propositions of the doctoral dissertation entitled catalysis engineering of light induced dye
TRANSCRIPT
Propositions of the doctoral dissertation entitled
Catalysis Engineering of Light Induced Dye Degradation
and Cyclohexane Photo-oxidation
by Peng Du
1. Some of the earlier work on semiconductor photo-systems proved to be highly
irreproducible; this has not helped the subject to develop as rapidly as it might have, and
may have generated some degree of skepticism in the scientific community about
subsequent developments in the field.
A. Mills et al., J. Photochem. Photobiol. A, 108 (1997) 1
2. Methylene blue, which is a representative of the thionine dyes resistant to biodegradation,
has been proven to be little representative for photocatalytic degradation of organic dyes
and contaminants in general.
X.L.Yan et al. Chem. Phys. Lett., 429 (2006) 606
I.K. Konstantinou et al., Appl. Catal.B: Env. 49 (2004) 1.
Chapter 7 of this thesis.
3. Surface area is for conventional catalytic processes often one of the most important scaling
parameters. The photocatalytic activity, however, usually does not scale with the catalytic
surface area, due to the complex nature of photon-induced catalytic processes.
A. Sclafani et al., J. Phys. Chem., 100 (1996) 13655
Chapter 3 of this thesis.
4. Because in the studies of photocatalytic oxidation of organic compounds in non-aqueous
media, the effect of the applied wavelength on catalyst performance and selectivity was
typically not addressed, photolysis (not a catalytic process) was in various cases mixed up
with catalytic action.
P. Du et al., J. Catal. 238 (2006) 207
Chapter 5 of this thesis.
5. For a satisfactory industrial application of monolith based photocatalytic processes the two
major challenges left are the preparation of high-quality TiO2 coatings and a smart
introduction of light into the monolith channels.
Chapter 6 of this thesis.
6. At TU Delft, the 2nd
years course “Transport Phenomena” is a sustainable headache for
most students because of its high failing rates. The importance of the course will be
realized after entering the real world of the chemical industry, where mass and heat
transfer are the fundamentals of the engineering discipline.
7. The purpose of models is not to fit the data but to sharpen scientific thinking.
8. Gasification would be an important technology for renewable energy, as it can apply
practically all types of organic feedstock such as coal, oil and biomass as raw material,
while particularly for biomass it can be carbon neutral.
9. The word “research” is originating from old French, with the prefix ‘re-‘ meaning ‘really
intensively’. Understanding the proper meaning of “research” should minimize the efforts
of scientists to do “re-search”.
10. People who invented shoes must have never thought of “flying shoes” being the “weapon
of mass destruction”, but we can only hope that this new form of “terrorism” will be
replacing the traditional more violent form.
11. An expert is a person who has made all the mistakes which can be made in a very narrow
field.
12. China’s rise might have induced fear in the time of Napoleon who uttered the phrase
"quand la Chine s'éveillera, le monde tremblera". History has proven him wrong since
most Chinese are focused on the improvement of personal welfare, and show little interest
in becoming a rising superpower.
These propositions are considered opposable and defendable and as such have been
approved by the supervisors, Prof. Dr. J.A. Moulijn and Dr. G. Mul.
Stellingen behorende bij het proefschrift
Catalysis Engineering of Light Induced Dye Degradation
and Cyclohexane Photo-oxidation
door Peng Du
1. Vroeger onderzoek naar halfgeleider fotosystemen is bewezen zeer onreproduceerbaar te
zijn; dit heeft de ontwikkeling van het vakgebied fotokatalyse vertraagd, en kan een zekere
mate van scepticisme in de wetenschappelijke gemeenschap verklaren.
A. Mills et al., J. Photochem. Photobio. A, 108 (1997) 1
2. Methyleen Blauw, een veel onderzochte kleurstof op basis van thionine die moeilijk om te
zetten is door middel van biodegradatie, is niet erg representatief voor de fotokatalytische
afbraak van organische (kleur)stoffen in het algemeen.
X.L.Yan et al. Chem. Phys. Lett., 429 (2006) 606
I.K. Konstantinou et al., Appl. Catal.B: Env. 49 (2004) 1.
Hoofdstuk 7 van dit proefschrift.
3. Oppervlak is voor conventionele katalytische processen één van de meest belangrijke
parameters die de activiteit per gram katalysator bepaald. De fotokatalytische activiteit is
echter niet noodzakelijk afhankelijk van oppervlak vanwege de complexiteit van foton-
geïnduceerde katalytische processen.
A. Sclafani, et al., J. Phys. Chem., 100 (1996) 13655
Hoofdstuk 3 van dit proefschrift.
4. In onderzoek naar fotokatalytische oxidatie van organische verbindingen in niet-waterige
media is het effect van de toegepaste golflengte op katalysator activiteit en selectiviteit
doorgaans niet meegewogen.
P. Du et al., J. Catal. 238 (2006) 207
Hoofdstuk 5 van dit proefschrift.
5. Aanbrengen van hoog oppervlakkig titania op de wand van monoliet kanalen en efficiënte
introductie van licht hierin zijn de belangrijkste uitdagingen om fotokatalytische
conversies te introduceren in de industrie.
Hoofdstuk 6 van dit proefschrift.
6. Bij de TU Delft is de 2e jaars cursus "Fysische Transportverschijnselen" een duurzaam
‘hoofdpijnvak’ voor de meeste studenten door het lage slagingspercentage. Het belang van
de cursus realiseert men pas wanneer men gaat werken in de echte wereld van de
chemische industrie, waar massa-en warmte-overdracht de fundamenten van de
engineering discipline blijken te zijn.
7. Het doel van modelleren is niet zozeer om data te verklaren, maar met name om de geest te
scherpen.
8. Vergassing zou een belangrijke technologie kunnen worden voor hernieuwbare energie en
koolstofneutrale operatie, omdat het van toepassing kan zijn voor vrijwel alle typen
biologische grondstoffen en biomassa.
9. Het woord "research" is ontstaan uit het oude Frans, waarin het voorvoegsel 're-' ‘heel
intensief’ betekent. Inzicht in deze betekenis van "research" zou de inspanningen van
wetenschappers om "re-search" te doen, kunnen minimaliseren.
10. Mensen die schoenen hebben uitgevonden zullen nooit gedacht hebben dat "vliegende
schoenen" als een soort "massavernietigingswapen" zouden kunnen worden toegepast. We
kunnen alleen maar hopen dat deze nieuwe vorm van ‘terrorisme’ de traditionele
geweldadige vorm zal gaan vervangen.
11. Een deskundige is iemand die alle fouten heeft gemaakt, die kunnen worden gemaakt in
een zeer smal onderzoeksveld.
12. De opkomst van China heeft wellicht tot angst geleid in de tijd van Napoleon, die de
zinsnede uitte: "quand la Chine s'éveillera, le monde tremblera". De geschiedenis heeft
aangetoond dat hij het bij het verkeerde eind had, aangezien de meeste Chinezen gericht
zijn op verbetering van persoonlijk welzijn, en weinig interesse tonen om een nieuwe
supermacht te worden.
Deze stellingen worden opponeerbaar en verdedigbaar geacht en zijn als zodanig goedgekeurd
door de promotoren, Prof. Dr. J.A. Moulijn and Dr. G. Mul.
Catalysis Engineering of Light Induced Dye Degradation and
Cyclohexane Photo-oxidation
Catalysis Engineering of Light Induced Dye Degradation and
Cyclohexane Photo-oxidation
Proefschrift
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof.dr.ir. J.T. Fokkema,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen op 24 februari 2009 om 15:00 uur
door
Peng Du
scheikundig ingenieur
geboren te Zhejiang, China
Dit proefschrift is goedgekeurd door de promotor:
Prof. Dr. J.A. Moulijn
Samenstelling promotiecommissie:
Rector Magnificus voorzitter
Prof. dr. J.A. Moulijn Technische Universiteit Delft, promotor
Dr. G. Mul Technische Universiteit Delft, copromotor
Dr. R. van de Krol Technische Universiteit Delft
Prof. dr. A.I. Stankiewicz Technische Universiteit Delft
Prof. dr. L.Lefferts Technische Universiteit Twente
Prof. dr. H.J. Heeres Rijksuniversiteit Groningen
Prof. dr. D. Bahnemann Universiteit Hannover
This research reported in this thesis was carried out at the Catalysis Engineering group,
DelftChemTech, Faculty of Applied Science, Delft University of Technology (Julianalaan 136,
2628 BL, Delft, The Netherlands), with financial support of the Stichting Technologische
Wetenschappen (STW, the Simon Stevin Meesterschap awarded to Prof. Dr. J.A. Moulijn).
Proefschrift, Technische Universiteit Delft
met samenvatting in het Nederlands
Copyright © 2008 by Peng Du
All rights reserved
Contents
Chapter 1 Introduction 1
1.1 Background 2
1.2 Heterogeneous photocatalysis – mechanistic aspects 4
1.3 Photocatalysts 7
1.4 Kinetics of photocatalysis 11
1.5 Photocatalytic reactors 13
1.6 Objectives & approach 14
Chapter 2 A combinatorial approach towards photocatalytic oxidative
decolorization of methylene blue over titania materials 23
2.1 Introduction 24
2.2 Experimental 25
2.3 Results and discussion 27
2.4 Conclusions 34
Appendix 2.1 Determination of mass transfer parameters in slurry reactor 36
Chapter 3 Effect of TiO2 source and thermal pre-treatment on photoactivity
for methylene blue degradation in water 45
3.1 Introduction 46
3.2 Experimental 47
3.3 Results 48
3.4 Discussion 62
3.5 Conclusions 69
Appendix 3.1 Photocatalytic decolorization of Erythrosine B (EB) and Congo Red (CR) 71
Chapter 4 The effect of surface OH-population on the photocatalytic activity of
rare earth doped P25-TiO2 in methylene blue degradation 77
4.1 Introduction 78
4.2 Experimental 79
4.3 Results 81
4.4 Discussion 91
4.5 Conclusions 95
Chapter 5 Effect of irradiation energy and TiO2 structure on the rate of photo
-oxidation of cyclohexane and side product formation 99
5.1 Introduction 100
5.2 Experimental 102
5.3 Results 105
5.4 Discussion 119
5.5 Conclusions 123
Chapter 6 A novel photocatalytic monolith reactor for multiphase heterogeneous
photocatalysis 127
6.1 Introduction 128
6.2 Experimental 129
6.3 Results 135
6.4 Discussion 139
6.5 Conclusions 143
Chapter 7 Conclusions and outlook 147
7.1 Conclusions 148
7.2 Outlook 150
Samenvatting 155
Acknowledgements 159
Publications and oral presentations 161
Curriculum Vitae 163
1
1
Introduction
Abstract
Looking into the history of chemistry, one of the fascinating facts is the discovery and utilization
of solar irradiation as a clean and safe energy supply. Realizing that light plays a crucial role in our
daily life, we are moving steadily in a constructive and positive direction in the establishment and
development of clean photofunctional systems. Photocatalysis, which in its most simplistic description
denotes the acceleration of a photoreaction by the action of a catalyst [1], has been widely studied as a
mean of air and water purification treatment and organic synthesis. Semiconductors, with in special
Titania (TiO2), is by far the most attractive and promising photocatalyst in view of photo-oxidation
potential and chemical stability. In this introduction the mechanism of photocatalytic process in liquid
phase and properties of catalyst are discussed. Being widely applied as a standard test reaction of
wastewater treatment, photocatalytic degradation of methylene blue is described in detail. In case of
organic synthesis, direct oxidation of cyclohexane by molecular oxygen represents a large class of
commercial oxidation processes. A description of the attempts and possibilities to a photocatalytic
alternative route of this reaction is also provided. Furthermore, an overview of the reactor design with
regards to the commercial application of photocatalysis is presented.
Chapter 1
2
1.1 Background
Photocatalysis is a fast growing area with respect to both applied and fundamental research. The
increasing scientific interest in this field is reflected by the expanding number of publications that deal
with theoretical and practical applications of these reactions (Fig. 1). In the early seventies, Fujishima
and Honda discovered that water could be photocatalytically split into hydrogen and oxygen on TiO2
electrodes [2]. This marks the beginning of the development of heterogeneous photocatalysis. Since
then, the photocatalytic activities of semiconductors, mainly titania based, are studied in a manifold
ways and various applications have been developed.
Figure 1. Number of publications regarding TiO2 based heterogeneous photocatalysis in English
journals (CAplus source). Hydrodesulfization (HDS) is listed for comparison.
Figure 2 indicates most active fields and their current status in the researches on TiO2
photocatalysis, which is the most widely applied photocatalyst. The story began with
photoelectrochemical solar energy conversion and then shifted into the area of environmental
photocatalysis, including both air purification and wastewater abatement, and most recently into the
area of the self-cleaning surfaces due to the photoinduced hydrophilicity. Several excellent reviews
have been written over various aspects of photocatalysis, especially on the topic of environmental
cleaning in both air and aqueous phases [3-21].
By far, the most active field of TiO2 photocatalysis is the photodegradation of organic compounds
in air and water. TiO2 has become a photocatalyst in environmental decontamination for a large variety
of organics, viruses, bacteria, fungi and cancer cells, which can be totally degraded to CO2 and H2O,
and harmless inorganic ions. The superior performance is attributed to the formation of highly active
oxidizing holes and hydroxyl radicals. Hydroxyl radicals are almost the most powerful of all the
available oxidants in terms of oxidation potential. The oxidation potential of this radical is 2.80 V
versus NHE, being only slightly exceeded by fluorine.
Heterogeneous photocatalysis in organic synthesis is a less explored field. However, the
possibility to induce selective, synthetically useful redox transformations has become increasingly
more attractive and promising. Studies demonstrated that photocatalysis could yield different product
distributions compared with other oxidation means, although the productivities were extremely low.
0
400
800
1200
1976
1982
1988
1994
2000
2006
Nu
mb
er
of
pu
bli
cati
on
s
Photocatalysis
HDS
Introduction
3
Figure 2. Application fields of photocatalysis
Along with the development of commercial photocatalysts, the efficient utilization of solar energy
becomes one of the major goals that will have a great impact on technological applications of
photocatalysis.[22-25]. The widespread technological use of TiO2 is, however, hampered by its wide
band gap, which requires ultraviolet irradiation for photocatalytic activation. Because of the limited
fraction of UV in solar light (8%) compared to visible spectra (45%), any shift in the optical response
of TiO2 from the UV to the visible spectral range will have a profound positive impact on the
photocatalytic efficiency of the material [26,27]. Early approach towards photocatalysis using visible
light was the doping of TiO2 with transition-metal elements [28-34]. These studies show some positive
results, especially within certain dopant concentrations. However, metal doping has several intrinsic
drawbacks. The doped materials have been shown to suffer from thermal instability, and the metal
centers act as electronic traps, which reduces the photocatalytic efficiency. Furthermore, the
preparation of transition-metal doped TiO2 requires more expensive ion-implantation techniques
[35,36].
Recent research advances have been made in the design and development of highly reactive and
functional titanium oxide photocatalysts for utilization of only UV but also visible or solar light by
using anionic dopant species [26,27,37,38], and a clarification of the active sites as well as the
detection of the reaction intermediates at the molecular level.[39-42]. Highly dispersed titanium oxide
species prepared within zeolite frameworks as well as SiO2 or Al2O3 matrices showed much higher and
unique photocatalytic performances as compared to bulk TiO2 photocatalysts.
Along with these lines, detailed studies into the characterization of TiO2 nano-particles and
various TiO2 based molecular catalytic systems have been carried out using molecular spectroscopy
techniques. Two main objectives were sought: improving the photocatalytic reactivity and its
efficiency [43-47], and the design and development of TiO2 photocatalysts which are able to absorb
and work not only under UV but also visible or solar light irradiation [48-52].
All these studies paved a new path towards the improvement of the photocatalytic reactivity and
its efficiency, and the design and development of TiO2 photocatalysts which are able to absorb and
work not only under UV but also visible or solar light irradiation.
+TiO2 Light
Decomposition of aldehyde
Removal of NOx
Air purificationDecomposition of organics
Municipal water sterilization
Decomposition of virus
Water purification
Decomposition of oil
Superhydrophilic effect
Self-cleaninganti-fogging
Anti-contamination
Hydrogen production
Artificial synthesis
Energy conversion
Production of monomers
Selective oxidation
Organic synthesis
Lab scale
Commercial
Chapter 1
4
1.2 Heterogeneous photocatalysis – mechanistic aspects
Heterogeneous photocatalysis can be carried out in various media: gas phase, pure organic phase
or aqueous solutions. The overall process is controlled by several steps: mass transfer of reactants to
catalyst surface, adsorption of the reactants, light absorption creating electrons (e-) and holes (h+),
transport of photogenerated charges to the adsorption sites, reaction of the adsorbed species,
desorption of products and removal of the products from the catalyst surface [14]. It is of crucial
importance for a photoinduced catalytic activity that the photocatalysts absorb photons and adsorb
reactants simultaneously.
Figure 3. Major processes occurring on a photocatalyst particle [5,14].
Detailed surface reaction mechanism of the photocatalytic process is very complicated and
remains far from clear, particularly that concerning the initial steps involoved in the reaction of
reactive oxygen species and organic molecules. Despite of the debates on the surface reactive species
and the localization of various reactions, all photocatalytic reactions proceed through the primary
excitation process resulting in charge separation of electorn-hole pairs. When a photocatalyst, typically
a semiconductor material of the chalcogenide type (oxides TiO2, ZnO, ZrO2, CeO2, etc.) or sulfides
(CdS, ZnS, etc.), is illuminated with photons with an energy exceeding the bandgap energy Egap (hν ≥
Egap), an electron (e-) is promoted from the valance band to the conduction band. At the valance edge,
an electronic vacancy or hole (h+) is created. In the following TiO2 is taken as an example:
)(22
+− +→ν+ heTiOhTiO (1.1)
The holes and electrons formed after the charge-carrier generation participate in several pathways
in the photocatalytic catalysis. The electron-hole pair can rapidly recombine, especially when the
concentrations of e- and h+ in the catalyst particle are high. This crucial reaction reduces the efficiency
of photocatalytic processes as the energy is lost as heat:
heathe →+ +− (1.2)
----
+
hνννν
----
+
conduction band
valence band+
---- hννννvolume recombination
surface recombination
D
D+
A
A+
Introduction
5
In order to proceed photocatalysis effectively, the photogenerated electrons must be removed from
the catalyst particle. A good example is photocatalytic oxidation in the aqueous phase. In the presence
of molecular oxygen, the photo-generated electrons are sufficiently strong in the reduction power to
produce superoxide (O2-) with adsorbed oxygen.
⋅→+ −−22 OOe (1.3)
The superoxide is an effective oxidizing agent that attacks neutral substrates as well as
surface-adsorbed radicals and/or radical ions. Theoretically, the redox potential of the electron-hole
pair permits H2O2 formation, either by water oxidation by holes or by the reduction of the adsorbed
oxygen involving two conduction band electrons.
++ +→+ HOHhOH 222 222 (1.4)
222 22 OHHeO →++ +− (1.5)
Hydrogen peroxide contributes to the photocataytic degradation pathways through hemolytic
scission yielding hydroxyl radicals.
Independent on the absence of acceptors, electrons can also be trapped by coordination defects at
the surface (shallow trap), which could still participate in photocatalysis by hopping or by thermal
emission of free carriers, or by lattice defects in the bulk (deep trap) inevitably leading to
recombination with a hole [53,54].
Photon-activation of electrons creates vacancies (holes) on the valence band of TiO2 that can
receive electrons from donors with the potential level to be above (more negative than) the valence
band edge of TiO2. Due to the low band edge, the hole is a strong oxidant and can oxidize organic
molecules at the surface through surface bound hydroxyl radicals, eventually mineralizing them to
CO2.
++ ⋅>→>+ }{ OHTiOHTihIVIV
(surface-bound hydroxyl radical) (1.6)
The surface-bound hydroxyl radicals are assumed to be the primary oxidizing species in the
photocatalytic oxidation of organics [55,56]. As illustrated in Table 1, the hydroxyl radicals is one of
the most powerful oxidizing species available. Utilization of this oxidation power results in reactions
that are a billion times faster than reactions with typical oxidants such as ozone (O3) or hydrogen
peroxide (H2O2) [57,58].
Similarly, the hole can oxidize water or hydroxide ions to form hydroxyl radicals, which are also
efficient oxidants of organic molecules.
++ +⋅→+ HOHOHh 2 (1.7)
OHOHh ⋅→+ −+ (1.8)
Both surface-bound hydroxyl radicals (eqs. 1.6) and free hydroxyl radicals (eqs. 1.7, 1.8) can
react with adsorbed organic compound, via abstraction of H atoms by ·OH radicals by C-H bond
cleavage. The resulting radical carbon can react with oxygen to form oxygenated compounds, or
proceed further with adjacent species through radical transfer.
OHROHRH 2+⋅→⋅+ (1.9)
Chapter 1
6
Equations (1.1) through (1.9) summarize the important initial steps of catalyst activation by
photons. Further reaction of the photo-generated active species with surface adsorbed organic
compounds proceeds through a radical reaction chain mechanism. Depending upon the reaction
conditions, the holes, ·OH radicals, O2-·, H2O2 and O2 can play important roles in photocatalytic
reactions.
Table 1. Oxidation potentials of some oxidants [57]
Species Oxidation potential [V] Species Oxidation potential [V]
Fluorine 3.03 Hypobromous acid 1.59
Hydroxyl radical 2.80 Chlorine dioxide 1.57
Atomic oxygen 2.42 Hypochlorous acid 1.49
Ozone 2.07 Hypoiodous acid 1.45
Hydrogen peroxide 1.78 Chlorine 1.36
Perhydroxyl radical 1.70 Bromine 1.09
Permanganate 1.68 Iodine 0.54
Recently Nakamura et al. studied the surface intermediates of photocatalytic reactions on
nanocrystalline TiO2 films in contact with aqueous solutions using multiple internal reflection infrared
spectroscopy [59-61]. Characteristic IR bands were assigned to short lifetime intermediates, i.e.
surface peroxo and surface hydroperoxo species. On the basis of the IR studies, they proposed an
alternative reaction scheme for the photocatalytic reduction of O2 at the TiO2 surface, initiated by
electron capture at H2O-adsorbed Ti4+ sites. The surface peroxo species, Ti(O2), is primarily produced,
probably via Ti-OO⋅ as a precursor, which is then transformed to the surface hydroperoxo, TiOOH by
protonation in the dark.
Table 2. Characteristic timescales for TiO2-sensitized photooxidative mineralization of organic
compounds [9]
Characteristic times for the various initial surface reaction steps have been determined by laser
flash photolysis experiments. Results are summarized in Table 2. Election-hole pair generation upon
absorption of a photon is extremely fast (fs). On the basis of the measurements by Martin et al. [62,63],
it was determined that trapping of electrons and holes happens on the nanosecond scale (~ 0.1-10 ns).
Primary process Characteristic time
Generation of electron/hole pair
)(22
+− +→ν+ heTiOhTiO
fs (very fast)
Trapping of electron/hole pairs
++ ⋅>→>+ }{ OHTiOHTihIVIV
−− >↔>+ }{ OHTiOHTieIIIIV
−− >→>+ }{ IIIIVTiTie
10 ns (fast)
100 ps (shallow trap: dynamic equilibrium)
10 ns (deep trap)
Electron/hole recombination
}{}{ OHTiOHTieIVIV >→⋅>+ +−
}{}{ OHTiOHTihIVIII >→>+ −+
100 ns (slow)
10 ns (fast)
Reaction at the surface
{ } organic pollutant { } oxidized pollutantIV IVTi OH Ti OH
+> ⋅ + → > +
⋅+>→+> −−22 }{}{ OOHTiOOHTi
IVIII
100 ns (slow)
ms (very slow)
Introduction
7
Recombination has a characteristic time of 10 to 100 ns. Surface reaction of the holes is slow (~ 100
ns), but the slowest step is the interfacial charge transfer of electrons to the electron acceptor (ms).
Ohko et al. [64] also deduced similar characteristic times, based on the photocatalytic decomposition
of gaseous 2-propanol on titanium dioxide thin films under very weak UV light.
In order for photocatalysis to be efficient, electron/hole pair recombination must be suppressed
before the surface reactions occur at the interface. The recombination reaction occurs relatively fast
with respect to surface reaction (microseconds to milliseconds), and the resulting low quantum
efficiency is one of the main impediments for the use of photocatalysis. It has been observed that the
photocatalytic activity is completed suppressed in the absence of an electron scavenger such as oxygen.
An increase in either electron/hole lifetime or the interfacial electron-transfer rate is expected to lead to
higher quantum efficiency of photocatalysis. Gerischer and Heller have suggested that reduction of
oxygen is the rate-determining step for most surface limited photocatalytic reactions [65-67].
1.3 Photocatalysts
A photocatalyst is characterized by its capacity to simultaneously adsorb reactants and absorb
photon energy. Two reactants can be reduced and oxidized respectively by a photonic activation
through an efficient absorption (hν ≥ Eg). Figure 4 shows the band gap of several semiconductors and
the standard redox potentials of water. The ability of a semiconductor to undergo photoinduced
electron transfer to an adsorbed particle is governed by the band energy positions of the semiconductor
and the redox potential of the absorbates. From the thermodynamic point of view, adsorbed couples
can be reduced photocatalytically by conduction band electrons if the lower redox potential is more
negative than the conduction band, and the higher redox potential is more positive than the
semiconductor valence band [68,69].
Figure 4. Band gap positions (top of valence band and bottom of conduction band) in various
semiconductors. The energy scale is indicated in electron volts using either the vacuum level (left) or
the normal hydrogen electrode (NHE) (right) as a reference. [77,78]
Vacuum level
-3.0
-4.0
-5.0
-6.0
-7.0
-8.0
0
-4.5
TiO2Rutile
TiO2Rutile
3.0
TiO2Anatase
3.2
SrTiO3
3.2
FeTiO3
2.7
2.8
MnTiO3
3.2
ZrO2
BaTiO3
5.0
Nb2O5
3.4
3.4
KTaO3
WO3
2.8
2.2
ZnO2
3.2
Fe2O3
SnO2
3.8
GaP
2.3
1.1
Si
SiC
3.0
CdSe
CdS
1.7
2.5
E vs. NHE
@ pH = 0
H+/H2
-1.0
+2.0
+1.0
+3.0
0
eV
O2/OH-
Chapter 1
8
Of the semiconductors tested in literature (ZnO, CdS, Fe2O3, ZnS, WO3, SrTiO3 and TiO2), TiO2
has been found to be the best catalyst for photocatalytic degradation of organic substances in water,
although some reports do show that sometimes ZnO seems to have a higher activity [70-73]. It
appeared to be a suitable alternative to TiO2; however, ZnO is unstable due to dissolution and reaction
in water [74-76] to yield Zn(OH)2 on the ZnO particle surfaces and thus leading to catalyst
deactivation over time.
Figure 5. Spectra classification and solar irradiance spectra
Titanium dioxide is widely used as a white paint pigment, as a sunblocking material, as a
cosmetic, and as a builder in vitamin tablets among many other uses. It is almost an ideal class of
photocatalysts, due to its high photoactivity, ability to uitilize UV or visible light (Figure 5),
biologically and chemically inertness, photostability and low price. Among the commercial types of
TiO2, Degussa P25, which is a nonporous 70-30% anatase-to-rutile mixture with a BET surface area of
55 ±15 m2/g and crystalline size dp = 20-30 nm in a primary particle with d(aggregate) = 0.1-3 microns
[6,79,80], is found to be highly active in most researches and has been set as the standard photocatalyst
[81,82]. Values for the flat band potential of the conduction band and valance band of Degussa P-25
have been calculated as −0.3 and +2.9V (pH 0), respectively [62]. Despite of its microcrystalline
nature, Degussa P25 exhibits a fairly regular morphology, and that its early thermal transformation
into rutile is likely due to some rutile microcrystallites present, as an overlayer, on some of the anatase
crystallites [83-85].
Others have reported that Hombikat UV100 from Sachtleben GmbH exhibited higher activity in
certain photocatalytic reactions [86-91]. These ambient temperature photocatalysts, i.e. Degussa P25
and Hombikat UV100, are now known to oxidize virtually all organic water contaminants, given a
dissolved oxygen supply, including both common molecular solutes and microbial cells, viruses,
biopolymers, and oils [92-94].
Titania crystalline structure
Titania (TiO2) is industrally produced by two basic processes. Both use mineral ilmenite (FeTiO3)
as a raw material. The first is the ‘sulphate’ technology where ilmenite is leached by sulphuric acid and
engendering TiOSO4 is decomposed by steam to TiO2 (used in Hombikat UV100 production). The
second ‘chloride’ technology (used in Degussa for production of P25) is based on chlorination of
ilmenate and resulting TiCl4 is after purification oxidised by oxygen to TiO2 [95-97]. Selection of
either of these two processes is based on a number of factors, including raw material availability,
X-Rays UV Visible IR Radio
VUV UV-C UV-B UV-A λλλλ
40 nm 400 nm
400 nm320 nm280 nm200 nm40 nm
750 nm 1 mm
Solar Irra
diance
[W
/m
2/nm]
0
1
2
Wavelength [nm]
0 1000 2000 3000
Introduction
9
transportation and waste disposal costs. The chloride process is less environmental invasive. On the
other hand, the sulfate route presents the advantage that different TiO2 phases and titanium chemicals
can be made from one process. Currently, approximately 47% of TiO2 pigments are made by the
sulfate manufacturing process and 53% by the chloride process.
Three major crystalline configurations of titanium dioxide exist: rutile (tetragonal, a = b = 4.584
Å, c = 2.953 Å), anatase (tetragonal, a = b = 3.782 Å, c = 9.502 Å), and brookite (rhombohedrical, a =
5.436 Å, b = 9.166 Å, c = 5.135 Å) [98]. Other structures exist as well, however, only rutile and
anatase play any role in photocatalytic applications and are of the interest here. Thermodynamic
calculations based on calorimetric data indicate that rutile is the most stable phase at all temperatures
and pressures up to 60 kbar [99]. However, anatase is kinetically stable under normal conditions
because the phase transformation into rutile at room temperature is too slow to be detectable. Only at
temperatures > 600°C, the transformation reaches a measurable speed [100,101].
The phase stability is also affected by the crystalline size. Zhang et al. indicated that the relative
phase stability may reverse for small crystallites due to the surface-energy effect [102,103]. Other
factors that may influence the phase transformation from anatase to rutile are the lattice and surface
defects, alien ions, and pressure. An increase of surface defects and bulk oxygen vacancies enhances
the rutile transformation rate, as theses defects act as nucleation sites. Interstitial ions, whose sizes are
too large to substitute the lattice Ti4+
, decrease the concentration of oxygen vacancies and inhibit the
transformation. It is generally observed that substitutional ions with valence less than 4 (i.e., Cu2+
, Cr3+
,
Co2+, Li+) facilitate the anatase-to-rutile phase transformation [104-106]. Contrary effects were found
in anatase with Ti4+ substituted with ions of valance greater than four, as well as for the substitution of
an oxygen ion with two F- or Cl- ions [107,108]. This ‘substituted ion effect’ can be explained by the
changing in the strain energy which must be overcome before structural rearrangement can occur.
Photocatalytic activity of titania
The overall photocatalytic activity of titania is determined by the interplay of properties like
crystalline structure, catalyst surface area,, density of surface hydroxyl groups, surface acidity, number
of defects and adsorption/desorption characteristics. Moreover, the way of catalyst utilization, either in
slurry or fixed on a catalyst support, and the manner of light harvesting and reaction arrangement have
complicated influences on the apparent photocatalytic efficiencies [68]. The profound impact of these
factors will be discussed in following paragraphs.
Tanaka has described the relationship between the crystallographic phase of titania and its
catalytic activity during the decomposition of many organic compounds such as aromatics, commonly
present in contaminated water. In principle, anatase has always been found as the best photocatalyst for
use in aqueous solution [109-111]. However, rutile has been shown to be effective at both oxidative
and reductive chemistry in specific applications [112]. Sopyan et al. synthesized efficient TiO2 powder
with the rutile structure which showed much higher photoactivity than Degussa P25 in
photodegradation of acetaldehyde [113]. It is worthwhile to report that the photocatalytic activity of
amorphous TiO2 is negligible indicating that crystallinity is an important requirement [114].
Sclafani and Hermann pointed out that, unlike conventional catalytic processes, the photocatalytic
activity is not necessary dependent on catalyst surface area but rather on availability of active sites
[115]. A large surface area can be the determining factor in certain photocatalytic reactions, as a large
amount of adsorbed species promotes the surface reaction rate [116-120]. However, powders with a
Chapter 1
10
large surface area are usually associated with large amount of crystalline defects, which facilitate the
recombination of electrons and holes, leading to a poor photocatalytic activity [117,121,122].
The surface hydroxyl groups have been recognized to play an important role in the
photodegradation process due to direct involvement in reaction to generate principle oxidizing
agent ·OH, and indirect participation as the adsorption sites for reactants. The knowledge of this
quantity is of great interest in view of the overall photocatalytic activity. Van Veen et al. has developed
a method for the quantitative determination of the basic, acidic and total surface hydroxyl contents of
TiO2 [123]. Chhor et al. reported the surface hydroxyl group concentration of 8.7 µmol/m2 for Degussa
P25 and 2.9 µmol/m2 for Hombikat respectively [124].
Many infrared (IR) spectroscopy studies revealed the existence of two types of hydroxyl groups
and surface chemisorbed water [125-127]. At thermodynamic equilibrium, the morphology of anatase
was found to be a truncated bipyramid exhibiting only the (101) and (001) facets, whatever the
pressure and temperature [128,129]. Arrouvel et al. made a systematic approach of the anatase surface
hydration process as a function of temperature and pressure [130]. It is found that the mode and
amount of the surface coverage by hydroxyl groups are strongly influenced by the equilibrium
conditions. The concentration of surface OH/chemisorbed H2O decreases with increasing temperature.
At (001) surface, H2O adsorbs dissociatively leading to the surface Ti-O bond breaking, and the
simultaneous formation of two hydroxyl groups with a strong intramolecular hydrogen bond. On the
contrary, water molecules are chemisorbed on (101) surface without dissociation. The chemisorbed
water molecules are relatively unstable, as in example at surface coverage of 5.0 H2O/nm2, the fully
non-dissociated state and the fully dissociated state can compete in energy within less than 7 kJ/mol.
Kozlov et al. studied the effect of the acidity of TiO2 surface on its photocatalytic activity in
acetone and benzene gas-phase oxidation reactions [131,132]. It was shown that the TiO2 activity
strongly depends on the concentration of acidic and basic sites on the surface. Samples characterized
by strong acidity of the surface are more active in these reactions, as is contributed to the change in the
adsorption energy of the reactants on their surface.
Improving Photocatalytic activity by Catalyst Modifications
The surface characteristics of the photocatalyst can be modified by several pre-treatments such as
doping with transition and/or noble metals, Sensitizing the catalyst with a dye, forming composite
semiconductors and subsitituting oxygen with anions. The purposes of these modifications are:
- to increase the light absorption capability on the TiO2 catalyst;
- to enhance reactant adsorption capacity at the catalyst surface;
- to prevent recombination / enhance interfacial charge transfer as much as possible.
The effect of transition metal doping on the photocatalytic activity is a complex matter. Many
controversial results exist since even the method of preparation can lead to different morphological
and crystalline structures of the photocatalyst, hence the corresponding photocatalytic activity
[133,134].
Carp et al. and Litter made excellent reviews on the effects of metal doping on photocatalytic
activity [18,21]. The main objective of doping is to narrow the semiconductor band gap or introduce
intra-band gap states, which results in more UV-to-visible light absorption. The doped ions also act as
trapping sites for electrons and hole, hence altering the recombination rate. Since metal ion dopants
Introduction
11
occupy surface sites, the surface properties as well as the point of zero charge (PZC) may be altered by
doping. Consequently a modification of adsorption properties takes place. In case metal oxide exits as
an over-layer on the photocatalyst surface, it induces the charge separation which is beneficial for
photocatalysis. On the other hand, when the concentration of doping is high, the photon adsorption
efficiency of semiconductor photocatalyst will be affected. An optimum concentration of dopant ions
is frequently observed as the result of aforementioned doping effects.
Another modification of TiO2 is by sensitizing the catalyst with a dye to extend the light
absorption to visible spectra region. Dyes with high absorptivity in the visible region that have been
used as sensitizers include ruthenium(II) trisbipyridine, erythrosine B, rhodamine, thionine, and
phtahlocyanines [135-142]. The dye is firstly excited by visible light to the metastable state, which in
its turn injects electron into the semiconductor photocatalyst conduction band. The injected electron
reacts with surface adsorbed O2 to yield O2-·. It produces HO2
-· on protonation leading to the reduction
of the organic molecules or of the dye itself.
Furthermore, an increase in activity can be obtained when using composite
semiconductor-semiconductor photocatalysts [5,143-145]. The coupling of two semiconductors,
possessing different energy levels for their corresponding conduction and valence bands, provides an
approach to achieve a more efficient charge-separation, suppressed electron/hole recombination rate,
and extended light absorption range.
One new approach to enhance photocatalytic activity is to substitute oxygen with inorganic ions
(N3+,. C4+, S2-, F-) which induces visible light activation due to the band gap narrowing
[26,38,146,147]. There has been a fast growing interest in this area. Many techniques have been used
to produce visible light active TiO2-xNx photocatalyst such as laser sputtering [26,148,149], CVD [150],
mechanochemical reaction [151], ion implantation [152], sol-gel [153,154] and NH3 annealing
[155-157]. Activity in the visible-light region of these doped TiO2 samples has been demonstrated,
together with the shift of absorption edge of the photocatalyst.
1.4 Kinetics of photocatalysis
Most kinetic models used in photocatalysis are based on the Langmuir-Hinshelwood mechanism
confirming the heterogeneous catalytic character of the system [3,6,14,55,158-160]. This law
successfully explains the kinetics of reactions that occur between two adsorbed species, a free radical
and an adsorbed substrate, or a surface-bound radical and a free substrate. The initial rate of substrate
removal Ri varies proportionally with the surface coverage (θ), and the adsorption equilibrium of the
substrate follows a Langmuir isotherm giving as the result:
i
i
iKC
kKCkR
+=θ=
1 (1.10)
where Ci is the initial concentration of the substrate; t is the reaction time; k is the
Langmuir-Hinshelwood specific reaction rate constant; and K is the adsorption equilibrium constant.
Both k and K depend on the catalyst utilized and the nature of the substrate.
Although in many cases the Langmuir-Hinshelwood model can be applied for long time span, it is
worthwhile to mention that a generalized model is absent for dependency of the photocatalytic
degradation rate on the experimental parameters for the whole treatment time, just due to the
complexity of the photocatalytic reaction mechanism.
Chapter 1
12
Besides the nature of the photocatalyst and the substrate, the rate of photocatalytic reactions is
influenced by various parameters, among others but not exclusively, catalyst loading, irradiation
wavelength, light intensity, temperature and pH. The effect of varying these parameters will be
discussed in the following paragraphs.
Catalyst loading
At low catalyst concentration, the initial rates of the reaction were found to be directly
proportional to the loading of the photocatalyst, whatever the catalyst is suspended in slurry or fixed
on a support, indicating the true heterogeneous catalytic regime. However, it was observed that above
a certain concentration, the reaction rate levels off and becomes independent of the catalyst
concentration. Most of the studies reported the optimal catalyst loading lies in range between 0.15 and
8 g/l, increasing with increasing light intensity [72,161-165]. These limits depend on the reaction
geometry and on the operating conditions, and correspond to the maximum amount of photocatalyst
being totally illuminated [14,19,166].
Irradiation wavelength
Irradiation source provides the photons required for the electron transfer from valence band to
conduction band of the photocatalyst. The actual band gap determines the threshold of light absorption.
For TiO2 with a band gap of 3.0 eV, it requires irradiation source with wavelength < 400 nm to activate
the catalyst. It should be noted that, although the photocatalyst itself is non-active under low energy
irradiation (high wavelength), the reactants might be able to absorb the light initiating homogeneous
photocatalysis and/or photolysis.
Light intensity
For a simple set of photocatalytic reaction that includes only electron/hole pair generation,
charge-carrier recombination and surface reactions, it is easily derived that at low intensities and
correspondingly low carrier concentrations, the rate of photocatalysis is proportional to the light
intensity. While at higher light intensity regime, the rate is dominated by the recombination of electron
and holes, hence a square-root dependence on the light intensity. Many researchers have verified this
behavior experimentally, and found some typical threshold value of 25 mW/cm2 for the transition from
linear to square-root dependence regime [167-171].
Increasing the incident photon rate results in an increase in the overall photocatalytic reaction rate,
until the mass transfer limitation is reached. This transition depends on the catalyst configuration and
on the flow regime in the reactor, and varies with each application [167].
Temperature
It is well known that the photocatalytic reaction rate is not much affected by temperature due to
the photonic activation [56]. The reported apparent activation energies usually lie in the low region of
a few kJ/mol compared to ordinary thermal reactions [9,168,171,173].
Herrmann gave a theoretical consideration on the effect of temperature on photocatalysis [14].
Introduction
13
The true activation energy is nil, as is confirmed by the low apparent activation energy measured in the
medium temperature range of between 20°C and 80°C. At very low temperatures (<0°C), the rate
limiting step becomes the desorption of the final product, resulting in an increase in the apparent
activation energy. On the opposite, for high temperature photocatalysis (>80°C) and tends to the
boiling point of water, the exothermic adsorption of reactant becomes disfavored and tends to become
the rate limiting step. Correspondingly, the apparent activation energy becomes negative.
pH
The pH of an aqueous solution significantly influences the overall efficiency of the photocatalytic
process, including the surface charge on the semiconductor particles, the size of the aggregates, and
the ability to adsorb reactants. [174-178].
TiO2 particles suspended in water are known as amphoteric due to the “titanol” moeity at the
surface, >TiOH. These “titanol” groups undergo the following surface acid-base equilibria:
++ +−↔− HOHTiOHTi 2 pKa1 (1.11)
OHOTiOHOHTi 2+−↔+− − pKa2 (1.12)
For Degussa P25, the pKa values have been measured as 4.5 for pKa1 and 8 for pKa2, which results in a
pH of zero point charge, pHZPC = 0.5 (pKa1 + pKa2) of 6.25 [169]. At pH < pHZPC, the TiO2 surface
accumulates a net positive charge due to the increasing fraction of total surface site presented as Ti-OH2+.
At high pH, equilibrium in eqs.(1.12) will shift to right towards a net negative surface charge due to a
significant fraction of total surface sites present as Ti-O-.
Another effect of the pH is the shift of the energies of the valence and conduction band edges by 59
mV per unit pH at ambient temperature, in accordance with Nernst’s law [179,180]. This shift will change
the ability of the electrons and holes to participate in redox reactions, namely valence band electrons more
potent and the conduction band holes less potent at higher pH.
1.5 Photocatalytic reactors
Whereas much research has been performed in the field of photocatalysis since the seventies, few
large-scale applications in chemical industrial exist. Main obstacle in this field remains to be the lack
of reaction engineering insights that results in the absence of an efficient reactor design [57,69,167].
Compared with conventional reactor systems, photocatlysis brings two additional variables,
namely photon and catalyst. The optimized combination of the incident light energy and the catalyst
system is of crucial importance for the reactor design. A well-designed photocatalytic system should be
able to achieve a maximized light efficiency, high illuminated surface area, and good mass transfer.
Furthermore, the pH, temperature, UV-source and the presence of foreign matters influence the
photocatalytic reaction rate.
The catalyst can be applied in a suspended or immobilized configuration. The use of TiO2 in
suspension could be efficient due to the large surface area available for the reaction. This system has
advantages, however, that the catalyst particles need to be separated from the products and that the UV
light penetration in the slurry is limited. With an immobilized system, catalyst separation is no more
required and the recovery of catalyst is relatively simple. Another advantage is the possibility to design
Chapter 1
14
a reactor that the total catalyst surface area can be illuminated. A few technical challenges remain to be
the obstacles for developing such system. To name a few, the mechanical strength of the immobilized
catalyst, uniform catalyst illumination, and the low load of photocatalyst are the drawbacks of this kind
of photocatalytic system.
1.6 Objectives and Outline of the Thesis
Despite of the rapid development in the photocatalytic research, quite a few uncertainties remain
in the scientific world around the established area of interest. Furthermore it is often seen that the
literature data bring ambiguous and sometimes contradictory results indicating poor consistencies
between different photocatalytic reactions as well as reactor systems. All of these can be attribute to
the two additional parameters, photon and catalyst, and its interaction that bring an extra dimension of
complexity and additional difficulties for the standard catalytic research approaches. Therefore it has
been one of the objective of this research to revisit the photocatalytic system from mainly experimental,
engineering point of view whilst tackling some of the theoretical fundamentals of photocatalysis.
Another objective is to make the first movement towards the commercialization of photocatalytic
system in the conventional chemical industry, of which the application of photon might propose a
potential breakthrough.
Due to the large varieties of applied photocatalysts, it is a priori to apply a fast and quantitative
catalyst screening system of which multiple photocatalytic reactions can proceed in parallel under
comparable conditions. The outline of this effort as well as the results has been presented in Chapter 2.
