1 TiO2/Cu(II) photocatalytic production of benzaldehyde from benzyl 1 alcohol in solar pilot plant reactor 2 Danilo Spasiano a* , Lucia del Pilar Prieto Rodriguez b , Jaime Carbajo Olleros c , Sixto Malato b , 3 Raffaele Marotta a , Roberto Andreozzi a 4 5 a Dipartimento di Ingegneria Chimica, dei Materiali e della Produzione Industriale, Università 6 di Napoli “Federico II”, p.le V. Tecchio, 80 – 80125 – Naples, Italy. 7 b Plataforma Solar de Almería-CIEMAT, Carretera de Senés Km 4 - 04200 - Tabernas, Almería, 8 Spain. 9 c Departamento de Ingeniería de Procesos Catalíticos Instituto de Catálisis y Petroleoquímica, CSIC, 10 C/ Marie Curie 2 - 28049 - Cantoblanco, Madrid, Spain. 11 12 * Corresponding author. Tel +390817682968 fax +390815936936. E-mail danilo.spasiano@unina.it 13 14 ABSTRACT 15 The technical feasibility of selective photocatalytic oxidation of benzyl alcohol to benzaldehyde, in 16 aqueous solutions, in presence of cupric ions has been investigated in a solar pilot plant with 17 Compound Parabolic Collectors. Aldrich (pure anatase) and P25 Degussa TiO2 have been used as 18 photocatalysts. The influences of cupric species concentrations, solar irradiance and temperature are 19 discussed too. The oxidation rates were strongly influenced by the initial cupric ions concentration, 20 incident solar irradiance and temperatures. 21 The best results found, in terms of yields and selectivities to benzaldehyde under acidic conditions 22 were higher than 50% and 60% respectively under acidic conditions. 23 manuscript revised.doc Click here to view linked References http://ees.elsevier.com/apcatb/viewRCResults.aspx?pdf=1&docID=10331&rev=1&fileID=327008&msid={0C3BACAB-6896-4BEA-BAE0-870D59CE1610} 2 Under deaerated conditions, the presence of reduced copper species was proved by XPS analysis. 1 The results indicated that, at the end of the process, cupric species can be easily regenerated and 2 reused, through a re-oxidation of reduced copper, produced during the photolytic run, with air or 3 oxygen in dark conditions. 4 A figure-of-merit (ACM), proposed by the International Union of Pure and Applied Chemistry 5 (IUPAC) and based on the collector area, has been estimated, under the proposed conditions, with 6 the aim to provide a direct link to the solar-energy efficiency independently of the nature of the 7 system. Generally speaking, it can be considered that the lower ACM values the higher the system 8 efficiency. 9 10 Keywords: selective oxidation, benzyl alcohol, benzaldehyde production, TiO2 photocatalysis, solar 11 photocatalytic plant, figure-of-merit. 12 13 1. Introduction 14 The use of TiO2, as readily available and environmentally friendly photocatalyst, was largely 15 investigated for the removal of non-biodegradable or undesiderable organic substances from 16 wastewater [1-3] and TiO2 based on solar technologies were developed during the past years for 17 reducing the cost of large-scale aqueous-phase applications to treat industrial wastewater [4,5]. 18 However, only in recent years the research has pointed its attention on the possibility to use the 19 TiO2 as a photocatalyst for the selective oxidation of organic molecules under UV radiation in non 20 aqueous media [6]. For example, the selective oxidation of aromatic alcohols in the respective 21 aldehydes can be easily gained in acetonitrile or in solvent free systems at room temperature [7,8] 22 due to the reaction of the alcoholic substrate with the photogenerated positive holes (  vbh ) and using 23 oxygen as acceptor of photoelectrons [9]. In particular, the addition of an organic solvent, as 24 acetonitrile, allows the oxidation of benzyl alcohol and its derivates with selectivities to the 25 3 respective aldehydes of over 90% [8,10], values well higher than those reported (10-60%) when the 1 same process is carried out in aqueous solutions using TiO2 nanoparticles [11,12]. 2 Moreover, selectivities of 41-74% were reported for the oxidation of 4-substituted aromatic 3 alcohols to the corresponding aldehydes, using aqueous media, over rutile or anatase TiO2 catalysts 4 [11,13]. The decrease of the selectivity in aqueous solution with respect the use of organic non 5 aqueous solvent is mainly due to [9]: 6 a) formation of very reactive species such as hydroxyl radicals (HO • ) through the reaction of 7 adsorbed water molecules or hydroxyl groups with positive holes: 8    HHOHOOH VBh adsads )(/2 (r1) b) generation of radical superoxide (O2 •- ) by the reaction of molecular oxygen with photogenerated 9 electrons (r2). In presence of water, O2 •- species gives rise to hydrogen peroxide (r3) whose 10 electron trapping generates HO • radicals (r4), which oxidizes not selectively the organic substances 11 (r5): 12 ads e ads OO CB   22 )( (r2) ads e ads OHHO CB 222 )(2    (r3)    HOHOOH CBe ads )( 22 (r4) oxSSHO  (r5) On the other side, water is considered the best solvent for environmental friendly process and, for 13 this reason, other new approaches have arisen in order to avoid the use the organic solvents. In 14 particular, very recent studies demonstrated that the use of Cu(II) ions, as electron acceptor, instead 15 of molecular oxygen, under deaerated conditions, leads to an increase of selectivity to benzaldehyde 16 up to 72%, starting from benzyl alcohol [14], since the formation of hydroxyl radicals takes place 17 just by the reaction of water molecules or surface adsorbed hydroxyl groups with positive holes. In 18 4 this case, cupric ions reduce to a lower state of oxidation by capturing the photogenerated electron 1 on the TiO2 surface and precipitate from the solution: 2 )0()()( )()( CuICuIICu CBCB ee   (r6) whereas adsorbed benzyl alcohol (BA) is oxidized to the benzaldehyde (BHA), through a direct 3 reaction with the positive holes: 4 ads )( ads BHA 2 BA    VBh (r7) At the end of the process, cupric ions could be regenerated and reused, through the reoxidation of 5 zero valent precipitated copper with an air flow at dark conditions. 6 Chemical state of solid copper after the photocatalytic run was also intensively investigated, but 7 controversial results were reported. Most of the researchers reported that the solid is composed by a 8 mixture of zero valent copper and cupric oxide (and, in some case, cuprous oxide) [15-19]. 9 In some studies it was concluded that the reduced Cu species was zero valent copper [20-21]. 10 However, the possibility of the presence of both Cu(I)/Cu(II) species, due to a reoxidation of metal 11 copper during the preparation of the analytical samples, was not completely ruled out [22]. 12 At best of the Authors’ knowledge, only lab-scale lamp-driven investigations on the selective TiO2 13 photocatalytic oxidation of benzyl alcohol and its derivatives, in aqueous or acetonitrile or solvent 14 free systems, were reported in the literature [11, 12, 14, 23-27]. 15 It could be very interesting, for economic and environmental reasons, to evaluate the possibility of 16 exploiting the use of solar radiations daily arriving on the earth surface, through the use of pilot 17 plants, for the production of benzaldehyde from benzyl alcohol. 18 In the present work, the development of the TiO2/Cu(II)/solar radiation system for the selective 19 oxidation of benzyl alcohol to benzaldehyde in water is studied, by using a solar photocatalytic pilot 20 plant at different operating conditions (TiO2 photocatalyst type, Cu(II) initial concentration and 21 irradiance of solar radiation). The possibility of reuse copper as catalyst is also tested by oxidizing 22 the precipitated zero valent copper with an air flow blown into the pilot-plant in dark conditions. 23 5 Optimal conditions of TiO2 load and pH values are fixed according to the ones found in previous 1 studies [14,28]. 2 Moreover, analyses of solid, after the oxidation runs, are carried out to better clarify the nature of 3 copper species at the end of the process. 4 5 2. Material and methods 6 2.1 Experimental set-up and procedures 7 2.1.1 Pilot-Plant experiments 8 Experimental investigations were carried out, from May to July 2012, in a solar pilot-plant 9 consisting of twelve Compound Parabolic Collectors (CPC), installed in the Plataforma Solar de 10 Almería (37° latitude N, Spain) and elsewhere described in its geometrical and functional 11 characteristics [29]. The total solution volume is 39 l, where only 22 l are exposed to solar 12 radiation; the rest is distributed between the recirculation tank (9 l) and the hydraulic connections (8 13 l). The recirculation tank of the adopted CPC was modified, as reported in figure 1, in order to 14 ensure the absence of oxygen, essential for performing the proposed process. For this purpose, the 15 solutions, containing the catalysts and the substrate, were preventively purged with nitrogen 16 gaseous stream through two porous ceramic spargers, by closing the valve Vn, to remove the 17 dissolved oxygen. During the experimental runs a continuous nitrogen flow was guaranteed to the 18 reactor to prevent the entry of air in the reactor but it was switched in the freeboard of the 19 recirculation tank, by opening the valve Vn, to inhibit the stripping of organic substances from the 20 solution. In the last case, the bubbling of nitrogen is strongly limited by the pressure drop due to the 21 spargers. 