Chemical Engineering and Processing 46 (2007) 781–789

A review of intensification of photocatalytic processes

Tom Van Gerven a,b,∗ , Guido Mul c , Jacob Moulijn c , Andrzej Stankiewicz b
aDepartment of Chemical Engineering, K.U.Leuven, de Croylaan 46, 3001 Leuven, Belgium
b Process & Energy Department, Delft University of Technology, Leeghwaterstraat 44, 2628 CA Delft, The Netherlands
c DelftChemTech Department, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands

Received 30 March 2007; received in revised form 16 May 2007; accepted 26 May 2007
Available online 6 June 2007

Abstract
Photocatalysis is an attractive technology with potential applications in various disciplines, such as chemical synthesis, environmental technology
and medicine, and receives an impressive amount of exposure in the open literature. However, industrial implementation remains limited due to
scale up problems and the design of photoreactors. In this paper an overview is presented of recent advances in the design and application of novel
reactors and devices.
Two issues are essential: photon transfer limitations and mass transfer limitations (in the case of liquid phase reactions). In the field of mass
transfer optimisation, spinning disc reactors, monolithic reactors and microreactors have been investigated for their use in photocatalysis. Significant
advances are reported compared to conventional reactors. Studies focusing on performance improvement by optimising photon transfer, however,
remain limited. While optical fibers and LEDs have been explored, major breakthroughs are still lacking. More focus on the introduction of a
multitude of micro- or even nanoscale light emitting sources close to the catalyst particles is likely to be the way forward.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Photocatalysis; Photoreactor; Process intensification; Microreactor; Spinning disc reactor; Monolithic reactor; Optical fibers; Nanoscale illumination

1. Introduction as indicated in the following equation:

Photocatalysis implies the acceleration of a photoinduced
reaction by the presence of a catalyst [1]. Photoinduced reactions
are activated by absorption of a photon with sufficient energy,
i.e., equal or higher than the band-gap energy (Ebg ) of the cata-
lyst [2]. The absorption leads to a charge separation due to the
promotion of an electron (e− ) from the valence band of the semi-
conductor catalyst to the conduction band thus generating a hole
(h+ ) in the valence band. In order to have a photocatalyzed reac-
tion, the e− –h+ recombination, subsequent to the initial charge
separation, must be prevented as much as possible. Desired is the
reaction of the activated electron with an oxidant (Ox1 ), yielding
a reduced product (Red1 ), and the reaction of the generated hole
with a reductant (Red2 ) to produce an oxidized product (Ox2 ),
∗ Corresponding author at: Department of Chemical Engineering,
semiconductor
Ox1 + Red2 −→ Red1 + Ox2 (1)
light≥Ebg
The most widely used semiconductor catalyst in photoin-
duced processes is titanium dioxide (TiO2 ), because it is
chemically and biologically inert, photocatalytically stable, rel-
atively easy to produce and to use, able to efficiently catalyse
reactions, cheap and without risks to environment or humans
[2]. To enhance the performance of TiO2 , promoters have been
used, such as Pt and Ru [1,3]. The only disadvantage of TiO2 is
that it is not activated by visible light, but only ultraviolet (UV)
light. Other photocatalysts include ZnO, ZnS and CdS, but these
are much less active [3].
The use of light, either artificial or solar, for carrying out
chemical and biochemical reactions may render the process
more sustainable for two reasons [2,4]:

K.U.Leuven, de Croylaan 46, 3001 Leuven, Belgium. • process selectivity to the required products can drastically
Tel.: +32 16 32 23 42; fax: +32 16 32 29 91. increase (e.g., due to a different chemistry, or low/ambient
E-mail address: tom.vangerven@cit.kuleuven.be (T. Van Gerven). process temperature);

0255-2701/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.cep.2007.05.012
782 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789

• energy consumption in the process can drastically decrease Table 1

(e.g., due to the low temperature processing or use of solar Comparison of different photoreactor configurations
light). Photocatalytic Reference Catalyst coated surface
reactor per reaction liquid
Examples of photocatalysis can be found in many disci- volumea (m2 /m3 )
plines, from direct H2 production from water, selective organic [9] 2631c
Slurry reactorb
synthesis, water treatment and air cleaning to disinfection and [58] 8500–170,000d
anti-tumoral applications. An overview of the many possibilities [9] 27c,e
is given in recent reviews [1,2]. In addition, each year hundreds [8] 69c,e
of research papers are published reporting on promising results [9] 133c,e
(Multi)annular/immersion reactor
[38] 170
of photocatalytic processes. However, industrial applications
[47] 340
remain limited to date. The few examples include commercial [9] 2667c,e
wastewater treatment installations from Zentox Corporation [5],
[59] 46
Matrix Photocatalytic Inc. [6], Clearwater Industries, Photox [60] 53
Bradford Ltd., Lynntech Inc., and Purifics Environmental Tech- [26] 112
nologies Inc. [7]. The current lack of industrial applications is Optical fiber/hollow tube reactorf [19] 210
mainly due to two reasons: [8] 1087c,e
[18] 1920
[9] 2000c,e
• the low photocatalytic efficiency, and related to that the lack
of agreement on how to quantify this efficiency, in particular [48] 943
Monolith reactor
[12] 1333f
with respect to the photoreactor configuration;
• the absence of examples where the laboratory photocataly- Spinning disc reactorg
[10] 50–130
sis set-up successfully has been scaled up to an industrially [43] 20,000–66,000
relevant scale. [53] 7300
[29] 12,000c
Microreactor
[31] 14,000c
Overall the authors of this paper suggest that the reactivity
[54] 250,000c
of photocatalysts in combination with photoreactors, which is
a The catalyst coated surface does not take into account the catalyst specific
currently of the order of 0.05 − 0.1 mol/m3reactor s, should be
surface.
improved by a factor of 100–1000 before industrial implemen- b The illuminated surface per volume (κ) will be much lower than this value
tation in a wide variety of processes is feasible. due to the small penetration depth of light.
This paper focuses in detail on the essential factors that are c This value is explicitly given in the reference, all other values are calculated

limiting scale-up of photoreactors and reviews recent efforts to
tackle these issues.
2. Identification of the current engineering limitations
of photocatalytic processes
by the authors of the present article based on the data given in the reference.
d The ratio of values is due to the variety of tested catalyst concentration in
the slurry.
e This value is explicitly equalled to the illuminated surface per volume (κ) in
the reference.
f The illuminated surface per volume (κ) will be lower than this value due to

the progressive extinction of light along the fiber length.

