Photocatalytic Reactors

Vol.
05 INTERNATIONAL JOURNAL OF PHOTOENERGY 2003

Photocatalytic reactors
M. Bouchy
and O. Zahraa
Dpartement de Chimie Physique des Ractions - UMR 7630 CNRS-INPL

ENSIC, 1 rue Grandville, BP 451, F-54001 - Nancy Cedex, France
Abstract. Fundamental concepts for the design of photocatalytic reactors are recalled: specic rate, mod-
eling of the distribution of the irradiation, description and modeling of transport phenomena on the macro-
scopic and macroscopic scales. These are required for establishing the reactor mass balance. A brief survey
of various reactors is given for solar or articial light, gas- or liquid-phase, slurry or xed-bed.
1. INTRODUCTION
Photocatalysis has been a subject of increasing interest
during the last twenty years. It has indeed various po-
tential applications such as metal recovery, abatement
of NO
x
, synthesis of ammonia but the recent develop-
ments have been induced mainly by the application to
the abatement of organic pollutants both in the aque-
ous and the gaseous phase. This development partic-
ipates to the general concern about the environment
and the increasing constraints on the toxicity of wastes.
The academic studies have been much developed in the
elds of organic chemistry, physical chemistry, chem-
ical kinetics and catalyst preparation, but few studies
have been dedicated to chemical engineering and the
development of reactors. Concomitantly, fewindustrial
processes have been developed.
The aimof this presentation is to give some aspects
of studies related to chemical engineering of photo-
catalysis; far from being exhaustive, it aims at giving
an insight into the extent and importance of this topic.
The presentation will deal with three selected sub-
jects: light distribution within a reactor, mass trans-
fer processes and a brief survey of photochemical
reactors.
2. BASIC CONSIDERATIONS
2.1. Photocatalytic process. The photocatalytic
process results from the excitation by UV-visible light
of a solid semiconductor, generating free electrons and
holes, which leads to redox processes on the surface
and the attack of adsorbed molecules. Direct attack
of organic molecules on the surface, or formation
of highly reactive hydroxyl radicals results, in the
presence of oxygen, in the oxidation of the organic
molecules, which, in most cases, leads to a complete
mineralization.
E-mail: [email protected]
E-mail: [email protected]
2.2. Rate of photocatalytic degradation. The
rate of photocatalytic degradation depends on several
parameters for a given pollutant and catalyst which are
mainly:
pollutant, oxygen (and other compounds) con-
centrations,
temperature,
irradiation wavelength and ux.
2.3. Specic rate. These parameters are not usu-
ally uniform within the reactor. At any given point, the
rate per unit catalyst amount q, or specic rate r
spe
,
takes a local value so that the total rate in the reactor
is the integration of the elementary rates dr = r
spe
dq
over elementary amounts dq of catalyst.
The eect of the irradiation on this elementary cata-
lyst amount is determined by the specic rate of energy
absorption i
abs
= dI
abs
/dq, where I
abs
is the absorbed
energy (E s
1
or W), also called LVREA (Local Volumet-
ric Rate of Energy Absorption).
Note that this amount of catalyst can be chosen
as catalyst mass, suspension volume or support area
(immobilised catalyst), depending on the type of reactor
used.
2.4. General system. Another problem is that
mass transfer phenomena can intervene in the process
by reducing the eective local concentration so that the
analysis of the reactor must take into consideration:
the local specic rate,
the irradiation distribution (the space-dependent
LVREA),
the laws of mass-transfer phenomena (external,
internal transfers),
the mass-balance over the whole of the reactor.
A complete analysis allows accurate information on the
laboratory scale to be obtained and the scaling-up of an
eective reactor to be carried out. However, the whole
192 M. Bouchy and O. Zahraa Vol. 05
problemis complex and requires knowledge of the con-
stitutive laws and of the operative parameters so that
simplifying assumptions are often made so as to get a
feasible study and practical reactors are designed with-
out a complete ab-initio analysis.
3. SPECIFIC RATE
The basic information which allows a photocatalytic re-
actor to be designed is an analytical expression for the
specic rate of reaction. Data can be found in the liter-
ature but are often given as the total rate within the re-
actor, resulting froman average value of the irradiation
rate instead of taking the irradiation distribution into
account [1]. A general picture is as follows, although
some more elaborate rate expressions are also found:
a concentration dependence of the Langmuir-
Hinshelwood type [2] which leads to rst-order
kinetics at low concentration and zero-order
kinetics at high concentration, with a similar
additionnal eect of oxygen concentration, of-
ten referred to as the LHHW model (Langmuir-
Hinshelwood-Hougen-Watson),
a low activation energy,
an order with respect to absorbed irradiation
varying from 1 at low ux to 0.5 at high ux,
which is attributed to recombination of reactive
species (especially hole-electron recombination).
4. IRRADIATION DISTRIBUTION
There is no feed-back eect between irradiation and
chemical conversion as long as the reactants (or prod-
ucts) do not absorb the irradiation in the wavelength
range where the catalyst is excited. The catalyst con-
centration is therefore the only parameter and as this
catalyst is by essence a stable compound, the light
distribution does not depend on the extent of re-
action. This is indeed generally the case in low-
concentrated uids and in near-UV range (wavelength
of 350400nm). If this were not the case, the concentra-
tions and light distributions would be linked together
through the absorption law for the uid phase, which
would add to the complexity of the system.
4.1. Suspension. In the case of a suspension (or a
nearly-continuous xed bed), the LVREA depends on
the emission of the light source, both in intensity and
spatial distribution, and on the absorption and scatter-
ing properties of the reactor lling. If there is no ab-
sorption by the compounds in the uid, the LVREA is
dependent on the catalyst only. However, this calcula-
tion is much complicated by the scattering.
The master equation is an energy balance along a
path ds on any energy ux per unit surface, time and
solid angle I() (E m
2
sr
1
s
1
) along the direction
dened by the unit vector , which includes:
loss by absorption (coecient ),
loss by scattering-out (coecient ),
contribution by scattering-in from uxes in all
the directions
in space (probability dependent

