You are on page 1of 12

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Contents lists available at ScienceDirect

Chemical Engineering Research and Design


journal homepage: www.elsevier.com/locate/cherd

Effect of hydrodynamics during solgel synthesis of TiO2


nanoparticles: From morphology to photocatalytic properties
Mlisa Hatat-Fraile, Julie Mendret , Matthieu Rivallin, Stephan Brosillon
IEM (Institut Europen des Membranes), UMR 5635 (CNRS-ENSCM-UM2), Universit Montpellier 2, Place E. Bataillon,
F-34095 Montpellier, France

a b s t r a c t
In this study, the role of mixing hydrodynamics during the solgel synthesis of titania nanoparticles and the consequences on their photocatalytic properties were investigated. For the rst time three different T-mixer geometries
were tested. Alcoholic solutions of titanium tetra-isopropoxide and water were mixed in three different T-mixers
with turbulence promoters and thus different mixing characteristics. The changes of nanoparticle sizes during the
induction time of the solgel process were followed by dynamic light scattering and velocity and turbulence elds
were simulated by Computational Fluid Dynamics (CFD) for the three T-mixer geometries. The results indicated that
macro-mixing is crucial during the rst step as it determines the nucleation rate and then the primary particle size.
The micro-mixing has an inuence on particle properties, especially on particle stability. Titanium dioxide nanoparticles synthesized by the solgel process were deposited on alumina supports. A homogeneous lm of about 200 nm
was deposited in all cases. Degradation of Acid Orange 7 (AO7) was used to evaluate the photocatalytic activity of
TiO2 coatings. No difference was observed between the photoactivity of synthesized TiO2 . Total mineralization of the
dye occurred after 24 h irradiation.
2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Water treatment; Titanium dioxide; Solgel synthesis; Photocatalysis; Mixing

1.

Introduction

Advanced Oxidation Processes (AOPs) are those which allow


the degradation of organic bio-recalcitrant pollutants in water.
Broadly speaking, AOPs can be used as oxidation methods
in aqueous medium based on the use of highly reactive and
non-selective species, such as hydroxyl radicals OH for the
destruction of organic compounds (Comninellis et al., 2008;
Oturan et al., 2011). Due to the highly unselective reactions
involved in AOPs, the application of these techniques was
mainly directed to the removal of hazardous compounds in
polluted efuents. Photocatalysis is one of the most commonly used AOPs. In this method, the generation of the
oxidizing species is achieved by the irradiation in the near UV
of a semiconductor material.
Among semiconductor materials, titanium dioxide TiO2 is
considered as one of the best for photocatalysis application
(Fujishima et al., 1999; Chebli et al., 2011; Brosillon et al., 2008;
Molinari et al., 2009; Faisal et al., 2011; Lin et al., 1998, 2011;

Abramovic et al., 2011). The use of aqueous nanometric TiO2


particles in the aqueous phase for pollutant treatment by photodegradation has been widely studied (Faisal et al., 2011; Lin
et al., 1998, 2011; Abramovic et al., 2011; Molinari et al., 2002)
and provides very good elimination of organic species compared with conventional treatment methods. However, one of
the main drawbacks of this technology pertains to the recovery
of the catalyst from the reactive environment. Immobilization
of TiO2 on a support enables this drawback to be mitigated and
offers the possibility of preparing thin TiO2 lms of a desired
thickness and homogeneity with the power to degrade organic
compounds (Antoniou et al., 2009; Romanosa et al., 2012). Nevertheless, according to several studies (Molinari et al., 2009;
Zhang et al., 2003; Plantard et al., 2011) the deposited catalyst tends to (i) diminish the efciency of the photocatalysis
because the catalyst structure and morphology induce a nonoptimal distribution of the irradiating light and (ii) be released
over time, thus decreasing the efciency of the photocatalytic
process. Furthermore, the level of photoactivity of the coated

Corresponding author. Tel.: +33 467144624; fax: +33 467149119.


E-mail address: julie.mendret@um2.fr (J. Mendret).
Received 17 July 2012; Received in revised form 5 April 2013; Accepted 25 April 2013
0263-8762/$ see front matter 2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cherd.2013.04.022

2390

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

TiO2 is highly dependent on the lm morphology and structure.


A promising approach to improve the coating integrity
and to reduce photocatalyst release can be to elaborate the
photoactive coating from nanometric TiO2 particles obtained
by solgel process. Indeed, these nanoparticles are highly
reactive (toward a support) and offer the possibility to form
resistant thin lms (Pucher et al., 2007). The efciency of
solgel synthesized titanium dioxide to degrade organic compounds has been widely studied and it has been demonstrated
that its photocatalytic activity is greatly inuenced by its
crystalline form (Henderson, 2011). In addition, nano-scale
materials have a larger relative surface area and stronger
organic adsorption capacity, which are both favorable for photocatalytic reaction (Nakata et al., 2012).
The solgel synthesis of titanium dioxide is carried out in
alcoholic solutions containing titanium alkoxide Ti(OR)4 and
water. The basic chemistry of the solgel process includes
hydrolysis (1) and condensation by oxolation (2a) and alcoxolation reactions (2b):
Ti(OR)4 + H2 O Ti(OR)3 OH + ROH

(1)

(OR)3 Ti OH + HO Ti(OR)3 (OR)3 Ti O Ti(OR)3 + H2 O

(2a)

(OR)3 Ti OH + RO Ti(OR)3 (OR)3 Ti O Ti(OR)3 + ROH

(2b)

In terms of particle formation, there are two steps: the


rst one is nucleation and the second one is an aggregation step during which particles collide and bind together
(Soloviev et al., 2003). The relative rates of these mechanisms
are very important since they have a great inuence on the
nal product characteristics such as particle size distribution,
morphology and structure.
The solgel process has been studied for a long time and
the chemical aspects have been made clear but there still
exists a key issue for optimizing synthesis conditions in order
to obtain a powder with improved morphological properties
compared with current commercial products. This key issue
mainly resides in the optimization of mixing, the effects of
which have not been thoroughly investigated.
Precipitation is the consequence of mixing of two reactive
uids and the driving force is super-saturation. Mixing has a
direct effect on the generation of super-saturation as it modies its distribution and transports previously formed particles
in regions of high super-saturation. High particle nucleation rates can be obtained with high mixing rates. Moreover,
nucleation and growth are competing phenomena, as both
consume solute molecules. Particle size is the result of this
competition. Hence, it is of the greatest importance to identify both characteristic time and length scales for each step
in the solgel process: macro and micro-mixing, hydrolysis,
further condensation/aggregation. Analyzing and comparing
these characteristics would then allow better control of the
particle properties.
On the one hand, the characteristic time for hydrolysis is
related to the induction time; i.e. the time required for the
formation of a visible solid phase. Some studies have shown
that hydrolysis is faster than condensation and that the rst
hydrolysis step corresponds to the creation of a supersaturated solution (Harris and Byers, 1988). Nonetheless, kinetic
studies of titania solgel synthesis demonstrated that both
hydrolysis and condensation are quite fast and that condensation occurs after only 2550% of the original alkoxyde groups