With respect to scale up, several reactor designs have been
proposed [3]. Slurry reactors, annular reactors, immersion reac-
tors, optical tube reactors and optical fiber reactors are among
the most cited. These reactors are often evaluated on their perfor-
mance to carry out the photocatalytic reaction. Several efficiency
parameters (e.g., quantum yield, photonic efficiency, photo-
chemical thermodynamic efficiency factor) have been proposed
[2,3]. However, these parameters mainly focus on the efficiency
of the process from the activated catalyst to the formation of the
reaction products. The transport of light to the catalyst, being the
first part of the photocatalytic process, is not considered in these
parameters. The most important issues in large-scale designs
are discussed in the following, as well as means to evaluate
them.
2.1. Low efficiency of illumination (photon transfer
limitations)
In a photoreactor the catalyst has to be activated by incident
light. The major parameter that is currently assessed with respect
g The ratio of values is due to the varying film thickness of the reaction liquid
as a function of the spinning velocity.
to illumination is the amount of illuminated surface per unit
of reaction liquid volume inside the reactor (m2ill m−3reactor ). This
value has been denoted as kappa, κ, by Ray and Beenackers
[8]. They calculated κ for different photoreactor configurations
[8,9]. This value has been used in literature quite often (Table 1),
although it sometimes may be questioned whether the surface
area of the reactor per volume that is most often reported can be
instantaneously equalled to the illuminated surface per volume.
Care should be taken that κ is not overestimated by neglecting
the non-illuminated surface in the reactor.
Unfortunately the proposed κ does not take into account two
other aspects related to illumination: the ratio incident/emitted
light power and the uniformity of light incidence. The average
ratio incident/emitted light power (average power efficiency) is
dependent on two physical phenomena. First, the irradiance of
light decreases inversely with the square of distance from the
 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789
source following the inverse square law in Eq. (2):
Plamp
Hc = (2)
l2
with Hc is the average irradiance of light on the catalyst surface
(W m−2 ), Plamp the emitted radiant power of the lamp (W) and
l is the distance from lamp to the catalyst surface (m).
A small distance between light source and catalyst sites is
thus preferred.
Second, light can be absorbed on the way to its final destina-
tion. This process is governed by the law of Lambert-Beer (Eq.
(3)):
P0
log10 = εlc (3)
P This issue mainly relates to the application for process fluids
with P0 is the radiant power before absorption (W), P the radi-
ant power after absorption (W), ε the molar absorptivity of the
absorbing species (l mol−1 m−1 ), l the path length through the
material containing the absorbing species (m) and c is the con-
centration of the absorbing species (mol l−1 ).
In many photoreactor designs the light has to travel through
the fluid or gas containing reagents and other species, and also
through a transparent wall. Part of the originally emitted light can
be absorbed by these components before it reaches the catalyst.
In addition, it is very difficult to achieve uniform irradiance
of the entire catalyst surface. Uniformity is important because a
minimum energy level is needed to activate the catalyst. Diony-
siou et al. [10] have measured the light irradiance distribution
of two low pressure mercury UV tubes positioned horizon-
tally at a distance of 10 cm from a TiO2 coated surface of a
disc in a spinning disc reactor. They observe a highly non-
uniform illumination with light irradiance varying from 30 to
1500 ␮W/cm2 and an average value of 895 ␮W/cm2 . Raupp et
al. [11] have modelled the UV light irradiance field from three
UV lamps positioned horizontally at 5 cm from a monolith sur-
face of 32 cm × 48 cm. It appears that the variation on the surface
is substantial. Greater uniformity can be achieved by increasing
the distance between the lamps and the monolith surface (but
decreasing average irradiance) or by adding more lamps (but
increasing energy costs).
We propose the illumination efficiency, ηill , as a new param-
eter (Eq. (4)), taking into account the illuminated surface per
volume (κ, m−1 ), the average power efficiency (defined as a ratio
of average incident radiant power on the catalyst, measured with
a radiometric probe at different sites, to emitted radiant power),
and the incident uniformity. The latter issue needs further inves-
tigation to measure and quantify. One might think of defining
uniformity as a factor of the catalyst surface that receives at least
the minimum energy (i.e., the band-gap energy), to the total cat-
alyst surface area. Clearly, to date quantifiable results for this
parameter cannot be calculated from the data given in literature
for the specific designs proposed, due to the fact that not all data
has been measured for each reactor configuration. However, it
is anticipated that its use would make a comprehensive compar-
ison possible between the different configurations proposed in
783
literature:
  
