J. Phys. Chem. B 2005, 109, 977-980
977
Recombination Pathways in the Degussa P25 Formulation of TiO2: Surface versus Lattice Mechanisms Deanna C. Hurum and Kimberly A. Gray* Institute for EnVironmental Catalysis and Department of CiVil and EnVironmental Engineering, Northwestern UniVersity, EVanston, Illinois 60208
Tijana Rajh and Marion C. Thurnauer Chemistry DiVision, Argonne National Laboratory, Argonne, Illinois 60439 ReceiVed: October 8, 2004
Charge migration between electron trapping sites within the mixed-phase titania photocatalyst Degussa P25 has been studied. In addition to previously described lattice electron trapping sites on both anatase and rutile phases, surface electron trapping sites and an anatase-rutile interface trapping site specific to Degussa P25 are identified. The relationship between these sites and recombination with surface hole trapping sites is also determined. It is experimentally shown that upon band-gap illumination holes appear at the surface and preferentially recombine with electrons in surface trapping sites. These findings indicate that in mixed-phase TiO2, such as Degussa P25, photogenerated holes are trapped exclusively on the particle surface, while photogenerated electrons are trapped within the nanoparticle lattice. Recombination reactions are dominated by surface reactions that follow charge migration. These findings indicate that, in mixed-phase TiO2, such as Degussa P25, a random flight mechanism of recombination predominates. Such knowledge simplifies the mechanistic mathematical models used for process design and points the way for improving future oxidative titania catalysts.
Introduction Titania photocatalysts have been widely studied for use in industries ranging from chemical synthesis,1-5 energy production and storage,6,7 environmental remediation,8,9 and sensors and odor control.10-13 TiO2 catalysts are commercially advantageous due to a band gap that can be activated by sunlight and spans the redox properties of water.14 Furthermore, while the activated catalyst has biocidal properties, the titania material itself has low biological toxicity.15 Pure mineral phases of titania have been shown to have different photoefficiencies and chemical selectivities. The anatase phase of TiO2 is generally the higher activity oxidative photocatalyst.16,17 Comparatively, rutile is less active. However, rutile has been shown to be effective at both oxidative and reductive chemistry in specific applications.18 In addition to the pure-phase TiO2 catalysts, several formulations of mixed-phase titania are also commercially produced. The mixing of an active oxidizing phase (anatase) with a comparatively inactive phase (rutile) produces a class of photocatalysts with unusually high activity.14,19 Factors contributing to the increased activity include high surface areas, high adsorption affinity for organic compounds,20 and lower recombination rates.21 While each of these increases the likelihood of chemical reaction with the photogenerated holes and electrons, recombination is critical to catalyst activity and the understanding of recombination can greatly improve catalyst activity. Many researchers have focused on theoretical modeling of how recombination affects the quantum efficiency of photocatalysts. Several models have been proposed, on the basis of a variety of recombination mechanisms. Direct recombination models describe the reaction of the geminate ion pair in the lattice before any transfer to trapping sites occurs.22 This type
of recombination reduces efficiency as neither charge has a chance to migrate to the surface, where they may react or be trapped in lower energy sites. The multiple-trapping-site recombination mechanism models the photoinduced separated charges as they recombine from multiple trapping sites of similar energy and geometry.23 In this case, a surface-trapped hole recombines with one of many surface-trapped electrons that are located in similar trapping sites. This model of recombination from monoenergetic surface sites has been confirmed experimentally in single-crystal systems.24,25 An extension of this explanation is random-flight recombination. This model allows trapping sites of different energies to be involved in recombination to more realistically describe non-single-crystal materials. A surface hole may recombine with any electron that, on the basis of the thermal energy in the system, may be sampling trapping sites that are in the lattice and on the surface until recombination occurs. In such a model many trapping configurations must be considered to account for the migration of the charges. This approach has been successful in describing recombination in TiO2/sensitizing dye systems.26 Several methods have been used to study recombination. Of these methods, electron paramagnetic resonance (EPR) is frequently utilized due to its sensitivity and the ability to directly and indirectly, through spin-trapping methods, observe photoinduced radicals.27-29 Using EPR, the surface and lattice trapping sites can be identified for both electrons and holes. Additionally, by working at low temperatures the migration between these sites can be monitored.30,31 In this work the recombination behavior of a mixed-phase TiO2 photocatalyst, Degussa P25, is studied by using EPR to track the population of electron trapping sites.