are substituted by hydroxides (Turova et al., 2002). However,


there is very little quantitative data available in the literature
about the typical particle formation time and even less about
typical reaction time.
On the other hand, turbulent mixing occurs at several time
and length scales, beginning on a scale of the same order
of magnitude as that of the dimensions of the equipment
and ending on a very small scale (Batchelor scale) that is
determined by the uid transport properties (viscosity and diffusivity) and by turbulence intensity (turbulence dissipation
rate). The mixing characteristic time is thus given by the evaluation of the time needed to reduce the scale of segregation
from the macro (macro-mixing) down to the molecular level
(micro-mixing).
Previous studies have shown that the ratio between the
convective time scale and the typical reaction time scale,
represented by the Damkhler number, Da, controls the
size characteristics of the solid product: only a very fast
mixing (Da < 1) enables ultra-ne particles with a narrow sizedistribution to be obtained (Rivallin et al., 2005; Azouani et al.,
2010; Marchisio et al., 2009). These studies demonstrated that
mixing plays a very important role and clearly affects not only
the particle size distribution but also the time evolution of
the particles formed immediately after mixing. Thus, it has
been demonstrated that mixing is a key parameter during
TiO2 particle synthesis. However, it has been difcult to distinguish the relative effect of macro and micro-mixing on the
nal particle quality. Moreover, previous studies investigated
the effect of mixing on particle size distribution but not the
effect on their catalytic performance. In this work, we aim at
bringing a contribution to the understanding of hydrodynamic
effects on both particle morphology and photocatalytic properties. Acid Orange 7 (AO7) was chosen as a model compound
to evaluate the photocatalytic activity because it is the most
studied compound among the azo dyes (Konstantinou and
Albanis, 2004; Satuf et al., 2011; Mills et al., 2012; Zita and

Krysa,
2011; Velegraki et al., 2006). So, the aim of this study
is not to establish the degradation pathways and the formation of by-products which have already been fully described
(Vinodgopal et al., 1996; Bauer et al., 2001; Stylidi et al., 2003).
The novelty of this paper lies within the two following
aspects: rst, the effects on particle size evolution of several mixing-reactor geometries with the same inlet ow rates
were compared. Indeed, to our knowledge, most of the studies
investigated the effect of the inlet ow rates on the particle distribution size in the same reactor geometry. Second,
the consequences of the mixing conditions employed during
solgel synthesis on subsequent photocatalytic properties of
TiO2 coatings were investigated. The originality of this work
is to focus on the role of uid dynamics on both particle
size/stability of the sol (local scale) and photocatalytic performance (pilot scale).
In this present work, particle characterization and photocatalysis process performances were investigated from an
experimental point of view and ow behavior in the T-mixer
was characterized using Computational Fluid Dynamics (CFD).

2.

Materials and methods

2.1.

Solgel reactions

Titanium dioxide was synthesized by mixing a solution of


milli-Q water and 2-propanol with a solution of titanium

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

2391

Fig. 1 Sketch of the solgel reactor.

tetra-isopropoxide (TTIP) as alkoxide. 2-Propanol 98% (from


VWR) and TTIP 97% (from Aldrich) were used without further
purication. Equal volumes of reactant solutions (i.e. 100 ml)
were mixed at 293 K in a static T-mixer. The TTIP/2-propanol
solution (A) at 0.153 mol TTIP/L and H2 O/2-propanol solution
(B) at 0.376 mol water/L were injected with constant equal ow
rates into a static T-mixer via two peristaltic pumps. The
hydrolysis ratio is H = CW /CTi = 2.46. The air for the experiment
was dried using a silica gel desiccator. Solutions exiting the
reactor were collected in a small vessel at a thermostatically
controlled temperature of 273 K (Fig. 1).
Static T-mixers with an inlet Reynolds number Rein ranging from 6000 to 8000 offer all the characteristics necessary
in order to obtain an optimal granulometry and a quasi-monodispersed size distribution of particles (Azouani et al., 2010).
In this study, three different T-mixer congurations were chosen to investigate the inuence of the hydrodynamics on the
TiO2 nanoparticles solgel elaboration: a simple T-mixer (Ts ),
a T-mixer with bafes (Tb ), and a T-mixer composed of a large
pipe connected to a narrow pipe (Tn ). Fig. 2 gives a detailed
schematic of the T-mixers. Bafes and narrow pipes were chosen to create turbulence and therefore various mixing times
shorter than the typical reaction time based on Rivallin et al.,
giving a primary hydrolysis time of several tens of milliseconds
for the same operating conditions (Rivallin et al., 2004). For the
three T-mixers, the two reactant streams were fed tangentially
into opposite sides of the reactor through two 2 mm inner
diameter eccentric pipes; the mixed solution left the reactor
from the bottom through an outlet pipe of 4 mm inner diameter for Ts and Tb and 2 mm for Tn . Eccentric injection of the
reactive uids created a vortex which promoted mixing. The
experimental inlet velocity chosen in this study is the highest
one which enables the use of the three mixers (Tn is ow-rate
limited) with a turbulent inlet Reynolds number.

Fig. 2 T-mixers. Top view (a), front view of Ts (b), Tb (c)


and Tn (d).

2.2.

Dip-coating reactor

During the induction period, the temporal stabilization of


extremely reactive TiO2 clusters and particles enabled the
efcient doping and deposition processes. Chemically active
sols reacted with the active sites of the support surface. In
this study, solgel TiO2 nanoparticles were immobilized during the induction period on non-porous alumina supports
(100 mm 50 mm 4 mm 75 g Technical Glass Company,
UK) using a dip-coating apparatus (Dip Master 201, Chemat
Technology, Inc.). Supports were washed in an acid solution
and then submerged for 99 s in the TiO2 nanoparticle suspension. The withdrawal speed was 10 cm/min. TiO2 coating was
realized 6 times on one face only (the other faces were hidden).
The support was dried at 343 K for 24 h. The thermal treatment
was continued by calcinations at 673 K for 3 h to enable the
anatase formation with heating at 2 K/min. Finally, the support was washed in order to remove excessive non-reacted
nano-particles.

2.3.