on and on the angle between
and , with
the probability per unit solid angle (or phase-
function) p(
, ) [1, 3]:
dI()
ds
= ( +) I()
+
_
1
4
_
_
4
p(
, ) I(
)d
,
where d
is an element of solid angle in the di-

rection
.
To solve this equation requires experimentally deter-
mining or modelling the absorption and scattering co-
ecients and also the phase function p. On principle,
a theoretical expression for p can be derived from
MIE theory but due to the complexity of the deriva-
tion, approximate functions have been used (for large
spheres [4]) and also simple isotropic scattering.
In addition, solving of this integro-dierential equa-
tion requires knowledge of a boundary condition. This
is the incident ux I
w
() on the reactor walls. The
LVREA is then:
i
abs
=
_
4
Id.
Note also that the parameters are dependent on the
wavelength and that, for a polychromatic source
(sun, uorescent lamp) the spectral density I
() =
dI()/d and corresponding i
abs,
should be used in-
stead of I() and i
abs
, which would then be integrated
over the wavelength range.
The lamp irradiation ux on the reactor wall I
w
()
can be known from the geometry of the reactor and
the distribution of the emission from the light source.
The solar emission is well dened as a one-direction
uniform ux, but the emission of an articial lamp is
more dicult to assess. The lamps being considered
are of a cylindrical shape and have been simulated
as an isotropically emissive medium, either a cylindri-
cal volume (which could ultimately be considered as a
thread [5]) or a cylindrical surface (uorescent lamp).
The emission is on principle isotropic in the whole
space but a simplied model assumes emission to oc-
cur perpendicularly to the lamp axis only.
Once the system is dened, it is a dicult mathe-
matical problem to solve it. If no simplifying assump-
tions are made, a numerical method is necessary. The
Discrete Ordinate Method is a convergent algorithm
based on exchange between nite volume elements [1]
Vol. 05 Photocatalytic reactors 193
or discrete directions in space [3] (about 220). Alter-
natively, a Monte Carlo approach can be used [3, 5]:
in this method, the fate of individual photons is sim-
ulated then all these events are averaged over a large
number of photons to get the LVREA, a rigorous but
time-consuming method.
Analysis of the light distribution is of importance:
although a pseudo Beer-Lambert can be used to sim-
ulate absorption under some conditions (see below), it
can be observed that the increase in catalyst loading can
result in a maximum in the rate of the reaction instead
of a limiting value [6, 7].
4.2. Thin lm- deposited lm. A simplication of
the general treatment has been suggested [8] where the
contribution of scattering-in to the energy intensity can
be neglected by a careful design of the reactor (with
small light path), which is appropriate for laboratory
kinetic studies leading to a pseudo Beer-Lambert law.
Such an approximation, has also been used for thin de-
posits [5, 9], which gives a satisfactory representation
of the global properties of the lm as measured by an
integrating sphere [10]. A simple model [11] considers
a nite number of elementary slabs having the same
properties of absorption and reection and exchang-
ing light ows. In an alternative model, originally de-
signed for suspension lms [12], the light is assumed
to propagate normally to the lm plan only: scattering-
in is neglected and the total light ow is assimilated to
opposite ows coupled to each other by scattering-out.
This latter model has the advantage of leading to an
analytical expression for the LVREA and has also been
applied with some renement to lm deposits [13].
This model has proved satisfactory although out- and
in-scattering by thin deposits has been found to be
nearly isotropic [14]. Note that the determination of
the global properties of a deposited lm can also be
used to carry out the simulation of reactors equipped
with sets of transparent supports of moderately
absorbing lms.
5. MASS-TRANSFER PHENOMENA
5.1. Coupling with mass transfer. Any consump-
tion of reactant results in a decrease in the local
reactant concentration. A concentration gradient then
appears and results in a ux of matter. It can be as-
sumed that the QSSA (quasi-stationary-state approxi-
mation) is obeyed so that the photographic picture of