22 To better follow the concentration profiles of the compounds involved in the process, some 23 experimental runs were carried out during two days. In these cases, at the end of the first day, the 24 irradiated part of the reactor was covered and the recycling tank was capped to stop the 25 photochemical reactions and to prevent the entry of oxygen respectively. 26 6 Samples were collected in a glass bottle at different reaction times, by opening the valve Vs, rapidly 1 filtered to prevent the reoxidation of precipitate copper and finally analyzed. 2 All the experimental data were reported in function of the accumulated energy, per unit of volume 3 (kJ/l), incident on the reactor at the corresponding time of the withdrawn sample (Qn) [30]: 4 11 ;   nnn t r nnnn ttt V A UVtQQ (eq. 1) where Qn-1 represent the accumulated energy (per unit of volume, kJ/l) taken during the experiment 5 relative to the previous sampling; ∆tn, nUV and Vt are, respectively, the elapsed time, the average 6 UV-irradiance (measured by a global UV radiometer KIPP&ZONEN, model CUV 3 mounted on a 7 platform tilted the same angle as the CPCs) which reaches the collector surface (Ar) between the 8 two samplings and the total solution volume. 9 10 2.1.2 Laboratory-scale experiments 11 Experiments were carried out in a Suntest solar simulator (Suntest XLS+ photoreactor, Atlas) 12 equipped with a 765-250 W/m 2 Xenon lamp (61-24 W/m 2 from 300 nm to 400 nm, 1.4∙10 20 -13 5.5∙10 19 photons/m 2 ∙s) and a cooler to keep the temperature at 35°C. The UV irradiation, in the 14 range of 300-400 nm, was monitored by using a portable radiometer (Solar Light CO PMA 2100) 15 fixed on the shaker inside the lamp influence zone, like shown in figure 2. 16 In this case, a volume of 700 ml of reacting solution was prepared in a one liter flask and, after 30 17 minutes of stripping with nitrogen, was rapidly poured in ten cylindrical glass vials with the 18 capacity of 42 ml and a diameter of 25 mm. The vials were rapidly closed with a screw cap, to 19 prevent the entry of oxygen, and were simultaneously exposed to the lamp radiation and agitated by 20 a shaker like shown in figure 2. The temperature was set at 35 °C. Each vial represented a sample to 21 remove from the solar box at different reaction times. Once collected, all the samples were rapidly 22 filtered and analyzed. 23 7 The pH of the solutions was adjusted at the value of 2.0, for all the runs, by using an aqueous 1 solution of 85% of phosphoric acid. 2 2.2 Analytical methods 3 The concentrations of benzyl alcohol, benzaldehyde and benzoic acid at different reaction times 4 were evaluated by HPLC analysis. For this purpose, the HPLC apparatus (Agilent 1100) was 5 equipped with a diode array UV/Vis detector (λ= 215, 230, 250 nm) and Phenomenex (Gemini 5u 6 C18 150x3 mm) column, using a mobile phase flowing at 0.7 ml min −1 . The mobile phase was 7 prepared by a buffer solution (A), H2O (B) and CH3CN (C). A linear gradient progressed from 15% 8 C to 28% C and from 45% B to 32% B in 10 minutes with a subsequent re-equilibrium time of 3 9 minutes. One liter of buffer was made by 10 ml of phosphoric acid solution (5.05 M), 50 ml of 10 methyl alcohol and water for HPLC. 11 The concentration of dissolved copper ions was measured by means of a colorimetric method using 12 an analytical kit (based on oxalic acid bis-cyclohexylidene hydrazide, cuprizone®) purchased from 13 Macherey-Nagel. An UV/Vis spectrometer (UNICAM-II spectrophotometer) has been used for the 14 measurements at a wavelength of 585 nm. 15 Total organic carbon (TOC) was monitored by Shimadzu Total Organic Carbon analyser model 16 TOC-5050A, equipped with an auto sampler ASI 5000A. The pH was monitored by a portable pH-17 meter (Crison pH 25). 18 The Cu/Ti ratio for unknown solids, withdrawn at the end of some photocatalytic runs, was 19 estimated by an Energy Dispersive X-ray spectrometer system (SwiftED, Oxford Instruments) 20 attached to a Scanning Electron Microscope (TM-1000, Hitachi). 21 Powder X-ray diffraction (XRD) patterns were estimated using a X´PertPRO (PANalytical) 22 diffractometer with nickel-filtered Cu Kα radiation. The X-ray generator was operated at 45 kV and 23 40 mA. The powders were scanned from 2θ = 4º to 90º with a 0.02 step size and accumulating a 24 total of 5 s per point. 25 8 X-ray photoelectron spectroscopy (XPS) analysis was carried out under high vacuum chamber with 1 a base pressure below 9x10 -7 Pa at room temperature. Photoemission spectra were recorded using a 2 SPECS Gmbh system equipped with an UHV PHOIBOS 150 analyser with Al monochromatic 3 anode operated at 12 kV and 200 W with a photon energy of h = 1486.74 eV. A pass energy of 25 4 eV was used for high-resolution scans. Binding energies (BE) were referenced to C1s peak (284.6 5 eV) to take into account charging effects. 6 The XPS spectra obtained were then curved fitted using the XPS PeakFit software. The areas of the 7 peaks were computed by fitting the experimental spectra to Gaussian/Lorentzian curves after 8 subtracting the background (Shirley function). 9 10 2.3 Materials 11 Two commercial microcrystalline TiO2 powders were tested: TiO2 Degussa P25 (80% anatase, 20% 12 rutile, BET specific surface area 50 m 2 g −1 ) and TiO2 Aldrich (pure anatase phase, BET specific 13 surface area 9.5 m 2 g −1 ). 14 Cupric ions were introduced in the system as anhydrous cupric sulphate (Sigma-Aldrich) with a 15 purity >99.0% (w/w). Benzyl alcohol (BA), benzaldehyde (BHA) and benzoic acid (BAC), with a 16 purity >99.0% (w/w), were purchased from Sigma Aldrich and used as received. Phosphoric acid 17 with a purity 85% from Merck as used as received 18 19 3. Result and discussion 20 3.1 Effect of TiO2 type 21 The results obtained during different experimental runs of solar photo-oxidation of benzyl alcohol, 22 with two different typologies of commercial TiO2 samples, at the same load (200 mg/l), are shown 23 in figures 3a-b. 24 The runs were carried out in two days. At the end of first day, when the reactor was covered, the 25 UV-irradiances (wavelength 300-400 nm) and temperatures decreased (see figs 3a,b). During the 26 9 runs, measured UV-irradiances ranged between 40 and 50 W/m 2 approximately (fig. 3b, 1 continuous and dashed lines). 2 As shown in the diagrams, the reactivity of the system is higher when TiO2 P25 by Degussa is used 3 than the TiO2 Aldrich catalyst. In particular, in presence of P25 Degussa sample (empty symbols), 4 for a Qn value of 123 kJ/l, the BA concentration approached to zero (figure 3a) and Cu(II) ions were 5 totally reduced (figure 3b), whereas, if Aldrich TiO2 sample is used (full symbols), about 27% of 6 initial benzyl alcohol and Cu(II) ions were still present in the solution. 7 The higher reactivity observed for the system in presence of P25 Degussa TiO2 is in disagreement 8 with the results, previously reported [14], that show a similar reactivity when P25 Degussa and 9 Aldrich TiO2 catalysts were used in presence of an UV radiation emitted by a laboratory 10 thermostated lamp. 11 This discrepancy may be probably due to the different emission spectra between lamp and the 12 sun and (different) light absorption characteristics, in the range of 300-800 nm, of the two 13 used catalysts. 14 Moreover, the different reactivities, recorded in the solar experiments, could be also 15 attributed to the differences between the averages of the measured temperatures: 38.6 °C and 16 34.3 °C for runs carried out in presence of Degussa P25 (dotted lines) and Aldrich TiO2 17 (dashed and dotted lines) catalysts respectively (fig 3a). 18 As shown in figure 4, the BA solar photocatalytic oxidation resulted in the production of BHA 19 (diamonds) for Aldrich (full symbols) and P25 Degussa (empty symbols) TiO2 catalysts, and 20 BAC an undesired product that derives from the reaction between the positive photogenerated holes 21 on the TiO2 surface and benzaldehyde. BAC yields are also reported in the same figure (triangles). 