Article

pubs.acs.org/IECR

Laboratory Scale Water Circuit Including a Photocatalytic Reactor

and a Portable In-Stream Sensor To Monitor Pollutant Degradation
Patrick Nickels,, Hang Zhou,, Sulaiman N. Basahel, Abdullah Y. Obaid, Tarek T. Ali,
Ahmed A. Al-Ghamdi, El-Sayed H. El-Mossalamy, Abdulrahman O. Alyoubi, and Stephen A. Lynch*,

London Centre for Nanotechnology, University College, London, U.K.

Bio Nano Consulting, U.K.

Chemistry Department, King Abdulaziz University, Saudi Arabia

Physics Department, King Abdulaziz University, Saudi Arabia

ABSTRACT: We describe a lab-scale closed-circulating test system for photocatalytic wastewater treatment. The system
comprises a UV-LED photoreactor, a microcirculating fluid pump, and an in-stream sensor unit. The reactor can hold volumes
up to 250 mL and is optimized to study the degradation of pollutant concentrations in the microgram to milligram per liter range
using photocatalysts fixed to a planar surface within the reactor vessel. The test pollutant used was methyl orange. The in-stream
sensor unit consists of a liquid flow cell with transparent windows, allowing the transmission of light from an LED to be
monitored by a photodiode. The concentration of the pollutant is evaluated in real-time. The system is lightweight, cheap,
portable, and flexible, ideal for laboratory or fieldwork use, and could be easily up-scaled and used for in-line quality control
monitoring in a wastewater treatment plant.

1. INTRODUCTION
Water pollution is a global problem. Compounds including
natural organic matter and synthetic organic microcontaminants, for example, hydrocarbons, pharmaceuticals, endocrinedisrupting compounds like polychlorinated biphenyls, fertilizers
and pesticides, are released constantly into the environment
by industry, households, and agriculture.1 Regular wastewater
plants help to remove most of the pollutants via regular and
cost-effective treatment steps like sedimentation, filtration, and
biological processes, all of which are deemed relatively effective
for the treatment of wastewater. However, biologically toxic
and nondegradable organics can still remain. Advanced treatment processes such as activated carbon and advanced oxidation processes are being adopted;2 but these can be expensive
to run and result in increased water costs.3 The use of semiconductor photocatalysts to generate reactive oxygen species
for advanced oxidation processes in water treatment technology
has become one of the most promising techniques to provide
a cheap and energy efficient method for the disinfection of
water.46 Other advantages are that fouling can possibly be
inhibited by the photocatalytic activity, and ideally the catalytic
material does not need refueling or replacement and can, therefore,
run continuously. Thus, the investigation into, and development of,
efficient photocatalysts and reactors has become a worldwide
challenge.
Titanium dioxide is the most widely studied photocatalytic
material to date. Crystalline TiO2 is a compound semiconductor
and has a bandgap that lies in the range 3.13.4 eV, depending on the exact crystal structure (anatase, rutile, or brookite).7
Bandgap excitation is achieved using photons with wavelengths
lying in the near-UV band (shorter than 380 nm). TiO2 is widely
available and, due to its ubiquitous use as a white pigment,
is inexpensive. It is biologically compatible and very stable;
XXXX American Chemical Society

such properties have brought it accreditation even as a food

additive.8
There are two methods to treat wastewater in a photocatalytic process: either to suspend the catalyst in a powder or
granule form in the water, (a so-called slurry system) or coat
the catalyst on a surface over which the water flows (commonly
referred to as a fixed bed system).9 A possible advantage of the
slurry system is that there is a much higher surface-to-liquid
interface area and, therefore, a more efficient generation of
reactive oxygen species or direct interaction with pollutants.10
However in this study, a fixed bed reactor system has been
investigated to avoid the possible need for a post-reaction separation of catalyst from the water.
A key step in the development process is to understand how
well the photocatalyst behaves under different environmental
conditions.11 The efficacy of a photocatalyst is usually evaluated
by monitoring the degradation rate of a specific compound in
aqueous solution under controlled conditions; these include
concentration of solution and photocatalyst, irradiance, pH,
and volume.12 One compound that is commonly used as a test
model pollutant is the relatively benign chemical methyl
orange13 due to its strong color visible to the naked eye and
the use of conventional spectrometers to assess the concentration by absorption spectroscopy. Additionally, it has many
properties of common organic pollutants such as benzene rings,
sulfonate, and amine groups. There are several methods that
can be used to measure the concentration of such an agent in
solution. The most obvious of these is conventional spectroscopy
Received: October 15, 2011
Revised: December 12, 2011
Accepted: January 5, 2012

