You are on page 1of 8

Ind. Eng. Chem. Res.

2007, 46, 7637-7644

7637

A Rational Approach to the Design of Photocatalytic Reactors


Giovanni Camera Roda* and Francesco Santarelli
Dipartimento di Ingegneria chimica, mineraria e delle tecnologie ambientali, UniVersita degli studi di Bologna, Viale Risorgimento 2, I-40136 Bologna, Italy

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

A model to assess the performances of an annular photocatalytic reactor has been developed by investigating different operational conditions through an analysis based on the proper dimensionless parameters (namely, the optical thickness, the Thiele modulus and two among the well-known Damkohler and Peclet numbers, which are properly redefined). Different dependences of the reaction kinetics on the local rate of radiant energy absorption are also considered. Because the progress of the reaction is affected by the radiation field, all these parameters are dependent on the catalyst concentration and then indirectly on the catalyst load. The conditions under which an optimal value for the catalyst concentration may exist are determined, thus contributing to provide insight to one aspect that is quite controversial in the literature.
Introduction It is expected that, soon, photocatalytic processes will be widely used as an effective tool for the treatment of water that has been polluted by traces of toxic and/or persistent chemicals.1 Therefore, a large amount of attention has been given to studying the kinetics of the degradation of the organic pollutant substrate, as well as investigating the parameters that affect the performance of the systems (lamp/reactor) used. Results are undoubtedly interesting but suffer from a lack of generality,2,3 because they are presented in terms of dimensional parameters, which are significant to the specific situations investigated and, consequently, are valid under the same limitations. More-general criteria then are needed to assess the role that the operational variables have in determining and possibly optimizing the performance of a reactor. A significant example of the previous statement is given by the different conclusions presented by the authors who investigated, in slurry systems, the effects of catalyst load on the conversion and possible existence of an optimal value for this quantity. According to a large number of authors,3-11 the rate of removal of the substrate increases monotonically with the catalyst load and approaches an asymptotic value. The value of the catalyst load at which the asymptotic behavior appears is obviously dependent on the specific situation investigated; however, in any case, it is related to situations where a strong attenuation of the radiation field occurs within the reactor. In other experiments and works,2,3,11-19 a maximum of the rate of removal results at a specific value of the catalyst load. This behavior is generally justified by the significant attenuation of the radiation by the photocatalyst particles when their concentration is increased beyond a given value. This shielding effect might thwart the positive effect of an increase of the available catalyst sites, obtained by increasing the catalyst load. However, the maximum is strangely observed by the various researchers at catalyst concentrations that give very dissimilar attenuations of the radiation in the reactor, and this disagreement has not yet been justified. In a similar way, the dependence on the particular investigated situation limits the general validity of the analysis of the effects,
* To whom correspondence should be addressed. E-mail address: giovanni.camera.roda@mail.ing.unibo.it.

which are caused by (i) the flow rate of the stream to be treated2,20 and (ii) the order of the reaction, with respect to the local rate of radiant energy absorption. The discrepancies that emerge in the conclusions of the cited investigations are not considered as a sign of inaccuracy but are simply the outcome of attention to operational conditions with a different relative weight of the relevant parameters. Therefore, the present contribution is an effort to investigate the behavior of a photocatalytic reactor on the basis of the dimensionless parameters that are currently used in the transport phenomena and/or in the reaction engineering analysis to find answers to some open questions. Mathematical Model: Basic Equations A photocatalytic reaction has been considered to occur within an annular reactor with a fluorescent UV-A lamp placed on its axis. Because of the axial symmetry of the investigated situations, a two-dimensional problem results for both the radiation and the pollutant concentration fields. The geometry of the system can be represented by the following dimensionless variables:

k)

Rint R Rlamp R L R

R* ) lamp L* )

where Rint is the internal radius of the annular region, R the external radius of the annular region, Rlamp the radius of the emitting lamp, and L the length of the irradiated annular reactor. The investigation has been developed through the following steps of the analysis: (a) radiative field and distribution of the rate of radiant energy absorption within the reactor; (b) conversion per pass when the reactor operates in a continuous mode; and (c) overall conversion when the reactor is inserted in a recycle/ batch system. Radiative Field and Distribution of the Rate of Radiant Energy Absorption. It is well-known that a peculiar feature of the photocatalytic processes is the essential role of the

