摘要:
Paper Surface diffusion control of the photocatalytic oxidation in air/TiO2 heterogeneous reactors R. Tsekov,a E. Evstatievab and P. G. Smirniotisc aDepartment of Physical Chemistry, University of Sofia, 1126 Sofia, Bulgaria, bDepartment of Water Technology, ITC, Karlsruhe Research Center, 76021 Karlsruhe, Germany cDepartment of Chemical Engineering, University of Cincinnati, Cincinnati, OH 45221, USA Received 14th August 2002, Accepted 9th October 2002 First published as an Advance Article on the web 24th October 2002 The diffusion of superoxide radical anions on the surface of TiO2 catalysts is theoretically considered as an important step in the kinetics of photocatalytic oxidation of toxic pollutants. A detailed analysis is performed to discriminate the effects of rotation and adsorption bond vibrations on the diffusion coefficient.A resonant dependence of the diffusivity on the lattice parameters of the TiO2 surface is discovered showing that the most rapid diffusion takes place when the lattice parameters are twice as large as the the bond length of the superoxide radical anions. Whereas the rotation and vibrations normal to the catalyst surface are important, the anion bond vibrations do not affect the diffusivity due to their low amplitudes as compared to the lattice parameters. Introduction Photocatalytic oxidation of organic compounds is an alternative to the conventional methods for removal of toxic pollutants from air and water.1,2 Among the applied semiconductor photocatalysts3 the most promising one is the titanium dioxide4 because of its high photocatalytic activity, resistance and low costs.It is well known that valence band holes and conduction band electrons are generated by illumination of TiO2 samples with light energy greater than the band gap (3.2 eV) TiO2 /? Bn hzze{ Here a possible recombination of the generated carriers with an emission of heat is also described. The holes h1 and electrons e2 act on the TiO2 surface as powerful oxidizing and reducing agents, respectively. The holes oxidize water molecules, adsorbed at the TiO2 surface from the air, to hydroxyl radicals H2Ozhz?. OHzHz while the electrons react with oxygen molecules adsorbed on the Ti(III) sites reducing them to superoxide radical anions O2ze{?. O{ 2 The hydroxyl and superoxide radicals are responsible for the oxidizing steps in photocatalysis.5 The generally agreed mechanism involves an initial attack of the organic molecule by a hydroxyl radical followed by a reaction of the resulting radical with superoxide radicals.The problem is that several ?O22 are required for complete mineralization of a toxic molecule. Hence, the oxidation process can be limited by the mobility of the superoxide radical anions on the TiO2 surface and, for this reason, the description of their surface diffusion is very important for the kinetic modeling of the air/TiO2 reactors.6 The diffusion of particles in solids and on their surfaces is of current interest both for the pure science and technology. The diffusion coefficient of molecules in solids exhibits nonmonotonous dependence on the molecular size, a phenomenon called resonant diffusion.7 An interfacial example of this phenomenon represents the Brownian motion of rhenium dimers on tungsten8 and ruthenium clusters on rhenium.9 The resonant diffusion is theoretically described via molecular dynamics simulations10,11 and stochastic modeling.12–14 Some general treatments of the diffusion in modulated structures are also developed.15–20 The problem of diffusion on solid surfaces is equivalent to analysis of the effect of the thermal phonons in the solid on the behavior of the adsorbed molecule.A rigorous consideration of the stochastic dynamics of a particle coupled with a harmonic bath is possible.16,19 As a result, one yields a dynamic equation with a position dependent friction and a random force, which are coupled by the fluctuation–dissipation theorem.The comparison of the theory with experiments for diffusion of rhenium atoms and dimers on tungsten surfaces is excellent.13 The effects of dimmer rotation21 and vibration22 on the barrier crossing are also taken into account. The present study aims to apply this model for description of the diffusion of superoxide radical anions on TiO2 surface in order to discover the optimal conditions for photocatalytic oxidation of pollutants in air/TiO2 reactors. Surface diffusion of a rigid superoxide radical anion Let us consider first the model of a rigid