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

(be it UVvis or FTIR); however, other possibilities include high

pressure (performance) liquid phase chromatography (HPLC) or
liquid chromatographymass spectroscopy (LCMS). Common
disadvantages to these methods are the equipment needed is often
not easily portable and it is usually very expensive, often requiring
a trained user. Furthermore, in a real world application, water will
be circulating through a photocatalytic reactor with either artificial
light or natural sunlight as the radiation source. In such a water
circuit, it is important to monitor the concentration of the
chemical(s) requiring removal. Therefore, it is of great interest to
have sensor systems in place that can record the concentration in
real-time. Of the existing analysis methods previously mentioned,
some could be adapted to perform real-time monitoring, but the
modifications would be expensive.
The setup described in this study is a compact, robust, and
cheap solution designed to understand the efficacy of photocatalytic reactions in real-time. It is a closed water circuit that
integrates a real-time in-stream sensor and photocatalytic reactor.
The reactor consists of a vessel with inlets and outlets, and a substrate coated with photocatalytic material covers the base. Water is
circulated through the reactor using a centrifugal pump, providing
constant mixing and flow. To initiate a photocatalytic reaction
UV-LEDs, mounted on the cover of the reactor, illuminate the
photocatalyst. UV-LEDs have recently become a popular choice as
a UV light source in reactors because of their cheap price, long
lifetime, high quantum yield, and small size,1418 and importantly
they have been shown to be effective for chemical degradation.1922 A liquid cell that measures light transmittance is used to
enable the concentration of any selected chemical in the system to
be monitored. This setup enables the study of a range of parameters
and optimal conditions for photocatalytic reactions accordingly.

measurement of the light absorption can be made. The signal

from a photodiode is then processed by analogue electronic
circuitry, and the resulting signal corresponds to the concentration of the absorbing chemical, which, in this instance, is
methyl orange.
2.2. Chemicals and Photocatalyst Preparation. Methyl
orange (MO) sourced from Sigma-Aldrich was dissolved in
deionized (DI) water in typical concentrations ranging from
100 to 10 ppm. Drops of the MO solution were added into the
reactor containing DI water. We monitored the real-time photodiode signal during this process and observed that a homogeneous
mixture was produced on a time scale of seconds. This time scale
was negligible when compared to the rate-constant of any of the
reactions we studied.
The photocatalyst used for these experiments was Evonik
TiO2 Aeroxide P25, which we will subsequently refer to in this
paper as P25. P25 consists of a mixture of 20% rutile and 80%
anatase TiO2. The responsible photocatalytic mechanism that
makes P25 one of the most photocatalytically active materials
on the market23 is under constant debate; some believe the
rutile TiO2 acts as an antenna, which due to a smaller bandgap
absorbs a larger range of wavelengths, while others claim that
at the interface between the two materials charge separation
and prolongation of lifetimes enhance the photocatalysts
activity.24,25
For coating, we adapted a spin-casting method where TiO2
nanoparticle suspensions were formed by mixing TiO2 (400 mg)
with ethanol (4 mL) and Triton X-100 surfactant (250 L).26
Thin films were fabricated by spin-casting the TiO2 suspension
onto 3 in. glass wafers. For several cycles, 0.5 mL of suspension
was drop cast onto the substrate surface and then spun at 300 rpm
for 20 s. The wafer was rapidly heated to 450 C for 10 min. The
function of this processing step was to remove any traces of the
organic surfactant used for spin coating. We have chosen a
temperature of 450 C because this is well below the annealing
temperature required for microstructural transformation of the
film, as discussed by Zhang et al.27
2.3. Characterization. Powder X-ray diffraction (XRD) experiments on the TiO2 powder and coated TiO2 samples were
performed at room temperature using a Philips PW 3040 DY640
diffractometer equipped with a graphite monochromator using
Cu K radiation ( = 0.1541 nm). The samples were scanned
over a 2 range of 1080o in steps of 0.02o. To verify the surface
coverage and morphology of the TiO2 on the glass wafers both
before and after reaction, field-emission scanning electron
microscopy (SEM, Carl Zeiss XB 1540) at 5 kV acceleration
voltage was employed. The thicknesses of the films were measured using a Dektak profilometer.
2.4. Reactor Assembly. The water circuit was assembled
by connecting a small batch reactor, made from a glass with
inlets and outlets, in series with a small centrifugal pump and
the liquid cell. Figure 2 shows a schematic of the reactor vessel
on the left panel. The glass reactor vessel has a height of 70 mm
and has an inner diameter of 84 mm. The reactor cover holds
15 UV-LEDs and the coated wafer is fixed to the reactor base.
The distance of the UV-LEDs to the photocatalyst is 65 mm.
The inlets and outlets are 4 mm diameter glass tubes situated
10 mm from the base of the reactor. The reactor was filled with
the test liquid in volumes ranging from 100 to 250 mL. In most
experiments 100 or 150 mL was used, which give water depths
of approximately 20 or 30 mm, respectively.
For the UV light source, we have used Ultra Bright
Deep Violet LED370E UV-LEDs sourced from Thorlabs. The

