Packed Bed Photoreactor for the Removal of Water Pollutants Using Visible Light Emitting Diodes

A packed bed photoreactor was developed using a structured photocatalyst active under visible light. The packed bed reactor was irradiated by visible light-emitting diodes (LEDs) for the evaluation of photocatalytic activity in the removal of different types of water pollutants. By using a flexible LEDs strip as the external light source, it was possible to use a simple cylindrical geometry for the photoreactor, thereby enhancing the contact between the photocatalyst and the water to be treated. The visible light active structured photocatalyst was composed by N-doped TiO2 particles supported on polystyrene spheres. Photocatalytic results showed that the almost total methylene blue decolorization was achieved after 120 min of irradiation. Moreover, the developed packed bed photoreactor was effective in the removal of ceftriaxone, paracetamol, and caffeine, allowing it to reach the almost total degradation of the pollutants and a total organic carbon removal above 80% after 180 min of visible light irradiation.


Introduction
The main causes of water contamination are industrial discharges, excess use of pesticides and fertilizers, the presence of pharmaceutical residues, and landfilling of domestic wastes [1].Traditional water treatment processes are mainly based on mechanical (sedimentation and filtration), biological, and physical-chemical processes (flocculation, sterilization, or chemical oxidation) [2,3].Advanced treatment technologies are based on the pollutants oxidation by potassium permanganate, ozone, hydrogen peroxide, high-energy ultraviolet light, photo-Fenton, electro-Fenton, and photo-electro-Fenton processes [4][5][6][7][8].Typically, these technologies are not cost-effective [9].On the other hand, the conventional wastewater treatments have different disadvantages, such as use of chemicals and production of sludge, as well as low efficiency in removal of bio-recalcitrant compounds [10].Heterogeneous photocatalysis is one of the advanced oxidation processes that couples low-energy light with semiconductors acting as photocatalysts and, therefore, it could be considered as sustainable process for the treatment and purification of water and wastewater [11,12].Although in recent years, a large number of publications have shown the potential of photocatalytic treatment for the removal of different categories of toxic compounds in water [13][14][15] using also innovative photoreactors configurations [16][17][18], the research is still limited to the development of photocatalytic laboratory reactors or to photoreactors very difficult to scale up.It must be considered that the design of a photocatalytic reactor is not an easy step; in particular, the development of a photocatalytic reactor is complicated because it is necessary to have an adequate irradiation of the overall reaction volume for the activation of the entire mass of photocatalyst and to favor an easy catalyst separation after treatment [19].Typically, these reactors are designed to study the photocatalytic degradation of pollutants through the application of conventional lamps as light sources [20].However, since conventional lamps are expensive in terms of manufacturing, as well as electricity consumption, many studies have been devoted to photocatalytic reactors that exploit the solar spectrum [21].
But it is necessary to take into account the many limitations related to solar photoreactors.For example, in a solar compound parabolic collector (CPC), despite the fact that the reactor is formed by small tubes, the area equipped by the parabolic collector for light harvesting is huge.So, the installation of these systems also requires a large area [22] to avoid that the single modules projecting shadows upon each other [23].Another limitation this kind of system is connected to the use of only direct sunlight [24,25].Thus, the solar photoreactors do not work efficiently during cloudy days and night time.The recent rapid advances in light-emitting diodes (LEDs) technology have made possible a new frontier for a wide range of lighting applications, including photocatalysis and, consequently, photoreactor design [26,27].In fact, the energy conversion efficiency of LEDs has increased exponentially in recent years, making them a suitable light source for the design of compact photocatalytic reactors for water and wastewater treatment [28,29], as well as for the removal of hazardous compounds from gaseous streams [30], overcoming the limitations related to the developing of photocatalytic reactors into pilot scale level.With the use of LEDs, the design of photoreactors could be significantly facilitated and, by using a flexible LEDs strip as the external light source, it is possible to use simple cylindrical geometry for photoreactor, thereby enhancing the contact between the photocatalyst and the water to be treated.
Moreover, cylindrical geometry for photocatalytic reactors is the most used in solar through collectors [31], and it is the simplest configuration, allowing a possible scale-up of photocatalytic systems for water and wastewater treatment.
Considering all the aspects describe above, the aim of this paper is to develop a cylindrical packed bed photoreactor at semi-pilot scale irradiated by visible LEDs and using a visible light active structured photocatalyst.The visible light active structured photocatalyst is composed by N-doped TiO 2 supported on polystyrene (PS) spheres [32,33] in order to avoid a post-separation step necessary for the separation of the photocatalyst from the treated water.The photocatalytic system has been tested in the removal of different types of water pollutants (methylene blue (MB), caffeine (CAF), paracetamol (PC), and ceftriaxone (CFX)).