10.1021/ie070302a CCC: $37.00 2007 American Chemical Society Published on Web 06/09/2007

7638

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

radiation field as an inherent element of the process. This feature is common to all the photochemical processes and requires a specific treatment that has a scaring effect on most of the people interested in photochemical and photocatalytic reactions and is often handled with simplified and questionable approaches. A rigorous analysis of the radiative transfer can be developed starting from the basic treatment presented in the textbooks on this subject21,22 or from the comprehensive analysis of the role of radiative transfer in photochemical processes.23-26 In the case of photocatalytic processes, the radiation field is dependent on the catalyst concentration after the geometry of the reactor and the location and the emission properties of the lamp are assigned. Even if the catalyst load is currently used to account for the radiation absorption, its intensive value, the catalyst concentration, is definitely more appropriate for this purpose, because it enters the constitutive equations for the optical properties of the participating particles. After the radiative transfer equation (RTE) has been solved and the distribution of the radiation intensity has been obtained, it is possible to determine the local rate of radiant energy absorption (e) as

The following dimensionless variables have been defined:

C* ) A

CA CA1

where the characteristic concentration, CA1, is the concentration at the inlet of the reactor;

(e)* )

e e ) e I0 0

where I0 is the reference intensity of the radiation, which is defined through the UV-A radiant power P emitted by the lamp as I0 ) P/(2RintL);

z* ) r* )

z R r R R D

V* ) V

e )

4 I d

where represents any direction of the travelling radiation. The relevant optical parameters that affect the radiative transfer are (1) the optical thickness, given as ) R(1 - k), where ) + is the extinction coefficient (here, is the absorption coefficient, and is the scattering coefficient), (2) the single scattering albedo, 0 ) /, and (3) the phase function p( f ) for elastic scattering from the -direction to the -direction. After the catalyst has been chosen, the albedo 0 and the phase function are assigned, because the absorption and scattering coefficients (and, then, the total extinction coefficient ) are dependent linearly on the catalyst concentration.27 Therefore, the effects on the photons distribution due to the catalyst load can be investigated more properly, only in terms of the optical thickness, because this parameter accounts for the combined effect of the catalyst concentration (and then, indirectly, the catalyst load) and of the dimensions of the reactor.28 Continuous Annular Photoreactor-Conversion per Pass. Two flow conditionss fully developed laminar flow and turbulent flowshave been considered for the annular photoreactor when it operates in a continuous mode, assuming that the reaction rate is given by the kinetic equation

where D is the diffusivity of the organic pollutant substrate. Thus, the dimensionless mass balance equation can be written as

C* 1 A 1 C* r* + (r*V* A) + (V* ) ) rC* r* r* z* z A r* r* r* C* A - 2(e*)RC* (2) A z* z*


where

( ) ( )

V*) 2 z

( )
L* Da
2

1 - (r*)2 +

1 + k2 V*) 0 r

[ ] [ ]
1 - k2 ln(1/k)

1 - k2 ln(r*) ln(1/k)

(3)

(3)

Equation 2 has been solved for the following boundary conditions:

C* ) 1 A C* A )0 z* C* A )0 r* C* A )0 r*

(at z ) 0 and k < r* < 1) (at z ) L* and k < r* < 1)

R ) K(e)RCA

(1)

(at r ) k and 0 < z* < L*) (at r ) 1 and 0 < z* < L*)

where e is the local rate of radiant energy absorption (here, e will be considered to be due to a monochromatic radiation or representative of the value averaged in a range of wavelengths), R is the order of the reaction (with respect to e), and CA is the substrate concentration. The assumption of a firstorder reaction, with respect to the pollutant concentration, is consistent with the low concentration values that are typical of most of these processes. (a) Laminar Flow in the Reactor. When the Reynolds number (which is defined as Re ) [2VzR(1 - k)F]/) is <2000, the flow in the reactor is laminar.

