Photocatalytic and degradation mechanisms of anatase TiO₂: a HRTEM study

Abstract

The photocatalytic process for degradation of methylene blue with P25 TiO₂ was directly observed using a high-resolution transmission electron microscopy (HRTEM). It was found that: (1) the pristine anatase TiO₂ nanoparticles exhibited a perfect crystal lattice with a clear HRTEM image; (2) after adsorption and degradation, there were many methylene blue crystals as 1 nm molecular adsorbed at the surface of TiO₂ nanoparticles, which therefore resulted in a fuzzy HRTEM image; (3) when the TiO₂ was exposed in air for a period of time, the methylene blue molecules disappeared and the TiO₂ lattice image again became integrated as the pristine one; (4) however, if the TiO₂ nanoparticles became deactivated after degradation of methylene blue for more than 20 cycles, the HRTEM lattice image was fuzzy fully and could not recover even it was exposed in air for a long time. The results reveal that the lattice distortion on the anatase TiO₂ (101) surface induced by chemical adsorption of methylene blue molecules is a crucial intermediate step during photocatalyzing. The distorted lattice atoms absorb photons under illumination of light and tend to recover and cut the molecule bond which causes the degradation of methylene blue. Furthermore, we propose that there exists a surface lattice driving force forming on the surface distortion leads to the TiO₂ lattice HRTEM image from fuzzy to integrate, and it decides the photocatalytic ability.

---

1. Introduction

As an excellent and highly efficient photocatalyst, TiO₂ has triggered broad interest studies. Most of the studies have been devoted to enhancing the photocatalytic activity of TiO₂ by doping, semiconductor compounding, dye sensitization and exposure of reactive facets.^1–3 However, there are less works related to the photocatalytic progress and deactivation mechanism from atomic-scale.

Some of the research tends to speculate the chemical reaction path of photocatalysis indirectly by “in situ” monitoring the transformation of functional groups on TiO₂ during photocatalytic process by using a Fourier transform infrared (FTIR) spectroscopy.^4–13 But the variation of structure and electronic states of TiO₂ during photocatalyzing can not be known from this indirect method. One the other hand, the changes in the surface lattice of TiO₂ by photo-induced had been researched. Nakato Y et al.^14,15 reported the photo-induced changes in the surface of a single-crystal TiO₂ using AFM and that was proposed to the key process in photocatalysis. Hashimoto K et al.¹⁶ found the changes in the surface lattice of TiO₂ by photo-induced caused supper-hydrophilic conversion. Recently, the more precise works have been processed for characterizing local ordering or chemical reaction of submonolayer at the TiO₂ surface in an atomic-scale by using a scanning tunnelling microscope (STM). Bikondoa O et al.¹⁷ utilized a STM to image the reaction of water molecules with bridging-oxygen vacancies on a model oxide surface of rutile TiO₂ (110). He Y et al.¹⁸ found the locally ordered (2 × 2) superstructure of molecular water on the (101) surface of anatase TiO₂ by using a STM and the simulation calculation confirmed that the ordered structure was of the lowest adsorption energy and the most stability. In addition to the absence of research on photocatalytic mechanism, the main difficulty is the high requirement of STM samples, which limits the further research work.

Comparing to STM, high-resolution transmission electron microscopy (HR-TEM) is an effective approach for directly observing crystal lattice in an atomic-scale with simple sampling preparation for nanoparticles.¹⁹ In this work, in order to reveal the microstructural evolution of P25 TiO₂ during photocatalyzing and the deactivation mechanism, six samples were examined directly by using a HRTEM, including pristine P25 TiO₂ nanoparticles, methylene blue adsorbed nanoparticles in dark environment, methylene blue degraded nanoparticles under UV-vis light irradiation, nanoparticles exposed in air for 30 days, deactivation photocatalyst after degradation of methylene blue for 20 times, and deactivated photocatalyst exposed in air for 30 days. It is expected to further understand the photocatalytic progress and deactivation mechanism from an atomic scale.