Discoloration and Mineralization of Methylene Blue with the Packed Bed Photoreactor Operating in Batch Mode
Preliminary experiments were carried out while operating in batch mode (Figure 1) in order to verify that methylene blue (MB) was degraded by the heterogeneous photocatalytic process.
Appl.Sci.2019, 9, x; doi: 2 of 15 conventional lamps are expensive in terms of manufacturing, as well as electricity consumption, many studies have been devoted to photocatalytic reactors that exploit the solar spectrum [21].But it is necessary to take into account the many limitations related to solar photoreactors.For example, in a solar compound parabolic collector (CPC), despite the fact that the reactor is formed by small tubes, the area equipped by the parabolic collector for light harvesting is huge.So, the installation of these systems also requires a large area [22] to avoid that the single modules projecting shadows upon each other [23].Another limitation this kind of system is connected to the use of only direct sunlight [24,25].Thus, the solar photoreactors do not work efficiently during cloudy days and night time.The recent rapid advances in light-emitting diodes (LEDs) technology have made possible a new frontier for a wide range of lighting applications, including photocatalysis and, consequently, photoreactor design [26,27].In fact, the energy conversion efficiency of LEDs has increased exponentially in recent years, making them a suitable light source for the design of compact photocatalytic reactors for water and wastewater treatment [28,29], as well as for the removal of hazardous compounds from gaseous streams [30], overcoming the limitations related to the developing of photocatalytic reactors into pilot scale level.With the use of LEDs, the design of photoreactors could be significantly facilitated and, by using a flexible LEDs strip as the external light source, it is possible to use simple cylindrical geometry for photoreactor, thereby enhancing the contact between the photocatalyst and the water to be treated.
Moreover, cylindrical geometry for photocatalytic reactors is the most used in solar through collectors [31], and it is the simplest configuration, allowing a possible scale-up of photocatalytic systems for water and wastewater treatment.
Considering all the aspects describe above, the aim of this paper is to develop a cylindrical packed bed photoreactor at semi-pilot scale irradiated by visible LEDs and using a visible light active structured photocatalyst.The visible light active structured photocatalyst is composed by N-doped TiO2 supported on polystyrene (PS) spheres [32,33] in order to avoid a post-separation step necessary for the separation of the photocatalyst from the treated water.The photocatalytic system has been tested in the removal of different types of water pollutants (methylene blue (MB), caffeine (CAF), paracetamol (PC), and ceftriaxone (CFX)).