Two dimensionless groups appear in the previous equations, namely, the Damkohler number,

Vr Da ) K(e)R V
(where Vr is the volume of the reactor and V ) VzR2(1 - k2)

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007 7639

is the volumetric flow rate), and the Thiele modulus,

)R

K(e )R 0 D

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

The choice of the two intervening groups is not unique, because, for instance, the Damkohler number Da, the Thiele modulus , and the mass Peclet number (defined as Pem ) Re(/D) ) Re Sc) are linked each other through the relationship Da Pem ) 22(1 - k)L*. (b) Ideal Case of Plug-Flow Reactor. When the Re value is high, the assumption of laminar flow does not hold any longer and a plug-flow reactor can be considered with a distribution of e, as it is given by the solution of the RTE. Indeed, at high values of Re, a negligible radial gradient of the substrate concentration and of the velocity can be assumed, because the mass boundary layer and the velocity boundary layer are both very thin.29 Also, the transport by diffusion in the z-direction can be safely neglected, with respect to the convective transport. In this case, the mass balance equation for substrate A is

Figure 1. Schematic setup of the batch/recycle system.

where

(e)* ) A)

e e 0

c e 0 KVr V

dCA ) dz

R kR R2r dr

(4)

Da )

Assuming the kinetic equation as given in eq 1, the integration of eq 4, with the inlet condition (at z ) 0) CA ) CA1, gives the outlet concentration CA2 as

CA2 ) CA1 exp or, in dimensionless form,

K V

V (e )R dV]
r

CA2 CA1

Da ) exp (1 - k2)L*

V*(e *)R dV*


r

]
(5)

It must be observed that the conversion, P ) 1 - (CA2/CA1), is a function of Da, k, L*, and the e distribution. A More-General Kinetic Equation. A more-general equation for the reaction rate can be written as

Recycle/Batch System: Transient Behavior with the Reactor in Pseudo-Steady State. The dynamic behavior of a recycle/batch system (see Figure 1) has been investigated, assuming that the reactor is operating in a pseudo-steady state. Under this assumption, because the reaction rate is first order with respect to the substrate concentration, the conversion per pass in the reactor is independent of the inlet concentration and, consequently, does not change during the process. The condition of pseudo-steady state for the reactor can be safely assumed if the residence time in the reactor (tr) is much shorter than the characteristic time of disappearance of the substrate from the system (td). Because of the fact that these two characteristic times can be estimated as

R)K

(ee c)C +

tr )
and

Vr V

This equation accounts for a continuous change of the order of reaction,28 which is 1 at a low level of the intensity of radiation (c . e) and, through a value of 0.5 at intermediate levels of the intensity, reaches the limiting value of 0 at a high level of intensity (c , e). This occurrence has been experimentally observed and discussed by many researchers,9,28,30,31 so that, even if the underlying reasons are still under discussion, it can be stated that eq 5 is able to represent at least the phenomenological behavior of many real systems. In this case, the mass balance equation in the reactor, in the hypothesis of no radial gradient of the substrate concentration or of the velocity, becomes

td ) )

CA0(Vr + Vt) rate of disappearance of A from the system CA0(Vr + Vt) K(e )RCA0Vr 0 V r + Vt K(e )RVr 0

dCA ) -KCA V dz

R kR

e 2r dr e + c

(6)

where Vt is the volume of the perfectly mixed tank and CA0 is the initial concentration of the substrate in the system, the reactor works in pseudo-steady state if

Integrating this equation with the boundary condition that, at z ) 0, CA ) CA1, the conversion turns out to be

Da P ) 1 - exp (1 - k2)L*

V*(e )* + A dV*
r

(e)*

0 tr K(e )RVr Vr ) td V Vr + Vt ) Da

1 ,1 1 + Vt/Vr

7640

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007

i.e., it is required that Da , 1 and/or Vr/Vt , 1. Note that these latter inequalities are very often satisfied in real cases. The mass balance of A in the stirred tank is

Vt

dCA1 ) V(CA2 - CA1) dt

(7)