Pcat Amin E
ηill = κ (4)
Plamp Acat
with ηill is the illumination efficiency (m−1 ), κ the illuminated
surface per unit of reaction liquid volume inside the reactor
(m2ill m−3 −1
reactor or m ), Pcat the radiant power incident on the
catalyst surface (W), Plamp the radiant power emitted from the
lamp (W), Amin E the catalyst surface that receives at least the
band-gap energy (m2 ) and Acat is the total catalyst surface (m2 ).
2.2. Limited contact between activated catalyst and
reagents (mass transfer limitations)
as mass transfer limitations will be of minor importance in gas
streams. The presence of a gas phase in a liquid environment
can affect the optimal reactor design but this will not be further
investigated in this paper, as no publications report on this issue
in detail.
Once the catalyst is activated by the incident light, maxi-
mized contact between catalyst and reagents in the process fluid
should be achieved. Furthermore, removal of the formed reaction
products should be maximised as well. The issue of improved
mass transfer obviously transcends photocatalysis. It is widely
researched in chemical engineering and various devices have
been developed or proposed to reach that goal. Some of them
have been applied to photoreactors, such as the spinning disc
reactor, the monolith reactor and the microreactor. The issue
therefore appears less critical than the photon-related issues, as
more solutions have been proposed. However, an ideally inten-
sified reactor should be able to integrate both maximized light
efficiency and mass transfer within one piece of equipment.
Mass transfer optimisation in catalysed reactions is mostly
quantified by the amount of catalyst surface area per unit or
reactor volume. Table 1 gives an overview of this value for
various reactor configurations. A wide range of this value is
observed within each of the reactor types. The lower values in
these ranges indicate that the aim of the specific research was not
to maximize mass transfer but rather to explore a specific part
of the reactor, such as photon transport. This value is already
incorporated in the κ value related to the illuminated surface per
volume, described in the previous paragraph. Furthermore, in
immobilized catalytic reactors, the ratio between the amount of
reactants in the reacting liquid and the amount of reactants that
actually come in contact with the catalyst coating is controlled
by the degree of turbulence. Lin and Valsaraj [12] define in that
respect external mass transfer (diffusion of reactants from the
bulk liquid through a boundary layer to reach the liquid–catalyst
interface) and internal mass transfer (inter-particle diffusion of
reactants within the catalyst film to the active surface sites where
they can adsorb and eventually react). Intra-particle diffusion
was neglected by these authors since their TiO2 was considered
non-porous. Increasing the flow velocity, and thus the Reynolds
number, could reduce the external mass transfer resistance [12].
As the latter parameters (flow velocity, Reynolds number) are
784 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789
not fixed by the reactor configuration itself, we do not select
them as assessment parameters for the optimal configuration.
2.3. Other engineering problems in reactor design
A list of other issues can be compiled. On an individual basis,
they will contribute less substantially to the global performance
of a photoreactor. However, all together they can intensify the
eventual design significantly. Moreover, they are often related
to the issues described earlier, and will be affected by intensifi-
cation developments that focus primarily on those issues.
Figuring in this list are:
• maximized light efficiency, which is related to the efficiency
of converting activating power into light, as well as to the
minimized energy consumption of the light source;
• optimal reactor design with for example the presence of a
transparent wall (posing size limitations along with breakage
risks);
• post-reaction separation of the catalyst in the case of slurry
reactors, this step being a very energy-intensive part of the
overall process;
• improved catalyst activation, which is related to catalyst dop-
ing to increase activity or to shift the activating wavelength
range (e.g., from UV to the visible spectrum), and prevention
of deactivation of the catalyst;
• promotion of adsorption of the reagents on the catalyst, which
can be achieved by coating the catalyst on a highly porous
support;
• promotion of desorption of the reaction products from the
catalyst, which for example may be improved by the use of
ultrasound.
3. Overcoming photon transfer limitations
Photons can be acquired either from the sun or from artificial
sources. Solar illumination will decrease energy costs consider-
ably, but only 4–5% of the sunlight reaching the earth’s surface
is in the 300–400 nm near-ultraviolet range, and can conse-
quently be used for activation of TiO2 [3]. Doping of TiO2 is
widely investigated in order to shift the activating wavelength
range from UV to the visible spectrum [2]. Although the use of
solar energy for photocatalytic purposes is investigated at large
research sites such as the European Test Centre for Solar Energy
Applications in Almeria, Spain, and at some institutions in the
developing countries (e.g., [12–14]), the intermittency and vari-
ability of sunlight represent major drawbacks [3]. This paper
will therefore focus on artificial photon sources.
The main development in improving the illumination effi-
ciency is the introduction of optical fibers. In these optical fibers
light is propagated along the fiber length by reflection on the fiber
wall (Fig. 1). Depending on the refractive index of the fiber wall,
a portion of the light intensity is not reflected but refracted. In
an optical fiber reactor the catalyst is typically coated on the
stripped fibers. The refracted light will then be absorbed by
the catalyst which is subsequently activated. There has been
an enormous amount of work put in optical fiber reactors since
Fig. 1. Propagation of light through optical fibers (after [24]).
Marinangeli and Ollis [15–17], Hofstadler et al. [18], and Peill
and Hoffmann [19–21].
Although this development is an interesting route, major
problems still persist. A huge problem with optical fibers is that
the light intensity decays exponentially along the axial direction
of the coated fiber [19,22,23]. Although this might be partly
overcome by changing the refractive index of the fiber down-
ward the length of the fiber [22], the issue remains the largest
drawback of the use of optical fibers. Wang and Ku [24] report a
99% refraction after 10 cm, whereas an uncoated stripped fiber
still transmits about 40% after 20 cm. Choi et al. [23] report
about 70% of refracted light after 5 cm in a 30 cm long fiber. Lin
and Valsaraj [12] report that the light from optical fibers inserted
in the 30 cm long channels of a monolith reactor is extinct after
5 cm. Peill and Hoffmann [19] present similar results, with vari-
ations due to coating thickness and incident light angle. The
amount of refraction in the fiber depends on the thickness of the
TiO2 coating. The thicker the coating, the more light is refracted,
the less light reaches the end of the fiber [19,24]. The thicker the
coating, also the more distance the reagents have to diffuse into
the TiO2 layer to reach the activated catalyst [24]. This is due
to the so-called phenomenon of ‘back-irradiation’, typical for
optical fiber reactors, compared to ‘front-irradiation’ in annular
and other reactors. A trade-off exists between these two down-
sides of thick coatings and the advantage of having more catalyst