Discoloration and Mineralization of Methylene Blue with the Packed Bed Photoreactor Operating in Batch Mode
Preliminary experiments were carried out while operating in batch mode (Figure 1) in order to verify that methylene blue (MB) was degraded by the heterogeneous photocatalytic process.The obtained results are reported in Figure 2. In the absence of a structured catalyst, no significant decrease in MB concentration was observed during the overall irradiation time, evidencing that the visible light was not able to degrade the MB molecule.A completely different behaviour was observed Appl.Sci.2019, 9, 472 3 of 14 when N-TiO 2 /PS structured photocatalyst was placed into the photoreactor.In fact, the MB relative concentration gradually decreased in the run time and almost complete decolorization was achieved after 120 min of visible light irradiation.Several published papers have reported the removal of MB with fixed bed photoreactors irradiated by artificial light sources, as well as with solar photocatalytic reactors using structured photocatalysts (Table 1) [34][35][36][37][38].For example, almost total MB decolorization was achieved after 48 h in the solar photoreactor proposed by Sutisna et al. [34], in which TiO 2 nanoparticles were placed as coating on transparent granules and used in the prototype of a flat-panel photoreactor.El-Mekkawi et al. [35] observed the complete decolorization after 5 h under the sunlight irradiation using a solar concentrator photoreactor prototype with TiO 2 deposited on polyethylene terephthalate (PET) textile substrate sheets.In some cases, it was reported that MB degraded more than 96% in up to 90 min of reaction using TiO 2 immobilized on glass spheres [36].However, the same authors used a lab scale compound parabolic collector reactor irradiated by lamp at high electrical power (300 W) simulating the solar spectrum.The obtained results are reported in Figure 2. In the absence of a structured catalyst, no significant decrease in MB concentration was observed during the overall irradiation time, evidencing that the visible light was not able to degrade the MB molecule.A completely different behaviour was observed when N-TiO2/PS structured photocatalyst was placed into the photoreactor.In fact, the MB relative concentration gradually decreased in the run time and almost complete decolorization was achieved after 120 min of visible light irradiation.Several published papers have reported the removal of MB with fixed bed photoreactors irradiated by artificial light sources, as well as with solar photocatalytic reactors using structured photocatalysts (Table 1) [34][35][36][37][38].For example, almost total MB decolorization was achieved after 48 h in the solar photoreactor proposed by Sutisna et al. [34], in which TiO2 nanoparticles were placed as coating on transparent granules and used in the prototype of a flat-panel photoreactor.El-Mekkawi et al. [35] observed the complete decolorization after 5 h under the sunlight irradiation using a solar concentrator photoreactor prototype with TiO2 deposited on polyethylene terephthalate (PET) textile substrate sheets.In some cases, it was reported that MB degraded more than 96% in up to 90 min of reaction using TiO2 immobilized on glass spheres [36].However, the same authors used a lab scale compound parabolic collector reactor irradiated by lamp at high electrical power (300 W) simulating the solar spectrum.As observed in our previous work [33], the performances of the photocatalytic system were not influenced by mass transfer limitation phenomena of MB dye from the aqueous phase to the photoactive surface, evidencing that the MB photocatalytic degradation rate is controlled by kinetics.In addition, after 300 min of visible light irradiation, the system allowed to achieve a total organic carbon (TOC) removal of 85±2% for all the liquid flow rates (Figure 3).Similar behaviours were observed for other photocatalytic systems in which the photoactive phases were immobilized on macroscopic supports [38,39].In order to verify the stability of the structured photocatalyst, the photocatalytic degradation of MB was replicated up to four cycles.Before each reuse cycle, any cleaning or regenerating step of the structured photocatalyst was performed.The treated solution was removed from the photoreactor, and a fresh solution with the same initial MB concentration was fed for a new test.The MB decolorization and total organic carbon (TOC) removal was analyzed after 300 min visible light irradiation and the obtained results are shown in Figure 4.The treated solution was removed from the photoreactor, and a fresh solution with the same initial MB concentration was fed for a new test.The MB decolorization and total organic carbon (TOC) removal was analyzed after 300 min visible light irradiation and the obtained results are shown in Figure 4.
From the comparison between the first cycle (when N-TiO 2 /PS was fresh) and the fourth cycle, it was possible to observe that the photocatalytic activity was similar.This result indicated that N-TiO 2 /PS did not deactivate after its reuse and no loss of photocatalyst particles from the polymer surface occurred, in agreement with previous results observed on structured photocatalysts prepared through solvent-casting method [32,33,40].
The treated solution was removed from the photoreactor, and a fresh solution with the same initial MB concentration was fed for a new test.The MB decolorization and total organic carbon (TOC) removal was analyzed after 300 min visible light irradiation and the obtained results are shown in Figure 4.