In pseudo-steady state and in the hypothesis that the volume inside the tubes is negligible, with respect to Vt, the concentration CA2 is given, at any time, by CA2 ) CA1 (1 - p), where p is the conversion per pass in the reactor. Introducing the dimensionless time, t*) t R2/, the mass balance equation becomes

dCA1
Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

dt*

) -KdCA1

(8)
Figure 2. Radial distribution of e* at z* ) L*/2, at different values of .

where Kd is the dimensionless constant of the disappearance of A in the system (Kd ) {Re/[2(1 - k)L*]}(Vr/Vt)p). Therefore, only one additional parameter, the ratio Vr/Vt, must be considered to characterize the transient behavior. The result of the integration of the mass balance equation with the initial condition that CA1 ) CA0 at t ) 0 is

C A1 C A0

) exp(-Kdt*)

Discussion of the Results The values of the geometrical dimensionless variables have been assumed, with reference to an experimental annular reactor available in the authors laboratory, to be as follows:

R ) 20.5 mm R* ) lamp k) 8 20.5


Figure 3. Normalized absorbed radiant power, as a function of .

12 20.5 265 20.5

L* )

the absorbed photons affects the local rate of disappearance of the material reactant, it turns out that, when a large nonuniformity occurs in the distribution of e (i.e., for large values of ), the diffusion of the reacting species may have a significant role. After the effect of the optical thickness on the radiation distribution has been evaluated, the influence of on the conversion per pass has been investigated for two values of R (R ) 0.5 and 1) and for the two flow regimes considered. In Figure 4, the conversion per pass is given, relative to the optical thickness , at various values of the Reynolds number (Re) for R ) 0.5 and 1. Note that, to make the role of the flow regime more evident, the more familiar Re has been assumed to be one of the two independent dimensionless groups upon which the solution of the problem is dependent. This choice is consistent with the possibility outlined in the previous section and considering that Re and Pem differ by a multiplying constant, as the Schmidt number (Sc) is constant for a given fluid. The case of Re ) 10000 has been considered to be representative of turbulent flow conditions. Because accounts for the catalyst concentration that, through e affects also Da and , it turns out that, if the geometrical 0, dimension is kept constant, a different value is obtained at any for Da and for . The curves in the figures have been obtained

For the optical properties of the photocatalyst particles, reference has been made to Degussa P25 titanium dioxide, assuming isotropic scattering and averaged values in the wavelength range of 300-400 nm for the absorption and the scattering coefficients, based on the data measured by Cabrera et al.27 The RTE (radiative transfer equation), with the proper hypothesis and boundary conditions, which are illustrated by Sgalari et al.,25 has been solved by the finite volume method.32,33 The obtained values of e are then utilized in the source term of eq 2, which has been discretized by the finite volume method34 and numerically solved. The role of in affecting the radiation field can be clearly understood by a combined scrutiny of Figures 2 and 3, as the value of this parameter increases when the catalyst load and/or the annular gap (R (1 - k)) of the reactor increase. When increases, the fraction of the emitted power that is absorbed within the reactor increases, but, at the same time, an increasing nonuniformity results for the distribution of e and then, even if the absorbed power increases, a less-effective exploitation of the reaction volume occurs. Therefore, it can be expected that the resulting effect on the conversion is dependent on which of the two competitive occurrences is the prevailing one. Because the distribution of

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007 7641

Figure 4. Conversion per pass, as a function of at different values of Re: (a) for R ) 0.5 and (b) for R ) 1.

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

Figure 5. Radial concentration profiles at z* ) L*: (a) at different values of for Re ) 250 and R ) 0.5 and (b) at different values of Re for ) 10 and R ) 1.