available [12,23,25]: an optimum thickness has been reported in
the cited papers to be between 0.4 and 1 ␮m. Another parame-
ter affecting the amount of light reaching the end of the fiber is
the fiber diameter. Essentially, the larger the diameter, the less
reflections the light undergoes and subsequently the less losses
refraction induces [20,26,27]. Lin and Valsaraj [27] report an
effective light propagation length of 20 cm for a 1 mm diameter
fiber compared to a propagation length of 5 cm for a 0.4 mm
diameter fiber. The amount of reflections is also influenced by
the angle at which the light is introduced in the fiber. The more
parallel the light rays travel with the fiber axis, the less reflections
[19].
Another major problem with coated optical fibers is that the
charge carriers are generated far from the liquid–catalyst inter-
face (because of the back-irradiation) and, consequently, are
more susceptible to recombination loss [12,23].
A third problem with optical fiber reactors is that the fibers
can take up to 30% of the reactor volume, thus decreasing the
flow rate and increasing the pressure drop. An improvement
could be the optical fiber monolith reactors with high light effi-
ciency and availability, combined with high catalyst surface area
 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789
per unit reactor volume and scale-up potential [12]. The optical
fiber monolith reactor that Lin and Valsaraj [12] developed gives
a 10-fold increase on the illuminated catalyst surface per unit
of reactor volume compared to an annular reactor, and a 100
times increase of the apparent quantum efficiency. An improved
variation of the design of Lin and Valsaraj [12] could be that
the catalyst is only coated on the monolith wall and not on the
fibers, in order to maximize light emission from the fibers and
avoiding back-irradiation. This is currently under evaluation in
the Delft laboratories.
Finally it has been reported that the light intensity is inversely
related to the quantum efficiency of photocatalytic reactions
[20,28], due to the high recombination rate of charge carri-
ers at high intensity. A lower intensity is thus preferred to
decrease recombination, but this obviously will decrease cat-
alyst activation as well, with consequently a lower rate of
charge pair generation and thus will lower overall photocat-
alytic reaction rate. It appears that optical fibers can overcome
this problem by decreasing the light intensity but increas-
ing the number of fibers [20]. However, this will then again
have a negative effect on the reactor volume available for the
reagents.
Research on the specific area of micro- or nanoscale illu-
mination in photocatalysis is rather scarce. The concept can
be considered as new with only a handful papers present in
the literature. Microscale illumination has been investigated in
the papers of Gorges et al. [29], Chen et al. [30] and Mat-
sushita et al. [31] with encouraging results for the oxidation
of 4-chlorophenol and PCE, and the reduction of benzaldehyde
and nitrotoluene. They use commercially available UV emitting
LED devices. In Gorges et al. [29] the light source consists of
11 LEDs producing a current of 30 mA each and a potential of
3.7 V, whereas Chen et al. [30] use 16 LEDs producing a current
of 20 mA each and a potential of 3.7 V. Matsushita et al. [31] do
not specify details of the used array. Chen et al. [30] report that
the LEDs have an optical output power of 12 mW each, which is
16% of the electric input power (i.e., current intensity times elec-
tric potential, 74 mW each). This is comparable to the ratio range
of 7% (black light) to 23% (germicidal lamp) for traditional gas
charge UV light sources. However, when taking into account the
absorbed photon energy on TiO2 to electric input ratio the figure
of 23% decreases to 15%, because one-third of the optical output
in the germicidal lamp is lost as heat (due to the unnecessary high
energy level (λ = 254 nm), much higher that that used for cata-
lyst activation (λ = 387 nm, equivalent to the 3.2 eV band-gap)).
Other advantages are that these devices are small (miniaturisa-
tion of equipment), robust, and long lasting (hundred thousands
of hours compared to thousands of hours in the case of classi-
cal lamps). The use of microscale illumination in microreactors
provides both a large catalyst surface area per unit or reactor
volume and a high illumination efficiency, also related to the
small angle of emittance of the UV-LEDs. However, catalyst
surface and light source remain physically separated, thus still
leaving potential for improvement (even more increase of the
illumination efficiency).
Nanoscale illumination does explore this as yet unharvested
potential [32–35]. In addition, the integration of light source and
785
Fig. 2. Nanoscale illumination reactor (after [32]).
catalyst surface enhances the options for other reactor types (e.g.,
monolithic reactor, spinning disc reactor), which are otherwise
difficult to combine with microscale illumination. It remains
to be verified whether nanoscale UV sources can provide suf-
ficient energy (i.e., the required wavelength and intensity) for
the desired reaction to occur. Gole et al. [34] suggest the intro-
duction of nitrogen doped titania nanostructures into the pores
of porous silicon (PS) in order to develop a device to produce
visible light by electroluminescence of PS thus activating the
photocatalyst particles. This device could then be incorporated
in a microreactor. A scheme of this promising photoreactor prin-
ciple is shown in Fig. 2. Porous silicon emits visible light, hence
the necessity of using nitrogen doped TiO2 samples to shift the
absorption spectrum from UV to the visible spectrum. Stability
of this modified titania is, however, an issue to be evaluated more
in detail.
4. Overcoming mass transfer limitations
In order to overcome the mass transfer limitations in fluid sys-
tems, several intensified reactor types have been proposed. The
most reported and investigated types of reactor for photocataly-
sis are slurry systems, spinning disc reactors, monolithic reactors
and microreactors. Slurry systems are by far the most used in
photocatalysis research, mainly with respect to the potential of
photocatalysis for specific oxidation or reduction reactions. An
obvious advantage of these systems is the good contact between
reactants and catalyst, illustrated by the high surface area per vol-
ume that can be achieved (Table 1). However, scaling up such a
device implies a separation step of the catalyst from the reaction
products, with major technical and economical problems as a
result. Also, it is difficult to uniformly irradiate suspended par-
ticles, although some configurations may (partly) overcome this
disadvantage (Fig. 3). In the fluidised bed reactor, for example,
the catalyst is placed on supporting beads. These beads form the
bed material and are fluidised by way of an air stream. An aque-
ous or gaseous pollutant flow can be introduced in the fluidised
reactor, either in batch or in continuous set-up [36–38]. Although
the fluidised bed reactor is a dispersed-phase reactor with all its
associated problems with respect to illumination [9], the gener-
ation of bubbles in the fluidised bed reactor will enhance light
penetration compared to the conventional slurry reactor [38]. A
similar aim is achieved in a cocurrent downflow contactor pho-
tocatalytic reactor [39]. The fountain photocatalytic reactor is
another configuration building on the conventional slurry reac-
tor design that combines minimal mass transfer limitations with
a very large illumination surface area per unit volume catalyst
[40].
786 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789