22 According to the previous results, the highest BHA production rates were gained in presence of P25 23 Degussa TiO2. Moreover, for this catalyst, the maximum BAC production rate was reached at 24 the highest yields of BHA. 25 10 However, for accumulated energy values higher than 80 kJ/l, when the unconverted BA 1 concentration was less than 20% of its initial concentration (fig. 3a, empty squares), a decrease 2 in BHA yield with respect of initial BA amount, from the value of 53.3% to 45.5%, was observed 3 using P25 Degussa TiO2 samples. This result can be explained by considering a competition in the 4 reaction between the BA and BHA molecules for the photogenerated positive holes on the TiO2: 5 BAC 2 BHA 2 BA )( ads )( ads     VBVB hh (r8) It is interesting to remark that for accumulated energy values higher than 120 kJ/l, in the case 6 of Degussa TiO2 catalyst, when Cu(I)/Cu(II) species are totally reduced (fig. 3b, empty 7 squares), being the predominant reaction the recombination between photoinduced positive 8 holes and electrons: 9 heateh CBVB   )()( (r9) the BA (fig. 3a, empty squares) and BHA (fig. 4, empty diamonds) consumption and BAC 10 production rates (fig. 4, empty triangles) stopped. 11 The BHA selectivity, with respect to BA consumption, was calculated for both TiO2 types (fig. 5). 12 As shown in the diagram, it seems that the use of Aldrich TiO2 sample renders the system more 13 selective, reaching, for Qn values of 130 kJ/l, BHA selectivity values close to 70% (in presence of 14 P25 Degussa, only a value of 50% was obtained). 15 The highest selectivity achieved in presence of Aldrich TiO2 sample is correlated at the lower 16 degree of conversion (fig. 3a, full squares) when compared with the test in which Degussa P25 17 is used (fig. 3a, empty squares) 18 TOC measurements collected during the runs are also reported in figure 5, in terms of 19 mineralization degrees. Due to the HO radicals generation from the reaction of water molecules 20 or surface adsorbed hydroxyl groups with positive holes [37]: 21 22 11     HHOOH ads2 VBh (r10)     HOHO VBads h (r11) the two trends are very similar, with a degree of mineralization at the end of the experimental runs 1 of 9% and 7% in presence of P25 Degussa and Aldrich TiO2 respectively, 2 However, the low degree of mineralization achieved demonstrates the possibility of using such 3 system to carry out selective oxidations in aqueous media. 4 5 3.2 Effect of cupric ions concentration 6 To evaluate the effect of Cu(II) initial concentration, some experimental runs of solar photoxidation 7 of benzyl alcohol were carried out with Aldrich TiO2 at the load of 200 mg/l and pH = 2.0, varying 8 the cupric ions starting concentration (0.5 mM, 1.0 mM and 1.5 mM). 9 As shown in figures 6a-b, the temperatures and UV-irradiances profiles (solid, dashed and dotted 10 lines) are so similar as to be considered equals for the three runs. 11 With Cu(II) and BA starting concentrations of 0.5 mM and 1.5 mM respectively, the BA oxidation 12 stopped for Qn values close to 35 kJ/l where a complete reduction of cupric ions was observed (full 13 triangles). 14 At the highest Cu(II) initial concentration (1.5 mM), it resulted into a decrease of the system 15 reactivity and BA conversion (fig. 6a, full circles) and, at the same time, the Cu(II) reduction rates 16 decrease. The observed results, according to those previously reported [14], are probably due to a 17 partial catalyst deactivation by the adsorbed sulphate ions which may block the TiO2 active sites 18 (s*) [31]: 19 *s+ 4OS (ads) -2 SO4 -2 (r12) In fact, since Cu(II) species were added to the reactive solution as cupric sulphate (CuSO4), 20 increasing the Cu(II) ions initial concentration results into an increase of sulphate concentration too. 21 12 Moreover, the sulphate anions, at concentration level higher than 1.0 mM, exerted a negative effect, 1 inhibiting the BA photo-oxidation rates because they are in competition with BA molecules in the 2 reaction with the positive holes [14]: 3   4 2 4 SOSO VBh (r13) In figure 7 are reported the experimental concentration profiles of BHA and BAC (full and empty 4 symbols) that are in agreement with BA concentration trends shown in figure 6a. In particular the 5 best result found, in term of yield, was of 43% for BHA, starting with [Cu(II)]o = 1.0 mM, for a 6 accumulated energy value of 67 kJ/l. For all the runs, the selectivity was always higher than 67% 7 and the mineralization degrees were lower than 4.5% (data not shown). 8 9 3.3 Effect of the irradiance and temperature 10 As previously shown in the figures 3b and 6b, the Cu(II) concentrations are characterized by a s-11 shaped profile and, in particular, when the solar UV radiations and reactor temperatures reached the 12 top, a marked increase of Cu(II) reduction rate was observed. With the aim to better assess the 13 relationship between the variability of both irradiances and temperatures during a solar run and the 14 s-shaped of cupric ions concentration profile, a set of three laboratory photolytic experiments, with 15 Cu(II) initial concentration equal to 0.5 mM, were carried out at three different irradiances, kept 16 constant during a single run and under controlled solution temperature (T=35 °C). For this purpose, 17 a solar box apparatus, described in the experimental section, was used. Since the internal diameters 18 of CPC solar reactor tubes and the solar box vials are 31 mm and 24 mm respectively, the TiO2 load 19 that maximizes the adsorption of UV radiation emitted by solar lamp is higher than that used for 20 solar experiments carried out in CPC reactor. 21 The optimum TiO2 concentration (ccat), for which the optical thickness equals that of CPC reactor 22 configuration (=9.12) can be easily calculated as suggested by others [28,32]: 23 13   mg/l258   dk ccat   eq. 2 where  is the scattering coefficient (1.29510 3 m 2 /kg), k is the catalyst specific mass absorption 1 (1.7510 2 m 2 /kg) and d the internal tube diameter (24 mm). 2 As shown in figure 8, by increasing the UV irradiance from 39.5 W/m 2 to 59.7 W/m 2 , the Cu(II) 3 photoreduction rate increases. 4 A parallel increase of BA oxidation rate was also observed (data not shown). The results 5 collected during these runs indicate that, under controlled temperature and irradiance, no S-6 shaped concentration profile was recorded. 7 8 3.4 Figure-of-merit 9 To estimate the operating costs of sole natural radiation, the figure-of-merit concept was used [33]. 10 For solar-energy-driven systems, the figure-of-merit allows the assessment of the solar technology 11 efficiency used for the investigated process. In fact, even if there is no cost for solar radiation, there 12 could be a non-marginal capital cost for the collector. Being the capital cost of a solar collector 13 generally proportional to its area, a figure-of-merit, based on the solar collector area, is appropriate. 14 For the adopted experimental batch conditions, the appropriate figure-of-merit is the collector area 15 per mass (ACM), defined as the collector area required to reduce of a unit mass of the substrate in the 16 reacting system in a time of 1 hour (to) for an incident solar irradiance of 1000 W m −2 ( o SE ): 17  fi o Sot Sr CM ccEtVM EtA1000 A    eq. 3 where Ar (3.19 m 2 ) is the real collector area, M is the molar mass of the substrate (108.14 g/mol), Vt 18 (39 l) is the volume of treated solution, SE (814.6 W/m 2 ) is the average direct solar irradiance over 19 the reaction time t (4.83 h), ci and cf are respectively the initial and final substrate concentrations 20 14 (mol/l), o SE is the standardized irradiance (1000 W/m 2 , based on the AM1.5 standard solar spectrum 1 on a horizontal surface) [34] and to is the reference time (1 h). 2 On the basis of the previous formula an average value of ACM = 3.08∙10 3 m 2 per kilogram of benzyl 3 alcohol and per hour consumed was thus calculated for the collector area per mass (200 mg/l of 4 Aldrich TiO2, pH=2.0, [Cu(II)]0 = 1.0 mM, [BA]0 = 1.5 mM). 5 6 3.5 Copper reuse 7 The possibility of reuse the reduced copper, as catalyst, was tested by carrying out an experimental 8 run, by using the CPC solar reactor, consisting in three cycles of BA photo-oxidation (fig. 9). When 9 all Cu(II) species was totally reduced as precipitate copper, its reoxidation was carried out, in dark 10 conditions, under air bubbling in the recirculation tank for 30 minutes. At the end of the first cycle, 11 the reactive solution was re-purged with nitrogen gas, for other 30 minutes, to remove the dissolved 12 oxygen and a new photocatalytic cycle was started. The experimental results, reported in figure 9, 13 pointed out that it’s possible, for each cycle, to reoxidize completely the precipitate copper to cupric 14 species (empty diamonds). 