Solar Energy 77 (2004) 503–512 www.elsevier.com/locate/solener
Solar photoreactors comparison based on oxalic acid photocatalytic degradation Erick R. Bandala a, Camilo A. Arancibia-Bulnes Claudio A. Estrada b,1 b
b,*,1
, Sayra L. Orozco
b,2
,
a Instituto Mexicano de Tecnologıa del Agua, Paseo Cuauhnahuac 8532, Jiutepec, Morelos 62550, Mexico Centro de Investigaci on en Energıa, Universidad Nacional Autonoma de Mexico, Apdo. Postal 34, Temixco, Morelos 62580, Mexico
Received 2 December 2003; received in revised form 9 March 2004; accepted 31 March 2004 Available online 4 May 2004 Communicated by: Associate Editor Sixto Malato-Rodrıguez
Abstract Solar heterogeneous photocatalytic degradation of oxalic acid in water is carried out in four different solar photoreactors: a parabolic trough concentrator (PC), a tubular collector (TC), a compound parabolic collector (CPC), and a V-trough collector (VC). The reactors operate under equal conditions of solar irradiance, collection surface and fluid flow rate to ensure a better comparison between the systems. The effects of TiO2 catalyst concentration and radiation incidence angle on the degradation are studied. Oxalic acid degrades without appreciable generation of intermediates, and a simple kinetic model is proposed to describe the process. There are differences in the degradation rates depending on the collector geometry. The CPC shows the best overall performance in terms of accumulated energy, followed closely by the VC. Incidence angle affects the total amount of energy collected but does not reduce very much the efficiency of the reactors to use this energy in the photocatalytic process. Ó 2004 Elsevier Ltd. All rights reserved. Keywords: Photocatalysis; Oxalic acid; Photoreactor; Solar collector
1. Introduction Photocatalytic detoxification is a promising technique for water remediation, which allows destroying a wide variety of pollutants that are difficult to treat by conventional methods. The use of solar irradiation to power this process is environmentally appealing, and has a potential to reduce the costs of this technology. In * Corresponding author. Tel.: +52-777-3250052; fax: +52777-3250018. E-mail address:
[email protected] (C.A. Arancibia-Bulnes). 1 ISESÒ members. 2 Present address: Unidad Academica de Ciencias Quımicas, Universidad Aut onoma de Zacatecas, Km 0.5 Carretera a Cd. Cuahutemoc, Guadalupe, Zacatecas 98600, Mexico.
addition to small scale studies (e.g., Jimenez et al., 2000; Arancibia-Bulnes et al., 2002a; Bandala et al., 2002), several pilot plant experiments have been performed in order to test the applicability of solar photocatalysis for wastewater depuration (Alpert et al., 1991; Minero et al., 1993; Gimenez et al., 1999; Malato et al., 1999, 2003). With scaling-up of the processes in mind, the comparison of results obtained in different types of solar photocatalytic reactors is an important issue, in order to optimize the degradation of pollutants. However, comparison between results obtained by different authors may not be possible, due to the variety of experimental conditions and variables considered in each case. This motivates the research on the direct comparison of photoreactors. There has been some work on the comparison of the performance of different solar photocatalytic reactors
0038-092X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.solener.2004.03.021
504
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
for the degradation of pollutants (Hilgendorff et al., 1992 ; Bahnemann et al., 1994; Blanco et al., 1994; Bockelmann et al., 1995; Curc o et al., 1996a; Goslich et al., 1997; Malato et al., 1997; Gimenez et al., 1999). The emphasis has been on the comparison of concentrating with non-concentrating solar reactors. For instance, Bockelmann et al. (1995) carried out a comparison of the degradation of dichloroacetic acid, as well as samples of wastewaters, between a parabolic trough reactor (concentrating) and a thin film fixed bed reactor (1 sun). They observed that both concentrating and nonconcentrating collectors had advantages and disadvantages, and it was not possible to decide unequivocally which system is better. More recently, reactors based on non-imaging collectors, like the compound parabolic collector, have attracted interest (Blanco et al., 1994). These reactors share some of the advantages of both parabolic troughs and non-concentrating reactors (Malato et al., 1997), which has been confirmed by several studies comparing the CPCs with other non-concentrating reactors as well as parabolic troughs (Curc o et al., 1996a; Malato et al., 1997; Gimenez et al., 1999). The more frequently quoted advantages of nonconcentrating CPC reactors are: the possibility of using solar UV radiation coming from all directions in the sky (global UV radiation), simplicity of construction and operation, turbulent flow regime (which improves mass transfer), and high reduction of the vaporization of volatile pollutants. However, all non-concentrating slurry reactors of the tubular type offer these advantages in principle, to different degrees. There are many possibilities for the geometry of the reflectors that illuminate the tubes. In order to gain a better understanding of how the reflector geometry affects the performance of the reactors, we have carried out the simultaneous comparison between four different solar reactors of tubular geometry. These reactors are a parabolic trough concentrator (PC), a compound parabolic collector (CPC), a tubular collector (TC; no reflectors), and a V shaped
trough collector (VC). This comparison considers the photocatalytic degradation of oxalic acid as a model pollutant. Oxalic acid, like other short chain carboxylic diacids, is among the frequently encountered intermediates of the degradation of aromatic compounds (Franch et al., 2002). Besides the intrinsic importance of this fact, oxalic acid degradation also provides a mean to compare different reactors in a simple basis; due to the lack of structural complexity of this substance, the degradation kinetics is much simpler than for most pollutants.