2. EXPERIMENTAL SECTION
2.1. Overview. Figure 1 shows a schematic of the closed
water circuit system containing a reactor where the photocatalytic

Figure 1. Schematic of the closed water circuit for evaluation of

photocatalytic reactions. The reactor is driven by UV-LEDs
illuminating a substrate coated with photocatalytic material on the
base of the reactor vessel. A micropump provides constant mixing and
mass flow over the catalyst and serves the in-stream sensor unit to
monitor the concentration of pollutants in real time by detecting light
absorption.

degradation of chemicals or organic pollutants is carried out.

The water flow is driven through a centrifugal micropump to
guarantee constant mixing in the reactor vessel. The heart of
the monitoring system can be seen on the left side in Figure 1.
A liquid cell is placed in the flow circuit; here a light (LED
source) passes through the water/pollutant stream so that a
B

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

emitting semiconductor chips. The intensity distribution was

measured with a Newport 918D-UV-OD3 detector and power
meter (results are shown in Figure 3d) at a step distance of
1 cm. The maximum irradiance is 2.1 W/m2, with the peak
center shifted slightly to the right of the ideal position. The
integrated power of the measured irradiated field from the
measurement is 31.2 mW, which has to be corrected by a factor
of 4/ due to the circular aperture of the detector and the
square-type measurement matrix and amounts to 24.5 mW of
total irradiant power. The intensity of the light can be changed
by a potentiometer set in series to the UV-LEDs. For measurements of the light intensity, we have plotted the irradiance
observed in the center of the light field and assumed a linear
relationship with total power. The total area of the coated wafer
is 45.6 cm2. All 15 UV-LEDs irradiate approximately threequarters of the coated surface.
The reactor vessel design was chosen to ensure both efficient
mixing of the MO solution and at a steady but controlled mass
flow rate over the photocatalytic surface. In Figures 4a,b we

Figure 2. (a) Schematic showing the reactor, which consists of a glass

vessel equipped with inlet and outlet, with the photocatalyst fixed on
its base. UV-LEDs are fixed into the cover of the reactor. (b) Emission
spectrum of the illuminating LEDs showing a peak emission
wavelength centered on 375 nm.

emission spectrum is indicated on the right panel in Figure 2

with a main emission peak at 375 nm and a line width of
approximately 10 nm. Thus, the emitted light lies well in the
absorption spectrum of P25. Each UV-LED has a half viewing
angle of 19 and a forward optical power of 2 mW at the drive
current of 20 mA.
The arrangement of the 15 UV-LEDS is shown in Figure 3a
with a slight prolongation along one axis. The real light field

Figure 4. (a,b) Two investigated chamber geometries in. The design

and results of two-dimensional CFD simulations are presented. For
illustration purposes we have included arrows indicating the speed and
flow direction in the vector diagrams. The first design in panel a was
implemented in the reactor. (c) Mixing of methyl orange in the reactor
vessel monitored by the in-stream sensor. Each step represents adding
one drop (ca. 0.05 mL) of a 100 ppm MO solution to 150 mL of clear
water.

show two-dimensional computational fluid dynamics (CFD)

simulations and subsequent distribution of flow rates indicated by velocities for two different designs, respectively. Both
designs have a central circular chamber, the design in Figure 4a
has opposing inlets and outlets, whereas the design in Figure 4b
has a linear arrangement for the inlet and outlet. CFD simulations were performed with EasyCFD in the steady state regime
with turbulent flow, isothermal, and nonbuoyant settings. A fast
converging steady state solution depending on the grid size was
confirmed. The in and out mass flow rate was set to 0.5 L/min
similar to the real pump rate. The first design (Figure 4a)
shows a slow flow in the middle of the reactor and faster flow
at the edge and also has turbulence due to the direction change
at the outlet from the water stream coming from the inlet.
It, therefore, gives better mixing properties as opposed to the

Figure 3. (a) Photograph of the UV-LED pattern in the reactor cover,

(b) photograph of the resulting illuminated area on the photocatalyst
coated wafer, (c) simulated irradiant power distribution, based upon
geometric arrangement assuming Gaussian distribution of the power
for each LED on the photocatalytic disk, and (d) measured irradiant
power distribution reaching up to 2.1 mW/cm2.

was photographed and is shown in Figure 3b. The ideal light

field generated, at a distance of 65 mm, (the position of the
photocatalyst surface) gives an almost circular illumination, as
shown in the simulation in Figure 3c. The real illuminated area
deviates due to nonideal soldering of the UV-LEDs on the
lid plate and possible inhomogeneous molding of the light
C