A novel photocatalyst screening system was assembled and verified for successful application of dye
degradation process in water, despite of its inherent constraints limiting its applications in a wider area
of interest. The positive outcome of the novel photocatalyst screening system provides a solid basis for
the further photocatalyst activity studies that show more insights into the complex interactions of
reaction intrinsic kinetics, surface chemistry with reactant and photons, and the transportation
characteristics of reactants/products.
As is discussed before, the activity of photocatalyst is affected by multiple parameters, i.e.,
morphology, surface area, and surface hydroxyl groups. It could differ from one vendor to another, or
even in extreme cases from one batch to another, regardless of the same chemical formula and similar
apparent properties. The surface characteristics, and consequently the apparent photocatalytic activity,
could be modified by various pre-treatment procedures. Chapter 3 investigated the effect of TiO2
source and thermal pretreatment with calcined samples on the apparent photocatalytic activity,
applying methylene blue decolorization as the test reacction. Other two test reactions were investigated
as well, namely the photocatalytic decolorization of Erythrosine B and Congo red. Results were shown
in appendix 3.1, and the discussions were covered in Chapter 3.
Another surface modification method, doping with rare earth metal, was applied for the
commercial TiO2 P25 from Degussa. Results were discussed in Chapter 4. Photocatalytic degradation
of methylene blue was selected to be the test reaction in general, of which the chemistry as well as
engineering parts were found to be rather complicated. Various analytical methods were applied to the
modified catalysts, in order to evaluate dominant factors that control the apparent photocatalyst
performance.
In Chapter 5, the photocatalytic oxidation of organics in the absence of water was studied, which
is a less explored area in photocatalysis. The liquid phase photolytic oxidation of cyclohexane was
Introduction
15
compared with photocatalytic oxidation over TiO2 with varying wavelengths of light exposure, slurry
densities, and sources and pretreatments of catalyst material. As was discovered in a thorough
literature survey, large discrepancies exist in this specialized field of photocatalysis. Current study
targeted to clarify the disagreements, and establish a kinetic model and the most influencing factors in
the photocatalytic oxidation of cyclohexane.
Being a first step towards the industrial application of photocatalysis in conventional chemical
conversion processes, Chapter 6 described the novel concept of photocatalytic design and the
realization of the so-called Internally Illuminated Monolith Reactor (IIMR). Comparison with
conventional slurry reactors were performed based on the concept of photonic efficiency. The results
were discussed on the basis of differences in photon flows entering the reactors, and the related
magnitude of product concentrations. Chapter 7 contains the conclusions to be drawn from this
research, as well as the outlook and recommendations for further studies.
Chapter 1
16
References
1. Serpone, N., & Emeline, A.V., Int. J. Photoenergy, 2002, 4, 91
2. Fujishima, A., & Honda, K., Nature, 1972, 238, 37
3. Mills, A., Davies, R.H., Worsley, D., Chem. Soc. Rev., 1993, 22, 417
4. Blake, D.M., Technical Report NREL/TP-430-6084, 1994
5. Linsebigler, A.L., Lu, G., Yates, J.T. Jr., Chem. Rev., 1995, 95, 735
6. Hoffmann, A.J., Martin, S.T., Choi, W., Bahnemann, D.W., Chem. Rev., 1995, 95, 69
7. Blake, D.M., Technical Report NREL/TP-473-20300, 1995
8. Blake, D.M., Technical Report NREL/TP-430-22197, 1996
9. Mills, A., Le Hunte, S., J. Photochem. Photobiol. A, 1997, 108, 1
10. Peral, J., Domenech, X., Ollis, D.F., J. Chem. Technol.. Biotechnol., 1997, 70, 117
11. Pozzo, R.L., Baltanas, M.A., Cassano, A.E., Catal. Today, 1997, 39, 219
12. Blake, D.M., Technical Report NREL/TP-570-26797, 1999
13. Blake, D.M., Sep. Purif. Methods, 1999, 28, 1
14. Herrmann, J.M., Catal. Today, 1999, 53, 115
15. Alfano, O.M., Bahnemann,, D., Cassano, A.E., Dillert, R., Goslich, R., Catal. Today, 2000,
58, 199
16. Blake, D.M., Technical Report NREL/TP-510-31319, 2001
17. Anpo, M., Takeuchi, M., J. Catal., 2003, 216, 505
18. Carp, O., Huisman, C.L., Reller, A., Prog. Solid State Chem., 2004, 32, 33
19. Konstantinou, I.K., Alabanis, T.A., Appl. Catal. B, 2004, 49, 1
20. Gogate, P.R., Pandit, A.B., AIChE J., 2004, 50, 1051
21. Litter, M.I., Appl. Catal. B, 1999, 23, 89
22. Schiavello, M., Photoelectrochemistry, Photocatalysis, and Photoreactors : Fundamentals
and Develpments; NATO ASI Series; D. Reidel Publishing Company, Dordrecht, The
Netherlands, 1985
23. Ollis, D.S., Al-Ekabi, H., Photocatalytic Purification and Treatment of Water and Air;
Elsevier, Amsterdam, The Netherlands, 1993
24. Khan, S.U.M., Akikusa, J., J. Phys. Chem. B, 1999, 103, 7184
25. Licht, S., Wang, B., Mukerji, S., Soga, T., Umeno, M., Tributsch, H., J. Phys. Chem. B, 2000,
104, 8920
26. Asahi, R., Morikawa, T., Ohwaki, T., Aoki, K., Taga, Y., Science, 2001, 293, 269
27. Irie, H., Wanatabe, Y., Hashimoto, K., J. Phys. Chem. B, 2003, 107, 5483
28. Shah, S.I., Li, W., huang, C.P., Jung, O., Ni, C., Proc. Natl. Acad.. Sci. USA, 2002, 99, 6482
29. Xu, A., Zhu, J., Gao, Y., Liu, H., Chem. Res. Chin. Univ. 2001, 17, 281
30. Wang, C., Bahnemann, D.W., Dohrmann, J.K., Chem. Commun. 2000, 16, 1539
31. Wang, Y., Hao, Y., Cheng, H., Ma, H., Xu, B., Li, W., Cai, S., J. Mater. Sci. 1999, 34, 2773
32. Coloma, F., Marquez, F., Rochester, C.H., Anderson, J.A., Phys. Chem. Chem. Phys. 2000, 2,
5320
33. Altynnikov, A.A., Zenkovets, G.A., Anufrienko, V.F., React. Kinet. Catal. Lett. 1999, 67,
273
34. Umebayashi, T., Yamaki, T., Itoh, H., Asai, K., J. Phys. Chem. Solids 2002, 63, 1909
Introduction
17
35. Wang, Y., Cheng, H., Hao, Y., Ma, J., Li, W., Cai, S., Thin Solid Films 1999, 349, 120
36. Yamashita, H., Honda, M., Harada, M., Ichihashi, Y., Anpo, M., Hirao, T., Itoh, N., Iwamoto,
N., J. Phys. Chem. B, 1998, 102, 10707
37. Yu, J.C., Yu, J.G., Ho, W.K., Jiang, Z.T., Zhang, L.Z., Chem. Mater. 2002, 14, 3808
38. Khan, S.U.M., Al-Shahry, M., Ingler, W.B. Jr., Science 2002, 297, 2243
39. Kamat, P.V., Pure Appl. Chem. 2002, 74, 1693
40. Anandan, S., Yoon, M., J. Photochem. Photobiol. C, 2003, 4, 5
41. Corma, A., Garcia, H., Chem. Commun. 2004, 20, 1443
42. Anpo, M., Bull. Chem. Soc. Jpn. 2004, 77, 1427
43. O’Regan, B., Gratzel, M., Nature 1991, 353, 737
44. Nasr, C., Vinodgopal, K., Hotchandani, S., Chattopadhyaya, A., Kamat, P.V., J. Phys. Chem.
1996, 100, 8436
45. Ziolkowski, L., Vinodgopal, K., Kamat, P.V., Langnuir, 1997, 13, 3124
46. Hara, K., Sugihara, H., Tachiana, Y., Islam, A., Yanagida, M., Sayama, K., Arakawa, H.,
Fujihashi, G., Horiguchi, T., Kinoshita, Y., Langmuir 2001, 17, 5992
47. Yamashita, H., Kawasaki, S., Ichihashi, Y., Harada, M., Takeuchi, M., Anpo, M., J. Phys.
Chem. B, 1998, 102, 5870
48. Anpo, M., Zhang, S.G., Higashimoto, S., Matsuoka, M., Yamashita, H., Ichihashi, Y.,
Matsumura, Y., Souma, Y., J. Phys. Chem. B, 1999, 103, 9295
49. Zhang, J., Minagawa, M., Matsuoka, M., Yamashita, H., Anpo, M., Chem. Lett. 2000, 66, 241
50. Ikeue, K., Yamashita, H., Anpo, M., J. Phys. Chem. B 2001, 105, 8350
51. Matsuoka, M., Anpo, M., J. Photochem. Photobiol. C 2003, 3, 225
52. Ikeue, K., Nozaki, S., Ogawa, M., Anpo, M., Catal. Today 2002, 18, 85
53. Warman, J.M., De Haas, M.P., Pichat, P., Koster, T.P.M., Van Der Zouwen-Assink, E.A.,
Mackor, A., Cooper, R., Radiat. Phys. Chem. 1991, 37, 433
54. Szczepankiewicz, S.H., Colussi, A.J., Hoffmann, M.R., J. Phys. Chem. B 2000, 104, 9842
55. Turchi, C.S., Ollis, D.F., J. Catal., 1990, 122, 178
56. Fox, M.A., Dulay, M.T., Chem. Rev., 1993, 93, 341
57. Legrini, O., Oliveros, E., Braun, A.M., Chem. Rev., 1993, 93, 671
58. Environmental Protection Agency, Advanced Photochemical Oxidation Processes Handbook,
U.S. Government Printing Office, Washington DC, 1998
59. Nakamura, R., Imanishi, A., Murakoshi, K., Nakato, Y., J. Am. Chem. Soc. 2003, 125, 7443
60. Nakamura, R., Nakato, Y., J. Am.. Chem. Soc. 2004, 126, 1290
61. Nakamura, R., Sata, S., J. Phys. Chem. B 2002, 106, 5893
62. Martin, S.T., Herrmann, H., Choi, W., Hoffmann, M.R., J. Chem. Soc. Faraday Trans. 1994,
90, 3315
63. Martin, S.T., Herrmann, H., Hoffmann, M.R., J. Chem. Soc. Faraday Trans. 1994, 90, 3323
64. Ohko, Y., Hashimoto, K., Fujishima, A., J. Phys. Chem. A 1997, 101, 8057
65. Gerischer, H., Heller, A., J. Phys. Chem. 1991, 95, 5261
66. Gerischer, H., in Photocatalytic Purification and Treatment of Water and Air, Ollis, D.F.,
Al-Ekabi, H. (Eds.), Elsevier Science Publishers, Amsterdam, 1993
67. Gerischer, H., Heller, A.J., J. Electrochem. Soc. 1992, 139, 113
68. Serpone, N., J. Photochem. Photobiol. A 1997, 104, 1
69. Rajeshwar, K., J. Appl. Electrochem 1995, 25, 1067
Chapter 1
18
70. Poulios, I., Tsachpinis, I., J. Chem. Technol. Biotech. 1999, 74, 349
71. Poulios, I., Avranas, A., Rekliti, E., Zouboulis, A., J. Chem. Technol. Biotech. 2000, 75, 205
72. Sakthivel, S., Neppolian, B., Shankar, M.V., Arabindoo, B., Palanichamy, M., Murugesan, V.,
Solar Energy Mater. & Solar Cells 2003, 77, 65
73. Lizama, C., Freer, J., Baeza, J., Mansilla, H.D., Catal. Today 2002, 76, 235
74. Carraway, E.R., Hoffmann, A.J., Hoffmann, M.R., Envrion. Sci. Technol. 1994, 28, 776
75. Hoffmann, A.J., Carraway, E.R., Hoffmann, M.R., Environ. Sci. Technol. 1994, 28, 786
76. Bahnemann, D.W., Kormann, C., Hoffmann, M.R., J. Phys. Chem. 1987, 91, 3789
77. Serpone, N., Solar Energy Mater. Solar Cells 1995, 38, 369
78. Gratzel, M., Nature 2001, 414, 338
79. Nargiello, M., Herz, T., in Photocatalytic Treatment and Purification of Water and Air, Ollis,
D.F., Al-Ekabi, H. (Eds.), Elsevier Science Publishers, Amsterdam, 1993
80. Suzuki, Y., Warsito, A.M., Uchida, S., Chem. Lett. 2000, 29, 130
81. Serpone, N., Salinaro, A., Pure Appl. Chem. 1999, 71, 303
82. Serpone, N., Sauve, G., Koch, R., Tahiri, H., Pichat, P., Piccinini, P., Pelizzetti, E., Hidaka,
H., J. Photochem. Photobiol. A 1996, 94, 191
83. Spoto, G., Morterra, C., Marchese, L., Orio, L., Zecchina, A., Vaccum 1990, 41, 37
84. Bickley, R.I., Gonzales-Carreno, T., Lees, J.S., Palmisano, L., Tilley, R.J.D., J. Solid State
Chem. 1991, 92, 178
85. Cerrato, G., Marchese, L., Morterra, C., Appl. Surf. Sci. 1993, 70/71, 200
86. Colon, G., Hidalgo, M.C., Navio, J.A., Langmuir 2001, 17, 7174
87. Wang, C.Y, Rafani, J., Bahnemann, D.W., Dohrmann, J.K., J. Photochem. Photobiol. A 2002,
148, 169
88. Lindner, M., Theurich, J., Bahnemann, D.W., Water Sci. Technol. 1997, 35, 79
89. Ray, A.K., Beenackers, A.A.C.M., AIChE J. 1997, 43, 2571
90. Chen, D., Ray, A.K., Water Res. 1998, 32, 3223
91. Alfano, O.M., Bahnemann, D., Cassano, A.E., Dillert, R., Goslich, R., Catal. Today 2000, 58,
199
92. Jacoby, W.A. Maness, P.C., Wolfrum, E.J., Blake, D.M., Fennell, J.A., Environ. Sci. Technol.
1998, 32, 2650
93. Stikiewitz, S., Heller, A., New J. Chem. 1996, 20, 233
94. Romeas, V., Pichat, P., Guillard, C., Chopin, T., Lehaut, C., New J. Chem. 1999, 23, 3365
95. Krysa, J., Keppert, M., Jirkovsky, J., Stengl, V., Subrt, J., Mater. Chem. Phys. 2004, 86, 333
96. Gesenhues, U., Chem. Eng. Technol. 2001, 24, 7
97. Heine, H., Inorganic Pigments in Ullmann’s Encyclopedia of Industrial Chemistry, 5th Ed.,
Vol. A20, VCH Publishes, Weinheim, 1992
98. Diebold, U., Surf. Sci. Reports 2003, 48, 53
99. Norotsky, A., Jamieson, J.C., Kleppa, O.J., Science 1967, 158, 338
100. Kumar, P.N., Keizer, K., Burggraff, A.J., Okubo, T., Nagomoto, H., Morooka, S., Nature
1992, 358,48
101. Basca, R.R., Gratzel, M., J. Am. Ceram. Soc. 1996, 79, 2185
102. Zhang, H.Z., Banfield, J.F., J. Mater. Chem. 1998, 8, 2073
103. Zhang, H.Z., Banfield, J.F., J. Phys. Chem. B 2000, 104, 3481
104. Arrayo, R., Codoba, G., Padilla, J., Lara, V.H., Mater. Lett. 2002, 54, 397
Introduction
19
105. Karvinen, S., Solid State Sci. 2003, 5, 811
106. Kim, S., Gislason, J.J., Morton, R.W., Pan, X.Q., Sun, H.P., Laine, R.M., Chem. Mater. 2004,
16, 2336
107. Arbiol, J., Cerda, J., Dezanneau, G, Cirera, A., Peiro, F., Comet, A., Morante, J.R., J. Appl.
Phys. 2002, 92, 853
108. Yu, J.C., Yu,, J., Ho, W., Jiang, Z., Zhang, L., Chem. Mater. 2002, 14, 3808
109. Tanaka, K., Hisanada, T., Rivera, A.R., in Photocatalytic Purification and Treatment of
Water and Air, Ollis, D.F., Al-Ekabi, H. (Eds.), Elsevier Science Publishers, Amsterdam,
1993
110. Watanabe, T., Nakajima, A., Wang, R., Minabe, M., Koizumi, S., Fujishima, A., Hashimoto,
K., Thin Solid Films 1999, 351, 260
111. Sumita, T., Yamaki, T., Yamamoto, S., Miyashita, A., Appl. Surf. Sci. 2002, 200, 21
112. Park, N., Van de Lagemaat, J., Frank, A.J., J. Phys. Chem. B 2000, 104, 8989
113. Soypan, I., Watanabe, M., Murasawa, S., Hashimoto, K., Fujishima, H., Chem. Lett., 1996, 69
114. Ohtani, B., Ogawa, Y., Nishimoto, S., J. Phys. Chem. B 1997, 101, 3746
115. Sclafani, A., Hermann, J.M., J. Phys. Chem. 1996, 100, 13655
116. Watson, S.S., Beydoun, D., Scott, J.A., Amal, R., Chem. Eng. J. 2003, 95, 213
117. Saadoun, L., Ayllon, J.A., Jimenez-Becerril, J., Peral, J., Doménech, X., Rodríguez-Clemente,
R., Appl. Catal. B: Environ. 1999, 21, 269
118. Wang, K.H., Hsieh, Y.H., Wu, C.H., Chang, C.Y., Chemosphere 2000, 40, 389
119. Chen, Y., Wang, K., Lou, L., J. Photochem. Photobiol A: Chem. 2004, 163, 281
120. Bahnemann, D.W., Kholuiskaya, S.N., Dillert, R., Kulak, A.I., Kokorin, A.I., Appl. Catal. B:
Envrion. 2002, 36, 161
121. Tanaka, K., Capule, M.F.V., Hisanaga, T., Chem. Phys. Lett. 1991, 187, 73
122. Zhang, Z.B., Wang, C.C., Zakaria, R., Ying, J.Y., J. Phys. Chem. B 1998, 102, 10871
123. Van Veen, J.A.R., Veltmaat, F.T.G., Jonkers, G., J. Chem. Soc. Chem. Commun. 1985, 1656
124. Chhor, K., Bocquet, J.F., Colbeau-Justin, C., Mater. Chem. Phys. 2004, 86, 123
125. Busca, G., Saussey, H., Saur, O., Lavalley, J.C., Lorenzelli, V., Appl. Catal. 1985, 14, 245
126. Morterra, C., J. Chem. Soc. Faraday Trans. 1988, 84, 1617
127. Travert, A., Manoilova, O.V., Tsygenenko, A.A., Mauge, F., Lavalley, J.C., J. Phys. Chem. B
2002, 106, 1350
128. Ziolkowski, J., Surf. Sci. 1989, 209, 356
129. Olivier, P.M., Watson, G.W., Kelsey, E.T., Parker, S.C., J. Mater. Chem. 1997, 7, 563
130. Arrouvel, C., Digne, M., Breysse, M., Toulhoat, H., Raybaud, P., J. Catal. 2004, 222, 152
131. Kozlov, D.V., Panchenko, A.A., Bavykin, D.V., Savinov, E.N., Smirniotis, P.G., Rus. Chem.
Bull. Int. Ed. 2003, 52, 1100
132. Kozlov, D.V., Bavykin, D., Savinov, E., Catal. Lett. 2003, 86, 169
133. Soria, J., Conesa, J.C., Augugliaro, V., Palmisano, L., Sciavello, M., Scalfani, A., J. Phys.
Chem. 1991, 95, 274
134. Palmisano, L., Schiavello, M., Scalfani, A., Martin, C., Martin, I., Rives, V., Catal. Lett. 1994,
24, 303
135. Ranjit, K.T., Willner, I., Bossmann, S., Braun, A., J. Phys. Chem. B 1998, 102, 9397
136. Cheung, S.T.C., Fung, A.K.M., Lam, M.H.W., Chemosphere 1998, 36, 2461
137. Kamat, P.V., Fox, M.A., Chem. Phys. Lett. 1983, 102, 379
Chapter 1
20
138. Patrick, B., Kamat, P.V., J. Phys. Chem. 1992, 96, 1423
139. Mills, A., Belghazi, A., Davis, R., Worsley, D., Morris, S., J. Photochem. Photobiol. A: Chem.
1994, 79, 131
140. Yu, J.C., Xie, Y., Tang, H.Y., Zhang, L., Chan, H.C., Zhao, J., J. Photochem. Photobiol. A:
Chem. 2003, 156, 235
141. Iliev, V., J. Photochem. Photobiol. A: Chem. 2002, 151, 195
142. Mele, G., Bisccarella, G., Vassapollo, G., Garcia-Lopez, E., Palmisano, L., Schiavello, M.,
Appl. Catal. B: Environ. 2002, 33, 309
143. Serpone N., Maruthamuthu, P., Pichat, P., Pelizzetti, E., Hidaka, H., J. Photochem. Photobiol.
A: Chem. 1995, 85, 247
144. Shang, J., Yao, W., Zhu, Y., Wu, N., Appl. Catal. A: Gen. 2004, 257, 25
145. Shi, L., Li., C., Gu, H., Fang, D., Mater. Chem. Phys. 2000, 62, 62
146. Zhang, Q., Wang, J., Yin, S., Sato, T., Saito, F., J. Am. Ceram. Soc. 2004, 87, 1161
147. Yu, J.C., Yu, J.G., Ho, W.K., Jiang, Z.T., Zhang, L.H., Chem. Mater. 2002, 14, 3808
148. Lindgrena, T., Lua, J., Hoela, A., Granqvista, C.G., Torresb, G.R., Lindquist, S.E., Solar
Energy Mater. Solar Cells 2004, 84, 145
149. Diwald, O., Thompson, T.L., Goralski, E.G., Walck, S.D., Yates Jr., J.T., J. Phys. Chem. B
2004, 108, 52
150. Suda, Y., Kawasaki, H., Ueda, T. Ohshima, T., Thin Solid Films 2004, 453, 162
151. Yin, S., Yamaki, H., Zhang, Q., Komatsu, M., Wang, J., Tang, Q., Saito, F., Sato, T., Solid
State Ionics 2004, 172, 205
152. Zheng, S.K., Wang, T.M., Hao, W.C., Shen, R., Vacuum 2002, 65, 155
153. Gole, J.L., Stout, J.D., Burda, C., Lou, Y., Chen, X., J. Phys. Chem. B 2004, 108, 1230
154. Burda, C., Lou, Y., Chen, X., Samia, A.C.S., Stout, J., Gole, J.L., Nano Lett. 2003, 3, 1049
155. Miyauchi, M., Ikezawa, A., Tobimatsu, H., Irie, H., Hashimoto, K., Phys. Chem. Chem. Phys.
2004, 6, 865
156. Diwald, O., Thompson, T.L., Zubkov, T., Goralski, E. G., Walck, S.D., Yates Jr., J.T., J. Phys.
Chem. B 2004, 108, 6004
157. Tokudome, H., Miyauchi, M., Chem. Lett. 2004, 33, 1108
158. Matthews, R., Water. Res. 1990, 24, 653
159. Peral, J., Ollis, D.F., J. Catal. 1990, 122, 178
160. Mills, A., Moris, S., J. Photochem. Photobiol. A: Chem. 1992, 71, 75
161. Yu, Y., Leu, R.M., Lee, K.C., Water Res. 1996, 30, 1169
162. Chen, D., Ray, A.K., Appl. Catal. B: Environ. 1999, 23, 143
163. Crittenden, J.C., Liu, J., Hand, D.W., Perran, D.L., Water Res. 1997, 31, 429
164. So, C.M., Cheng, M.Y., Yu, J.C., Wong, P.K., Chemosphere 2002, 46, 905
165. Grzechulska, J., Morawski, A.W., Appl. Catal. B: Environ 2002, 36, 45
166. Turchi, C.S., Ollis, D.F., J. Catal. 1989, 119, 483
167. Ollis, D.F., Pelizzetti, E., Sermone, N., Environ. Sci. Technol. 1991, 25, 1523
168. Okamoto, K., Yamamoto, Y., Tanaka, H., Itaya, A., Bull. Chem. Soc. Jpn. 1985, 58, 2023
169. Kormann, C., Bahnemann, D.W., Hoffmann, M.R., Environ. Sci. Technol. 1991, 25, 494
170. Ohko, Y., Ikeda, K., Rao, T.N., Hashimoto, K., Fujishima, A., Zeitschrift fur Physikalische
Chemie 1999, 213, 33
171. Mills, A., Wang, J., Zeitschrift fur Physikalische Chemie 1999, 213, 49
Introduction
21
172. Terzian, R., Serpone, N., J. Photochem. Photobiol. A 1995, 89, 163
173. Al-Sayyed, G., D’Olivera, J.C., Pichat, P., J. Photochem. Photobiol. A 1991, 58, 99
174. Watts, R.J., Kong, S., Orr, M.P., Miller, G.C., Henry, B.E., Water Res. 1995, 29, 95
175. Lu, M.C., Roam, G.D., Chen, J.N., Huang, C.P., Chem. Eng. Comm. 1995, 139, 1
176. Gupta, H., Tanaka, S., Water. Sci. Technol. 1995, 31, 47
177. Augugliaro, V., Loddo, V., Marci, G., Palmisano, L., Lopez-Munoz, M.J. J. Catal. 1997, 166,
272
178. Bangun, J., Adesina, A.A., Appl.. Catal. A 1998, 175, 221
179. Nozik, A.J., Memming, R., J. Phys. Chem. 1996, 100, 1306
180. Matsumoto, Y., J. Solid State Chem. 1996, 126, 227
Chapter 1
22
23
2
A Combinatorial Approach Towards Photocatalytic Oxidative
Decolorization of Methylene Blue over Titania Materials
Abstract
A novel photocatalyst testing system was constructed. The activity of photocatalysts can be
investigated batchwise in 10 parallel photoreactors under comparable reaction conditions. Light
intensity measurements showed that, within an error range of 10%, all 10 reactors receive a uniform
light flux at the height of reaction liquid. The reactor assembly was successfully applied to the dye
degradation processes in water, namely the photocatalytic oxidative decolorization of methylene blue.
The decolorization profile followed a pseudo first order reaction mechanism. The overall reaction rate
is not limited by the external mass transfer of methylene blue, nor does the transport of oxygen from
gas phase to the photocatalyst surface restrict the reaction.
Chapter 2
24
2.1 Introduction
Combinatorial organic synthesis through high throughput experimentation is one of the most
important new methodologies developed in recent years [1,2]. It represents a technique by which large
numbers of structurally distinct molecules may be synthesized and analyzed much more rapidly and at
lower cost than traditional synthetic chemistry. In the modern form, it is broadly applied in the
pharmaceutical discoveries as well as biological industry. Its application to other fields, such as
catalysis and photochemistry, has not been explored to certain extent.
Photocatalysis, i.e. using semiconductor particles under sufficient irradiation energy for the
simultaneous reduction and oxidation of different redox systems, has been intensively studied during
the last 30 years [3-7]. The major focuses are on the investigation of commercial applications of
photocatalytic systems for the efficient treatment of water and air streams polluted with toxic
substances. In some cases, pilot-scale or even commercially available reactors have already been
constructed, especially with titanium dioxide as the photocatalyst [8-9].
Methylene blue is an organic dye frequently chosen as a model material in photocatalytic activity
testing. The intrinsic kinetics was already discussed in previous studies [10-13], which are mainly
based on a benchmark catalyst of Degussa P25. This reaction has been suggested by the Photocatalyst
Standardization Committee of Japan as the national standard for testing of water purification
performance of photocatalytic materials. It is well known that the photocatalytic activity strongly
depends on the nature of the catalyst, the reactants and the their interaction. Various parameters can
also affect the photocatalytic efficiency because of the high complexity of the system. The surface
morphology and physiochemical properties of photocatalyst, namely crystal size, surface –OH groups,
agglomerate structure and impurities differ per catalyst from different supplier, or even per batch of
catalyst from the same vendor. Also the mass transfer characteristics of the testing reaction system are
poorly investigated, and in many cases, not taken into account [14].
In recent years, the potential power of combinatorial approaches and high throughput (HT)
screening for materials and catalysts has already been amply demonstrated [15]. Its application into
photocatalysis has been rarely explored, however, partly due to the irregular irrradiation field resulted
from the light source geometry. Lettmann et al. made a “laboratory design” of 45 transparent glass
flasks of 2 ml each, arranged in an rack of nine by five, with eight conventional fluorescence lamps
above the rack providing homogeneous light intensity [16]. The reliability of their design was
confirmed by the excellent agreement between the conventional and the high-throughput results found
for Ti-based mixed oxides.
Because of the complexity of photocatalytic systems involving reactant, photocatalyst, activation
medium, and the interaction between all these players, it is of great importance to develop a system
that provide the possibility and reliability to perform catalyst screening and quesi-kinetic studies
within limited timeframe and with reduced cost. A parallel testing assembly for high-throughput
activity screening of photocatalysts was developed for this purpose. The photon energy field inside the
reaction zone was measured using calibrated spectrophotometer coupled with irradiance collector.
Parallel photocatalytic decolorization of methylene blue was applied to investigate the system
applicability and reproducibility. Due to the porous nature of photocatalyst agglomerates in water,
special attention was given to the mass transfer behaviour of reactants to the catalyst surface.
Combinatorial reaction approach
25
2.2 Experimental
High-Throughput Photocatalytic Reactors (HTPR) set-up
Figure 1. Reactor assembly for parallel photocatalytic testing (left) and an example of sample
sets taken during reaction (right).
Photocatalytic acitivity measurements were carried out in a combinatorial way of experimentation.
A home-build multi-reactor assembly was able to handle up to 10 photo-reactions simultaneously (Fig.
1). 10 flasks of 250 ml with pyrex glass covers were used as reactor vessels, in which the suspensions
were agitated by high-performance multi-position magnetic stirrers (IKA, RT10) with an equal stirring
rate of 600 rpm. The UV-irradiation was provided with 8 blacklight tubes (18 W, Philips) located 20
cm above the liquid level. The reaction assembly walls and internals were covered with light reflective
aluminum paper to minimize the light absorption. It was designed so that a uniformed light flux can be
achieved, even with reduced irradiance by switching off certain lamps.
During the reaction, the reactor housing was continuously aerated with a fan and the humidity in
the reactor assembly was monitored. Temperature was controlled at 305±2 K by water flow through
the cooling coil at the back of the reactor housing using a thermostat. The TiO2 powder, typically
sieved to a fraction of 53-75 µm if not specified, was mixed for 2 hrs with 100 ml of methylene blue
solution (0.03 mmol·l-1) in dark to assure a saturated adsorption. Dark samples were taken to
investigate the adsorption characteristics. Afterwards UV lights were switched on intermittent samples
were taken for analysis. Figure 1 show a typical sample set taken during adsorption and reaction. The
samples were filtered through 0.45 mm PTFE Millipore membrane filters to remove suspended titania
agglomerates. A UV-VIS spectrometer (Avantes Avaspec-1024-UV/VIS) registered the absorbance
spectra of the clean solution over the 400-1000 nm range with a spectra resolution of 0.33 nm.
Calibrations were taken at 10 wavelengths adjacent to the maximum absorbance of methylene blue,
which is determined at 667 nm. A Beer-Lambert diagram was established to correlate the absorbance
to MB concentration. After photocatalytic reactions, solutions were collected and subjected to
agglomerate size analysis, which was taken with a laser Mastersizer S equipped with a 300 RF lens.
Spectral Irradiance measurements were performed using a spectrophotometer (Avantes,
S-2000-UV) with a fiber optic cosine collector. For absolute irradiation, the spectrophotometer was
firstly configured with the fiber optics and radiometrically calibrated in the Avantes calibration
laboratory with a range from 200 to 400 nm. With the help of a check board and a optical post mount,
the actual irradiation inside the HTPR can be measured (Fig.2). The check board was placed on the
bottom of the reactor assembly, of which the optical post mount can be fixed on each pins. By moving
8 UV-lamps (blacklight)
Sampling
Port
Cooling coil
Reaction flasks
10 head stirrer
Fan
Hygrometer
Flask Nr.
Sampling
Time
1
2
3
4
5
6
7
8
9
10
t0
Chapter 2
26
the optical post mount through the check board, as well as the vertical location of the cosine collector
on the post mount, the light intensity inside the whole reactor assembly can be accessed. Normally 100
measuring points were taken to construct a 2D light intensity map at certain height.
Figure 2. Light intensity measurement line-up.
Chemicals
Eight commercial titania photocatalysts were used without further purification. The suppliers and
denotations are as follows: Hombikat UV100 from Sachtleben (Hombikat), P25 from Degussa (P25),
titania nanopowder 99.7% from Aldrich cat. 637254 (Aldrich_A), titania nanopowder 99.9% from
Aldrich cat. 634662 (Aldrich_B), titania from Merck cat. 1.00808 (Merck), titania from Fluka cat.
71615 (Fluka), titania from Riedel de Haen cat. 14027 (RDH), and titania from Sigma cat. T8141
(Sigma). All samples were found to be over 99% pure except for the Hombikat, which showed 8 % weight
loss in TGA analysis mainly due to the decomposition of sulphates. Ultrapure (distillated and deionised)
water was used to prepare the methylene blue solutions, of which the organic dye was purchased from
Merck (art. 1.15943, 97%).
The absorption spectra of solid samples were measured using a Varian Cary 1 UV-Vis spectrometer
equipped with diffuse reflection accessories. BaSO4 was used as the reference material. Samples were
scanned with a light beam ranging from 190 nm to 500 nm with a scanning rate of 10 nm·s-1
.
Photocatalytic reaction kinetics
The photoactivity of each TiO2 powder was determined with an apparent first order kinetics of MB
decolorization. It followed from the Langmuir-Hinselwood mechanism (Eq. 2.1) and the generalized mass
balance over the reactor volume (Eq. 2.2). In aqueous systems, water is frequently a competitor for the
adsorption of organics. Due to the strong inhibition effect of water and the low concentration of methylene
blue, simplifications can be made resulting in a first order kinetic model in which the apparent kinetic
constant kapp
is the only variable to be determined from the decay curve of methylene blue (Eq. 2.3).
MBin
WWMBMB
MBMB
MBin ckcKcK
ckKkr ≈
++=θ=
1 (2.1)
catin
MB Wrdt
dcV ⋅−= (2.2)
tk
MBMB
appecc⋅−
⋅= 0, (2.3)
z
y
x
Spectrophotometer
Cosine collector
Optical post
mount
10
2 4 6 8 10 12 14 16 18
2
4
6
8
10
Check board
Combinatorial reaction approach
27
2.3 Results and Discussion
Irradiation field inside the reactor assembly
Figure 3. Measured photon flux to the reaction liquid in case of different amount of lamps
switched “on”. Measurement performed at the height of liquid reactant.
The UVA (320 – 400 nm) photon flux to individual reaction flasks was determined using the light
intensity measurement setup (Fig. 2). The results of the measurement, performed at the height of the
liquid reactant, are shown in Figure 3 in the form of surface plots. The top surface plot represents the
irradiation intensity with all 8 lamps switched “on”. The reactor assembly is designed with the
possibility to vary the photon flux, by switching certain lamps “off”, and this feature is evaluated using
the light intensity measurement. The middle surface plot and the bottom one show the measured
photon flux in case that only 4 lamps (number bcfg) or 2 lamps (number cf) are turned “on”.
Statistic analysis on the measurements show that with a normal distribution, all the light intensity
measurements show sharp peaks on the mean value with very low standard deviation. The average
UVA photon flux and the corresponding standard deviation are, 456 µW/cm2, 27 µW/cm2; 231
µW/cm2, 20 µW/cm2; and 117 µW/cm2, 11 µW/cm2, for 8, 4, and 2 lamps “on” respectively. It is clear
that a homogeneous photon flux to each reaction flasks can be achieved with the designed reaction
configuration, equipped with the light switching ability.
During photocatalytic experiments, all flasks were stirred with magnetic stirrers of 600 rpm. A
whirlpool vortex in the middle of the liquid will appear under the constant mechanical stirring.
Although all flasks will show similar contour of the liquid surface, which will not be on the same
horizontal plane, it is worthwhile to check the effect of liquid height on the photon flux received.
abcdef
gh
1 2 3 4 5
6 7 8 9 10
abcdefgh “on”
bcfg “on”
cf “on”
Lamp
Flask
x
y
Chapter 2
28
Furthermore, liquid samples were taken during the course of photocatalytic reaction, resulting in the
decrease of normal liquid levels. For both purposes we performed light intensity measurements at
different height.
Figure 4 shows the photon flux along the x-axis, measured at different height. The y axis is 15 cm,
the middle line of the flasks 6-10. The light intensity is slight higher at the middle of the x axis than on
both side, which can be the natural emitted energy distribution of the tubular blacklight lamps. With
regard to the height where the measurement took place, the light intensity is not very sensitive on the
varied height of the liquid surface by stirring. Only when the measured point is elevated by 16 cm, the
intensity will be increased by ~30%. It is because of the shortened distance between measurement and
the irradiation light source. The contour of photon flux will be more parabolic than parallel. During
normal operation, the maximum height difference between the tip of the vortex and normal liquid level
will never exceed 4 cm, of which the incidental light effect is negligible.
Combined with Fig.3, it can be concluded that the irradiation from the top of the reactor assembly
to the liquid reactant can be considered as parallel and uniform, despite of the cylinder form of each
individual light source.
Figure 4. Measured photon flux at different height (see legend). The light intensity is measured
with all 8 lamps “on”
Photocatalytic degradation of methylene blue – test runs
A prerequisite for the photocatalytic reaction to take place is that the photocatalyst is able to
absorb incident light with simutanous adsorption of the reactive species. Figure 5 show the absorbance
of TiO2 photocatalyst (Hombikat),measured using a Varian Cary 1 UV-Vis spectrometer equipped with
diffuse reflection accessories, along with the MB absorbance, and the irradiance of the blacklight lamp.
It can be seen from Figure 5 that the absorbance of TiO2 (Degussa P25) and methylene blue covers two
different regions. P25, the photocatalyst, mainly absorb UV-light, as the absorption edge ends at ~410 nm.
Methylene blue, the reactant, on the contrary only absorbs visible light with the wavelength between 450
nm and 750 nm, over the measured spectra range of 300 nm – 800 nm. The applied irradiation source,
blacklight lamp, shows a sharp irradiance peak at 370 nm. It is interesting to note that there are hardly any
0
200
400
600
800
5 15 25 35 45
Distance to left end [cm]
Lig
ht
inte
nsit
y [
µµ µµW
/cm
2]
Liquid height
Liquid height + 4 cm
Liquid height + 8 cm
Liquid height + 12 cm
Liquid height + 16 cm
x
Combinatorial reaction approach
29
overlaps between the absorption spectra of methylene blue and the emission spectra of applied light source.
Figure 5. Measured TiO2 (P25) and methylene blue absorbance, overlay on the emission spectra
of the blacklight lamps.
Figure 6 shows a typical curve of methylene blue concentration, calculated from the methylene
blue visible light absorption, as a function of the reaction time. Initial drop of methylene blue
concentration corresponds to the dark adsorption of methylene blue, in this specific case of TiO2 P25
as the photocatalyst, 0.003 mmol/gcatalyst. The adsorption capacity stabilized after agitating the reaction
solution with suspended photocatalyst for ~60 min in dark. Once the light was switched “on”, a fast
drop in the methylene blue concentration was observed, from which the apparent first order kinetic
constant kapp can be derived.
Figure 6. Time course of photocatalytic decolorization of methylene blue in the absence of
photocatalyst and with the presence of 0.05 g TiO2 (P25, pre-sieved to 75-53 µm). Timer counting
starts at the moment that all 8 lamps were switched “on”.
300 400 500 600 700 800
Wavelength [nm]
Ab
so
rban
ce /
Irr
ad
ian
ce [
A.U
.] TiO2
MBBlacklight
0
0.01
0.02
0.03
-100 0 100 200
Irradiation time [min]
CM
B [
mm
ol/
l]
DarkLight
Without catalyst
With catalyst
Chapter 2
30
A comparative study was performed in the absence of photocatalyst (Fig. 6). The methylene blue
concentration did not drop in dark experiment as no absorbent was present. The decolorization reaction
proceeds with a negligible rate when the light was switched “on”. It was mainly the simultaneous
reaction of methylene blue with the incident photons. As can be seen from Figure 5, methylene blue
can hardly be activated to the excited states by the blacklight irradiation. Due to the lack of overlap
between MB absorbance and blacklight irradiance, the light reaction in oxygen rich atmosphere
without catalyst, frequently called “photolysis”, can be considered negligible.
Photocatalytic reactor screening
In order to evaluate the performance of these 10 parallel reactor flasks in the photocatalytic
reaction assembly, several screening tests under reactive conditions were conducted. Figure 7 shows a
typical result of 10 identical photocatalytic decolorization tests performed in one run. The points
represent the samples taken from individual flask at certain reaction time.
Figure 7. Time course of ten identical photocatalytic experiments performed in one run. Line is
for guide the eyes.
From the curve fitting of the decolorization profile for each reactor flask, the apparent 1st order
kinetic constants were calculated. Figure 8 shows the photocatalytic activities as were derived from
Fig.7, given as the 1st order kinetic rate constant of all the 10 photocatalytic tests. Error bars represent
the standard deviation from the 1st order curve fitting. Besides the experiment with Hombikat, another
two equivalency runs were carried out with Aldrich A samples (0.050 g catalyst) and Fluka samples
(0.050 g catalyst) respectively. For all these three independent equivalency runs, the relative error
between the highest reaction rate and the lowest one never exceeded the mean photocatalytic activity ±
8 %. Even for the fastest reaction catalyzed by Aldrich A, of which the reaction configuration could
play an important role in the apparent reaction rate due to the possible transport limitations, all 10
reactor flasks behave similar. Seen the slightly different photon flux each reactor flask received (Fig.
3), it is confident to state that the photocatalytic reaction rates measured in separate reactor flasks can
be compared within a error range of ± 8 %, with the error mainly caused by the inconsistent indicent
light intensity.
0
0.01
0.02
0.03
0.04
-200 0 200 400
Time [min]
Co
nc
. MB [
mm
ol/l]
flask 1
flask 2
flask 3
flask 4
flask 5
flask 6
flask 7
flask 8
flask 9
flask 10
Dark Light
Hombikat 75-53 µm
Wcat: 0.020 g
Vliq: 0.100 l
Illumi.: 8 lamps
Combinatorial reaction approach
31
Figure 8. Apparent reaction rates obtained from different reaction flasks, in case identical
reactions were performed in all flasks.
Experiments were also carried out to investigate the possible variations in photocatalytic activity
measured at different runs. As can be seen from Figure 9, 6 identical photocatalytic decolorization of
methylene blue runs using Hombikat and 5 identical runs using Merck were conducted. Both graphs
show, again, comparable results between the runs. Together with the findings described in previous
paragraph, the conclusion can be drawn that it is acceptable and validated to cross-compare
photocatalytic reaction rates in the same run as well as between runs.
Figure 9. Apparent reaction rates obtained from different runs using same reaction conditions:
catalyst amount: 0.050 g (75-53 µm); liquid volume: 0.10 l; initial MB concentration 0.030 mmol/l;
illumination source: 8 blacklight lamps. Run numbers were given in the x-axis.
Photocatalytic reaction parameter studies
In slurry photocatalytic processes, the catalyst amount is an important parameter that has been
extensively investigated [17]. The effect of the photocatalyst amount was studied using Hombikat
0
0.01
0.02
0.03
0.04
1 2 3 4 5 6 7 8 9 10
Flask number
ka
pp
[m
in-1
]Hombikat, 0.020 g
Aldrich A, 0.050 g
Fluka, 0.050 g
0
0.01
0.02
0.03
1 2 3 4 5 6
kap
p [
min
-1]
Hombikat
0
0.03
0.06
0.09
1 2 3 4 5
kap
p [
min
-1]
Merck
Chapter 2
32
catalyst. Results are given in Figure.10 as points with their corresponding standard error range. It is
found that the apparent 1st order reaction rate constant initially increases proportionally with the
catalyst amount. However, as the catalyst concentration increases to above 0.5 g/l, the measured
overall photocatalytic decolorization rate began to level off towards a constant value of 0.025 min-1
.
This finding is in consistence with what previously described by other researchers, that above a certain
catalyst concentration, the reaction rate becomes independent of the catalyst loading [4,11,18,19]. It is
contributed to the “light shielding effect”. The linear part indicated a true heterogeneous catalytic
regime in which the effective optical penetration length exceeded the solution geometry in the light
penetration direction. Below the optimal concentration all titania particles can sufficiently absorb the
incoming photons. Increasing catalyst loading above the critical value results in a shielding effect of
excess particles, which reduces the total amount of photosensitive surface. The optimal concentration
we found is higher than that reported by Lakshmi et al. [11], probably due to the different geometry
and operation conditions. Another point worth mentioning is that photolysis is negligible, as can be
seen from the point at 0 mg/l catalyst.
Figure 10. Effect of catalyst amount on the photocatalytic decolorization of methylene blue.
Catalyst: Hombikat 75-53 µm; liquid volume: 0.10 l; initial MB concentration 0.030 mmol/l;
illumination source: 8 blacklight lamps. Lines are for guide the eyes. Error bars correspond to 95%
confidence interval.