Characterization

The characterization of the particulate systems obtained in


the different experiments was carried out assuming an instantaneous formation of solid particles and a subsequent slow
aggregation/condensation process. The mean particle size of
the sol obtained after mixing was measured by dynamic light
scattering (DLS) with a Malvern ZetaSizer Nanoseries (NanoS,
Malvern instruments) providing reliable information in the
size range of 16000 nm. This device enables the measurement
of the mean particle size and the particle size distribution

2392

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

no signicant difference was observed between the initial and


the nal pH.

3.
Governing equations and numerical
details

Fig. 3 Chemical structure of Acid Orange 7 (AO7).


(PSD). The mean particle size and particle size distribution
were followed with time by making regular measurements.
Scanning electron microscopy (SEM, S-4800, Hitachi) was carried out before and after thermal treatments. Each experiment
was repeated three times in order to verify the synthesis and
characterization protocol reproducibility.

2.4.

Photocatalytic reactor

Photocatalysis properties of the synthetized TiO2 coating were investigated by following the degradation of
Acid Orange 7 (AO7) as a model pollutant (Fig. 3). AO7
(purity > 85%) was purchased from Aldrich and used without further purication. The photocatalytic reactor consists
in a batch glass reactor composed of two tanks separated
by a UV transparent Plexiglas wall (Fig. 4). The rst one
(height = 200 mm, width = 50 mm, depth = 110 mm) contains a
UV light ( = 365 nm, Philips pL-L 24W/10/4P) and the second
one (height = 200 mm, width = 85 mm, depth = 110 mm) contains contaminated water, an alumina support with TiO2
coating and an air diffuser. The irradiance after the plexiglass
wall was about 22.76 W/m2 . The dissolved oxygen reached
a value of 8 mg O2 /L and only a slight decrease of the dissolved O2 (10%) was observed after 16 h of irradiation. The
AO7 degradation under UV irradiation was monitored using
a UVvis spectrophotometer (UV-2401 PC Shimadzu). The
concentration of the model pollutant is calculated by the
calibration curve obtained from the absorbance of solutions
(max = 485 nm) at different concentrations. The photocatalytic
activity of the TiO2 coatings elaborated in the three solgel
reactors was determined by studying the AO7 degradation
under UV irradiation. Firstly, alumina supports with or without TiO2 coating were placed in a 10 mg/L solution of AO7
under dark conditions for 30 min with bubbling air in order to
reach adsorption equilibrium. Then the UV-lamp was turned
on and at scheduled times, 1 mL of sample was taken from
the reactor. The pH was the natural pH of the dye (7.2) and

Fig. 4 Scheme of the batch photocatalytic reactor.

Flow behavior in T-mixers was investigated by running CFD


simulations. Literature concerning nanoparticle precipitation
reports that the uid ow in a T-mixer is turbulent for
Reynolds numbers above 500 (Gradl et al., 2006). In this study,
the inlet velocity leads to a Reynolds number of around
2000 and therefore a turbulent uid ow. The ow eld
and the turbulence eld were described using the standard
Reynolds-average NavierStokes (RANS) approach and the k
turbulence model. This model relies on several assumptions in
particular that Reynolds is high enough to use RANS approach
and that turbulence is in equilibrium in the boundary layer
meaning that production equals dissipation. These assumptions limit the accuracy of the model as in our case Reynolds
number is quite low. However, it was considered that the
limited accuracy is a fair trade-off for the amount of computational resources saved compared with more complicated
turbulence models.
The problem is described in terms of a Reynolds-averaged
relative concentration c which is an inert scalar ranging from
0 to 1 and is assumed to be equal to 1 in one feed stream and to
0 in the other. The mixing at macro-scale was considered by
resolving the transport equation for steady-state conditions
that is:
UC [(D + Dt )C] = 0

(3)

where D and Dt are the molecular and turbulent diffusion


coefcients respectively and the source term is null since the
mixture fraction is a non-reacting scalar. The turbulent diffusion coefcient can be calculated using a turbulent Schmidt
number and the Boussinesq hypothesis (Marchisio et al., 2006).
Dt =

C k2
Sct

(4)

where C is a numerical constant equal to 0.09, Sct is the turbulent Schmidt number equal to 0.7, k is the turbulence kinetic
energy and the turbulence energy dissipation rate which is
computed from the CFD turbulence model.
Mixing at the micro-scale was characterized by estimating the spacial distribution of the local Reynolds number, Rel ,
which uses the local turbulence level and is given by:
Rel =

(5)

(v)

where  is the kinematic viscosity. k and are obtained from


the RANS model.
Mixing in the reactor was investigated by computing CFD
simulations using the nite element software Comsol multiphysics 4.2. Simulations in the turbulent ow regime were
run using the RANS approach and employing the k turbulence model and wall functions for the near wall treatment.
3-D geometries corresponding to the three different T-mixers
(Fig. 2) were meshed with tetrahedral elements generated by
the software Comsol. Wall functions were used to describe
the ow at the walls. The wall functions in Comsol are such
that the computational domain is assumed to start a distance

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Fig. 5 Change of the TiO2 nanoparticle size during the


induction period of the solgel process using (a) Ts , (b) Tb
and (c) Tn .
w from the wall. This distance is automatically compute so
1/4
k is the friction velocity,
that +
w = u w /, where  = C
becomes 11.06. This value corresponds to the distance from
the wall where the logarithmic layer meets the viscous sublayer. w is limited from below so that it never becomes smaller
than half of the height of the boundary mesh cell. This means
that +
w can become higher than 11.06 if the mesh is relatively
coarse. It was thus checked that +
w is 11.06 on all walls. Different computational grids were tested until a grid-independent
solution was obtained. The nal mesh contained 361 336 computational cells.
The ow eld was simulated assuming a ow of the main
reagent only (2-propanol) as the two other reagents, water and
TTIP, were strongly diluted. Diffusivity of 2-propanol in water
was estimated from Poling et al. and is 2.41 109 m2 /s (Poling
et al., 2004). The density is then 785 kg/m3 and the kinematic
viscosity 2.89 105 m2 /s. The temperature was assumed to be
298 K and kept constant. Moreover, it is supposed that particles
do not affect the liquid phase. Indeed, as residence time in
the T-mixers is very short, when the reacting solution exits
the T-mixer, it contains very small particles. The boundary
conditions are c = 0 in one inlet and c = 1 in the other one.
The inlet velocity condition is a at prole with u = 3.64 m/s as
during experiments. The outlet pressure is P = 0 Pa.

4.

Results and discussion

4.1.