assuming ) 50 at ) 10, based on realistic values for the kinetic constant K and for the substrate diffusivity D. The results show the following: (1) Conversion obviously decreases when Re increases, if the other conditions remain unchanged, because of the reduced mean residence time. Mean residence time is actually the most significant fluid dynamics parameter that affects the performance of the reactor, whereas the residence time distribution (RTD) has a marginal role on the conversion, as will be shown later. (2) Different situations occur for different values of R: for R ) 0.5, a maximum always results for the conversion when increases. Even if this seems to be evident only for Re ) 250 in the R ) 0.5 case, this conclusion is, on the contrary, quite general and it would be clearly apparent if the conversion is normalized with respect to its maximum value. Some relevant radial concentration profiles at an axial position close to the exit of the reactor are examined in Figure 5. The disappearance of the substrate is higher in the positions where the residence time is higher; that is, the axial velocity is low, and e is larger (close to the inner wall). As a consequence, because the diffusion is rather ineffective in transporting the substrate in the radial direction, the profiles do not seem to be uniform. This effect is particularly evident when the extent of reaction is high; that is, when is high or Re is low, as it is apparent in Figure 5. A utilization of the reactor with a single pass in a continuous process is often impracticable, because of the very low conver-

sion per pass. For this reason, in many applications, the outlet stream is recycled to the reactor so that a reasonable conversion can be obtained after a sufficient time of batch operation of the system.35 Very often, a tank is added to the system, as shown in Figure 1. The function of the tank can be (i) to control the conditions of the suspension during the operation by agitating, aerating, and thermostating it, and/or (ii) to increase the volume of the aqueous solution to be processed which, otherwise, should be limited to the usually small capacity of the reactor and of the tubes. The transient behavior has been simulated by the model presented in the previous section to obtain the variation with time of the concentration of the substrate in the tank with a ratio Vr/Vt ) 0.05. The results obtained in the simulations show that, in the case of no radial gradient of concentration and of velocity within the reactor, the rate of disappearance of the substrate in the system is almost independent of the recycle flow rate. Therefore, the results obtained for Re ) 10 000 are actually representative of whatever the flow rate in turbulent flow. They are plotted in Figure 6 for different values of , which, again, represent different values of the concentration of the photocatalyst. When R ) 0.5, the fastest decay is observed at a certain value of ( = 5), whereas when R ) 1, the decay is faster with larger . The results can be examined also through a scrutiny of the constant of disappearance Kd, which has been previously

7642

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

Figure 6. Tank bulk concentration, as a function of time, in the case of turbulent flow at different values of : (a) for R ) 0.5 and (b) for R ) 1.

Figure 7. Dimensionless kinetic constant of the disappearance of the substrate, as a function of , at different values of Re: (a) for R ) 0.5 and (b) for R ) 1.

introduced and characterizes the exponential decay of the substrate in the system. In Figure 7, Kd is plotted against for different values of Re in the reactor, i.e., for different recycle flow rates. The following observations can be observed: (1) For R ) 0.5, a relative maximum of Kd is always present and then an optimal value of the optical thickness results, i.e., an optimal value of the concentration of the photocatalyst. (2) For R ) 1, a relative maximum of Kd is present only for laminar flow, when the radial profile of the substrate concentration is not flat, because of limitations in the mass transport of the substrate in the radial direction. Otherwise, Kd is monotonically increasing with toward an asymptotic value. (3) At a given , the variation of Kd and, hence, of the rate of disappearance of the substrate with Re is more remarkable at the highest values of and R; however, in any case, the effect of the flow rate is not very important. As a matter of fact, even if the substrate concentration profiles can greatly diverge at different Re (see, e.g., Figure 5b), this variation, nonetheless, is mainly confined in a zone where the convective transport in the z-direction is low, because the velocities here are low and, thus, the contribution to the conversion is limited. For the sake of comparison in Figure 5b also, the concentration profile at Re ) 250 is plotted for the hypothetical case of no radial

gradient of the velocity and of the concentration. The molar rate of the substrate at the exit is proportional to the integral from the inner radius to the outer radius of the local concentration weighted by the local velocity and the local radius. The final result of this weighted integration is that the differences are smoothed and, hence, the conversion also is not greatly affected by the transport limitations of the reagent in a radial direction. In other words, diffusion can significantly affect the concentration profiles inside the reactor, but it appears to be unable to severely limit the rate of disappearance of the substrate. Besides, the limitation of the diffusion on the rate of disappearance is assessable only at rather high values of the catalyst concentration, which represent cases of less-practical importance, because it is useless to increase the catalyst load additionally when no appreciable further enhancement can be obtained. The analysis of the effect of the order of reaction can be extended by considering the more-general kinetic eq 5 previously presented. It has been used to analyze the effect of a change of the concentration of the photocatalyst and of the reference intensity (I0) on the observed rate of disappearance of the substrate, which is soundly reproduced by the value of Kd in the batch apparatus. Only the case of turbulent flow in the reactor has been considered (without any radial gradient of