10.1021/jp045395d CCC: $30.25 © 2005 American Chemical Society Published on Web 12/21/2004
978 J. Phys. Chem. B, Vol. 109, No. 2, 2005
Hurum et al.
Recent work in our laboratory and others indicates that electron transfer from rutile to anatase affects catalyst activity and recombination.32-34 In these studies, it has been shown that rutile both extends the wavelength range of activity of the catalyst and shuttles electrons to anatase away from holes leading to a more active catalyst. Specifically, by using EPR, it was determined that under visible illumination conditions, electron transfer occurs from rutile to anatase.32 This observed electron transfer leads to a more active catalyst by two mechanisms. The lower band gap of rutile allows for greater absorption of solar radiation. In this mechanism, rutile acts as an antenna for anatase, increasing the spectral range of activity of the catalyst. In addition to the increase in the spectral range, this charge transfer also further separates the photogenerated holes and electrons spatially. This increased charge separation, relative to a pure phase material, reduces recombination and increases the efficiency of the catalyst. The existence of an interface, then, between the two phases is critical to enabling the observed charge separation. Materials and Methods Degussa P25 was kindly donated by Degussa Corp. Samples were prepared by making a titania slurry of 40 g/L of Degussa P25 in either 18 MΩ water (MilliQ) or a solution of 0.8 mM 2,4,6-trichlorophenol (TCP) (aqueous; Aldrich) as previously described.32 After removing oxygen from the sample by nitrogen purge, the sample was illuminated with ambient fluorescent lighting. After several hours these samples were frozen in liquid nitrogen to form a water glass and then transferred to the EPR microwave cavity. EPR spectra were collected on a Varian E-9 spectrometer equipped with a helium cryostat. Samples were cooled to 10 K and further illuminated within the cavity at that temperature while spectra were acquired. A 300 W xenon lamp (ILC Inc.) was used. Illumination while taking spectra was continuous and typically lasted for 30 min. Results and Discussion Charge trapping in TiO2 has been well-studied by previous work on colloidal anatase particles.30,31 Electrons are initially held in lattice trapping sites, while holes appear at surface trapping sites. No lattice hole trapping site has been identified for nanoparticle TiO2. Similarly, with Degussa P25 the initial electron trapping sites are in the lattice while the holes are on the surface trapping sites. Previously we have investigated the mechanism of charge separation in suspensions and colloidal solutions obtained from Degusa P25 to study the relationship between the aggregate particle size and the composition of the particles that constitute Degusa P25.32 In this work we are investigating the charge separation mechanism in P25 slurry that is composed of broad distribution of the aggregate particle sizes and large concentrations of TiO2 (40 g/L) using EPR spectroscopy. Illumination of a P25 slurry at 10 K results in the formation of electron-hole pairs that have characteristic EPR spectra depending on their coordination and environment. As seen in Figure 1a both anatase (g⊥ ) 1.990, g| ) 1.957) and rutile (g⊥ ) 1.975, g| ) 1.940) lattice electron trapping sites are observed for a 40 g/L slurry of P25.32 Additionally, a signal from an oxygen centered radical is seen that can be associated with surface trapped holes in an anatase environment.30,31 Upon warming this sample to 120 K, recombination of charges occurs and both the hole trapping site and the majority of the lattice electron trapping site signals are lost (Figure 1b). Few electron-hole pairs survive recombination, preventing full observation of the surface electron trapping sites
Figure 1. (a) EPR spectrum illustrating the trapping sites in a slurry sample of Degussa P25. Both anatase and rutile electron trapping sites are observed. Additionally a weak, broad signal characteristic of surface electron trapping sites is present. (b) EPR spectrum of the sample after it is warmed to 120 K to allow recombination to occur and then returned to 10 K. A weak anatase lattice signal is left along with a signal similar to interfacial electron trapping sites.