superoxide anion with a fixed distance between the two oxygen atoms. The potential w is the fundamental measure of the interaction between an oxygen atom and the TiO2 surface, which can be rigorously calculated by ab initio methods.However, since the catalysts surfaces are not well defined and covered by various equilibrium and non-equilibrium interfacial structures, it is necessary to introduce an effective model for the potential w. Since the latter is a periodic function, which possesses the symmetric properties of the local TiO2 lattice, w can be naturally expanded in a Fourier series. Assuming that there is a dominant term in this series, hereafter the following model well known from the literature12,21 will be employed w~Acos (2px=a)zAcos (2py=b) (1) where x and y are the coordinates of the oxygen atom and A is the magnitude of the interaction. By variation of the parameters a and b one is able to model different lattices on the DOI: 10.1039/b207965h PhysChemComm, 2002, 5(24), 161–164 161 This journal is # The Royal Society of Chemistry 2002TiO2 surface.The basic advantage of the decoupled cosine potential (1) is that the corresponding forces split to independent components, i.e. the forces acting of x and y axes depend on x and y, respectively. It is easy now to write the potential energy of the superoxide radical anion in the form U~w1zw2~2Acos (p cos QL=a) cos (2pX=a) z2Acos (p sin QL=b) cos (2pY=b) (2) where X and Y are the coordinates of the mass center of the anion, Q is the angle between the anion and x axis and L is the anion length. Another characteristic important for the Brownian motion of the anions is the friction coefficient B. It is possible starting from classical mechanics to derive the following expression17 for the friction coefficient acting on the x axis 4prc3BX~(L2w1=Lx2 1)2z(L2w2=Lx2 2)2 ~(16p4A2=a4)½1zcos (2p cos QL=a) cos (4pX) (3) where r and c are the mass density and sound velocity in TiO2, respectively. The friction coefficient on the y axis can be calculated analogically. Due to the simplicity of the potential (1), one can directly apply the well-known formula DX~ a2 b � a 0 BX exp (bU)dX � a 0 exp ({bU)dX (4) for calculation of the diffusion constant, which is rigorously derived for the case of one dimensional motion of a particle in a periodic potential.15 Here b ~ (kT)21 is the inverse thermodynamic temperature.Eqn. (4) shows a relative asymmetric effect of the friction coefficient to diffusion.Since the exponential term in the first integral is mainly determined by the maximum of U, the friction on the top of the potential barriers is more essential for the value of DX, while the value of BX in the potential wells is of second order of importance. Substituting the expressions (2) and (3) in eqn. (4) one is able to calculate analytically the diffusion coefficient of the superoxide radical anion DX~ rc3a4 4p3AI0½2bAcos (p cos QL=a) cos (p cos QL=a)= f2bAcos3 (p cos QL=a)I0½2bAcos (p cos QL=a) {cos (2p cos QL=a)I1½2bAcos (p cos QL=a)g (5) where I0 and I1 are the modified Bessel functions of zero and first order, respectively. In Fig. 1 the dependence of the diffusion coefficient from eqn. (5) is plotted as a function of the angle between the anion and x axis and the corresponding lattice parameter a.The specific material constants are equal to r~4260 kg m23, c~6900 m s21, T~350 K and L~1.35 A° . The value 2A~5610220 J is estimated as the adsorption heat of oxygen on TiO2’s surface.23 As seen from Fig. 1 there is a strong resonant dependence of the diffusion coefficient, which exhibits a maximum when the lattice parameter a wice larger than the x projection of the anion. In this case the activation energy is zero. A further problem is the influence of the rotation of the adsorbed superoxide radical anions on the resonant dependence of the diffusion coefficient. The complete investigation of the dynamics of a rotating anion requires a stochastic description of both the translation and rotation.A basic approximation here follows from the difference between the characteristic timescales of these two processes.21 In this case one can apply adiabatic separation of the dynamics to slow (translation) and quick (rotation) motions. This means, that from the viewpoint of translation the rotation is always at equilibrium. Thus, using the potential energy of the anion from eqn. (2) one can calculate the rotational integral of state ZR~ � 2p 0 exp ({bU)dQ (6) The knowledge of ZR allows calculation of the