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

from OSRAM Opto Semiconductor GmbH). Here the spectral

range is chosen to match the illuminating LED.
The fraction of illuminating light not absorbed by the MO
solution registers on the photodiode in the form of an electrical
signal I (Figure 7). A calibrated reference signal representing

design suggested in Figure 4b, where most of the liquid flows

directly in a linear stream from the inlet to the outlet. In
addition, the design in Figure 4a guarantees a constant flow rate
and mass exchange in the center of the photocatalytic wafer, the
area that receives the highest photon flux and is expected to
have the highest photocatalytic activity, while at the same time
providing a fast mixing and an instant sensor reading of the real
concentration. Figure 4c shows the concentration of MO,
measured by the sensor upon dropwise addition of approximately 0.05 mL of 100 ppm MO solution into the reactor
containing 150 mL of water. The concentration increases in a
stepwise manner and demonstrates fast and homogeneous
mixing in three to five seconds.
2.5. In-Stream Sensor Unit. The sensor system consists of
an aluminum milled liquid flow cell with front and back quartz
observation windows (see Figure 5). Quartz has been chosen

Figure 7. Flow diagram of the read out and signal processing

electronics. The signal from the photodiode (I) is amplified and passed
through comparator electronics to produce an output directly
proportional to the concentration. The other input to the differential
amplifier is the calibrated reference intensity (I0).

the intensity I0 of a total absence of MO is produced either by

the photodiode on a similar reference cell containing pure
water or alternatively by an adjustable constant voltage source.
Both signals are then passed first through a trans-impedance
amplifier and second a logarithmic amplifier. In the last stage, a
differential amplifier compares the amplified logarithmic signals.
In this way, the output signal produced is proportional to the
quotient of the sample I and reference signal I0:

Figure 5. (a) The absorption measurement is performed in special

flow-through cells, which have fittings for the tubing in one direction
and two windows on each side on one of the orthogonal axes. On the
windows attached are holders for the light source (LED) and the
photodetector to measure the light absorption in the liquid. (b) A
photograph of the cell.

for this application because it is transparent in the range 200

2500 nm. Other window materials could be used to access
alternative spectral bands. At one of the windows there is a
light-tight tube containing an illuminating LED with spectral
properties matching the visible absorption of methyl orange,
along with collimation optics. The emission spectrum of the
LED (Hyper blue LED LB3333 from OSRAM Opto Semiconductor GmbH) is taken from the datasheet and presented in
Figure 6. When compared to the measured absorption of MO

log

I
= log I log I0
I0

(1)

which, in turn, is proportional to the concentration C of the

monitored chemical, according to the BeerLambert law:

I = I010alC

(2)

where I and I0 are the intensities of the transmitted and

incident light, is the absorption coefficient, l the path length,
and C the concentration. The resultant logarithmic quotient is
therefore directly proportional to the concentration of the
monitored chemical.

3. RESULTS AND DISCUSSION

Before considering the properties of the catalytic reactor/sensor
system, some basic material characterization of the TiO2 film
was performed. The aim of this exercise was to establish
whether the coating process itself affected the catalytic
properties of the TiO2. Factors such as the microcrystalline
structure and the film uniformity has been studied in previous
investigations.47,25 XRD results before and after confirm that
no major phase transitions have occurred after coating. Some
broadening of the peaks was observed (Figure 8) but this was
attributed to the decreased sample volume in the film, compared to the powdered state. Analysis of the SEM images
(Figure 9) shows that the coating method resulted in a uniform
coverage with an average thickness of about 40 nm (measured
by Dektak). This measurement was repeated after several
catalytic reactions had been performed. While the later SEM

Figure 6. Absorption spectrum showing the absorption values for

methyl orange and the matching emission of the blue LED, which is
used in the concentration sensor. The UV-LED spectrum is plotted as
a reference.

in Figure 6a, both spectra have the same maximum wavelength

at 465 nm. At the other window of the liquid cell, is a light-tight
tube containing a photodiode (visible light photodiode BPW21
D

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

Figure 8. Powder X-ray diffraction patterns of (a) the pristine P25

TiO2 and (b) P25 TiO2 coated onto the glass wafer.

images did show some minor changes to the surface morphology,

showing some additional agglomeration of the particles, the average thickness of the film remained more or less constant at 40 nm.
From this, we concluded that the films had remained relatively
stable during the catalytic reactions, and consequently leaching of
the TiO2 nanoparticles into the water was negligible.
The batch reactor, micropump and liquid cell were
connected by flexible 3 mm diameter tubing and loaded with
DI water. The sensor system was calibrated to zero output
before mixing the MO solution into the reactor. To initiate the
photocatalytic reaction, the TiO2 coated glass wafer was fixed
on the bottom and was illuminated by the UV-LEDs.
Figure 10 depicts typical results from an experiment measuring the degradation of MO, where the initial concentration of
MO in the solution is 0.6 ppm, and after approximately six to
eight hours the orange solution becomes colorless. Control
experiments using a wafer prepared without TiO2 and experiments with no UV illumination confirmed that the decolorization is due to the photocatalytic reaction. Figure 10a shows the
measured data from the output of the sensor unitan almost
perfect exponential decay. Hence, we are assuming first order
kinetics, where the concentration C at time t is described by