that would be populated at higher temperatures. The remaining features consist of two weak signals. The first at g ) 1.990 is characteristic of anatase lattice electron trapping sites. The second signal at g ) 1.979 is very similar to those of distorted four-coordinated titania sites previously observed at interfaces between silica and anatase phase titania coatings.35,36 Our previous work has postulated the existence of an interface between rutile and anatase in Degussa P25 that allows lowenergy electron transfer between the particles.32 Due to the strain that must exist at an interface between rutile and anatase nanoparticles, this signal (g ) 1.979) is interpreted as an electron trapping site in the distorted interfacial region. Distorted anatase structures have been observed and monitored during rutile nucleation, leading to phase change between anatase and rutile by HR-TEM.37 To better observe and analyze these trapping sites, we have used an efficient hole scavenger to prevent electron-hole recombination. In our previous work 2,4,6-trichlorophenol was shown to be an efficient hole scavenger and electron injector.38 An aqueous TCP solution (0.8 mM) was added to a sample of P25 to generate a 40 g/L slurry. This suspension was illuminated using fluorescent lighting at room temperature while under a nitrogen atmosphere. As TCP scavenges the holes in the absence of oxygen, surface electron trapping sites become populated. We monitored the outcome of this room-temperature reaction using low-temperature (10 K) EPR. The spectrum of this sample after cooling is shown in Figure 2a. The relative lack of an organic radical signal indicates that the TCP has been degraded during the room-temperature illumination, leaving behind stable, diamagnetic oxidation products. Two features are identifiable in the spectrum that results from cooling this sample to 10 K with no additional illumination. (Figure 2a) The very intense broad signal centered at g ) 1.93 has been previously observed in anatase nanoparticulate systems and is assigned to anatase surface electron trapping sites. The broad nature of this signal is indicative of a wide distribution of surface sites with differing geometries and therefore varying trapping energies.39,40 The second signal at g ) 1.979 is the same as the signal obtained by illumination of P25 slurry at
Degussa P25 Formulation of TiO2
Figure 2. (a) EPR spectrum from a Degussa P25 slurry sample illuminated at room temperature with TCP as a hole scavenger. Two signals are strongly present. Both an anatase surface signal (g ) 1.930) and a signal characteristic of interfacial titania (g ) 1.979) are seen. The interfacial titania signal (dotted line) is close in g value to that of the lattice rutile signals (solid line). After additional illumination, the spectrum in b is observed. Both lattice rutile and anatase trapping site signals are present. Additionally both the surface site signals and the interfacial site signals have dropped in intensity (arbitrary units) to 32% and 57% of the original intensity, respectively. The inset illustrates the presence of two overlapping signals. One at g ) 1.979 (interface) and a second at g ) 1.975 (rutile lattice).
elevated temperatures (Figure 1b), which was assigned to the newly identified distorted four-coordinated interfacial sites. These data indicate that there are two distinct classes of electron surface trapping sites: the anatase surface trapping sites and the interfacial sites. Subsequent additional illumination of this sample at 10 K produces additional e--h+ pairs that react with existing trapped species in the location of their formation due the low mobility of species at liquid helium temperatures. The EPR spectrum of the sample after additional illumination at 10 K is presented in Figure 2b. Illumination at 10 K does not generate more carbon centered radicals as are observed in solutions that have not been previously illuminated,38 another indication that the surface adsorbed TCP has been transformed by the scavenged surface holes at room temperature. Instead, newly produced photogenerated holes are trapped as surface oxygen centered radicals on TiO2. The weak spectral intensity of these hole trapping sites indicates that as the holes are formed, they recombine with surface accumulated electrons. The mechanism of recombination of newly formed charges with existing trapped species can be deduced from the change of the electron population at the four identified electron trapping sites: anatase and rutile lattice sites, anatase surface sites, and the fourcoordinated interfacial sites. The first difference to be noticed is reduction of both signals associated with surface electrons (anatase surface electrons g ) 1.93 as well as interfacial fourcoordinated sites g ) 1.979) to approximately half of their original size before illumination. This result indicates that surface electrons disappear due to recombination with photogenerated holes. Concurrently, the signals for lattice electron trapping sites appear and become the dominant electron spectral feature. The increase in the population of lattice sites indicates that the population of photogenerated holes formed within the lattice does not recombine with lattice formed electrons, and, therefore, recombination is dominated by surface mechanisms.