the local situation at any time reveals an exact mass-
balance i.e. an equality between transported reactant
through mass-transfer phenomena and consumed re-
actant through the photocatalytic process. The conse-
quence is that the local concentration in the immediate
vicinity of the catalyst surface is lower than it is in the
bulk of the uid.
Note that increasing the surface area of the catalyst
is protable as this increases the pollutant adsorption
and therefore its degradation and usual catalysts are
therefore of a porous or dispersed nature although this
increases the possibility of mass-transfer limitation.
Mass-transfer phenomena can be classied as exter-
nal, within the uid in the vicinity of the catalyst pel-
lets or deposits (then enhanced by uid stirring), and as
internal within the catalyst pores (independent of the
stirring).
5.2. External mass transfer. The rate of external
mass transfer is generally expressed according to the
model of the boundary layer on the catalyst surface.
The rate is then expressed as:
r = k
ext
(C
b
C
s
) (mol m
2
s
1
),
where C
b
and C
s
represent the bulk and surface con-
centrations, respectively.
The rate constant k
ext
depends on the ow regime
of the uid, generally characterised by the non-
dimensional Reynolds number Re. It is given by various
empirical expressions such as [9]:
k
ext
= 3.49 10
7
Re
0.77
(m s
1
).
As the rate constant k
ext
increases with the ow turbu-
lence, a good stirring is usually enough for the eect of
external transfer to vanish, i.e. (C
b
C
s
)/C
b
1. Exper-
imentally, this can be veried by checking that the rate
is not dependent on the ow rate or stirring beyond
some minimum value.
In some instances, the oxygen supply has been sug-
gested as a source of rate limitation. Note that the ow
in oxygen to the solid particle or deposit results not
only from external transfer but also, for aqueous solu-
tion treatment with bubbling gas, from dissolution at
the gas-liquid interface. This can to be taken into ac-
count [5] in mass balance with a specic rate of disso-
lution given by the basic expression:
r = k
G/L
_
C
G/s
C
b
_
(mol m
2
s
1
),
where k
G/L
is the rate constant and C
G/s
the oxygen con-
centration in the liquid phase in equilibrium with the
gas phase by the gas-liquid interface.
5.3. Internal mass transfer. Internal mass-transfer
is more likely to result in rate limitation in a deposited
catalyst than it does on a suspended catalyst [15]. Con-
trary to external mass-transfer, it is not possible to ob-
tain direct evidence of a limitation as the parameters
involved (dimension of pores, molecular diusion coef-
cient) cannot be varied independently of other kinetic
parameters. An indirect way of investigation is to vary
the thickness of the catalyst lm (an increasing eect
of internal mass transfer would lead to a limiting value
of the quantum yield of degradation), or to compare
the eect of irradiation from the uid side and from
the (transparent) support side in which case an eect
of internal mass transfer would lead to a lower quan-
tum yield of degradation. A measurable [9] or negligi-
ble eect [11] has been observed by this technique. The
comparison of the rate of adsorption r
ads
and of pho-
tocatalytic degradation r
deg
can also be compared as
r
ads
is mainly diusion controlled (in physical adsorp-
tion). A partial limitation in suspension of aggregates
has thus been shown [6].
The consequence of a mass transfer rate limitation
is to alter the true value of the orders with respect to
the reactant and the irradiation rate (with a limiting
value of 1 for reactants and 0 for the irradiation rate).
As a consequence, the eect of internal transfer, if not
specically studied, is often included in an apparent
specic rate of reaction.
The phenomenon has been modelled either by as-
suming an exponential expression for the decrease in
concentration with the increase in depth within the cat-
alyst [9].
In thermal catalysis, the non-dimensional Thiele
and Weisz moduli and
are dened:
2
=
r
s
L
2
DC
s
,
= r
L
2
DC
s
,
where r
s
and C
s
are the specic volume rate and the
concentration at the surface of the lm, respectively,
D is the diusion coecient and L is the characteristic
length (ratio of volume to outer area), and where
makes use of the eective mean rate r which is an