The influence of the UV irradiation intensity on the reaction rate was investigated at two different
catalyst loadings with various photocatalysts. Figure 11 revealed a linear relationship between the light
intensity and the rate constant for all tested commercial samples, especially with 0.5 g/l of
photocatalysts. Half order dependency was generally expected at high light intensity or elevated
temperature, which was not discovered under our experimental conditions with highest photon flux of
456 µW/cm2 achieved with 8 UVA lamps. The excellent linearity implies that the photon-induced
charge separation at the catalyst surface was dominant over the recombination process of generated
electrons and holes.
0
0.01
0.02
0.03
0 0.4 0.8 1.2 1.6
Conc.cat [g/l]
kap
p [
min
-1]
Light Shielding Effect
Combinatorial reaction approach
33
Figure 11. Dependence of apparent photocatalytic reaction rate on the UVA photon flux at two
different catalyst loadings of 0.5 g/l (left) and 0.2 g/l (right)
The primary particles of many TiO2 photocatalysts are small (nm), which intend to form
agglomerates in aqueous phase. Due to the porous structure of agglomerates it can have great influence
on the apparent photocatalytic reaction rate. Some factors that can be thought but not exclusively are:
light penetration inside the agglomerates via absorption/scattering, charge carriers (electrons and holes)
migration and hopping through primary particles, and adsorption/desorption and mass transfer of
reactants/products from agglomerate internals to bulk liquid. The effect of agglomerates in aqueous
phase photocatalysis is, however, poorly understood in general. To study the agglomerate effect,
experiments will pre-sieved TiO2 photocatalysts were carried out. It can be envisaged that TiO2
photocatalyst with large pre-sieved fraction will also form bigger agglomerates in water.
Figure 10. Effect of pre-sieving on the photocatalytic decolorization of methylene blue. Catalyst:
Hombikat; liquid volume: 0.10 l; initial MB concentration 0.030 mmol/l; illumination source: 8
blacklight lamps. Error bars correspond to 95% confidence interval.
0
0.02
0.04
0.06
0.08
0 100 200 300 400 500
UVA Irradiance [µµµµW/cm2]
ka
pp
[m
in-1
]
P25
Hombikat
Merck
Fluka
Aldrich A
0
0.02
0.04
0.06
0.08
0 100 200 300 400 500
UVA Irradiance [µµµµW/cm2]
ka
pp
[m
in-1
]
P25
Hombikat
Merck
Fluka
Aldrich A
0
0.01
0.02
38-45 45-53 53-63 63-75 75-90
Pre-sieved fraction [µµµµm]
ka
pp
[m
in-1
]
Chapter 2
34
Figure 10 indicates that the photocatalytic
decolorization rates of methylene blue increases
slightly with increasing pre-sieved size of Hombikat.
This phenomena could be explained by the light
absorption and scattering characteristics within the
agglomerate particles. As can be seen in the
schematic representation of light penetration through
agglomerates (right), once the incident light reaches
the surface of an primary catalyst particle, certain
amount of photons will be absorbed, whilst the rest
are scattered back into the surrounding environment.
The wavelength-dependent light absorption and
scattering coefficients have been determined by
Cabrera et al. for various Titanium dioxide
particulate suspensions in water. As is illustrated in the graph, a relatively large fraction of light can be
scattered out of the small agglomerates than large ones, because of the shorter light path and less
possibilities of attenuation by adjacent primary particles. This “scattering-out” effect will result in a
net high photon absorption for large agglomerates while applying same photocatalyst, hence the
enhanced apparent photocatalytic activity.
Table 1. Determination of external mass transfer limitations
Nm D cbulk rV,obs kovaov Ca
[rpm] [cm2/s] [mol/cm
3
liq] [mol/(cm3
cats)] [1/s] [-]
Oxygen 161 2.70×10-5
2.29×10-7
1.40×10-7
8.95×10-2
1.42×10-3
Methylene blue 161 1.66×10-6
3.00×10-8
1.40×10-7
6.81×10-2
1.43×10-2
The remaining uncertainties of applying HTPR setup for catalyst screening using photocatalytic
decolorization of methylene blue is the mass transfer of reactants, namely methylene blue and oxygen
to the photocatalyst surface. Mass transfer characteristics in agitated slurry reactor is discussed in
Appendix 2.1. The results with regard to the external mass transfer characteristics in HTPR setup are
listed in Table 1.
The applied agitation speed of 600 rpm is sufficiently large to keep the entire solid mass
suspended for maximum utilization of the catalyst, as compared to the minimum agitation speed for
complete particle suspension Nm. Diffusion coefficients for oxygen and methylene blue D are obtained
from literature [21]. It can be seen in this worst-case scenario, i.e. fastest reaction rate measured per
volume of the photocatalyst, the Carberry numbers for both oxygen and methylene blue are less than
0.05, indicating the absence of external mass transfer limitations.
2.4 Conclusions
A novel reaction assembly for high throughput photocatalytic experimentation (HTPR) was
constructed., which allows parallel catalyst screening for up to 10 different photocatalysts. Light
irradiance measurements inside the reaction assembly indicate uniform light distribution to all reaction
flasks, even with reduced light flux. The equivalency between the reaction flasks was verified by three
Large agglomerate
small agglomerate
Incident light
light scattere
d out
Combinatorial reaction approach
35
series of independent photocatalytic methylene blue decolorization experiments applying each time
different photocatalyst. Results between decolorization experiments performed in different runs were
also comparable, as is proved by the experiments with same photocatalyst under similar reaction
conditions. The optimum testing conditions for photocatalytic catalyst screening using methylene blue
decolorization were determined experimentally, to be 0.5 g/l photocatalyst of 53-75 µm with all
irradiation sources “on”. The practical application of HTPR for catalyst selection purposes will be
further discussed in chapter 3.
References
1. Jung, G., Combinatorial Chemistry: Synthesis, Analysis, Screening, Wiley-VCH, 2000
2. Bannwarth, W., Hinzen, B., Combinatorial Chemistry: From Theory to Application (Methods
and Principles in Medicinal Chemistry), 2nd ed., Wiley-VCH, 2006
3. Hoffmann, M.R., Martin, S.T., Choi, W., Bahnemann, D.W., Chem. Rev., 1995, 95, 69
4. Hufschmidt, D., Lium L., Selzer, V., Bahnemannn, D., Water Sci. Technol.., 2004, 49, 135
5. Konstantinou, I.K., Albanis, T.A., Appl. Catal. B: Environ., 2004, 49, 1
6. Peral, J., Domenech, X., Ollis, D.F., J. Chem. Technol. Biotechnol. 1997, 70, 1
7. Bhatkhande, D.S., Pangarkar, V.G., Beenackers, A.A.C.M., J. Chem. Technol. Biotechnol.,
2001, 77, 102
8. Dillert, R., Cassano, A.E., Goslish, R., Bahnemann, D., Catal. Today, 1999, 54, 267
9. Alfano, O.M., Bahnemann, D., Cassano, A.E., Dillert, R., Goslish, R., Catal. Today, 2000, 58,
199
10. Mattews, R.W., Water Res., 1991, 29, 1169
11. Lakshami, S., Renganathan, R., Fujita, S., J. Photochem. Photobiolo. A, 1995, 88, 163
12. Xu, N., Shi, Z., Fan, Y., Dong, J., Shi, J., Hu, M.Z.C., Ind. Eng. Chem. Res., 1999, 38, 373
13. Houas, A., Lachhab, H., Ksibi, M., Elaloui, E., Guillard, C., Herrmann, J.M., Appl. Catal. B,
2001, 31, 145
14. Ollis, D.F., Pelizzetti, E., Sermone, N., Environ. Sci. Technol. 1991, 25, 1523
15. Jandeleit, B., Schaefer, D.J., Powers, T.S., Turner, H.W., Weinberg, W.H., Angew. Chem. Int.
Ed. 1999, 38, 2494
16. Lettmann, C., Hinrichs, H., Maier, W.F., Angew. Chem.Int. Ed. 2001, 40, 3160
17. Chen, D., Ray, A.K., Appl. Catal. B, 1999, 23, 143
18. Herrmann, J.M., Catal. Today, 1999, 53, 115
19. Turchi, C.S., Ollis, D.F., J. Catal. 1989, 119, 483
20. Cabrera, M.I., Alfano, O.M., Cassano, A.E., J. Phys. Chem. 1996, 100, 20043
21. Murov, S.L., Carmichael, I., Hug, G.L., Handbook of photochemistry, Dekker, New York, 1993
Chapter 2
36
Appendix 2.1 Determination of mass transfer parameters in slurry reactor
In this appendix, mass transfer characteristics in mechanically agitated slurry is discussed. For
convenience, the first part lists different correlations for predicting various parameters in reactor
design. Most correlations are taken from references of Ramachandran & Chaudhari (1983) and
Beenackers & van Swaaij (1993). If not mentioned, properties are in CGS units.
A2.1.1 Particle suspension
Minimum agitation speed for complete particle suspension is given by:
Zwietering (1958)
85.055.0
13.045.045.01.02.033.1 ')()/(2
IL
LpLpIT
md
wgdddN
ρ
ρ−ρη= (A2.1)
and Baldi et al. (1978)
89.058.0
125.014.042.042.017.0
2 ')(
IL
pLpL
md
wdgN
ρ
ρ−ρηβ= (A2.2)
The parameter β2 can be estimated using the correlation of Nienow (1968,1975):
33.1
2 )(2I
T
d
d=β (A2.3)
A2.1.2 Power consumption for agitation
Power number NP
turbine)blade-(flat 10000Refor 3.62
53
0 >η
ρ==
ρ=
L
LII
Ll
P
Nd
dN
PN (A2.4)
Different correlations are used for calculating the power consumption in presence of gas bubbles.
Michel & Miller (1962):
45.0
56.0
32
0 )(812.0G
l
Q
NdPP = SI units (A2.5)
Calderbank (1958):
2
33
2
33
0
105.3for ,85.162.0
105.3for ,26.10.1
−
−
×>−=ψ
×<−=ψ
ψ=
I
G
I
G
I
G
I
G
Nd
Q
Nd
Q
Nd
Q
Nd
Q
PP
(A2.6)
Combinatorial reaction approach
37
Luong & Volesky (1979)
18.032
38.0
3
0
)()(497.0−−
σ
ρ=
L
Ll
I
G dN
Nd
Q
P
P (A2.7)
A2.1.3 Average gas holdup
Loiseau et al. (1977)
27.0056.036.036.0 )(011.0L
G
L
LLGGV
P
V
Pu +ησ=ε −−
SI units (A2.8)
Yung et al. (1979)
4.165.032
5.0
3
3)()()(108.6
T
I
L
IL
I
GG
d
ddN
Nd
Q
σ
ρ×=ε −
(A2.9)
Van Dierndonck et al. (1968,1970) (bubble column)
125.075.0)(2.1 Mou
L
GLG
σ
η=ε (A2.10)
A2.1.4 Bubble diameter
Calderbank (1958)
09.0)/(
15.42.04.0
5.06.0
+ρ
εσ=
LL
GLB
VPd (A2.11)
Mersmann (1977) (bubble column)
2/1])(
[8.1g
dGL
LB
ρ−ρ
σ= (A2.12)
Van Dierendonk (1968,1970) (bubble column)
4/12/12 ))((25.6 −−
σ
η
ρ
σ= Mo
u
gd
L
GL
L
LB (A2.13)
A2.1.5 Critical stirring speed
Westerterp et al.(1963) found a critical speed Nc, above which kLaGL is not a strong function of gas
velocity and is dependent mainly on the speed of agitation.
I
T
LL
Ic
d
d
g
dN25.122.1
)/( 25.0+=
ρσ (A2.14)
Chapter 2
38
A2.1.6 Gas-Liquid mass transfer
Calderbank (1958,1961)
5.03/1
2)(]
)([42.0
L
L
L
LGLL
Dgk
η
ρ
ρ
ηρ−ρ= (A2.15)
and
20000)()( if )()(1095.13.2
log
20000)()( if )/()/(
44.1
3.07.02
3.07.02
5
0
3.07.02
6.0
5.02.044.0
>η
ρ
η
ρ×=
<η
ρ
σ
ρ=
−
G
B
L
LI
G
B
L
LI
GL
GL
G
B
L
LI
L
BGLLGL
u
NdNd
u
NdNd
a
a
u
NdNduuVPa
(A2.16 & A2.17)
aGL0 is the value of aGL calculated from Eq. (A2.16).
Yagi and Yoshida (1975)
32.06.019.05.05.1Re06.0 AIIB NGFNFrScSh−= (A2.18)
Oguz et al. (1987)
LOH
AIIB NGFNFrScShσσ⋅−=
/09.06.019.05.05.1 2Re162.0 (A2.19)
Bern et al. (1976)
521.032.0979.116.1210099.1 −−×= LGIGLL VudNak (A2.20)
Litmans et al. (1972)
67.0")( G
m
L
GLLV
Pak εα= (A2.21)
where α = 0.618 and m” = 0.605 for P/VL < 8 W/liter, and α = 1.215 and m” = 0.315 for P/VL >10
W/liter.
Kawase & Moo-Young (1990)
4.12.1 )(3.0L
LGL
guSck
ρ
η= −
(A2.22)
Dietrich et al. (1992)
4.1/ when 105.1 ,1/ when103
Re
44
5.05.045.1
=×==×=
=
−−TT
IB
dHBdHB
WeScBSh(A2.23)
Yagi and Yoshida’s correlation is developed from the adsorption tests in the aqueous system.
Although Bern’s correlation is based on data for three-phase systems, it does not provide the
dependency of kLaGL on the solution viscosity and surface tension. Calculation of kLaGL from either
Calderbank or Litmans correlation requires the information of gas holdup and power input, which is
generally less accessible. The correlation of Oguz is the modified form of Yagi-Yoshida’s correlation
which comprehends the data of aqueous slurry systems and organic liquids as well. Kawase and
Combinatorial reaction approach
39
Moo-Young developed their theoretical relation for kL using the pseudo-homogeneous-liquid approach,
for particles with densities close to the liquid density. Dietrich’s correlation is specially used for
6-blade self-gas-inducing agitator of Rushton type. For practical purpose, the correlation of Oguz and
Dietrich are recommended.
A2.1.7 Specific interfacial area (bubble column)
Van dierendonck et al. (1968,1970)
4/12/1))((2 Mogu
aL
L
L
GLGL
σ
ρ
σ
η= (A2.24)
Yoshida and Akita (1973)
1.11131.062.05.06.0 GTRGL dShGaBoSca ε= −− (A2.25)
A2.1.8 Liquid-Solid mass transfer
Relations for the mass transfer coefficient around the solid particles are usually presented in the
form of:
21Re2 nn
pp ScCSh += (A2.26)
Many attempts have been made using the Kolmogoroff’s theory of local isotropic turbulence as a
basis for the correlation of liquid-solid mass transfer in agitated vessels. This leads to a Reynolds
number based on the velocity of the critical eddies responsible for most of the energy dissipation.
3/1
3
34
)(ReL
Lp
p
ed
η
ρ= (A2.27)
The specific local energy dissipation rate per unit mass of liquid is defined as:
LL
GV
Pgue
ρ+= (A2.28)
The approach based on the energy dissipation rate as outlined above is not limited to a particular
type of slurry reactor. The parameters in equation A2.26 proposed in literature are listed in table
A2.1.1.
Table A2.1.1 Constants in Sherwood relationship for L-S mass transfer
Reference C n1 n2
Sano et al. (1974) 0.400 0.75 0.33
Levins & Glastonbury (1972) 0.47·(dI/dT)0.17 0.62 0.36
Sänger & Deckwer (1981) 0.545 0.80 0.33
Lazaridis (1990) 0.368 0.69 0.33
Marrone & Kirwan (1986) 0.36 0.75 0.33
Chapter 2
40
Other correlations in different form are listed here.
Boon-Long et al. (1978)
461.0019.0011.0
3
173.0
2
32
283.0
2
)()()()()2
(046.0Dd
d
d
wVgdNdd
D
dk
L
L
p
T
pL
L
L
pL
L
TLppS
ρ
η
ρη
ρ
η
ρπ= −
(A2.29)
Asai et al. (1989)
8.5/18.53/158.08.5])Re61.0(2[ ScSh pp += (A2.30)
Kobayashi & Saito (1965)
112.03/1
3
)())(
(212.02L
LGp
L
Lpp
p
ud
D
gdSh
η
ρ
η
ρ−ρ+= (A2.31)
Calderbank & Jones (1961)
500for ))(
(34.03/1
2
3/2 >ρ
ρ−ρη= −
Peg
SckL
LpL
S (A2.32)
A2.1.9 Overall external Gas-Solid mass transfer
Based on resistance in series model (Westerterp, 1984), The overall external mass transfer can be
expressed as:
1)11
( −+=PSGLL
ovovakak
ak (A2.33)
A2.1.10 Determination of rate limiting step
Carberry criterion: [Moulijn et al. (1999)]
The external mass transfer limitation can be neglected if :
05.0)//( ,
,,
,
,<
−=
ρ=
bx
sxbx
pbxovov
obsV
xc
cc
wcak
rCa (A2.34)
Weisz-Prater criterion: [Moulijn et al. (1999)]
The internal diffusion limitation can be neglected if:
15.02
1
,,
2
,<
+⋅=Φn
cD
Lr
sxeffx
obsV
x (nth order reaction) (A2.35)
The characteristic length L is defined as:
sp
p
aS
VL
1== (A2.36)
In these correlations, x represents reactant, which can be either methylene blue or oxygen.
Combinatorial reaction approach
41
References in Appendix 2.1
Ramachandran, P.A., Chaudhari, R.V., Three-phase catalytic reactors, topics in chemical engineering
vol. 2, Gordon & Breach, Philadelphia (1983)
Beenackers, A.A.C.M., van Swaaij, W.P.M., Chem. Eng. Sci., 48, 3109 (1993)
Westerterp, K.R., van Swaaij, W.P.M., Beenackers, A.A.C.M., Chemical reactor design and operation,
Wiley & sons, UK (1984)
De Blok, W.J., Mass transfer in three-phase slurry reactors, Ph.D. Thesis, University of Amsterdam,
Amsterdam, The Netherlands (1984)
Oguz, H., Brehm, A., Deckwer, W.D., Deckwer, in Recent trends in chemical reactor engineering,
Vol.2, Wiley Eastern, New Delhi, 484 (1987)
Van Dierendonck, L.L., Vergrotingsregels voor gasbelwassers, Ph.D. Thesis, University of Twente,
Enschede, The Netherlands (1970)
Van Dierendonck, L.L., Fortuin, J.M.H., Venderbos, D., in Chemical Reaction Engineering, 4th
European symposium and 81st meeting, Brussels, 205 (1968)
Akita, K., Yoshida, F., Ind. Eng. Chem., Process Des. Develop., 12, 76 (1973)
Mersmann, A., Chem.-Ing.-Techn., 49, 679 (1977)
Yagi, H., Yoshida, F., Ind. Eng. Chem. Process Des. Develop., 14, 488 (1975)
Zwietering, T.N., Chem. Eng. Sci., 8, 244 (1958)
Calderbank, P.H., Trans. Instn. Chem. Engrs., 36, 443 (1958)
Calderbank, P.H., Moo-Young, M.B., Chem. Eng. Sci., 16, 39 (1961)
Baldi, G., Conti, R., Alaria, E., Chem. Eng. Sci., 33, 21 (1978)
Nienow, A.W., Chem. Eng. Sci., 23, 1453 (1968)
Nienow, A.W., Chem. Eng. J., 9, 153 (1975)
Yung, C.N., Wong C.W., Chang, C.L., Can. J. Chem. Eng., 57, 672 (1979)
Loiseau, B., Midoux, N., Charpentier, J.C., AIChE J., 23, 931 (1977)
Michel, B.J., Miller, S.A., AIChE J., 8, 262 (1962)
Luong, H.T., Volesky, B., AIChE J., 25, 893 (1979)
Westerterp, K.R., van Dierendonck, L.L., de Kraa, J.A., Chem. Eng. Sci., 18, 157 (1963)
Bern, L., Lidefelt, J.O., Schoon N.H., J. Am. Oil Chem. Soc., 53, 463 (1976)
Joosten, G.E.H., Schilder, J.G.M., Janssen, J.J. Chem. Eng. Sci., 32, 563 (1977)
Sano, Y., Yamaguchi, N., Adachi, T., J. Chem, Eng. Japan, 7, 255 (1974)
Kobayashi, T., Saito, H., Kagaku Kogaku, 3, 210 (1965)
Levins, D.M., Glastonbury, J.R., Chem. Eng. Sci., 27, 537 (1972)
Boo-Long, S., Laguerie, C., Couderc, J.P., Chem. Eng. Sci., 33, 813 (1978)
Sänger, P., Deckwer, W.D., Chem. Eng. J., 22, 179 (1981)
Kawase, Y., Moo-Young, M., Chem. Eng. Commun., 96, 177 (1990)
Lazaridis, S., Stoffübergang in einem blasensäulen-reaktor mit suspediertem feststoff an der
phasegrenze fest-flüssig in Newton’schen und nicht-Newton’schen flüssigkeiten, Ph.D. Thesis,
Technical University Aachen, Germany (1990)
Marrone, G.M., Kirwan, D.J., AIChE J., 32, 523 (1986)
Asai, S., Konishi, Y., Kajiwara, T., J. Chem. Eng. Japan, 22, 96 (1989)
Calderbank, P.H., Jones, S.J.R., Trans. Instn. Chem. Eng. (London), 39, 363 (1961)
Moulijn, J.A., Xu, X., Kapetijn, F., van Langefield, A.D., Lecture notes on “Catalysis and Catalysts”,
Chapter 2
42
Technical University of Delft, The Netherlands (1999)
List of symbols
aGL gas-liquid mass transfer surface area, cm2/cm3
aP external area of particles per unit volume of reactor, cm2/cm3
as specific surace area of catalyst particle, cm-1
cx,b concentration of reactant x in bulk liquid phase, mol/cm3
cx,s concentration of reactant x at catalyst surface, mol/cm3
dB average diameter of the gas bubbles in the slurry reactor, cm
dI diameter of the impeller, cm
dp average diameter of the catalyst particles, cm
dT diameter of the reactor, cm
e energy supplied (by agitator or gas bubbling) to the liquid per unit mass, cm2/s
3
g acceleration due to gravity, cm/s2
kL gas-liquid mass transfer coefficient, liquid side, cm/s
kovaov overall volumetric mass transfer coefficient based on reactor volume, 1/s
kS liquid-catalyst mass transfer coefficient, cm/s
L characteristic length of particle, cm
N speed of agitation employed, s-1
Nc critical agitation speed in eq, (A2.14), s-1
Nm minimum speed of agitation for suspension of particles, s-1
P power consumption for agitation for an aerated liquid, erg/s
P0 power consumption for agitation of a gas-free liquid, erg/s
PG power supplied to the liquid by gas phase
QG volumetric flow rate of the gas, cm3/s
rV,obs reaction rate based on unit volume of catalyst, mol/cm3/s
Sp particle surface area, cm2
uG superficial velocity of the gas phase in the reactor, cm/s
Vp particle volume, cm3
VL volume of the liquid in the reactor, cm3
w catalyst mass per unit volume of the reactor, g/cm3
w’ percentage of catalyst loading, g/100 g solution
D diffusion coefficient of reactant x in liquid, cm2/s
Dx,eff effective diffusion coefficient of reactant x in porous catalyst, cm2/s
εG gas holdup
ηL viscosity of the liquid, g/cm/s
ρp density of the catalyst particle, g/cm3
ρL density of the liquid, g/cm3
σL surface tension of the liquid, dyne/cm
σH2O surface tension of water, dyne/cm
Combinatorial reaction approach
43
Dimensionless groups
ShR reactor Sherwood number, kLdT/D
ShB bubble Sherwood number, kLaGLdI2/D
Shp particle Sherwood number, kSdp/D
ReI impeller Reynolds number, NdI2ρL/ηL
Rep partilcle Reynolds number, (edp4ρL
3/ηL3)1/3
Sc Schmidt number, ηL/(ρLD)
FrI impeller Froude number, dIN2/g
GFN gas flow number, σL/(ηLuG)
NA aeration number, NdI/uG
Mo Morton number, σL3ρL/(ηL
4g)
Bo Bond number, gdT2ρL/σ
Ga Galilei number, gdT3ρL
2/ηL
2
We Weber number, ρLN2dI
3/σL
Pe Péclet number, utdP/D, and ut = gdp2(ρp-ρL)/18ηL
Chapter 2
44
45
3
Effect of TiO2 Source and Thermal Pre-Treatment on Photoactivity
for Methylene Blue Degradation in Water
Abstract
The photocatalytic activity of various commercial titania catalysts was studied in the
photodegradation of methylene blue (MB) in water, using the high throughput photocatalytic reaction
setup. P25 from Degussa exhibits the highest apparent activity per gram catalyst among eight
commercial titania catalysts. Hombikat and one of the Aldrich samples were the worst catalysts in
terms of activity normalised to total surface area, for which the diffusion of the methylene blue into
meso-porous agglomerates appears to be the rate limiting step. Titania from Merck expressed highest
activity per surface area, presumably due to a synergetic effect of traces alumina present in this sample.
Different reaction intermediates were formed in the case of Meck titania catalysed reaction, indicating
modified selectivity by altered adsorption modes of methylene blue.
The effect of thermal pre-treatment was also investigated. Unlike frequently presumed, a strong
change in photoactivity does not coincide with the anatase to rutile phase transformation, with the
exception for P25 photocatalyst. The maximum in activity of thermal samples is due to the interplay of
various factors, including but not exclusively, photon absorption, surface area, pore diameter, phase
composition, crystal size and surface hydroxyl group density.
Chpater 3
46
3.1 Introduction
Nowadays conventional water and wastewater treatment processes are confronted with major
challenges of increasing number and varieties of identified contaminants, growth of world population
and industrial activities, and the diminishing availability of clean water resources. Photocatalytic water
purification is an emerging technology, which leads to the total demineralization of organic pollutants
in an energy-saving and cost-efficient way [1-3]. Among various photocatalysts, titania (TiO2) is a
promising material due to its strong oxidizing power, high photochemical stability and low cost. It has
been investigated for different environmental applications, i.e., photocatalytic degradation of organic
pollutants and heavy metal ions, and photodisinfection of water [4-7].
Organic dyes represent one of the major industrial water contaminants, as 15% of the total world
production of dyes is lost during the dye processing and is released in textile effluents [8,9].
Photocatalytic decolorization of textile dyes and other industrial dyestuff has received much attention
from the last decades, because it appears as the most cost-effective destructive technology at ambient
conditions [10-12]. There are many studies dealing with the photocatalytic decolorization of specific
dyes from different chemical categories. Among others methylene blue, a typical cationic thionine dye,
has been the focus of various studies [13-17]. It is also widely applied as the test reaction for
photocatalyst activity as well as for photo-reactor development studies [18-24]. The photocatalytic
bleaching of methylene blue follows an apparent first-order kinetics, as is described in chapter 2. The
apparent reaction rate constant kapp can be used as the single measure of photocatalyst activity.
The overall process of heterogeneous photocatalysis is controlled by several steps, i.e. reactant
adsorption, catalyst activation by photons, surface reaction and desorption of product. Among others,
the mass transfer characteristics of reactants and products in photocatalysis are rarely investigated, and
in many cases, not taken into account. It is well known that titania exhibits a strong tendency to
aggregate due to its high hydrophilicity and the natural influence of van der Waals interactions. The
surface morphology of photocatalyst, namely crystal size and agglomerate structure differs with
different titania sources.
Another method to modify the physicochemical properties of titania photocatalysts is based on
thermal pre-treatment of commercially available photocatalyst samples. Recently big efforts were
made preparing photocatalyst of nano-sized particles, and subsequently modify the surface
morphology by (hydro-)thermal treatment [25-27]. Varied conclusions were drawn reflecting the
complicated impact of thermal pre-treatment on the photocatalytic behaviors. Possible parameters
include photon absorption, surface area, pore diameter, phase composition, crystal size and surface
hydroxyl group density.
This work is devoted to the investigation of the catalytic behaviour of various commercial titanias
in photo-decolorization of methylene blue in water. Also attempts were made to understand the
complex behavior of thermal treatment on the surface morphology of the photocatalyst. The following
analysis techniques have been applied, X-ray diffraction, Raman spectroscopy, UV-VIS absorption
spectra, thermogravimetry, X-ray fluorescence, surface hydroxyl group density determination and pore
distribution analysis. In view of the porous nature of agglomerates in water, particular attention was
given to the mass transfer of reactants to the catalyst surface.
Effect of TiO2 source and thermal pre-treatment
47
3.2 Experimental
Eight commercial titania photocatalysts were used without further purification. The suppliers and
denotations are as follows: Hombikat UV100 from Sachtleben (Hombikat), P25 from Degussa (P25),
titania nanopowder 99.7% from Aldrich cat. 637254 (Aldrich_A), titania nanopowder 99.9% from
Aldrich cat. 634662 (Aldrich_B), titania from Merck cat. 1.00808 (Merck), titania from Fluka cat.
71615 (Fluka), titania from Riedel de Haen cat. 14027 (RDH), and titania from Sigma cat. T8141
(Sigma). All samples were found to be over 99% pure except for the Hombikat, which showed 8 % weight
loss in TGA analysis mainly due to the decomposition of sulphates. Calcinations were performed under
static air, at temperatures ranging from 120°C to 1100°C. Ultrapure (distillated and deionised) water
was used to prepare the methylene blue solutions, of which the organic dye was purchased from Merck
(art. 1.15943, 97%).
Catalysts as received were subjected to thermalgravimetric analysis (TGA) in a Mettler Toledo
TGA/SDTA 851e apparatus. Solid samples, typically 10 mg, were heated to 1000°C after dehydration
at 120°C for 8 hours.
Trace elements in commercial samples were determined by X-ray fluorescence (XRF). A Philips
X-ray fluorescence spectrometer (PW1480) scanned the sample for 76 trace elements. Results were
analyzed with a quantitative analytical software package, UniQuant 4.
Various properties, such as the Brunauer-Emmett-Teller (BET) surface area, the pore dimension
and the pore volume, were obtained by the measurement of nitrogen physisorption capacity at 77K,
applying a Quantachrome Autosorb 6B apparatus. All samples were pre-treated in vacuum at 383 K for
16 hrs.
The X-ray diffraction (XRD) pattern was used to identify the crystal phase and their
corresponding crystallite size. It was recorded on a Philips PW1840 X-ray diffractometer using Cu Kα
radiation at a scan rate of 2θ = 0.01°s-1
. The accelerating voltage and the applied current were 40kV
and 50 mA, respectively.
The absorption spectra of solid samples were measured using a Varian Cary 1 UV-Vis
spectrometer equipped with diffuse reflection accessories. BaSO4 was used as the reference material.
Samples were scanned with a light beam ranging from 190 nm to 500 nm with a scanning rate of 10
nm·s-1.
Raman analysis was performed using a Renishaw Ramascope System 2000 instrument linked to a
Leica microscope. A 514 nm, 20 mW Ar+ laser was used as excitation source. The backscattered light
was filtered for Rayleigh scattering using a holographic notch filter. The spectrograph uses a grating to
disperse the light over the CCD detector, which is coupled to a PC to obtain the Raman spectrum with
a resolution of 4 cm-1. The Raman mapping procedure was fully automated; sample positioning and
laser focusing were handled by a Prior H101 motorized XYZ-stage connected to the Raman software.
The agglomerate size and porous structure of the samples in dry solid was studied using scanning
electron microscopy (SEM) on a JEOL JSM-6400F equipped with a Pioneer EDX. Suspended solid
agglomerate sizes were measured by forward light scattering, using a Mastersizer S, 300 mm RF lens
and a sample dispersion unit.
The amount of surface hydroxyl groups was determined by the method described by Van Veen et
al. [28], using Fe(AcAc)3 as the organic ligand. Typically 0.005 gram of catalyst was added to 10 ml of
0.25 mmol/l Fe(AcAc)3 solution in toluene and stirred in the dark overnight. Afterwards the solid was
removed by centrifugation and the supernatant solution was subjected to UV absorption measurements.
Chpater 3
48
The amount of adsorbed Fe(AcAc)3 was determined by comparing the UV absorption at 355 nm with
calibrated samples.
Photocatalytic bleaching experiments of methylene blue were performed in the high throughput
photocatalytic reaction assembly (HTPR) discussed in chapter 2. The apparent decolorization kinetics
is of first-order, therefore the apparent kinetic rate constant kapp [1/min] is used as the parameter to
compare photocatalyst performance.
3.3 Results
Large variations were found in the photocatalyst activities of commercial TiO2 samples in
methylene blue decolorization. It can be noticed in Fig.1 that an order of magnitude difference in
activity exists for various commercial catalysts. The apparent 1st order reaction rate ranges from 0.058
min-1
to less than 0.010 min-1
for Fluka, Sigma and Riedel de Haën samples.
Figure 1. Comparison of photocatalyst activity in methylene blue decolorization. Error bars
represent the 95% confidence interval of a fit of the 1st order reaction kinetics. catalyst amount: 0.050
g (75-53 µm); liquid volume: 0.10 l; initial MB concentration 0.030 mmol/l; illumination source: 8
blacklight lamps.
In conventional catalytic surface reactions, the reaction rate is directly proportional to the surface
area of the catalyst, i.e. the total amount of surface active sites accessible for reactants. A fair
comparison of photocatalyst activity can be made based on the apparent reaction rate per
photon-activated catalyst surface. Wang et al. proposed an antenna mechanism that energy transfer
between primary particles took place in the three dimensional internal networks of photocatalyst
agglomerates [29]. The high migration ability of electrons and holes in the semiconductor framework
makes it sensible to correlate the photo-activated surface area with the physical surface area
determined by nitrogen adsorption. Figure 2 depicts the apparent activity per surface area of the
commercial TiO2 samples ks,app (= kapp/(SBET⋅wcat)). It can be clearly seen that for photocatalytic
decolorization of methylene blue, no linear trend between the apparent activity and the total amount of
0
0.02
0.04
0.06
0.08
Hom
bikat
Aldrich
_A
P25
Aldrich
_B
Merc
k
Sigm
a
Fluka
Riedel d
e h
aen
kap
p [
1/m
in]
Effect of TiO2 source and thermal pre-treatment
49
surface area exists for commericial photocatalysts.
Figure 2. Apparent photocatalyst activity (Figure 1) normalized to catalyst surface, together with
measured values of BET surface area. Error bars represent 95% confidence interval.
Methylene blue uptake capacity was determined by the decrease of methylene blue concentration
in dark aqueous solution until the equilibrium has been reached (Fig.3). For all photocatalysts, the total
methylene blue uptake was rather insensitive to the temperature at which the equilibrium was
established. Aldrich_A and Fluka exhibit the highest capacity, whereas the lowest amount of
methylene blue adsorbed on the surface was found for Hombikat photocatalyst. There is clearly no
direct relationship between the amount of adsorbed methylene blue on TiO2 and its corresponding
photocatalytic activity in decolorization of methylene blue.
Figure 3. Methylene blue uptake capacity of different photocatalysts as function of temperature
Based on the BET surface area and the surface specific apparent photocatalytic activity (Fig.2),
the eight photocatalysts can be classified into four groups. Both Hombikat and Aldrich_A samples
0
0.03
0.06
0.09
0.12
Hom
bikat
Aldrich
_A
P25
Aldrich
_B
Merc
k
Sigm
a
Fluka
Riedel d
e h
aen
Su
rface s
pecif
ic r
eacti
vit
y k
s,a
pp
[1/m
2/m
in]
0
100
200
300
400
SB
ET [
m2/g
]
ks,app
SBET
0
0.005
0.01
0.015
0.02
0.025
20 30 40 50 60 70
Temperature [ ]
Up
tak
e c
ap
ac
ity
[m
mo
l/g
]
Hombikat
Aldrich_A
P25
Merck
Fluka
[°C]
Chpater 3
50
exhibit very low specific activity whilst they possess the highest BET surface areas among all samples,
with a value over 200 m2/g. P25 and Aldrich_B samples have a surface area of around 20-50 m2/g,
marginal higher than the low surface area catalysts such as Merck, Sigma, Fluka and Riedel de Haen
(~10 m2/g). Correspondingly their photocatalytic activity is higher than that of Sigma, Fluka and
Riedel de Haen samples. The Merck sample has an extraordinary high specific photocatalytic activity,
despite of its low BET surface area.
Figure 4. N2 adsorption/desorption isotherm of various photocatalysts (left) and their
corresponding pore size distribution (right).
The low specific photocatalytic activities of Hombikat and Aldrich_A indicate that the surfaces of
these catalysts were apparently not utilized in the most efficient way. The nitrogen adsorption analysis
provided information on the pore size of the agglomerates, of which both featured mesoporous
characteristics (Fig.4). All samples show type II or III adsorption isotherms, indicating the adsorption
on irregular meso- to macro-porous adsorbents with strong and weak adsorbate-adsorbent interactions,
typical for inter-particle spaces of agglomerates [30]. The mean pore diameter of Hombikat and
Aldrich_A is 2.2 nm and 4.5 nm, respectively. The molecular size of methylene blue is 1.5 nm as
reported previously [31]. With these parameters, the effect of diffusion in agglomerates of reactants
can be evaluated.
Kapteijn et al. provided a systematic approach towards the assessment of mass transfer effects on
measured reaction rates [32]. Criteria for external and internal mass transfer limitations were derived
so that deviations from the ideal situation were not larger than 5%, see appendix 2.1. In case of
external mass transfer, the Carberry number can be derived, which assures that the observed rates do
not deviate more than 5% from the ideal state.
05.0)//(,
,<
ρ=
pbxovov
obsV
xwcak
rCa (3.1)
0
100
200
300
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
P/P0
Va
ds [
cm
3/g
]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
0
0.2
0.4
0.6
0.8
10 100 1000 10000
Pore diameter [
Deso
rpti
on
(d
V/d
log
d)
[cm
3/?
g]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
0
100
200
300
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
P/P0
Va
ds [
cm
3/g
]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
0
0.2
0.4
0.6
0.8
10 100 1000 10000
Pore diameter [
Deso
rpti
on
(d
V/d
log
d)
[cm
3/?
g]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
[°C]
Effect of TiO2 source and thermal pre-treatment
51
The Weisz-Prater criterion was applied to access the importance of internal diffusion limitations.
The effect of internal diffusion can be neglected if the calculated Wheeler-Weisz modulus satisfies this
criterion.
15.02
1
,,
2
,<
+⋅=Φn
cD
Lr
sxeffx
obsV
x (nth
order reaction) (3.2)
The explanation of symbols is given in Appendix 2.1. In these correlations, x represents reactant,
which can be either methylene blue or oxygen.
The agglomerate size can be estimated from SEM analysis, taken for dry samples, or from the
forward laser scattering technique in aqueous suspension (Fig. 5). The agglomerates are clearly of
random form with irregular shapes, sometimes forming big lumps.
Figure 5. TiO2 agglomerates shown as SEM pictures (left) and the size determination in aqueous
suspension by forward laser scattering (right)
The most important and major uncertainty in the latter equation is the effective diffusivity Dx,eff.
Unlike binary diffusion described in various models, the diffusion inside a porous material is strongly
restricted by the size confinement effect and the interaction between molecules and walls. Due to the
small pore size as compared to the bulky molecule of methylene blue, the effective diffusion
coefficient, Dx,eff, for the diffusion of large molecules in relatively small catalyst pores can be estimated
by the following equation:
p
s
bxeffxd
deFwithFDD =λ=λλ
τ
ε= λ−
,)()(6.4
,, (3.3)
where Dx,b is the bulk diffusion coefficient of the solute x (methylene blue), ε is the catalyst pellet
porosity, τ is the catalyst pellet tortuosity and F(λ) is the restrictive factor, which is the factor of λ, the
ratio of the molecular diameter of methylene blue, ds, and the pore diameter, dp [33].
P25Hombikat
Merck Fluka
0
2
4
6
8
10
0.1 1 10 100
Agglomerate diameter [µµµµm]
Fre
qu
en
cy [
%]
Hombikat
Aldrich_A
P25
Merck
Fluka
Chpater 3
52
The evaluation of internal and external mass transfer limitations for all these photocatalysts is
given in table 1. On the basis of Caberry number and Wheeler-Weisz modulus, conclusions can be
drawn that for Hombikat and Aldrich_A samples, the internal diffusion of methylene blue into the
porous agglomerates is most likely to limit the overall reaction rate. In all other cases, the mass
transfer effects on the apparent reaction activities can be assumed negligible.
Table 1. Determination of mass transfer limitations on measured apparent reaction rates
Methylene blue Oxygen Catalyst
Ca Φ Ca Φ Mass transfer limitation?
Hombikat 3.96×10-3
3.87×10-1
3.94×10-4
2.65×10-3 Internal diffusion MB
Aldrich_A 2.89×10-3
2.30×10-1
6.58×10-4
1.57×10-3 Internal diffusion MB
P25 1.43×10-2
1.01×10-1
1.42×10-3
6.91×10-4 No
Aldrich_B 8.22×10-3
4.69×10-2
9.35×10-4
3.21×10-4 No
Merck * 1.33×10-2 - 1.38×10
-3 - No
Fluka * 3.77×10-3 - 2.10×10
-4 - No
* No internal diffusion limitations due to the absence of micro- and meso-pores (<100 nm).
Both P25 and Aldrich_B samples express higher specific reaction rates per surface area than the
low surface area photocatalysts (Sigma, Fluka, and Riedel de Haen). XRD and Raman spectra reveal
that in both samples, anatase and rutile phases co-exist (Fig. 6). The rutile phase gives characteristic
peaks at 2θ of 27º and 35º in XRD pattern, and a characteristic Raman shift of 430 cm-1 in the Raman
spectra. It has been reported that mixed-phase catalysts exhibit enhanced activity due to prolonged
separation of photogenerated electrons and holes through interfacial electron transfer from the
conduction band of the rutile phase to the trapping states of the anatase phase [34,35]. Morever, rutile
acts as an antenna for photon absorption that extends the photocatalyst activity into visible
wavelengths (Fig. 7).
Figure 6. XRD (left) and Raman spectra (right) of various TiO2 photocatalysts
Concluding, the intermediate performance is most likely the results of absence of transfer
limitations, and the beneficial effect of the presence of both rutile and anatase phases.
0 20 40 602θθθθ-degree
Inte
nsit
y [
A.U
.]
P25
Hombikat
Merck
Riedel de Haen
Fluka
Sigma
A A
A
A
R R
Aldrich_A
Aldrich_B
250 450 650 850
Raman shift [cm-1
]
Re
lati
ve
in
ten
sit
y [
-]
R
Hombikat
P25
Merck
Sigma
Fluka
Riedel de Haen
A A
A
Effect of TiO2 source and thermal pre-treatment
53
Figure 7. Diffusion reflection UV-VIS spectra of various TiO2 photocatalysts
The high reaction rate of the Merck sample is remarkable, as its excellent photoactivity has never
been mentioned before in studies on methylene blue decolorization. The BET surface area of Merck as
determined by N2 physisorption is only 11 m2/g, nor is the surface morphology structurally different
from other low surface area photocatalysts like Sigma, Fluka and Riedel de Haen. We investigated the
bulk properties of Merck and other low surface area TiO2 samples and discovered little differences in
XRD, Raman, and UV-VIS absorption spectra. Apparently the extraordinary high activity of the Merck
sample in photocatalytic decolorization of methylene blue cannot be explained by the optical-physical
or textural properties.
Figure 8. Visual inspection of sampling solutions(left) and their corresponding absorbance
spectra. Sample set 1 is taken at the start of irradiation. Correlated samples of Merck and P25 in
sample sets were taken after the same period of irradiation.
A visual inspection of the sampling solutions revealed that the decolorization of methylene blue
on Merck catalyst proceeds differently as compared to other catalysts. As can be seen in figure 8,
unlike the P25 catalyst on which the blue color of methylene blue simply vanishes on irradiation, the
0
0.5
1
1.5
300 350 400 450
Wavelength [nm]
Ab
so
rba
nc
e [
-]
Hombikat
P25
Merck
Sigma
Fluka
Riedel de Haen
P25
400 600 800
Wavelength [nm]
Ab
so
rban
ce [
-]
400 600 800
Wavelength [nm]
Ab
so
rban
ce [
-]
Merck
P25
Sample set number
1 2 3 4 5 6
Blue shift
Chpater 3
54
decolorization process on the Merck catalyst goes through an intermediate stage with the appearance
of a violet colored solution. A blue shift is clearly shown in the visible light absorption spectra of the
sampling solutions 3, 4 and 5 in photocatalytic decolorization of methylene blue on the Merck catalyst.
The blue shift has a maximum absorption peak of around 600 nm with a shoulder extended into the
UV absorption region.
Horikiri et al. [22] investigated the decomposition of methylene blue on anatase type TiO2 loaded
onto Al2O3. The derivatives were analysed and, as a result, azure A (AA) and azure B (AB) were
observed. AA and AB absorption spectra were measured in our study, and it can be clearly seen that
the maximum absorption peak shifts from 670 nm for MB to 650 nm for AB and to 535 nm for AA.
Separation and identification of reaction intermediates was a difficult task and remained partly
unsolved, due to the complexity of the methyelene blue degradation kinetics. Preliminary HPLC and
HPLC-MS analysis do qualitatively indicate the presence of AA and AB, both contributing to the blue
shift with Merck samples.