2. Methodology 2.1. Solar collector layout Four different solar collector geometries (parabolic trough, PC; V trough, VC; tubular, TC; and compound parabolic, CPC) were tested, as mentioned before. The PC reactor consists of a single glass tube of 2.54 cm diameter, mounted on the focal axis of a parabolic trough for a concentration ratio of 13 suns. The reflector is an aluminum sheet curved to a parabolic shape over a rigid steel structure. Both the CPC and the VC consist of a parallel row of eight glass tubes of 3 cm diameter. Each tube has a back reflector to enhance solar incidence. The reflectors were shaped from aluminum sheets as CPC and V troughs, respectively, with a concentration ratio of one sun. They are therefore, non-concentrating reflectors whose only function is to improve the distribution of solar irradiation around the tube walls. The fourth collector (TC) is a row of 14 parallel glass tubes without any kind of reflectors. All the collector areas were set to 0.72 m2 . For this purpose a portion of the area of an existing parabolic collector, originally built to a larger size, had to be covered with a non-reflecting material. Also, similar flow rates were used in all of them to ensure a better comparison. Fig. 1 shows the cross
beam radiation
VC
θ
collector normal
CPC
TC PC
Fig. 1. Cross sections of the different photoreactors tested, illustrating the incidence angle of beam radiation.
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
505
Table 1 Main parameters of the photoreactors Optical system
Acceptance half-angle
Mirror reflectance
Photorreactor ID (m)
Collection surface (m2 )
Total system volume (l)
Number of tubes
CP CV CPC CT
0.27° 90° 90° 90°
0.8 0.85 0.85 0.85
0.0254 0.03 0.03 0.03
0.72 0.72 0.72 0.72
10 10 10 10
1 8 8 14
sections of the four reactor configurations, and Table 1 presents some of the main parameters that characterize each reactor configuration used in this work. Except for the PC, all the collectors have acceptance half-angles of 90°. Therefore, they do not require tracking of the sun to operate. However, the four collectors were mounted together in a two axis (elevation and azimuth) solar tracking system. The objective of this was to ensure that the irradiance values impinging on each of the collectors were the same, to obtain the fairest possible comparison. Fig. 2 shows a schematic representation of the layout of the collectors on the suntracking system. Not all the collectors we employ are able to use the same fraction of solar radiation; The PC collects only beam UV radiation (directly coming from the solar disc), while the others collect also diffuse UV radiation (coming from all directions in the sky). Therefore, it is desirable to carry out radiation measurements which allow us to obtain the values of these two components separately. Two UV radiometers (Eppley, model TUVR), which work in the wavelength range between
295 and 385 nm, were mounted on the tracking system in order to measure radiation incident on the moving aperture of the collectors (Fig. 2). One of the radiometers measured global radiation (beam and diffuse), while the other measured diffuse radiation by blocking the beam component with a shading disk. Beam radiation was obtained by subtraction of the diffuse component from global UV radiation. In order to compare the experimental results from different days and collectors, the accumulated energy was used instead of time as the independent variable. The accumulated energy QUV [kJ/l] is defined (Curc o et al., 1996a) as the total amount of radiative energy reaching the collector from the beginning of an experiment up to a given time t, per unit solution volume VR QUV ¼
Z
t
ðAc GUV =VR Þ ds
ð1Þ
0
where Ac is the collection area, and GUV [W/m2 ] is the measured value of UV radiation (either global or beam). Two experiments with the same duration may have
Fig. 2. Schematic representation of the collector layout on the solar tracking system.