15 During the first two cycles of BA solar photoxidation, not particular changes were observed on the 16 reactivity of the system, whereas during the third cycle a decrease of both BA consumption and 17 BHA production rates were observed. This behaviour could be explained by considering that the 18 solution composition changes during the experimental run. In particular, at the 3 rd cycle beginning, 19 BA conversion and BHA yield were 43% and 33% respectively thus favouring a competition 20 kinetics between BA and BHA, both adsorbed on TiO2 surface, towards the reaction with the 21 photogenerated holes [35]: 22 ads sMk ads BHAhBA    /108.3 5 2 ads sMk ads BAChBHA    /100.3 4 2 15 The BHA selectivities of the process for the three photocatalytic cycles are reported in figure 10. As 1 shown in the diagram, the highest selectivities (up to 100%) were obtained in the first cycle, 2 whereas lower selectivities were observed during the second (close to 75%) and the third cycle 3 (close to 40%), thus supporting, with the progress of the reaction time, the intervention of undesired 4 oxidation reactions of BHA once produced. 5 After each photocatalytic cycle under anoxic conditions, the solid was filtered, stored under inert 6 atmosphere and submitted to EDS, XRD and XPS analysis in order to better investigate the 7 distribution of copper deposited on the solid and its oxidation states. 8 From the EDS investigations, Ti/Cu atom ratios, for the filtered solids, resulted 97.8/2.2, 96.5/3.5 9 and 85.2/14.8 (w/w) for the samples withdrawn at the end of the first, second and third cycle 10 respectively. These results supported the idea that reduced Cu species, formed during the photo-11 oxidative runs, accumulated on TiO2 surface increasing the amount of deposited copper from the 12 first cycle to third one up to 17.4% by weight. The increase of Cu amount on the TiO2 is probably 13 due to the deposition of a part of photocatalyst powders, during the experimental run, in the 14 recirculation tank and/or along the hydraulic connections of the plant, thus decreasing the active 15 load of it available in the solar photoreactor with a consequent reduction of BA consumption and 16 BHA production rates, really observed, in particular during the third cycle. 17 A typical XPS spectra for the solid samples shows different peaks (fig. 11). The peaks at 932.8 eV 18 and 952.6 eV indicate the predominant presence of copper reduced species (+1/0) as previously 19 reported by others [19,22,36] whereas the existence of peaks at 935.4 eV and 955.2 eV can be 20 attributed to the presence of cupric species [22], such as CuO, probably an artifact produced during 21 the preparation of the sample before the XPS analysis. 22 Unfortunately, XRD analysis did not give any result being the amount of Cu reduced species 23 accumulated on the TiO2 surface (max 17.4%) below XRD detection limit. 24 25 26 16 4. Conclusion 1 The possibility to convert benzyl alcohol to benzaldehyde by photocatalytic oxidation was 2 demonstrated in aqueous solution under natural solar radiation at pilot plant scale. The oxidation 3 rates were strongly influenced by the initial cupric ions concentration, incident solar irradiance and 4 temperatures. The best result found, in terms of yield, was of 53.3% for benzaldehyde with respect 5 to the initial benzyl alcohol concentration (63.4 % of selectivity) for an accumulated energy value 6 (Qn) of 78.9 kJ/l (reaction time of 385 min) and operating with an average temperature of 38.6 °C. 7 EPS investigations, carried out on the solids, withdrawn during different photocatalytic cycles, 8 confirmed the existence of both Cu reduced (0/+1) and oxidized species, the latter probably 9 produced during the sample preparation before the analysis. 10 The results also indicated that cupric species can be easily regenerated and reused with air or 11 oxygen in dark conditions. 12 A figure-of-merit (ACM) was calculated be equal to 3.08∙10 3 m 2 per kilogram and hour of benzyl 13 alcohol converted. 14 15 References 16 [1] M.N. Chong, B. Jin, C.W.K. Chow, C. Saint, Water Research 44 (2010) 2997 – 3027. 17 [2] K. Rajeshwara, M.E. Osugi, W. Chanmanee, C.R. Chenthamarakshan, M.V.B Zanoni, P. 18 Kajitvichyanukul, R. Krishnan-Ayer, Journal of Photochemistry and Photobiology C: 19 Photochemistry Reviews 9 (2008) 171–192. 20 [3] J.M. Herrmann, Applied Catalysis B-Environmental 99(3-4) (2010) 461-468. 21 [4] R.J. Braham, A.T. Harris, Industrial & Engineering Chemistry Research, 48 (2009) 8890–8905. 22 [5] S.R. Malato, P. Fernandez-Ibanez, M.I. Maldonado, G.J. Blanco, W. Gernjak, Catalysis Today 23 147 (2009) 1–59. 24 [6] Y. Shiraishi, T. Hirai, Journal of Photochemistry and Photobiology C: Photochemistry Reviews 25 9 (2008) 157–170. 26 17 [7] S. Higashimoto, N. Suetsugu, M. Azuma, H. Ohue, Y. Sakata, Journal of Catalysis 274 (2010) 1 76–83. 2 [8] W. Feng, G. Wu, L. Li, N. Guan, Green Chemistry 13 (2011) 3265–3272. 3 [9] C.S. Turchi, D.F. Ollis, Journal of Catalysis, 122 (1990) 178–192. 4 [10] S. Higashimoto, N. Kitano, N. Yoshida, T. Sakura, M. Azuma, H. Ohue, Y. Sakata, Journal of 5 Catalysis 266 (2009) 279–285. 6 [11] G. Palmisano, S. Yurdakal, V. Augugliaro, V. Loddo, L. Palmisano, Adv. Synth. Catal. 349 7 (2007) 964-970. 8 [12] S. Yurdakal, G. Palmisano, V. Loddo, V. Augugliaro, L. Palmisano, Journal of American 9 Chemical Society 130 (2008) 1568-1569. 10 [13] V. Augugliaro, V. Loddo, M.J. Lopez-Munoz, C. Marquez-Alvarez, G. Palmisano, L. 11 Palmisano, S. Yurdakal, Photochemical & Photobiological Sciences 8 (2009) 663–669. 12 [14] R. Marotta, I. Di Somma, D. Spasiano, R. Andreozzi, V. Caprio, Chemical Engineering 13 Journal 172 (2011) 243–249. 14 [15] J.W.M. Jacobs, F.W.H. Kampers, J.M.G Rikken, C.W.T. Bulle-Lieuwma, D.C. Doningsberger, 15 Journal of The Electrochemical Society 136 (1989) 2914-2023. 16 [16] M. Bideau, B. Claudel, L. Faure, M. Rachimoellah, Chemical Engineering Communications 93 17 (1990) 167–179. 18 [17] S. Morishita, Chemistry Letters 10 (1992) 1979-1982. 19 [18] N.S. Foster, R.D. Noble, C.A. Koval, Environmental Science & Technology 27 (1993) 350–20 356. 21 [19] S.W. Zou, C.W. How, J.P. Chen, Industrial & Engineering Chemistry Research 46(20) (2007) 22 6566–6571. 23 [20] M. Canterino, I. Di Somma, R. Marotta, R. Andreozzi, . Water Res., 42 (2008) 4498 -4506. 24 [21] H. Reiche, W.W. Dunn, A.J. Bard, Journal of Physical Chemistry 83 (1979) 2248-2251. 25 18 [22] S. Xu, J. Ng, A.J. Du, J. Liu, D.D. Sun, International Journal of Hydrogen Energy 36 (2011) 1 6538-6545. 2 [23] D.I. Enache, J.K. Edwards, P. Landon, B. Solsona-Espriu, A.F. Carley, A.A. Herzing, M. 3 Watanabe, C.J. Kiely, D.W. Knight, G.J. Hutchings, Science 311 (2006) 362–365. 4 [24] S. Yurdakal, G. Palmisano, V. Loddo, O. Alagoz, V. Augugliaro, L. Palmisano, Green Chem. 5 11 (2009) 510–516. 6 [25] V. Augugliaro, T. Caronna, V. Loddo, G. Marcì, G. Palmisano, L. Palmisano, S. Yurdakal, 7 Chemistry - A European Journal. 14 (2008) 4640–4646. 8 [26] S. Higashimoto, K. Okada, T. Morisugi, M. Azuma, H. Ohue, T.H. Kim, M. Matsuoka, M. 9 Anpo, Topics in Catalysis 53 (2010) 578–583. 10 [27] C.J. Li, G.R. Xu, B. Zhang, J.R. Gong, Applied Catalysis B: Environmental 115–116 (2012) 11 201–208. 12 [28] L. Prieto-Rodriguez, S. Miralles-Cuevas, I. Oller, A. Agüera, G. Li Puma, S. Malato, Journal 13 of Hazardous Materials 211– 212 (2012) 131– 137. 14 [29] M. Kositzi, I. Poulios, S. Malato, J. Cáceres, A. Campos, Water Research, 38 (2004) 1147-15 1154. 16 [30] S. Malato, J. Blanco, C. Richter, M.I. Maldonado, Applied Catalysis B: Environmental 25 17 (2000) 31–38. 18 [31] M. Abdullah, G.K.C. Low, R.W. Matthews, Journal of Physical Chemistry 94 (1990) 6820–19 6825. 20 [32] J. Colina-Márquez, F. Machuca-Martínez, G. Li Puma, Environmental Science & Technology 21 44 (2010) 5112–5120. 22 [33] J.R. Bolton, K.G. Bircher, W. Tumas, C.A. Tolman, Pure and Applied Chemistry, 73(4) 23 (2001) 627–637. 24 [34] R. Hulstrom, R. Bird, C. Riordan, Solar Cells 15 (1985) 365–391. 25 19 [35] R. Marotta, D. Spasiano, I. Di Somma, R. Andreozzi, V. Caprio, Chemical Engineering 1 Journal, 209 (2012) 69-78. 2 [36] C.C. Chusuei, M.A. Brookshier, D.W. Goodman, Langmuir 15 (1999) 2806-2808. 3 [37] D. Chen, A.K. Ray, Chemical Engineering Science 56 (2001) 1561–1570. 4 1 TiO2/Cu(II) photocatalytic production of benzaldehyde from benzyl 1 alcohol in solar pilot plant reactor 2 Danilo Spasiano a* , Lucia del Pilar Prieto Rodriguez b , Jaime Carbajo Olleros c , Sixto Malato b , 3 Raffaele Marotta a , Roberto Andreozzi a 4 5 a Department of Chemical Engineering, Faculty of Engineering, University of Naples “Federico II”, 6 p.le V. Tecchio, 80 – 80125 – Naples, Italy. 7 b Plataforma Solar de Almería-CIEMAT, Carretera de Senés Km 4 - 04200 - Tabernas, Almería, 8 Spain. 9 c Departamento de Ingeniería de Procesos Catalíticos Instituto de Catálisis y Petroleoquímica, CSIC, 10 C/ Marie Curie 2 - 28049 - Cantoblanco, Madrid, Spain. 11 12 13 * Corresponding author. Tel +390817682968 fax +390815936936. E-mail danilo.spasiano@unina.it 14 15 ABSTRACT 16 The technical feasibility of selective photocatalytic oxidation of benzyl alcohol to benzaldehyde, in 17 aqueous solutions, in presence of cupric ions has been investigated in a solar pilot plant with 18 Compound Parabolic Collectors. Aldrich (pure anatase) and P25 Degussa TiO2 have been used as 19 photocatalysts. The influences of cupric species concentrations, solar irradiance and temperature are 20 discussed too. The oxidation rates were strongly influenced by the initial cupric ions concentration, 21 incident solar irradiance and temperatures. 