C = C0 exp( kt )

Figure 10. (a) Degradation measurement of methyl orange solution in

the reactor, showing an exponential decay of concentration (C) with
decay rate (k). (b) Negative logarithm of the concentration divided by
the initial concentration (C0) and the observed decay rate (k).

0.52 ppm, we can estimate the cleaning capacity to be in the

range of 0.0036 mol L1 h1. This rate depends on the geometry of the reactor, which includes the ratio of photocatalytic surface area and water volume. To achieve an improved cleaning rate, this ratio has to be optimized through
the reactor design.
For continuous operation it is essential to demonstrate that the
reactor is stable and can be operated for many cycles. Figure 11a
shows four consecutive runs, where in each run the concentration
was set to 0.35 ppm in the reactor vessel at a filling level of
100 mL. As can be seen in the plot, the rate gives similar results for
each run. This demonstrates, therefore, that the system is stable
and the efficacy of the photocatalyst is conserved.
In Figure 11b the UVvis spectra of the contaminated model
water (MO solution) and the clean water after the photocatalytic
reaction is shown. The water containing MO exhibits the typical
peak around 465 nm. After the reaction, the peak disappears,
demonstrating complete removal. Baiocchi et al.28 have shown
that in the photocatalytic process the MO molecule is decomposed into smaller molecules. The reaction proceeds through a
number of steps including demethylation, hydroxyl attack on the
phenyl ring, and eventually cleavage of the azo bond. The final
end products are sulfate, water, and carbon dioxide.29

(3)

with initial concentration C0, and observed decay rate k. The

rate k is determined by the slope of a linear fit to ln(C/C0)
over t (see Figure 10b). From the data, we observe a decay
rate k of 0.5 h1. Taking into account the amount of water
(150 ML in this instance) and the initial concentration of

Figure 9. Representative SEM images of the coated wafer surface before (a) and after (b) photocatalytic reaction.
E

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

can take place.32 An important parameter is the incident light

intensity, which will influence the charge carrier dynamics in
the semiconductor and can affect both constants.33 To see the
dependence of the irradiant power of the UV in our system,
the light intensity was varied at constant initial concentration
(0.35 ppm) and constant filling levels of 100 mL. In Figure 13,

Figure 11. (a) Several cleaning cycles of freshly added polluted water
(methyl orange) demonstrates the possibility of continuous operation.
(b) UVvis of the prepared solution before and after cleaning shows
the complete removal. In the contaminated water methyl orange is
expressing the typical peak around 465 nm.

To understand the influence of essential parameters and to

demonstrate the capability of our setup, a series of experiments
were performed with variations in the initial MO concentration,
the irradiance, and the liquid volume.
Figure 12 presents observed decay rates from measurements
at varied initial concentrations. All measurements were performed

Figure 13. (a) Measurements at constant initial concentrations

(0.35 ppm) varying the UV irradiance in the center of the light field
illuminating the fixed photocatalyst. (b) A plot showing the obtained
decay rates against irradiance. The line is a guide, demonstrating the
linear increase of the rate with an increase of the irradiance.

measurements and derived rate constants for light intensities

from 50 to 200 W per cm2 are plotted. The rates follow an
almost linear increase as indicated by the dotted line in
Figure 13b. The small deviation of the point measured at
100 W/cm2 can be attributed to measurement errors.
To test the linear increase of the rate with light intensities
further, measurements at higher power intensities up to a
maximum of 2000 W/cm2 were performed. In Figures 14a,b
measurements on two different wafers are shown. We found
that our coating process resulted in inhomogeneous thickness and together with the nonuniform light field (as seen in
Figure 3) the system is sensitive to the exact position of the
wafer resulting in large fluctuations in the decay rates. Depending on the position of the wafer relative to the illuminating
UV-LEDs the rates can double as can be seen in Figure 14a.
An overall trend in all the measurements is the linear increase
in the irradiance range below 1000 W/cm2 and a saturation
effect that appears at higher UV power. We have also tested a
reduced set of five UV-LEDs and found that a similar effect
occurs. There appears to be a transition where the rate and its
dependence on the light intensity saturate at a similar position
around 1000 W/cm2. It was identified that the reaction rate
follows a linear relationship when the process is dominated
by the chemical reaction. If, however, the light flux reaches a
threshold the internal processes in the semiconductor can
become dominant and recombination of charges controls the
reaction. Similar behaviour with respect to light intensities has
been reported by Stefanov et al.34 and Wang et al.35
In a third series of experiments we tested the dependence of
decay rate on water volume in the reactor (see Figure 15). As
expected, there was a clear reduction in degradation rate when
the volume and, therefore, the total number of molecules which
have to be degraded increases. The measured rates follow an
exponential curve, as can be seen in Figure 15b. This can be
reasonably explained by changes in the mass flow rate over the
photocatalytic surface, which is kept constant.