506
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
different values of this variable depending on the time of the year or weather conditions. Moreover, two reactors that do not collect the same component of radiation may reach different values of the accumulated energy in the same experiment. Therefore, this variable is more meaningful for solar photocatalysis than the duration of experimental runs. 2.2. Materials and reagents Reagent grade chemicals were used in all cases. Oxalic acid (C2 H2 O4 ) and sodium permanganate (KMnO4 ) were obtained from Baker, titanium dioxide (TiO2 , specific surface area >99% anatase, particle size 180 A, 23 m2 /g) from Aldrich and uranil salt (UO2 (NO3 )2 Æ 6H2 O, 98%) from Panreac. 2.3. Actinometric characterization The optical efficiency of each of the collectors was obtained from chemical actinometry experiments. In this method the amount of radiation reaching the reactor (which is the receiver of the solar collector) is evaluated by monitoring the degradation of a chemical reagent in a well characterized photoreaction. The actinometric pair employed in this work was the widely used uranil-oxalic acid actinometer, according to the procedure previously reported by Curc o et al. (1996b). The data obtained from these experiments were used for the calculation of the optical efficiency factor ð/eff Þ, according to the formulas presented by these authors. This factor is actually the ratio from the number of UV photons entering the reactor tubes to the number of photons falling into the collecting area. It accounts for optical losses, which depend on the reflectance and optical errors of the reflectors, and the transmittance of the glass tubes. 2.4. Photocatalytic experiments Experiments for the photocatalytic degradation of oxalic acid were performed simultaneously in the four collection geometries. Four catalyst concentrations, 0.05, 0.2, 0.5, and 1 g/l, were tested, in order to study the influence of this parameter on the reaction rate. The procedure followed during the experimental runs was the following: 10 l of milli-Q water were spiked with enough oxalic acid to get a final concentration of 10 mM. The resulting solution was circulated in the corresponding reactor, in the dark, for approximately 5 min. At this point, the catalyst was added and the suspension was circulated for another 5–7 min, which is long enough to reach adsorption equilibrium (Franch et al., 2002). The first sample (t ¼ 0) was taken after this period and irradiation started; the covers were retired simultaneously from all the solar collectors.
The sun-tracking system operated continuously during irradiation, and samples were taken every nine minutes from each reactor, in order to monitor oxalic acid concentration. This concentration was determined by volumetric titration with sodium permanganate. In a couple of experiments the total organic carbon (TOC) of the solution was also monitored by means of a TekmarDohrmann Apollo 9000 TOC Analyzer. Experiments at several radiation incidence angles (0°, 30°, 48° and 60°) were also carried out for the CPC, VC, and TC, to have an idea of the effect of this parameter on the performance of the reactors. This is important because non-concentrating collectors usually operate without sun tracking. The angle was measured with respect to the normal to the collector’s aperture, in a plane perpendicular to the tube receivers (Fig. 1). In each experiment this angle was kept constant, despite of the apparent movement of the sun in the sky, by means of an adequate manipulation of the sun tracker. A catalyst concentration of 0.5 g/l and an oxalic acid concentration of 10 mM were used in these runs. The general procedure followed in these experiments is the same as mentioned in the above paragraph.
3. Degradation kinetics The rate of change of a chemical species with time is one of the most relevant variables from the point of view of the design and scaling-up of reactors and processes. Several parameters that influence this rate can be taken into account by kinetic models. In particular, the mechanisms for the oxidation of organic compounds by TiO2 photocatalysis are known to involve the reaction with hydroxyl radicals. These radicals are produced by the interaction of water molecules with the photogenerated holes, and are known to react rapidly with most organic solutes. We will take into account the following steps, as proposed by Turchi and Ollis (1990): 1. Catalyst activation by radiation of adequate wavelength, forming electron–hole pairs. 2. Electron–hole recombination, which reduces the quantum efficiency of the process. 3. Reaction between water molecules and holes to produce hydroxyl radicals. 4. Reaction of electrons with oxygen, which reduces recombination rate. 5. Hydroxyl radical attack to the oxalic acid molecules. The above translates into the following set of reactions ka
TiO2 þ hvð P Ebg Þ!e þ hþ
ð2Þ
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512 kr
e þ hþ !heat
ð3Þ
ke
e þ O2 !O 2
ð4Þ
kh
hþ þ H2 O!OH þ Hþ
½eþ ¼
ka eL ke þ kr ½hþ
½hþ ¼ A þ
ð5Þ ½OH ¼
ki;1
C2 H2 O4 þ OH !Rint ki;2
Rint þ OH !CO2 þ H2 O kd
C2 H2 O4 þ OH !2CO2 þ H2 O þ H
ð14Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi A2 þ beL
kh ½hþ ki;1 ½Ox þ ki;2 ½Rint þ kd ½Ox
ð15Þ
ð16Þ
ð6Þ ð7Þ ð8Þ