Photocatalytic Results on Methylene Blue Degradation with the Packed Bed photoreactor: Kinetic Evaluation and Mathematical Modelling
The mathematical modeling of the photocatalytic system, based on the use of the developed packed bed photoreactor at semi-pilot scale, has been performed using first order kinetics for MB photodegradation [41] and neglecting the external mass transfer phenomena.In order to verify this last hypothesis, the apparent kinetic constant was estimated from the experimental data collected using different liquid flow rates when the photoreactor operated in continuous mode (Figure 5) and at steady-state condition (Table 2).
Appl.Sci.2019, 9, x; doi: 6 of 15 From the comparison between the first cycle (when N-TiO2/PS was fresh) and the fourth cycle, it was possible to observe that the photocatalytic activity was similar.This result indicated that N-TiO2/PS did not deactivate after its reuse and no loss of photocatalyst particles from the polymer surface occurred, in agreement with previous results observed on structured photocatalysts prepared through solvent-casting method [32,33,40].

Photocatalytic Results on Methylene Blue Degradation with the Packed Bed photoreactor: Kinetic Evaluation and Mathematical Modelling
The mathematical modeling of the photocatalytic system, based on the use of the developed packed bed photoreactor at semi-pilot scale, has been performed using first order kinetics for MB photodegradation [41] and neglecting the external mass transfer phenomena.In order to verify this last hypothesis, the apparent kinetic constant was estimated from the experimental data collected using different liquid flow rates when the photoreactor operated in continuous mode (Figure 5) and at steady-state condition (Table 2).Considering the plug-flow behavior inside the packed bed, the MB mass balance can be written as: where:  Considering the plug-flow behavior inside the packed bed, the MB mass balance can be written as: where: Q 0 = liquid flow rate, L/min; K = apparent kinetic constant, L/(g min); X = MB conversion; W cat = structured catalyst amount, g.
The boundary condition for Equation ( 1) is: Equation ( 1) is easily integrated obtaining the following relationship: Figure 6 reports the behavior of (-ln(1 − X)) as a function of W cat /Q 0 .The slope of the obtained straight line is the value of the apparent kinetic constant K (3 × 10 −5 L/(g min)), which can be seen as the product between the kinetic constant and the light intensity at the external surface of the photoreactor (Equation ( 3)) [38]: k kinetic = kinetic constant, L cm 2 /(g min mW); I = light intensity, mW/cm 2 .
Appl.Sci.2019, 9, x; doi: 7 of 15 the product between the kinetic constant and the light intensity at the external surface of the photoreactor (Equation ( 3)) [38]: where: kkinetic = kinetic constant, L cm 2 /(g min mW); I = light intensity, mW/cm 2 .Considering the light intensity of 78 mW/cm 2 , it was possible to evaluate the value of kkinetic from Equation (3), obtaining a value equal to 3.85×10 -7 L cm 2 /(g min mW).
The behaviour shown in Figure 6 confirmed the hypothesis that external mass transfer phenomena can be neglected since only one value of K is able to described the results obtained using Considering the light intensity of 78 mW/cm 2 , it was possible to evaluate the value of k kinetic from Equation (3), obtaining a value equal to 3.85×10 −7 L cm 2 /(g min mW).
The behaviour shown in Figure 6 confirmed the hypothesis that external mass transfer phenomena can be neglected since only one value of K is able to described the results obtained using different liquid flow rates [38,42].
The accuracy of the obtained value for k kinetic was verified by performing another series of experiments in which the visible light intensity was varied from 17.2 up to 78 mW/cm 2 to verify the ability of the model to predict experimental behaviour.In this case, the photoreactor operated in in batch mode, according to the schematic picture reported in Figure 1.In fact, the packed bed photoreactor was coupled with a tank, and the aqueous solution is continuously recirculated by means of the peristaltic pump for feeding the polluted water into the photoreactor.In this way, the reactor can be modeled like a batch system, as also reported in the literature concerning different photoreactor configurations for water and wastewater treatment [43].
Therefore, the MB mass balance can be written as: The initial condition for Equation ( 4) is: where: C 0 = MB initial concentration, mg/L; C = MB concentration at the generic irradiation time, mg/L; V = treated solution volume, L; W cat = structured catalyst amount, g.
Considering the mass balance on MB (Equation ( 4)) and integrating it between initial time (t = 0) and generic irradiation time t, it was obtained: The value of k kinetic , calculated from Equation (3) was used in Equation ( 5), and the calculated values of MB relative concentration as a function of irradiation time were compared with the experimental results collected at different light intensities (Figure 7).
As expected, the photocatalytic performances increased with the visible light intensity and the results obtained by the mathematical model were in agreement with the experimental results, evidencing the ability of the mathematical model to be predictive.
The value of kkinetic, calculated from Equation ( 3) was used in Equation ( 5), and the calculated values of MB relative concentration as a function of irradiation time were compared with the experimental results collected at different light intensities (Figure 7).As expected, the photocatalytic performances increased with the visible light intensity and the results obtained by the mathematical model were in agreement with the experimental results, evidencing the ability of the mathematical model to be predictive.