A major distinction between the Merck catalyst and other titania arises from the trace element
analysis by the XRF technique. In contrast to other titania samples, which are all alumina-free, Merck
titania contains 0.2 wt% of alumina. Based on the study of Hirikiri using TiO2 loaded Al2O3 [22], the
blue shift of the Merck titania catalyzed photo-decolorization of methylene blue could be related to the
existence of alumina. Although not fully proved, the improved photocatalytic activity and altered
selectivity of the Merck catalyst in photocatalytic decolorization of methylene blue can at least
partially be attributed to the existence of aluminum ions that replace titanium in the metal oxide
framework.
Table 2. Determination of trace elements in TiO2 photocatalysts by XRF technique
The effect of alumina on the photocatalytic activity and selectivity in methylene blue
decolorization might be explained by assuming different adsorption characteristics of methylene blue
on aluminum-modified titania. The methylene blue molecule is a flat molecule with different
functional groups, which enables both horizontal and vertical orientation on the catalyst surface.
0.316
0.276
0.527
0.105
0.379
0.347
0.284
I
-
-
-
-
-
0.147
0.295
SO3
0.021
0.017
0.014
0.052
-
0.084
0.051
Nb2O5MoO3V2O5ZrO2P2O5SiO2Al2O3Na2O
-0.402-0.339--0.045RDH
-0.399-0.2760.052--Fluka
0.0210.3960.0260.3330.064-0.090Sigma
-0.4100.0350.3640.2690.2040.046Merck
-0.387-0.0110.035--P25
0.0140.3660.0210.2840.096-0.047Aldrich_A
-0.3660.0290.3130.035-0.056Hombikat
Trace compounds [wt%]Catalyst
0.316
0.276
0.527
0.105
0.379
0.347
0.284
I
-
-
-
-
-
0.147
0.295
SO3
0.021
0.017
0.014
0.052
-
0.084
0.051
Nb2O5MoO3V2O5ZrO2P2O5SiO2Al2O3Na2O
-0.402-0.339--0.045RDH
-0.399-0.2760.052--Fluka
0.0210.3960.0260.3330.064-0.090Sigma
-0.4100.0350.3640.2690.2040.046Merck
-0.387-0.0110.035--P25
0.0140.3660.0210.2840.096-0.047Aldrich_A
-0.3660.0290.3130.035-0.056Hombikat
Trace compounds [wt%]Catalyst
Effect of TiO2 source and thermal pre-treatment
55
In case of aluminium-free titania, the dye molecule attaches itself onto the catalyst surface in a
flat manner, as is proposed by Houas et al. [14]. Surface rearrangement of adsorbed methylene blue
could occur on the photo-activated surface. The C-S=O functional group in the reaction intermediate
or the C-S=C functional group in the original methylene blue is then attacked by the photon-generated
OH· radicals, resulting in the direct opening of the central aromatic ring. This is the first step in the
photocatalytic degradation of methylene blue and decolorization occurs directly due to the destruction
of the resonance structure. In the case of the Merck sample, methylene blue adsorbs preferably on
aluminium, in a mode that the nitrogen atom in the side dimethylamino group donates the lone pair
electrons to the semiconductor to fill the depopulated valence band. The attachment of the dye is
perpendicular to the titania surface at one point only, hence it can be expected that the more spatial
orientation of dye molecules favors a higher adsorption capacity, which is directly reflected in an
enhanced degradation rate. Furthermore OH· radicals adjacent to the adsorbed species can attack the
C-N bond in the dimethylamino group. As the result of demethylation, azure A and azure B are formed
while the resonance structure of the original molecule remains intact.
Figure 9 Schematic implication of adsorption and degradation pathways of methylene blue on the
photo-activated titania surface. Top: P25 TiO2; bottom: Merck TiO2
The effect of calcination temperature on the photocatalytic activity of various TiO2 catalysts was
investigated. Figure 10 depicts the apparent reaction rates of methylene blue photo-decolorization on
the calcined titania samples up to 1100°C.
Ti4+Al3+ O
·· OHννννhννννh
Ti4+Al4+ O
-·OH
Perpendicular adsorption of MB
- CH 3OH, ·O
H+H 2
O -C
H2 ·
Ring structure preserved
(Blue shift)
Merck
Azure B
Azure A
Methylene
Blue
Ring opening intermediates
(Colorless)
P25
Planar adsorption of MB
ννννhννννh
·OH
Chpater 3
56
Figure 10. Effect of calcination temperature on the apparent reaction rate of photocatalytic
decoloriation of methylene blue. (a) Hombikat: (b) Aldrich_A; (c) P25; (d) Merck; (e) Fluka. Error
bars represent the 95% confidence interval of the 1st order reaction kinetics. Catalyst amount: 0.050 g
(75-53 µm); liquid volume: 0.10 l; initial MB concentration 0.030 mmol/l; illumination source: 8
blacklight lamps.
With the exception of P25, all titania photocatalysts exhibit an optimal calcination temperature, at
0
0.02
0.04
0.06
0.08
0 400 800 1200
Calcination temperature [ ]
ka
pp
[1/m
in]
(a)
0
0.02
0.04
0.06
0.08
0 400 800 1200
Calcination temperature [ ]
ka
pp
[1/m
in]
(b)
0
0.02
0.04
0.06
0.08
0 400 800 1200
Calcination temperature [ ]
ka
pp
[1/m
in]
(c)
0
0.02
0.04
0.06
0.08
0 400 800 1200
Calcination temperature [ ]
ka
pp
[1/m
in]
(d)
0
0.02
0.04
0.06
0.08
0 400 800 1200
Calcination temperature [ ]
ka
pp
[1/m
in]
(e)
[°C] [°C]
[°C] [°C]
[°C]
Effect of TiO2 source and thermal pre-treatment
57
which the photocatalytic activity of the thermally pre-treated sample reaches a maximum. This optimal
temperature differs per commercial catalyst, i.e., for Hombikat it is 700°C, for Aldrich_A ~750°C, and
for Merck and Fluka 800°C. Below this temperature, the photocatalytic activity vs. calcination
temperature curve either shows a U-from with a minimum around 350°C (Aldrich_A, Merck), or
depicts a continuous increase of activity with the temperature of thermal pre-treatment (Hombikat,
Fluka). However, regardless of the catalyst source, its photocatalytic activity was shown to decline
dramatically if the calcination temperature is increased to above the optimal. For P25, although no
optimal calcination conditions can be found due to the concomitant decrease in photocatalytic activity
upon thermal pre-treatment, a steep drop can be observed at 600-700°C as well.
Figure 11. UV-VIS absorption spectra of Hombikat samples after thermal pre-treatment.
Photocatalysis utilizes light to certain absorption edge to be able to generate electron-hole pairs. A
shift in the absorption band towards visible light region, so called ‘red shift’, allows the photocatalyst
to use a greater portion of the solar spectrum to drive the charge-carrier initiation step. Interestingly,
the ‘red shift’ in absorption to longer wavelengths runs contrary to the quantum size effect, which is
beneficial for the selectivity enhancement [36,37]. As semiconductor nanoparticles decrease in size,
their excitation energy generally increases, which results in a ‘blue shift’ of their absorption band to a
shorter wavelength region.
Figure 11 depicts the light absorption spectra of Hombikat samples pre-treated at different
temperature. The absorption edge shifts towards the visible light region with increasing calcination
temperature, which is the result of a continuous increase of the rutile absorption band at ~390 nm.
Apparently, the light absorption characteristics can be improved significantly in case the thermal
pretreatment temperature is sufficiently high. On the contrary, this is not reflected in the corresponding
photocatalytic activities. As can be seen for the Hombikat samples (fig. 10a), the apparent reaction rate
decreases monotonically with increasing calcination temperature above 700°C.
0
0.5
1
1.5
300 350 400 450
wavelength [nm]
ab
so
rban
ce [
-]
120 350
500 600
700 800
900 925
950 975
1000
Increase
calcination temperature
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
Chpater 3
58
Figure 12. BET surface area of calcined samples as determined by N2 physisorption.
(a) Hombikat; (b) Aldrich_A; (c) P25.
Figure 13. Pore size distribution of calcined samples as determined by N2 physisorption.
(a) Hombikat; (b) Aldrich_A; (c) P25.
Surface area is for conventional catalytic processes one of the most important parameters. The
photocatalytic activity, however, is not necessarily dependent on catalytic surface area due to the
complex nature of photon-induced catalytic processes [38]. This is again proven by the photocatalytic
decolorization of methylene blue on Hombikat and Aldrich_A samples. As expected, high temperature
pre-treatment causes the collapse of porous structure and a reduction in surface area (fig. 12a, 12b).
The reaction rates, however, pass through maximum at a calcination temperature around 700-800°C,
see figure 9a, 9b. No general relationship can be derived for TiO2 samples between calcination
temperatures up to 700°C and photocatalytic activity in methylene blue decolorization. Other factors
apparently play important to dominant roles for the optimum calcination temperature at 700-800°C. An
exception is the P25 sample. The photocatalytic activity of P25 does follow neatly the trend of BET
surface area on thermal pre-treatment temperature, showing a sharp drop at above 600°C.
0
100
200
300
400
0 400 800 1200
Calc. temp. [ ]
SB
ET [
m2/g
]
(a)
0
60
120
180
240
0 400 800 1200
Calc. temp. [ ]
SB
ET [
m2/g
]
(b)
0
15
30
45
60
0 400 800 1200
Calc. temp. [ ]
SB
ET [
m2/g
]
(c)
0
0.002
0.004
0.006
0.008
0.01
10 100 1000
Pore diameter [
Deso
rpti
on
(d
V/d
d)
[cm
3/?
g]
120
500
600
650
700
750
800
1100
increase
calcination temperature
(a)
0
0.002
0.004
0.006
0.008
0.01
10 100 1000
Pore diameter [
Deso
rpti
on
(d
V/d
d)
[cc/?
g]
120
500
600
700
800
900
1000
increase
calcination temperature
(b)
0
0.2
0.4
0.6
0.8
1
10 100 1000 10000
Pore diameter [
Deso
rpti
on
(d
V/ d
log
d)
[cm
3/?
g]
120
500
600
700
800
900
1000
increase
calcination temperature
(c)
[°C][°C] [°C]
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
°C
[nm] [nm] [nm]
Effect of TiO2 source and thermal pre-treatment
59
The pore diameter and textile structure of titania agglomerates in water will not have impact on its
photocatalytic activity, unless internal diffusion or light penetration depth could influence the apparent
overall reaction rate. For untreated Hombikat and Aldrich_A samples, measured reaction rates were
indeed limited by the interal diffusion of methylene blue, as is indicated in table 1. An increase in pore
diameter will facilitate the effective diffusion of methylene blue into agglomerate pores, hence
enhance the overall reaction rate.
Figure 13 depicts the pore size distribution of Hombikat, Aldrich_A and P25, obtained after
different thermal pre-treatments. As expected, the peaks of the pore size distribution shift towards large
pore diameters with increasing calcination temperature. Therefore the mass transfer constraints will
less likely to occur with calcined samples. It is certainly beneficial for Hombikat and Aldrich_A
samples, apart from other factors with thermal treatment that affect the photocatalytic activity in
negative ways.
As mentioned before, mixed phase titania could exhibit certain synergetic effect on their
photocatalytic activities. Figure 14 and 15 revealed the phase transfer from anatase titania to rutile
phase on thermal treatment, as is examined by X-ray diffraction and Raman spectroscopic analysis for
Hombikat catalyst. Bulk anatase and rutiles phases are characterized by their corresponding diffraction
angles. However, this method is not suitable to identify the crystalline phases for nano-sized crystals.
Raman spectroscopy was used to overcome this limitation.
Rutile phase shows characteristic XRD peaks of 2θ = 27.5, 36.2 and 54.4°, which occur at
samples calcined at above 900°C (fig. 14). The XRD results are consistent with the finding by Raman
spectroscopy (fig. 15). The characteristic Raman shift of 440 and 605 cm-1 for rutile phases were only
found for those samples with a pre-treatment temperature higher than 900°C. Both analysis pointed out
to the onset temperature of the phase transformation from anatase to rutile at 900°C, far higher than the
temperature at which the photocatalytic activity reached the epics (fig. 10a). Therefore the enhanced
photocatalytic activity of Hombikat at 700°C and the strong drop thereafter cannot be attributed to the
phase transformation from anatase and rutile and the mixed-phase synergetic effect.
Figure 14. XRD spectra of Hombikat samples before and after thermal pre-treatment.
0 20 40 60
2θ−θ−θ−θ−Degree
Inte
nsit
y
25
120
500
600
700
800
850
900
925
950
1000
1100
1100 24hrs
Anatase
Rutile
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
Chpater 3
60
Figure 15. Raman spectra of Hombikat samples before and after thermal pre-treatment.
The relative abundance of anatase to rutile in the samples was calculated by using the equation
[39]:
rr
a r
1.26 IF
I 1.26 I
⋅=
+ ⋅ (3.4)
where Fr is the rutile fraction, and Ir and Ia are the strongest intensities of the rutile (110) and
anatase (101) diffraction angles, respectively.
The crystal sizes of anatase and rutile were determined by employing the Scherrer equation:
·
os=
KD
λ
β θ (3.5)
where λ is the wavelength of the Ni-filtered CuKα radiation used (λ = 0.15418 nm), β is the full
width at half-maximum of the diffraction angle considered, K is a shape factor (0.9) and θ is the angle
of diffraction. For these calculations, the indices (101) for anatase and (110) for rutile were used.
Smaller crystal size means a high specific surface area and relatively large number of active sites
being available on the catalyst surface. It is also in favor of higher photoactivity due to smaller
distances for electrons and holes to migrate to the surface. On the other hand, the charge-carrier
density will be relatively high on smaller crystals, combined with increased density of surface defects
as recombination sites, short separation distance of electrons and holes, the electron-hole
recombination can occur more often. When the dimensions of semiconductor particles further decrease
to nano scale, the energy levels shift according to the quantum size effect. The shift of the conduction
band may accelerate the reduction, while that of the valence band may increase the oxidation reaction,
which could counteract the reduced light absorption due to ‘blue shift’.
Table 3 summarizes the phase composition and crystal sizes of calcined Hombikat samples. The
200 400 600 800
Raman shift [cm-1
]
Rela
tive i
nte
nsit
y [
-]
25 120
500
600
700
800 850 900 925 950 975 1000 1100
anatase
rutile
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
[°C]
·cos
Effect of TiO2 source and thermal pre-treatment
61
whole phase transformation process occurrs within a temperature range of 900-1000°C. No rutile
phase is detected for samples calcined up to 900°C, nor anatase for 1000 and 1100°C calcined samples.
The anatase crystal size increases with increasing calcination temperature, which is nicely reflected by
the reduction of BET surface area, see figure 12a. Again, the optimal calcination temperature of 700°C
for photocatalytic decolorization of methylene blue cannot be solely explained by the changes in the
crystal size.
Table 3. Phase composition and crystal sizes of calcined Hombikat samples, as is determined
from XRD
Pre-treatment temperature Anatase fraction Anatase crystal size Rutile crystal size
[°C] [-] [nm] [nm]
Room temperature 100% 8
120 100% 9
500 100% 17
600 100% 23
700 100% 27
800 100% 37
850 100% 40
900 99% 42 40
925 95% 42 45
950 77% 44 >50
1000 0% >50
1100 0% >50
Surface hydroxyl groups can be envisaged to play an important role in photocatalytic
decolorization of methylene blue, because of its direct involvement in generation of the oxidizing
agent ·OH, and the surface chemisorption of methylene blue. Figure 15 compares the surface
properties of various photocatalysts, as well as the Hombikat catalyst calcined at different
temperatures. The total amount of surface –OH groups is in good agreement with those reported by
Chhor et al. [40], Van Veen et al. [28] and Boehm [41], the latter applied various probe molecules to
characterize surface hydroxyl groups of the P25 photocatalyst.
The comparison of surface –OH groups for different photocatalysts shows a similar trend as the
measured BET surface area, although the surface hydroxyl group densities (total amount of surface
–OH group normalized to the BET surface area) varies from 2.2 µmol/m2 for Aldrich_A sample to 7.7
µmol/m2 for Aldrich_B. The fluctuation in surface –OH group density could be explained by its
dependence on the exposed crystalline phase plane, surface defect, and the crystal size [42]. With
regard to the effect of thermal pre-treatment, the amount of surface –OH group on Hombikat catalyst
decreases monotonically with increasing calcination temperature. Hence, the increase of the apparent
photocatalytic activity at low calcination temperature region cannot be attributed to the surface –OH
group quantity.
Chpater 3
62
Figure 16. Surface hydroxyl group concentration of different photocatalyst (left) and its
dependency on calcination temperature for Hombikat catalyst (right).
Figure 17. Apparent photocatalytic decolorization activity of methylene blue (MB) as function of
the methylene blue uptake
Figure 17 summarizes the influence of methylene blue uptake capacity on apparent photo-activity
for different photocatalysts with various thermal pre-treatment procedures, determined by the decrease
of methylene blue concentration in dark aqueous solution until the equilibrium has been reached. The
points are scattered over the whole graph, from which hardly any trend can be derived.
3.4 Discussion
Comparison of photocatalysts from different sources in photocatalytic dye decolorization
Unlike conventional catalytic procesesses for which the apparent reaction rate is directly
correlated with the amount of accessible sites, photocatalytic decolorization of methylene blue exhibits
a complex behavior that is influenced by various factors. For a general application of photocatalysis, it
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Hom
bikat
Aldrich
AP25
Aldrich
B
Merc
k
Fluka
Su
rface -
OH
[m
mo
l/g
]
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 200 400 600 800 1000 1200
Calcination temperature [ ]
Su
rface -
OH
[m
mo
l/g
]
Hombikat
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0 0.005 0.01 0.015 0.02 0.025
Uptake capacity [mmol/gcat]
kap
p [
1/m
in]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
[°C]
Effect of TiO2 source and thermal pre-treatment
63
is of great importance to establish a minimum set of factors that can describe the photocatalytic
behavior of TiO2 in aqueous solutions, independent on the catalyst source and reactor configuration.
Total surface area is, among others, one of these determining parameters. As can be seen from
Appendix 3.1, the photocatalytic decolorization rates of Erythrosine B and Congo Red on different
TiO2 catalysts correspond nicely with their BET surface areas (Fig. A1, A2). It is well known that the
adsorption of organic molecules on TiO2 occurs solely on the surface hydroxyl groups. Comparing the
surface –OH measurement results (Fig.16, left) with the N2-physisorption determined surface area
(Fig.2), both show similar trends indicating the surface –OH density does not vary significantly per
catalyst source. Hence, it appears that for both Erythrosine B and Congo red, the surface area is one
and probably the most important parameter that determines the photocatalytic dye disappearance.
The nice correlation between SBET and kapp, however, is not observed for the methylene blue
decolorization (Fig. 1,2). Although Hombikat and Aldrich_A possess micro-/meso-porous structure
and high BET surface area, their surface specific activity ks,app lie far below those of low surface area
photocatalysts such as Fluka, Riedel de Haen and Sigma. Moreover the Merck sample exhibits
extraordinary high activity qua surface area, despite of its low SBET. As is explained previously (Table
1), both for Hombikat and Aldrich_A samples, the internal diffusion of methylene blue into the porous
agglomerates is most likely to limit the overall reaction rate. In all other cases, the mass transfer effects
on the apparent reaction activities are negligible. The high activity of the Merck sample in methylene
blue decolorization might be explained by the presence of Alumina, on which the selectivity is altered
as well due the different adsorption modes of methylene blue (Fig. 8, 9, Table 2).
Figure 18. Calculated molecular size of methylene blue, erythrosine B and Congo red and their
light absorption spectra
The internal diffusion limitation into Hombikat and Aldrich_A agglomerates probably occurs in
the Erythrosine B and Congo red photocatalytic decolorization as well. As is calculated using
ChemOffice 3D molecular simulation software, the molecular diameters of EB and CR are 1.8 and 2.3
nm respectively, larger than 1.5 nm of methylene blue. As a consequence, the effective diffusion
coefficients D,eff for EB and CR into TiO2 agglomerates should be less than that of MB. Combined
Methylene blue (MB)
0
0.4
0.8
1.2
190 390 590 790
Wavelength [nm]
Ab
so
rba
nc
e [
-]
Erythrosine B (EB) Congo Red (CR)
1.8 nm1.5 nm
0
1
2
3
4
190 240 290 340 390 440 490 540 590
Wavelength [nm]
Ab
so
rba
nc
e [
-]
2.3 nm
0
1
2
3
190 290 390 490 590
Wavelength [nm]
Ab
so
rban
ce [
-]
Chpater 3
64
with higher observed volumetric apparent reaction rates rV,obs, the calculated Wheeler-Weisz module
for Erythrosine B and Congo red according to equation 3.2 will be much higher than 0.15, that means a
strong internal diffusion limitation. Nevertheless, the influence of pore texture on the apparent
photocatalytic activity is much less pronounced than in the case of methylene blue. As can be seen in
Figure 19, the surface specific activities of Hombikat and Aldrich in Erythrosine B and Congo red
degradation only differ marginally with that of the benchmark low surface area, macroporous
photocatalyst Fluka.
Figure 19. Surface specific activity of photocatalysts normalized to that of Fluka. The surface
specific photocatalytic activities of Fluka were taken as unity in all three dye degradation cases.
Figure 20. Time course of photocatalytic decolorization of methylene blue, Erythrosine B and
Congo Red on Hombikat catalyst. Catalyst amount: 0.050 g (75-53 µm); liquid volume: 0.10 l; initial
dye concentration differs to get proper absorption measurements; illumination source: 8 blacklight
lamps
0
0.01
0.02
0.03
0.04
-150 -100 -50 0 50 100 150 200 250 300
Time [min]
Co
nc
en
tra
tio
n [
mm
ol/
l]
Methylene blue
Erythrosin B
Congo red
Dark
Light
0
1
2
3
4
5
6
Hombikat Aldrich_A P25 Aldrich_B Merck Fluka
Re
lati
ve
su
rfa
ce
sp
ec
ific
e a
cti
vit
y [
-]
Methylene blue
Erythrosine B
Congo Red
Effect of TiO2 source and thermal pre-treatment
65
It is therefore worthwhile to take a closer look at one catalyst in different photocatalytic dye
degradation processes. Figure 20 shows the time course of 3 separate runs on the Hombikat catalyst,
on methylene blue, erythrosine B and Congo red photocatalytic decolorization, respectively. The dark
adsorption of methylene blue on Hombikat is the least, followed by a slow photocatalytic bleaching
process as compared with Erythrosine B and Congo red decolorization. Highest dye uptake is achieved
with Congo red, whose degradation process on Hombikat apparently proceeds with higher rates than
the another two organic dyes as well.
Figure 21. Influence of the dye uptake capacity on their corresponding photocatalytic
decolorization rates on various TiO2 photocatalysts. (a) Hombikat: (b) Aldrich_A; (c) P25; (d) Merck;
(e) Fluka. Catalyst amount: 0.050 g (75-53 µm); liquid volume: 0.10 l; initial MB concentration 0.030
mmol/l; illumination source: 8 blacklight lamps
Figure 20 can be simplified into a scattered plot, that sketches the relationship between the dye
uptake and the corresponding 1st order photocatalytic decolorization rate, see Fig. 21(a). A monotonic
increase of the apparent reactivity with increasing dye uptake was found. Without further implication,
it can be envisaged that for Hombikat photocatalyst, the dye uptake plays an important to determining
0
0.01
0.02
0 0.005 0.01 0.015 0.02 0.025
Uptake capacity [mmol/gcat]
ka
pp
[1/m
in]
Methylene blue
Erythrosine B
Congo red
(e)
0
0.1
0.2
0.3
0.4
0.5
0 0.01 0.02 0.03 0.04 0.05 0.06
Uptake capacity [mmol/gcat]
ka
pp
[1
/min
]
Methylene blue
Erythrosine B
Congo red
(a)
0
0.1
0.2
0.3
0 0.005 0.01 0.015 0.02 0.025
Uptake capacity [mmol/gcat]
ka
pp
[1
/min
]
Methylene blue
Erythrosine B
Congo red
(b)
0
0.03
0.06
0.09
0 0.01 0.02 0.03 0.04Uptake capacity [mmol/gcat]
kap
p [
1/m
in]
Methylene blue
Erythrosine B
Congo red
(c)
0
0.02
0.04
0.06
0 0.005 0.01 0.015 0.02
Uptake capacity [mmol/gcat]
ka
pp
[1
/min
]
Methylene blue
Erythrosine B
Congo red
(d)
Chpater 3
66
role in the initial photocatalytic decolorization step. The linear trend was not found for the other four
photocatalysts tested (Fig.21 (b-e)). Probably the importance of the dye uptake capacity is suppressed
by other factors such as surface morphology, adsorption characteristics and surface chemistry.
With the word ‘uptake’ three separate contributions should be considered. The major contribution
is probably, in the case of methylene blue on Hombikat, from dyes molecules that chemically bond to
the TiO2 photocatalyst through surface rearrangement or hydrogen bonding. Weakly chemisorbed
overlayer and physisorption could contribute for a great part to the total uptake, especially on the low
surface area titania samples. Non-adsorbed dyes remaining in the solutions entrapped in the pores are
negligible in all TiO2 photocatalysts examined, Hombikat being the worst case with largest pore
volume of 0.46 ml/gcat. With the initial MB concentration of 0.03 mmol/l it means that total amount of
entrapped non-adsorbed MB is estimated to be 1.4×10-5 mmol/gcat, orders of magnitude lower than the
measured total uptake.
Figure 22. Dyes uptake capacity as function of BET surface area
It is noteworthy that the uptake capacity of different photocatalysts does not show a direct
relationship with the measured BET surface area of the photocatalyst. As is depicted in the Figure 22,
the uptake capacity varies per catalyst source and the nature of dyes adsorbed onto the photocatalyst.
Interestingly, the uptake of methylene blue onto photocatalyst shows an opposite trend as that of
Erythrosine B and Congo red. This could indicate a different adsorption mode that may have profound
impact on the apparent photocatalytic activity.
Assuming a planar alignment of dye molecules on the catalyst surface, the monolayer coverage of
photocatalyst by methylene blue, erythrosine B and Congo red can be estimated. It can be seen from
Figure 23, that for low surface area photocatalysts such as Merck and Fluka, the measured dye uptake
exceeds the estimated monolayer coverage, indicating the occurrence of multi-layer adsorption and/or
physisorption. The fact that for Hombikat, Aldrich_A and P25, the uptake is below the monolayer
converage can be explained by the low accessibility of micropores of the agglomerates for the dye
molecules. Hence the area associated with these micropores is not contributing to adsorption. The
uptake values of Erythrosine B and Congo red are in general, with the exception of Fluka and Merck,
0
0.02
0.04
0.06
0 100 200 300SBET [m2/g]
Up
tak
e c
ap
ac
ity
[m
mo
l/g
ca
t]
Methylene blue
Erythrosine B
Congo red
HombikatAldrich_AP25Merck
Fluka
Effect of TiO2 source and thermal pre-treatment
67
higher than that of methylene blue. The uptake capacity is, however, not necessarily linked to high
specific photocatalytic activity for most of the catalysts, as is shown in Figure 21(b) through Figure
21(e). Other factors as surface morphology, interaction of dyestuffs with the photocatalyst surface,
chemistry of the photo-degradation process and light absorption/scattering behavior could all have
their shares in determining the apparent reaction rate. It is not possible to point to a single factor that
dominates the photocatalytic decolorization process on TiO2 photocatalysts.
Figure 23. Specific uptake of dyestuffs by different TiO2 photocatalysts, of which the specific
uptake is defined as the total uptake normalized to BET surface area. (a): Methylene blue; (b):
Erythrosine B; (c): Congo red.
Unlike other photocatalysts, the photocatalytic dye disappearance on Hombikat could be
explained in a simplified matter. Photocatalytic processes require the simultaneous adsorption of
substrate, in this study dye stuffs, and the absorption of photon generating initial reactive species, the
surface bonded hydroxyl radicals. Hydroxyl radicals are generated by the electron-deficient hole attack
on the surface –OH groups or surface water. The reactive species generation is directly linked to the
total number of available site on the TiO2 surface, more explicitly the total surface area.
In case of methylene blue on Hombikat catalyst, the uptake is far below the monolayer coverage,
most of which is likely chemisorbed on the catalyst surface. The amount of reactive species is only a
function of the incident photon flux. Due to the excessive presence of surface hydroxyl groups and
surface water, the adsorption of dye molecules has only a marginal influence on the photon-assisted
generation process of reactive species. Assuming that the amount of photon-generated hydroxyl
radicals is prevailing over the adsorbed methylene blue, as is shown in Figure 24, it can be envisaged
that Methylene blue molecules are “floating” in the sea of hydroxyl radicals, and get degraded in their
first good-ever encounter. Therefore, the diffusion of methylene blue into the Hombikat agglomerates
is the most important and probably the rate-determining step.
The situation could be different in case of the other two dyes, erythrosine B and Congo red.
0
1
2
3
Hom
bikat
Aldrich
_A
P25
Merc
k
Fluka
Sp
ecif
ic u
pta
ke M
B [
µµ µµm
ol/m
2c
at]
Monolayer
(a)
0
0.3
0.6
0.9
Hom
bikat
Aldrich
_A
P25
Merc
k
Fluka
Sp
ecif
ic u
pta
ke E
B [
µµ µµm
ol/m
2c
at]
Monolayer
(b)
0
0.6
1.2
1.8
Hom
bikat
Aldrich
_A
P25
Merc
k
Fluka
Sp
ecif
ic u
pta
ke C
R [
µµ µµm
ol/m
2c
at]
Monolayer
(c)
Chpater 3
68
Strong adsorption of the dye molecules on the Hombikat catalyst surface is observed, which is likely to
be related to the functional groups of the molecules. Even multilayer coverage at the outer shell of the
agglomerates is possible (Fig. 24). Therefore the surface reaction of dye molecules with hydroxyl
radicals proceeds much faster, resulting in the depletion of reactive species. Surface bonded hydroxyl
radicals, once generated by the photon-absorption, will immediately react away with the first dye
molecule in the vicinity. Hence, the overall photocatalytic decolorization process is limited by the
charge-carrier separation and generation of surface reactive species, of which the total surface area is
the determining factor.
To summarize, photocatalytic decolorization of methylene blue on Hombikat is most likely
limited by the diffusion of dye molecules into the TiO2 agglomerates. The decolorization of
Erythrosine B and Congo red, on the other hand, proceeds much quicker and the rate-limiting factor
becomes the generation of surface reactive species by photon-absorption. This process is directly
linked to the total surface area. It explains the peculiar behavior we found in Fig. 19, that for the
methylene blue decolorization, the surface specific activity of Hombikat is much lower than the low
surface area TiO2 of Fluka, whilst for the decolorization processes of Erythrosine B and Congo red,
Hombikat exhibits comparable surface specific activity with Fluka.
Figure 24. Schematic implication of the adsorption/reaction of dye stuffs on Hombikat TiO2
surface activated by UV irradiation. Left: methylene blue; right: Erythrosine B & Congo red.
Thermal pre-treatment of commercial photocatalysts influences their corresponding apparent
reactivity in a profound way. By calcining commercial samples, the following parameters and catalyst
characteristics are modified, photon absorption, total surface area, surface morphology, phase
composition, grain size, surface hydroxyl groups, and defects and impurities. The influence of most of
these parameters on the apparent photocatalytic acivity has been elucidated in paragraph 3: Results.
Table 4 summarizes the discussion on the possible consequences of these modifications by thermal
Methylene blueErythrosine B
Congo red
Weak surface adsorption Strong surface adsorption
Diffusion limited reaction Radical generation determining
Dye molecule OH· radical
Effect of TiO2 source and thermal pre-treatment
69
treatment on the apparent photocatalytic activity. Please note that the list is far from exclusive.
Table 4. Influential factors with regard to photocatalyst thermal pre-treatment
Modifications caused by thermal treatment Possible consequences on photocatalytic activity
Increased photon absorption +
Less surface area -
Enlarged agglomerate pore diameter = / +
Phase transition from anatase to rutile - (rutile less active than anatase) / + (combination of
phases could be positive)
Larger crystal size + (primary charge segregation) / - (less surface area)
Reduced amount of surface hydroxyl groups -
Burning off impurities = / + / -
Remarks: +: activity increases;
-: activity decreases;
=: no influence on photocatalytic activity.
3.5 Conclusions
• Photocatalytic decolorization of methylene blue on TiO2 photocatalyst is a very complex
reaction. Total surface area and the associated surface hydroxyl groups are, among others, the
most important parameters determining catalyst effectivity.
• The apparent decolorization rate of methylene blue on Hombikat is most likely limited by the
internal diffusion of methylene blue into the porous agglomerates.
• TiO2 photocatalyst supplied from Merck exhibits an extraordinary high reaction rate in methylene
blue decolorization, possibly due to the presence of alumina impurities. It is also possible that the
mode of methylene blue adsorption and the degradation path are altered by the replacement of Ti
atoms in the titania framework by Al atoms.
• Thermal treatment of commercial TiO2 samples has a complicated impact on their apparent
photocatalytic activity. Complete understanding of the experimental results requires further study
and deep knowledge on the surface chemistry, transport phenomena and optical properties.
• With the exception of P25, the photocatalytic activity of Hombikat, Aldrich_A, Merck and Fluka
samples can be improved by an appropriate thermal treatment.
References
1. Ollis, D.F., Al-Ekabi, H., Photocatalytic Purification and Treatment of Water and Air,
Elservier, Amsterdam, 1993
2. Alfano, O.M., Bahnemann, D., Cassano, A.E., Dillert, R., Goslich R., Catal. Today, 2000, 58,
199
3. Bhakhande, D.S., Pangakar, V.G., Beenackers, A.A.C.M., J. Chem. Tech. Biotech., 2002, 77,
102
4. Legrini, O., Oliverous, E., Braun, A.W., Chem. Rev., 1993, 93, 671
5. Hoffmann, M.R., Martin, S.T., Choi, W., Bahnemann, D.W., Chem. Rev., 1995, 95, 69
6. Halmann, M.M., Photodegradation of water pollutants, CRC press, Boca Raton, 1996
Chpater 3
70
7. Blake, D.M., Maness, P.C., Huang, Z., Wolfrum E.J., Huang, J., Jacoby, W.A., Separ. Purif.
Meth., 1999, 28, 1
8. Zollinger, H., Color Chemistry, Synthesis, Properties and Applications of Organic Dyes and
Pigments, 2nd
eds., VCH, 1991
9. Perelta-Zamora, P., Kunz, A., Moraes, S.G., Pelegrini, R., Moleiro, P.C., Reyes, J., Duran, N.,
Chemosphere, 1999, 38, 835
10. Linsbigler, A.L., Guangquan, L., Yates, J.T., Chem. Rev., 1995, 95, 735
11. Liu, G., Wu, T., Zhao, J., Hidaka, H., Serpone, N., Environ Sci. Technol., 1999, 33, 2081
12. Konstantinou, I.K., Albanis, T.A., Appl. Catal. B: Envrion. 2003,, 42, 319
13. Matthews, R.W., J. Chem. Soc. Faraday Trans., 1989, 1, 1291
14. Houas, A., Lachheb, H., Ksibi, M., Elaloui, E., Guillard, C., Herrmann, J.M., Appl. Catal. B:
Environ. 2001, 31, 145
15. Lakshmi, S., Renganathan, R., Fujita, S., J. Photochem. Photobiol. A: Chemistry 1995, 88, 163
16. Wu, C.H., Chern, J.M., Ind. Eng. Chem. Res. 2006, 45, 6450
17. Kuo, W.S., Ho, P.H., Chemosphere 2001, 40, 77
18. Awati, P.S., Awate, S.V., Sheh, P.P., Ramaswamy, V., Catal. Comm. 2003, 4, 393
19. Burda, C., Lou, Y., Chen, X., Samia, A.C.S., Stout, J., Gole, J.L., Nanoletters 2003, 3, 1049
20. Khan, S.U.M., Al-Shahry, M., Ingler, W. B. Jr., Science 2002, 297, 2243
21. Inagaki, M., Imai, T., Yoshikawa, T., Tryba, B., Appl. Catal. B: Environ. 2004, 51, 247
22. Horikiri, S., Teshima, N., Saruki, Y., Nishikawa, H., Sakai, T., Bunseki Kagaku 2003, 52, 881
23. An, T.C., Zhu, X.H., Xiong, Y., Chemosphere 2002, 46, 897
24. Reddy, K.M., Guin, D., Manorama, S.V., J. Mater. Res. 2004, 19, 2567
25. Krysa, J., Keppert, M., Jirkovsky, J., Stengl, V., Subrt, J., Mater. Chem. Phys. 2004, 86, 333
26. Tanaka, Y., Suganuma, M., J. Sol-Gel Sci. Technol. 2001, 22, 1573
27. Chan, A.H.C., Porter, J.F., Barford, J.P., Chan, C.K., J. Mater. Res. 2002, 17, 1758
28. Van Veen, J.A.R., Veltmaat, F.T.G., Jonkers, G., J. Chem. Soc. Chem. Commun. 1985, 1656
29. Wang, C.Y., Bottcher, C., Bahnemann, D.W., Dohrmann, J.K., J. Nanopart. Res.2004, 6, 119
30. Brunauer, S., Deming, L., Deming, W., Teller, E., J. Am. Chem. Soc. 1940, 62, 1723
31. Krupa, N., Cannon, F., J.Am. Water Works Asso. 1996, 6, 94
32. F. Kapetijn, Marin, G.B., Moulijn, J.A., in Catalysis: A Intergrated Approach, 2nd
Ed., Van
Santen, R.W., Van Leeuwen, P.W.N.M., Moulijn, J.A., Averill, B.A., eds., Elservier,
Amsterdam, 1999, pp 375
33. Satterfield, C.N., Colton, C.K., Pitcher, W.H., AIChE J. 1973, 19, 628
34. Hurum, D.C., Agrios, A.G., Gray, K.A., Rajh, T., Thurnauer, M.C., J. Phys. Chem. B 2003, 107,
4545
35. Ohno, T., Tokieda, K., Higashida, S., Matsumura, M., Appl. Catal. A: Gen. 2003, 244, 383
36. Yoshida, H., Curr. Opin. Solid State Mater. 2003, 7, 435
37. Stroyuk, A.L., Kryokov, A.I., Kuchmii, S.Y., Pokhodenko, V.D., Theo. And Experim. Chem.
2005, 41, 67
38. Sclafani, A., Herrmann, J.M.; J. Phys. Chem. 1996, 100, 13655
39. Su, C., Hong, B.Y., Tseng, C.M., Catal. Today 2004, 96, 119
40. Chhor, K., Bouquet, J.F., Colbeau-Justin, C., Mater. Chem. Phys. 2004, 86, 123
41. Boehm, H.P., Discuss. Faraday Soc., 1971, 52, 264
42. Arrouvel, C., Digne, M., Breysse, M., Toulhoat, H., Raybaud, P., J. Catal. 2004, 222, 152
Effect of TiO2 source and thermal pre-treatment
71
Appendix 3.1 Photocatalytic decolorization of Erythrosine B (EB) and Congo Red (CR)
In a parallel study to methylene blue degradation, the photocatalytic decolorization of two other
commonly applied dyes was investigated. Similar to Methylene blue, Erythrosin B is widely applied as
a photo-sensitizer, upon visible light irradiation its excited state can inject an electron into the
conduction band of semiconductor particles [1-3]. Congo red represents one type of the diazo dyes,
those dyes and related compounds are widely used as industrial dyes for foods, drugs, cosmetics,
textile, printing inks, and laboratory indicators [4]. It has been widely studied for the TiO2 mediated
photocatalytic degradation process using UV as well as visible light [5-8]. Generally, the sites near the
azo bond (C-N=N- bond) form the attack area in the photocatalytic degradation process, whilst the
TiO2 photocatalytic destruction of the C-N= bond and -N-N- bonds leads to fading of the dyes [9].
Photocatalytic decolorization reactions were performed in high throughput photocatalytic reaction
assembly (HTPR) applying photocatalysts as is characterized in chapter 3. 0.05 g of commercial
catalyst was suspended in 100 ml of aqueous solution. 8 blacklight lamps were applied to facilitate the
photocatalytic processes. After checking that no detectable degradation occurred without titania nor
UV-irradiation, the photocatalytic disappearance of the dyes was monitored by measuring the light
absorption of the aqueous solution at the absorption peaks, 470 nm and 450 nm for EB and CR
respectively, referring to the experiment section of chapter 2. The apparent decolorization kinetics is
assumed to be first order due to low reactant concentration [10], therefore the apparent kinetic rate
constant kapp [1/min] is used as the single parameter to compare photocatalyst performances.
Figure A1 shows the apparent reaction rates of the commercial TiO2 photocatalysts in Erythrosine
B photocatalytic decolorization, together with their corresponding BET surface area. A monotonically
decrease of the apparent reaction rate with the BET surface area was revealed, as Hombikat TiO2
shows substantially higher activity than the more universally applied photocatalyst P25 and bulky TiO2
of Merck and Fluka. Similar trend was found for the photocatalytic decolorization of Congo Red, as is
shown in Figure A2.
Figure A1. Apparent photocatalytic decolorization activity of Erythrosine B (EB), together with
measured values of BET surface area. Error bars represent 95% confidence interval.
The uptake capacity of Erythrosine B on the TiO2 photocatalyst was measured in dark, by
reaching adsorption equilibrium after 2 hrs. Results with different commercial photocatalysts as well
as same catalyst modified at different pre-treating temperature were plotted together to construct the
0
0.04
0.08
0.12
0.16
0.2
Hombikat Aldrich__A P25 Aldrich_B Merck Fluka
ka
pp
[1/m
in]
0
100
200
300
400
SB
ET [
m2/g
]
Chpater 3
72
collaborative figure of apparent 1st order reaction rate constant with the corresponding dark adsorption
(Fig. A3). Although the measured points are fairly scattered, figure A3 shows a trend of enhanced
apparent photocatalytic activity with increasing amount of Erythrosine B adsorbed.
Figure A2. Apparent photocatalytic decolorization activity of Congo Red (CR), together with
measured values of BET surface area. Error bars represent 95% confidence interval.
Figure A3. Apparent photocatalytic decolorization activity of Erythrosine B (EB) as function of
the Erythrosine B uptake
Effect of calcination temperature on the photocatalyst activity was also investigated with
Erythrosine B and Congo Red as the probe molecules. Thermal pre-treatment procedure was described
in the experimental section of chapter 3. Figure A4 summarized the apparent reaction rates of
erythrosine B photo-decolorization on the calcined Hombikat, Aldrich_A, P25, Merck, and Fluka
samples up to 1000°C. All catalysts show a critical temperature above which the photocatalytic
activity drops dramatically. The onset of the drop is at 500°C for Hombikat and P25, at 600°C for
Aldrich_A and at above 800°C for Merck and Fluka samples. Unlike for the cases of methylene blue
and ErythrosineB, the apparent 1st order photocatalytic reaction rate of Congo Red decolorization
decreases monotonically with increasing calcination temperature (Fig A5).
0
0.1
0.2
0.3
0.4
0.5
Hombikat Aldrich_A P25 Aldrich_B Merck Fluka
ka
pp
[1/m
in]
0
100
200
300
400
SB
ET [
m2/g
]
0
0.05
0.1
0.15
0.2
0.25
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
Adsorption capacity [mmol/gcat]
ka
pp
[1
/min
]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
Effect of TiO2 source and thermal pre-treatment
73
Figure A4. Effect of calcination temperature on the apparent reaction rate of photocatalytic
decoloriation of erythrosine B. (a) Hombikat: (b) Aldrich_A; (c) P25; (d) Merck; (e) Fluka. Lines are
for guide the eyes. Error bars represent the 95% confidence interval of 1st reaction kinetic fitting.
catalyst amount: 0.050 g (75-53 µm); liquid volume: 0.10 l; initial EB concentration 0.030 mmol/l;
illumination source: 8 blacklight lamps.
0
0.05
0.1
0.15
0.2
0.25
0.3
0 200 400 600 800 1000 1200
Temperature [ ]
ka
pp
[1/m
in]
(a)
0
0.05
0.1
0.15
0.2
0.25
0.3
0 200 400 600 800 1000 1200
Temperature [ ]k
ap
p [
1/m
in]
(b)
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0 200 400 600 800 1000
Temperature [ ]
kap
p [
1/m
in]
(c)
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0 200 400 600 800 1000 1200
Temperature [ ]
kap
p [
1/m
in]
(d)
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0 200 400 600 800 1000 1200
Temperature [ ]
kap
p [
1/m
in]
(e)
[°C]
[°C][°C]
[°C]
[°C]
Chpater 3
74
Figure A5. Effect of calcination temperature on the apparent reaction rate of photocatalytic
decoloriation of Congo Red. (a) Hombikat: (b) Aldrich_A. Lines are for guide the eyes. Error bars
represent the 95% confidence interval of 1st reaction kinetic fitting. catalyst amount: 0.050 g (75-53
µm); liquid volume: 0.10 l; initial CR concentration 0.030 mmol/l; illumination source: 8 blacklight
lamps.
Figure A6. Apparent photocatalytic decolorization activity of Congo red (CR) as function of the
Congo red uptake
The uptake capacity of Congo red on the TiO2 photocatalyst is shown in Figure A6. It can be seen
that the apparent photocatalytic activities are rather low and increase with increasing uptake capacity,
with 2 outstanding exceptions of Hombikat and Aldrich_A, both being uncalcined samples.