2. Experimental and computational methods

Commercial P25 TiO₂ nanoparticles were purchased from Degussa. The crystal phases of the pristine P25 TiO₂ nanoparticles were measured using an X-ray diffractometer (XRD) (D8 Advanced XRD, Bruker AXS, Germany) with CuKα radiation. The photocatalytic experiments were executed by the decomposition of methylene blue with relatively large mono-molecule in order to be characterized by HRTEM. Firstly, 100 mg P25 TiO₂ were put into 100 mL of a 12 mg L⁻¹methylene blue solution. The mixed solution was stirred incessantly and put into a black room for 10 h to mark the full adsorption of methylene blue. The methylene blue adsorbed P25 TiO₂ nanoparticles were measured by an X-ray photoelectron spectroscopy (XPS) (VG Multilab2000 spectrometer). Then the mixed solution was irradiated by a 250 W high pressure mercury lamp, which generates light in the 350–450 nm range with a maximum intensity at 365 nm. The average light intensity at the position of the sample surface was about 23 mW cm⁻². The photocatalytic reaction is aerobic, so the solution was exposure in the air and stirred continuously during the photocatalytic reaction. The temperature was 22 °C in lab room. The methylene blue was fully degraded after irradiation for 4 h. In the degradation cycles, P25 TiO₂ nanoparticles were used repeatedly until to be deactivated.

Totally, six samples were examined by a HRTEM (JEM 2010FEF HRTEM, JEOL, Japan) including (1) pristine P25 TiO₂ nanoparticles, (2) methylene blue adsorbed nanoparticles in dark environment, (3) methylene blue degraded nanoparticles under UV-vis light irradiation, (4) methylene blue degraded nanoparticles exposed in air for 30 days, (5) the deactivated nanoparticles after degradation for 20 cycles, (6) deactivated nanoparticles exposed in air for 30 days. The TEM specimens were prepared by separating the TiO₂ nanoparticles with a centrifugal machine, and then dispersing them in an ethanol ultrasonic bath for 10 min and dropping the suspension onto 3 mm diameter copper microgrids.

Theoretical calculations

In order to verify the experimental results, the theoretical calculations were carried out for simulating the lattice variation when methylene blue molecules were adsorbed upon the TiO₂ surface. The theoretical calculations were based on the semi-empirical self-consistent field (SCF) molecular orbital (MO) program, MSINDO.²⁰ The modified SCF MO method, SINDO1,²¹ was used. Pseudo-minimal Slater basis set (3d, 4s, 4p) was used to describe Ti, while C, N, O by a (2s, 2p) basis set and S by a (3s, 3p) basis set with 3d for polarization.

Restricted Hartree–Fock (RHF) formalism was used in calculations, since all the particles and molecules are all close shell system. Pongor's level-shifting algorithm²² was used to accelerate the SCF processes with the convergence criterion of 10⁻⁸ Hartree. The Broyden–Fletcher–Goldfarb–Shanno (BFGS) method together with Pulay's convergence accelerator²³ was used to optimize the geometries structures. The relaxation convergence criterion is that the maximal force acting on atoms is less than 10⁻³ au; the maximal displacement of atoms is less than 10⁻² au, and the root-mean-square of the force and the displacement less than a third of the average.

Considering the surface energy of the anatase,^24,25 two nano-metric particles was constructed. One is with 132 atoms and another is with 576 atoms. Each particle has four (101) surface and two (001) surface, and two of the (101) surface has a large area that can adsorb methylene blue molecule. Both the particles and the molecule are optimized separately, and compared with the optimized adsorbed structures. The relative position of Ti atom was checked to see if there exists any visible crystal distortion.

3. Results and discussion

Fig. 1 illustrates a XRD pattern of the pristine P25 TiO₂ nanoparticles. It indicates a mixture of anatase and rutile phases which has been reported elsewhere.²⁶HRTEM observations revealed that the pristine cuboid or polyhedron P25 TiO₂ nanoparticles were well dispersed with a size range of 10–30 nm and few large size of about 80 nm; and the anatase TiO₂ nanoparticles can be discriminated and exhibited a perfect lattice, as shown in Fig. 2. Due to the large size around 80 nm; the rutileTiO₂ nanoparticles caused difficult imaging in HRTEM.

Fig. 1 XRD patterns of pristine P25 TiO₂.

Fig. 2 The (a) low resolution and (b) high resolution HRTEM images of pristine P25 TiO₂ nanoparticles.

However, it was interested to note that after adsorption of methylene blue in a black room, many small dots in size about 1 nm were found on the TiO₂ surface as black arrows indicate and its atomic lattice HRTEM images became fuzzy, as shown in Fig. 3. The similar images were also observed in the degraded sample under UV-vis light irradiation, as shown in Fig. 4. It should be indicated that the fuzzy images were not caused by wrong focus or high variance of the particles in the same image during observations. We actually examined several areas and no clear HRTEM image was obtained. Exactly and obviously, these changes were induced from the lattice distortion happened at the TiO₂ surface. Occasionally, when the degraded sample was exposed in the air and irradiated under solar light for about 30 days, HRTEM observation found that the atomic lattice of the TiO₂ nanoparticles again became integrated as the pristine one, and the dots has been completely disappeared from the surface, as shown in Fig. 5.