Degradation and Mineralization of Ceftriaxone, Paracetamol, and Caffeine with the Packed Bed Photoreactor
The efficiency of the packed bed photoreactor was also evaluated in the removal of water pollutants of emerging concern, like pharmaceuticals and personal care products (PPCPs) [44][45][46].The molecules tested in the present study were selected among the emerging pollutants detected in some Italian rivers and lakes and to represent different classes of PPCPs [47].More precisely, ceftriaxone (CFX), caffeine (CAF), and paracetamol (PC) were chosen for photocatalytic tests.CFX is a long-acting, broad-spectrum cephalosporin antibiotic for parenteral use [48]

Degradation and Mineralization of Ceftriaxone, Paracetamol, and Caffeine with the Packed Bed Photoreactor
The efficiency of the packed bed photoreactor was also evaluated in the removal of water pollutants of emerging concern, like pharmaceuticals and personal care products (PPCPs) [44][45][46].The molecules tested in the present study were selected among the emerging pollutants detected in some Italian rivers and lakes and to represent different classes of PPCPs [47].More precisely, ceftriaxone (CFX), caffeine (CAF), and paracetamol (PC) were chosen for photocatalytic tests.CFX is a long-acting, broad-spectrum cephalosporin antibiotic for parenteral use [48].CF is a well-known compound used in pharmaceutical formulations as a stimulant and, moreover, is the most widely consumed psychoactive drug [49].Finally, PC, also known as acetaminophen, is the compound used in many analgesic and antipyretic medicines [49].For the photocatalytic tests on CFX, CAF, and PC, the packed bed photoreactor was used operating in batch mode.The initial concentration value for all PPCPs was mg/L.The obtained results are depicted in Figure 8.It was possible to observe a progressive decrease of the pollutants concentration during the irradiation time, evidencing that the photocatalytic system was effective in the removal of all the tested PPCPs.
In particular, for CFX, CAF, and PC, similar degradation degree (about 99% for CFX, 95% for CAF, and 99% for PC) and TOC removal (about 85% for CFX, 80% for CAF, and 83% for PC) were achieved after 180 min of visible light irradiation (Figure 9).Considering the literature concerning the degradation of CFX, PC, and CAF, it must be underlined that the most studied photocatalysts are used in powder form [49,50], making their use on an industrial scale very difficult.However, some studies have also reported the immobilization of photocatalyst particles on different macroscopic supports [50,51].In particular, in the case of CFX, it has been shown that the studied structured photocatalyst is able to degrade the target molecule under UV irradiation for treatment time similar to those observed in our system, but the mineralization degree similar to our paper was achieved for 420 min irradiation time [52].In addition, Miranda-Garcia et al., studying the degradation of the CAF and PC in a pilot compound parabolic collector (CPC) solar plant, observed that the photoactivity progressively worsened in five reuse cycles [51].On the contrary, in our work, the same batch of N-TiO 2 /PS catalyst was instead used for the different PPCPs removal tests, indicating a good stability of the structured photocatalyst.
compound used in pharmaceutical formulations as a stimulant and, moreover, is the most widely consumed psychoactive drug [49].Finally, PC, also known as acetaminophen, is the compound used in many analgesic and antipyretic medicines [49].For the photocatalytic tests on CFX, CAF, and PC, the packed bed photoreactor was used operating in batch mode.The initial concentration value for all PPCPs was mg/L.The obtained results are depicted in Figure 8.It was possible to observe a progressive decrease of the pollutants concentration during the irradiation time, evidencing that the photocatalytic system was effective in the removal of all the tested PPCPs.In particular, for CFX, CAF, and PC, similar degradation degree (about 99% for CFX, 95% for CAF, and 99% for PC) and TOC removal (about 85% for CFX, 80% for CAF, and 83% for PC) were achieved after 180 min of visible light irradiation (Figure 9).Considering the literature concerning the degradation of CFX, PC, and CAF, it must be underlined that the most studied photocatalysts are used in powder form [49,50], making their use on an industrial scale very difficult.However, some studies have also reported the immobilization of photocatalyst particles on different macroscopic supports [50,51].In particular, in the case of CFX, it has been shown that the studied structured photocatalyst is able to degrade the target molecule under UV irradiation for treatment time similar to those observed in our system, but the mineralization degree similar to our paper was achieved for 420 min irradiation time [52].In addition, Miranda-Garcia et al., studying the degradation of the CAF and PC in a pilot compound parabolic collector (CPC) solar plant, observed that the photoactivity progressively worsened in five reuse cycles [51].On the contrary, in our work, the same batch of N-TiO2/PS catalyst was instead used for the different PPCPs removal tests, indicating a good stability of the structured photocatalyst.