experimental value.
Models in thermal catalysis correlate to these mod-
uli the eectiveness of the catalyst, that is the ratio
= r/r
s
. The eectiveness is equal to or lower than
unity as the concentration within the catalyst is lower
than C
s
if internal diusion interferes. This description
cannot be directly applied to photocatalysis as a gradi-
ent in LVREA can take place within the catalyst, but a
numerical treatment has been derived for cylindrical
pores in the form of an abacus () for various values
of the optical density [16].
6. REACTOR MASS BALANCE
The local specic rate of reaction r
spe
is a function of
the local bulk concentration C
b
and LVREA. The global
mass-balance on the reactants then couples the reac-
tant disappearance with the mass transfer over the
whole reactor, and leads to the conversion yield.
Ideal reactors can be classied into three categories:
perfectly-stirred closed reactor (batch reactor),
perfectly-stirred open reactor (CSR),
plug-ow reactor (PFR, or integral reactor).
Concentrations are assumed to be uniform in the
rst two types. On the contrary, they vary during the
progress of the uid within the PFR reactor so that the
mass balance must take into account the ow of the
reactive medium. Non-ideal reactors will include addi-
tional features such as lateral mass transport, or seen
as a compound system (reactor with mixing tank). In
real reactors the mass transport is bound to the uid ve-
locity distribution. Perfectly laminar or turbulent ows
are often considered, as real ow computation through
Navier-Stokes equation is a very complex problem on
its own.
A complete ab-initio modelling of the reactor has
seldom been carried out but a chosen example will be
given as an illustration in the case of a xed-bed reactor
where the catalyst is deposited on packed rings in an
annular reactor of cylindrical symmetry [5]. Note that if
the rings are small enough, the bed is quasi-continuous
so that the treatment is identical to that of a suspen-
sion.
The master equation (1) of mass balance for each
reactant is written as follows (bulk concentration C, ra-
dial variable r and longitudinal variable z):
_
v
L
dC
b
dz
D
__
1
r
dC
b
dr
+
d
2
C
b
dr
2
__
h
up
=r
spe
a
cat
k
G/L
(C
G/s
C
b
)a
gas
(mols
1
m
3
). (1)
The left-hand side represents the dierential ow of
matter in an tore-shaped element of section dz dr.
According to the QSSA, this is equal to the consump-
tion in the same element of space given in the right-
hand side expression. The terms of the left-hand side
represent the longitudinal transport of matter (inter-
stitial longitudinal rate v
L
and the radial diusion
(eective diusion constant D). The terms of the right-
hand side represent the reactant consumption (where
r
spe
(mol s
1
m
2
) is the specic supercial rate and
the stoichiometric coecient) and the appearance (in
the case of oxygen only) through gas-liquid transfer.
The balance includes weighing terms: a
cat
the catalyst
area, a
gas
the gas-liquid interface area and h
up
the liquid
hold-up per unit volume. Note that r
spe
is a function of
the concentrations C
s
of reactant and oxygen by the sur-
face catalyst and of the LVREA (modelled separately).
In addition, as seen above, external mass-transfer
in the boundary layer around the particles veries the
QSSA:
k
ext
(C
b
C
s
) = r
spe
(mol s
1
m
2
). (2)
Solving (by a numerical method) the whole system of
equations of types (1) and (2) for reactant and oxy-
gen, combined with the expression for the r
spe
includ-
ing the LVREA requires boundary conditions which are
the concentration at the entrance to the reactor and a
zero value of (dC
b
/dr) on the reactor walls (no ux of
matter).
The authors note the importance of the radial dif-
fusion as the LVREA generates a sharp decrease in the
rate of the reaction from inner to outer walls. Due to