Figure 12. Degradation rates of methyl orange depending on the

initial concentration of methyl orange used. The graph shows derived
values (black squares) from measurements and a fit (dotted line) using
the LangmuirHinshelwood kinetic rate model.

with a filling volume of 150 mL. It can be seen that there is a

steep increase in decay rate with increasing initial concentration, which plateaus at higher initial concentrations. A reaction model often used to explain this kinetic behavior is the
LangmuirHinshelwood (LH) model30 where the adsorption
constant Kads describes the rate of ad- and desorption of the
chemical under investigation on the surface and the constant
kLH describes other influences such as the light intensity. In the
model it follows that the degradation rate ri depends on the
initial concentration C0 in the form of

ri = kLH

K ads[C0]
1 + K ads[C0]

(4)

This model was fitted to our data (see Figure 12 dotted line)
resulting in values of kLH = 0.32 mol L1 h1 and Kads =
0.45 mol1 L. These values give an indication of the photocatalytic efficiency of our coated surface.
The LH model has been criticized as an oversimplification31 owing to the very complex nature of photocatalytic processes involving a series of steps from light absorption, transfer
of excited states to the surface, and production of active oxygen
species before a reduction/oxidation of a given molecule
F

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

have also developed a photocatalytic test reactor and demonstrated its function by measuring the degradation of methyl
orange. The sensor system allows us to monitor the degradation of the concentration in real-time and also records degradation
curves. From the data, we can calculate the first-order rate constant, which is a measure of the efficiency of the reaction. Our
system provides the possibility to investigate a range of important
parameters that can affect the reaction rate. We have demonstrated
its ability by showing its stability in operation and by investigating
the dependence of the reaction rate on initial concentration, light
intensity, and liquid volume to catalyst surface.
While our setup is designed specifically to study photocatalytic degradation of methyl orange, in principle, it could be
used to monitor any liquid-phase chemical or biochemical
reaction in real time. Some alternative reactions that our system
may be able to be adapted and optimized to study include monitoring fermentation reactions to detect changes in turbidity, detecting changes in metabolic product concentrations, and assessing
the effect of antibiotics on bio-organisms.
Recently, efforts have been made to introduce standards
(e.g., BSI:ISO 10678:2010) to enable comparison of the
efficiency of new photocatalysts developed in different laboratories or companies. Our cheap and simple setup could potentially be incorporated into standard procedures which would
allow different laboratories and companies to benchmark their
new photocatalysts against a competitor.
Because our system is very cheap it would be easy to scale-up
by purchasing additional units. For example, several tens of our
invention could be operated in parallel for the price of one
UVvis spectrometer.
In a real water purification plant, the water is assessed
throughout the treatment process as part of quality control. Our
sensor could be easily adapted as an in-line quality-testing tool that
could raise an alarm should the water quality fall outside sample
limits.

Figure 14. Rate constants were derived from measurements made at

constant initial concentrations, varying the UV irradiance in the center
of the light field illuminating the fixed photocatalyst on two different
wafers (a,b). The first wafer in panel a shows two series where the
position of the wafer is changed leading to large fluctuations in the
rate. The second wafer in panel b shows two series at lower and higher
irradiant power. The lines in both graphs are guides to the eye showing
trends in the change of rates with increasing UV power. Both wafers
show a region of linear increase at lower power but the rate reaches
saturation at higher power, the threshold being around 1000 W/cm2.

AUTHOR INFORMATION

Corresponding Author

*Tel.: +44 (0)29 208 75315. Fax: +44 (0)29 208 74056.
Address: School of Physics and Astronomy, Cardiff University,
Queens Buildings, The Parade, Cardiff CF24 3AA, United
Kingdom

ACKNOWLEDGMENTS
The authors acknowledge Hanbin Ma and Jun Yu for valuable
input on the circuit design, Deena Modeshia and Maurice
Mourad for assisting with the characterisation studies, Felicity
Sartain for project management of this work, and the Deanship
of Scientific Research at King Abdulaziz University for the
support of this project (T/80/429).

Figure 15. (a) Measurements at different volumes in the reactor

vessel. (b) Graph to show the observed decay rate against volume
showing an exponential slowdown of the rate as the volume of
polluted water increases.