TiO2 nanoparticle solgel synthesis

In Fig. 5, the mean particle size prepared under the same operating conditions in the three T-mixers is reported, versus time.
It can be seen that the mean particle size stabilizes between

2393

2 and 4 nm and then increases rapidly when the induction


period nishes. Both duration of the rst period and mean particle size before rapid growth depend on the T-mixer geometry.
The induction step duration is 1500 s for particles synthesized
in Tb whereas those from Ts and Tn have a long rst period
duration of around 2800 s. In addition, mean particle size is
slightly higher for the particles synthesized in Tb (rp = 2.98 nm)
than for the particles synthesized in Ts (rp = 2.69 nm) and in
the Tn (rp = 2.65 nm). The mean particle size during the induction time is higher than that (rp = 1 nm) obtained by other
authors. This is probably due to difference in operating conditions: Azouani et al. (2010) worked at 20 C with H = 1.90 and
Marchisio et al. (2009) worked at 30 C with different hydrolysis ratios (H = 0.250.5) with hydrochloric acid added in the
reactive medium.
The effect of the T-mixer geometry on particle size distribution is presented in Fig. 6 for two values of the ratio time. The
time ratio is calculated by dividing the time after exiting the
T-mixer, t, by the induction time (t/tind = 0.1 and t/tind = 1). As
can be seen, it is difcult to conclude about the effect of mixing conditions on particle size distribution and it seems that
at the end of the induction period, there is no effect of mixing
geometry. Several authors have demonstrated that the hydrodynamics have a strong inuence on particle size distribution
which would become narrower and centered on a smaller size
as Re increases (Azouani et al., 2010; Marchisio et al., 2009).
Nonetheless, in these studies, hydrodynamics were modied
by varying the inlet Reynolds number. In the present study,
the inlet Reynolds number was maintained constant and only
mixer geometry was changed. Hence, particle size distribution
seems to be more dependent on inlet Re number than on the
mixer geometry. However, comparison between studies must
be considered with care as hydrolysis ratios are often different and some authors have shown that this parameter affects
particle size distribution (Hazah and Sopyan, 2009). Synthesized particles were coated on alumina supports which were
then dried and calcined for further characterization and used
as a photocatalyst. Mixing does not seem to have an effect
on the crystalline form as after calcinations at 673 K, the TiO2
anatase form was obtained for all reactor geometries tested
(Hatat Fraile, 2013).

4.2.

TiO2 coating on alumina support

In order to optimize the dip-coating technique, solgel TiO2


nanoparticles were rstly immobilized on a non-porous alumina support during the induction period. The obtained

Fig. 6 Change of the TiO2 nanoparticle size distribution during the induction period of the solgel process using Ts , Tb and
Tn .

2394

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Fig. 7 SEM micrographies of the alumina support with and without TiO2 coating. Top view without TiO2 coating (a), top
view with TiO2 coating (b) and front view with TiO2 coating (c).
coatings were observed using SEM (Fig. 7) in order to examine
their morphology and defects.
With all the samples obtained, similar TiO2 coatings with a
homogeneous thickness of about 200 nm were observed. TiO2
completely enveloped grains composing the alumina support
(Fig. 7c). However, cracks appeared on the lm surface. This
phenomenon was also noticed by Tian et al. and is attributed
to the removal of residual hydroxyl and organic groups during the heating process (Tian et al., 2009). No difference was
observed between coating morphology obtained from sols
synthesized in the three T-mixers.

4.3.

Photocatalytic activity of the TiO2 coatings

Fig. 8 shows the normalized AO7 concentration versus time


for the different coatings. It emerges that the degradation
of AO7 by photolysis is weak since only 20% of the dye was
bleached after 16 h of irradiation. The presence of the alumina sheet alone (without TiO2 ) does not change the AO7
degradation rate, thus it can be concluded that the alumina
support does not participate in the photodegradation. In the
presence of synthesized TiO2 coatings, the yield of degradation
of AO7 ranges from 73% to 79% after 16 h of irradiation. Consequently, the decrease of AO7 concentration is mainly due
to the heterogeneous photocatalytic degradation. It appears
that the three degradation rates are very similar. This result
is consistent with the measurement of the thickness of the
TiO2 lm for the three cases which shown a similar thickness
(200 nm) and the evaluation of the quantities of TiO2 coated on
the alumina. Indeed the TiO2 lm was dissolved by an acid following the protocol established by Coronado et al. (2003). The

Fig. 8 AO7 concentration under UV irradiation: without


any support (a), with uncoated alumina support (b), with
supports coated with TiO2 elaborated in Ts (c), Tn (d) and Tb
(e).

amount of TiO2 coated ranged from 4.4 to 6.6 mg. Such a small
difference in the mass of TiO2 could explain the very similar
observed kinetics. So the preparation of TiO2 nanoparticles
with three different T-mixers gives particles which present
the same photocatalytic activity. The evolution of the AO7
UVvisible spectrum during irradiation is shown in Fig. 9.
It appears that the visible chromophor ( = 485 nm) strongly
decreases and corresponds to the breakdown of the N N bond
or C N bond as mentioned by Konstantinou and Albanis
(2004) and Zhang et al. (1998). During the rst step of AO7
oxidation by photocatalysis, the main organic by-products
identied were benzene sulphonic acid, sulphoanilic acid,
1,4-naphthoquinone, phtalic acid (Vinodgopal et al., 1996) quinine and 4-hydroxybenzene sulphonic acid (Bauer et al., 2001;
Stylidi et al., 2003). The last organic by-products before complete mineralization were aliphatic acids. Since the majority
of the by-products contain a naphthalene group or a benzene
ring, which both strongly adsorb light at  = 254 nm, it is consistent to monitor the variation of the absorbance at  = 254 nm
(A254nm ) versus time. Indeed, the decrease of A254nm indicates
the opening of the rings. Then, as suggested by Shinde et al.
(2009), a simplication could be made by correlating the rate
of decrease of A254nm to the rate of mineralization:
d(TOC)
d(A254nm )

dt
dt

(7)

Fig. 10 presents the variation of absorbance of the visible


chromophor A486nm and A254nm during photocatalysis of AO7.
Logically, AO7 decoloration is more rapid at the beginning than

Fig. 9 Change of the AO7 absorption spectrum.


Experiment was carried out with an alumina support
coated with TiO2 elaborated in the solgel reactor with Ts .

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Fig. 10 Change of the absorbance of the solution for two


wavelengths: 254 nm and 485 nm during photocatalytic
reaction.

at the opening of the aromatic ring which necessitates several


oxidation steps (Lhomme et al., 2005).
However, after 24 h of irradiation, the two absorbances,
A254nm and A486nm were close to zero indicating a concomitant
decoloration and mineralization. These results have already
been observed by other authors (Bauer et al., 2001; Lhomme
et al., 2006).

4.4.