Fig. 3. Examples of improved slurry reactors: (a) fluidised bed reactor and (b) fountain reactor (after [38,40]).

Fig. 4. Different configurations of spinning disc photoreactors (after [42,44]).

Other researchers, focusing on scaling-up of photocatalysis,
advocate the use of immobilised catalysts. Dijkstra et al. [41]
have shown that immobilized systems can attain comparable
quantum yields as slurry systems. A spinning disc reactor has
attractive features such as limited mass transfer limitations due
to the combination of thin film and turbulent flow with subse-
quent high mass transfer coefficient [42], increase of conversion,
better and more reliable product quality [43], introduction of the
reaction liquid in a gentle non-oscillating motion thus diminish-
ing the occurrence of the phenomenon of catalyst attrition [10],
which has been reported in, e.g., annular reactors. Table 1 shows
that a high catalyst surface area per reaction liquid volume can
potentially be attained. It is also easy to scale up [10], either by
increasing the disc diameter (although this is technically not an
infinite option) or by installing multiple discs (although costs
would increase simultaneously). The spinning disc reactor can
be operated in two ways (Fig. 4): horizontally [42,43] with the
liquid introduced in the center of the disc, which then spreads
over the top surface of the disc until it falls off at the edges of
the disc, or vertically [10,44] with half of the disc immersed
in the liquid and the top half entraining the reactor liquid and
exposed to the UV-source. The difference in efficiency between

Fig. 5. Monolith photoreactor (after [46]).