Although energy efficient, due to UV-LEDs, the reactor design

would need to be enhanced in order to reach performances of a
suspension based system.21 To enhance the reactors cleaning
capacity the geometric arrangement of light source, liquid and
catalyst or periodic illumination has to be optimized.14 Another
possibility is to increase the ratio of coated surface to water volume,
introducing coated light guides.22

REFERENCES

(1) Corcoran, E.; Nellemann, C.; Baker, E.; Bos, R.; Osborn, D.;
Savelli, H. Sick Water? The Central Role of Wastewater Management
in Sustainable Development; United Nations Environment Programme, UN-HABITAT, GRID-Arendal: Arendal, Norway, 2010;
www.grida.no.
(2) Cheremisinoff, N. P. Handbook of Water and Wastewater
Treatment Technologies; Butterworth-Heinemann: Oxford, 2002.
(3) Matsuo, T., Hanaki, K., Takizawa, S.; Satoh, H. Advances in Water
and Wastewater Treatment Technology; Molecular Technology, Nutrient
Removal, Sludge Reduction and Environmental Health: Elsevier Science
BV: Amsterdam, 2001.

4. CONCLUSIONS
We have assembled a cheap, robust, and small closed circulating water system and have integrated a sensor unit that can
measure the concentration of chemicals in a water stream. We
G

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX

Industrial & Engineering Chemistry Research

Article

Surface versus lattice mechanisms. J. Phys. Chem. B 2005, 109, 977

980.
(25) Li, G. H.; Gray, K. A. The solidsolid interface: Explaining the
high and unique photocatalytic reactivity of TiO2-based nanocomposite materials. Chem. Phys. 2007, 339, 173187.
(26) Gan, X.; Li, X.; Gao, X.; Zhuge, F.; Yu, W. ZnO nanowire/TiO2
nanoparticle photoanodes prepared by the ultrasonic irradiation
assisted dip-coating method. Thin Solid Films 2010, 518, 48094812.
(27) Zhang, J.; Xu, Q.; Feng, Z.; Li, M.; Li, C. Importance of the
relationship between surface phases and photocatalytic activity of
TiO2. Angew. Chem., Int. Ed. 2008, 47, 1766 1769.
(28) Baiocchi, C.; Brussino, M. C.; Pramauro, E.; Prevot, A. B.;
Palmisano, L.; Marci, G. Characterization of methyl orange and its
photocatalytic degradation products by HPLC/UVVIS diode array
and atmospheric pressure ionization quadrupole ion trap mass
spectrometry. Int. J. Mass Spectrom. 2002, 214, 247256.
(29) Dvininov, E.; Joshi, U. A.; Darwent, J. R.; Claridge, J. B.; Xu, Z.;
Rosseinsky, M. J. Room temperature oxidation of methyl orange and
methanol over Pt-HCa2Nb3O10 and Pt-WO3 catalysts without light.
Chem. Commun. 2011, 47, 881883.
(30) Ollis, D. F. Kinetic disguises in heterogeneous photocatalysis.
Top. Catal. 2005, 35, 217223.
(31) Emeline, A. V.; Ryabchuk, V. K.; Serpone, N. Dogmas and
misconceptions in heterogeneous photocatalysis. Some enlightened
reflections. J. Phys. Chem. B 2005, 109, 1851518521.
(32) Ollis, D. F. Kinetics of liquid phase photocatalyzed reactions: An
illuminating approach. J. Phys. Chem. B 2005, 109, 24392444.
(33) Mills, A.; Wang, J. S.; Ollis, D. F. Kinetics of liquid phase
semiconductor photoassisted reactions: Supporting observations for a
pseudo-steady-state model. J. Phys. Chem. B 2006, 110, 1438614390.
(34) Stefanov, B. I.; Kaneva, N. V.; Puma, G. L.; Dushkin, C. D.
Novel integrated reactor for evaluation of activity of supported
photocatalytic thin films: Case of methylene blue degradation on TiO2
and nickel modified TiO2 under UV and visible light. Colloid Surface A.
2011, 382, 219225.
(35) Wang, Z.; Liu, J.; Dai, Y.; Dong, W.; Zhang, S.; Chen, J.
Dimethyl sulfide photocatalytic degradation in a light-emitting-diode
continuous reactor: Kinetic and mechanistic study. Ind. Eng. Chem. Res.
2011, 50, 79777984.

(4) Hoffmann, M. R.; Martin, S. T.; Choi, W. Y.; Bahnemann, D. W.