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007 7643

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

Figure 8. Dimensionless kinetic constant of the disappearance of the substrate as a function of in the case of the kinetic equation (eq 5) and turbulent flow at different values of A evaluated at ) 10.

concentration and velocity). An increase of the concentration of the catalyst implies an increase of the absorption coefficient , and, as a consequence, as previously discussed, it can be reproduced by a variation of the optical thickness . Different values of I0 can be represented by the corresponding values of the parameter A ) c/e) c/(I0) at a given optical thickness. 0 The results obtained are plotted in Figure 8 versus the optical thickness for different values of the parameter A evaluated at ) 10. At very low values of I0 (high values of A), the kinetics is obviously slow and Kd is very low. Moreover, in this case, only a first-order relationship is effective throughout the entire reactor, because of the low level of e with respect to c at any radial position. The consequence is that no relative maximum is found by varying the optical thickness, that is, by varying the concentration of the substrate. These results are qualitatively similar to those previously obtained for R ) 1. In contrast, at low values of A, the order of the reaction may change inside the reactor with the radial position, in particular, if the optical thickness is high. In fact, in this latter case, by taking into consideration also the radial distribution of e reported in Figure 3, at any radial position a different order of reaction is actually effective. In particular, at a radial position close to the emitting lamp, e is much higher than c and the order of the reaction is zero, whereas, at a radial position far from the lamp, e is much lower than c and the order of the reaction is 1. At intermediate radial positions, the entire range of order of reaction, from 0 to 1, is experienced. The final result of the complex interaction between the absorption of the radiant energy and the kinetics of the reaction is that, at high values of the reference intensity I0, a relative maximum of Kd appears, as is evident in Figure 8. This maximum is more pronounced with lower A (or higher I0) values and the corresponding optimal value of the catalyst concentration diminishes by increasing I0. Conclusions It has been demonstrated that the rate of disappearance of an organic pollutant substrate in a photocatalytic slurry is dependent on a reduced number of dimensionless parameters. Specifically, the geometric ratios of the system, the optical thickness, the Damkohler number (or, alternatively, the Reynolds number (Re)), and the Thiele modulus are the dimensionless groups that define the performances of the system, together with an