J. Phys. Chem. B, Vol. 109, No. 2, 2005 979
Figure 3. (a) Low-temperature EPR spectrum from a deoxygenated Degussa P25 slurry sample illuminated at room temperature with methanol as a hole scavenger. Two signals are strongly present. Both an anatase surface signal marked by the dashed line (g ) 1.930) and a signal characteristic of interfacial titania, marked by a dotted line (g ) 1.979) are seen. (b) Upon further illumination at low temperature, an both lattice rutile and lattice anatase electron trapping site signals are observed, while surface and interfacial trapping site signals are reduced in intensity.
These results are not specific to the reaction of TCP on the surface of TiO2. When methanol is used as a hole scavenger at room temperature, the same spectral features are present, as can be seen in Figure 3. All three experimental results, i.e., disappearance of accumulated electrons from the surface trapping sites during illumination at liquid helium temperatures, formation of lattice trapped electrons, and weak signals for photogenerated holes, provide evidence that photogenerated holes in particulate titania particles are formed exclusively on the particle surface, while photogenerated electrons are rapidly trapped within the nanoparticles’ lattice. This result is the first experimental evidence that the large surface-to-bulk ratio in particulate systems leads to a change of the excitation mechanism in particulate systems. As surface atoms experience a distortion in symmetry and restructure the crystalline surface, the electronic properties, the selection rules, and the relaxation dynamics of the charge carriers are altered, either due to a change in the energy of valence band states at the surface or a change in the selection rules accompanying the distortion of the crystalline symmetry; as has been previously observed in X-ray absorption spectra in nanoparticulate systems,41 the absorption of light in particulate systems results in the excitation of the electrons from surface atoms to the conduction band of TiO2. Additionally, these results provide direct experimental evidence that two processes are involved during recombination within particulate titania particles. The first, rapid disappearance of the surface electrons confirms that recombination in particulate systems is dominated by surface phenomena. Second, these results also provide direct evidence of the dominance of a random-flight mechanism for recombination in these materials. Recombination from lattice sites is not the major mechanism of recombination in this material; however, recombination from multiple sites on the catalyst dominates the overall process. This system is far more complicated than the monoenergetic surface present in singlecrystal systems. Not only is there a distribution of sites on each mineral phase but also interfacial sites actively contribute sites
980 J. Phys. Chem. B, Vol. 109, No. 2, 2005 for recombination. The dominance of this recombination mechanism indicates many strategies by which recombination can be effectively controlled. In a catalyst where surface recombination controls eventual activity, surface modification that removes holes or electrons from the surface of the photocatalyst will dramatically improve the effectiveness of the catalyst. Indeed, surface-modified TiO2 systems have been studied and do show such an improvement in activity.42 Conclusions The relationships between the lattice electron trapping sites, surface electron trapping sites, a newly identified interface trapping site specific to Degussa P25, surface hole trapping sites, and recombination in Degussa P25 formulation TiO2 have been studied. Surface and newly discovered interfacial electron trapping sites have been identified. We consider the identification of this new interfacial electron trapping site to be an important and unique feature of the rutile/anatase morphology within Degussa P25 that is critical to the enhanced photocatalytic behavior of mixed-phase TiO2. It has been experimentally shown that upon band-gap illumination holes appear at the surface and preferentially recombine with electrons in surface trapping sites. These findings indicate that in mixed-phase TiO2, such as Degussa P25, the random flight mechanism of recombination predominates. Such knowledge simplifies and improves the accuracy of the mechanistically based kinetic models43 of photocatalytic reactions used for process and reactor design and points the way for improving future oxidative titania catalysts. References and Notes (1) Agrios, A. G.; Gray, K. A.; Weiss, E. Langmuir 2003, 19, 1402. (2) Schoumacher, K.; et al. J. Photochem. Photobiol. A: Chem. 2002, 152, 147. (3) Lin, C. H.; et al. Catal. Lett. 2001, 73, 121. (4) Jia, J. G.; Ohno, T.; Matsumura, M. Chem. Lett. 2000, 8, 908. (5) Fox, M.; Dulay, M. T. Chem. ReV. 1993, 93, 341. (6) Hagfeldt, A.; Gratzel, M. Chem. ReV. 1995, 95, 49.