where Rint is any possible reaction intermediate appearing in the degradation of oxalic acid. The following assumptions are made to simplify the kinetic equations: 1. The reaction rate is not limited by the concentration of TiO2 , H2 O, or O2 , which are considered to be in excess in the suspension. 2. It is possible to consider a steady state for the concentrations of unstable species like OH , holes, and electrons. 3. The properties of the catalyst remain unchanged. The first of these conditions has been discussed by Turchi and Ollis (1990), while the last two are reasonable for catalytic processes, particularly when there is no poisoning of the catalyst. Considering the preceding reaction scheme and taking into account assumption 1, a mass balance produces the following equations d½eþ ¼ ka eL kr ½eþ ½hþ ke ½eþ dt
ð9Þ
d½hþ ¼ ka eL kr ½eþ ½hþ kh ½hþ dt
ð10Þ
d½OH ¼ kh ½hþ ki;1 ½OH ½Ox ki;2 ½OH ½Rint dt kd ½OH ½Ox
507
where A ¼ ke =2kr and b ¼ ka ke =kh kr . Substituting Eqs. (15) and (16) in Eq. (12) pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d½Ox Ko ½Ox ðA þ A2 þ beL Þ ¼ kh dt Ko ½Ox þ ki;2 ½Rint ð17Þ where Ko ¼ ki;1 þ kd . The expression (17) describes a degradation kinetics that follows a Langmuir–Hinshelwood equation. This kinetics is usually associated with an adsorption of the reactant in the catalyst surface prior to the reaction. However, in the present case is obtained due to the assumption of an intermediate compound in the degradation. In principle, the above expression is valid on a local basis, while global concentrations are measured in the experiments. If we assume perfect mixing, we can consider the concentrations independent of location. However, such an assumption is not possible for the LVREA eL , which tends to have a highly inhomogeneous distribution in the reactor. Therefore, it is necessary to average Eq. (17) over the system volume in order to obtain a global kinetic expression; i.e. ! d½Oxg Ko ½Oxg ¼ kh dt Ko ½Oxg þ ki;2 ½Rint g Z Z Z pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ð18Þ ðA þ A2 þ beL Þ dV VR VR
ð11Þ
d½Ox ¼ ki;1 ½OH ½Ox kd ½OH ½Ox dt
ð12Þ
d½Rint ¼ ki;1 ½OH ½Ox ki;2 ½OH ½Rint dt
ð13Þ
where eL is the local volumetric rate of energy absorption (LVREA); i.e., the amount of photons moles absorbed for unit time and unit volume. [Ox] stands for oxalic acid concentration. By taking into account assumption 2 and carrying out some algebra, we obtain
In general, it is not possible to carry out this integral without knowing the distribution of the LVREA, and the determination of this distribution is an involved mathematical task (e.g., Arancibia-Bulnes et al., 2002a; Arancibia-Bulnes and Cuevas, 2004). Nevertheless there is one case, where the integral can be estimated in a simple fashion, which is when the integrand can be approximated by a linear function of eL . This is the case observed for low radiation levels, typically for nonconcentrating or low concentration collectors. More precisely, this occurs when beL A2 , which does not only depend on radiation levels but also on the kinetic constants of the involved reactions, through the parameters A and b. Because of this dependence, it is not possible in general to determine a definite value of radiation intensity for the crossover from linear to nonlinear behavior. The hypothesis of linear behavior can
508
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
only be tested by comparison with the experimental data. Assuming that the linear dependence is valid in the present case (Alfano et al., 1997) A þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b A2 þ beL ffi eL 2A
ð19Þ
By using this expression, the integral in Eq. (18) becomes proportional to the total amount of radiation absorbed in the reactor. To simplify the analysis we assume this quantity proportional to the amount of radiation reaching the reactor /eff Ac GUV [W], as measured by a radiometer. This approximation is a rough simplification, which ignores the complex optical processes undergone by radiation inside the reaction space. Consequently, the proposed approximation overestimates the absorbed radiation to a degree that depends on catalyst concentration (Brandi et al., 2000; ArancibiaBulnes et al., 2002a; Salaices et al., 2002). However, the analysis of these processes is an involved task (Arancibia-Bulnes and Cuevas, 2004), which is not the goal of the present work. Taking into account the above, we can express Eq. (18) as ! d½Oxg Ko ½Oxg ¼K /eff Ac GUV =VR dt Ko ½Oxg þ ki;2 ½Rint g ð20Þ where Ac is the collector area and VR is the reactor volume, and K is a phenomenological lumped kinetic constant. As we will see in the following section, no appreciable intermediate formation occurs in the degradation of oxalic acid. Direct mineralization with the formation of two CO2 molecules is the dominant mechanism. When no intermediates are present (½Rint g ¼ 0), which is a particular case of the reaction scheme (2)–(8), the processes of Eqs. (6) and (7) can be neglected and the kinetic equation (20) becomes
d½Oxg ¼ KAc /eff GUV =VR dt
ð21Þ
The integration of this equation with respect to time gives a linear relationship between the instantaneous oxalic acid concentration and accumulated energy ½Oxg;0 ½Oxg ¼ K/eff QUV
ð22Þ
4. Results 4.1. Actinometric characterization Table 2 shows the values of the efficiency factor (/eff ) determined for each of the photoreactors by means of
Table 2 Results of the actinometric characterization of the solar reactors for different incidence angles, optical efficiency factor (/eff ) Reactor
