propositions of the doctoral dissertation entitled catalysis engineering of light induced dye

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


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


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.


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


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


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.


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



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


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


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


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




x


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


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


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


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)


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


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.


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


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]


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]


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


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


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


ka

pp

[1/m

in]

(b)

0

0.02

0.04

0.06

0.08

0 400 800 1200


ka

pp

[1/m

in]

(c)

0

0.02

0.04

0.06

0.08

0 400 800 1200


ka

pp

[1/m

in]

(d)

0

0.02

0.04

0.06

0.08

0 400 800 1200


ka

pp

[1/m

in]

(e)

[°C] [°C]

[°C] [°C]

[°C]


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


(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


(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


(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]


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


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


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]


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


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


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


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


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


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


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


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


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


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


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


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


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


ka

pp

[1

/min

]

Hombikat

Aldrich_A

P25

Aldrich_B

Merck

Fluka

[°C][°C]


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


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


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)


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


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


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.


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)


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]


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


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


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


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


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


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


Yie

ld [

mo

l/l]

cyclohexanone

1,1'-oxybis(cyclohexane)

cyclohexyl hexanoate

1/6 CO2

(b)


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


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


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


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


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

]


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


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


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


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


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


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]


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


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


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.


123

5.5 Conclusions


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


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


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


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.


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.


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

[ °]


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


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


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)


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


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


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


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.

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.

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.

propositions of the doctoral dissertation entitled catalysis engineering of light induced dye

Documents