Fig. 3 The (a) low resolution and (b) high resolution HRTEM images of methylene blue adsorbed nanoparticles in dark environment.

Fig. 4 The (a) low resolution and (b) high resolution HRTEM images of methylene blue degraded nanoparticles under UV-vis light irradiation.

Fig. 5 The (a) low resolution and (b) high resolution HRTEM images of nanoparticles exposed in air for 30 days.

In order to reveal the character of the mall dots, XPS experiment was performed for the methylene blue adsorbed P25 TiO₂ sample, as shown in Fig. 6. The XPS survey spectrum demonstrates that the sample contains six elements Ti, O, C, N, S and Cl, in which the chemical binding energies for Ti 2p_3/2, O 1s, and C 1s are 458.6, 530.4, and 284.7 eV respectively. The N 1s, S 2p and Cl 2p core level measurements show that the chemical binding energies are 399.7, 167.5, and 198.3 eV, respectively, as illustrated in Fig. 6b, c and d. And the atom contents of N, S, and Cl was estimated to be 1.17%, 0.49% and 0.50% by comparing the product of the peak area multiplied by the corresponding element sensitive factor to the product of all peaks multiplied by respective element sensitive factor. Then, the atom content ratio of N, S and Cl is about 3 : 1 : 1, which is accord with the atom ratio in a methylene blue molecular of C₁₆H₁₈ClN₃S. The atom content of C element is unreliable because of adventitious elemental carbon in XPS measure. Consequently, the XPS results confirm that the ∼1 nm dots adsorbing on the surface of P25 TiO₂ in Fig. 3 were the methylene blue molecules, which also played a key role for microstructure variations of the anatase TiO₂ samples in a sequence clear lattice–fussy–clear lattice during photocatalyzing. Though the separate methylene blue molecules exhibited a weak HRTEM contrast comparing with that of TiO₂, the dots could be observed under an appropriate focus.

Fig. 6 XPS spectra of (a) survey spectrum, (b) N 1s, (c) S 2p and (d) Cl 2p for TiO₂ after adsorption of methylene blue.

Further theoretical calculations and simulations were carried out for exploring the anatase TiO₂ surface state when a methylene blue molecular was adsorbed on the TiO₂ (101) surface. In the present models, two sizes of anatase TiO₂ particles wit 44 and 125 TiO₂ molecules were simulated to calculate different surface energy of anatase TiO₂ particle. There are four (101) and two (001) surface in both of the two sizes of TiO₂ particles and two (101) surface is large enough to adsorb a methylene blue molecule. Although there are isolated –OH and H₂O_ad adsorbed on the surface of TiO₂ nanoparticles, methylene blue molecules can direct contact with the TiO₂ nanoparticles in displacement of isolated –OH and H₂O_ad through a wetting and drying process.¹² The anatase TiO₂ particles and the methylene blue molecule were optimized severally and then compared with the structure after adsorption which was also optimized. The distances between two adjacent Ti atoms of anatase TiO₂ particles, which form chemical bond with the methylene blue molecule before and after adsorption were measured to judge the lattice distortion of anatase TiO₂. The model and the structure of methylene blue molecule are showed as Fig. 7. In a methylene blue molecule, all C, N and S atoms are on the same surface and the Cl atom is ignored because it does not form chemical bond in aqueous solution.

Fig. 7 The model and the structure of methylene blue.

The simulation results are showed as Fig. 8. A methylene blue molecule adsorbs on the (101) surface of the anatase TiO₂ particle, and the N atom in the middle ring of the methylene blue molecule form a chemical bond with a Ti atom A of the anatase TiO₂, at the same time, two corresponding C atoms in other two rings form chemical bonds with the two O atom next to the Ti atom A. The chemical bonds induced by the adsorption cause lattice distortion in a region of near the Ti and O atoms forming chemical bonds with the methylene blue molecule.

Fig. 8 The structures of anatase TiO₂ particles with the size of (a) pristine 44 TiO₂ molecules, (b) 44 TiO₂ molecules after adsorption of methylene blue, (c) pristine 125 TiO₂ molecules, (d) 125 TiO₂ molecules after adsorption of methylene blue.

Because the contrast of metal atom Ti is much higher than that of nonmetal atom O in HRTEM, the distances between Ti atom A and other two Ti atoms B and C next to the A were measured to speculate the lattice distortion. The distances in different states were listed as Table 1. It can be seen that the distances between two adjacent Ti atoms were prolong at a ratio of about 5% after the adsorption of a methylene blue molecular both in the two sizes of anatase TiO₂ particles. These theoretical calculations demonstrate that the adsorption of methylene blue molecule can change the atom array and induce lattice distortion in a region of several atom layers on the surface of anatase TiO₂ nanoparticles. This lattice distortion makes Ti atoms on a same crystal face deviate from the face which will reduce the resolution of HRTEM images. The more deviation of surface atoms from its crystal face, the lower resolution will be until the HRTEM images become fuzzy. So the lattice distortion on the surface of TiO₂ can produce fuzzy HRTEM images.