 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789
Fig. 6. Microreactor for photocatalytic reactions (after [54]).
both configurations is not clear. It can be anticipated that it is
more difficult to devise a continuous system with vertical discs
and that less surface is illuminated (i.e., only half discs), although
multiple vertical discs can be combined in stacks, thus increas-
ing capacity with limited increase of reactor volume. A deviation
to this stack-configuration of rotating discs is the rotating drum,
as reported by Zhang et al. [45]. The drum can carry multiple
teeth (a screw-like configuration) to increase the surface area to
be coated with catalyst. Problems with a uniform illumination
of the discs were reported by Dionysiou et al. [10] as described
earlier.
The monolithic photoreactor (Fig. 5) provides a high surface-
to-volume ratio (10–100 times more than plates or beads
substrates with the same outer dimensions) [12] although from
Table 1 it appears less attractive than spinning disc reactors or
microreactors. It allows high flow rates with low-pressure drop,
and is easy to scale-up. It can also be combined with other reac-
tor configurations such as a cocurrent downflow bubble column
reactor [39]. Disadvantage of the monolithic photoreactor is the
low light efficiency. Hossain et al. [46] predict the light flux
in the monolith channels by modelling. At the entrance of the
channels, the light flux already decreases by 50% because of
shadowing of the incoming diffuse light at the channel entrance
by the wall. At a distance of 1–2 aspect ratios (length per chan-
nel width) in the square channels, only 10% of the initial light
flux remains, and at a distance of 3–4 aspect ratios the light flux
has decreased to 1% of the initial flux. In order to overcome this
shortcoming of monolith reactors several improved designs have
been suggested. Imoberdorf et al. [47] conceived a multiannular
reactor with four tubes from UV-transparent walls concentrically
arranged, which resembles a very basic transparent monolith.
The UV source is placed at the centreline and irradiates the three
annular channels where the polluted air flows. TiO2 is coated on
the inner and outer surfaces of the annular walls. This geometry
has three disadvantages. First, the TiO2 coating on the inner wall
of the duct is illuminated on its backside, whereas the reagent
pollutants adsorb on the coating from the other side. Second, due
to the TiO2 coating, the UV flux is bound to be diminished, and
this more in the outer annular ducts. Calculations by the authors
of the present publication on the basis of the reported results
indicate that only 10% of the initial photon flux reaches the
inner TiO2 coating of the second duct, and less than 1% reaches
the inner catalyst coating of the third duct. Of course the amount
of light flux going through the catalyst coated wall depends on
787
the thickness of the TiO2 coating. Sauer and Ollis [48] refer to
a reported light absorption of 99% within a 4.5 ␮m thick TiO2
coating. Third, although limited absorption was present in the
investigated case of PCE pollution in air, this geometry is dis-
advantageous when UV-radiation is absorbed by the pollutant
or by the medium in which they are present (e.g., liquid). Sauer
and Ollis [48] suggest using two light sources, one on either
end of the monolith reactor. This would roughly double the illu-
mination efficiency. The optical fiber monolith reactor that Lin
and Valsaraj [12] developed gives a 10-fold increase on the illu-
minated catalyst surface per unit of reactor volume compared
to an annular reactor, and a 100 times increase of the apparent
quantum efficiency.
Only very recently microreactors (Fig. 6) have been investi-
gated in performing photochemical and photocatalytic reactions.
The main advantage of microreactors is the high surface-to-
volume ratio. In the case of photochemical reactions this leads to
an efficient illumination [49–52]. In the case of photocatalytic
reactions, this implies efficient catalytic exposure to radiation
[29], but in addition it also leads to maximized reagent/catalyst
contact [53,54]. The small size of the channel also provides
greater control over variables such as temperature control and
flow rates [53], due to the fast heat and mass transfer, and
the presence of laminar flow [49,50]. The light source itself
can also be UV-LED, thus miniaturizing the whole photocat-
alytic set-up [29,31]. The patented configuration of Barthe et al.
[53] can contain 23 g/l TiO2 , which is 11 times more catalyst
per unit of volume than in a conventional batch reactor [55].
Gorges et al. [29] and Matsushita et al. [31] report a (illumi-
nated) catalyst area-to-reactor volume of almost 12,000 m2 /m3
and 14,000 m2 /m3 , respectively, while this value in Takei et al.
[54] even reaches 0.25 m2 /ml (i.e., 250,000 m2 /m3 ). These num-
bers are among the highest in Table 1. Takei et al. [54] found a
70-fold increase in conversion rate compared to a slurry reac-
tor with the same selectivity and yield. Li et al. [56] report
a 60 (530 ␮m diameter capillary)- to 160 (200 ␮m diameter
capillary)-fold increase in conversion rate compared to a slurry
reactor. Obvious drawback is the relatively small throughput
of microreactors [57]. The flow-rate in Takei et al. [54] is
1 ␮l/min (i.e., 526 ml/year), although this was evidently per-
formed in a lab set-up without focus on scale up. Microreactors
currently implemented in industry can produce up to 5000 or
even 20,000 tonnes/year. It is clear that microreactors have large
potential.
788 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789
5. Conclusions
The most important developments in the last 10 years in the
area of photoreactors were the novel configurations for maxi-
mizing the reactant-catalyst contact, which is also the subject of
research outside photocatalysis, with fruitful cross-fertilization
thus enhancing development. Overall it appears that the microre-
actor, the spinning disc reactor, and, to a lesser extent, the (optical
fiber) monolith reactor are the most promising configurations to
date reported in literature. Less successful results were achieved
with respect to the maximization of the illumination efficiency,
most likely because this research line is undertaken only by