Hurum et al. (7) Bak, T.; Nowotny, J.; Rekas, M.; Sorrell, C. C. Int. J. Hydrogen Energy 2002, 27, 991. (8) Konstantinau, I. K.; Albanis, T. A. Appl. Catal., B 2003, 42, 319. (9) Pirkanniemi, K.; Sillanpaa, M. Chemosphere 2002, 48, 1047. (10) Zorn, M. E.; Tompkins, D. T.; Zeltner, W. A.; Anderson, M. A. Appl. Catal., B 1999, 23, 1. (11) Peral, J.; Domenech, X.; Ollis, D. F. J. Chem. Technol. Biotechnol. 1997, 70, 117. (12) Anderson, M. A. Stud. Surf. Sci. Catal. 1997, 103, 445. (13) Pichat, P.; et al. Catal. Today 2000, 62, 363. (14) Linsebigler, A. L.; Lu, G.; Yates, J. T. Chem. ReV. 1995, 95, 735. (15) Blake, D. M.; et al. Sep. Purif. Methods 1999, 28, 1. (16) Watababem, T.; et al. Thin Solid Films 1999, 351, 260. (17) Sumita, T.; et al. Appl. Surf. Sci. 2002, 200, 21. (18) Park, N.; van de Lagemaat, J.; Frank, A. J. Phys. Chem. B 2000, 104, 8989. (19) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Chem. ReV. 1995, 95, 69. (20) Stafford, U.; Gray, K. A.; Kamat, P. V.; Varma, A. Chem. Phys. Lett. 1993, 205, 55. (21) Riegel, G.; Bolton, J. R. J. Phys. Chem. 1995, 99, 4215. (22) Bisquert, J.; Zaban, A.; Salvador, P. J. Phys. Chem. B 2002, 106, 8774. (23) Adriaenssens, G. J.; et al. Phys. ReV. B 1995, 51, 9661. (24) Vandermolen, J.; Gomes, W. P.; Cardon, F. J. Electrochem. Soc. 1980, 127, 324. (25) Salvador, P.; Gutierrez, C. J. Electrochem. Soc. 1984, 131, 326. (26) Baarzykin, A. V.; Tachiya, M. J. Phys. Chem. B 2002, 106, 4356. (27) Riegel, G.; Bolton, J. R. J. Phys. Chem. 1995, 99, 4215. (28) Howe, R.; Gratzel, M. J. Phys. Chem. 1985, 89, 4495. (29) Liu, G.; et al. EnViron. Sci. Technol. 1999, 33, 2081. (30) Rajh, T.; et al. Chem. Phys. Lett. 2001, 5, 31. (31) Micic, O.; et al. J. Phys. Chem. 1993, 97, 7277. (32) Hurum, D. C.; et al. J. Phys. Chem. B 2003, 107, 4545. (33) Ohno, T.; et al. Appl. Catal., A 2003, 244, 383. (34) Sun, B.; Smirniotis, P. G. Catal. Today 2003, 88, 49. (35) Anpo, M.; et al. Chem. Lett. 1987, 10, 1997. (36) Yamashita, H.; et. al. J. Phys. Chem. B 1998, 102, 5870. (37) Penn, R. L.; Banfield, J. F. Am. Miner. 1999, 84, 871. (38) Hurum, D. C.; et al. J. Phys. Chem. B 2004, 108, 16483. (39) Howe, R. F.; Gratzel, M. J. Phys. Chem. 1985, 89, 4499. (40) Rajh, T.; et al. J. Phys. Chem. 1996, 100, 4538. (41) Chen, L. X.; et al. J. Synchrotron Radiat. 1999, 6, 445. (42) Rajh, T.; Teide, D. M.; Thurnauer, M. C. J. Non-Cryst. Solids 1996, 205, 815. (43) Stafford, U.; Gray, K. A.; Kamat, P. V. Res. Chem. Intermed. 1997, 23, 355.