additional parameter A, which takes into account the kinetic equation of the rate of reaction. When the dimensions of the reactor are fixed, the optical thickness becomes dependent linearly on the catalyst concentration. With the present dimensionless approach, it has been shown that the rate of disappearance of the substrate from the system exhibits a maximum at an optimal value of the catalyst concentration only at low Re values or at low A values. Otherwise, the rate of disappearance is continuously increasing with the catalyst concentration toward an asymptotic value. In other words, it has been demonstrated that the occurrence of the maximum and also its possible position versus the catalyst concentration are affected by the following parameters: (1) The intensity of the entering radiation, because, ultimately, the intensity may affect the order of the reaction, with respect to the local rate of radiant energy absorption. In particular, the higher the intensity of the impinging radiation, the higher the peak of the rate of removal of the substrate and the lower the catalyst load that gives this maximum. (2) The flow rate. However, the effect of the flow rate is minor. The previous observations demonstrate that the occurrence of a maximum for the rate of substrate disappearance cannot be explained only based on the shielding effect, but it is due also to other concurrent phenomena, namely, the diffusion of the substrate and the influence that the radiation power entering the reactor has on the kinetics of the reaction. As a consequence, the optimal value of the catalyst concentration is dependent on these phenomena. Literature Cited
(1) Blanco-Galvez, J.; Fernandez, P.; Malato-Rodriguez, S. Solar Photocatalytic Detoxification and Disinfection of Water: Recent Overview. J. Solar Energy Eng. 2007, 129, 4. (2) Mehrotra, K.; Yablonsky, G. S.; Ray, A. K. Kinetic Studies of Photocatalytic Degradation in a TiO2 Slurry System: Distinguishing Working Regimes and Determining Rate Dependences. Ind. Eng. Chem. Res. 2003, 42, 2273. (3) Bickley, R. I.; Slater, M. J.; Wang, W.-J. Engineering Development of a Photocatalytic Reactor for Waste Water Treatment. Process Saf. EnViron. Prot. 2005, 83, 205. (4) Reutergardh, L. B.; Iangphasuk, M. Photocatalytic Decolourization of Reactive A20 Dye: a Comparison between TiO2 and CdS Photocatalysis. Chemosphere 1997, 35, 565. (5) Chen, D.-W.; Ray, A. K. Photodegradation Kinetics of 4-nitrophenol in TiO2 Suspensions. Wat. Res. 1998, 32, 3223. (6) Chen, D.-W.; Ray, A. K. Photocatalytic Kinetics of Phenol and its Derivatives over UV Irradiated TiO2. Appl. Catal., B 1999, 23, 142. (7) Herrmann, J. M. Heterogeneous Photocatalysis: Fundamentals and Applications to the Removal of Various Types of Aqueous Pollutants. Catal. Today 1999, 53, 115. (8) Nam, W.; Kim, J.; Han, G. Photocatalytic Oxidation of Methyl Orange in a Three-Phase Fluidized Bed Reactor. Chemosphere 2002, 47, 1019. (9) Terzian, R.; Serpone, N. Heterogeneous Photocatalyzed Oxidation of Creosote Components: Mineralization of Xylenols by Illuminated TiO2 in Oxygenated Aqueous Media. J. Photochem. Photobiol. A: Chem. 1995, 89, 163. (10) Brandi, R. J.; Alfano, O. M.; Cassano, A. E. Rigorous Model and Experimental Verification of the Radiation Field in a Flat-Plate Solar Collector Simulator Employed for Photocatalytic Reaction. Chem. Eng. Sci. 1999, 54, 2817. (11) Minero, C.; Vione, D. A Quantitative Evaluation of the Photocatalytic Performance of TiO2 Slurries. Appl. Catal., B 2006, 67, 257. (12) Serpone, N. Relative Photonic Efficiencies and Quantum Yields in Heterogeneous Photocatalysis. J. Photochem. Photobiol. A: Chem. 1997, 104, 1. (13) Wang, H.; Adesina, A. A. Photocatalytic Causticization of Sodium Oxalate using Commercial TiO2 Particles. Appl. Catal., B 1997, 14, 241.

7644

Ind. Eng. Chem. Res., Vol. 46, No. 23, 2007


(26) Pasquali, M.; Santarelli, F.; Porter, J. F.; Yue, P.-L. Radiative Transfer in Photocatalytic Systems. AIChE J. 1996, 42, 532. (27) Cabrera, M. I.; Alfano, O. M.; Cassano, A. E. Absorption and Scattering Coefficients of Titanium Dioxide Particulate Suspensions in Water. J. Phys. Chem. 1996, 100, 20043. (28) Camera-Roda, G.; Santarelli, F.; Martn, C. A. Design of Photo catalytic Reactors Made Easy by considering the Photons as Immaterial Reactants. Solar Energy 2005, 79, 343. (29) Kaneda, M.; Yu, B.; Ozoe, Y.; Churchill, S. W. The Characteristics of Turbulent Flow and Convection in Concentric Circular Annuli. Part I: Flow. Int. J. Heat Mass Transfer 2003, 46, 5045. (30) Turchi, C. S.; Ollis, D. F. Photocatalytic Degradation of Organic Water Contaminants: Mechanisms Involving Hydroxyl Radical Attack. J. Catal. 1990, 122, 178. (31) Ollis, D. F. Solar-Assisted Photocatalysis for Water Purification: Issues, Data, Questions. In Photochemical ConVersion and Storage of Solar Energy; Pelizzetti, E., Schiavello, M., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1991; p 593. (32) Chui, E. H.; Rathby, G. D.; Hughes, P. M. J. Prediction of Radiative Transfer in Cylindrical Enclosures with the Finite Volume Method. J. Thermophys. Heat Transfer 1992, 6, 605. (33) Chai, J. C.; Lee, H.; Patankar, S. V. Finite Volume Method for Radiation Heat Transfer. J. Thermophys. Heat Transfer 1994, 8, 419. (34) Patankar, S. V. Numerical Heat Transfer and Fluid Flow; Hemisphere Publishing: Washington, DC, 1980. (35) Klausner, J. F.; Martin, A. R.; Goswami, D. Y.; Schanze, K. S. On the Accurate Determination of Reaction Rate Constants in Batch-Type Solar Photocatalytic Oxidation Facilities. J. Solar Energy Eng. 1994, 116, 19.