the smaller photo(catalyst) community, with limited input from
other engineering disciplines. Use of fibers is the main develop-
ment to date, but suffers from obvious drawbacks. Availability of
a multitude of low-intensity light emitting sources on the micro-
or even nanoscale near the catalyst particles appears to be a novel
and promising approach to achieve uniform and maximized illu-
mination. Combination of the latter approach with equipment
to overcome the mass transfer limitations may prove to be the
significant improvement that photoreactors need for industrial
implementation.
Acknowledgements
T. Van Gerven acknowledges a postdoctoral research grant
from the Research Fund K.U.Leuven. He also would like to
express his gratitude to the Research Council of the K.U.Leuven
and to Prof. B. Van der Bruggen and Prof. J. Van Impe from the
Department of Chemical Engineering (K.U.Leuven) for sup-
porting his research visit at the Delft University of Technology.
Appendix A. Nomenclature
Acat total catalyst surface (m2 )
Amin E catalyst surface that receives at least the band-gap
energy (m2 )
c concentration of the absorbing species (mol l−1 )
e− electron
Ebg band-gap energy of the catalyst, i.e., energy difference
between the valence band and the conduction band (J)
h+ hole
H irradiance, i.e., the power per unit area incident on
a surface (illumination or illuminance in photometry)
(W m−2 )
Hc irradiance on the reactor wall (W m−2 )
l distance (m)
P radiant power, i.e., the rate of transfer of energy (lumi-
nous flux in photometry) (W)
Pcat radiant power incident on the catalyst surface (W)
Plamp radiant power emitted from the lamp (W)
P0 radiant power before absorption (W)
Greek symbols
ε molar absorptivity of the absorbing species
(l mol−1 m−1 )
ηill illumination efficiency (m−1 )
κ illuminated surface per unit of reaction liquid volume
inside the reactor (m2ill m−3 −1
reactor or m )
λ radiation wavelength (nm)
References
[1] A. Mills, S. Le Hunte, An overview of semiconductor photocatalysis, J.
Photochem. Photobiol. A: Chem. 108 (1997) 1–35.
[2] O. Carp, C.L. Huisman, A. Reller, Photoinduced reactivity of titanium
oxide, Progr. Solid State Chem. 32 (2004) 33–177.
[3] H. de Lasa, B. Serrano, M. Salaices, Photocatalytic Reaction Engineering,
Springer, New York, 2005.
[4] A. Stankiewicz, Energy matters, alternative sources and forms of energy
for intensification of chemical and biochemical processes, Trans IChemE:
A Chem. Eng. Res. Des. 84 (2006) 511–521.
[5] D.M. Blake, Bibliography of work on the heterogeneous photocatalytic
removal of hazardous compounds from water and air, update number 2 to
October 2006, NREL/TP-430-22197, 2006.
[6] P.R. Gogate, A.B. Pandit, Sonophotocatalytic reactors for wastewater treat-
ment: a critical review, AIChE J. 50 (2004) 1051–1079.
[7] A. Mills, S.K. Lee, in: S. Parsons (Ed.), Advanced Oxidation Processes
for Water and Wastewater Treatment, IWA Publishing, London, 2004, pp.
137–166.
[8] A.K. Ray, A.A.C.M. Beenackers, Development of a new photocatalytic
reactor for water purification, Catal. Today 40 (1998) 73–83.
[9] P.S. Mukherjee, A.K. Ray, Major challenges in the design of a large-scale
photocatalytic reactor for water treatment, Chem. Eng. Technol. 22 (1999)
253–260.
[10] D.D. Dionysiou, G. Balasubramanian, M.T. Suidan, A.P. Khodadoust, I.
Baudin, J.M. Laı̂né, Rotating disk photocatalytic reactor: development,
characterization, and evaluation for the destruction of organic pollutants in
water, Water Res. 34 (2000) 2927–2940.
[11] G.B. Raupp, A. Alexiadis, M.M. Hossain, R. Changrani, First-principles
modeling, scaling laws and design of structured photocatalytic oxidation
reactors for air purification, Catal. Today 69 (2001) 41–49.
[12] H. Lin, K.T. Valsaraj, Development of an optical fiber monolith reactor
for photocatalytic wastewater treatment, J. Appl. Electrochem. 35 (2005)
699–708.
[13] A.A. Yawalkar, D.S. Bhatkhande, V.G. Pangarkar, A.A.C.M. Beenackers,
Solar-assisted photochemical and photocatalytic degradation of phenol, J.
Chem. Technol. Biotechnol. 76 (2001) 363–370.
[14] D.S. Bhatkhande, V.G. Pangarkar, A.A.C.M. Beenackers, Photocatalytic
degradation for environmental applications—a review, J. Chem. Technol.
Biotechnol. 77 (2001) 102–116.
[15] R.E. Marinangeli, D.F. Ollis, Photoassisted heterogeneous catalysis with
optical fibers. 1. Isolated single fiber, AIChE J. 23 (1977) 415–426.
[16] R.E. Marinangeli, D.F. Ollis, Photoassisted heterogeneous catalysis with
optical fibers. 2. Non-isothermal single fiber and fiber bundle, AIChE J. 26
(1980) 1000–1008.
[17] R.E. Marinangeli, D.F. Ollis, Photoassisted heterogeneous catalysis with
optical fibers. 3. Photo-electrodes, AIChE J. 28 (1982) 945–955.
[18] K. Hofstadler, R. Bauer, S. Novalic, G. Heisler, New reactor design for
photocatalytic wastewater treatment with TiO2 immobilized on fused-silica
glass fibers: photomineralization of 4-chlorophenol, Environ. Sci. Technol.
28 (1994) 670–674.
[19] N.J. Peill, M.R. Hoffmann, Development and optimisation of a TiO2 -coated
fiber-optic cable reactor: photocatalytic degradation of 4-chlorophenol,
Environ. Sci. Technol. 29 (1995) 2874–2981.
[20] N.J. Peill, M.R. Hoffmann, Chemical and physical characterisation of a
TiO2 -coated fiber-optic cable reactor, Environ. Sci. Technol. 30 (1996)
2806–2812.
[21] N.J. Peill, M.R. Hoffmann, Mathematical model of a photocatalytic fiber-
optic cable reactor for heterogeneous photocatalysis, Environ. Sci. Technol.
32 (1998) 398–404.
[22] R.D. Sun, A. Nakajima, I. Watanabe, T. Watanabe, K. Hashimoto, TiO2 -
coated optical fiber bundles used as a photocatalytic filter for decomposition
 T. Van Gerven et al. / Chemical Engineering and Processing 46 (2007) 781–789
of gaseous organic compounds, J. Photochem. Photobiol. A: Chem. 136
(2000) 111–116.
[23] W. Choi, J.H. Ko, H. Park, J.S. Chung, Investigation on TiO2 -coated optical
fibers for gas-phase photocatalytic oxidation of acetone, Appl. Catal. B:
Environ. 31 (2001) 209–220.
[24] W. Wang, Y. Ku, The light transmission and distribution in an optical fiber
coated with TiO2 particles, Chemosphere 50 (2003) 999–1006.
[25] H. Joo, H. Jeong, M. Jeon, I. Moon, The use of plastic optical fibers in
photocatalysis of trichloroethylene, Sol. Energy Mater. Sol. Cells 79 (2003)