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0 200 400 600 800 1000 1200
Temperature [ ]
kap
p [
1/m
in]
(a)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0 200 400 600 800 1000 1200
Temperature [ ]k
ap
p [
1/m
in]
(b)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0 0.01 0.02 0.03 0.04 0.05 0.06
Uptake capacity [mmol/gcat]
ka
pp
[1
/min
]
Hombikat
Aldrich_A
P25
Aldrich_B
Merck
Fluka
[°C][°C]
Effect of TiO2 source and thermal pre-treatment
75
References used in Appendix 3.1
1. Linsebigler, A.L., Lu, G., Yates, J.T.Jr., Chem. Rev. 1996, 95, 636
2. Kamat, P.V., Fox, M.A., Chem. Phys. Lett. 1983, 102, 379
3. Zhang, F., Zhao, J., Zang, L., Shen, T., Hidaka, H., Pelizzetti, E., Serpone, N., J. Mol. Catal. A:
Chem. 1997, 120, 173
4. Salem, I., Catal. Rev. 2003, 45, 205
5. Lachheb, H., Puzenat, E., Houas, A., Ksibi, M., Elaloui, E., Guillard, C., Herrmann, J.M., Appl.
Catal. B: Environ. 2002, 39, 75
6. Guillard, C., Lachheb, H., Houas, A., Ksibi, M., Elaloui, E., Herrmann, J.M., J. Photochem.
Photobiol. A : Chem. 2003, 158, 27
7. Tanaka, K., Padermpole, K., Hisanaga, T., Water Res. 2000, 34, 327
8. Xu, Y., Langford, C.H., Langmuir 2001, 17, 897
9. Zhang, F., Zhao, J., Shen, T., Hidaka, H., Pelizzetti, E., Serpone, N., Appl. Catal. B: Envrion.
1998, 15, 147
10. Konstantinou, I.K., Albanis, T.A., Appl. Catal. B: Environ 2004, 49, 1
Chpater 3
76
77
4
The Effect of Surface OH-population on the Photocatalytic Activity
of Rare Earth doped P25-TiO2 in Methylene Blue Degradation
Abstract
Commercial TiO2 (P25, from Degussa) was doped with La, Ce, Zr, Y, Pr and Sm, and the
activity of the samples as a function of calcination temperature was tested in methylene-blue
photocatalytic degradation. Samples were characterised by N2 adsorption, Raman spectroscopy, XRD
and UV-Vis absorption. Doping of P25 with rare earth metals (RE), combined with calcination at 600
ºC or 800 ºC, yields materials with surface areas ranging from ~10 to 50 m2/g, and an anatase to rutile
phase ratio ranging from ~0.03 to 0.7, as determined from XRD data. After pretreatment of P25 at 600
ºC compared to the other catalysts studied exhibits the highest activity in methylene-blue degradation
in a combinatorial reactor, while rare earth metal modification decreases the activity. After
pretreatment at 800 ºC, RE modified catalysts perform better in methylene blue degradation than
unpromoted P-25, La being the preferred RE. From the extensive data set, nor the anatase to rutile
ratio, nor the BET area was found to correlate with the observed methylene-blue decomposition rate.
Rather, by evaluation of the DRIFT spectra of the various catalysts, a linear correlation between the
number of a specific Ti-OH group and the methylene blue degradation rate was determined, suggesting
that this OH-group is an important precursor for the reactive site in aqueous phase Methylene Blue
degradation, and a dominant factor in controlling performance.
Chapter 4
78
4.1 Introduction
TiO2 as a raw material has grown to a 4 million-ton business from its discovery in 1791. It is
widely used in the pigment industry for paints and varnishes, papers, cosmetics, and plastics, and has
found potential applications in catalysis and ceramic membranes [1]. Over the last decades, TiO2 based
materials have been intensively studied as photocatalysts. Commercial explorations are focused on the
applications in destruction of pollutants in water and air, self-cleaning windows and buildings, and
self-sterilization by the oxidation of hydrocarbons into CO2. Recent development is the
NOx-abatement in crowded populated environments [2-4].
The three polymorphs of TiO2 are anatase, rutile and broskite, and most commercial powders are
composed of anatase, rutile or a mixture of these two phases. The difference in crystal form of rutile
and anatase has been reflected in their physical, physiochemical, and optical properties [5]. Anatase is
generally considered as the photoactive phase, whereas rutile is commonly thought to have low
photocatalytic activity or even to be inactive [6-8]. However, reactions in which both crystalline
phases have the same photoreactivity or rutile behaves more active are also reported [9-11]. Other
properties affecting the photocatalytic activity of TiO2 are particle size, crystal structure, nature of the
pollutants and the surface chemistry determining, for instance, its adsorption properties.
The photocatalytic activity of various photocatalysts is often compared to that of a commercial
reference catalyst, P25 from Degussa. This photocatalyst has been established as a benchmark in
photocatalysis because of its high photocatalytic activity, well-known structure, and commercial
availability. It is a mixed phase TiO2 made of approximately 70% of anatase and 30% of rutile. Both
phases exist separately according to a morphology study based on TEM [12].
To enhance the quantum yield of commercially available TiO2, which is typically below 1%, TiO2
particles have often been chemically modified. While many reports exist on modification with
transition metal oxides to enhance the quantum yield and visible light sensitivity, the effect of doping
with lanthanides has been less extensively investigated. Generally a positive effect of La-doping on
photocatalytic activity of TiO2 is reported. Inhibition of recombination of electrons and holes [13-15],
or beneficial surface adsorption properties [16-18] has been proposed to explain this positive effect.
Both explanations are based on a direct involvement of the promoter in the reactions studied, which
include Methylene Blue degradation [13], Nitrite degradation [14], 2-Mercaptobenzothiazole
decomposition [15], Rhodamine B degradation [16], and Salicylic, t-Cinnamic Acid, and
p-Chlorophenoxyacetic acid degradation [17,18].
The main objective of the present study was to further evaluate the performance of a commercial
TiO2 photocatalyst P25 from Degussa after modification with rare earth oxides (La, Ce, Y, Pr, and Sm)
in the Methylene Blue decomposition reaction at 370 nm. We compared performance after calcination
at 600 ºC or 800 ºC, representative of pretreatment conditions applied in other studies [13-18].
Furthermore we specifically tried to correlate the degradation rate to material properties, i.e. the
surface area, anatase/rutile ratio of the samples, and the nature of the hydroxyl groups present on the
catalyst surface as determined by Diffuse Reflectance infrared spectroscopy. From the extensive data
set it can be derived that the number of surface Ti-OH groups available in the light exposed reactor
volume shows a strong correlation with the observed methylene blue decomposition rate, suggesting
that these groups are largely determining the P-25 reactivity for this specific reaction.
Rare-earth doped P25-TiO2 in methylene blue degradation
79
4.2 Experimental
Photocatalyst preparation
Rare Earth (RE)-doped TiO2 samples were prepared by using La(NO3)3·6H2O (Merck, 99%),
Ce(NO3)3·6H2O (Aldrich, 99%), Y(NO3)3 (Aldrich, 99%), Pr(NO3)3 (Aldrich, 99%), Sm(NO3)3
(Aldrich, 99%), and TiO2 (Degussa P25). The required amounts of the nitrate precursors were
physically mixed with TiO2 in a mortar and were calcined overnight in static air in a furnace at 600 °C
or 800 °C, applying a heating rate of 10 °C/min. The target RE loading was 0.2, 1, or 2 wt-%,
respectively. For comparison, undoped P25 was heat-treated under the same conditions, and a series of
La2O3, CeO2, Y2O3, ZrO2, PrO2, and Sm2O3 was prepared by calcination of the nitrate precursors in the
absence of TiO2, also in similar conditions. Doped samples are denoted as “P25_%RE_Temp”, “%”
being the target percentage of RE and “Temp” the calcination temperature. The undoped P25
photocatalysts are named P25_600 and P25_800.
Characterization
Textural properties of the samples, i.e., Brunauer-Emmett-Teller (BET) surface area, the pore
dimension and the pore volume, were obtained by N2 adsorption at –196°C in a Quantachrome
Autosorb 6B apparatus. Before the N2 adsorption measurements, the samples were pre-treated in
vacuum at 110 °C for 16 hours.
The X-ray diffraction (XRD) patterns were recorded on a Philips PW1840 X-ray diffractometer
using CuKα radiation at a scan rate of 2θ = 0.01°s-1 and used to identify the crystal phase and their
corresponding crystallite size. The accelerating voltage and the applied current were 40kV and 50 mA,
respectively.
The relative abundance of anatase to rutile in the samples was calculated by using the following
equation [19]:
rr
a r
1.26 IF
I 1.26 I
⋅=
+ ⋅ (1)
where Fr is the rutile fraction, and Ir and Ia are the strongest intensities of the rutile (110) and
anatase (101) diffraction angles, respectively.
The crystal sizes (D) of anatase and rutile were determined by employing the Scherrer equation:
·
os=
KD
λ
β θβ•cosθ (2)
where λ is the wavelength of the Ni-filtered CuKα radiation used (λ = 0.15418 nm), β the full
width at half-maximum of the diffraction angle considered, K a shape factor (0.9) and θ the angle of
diffraction. For these calculations the indices (101) for anatase and (110) for rutile were used.
The absorption spectra of solid samples were measured using a Varian Cary 1 UV-Vis
Chapter 4
80
spectrometer equipped with a diffuse reflection accessory. BaSO4 was used as the reference material.
Samples were scanned with a light beam ranging from 190 nm to 500 nm with a scanning rate of 10
nm·s-1
.
Raman analysis was performed using a Renishaw Ramascope System 1000 instrument linked to a
Leica microscope. A 514 nm, 20 mW Ar+ laser was used as excitation source. The backscattered light
was filtered for Rayleigh scattering using a holographic notch filter. A CCD detector coupled to a PC
was used to obtain the Raman spectrum with a resolution of 4 cm-1.
The IR absorption spectra of the solid samples were recorded using a Thermo Nicolet Nexus
spectrometer with a MCT detector and a Spectratech Diffuse Reflectance Accessory equipped with a
high temperature cell. The spectrum of KBr at 120 oC in flowing He (25 ml/min) was used as
background. Water was removed from the catalyst surface to facilitate the analysis of the OH-group
composition by recording the spectra at 120 oC after equilibration for 15 minutes in flowing He (25
ml/min), applying a ramp rate of 10 oC/min. All spectra were recorded from 4000-700 cm-1 by
collecting 64 scans with a resolution of 4 cm-1. The Kubelka-Munk and pseudoabsorbance (noted here
as absorbance, for the sake of brevity) transformations were considered in representing the data, and in
view of a recent evaluation by Meunier et al. [19], we chose to use absorbance as a measure for the
relative contributions of the various OH-groups to the IR spectra of the investigated titanias.
Photocatalytic tests
Methylene Blue (MB) was obtained from Merck (97%) and used without further treatment.
Photocatalytic activity measurements were carried out in a combinatorial screening assembly, outlined
in chapter 2. In each run up to 10 parallel experiments could be performed simultaneously. Preliminary
photocatalytic experiments proved that all 10 reactors behaved identically and the results between runs
were comparable within ±8% error range [21]. The UV irradiation was delivered by 8 blacklight lamps
(18 W, Philips) maximizing at 370 nm, providing a light flux of 470±20 µW/cm2 entering the
TiO2/Methylene Blue suspension. For each experiment 50 mg of photocatalyst, sieved to a fraction of
53-75 mm, was added to a 100 ml aqueous solution of MB (0.03 mmol/l). Before the start of the
reaction, the mixture was stirred using a magnetic stir-bar in the dark for 2 hours to establish MB
adsorption-equilibrium. During the reaction the reactor housing was continuously purged with a fan
and the temperature was controlled at 32 ± 2 °C. Samples were withdrawn at constant time intervals
and filtered through a 0.45 µm PTFE Millipore membrane filter to remove suspended agglomerates.
Experimental checks proved that the amount of MB retained by the filter was negligible. Furthermore,
reference experiments indicated that the photosensitized degradation of MB did not take place in the
absence of photocatalysts. A UV-VIS spectrometer was used to record the absorbance spectra of the
solutions in the 400-1000 nm range with a spectral resolution of 0.33 nm. Calibrations were taken at
10 wavelengths adjacent to the maximum absorbance of MB at 667 nm. A Lambert-Beer diagram,
typically in the form of absorbance:
A = -log(I/I0) = ε·b·c (3)
was established to correlate the absorbance to MB concentration, where ε is the
wavelength-dependent molar absorption coefficient with units of m2·mol
-1, b is the light path length
Rare-earth doped P25-TiO2 in methylene blue degradation
81
(m), and c is the MB concentration (mol·m-3).
4.3 Results
Textural analysis
The BET surface areas of the different samples are compiled in Table 1. Figure 1 gives three
examples of the N2 adsorption-desorption isotherms All showed type II isotherms (Figure 1a)
indicating some meso- and macro-porosity. It is well-known that P25 consists of non-porous
nanoparticles [22]. The absence of a plateau at high relative pressure (p/p0) could indicate the filling of
inter-particle voids and the presence of surface roughness. The BET surface area of the TiO2
photocatalyst P25 was 51 m2/g, and little difference was observed in total surface area after calcination
at 600 °C, both for P25_600, and all rare earth modified samples (P25_0.2RE_600). However, further
increase of the thermal treatment temperature resulted in a significant reduction of surface area of
un-promoted P25 to 16 m2/g (P25_800). Comparison of this value with the remaining surface area of
the 0.2 wt-% RE-doped samples calcined at 800 °C shows that doping of P25 partially stabilizes the
textural properties, resulting in surface areas in the range of 17-24 m2/g. Enhancing the rare earth
amount to 1 wt-% showed a somewhat enhanced stabilization effect, with the BET area ranging from
16-30 m2/g. Further enhancement of the RE-content to 2 wt-% was detrimental.
Figure 1. N2 adsorption-desorption isotherm of P25 calcined at 600 °C and 800 °C and P25
doped with 1% La, calcined at 800 °C (left), and their corresponding pore diameter distribution
(right).
P/P0
0.0 0.2 0.4 0.6 0.8 1.0
Adso
rbe
d v
olu
me
[cc/g
]
0
50
100
150
200
250
300
P25_600
P25_1La_800
P25_800
Pore diameter [Å]
10 100 1000
De
sorp
tion (
dV
/dlo
gd)
[cm
3/g
]
0.0
0.2
0.4
0.6
0.8
Chapter 4
82
Table 1. Characterization of samples
Sample Anatase
fraction(a)
Anatase
crystal size(a)
[nm]
Rutile crystal
size(a)
[nm]
Band gap
energy(b)
[eV]
SBET
[m2/g]
P25 0.70 22 37 3.25 51
P25_600 0.70 25 36 3.23 47
P25_0.2La_600 0.71 28 41 3.23 46
P25_0.2Ce_600 0.71 27 50 3.19 47
P25_0.2Y_600 0.72 28 39 3.16 46
P25_0.2Zr_600 0.68 27 41 3.15 42
P25_0.2Pr_600 0.71 27 50 3.14 47
P25_0.2Sm_600 0.70 26 47 3.16 46
P25_800 0.05 - 43 3.04 16
P25-0.2La-800 0.22 31 43 3.03 23
P25-0.2Ce-800 0.15 35 45 3.04 19
P25-0.2Y-800 0.13 30 50 3.02 20
P25-0.2Zr-800 0.02 - 50 3.01 11
P25-0.2Pr-800 0.31 35 47 3.02 24
P25-0.2Sm-800 0.08 29 50 3.01 17
P25-1La-800 0.31 35 45 3.05 25
P25-1Ce-800 0.48 31 47 3.07 30
P25-1Y-800 0.05 - 50 3.04 16
P25-1Zr-800 0.03 - 47 3.03 13
P25-1Pr-800 0.15 31 45 3.03 21
P25-1Sm-800 0.37 33 50 3.06 29
P25_2La_800 0.14 34 47 20.5
P25_2Ce_800 0.15 35 49 18 (a)
Determined from XRD. (b)
Determined from UV-VIS.
X-ray diffraction
The measured XRD patterns and the derived crystal sizes and anatase fractions of the investigated
photocatalysts are given in Figure 2 and Table 1, respectively. The XRD characterization showed that
both anatase and rutile phases were present in the commercial P25 sample. Both the anatase to rutile
ratio (70:30) and the crystallite sizes (22 nm for anatase and 37 nm for rutile) are in good agreement
with results found by other authors [20,23].
Characteristic diffraction lines of La2O3, CeO2, Y2O3, ZrO2, PrO2, and Sm2O3 were not detectable
Rare-earth doped P25-TiO2 in methylene blue degradation
83
in RE-doped P25 up to 1 wt-% loading. On the contrary, the diffractograms of P25_2La_800 and
P25_2Ce_800 contained characteristic lines at 2θ ~ 39.8° and 28.7°, respectively, indicating the
formation of segregated phases of La2O3 or CeO2.
Figure 2. XRD characterization. (a) P25-0.2RE-800, (b) P25-1RE-800 and (c) P25-0.2RE-600.
20 30 40 50 60 70
2θ (deg.)
Inte
nsi
ty (
a.u.)
P25_0.2Ce_800
P25_0.2La_800
P25_0.2Pr_800
P25_0.2Sm_800
P25_0.2Y_800
P25_0.2Zr_800
P25_800
A (101)
R (110)A (004)
R (200)
R (101) R (111)
A (200)
A (105)
R (211) A (211)R (220)
R (210)(a)
20 30 40 50 60 70
2θ (deg.)
Inte
nsi
ty (
a.u
.)
P25_1Ce_800
P25_1La_800
P25_1Pr_800
P25_1Sm_800
P25_1Y_800
P25_1Zr_800
P25_800
A (101)
R (110)R (101) R (200)
A (004) R (111)
R (210)
A (200)A (105)
R (211) A (211)
R (220)(b)
20 30 40 50 60 70
2θ (deg.)
Inte
nsi
ty (
a.u.)
A (101)
R (110)A (103)
A (004)
A (112)
R (101)R (111)
A (200)
A (105)
R (211)
A (211)
R (220)
TiO2_0.2Ce_600
TiO2_0.2La_600
TiO2_0.2Pr_600
TiO2_0.2Sm_600
TiO2_0.2Y_600
TiO2_0.2Zr_600
TiO2_600
(c)
P25
P25
P25
P25
P25
P25
P25
Chapter 4
84
( A: Anatase, R: Rutile).
As indicative from the results of the N2 adsorption measurements, thermal treatment at 600 °C did
not change the composition of P25, nor did the dopants affect the relative phase composition.
Regardless of the type of doping, the anatase fraction in the photocatalysts calcined at 600 ºC was
about 0.70. In contrast, thermal treatment at 800 °C did change the morphology of P25, converting
anatase to rutile, the remaining anatase fraction in P25_800 being only 0.05. For the RE-doped
samples calcined at 800 ºC, significant inhibition of phase transformation from anatase to rutile was
observed, in agreement with the data of Zhang [16]. Doping of P25 with 0.2 % RE retarded rutile
formation, the anatase fraction following the decreasing order Pr > La > Ce > Y > Sm.
Unlike the findings presented in other studies [24,25], the Zr additive had trivial or even a slightly
accelerating effect in the anatase to rutile phase transformation. The inhibiting effect in phase
transformation of P25 becomes more dominant by enhancing the doping concentration to 1 wt-% of La,
Ce or Sm, while being less effective for Y and Pr. An increase in the RE concentration to 2 wt-% was,
however, less effective in the inhibition of phase transformation, as shown in Table 1 for the La and Ce
doped samples.
Raman characterization
Figure 3 shows the Raman spectra of the different photocatalysts. According to factor group
analysis, anatase has six Raman active modes (A1g + 2B1g + 3Eg). These allowed modes of anatase
appeared in the Raman spectrum at 144 cm-1 (Eg), 197 cm-1 (Eg), 399 cm-1 (B1g) 513 cm-1 (A1g), 519
cm-1
(B1g), and 639 cm-1
(Eg). The bands near 608 cm-1
and 446 cm-1
were identified as the A1g and Eg
modes for the rutile phase, respectively [26,27].
All photocatalysts prepared at 600 ºC showed the anatase absorption bands, and the Eg rutile band
was only observed in a few cases. On the contrary, P25_800 presents the rutile structure and 1 % Ce,
La and Sm-doped P25 and 0.2% Ce, La, Pr and Y-doped P25 (800 ºC-calcined) present bands
attributed both to rutile and anatase. No other peaks besides those of anatase and rutile were found in
Raman, in agreement with the XRD results.
100 300 500 700 900
Raman shift (cm-1)
Inte
nsi
ty (
arb
.)
P25_0.2Ce_800
P25_0.2La_800
P25_0.2Pr_800
P25_0.2Sm_800
P25_0.2Y_800
P25_0.2Zr_800
P25_800
A (Eg)A (B1g) A (A1g)
A (Eg)
R (A1g)R (Eg) (a)
Rare-earth doped P25-TiO2 in methylene blue degradation
85
100 300 500 700 900
Raman shift (cm-1)
Inte
nsi
ty (
arb.)
P25_1Ce_800
P25_1La_800
P25_1Pr_800
P25_1Sm_800
P25_1Y_800
P25_1Zr_800
P25_800
A (Eg) A (B1g) A (A1g)A (Eg)
R (A1g)R (Eg) (b)
100 300 500 700 900
Raman shift (cm-1)
Inte
nsi
ty (
arb
.)
P25_1Ce_800
P25_1La_800
P25_1Pr_800
P25_1Sm_800
P25_1Y_800
P25_1Zr_800
P25_800
A (Eg) A (B1g) A (A1g)A (Eg)
R (A1g)R (Eg) (c)
Figure 3. Raman characterization. (a) P25-0.2RE-800, (b) P25-1RE-800 and (c) P25-0.2RE-600.
(A: Anatase, R: Rutile).
UV-VIS characterization
UV-VIS spectra of selected photocatalysts (P25_T and P25_%La_T) are shown in Figure 4. The
absorbance spectrum of P25_800 shifted significantly towards the visible region compared to that of
P25_600. La doping at 600 °C altered the absorbance of P25 slightly, while the shape and onset in
UV-absorption remained unchanged. Similar behaviour was found for all the 0.2 % RE-doped P25
prepared at 800 °C. Increasing the RE content to 1% resulted in a decrease of the absorbance between
340–400 nm for all samples prepared in this work.
Band gap energies of all the photocatalysts were determined from the maximum of the first
derivative of the absorbance around the absorption edge and are listed in Table 1 [28]. It can be seen
that in the original P25 samples the energy of band gap is about 3.2 eV, corresponding to the dominant
Chapter 4
86
anatase phase. UV-VIS spectra indicated that RE-doping at 600 °C changed the UV adsorption edge to
the visible region. Accordingly, the band gap energies of the doped catalysts were reduced. The
samples prepared at 800 °C give even lower energies of band gap of about 3.0 eV, which is attributed
to the enrichment in rutile phase.
Figure 4. UV-VIS spectra of La-doped P25 (top) and RE precursors calcined at 800 °C (bottom).
F(R∞) represents the Kubelka-Munk function.
UV-VIS spectra of pure oxides prepared with the same procedure as applied for the dopants were
also measured (data not shown for brevity). La2O3, Y2O3, ZrO2, and Sm2O3 hardly absorbed any
photons in the region of blacklight lamp emission (340 – 400 nm). CeO2, on the contrary, is a slight
yellowish powder with an absorption edge extended to 430 nm.
DRIFT characterization
The IR spectra of selected samples in the region of 4000-2800 cm-1, where O-H stretching modes
are expected, are shown in Figure 5. It should be noted that the measurements have been done at 120
°C. Upon heating fresh samples from Room Temperature up to 120 °C in He, the amount of adsorbed
water on the surface of the catalysts decreases, allowing a better evaluation of the nature of the various
Rare-earth doped P25-TiO2 in methylene blue degradation
87
OH-groups. The absorption bands for O-H stretching modes representing Rutile are located at 3650
cm-1 and at 3415 cm-1, respectively [23,29-32]. For the band at 3415 cm-1 an assignment has been
proposed to water molecules strongly adsorbed to TiO2 via interactions with coordinatively unsatured
Ti4+
surface cations [31]. As expected on the basis of the XRD data, these bands contribute
significantly to the spectra of the samples calcined at 800 °C. Without calcination at 800 °C, a different
OH spectral signature is obtained, characterized by a series of components in the 3800-3600 cm-1
range [23,29-32], with Rutile induced vibrational modes overlapping with O-H stretching modes of
Anatase-OH. The complicated spectral signature is the result of OH being present on different defect
sites, as well as the result of contributions of hydrated sites [23]. The absorption at 3677 cm-1
is
assigned to an isolated Anatase vibration [30-32]. For the band at 3640 cm-1 we follow the assignment
of Surca Vuk et al. [29], to Anatase bridging (Ti)2-OH [29].
Figure 5. DRIFT spectra for different RE-doped and non-doped titanium dioxide photocatalysts.
Spectra were recorded at 120 °C in 30 ml/min He.
To semi-quantify the intensity of each OH-group, peak areas were calculated after deconvolution
with the program PeakFit v4.12. The deconvoluted spectra for a selected sample is shown in Figure 6.
Please note that the x-axis has been inversed as compared to Figure 5, as a result of the Peak-Fit
procedure. From the deconvoluted spectra, areas were determined for the contributions of Rutile
associated OH-groups (3650 cm-1
, 3415 cm-1
), and the contribution of the Anatase associated
OH-groups (3677, 3640 cm-1
). While other contributions, centered at 3610, 3520, and 3350 cm-1
, were
taken into account to obtain the best fit, these were considered to be related to remaining adsorbed
water [31,32]. Unfortunately it was not possible to completely remove these water bands in the
conditions of the measurements (Flowing He, 120 °C). Consequent perturbations of the vibrational
patterns and intensity of the hydroxyl groups, depending significantly on the relative location of
adsorbed water molecules and type of hydroxyl group, do not allow a full quantitative analysis. Still,
the calculated areas for selected samples are compiled in Figure 7. Comparing P25_600 with the 0.2
Abso
rban
ce
3000 3200 3400 3600 3800
Wavenumbers (cm-1)
0.1
34153650
3677 3640
P25_2La_800
P25_2Ce_800
P25_0.2Sm_800
P25_0.2Ce_800
P25_0.2La_800
P25_0.2Pr_800
P25_800
P25_600
P25_0.2Sm_600
Chapter 4
88
wt-% Sm-doped analogue, shows that the OH-group population is hardly affected by the presence of
Sm, in agreement with the data shown in Table 1 (XRD and surface area).
Comparing P25_600 and P25_800, clearly the high temperature treatment has almost completely
decomposed the surface Anatase-OH, while the Rutile-OH groups have slightly increased in intensity.
Again the stabilizing effect of the RE on the anatase phase is evident from the relatively large amount
of Anatase-OH still present on the doped-P-25 surface (0.2 wt-%) after calcination at 800 °C. While
the absolute values of the intensities should be considered semi-quantitative [20], the amount of
surface hydroxyl-groups seems to be best preserved after pretreatment at 800 °C by modification with
La, relative to other RE-dopants. Applying 2 wt-% doping, instead of 0.2 wt-%, decreases the amount
of Anatase-OH-groups, most likely a result of more extensive surface converage with RE and the 20%
reduction in surface area (compare Table 1).
Figure 6. Example of spectral deconvolution for the La-promoted sample pretreated at 800 °C, as
achieved by the Peak-Fit program.
Figure 7. The dimensionless intensities of various OH-vibrations for the series of RE-doped TiO2
samples, as determined after spectral deconvolution.
3677
3650
3640
3677
3650
3640
Comparison Original and calculated spectra
Rare-earth doped P25-TiO2 in methylene blue degradation
89
The Rutile-OH intensities (3415, 3650 cm-1
) of the doped-P-25 surface (0.2 wt-%, after
calcination at 800 °C) are slightly increased compared to P25-600, while the relative ratio of the two
Rutile-OH groups is quite independent on the nature of doping. This increase in Rutile-OH intensity of
the doped samples is in agreement with the higher Rutile content of these samples as compared to
P25-600. Similar to the trend in Anatase-OH intensity, applying 2 wt-% doping, instead of 0.2 wt-%,
decreases the amount of Rutile-OH-groups, again explained by surface converage of P-25 with RE.
Generally one can conclude for the 0.2 wt-% samples (800 °C) that, relative to P25-600,
qualitatively a decrease in Anatase-OH intensity results in an increase in the Rutile-OH intensity, in
good agreement with the XRD data on the Anatase to Rutile fractions of the respective catalysts (Table
1).
Photocatalytic activity
The photocatalytic decoloration of MB over all prepared samples was evaluated, and Figure 8
shows examples for selected samples of dark adsorption (in the Figure 5 time below 0 min, before the
light was switched on) and the decay curve of MB concentration under UV-irradiation (in the Figure 8
time above 0 min). The lines in Figure 8 represent the fitted curves of a first order kinetic model in MB
degradation by light exposure:
appk t
MB MB,0C C e− ⋅
= ⋅ (4)
in which CMB and CMB,O are the MB concentrations at time (t), and (t=0), respectively, and kapp the
apparent first order rate constant. In all experiments first order kinetics was observed.
Figure 8. MB photocatalytic degradation curves for two photocatalytic systems with variable RE
content. (a) P25_%La_800, and (b) P25_%Ce_800.
In separate sets of experiments it was found that the dark adsorption followed a Langmuir-type
Time [min]
-100 0 100 200 300 400 500
C [
mm
ol/l]
0.00
0.01
0.02
0.03
P25_800
P25_0.2La_800
P25_1.0La_800
P25_2.0La_800
Time [min]
-100 0 100 200 300 400 500
C [
mm
ol/l]
0.00
0.01
0.02
0.03
P25_800
P25_0.2Ce_800
P25_1.0Ce_800
P25_2.0Ce_800
Chapter 4
90
adsorption isotherm and the equilibrium was established within 2 hours depending on the morphology
of the photocatalyst of which the adorption is in agreement with previously reported by Houas et
al.[38], and it was deduced that the typical adsorbed amount of MB is 1×10-6
mol/l, which corresponds
to 2×10-6
mol MB per gram of photocatalyst. As it is observed in the Figure 8, P25 photocatalytic
activity was affected by RE, this effect depending on the nature of RE, the RE loading and the
preparation temperature. For instance, profiles in Figure 8a showed that 0.2 and 1 %-La loading
improved P25_800 photocatalytic activity while 2 %-La loading decreased the activity. Figure 8b
shows that P25_1.0Ce_800 presented the best activity among the 800 °C-calcined samples.
Experiments performed with La2O3, CeO2, Y2O3, ZrO2, PrO2, and Sm2O3 showed that these oxides
hardly degraded MB under our experimental conditions and, therefore, differences in activity between
non-doped and doped samples should be not attributed to the additional contribution of the oxides, but
to the modification of the P25 properties itself.
P25_Tem
p
P25_0.2
La_Temp
P25_0.2
Ce_Tem
p
P25_0.2
Zr_Tem
p
P25_0.2
Y_Temp
P25_0.2
Pr_
Temp
P25_0.2
Sm
_Temp
kap
p [
1/m
in]
0.00
0.02
0.04
600°C
800°C
Figure 9. Apparent first order rate constants as a function of RE amount and calcination
temperature.
In Figure 9, the kinetic rate constants of the 0.2% RE-doped photocatalysts calcined at 600 and
800 ºC, and the constants corresponding to P25_600 and P25_800 are compiled. The errors as
determined from the 95 % confidence interval of the apparent first order rate constant (Figure 5) were
lower than 10 %. Comparing the apparent first order rate constants of P25_600 and P25_800, thermal
treatment has a significant deteriorating effect on the photocatalytic activity. In case of the 600
°C-calcined samples, Figure 6 indicates that all the RE-doped samples have lower photocatalytic
activity than pure P25_600. On the contrary, in case the photocatalysts were calcined at 800 °C, the
photocatalytic activity of the RE-doped samples depends on the nature and loading of the RE. La, Y, Pr,
and Sm yield higher first-order rate constants in MB degradation than undoped P25_800, whereas for
Ce the positive effect is less dramatic. Without presenting all the details, increasing the loading of RE
above 0.2 wt-% generally does not improve the performance of P25 after calcination at 800 °C.
Rare-earth doped P25-TiO2 in methylene blue degradation
91
0.000
0.005
0.010
0.015
0.020
0.0 0.5 1.0 1.5 2.0 2.5
RE content [%]
kap
p [
1/m
in]
P25_xLa_800
P25_xCe_800
P25_xPr_800
P25_xSm_800
P25_xY_800
P25_xZr_800
Figure 10. Apparent first order rate constant for photocatalysts prepared at 800 °C and with
different RE contents.
In Figure 10, the effect of RE loading on the activity of samples calcined at 800 °C is shown. The
optimum value for La, Y, or Pr doping was around 0.2% while for Ce and Sm, 1% doping exhibited
the highest activity. Zr-loaded P25 was always less active than pure P25_800, regardless of the amount
of Zr.
4.4 Discussion
As was stated in the introduction the aim of the present study was to contribute to the evaluation
of the effect of Lanthanides and high temperature treatment on the photocatalytic activity of TiO2, by
monitoring changes in phase composition, surface area, and surface hydroxyl-group composition of
P-25. In the following the effect of these parameters on the photocatalytic activity will be discussed.
Activity of Rare Earth oxides
Pure rare earth oxides have rarely any activity as is measured (results not shown) in photocatalytic
methylene-blue decomposition, in agreement with the reported low activity in Salicylic Acid
decomposition reported by Ranjit et al. [17,18]. This is in agreement with the absence of absorption
bands at the wavelength of the light emitted (370 nm) by the ‘black-light’ sources, used to stimulate
methylene blue decomposition. Hence, activity differences are most likely the result of structural
changes of TiO2, induced by the applied thermal treatments in the presence or absence of the dopants,
Chapter 4
92
rather than reaction of methylene blue over RE-oxides.
Phase composition and surface area
When pure P25 is heated to 800 °C, the content of rutile, which is the thermodynamically favored
phase, increases from 30 % to 95 %, as was deduced both from XRD, Raman, and DRIFT
characterization. As a consequence of the anatase to rutile transformation, the BET areas of the
photocatalysts decreased from about 50 m2/g for P25 and 600 ºC-calcined samples to 16 m
2/g for
P25_800. Although the presence of RE may partially inhibit the anatase to rutile transformation, rutile
formation was inevitable. The decrease of the BET surface area when P25 was calcined at 800 °C
(with or without RE) was a consequence of the larger rutile particle size in comparison to anatase, as it
is observed in Figure 11 in which a linear relationship between rutile fraction and BET area of the
different photocatalysts is illustrated.
0
10
20
30
40
50
0 0.2 0.4 0.6 0.8 1
Rutile [fraction]
BE
T [
m2/g
]
P25_T
P25_0.2RE_600
P25_0.2RE_800
P25_1.0RE_800
P25_2.0RE_600
Figure 11. Relationship between rutile fraction and BET area of the different photocatalysts.
The anatase to rutile phase transformation is generally considered to be a nucleation and growth
process during which rutile nuclei form within the anatase phase [33,34]. The stabilization of doped
anatase phase [22,36,37], has been attributed to the formation of surrounding metal oxides on the TiO2
particles. At the interface, Ti atoms can substitute the RE element in the lattice of RE oxide films, to
create tetrahedral Ti sites. The interaction between the tetrahedral Ti species and the octahedral Ti sites
in the anatase is thought to prevent the phase transformation to rutile [35]. The formation of solid
solutions into the TiO2 lattice bulk seems not to be possible due to the differences in the radii of the
different cations potentially present in our samples. The ionic radii (in nm) of these cations are:
Ti4+ (6.9) < Zr4+ (8.7) < Pr4+ (9.2) < Ce4+ (9.4) < Sm3+ (10.0) < Pr3+ (10.6) = Y3+ (10.6) < Ce3+
(10.7) < Sm2+
(11.1) < La3+
(12.2)
Rare-earth doped P25-TiO2 in methylene blue degradation
93
The ionic radius of Ti4+ is much smaller than the ionic radii of the RE and, therefore, it is difficult
for these cations to enter into the TiO2 lattice.
Figure 12. Correlation between the apparent first order rate constants of the various
photocatalysts, and the corresponding Anatase fraction. Legend as indicated in the Figure. Not show
as the reference, P25_800 has similar activity of P25_0.2Ce_800 (Fig 9). Clearly the correlation is
poor.
Clearly the Rutile phase has a significantly lower activity than the mixed Rutile/Anatase
composition of P-25_600. The data in Table 1 show that the phase transformation is largely prevented
by doping, as is discussed in previous paragraph. An attempt to correlate the Anatase fraction of 4
selected photocatalysts, containing 0.2 wt.% Ce, Pr, La, and Sm, with the photocatalytic activity is
shown in Figure 12. Clearly there is no direct correlation. In agreement with the statements of Ranjit et
al, significant differences in photocatalytic activity can apparently not be attributed to the differences
in phase composition alone [17,18]. A treatment at 800 °C might produce novel dopant/TiO2
interactions that a treatment of the doped materials limited at 600 °C does not induce.
Photoluminescence studies might show if there is an affect of the rare earth ions on physical properties
of TiO2 after treatment at 800 °C. It is, however, not straight forward to correlate luminescence
properties to photocatalytic activity, in view of the accurate energy balance that is needed: heat could
just as well be generated by the recombination of electrons and holes as luminescence. Still, we
conclude that the physical properties have a less pronounced affect on performance than the surface
OH-group population, as will be discussed in the following.
The linear correlation in Figure 8 clearly shows that the phase composition and the surface area
are coupled. This correlation indicates that the different statements on the effect of the anatase to rutile
ratio on photocatalytic activity made in the literature, at least for P-25 modified by heat treatment,
cannot be discussed independently from an effect of the available surface area. Since we did not find a
good correlation between the Anatase fraction and the first order kinetic rate constant of methylene
blue decomposition, as expected we did not obtain a good correlation between the first order kinetic
0.000
0.004
0.008
0.012
0.016
0.020
0 0.1 0.2 0.3 0.4
P25_0.2Ce_800
P25_0.2La_800
P25_0.2Pr_800
P25_0.2Sm_800
Chapter 4
94
rate constant and the BET area. Another factor must play an important role in determining catalytic
activity [17,18].
Surface hydroxyl groups
Photo stimulation of TiO2 generates electrons and holes, and Ti4+-OH entities on the surface trap
the hole by formation of surface hydroxyl radicals (Ti4+-OH·) [36]. It has been proposed that the first
step in the oxidation of organic compounds is the reaction of these (surface) OH· radicals with the
organic molecule. The deconvoluted DRIFT data allow us to distinguish between the various surface
OH-groups. Plotting the first order kinetic rate constant of MB decomposition against the sum of
intensities of all OH-groups, or the Rutile-OH groups, does not give a good correlation. On the
contrary, the amount of Anatase hydroxyl-groups show a good correlation with the degradation rate of
MB, and in particular the quantity of the bridging (Ti)2-OH (3635 cm-1
) as illustrated in Figure 13. It is
to be assumed that this OH-group has the highest efficiency in trapping the photo-generated holes by
formation of surface hydroxyl radicals (Ti4+-OH·) [38], and/or the highest affinity for methylene blue
adsorption [17,18] in reaction conditions. It should be noticed, however, that the trendline in Figure 13
does not go through the origin, suggesting that other (hydrated) TiO2 sites also contribute to
photocatalytic activity.
Figure 13. Correlation between the apparent first order rate constants of the various
photocatalysts, and the corresponding dimensionless anatase related (Ti)2-OH intensity obtained after
spectral deconvolution. Legend as indicated in the Figure. The correlation is rather good.
The photocatalytic tests were carried out on TiO2 materials suspended in an aqueous medium. The
(Ti)2-OH (3635 cm-1) site observed in the DRIFT spectra of the partially dehydrated systems should
therefore be considered as a precursor for the actual site during the reaction, which is largely altered by
the extensive water population on the catalyst surfaces in aqueous conditions. It is to be assumed that
this site has the highest efficiency in trapping the photo-generated holes by formation of surface
hydroxyl radicals (Ti4+
-OH·) [38]. Unfortunately, the exact nature of the active site will be extremely
0
0.004
0.008
0.012
0.016
0.02
0 2 4 6
(Ti)2OH Amount [cm-1]
Rare-earth doped P25-TiO2 in methylene blue degradation
95
difficult to assess by IR spectroscopy, even if the ATR technique is applied, in view of the large and
broad spectral contribution of water, overlapping the OH-vibrations.
The questions remaining are i) why the different RE dopants affect the remaining Anatase
fraction and (partially dehydrated) hydroxyl group intensity differently (by calcination at 800 ºC), and
ii) why the hydroxyl group intensity does not change linearly with a change in Anatase fraction
(compare Table 1 and Fig. 13). The answer to both questions is related to the temperature and rate at
which the RE nitrate precursor decomposes, which determines to what extent the RE-oxide becomes
dispersed over the TiO2 surface. Clearly, the dispersion of the RE-oxide will in turn determine the
extent of the Anatase to Rutile conversion at 800 °C, as well as the amount of remaining surface
exposed Ti-OH groups, not necessarily to the same degree. Changing the catalyst preparation
procedure (a.o. ramp rate, flow vs static conditions) might thus dramatically affect the outcome of the
results presented in this study.
Summarizing, rather than a beneficial effect by retarding electron-hole recombination, or
RE-assisted adsorption of MB, which should have led to better performance of our RE-doped P25
samples pretreated at 600 °C, it is more likely that the positive effect of the addition of RE after
treatment at 800 °C is the result of a stabilizing effect on the amount of remaining Anatase
(Ti)2-OH-groups. This is affected by the extent of dispersion of the RE-oxide. As was recently shown
by Ryu and Choi, other properties of TiO2 might be more important in controlling reactivity towards
other substrates [39]. In this respect it is difficult to evaluate our data in relation to literature data on
the effect of La-addition, since a broad range of substrates has been used [13-18]. Further
investigations using ATR-FTIR spectroscopy to reveal the dynamics of the hydroxyl groups in
operando conditions are ongoing in the group of Industrial Catalysis.
4.5 Conclusions
The following conclusions can be derived from the work described in this chapter:
Doping of P25 with rare earth oxides such as La, Ce, Y, Pr, and Sm prevents the anatase to
rutile phase transformation upon calcination at 800 ºC, positively affecting the remaining BET surface
area. A linear correlation was found between the BET surface area and the anatase/rutile ratio in P25.
The photocatalytic degradation of methylene blue over rare earth oxide modified TiO2 follows
first order kinetics, and is mainly dependent on the quantity of the specific bridging anatase (Ti)2-OH
group in the applied P-25 series. This quantity is a function of the BET surface area (and hence
anatase/rutile ratio), and the quantity and extent of dispersion of the rare earth oxide.
Phase transition from anatase to rutile is largely prevented by doping. Varied photocatalytic
activity of doped TiO2 cannot be attributed to the differences in the phase composition alone.
A linear corrleation between the number of a specific Ti-OH group and the methylene blue
degradation rate suggesting that anatase OH-groups are important reactive sites for methylene blue
adsorption and degradation, and a dominant factor in controlling performance.
