. Chemistry of Materials 11, 2813-2826 (1999).
120. Hung, I. M., Hung, D. T., Fung, K. Z. & Hon, M. H. Synthesis and characterization of
highly ordered mesoporous YSZ by tri-block copolymer. Journal of Porous Materials 13,
225-230 (2006).
121. Sing, K. S. W. et al. Reporting physisorption data for gas/solid systems. Pure & applied
chemistry 54, 603-619 (1985).
122. Benes, N. E. Mass transport in thin supported silica membranes (PhD thesis, University
of Twente, Enschede, 2000).
123. De Bruijn, F. T. Pervaporation and vapour permeation of methanol and MTBE through a
microporous methylated silica membrane (PhD thesis, Technical University of Delft,
Delft, 2006).
124. Vroon, Z., Keizer, K., Burggraaf, A. J. & Verweij, H. Preparation and characterization of
thin zeolite MFI membranes on porous supports. Journal of Membrane Science 144, 65-
76 (1998).
125. Nakayama, K., Suzuki, K., Yoshida, M., Yajima, K. & Tomita, T. 33 (Patent, NGK
insulators, Ltd, Japan, 2005).
126. Van der Donk, G. J. W., Van Gestel, T. & Pyckhout-Hintzen, W. in Neutron scatteringg
at FRJ-2, Experimental reports 2005/6 (eds. Brückel, T., Richter, D. & Zorn, N.)
(Forschungszentrum Jülich GmbH, Jülich, 2006).
119

6 References
127. Suh, Y. W., Lee, J. W. & Rhee, H. K. Synthesis of thermally stable tetragonal zirconia
with large surface area and its catalytic activity in the skeletal isomerization of 1-butene.
Catalysis Letters 90, 103-109 (2003).
128. Diaz-Parralejo, A., Macias-Garcia, A., Cuerda-Correa, E. M. & Caruso, R. Influence of
the type of solvent on the textural evolution of yttria stabilized zirconia powders obtained
by the sol-gel method: Characterization and study of the fractal dimension. Journal of
Non-Crystalline Solids 351, 2115-2121 (2005).
129. Kaneko, K. & Ishii, C. Superhigh surface-area determination of microporous solids.
Colloids and surfaces 67, 203 (1992).
130. Barrett, P. A., Joyner, P. H. & Halenda, J. The determination of pore volume and area
distribution in porous substances: The computations from nitrogen isotherms. Journal of
the American Chemical Society 73, 373 (1951).
131. Horvath, G. & Kawazoe, K. Method for the Calculation of Effective Pore-Size
Distribution in Molecular-Sieve Carbon. Journal of Chemical Engineering of Japan 16,
470-475 (1983).
132. De Lange, R. S. A. (PhD thesis, University of Twente, Enschede, 1993).
133. Caro, J. & Noack, M. Zeolite Membranes - Status and Prospective. Advances in Porous
Materials 1 (In press).
134. Palimino, M. et al. Pure silica ITQ-32 zeolite allows separation of linear olefins from
paraffins. Chemical Communications 12, 1233-1235 (2007).
135. Van Gestel, T. et al. in International conference on porous ceramic materials (eds.
Luyten, J. & Snijkers, F.) (VITO, Brugge-Belgium, 2005).
136. Cushing, B. L., Kolesnichenko, V. L. & O'Connor, C. J. Recent advances in the liquidphase
syntheses of inorganic nanoparticles. Chemical Reviews 104, 3893-3946 (2004).
137. Saito, A. & Foley, H. C. Argon Porosimetry of Selected Molecular-Sieves - Experiments
and Examination of the Adapted Horvath-Kawazoe Model. Microporous Materials 3,
531-542 (1995).
138. http://www.geocities.com/ostroum/FTRAMAN.htm (March 2007).
139. Wan, Y. et al. Preparation of titania-zirconia composite aerogel material by sol-gel
combined with supercritical fluid drying. Applied Catalysis-A 277, 55-59 (2004).
140. Xomeritakis, G., Tsai, C. Y. & Brinker, C. J. Microporous sol-gel derived aminosilicate
membrane for enhanced carbon dioxide separation. Separation and Purification
Technology 42, 249-257 (2005).
141. Gu, Y. F., Kusakabe, K. & Morooka, S. Sulfuric acid-modified zirconia membrane for
use in hydrogen separation. Separation and Purification Technology 24, 489 (2001).
142. Gu, Y. F., Kusakabe, K. & Morooka, S. The separation of hydrogen from carbon dioxide
using platinum-loaded zirconia membranes. Journal of Chemical Engineering of Japan
35, 421-427 (2002).
120

Curriculum Vitae – George van der Donk
George van der Donk was born in Nijmegen at 1978. He started his study
chemical engineering at the University of Twente in 1997. His traineeship on
solid oxide fuels cells was performed at Risø National Laboratory in Denmark.
George graduated, as a member of the Inorganic Materials Science research
group, from the University of Twente. His thesis: “Oxygen coulometric titration of
mixed conducting perovskites”. After university he went to Forschungszentrum Jülich
GmbH (Germany) to start his Ph.D. For the Institute for materials and processes in
energy systems, in cooperation with the Ruhr University of Bochum, he wrote this
Ph.D thesis and other publications. Currently he is working as a Design Engineer at
Sensata Technologies Holland B.V. in Almelo (the Netherlands), where he develops
electrochemical sensors for the automotive industry.