/eff (h ¼ 0°)
/eff (h ¼ 30°)
/eff (h ¼ 48°)
PC VC CPC TC
0.47 0.36 0.41 0.44
– 0.32 0.39 0.44
– 0.29 0.34 0.34
the method proposed by Curc o et al. (1996a,b). For the case of the PC the efficiency determination was performed only at normal irradiation, because this geometry is able to focus only beam radiation at zero incidence angle. The efficiency of this collector is calculated with respect to the number of photons impinging on its aperture due to beam radiation only; so, a high efficiency refers only to the ability to make use of this component of radiation. At first sight, the values from Table 1 may appear too low. However, we must take into account that neither the reflectance of the aluminum concentrators, nor the transmittance of glass tubes, is very high in the UV region. The highest efficiency factor at normal incidence corresponds to the PC, followed by the TC, CPC, and VC, in that order. Among the tested non-concentrating reactors, the TC has the highest optical efficiency, due to the absence of a mirror in this system. This avoids loses in the rays reaching the reactor tubes through reflection in the mirror. We must take into account however, that the efficiency factor for this reactor has been calculated with respect to the radiation incident in the tubes only, ignoring the optical losses that occur in the gaps separating the tubes. We think this is the correct procedure, because the gaps can be designed to be arbitrarily small, and therefore these losses are not an intrinsic characteristic of the TC geometry. For all the collectors the value of /eff decreases when the incidence angle increases. For instance, in the case of CPC, the efficiencies for incidence angles 0°, 30°, and 48°, are 0.41, 0.39, and 0.34, respectively; an overall variation of 17%. The VC and TC show variations around 19% and 22%, respectively. The variation of optical efficiency with incidence angle reflects the ability of each of the reactors to collect solar energy coming from different directions. All of the non-concentrating reactors, make use of global radiation; i.e., they accept in principle radiation coming from the whole hemisphere above them. However, their acceptance is not the same for all angles. In principle, the CPC has the best acceptance function from any direction, except for a small portion of radiation at very inclined angles, which is lost due to surface errors (Rabl, 1976). Therefore, its efficiency does not vary very rapidly with incidence angle, as observed in Table 2. The VC
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
4.2. Photocatalytic degradation As described in Section 2.4, the degradation of oxalic acid was carried out simultaneously in the four collectors, for different catalyst concentrations and beam radiation incidence angles. Fig. 3 shows an example of the variation of the molar concentration of oxalic acid as a function of the accumulated energy, for 0.5 g/l of catalyst. We want to point out that no variation in the oxalic acid concentration was determined in the absence of TiO2 . Once this catalyst was added, even in small amounts, important variations in the substrate concentration were detected. As shown in Fig. 3, the performance of the four collectors is quite similar in terms of accumulated energy. It is possible to observe a slight advantage for the CPC, while the TC is the collector that produced the least degradation. In all cases, the collectors behave in a similar way relative to each other, but the degradation rate increases with catalyst concentration. As mentioned earlier, total organic carbon was measured in a couple of experiments, in addition to oxalic acid concentration. We compared the TOC data with the amount of carbon present in the form of oxalic acid, calculated from the measured concentration of the substrate. We found that the organic carbon present in the form of oxalic acid does not deviate from the TOC by more than 5%, which is attributable to the uncertainty of the method used to determine oxalic acid concentration. Moreover, the evolution of both quanti-
ties follows the same trend. This led us to conclude that all the organic carbon present in the samples comes from oxalic acid, without appreciable formation of degradation intermediates. This direct degradation of oxalic acid is not surprising, because of the simple chemical structure of this molecule. This coincides with the results of Franch et al. (2002) who observed, working with a lamp reactor, that direct mineralization of oxalic acid was the predominant process. Formic acid at very low concentrations was the only intermediate detected by these authors. According to the previous discussion, we will fit the experimental data by using the kinetic model of Eq. (22), which does not consider the presence of intermediates. Figs. 4 and 5 show two typical examples of this fitting. The majority of cases are similar to Fig. 4, where the data correlates very well with a linear function. Nevertheless, some cases behave like Fig. 5, where a straight
12 10
[Ox] (mM)
acceptance function has lower values as compared to the CPC, and also its angular dependence is not so regular. The tubular collector, not having a reflector, accepts the same amount of radiation from any direction, except at large incidence angles, where shadowing of beam radiation by neighboring tubes starts to play a role.