22 The best results found, in terms of yields and selectivities to benzaldehyde under acidic conditions 23 were higher than 50% and 60% respectively under acidic conditions. 24 manuscript.doc Click here to view linked References http://ees.elsevier.com/apcatb/viewRCResults.aspx?pdf=1&docID=10331&rev=1&fileID=327009&msid={0C3BACAB-6896-4BEA-BAE0-870D59CE1610} 2 Under deaerated conditions, the presence of reduced copper species was proved by XPS analysis. 1 The results indicated that, at the end of the process, cupric species can be easily regenerated and 2 reused, through a re-oxidation of reduced copper, produced during the photolytic run, with air or 3 oxygen in dark conditions. 4 A figure-of-merit (ACM), proposed by the International Union of Pure and Applied Chemistry 5 (IUPAC) and based on the collector area, has been estimated, under the proposed conditions, with 6 the aim to provide a direct link to the solar-energy efficiency independently of the nature of the 7 system. Generally speaking, it can be considered that lower ACM values higher the system 8 efficiency. 9 10 Keywords: selective oxidation, benzyl alcohol, benzaldehyde production, TiO2 photocatalysis, solar 11 photocatalytic plant, figure-of-merit. 12 13 1. Introduction 14 The use of TiO2, as readily available and environmentally friendly photocatalyst, was largely 15 investigated for the removal of non-biodegradable or undesiderable organic substances from 16 wastewater [1-3] and TiO2 based on solar technologies were developed during the past years for 17 reducing the cost of large-scale aqueous-phase applications to treat industrial wastewater [4,5]. 18 However, only in recent years the research has pointed its attention on the possibility to use the 19 TiO2 as a photocatalyst for the selective oxidation of organic molecules under UV radiation in non 20 aqueous media [6]. For example, the selective oxidation of aromatic alcohols in the respective 21 aldehydes can be easily gained in acetonitrile or in solvent free systems at room temperature [7,8] 22 due to the reaction of the alcoholic substrate with the photogenerated positive holes (  vbh ) and using 23 oxygen as acceptor of photoelectrons [9]. In particular, the addition of an organic solvent, as 24 acetonitrile, allows the oxidation of benzyl alcohol and its derivates with selectivities to the 25 3 respective aldehydes of over 90% [8,10], values well higher than those reported (10-60%) when the 1 same process is carried out in aqueous solutions using TiO2 nanoparticles [11,12]. 2 Moreover, selectivities of 41-74% were reported for the oxidation of 4-substituted aromatic 3 alcohols to the corresponding aldehydes, using aqueous media, over rutile or anatase TiO2 catalysts 4 [11,13]. The decrease of the selectivity in aqueous solution with respect the use of organic non 5 aqueous solvent is mainly due to [9]: 6 a) formation of very reactive species such as hydroxyl radicals (HO • ) through the reaction of 7 adsorbed water molecules or hydroxyl groups with positive holes: 8    HHOHOOH VBh adsads )(/2 (r1) b) generation of radical superoxide (O2 •- ) by the reaction of molecular oxygen with photogenerated 9 electrons (  cbe ). This species, once formed, in presence of water, gives rise hydrogen peroxide 10 whose electron trapping generates HO • radicals, which, in absence of non aqueous solvent, oxidizes 11 not selectively the organic substances: 12 ads e ads OO CB   22 )( (r2) ads e ads OHHO CB 222 )(2    (r3)    HOHOOH CBe ads )( 22 (r4) oxSSHO  (r5) On the other side, water is considered the best solvent for environmental friendly process and, for 13 this reason, other new approaches have arisen in order to avoid the use the organic solvents. In 14 particular, very recent studies demonstrated that the use of Cu(II) ions, as electron acceptor, instead 15 of molecular oxygen, under deaerated conditions, leads to an increase of selectivity to benzaldehyde 16 up to 72%, starting from benzyl alcohol [14], since the formation of hydroxyl radicals takes place 17 just by the reaction of water molecules or surface adsorbed hydroxyl groups with positive holes. In 18 4 this case, cupric ions reduce to a lower state of oxidation by capturing the photogenerated electron 1 on the TiO2 surface and precipitate from the solution: 2 )0()()( )()( CuICuIICu CBCB ee   (r6) whereas adsorbed benzyl alcohol (BA) is oxidized to the benzaldehyde (BHA), through a direct 3 reaction with the positive holes: 4 ads )( ads BHA 2 BA    VBh (r7) At the end of the process, cupric ions could be regenerated and reused, through the reoxidation of 5 zero valent precipitated copper with an air flow at dark conditions. 6 Chemical state of solid copper after the photocatalytic run was also intensively investigated, but 7 controversial results were reported. Most of the researchers reported that the solid is composed by a 8 mixture of zero valent copper and cupric oxide (and, in some case, cuprous oxide) [15-19]. 9 In some studies it was concluded that the reduced Cu species was zero valent copper [20-21]. 10 However, the possibility of the presence of both Cu(I)/Cu(II) species, due to a reoxidation of metal 11 copper during the preparation of the analytical samples, was not completely ruled out [22]. 12 At best of the Authors’ knowledge, only lab-scale lamp-driven investigations on the selective TiO2 13 photocatalytic oxidation of benzyl alcohol and its derivatives, in aqueous or acetonitrile or solvent 14 free systems, were reported in the literature [11, 12, 14, 23-27]. 15 It could be very interesting, for economic and environmental reasons, to evaluate the possibility of 16 exploiting the use of solar radiations dayling arriving on the earth surface, through the use of pilot 17 plants, for the production of benzaldehyde from benzyl alcohol. 18 In the present work, the development of the TiO2/Cu(II)/solar radiation system for the selective 19 oxidation of benzyl alcohol to benzaldehyde in water is studied, by using a solar photocatalytic pilot 20 plant at different operating conditions (TiO2 photocatalyst type, Cu(II) initial concentration and 21 irradiance of solar radiation). The possibility of reuse copper as catalyst was also tested by 22 oxidizing the precipitated zero valent copper with an air flow blown into the pilot-plant in dark 23 5 conditions. Optimal conditions of TiO2 load and pH values were fixed according to the ones found 1 in previous studies [14,28]. 2 Moreover, analyses of solid, after the oxidation runs, were carried out to better clarify the nature of 3 copper species at the end of the process. 4 5 2. Material and methods 6 2.1 Experimental set-up and procedures 7 2.1.1 Pilot-Plant experiments 8 Experimental investigations were carried out, from May to July 2012, in a solar pilot-plant 9 consisting of twelve Compound Parabolic Collectors (CPC), installed in the Plataforma Solar de 10 Almería (37° latitude N, Spain) and elsewhere described in its geometrical and functional 11 characteristics [29]. The total solution volume is 39 l, where only 22 l are exposed to solar 12 radiation; the rest is distributed between the recirculation tank (9 l) and the hydraulic connections (8 13 l). The recirculation tank of the adopted CPC was modified, as reported in figure 1, in order to 14 ensure the absence of oxygen, essential for performing the proposed process. For this purpose, the 15 solutions, containing the catalysts and the substrate, were preventively purged with nitrogen 16 gaseous stream through two porous ceramic spargers, by closing the valve Vn, to remove the 17 dissolved oxygen. During the experimental runs a continuous nitrogen flow was guaranteed to the 18 reactor to prevent the entry of air in the reactor but it was switched in the freeboard of the 19 recirculation tank, by opening the valve Vn, to inhibit the stripping of organic substances from the 20 solution. In the last case, the bubbling of nitrogen is strongly limited by the pressure drop due to the 21 spargers. 22 To better follow the concentration profiles of the compounds involved in the process, some 23 experimental runs were carried out during two days. In these cases, the irradiated part of the reactor 24 was covered and the recycling tank was capped to stop the photochemical reactions and to prevent 25 the entry of oxygen respectively. 26 6 Samples were collected in a glass bottle at different reaction times, by opening the valve Vs, rapidly 1 filtered to prevent the reoxidation of precipitate copper and finally analyzed. 