Water puri ﬁ cation in a solar reactor incorporating TiO 2 coated mesh structures

The rate of photocatalytic oxidation of contaminants in drinking water using an immobilized catalyst can be increased by properly designing the catalyst structure. By creating a solar reactor in which meshes coated with TiO 2 were stacked, we demonstrated that degradation of humic acids with four superimposed stainless steel meshes was up to 3.4 times faster than in a single plate ﬂ at-bed reactor. Incorporation of TiO 2 coated mesh structures resulted in a high speci ﬁ c photocatalytically active surface area with suf ﬁ cient light penetration in the reactor, while the coated area for one mesh was 0.77 m 2 per m 2 projected area. This brought the photocatalytic ef ﬁ ciency of such reactors closer to that of dispersed-phase reactors, but without the complex separation of the very ﬁ ne TiO 2 particles from the treated water.


INTRODUCTION
Heterogeneous solar photocatalysis uses solar UV light in the range between 300 nm and 400 nm to photo-excite the catalyst in contact with water and in the presence of oxygen (Malato et al. ) to generate hydroxyl radicals ( • OH).
TiO 2 is the most-used photocatalyst because of its biological and chemical inertness, photo-stability, availability, and nontoxicity (McCullagh et al. ). The largest exposed active catalyst surface area, relative to the reactor volume, can be obtained in a so-called dispersed-phase reactor. However, a complex and expensive separation step is required to remove the fine catalyst particles from the water. This can be avoided by using immobilized photocatalyst reactors (Feitz et al. ; Malato et al. ; McCullagh et al. ).
For efficient contaminant degradation such reactors require a large reactor surface with low throughput because of their low catalyst surface to reactor volume ratio. The degradation potential can be increased by changing the configuration of such reactors. Zhang et al. () reported that the rate of degradation with a corrugated plate was a factor 1.5 faster than with the flat-plate for the same exposed reactor area.
More complex configurations can be used to increase even further the catalyst surface to reactor volume ratio. However, although the catalyst surface to reactor volume ratio may increase, light penetration in the reactor can then become a limiting factor (Feitz et al. ; Dijkstra et al. ). used as catalyst support material and TiO 2 coatings were applied by electrophoretic deposition (EPD) using a commercial sol gel (O5100) as a stable suspended electrolyte.
In this paper, the TiO 2 coated mesh reactor was used to study the adsorption and solar photocatalytic degradation efficiency of humic acids (HA) as a model compound for surface water contaminants. The performance of the catalyst structure was investigated at different scales, including a batch system in a beaker and a recirculating flow fixed-bed reactor. Results were compared with those obtained with a similarly coated flat-plate reactor of the same size.

Preparation of the immobilized TiO 2 catalyst
Immobilization of the TiO 2 photocatalyst on grade 304 stainless steel woven meshes and flat plates was done using the EPD technique described in El-Kalliny et al.