the complexity of the system model simplications are
suggested and validated. A rst simplication is done
for a highly absorbing medium: the reactor is divided in
two zones, a zone by the inner wall where the LVREA is
high and the reaction occurs, supplied by mass transfer
fromthe outer second zone. Asecond simplication for
a lowabsorbing mediumis to assume a uniformLVREA.
This example deals with a turbulent owwhere the uid
velocity is uniform. This is not the case for a laminar
ow where the velocity follows a parabolic prole (slit
ow or falling lm reactor). An analysis has been de-
veloped [17] for these cases assuming the annular gap
to be small enough to be treated as a planar system
(i.e. cancelling of the rst diusion term in (1)), which
showed the advantage of a falling lm on a slit ow.
Note that the treatments shown here assume that
only a surface reaction occurs. This is only approximate
as the actual reaction can have a homogeneous part,
and be partly photolytic if a reactant absorbs the irra-
diation light, which requires additional terms of com-
pound consumption in the right-hand side of (1) [17].
7. GEOMETRY OF THE REACTOR
Two main features determine the design of the reactor:
suspended or immobilised catalyst,
source of irradiation.
The obvious advantage of an immobilised catalyst is to
avoid the separation process after the chemical treat-
ment, a process which can be an expensive (for the
ltration) and time consuming (for sedimentation) con-
sidering the small size of the catalyst crystals or aggre-
gates (0.05m to a few m). However, mass transfer
limitations are more likely to occur within a coat of im-
mobilised catalyst. Various types of support have been
used, made mainly of glass (plate, beads, breglass mat)
but also paper, cotton, bres.
The source of irradiation can be either natural (so-
lar irradiation) or articial (lamp). The geometry of the
reactor is strongly related to the source of irradiation,
in particular it has to be designed so as to collect the
maximum of emitted light so as to reduce the energy
and investment cost.
7.1. Solar reactors. Considering the unidirec-
tional solar irradiation, the collection of the irradiation
is carried out in two ways [18]:
use of a xed reactor where a large surface of
the solution or suspension is exposed to the
sun: trickling on a at plate supporting a cata-
lyst mesh, owing a sinuous path [19], owing
through an array of pipes so as to increase the
residence time, or through a at xed-bed [20] or
exposure within a tank such as those designed
for water stations, which can be equipped with
impellers designed to ensure a good mixing and
aerating of the slurry [7, 21],
focusing with a mirror (parabolic trough [22])
or more elaborate system (Fresnel lenses, holo-
graphic collector) on a reactor of small dimen-
sion.
When using a focusing system, the volume of uid is
reduced, which makes it easier to build, and is also a
necessity as the whole system(collector + reactor) must
be placed on a motorised support so as to track the sun.
However, it leads to a high irradiation ux, which may
reduce the quantum yield of degradation. In addition,
depending on the weather, an important fraction of the
sunlight can be diuse instead of direct, which reduces
the eciency of the focusing system [7, 22].
An interesting compromise is the Compound
Parabolic Collector [22] which is xed, but ensures a
partial focusing and a moderate irradiation ux on the
catalyst.
7.2. Articial light reactors. Articial light sources
are on the contrary multidirectional. Two main solution
are:
to reect or focus the emitted light to a reac-
tor [4],
to wrap the light source with the reactor, such
as in the coaxial lamp system; this has the addi-
tional advantage that the back-scattering of light
by the catalyst is not a loss as this scattered light