8 6 4 2 0 0
3
6
9
12
15
Q UV (kJ/litre) Fig. 4. Degradation of oxalic acid in the TC reactor, for a 1 g/l initial TiO2 concentration (points); fitting of these results with the kinetic model (line).
12
12
VC
10
PC
CPC
TC
10
8
[Ox] (mM)
[Ox]g (mM)
509
6 4
8 6 4 2
2
0
0 0
10
20
30
40
Q UV (kJ/ litre) Fig. 3. Variation of oxalic acid concentration as a function of accumulated energy for different solar reactors. The catalyst concentration was 0.5 g/l.
0
5
10
15
20
25
30
Q UV (kJ/litre) Fig. 5. Degradation of oxalic acid in the CPC reactor, for a 0.05 g/l initial TiO2 concentration (points); fitting of these results with the kinetic model (line).
510
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
line can be fitted with an acceptable correlation only to the first part of the process. We calculate the lumped kinetic constant K as the slope of the linear fits to the experimental data, as illustrated in the previous graphs. The values obtained for each collector, as a function of catalyst concentration cp , are presented in Fig. 6. These data exhibit two different kinds of behavior: while the kinetic parameter increases monotonically with concentration for the PC and TC reactors, it reaches a maximum at 0.5 g/l of TiO2 for the CPC and VC. It is interesting to note that, despite of the fact that the highest values of /eff were determined for the PC and the TC, the CPC and VC surpassed the performance of these systems for the photocatalytic degradation of oxalic acid. However, let us recall that the optical efficiencies were obtained by chemical actinometry, which involves a homogeneous photoreaction. We cannot expect that such efficiencies determine which of the reactors is better to carry out heterogeneous photoreactions, where the optical phenomena are considerably more involved (Arancibia-Bulnes and Cuevas, 2004). The poor performance obtained from the PC and TC, as compared to their relatively high optical efficiencies /eff is very likely a consequence of the way these collectors distribute radiation around the walls of the reactor tubes. While the CPC and the VC tend to create a more even distribution of radiation in the tube wall, the other two collectors have one side of the tube much better irradiated than the other; the rear side in the case of the PC, and the front side for the TC. As it is clear from Fig. 6, the parameter K is not a true kinetic constant. This is because this parameter does not only contain kinetic information, but also other information not explicitly considered in the kinetic model. Actually, K is roughly proportional to the fraction of radiation absorbed by the reactor, which is always less than one, due to the pronounced radiation
backscattering effects produced by the catalyst particles (Brandi et al., 2000; Arancibia-Bulnes et al., 2002a; Salaices et al., 2002). The fraction of absorbed radiation is dependent on catalyst concentration: it increases rapidly with this concentration at first, and converges asymptotically to a maximum value for large catalyst loads (Arancibia-Bulnes et al., 2002a). This variation could explain in part the dependence of the obtained values of K with cp . However, to make detailed radiation transfer calculations for the different reactors is out of the scope of the present work. The catalyst concentration corresponding to the maxima observed for the CPC and VC optimizes the degradation process, according to Eq. (22), by using less accumulated energy to carry out the degradation. These maxima occur around 0.5 g/l, which agree with the results for carbaryl degradation in a CPC (ArancibiaBulnes et al., 2002b). No maxima are observed for the PC in this region, contrary to the situation with carbaryl (Arancibia-Bulnes et al., 2002a) or DBSNa (Jimenez et al., 2000), for instance. 4.3. The effect of the radiation incidence angle Figs. 7 and 8 show the influence of the radiation incidence angle on the overall performance of the TC and CPC, respectively, for the photocatalytic degradation of oxalic acid. The results obtained with the VC are not shown, but are very close to those obtained with the CPC (e.g., Fig. 3). As we can see, the degradation curves for different incidence angles tend to overlap for a given reactor; there is little difference in the slope of the different curves. In particular, in the case of the TC reactor the only significant difference among the results for different angles is the amount of energy collected. As the angle increases, the value of accumulated energy reduces, which is a consequence of the cosine law that applies to the
1.6
10
1.2
[Ox]g (mM)
K (mM · litre/kJ)
12
0.8 CPC VC 0.4
PC
60°
48°
30°
0°
8 6 4 2
TC
0
0 0
0.2
0.4
0.6
0.8
1
1.2
c p (g/litre) Fig. 6. Lumped kinetic constant for the different reactors, as a function of catalyst concentration (lines are drawn for visualization purposes only).