Fixed-bed reactor and photocatalytic activity evaluation
The batch experiments with small circular meshes (ø 5.5 cm) were carried out in 250 mL Pyrex glass beakers (internal diameter 7 cm) using a magnetic stirrer (IKA, RT15) with a stirring rate of 550 rpm. The meshes were placed on top of each other and separated by 2-mm-thick silicone rubber rings. The temperature was kept at 32 ± 1 C by using a recirculation cooler (Julabo, FL300). This temperature is a simulation for the temperature of the system subjected to solar light. Working solutions of HA were prepared from a 1,000 mg·L À1 stock solution of HA sodium salt (Sigma Aldrich). The HA sodium salt was dissolved in aerated deionized water and filtered through a 0.45 μm syringe-driven filter unit (Millex) to remove suspended solids (0.03 g·L À1 ).  respectively, at constant temperature 32 C ±1 C. The adsorbed amount of HA (q t , mg·m À2 ) on the coated meshes was determined by: is the volume of the HA solution, and A S (m 2 ) is the surface area of the stainless steel woven mesh coated by the photocatalyst.
A S is half the value of the total surface area of the mesh (A Mesh ) as the TiO 2 film was coated on one side of the mesh which was facing the anode in the EPD process. Hence: where L (m) is the total length of the wire from which the mesh is made and D (m) is the diameter of the wire. No corrections for the wire diameter due to the coating were carried out as it was small compared with the diameter of the virgin wire. The length of the wire was determined by: where W (kg) is the weight of the wire mesh and ρ is the density of the stainless steel (7,977 kg·m À3 ). By weighing a mesh, the surface area of that mesh was calculated knowing its wire diameter (0.355 mm according to the manufacturer). The 5.8 cm diameter disk used in the batch experiments weighed 2.88 g, so its coated surface area was 20.3 cm 2 for a projected area A P of 26.4 cm 2 . The coated area was thus 0.77 m 2 per m 2 projected area.
Because of the grid structure of the mesh, there was wire overlap at the contact points. From basic geometrical considerations it can be calculated that for the given wire diameter and mesh aperture the overlap was 16.25% of the total wire surface. The exposed coated surface area A S was therefore corrected accordingly.
The adsorption experiments were carried out in the dark for different starting concentrations in the range of 6 mg·L À1 to 14 mg·L À1 . Assuming that the adsorption kinetics of HA where q e (mg·m À2 ) and q t (mg·m À2 ) are the amounts of HA adsorbed at equilibrium and at time t (min), respectively, and k 1 (min À1 ) is the pseudo-first-order rate constant.
The adsorbed amount at equilibrium, q e (mg·m À2 ), can be obtained through the empirical equation proposed by Freundlich (Zhao et al. ): where K f (L·m À2 , in case of n ¼ 1) is the Freundlich isotherm constant and is a measure of adsorptive capacity, and n determines the intensity of adsorption.
Precise experimental determination of both q e and C e is difficult because of mass conservation; both are related to the initial concentration of HA in the solution (C 0 ). By combining Equations (4) and (5) and substituting C 0 for C e , with an assumption that q e is C 0 -dependent in order to simplify the calculations, the amount of HA adsorbed q t at t t can be obtained by the following equation:

Analytical methods
The HA concentration was determined by the UV absorption at 254 nm, which is representative of the aromatic moieties. This was done with a Hach Lange DR 5,000 spectrophotometer as described in El-Kalliny et al. ().

Solar radiation evaluation
The intensity of the UV solar light (300-400 nm) was measured with a Xenocal UV-sensor with a resolution of 0.1 J·m À2 ·s À1 . The exposed energy Q UV,N (kJ·L À1 ) is the total radiation energy absorbed per unit volume in the reactor from the beginning of an experiment up to a given time ( with where t n (s) is the experimental time for each sample, I N (kJ·m À2 ·s À1 ) is the intensity of solar UV 300-400 nm irradiation projected on top of a mesh layer during the time interval t n , A P (m 2 ) is the projected area for the mesh layer, N is the number of the mesh layers, and V (L) is the reactor volume. In the case of a first-order reaction, the reaction rate is expressed in units of mg·kJ À1 of UV irradiated on the catalyst surface, as the exposed energy was used instead of time to describe the process.
where For the large-scale fixed-bed reactor, the separators were taken into account and the intensity of solar UV 300-400 nm I N (kJ·m À2 ·s À1 ) on top of the N th mesh was determined by: where A T (m 2 ) is the total area of the mesh layer, and A Sep (m 2 ) is the area of the separator that blocks the light.
The applicability of the Langmuir-Hinshelwood (LH) model gives an interpretation for the mechanism of the photocatalytic degradation of HA using coated meshes.
The reaction rate in exposed energy Q is obtained by using the LH expression given by Kumar et al. (): where k r is the rate constant of photochemical reaction, and K LH is the adsorption coefficient of HA onto the catalyst during the irradiation period. If the LH obeys pseudo-firstorder kinetics, then r and k r are in mg·kJ À1 and K LH is in L·mg À1 . The reaction rate r is represented as a function of the initial concentration of HA (C o ) as follows: The parameters k r and K LH were predicted by linearizing Equation (12) as follows:

RESULTS AND DISCUSSION
Adsorption of HA on coated meshes  Table S1, available with the online version of this paper).
It was found that the calculated q e (via Equation (5)

Solar photocatalytic degradation of HA
Before starting photocatalytic degradation experiments, the systems were preconditioned for 2 h in the dark to attain adsorption equilibrium. When exposure to solar light started, the HA concentration was equal to the C e . Therefore, in Figure 2, the decrease of HA concentration due to exposed solar energy Q (in kJ·L À1 ) is given as C/C e for different C 0 values. An additional photolysis experiment was done in the batch reactor by irradiating a HA solution of 10 mg·L À1 without any photocatalyst present. The results in Figure 2(a) show that no noticeable degradation of HA occurred up to 12.5 kJ·L À1 of solar energy irradiation without the catalyst. However, when applying the small coated meshes, HA concentrations were reduced to less than half of those when solar irradiation started. By using a flat plate in the fixed bed reactor at an initial HA concentration of 10 mg·L À1 the solar photocatalytic degradation of HA was 3.4 times lower than in the case of the coated mesh structure at the same exposed energy (Figure 2(b)). Even though the light intensity was not homogeneously distributed over the expanded surface area (A S for four coated meshes was increased from 1 m 2 to 3 m 2 , see above), the attenuation factor was still higher than 3, indicating that a more spatial distribution of light over the surface was beneficial. Not only were more active sites of TiO 2 photocatalyst available on the mesh structure than on the flat plate for the same volume of water, but these sites were also more efficient due to better light distribution. In order to investigate the dependence of photocatalytic degradation on the adsorption step by checking the applicability of the LH model, it is essential to calculate the apparent rate constants (k app ) for the removal of HA for different initial HA concentrations (Equation (12)). The relations between ln(C) and the exposed energy (Q), including light absorption, resulted in straight lines for different C o .
The slopes of these lines represent the apparent rate constants k app . These pseudo-first-order rate constants are inserted in Figure 2 together with their correlation coefficient R 2 , which is always greater than 0.98. The k app was calculated for the first stage of the photocatalytic reaction. In this stage, the concentration of HA is predominant relative to the by-products. Accordingly, the rate of reaction depends on the HA concentration, not on by-products.
Therefore, the simplified form of the LH model (Equation (12)) was used. Consequently, the experimental data are fitted to the pseudo-first-order kinetic model, as shown in Figure 2. It is noticeable that k app for the degradation of HA in the fixed bed reactor is higher than that in the glass beaker reactor, as is shown in Figure 2. This can be attribu-  (9) and (10)), therefore, this behaviour could be due to the higher number of adsorbed HA molecules (i.e. q e ) at a high HA C o . These HA molecules may block some active sites on the TiO 2 surface, which are responsible for the production of active radicals. Hence, the number of produced • OH decreases with increasing HA C o , leading to a slower rate of photodegradation. This indicates that availability of • OH is rate-limiting by hindering the reaction of HA with the produced • OH.
The k app for the degradation of HA by coated meshes was 3.4 times higher than that over a flat plate at the same absorbed energy, the same projected area, and at the same initial HA concentration of 10 mg·L À1 . (1.495 mg·kJ À1 ) was close to that for small coated meshes in the batch system (1.462 mg·kJ À1 ), which proves that up-scaling was feasible.

CONCLUSIONS
• The adsorption kinetics of HA onto TiO 2 film coated over stainless steel woven meshes obeyed pseudo-firstorder adsorption kinetics according to the Freundlich adsorption isotherm.
• The degradation kinetics of HA using small coated meshes in the batch reactor and large coated meshes in the recirculating flow fixed-bed solar reactor obeyed the LH kinetic model.
• Up-scaling of mesh size was possible as was shown by the comparable adsorption and photodegradation kinetics for the two investigated reactor sizes.
• The photocatalytic degradation of HA in a solar reactor with four superimposed stainless steel meshes was 3.4 times faster than the flat-plate reactor. This brought the photocatalytic efficiency of such reactors closer to that of dispersed-phase reactors without the complex separation of the TiO 2 photocatalyst.