re-enters the reactor.
The main articial sources are mercury lamps:
low pressure lamps (254nm) of high yield at the
plug (up to 0.5) of low power (about 30W/m)
which require quartz windows,
uorescent lamps (around 365nm), i.e. low pres-
sure lamps equipped with a uorescent coating,
with a decrease of the emitting power of about
50% (so-called black-light lamps),
mediumpressure lamp (UV-visible range), higher
in power (about to 3kW/m) but lower in emission
yield,
high-pressure lamp, nearly punctual, with low
part of the emission in the UV range.
High-pressure lamps have a high cost in energy as along
as UV light is looked for. Medium-pressure lamps have
a very high power which leads to a low quantum yield
of degradation unless the catalyst is placed at a large
enough distance to ensure a low enough irradiation
ux, which would be space consuming. According to
these considerations, uorescent lamps working with
conventional glass vessels are often suggested.
An alternative is the distribution of light over a
large area of catalyst by the use of coated optical -
bres or hollow tubes [23]. The diculty of these de-
signs is to collect the maximum amount of light from
the source and to ensure a distribution of irradiation of
the catalyst along the whole of the light conductor by a
careful design of the interface loss through refraction.
A summary of reactors is presented below, which al-
though not exhaustive gives an insight into the varia-
tion of designs, some of which have been the object of
advanced modelling when some others are mainly of an
intuitive design.
7.3. Aqueous suspension (slurry). Reactors with
suspended catalyst have been the object of fundamen-
tal studies (e.g. [1, 4, 5, 24]) with the aim of either get-
ting true kinetic data or allowing large reactors to be
designed. Several designs have been studied or sug-
gested. Suspended catalyst has often been used as a
ow in an annular space [1, 4] around a linear lamp
but a thorough study has been dedicated to a falling
lm reactor [24], where a thin lm runs freely on the
outer wall within the annular space. If the reaction is
fast enough, a radial concentration gradient in the re-
actants will take place, this can be reduced by creating
a turbulent ow by the presence of baes of various
design. A completely dierent design is the fountain
reactor [25] where a nearly planar horizontal lm is ex-
posed to articial or solar irradiation. These reactors
are meant for water treatment where conversion in the
reactor is usually very lowso that they are coupled with
a tank with a continuous recirculation of the uid. Mass
balance for the whole reactor has to take this situation
into account although in the case of a low conversion
and good stirring of the tank, the reactor is formally
analogous to a batch reactor.
7.4. Liquid-phase xed-bed reactor. Another de-
sign for cylindrical reactors is a xed-bed reactor where
the catalyst is xed onto a support so as to avoid a sep-
aration process. The support can be packed Rashing
rings [5], bre mesh whereas an intermediate design
between suspended and xed catalyst uses freely sus-
pended glass beads as catalyst support. Other reactors
use large sized supports such as the inner or outer
wall of an annular reactor [20]. Connement of the
catalyst onto a practically two-dimensional surface re-
quires a good mixing of the uid phase in the annular
space. Creating a turbulent ow by the use of baes
or foils [26], of a catalyst-coated ridged helix [27], or
by inducing a helicoidal swirl [20] is then important.
An original solution to this problem is also a vortex re-
actor where the rotating inner wall induces instability
of the ow due to centrifugal forces and the building-
up of successive vortex cells within the annular space
as shown in a detailed study [28]. Improving the con-