Preparation of the Structured Photocatalyst
The Nitrogen-doped TiO2 (N-TiO2) photocatalyst was obtained by direct nitration during the hydrolysis of titanium tetraisopropoxide with ammonia aqueous solutions and calcination in air at 450 °C for 30 min.The N/Ti molar ratio in the prepared sample was 18.6 and corresponded to an

Preparation of the Structured Photocatalyst
The Nitrogen-doped TiO 2 (N-TiO 2 ) photocatalyst was obtained by direct nitration during the hydrolysis of titanium tetraisopropoxide with ammonia aqueous solutions and calcination in air at 450 • C for 30 min.The N/Ti molar ratio in the prepared sample was 18.6 and corresponded to an optimized catalyst formulation found in our previous work [41].The N-TiO 2 catalyst has been already characterized evidencing that the main crystallographic phase is antase and the band-gap energy is equal to 2.5 eV [41,53].
The polymeric supports used for the deposition of N-TiO 2 particles were PS spheres (mean diameter: 12 mm) [33].The PS spheres were used as substrate for the deposition of N-TiO 2 powder using the solvent-casting method [54], according to the method reported in our previous work [32,33].In particular, a certain quantity of N-TiO 2 photocatalyst was added to pure acetone and the suspension vigorously mixed until a uniform dispersion was obtained.Subsequently, the PS spheres were added to the obtained suspension in order to immobilize N-TiO 2 particles on the PS surface (N-TiO 2 /PS).The spheres were then removed from the solution and dried at room temperature.In order to remove the N-TiO 2 particles that were not adhering to the PS spheres, several cycles in an ultrasonic bath (CEIA-CP104) were performed until a stable loading of the sample was reached.The amount of N-TiO 2 on the PS spheres was measured using precision balance (Mettler Toledo), and the final loading of N-TiO 2 in N-TiO 2 /PS sample was found to be 1.7 wt% [33].The structured N-TiO 2 /PS photocatalyst was characterized by spectroscopic and morphological techniques in our previous work, evidencing that the preparation method was able to uniformly disperse N-doped TiO 2 particles in anatase phase on PS spheres and that the structured catalyst was active under visible and solar light [32,33].No alteration of the N-TiO 2 photocatalytic activity was found after the supporting procedure.