Chapter 4
96
References
1. Carp, O., Huisman, C.L., Reller, A., Prog. Solid State Chem., 2004, 32, 33
2. Legrini, O., Oliveros, E., Braun, A.M., Chem. Rev., 1993, 93, 671
3. Mills, A., Le Hunte, S., J. Photochem. Photobiol. A, 1997, 108, 1
4. Devahasdin, S., Fan, C., Li, K., Chen, D.H., J. Photochem. Photobiol. A, 2003, 156, 161
5. Diebold, U., Surf. Sci. Reports, 2003, 48, 53
6. Tanaka, K., Hisanada, T., Rivera, A.P., in Photocatalytic Purification and Treatment of Water
and Air, Elsevier, Amsterdam, 1993, 169
7. Agrios, A.G., Gray, K.A., Weitz, E., Langmuir, 2003, 19, 1402
8. Cheng, S., Tsai, S.J., Lee, Y.F., Catal. Today, 1995, 26, 87
9. Deng, X.Y., Yuw, Y.H., Gao, Z., Appl. Catal. B, 2002, 39, 135
10. Lee, S.K., Robertson, P.K.J., Mills, A., McStay, D., Elliott, N., McPhail, D., Appl. Catal. B,
2003, 44, 173
11. Watson, S.S., Beydoun, D., Scott, J.A., Amal, R., Chem. Eng. J., 2003, 95, 213
12. Cerrato, G., Marchese, L., Morterra, C., Appl. Surf. Sci., 1993, 70/71, 200
13. Gao, L., Liu, H., Sun, J., Mat. Science Forum, 2005, 486-487, 53
14. Xu, A.W., Gao, Y., Liu, H.Q., J. Catal., 2002, 207, 151
15. Li, F.B., Li, X.Z., Hou, M.F., Appl. Catal. B, 2004, 48, 185
16. Zhang, Y., Xu, H., Xu, Y., Zhang, H., Wang, Y., J. Photochem. Photobiol. A., 2005, 170, 279
17. Ranjit, K.T., Willner, I., Bossmann, S.H., Braun, A.M., J. Catal., 2001, 204, 305
18. Ranjit, K.T., Willner, I., Bossmann, S.H., Braun, A.M., Environ. Sci. Technol., 2001, 35, 1544
19. Su, C., Hong, B.Y., Tseng, C.M., Catal. Today, 2004, 96 ,119
20. Sirita, J., Phanichphant, S., Meunier, F.C., Anal. Chem., 2007, 79, 3912
21. Du, P., Shibata, H., Mul, G., Moulijn, J.A., in Proceedings of the TiO2-9/AOT-10 Conference,
San Diego, USA; D.Ollis, and H. Al-Ekabi (Eds.), Univ. West Ontario Press, 2005
22. Datye, A.K., Riegel, G., Bolton, J.R., Huang, M., Prairie, M.R., J. Solid State Chem., 1995, 115,
236
23. Martra, G., Appl. Catal. A, 2000, 200, 275
24. Colon, G., Hidalgo, M.C., Navio, J.A., Appl. Catal. A., 2002, 185
25. Yang, J., Ferreira, J.M.F., Mater. Res. Bull., 1998, 33, 389
26. Choi, H.C., Jung, Y.M., Kim, S.B., Vibrational Spectroscopy, 2005, 37, 33
27. Dhage, S.R., Choube, V.D., Samuel, V., Ravi, V., Mater. Lett., 2004, 58, 2310
28. Inagaki, M., Imai, T., Yoshikawa, T., Tryba, B., Appl. Catal. B, 2004, 51, 247
29. Surca Vuk, A., Jese, R., Gaberscek, M., Orel, B., Drazic, G., Sol. Energy Mater. Sol. Cells,
2006, 90, 452
30. Davydov, A., Molecular Spectroscopy of Oxide Catalyst Surfaces, Wiley, New York, 2003
31. Morterra, C., J. Chem. Soc. Faraday Trans., 1988, 84, 1617
32. Mul, G., Zwijnenburg, A., Van der Linden, B., Makkee, M., Moulijn, J.A., J. Catal., 2001, 201,
128
33. Hishita, S., Mutoh, I., Koumoto, K., Yanagida, H., Ceram. Int., 1983, 9, 61
Rare-earth doped P25-TiO2 in methylene blue degradation
97
34. Nair, J., Nair, P., Mizukami, F., Oosawa, Y., Okubo, T., Mater. Res. Bull., 1999, 34, 1275
35. Anderson, C., Bard, A.J., J. Phys. Chem. B, 1997, 101, 2611
36. Zhang, Y., Zhang, H., Xu, Y., Wang, Y., J. Solid State Chem., 2004, 177, 3490
37. Lin, J., Yu, J.C., J. Photochem. Photobiol. A, 1998, 116, 63
38. Houas, A., Lachheb, H., Ksibi, M., Elaloui, E., Guillard, C., Herrmann, J.M., Appl. Catal. B,
2001, 31, 145
39. Ryu, J., Choi, W.Y., Environ. Sci. Technol., 2008, 42, 294
Chapter 4
98
99
5
Effect of Irradiation Energy and TiO2 Structure on the Rate of
Photo-oxidation of Cyclohexane and Side Product Formation
Abstract
The liquid phase photolytic oxidation of cyclohexane was studied and compared with
photocatalytic oxidation over TiO2 with varying wavelengths of light exposure, slurry densities, and
sources and pretreatments of catalyst material. Photolytic oxidation at λ < 275 nm (i.e., in the absence
of catalyst) yielded a high selectivity to cyclohexanol (>85%). By adding a TiO2 catalyst to
cyclohexane exposed at λ < 275 nm, the selectivity shifted to the ketone, with the amount of catalyst
added determining the obtained cyclohexanone:cyclohexanol ratio. When a combination of a TiO2
catalyst and a Pyrex reactor was used (the latter preventing photolytic formation of cyclohexanol), an
almost complete selectivity to cyclohexanone was obtained (>95%). The activity toward ketone
formation was affected by catalyst structure, with surface hydroxyl group density being the most
important parameter. Based on the observed correlation between the hydroxyl group density and
activity, as well as the observed negative effect of cyclohexanol addition on cyclohexanone production
rate, a preliminary reaction mechanism is proposed involving the light-induced formation of surface
cyclohexyl radicals, followed by formation of a peroxide intermediate and decomposition and
desorption to cyclohexanone. Accumulation of cyclohexanol on the TiO2 surface is proposed to
deteriorate the photocatalytic activity and to contribute to CO2 formation.
Chapter5
100
5.1 Introduction
Heterogeneous photocatalysis has been a subject of various studies since the discovery of
photochemical water splitting on TiO2 electrodes in 1972 [1]. Many researchers have focused on
environmental abatement, such as air cleaning and water purification, in which organic pollutants are
totally degradated to carbon dioxide and water over TiO2 photo-catalysts [2-4]. In contrast, relatively
few studies were conducted on the application of photocatalysis for product synthesis by selective
oxidation. The high oxidation potential of TiO2 and non-selective nature of radical reactions could, at
least partially, explain this apparent ignorance. Nevertheless, photocatalytic selectivity for the desired
product in the oxidation of hydrocarbons is attracting attention in recent years. These studies
demonstrated that different product distributions could be obtained in TiO2 photo-oxidation as
compared with conventional oxidation processes [5-7]. Another approach to improve the selectivity is
based on the stabilization of charge-transfer complex in a confined environment. Pioneering work has
been performed by the group of Frei and the group of Larsen, who revealed high selectivity of ethane,
propane, isobutane and cyclohexane oxidation with molecular oxygen to corresponding oxygenates
under photochemical conditions in cation exchanged zeolites [8-11].
Liquid phase oxidation of cyclohexane is an important commercial reaction in the conversion of
cyclohexane via cyclohexanone in caprolactam, a monomer for nylon-6 production. A low
cyclohexanol:cyclohexanone ratio is prefered by caprolactam producers, because the consecutive step
of cyclohexanol dehydration to cyclohexanone is a costly and energy-consuming process proceeding at
elevated temperature of 400-450°C. Typical cobalt-catalyzed air oxidation gives an alcohol-ketone
ratio of 2.5-4:1. Because the reaction intermediate and products are more readily oxidized than
cyclohexane, the conversion must be kept low (usually under 10%) in order to maximize yield.
Several previous studies were performed applying photon energy to oxidize liquid cyclohexane
selectively, as is listed in table 1. A first indication of selective photo-oxidation of cyclohexane by
semiconductor oxides was presented in a brief report of Giannotti et al. [12], in which the
photocatalytic activities of anatase as well as rutile phases were mentioned. Although the conversion is
less than 0.1% for 3 hrs of reaction, high ketone selectivities were reported of 100% (no CO2 and
cyclohexanol) for anatase and 90% (only 10% CO2) for rutile. The type of reactor was not discussed,
however. In the late 1980s, Mu et al. [13] performed a comprehensive study on this reaction using
Degussa P25 TiO2, which consists of approximately 70% of anatase and 30% of rutile. Again, high
selectivities to cyclohexanone were observed, namely 83% selectivity to the ketone, 5% to the alcohol,
and 12% to CO2. This study was extended to other hydrocarbons and further evaluated by Hermann et
al. in 1991 [14]. Similar product selectivities, with ketone being the major product, were reported in
more recent studies of Lu et al. [15], Boarini et al. [18] and Almquist and Biswas [20]. Lu et al. [15]
compared the performance of TS-1 and TiO2, and Boarini et al. [18] and Almquist and Biswas [20]
focused mainly on the effect of different solvents on catalyst activity and selectivity in cyclohexane
photooxidation over TiO2. Generally the high ketone selectivity is explained by strong adsorption of
cyclohexanol on titania and high reactivity of cyclohexanol versus cyclohexane, which undergoes
further oxidation towards cyclohexanone or CO2 as the final product [13-15,18,20]. According to
Almquist and Biswas [20], ketone was mainly formed through alkylperoxy radicals, a parallel route of
the cyclohexanol oxidation. Addition of cyclohexanol inhibited the formation of cyclohexanone,
blocking selective oxidation sites and undergoing deep oxidation.
Photo-oxidation of cyclohexane
101
Table 1. Summary of open resources on photocatalytic oxidation of cyclohexane over titania
Reference Catalyst Reactor/
filter
Illumination
source C6H12 [ml] Solvent
Time
[h]
Conversion
[%]
Selectivity
C6H12OH [%]
Selectivity
C6H10O [%]
Selectivity
1/6 CO2 [%]
[12] TiO2(anatase) / 0.1g - 1000W Hg-Xe 3 - 20 0.09 0 100 0
[12] TiO2(rutile) / 0.1g - 1000W Hg-Xe 3 - 20 0.095 0 91.1 8.9
[13,14] TiO2 (P25) λ>300 nm 125W Hg 10 3 0.3 5 83 12
[15] TiO2(anatase) / 30mg Quartz 250W Hg 10 - 3 0.05 5.5 92.2 2.3
[16] Ultrafine TiO2 /
2.5 mmol/l ‡
Pyrex 2000W Xe 20 - 18 0.012 37.3 32.7 -
[17] Ultrafine TiO2 / 30mg Quartz 250W HP Hg 10 CH3CN/10ml +
HNO3/1mol 8 0.096 85.3 14.7 -
[18] TiO2 (P25) / 4 g/dm3
400W MP Hg
(λ>360nm) - - 1.5 0.078 0 99.1 0.9
[19] TiO2 (P25) / 1g Quartz 5.5W LP Hg 2 H2O/13 ml +
30%H2O2/3 ml 2 4.24 * 30.08 44.03 0
[20] TiO2 (P25) / 20 mg Pyrex 450W Xe 20 - 0.75 0.035 19 82 -
[6] TiO2 (P25) / 20 mg Pyrex 500W Xe 10 H2O / 10ml 12 0.094 0 63.5 36.5
[21,22] Nanosized TiO2 /
30 mg Quartz 250W Hg 10 CH3CN / 10ml 3 0.097 84.5 14.9 0.6
[23] TiO2 / 100 mg Pyrex 500W HP Hg 30 - 24 0.76 1 66 33
* No oxygenates were found in the absence of hydrogen peroxide.
‡ 2.5 mmol/l of titanium(IV) tetrabutoxide solution.
Chapter5
102
Contradictory results also exist in the open literature. Su et al. [17] and Li et al. [21,22] obtained
mainly cyclohexanol by photoactivation of cyclohexane with molecular oxygen, applying nanosized
TiO2 particles. Acetonitrile was used as solvent instead of pure cyclohexane. Solvent effects cannot
explain the result satisfactorily, as is discussed in the study of Mu et al. [13]. They investigated the
influence of the concentration of cyclohexane in acetonitrile and detected cyclohexanone as the main
product. Li et al. [22] discussed the discrepant performance compared with previous studies on the
basis of a quantum size effect and a different surface structure of the used nanosized TiO2, which
supposedly prevents consecutive reaction of cyclohexanol to cyclohexanone [21,22]. From these
previous studies, we can conclude that the product distributions reported for photooxidation of neat
cyclohexane with TiO2 catalysts are far from consistent and need further evaluation.
It is remarkable that none of the aforementioned studies systematically compared the photolytic
(i.e., without catalyst) and photocatalytic oxidation products in terms of photon efficiency and
selectivity. The effect of the applied wavelength on catalyst performance also was typically not
addressed [24,25]. Our goal in this research is to reveal the origin of the discrepancy between previous
studies and to provide a first step towards the optimization of operation conditions for industrial
application. The present study investigates neat cyclohexane photooxidation, varying the reactor setup
(quartz, Vycor, or Pyrex immersion wells), affecting the wavelengths available for reaction, as well as
the amount and constitution of the applied titania (Degussa P25 and Hombikat pretreated at various
calcinations temperatures). We show that photolysis and photocatalysis lead to very different product
distributions, and that the surface hydroxyl group density on TiO2 is an important factor in controlling
the reaction rate.
5.2 Experimental
Applied catalyst materials and catalyst characterization
Degussa P25 titanium dioxide and Hombikat UV100 titania (Sachtleben) were used as
photocatalysts. Characteristics of the Hombikat TiO2 (100% anatase as determined by XRD), include a
SBET of 337 m2/g and primary particle size of ~5 nm (determined using Scherrer’s equation), with a
mean agglomerate size in cyclohexane after ultrasonication of ±3 µm (as determined by forward
light-scattering). The Hombikat was further pretreated at various temperatures in the range of
400–1273 K in a static oven in air for 12 h, typically at a heating rate of 10 K/min. The various
catalyst samples were analysed by various techniques, including UV/vis, XRD, pore texture analysis,
and ammonia TPD, to allow evaluation of the structure of the applied TiO2 and the resulting
performance in cyclohexane oxidation.
Cyclohexane UV absorption spectra were measured on a Cary-5 UV–vis spectrometer using a
1-cm quartz transmission cell. The spectra of the solids were recorded at ambient temperature in
diffuse reflectance mode, using BaSO4 as a reference. Samples were ground, heated overnight at
180◦C, and scanned from 190 to 800 nm. Powder X-ray diffraction (XRD) patterns were measured on
a Philips PW 1840 diffractometer equipped with a graphite monochromator using Cu-Kα radiation (λ
= 0.1541 nm). Nitrogen adsorption and desorption isotherms were recorded on a QuantaChrome
Autosorb-6B at 77 K. Samples were previously evacuated at 623 K for 16 h (at a ramp rate of 10
Photo-oxidation of cyclohexane
103
K/min). The BJH model was used to calculate the pore size distribution from the adsorption branch,
and the BET method was used to calculate the surface area (SBET) of the samples.
Temperature-programmed desorption of ammonia (NH3-TPD) was carried out on a Micrometrics
TPR/TPD 2900 apparatus equipped with a thermal conductivity detector (TCD). Approximately 25 mg
of TiO2 was flushed with helium at 773 K for 1 h (at a heating rate of 10 K/min), except for the sample
activated at 398 K, which was pretreated at this temperature in the ammonia TPD setup. After
pretreatment, the sample was rapidly cooled to 373 K and loaded with ammonia, applying a flow of 30
ml/min for about 60 min, after which a helium flow of 30 ml/min was applied to remove weakly
adsorbed NH3. A linear temperature program was started (373–873 K at 10 K/min), and the desorbed
amount of ammonia was analyzed by the TCD. The TPD spectra were used to determine the nature
and amount of hydroxyl groups present on each catalytic material.
A second procedure to determine OH group density on the surface of the applied TiO2 was based
on the Fe(acac)3 method described by Van Veen et al. [26]. Typically, 0.005 g of catalyst was added to
10 ml of 0.25 mmol/l Fe(acac)3 solution in toluene and stirred in the dark overnight. After this, the
solid was centrifuged off, and the supernatant solution was subjected to UV absorption measurements.
The amount of adsorbed Fe(acac)3 was determined by comparing the UV absorption at 355 nm with
calibrated samples.
Reactants and solvents
Cyclohexane, cyclohexanol, and cyclohexanone were purchased from Merck. Cyclohexyl
hexanoate was purchased from Alfa Aesar, and 1,1’-oxybis(cyclohexane) was synthesized and purified
following the procedure of Olah et al. [27]. Anhydrous hexadecane, used as the internal standard for
gas chromatography, was purchased from Aldrich. All commercial chemicals were of analytically pure
grade and were dried on silica gel before the experiments.
Photo-activity measurements
To evaluate the effect of wavelength on the selectivity of the reaction, reactions were carried out
in a 1000-ml semibatch slurry-type photochemical reactor with an immersion well (ACE Glass)
located in a dark fume hood. The reactor vessel was covered with aluminium foil to prevent the
influence of stray light. Illumination was provided by a 450 W medium-pressure mercury-vapor lamp
with 39% of total radiated energy in the UV spectrum, also supplied by ACE Glass. During operation,
distilled water was circulated through the immersion well for cooling purposes. The temperature inside
the reaction vessel was regulated at 333 K through a circulating bath. A Pyrex, Vycor, or quartz
cooling jacket was used, with the choice affecting the wavelengths that were available to illuminate
the reaction mixture. UV transmission of the applied reactor materials (Pyrex, Vycor, and quartz) was
measured using a calibrated Avantes spectrophotometer S-2000 with a UV/vis cosine collector.
In a typical experiment, 600 ml of cyclohexane, along with 1 g of hexadecane as the internal
standard, was mechanically stirred together with a desired amount of catalyst, typically 1 g/l. Air was
bubbled through the liquid at a rate of 300 ml/min through a gas sparger. Evaporative losses of
organics were minimized by applying a reflux condenser. A carbon dioxide trap with saturated barium
hydroxide solution was installed to determine the amount of carbon dioxide produced in the form of
precipitated barium carbonate.
Chapter5
104
GC samples were taken from gas and organic phases separately, applying the appropriate syringes.
Organic compounds were identified by GC-MS (Chromopack, CP Sil-5) and quantatively analyzed
twice using a gas chromatograph with a flame ionization detector (Chromopack, CPwax52CB).
Quantification of the oxygenated products in the liquid phase was derived from a multipoint
calibration against the internal standard. The following products were thus analyzed quantitatively:
cyclohexanol, cyclohexanone, 1,1’-oxybis-cyclohexane, and cyclohexyl hexanoate. CO2 was analyzed
by a gas chromatograph equipped with a TCD, using a Poraplot column. Comparable results were
obtained with the BaCO3 precipitation method and the GC analysis. GC quantification is preferred,
because it results in more data points and is less labor-intensive.
Cyclohexyl hydroperoxide (CHHP) was detected indirectly according to the procedure of
Shul’pin et al. [27]. It was assumed that CHHP was totally converted to the corresponding alcohols
and ketones by the addition of an excess of triphenolphosphine. The measurement is useful mainly for
qualitative analysis, because of the partial decomposition of CHHP in the gas chromatograph injector
and column.
Because of the large amount of liquid reagent in the commercial slurry reactor and the need for
cooling to control the reaction temperature, a small slurry system, containing 100 ml of cyclohexane
and consisting of a “top illumination reactor” with sophisticated temperature and flow control, was
used to evaluate the effect of the catalyst constitution. The solution was illuminated from the top of the
reactor through a Pyrex window that cut off the undesired UV radiation. The lamp used in the top
illumation reactor was a 35 W Xe–Hg high-intensity discharge lamp (Philips D2/D2S) equipped with
an incandescent reflector. The catalyst amount was varied between 0 and 2 g/l. Air, presaturated with
cyclohexane at the reaction temperature, was bubbled through the liquid at a rate of 30 ml/min. During
reactions, both gas and liquid samples were withdrawn and analyzed by GC.
Further studies on photocatalytic
reaction kinetics were conducted by FT-IR
spectroscopy (right). TiO2 samples were
pressed into self-supporting wafers of 10
mm in diameter and analyzed in-situ by
transmission Fourier-transform infrared
spectroscopy using a Nicolet Nexus
spectrometer equipped with a MCT detector.
The pellet was mounted into a home-built
miniature stainless steel cell equipped with
transparent CaF2 windows. Prior to loading
of cyclohexane and oxygen from the gas
phase, the catalyst was dehydrated in vacuum (<10-6 mbar) for 2 hours using a turbomolecular pump.
Loading of reactants was controlled by gas pressure. Cyclohexane was introduced into the IR cell until
equilibrium was reached at 3 mbar in the gas phase, followed by addition of 12 mbar of oxygen. The
UV irradiation energy of a mercury lamp (HBO100, Osram) was focused onto one end of a fiber
optical light guide and transmitted to the catalyst by the light guide and a mirror in the sample
compartment of the IR spectrometer. Visible light was filtered using a special UV filter that cuts off
the irradiation above 400 nm.
Organics
V1
V2
V3V4 V5
V8 V7
PI PI V6O2
UV-VIS
IR Detector
Low-vacuum
pump
Turbomolecular
pump
Organics
V1
V2
V3V4 V5
V8 V7
PI PI V6O2
UV-VIS
IR Detector
Low-vacuum
pump
Turbomolecular
pump
Photo-oxidation of cyclohexane
105
5.3 Results
Characteristics of the applied reactor materials: wavelength variation
Fig. 1 depicts the onset of the UV absorption by neat cyclohexane at 270 nm. Below this
wavelength, high light absorption is observed. The figure also shows the transmittance of different
glass types and the emission spectrum of the Hg lamp. A sharp increase in absorption was found at
wavelengths below 230 nm, which is to a very large extent due to dissolved oxygen [28]. A large
transmittance for quartz at wavelengths in the UV-C region (λ < 280 nm) was found; thus, a
considerable amount of the UV-C irradiation from the Hg lamp can be absorbed directly by
cyclohexane in the event that a quartz immersion well is applied. In contrast, transmittance of the
Vycor glass and Pyrex starts at values of 220 and 275 nm, respectively, the latter removing the
radiation that would activate the cyclohexane directly, thus eliminating photolysis processes. In what
follows we show that this has a dramatic influence on the selectivity of the products observed in the
catalytic cyclohexane oxidation.
Figure 1. Comparison of the absorption spectrum of liquid cyclohexane and the UV
transmittance of different glass types. From left to right, quartz, Vycor, and Pyrex. Also shown is the
line spectrum of the applied Mercury lamp. Intensities are normalized to the maximum emission at 366
nm.
Cyclohexane photolysis (no catalyst)
Dark reaction indicates that no thermal induced oxidation takes place under reaction conditions.
In theory the oxidation of cyclohexane proceeds through an energetically most favourable radical
chain mechanism, as is confirmed by a large number of experimental studies. Being aware of the fact
that auto-oxidation is likely to occur after the radical formation step, we performed dark experiments
after the photo-assisted oxidation. The results show negligible auto-oxidation rates after switching off
the irradiation source, regardless of the presence of suspended catalysts. Thus the chain termination,
0
0.2
0.4
0.6
0.8
1
200 250 300 350 400 450 500
λλλλ [nm]
Ab
so
rba
nc
e [
A.U
.]
0
0.2
0.4
0.6
0.8
1
Irra
dia
tio
n [
-] / T
ran
sm
iss
ion
[-]
Chapter5
106
which is a result of collisions between two free radicals or trapping of a radical on catalyst surface
defects, is dominant over the propagation processes.
The results of photolytic oxidation of cyclohexane in the absence of catalyst, using the quartz
immersion well, are illustrated in Figs. 2a, 2b and Fig. 3. Cyclohexanol was formed with an order of
magnitude higher yield over cyclohexanone and other products, as can be seen by comparing the
vertical scales in Figs 2a and 2b. The slightly S-shape conversion plot (Fig.2a) can indicate a short
induction period, most likely related to initiation of the radical reaction and/or heating time of the
applied lamp. This is followed by a constant rate of cyclohexanol production for up to about 3 h of
reaction time, then a significant leveling off of the production rate occurs, related to chain propagation
reactions inducing the formation of oligomeric carbon deposits, appearing as a brownish layer on the
outer sleeve of the immersion well. Thus, the drop in oxidation rate can be attributed to a reduced
photon flux to liquid cyclohexane.
Figure 2. (a) The yield of cyclohexanol compared to total yield as a function of illumination time.
Legend as indicated in the figure. (b) The yield of respectively, cyclohexanone,
1,1’-oxybis(cyclohexane), cyclohexylhexanoate, and CO2 (1/6). Legend as indicated in the figure.
0
0.04
0.08
0.12
0.16
0.2
0 100 200 300 400 500
Irradiation time [min]
Yie
ld [
mo
l/l]
cyclohexanol
total yield
(a)
0
0.004
0.008
0.012
0.016
0.02
0 100 200 300 400 500
Irradiation time [min]
Yie
ld [
mo
l/l]
cyclohexanone
1,1'-oxybis(cyclohexane)
cyclohexyl hexanoate
1/6 CO2
(b)
Photo-oxidation of cyclohexane
107
Figure 3. Ketone/Alcohol ratio and selectivity towards all organic products during the irradiation
with quartz immersion well. Neat pre-dried cyclohexane, 333K.
The formation rates of the other products merits further discussion. 1,1’-Oxybis(cyclohexane) and
cyclohexyl hexanoate, which is formed by the reaction of ketene and cyclohexanol [23], are produced
in the largest fraction besides cyclohexanol and cyclohexanone (Fig.2b). The formation of
1,1’-oxybis(cyclohexane) is explained by the etherification reaction of two cyclohexanol molecules,
producing water. This ether formation is probably the result of radical processes involving
C6H11O· free radicals.
Cyclohexyl formate, cyclohexyl acetate, 3-cyclohexyl-1-propanol and bicyclohexyl are detected
in trace amounts. All compounds are the oxidative-coupling products of cyclohexane and its partially
oxidized species, except for bicyclohexyl, which is formed by the direct coupling of cyclohexane
radicals. Unlike the direct photolysis of cyclohexane in vacuum yielding mainly cyclohexene [29,30],
the fact that photolysis in air favours the oxygenate formation indicates that alkoxy and alkylperoxy
radicals are more abundantly present in the solution than alkyl radicals. This could have been expected
because the reaction of alkyl radical with oxygen is extremely rapid and requires practically no
activation energy [31].
It is visualized in Fig.2(b) that both side products, 1,1’-oxybis(cyclohexane) and cyclohexyl
hexanoate, and carbon dioxide evolve much later than cyclohexanone. The high initial rate of
cyclohexanone formation demonstrates that cyclohexanone is a primary oxidation product of
cyclohexane. Further photooxidation of the ketone proceeds much more rapidly than that of
cyclohexane [31], explaining the rapid leveling off of the yield as a function of irradiation time, as
shown in Fig. 2b. It is well established that the most reactive bonds of the ketone molecule are the C-H
bonds in α-position to the carbonyl group due to the inductive effect of oxygen atoms and the s-p
conjugation effect with the electrons of the C=O bond. The main products of cyclohexanone thermal
oxidation are α-keto hydroperoxide, adipic acid, adipic aldehyde and ε-hydroxycaproic acid, under
which the α-keto hydroperoxide is the primary product of the homolytic cleavage of the α C-H bond.
In contrast to thermal processes, cyclohexyl hexanoate, the esterification product of cyclohexanol
0
0.1
0.2
0.3
0.4
0.5
0 100 200 300 400 500
Irradiation time [min]
K/A
ra
tio
[-]
0.9
0.92
0.94
0.96
0.98
1
Org
an
ics
se
lec
tiv
ity
[-]
Chapter5
108
and hexanoic acid, arises after the cyclohexanone formation in photon-induced cyclohexane oxidation.
It is speculated that hexanoic acid evolves from the photo-excitation of cyclohexanone through an
unstable ketene intermediate [23]. Hence the probable primary process is the dissociative splitting of a
C-C bond adjacent to the absorbing carbonyl group, forming a diradical. UV spectra of cyclohexanone
solution in cyclohexane with varied concentration reveal the development of a new absorption band at
near 280 nm, which is assigned to a singlet-singlet n�π* transition involving the non-bonding
electrons of the oxygen atom. Apparently photolysis proceeds through an intermediate vibrationally
excited ground state, different from the thermally activated transition state intermediate.
The by-product formation of 1,1’-oxybis(cyclohexane) is reported for the first time. It can be
envisaged as the etherification product of cyclohexanol, the primary product of cyclohexane oxidation.
Chatterjee suggested a reaction scheme of this etherification reaction over brønsted acid catalysts [32].
However in our case the ether formation can be better understood as a radical process involving
C6H11O· free radicals.
The origin of carbon dioxide has never been stated clearly, despite the fact that it is the final
degradation product found in most of the previous studies. Kinetics of thermal oxidation of
cyclohexane indicates that CO2 is formed after the carbon-carbon bond cleavage of acryl radicals,
and/or organic acid intermediates [31]. Shimizu et al. [6] investigated an industrial process for adipic
acid production by the liquid phase oxidation of cyclohexanone with molecular oxygen. Delayed
appearance of CO2 at the start of photolysis also indicates the evolution of CO2 proceeds from a
consecutive reaction of cyclohexanol and, at least partially, through the route of cyclohexanone
formation and degradation.
The ketone to alcohol ratio and the organics selectivity, as plotted against the irradiation time (Fig.
3), are deduced from the kinetic curve of cyclohexane oxidation. The K/A ratio is highest at the start of
the reaction due to the high evolution rate of cyclohexanone. However, the accumulation of
cyclohexanone is largely inhibited by the quick consecutive reactions forming organic acids and esters,
resulting in a sharp decrease in K/A ratio. After 200 minutes of reaction equilibrium is established and
the K/A ratio reaches a stable value of around 0.07. On the other hand, the organics selectivity exhibits
a monotonic decrease with time. Unlike the formation of cyclohexanol and cyclohexanone, which
shows a levelling-off behaviour for elongated illumination, the formation rate of CO2 is rather constant
with the exception of a slow induction period. It might be attributed a stable concentration of acid
intermediates and acryl radicals during photolysis. Further kinetic and mechanistic studies are required
to evaluate this hypothesis.
Effect of reactor material on photolysis and photolysis efficiency
Comparing the photolysis in different reactor materials, it can be noted that the highest rate was
achieved when a quartz immersion well was applied (Fig. 4). The total yield was reduced by almost
half when a Vycor glass well was used, with the product distribution remaining largely unmodified.
After most of the UV-B and UV-C radiation was eliminated with the Pyrex immersion well, direct
photolysis became negligible. In the latter case, cyclohexanol and cyclohexanone were detected only
after 150 min of reaction, and the concentration of CO2 in the exhaust gas remained below the TCD’s
detection limit during the entire experiment. Clearly, the energy of individual photons after light
filtration was too low to activate cyclohexane molecules to a photoreactive excited state.
At the beginning of the reaction the photo-oxidation of cyclohexane is dominated by the chain
Photo-oxidation of cyclohexane
109
initiating process. Cyclohexane activation by free radicals can be neglected in comparison with the
radical initiation step. Hence initially photo-oxidation of cyclohexane follows a first-order kinetics
with respect to the organic reactant. The correspondence between the conversion curves in Fig. 4a and
the first-order mechanism is confirmed by the R-squared value from the regression analysis, which is
in all cases greater than 0.99.
Figure 4. Evolution of cyclohexanol, cyclohexanone, CO2 and total conversion in direct
photolysis of cyclohexane with molecular oxygen under modified irradiation with various light filters.
Figure 5 depicts a linear relationship between the oxidation rate of cyclohexane and the effective
radiant flux, correlated to the absorbed photon energy without light filtration. The effective photon
flux was calculated from the light intensities at each specified rays of the UV lamp, the transmittance
of the optical filter and the absorbance of cyclohexane solution, being aware that photon energy at
-0.001
0.001
0.003
0.005
0.007
0.009
0 50 100 150 200 250 300 350 400
Irradiation time [min]
Yie
ld [
mo
l/l]
ChNON, Quartz
ChNON, Vycor
ChNON, Pyrex
1/6 CO2, Quartz
1/6 CO2, Vycor
1/6 CO2, Pyrex
(b)
-0.02
0.02
0.06
0.1
0.14
0.18
0 50 100 150 200 250 300 350 400
Irradiation time [min]
Yie
ld [
mo
l/l]
-0.2
0.2
0.6
1
1.4
1.8
Co
nv
ers
ion
[%
]
ChNOL, Quartz
ChNOL, Vycor
ChNOL, Pyrex
Conv., Quartz
Conv., Vycor
Conv., Pyrex
(a)
Chapter5
110
various wavelengths are not equally efficient in creating photo-reactive species (Fig 1). The excellent
linearity implies that the majority of photo-produced active species is directly involved in the
transformation and do not simultaneously return to the original ground state. We evaluated the
quantum yield in our reactor configuration, based on the ratio of the reaction rate r (in mol produced
per second) and photonic flux expressed by the number of efficient moles of photons. Under the
reaction conditions we applied, a quantum yield of 0.15 was derived corresponding to the linear
correlation of the plot in fig. 5.
Figure 5. Influence of effective photon flux on the photolytic oxidation of cyclohexane.
Effect of cyclohexanol or cyclohexanone addition on photolysis
For a better understanding of the reaction mechanism, it is worthwhile to discuss the generation of
intermediates and side products from the primary oxidation products. Experimentally this problem can
be clarified by carrying out the oxidation in the presence of different additives. In this study we
performed experiments with addition of 12 mmol of cyclohexanol or cyclohexanone respectively, the
amount of which is too tiny to be dominant in the radical processes. The kinetic curve of cyclohexane
photolysis with spiking of cyclohexanol exhibits little difference to pure cyclohexane photo-oxidation,
abstaining from the absolute yield of individual reaction products. Figure 6 illustrates the productivity
after the addition of cyclohexanol as compared to pure cyclohexane oxidation. The effect of
cyclohexanol on the cyclohexane oxidation kinetics is characterised by a strong increase of ester
formation. The yield of other major products, cyclohexanone, cyclohexanol, dicyclohexyl ether and
carbon dioxide are practically unaffected. The more or less constant ketone production indicates that it
might not be formed from cyclohexane solely, as raised ester formation is at the expense of
cyclohexanone. Berezin [31] proposed a reaction scheme in which cyclohexanol is oxidized by
molecular oxygen to yield equal amount of cyclohexanone and hydrogen peroxide.
Different results are obtained in the test with cyclohexanone addition (Fig. 7). One can observe a
sharp decrease in cyclohexanone concentration at the start of the reaction. It reaches a steady state at
0.008 mol/l after 100 mins of reaction. Analogous to photo-oxidation with cyclohexanol spiked, a
higher yield of cyclohexyl hexanoate is found after addition of cyclohexanone. It has little effect on
0
2
4
6
8
10
0 0.2 0.4 0.6 0.8 1 1.2
ΦΦΦΦ /ΦΦΦΦ 0 [-]
ko
bs [
×× ××1
0-7
mo
l/s
]
Photo-oxidation of cyclohexane
111
the course of cyclohexanol and ether formation. However, the emission of carbon dioxide is improved,
especially at the beginning of the reaction. It indicates that cyclohexanone is the intermediate product
of photolysis, which is more reactive and prone to be further oxidized to other oxygenates, i.e.,
cyclohexanol, cyclohexyl hexanoate and CO2.
Figure 6. Effect of cyclohexanol addition on productivity (reaction time = 400 min). Dark bars
represent the yield without the addition and the light bars show the product distribution with
pre-addition of cyclohexanol. Points with error bars represent the relative productivity after
cyclohexanol addition with respect to pure cyclohexane oxidation.
Figure 7. Kinetics of photolytic cyclohexane oxidation product formation after addition of
cyclohexanone.
0
0.1
0.2
0.3
0.4
ChNON ChNOL Ether Ester 1/6 CO2 Total
Yie
ld [
mo
l/l]
0
0.5
1
1.5
2
Re
lati
ve
pro
du
cti
vit
y [
-]
0
0.01
0.02
0.03
0.04
0 100 200 300 400 500
Irradiation time [min]
Yie
ld [
mo
l/l]
0
0.03
0.06
0.09
0.12
Yie
ld C
hN
OL
[m
ol/l]
cyclohexanone
1,1'-oxybis(cyclohexane)
cyclohexyl hexanoate
1/6 CO2
cyclohexanol
Chapter5
112
Photocatalytic oxidation of cyclohexane
The effect of the applied reactor material on the product formation of cyclohexane oxidation
without (photolysis) and with catalyst (Degussa P25, photocatalysis) is further illustrated in Fig. 8. For
simplicity, this figure shows only cyclohexanol and cyclohexanone production in the case of
photolysis, neglecting the minor products of consecutive radical chemistry, which generally were not
observed in photocatalysis. As stated previously, in the photolysis reaction conducted with the quartz
immersion well, cyclohexanol was the major product, with little cyclohexanone formed. When using
the Pyrex well (Fig. 8D), photolysis was negligible. When P25 was introduced into the reactor
equipped with a quartz well, the product distribution changed dramatically (compare Figs. 8A and 8B).
In the presence of the catalyst particles, cyclohexanone became the major product, whereas
cyclohexanol production was largely suppressed. When using the Pyrex well (Fig. 8C), cyclohexanol
production was practically nil, and cyclohexanone was obtained with high selectivity.
Figure 8. Effect of the experimental conditions on the cyclohexanone and cyclohexanol amounts
produced. (A) Quartz reactor, no catalyst (pure photolysis, cf. Fig. 2), (B) Quartz reactor, with catalyst
(1 g/l of P25), (C) Pyrex reactor with catalyst (1 g/l of P25), and (D) Pyrex reactor, no catalyst.
It should be noted that there is no apparent linear initial part in the production curve of
cyclohexanone (cf. Fig. 8), which, combined with the fact that the kinetics of the reaction are not
known, makes the determination of the specific activity (per g of catalyst) or intrinsic activity (per m2
of catalyst) tedious and possible only with insufficient accuracy. By a rough comparison of the
production curves of cyclohexanol and cyclohexanone in Figs. 8a and 8b, the catalytic rate is at least
one order of magnitude lower than the photolysis rate, and quantum efficiency is estimated to be in the
order of 1–2% in these specific reaction conditions.
Fig. 9 shows the effect of the amount of catalyst in the quartz reactor on the product distribution
(cyclohexanone/cyclohexanol). Increasing the amount of catalyst results in increased cyclohexanone
production and decreased cyclohexanol production, up to a catalyst density of about 1 g/l, after which
the addition of more catalyst has little effect on the quantities produced.
0
0.02
0.04
0.06
-100 0 100 200 300
Time [min]
Yield [mol/l]
0
0.02
0.04
0.06
-100 0 100 200 300
Time [min]
Yield [mol/l]
cyclohexanone
cyclohexanol
0
0.02
0.04
0.06
-100 0 100 200 300
Time [min]
Yield [mol/l]
0
0.02
0.04
0.06
-100 0 100 200 300
Time [min]
Yield [mol/l]
A B
C D
Photo-oxidation of cyclohexane
113
Figure 9. Effect of the TiO2 slurry density on the amounts of cyclohexanol and cyclohexanone
formed after 60 min of irradiation time in the quartz immersion well reactor.
The evolution of CO2 was evaluated using the optimized amount of 1 g TiO2/l. Fig. 10a shows the
development of the product constitution as a function of reaction time. A significant decrease in
reactivity can be observed after about 45 min of illumination. Although not directly apparent from the
product distribution shown in Fig. 10a, the evolution of CO2 is somewhat retarded in the first hour of
reaction, leading to an apparent decreasing selectivity of total selective oxidation products as a
function of time. This is further illustrated in Fig. 10b, which shows a decrease in selective oxidation
products (ketone and alcohol) from >95% to about 85%.
Figure 10. Product development as a function of irradiation time (1 g/l slurry density of
Hombikat catalyst, top illumination reactor). (a) Cyclohexanol, cyclohexanone and CO2 production
(legend as indicated in the figure); (b) Selectivity of ketone and alcohol as a function of time.
Fig. 11 illustrates that adding cyclohexanol to the reaction mixture decreased cyclohexanone
production. Increasing the amount of cyclohexanol from 0.05 to 0.11 g had no further deteriorating
effect on the cyclohexanone formation, however.
0
0.003
0.006
0.009
0.012
0.015
0 0.5 1 1.5 2 2.5
TiO2 concentration [g/l]
Yie
ld [
mo
l/l]
Cyclohexanone
Cyclohexanol
Irradiation time [min]
0 100 200 300 400
Ke
ton
e+
alc
oh
ol
se
lec
tivit
y [
-]
0.80
0.85
0.90
0.95
1.00
Irradiation time [min]
0 100 200 300 400
Yie
ld [
mo
l/l]
0.000
0.002
0.004
0.006
0.008
cyclohexanone
cyclohexanol
1/6 CO2
(a) (b)
Chapter5
114
Figure 11. Effect of preaddition of cyclohexanol (0.5 and 1.1 g) on the cyclohexanone production
curve. Reaction conditions: top illumination reactor, 100 ml cyclohexane, 1 g/l slurry density of
Hombikat catalyst.
Effect of catalyst constitution on performance
Besides the amount of catalyst added to the reactor system, the composition of the catalyst also
affects the obtained reaction rates. This is illustrated in Fig. 12, which shows the effect of Hombikat
pretreatment temperature on performance. In principle, Hombikat TiO2 is more active than P25,
whereas pretreatment of Hombikat at 773 K results in similar activity. Further increase in pretreatment
temperature results in further deterioration of activity, with activity reduced by a factor of 2 at 1073 K
and almost completely eliminated at 1373 K. Remarkably, the selectivity of the reaction was hardly
affected. High-temperature treatment induced various modifications in Hombikat, the most important
of which were the reduction of the surface area and hydroxyl group density and a phase transition
from anatase to rutile above ~1000 K. We discussed this in more detail in Section 3.7.
Time [min]
0 50 100 150 200
Cyclo
hex
an
on
e y
ield
[m
ol/
l]
0.000
0.002
0.004
0.006
0 g cyclohexanol
0.05 g cyclohexanol
0.11 g cyclohexanol
Photo-oxidation of cyclohexane
115
Figure 12. Comparison of the performance of various TiO2 samples in the production of
cyclohexanone (Hombikat pretreated at respectively 393, 773, 1073 and 1373 K, and P25 treated at
393 K). Legend as indicated in the figure.
Catalyst morphology as a function of pretreatment temperature
From the XRD diffraction patterns (Chapter 3), it follows that the composition of Hombikat
changed from a purely crystalline anatase phase to a rutile phase starting at about 1000 K. At this
temperature, a mixed composition of anatase and rutile phases was obtained, whereas above 1273 K,
the catalyst consisted predominantly of rutile. Fig. 13 shows the UV absorption spectra of the
temperature-pretreated Hombikat samples. Temperature treatments up to about 1000 K had little affect
on the absorption spectra. At pretreatment temperatures above 1000 K, the absorption maximum at
about 385 nm gradually increased as a function of increasing pretreatment temperature. This enhanced
absorbance is related to formation of the rutile phase in the catalysts, as was observed in the XRD
patterns of the corresponding materials.
0
0.002
0.004
0.006
0.008
0 100 200 300 400
Time [min]
Cycl
ohex
anone
yie
ld [
mol/
l]
Hombikat, 393 K
Hombikat, 773 K
Hombikat, 1073 K
Hombikat, 1373 K
P25
Chapter5
116
Figure 13. Effect of the pretreatment temperature of Hombikat on the corresponding UV/vis
absorption spectra. Starting from a pretreatment temperature of about 1373 K, a gradual increase of
the absorption maximum at 385 nm is induced.
Fig. 14a gives examples of nitrogen adsorption–desorption isotherms of treated Hombikat UV100
samples. All show type II isotherms, indicating some mesoporosity and macroporosity. It is very
unlikely that cyclohexane will suffer from restrictions diffusing into these open structures. The
absence of a plateau at high relative pressures (p/p0) indicates the filling of interparticle voids and the
presence of surface roughness. Fig. 10b illustrates the decreased surface area of Hombikat as a
function of pretreatment temperature, with BET surface area decreasing from 330 m2/g to only a few
m2/g.
Figure 14. Nitrogen adsorption–desorption isotherms (a) and the corresponding surface areas as
calculated from the BET method (b) of Hombikat pretreated at 393, 773, 1073 and 1373 K, and P25
pretreated at 393 K, respectively.
Wavelength [nm]
300 350 400 450
Ab
so
rba
nce
[-]
0.0
0.5
1.0
1.5
Hombikat, 393 K
Hombikat, 773 K
Hombikat, 1073 K
Hombikat, 1373 K
P25
0
100
200
300
400
0 0.2 0.4 0.6 0.8 1
p /p 0 [-]
Va
ds
[c
m3/g
]
Hombikat, 393 K
Hombikat, 773 K
Hombikat, 1073 K
Hombikat, 1373 K
P25
0
100
200
300
400
273 673 1073 1473
Calc. temp. [ ]
SB
ET [
m2/g
]
(a) (b)
[ C]
Photo-oxidation of cyclohexane
117
TPD of ammonia (NH3-TPD) is a very common method for characterizing acidic OH groups in
microporous and mesoporous materials [33]. Fig. 15 shows NH3-TPD spectra of various thermally
pretreated titania catalysts. Two desorption maxima can be distinguished at approximately 475 and
600 K. The low-temperature maximum is usually assigned to the removal of ammonia interacting with
surface adsorbed water molecules, whereas the high-temperature maximum is correlated with NH3
associated with the sites created by dehydroxylation of surface OH groups [34]. According to
Almquist and Biswas [20], both surface water and hydroxyl groups evenly contribute to photocatalytic
oxidation reactions in organic solvent, because in both cases hydroxyl radicals are formed by
photogenerated holes. This hypothesis is plausible, because at a surface coverage of 5 H2O/nm2 (as is
Figure 15. Ammonia TPD spectra of the applied Hombikat catalysts. Legend as indicated in the
figure.
the case for most titania under normal conditions), the fully nondissociated state of water and the
fully dissociated configurations can compete in energy within <7 kJ/mol [35]. Therefore, in the
discussion that follows, “surface OH groups” is used to designate an OH mode from either Ti–OH or
surface-adsorbed H2O.
Table 2 compares the surface properties of the various catalysts. The total surface-OH group
quantity of P25 is in good agreement with the quantities reported by Van Veen et al. [26], Chhor et al.
[36], and Boehm [37] (the latter of which applied various probe molecules to characterize surface
hydroxyl groups on P25 catalysts). As discussed earlier, the performance of the Hombikat pretreated at
773 K is very similar to that observed for P25, pretreated at 393 K. Comparing the data in Table 2
indicates that the number of hydroxyl groups is in much better agreement in the Hombikat 773 K
catalyst and the P25 catalyst compared with the BET surface area, and thus is a more important
parameter than surface area per se.
Temperature [K]
373 473 573 673 773
TC
D s
ign
al
[A.U
.]