Hydrodynamics in T-mixers

Fig. 5 shows that sol synthesized with Tb has a different behavior than sol synthesized with Ts and Tn . The only difference
between the three experiments stands in the mixer geometry
and could therefore be explained by differences in hydrodynamics and characteristic times in the mixer.
The detailed mechanisms of mixing are difcult to observe
experimentally due both to spatial limitations and to the very
short time scales involved. At this point, CFD can be used to
investigate the characteristics of mixing in the chemical reactors. Although a direct validation through comparison with
experimental data would be preferred, it has been proved that
CFD is a reliable tool that can be trusted in order to evaluate
trends and orders of magnitude for quantities such as ow
eld and concentration gradient in such geometries (Marchisio
et al., 2006; Liu and Fox, 2005; Bothe et al., 2006).
Under perfect mixing conditions, the output local relative concentration would be c0 /2 =0.5. Concentration gradients
in a (x, y) plane at a z position can then be represented by

I = (x,y) (c(x, y) (c0 /2))2 . Simulation results enable concentration gradients to be plotted versus the z-axis (vertical pipe)
as shown in Fig. 11. I is an image of macro-scale segregation,
the smaller it is, the better mixing is. For all geometries, concentration gradients appear at the inlet and then disappear
progressively by convection and turbulent mixing along the
z-axis. It rapidly reaches a very low level and macro-scale segregation disappears in the rst section from z = 0 to z = 20 mm
along the z-axis and the uid leaving the reactors is completely
mixed. This rapid mixing is mainly due to the vortex initiated
by eccentric arms (Fig. 12). The relative concentration in the
(x, y) plane sections along the z-axis is shown for the three
T-mixers in Fig. 13. As can be seen, from z = 30 mm, mixing
is completed and there is no longer any concentration gradient and segregation. Before z = 20 mm, c ranges from 0.3
to 0.7. Macro-mixing characteristics in the T-mixers seems
comparable.

2395

Fig. 11 Concentration gradients along z-axis.


The present experimental results show that characteristic induction time is equivalent when sols are synthesized
in Ts or Tn but shorter when synthesized in Tb (Fig. 5). In
addition, particles synthesized in Tb appear to be larger, less
stable and to agglomerate more rapidly. Mixing and particle
formation usually occur on a similar time-scale, as precipitation is generally limited by mixing time-scale (Schwarzer and
Peukert, 2002). Thus, initial intensity of macro-mixing determines the initial level of supersaturation. This was conrmed
by Wang et al. who performed direct numerical simulation of
the formation and growth of TiO2 nanoparticles in the twodimensional temporal mixing layer (Wang and Garrick, 2006).
Their numerical results show that macro-mixing has a strong
inuence on particle size as the largest particles are found in
the eddy core, while a high concentration of smaller particles
is found along the interface between reactant streams where
chemical reaction and nucleation takes place. In the present
study, mixing on a macro-scale seems the same in the three
mixers and is rapidly effective because of the vortex generated
in the upper region (Fig. 11). Then, one might suppose that
reaction and nucleation, which are known to be very rapid,
are initiated in the upper region of T-mixers. According to
Marchisio et al., better mixing favors nucleation with respect
to particle growth until a constant value is reached when the
characteristic mixing time becomes smaller than the typical
reaction time (Da < 1). When this mixing time is achieved, the

Fig. 12 Velocity eld (m/s) in the upper region of T-mixer:


vortex initiated by eccentric arms.

2396

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Fig. 13 Relative concentration evolution in (x, y) planes at several z locations.


authors explain that further improvement in mixing does not
affect the nal particle size distribution (Marchisio et al., 2009).
In this study, the situation is different as several geometries,
thus different hydrodynamic structures, were tested. The Da
number should be the same in the three T-mixers, as mixing
efciencies are the same in the three mixers, and could not
explain results presented in Fig. 5.
With comparable mixing efciency, the degree of supersaturation and thus the number of nuclei should be equivalent
in the three T-mixers. The primary particles nucleated from
solutions are known to grow by molecular addition or aggregation with small subunits. This last phenomenon and the
next one, namely agglomeration, are encouraged by increasing the number of collisions among suspended nanoscaled
particles. Indeed, particle growth and agglomeration can be
considered as transport-controlled for sufcient supersaturation levels. In this way, shear forces in the turbulent ow-eld
can have a signicant effect. Indeed, high turbulence level,
after primary particles are formed, can have a negative effect
on particle size as it enhances aggregation. Then, differences
observed in Fig. 5 could come from differences in hydrodynamics after nucleation. Fig. 14 is a map of several relevant
quantities regarding the hydrodynamic characterization in the
central plane (Oyz): turbulence kinetic energy (k), turbulence
energy dissipation rate () and local turbulence Reynolds number (Rel ). This gure conrms that hydrodynamics in the upper

region is the same for Ts and Tn . For these geometries, both


and k present high values in the vortex region and decrease
suddenly from z = 6 mm where it reaches lower values. Rel is
around 9 in the middle part of the region and then stabilizes
around 7. Concerning hydrodynamics in Tb , the presence of
a bafe at z = 11 mm has a clear effect. It creates local turbulence leading to bursts of and k and a rather homogeneous
Rel around 9. The local Reynolds number is an indicator of the
turbulence level, it is the ratio of the turbulence time scale (k/)
to the Kolmogorov time scale (/)1/2 . Rel = 1 thus corresponds
to a ow with only one time scale, that is to say laminar ow.
Here, there are differences in hydrodynamics in Tb compared
with hydrodynamics in Ts and Tn which present less smallscale turbulence. As some authors explained, the presence
of turbulence after the mixing is completed, enhances shearinduced diffusion (Scwarzer and Peukert, 2004). To obtain the
nest particles, solid formation should takes place in a perfectly mixed suspension and a slightly moving ow. Thus,
differences observed in Fig. 5 could be explained by a higher
level of turbulence after nucleation in Tb . On one hand it
increased the number of collisions between particles and on
the other hand, it could provide more water around the nuclei
(due to higher mixing rate) which enhanced the dielectric constant around the nuclei and developed an environment for
the nuclei to aggregate making the median diameter larger
(Nagasawa et al., 2007).