2 All the experimental data were reported in function of the accumulated energy, per unit of volume 3 (kJ/l), incident on the reactor at the corresponding time of the withdrawn sample (Qn) [30]: 4 11 ;   nnn t r nnnn ttt V A UVtQQ (eq. 1) where Qn-1 represent the accumulated energy (per unit of volume, kJ/l) taken during the experiment 5 relative to the previous sampling; ∆tn, nUV and Vt are, respectively, the elapsed time, the average 6 UV-irradiance (measured by a global UV radiometer KIPP&ZONEN, model CUV 3 mounted on a 7 platform tilted the same angle as the CPCs) which reaches the collector surface (Ar) between the 8 two samplings and the total solution volume. 9 10 2.1.2 Laboratory-scale experiments 11 Experiments were carried out in a Suntest solar simulator (Suntest XLS+ photoreactor, Atlas) 12 equipped with a 765-250 W/m 2 Xenon lamp (61-24 W/m 2 from 300 nm to 400 nm, 1.4∙10 20 -13 5.5∙10 19 photons/m 2 ∙s) and a cooler to keep the temperature at 35°C. The UV irradiation, in the 14 range of 300-400 nm, was monitored by using a portable radiometer (Solar Light CO PMA 2100) 15 fixed on the shaker inside the lamp influence zone, like shown in figure 2. 16 In this case, a volume of 700 ml of reacting solution was prepared in a one liter flask and, after 30 17 minutes of stripping with nitrogen, was rapidly poured in ten cylindrical glass vials with the 18 capacity of 42 ml and a diameter of 25 mm. The vials were rapidly closed with a screw cap, to 19 prevent the entry of oxygen, and were simultaneously exposed to the lamp radiation and agitated by 20 a shaker like shown in figure 2. The temperature was set at 35 °C. Each vial represented a sample to 21 remove from the solar box at different reaction times. Once collected, all the samples were rapidly 22 filtered and analyzed. 23 7 The pH of the solutions was adjusted at the value of 2.0, for all the runs, by using an aqueous 1 solution of 85% of phosphoric acid. 2 2.2 Analytical methods 3 The concentrations of benzyl alcohol, benzaldehyde and benzoic acid at different reaction times 4 were evaluated by HPLC analysis. For this purpose, the HPLC apparatus (Agilent 1100) was 5 equipped with a diode array UV/Vis detector (λ= 215, 230, 250 nm) and Phenomenex (Gemini 5u 6 C18 150x3 mm) column, using a mobile phase flowing at 0.7 ml min −1 . The mobile phase was 7 prepared by a buffer solution (A), H2O (B) and CH3CN (C). A linear gradient progressed from 15% 8 C to 28% C and from 45% B to 32% B in 10 minutes with a subsequent re-equilibrium time of 3 9 minutes. One liter of buffer was made by 10 ml of phosphoric acid solution (5.05 M), 50 ml of 10 methyl alcohol and water for HPLC. 11 The concentration of dissolved copper ions was measured by means of a colorimetric method using 12 an analytical kit (based on oxalic acid bis-cyclohexylidene hydrazide, cuprizone®) purchased from 13 Macherey-Nagel. An UV/Vis spectrometer (UNICAM-II spectrophotometer) has been used for the 14 measurements at a wavelength of 585 nm. 15 Total organic carbon (TOC) was monitored by Shimadzu Total Organic Carbon analyser model 16 TOC-5050A, equipped with an auto sampler ASI 5000A. The pH was monitored by a portable pH-17 meter (Crison pH 25). 18 The Cu/Ti ratio for unknown solids, withdrawn at the end of some photocatalytic runs, was 19 estimated by an Energy Dispersive X-ray spectrometer system (SwiftED, Oxford Instruments) 20 attached to a Scanning Electron Microscope (TM-1000, Hitachi). 21 Powder X-ray diffraction (XRD) patterns were estimated using a X´PertPRO (PANalytical) 22 diffractometer with nickel-filtered Cu Kα radiation. The X-ray generator was operated at 45 kV and 23 40 mA. The powders were scanned from 2θ = 4º to 90º with a 0.02 step size and accumulating a 24 total of 5 s per point. 25 8 X-ray photoelectron spectroscopy (XPS) analysis was carried out under high vacuum chamber with 1 a base pressure below 9x10 -7 Pa at room temperature. Photoemission spectra were recorded using a 2 SPECS Gmbh system equipped with an UHV PHOIBOS 150 analyser with Al monochromatic 3 anode operated at 12 kV and 200 W with a photon energy of h = 1486.74 eV. A pass energy of 25 4 eV was used for high-resolution scans. Binding energies (BE) were referenced to C1s peak (284.6 5 eV) to take into account charging effects. 6 The XPS spectra obtained were then curved fitted using the XPS PeakFit software. The areas of the 7 peaks were computed by fitting the experimental spectra to Gaussian/Lorentzian curves after 8 subtracting the background (Shirley function). 9 10 2.3 Materials 11 Two commercial microcrystalline TiO2 powders were tested: TiO2 Degussa P25 (80% anatase, 20% 12 rutile, BET specific surface area 50 m 2 g −1 ) and TiO2 Aldrich (pure anatase phase, BET specific 13 surface area 9.5 m 2 g −1 ). 14 Cupric ions were introduced in the system as anhydrous cupric sulphate (Sigma-Aldrich) with a 15 purity >99.0% (w/w). Benzyl alcohol (BA), benzaldehyde (BHA) and benzoic acid (BAC), with a 16 purity >99.0% (w/w), were purchased from Sigma Aldrich and used as received. Phosphoric acid 17 with a purity 85% from Merck as used as received 18 19 3. Result and discussion 20 3.1 Effect of TiO2 type 21 The results obtained during different experimental runs of solar photooxidation of benzyl alcohol, 22 with two different typologies of commercial TiO2 samples, at the same load (200 mg/l), are shown 23 in figures 3a-b. 24 9 The runs were carried out in two days. At the end of first day, when the reactor was covered, the 1 UV-irradiances (wavelength 300-400 nm) and temperatures decreased (see figs 3a,b). During the 2 runs, measured UV-irradiances ranged between 40 and 50 W/m 2 approximately (fig. 3b). 3 As shown in the diagrams, the reactivity of the system is higher when TiO2 P25 by Degussa is used 4 than the TiO2 Aldrich catalyst. In particular, in presence of P25 Degussa sample (empty symbols), 5 for a Qn value of 123 kJ/l, the BA concentration approached to zero (figure 3a) and Cu(II) ions were 6 totally reduced (figure 3b), whereas, if Aldrich TiO2 sample is used (full symbols), about 27% of 7 initial benzyl alcohol and Cu(II) ions were still present in the solution. 8 The higher reactivity observed for the system in presence of P25 Degussa TiO2 is in disagreement 9 with the results, previously reported [14], that show a similar reactivity when P25 Degussa and 10 Aldrich TiO2 catalysts were used in presence of an UV radiation emitted by a laboratory 11 thermostated lamp. This discrepancy may be probably due to the different climatic conditions that 12 influenced the temperature trends of the reacting solution, as reported in figure 3a (dotted and dash-13 dot lines). Moreover, the different reactivities, recorded in the solar experiments, could be also 14 attributed to the differences between the averages of the measured temperatures: 38.6 °C and 34.3 15 °C for runs carried out in presence of Degussa P25 and Aldrich TiO2 catalysts respectively. 16 Moreover, since P-25 and Aldrich have not the same light absorption characteristics in the range of 17 300-800 nm, another reason could be the different emission spectra between lamp and the sun. 18 The BA solar phtocatalytic oxidation resulted in the production of BHA for both the catalysts, as 19 shown in figure 4 (empty and full diamonds) and BAC an undesired product that derives from the 20 reaction between the positive photogenerated holes on the TiO2 surface and benzaldehyde. BAC 21 yields are also reported the same figure (empty and full triangles). 22 According to the previous results, the highest BHA production rates were gained in presence of P25 23 Degussa TiO2. Moreover, for accumulated energy values higher than 80 kJ/l, a decrease in BHA 24 yield with respect of initial BA amount, from the value of 53.3% to 45.5%, was observed using P25 25 Degussa TiO2 samples. This result can be explained by considering a competition in the reaction 26 10 between the BA and BHA molecules with the photogenerated positive holes on the TiO2. This is in 1 agreement with the results reported in figure 3a (empty squares), when the accumulated energy 2 reached the value of 83 kJ/l, the unconverted BA concentration was only 16% of its initial 3 concentration. At the same time, a decrease of production rate of BAC (empty triangles, fig. 4) was 4 recorded, being the unreacted Cu(II) concentration only the 22% of the initial one (empty squares, 5 fig 3b). 6 The BHA selectivity, with respect to BA consumption, was calculated for both TiO2 types (Fig. 5). 7 As shown in the diagram, it seems that the use of Aldrich TiO2 sample renders the system more 8 selective, reaching, for Qn values of 130 kJ/l, BHA selectivity values close to 70% and 50% in 9 presence of Aldrich TiO2 and P25 Degussa respectively. TOC measurements collected during the 10 runs are also reported in figure 5, in terms of mineralization degrees. The trends are very similar, 11 with a degree of mineralization at the end of the experimental runs of 9% and 7% in presence of 12 P25 Degussa and Aldrich TiO2 respectively. 