Acknowledgement
First of all, I would like to thank my Doktorvater Professor Stöver for giving me the
opportunity to do this PhD. I am grateful to Buchkremer for his enthousiasm for this topic
about inorganic membranes and for his efforts to setup the membrane group in Jülich. I
would also thank Robert Vassen and his Thermal barrier layer group for their support.
The person who answered most of my scientific and social questions is my supervisor
Willi Meulenberg. With his motivation, we were able to create, on short notice, a wellaccepted
inorganic membrane group. Besides he also formed a bridge between the
German work culture and my Dutch culture.
The inorganic membrane group including Vicente, Martin Bram, Kampel, Hansch,
Wiener, Betz, Willi, Tim, Oli and Jose are gratefully acknowledged. Many of the
experiments and characterisations could not have been done without the help of e.g.
Sebold, Fischer, Lersch, Coenen, Pracht, Werner, Kicky, Mai, Hannelore, Vicky and
Kappertz from our institute and Wim Pyckhout-Hinzten, Michulitz, Esser, Peica,
Besmehn and Breuer from the other institutes of the Forschungszentrum Jülich. Professor
Gies, Bernd-“ich bin mahl gespannt”-Marler and Somsen from the Ruhr Universität
Bochum are thanked respectively for scientific zeolite discussions, cooking together D1H
and SEM/TEM analysis.
Mieke is thanked for introducing me in to the world of ceramic manufacturing. Rob
Delhez is gratefully acknowledged for the discussions on crystallinity. My joy in
inorganic membrane science was enhanced by the chats with Miquel Menendez, Joe Da
Costa, Frederic, Marcel den Exter and Rune Bredesen. The high quality of the article
containing the permeance of titania-zirconia membranes is mainly due to the effort of
Nieck Benes and Marcus van Schilt from the University of Eindhoven. Most of the
permeance measurements were performed by Marcus.
Due to the personalities of Nati and Jozef as well, our office quickly changed to the most
popular gossip and coffee room. Both roommates contributed to the positive work
atmosphere. Jürgen and Dirk, my carpool-mates, transported me daily to Jülich and back.
Thanks for that. It was an honour to defend several times the running-cup for the fastest
FZJ institute. This could not have been achieved without the frequent trainings with
Xabier, JB, Alexandra, Daniel and Ettler.
Special thanks are addressed to Henny Bouwmeester from the University of Twente. He
was constantly interested in my progress and in our project. This resulted finally in a
corporation between the University of Twente and FZJ. Many improvements to this
thesis are a result of scientific discussions with Henny. It was a great pleasure to write
two papers with Jose Manuel Serra Alfaro. His work mentality and scientific
thoroughness trigged me constantly.

Last but not least, I would like to thank all the helpful persons that I have forgotten to
mention by name. This thesis could not have been written without the support of my
friends, family and my girlfriend Jitka.
George van der Donk, November 2007

Schriften des Forschungszentrums Jülich
Reihe Energie & Umwelt / Energy & Environment
1. Einsatz von multispektralen Satellitenbilddaten in der Wasserhaushalts-
und Stoffstrommodellierung – dargestellt am Beispiel des
Rureinzugsgebietes
von C. Montzka (2008), XX, 238 Seiten
ISBN: 978-3-89336-508-1
2. Ozone Production in the Atmosphere Simulation Chamber SAPHIR
by C. A. Richter (2008), XIV, 147 pages
ISBN: 978-3-89336-513-5
3. Entwicklung neuer Schutz- und Kontaktierungsschichten für
Hochtemperatur-Brennstoffzellen
von T. Kiefer (2008), 138 Seiten
ISBN: 978-3-89336-514-2
4. Optimierung der Reflektivität keramischer Wärmedämmschichten aus
Yttrium-teilstabilisiertem Zirkoniumdioxid für den Einsatz auf metallischen
Komponenten in Gasturbinen
von A. Stuke (2008), X, 201 Seiten
ISBN: 978-3-89336-515-9
5. Lichtstreuende Oberflächen, Schichten und Schichtsysteme zur
Verbesserung der Lichteinkopplung in Silizium-Dünnschichtsolarzellen
von M. Berginski (2008), XV, 171 Seiten
ISBN: 978-3-89336-516-6
6. Politikszenarien für den Klimaschutz IV – Szenarien bis 2030
hrsg.von P. Markewitz, F. Chr. Matthes (2008), 376 Seiten
ISBN 978-3-89336-518-0
7. Untersuchungen zum Verschmutzungsverhalten rheinischer Braunkohlen
in Kohledampferzeugern
von Annette Schlüter (2008), 164 Seiten
ISBN 978-3-89336-524-1
8. Inorganic Microporous Membranes for Gas Separation in Fossil Fuel Power
Plants
by G. van der Donk (2008), VI, 120 pages
ISBN: 978-3-89336-525-8

Band | Volume 8
ISBN 978-3-89336-525-8