Photocatalytic Tests with the Packed Bed Photoreactor at Semi-Pilot Scale
The experimental setup (Figures 1 and 5) consists of a packed bed photocatalytic reactor, an external stirred aerated tank containing the polluted solution, and a peristaltic pump to provide a continuous flow of water along the photocatalytic reactor.
The packed bed reactor included a cylindrical Pyrex tube to maximize the incident radiation intensity (40 cm long, 5 cm external diameter) and N-TiO 2 /PS structured photocatalyst (325 g) as packing material.The photocatalytic experiments were conducted in both continuous mode (Figure 5) and batch mode (Figure 1).For both working conditions, the liquid flow rate was varied in the range 6.3-74.15mL/min.In the case of batch mode, the total volume of the treated solution was 500 mL.The external body of the photoreactor was irradiated by an LEDs strip (produced by LED lighthouse; electrical power: 81.6 W) closely enwrapped around the external surface of the photoreactor (Figure 10).The emission spectrum of the LEDs lies in the range from 400 to 700 nm (Figure 10 insert).
The photocatalytic tests were carried out using MB at 10 mg/L initial concentration.Additional photocatalytic tests were carried out using aqueous solutions containing CAF, PC, and CFX at 5 mg/L initial concentration.The reactor was left in dark condition for 2 h before switching on visible LEDs.
All the photocatalytic tests were performed at the spontaneous pH of the polluted solutions, and the temperature was in the range 20-30 • C.
5) and batch mode (Figure 1).For both working conditions, the liquid flow rate was varied in the range 6.3-74.15mL/min.In the case of batch mode, the total volume of the treated solution was 500 mL.The external body of the photoreactor was irradiated by an LEDs strip (produced by LED lighthouse; electrical power: 81.6 W) closely enwrapped around the external surface of the photoreactor (Figure 10).The emission spectrum of the LEDs lies in the range from 400 to 700 nm (Figure 10 insert).The photocatalytic tests were carried out using MB at 10 mg/L initial concentration.Additional photocatalytic tests were carried out using aqueous solutions containing CAF, PC, and CFX at 5 mg/L initial concentration.The reactor was left in dark condition for 2 h before switching on visible LEDs.
All the photocatalytic tests were performed at the spontaneous pH of the polluted solutions, and the temperature was in the range 20-30°C.
The TOC contained in a fixed volume of the solution was measured by the high temperature combustion method on a catalyst (Pt-Al2O3) in a tubular flow micro-reactor operated at 680 °C.The solution was injected in the catalytic reactor fed with air to oxidize the organic carbon into CO2 , whose concentration in gas-phase was monitored by a continuous analyzer (Uras 14, ABB) [6].

Conclusions
In this work, a packed bed photoreactor at semi-pilot scale was developed, using a visible light active structured photocatalyst.The packed bed reactor was irradiated by visible LEDs for the evaluation of photocatalytic activity in the removal of MB dye, as well as CFX, PC, and CAF, which are considered pollutants of emerging concern.The TOC contained in a fixed volume of the solution was measured by the high temperature combustion method on a catalyst (Pt-Al 2 O 3 ) in a tubular flow micro-reactor operated at 680 • C. The solution was injected in the catalytic reactor fed with air to oxidize the organic carbon into CO 2 , whose concentration in gas-phase was monitored by a continuous analyzer (Uras 14, ABB) [6].