13 14 3.2 Effect of cupric ions concentration 15 To evaluate the effect of Cu(II) initial concentration, some experimental runs of solar photoxidation 16 of benzyl alcohol were carried out with Aldrich TiO2 at the load of 200 mg/l and pH = 2.0, varying 17 the cupric ions starting concentration (0.5 mM, 1.0 mM and 1.5 mM). 18 As shown in figures 6a-b, the temperatures and UV-irradiances profiles (solid, dashed and dotted 19 lines) are so similar as to be considered equals for the three runs. 20 With Cu(II) and BA starting concentrations of 0.5 mM and 1.5 mM, the BA oxidation stopped for 21 Qn values close to 35 kJ/l where a complete reduction of cupric ions was observed (full triangles). 22 At the highest Cu(II) initial concentration than 1.5 mM resulted into a decrease of the system 23 reactivity and BA conversion (Fig. 6a, full circles) and, at the same time, the Cu(II) reduction rates 24 decrease. The observed results, according to those previously reported [14], are probably due to a 25 partial catalyst deactivation by the adsorbed sulphate ions which may block the TiO2 active sites 26 11 [31]. In fact, since Cu(II) species were add to the reactive solution as cupric sulphate (CuSO4), 1 increasing the Cu(II) ions initial concentration results into an increase of sulphate concentration too. 2 Moreover, the sulphate anions, at concentration level higher than 1.0 mM, exerted a negative effect, 3 inhibiting the BA photooxidation rates because they are in competition with BA molecules in the 4 reaction with the positive holes [14]. 5 In figure 7 are reported the experimental concentration profiles of BHA and BAC (full and empty 6 symbols) that are in agreement with BA concentration trends shown in figure 6a. In particular the 7 best result found, in term of yield, was of 43% for BHA, starting with [Cu(II)]o = 1.0 mM, for a 8 accumulated energy value of 67 kJ/l. For all the runs, the selectivity was always higher than 67% 9 and the mineralization degrees were lower than 4.5% (data not shown). 10 11 3.3 Effect of the irradiance and temperature 12 As previously shown in the figures 3b and 6b, the Cu(II) concentrations are characterized by a s-13 shaped profile and, in particular, when the solar UV radiations and reactor temperatures reached the 14 top, a marked increase of Cu(II) reduction rate was observed. With the aim to better assess the 15 relationship between the variability of both irradiances and temperatures during a solar run and the 16 s-shaped of cupric ions concentration profile, a set of three laboratory photolytic experiments, with 17 Cu(II) initial concentration equal to 0.5 mM, were carried out at three different irradiances, kept 18 constant during a single run and under controlled solution temperature (T=35 °C). For this purpose, 19 a solar box apparatus, described in the experimental section, was used. Since the internal diameters 20 of CPC solar reactor tubes and the solar box vials are 31 mm and 24 mm respectively, the TiO2 load 21 that maximizes the adsorption of UV radiation emitted by solar lamp is higher than the one used for 22 solar experiments carried out in CPC reactor. 23 The optimum TiO2 concentration (ccat), for which the optical thickness equals that of CPC reactor 24 configuration (=9.12) can be easily calculated as suggested by others [28,32]: 25 12   mg/l258   dk ccat   eq. 2 where  is the scattering coefficient (1.29510 3 m 2 /kg), k is the catalyst specific mass absorption 1 (1.7510 2 m 2 /kg) and d the internal tube diameter (24 mm). 2 As shown in figure 8, by increasing the UV irradiance from 39.5 W/m 2 to 59.7 W/m 2 , the Cu(II) 3 photoreduction rate increases. 4 The increase of Cu(II) reduction rate, by increasing UV irradiance, corresponded to the increase of 5 BA oxidation rate (data not shown). In particular, when the reacting system accumulated about 60 6 kJ/l, the BA conversion was of 17% and the BHA yield, was close to 14-16% for all the three runs. 7 Moreover, under controlled temperature and irradiance, no s-shaped concentration was observed. 8 9 3.4 Figure-of-merit 10 To estimate the operating costs of sole natural radiation, the figure-of-merit concept was used [33]. 11 For solar-energy-driven systems, the figure-of-merit allows to the assessment of the solar 12 technology efficiency used for the investigated process. In fact, even if there is no cost for the solar 13 radiation, there could be a non-marginal capital cost for the collector. Being the capital cost of a 14 solar collector generally proportional to its area, a figure-of-merit, based on the solar collector area, 15 is appropriate. 16 For the adopted experimental batch conditions, the appropriate figure-of-merit is the collector area 17 per mass (ACM), defined as the collector area required to reduce of a unit mass of the substrate in the 18 reacting system in a time of 1 hour (to) for an incident solar irradiance of 1000 W m −2 ( o SE ): 19  fi o Sot Sr CM ccEtVM EtA1000 A    eq. 3 where Ar (3.19 m 2 ) is the real collector area, M is the molar mass of the substrate (108.14 g/mol), Vt 20 (39 l) is the volume of treated solution, SE (814.6 W/m 2 ) is the average direct solar irradiance over 21 the reaction time t (4.83 h), ci and cf are respectively the initial and final substrate concentrations 22 13 (mol/l), o SE is the standardized irradiance (1000 W/m 2 , based on the AM1.5 standard solar spectrum 1 on a horizontal surface) [34] and to is the reference time (1 h). 2 On the basis of the previous formula an average value of ACM = 3.08∙10 3 m 2 per kilogram of benzyl 3 alcohol and per hour consumed was thus calculated for the collector area per mass (200 mg/l of 4 Aldrich TiO2, pH=2.0, [Cu(II)]0 = 1.0 mM, [BA]0 = 1.5 mM). 5 6 3.5 Copper reuse 7 The possibility of reuse the reduced copper, as catalyst, was tested by carrying out an experimental 8 run, by using the CPC solar reactor, consisting in three cycles of BA photo-oxidation (fig. 9). When 9 all Cu(II) species was totally reduced as precipitate copper, its reoxidation was carried out, in dark 10 conditions, under air bubbling in the recirculation tank for 30 minutes. At the end of the first cycle, 11 the reactive solution was re-purged with nitrogen gas, for other 30 minutes, to remove the dissolved 12 oxygen and a new photocatalytic cycle was started. The experimental results, reported in figure 9, 13 pointed out that it’s possible, for each cycle, to reoxidize completely the precipitate copper to cupric 14 species (empty diamonds). 15 During the first two cycles of BA solar photoxidation, not particular changes were observed on the 16 reactivity of the system, whereas during the third cycle a decrease of both BA consumption and 17 BHA production rates were observed. This behaviour could be explained by considering that the 18 solution composition changes during the experimental run. In particular, at the 3 rd cycle beginning, 19 BA and BHA normalized concentrations were 43% and 33% respectively thus favouring a 20 competition kinetics between BA and BHA, both adsorbed on TiO2 surface, towards the reaction 21 with the photogenerated holes [35]: 22 ads sMk ads BHAhBA    /108.3 5 2 ads sMk ads BAChBHA    /100.3 4 2 14 The BHA selectivities of the process for the three photocatalytic cycles are reported in figure 10. As 1 shown in the diagram, the highest selectivities (up to 100%) were obtained in the first cycle, 2 whereas lower selectivities were observed during the second (close to 75%) and the third cycle 3 (close to 40%) thus supporting, with the progress of the reaction time, the intervention of undesired 4 oxidation reactions of BHA once produced. 5 After each photocatalytic cycle under anoxic conditions, the solid was filtered, stored under inert 6 atmosphere and submitted to EDS, XRD and XPS analysis in order to better investigate the 7 distribution of copper deposited on the solid and its oxidation states. 8 From the EDS investigations, Ti/Cu atom ratios, for the filtered solids, resulted 97.8/2.2, 96.5/3.5 9 and 85.2/14.8 (w/w) for the samples withdrawn at the end of the first, second and third cycle 10 respectively. These results supported the idea that reduced Cu species, formed during the 11 photooxidative runs, accumulated on TiO2 surface increasing the amount of deposited copper from 12 the first cycle to third one up to 17.4% by weight. The increase of Cu amount on the TiO2 is 13 probably due to the deposition of a part of photocatalyst powders, during the experimental run, in 14 the recirculation tank and/or along the hydraulic connections of the plant, thus decreasing the active 15 load of it available in the solar photoreactor with a consequent reduction of BA consumption and 16 BHA production rates, really observed, in particular during the third cycle. 