rotational free energy of the anion FR~2b21lnZR, which plays the role of an average potential for translation. Unfortunately, the function FR is not a sum on independent components corresponding to the two axes of motion. For this reason we will employ further a second adiabatic procedure to take into account the influence of the motion on Y to the motion on X.The corresponding partial translation integral of state acquires the form ZT~ 1 b � b 0 exp ({bFR)dY~ � 2p 0 exp½{2bAcos (p cos QL=a) cos (2pX=a)I0½2bAcos (p sin QL=b)dQ (6) By means of ZT one can calculate the conditional free energy FT ~ 2b21lnZT for the anion motion on the x axis. Of course, FT is a periodic function of X and, consequently, one can apply once again the formula (4) to calculate the diffusion coefficient of the rotating anion DX (FT)~ rc3a6 8p3bA2 � a 0 exp (bFT)dX � a 0 exp ({bFT)dX (7) Note that in eqn. (7) the friction coefficient is replaced by its value at the maximum of the potential. The analytical integration of this formula is complex and time consuming and, for this reason, numerical calculations will be employed, where the values of the specific constants are fixed as mentioned above.The two lattice parameters a and b are varied in the process of calculations and the result is presented in Fig. 2. As seen there is a pronounced maximum of the diffusion constant at a ~ b ~2L. Another interesting feature is the asymmetric effect of the two lattice parameters: the effect of the lattice parameter on the complementary axis b is weaker as compared to the effect of the parameter a. Effect of vibrations on the superoxide radical anion diffusivity In the present part, a method based on eqn. (4) is developed for calculation of the diffusion coefficient of a vibrating anion on the TiO2 surfaces.In general, there are two kinds of vibrations relevant to the process of diffusion: internal vibrations of the anion’s bonds and vibrations of the anion normal to the TiO2. Fig. 1 Dependence of the diffusion coefficient of a superoxide radical anion on the angle Q and the lattice parameter a. 162 PhysChemComm, 2002, 5(24), 161–164According to eqn. (2), the potential energy of a superoxide radical anion with arbitrary length is U~2Acos (p cos QLj=a) cos (2pX=a) z2Acos (p sin QLj=b) cos (2pY=b) (8) where j is the relative bond length as compared to the equilibrium value L. As was mentioned above, owing to the difference in the characteristic timescales of the anion translation and vibration, it is possible to apply an adiabatic separation of the slow and quick variables. The latter should be considered in thermodynamic equilibrium.Therefore, the effect of the bond vibrations can be accounted for by the vibration partition function ZV~ � ? 0 � 2p 0 exp½{2bAcos (p cos QLj=a) cos (2pX=a) |I0½2bAcos (p sin QLj=b) exp ({bV)dQdj (9) where V is the internal potential energy of the anion. A convenient model for V is the harmonic potential V ~ 2m(pvL)2(j 2 1)2, where m ~ 8 g mol21 is the reduced mass of the anion and v ~ 1090 cm21 is the frequency of the superoxide anion vibrations.24 Using this expression one can calculate ZV from eqn. (9) and the conditional free energy of the vibrating anion FV ~ 2b21lnZV, which plays the role of an average potential acting on the anion translation. Hence, applying the formula (4) with a constant friction term leads to DX(FV) Thus the calculated diffusion coefficient does not differ from the result in Fig.2. This means that the interaction between the two oxygen atoms is very strong and the small bond vibrations do not affect the diffusion coefficient of the superoxide radical anion. One could argue that the value of v for an adsorbed anion is lower than the used value from the gas phase. Further calculations show, however, that the bond vibrations become important if v is below 200 cm21. Such an unrealistically low frequency could be observed only if the superoxide radical anion is in an exited state.25 Other possible vibrations are those normal on the TiO2 surface along the adsorption bond. In this case the potential energy of the superoxide radical anion acquires the form U~{W cos (p cos QL=a) cos (2pX=a) {W cos (p sin QL=b) cos (2pY=b) (10) where W is the potential energy of the adsorption bond depending on the z coordinate. If the interaction between an oxygen atom and the atoms of the solid obeys the Lennard-Jones potential, the adsorption van der Waals potential takes the form W~A(1=f9{3=f3) (11) where f is the relative distance between the TiO2 surface and the mass center of the superoxide radical anion as compared to the equilibrium distance.Obviously, W exhibits a minimum at f ~ 1, which is 2A deep. Using eqn. (11), one can calculate the vibration partition function for the present case as ZW~ � ? 