Hombikat, 393 K
Hombikat, 773 K
Hombikat 1073 K
Hombikat 1373 K
P25
Chapter5
118
Table 2. Comparison of surface area and OH group density of applied catalysts
Hombikat P25
393 K 773 K 1073 K 1373 K
Surface area (m2/g)
a 337 113 18 1.8 51
Total surface-OH (mmol/g)b 1.21 0.40 0.090 0.012 0.38
Acidic-OH (mmol/g)c 0.92 0.33 0.070 0.009 0.30
Acidic-OH/basic-OHb
3.3 4.7 3.6 3.4 3.6
Surface-OH density (1/nm2) 2.2 2.1 3.0 3.9 4.5
a Determined by nitrogen adsorption and desorption isotherm.
b Determined by Fe(acac)3 adsorption method as described by Van Veen et al. [26].
c Determined by ammonia TPD.
In-situ FT-IR studies
Figure 16. FI-IR spectra of cyclohexane photooxidation on Hombikat catalyst, with the increased
UV illumination time from bottom to top. Conditions described in Section Experimental.
The surface reaction of cyclohexane photocatalytic oxidation on TiO2 and the reaction
products/intermediates were investigated using an in-situ vacuum FT-IR transmission cell. Initially an
absorption spectrum was obtained of neat cyclohexane adsorbed on Hombikat surface. Figure 16
shows the evolution of the adsorption spectra of Hombikat sample loaded with cyclohexane and
oxygen, as function of the irradiation time. Product peaks are observed at 2350, 2150, 2130, 1670,
1550 and 1360 cm-1. The peak at 2350 cm-1 and the double peak at 2150 and 2130 cm-1 are the
characteristic peaks of gas phase CO2 and CO respectively. Carbon monoxide is a new species that
was not detected using the analytical methods mentioned in previous paragraphs. The formation of the
main organic product, cyclohexanone, can be clearly followed by the increased intensity of the peak at
1670 cm-1 representing the C=O mode in cyclohexanone. The broad peak at 1360 cm-1 is assigned to
the vibration mode of the reaction intermediate, cyclohexyl hydroperoxide [8,11,38]. The broad
shoulder at 1550 cm-1
can be attributed to the formation of various surface carbonates and carboxylates.
Furthermore the absorption spectra of surface-OH stretching mode at the high wavenumber region
60 s60 s60 s
600 s600 s600 s
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
-0.0
0.1
0.2
0.3
0.4
Arb
itra
ry u
nits
1500 2000 2500 3000 3500 4000
Wavenumbers (cm-1)
Surface O-HSurface O-H
0 s
CO
CO2
CO
CO2
Surface carbonates
C=O
6000 s6000 s6000 s
Peroxides
Illuminationtime
Photo-oxidation of cyclohexane
119
above 3500 cm-1 shows a clear shift towards lower wavenumbers with prolonged irradiation,
indicating more interaction of the products and intermediates with the surface hydroxyl groups.
Furthermore the formation of cyclohexanol was not detected in the FT-IR study [11,38].
5.4 Discussion
As is stated in the Introduction, various research groups have evaluated the performance of TiO2
in the selective oxidation of cyclohexane to cyclohexanol and cyclohexanone (Table 1). The
performance of the photocatalysts in the oxidation of cyclohexane is usually discussed only on the
basis of the applied catalyst material, with details of the reactors applied and thus the wavelengths to
which the reaction mixture is exposed not mentioned. The most important finding of the present study
is that varying the applied reactor material and the applied slurry density greatly affects the selectivity
of the photocatalytic oxidation of cyclohexane, which can vary from >85% selectivity to cyclohexanol
for photolytic oxidation to >95% selectivity to cyclohexanone for photocatalytic oxidation. This is
clearly illustrated by the results presented in Figs. 1, 2, 4, 8, 9.
When a quartz reactor and no catalyst are used, photolysis of cyclohexane is possible at
wavelengths λ < 275 nm. Along with the main product, cyclohexanol, various other products are
obtained in the photolysis of cyclohexane, which, as discussed previously, are most likely the result of
radical chemistry. The radical pathways, which have been discussed previously [18,20], can describe
the observations well. This radical chemistry is also responsible for the formation of oligomeric carbon
deposits on the walls of the immersed lamp, leading to reduced reaction efficiency.
When catalytic material (TiO2) is added to the reactor, depending on the amount, catalytic surface
reactions become dominant over photolysis radical reactions, and the various products of coupling
reactions are below the detection limit of the applied analytical procedures. The overall product
amount is decreasing as a result of inefficient light absorption by the catalyst particles, in which to a
large extent the generated holes and electrons (representative of the activated state) are recombining to
produce heat, rather than induce chemical conversion. As discussed in chapter 1, this results in a loss
of photoefficiency of at least one order of magnitude. To completely exclude the radical chemistry
induced by photolysis, and to obtain an as high a cyclohexanone selectivity as possible, Pyrex should
be used as the reactor material to prevent illumination of the reaction mixture to wavelengths <275 nm,
as indicated by the product distributions in Fig. 8. In view of this, along with the observed solvent
effects reported previously [18,20], it is likely that the nanoparticle effect to explain selectivity
changes claimed by Su et al. [17] and Li et al. [21,22] is nonexistent, and that the reversed selectivity
reported previously [17,21,22] is the result of the applied reactor material (quartz), and possibly the
addition of acetonitril to cyclohexane. The high selectivity in the photocatalysis of cyclohexane
oxidation to cyclohexanone has been extensively discussed in the literature. The consecutive reaction
of cyclohexanol to cyclohexanone has been proposed to explain the high selectivity to cyclohexanone
[13-15,18,20], whereas the high selectivity to cyclohexanone may also be related to a preferred direct
catalytic route of cyclohexane oxidation, as proposed by Boarini et al. [8]. From the results of the
present study, it can be postulated that the direct route (parallel formation of cyclohexanol and
cyclohexanone) is more likely, as discussed next.
As stated previously, the reaction rate significantly decreased as a function of reaction time using
TiO2 as a photocatalyst (Fig. 10). This cannot be the result of first-order cyclohexane behavior,
Chapter5
120
because the conversion of cyclohexane is very low. Rather, this reaction profile suggests that products
are accumulating on the catalyst surface, reducing the effectiveness of the photocatalyst. Results of the
experiments in which cyclohexanol was preadded to the reaction mixture show that cyclohexanol
inhibits cyclohexanone formation, in agreement with previous observations and with the results of
Almquist and Biswas [20]. Furthermore, in preliminary infrared studies, it was observed that
cyclohexanol is strongly adsorbed on the catalyst surface, as was also proposed by Almquist and
Biswas [20]. However, adsorbed cyclohexanol is photocatalytically converted mainly to carboxylates
and not to cyclohexanone, in agreement with IR studies on TiO2 catalysts that found adsorbed
alkoxygroups to be prone to formate and acetate formation on the surface, being precursors of CO2
[39]. This finding is in good agreement with the decreasing selectivity as a function of reaction time
observed in the present study (Fig. 10b). The coproduct cyclohexanol indeed accumulated on the
surface, but yielded mainly carboxylates and induced deactivation, rather than contributing to
cyclohexanone formation to any great extent. Besides cyclohexanol, the FT-IR spectra shown in this
Chapter (Fig. 16) show that cyclohexanone is also strongly adsorbed on the catalyst surface. It is
therefore likely that consecutive cyclohexane oxidation also contributes to carboxylate and, eventually,
CO2 formation.
Effect of catalyst pretreatment on cyclohexanone production
It is evident from the various characterization techniques discussed in chapter 3, that a
high-temperature treatment has a significant effect on both catalyst composition (rutile or anatase) and
catalyst texture. The transition of anatase to rutile induces enhanced light absorption at relatively high
wavelengths by the catalyst (Fig. 13); however, this does not lead to enhanced cyclohexanone
formation. Apparently, light absorption in rutile phases is less effective in inducing catalytic reaction
than anatase, because of the morphological changes accompanying the transformation of anatase to
rutile. A comparison of the performance and constitution of the Hombikat catalysts pretreated at
various temperatures with the performance of Degussa P25 is highly illustrative (Fig. 12, Table 2).
Comparing the data in Table 2 indicates that the number of hydroxyl groups is in much better
agreement between the Hombikat-773 K catalyst and the P25 catalyst, which show comparable
activity profiles, than the BET surface area. Thus, it seems that the number of hydroxyl groups is a
more important parameter than the surface area per se. Because the kinetic curves are similar, the
specific activity can be assumed to be similar, and thus the intrinsic activity (per m2 of catalyst) of P25
is about twice that of Hombikat-773 K. In other words, compared with Hombikat, P25 effectively
accommodates twice the amount of active hydroxyl groups at a comparable surface area (Table 2). At
the same time, it can be concluded that the intrinsic activity of each hydroxyl group is independent on
the catalyst used (Hombikat-773 K or P25). Based on the determined total amount of surface
Ti-hydroxyl groups in the reactor loaded with P25 (0.04 mmol) and a first rough approximation of the
corresponding initial photooxidation rate of 0.4 mmol/h (which, as stated earlier, is hard to determine
due to the absence of a linear part in the production curve (see Fig. 12)), a turnover of 10 h−1
can be
calculated for each OH group on the surface. This is a very low number compared with that for, say,
homogeneous catalysts with several orders of magnitude higher turnovers.
The importance of surface hydroxyl groups was further elucidated using vacuum FT-IR reaction
studies (Fig. 16). The interaction of surface-OH with the reaction products/intermediates increases
with prolonged reaction time. CO2, CO and surface carbonates appear to be the secondary products
Photo-oxidation of cyclohexane
121
after the cyclohexanone formation. A comparative study on photocatalytic reactions on ZnO also
revealed the crucial role of surface hydroxyls in cyclohexane oxidation (results not shown). ZnO was
experimentally proved to be more active than Homibicat catalyst during the photocatalytic degradation
of methylene blue. On the other hand, the photocatalytic activity of ZnO in cyclohexane oxidation was
found to be negligible both in photo-reactors and using FT-IR photocatalytic cells. The apparent
discrepancies can be explained by the crucial roles of hydroxyl species. ZnO is a bulk catalyst that
hardly contains any surface hydroxyls, as is identified using aforementioned procedures as well as
FT-IR. Therefore, unlike the methylene blue decolorization in aqueous solutions of which free
hydroxyl ions are abundant, the cyclohexane photocatalytic oxidation on ZnO can hardly proceed due
to a lack of surface hydroxyl groups on ZnO.
Figure 17. Reaction scheme proposed for the photo-catalytic production of cyclohexanone over
TiO2 catalysts.
It should be noted that from Fig. 12, it follows that the activity of the Hombikat catalyst
deteriorates continuously as a function of increasing pretreatment temperature. Hisanaga et al. [39]
investigated the effect of calcination temperature on photocatalytic performance in relation to
water-phase oxidation processes. In water-phase reactions, calcination in the temperature range of
300–773 K typically had little effect or improved photooxidation activity, depending on the solubility
of the substrates in water. To explain this improvement, the efficiency of electron and hole charge
separation was proposed to be higher in larger crystals (calcination temperature up to 773 K). The
effect of calcination temperature on the hydroxyl group density on the TiO2 surface was not discussed
by Hisanaga et al. [39] and appears to be less relevant in water-phase reactions than in neat
cyclohexane. Reconstruction of hydroxyl groups by immersion in water is likely, but is more difficult
to envisage in pure cyclohexane. If it exists for Hombikat, then apparently the enhanced efficiency of
electron hole separation in larger crystals cannot compensate for the reduced hydroxyl group density,
explaining the continuously deteriorating efficiency as a function of increasing calcination
Chapter5
122
temperature.
Combining all of the information presented herein, we can propose a mechanism for
cyclohexanone formation, as shown in Fig. 17. After adsorption of cyclohexane and the initial
activation by light, a reaction occurs between an activated hydroxyl group and cyclohexane, yielding
water and an adsorbed cyclohexyl radical. Subsequently, oxygen is activated by the thus-generated
Ti(III) center, yielding O2- . Recombination of the cyclohexyl radical and the surface O2
- anion results
in the formation of a peroxide intermediate that subsequently decomposes to cyclohexanone and
restores the hydroxyl group on the catalyst surface. The formation of peroxides as important
intermediates in zeolite-induced selective (photo)oxidation has been extensively described and
discussed by Frei and coworkers [8,9], including the reaction of cyclohexane to cyclohexanone [8].
Moreover, cyclohexylhydroperoxide is a common intermediate in the currently applied processes for
cyclohexanone production and is known to (catalytically) decompose to either alcohol or ketone [40].
Alternatively, the surface cyclohexylradical might form surface cyclohexanol via photon-induced
hydroxyl group activation, but this is speculative. Whatever the pathway to surface cyclohexanol
formation, consecutive oxidation, possibly through superoxide anions, leads to carboxylates on the
surface, which contribute to deactivation of TiO2 and also to CO2 formation. Carboxylates have been
previously shown to deactivate Au/TiO2 catalysts in the low-temperature selective propene
epoxidation reaction, using hydrogen and oxygen [41]. It should also be noted that this mechanism is
oversimplified; it was previously observed that, depending on the applied solvent, cyclohexanol can be
converted photocatalytically to cyclohexanone over TiO2. Further research, using ATR-FTIR and
DRIFT spectroscopy, is required to corroborate the mechanism proposed in Fig. 12, also taking into
account the more extensive reaction pathways proposed by other investigators [12-23]. In addition,
studies with deuterated cyclohexane and cyclohexyl hydroperoxide are recommended to evaluate the
kinetic isotope effect and products of peroxide decomposition, respectively, and thus reveal the
kinetically relevant step in the photooxidation of cyclohexane to cyclohexanone over TiO2 catalysts
and the extended network of surface photooxidation reactions. Further analysis of the reaction
mechanism and kinetically relevant steps may lead to design rules for significantly improved
photocatalysts and reactors, which are needed to bring the photocatalytic oxidation of cyclohexane
within the range of rates of interest to the chemical industry.
Photo-oxidation of cyclohexane
123
5.5 Conclusions
The following conclusions can be derived from the work described in this chapter:
Cyclohexanol is the major product of uncatalyzed photooxidation of cyclohexane at λ < 275
nm. Adding a catalyst suppresses cyclohexanol formation and enhances cyclohexanone formation
under these conditions.
If photolysis is prevented by the use of the proper light filters (e.g., Pyrex, λ > 275 nm), then
photocatalysis over TiO2 yields predominantly cyclohexanone (selectivity >95%).
In immersion well-type reactors, as well as in top illumination reactors, an optimized TiO2
slurry density of about 1 g/l was found. A higher amount shields part of the available reactor volume
from the light.
The textural and chemical composition of the applied TiO2 was found to have a significant
effect on the activity of the catalyst, but not to effect the selectivity. The main variable affecting
activity is the hydroxyl group density on the surface of the applied TiO2, suggesting that
hydroxylgroups are directly involved in the kinetically relevant step of the photooxidation of
cyclohexane to cyclohexanol.
Chapter5
124
References
1. Fujishima, A., Honda, K., Nature, 1972, 238, 37
2. Schiavello, M., Heterogeneous photocatalysis, Wiley Press, Chichester, 1997
3. Fox, M.A., Dulay, M.T., Chem. Rev., 1993, 93, 341
4. Hoffmann, M.R., Martin, S.T., Choi, W., Bahnemann, D.W., Chem. Rev., 1995, 95, 69
5. Suresh, A.K., Sharma, M.M., Sridhar, T., Ind. Eng. Chem. Res.,2002, 39, 3958
6. Shimizu, K., Kaneko, T., Fujishima, T., Kodama, T., Yoshida, H., Kitayama, Y., Appl. Catal.
A, 2002, 225, 185
7. Unnikrishnan, R.P., Sahle-Demessie, E., J. Catal., 2002, 211, 434
8. Sun, H., Blatter, F., Frei, H., J. Am. Chem. Soc., 1996, 118, 6873
9. Sun, H., Blatter, F., Frei, H., Catal. Lett., 1997, 44, 247
10. Blatter, F., Sun, H., Vasenkov, S., Frei, H., Catal. Today, 1998, 41, 297
11. Larsen, R.G., Saladino, A.C., Hunt, T.A., Mann, J.E., Xu, M., Grassian, V.H., Larsen, S.C., J.
Catal., 2001, 204, 440
12. Giannotti, C., Le Greneur, S., Watts, O., Tetrahedron Lett., 1983, 24, 5071
13. Mu, W., Herrmann,, J.M., Pichat, P., Catal. Lett., 1989, 3, 73
14. Herrmann, J.W., Mu, W., Pichat, P., Stud. Surf. Sci. Catal., 1991, 59, 405
15. Lu, G., Gao, H., Suo, J., Li, S., J. Chem. Soc., Chem. Commun., 1994, 2423
16. Shiojiri, S., Hirai, T., Komasawa, I., J. Chem. Eng. Jpn., 1997, 30, 137
17. Su, B., He, Y., Li, X., Lin, E., Indian J. Chem., 1997, 36A, 785
18. Boarini, P., Carassiti, V., Maldotti, A., Amadelli, R., Langmuir, 1998, 14, 2080
19. Gonzalez, M.A., Howell, S.G., Sikdar, S.K., J. Catal., 1999, 183, 159
20. Almquist, C.B., Biswas, P., Appl. Catal. A, 2001, 214, 259
21. Li., X., Chen, G., Yue, P.L., Kutal, C., J. Chem. Tech. Biotech., 2003, 78, 1246
22. Li., X., Quan, X., Kutal, C., Scripta Materialia, 2004, 50, 499
23. Teramura, K., Tanaka, T., Kani, M., Hosokawa, T., Funabiki, T., J. Mol. Catal. A, 2004, 208,
299
24. Brusa, M.A., Grela, M.A., J. Phys. Chem. B, 2005, 109, 1914
25. Matthews, R.W., McEvoy, S.R., J. Photochem. Photobiol. A: Chem., 1992, 66, 355
26. Van Veen, R.J.A., Veltmaat, R.J.G., Jonkers, G., J. Chem. Soc., Chem. Commun., 1985, 23,
1656
27. Olah, G.A., Shamma, T., Surya Prakash, G.K., Catal. Lett., 1996, 46, 1
28. Perkampus, H.H., Sandeman, I., Timmons, C.J., DMS UV Atlas of Organic Compounds.
Verlag Chemie, Weinheim, 1966-1971, Vol. I
29. Yang, J.Y., Servedio, F.M., J. Chem. Phys., 1968, 48, 1331
30. Nafisi-Movaghar, J., Hatano, Y., J. Phys. Chem., 1974, 78, 1899
31. Berezin, I.V., Denisov, E.T., Emanuel, N.M., The Oxidation of Cyclohexane, Pergamon Press,
Oxford, 1966
32. Chatterjee, D., Mody, H.M., Bhatt, K.N., J. Mol. Catal. A, 1995, 104, L115
33. Hunger, B., Heuchel, M., Clark, L.A., Snurr, R.Q., J. Phys. Chem. B, 2002, 106, 3882
34. Primet, M., Pichat, P., Mathieu, M.V., J. Phys. Chem., 1971, 75, 1216
35. Arrouvel, C., Digne, M., Breysse, M., Toulhoat, H., Raybaud, P., J. Catal., 2004, 222, 152
Photo-oxidation of cyclohexane
125
36. Chhor, K., Bocquet, J.F., Colbeau-Justin, C., Mater. Chem. Phys., 2004, 86, 123
37. Boehm, H.P., Discuss. Faraday Soc., 1971, 52, 264
38. Li, G., Xu, M., Larsen, S.C., Grassian, V.H., J. Mol. Catal. A, 2003, 194, 169
39. Rivera, A.P., Tanaka, K, Hisanaga, T., Appl. Catal. B: Environ., 1993, 3, 37
40. Hansen, C.H., Mul, G., Tabor, G.B.J., J. Royal Netherlands Chem. Soc., 1993, 112, 497
41. Mul, G., Zwijnenburg, A., Van der Linden, B., Makkee, M., Moulijn, J.A., J. Catal., 2001,
201, 128
Chapter5
126
127
6
A Novel Photocatalytic Monolith Reactor for Multiphase
Heterogeneous Photocatalysis
Abstract
A novel reactor for multi-phase photocatalysis is presented, the so-called Internally Illuminated
Monolith Reactor (IIMR). In the concept of the IIMR, side-light emitting fibers are placed inside the
channels of a ceramic monolith, equipped with a TiO2 photo-catalyst coated on the wall of each
individual channel. The photonic efficiency achieved with the IIMR reactor in the selective
photo-oxidation of cyclohexane is 0.062, which is lower than obtained with a top illumination slurry
reactor (0.151), but higher than the efficiency of an annular slurry reactor and reactor configuration
with side-light fibers immersed in a TiO2 slurry, reaching a photonic efficiency of 0.008 and 0.002,
respectively. The results are discussed on the basis of differences in photon flows entering the
reactors, and the related magnitude of product concentrations.
Chapter 6
128
6.1 Introduction
Photocatalysis applying semiconductor materials has attracted many researchers active in the
fields of physical chemistry, material science, catalysis, and reactor engineering. By far the most
research activity in photocatalysis is in the field of environmental abatement, such as air cleaning and
wastewater purification, in which organic compounds are totally oxidized into carbon dioxide and
water over mainly TiO2 based photocatalysts [1, 2, 3, 4]. Photocatalytic synthetic processes using
selective oxidation have been less well developed, partially due to the lack of a proper design of
multiphase photo-reactors. Currently photo-reactors for liquid phase oxidation are typically based on
slurry systems i.e., the solid phase is dispersed within the liquid in the reactor. Although this design
offers ease of construction and high catalyst loading, it has clearly drawbacks, such as the difficulty of
separation of catalyst particles from the reaction mixture, and low light utilization efficiencies due to
the scattering and shielding of light by the reaction medium and catalyst particles.
Catalyst separation difficulties can be avoided by immobilization of the photo-catalyst on a fixed
support. In immobilized bed reactors, the photocatalysts are coated on the walls of the reactor or on a
support matrix around the light source [5, 6, 7]. In this case, the total illuminated (surface) area is
largely limited by the geometry of standard light sources and the spatial distance between the catalyst
and light. The light is quickly attenuated by absorption and scattering by the reaction medium and
catalyst. Therefore, the challenge of photocatalysis consists of developing reactors, allowing an
increase in photonic efficiency [8].
Various attempts have been made to amend the aforementioned light distribution problem with
immobilized photocatalytic systems [9, 10]. One approach was to employ optical fibers as a light
distributing guide and support for photocatalysts. Light propagates through the fiber core, whilst
certain amount of photons is refracted into the coated titania layer. By this means, the optical fibers
enable the remote delivery of photon energy to the reactive sites of the photocatalyst. Ollis and
Marinangeli were the first to conduct studies on an optical fiber reactor (OFR) [1]. Various groups
have reported on the application of titania coated optical fibers in photocatalytic purification of air and
water. Hofstadler investigated 4-chlorophenol degradation in a multi-fiber photocatalytic reactor [11].
With their design, they obtained a degradation rate of 4-chlorophenol that is 1.6 times higher than
obtained with a conventional slurry reactor. In a different approach to a quite similar concept, Tada and
Honda [12] studied the performance of a titania film coated on a 10 mm diameter quartz rod, which
functioned as an internal light guide. Its efficiency was reported to be up to 50 times higher than that
yielded by a P25 TiO2 slurry reactor. Peil and Hoffmann have developed and modelled an optical fiber
reactor system for wastewater treatment [13]. Choi et al. investigated photocatalytic oxidation of
acetone in air using a single optical fiber reactor [14]. The light delivery and distribution phenomena
along optical fibers coated with P25 TiO2 particles were studied by Wang and Ku [15]. In the work of
Rice and Raftery [16], Sun et al. [17], and Wang and Ku [15], large numbers of titania coated optical
fibers were bundled together and applied for photocatalytic air treatment. The optical fiber reactor
concept was also evaluated by Danion et al. [18]. The photocatalytic activity was optimized for the
TiO2 film thickness as a function of fiber length. The most recent example is the development of an
optical fiber monolith reactor, as reported by Lin and Valsaraj [19]. They used a monolith for
photocatalytic wastewater treatment with the channels of the monolith completely filled with flowing
liquid. The monolith structure was used merely as the distributor of the optical fibers, while the
benefits of monoliths, such as low pressure drop and excellent mass transfer characteristics for
Novel photocatalytic monolith reactor
129
gas/liquid systems in certain hydrodynamic regimes, were not fully exploited to optimize the
photocatalytic oxidation reaction [20].
All these reactor designs based on coating of a TiO2 catalyst layer on quartz fibers possess several
intrinsic drawbacks. Firstly, the adhesion strength and layer thickness of catalyst coating on the fibers
strongly affect its durability and performance. As the adhesion of TiO2 particles on quartz fibers is
primarily due to electrostatic interaction, it is unlikely that the coating layer will withstand severe gas
and/or liquid flow conditions in large-scale continuous operation modes. To enhance the durability of
the titania coatings, fibers were often roughened before the immobilization of catalysts. However that
will inevitably result in an uneven distribution of catalyst and light along the axial direction of the
fibers. Other significant problems are the short light propagating length (less than 10 cm), and the heat
build-up in the bundled array [1], which might lead to local deactivation of the catalyst.
Here, we present a photocatalytic reactor system based on a modified design of the
abovementioned combination of side light emitting fibers and ceramic monoliths. The ‘side-light
fibers’ are evenly distributed inside a ceramic monolith structure, on the inner walls of which titania
photocatalyst is coated. The reaction system is so constructed that the hydrodynamic regimes of Taylor
flow and film flow can be realized. Because no catalyst is coated on the fibers, the emitted light can
reach the catalyst-reactant interface without being strongly attenuated by the solid particles. Compared
with conventional OFR’s, this unique configuration provides extra design flexibilities because the light
propagation process from the source to the catalyst-reactant interface is decoupled from the physical
properties of the catalyst. Furthermore, in contrast with reactors based on coatings directly on quartz
fibers this allows a much easier catalyst preparation. Furthermore, catalyst deactivation by a
potentially strongly localized light emission as a result of fiber roughening can be avoided. To
investigate its application in organic synthesis, we chose the selective photo-oxidation of cyclohexane
as a model reaction, a reaction we previously investigated using conventional slurry reactors [21].
Experiments were performed in the film flow regime, with other conditions being comparable to our
previous study [21]. The overall photonic efficiency achieved with the IIMR reactor is discussed on
the basis of that obtained with respectively a top illumination reactor, annular reactor, and reactor
configuration with side-light fibers immersed in a TiO2 slurry.
6.2 Experimental
Materials
Cyclohexane, cyclohexanol and cyclohexanone were supplied by Merck. Anhydrous hexadecane,
used as the internal standard for gas chromatography, was purchased from Aldrich. All commercial
chemicals were of analytically pure grade and dried over a molecular sieve, prior to experiments. The
titania photocatalyst, Hombikat UV100, was kindly supplied by Sachtleben GmbH, Duisburg,
Germany, and dried at 150°C in static air, overnight, before use. Double distilled water was applied
during the sol-gel coating of titania on the cordierite monolith structure. A 25 cpsi (cells per square
inch) ceramic monolith (cordierite, 2Al2O3·5SiO2·2MgO, 30% porosity), supplied by Corning, New
York, was used in the present study. The actual dimension of the monolith structure is 43 mm in
diameter and 250 mm in length. Titanium(IV) isopropoxide Ti(O-iC3H7)4 of 97% purity, which
transforms into a chemical binder for Hombikat after hydrolysis, was purchased from Aldrich.
Concentrated nitric acid of 65% purity was supplied by Merck. Nitrogen (99.99%, Hoek Loos) and
Chapter 6
130
instrumental air (99.95%, Hoek Loos) were passed through a silica gel bed, before use in reactor
performance testing.
TiO2 coating on monoliths
An all-titania washcoating method was developed and applied to immobilize titania on the inner
walls of monolith channels (Figure 1). Titanium(IV) isopropoxide Ti(O-iC3H7)4 was used as the
precursor for the titania sol where 0.3 mol of Ti(O-iC3H7)4 was added slowly to 1000 ml of double
distilled water at 40ºC. The speed of addition was adjusted to 1 mL/min using a precision peristaltic
pump. A transparent TiO2 colloidal solution was obtained by consecutively adding 0.15 mol of
concentrated HNO3 dropwise, to catalyse the hydrolysis. The solution was heated to 80°C and
maintained for 16 hours under vigorous stirring. The final pH of the gel was 1.6. Afterwards, 100 g of
the commercial TiO2 sample (Hombikat, UV100) was added to 500 g of the synthesized gel. A stable
slurry was obtained by homogenizing the mixture with a high velocity mixer at a speed of 10,000 rpm
for 15 minutes. The particle size distribution of the slurry thus obtained was examined by forward light
scattering with a Malvern 2600 Mastersizer M1.2, equipped with a He/Ne laser, showing a mean
particle size of approximately 640 nm.
Titanium isopropoxide
H2O, HNO3, pH = 2
Aging
Stable sol
Hombikat
Monolith dipping
Calcination
Slurry
Repeat
Drying at 150°CCatalyst SC2
Catalyst SC1Drying at 150°C
Figure 1. Immobilization procedure of commercial Hombicat catalyst on monolith channels
Prior to washcoating, the monolith block was dried at 150°C for 24 hours in static air. After
cooling down to room temperature, it was dipped into the titania solution and held for 10 min to
provide sufficient time for diffusion of sol into the porous monolith walls. The monolith was
withdrawn from the slurry, and the remaining solution inside the channels was gently blown out using
Novel photocatalytic monolith reactor
131
pressurized air, followed by applying a nozzle connected to a hair-dryer and blowing a warm airflow
from sequentially alternated directions into the channels. The final drying program of the monolith was
to elevate the temperature in a static air oven to 150°C at a heating rate of 0.2 K/min, followed by air
treatment at this temperature for 24 hours. The whole washcoating procedure was repeated several
times to vary the thickness of the deposited TiO2 layer [22]. Selected pieces of coated monoliths were
cut off and gold-sputtered for determination of coating thickness and surface morphology by scanning
electron microscopy (SEM).
Testing of the catalyst powders
In order to get more information on the physical and catalytic properties of the titania coating, two
powdered samples were prepared from two different stages of the slurry synthesis. One was obtained
by drying a fraction of the gel before addition of the Hombikat particles, under the aforementioned
heating conditions (sample SC1). The other sample, most representative of the structure of the layer
deposited on the monolith walls, was obtained from the thermal treatment of the slurry after addition
of the Hombikat catalyst, yielding sample SC2. Both were subjected to analysis with X-ray diffraction
(XRD) and nitrogen physisorption, and evaluated for catalyst activity in the so-called annular and “top
illumination” reactors, described elsewhere [21], holding a slurry density of 1 g/L. From previous
studies, [21], it was observed that Pyrex transmittance starts at 275 nm thus removing the highly
energetic UV radiation which activates cyclohexane directly, leading to photolysis. To allow a more
direct comparison of the use of optical fibers as light source with the IIMR and slurry based systems, a
slurry system based on the Top Illumination reactor was used, equipped with the optical fibers
immersed in the slurry. The reactor contained 100 mL of cyclohexane, hexadecane as the internal
standard, and 1g/L of catalyst. Air, pre-saturated with cyclohexane was bubbled through the liquid at a
flow rate of 30 mL/min. GC samples were taken from the organic phase and analysed in a gas
chromatograph with a flame ionization detector (Chromopack CPwax 52CB), as described elsewhere
[21].
Internally illuminated monolith reactor (IIMR)
The schematic diagram of the internally illuminated photocatalytic monolith reactor is shown in
Figure 2. The reactor consists of an UV/Vis light source, a standard quartz fiber guide, connected to
the specially designed side light fiber bundle, the titania coated ceramic monolith block, a liquid inlet
with spray nozzle, a gas inlet section and a bottom section for gas-liquid separation and outlet. The
titania photocatalyst was coated on the inside of the square channels of the applied monolith using the
aforementioned washcoating procedure. Two side light fibers were inserted through each full channel.
Figure 3 gives a visual impression of the fibers entering the monolith channels, together with the
properties of the monolith and side light fibers. The original side light fiber bundle, type SLS200T,
was supplied by Fibertech GmbH, Berlin, which was designed to have a significant side light emission
over a fiber length of 35 centimeters. In order to enhance the refracted light intensity, the tip of the
fiber was polished and coated with a reflective aluminium coating. The side light fibers were bundled
together and connected to a normal quartz fiber guide through a diffuser. The UV radiation source was
a 100 W mercury short arc lamp (HBO R103W/45, Osram) assembled in a closed case with air cooler,
shutter and timer.
Chapter 6
132
The liquid cyclohexane was recirculated through a reservoir and the reactor with a variable speed
gear pump and the flow was measured via a turbine flow sensor. The temperature was maintained at
50°C using an external thermostat. The spray nozzle distributed the liquid over the monolith, of which
the distance between nozzle tip and the monolith was so adjusted that an even distribution of liquid
over the channels was achieved. The monolith contained 16.7 g of TiO2, while a volume of 800 ml was
circulated through the system. Air and liquid flow rates were adjusted such that the reactor was
operating in the film flow regime. Air was pre-saturated with the liquid reactant and supplied with a
mass flow controller. During reaction liquid samples were taken and analyzed twice, using a gas
chromatograph with a flame ionization detector (Chromopack, CPwax52CB). Quantification of the
oxygenated products in the liquid phase was derived from a multipoint calibration against the internal
standard.
UV/Vis
source
Gas
Liquid inlet
Gas inlet Gas inlet
Liquid outlet
Gas outletGas outlet
B
C
Sid
e lig
ht fiber bundle
35 c
m
Diffu
ser
5 c
m
Quart
z fib
er
guid
e
60 c
m
A
fiber
cordierite
TiO2
washcoat
liquid
Figure 2. Internally illuminated photocatalytic monolith reactor: A) fiber optic bundle; B) IIMR;
C) monolith channel cross section view.
Novel photocatalytic monolith reactor
133
Side Light Fiber
Diameter [mm] 0.45
Length [mm] 350
Nr of fibers 100
Monolith
Cell density [cell.in-2] 25
Length [mm] 250
Diameter [mm] 43
Channel shape square
Void fraction [%] 66
Pitch [mm] 5.08
Wall thickness [mm] 0.89
Number of full channels 44
Surface/Volume ratio 650
Void fraction [%] 68.1
Figure 3. Properties of the side light fibers and monolith used. Side light fibers placement in
monolith channels.
Evaluation of reactor performance
For a proper quantitative comparison of the performance of the various reactors used in this study,
the light intensity of the light sources of the respective reactors was determined, using a calibrated
UV-Vis spectrophotometer S-2000 (Avantes) equipped with a cosine collector. For the Annular and
Top illumination reactor a single point measurement at the approximate distance between the lamp and
the liquid cyclohexane slurry was performed, and the assumption made that the light intensity was
constant over the whole illumination window in contact with the liquid cyclohexane slurry. The
emitted intensity of the optical fibers was determined by measuring the refracted light intensity along
the fiber length.
The output generated by the spectrophotometer consists of the so-called spectral irradiance (Iλ in
Chapter 6
134
J.s-1.m-2.nm-1), which is obtained as a function of wavelength. To determine the incident light intensity,
Iλ is multiplied by the corresponding wavelength, yielding the irradiance, I (J.s-1.m-2). The photon
intensity or photon irradiance (Ip in N. s-1
.m-2
) can then be calculated using the following equation:
p
p
I II
E hc
λ= = (1)
in which I is the irradiance (J.s-1.m-2), Ep (J/photon) the energy of one photon at the specific
wavelength, λ (m), c the speed of light (m.s-1), and h Planck’s constant (J.s). Notice that Equation
1 is the conversion of energy to the number of photons by dividing the irradiance by the energy of one
photon at each specific wavelength. Finally, dividing by Avogadro’s number (NAV) we obtain Ip in
Einst.m-2.s-1. Finally the total irradiance is obtained by summing Ip for each wavelength measured. In
order to compare between reactors we need to determine the photon flow, ρp (Einst.s-1), by multiplying
for each reactor with the area of the window that is used to transfer the light from the light source into
the reactor. In Figure 4 the four reactor configurations studied are shown. In the Top Illumination
reactor the area of the window that is in the top of the reaction vessel was used (black circle). For the
annular slurry reactor the area of the cooling vessel around the lamp was used. For the side light fiber
reactor and the IIMR the external area of the fiber bundle was used. The experimental photonic
efficiency (ξ) was calculated for each reactor and is defined as:
( )
( / )
nol none inin
AVp
d n n
R dtd N N
dt
ξρ
+
= = (2)
Where Rin (mol.s
-1) is the initial reaction rate (first 60 minutes, where catalyst deactivation is not yet
dominating performance) of the analyzed products (cyclohexanone and cyclohexanol) and ρp is the
photon flow (Einst.s-1). This equation defines the number of reacted molecules per number of photons
(quanta, N) [23].
lamp
a b clamp
Fiber
bundle
Figure 4. Different reactor configurations compared. From the left to the right, top illumination
slurry reactor, annular slurry reactor, side light fiber reactor and monolith channel of the IIMR: a)
fiber optic, b) liquid, c) catalyst layer and d) monolith wall. The grey area in the channel shows the
illuminated area of the reactor.
Novel photocatalytic monolith reactor
135
6.3 Results
Characterization of photocatalysts
The morphology and thickness of the titania coating on the monolith channels after each
consecutive coating step was analyzed by Scanning Electron Microscopy, yielding images as shown in
Figure 5a. Figure 5a shows the layer obtained after the final deposition step (i.e. the 82 µm coating).
From the top view image, it is clear that the coated layer consists of agglomerates of about 1-3 µm.
The side view in Fig. 5a shows the macroporosity of the cordierite monolith, with an average pore size
of 5 µm, and on top the coated TiO2 layer. The thickness of the deposited layer, as derived from the
respective SEM images, is plotted as a function of the weight fraction of catalyst in Figure 5b. The
points lie on a straight line, which does not intercept with the origin. The explanation is as follows.
The average particle size of the slurry used for coating was roughly 640 nm, as mentioned before,
being significantly smaller than the macropore size of the monolith walls. So, it has to be expected that
when the monolith was dipped in the titania slurry for the first time, a large amount of catalyst is
absorbed by the cordierite pores and is retained inside the monolith structure after calcination. This
fraction of the deposited catalyst does not contribute to the formation of the catalyst layer. Apparently,
the cordierite macropores are completely filled at a catalyst weight fraction of about 2%, followed by
the formation of over-layers and the linear trend observed in Figure 5b at higher catalyst weight
fractions.
a) b)
Figure 5. a) SEM micrograph of titania coating on monolith channels: in the top is the top view
of titania coating (6×) and in the bottom the side view of titania coating (6×); b) effect of repetition of
coating on the titania layer thickness as determined from SEM.
coatingcoating
coatingcoating
0
30
60
90
0 2 4 6 8
Catalyst Loading [%]
La
ye
r T
hic
kn
es
s [
µm
]
Pore filling
Chapter 6
136
Figure 6 shows the XRD patterns of: a) SC1, b) SC2, c) the starting material (Hombikat), which is
a pure anatase catalyst, and d) P25 TiO2, which consists of approximately 70% Anatase and 30%
Rutile [21]. Anatase is the only detectable phase in all samples, as can be derived from the
characteristic diffraction lines, and absence of a Rutile signature. The crystal size as determined by the
Scherrer equation is 5.4 nm for Hombikat, and 8.6 nm and 8.1 nm for SC1 and SC2, respectively. The
value of SC1, i.e. the dried gel formed by the acid catalyzed hydrolysis of Titanium(IV) isopropoxide
(Ti(O-iC3H7)4), is comparable with the value reported by Dey et al. [24], who applied a similar
procedure to obtain TiO2 with an average primary particle size of 8.5 nm. The nitrogen
adsorption-desorption isotherms, and the corresponding surface areas for SC1, SC2, and Hombikat
UV100 are shown in Figure 7. Of the three samples, SC1 has the lowest surface area (220 m2/g), while
addition of Hombikat UV100 to the hydrolysis leads to the observed higher surface area of 290 m2/g,
which is approximately the average of the value for Hombikat and the hydrolysed Titanium(IV)
isopropoxide material.
10 20 30 40 50 60 70
2θθθθ [
Inte
ns
ity
[A
.U.]
a
b
c
Figure 6. X-ray diffraction patterns of catalyst SC1 (a), SC2 (b), Hombikat UV100 (c) and
Degussa P25 (d).
Figure 7. Nitrogen adsorption-desorption isotherms (left) and the corresponding surface area
calculated through BET method (right) of catalyst SC1 (a), SC2 (b) and Hombikat UV100 (c).
p/p0
[-]
0.0 0.2 0.4 0.6 0.8 1.0
Vad
s [
cm
3/g
ST
P]
0
100
200
300
a
b
c
SB
ET [
m2/g
]
0
100
200
300
400
a
b
c
[ °]
Novel photocatalytic monolith reactor
137
Light emission characteristics of the side light optical fibers
While for the Annular and Top illumination reactor a single point measurement was performed to
determine the incident light flux, the side light emission from the fiber is a function of the position
along the fiber, as is illustrated in Figure 8. This figure shows that the relative light intensity emitted
from the side is quickly attenuated in the first 10 cm, while at the end of the fiber (at 35 cm), the light
intensity is relatively small. The light intensity change follows an exponential decay suggesting a
Beer’s law correlation between the side light intensity, Iside, and the input light intensity entering the
front tip of the fiber, Iinput, as expected. It should be mentioned that although the light emitted from the
side diminishes significantly, a considerable amount of light is emitted through the end point (Itrans).
The energy balance of the light flux in the optical fiber can be described by the following equation (3):
input side transI I I= + (3)
where Itrans is the residual intensity transmitted from the rear tip of the optical fiber. When the end-tip
of the fiber is coated with a reflective material the value of Itrans is almost zero and the light that enters
the input side of the fiber is almost all emitted through the sides. This is clear from the profile of the
side light intensity obtained after tip-coating with a reflective material shown in Figure 8. Not only
does the tip provide for a more even distribution of the light emission along the fiber length, the total
energy emitted through the sides of the fiber is enhanced from 40% to more than 95%. In this way
more light is available to the catalyst.
Figure 8. Side light emission as a function of the fiber length measured from diffuser. Side light
intensity at each distance is correlated to the initial side light intensity measured at the diffuser
entrance. Insert indicates the way of measurement.
Distance from diffuser [cm]
0 10 20 30
Re
lative
lig
ht
inte
nsity [
-]
0.0
0.5
1.0side light fibers without tip-coating
tip-coated side light fibers
Washer Washer
Detector
Side light
fiber bundle
Chapter 6
138
Photocatalytic oxidation of cyclohexane
Based on the high selectivity reported in earlier studies on the selective photo-oxidation of
cyclohexane [21], only cyclohexanone production was considered in the comparison in performance of
the powdered catalysts, SC1, SC2 and Hombikat, respectively. As can be seen from Figure 9,
Hombikat UV100 (c) gives the highest cyclohexanone yield, followed by SC2 (b) and SC1 (a).
Although various factors (crystal morphology, phase composition, hydroxyl group density) play a role,
for the present study it is sufficient to state that these results are in agreement with the trend observed
in surface area (compare with Figure 7b). Mixing Hombikat UV100 with the synthesizing gel, as well
as preparing TiO2 directly from the synthesizing gel, leads to a lower surface area and as a result a
corresponding lower activity.
The activity profiles to cyclohexanol and cyclohexanone for the Internally Illuminated Monolith
Reactor (IIMR) are shown in Figure 10. The cyclohexanone yield in this reactor is approximately 10
times lower as compared to the performance of the powdered catalyst representative of the coating
(SC2) in the Top illumination reactor (Figure 9b). Furthermore, the obtained cyclohexanol selectivity
is significantly higher in the IIMR, as compared to those typically reported for other reactor
configurations [21]. The activity curve of the immersed optical fiber reactor is not shown, but shows at
least an order of magnitude lower product yields. Table 1 provides an overview of the performance of
Hombikat in the photo-oxidation of cyclohexane achieved in the various reactors used in this study.
The Annular, Top illumination reactor, and the side light fiber reactor are slurry systems, and the IIMR
is an immobilized system. The incident light flux, Ip, as well as the photonic efficiency, ξ, (mol.Einst.-1)
are reported. By definition, the photonic efficiency of a radiation-induced process is the number of
times that a defined event (in this case a chemical reaction step) occurs per photon absorbed by the
system. The affectivity of the various reactors shows the following trend: top illumination reactor >
IIMR >> Annular reactor > the reactor configuration with side-light fibers immersed in the TiO2 slurry.
Figure 9. Photocatalytic production of cyclohexanone from neat cyclohexane on various slurry
catalysts. (a) SC1; (b) SC2; (c) Hombikat UV100. Lines are presented to guide the eyes.
Reaction time [min]
0 100 200 300 400
Cyclo
he
xa
no
ne f
orm
ati
on
[m
mo
l]
0.0
0.2
0.4
0.6
0.8
a
b
c
Novel photocatalytic monolith reactor
139
Figure 10. Product formation of photocatalytic oxidation of cyclohexane performed in internally
illuminated monolith reactor. Lines are presented to guide the eyes and have no fundamental
mechanistic meaning.
Table 1. Comparison of quantum yield in different reactor configurations.