17 A typical XPS spectra for the solid samples shows different peaks (fig. 11). The peaks at 932.8 eV 18 and 952.6 eV indicate the predominant presence of copper reduced species (+1/0) as previously 19 reported by others [19,22,36] whereas the existence of peaks at 935.4 eV and 955.2 eV can be 20 attributed to the presence of cupric species [22], such as CuO, probably an artifact produced during 21 the preparation of the sample before the XPS analysis. 22 Unfortunately, XRD analysis did not give any results being the amount of Cu reduced species 23 accumulated on the TiO2 surface (max 17.4%) below XRD detection limit. 24 25 26 15 4. Conclusion 1 The possibility to convert benzyl alcohol to benzaldehyde by photocatalytic oxidation was 2 demonstrated in aqueous solution under natural solar radiation at pilot plant scale. The oxidation 3 rates were strongly influenced by the initial cupric ions concentration, incident solar irradiance and 4 temperatures. The best result found, in terms of yield, was of 53.3% for benzaldehyde with respect 5 to the initial benzyl alcohol concentration (63.4 % of selectivity) for an accumulated energy value 6 (Qn) of 78.9 kJ/l (reaction time of 385 min) and operating with an average temperature of 38.6 °C. 7 EPS investigations, carried out on the solids, withdrawn during different photocatalytic cycles, 8 confirmed the existence of both Cu reduced (0/+1) and oxidized species, the latter probably 9 produced during the sample preparation before the analysis. 10 The results also indicated that cupric species can be easily regenerated and reused with air or 11 oxygen in dark conditions. 12 A figure-of-merit (ACM) was calculated be equal to 3.08∙10 3 m 2 per kilogram and hour of benzyl 13 alcohol converted. 14 15 References 16 [1] M.N. Chong, B. Jin, C.W.K. Chow, C. Saint, Water Research 44 (2010) 2997 – 3027. 17 [2] K. Rajeshwara, M.E. Osugi, W. Chanmanee, C.R. Chenthamarakshan, M.V.B Zanoni, P. 18 Kajitvichyanukul, R. Krishnan-Ayer, Journal of Photochemistry and Photobiology C: 19 Photochemistry Reviews 9 (2008) 171–192. 20 [3] J.M. Herrmann, Applied Catalysis B-Environmental 99(3-4) (2010) 461-468. 21 [4] R.J. Braham, A.T. Harris, Industrial & Engineering Chemistry Research, 48 (2009) 8890–8905. 22 [5] S.R. Malato, P. Fernandez-Ibanez, M.I. Maldonado, G.J. Blanco, W. Gernjak, Catalysis Today 23 147 (2009) 1–59. 24 [6] Y. Shiraishi, T. Hirai, Journal of Photochemistry and Photobiology C: Photochemistry Reviews 25 9 (2008) 157–170. 26 16 [7] S. Higashimoto, N. Suetsugu, M. Azuma, H. Ohue, Y. Sakata, Journal of Catalysis 274 (2010) 1 76–83. 2 [8] W. Feng, G. Wu, L. Li, N. Guan, Green Chemistry 13 (2011) 3265–3272. 3 [9] C.S. Turchi, D.F. Ollis, Journal of Catalysis, 122 (1990) 178–192. 4 [10] S. Higashimoto, N. Kitano, N. Yoshida, T. Sakura, M. Azuma, H. Ohue, Y. Sakata, Journal of 5 Catalysis 266 (2009) 279–285. 6 [11] G. Palmisano, S. Yurdakal, V. Augugliaro, V. Loddo, L. Palmisano, Adv. Synth. Catal. 349 7 (2007) 964-970. 8 [12] S. Yurdakal, G. Palmisano, V. Loddo, V. Augugliaro, L. Palmisano, Journal of American 9 Chemical Society 130 (2008) 1568-1569. 10 [13] V. Augugliaro, V. Loddo, M.J. Lopez-Munoz, C. Marquez-Alvarez, G. Palmisano, L. 11 Palmisano, S. Yurdakal, Photochemical & Photobiological Sciences 8 (2009) 663–669. 12 [14] R. Marotta, I. Di Somma, D. Spasiano, R. Andreozzi, V. Caprio, Chemical Engineering 13 Journal 172 (2011) 243–249. 14 [15] J.W.M. Jacobs, F.W.H. Kampers, J.M.G Rikken, C.W.T. Bulle-Lieuwma, D.C. Doningsberger, 15 Journal of The Electrochemical Society 136 (1989) 2914-2023. 16 [16] M. Bideau, B. Claudel, L. Faure, M. Rachimoellah, Chemical Engineering Communications 93 17 (1990) 167–179. 18 [17] S. Morishita, Chemistry Letters 10 (1992) 1979-1982. 19 [18] N.S. Foster, R.D. Noble, C.A. Koval, Environmental Science & Technology 27 (1993) 350–20 356. 21 [19] S.W. Zou, C.W. How, J.P. Chen, Industrial & Engineering Chemistry Research 46(20) (2007) 22 6566–6571. 23 [20] M. Canterino, I. Di Somma, R. Marotta, R. Andreozzi, . Water Res., 42 (2008) 4498 -4506. 24 [21] H. Reiche, W.W. Dunn, A.J. Bard, Journal of Physical Chemistry 83 (1979) 2248-2251. 25 17 [22] S. Xu, J. Ng, A.J. Du, J. Liu, D.D. Sun, International Journal of Hydrogen Energy 36 (2011) 1 6538-6545. 2 [23] D.I. Enache, J.K. Edwards, P. Landon, B. Solsona-Espriu, A.F. Carley, A.A. Herzing, M. 3 Watanabe, C.J. Kiely, D.W. Knight, G.J. Hutchings, Science 311 (2006) 362–365. 4 [24] S. Yurdakal, G. Palmisano, V. Loddo, O. Alagoz, V. Augugliaro, L. Palmisano, Green Chem. 5 11 (2009) 510–516. 6 [25] V. Augugliaro, T. Caronna, V. Loddo, G. Marcì, G. Palmisano, L. Palmisano, S. Yurdakal, 7 Chemistry - A European Journal. 14 (2008) 4640–4646. 8 [26] S. Higashimoto, K. Okada, T. Morisugi, M. Azuma, H. Ohue, T.H. Kim, M. Matsuoka, M. 9 Anpo, Topics in Catalysis 53 (2010) 578–583. 10 [27] C.J. Li, G.R. Xu, B. Zhang, J.R. Gong, Applied Catalysis B: Environmental 115–116 (2012) 11 201–208. 12 [28] L. Prieto-Rodriguez, S. Miralles-Cuevas, I. Oller, A. Agüera, G. Li Puma, S. Malato, Journal 13 of Hazardous Materials 211– 212 (2012) 131– 137. 14 [29] M. Kositzi, I. Poulios, S. Malato, J. Cáceres, A. Campos, Water Research, 38 (2004) 1147-15 1154. 16 [30] S. Malato, J. Blanco, C. Richter, M.I. Maldonado, Applied Catalysis B: Environmental 25 17 (2000) 31–38. 18 [31] M. Abdullah, G.K.C. Low, R.W. Matthews, Journal of Physical Chemistry 94 (1990) 6820–19 6825. 20 [32] J. Colina-Márquez, F. Machuca-Martínez, G. Li Puma, Environmental Science & Technology 21 44 (2010) 5112–5120. 22 [33] J.R. Bolton, K.G. Bircher, W. Tumas, C.A. Tolman, Pure and Applied Chemistry, 73(4) 23 (2001) 627–637. 24 [34] R. Hulstrom, R. Bird, C. Riordan, Solar Cells 15 (1985) 365–391. 25 18 [35] R. Marotta, D. Spasiano, I. Di Somma, R. Andreozzi, V. Caprio, Chemical Engineering 1 Journal, 209 (2012) 69-78. 2 [36] C.C. Chusuei, M.A. Brookshier, D.W. Goodman, Langmuir 15 (1999) 2806-2808. 3 Fig. 1 Figure 1 Fig. 2 Figure 2 Fig. 3 Figure 3 Q n (kJ/l) 0 20 40 60 80 100 120 140 [B H A ] / [B A ] 0 [B A C ] / [B A ] 0 ( % ) 0 10 20 30 40 50 60 Fig. 4 Figure 4 Fig. 5 Figure 5 Fig. 6 Figure 6 Fig. 7 Figure 7 Fig. 8 Figure 8 Fig. 9 Figure 9 Fig. 10 Figure 10 Fig. 11 Figure 11 Fig. 1: Solar photocatalytic CPC pilot plant reactor Fig. 2: Solar box Fig. 3: Effect of TiO2 type at pH=2.0: BA solar photoxidation (squares) and temperatures profiles (dotted and dashed-dotted lines) (3a); Cu(II) solar photoreduction (squares) and UV-irradiance (continuous and dashed lines) (3b). [BA]o = 1.5 mM. [Cu(II)]o = 1.5 mM. Initial TiO2 load = 200 mg/l. 3a: (, -•- ) Aldrich TiO2, (, •••) P25 Degussa TiO2. 3b: (, ___ ) Aldrich TiO2, (, ---) P25 Degussa TiO2. Fig. 4: Effect of TiO2 type on the BHA and BAC production. [BA]o = 1.5 mM. [Cu(II)]o = 1.5 mM. Initial TiO2 load = 200 mg/l. pH=2.0. Aldrich TiO2:  BHA,  BAC. P25 Degussa TiO2:  BHA,  BAC. Fig. 5: Effect of TiO2 type on the BHA selectivity and mineralization degree (dotted and dashed lines). [BA]o = 1.5 mM. [Cu(II)]o = 1.5 mM. Initial TiO2 load = 200 mg/l. pH=2.0. Aldrich TiO2: () BHA selectivity, (---) mineralization degree. P25 Degussa TiO2: () BHA selectivity, (•••) mineralization degree. Fig. 6: Effect of initial Cu(II) concentration at pH=2.0 on the BA and Cu(II) concentration profiles: BA solar photoxidation and temperatures profiles (6a); Cu(II) solar photoreduction and UV- irradiances profiles (6b). [BA]o = 1.5 mM. Initial TiO2 (Aldrich) load = 200 mg/l. [Cu(II)]o: () 0.5 mM, () 1.0 mM, () 1.5 mM. Temperatures and Irradiances: (•••) for [Cu(II)]o = 0.5mM, (---) for [Cu(II)]o = 1.0 mM, ( ___ ) for [Cu(II)]o = 1.5 mM. captions.doc Fig. 7: Effect of initial Cu(II) concentration on the BHA and BAC concentration profiles: BHA (full symbols) and BAC (empty symbols) productions. [BA]o = 1.5 mM. Initial TiO2 (Aldrich) load = 200 mg/l. pH = 2.0. [Cu(II)]o: (,) 0.5 mM, (, ) 1.0 mM, (,) 1.5 mM. Fig. 8: Effect of irradiance on the Cu(II) concentration profiles: [BA]o = 1.5 mM. [Cu(II)]o = 0.5 mM. Initial TiO2 (Aldrich) load = 258 mg/l. pH = 2.0 and T = 35 °C. UV Irradiance (solar box):  39.5 W/m 2 ,  49.0 W/m 2 ,  59.7 W/m 2 . Fig. 9: Normalized concentration profiles for Cu(II), BA, BHA and BAC with light on and nitrogen purge or light off and oxygen purge. [BA]o = 2.5 mM. [Cu(II)]o = 0.5 mM. Initial TiO2 (Aldrich) load = 200 mg/l. pH = 2.0.  Cu(II),  BA,  BHA,  BAC. Fig. 10: BHA selectivities during the three cycles: [BA]o = 1.5 mM. [Cu(II)]o = 0.5 mM. Initial TiO2 (Aldrich) load = 200 mg/l. pH = 2.0.  1 th cycle,  2 nd cycle,  3 rd cycle. Fig. 11: XPS spectra for the solid sample after the photocatalytic oxidation run. *Graphical Abstract (for review)  Use of the TiO2/Cu(II) photocatalytic system in solar pilot plant  Selective oxidation of benzyl alcohol to the corresponding aldehyde  Partial conversion of aldehyde to the corresponding benzoic acid  Copper regeneration  Evaluation of figure of merit *Highlights (for review)