Environmental Applications of Semiconductor Photocatalysis. Chem.
Rev. 1995, 95, 6996.
(5) Hashimoto, K.; Irie, H.; Fujishima, A. TiO2 photocatalysis, a
historical overview and future prospects. Jpn. J. Appl. Phys. 2005, 44,
82698285.
(6) Gaya, U. I.; Abdullah, A. H. Heterogeneous photocatalytic
degradation of organic contaminants over titanium dioxide: A review
of fundamentals, progress and problems. J. Photochem. Photobiol. C
2008, 9 (1), 112.
(7) Chen, X.; Mao, S. S. Titanium dioxide nanomaterials: Synthesis,
properties, modifications, and applications. Chem. Rev. 2007, 107,
28912959.
(8) The Food Standard Agency: Current EU approved additives and
their E Numbers: http://www.food.gov.uk/safereating/chemsafe/
additivesbranch/enumberlist, 2010 (Accessed September 2011).
(9) Feitz, A. J.; Boyden, B. H.; Waite, T. D. Evaluation of two solar
pilot-scale fixed-bed photocatalytic reactors. Water Res. 2000, 34,
39273932.
(10) Braham, R. J.; Harris, A. T. Review of major design and scale-up
considerations for solar photocatalytic reactors. Ind. Eng. Chem. Res.
2009, 48, 88908905.
(11) Chong, M. N.; Jin, B.; Chow, C. W.; Saint, C. Recent
developments in photocatalytic water treatment technology: A Review.
Water Res. 2010, 44, 29973027.
(12) Friedmann, D.; Mendive, C.; Bahnemann, D. TiO2 for water
treatment: Parameters affecting the kinetics and mechanisms of
photocatalysis. Appl. Catal. B 2010, 99, 398406.
(13) Brown, G. T.; Darwent, J. R. Methyl-orange as a probe for
photooxidation reactions of colloidal TiO2. J. Phys. Chem. 1984, 88,
49554959.
(14) Wang, W. Y.; Ku, Y. Photocatalytic degradation of Reactive Red
22 in aqueous solution by UVLED radiation. Water Res. 2006, 40,
22492258.
(15) Ghosh, J. P.; Langford, C. H.; Achari, G. Characterization of an
LED based photoreactor to degrade 4-chlorophenol in an aqueous
medium using coumarin (C-343) sensitized TiO2. J. Phys. Chem. A
2008, 112, 1031010314.
(16) Ghosh, J. P.; Sui, R.; Langford, C. H.; Achari, G.; Berlinguette,
C. P. A comparison of several nanoscale photocatalysts in the
degradation of a common pollutant using LEDs and conventional UV
light. Water Res. 2009, 43, 44994506.
(17) Natarajan, T. S.; Natarajan, K.; Bajaj, H. C.; Tayade, R. J. Energy
efficient UV-LED source and TiO2 nanotube array-based reactor for
photocatalytic application. Ind. Eng. Chem. Res. 2011, 50, 77537762.
(18) Wurtele, M. A.; Kolbe, T.; Lipsz, M.; Kulberg, A.; Weyers, M.;
Kneissl, M.; Jekel, M. Application of GaN-based ultraviolet-C light
emitting diodes-UV LEDs-for water disinfection. Water Res. 2011, 45,
14811489.
(19) Mori, M.; Hamamoto, A.; Takahashi, A.; Nakano, M.;
Wakikawa, N.; Tachibana, S.; Ikehara, T.; Nakaya, Y.; Akutagawa,
M.; Kinouchi, Y. Development of a new water sterilization device with
a 365 nm UV-LED. Med. Biol. Eng. Comput. 2007, 45, 12371241.
(20) Tayade, R. J.; Natarajan, T. S.; Bajaj, H. C. Photocatalytic
degradation of methylene blue dye using ultraviolet light emitting
diodes. Ind. Eng. Chem. Res. 2009, 48, 1026210267.
(21) Natarajan, T.S.; Thomas, M.; Natarajan, K.; Bajaj, H.C.; Tayade,
R.J.; , Study on UVLED/TiO2 process for degradation of rhodamine B
dye, Chem. Eng. J. 169 2011, 169, 126134.
(22) Kalithasan, N.; Natarajan, T. S.; Bajaj, H. C.; Tayade, R. J.
Photocatalytic reactor based on UV-LED/TiO2 coated quartz tube for
degradation of dyes. Chem. Eng. J. 2011, 178, 4049.
(23) Doll, T. E.; Frimmel, F. H. Kinetic study of photocatalytic
degradation of carbamazepine, clofibric acid, iomeprol, and iopromide
assisted by different TiO2 materials-determination of intermediates
and reaction pathways. Water Res. 2004, 38, 955964.
(24) Hurum, D. C.; Gray, K. A.; Rajh, T.; Thurnauer, M. C.
Recombination pathways in the Degussa P25 formulation of TiO2:
H

dx.doi.org/10.1021/ie202366m | Ind. Eng.Chem. Res. XXXX, XXX, XXXXXX