0 � 2p 0 exp½bW cos (p cos QL=a) cos (2pX=a) |I0½{bW cos (p sin QL=b) exp ({bW)dQdf (12) The internal anion vibrations are not considered in eqn.(12) since, according to the results above, they are not important due to the low amplitude of the O–O bond deviations. The free energy FW ~ 2b21lnZW for the translation follows directly from eqn. (12) and substituted in eqn. (7) it determines the diffusion coefficient of the superoxide anion. The numerical calculated DX(FW) is presented in Fig. 3 as a function of the lattice parameters a and b. As seen the normal vibration dramatically increases the diffusivity of the anions. The only exception is the case of the resonant maximum. Since there the activation energy is zero, it is not affected by the vibrations. Conclusions There are several effects important for the diffusion of superoxide radical anions on the TiO2 surface.While the rotation and vibrations normal to the catalyst surface are essential, the anion bond vibrations do not affect the diffusivity. The dependence of the diffusion coefficient on the lattice parameters of the TiO2 surface is strongly resonant and exhibits a pronounced maximum at lattice constants two times larger than the bond length of the superoxide radical anion. Hence, to increase the surface diffusion and thus the oxidation rate it is important to design the TiO2 surface during the catalyst’s engineering in order to fit as well as possible the resonant condition. Another interesting idea is to illuminate the media with specific light, which will excite the superoxide radical anions. Since the distance between the oxygen atoms in the exited state is larger than in the ground state it will be easier to achieve the resonant diffusion.Moreover, the bond vibrations in the exited state possess larger magnitudes and thus they could lead to an additional increase of the diffusion coefficient. Acknowledgements The authors wish to thank NATO for a grant (SfP-974209). Fig. 2 Dependence of the diffusion coefficient of a rotating superoxide radical anion on the lattice parameters a and b at the equilibrium distance from the TiO2 surface. Fig. 3 Dependence of the diffusion coefficient of a rotating superoxide radical anion on the lattice parameters a and b with account for the vibrations normal to the TiO2 surface. PhysChemComm, 2002, 5(24), 161–164 163References 1 Photocatalytic Purification and Treatment of Water an and H. Al-Ekabi, Elsevier, New York, 1993. 2 J. Peral, X. Domenech and D. Ollis, J. Chem. Technol. Biotechnol., 1997, 70, 117. 3 M. R. Hoffmann, S. T. Martin, W. Choi and D. W. Bahnemann, Chem. Rev., 1995, 95, 69. 4 A. Linsebigler, G. Lu and J. Yates, Chem. Rev., 1995, 95, 735. 5 E. Pelizzetti and C. Minero, Electrochim. Acta, 1993, 38, 47. 6 V. Zhdanov, Kinet. Katal., 1987, 28, 568. 7 E. Ruckenstein and P. S. Lee, Phys. Lett. A, 1976, 56, 423. 8 G. Ehrlich, Surf. Sci., 1995, 331–333, 865. 9 G. L. Kellog, Phys. Rev. Lett., 1994, 73, 1833. 10 C. Massobrio and P. Blandin, Phys. Rev. B, 1993, 47, 13687. 11 R. Wang and K. A. Fichthorn, Phys. Rev. B, 1993, 48, 18288. 12 R. Ferrando, R. Spadacini and G. E. Tommei, Surf. Sci., 1992, 265, 273. 13 E. Ruckenstein and R. Tsekov, J. Chem. Phys., 1994, 100, 7696. 14 G. Caratti, R. Ferrando, R. Spadacini and G. E. Tommei, Phys. Rev. E, 1996, 54, 4708. 15 R. Festa and E. G. d’Agliano, Physica A, 1978, 90, 229. 16 W. G. Kleppmann and R. Zeyher, Phys. Rev. B, 1980, 22, 6044. 17 K. Golden, S. Goldstein and J. L. Lebowitz, Phys. Rev. Lett., 1985, 55, 2629. 18 R. Ferrando, R. Spadacini, G. E. Tommei and G. Caratti, Physica A, 1993, 195, 506. 19 R. Tsekov and E. Ruckenstein, J. Chem. Phys., 1994, 100, 1450. 20 Y. Georgievskii and E. Pollak, Phys. Rev. E, 1994, 49, 5098. 21 R. Tsekov and E. Ruckenstein, Surf. Sci., 1995, 344, 175. 22 R. Tsekov, NATO ASI Ser., Ser. B, 1997, 360, 419. 23 T. Smith, Surf. Sci., 1973, 38, 292. 24 K. P. Huber and G. Herzberg, Molecular Spectra and Molecular Structure: Constants of Diatomic Molecules, Van Nostrand, New York, 1979. 25 E. Robinson, W. Holstein, G. Steward and M. Buntine, Phys. Chem. Chem. Phys., 1999, 1, 3961. 164 PhysChemComm, 2002, 5(24), 161–164
ISSN:1460-2733
DOI:10.1039/b207965h
出版商:RSC
年代:2002
数据来源: RSC