Reactor R
in
[mol.s-1
]
Ip
[Einst.m-2
.s-1
]
ρp
[Einst.s-1
]
ξ
[mol.Einst.-1
]
Annular reactor 1.59×10-6
3.90×10-3
1.96×10-4 0.008
Top illumination
reactor 1.20×10
-7 3.20×10
-4 7.95×10
-7 0.151
Side light fiber
reactor 7.68×10
-10 4.70×10
-5 3.69×10
-7 0.002
IIMR 2.28×10-8
4.70×10-5
3.69×10-7 0.062
6.4 Discussion
Coating procedure and catalyst activity
In the present paper a procedure is described with which a mechanically stable layer of Titania
Hombikat UV100, composed of pure anatase, can be obtained on the walls of a ceramic monolith
support. No loss of the titania coating was observed during the photocatalytic experiments in the
IMRR at loadings as high as 7 wt-%, proving that the TiO2 layer is sufficiently strong to withstand the
Time [min]
0 50 100 150 200 250 300
Pro
du
cti
on
[m
mo
l]
0.00
0.02
0.04
0.06
Cyclohexanone
Cyclohexanol
Chapter 6
140
most severe conditions in our operating window. As a result of the coating procedure, the surface area
of Hombikat TiO2 is reduced by about 15%, leading to a corresponding decrease in cyclohexanone
yield (Figure 9). In immobilizing the catalyst on the monolith ca 20 % of titania is lost in the
macropores of the cordierite structure, which will not contribute significantly to photocatalytic activity,
as will be discussed in the following.
Comparison of the Photonic Efficiencies
To explain the order in photonic efficiencies as observed in Table 1 for the various reactors
applied in the present study we should address to the various reactor configurations illustrated in
Figure 4. Clearly in the annular reactor the slurry is surrounding the light source while in the top
illumination reactor the light enters the slurry through a hole in the upper part of the reactor. Based on
this, the low photonic efficiency of the annular reactor compared to the top illumination reactor is
unexpected, with the top illumination reactor showing a photonic efficiency an order of magnitude
higher. This is explained by the very high light intensity applied in the annular reactor. It is to be
expected that the photon flow is far beyond the regime where the reaction rate varies linearly as a
function of light intensity, as assumed in comparing the photonic efficiencies [25]. In other words, a
large fraction of the photons sent into the reactor are not effective for reaction. This is related to the
higher probability of recombination of activated states in TiO2 (electrons and holes) at these high
photon flows, as has been reported by Hermann and coworkers [23]. Another reason which should not
be discharged is the fact that in the annular reactor a relatively thin slurry “layer” is present between
the lamp and the vessel wall. Considering this, it is probable that the light is by-passing the catalyst
and exiting the reactor without being used. Measurements are currently performed to quantify the
photon flow exiting the annular slurry reactor.
If the lamp illuminating the reactor from the top, is replaced by immersed side light fibers, as
illustrated in Figure 4, an order of magnitude drop in photonic efficiency is observed. Two factors are
responsible for this difference. First, the concentration as determined by the GC in the liquid is
multiplied by VR to establish Rin. These concentrations are significantly lower for the immersed fiber
reactor as compared to the top illumination reactor. We have established by ATR-FT-IR experiments
that a considerable amount of cyclohexanone is present as absorbed on the catalyst particles, with the
ratio of cyclohexanone in solution vs cyclohexanone adsorbed (chonsol/chonads) strongly increasing as a
function of increasing total cyclohexanone concentration. In other words, the rate of formation of
cyclohexanone, and reported photonic efficiency, is more extensively underestimated in the immersed
fiber reactor, and should be considered as a lower limit. It is estimated that this will account for about a
four times difference in photonic efficiency. The other reason for this difference in photonic efficiency
is again related to the reaction rate dependency on photon flow. At the value of the photon flow
reported for the top illumination reactor, we suspect that the dependency is of ½ order [23,25].
Obviously this will diminish the calculated photonic efficiency for the top illumination reactor.
Compared to the immersed fiber reactor, a significant improvement in performance is obtained by
structuring the fibers in the monolith channels, and immobilization of the catalyst on the monolith
walls. In the case of the IIMR, the catalyst is better exposed to the light coming from the light source,
and a much larger fraction of the reactor volume is effectively used to convert cyclohexane to
cyclohexanone.
Further comparison of the data of Table 1 shows that the photonic efficiency of the top
Novel photocatalytic monolith reactor
141
illumination reactor is the highest of all reactors studied. However, again the argument of product
adsorption should be considered in comparing the performance with the IIMR. A large amount of TiO2
(~17 g) was present on the monolith, and hence a large amount of produced cyclohexanone was not
quantified by just measuring the product concentration in solution. Furthermore, there is an optimal
ratio of layer thickness and light intensity [26]. Formenti et al. showed that 99% of the light absorption
occurred within a 4.5 µm powder layer of TiO2 [27], which suggests that the layer thickness applied in
the present study (~80 µm) can be significantly reduced without loosing photo-active TiO2, enhancing
the apparent photonic efficiency, and providing for the best reactor configuration. As a final note, we
like to state that in the present study the performance of the slurry reactors was compared on the basis
of a light source configuration (position, intensity) which was obviously far from optimized. Certainly
improvements in this configuration will also lead to improvements in the photonic efficiencies
reported for these reactors.
Selectivity
At first sight it is surprising that the cyclohexanone/cyclohexanol ratio is close to one in the IIMR,
whereas the ratio generally reported in the literature is typically approaching infinity [21]. Interference
of deep UV exposure (below 250 nm), which was shown to induce radical chemistry, leading to a high
selectivity to cyclohexanol, can be excluded, since cyclohexanol nor cyclohexanone were detected
after 300 min of reaction in a photolysis experiment with the optical fibers without catalyst present
(not shown). Although the very low product yields in the monolith make it difficult to draw firm
conclusions, the low photon flow of the fibers might be the key to the explanation of the observed
phenomena. At the resulting very low product concentrations, cyclohexanone adsorption phenomena
and possibly a different surface reaction selectivity might occur.
Coating fibers vs. monolith walls
A monolith configuration provides a high geometrical surface area to support catalysts on, but till
now only a design with coated fibers, using a monolith as a kind of straightener, has been described in
the open literature. A priori the design described in the present paper, i.e. coating the catalyst on the
walls of the monolith vs coating of the catalyst on the fiber surface, is expected to have significant
advantages, which is further illustrated in Figure 11. The governing rules to interpret the optics in OFR
are the Snell’s equation for light propagation along the fiber and the Lambert-Beer’s law for the
refracted light intensity inside the coated photocatalyst layer. In the titania layer coated on the surface
of the fibers (Figure 11a), the light transmission is in the opposite direction of the diffusion of reactants.
The light intensity is highest at the fiber-catalyst interface and attenuates with an exponential decay as
it approaches the catalyst-reactant interface. Therefore, an optimal layer thickness is typically
determined, where both sufficient light absorption and rapid reactant diffusion into the illuminated
layer are satisfied. Furthermore, the TiO2 coating on the optical fiber has two functions, namely to
catalyze the surface reaction and to reflect part of incident light back into the optical fiber. Depending
on the quartz fiber diameter and the coating material, different optimal coating layer thicknesses were
found. Furthermore, in most previous designs, the resulting effective fiber length for photocatalytic
reaction was often limited to less than 10 cm from the light incident point [14, 15]. Our design, with
Chapter 6
142
the fibers tip-coated with a reflective material, and the catalyst on the walls of the monolith, decouples
the light propagation process in the fiber from the physical properties of the catalytic layer. With the
side light fibers illuminating the coated walls of the monolith channels from the front (Figure 11b),
reactant concentration and light intensity decay in a similar direction, i.e. from the external surface
towards the ceramic wall of the monolith channel. It is to be expected that this will positively affect the
photonic efficiency. A last point of difference is related to the synthesis of the catalyst. It has been well
established that catalyst synthesis on monoliths is relatively easy. As for these types of reactors the
development of dedicated catalyst synthesis protocols is required, this advantage of the IIMR
compared to the optical fiber reactor with catalyst deposited on the fibers should not be
underestimated.
Figure 11. Schematic illustration of the catalyst activation mechanism a) in previous design of
titania coated quartz fiber and b) current design using side-light optical fiber.
Future work
Further research is ongoing within the group of Industrial Catalysis, TU Delft, to study the effect
of catalyst layer thickness, hydrodynamic properties (gas-, and liquid flow rates, Taylor flow), as well
as of reactor geometry (CPSI, cells per square inch) on the performance of the IIMR, including
quantification of the amounts of products adsorbed on the catalyst layer. Use of a smaller channel
diameter will not only benefit hydrodynamic properties, but will also reduce the distance between fiber
and catalyst, which can enhance the reactor performance. In order to reduce adsorption phenomena,
thinner coatings have to be used, and since we are facing strong product adsorption going to elevated
Reactant concentration
Light intensity
Reactant concentration
Light intensity
Optical
Fiber
Photocatalyst
Layer
Reaction
Medium
Optical
Fiber
Photocatalyst
Layer
Reaction
Medium
It
Ii
Ir It
Ii
Ir
a) b)
Novel photocatalytic monolith reactor
143
temperatures is also an option to consider. Furthermore, experiments are conducted to directly compare
the IIMR with a coating on optical fibers, to further quantify our discussion of the previous paragraph.
The concept of the IIMR will also be applied in other reactions and catalyst combinations.
6.5 Conclusions
The following conclusions can be derived from the work described in this chapter:
A novel reactor for multi-phase photocatalysis is presented, the so-called Internally
Illuminated Monolith Reactor (IIMR). In the concept of the IIMR, side-light emitting fibers are placed
inside the channels of a ceramic monolith, equipped with a TiO2 photo-catalyst coated on the wall of
each individual channel.
The applied coating procedure and Hombikat TiO2 as starting material lead to a high surface
area and pure anatase titania layer attached to the monolith walls.
The Photonic Efficiency achieved with the IIMR reactor in the selective photo-oxidation of
cyclohexane (0.062) is less than the one obtained with a top illumination reactor (0.151), whereas an
Annular reactor and reactor configuration with side-light fibers immersed in a TiO2 slurry reach a
photonic efficiency of only 0.008 and 0.002, respectively.
The Photonic Efficiency can be further optimized by reducing the layer thickness of TiO2 on
the monolith walls.
The new design introduces a broad range of possibilities in the photocatalytic reactor
research field.
Chapter 6
144
References
1. D.F. Ollis, H. Al-Ekabi (Eds.), Photocatalytic purification and treatment of water and air,
Elsevier, Amsterdam, 1993
2. Linsebigler, A.L., Lu, G., Yates, T., Chem. Rev., 1995, 95, 735
3. Mills, A., Le Hunte, S., J. Photochem. Photobiol. A: Chem., 1997, 108, 1
4. Carp, O., Huisman, C.L., Reller, A., Prog. Solid State Chem., 2004, 32, 33
5. Serpone, N., Borgarello, E., Harris, R., Cahill, P., Borgarello, M., Pelizzetti, E., Sol. Energy
Mat., 1986, 14, 121
6. Matthews, R.W., J. Catal., 1988, 113, 549
7. Ray, A.K., Beenackers, A.A.C.M., AIChE J., 1998, 44, 477
8. Dijkstra, M.F.J., Buwalda, H., De Jong, A.W.F., Michorius, A., Winkelman, J.G.M.,
Beenackers, A.A.C.M., Chem. Eng. Sci., 2001, 56, 547
9. De Lasa, H., Serrano, B., Salaices, M., Photocatalytic Reaction Engineering, Springer, New
York, 2005
10. Schiavello, M., Heterogeneous Photocatalysis, Wiley, Chichester, 1997
11. Hofstadler, K., Bauer, R., Novalic, S., Heisler, G., Environ. Sci. Technol., 1994, 28, 670
12. Tada, H., Honda, H., J. Electrochem. Soc., 1995, 142, 3438
13. Peill, N.J., Hoffmann, M.R., Environ. Sci. Technol., 1995, 29, 2974
14. Choi, W., Ko, J.W., Park, H., Chung, J.S., Appl. Catal. B, 2001, 31, 209
15. Wang, W., Ku, Y., Chemosphere, 2003, 50, 999
16. Rice, C.V., Raftery, D., Chem. Commun., 1999, 10, 895
17. Sun, R.D., Nakajima, A., Watanabe, I., Watanabe, T., Hashimoto, K., J. Photochem.
Photobiol. A: Chem., 2000, 136, 111
18. Danion, A., Disdier, J., Guillard, C., Abdelmalek, F., Jaffrezic-Renault, N., Appl. Catal. B:
Environ., 2004, 52, 213
19. Lin, H.F., Valsaraj, K.T., J. Appl. Electrochem., 2005, 35, 699
20. Kreutzer, M.T., Kapteijn, F., Moulijn, J.A., Heiszwolf, J.J., Chem. Eng. Sci., 2005, 60, 5895
21. Du, P., Moulijn, J.A., Mul, G., J. Catal., 2006, 238, 342
22. Yin, S., Hasegawa, H., Maeda, D., Ishitsuka, M., Sato, T., J. Photochem. Photobiol. A, 2004,
163, 1
23. Herrmann, J.M., Catalysis Today, 1999, 53, 115
24. Dey, S., Ray, M., Banerjee, P., Inorg. React. Mech., 2000, 24, 267
25. De Lasa H., Serrano, B., Salaices M., (Eds.), Photocatalytic Reaction Engineering, Springer,
USA, 2005
26. Aguado, M.A., Anderson, M.A., Hill, C.G., J. Molecular Cat., 1994, 89, 165
27. Formenti, M., Julliet, F., Meriaudeau, P., Teichner, S.J., Chem. Tech., 1971, 1, 670
Novel photocatalytic monolith reactor
145
Chapter 6
146
147
7
Conclusions and Outlook
Chapter 7
148
7.1 Conclusions
Heterogeneous photocatalysis describes a process whereby illumination with UV-visible light of
energies larger than the bandgap energy (≥Eg) of a semiconductor, most commonly TiO2 based,
generates thermalized conduction band electrons (e-) and valence band holes (h+) which, subsequent to
their separation process, are poised at the catalyst/reactant interface to initiate catalytic reactions. It has
been a fascinating field in the scientific world of catalytic research since the discovery of
photocatalytic properties in the early seventies, of which various topics in both fundamental and
applied chemistry have been studied, and rapid development has been booked. The objective of the
research described in this thesis was to revisit photocatalysis from mainly an experimental,
engineering point of view whilst tackling some of the theoretical fundamentals of photocatalyis, and to
make the first move towards the commercialization of photocatalytic system in the conventional
chemical industry, of which the application of photons might induce a potential breakthrough.
A novel reaction assembly for high throughput photocatalytic experimentation (HTPR) was
constructed, which allows parallel catalyst screening of up to 10 different catalysts. The equivalency
and the applicability in photocatalytic dye decolorization processes were verified by light irradiance
measurements as well as photocatalytic reactions. The optimal testing conditions for photocatalyst
screening were determined experimentally, which is specified to be methylene blue decolorization with
0.5 g/l of photocatalyst in 53-75 µm size with all irradiation sources “on”, equivalent to a power output
of 456 µW/cm2.
The effect of TiO2 source and thermal pre-treatment on photoactivity in dye degradation in water
was investigated. Photocatalytic decolorization of methlyene blue on TiO2 photocatalyst is found to be
a highly complex system. The reaction was found to follow apparent 1st order kinetics for all TiO2
materials studied, which is the simplified approach of Langmuir-Hinselwood mechanism with strong
competitive adsorption of competitor and low concentration of adsorbent (chapter 2). Combined with
the analysis of other dye decolorization experiments, it is concluded that the total surface area and the
associated amount of surface hydroxyl groups present, are the most important parameters for the
photocatalyst activity. Unlike the degradation of other dye molecules, the apparent decolorization rate
of methylene blue on high surface area TiO2 photocatalysts is limited by the internal diffusion of bulky
methylene blue molecules into the porous agglomerates. This also explains the positive effect of a
thermal pre-treatment temperature up to ~1100 K, reducing overall surface area, but enhancing
accessibility.
A TiO2 photocatalyst supplied from Merck exhibited extraordinary high reaction rates in
methylene blue decolorization. This reactivity was not found in the photocatalytic degradation of other
organic dyes, nor in the cyclohexane photo-oxidation reaction. Apparently, a certain specific
reactant/catalyst surface interaction plays an important role in enhancing the reaction rate for
methylene blue decomposition. The presence of alumina impurities in the Merck TiO2 could be the
explanation, altering the mode of methylene blue adsorption and the degradation pathway by
replacement of Ti atoms in the titania framework by Al atoms.
Another method to improve the photocatalytic activity of TiO2 is through doping of “foreign”
elements into the pure TiO2 crystalline structure. Commercial TiO2 (P25 from Degussa) was doped
with rare earth metals of La, Ce, Zr, Y, Pr and Sm, and the activity of the samples as a function of
calcination temperature was tested in methylene blue photocatalytic degradation. Results show that
doping of P25 with rare earth oxides such as La, Ce, Y, Pr, and Sm prevents the complete anatase to
Conclusions and outlook
149
rutile phase transformation upon calcination at 800 ºC, positively affecting the remaining BET surface
area. The photocatalytic degradation rate is mainly dependent on the quantity of a specific Ti-OH
group in the applied P25 catalyst, which is most likely the strongest adsorption-, and/or most effective
photo-reactive site. The quantity of this site is also affected by the extent of dispersion and loading of
the rare earth oxide.
Liquid phase selective photo-oxidation processes for organic synthesis have been much less
explored than water treatment technologies. In this thesis photo-oxidation of cyclohexane was used as
a test reaction to evaluate the potential of photocatalysis in selective oxidation. The wavelength of the
applied radiation was found to play an important role in determining both the reaction rates and
reaction pathways. Under the uncatalyzed photo-oxidation region (λ < 275 nm), cyclohexanol is the
major product. This is the result of a direct radical chain reaction, i.e. photolysis. If photolysis is
prevented by the use of the proper light filters (e.g., Pyrex, λ > 275 nm), the reaction rate is suppressed
unless TiO2 photocatalyst is added to the system and reaction proceeds through photocatalytic
pathways. Pure photocatalysis over TiO2 yields predominantly cyclohexanone with a ketone/alcohol
selectivity over 95%. The activity towards ketone formation was affected by catalyst structure, with
surface hydroxyl group density being the most important parameter. Based on the reactive studies
under various reactor configurations, reaction conditions, and varied light source, and numerous
associated catalyst/product analysis methods, a preliminary reaction mechanism is proposed involving
the light-induced formation of surface cyclohexyl radicals, followed by the formation of a peroxide
intermediate and decomposition to cyclohexanone and desorption. Accumulation of cyclohexanol on
the TiO2 surface is proposed to deteriorate the photocatalytic activity and to contribute to CO2
formation being a less desired product.
Another incentive to initiate the photocatalytic research in the catalysis engineering group is the
poor applicability of photocatalytic systems for chemical production processes on the industrial scale.
Conventional slurry type photo-reactors have typical drawbacks such as difficult catalyst separation,
low light utilization efficiency, and potential for scale-up. In this study we have developed a novel
reactor for multi-phase photocatalysis, the so-called Internally Illuminated Monolith Reactor (IIMR).
In the concept of IIMR, side-light emitting fibers are placed inside the channels of a ceramic monolith,
equipped with TiO2 photocatalyst coated on the wall of each individual channel. In this way, a high
illuminated catalyst surface area per reactor volume can be achieved. Furthermore, generally
recognized advantages of monotlith reactors, such as fast mass transfer rates and ease to scale up, can
be expoited. A novel washcoating procedure was developed that resulted in a steady and uniform TiO2
layer on the walls on monoliths channels. The preliminary estimated photonic efficiency achieved with
the IIMR reactor in the selective photo-oxidation of cyclohexane is less than the one obtained with a
small size top illumination slurry reactor, but higher than those obtained in an annular reactor of
similar size and reactor configuration with side-light fibers immersed in a TiO2 slurry. Based on SEM
micrographs and weight measurements of the TiO2 coatings on the monolith, it is evident that the total
amount of TiO2 on the monolith is far beyond optimal, so that it effectively reduced the apparent
photonic efficiency through light shielding effects and a large amount of products remaining adsorbed
on the TiO2 layer. The apparent photonic efficiency can be further optimized by reducing the layer
thickness of TiO2 on the monolith walls. Furthermore it should be addressed that the light source
configuration in the presented study is far from optimized (position, intensity). Improvements in
configuration will certainly lead to enhancements in the photonic efficiencies reported for the IIMR.
Chapter 7
150
7.2 Outlook
The results of this thesis show that heterogeneous photocatalysis is a promising field that has potential
in pollution control, wastewater treatment and possibly organic synthesis. Despite the tremendous
research efforts since its discovery, the effect of material composition and engineering parameters
(pressure, temperature, reactant concentrations, light intensity) on photocatalytic activity remain
poorly understood, due to the very diverse model reactions and process conditions applied in the
literature. In particular the extra variable compared to conventional catalytic systems – light – brings
an additional degree of freedom and complexity in both fundamental research and applied reactor
studies. Some of the earlier works on semi-conductor photosystems appear to describe dubious results,
sometimes even contradictory to earlier publications, without proper discussion. This has not helped
the subject to develop as rapidly as it might have, and may have generated to some degrees skepticism,
in particular in the catalytic community.
A typical example is the methylene blue decolorization process in water, which has been widely
applied as a standard test reaction for photocatalyst screening because of the simplicity and ease of the
dye bleaching measurement, i.e. it has been proposed by the Japanese standardization committee to be
the standard reaction for testing of photocatalyst activities in aqueous phase. However, the
decomposition of methylene blue is such a complicated reaction system, the apparent reaction rate
being affected by multiple factors, that a simple comparison of activity seems an over-simplification.
In chapter 3 of this thesis, it has been proposed that the large molecular size of methylene blue and the
complicated reactant/catalyst interaction could affect, or sometimes, dominate the reaction rate, which
makes its application for catalyst kinetic screening less plausible. Furthermore, in terms of degradation
mechanism, the observed fading of the blue color is not necessarily related to the oxidation of the dye,
especially if the reaction is carried out under conditions that favour the formation of LMB
(leuco-methylene blue), i.e., conditions which include: a low, easily depleted dissolved oxygen level
and a low pH. In a separate study (Fig. 1, not covered in this thesis) the complexity of the methylene
blue decolorization is addressed.
Figure 1. Effect of oxygen on the apparent reaction rate of methylene blue decolorization, and the
proposed reaction mechanism.
In terms of photocatalyst development, it is necessary to establish a standard catalyst testing
MB
hνννν
LMB
MB
hνννν
LMB
O2O2
Pox
CO2 + H2O
hνννν O2
hννννhνννν
Different active species /
activating mechanisms involved
0
0.01
0.02
0.03
0.04
-80 -40 0 40 80 120 160 200
Time [min]
Co
nc
.MB
[m
mo
l/l]
On-site
Samesample,dark in airovernight
Light off, N2
Light on, N2
Light off, N2
Light off, O2
Light on, O2
Hombikat 0.050 g (75-53 µm)
Vliq: 0.100 l
CMB,0: 0.030 mmol/l
Temp.: 316 K
Measured
immediately
Conclusions and outlook
151
system that applies a simple reaction with well-understood kinetics, a fixed light/catalyst/reactant
configuration and well-defined reactor characteristics. The combinatorial screening assembly for
aqueous photocatalytic systems described in this thesis is an early effort in this direction, although its
inherent constraints limit its applications in a wide area of interest. Further work towards
combinatorial photocatalysis is recommended that could largely reduce the amount of work in
preparation, and minimizing the discrepancies caused by the inherent characteristics of specified
photocatalytic systems.
As is shown in chapter 3 of this thesis, photocatalytic dye decolorization, although ease in
measurement, is in fact composed of various steps and complex in its way of interpretation. The results
with one dye component can simply not be transferred or extrapolated to other dye molecules without
thorough understanding of the individual photocatalytic system. Methylene blue, which is a
representative of the thionine dyes resistant to biodegradation, has been proven to be less
representative for photocatalytic degradation of organic dyes in general. The industrial application of
photocatalysis for wastewater treatment should, therefore, consider the uncomfortable facts of complex
dye molecules with the interaction with light and catalyst, as well as the potential problems for the
scaling up of the photo-reactors. Furthermore it is found that thermal treatment and/or doping with
foreign elements of commercial TiO2 samples could have a complicated impact on their apparent
photocatalytic activity. Complete understanding of the consequences requires further study and deep
knowledge on the surface chemistry, engineering and optophysical properties.
Figure 2. Window of reality for the commercialization of photocatalytic oxidation in chemical
industry. Current lab-scale study results are indicated.
With regard to the commercialization of photocatalytic oxidation in the chemical industry, more
efforts are required in the establishment of kinetic models, improving the understanding of the
photocatalytic reaction rate, and the ability to rationally develop suitable photo-reactors. Take for
example the photocatalytic oxidation of cyclohexane, which could be a process at ambient conditions
that offers engineering benefits over conventional processes, a simplified reaction scheme is proposed
in chapter 5. Further research, using ATR-FTIR and DRIFT spectroscopy, is required to corroborate
the proposed mechanism, also taking into account the reaction pathways proposed by other
investigators. It could point out the way of improvement in terms of physicochemical properties of
catalysts, as well as enhancing the intrinsic kinetics through more efficient application of
photon-generated charge carriers. Table 2 shows the window of reality of the photocatalytic
Top illumination reactorIIMR
Present results
Reactivity (mol / (mR
3⋅s))
10-9 10-6 10-3 1
Petroleum
geochemistry
Biochemical
processes
Industrial
catalysis
Reactivity (mol / (mR
3⋅s))
10-9 10-6 10-3 1
Petroleum
geochemistry
Biochemical
processes
Industrial
catalysis
Chapter 7
152
cyclohexane oxidation process, compared with typical industrial practices. Current lab-scale
experiments could reach the reactivity ranging from biochemical process to industrial catalytic process.
Further improvement in reactivity is required towards commercialization of this process.
Another challenge is the scaling up of the current lab-scale reactors and fit to industrial
engineering requirements. The results of this thesis show that an immobilized catalyst with high
illuminated surface area could be achieved and is suitable for large-scale photocatalytic processes,
which introduces a broad range of possibilities in the photocatalytic reactor research field. Still there is
a long way to go, however, to optimize the configuration and to adjust the system so that it could meet
the generally accepted industrial standard of reactor design. Certainly the monolith concept appears an
interesting direction to follow, in particular if transparent monoliths could be applied, simplifying the
introduction of light into the system as compared to the application of optical fibers. With regard to the
irradiation source, developments could be booked in the field of increasing the energy density in
usable range, i.e. UVA for TiO2, improved design on the shape of light sources so as to fit it better for
the industrial applications, and minimizing the energy loss from the light source to the catalyst
surfaces.
153
154
155
Samenvatting
Assessment Potentials of Fotokatalysis: Dye Degradation,
Cyclohexane Photo-oxidation and Reactors
Ruim dertig jaar geleden demonstreerden de Japanse onderzoekers Fujishima en Honda de
fotokatalytische effecten met behulp van titaandioxide (TiO2) als positieve pool (anode) in een
elektrochemische cel. Sindsdien is de heterogene fotokatalyse een snelgroeiend interessegebied voor
onderzoekers en waterzuiveringsbedrijven, in gebieden voor zowel lucht- en waterzuivering als
organische synthese. In een fotokatalytisch proces vindt reactie plaats onder invloed van een lichtbron
(b.v. UVA) in de aanwezigheid van een katalysator (b.v. TiO2). Blootstelling aan UV licht heeft als
gevolg dat er elektronen (e-) vrijkomen uit het TiO2. Tegelijkertijd worden positieve gaten gevormd
(h+). De elektronen en de positieve gaten veroorzaken de vorming van superoxide (O2-) en hydroxyl
radicalen (OH·) met waterdamp en lucht, die kunnen vervolgens reageren met de organische
verbindingen zodat een kettingreactie van radicaalvorming en oxidatie wordt gestart.
Fotokatalyse systemen worden op kleine schaal toegepast voor de behandeling van lucht en
waterstromen in de industrie, en bestaan ook als airconditionings units voor luchtbehandeling van
huizen en kantoren. Uit de onderzoeksresultaten van literatuur is er echter gebleken dat veel
onduidelijkheden en inconsistenties bestaan. Ook zijn de resultaten van verschillende onderzoeken
moeilijk met elkaar te vergelijken en soms niet herhaalbaar. Dat komt door het inbrengen van licht en
de interactie van licht met katalysator, waardoor er een extra dimensie van vrijheid en complexiteit
komt voor katalytische onderzoeken. Een van het doel van dit onderzoek is om fotokatalyse te
herbekijken waardoor meer duidelijkheid over de reakties en de interaktie van licht, reaktanten en
katalysatoren aan de licht komt.
Verder zijn er op dit moment nog geen commerciele toepassingen van heterogene fotokatalyse op
het gebied van conventionele chemische processen. Een van de problemen voor het toepassen van de
methode is het ontbreken van een efficiente reactor. Daardoor is het ook doel van dit onderzoek om
een technologische oplossing te ontwikkelen voor het ontwerp van een commercieel aantrekkelijke
fotokatalytische reactor, gebruikmakend van een nieuwe manier van licht inbrengen.
Vanwege de complexiteit van fotokatalyse met reactant, fotokatalysator, activeringsmedium, en de
interactie tussen al deze spelers, is het van groot belang een systeem te ontwikkelen dat de
mogelijkheid en de betrouwbaarheid heeft voor het uitvoeren van katalysator-screening en
quesi-kinetische studies in beperkt tijdsbestek en met beperkte kosten. In hoofdstuk 2 wordt een
parallelle reactie systeem voor het high-throughput screening van fotokatalysator ontwikkeld voor dit
doel. Ondanks de inherente beperkingen, heeft het systeem bewezen geschikt zijn voor succesvolle
toepassingen op de kleurstofafbraak in water. Dankzij deze combinatoriële benadering die het
uitvoeren van meerdere fotokatalytische experimenten mogelijk maakt in relatief korte tijd, zijn wij in
staat om onderzoek te doen naar de performance van verschillende (gemodificeerde) TiO2
fotokatalysators.
Methyleenblauw degradatie werd gekozen als de testreactie voor kleurstofafbraak in waterige
156
systemen, zoals het op grote schaal is toegepast in de wetenschappelijke wereld als een standaard test
reactie voor fotokatalysators. Het is echter gevonden dat deze reactie een zeer complex systeem is dat
wordt beïnvloed door verschillende factoren, dat wil zeggen, licht, katalysator en reactanten. In het
kader van de experimentele condities beschreven in hoofdstuk 3 en 4, de reactiekinetiek kan worden
vereenvoudigd als een schijnbare 1e orde voor alle titania gebaseerde materialen bestudeerd. De 1e
orde kinetische constante kan dus worden gebruikt als enige parameter voor de evaluatie van de
fotokatalysator activiteit. De totale activiteit van fotokatalytische titania wordt bepaald door het
samenspel van eigenschappen onder andere als kristallijne structuur, katalysatoroppervlakte, de
oppervlakhydroxylgroepen, en adsorptie/desorptie kenmerken. Bovendien hebben de manier van
katalysatorgebruik, hetzij in slurry en vaste op een van katalysatoren, en de wijze van het invangen van
licht invloeden op de schijnbare fotokatalytische efficiëntie. In combinatie met diverse
analysetechnieken, alsmede de resultaten van andere kleurstof degradatieprocessen, werd vastgesteld
dat de totale oppervlakte en de daarmee samenhangende oppervlakte hydroxyl groepen tot de meest
belangrijke parameters voor het bepalen van de fotokatalytische efficiëntie behoren.
Er zijn ook nog uitzonderingen op deze regel. Voor de fotokatalysator met het grotere oppervlakte,
Hombikat, is de schijnbare degradatiesnelheid van methyleenblauw relatief laag per katalysator
oppervlak. Integendeel, de fotokatalysator met geringe oppervlakte, Merck, toont buitengewoon hoge
activiteiten in methyleenblauw degradatie. In deze twee gevallen, spelen bijkomende factoren een rol.
Voor de Hombikat fotokatalysator, is de schijnbare reactiesnelheid zeer waarschijnlijk beperkt door de
interne diffusie van methyleenblauw in poreuze agglomeraat, terwijl voor de monsters van Merck, de
reactie mogelijk is versterkt door de aanwezigheid van verontreiniging aluminium. Het is mogelijk dat
de wijze van methyleenblauw adsorptie en het afbraakpad zijn gewijzigd door de vervanging van
sommige Ti atomen door Al atomen.
Andere methoden ter verbetering van de fotokatalytische activiteit van TiO2 zijn thermische
voorbehandeling of door middel van doping van "vreemde" elementen in de TiO2 kristallijne structuur.
In hoofstuk 4 zijn de commerciële TiO2 (P25 van Degussa) monsters gedoteerd met zeldzame
aardmetalen van La, Ce, Zr, Y, Pr en Sm, en de activiteit van de monsters als een functie van
calcineren temperatuur werd getest in de fotokatalytische degradatie van methyleenblauw. De
fotokatalytische reactiesnelheid is met name afhankelijk van de hoeveelheid van een specifieke
Ti-OH-groep in de toegepaste P25 katalysator, die waarschijnlijk de sterkste adsorptie en/of meest
effectieve foto-reactive site is. De hoeveelheid van deze site was ook beïnvloed door de mate van
verspreiding en het laden van de zeldzame aarde oxide.
Veel onderzoekers hebben zich gericht op het milieu te verbeteren, zoals lucht reiniging en
waterzuivering, waarin organische verontreinigende stoffen volledig worden afgebroken tot
kooldioxide en water over TiO2 foto-katalysatoren. Aan de ander kant, relatief minder studies zijn
uitgevoerd over de toepassing van fotokatalyse voor organische synthese. In dit proefschrift wordt
foto-oxidatie van cyclohexaan gebruikt als een test reactie voor de evaluatie van het potentieel van
fotokatalyse in selectieve oxidatie. De golflengte van de straling spelt een belangrijke rol bij het
bepalen van zowel de reactiesnelheid als het reactiepad. In het kader van de ongekatalyseerd
foto-oxidatie in regio (λ <275 nm), is cyclohexanol het hoofdproduct. Dit is het resultaat van een
directe radicale kettingreactie, dat wil zeggen fotolyse. Als fotolyse wordt voorkomen door het gebruik
van de juiste licht filters (bijvoorbeeld, Pyrex, λ> 275 nm), dan wordt de reactiesnelheid onderdrukt,
tenzij TiO2 fotokatalysator wordt toegevoegd aan het systeem en de reactie voortzet via fotokatalytic
trajecten. Pure fotokatalyse over TiO2 heeft voornamelijk cyclohexanon met een ketone / alcohol
157
selectiviteit meer dan 95%. De activiteit van ketoneforming werd beïnvloed door de
katalysatorstructuur, met de oppervlakte hydroxylgroep dichtheid als de belangrijkste parameter. Op
basis van de reactieve studies onder verschillende configuraties van reactor, reactie, en gevarieerde
lichtbron, en tal van bijbehorende katalysator / product analysemethoden, wordt een voorlopige
reactiemechanism voorgesteld. De licht-geïnduceerde vorming van de oppervlakte-cyclohexyl
radicalen speelt een essentiele rol, gevolgd door de vorming van een peroxide en tussentijdse
ontbinding van cyclohexanon en desorptie. De adsorptie van cyclohexanol op de TiO2 oppervlak wordt
voorgesteld als een verslechtering van de fotokatalytische activiteit en deze levert een bijdrage aan de
vorming van CO2, een ongewenste bijproduct.
Zoals eerder geschreven, zijn er op dit moment nog geen photokatalytische reactoren die op
industriele schaal worden gebruikt. Conventionele type slurry foto-reactoren hebben typische nadelen
zoals moeilijke afscheiding van de katalysator, laag licht-efficiëntie, en de moeilijkheid van opschaling.
In deze studie hebben we een nieuwe reactor voor multi-fase fotokatalyse gebouwd, de zogenaamde
Interne Verlichte Monoliet Reactor (IIMR). In het concept van IIMR, side-light fibers worden geplaatst
in de kanalen van een keramische monoliet, die uitgerust zijn met TiO2 fotokatalysator bekleed op de
wand van elk afzonderlijk kanaal. Op deze manier kan een grotere verlichte katalysatoroppervlakte per
reactor volume worden bereikt. Bovendien kunnen de algemeen erkende voordelen van monoliet
reactoren, zoals de snelle stofoverdracht en het gemak om op te schalen, worden geexploreerd. Een
nieuw washcoating methode werd ontwikkeld, dat heeft geresulteerd in een stabiele en uniforme TiO2
laag op de monoliet wanden. De voorlopige foto-efficentie met de IIMR reactor in de selectieve
foto-oxidatie van cyclohexaan is minder dan die verkregen is met een kleinere omvang top verlichting
slurry reactor, maar hoger dan die verkregen zijn in een ringvormige reactor van vergelijkbare grootte
en de configuratie van reactor met side-light fibers ondergedompeld in een TiO2 slurry. Op basis van
SEM fotos met de afmetingen en het afwegen van de TiO2 coatings op de monoliet, is het duidelijk dat
de totale hoeveelheid van TiO2 op de monoliet veel meer is dan optimaal, zodat de schijnbare
foton-effeciëntie daadwerkelijk laag is vanwege de afschermingseffecten en de grote hoeveelheid
producten geadsorbeerd aan de TiO2 laag. De schijnbare foton-efficiëntie kan verder worden
geoptimaliseerd door het verminderen van de laagdikte van TiO2 op de monolietwanden. Voorts dient
te worden opgemerkt dat de lichtbron configuratie in de gepresenteerde studie verre van optimal is
(positie, intensiteit). Verbetering van de configuratie zal zeker leiden tot verbeteringen in foton-
effeciëntie van de IIMR.
158
159
Acknowledgements
Doing four years of PhD research in TU Delft is a long long journey full of joy and happy
moments, but sometimes also fumbles, lessons-learned and frustrations. The success would never
come without the great help and support of all the people around me. Especially those who have made
their contributions to this thesis are gratefully acknowledged here.
At the very first, I would like to acknowledge my deepest gratitude to my PhD supervisors, Prof.
Dr. Jacob A. Moulijn and Dr. Guido Mul, who have given me the opportunity to carry out the
explorative research in photocatalysis.
Jacob, thanks for your guidance, broad scientific view, brilliant ideas and suggestions, and the
freedom to try new things. Your encouragement and persistency have proven to be essential to drive
me to completion of writing. I really enjoyed all these years working in your group.
Guido, it has been such a pleasant work atmosphere under your guidance. I am grateful for our
fruitful discussions, your expertise in spectroscopy, and your great help in writing. Guido, thank you
for the invaluable support of all these years.
I would like to thank all the present and former colleagues for their kind helps and contributions
so as to complete the thesis. Weidong introduced me to the group and gave me huge supports in both
scientific study and my stay in Holland. Michiel K. & Achim, my M.Sc. supervisors, although we have
not corroborated in the photocatalytic research, your learned me the real technical way of working in
the dark, that greatly helps me through my Ph.D work and later jobs, especially in hard times. Arjan,
we have been staying in the same office for four years, and rebuilt it multiple times, thanks for the
technical/engineering inputs. Hiro, my another roommate, it is a pleasure to work with you and thanks
for the fruitful discussions on the interactions of photon, electricity and catalysts. Semeh, although it is
not optimized yet, I really like your fancy ideas of TUD-1 catalyst, and do enjoyed our collaboration
on TUD-1 photocatalysis and cyclohexane oxidation. Bart, Harrie & Gerard R. provided great
technical supports on equipments such as GC, IR, TPR and TGA. Thanks to the colleagues in O&O
lab, Johan, Sander, Marcel, Loes, who supported in the HPLC, catalyst morphology determination,
PSD analysis, and many fruitful discussions. Marga, special thanks to your great work in
photocatalytic reactors in RU Groningen, and the generous sharing of the equipments. Without your
pioneer work, this thesis would not be complete. Wenjiang, I enjoyed our nice collaboration on the
work of mesoporous Ti-silica hollow spheres. Harald, it is a pleasure to supervise you in your
graduation project on photocatlytic swirl flow reactor. Sandra, Els, Lizzy, and Elly (the secretaries),
thank you for your management assistances.
160
Also I would like to express my gratefulness especially to the project leaders who provided
guidance in many projects I was involved, Prof. Freek Kapteijn, Prof. Marc-Olivier Coppens, and Dr.
Michiel Makkee. You do help me a lot in developing the scientific mind-set and the interest into
various topics that greatly broaden my view into the fascinating world of chemical industry.
Not to forgotten is my gratefulness to the all the colleagues of the former “Industrial Catalysis
group” and lateron RaCE, that we shared great times during my 5 years in the group, Bram, Bart Z.,
Bas, Edwin, Javier, Nari, Ronald, Xiaoding, Krishna, Agus, Agustin B., Agustin P., Ingrid, Brigitte,
Karen, Gerben, Xander, Martijn, Jorrit, Nakul, Joana and all others, to the technical staffs in the TOCK
group for HPLC-MS, in the inorganic chemistry group for SEM, and in the optics group for light
intensity measurements. Special moments I will always remember are the RaCE soccer team, to all
members for the fun and our success in NIOK soccer competition.
It is impossible to forget my families in this acknowledgement, who have be the ultimate support
for all the times no matter things go for or against the wind. Thanks for my wife Fang, my mother and
father, and my sister and brother-in-law, for your patience, understanding, endless support and love. I
am indebted to you all!
Peng
July 2008
YueYang, China
161
Publications and Oral Presentations
Publications
P. Du, A. Bueno-Lopez, M. Verbaas, A.R. Almeida, M. Makkee, J.A. Moulijn, G. Mul, The Effect of
Surface OH-Population on the Photocatalytic Activity of Rare Earth-Doped P25-TiO2 in Methylene
Blue Degradation, J. Catal., 260 (2008) 75-80
P. Du, J.T. Carneiro, J.A. Moulijn, G. Mul, A Novel Photocatalytic Monolith Reactor for Multiphase
Heterogeneous Photocatalysis, Appl. Catal. A: General, 334 (2008) 119-128
P. Du, J.A. Moulijn, G. Mul, Selective Photo(catalytic)-Oxidation of Cyclohexane: Effect of
Wavelength and TiO2 Structure on Product Yields, J. Catal., 238 (2006), 342-352
M. Baca, W.J. Li, P. Du, G. Mul, J.A. Moulijn, M.O. Coppens, Catalytic Characterization of
Mesoporous Ti-Silica Hollow Spheres, Catal. Lett., 109 (2006), 207-210
M.T. Kreutzer, P. Du, J.J. Heiszwolf, F. Kapteijn, J.A. Moulijn, Mass Transfer Characteristics of
Three-Phase Monolith Reactors, Chem. Engng. Sci., 56 (2001), 6015-6023
162
Oral presentations
J.T. Carneiro, P. Du, J.A. Moulijn, G. Mul, A Novel Photocatalytic Monolith Reactor for Multiphase
Heterogeneous Photocatalysis, presented at 2007AIChE Annual Meeting, Salt Lake City, USA, Nov.
2007
P. Du, J.A. Moulijn, G. Mul, Mechanistic Study of Photocatalytic Oxidation of Cyclohexane by TiO2:
Effect of Wavelength and Hydroxyl Groups, presented at 6th Netherlands Catalysis and Chemistry
Conference, Noordwijkerhout, The Netherlands, Mar. 2005
P. Du, H. Shibata, G. Mul, J.A. Moulijn, Effect of TiO2 Source and Thermal Treatment on
Photoactivity for Methylene Blue Degradation in Water, presented at 9th
International Conference on
TiO2 Photocatalysis, San Diego, USA, Nov. 2004
H. Shibata, P. Du, G. Mul, J.A. Moulijn, Hydrotalcite-Like Compounds Based Photocatalytic Water
Purification, presented at 9th
International Conference on TiO2 Photocatalysis, San Diego, USA, Nov.
2004
P. Du, G. Mul, J.A. Moulijn, Towards the Kinetic Study of Photocatalytic Oxidation of Cyclohexane by
TiO2, presented at 228th ACS National Meeting, Philadelphia, USA, Aug. 2004
P. Du, G. Mul, J.A. Moulijn, Liquid Phase Photocatalytic Oxidation of Cyclohexane by TiO2, presented
at 3rd Netherlands Process Technology Symposium, Veldhoven, The Netherlands, Nov. 2003
163
Curriculum Vitae
Peng Du was born on January 6, 1976, in Zhejiang, China. He moved to the Netherlands in 1995,
and followed the university study of Chemical Engineering at the Delft University of Technology,
from which he graduated for M.Sc. (Ir. In Dutch) in June 2001. His M.Sc. thesis, entitled “Mass
transfer in multiphase monolith reactors” was carried out in the group of Industrial Catalysis,
DelftChemTech, faculty of Applied Sciences. His Ph.D. project at Delft University of Technology
started in April 2001 at the same group, with thesis supervisors Prof. Dr. Jacob A. Moulijn and Dr. G.
Mul. His research interest is photocatalysis, which covers a wide scope from photocatalyst
development, applied studies of photocatalysis in water purification and selective photo-oxidation, and
novel photocatalytic reactor development. The main results are included in this dissertation.
After the PhD research he worked for Aker Kvaerner in the process department at Zoetermeer, the
Netherlands. From October 2005 to June 2007, he joined Shell Global Solutions in Amsterdam, as a
technologist in coal gasification. Since July 2007, he moved to Shell Gas and Power, and seconded to
Dongting Sinopec&Shell Coal Gasification JV for full-time technology support.