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

2397

Fig. 14 , k and Rel in the three T-mixers in the reaction region from z = 2 mm (bottom) to z = 18 mm (top).
Thus, the present results suggest that small-scale convection after nucleation, meaning micro-mixing, could have an
effect on particle stability and mean size. Indeed, particles
synthesized in the higher turbulence mixer are larger, less stable and tend to agglomerate more rapidly. Marchisio et al. have
investigated the effect of hydrodynamics on particle stability
with time with a quite similar vortex reactor. They showed for
a high Da number, the initial particle size is larger and particles tend to aggregate rapidly whereas when Da decreases,
initial size is smaller and particles are more stable over time
(Marchisio et al., 2009). In their case, increasing the inlet velocity leads to higher macro-mixing at the mixer entrance. Then
the nucleation of very ne and numerous particles, which
are more stable, occurs. With their geometry, the inuence of

turbulence after nucleation could not be tested. On the contrary, the present study enables the conrmation of the effect
of shear-induced diffusion. These experimental results underline the importance of a well-tailored mixer which is not easy
to produce, as reaction time-scales are not well-known.
Differences observed in Tb compared with Tn and Ts
suggest that nucleation occurred around the rst bafe (at
z = 11 mm) and then growth proceeded after this bafe (for the
current inlet velocity). Then, knowledge of residence time in
this region would enable the typical particle formation time
to be estimated.
The particle tracing function of Comsol software was used
in order to evaluate the residence time in the reaction region.
Here, particle tracing is used as a tool to simply estimate an

2398

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Table 1 Mean residence time and standard deviation


for z = 2, z = 11 mm, z = 20 mm and z = 200 mm for 200
particles.

tz = 2 (ms)
 (ms)
tz = 11 (ms)
 (ms)
tz = 20 (ms)
 (ms)
tz = 200 (ms)
 (ms)

Fig. 15 Example of particle trajectory in Tb .


order of magnitude of residence times in the T-mixer and
not as a residence time distribution prediction tool which
requires several thousand particles and long computation
times. Hence, about 200 massless particles were injected from
different locations at the two inlet (y, z) planes (x = 20, x = 20).
Fig. 15 represents an example of the trajectory for one particle
in Tb and Table 1 presents the synthesis of mean residence
time at the vortex outlet, z = 2 mm, tz=2 , at z = 11 mm (rst
bafe location), tz=11 , at z = 20 mm, tz=20 , and at the mixer outlet z = 200 mm, tz=200 , and the standard deviation  for these
values. Depending on the T-mixer geometry, residence time
in the mixer is between 37.0 and 113.3 ms. When particles
leave the reaction zone, residence time is between 13.6 and
15.7 ms. Mean residence time at z = 11 mm in Tb could be used
to estimate an order of magnitude of the characteristic particle formation time. After entering the T-mixer, the mean time
before molecules reach the vortex is 4.8 ms and thus characteristic particle formation time is estimated at around 6 ms.
Although such a value is very important for the process of
scale-up, only a few quantitative values are available in the

Ts

Tb

Tn

5.9
1.3
11.0
1.9
15.7
2.2
113.3
0.7

5.9
1.0
11.0
2.5
13.6
2.8
99.8
11.0

5.9
1.1
10.1
1.9
15.4
1.8
37.1
4.4

literature. Table 2 summarizes the characteristic particle formation time-scale found in the literature for solgel synthesis.
The value of this typical time is very dependent on operating
conditions. The order of magnitude found in this study seems
consistent with values available in the literature. Nonetheless,
it can be noted that the characteristic time found in this study
is not in accordance with the rst hydrolysis time of about several tens of milliseconds given by Rivallin et al. (2005) in similar
operating conditions. Based on this latter value, the reaction
zone, and thus the particle formation area, was expected after
z = 40 mm, which justied the initial bafe and narrow pipe
location.
These results conrm the rapidness of the solgel reaction
which is limited by the mixing rate. Adequate mixers which
can achieve an efcient mixing within a few milliseconds must
be found. T-mixers like the one presented could be optimized
for the velocity range tested. Other static mixers must be considered with well-located turbulence promoters. CFD can be
a reliable tool in order to calibrate appropriate mixers for a
specic nanoparticle precipitation application.
Precipitation is a complex phenomenon; the particle eld
and its evolution reect spatial variations of convection, diffusion and chemical reaction. It is well-known that solgel
synthesis is based on the hydrolysis and polycondensation of
metal compound precursors and that these two types of reactions occur simultaneously in a conventional solgel method,
i.e. polycondensation begins before complete hydrolyzation of
the metal compound. This leads to a complicated state around
the metal ion which depends on the synthesis parameters.
The way those parameters affect the process is still not clear.
Rare studies about the mixing effect during particle synthesis
have all underlined the importance of controlling mixing in
order to get the desired particle size distribution. The present
study proves that the mixing quality might have an effect not
only on particle size but also on their stability.

Table 2 Estimation of typical particle formation time under various operating conditions.
Reference

Azouani et al. (2010)


Marchisio
et al.
(2006)
Marchisio et al. (2009)
Marchisio et al. (2008)
Present study

Typical particle
formation time
(ms)
1.05

7.7
1.7
3.2
30
59
6

Operating conditions

TiO2 , CTTIP = 0.146 M, H = 1.9, T = 298 K


Precipitation of barium sulfate:
A + B solid
ca = 100 mol/m3 , cb = 100 mol/m3
ca = 500 mol/m3 , cb = 500 mol/m3
ca = 100 mol/m3 , cb = 500 mol/m3
TiO2 , CTTIP = 1 mol/l, H = 2, hypochloric acid concentration: 0.25 mol/l
TiO2 , CTTIP = 0.5 mol/l, H = 4, hypochloric acid concentration: 0.25 mol/l
TiO2 , CTTIP = 0.153 mol/l, H = 2.46

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Thus, when synthesizing nanoparticles, mixing, which is a


key parameter which must be optimized as a small difference
in the rst particle formation step, results in signicant differences in particle size, particle number density and hydrolyzed
state. Particle distribution size and stability could be determined by both kinetics of solid formation and residence time
distribution of individual particles.

5.

Conclusion

In this work, the role of hydrodynamics during production


of titanium dioxide nanoparticles via solgel reaction in Tmixers was investigated. For the rst time three different
T-mixer geometries were tested. This study was carried out
quantitatively by using three T-mixers with turbulence promoters and thus different mixing characteristics. The main
novelty of this works is in the investigation of the effect of
hydrodynamics on both nanoparticle morphology and photocatalytic activity. In addition, this analysis is supported by
CFD simulations of the velocity and turbulent elds in three
T-mixers.
Results show that mixing plays a different role according
to mixing-scale. All studies about nanoparticles precipitation
agree about formation steps: rst a burst of particles is produced in a very short time then the sol changes through
aggregation. The macro-mixing was comparable in the three
studied T-mixers. A high nucleation rate leads to small particles: on a large scale, the better the mixing is performed, the
smaller the created particles will be. Besides, micro-mixing
could affect the evolution of particles which will be less stable if shear induced diffusion is promoted during particle
growth. Our study showed, for the rst time, that increasing
turbulences after the nucleation had an effect on the particle
stability since the induction time for Tb was shorter (1500 s)
than for Ts and Tn (2800 s).
The particle distribution was the same for the three Tmixers which presented different hydrodynamics in the outlet
pipe but the same inlet Re number. Hence it was concluded
that the particle size distribution was probably more dependent on the inlet Re number than the T-mixer geometry.
The solgel TiO2 nanoparticles from the T-mixers were
used to coat alumina support. In all cases, a homogeneous
thin lm of about 200 nm was obtained. The deposited mass
of TiO2 was very similar.
The photocatalytic efciency of TiO2 coating was tested by
following the degradation of AO7 in an aqueous solution. The
degradation and mineralization kinetic rate were very close in
the three cases. So the difference in the hydrodynamics of the
T-mixers did not directly inuence the photocatalytic activity
of the TiO2 coating.
In this study, a typical particle precipitation time was
estimated at around 6 ms (CTTIP = 0.153 mol/L, H = 2.46). This
contribution to the particle precipitation time characterization could help during future process scale-up. Indeed, mixing
must be considered as an operating parameter during synthesis and so be actively used to produce the desired product:
a non-optimized mixing can greatly affect one of the main
advantages of TiO2 , the relatively high surface area to volume ratio. CFD calculation is an efcient tool, the use of which
could be of great benet during process scale-up so as to optimize the design of mixing geometries. The results presented
here also provide evidence for the need to perform population
balance calculations in parallel with particle size distribution
measurements in order to better understand and distinguish