tact between uid and catalyst can also be obtained
by allowing the uid to pass through a catalyst loaded
mesh on successive at or conical supports around the
lamp [29].
Basically these reactors are built around a single
lamp and large reactors can be obtained by a parallel ar-
ray of elementary reactors. The design can be improved
by using an array of light sources in a large tank. Hydro-
dynamics of such a system is much more complicated
than in the case of cylindrical symmetry. Design of the
reactor can be helped by the use of uid dynamics
simulation. Such a study, which is very demanding in
computational equipment, has been demonstrated on
a structure consisting of a cylindrical tank equipped
with seven catalyst-coated hollow tubes acting as light
conductors [23]. As a homogeneous dark reaction can
also take place, it can be of interest to accommodate
successive periods of irradiation and thermal evolution
by a special design of the reactor [26].
7.5. Gas-phase reactors. Gas-phase reactors are
most frequently xed-bed reactors. Crossing of the cat-
alyst bed such as coated glasbre mesh is often envis-
aged as it does not lead to a high pressure drop on the
one hand and the bed is more resistant to the ow of
uid on the other. A set of closely spaced ns perpen-
dicular to the lamp can also be used [30]. The catalyst
bed can be simply a at mat supported on a rigid grid
across a pipe with a normal or high incidence irradia-
tion by a set of lamps [31]. Alternatively a system of
annular geometry can be used where the mesh is of
cylindrical shape surrounding the lamp, the gas ow-
ing through the mesh outwards [32].
An equivalent of the liquid suspension system is
the uidised bed where the ow of gas through a par-
ticle bed creates a pseudo-liquid state when the ow is
strong enough to break the particle packing. This has
the advantage of creating a lower pressure drop than
a xed bed and provides a good mass transfer. In the
rotating uidised bed, the centrifugal forces take the
place of the gravitational forces [32, 33]; this systemhas
the advantage of having a cylindrical geometry which
is adapted to usual lamps, and allows, by adjusting the
rotating speed, higher ow rates to be used.
8. CONCLUSION
Modelling photocatalytic reactors is of importance in
order to get true intrinsic kinetic parameters and to
carry out the simulation of industrial-scale reactors so
as to select or to improve the reactor design. Many
phenomena are involved in the modelling which re-
sults in the need for eective parameters and in the
complexity of the solving of the problem by numerical
methods. Although a complete treatment appears to
be necessary, simplied models have been suggested
to avoid tedious numerical treatments. Applied patents
often suggest qualitatively ecient set-up so as to abate
mass-transfer limitations or make the best use of irradi-
ation sources. A gap still exists between modelling and
practical design, which will gradually be reduced with
the development of chemical engineering studies in the
eld of photocatalysis.
REFERENCES
[1] R. J. Brandi, O. M. Alfano, and A. E. Cassano, Envi-
ron. Sci. Technol. 34 (2000), 2623, 2631.
[2] C. S. Turchi and D. W. Ollis, J. Catal. 122 (1990),
178.
[3] G. Scaglari, G. Camera-Roda, and F. Santarelli, Int.
Comm. Heat Mass Transfer 25 (1998), 651.
[4] C. A. Martin, G. Camera-Roda, and F. Santarelli,
Catalysis Today 48 (1999), 307.
[5] A. Alexiadis, G. Baldi, and I. Mazzarino, Catalysis
Today 66 (2001), 467.
[6] H. Y. Chen, O. Zahraa, M. Bouchy, F. Thomas, and
J. Y. Bottero, J. Photochem. Photobiol. A: Chem. 85
(1995), 179.
[7] J. Gimenez, D. Curco, and M. A. Queral, Catalysis
today 54 (1999), 229.
[8] C. A. Martin, M. A. Baltanas, and A. E. Cassano, J.
Phototochem. Photobiol. A: Chem. 94 (1996), 173.
[9] D. Chen, F. Li, and A. K. Ray, Catalysis Today 66