Table 1 The distances between two adjacent Ti atoms of anatase TiO₂ particles with the size of 44 and 125 TiO₂ molecules before and after adsorption of methylene blue

Two adjacent Ti atoms	The distance between two adjacent Ti atoms/Å	The distance between two adjacent Ti atoms after adsorption of methylene blue/Å
A0, B0	3.4139	3.6370
A0, C0	3.3853	3.5483
A1, B1	3.6416	3.8192
A1, C1	3.6416	3.8153

---

From above experimental and theoretical results, we have to re-consider what happens in an atomic scale during photocatalyzing, that is to say, what is an actual photocatalytic process of TiO₂ to degrade methylene blue?

Obviously, the lattice distortion in a region of several atom layers at the surface of anatase TiO₂ nanoparticles induced by adsorption of methylene blue provides a key intermediate state during photocatalyzing. Therefore, we propose that the photocatalytic process can be described as follows: (1) firstly, methylene blue molecules adsorb on the surface of anatase TiO₂ and contact with the TiO₂ nanoparticles directly which induce surface lattice distortion in a region of several atom layers due to lattice matching; (2) under illumination by light, the distorted lattice atoms absorb photons and tend to be recovered which induce a kind of “surface lattice driving force”. Photo-induced surface-lattice distortion/relaxation of TiO₂ has also been observed.^14,15 Absorption of photons can increase the lattice energy of TiO₂ which causes the distorted lattice atoms recover. This surface lattice driving force can cut the molecule bond and resulting in the degradation of methylene blue; (3) finally, the distorted lattice is recovered along with the degradation of methylene blue molecules.

In the photocatalytic progress, TiO₂ can be visualized as a “lattice engine” according to the variation “distortion–recovery–distortion–recovery…”of the lattice which leads to the degradation of organic molecule. Similar work was also processed for rutileTiO₂ nanoparticles. The rutile TiO₂ obtained through heated P25 TiO₂ to 900 °C for 30 min and used to degrade methylene blue under the same experiment environment and parameter. The lattice changes were showed in Fig. 9. It can be found that small dots adsorbed on the surface of rutile TiO₂ which is similar as that on the surface of anatase TiO₂ in Fig. 3 and Fig. 4. However, the lattice image is fine in Fig. 9b indicates there is little lattice distortion even methylene blue molecules adsorbed on the surface. That phenomenon can explain why rutile TiO₂ have a poor perform compared with P25 TiO₂ in degradation of methylene blue as showed in Fig. 10. The surface lattice of rutile TiO₂ can be hardly distorted by adsorption of methylene blue molecules due to a poor photocatalytic activity.

Fig. 9 The high resolution HRTEM images of (a) pristine rutileTiO₂ nanoparticles and (b) rutileTiO₂ nanoparticles after degradation of methylene blue.

Fig. 10 Degradation of methylene blue with rutile and P25 TiO₂.

The multiple cycles for degrading methylene blue using P25 TiO₂ nanoparticles were carried out for further studying the deactivation mechanism of the photocatalyst. As shown in Fig. 11, the photocatalytic ability of the TiO₂ weakened gradually as the degradation cycles increasing and the photocatalyst became deactivated after 20 cycles. Fig. 12a shows the HRTEM image of the deactivated TiO₂. Obviously, the lattice image became completely fuzzy and exhibited an rough surface with many remained methylene blue molecular on the surface. However, different from one cycle, as shown in Fig. 5, the surface lattice of the deactivated TiO₂ could not be recovered even also it was exposed in air for 30 days, as shown in Fig. 12b.

Fig. 11 Cycle degradation of methylene blue with P25 TiO₂ for 20 times.

Fig. 12 HRTEM image of (a) deactivated TiO₂ and (b) deactivated TiO₂ after 30 days.

These results indicate that the surface lattice driving force decreases gradually as the degradation cycles increasing due to the formation of surface lattice vacancies and defects of TiO₂, which weakens the photocatalytic ability. The photocatalyst becomes deactivated when the surface lattice driving force disappeared and the surface lattice also can not become integrated. As we known, the well-known photo-induced carriers' theory can be used to describe the energy conversion during photocatalytic progress, but it is limited in the deactivation of photocatalyst. In this case, the present surface lattice driving force can be used to explain the deactivation mechanism of the photocatalyst and the photocatalytic progress in an atom-scale.