2399

the relative effect of macro and micro-mixing on the nal


product.

References
c,
B., Soji
D., Despotovic,
V., Vione, D., Pazzi, M.,
Abramovic,
Csandi, J., 2011. A comparative study of the activity of TiO2
Wackherr and Degussa P25 in the photocatalytic degradation
of picloram. Appl. Catal. B 105, 191198.
Antoniou, M.G., Nicolaou, P.A., Shoemaker, J.A., De la Cruz, A.A.,
Dionysiou, D.D., 2009. Impact of the morphological properties
of thin TiO2 photocatalytic lms on the detoxication of water
contaminated with the cyanotoxin, microcystin-LR. Appl.
Catal. B 91, 165173.
Azouani, R., Michau, A., Hassouni, K., Chhor, K., Bocquet, J.-F.,
Vignes, J.-L., Kanaev, A., 2010. Elaboration of pure and doped
TiO2 nanoparticles in solgel reactor with turbulent
micromixing: application to nanocoatings and photocatalysis.
Chem. Eng. Res. Des. 88, 11231130.
Bauer, C., Jacques, P., Kalt, A., 2001. Photooxidation of an azo dye
induced by visible light incident on the surface of TiO2 . J.
Photochem. Photobiol. A 140, 8792.
Bothe, D., Stemich, C., Warnecke, H.-J., 2006. Fluid mixing in a
T-shaped micro-mixer. Chem. Eng. Sci. 61, 29502958.
Brosillon, S., Lhomme, L., Vallet, C., Bouzaza, A., Wolbert, D., 2008.
Gas phase photocatalysis and liquid phase photocatalysis:
interdependence and inuence of substrate concentration
and photon ow on degradation reaction kinetics. Appl. Catal.
B 78, 232241.
Chebli, D., Fourcade, F., Brosillon, S., Nacef, S., Amrane, A., 2011.
Integration of photocatalysis and biological treatment for azo
dye removal application to AR183. Environ. Technol. 32,
507514.
Comninellis, A., Kapalka, A., Malato, S., Parsons, S.A., Poulios, I.,
Mantzavinos, D., 2008. Advanced oxidation processes for
water treatment: advances and trends for R&D. J. Chem.
Technol. Biotechnol. 83, 769776.
Coronado, J.M., Zorn, M.E., Tejedor, I., Anderson, M.A., 2003.
Photocatalytic oxidation of ketones in the gas phase over TiO2
thin lms: a kinetic study on the inuence of water vapor.
Appl. Catal. B 43, 329344.
Faisal, M., Tariq, M.A., Khan, A., Umar, K., Muneer, M., 2011.
Photochemical reactions of 2,4-dichloroaniline and
4-nitroanisole in aqueous suspension of titanium dioxide. Sci.
Adv. Mater. 3, 269275.
Fujishima, A., Hashimoto, K., Watanabe, T., 1999. TiO2
Photocatalysis Fundaments and Applications. BKC Inc., Tokyo.
Gradl, J., Schwarzer, H.S., Schwertrm, F., Manhart, M., Peukert,
W., 2006. Precipitation of nanoparticles in a T-mixer: coupling
the particle population dynamics with hydrodynamics. Chem.
Eng. Process. 45, 908916.
Hazah, N., Sopyan, L., 2009. Nanosized TiO2 photocatalyst
powder via solgel method: effect of hydrolysis degree on
powder properties. Int. J. Photoenergy,
http://dx.doi.org/10.1155/2009/962783.
Harris, M.T., Byers, H., 1988. Effect of solvent on the
homogeneous precipitation of titania by titanium ethoxide
hydrolysis. J. Non-Cryst. Solids 103, 4964.
Hatat Fraile, 2013. Etude des mthodes dlaboration et de la
mise en uvre de photocatalyseurs pour le traitement de la
micro-pollution bio-rfractaire dans leau. Universit
Montpellier 2, France (PhD thesis).
Henderson, M.A., 2011. A surface science perspective on TiO2
photocatalysis. Surf. Sci. Rep. 66, 185297.
Konstantinou, I.K., Albanis, T.A., 2004. TiO2 -assisted
photocatalytic degradation of azo dyes in aqueous solution:
kinetic and mechanistic investigations. A review. Appl. Catal.
B 49, 114.
Lhomme, L., Brosillon, S., Wolbert, D., Dussaud, J., 2005.
Photocatalytic degradation of phenylurea, chlortoluron, in
water using an industrial titanium dioxide coated media.
Appl. Catal. B 61, 227235.