0
10
20
30
40
Q UV (kJ/litre) Fig. 7. Degradation of oxalic acid in the TC reactor as a function of accumulated energy, for different incidence angles of solar radiation.
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512 12
60°
[Ox]g (mM)
10
48°
30°
0°
8 6 4 2 0 0
10
20
30
40
Q UV (kJ/litre) Fig. 8. Degradation of oxalic acid in the CPC reactor as a function of accumulated energy, for different incidence angles of solar radiation.
511
kinetic parameter are compared. The CPC and VC collectors have very similar values of this parameter as a function of TiO2 concentration, with an initial increase and a maximum for 0.5 g/l. On the other hand, the PC and TC present much smaller kinetic parameters that increase monotonically with concentration. In this sense, we can conclude that the CPC has the best performance in the present conditions, followed by the VC. Both share the characteristic of being non-concentrating collectors with geometry intended to produce a more or less homogeneous illumination of the perimeter of the reactor tubes, although the VC geometry is by no means optimal. The photocatalytic process was also carried out by defocusing the sun-tracking system, but changes in the reaction rates were not very important; the reactors collected less energy because of oblique incidence of beam solar radiation, but the degradation is roughly the same for a given amount of energy.
beam component of solar radiation. Clearly, a higher conversion of the substrate is achieved in the same period when the accumulated energy is larger. Nevertheless, the reactor uses the energy with very similar efficiency regardless of incidence angle. Larger differences appear in the case of the CPC and VC reactors, where the slope of the curves seems smaller for the intermediate angles. For radiation incident at angles different from zero linear fits can also be performed to the experimental data. The values of the lumped kinetics constants so obtained do not show a clear trend with incidence angle, rather it appears as if they were independent of angle with only statistical variations.
The authors want to acknowledge J.J. Qui~ nones Aguilar for technical support in the setup of the experiments, operation of the sun-tracking system, and radiation measurements. We also acknowledge O.M. Alfano and B. Serrano for useful discussions. S.L. Orozco was financially supported by Academia Mexicana de Ciencias within the program Veranos en la Ciencia. This work was partially supported by CONACYT (Mexico), under grant 37636-U.
5. Conclusions
References
The photocatalytic degradation of oxalic acid in an aqueous suspension of TiO2 particles was carried out in four solar photoreactors with different collector geometries: parabolic (PC), compound parabolic (CPC), Vtrough (VC) and tubular (TC). The reactors were mounted in a sun-tracking system to ensure equal irradiation conditions for all of them. A kinetic model is proposed, which predicts first order dependence of the degradation rate in accumulated energy and zero order in oxalic acid concentration. This model correlates very well with the experimental results, except in some cases for the last part of the process. From fits to the model, a lumped kinetic parameter, which is the constant of proportionality between degradation rate and accumulated energy, was obtained as a function of catalyst concentration. The four reactors exhibit similar degradation levels as a function of accumulated energy. However, larger differences are evident when the values of the lumped
Alfano, O.M., Cabrera, M.I., Cassano, A.E., 1997. Photocatalytic reactions involving hydroxil radical attack. I. Reaction kinetics formulation with explicit photon absorption effects. J. Catal. 172, 370–379. Alpert, D.J., Sprung, J.L., Pacheco, J.E., Prairie, M.R., Reilly, H.E., Milne, T.A., Nimlos, M.R, 1991. Sandia National laboratories’ work in solar detoxification of hazardous wastes. Solar Energy Mater. 24, 594–607. Arancibia-Bulnes, C.A., Bandala, E.R., Estrada, C.A., 2002a. Radiation absorption and rate constants for carbaryl photocatalytic degradation in a solar collector. Catal. Today 76, 149–159. Arancibia-Bulnes, C.A., Bandala, E.R., Estrada, C.A., 2002b. Radiation absorption in parabolic trough and CPC solar photocatalytic reactors. In: Steinfeld, A. (Ed.), Proceedings of the 11th Solar Paces Symposium on Concentrated Solar Power and Chemical Energy Technologies, Zurich, Switzerland, pp. 445–451. Arancibia-Bulnes, C.A., Cuevas, S.A., 2004. Modeling of the radiation field in a parabolic trough solar photocatalytic reactor. Solar Energy 76, 615–622.