Conclusions
In this work, a packed bed photoreactor at semi-pilot scale was developed, using a visible light active structured photocatalyst.The packed bed reactor was irradiated by visible LEDs for the evaluation of photocatalytic activity in the removal of MB dye, as well as CFX, PC, and CAF, which are considered pollutants of emerging concern.
Photocatalytic tests driven in batch configuration showed that the N-TiO 2 /PS photocatalyst was active in the photocatalytic degradation of MB, reaching almost total MB degradation after 120 min of irradiation, while no significant decrease in MB concentration was observed during the overall irradiation time in the absence of the structured catalyst.No influence of liquid flow rate was observed confirming that the overall MB photocatalytic degradation rate is mainly controlled by kinetics.A mathematical model for the photocatalytic degradation of MB was developed neglecting the external mass transfer phenomena.In order to verify this last hypothesis, the apparent kinetic constant was estimated from the experimental data collected, using different liquid flow rates when the photoreactor operated in continuous mode and at steady-state condition.The accuracy of the obtained value for the kinetic constant was verified by performing another series of experiments in which the visible light intensity varied.Finally, the efficiency of packed bed photoreactor at semi-pilot scale was also evaluated in the removal of CFX, CAF, and PC.The obtained results showed that the almost total degradation of the pollutants with a TOC removal more than 80% was achieved after 180 min of visible light irradiation.As concluding remarks, the obtained results are an encouraging step forward in the possible application of photocatalytic technology based on the use of visible LEDs and visible light structured photocatalysts in the elimination of hazard environmental pollutants, especially for the degradation of emerging bio-recalcitrant organic pollutants.

Figure 1 .
Figure 1.Schematic picture of the packed bed photoreactor working in batch mode.Figure 1.Schematic picture of the packed bed photoreactor working in batch mode.

Figure 1 .
Figure 1.Schematic picture of the packed bed photoreactor working in batch mode.Figure 1.Schematic picture of the packed bed photoreactor working in batch mode.

Figure 3 .
Figure 3. Behavior of MB relative concentration as function of irradiation time at different liquid flow rate.Light sources: visible LEDs; light intensity: 78 mW/cm 2 ; initial MB concentration: 7 mg/L; photoreactor configuration: batch mode; and total organic carbon (TOC).

Figure 5 .
Figure 5. Schematic picture of the packed bed photoreactor working in in continuous mode.

Figure 5 .
Figure 5. Schematic picture of the packed bed photoreactor working in in continuous mode.

Figure 6 .
Figure 6.Evaluation of kinetic constant using photoreactor in continuous mode PFR (Plug Flow Reactor) at different liquid flow rate (in the range 6.3-74.15mL/min).Light sources: visible LEDs; light intensity: 78mW/cm 2 ; and initial MB concentration: 7 mg/L.

Figure 7 .
Figure 7.Comparison between experimental data and model calculation at different light intensity.Light sources: visible LEDs; initial MB concentration: 7 mg/L; liquid flow rate: 50 mL/min; and photoreactor configuration: batch mode.

Figure 10 .
Figure 10.Schematic picture of the packed bed photoreactor configuration and the emission spectrum of the visible light LEDs strip (insert).

Figure 10 .
Figure 10.Schematic picture of the packed bed photoreactor configuration and the emission spectrum of the visible light LEDs strip (insert).

Table 1 .
Comparison with studies reported in the literature dealing with a different fixed bed photoreactor for methylene blue removal irradiated by artificial light sources, as well as with solar photocatalytic reactors using structured photocatalysts; PET (Polyethylene Terephthalate).
. Figure3.Behavior of MB relative concentration as function of irradiation time at different liquid flow rate.Light sources: visible LEDs; light intensity: 78 mW/cm 2 ; initial MB concentration: 7 mg/L; photoreactor configuration: batch mode; and total organic carbon (TOC).

Table 2 .
MB conversion at steady-state condition using different liquid flow rate when the photoreactor operated in continuous mode.Light sources: visible LEDs; light intensity: 78 mW/cm 2 ; and initial MB concentration: 7 mg/L.

Table 2 .
MB conversion at steady-state condition using different liquid flow rate when the photoreactor operated in continuous mode.Light sources: visible LEDs; light intensity: 78 mW/cm 2 ; and initial MB concentration: 7 mg/L.