93–101.
[26] A. Danion, J. Disdier, C. Guillard, F. Abdelmalek, N. Jaffrezic-Renault,
Characterization and study of a single TiO2 -coated optical fiber reactor,
Int. J. Appl. Electromagn. Mech. 23 (2006) 187–201.
[27] H. Lin, K.T. Valsaraj, An optical fiber monolith reactor for photocatalytic
wastewater treatment, AIChE J. 52 (2006) 2271–2280.
[28] D.F. Ollis, E. Pelizetti, N. Serpone, Photocatalytic destruction of water
contaminants, Environ. Sci. Technol. 25 (1991) 1523–1529.
[29] R. Gorges, S. Meyer, G. Kreisel, Photocatalysis in microreactors, J. Pho-
tochem. Photobiol. A: Chem. 167 (2004) 95–99.
[30] D.H. Chen, X. Ye, K. Li, Oxidation of PCE with a UV LED photocatalytic
reactor, Chem. Eng. Technol. 28 (2005) 95–97.
[31] Y. Matsushita, S. Kumada, K. Wakabayashi, K. Sakeda, T. Ichimura,
Photocatalytic reduction in microreactors, Chem. Lett. 35 (2006) 410–
411.
[32] A.G. Fedorov, J.L. Gole, C. Phillips, From energy to environment: unique
opportunities in nanoscale catalysis, in: Nanotherm: US-Japan Seminar on
Nanoscale Thermal Science & Engineering, Berkeley, California, USA,
June 24–26, 2002.
[33] J. Gole, C. Burda, A. Fedorov, M. White, Enhanced reactivity and phase
transformation at the nanoscale: efficient formation of active silica and
doped and metal seeded TiO2−x Nx photocatalysts, Rev. Adv. Mater. Sci. 5
(2003) 265–269.
[34] J.L. Gole, A. Fedorov, P. Hesketh, C. Burda, From nanostructures to
porous silicon: sensors and photocatalytic reactors, Phys. Stat. Sol. 1 (2004)
188–197.
[35] J.L. Gole, C. Burda, Z.L. Wang, M. White, Unusual properties and reac-
tivity at the nanoscale, J. Phys. Chem. Sol. 66 (2005) 546–550.
[36] L.A. Dibble, G.B. Raupp, Fluidized-bed photocatalytic oxidation of
trichloroethylene in contaminated airstreams, Environ. Sci. Technol. 26
(1992) 492–495.
[37] S. Rodrı́guez Couto, A. Domı́nguez, A. Sanromán, Photocatalytic degra-
dation of dyes in aqueous solution operating in a fluidised bed reactor,
Chemosphere 46 (2002) 83–86.
[38] T.H. Lim, S.D. Kim, Trichloroethylene degradation by photocatalysis in
annular flow and annulus fluidised bed photoreactors, Chemosphere 54
(2004) 305–312.
[39] I.J. Ochuma, R.P. Fishwick, J. Wood, J.M. Winterbottom, Optimisation of
degradation conditions of 1,8-diazabicyclo[5.4.0]undec-7-ene in water and
reaction kinetics analysis using a cocurrent downflow contactor photocat-
alytic reactor, Appl. Catal. B: Environ. 73 (2007) 259–268.
[40] G. Li Puma, P.L. Yue, A novel fountain photocatalytic reactor for water
treatment and purification: modelling and design, Ind. Eng. Chem. Res. 40
(2001) 5162–5169.
[41] M.F.J. Dijkstra, A. Michorius, H. Buwalda, H.J. Panneman, J.G.M. Winkel-
man, A.A.C.M. Beenackers, Comparison of the efficiency of immobilized
789
and suspended systems in photocatalytic degradation, Catal. Today 66
(2001) 487–494.
[42] H.C. Yatmaz, C. Wallis, C.R. Howarth, The spinning disc reactor—studies
on a novel TiO2 photocatalytic reactor, Chemosphere 42 (2001) 397–403.
[43] J. Dalglish, R. Jachuck, C. Ramshaw, in: A. Green (Ed.), Proceedings of the
Third International Conference on Process Intensification for the Chemical
Industry, BHR Group Conference Series Publication No. 38, Antwerp,
Belgium, October 25–27, Professional Engineering Publishing Ltd., Bury
St. Edmunds and London, UK, 1999.
[44] N.A. Hamill, L.R. Weatherley, C. Hardacre, Use of batch rotating photo-
catalytic contactor for the degradation of organic pollutants in wastewater,
Appl. Catal. B: Environ. 30 (2001) 49–60.
[45] L. Zhang, W.A. Anderson, Z. Zhang, Development and modeling of a
rotating disc photocatalytic reactor for wastewater treatment, Chem. Eng.
J. 121 (2006) 125–134.
[46] M.M. Hossain, G. Raupp, S.O. Hay, T.N. Obee, Three-dimensional devel-
oping flow model for photocatalytic monolith reactors, AIChE J. 45 (1999)
1309–1321.
[47] G.E. Imoberdorf, A.E. Cassano, O.M. Alfano, H.A. Irazoqui, Modeling of
a multiannular photocatalytic reactor for perchloroethylene degradation in
air, AIChE J. 52 (2006) 1814–1823.
[48] M.L. Sauer, D.F. Ollis, Acetone oxidation in a photocatalytic monolith
reactor, J. Catal. 149 (1994) 81–91.
[49] H. Lu, M.A. Schmidt, K.F. Jensen, Photochemical reactions and on-line
UV detection in microfabricated reactors, Lab. Chip 1 (2001) 22–28.
[50] R.C.R. Wootton, R. Fortt, A.J. de Mello, A microfabricated nanoreactor for
safe, continuous generation and use of singlet oxygen, Org. Process Res.
Dev. 6 (2002) 187–189.
[51] T. Fukuyama, Y. Hino, N. Kamata, I. Ryu, Quick execution of [2 + 2] type
photochemical cycloaddition reaction by continuous flow system using a
glass-made microreactor, Chem. Lett. 33 (2004) 1430–1431.
[52] H. Maeda, H. Mukae, K. Mizuno, Enhanced efficiency and regioselectiv-
ity of intramolecular (2π + 2π) photocycloaddition of 1-cyanonaphtalene
derivative using microreactors, Chem. Lett. 34 (2005) 66–67.
[53] P.J. Barthe, D.H. Letourneur, J.P. Themont, P. Woehl, Method and microflu-
idic reactor for photocatalysis, Patent B01J19/00 (2004).
[54] G. Takei, T. Kitamori, H.B. Kim, Photocatalytic redox-combined synthe-
sis of l-pipecolinic acid with a titania-modified microchannel chip, Catal.
Commun. 6 (2005) 357–360.
[55] J.M. Herrmann, C. Guillard, J. Disdier, C. Lehaut, S. Malato, J. Blanco,
New industrial titania photocatalysts for the solar detoxification of water
containing various pollutants, Appl. Catal. B: Environ. 35 (2002) 281–294.
[56] X. Li, H. Wang, K. Inoue, M. Uehara, H. Nakamura, M. Miyazaki, E. Abe,
H. Maeda, Modified micro-space using self-organized nanoparticles for
reduction of methylene blue, Chem. Commun. (2003) 964–965.
[57] A. Stankiewicz, J.A. Moulijn (Eds.), Re-engineering the chemical process-
ing plant: process intensification, Marcel Dekker Inc., New York, 2004.
[58] V. Auguglario, V. Loddo, G. Marci, L. Palmisano, M.J. Lopez-Munoz, Pho-
tocatalytic oxidation of cyanides in aqueous titanium dioxide suspensions,
J. Catal. 166 (1997) 272–283.
[59] W. Wang, Y. Ku, Photocatalytic degradation of gaseous benzene in air
streams by using an optical fiber photoreactor, J. Photochem. Photobiol. A:
Chem. 159 (2003) 47–59.
[60] J. Wu, M.H. Lin, C.L. Lai, Photo reduction of CO2 to methanol using
optical-fiber photoreactor, Appl. Catal. A: Gen. 296 (2005) 194–200.