2400

chemical engineering research and design 9 1 ( 2 0 1 3 ) 23892400

Lhomme, L., Brosillon, S., Wolbert, D., 2006. Photocatalytic


degradation of a triazole pesticide, cyproconazole, in water. J.
Photochem. Photobiol. A 188, 3442.
Lin, H., Huang, C.P., Li, W., Ni, C., Ismat Shah, S., Tseng, Y.H., 1998.
Size dependency of nanocrystalline TiO2 on its optical
property and photocatalytic reactivity exemplied by
2-chlorophenol. J. Phys. Chem. B 102, 1087110878.
Lin, Y., Ferronato, C., Deng, N., Chovelon, J.M., 2011. Study of
benzylparaben photocatalytic degradation by TiO2 . Appl.
Catal. B 104, 353360.
Liu, Y., Fox, R.O., 2005. CFD predictions for chemical processing
in a conned impinging-jets reactor. AIChE J. 52, 731744.
Marchisio, D.L., Rivautella, L., Barresi, A., 2006. Design and
scale-up of chemical reactors for nanoparticles precipitation.
AIChE J. 52, 18771887.
Marchisio, D.L., Omegna, F., Barresi, A.A., Bowen, P., 2008. Effect
of mixing and other operating parameters in solgel
processes. Ind. Eng. Chem. Res. 47, 72027210.
Marchisio, D.L., Omegna, F., Barresi, A.A., 2009. Production of
TiO2 nanoparticles with controlled characteristics by means
of a vortex reactor. Chem. Eng. J. 146, 456465.
Mills, A., ORourke, C., Kalousek, V., Rathousky, J., 2012.
Adsorption and photocatalytic and photosensitised bleaching
of acid orange 7 on multilayer mesoporous lms of TiO2 . J.
Hazard. Mater. 211/212, 182187.
Molinari, R., Borgese, M., Drioli, E., Palmisano, L., Schiavello, M.,
2002. Hybrid processes coupling photocatalysis and
membranes for degradation of organic pollutants in water.
Catal. Today 75, 7785.
Molinari, R., Caruso, A., Poerio, T., 2009. Direct benzene
conversion to phenol in a hybrid photocatalytic membrane
reactor. Catal. Today 144, 8186.
Nagasawa, H., Tsujiuchi, T., Maki, T., Mae, K., 2007. Controlling
ne particle formation processes using a concentric
microreactor. AIChE J. 53, 196206.
Nakata, K., Ochiaia, T., Murakamia, T., Fujishima, A., 2012.
Photoenergy conversion with TiO2 photocatalysis: new
materials and recent applications. Electrochim. Acta 84,
103111.
Oturan, M.A., Oturan, N., Edelahi, M.C., 2011. Oxidative
degradation of herbicide diuron in aqueous medium by
Fentons reaction based advanced oxidation processes. Chem.
Eng. J. 171, 127135.
Plantard, G., Goetz, V., Correia, F., Cambon, J.P., 2011. Importance
of a mediums structure on photocatalysis: using TiO2 -coated
foams. Sol. Energy. Mater. Sol. Cells 95, 24372442.
Poling, J.P., Prausnitz, J.M., OConnell, J., 2004. The Properties of
Gazes and Liquids. McGraw Hill, New-York.
Pucher, P., Benmami, M., Azouani, R., Krammer, K., Chhor, K.,
Bocquet, J.F., Kanaev, A., 2007. Nano-TiO2 sols immobilized on
porous silica as new efcient photocatalyst. Appl. Catal. A
332, 297303.
Rivallin, M., Benmami, M., Gaunand, A., Kanaev, A., 2004.
Temperature dependence of the titanium oxide sols
precipitation kinetics in the solgel process. Chem. Phys. Lett.
398, 157162.
Rivallin, M., Benmami, M., Kanaev, A., Gaunand, A., 2005. Solgel
reactor with rapid micromixing: modelling and
measurements of titanium oxide nano-particle growth.
Chem. Eng. Res. Des. 83, 6774.

Romanosa, G.E., Athanasekoua, C.P., Katsarosa, F.K.,


Kanellopoulosa, N.K., Dionysioub, D.D., Likodimosa, V.,
Falaras, P., 2012. Double-side active TiO2 -modied
nanoltration membranes in continuous ow photocatalytic
reactors for effective water purication. J. Hazard. Mater.
211212, 304316.
Satuf, M.L., Pierrestegui, M.J., Rossini, L., Brandi, R.J., Alfano, O.M.,
2011. Kinetic modeling of azo dyes photocatalytic
degradation in aqueous TiO2 suspensions. Toxicity and
biodegradability evaluation. Catal. Today 161, 121126.
Schwarzer, H.C., Peukert, W., 2002. Experimental investigation
into the inuence of mixing on nanoparticle precipitation.
Chem. Eng. Technol. 25, 657661.
Scwarzer, H.C., Peukert, W., 2004. Combined
experimental/numerical study on precipitation of
nanoparticles. AIChE J. 50, 32343247.
Shinde, P.S., Patil, P.S., Bhosale, P.N., Brger, A., Nauer, G.,
Neumann-Spallart, M., Bhosale, C.H., 2009. UVA and solar
light assisted photoelectrocatalytic degradation of AO7 dye in
water using spray deposited TiO2 thin lms. Appl. Catal. B 89,
288294.
Soloviev, A., Jensen, H., Sogaard, E.G., Kanaev, A.V., 2003.
Aggregation kinetics of solgel process based on titanium
tetraisopropoxide. J. Mater. Sci. 38, 33153318.
Stylidi, M., Kondarides, D.I., Verykios, X.E., 2003. Pathways of
solar light-induced photocatalytic degradation of azo
dyes in aqueous TiO2 suspensions. Appl. Catal. B 40,
271286.
Tian, J., Chen, L., Yin, Y., Wang, X., Dai, J., Zhu, Z., Liu, X., Wu, P.,
2009. Photocatalyst of TiO2 /ZnO nano composite lm:
preparation, characterization, and photodegradation activity
of methyl orange. Surf. Coat. Technol. 204, 205214.
Turova, N.Y., Turevskaya, E.P., Kessler, V.G., Yanovskaya, M.I.,
2002. The Chemistry of Metal Alkoxydes. Kluwer Academic
Publishers, Dordrecht.
Velegraki, T., Poulios, I., Charalabaki, M., Kalogerakis, N.,
Samaras, P., Mantzavinos, D., 2006. Photocatalytic and
sonolytic oxidation of acid orange 7 in aqueous solution. Appl.
Catal. B 62, 159168.
Vinodgopal, K., Wynkoop, D., Kamat, P., 1996. Environmental
photochemistry on semiconductor surfaces: photosensitized
degradation of a textile azo dye, acid orange 7, on TiO2
particles using visible light. Environ. Sci. Technol. 30,
16601666.
Wang, G., Garrick, S.C., 2006. Modeling and simulation of titania
formation and growth in temporal mixing layer. J. Aerosol Sci.
37, 431451.
Zhang, F., Zhao, J., Shen, T., Hidaka, H., Pelizzetti, E., Serpone, N.,
1998. TiO2 -assisted photodegradation of dye pollutants. II.
Adsorption and degradation kinetics of eosin in TiO2
dispersions under visible light irradiation. Appl. Catal. B 15,
147156.
Zhang, L., Kanki, T., Sano, N., Toyoda, A., 2003. Development of
TiO2 photocatalyst reaction for water purication. Sep. Purif.
Technol. 31, 105110.

c Stangar,
J.,
Zita, J., Krysa,
J., Cernigoj,
U., Lavrenci
U., Jirkovsky,
J., 2011. Photocatalytic properties of different TiO2
Rathousky,
thin lms of various porosity and titania loading. Catal. Today
161, 2934.

You might also like