Acknowledgements
512
E.R. Bandala et al. / Solar Energy 77 (2004) 503–512
Bahnemann, D.W., Bockelmann, D., Goslich, R., Hilgendorff, M., 1994. Photocatalytic detoxification of polluted aquifers: novel catalysts and solar applications. In: Helz, G.R., Zepp, R.G., Crosby, D.G. (Eds.), Aquatic and Surface Photochemistry. Lewis Publishers, Boca Raton, pp. 349–367. Bandala, E.R., Gelover, S., Leal, M.T., Arancibia-Bulnes, C., Jimenez, A., Estrada, C.A., 2002. Solar photocatalytic degradation of aldrin. Catal. Today 76, 189–199. Blanco, J., Malato, S., Bahnemann, D., Bockelmann, D., Weichgrebe D., Goslich R., 1994. Effective industrial waste water treatment by solar photocatalysis; application to fine chemicals spanish company. In: Proceedings of the 7th International Symposium on Solar Thermal Concentrating Technologies, vol. 3, Moscow, Russia, pp. 468–481. Bockelmann, D., Weichgrebe, D., Goslich, R., Bahnemann, D., 1995. Concentrating versus non-concentrating reactors for solar water detoxification. Solar Energy Mater. Solar Cells 38, 441–451. Brandi, R.J., Alfano, O.M., Cassano, A.E., 2000. Evaluation of radiation absorption in slurry photocatalytic reactors. 1. Assesment of methods in use and new proposal. Environ. Sci. Technol. 34, 2623–2630. Curc o, D., Malato, S., Blanco, J., Gimenez, J., Marco, P., 1996a. Photocatalytic degradation of phenol: comparison between pilot-plant-scale and laboratory results. Solar Energy 56, 387–400. Curc o, D., Malato, S., Blanco, J., Gimenez, J., 1996b. Photocatalysis and radiation absorption in a solar plant. Solar Energy Mater. Solar Cells 44, 199–217. Franch, M.I., Ayll on, J.A., Peral, J., Domenech, X., 2002. Photocatalytic degradation of short-chain organic diacids. Catal. Today 76, 221–233. Gimenez, J., Curc o, D., Queral, M.A., 1999. Photocatalytic treatment of phenol and 2,4-dichlorophenol in a solar plant in the way to scaling-up. Catal. Today 54, 229–243.
Goslich, R., Dillert, R., Bahnemann, D., 1997. Solar water treatment: principles and reactors. Water Sci. Technol. 35, 137–148. Hilgendorff, M., Bockelman, D., Nogueira, R.F.P., Weichgrebe, D., Jardim, W.F., Bahnemann, D., Goslich, R., 1992. Photocatalytic decomposition of aliphatic halogenated hydrocarbons in water: laboratory studies and solar applications. In: Proceedings of the 6th International Symposium on Solar Thermal Concentrating Technologies, vol. 2, Mojacar, Spain, pp. 1167–1181. Jimenez, A.E., Estrada, C.A., Cota, A.D., Roman, A., 2000. Photocatalytic degradation of DBSNa using solar energy. Solar Energy Mater. Solar Cells 60, 85–95. Malato, S., Blanco, J., Richter, C., Curco, D., Gimenez, J., 1997. Low concentrating CPC collectors for photocatalytic water detoxification: comparison with a medium concentrating solar collector. Water Sci. Technol. 35, 157–164. Malato, S., Blanco, J., Richter, C., Milow, B., Maldonado, M.I., 1999. Solar photocatalytic mineralization of commercial pesticides: methamidophos. Chemosphere 38, 1145–1156. Malato, S., Blanco, J., Vidal, A., Alarc on, D., Maldonado, M.I., Caceres, J., Gernjak, W., 2003. Applied studies in solar photocatalytic detoxification: an overview. Solar Energy 75, 329–336. Minero, C., Pelizzetti, E., Malato, S., Blanco, J., 1993. Large solar plant photocatalytic water decontamination: degradation of pentachlorophenol. Chemosphere 26, 2103–2119. Rabl, A., 1976. Optical and thermal properties of compound parabolic concentrators. Solar Energy 18, 497–511. Salaices, M., Serrano, B., de Lasa, H.I., 2002. Experimental evaluation of photon absorption in an aqueous TiO2 slurry reactor. Chem. Eng. J. 90, 219–229. Turchi, C., Ollis, D., 1990. Photocatalytic degradation of organic water contaminants: mechanism involving hydroxyl radical attack. J. Catal. 122, 178–192.