Downloaded by UNIV FED RURAL DO RIO DE JANEIRO UFRRJ on October 6, 2009 | http://pubs.acs.org Publication Date (Web): June 9, 2007 | doi: 10.1021/ie070302a

(14) Bangum, J.; Adesina, A. A. The Photodegradation Kinetics of Aqueous Sodium Oxalate Solution using TiO2 Catalyst. Appl. Catal., A 1998, 175, 221. (15) Goncalves, M. S. T.; Oliveira-Campos, A. M. F.; Pinto, E. M. M. S.; Plasencia, P. M. S.; Queiroz, M. J. R. P. Photochemical Treatment of Solutions of Azo Dyes containing TiO2. Chemosphere 1999, 39, 781. (16) Martin, C. A.; Camera-Roda, G.; Santarelli, F. Effective Design of Photocatalytic Reactors: Influence of Radiative Transfer on their Performance. Catal. Today 1999, 48, 307. (17) Li Puma, G.; Yue, P. L. Enhanced Photocatalysis in a Pilot Laminar Falling Film Slurry Reactor. Ind. Eng. Chem. Res. 1999, 38, 3246. (18) Dijkstra, M. F. J.; Buwalda, H.; de Jong, A. W. F.; Michorius, A.; Winkelman, J. G. M.; Beenackers, A. A. C. M. Experimental Comparison of Three Reactor Designs for Photocatalytic Water Purification. Chem. Eng. Sci. 2001, 56, 547. (19) Camera-Roda, G.; Santarelli, F. Intensification of Water Detoxification by Integrating Photocatalysis and Pervaporation. J. Solar Energy Eng. 2007, 129, 68. (20) Inel Y.; Okte, A. N. Photocatalytic Degradation of Malonic Acid in Aqueous Suspensions of Titanium Dioxide: an initial Kinetic Investigation of CO2 Photogeneration. J. Photochem. Photobiol. A:Chem. 1996, 96, 175. (21) Ozisik, M. N. RadiatiVe Transfer and Interactions with Conduction and ConVection; Wiley, New York, 1973. (22) Siegel R.; Howell, J. R. Thermal Radiation Heat Transfer, 3rd Edition; Hemisphere Publishing: Washington, DC, 1992. (23) Cassano, A. E.; Martn, C. A.; Brandi, R. J.; Alfano, O. M. Photoreactor analysis and design: Fundamentals and applications. Ind. Eng. Chem. Res. 1995, 34, 2155. (24) Brandi, R. J.; Alfano, O. M.; Cassano, A. E. Rigorous Model and Experimental Verification of the Radiation Field in a Flat Plate Solar Collector Simulator employed for Photocatalytic Reactions. Chem. Eng. Sci. 1999, 54, 2817. (25) Sgalari, G.; Camera-Roda, G.; Santarelli, F. Discrete Ordinate Method in the Analysis of Radiative Transfer in Photocatalytically Reacting Media. Int. Commun. Heat Mass Transfer 1998, 25, 651.

ReceiVed for reView February 28, 2007 ReVised manuscript receiVed May 2, 2007 Accepted May 3, 2007 IE070302A

You might also like