4. Conclusions

HRTEM characterization of the photocatalytic process in an atomic scale reveals that the surface lattice distortion of anatase TiO₂ nanoparticles induced by chemical adsorption of methylene blue molecules plays a crucial intermediate role during photocatalyzing. That is to say, the distorted lattice induces a kind of “surface lattice driving force”, which provides energy to cut the molecule bond and degrades methylene blue substance. This photocatalytic progress can be demonstrated from the HRTEM observations through the lattice image variations “distortion–recovery–distortion–recovery…”. The efficiency of this “lattice engine” will reduce gradually along with the increase times of degradation, and finally, the “lattice engine” is invalidated because the surface lattice can not be recovered. These direct evidences provide a novel point of view to further recognize the photocatalytic and deactivation mechanism in an atomic scale, that is, the atomic interaction between anatase TiO₂ and the degraded substances.

Acknowledgements

We would like to thank Mr Dongshan Zhao for his excellent technical assistance in HRTEM observations. This work was supported by the National Basic Research Program of China (973 Program) (no. 2009CB939705); and the Fundamental Research Funds for the Central Universities in 2008 (no. 20082020201000004).

References

1. S. U. M. Khan, M. Al-Shahry and W. B. Ingler, Science, 2002, 297, 2243.
2. J. Zhang, C. Pan, P. Fang, J. Wei and R. Xiong, ACS Appl. Mater. Interfaces, 2010, 2, 1173.
3. H. G. Yang, C. H. Sun, S. Z. Qiao, J. Zou, G. Liu, S. C. Smith, H. M. Cheng and G. Q. Lu, Nature, 2008, 453, 638.
4. G. N. Ekstrőm and A. J. McQuillan, J. Phys. Chem. B, 1999, 103, 10562.
5. S. Sato, K. Ueda, Y. Kawasaki and R. Nakamura, J. Phys. Chem. B, 2002, 106, 9054.
6. R. Nakamura, A. Imanishi, K. Murakoshi and Y. Nakato, J. Am. Chem. Soc., 2003, 125, 7443.
7. R. Nakamura and Y. Nakato, J. Am. Chem. Soc., 2004, 126, 1290.
8. Z. Yu and S. S. C. Chuang, J. Catal., 2007, 246, 118.
9. I. Dolamic and T. Bürgi, J. Catal., 2007, 248, 268.
10. C. H. Ho, C. Y. Shieh, C. L. Tseng, Y. K. Chen and J. L. Lin, J. Catal., 2009, 261, 150.
11. A. R. Almeida, J. A. Moulijn and G. Mul, J. Phys. Chem. C, 2008, 112, 1552.
12. Z. Yu and S. S. C. Chuang, J. Phys. Chem. C, 2007, 111, 13813.
13. Z. Yu and S. S. C. Chuang, Appl. Catal., B, 2008, 83, 277.
14. R. Nakamura, T. Okamura, N. Ohashi, A. Imanishi and Y. Nakato, J. Am. Chem. Soc., 2005, 127, 12975.
15. A. Imanishi, T. Okamura, N. Ohashi, R. Nakamura and Y. Nakato, J. Am. Chem. Soc., 2007, 129, 11569.
16. N. Sakai, A. Fujishima, T. Watanabe and K. Hashimoto, J. Phys. Chem. B, 2001, 105, 3023.
17. O. Bikondoa, C. L. Pang, R. Ithnin, C. A. Muryn, H. Onishi and G. Thornton, Nat. Mater., 2006, 5, 189.
18. Y. He, A. Tilocca, O. Dulub, A. Selloni and U. Diebold, Nat. Mater., 2009, 8, 585.
19. W. A. Daoud, J. H. Xin and G. K. H. Pang, J. Am. Ceram. Soc., 2005, 88, 443.
20. T. Bredow and K. Jug, Theor. Chem. Acc., 2004, 113, 1.
21. D. N. Nanda and K. Jug, Theor. Chim. Acta, 1980, 57, 95.
22. G. Pongor, Chem. Phys. Lett., 1974, 24, 603.
23. P. Császár and P. Pulay, J. Mol. Struct., 1984, 114, 31.
24. M. Lazzeri, A. Vittadini and A. Selloni, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 155409.
25. Y. Lei, H. Liu and W. Xiao, Model. Simul. Mater. Sci. Eng., 2010, 18, 025004.
26. T. Ohno, K. Sarukawa, K. Tokieda and M. Matsumura, J. Catal., 2001, 203, 82.

---