(2001), 475.
[10] S. M. Ould-Mame, O. Zahraa, and M. Bouchy, Int. J.
Photoenergy 2 (2000), 59.
[11] S. M. Ould-Mame, Contribution au gnie de la
dgradation photocatalytique de polluants or-
ganiques de leau sur TiO
2
en lit xe, Ph.D. Thesis,
Nancy, 1998.
[12] A. Brucato and L. Rizzuti, Ind. Eng. Chem. Res. 36
(1997), 4740, 4748.
[13] M. Corboz, private communication.
[14] H. Dora-Silva, M. Bouchy, and O. Zahraa, Entropie
228 (2000), 48.
[15] M. F. J. Dijksta, A. Michorius, H. Buwalda, H. J.
Panneman, J. G. M. Winkelmann, and A. A. C. M.
Beenackers, Catalysis Today 66 (2001), 487.
[16] M. E. Edwards, C. M. Villa, C. G. Hill Jr., and T. W.
Chapman, Ind. Eng. Chem. Res. 35 (1996), 712.
[17] D. F. Ollis and C. Turchi, Env. Prog. 9 (1990), 229.
[18] J. Blanco and S. Malato, Solar detoxication tech-
nology, Part B, Chapter 6, UNESCO Publication,
2001.
[19] B. Volker, M. Mller, D. Bahnemann, D. Weichgrebe,
and M. Brehm, Reaktoren fr die photokatalytische
Abwasserreinigung mit Stegmehrfachplatten als
Solarelementen, EP Patent 0, 738, 686, A1 (1996).
[20] A. Gonzalez-Martin, O. J. Murphy, and C. Salinas,
Photocatalytic oxidation of organics using a
porous titanium dioxide membrane and an e-
cient oxidant, US patent 6, 136, 186, 2000.
[21] G. Cooper and M. A. Ratcli, Method of decontam-
inating a contaminated uid by using photocat-
alytic particles, US patent 5, 294, 315, 1994.
[22] J. E. Pacheco, M. R. Prairie, and L. Yellowhorse, J.
Solar Energy Engineering 115 (1993), 123.
[23] U. Periyathamby and A. J. Ray, Chem. Eng. Technol.
22 (1999), 881.
[24] G. Li Puma and P. L. Yue, Chem. Eng. Sci. 53 (1998),
2993.
[25] G. Li Puma and P. L. Yue, Chem. Eng. Sci. 56 (2001),
2733.
[26] J. G. Sczechowski, C. A. Koval, and R. D. Noble,
Use of controlled periodic illumination for an im-
proved method of photocatalysis and an improved
reactor design, US patent 5, 439, 652, 1995.
[27] D. G. Ritchie, Photocatalytic uid purication ap-
paratus having helical non transparent substrate,
US patent 5, 069, 885, 1991.
[28] T. K. Sengupta, M. F. Kabir, and A. K. Ray, Ind. Eng.
Chem. Res. 40 (2001), 5268.
[29] H. L. De Lasa and J. Valladares, Photocatalytic re-
actor, US patent 5, 683, 589, 1997.
[30] J. Say, R. Bonnecaze, A. Heller, S. Sitkiewitz, E.
Heller, and P. Haugsjaa, Apparatus for photocat-
alytic uid purication, US patent 6, 063, 343,
2000.
[31] H. Ibrahim and H. De Lasa, Ind. Eng. Chem. res. 38
(1999), 3211.
[32] A. Tabatabaie-Raissi, N. Z. Muradov, and E. Martin,
Apparatus for low ux photocatalytic pollution
control, US Patent 6, 309, 611, B1 (2001).
[33] D. Firnberg and J. R. Fehlner, Rotating uidized
bed reactor with electromagnetic radiation source,
US patent 5, 374, 405, 1994.
Submit your manuscripts at
http://www.hindawi.com
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2013
Catalysts
Journal of
Carbohydrate
Chemistry
International Journal of
Analytical Chemistry
Volume 2013
ISRN
Organic Chemistry
The Scientifc
World Journal
Bioinorganic Chemistry
and Applications
Journal of
SPECTROSCOPY
ISRN
Physical Chemistry
Electrochemistry
Advances in
Physical Chemistry
ISRN
Chromatography
Spectroscopy
ISRN
Inorganic Chemistry
Journal of
Chemistry
Inorganic Chemistry
Photoenergy
Analytical Methods
in Chemistry
Journal of
Volume 2013
ISRN
Analytical
Chemistry
Chromatography
Research
International

Photocatalytic Reactors

Uploaded by

Copyright:

Available Formats

Photocatalytic Reactors

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Photocatalytic Reactors

Uploaded by

Copyright:

Available Formats

Vol.

05 INTERNATIONAL JOURNAL OF PHOTOENERGY 2003

Dpartement de Chimie Physique des Ractions - UMR 7630 CNRS-INPL

in space (probability dependent

is an element of solid angle in the di-

makes use of the eective mean rate r which is an

You might also like