UV Raman Spectroscopic Study on TiO2. I. Phase Transformation at

2.2.1. UV Raman Spectroscopy. UV Raman spectra were measured at room ..... of bands assigned to rutile phase (445 and 612 cm-1) increase after calcina...
0 downloads 0 Views 448KB Size
J. Phys. Chem. B 2006, 110, 927-935

927

UV Raman Spectroscopic Study on TiO2. I. Phase Transformation at the Surface and in the Bulk Jing Zhang, Meijun Li, Zhaochi Feng, Jun Chen, and Can Li* State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, P. O. Box 110, Dalian 116023, China ReceiVed: September 16, 2005; In Final Form: NoVember 4, 2005

Phase transformation of TiO2 from anatase to rutile is studied by UV Raman spectroscopy excited by 325 and 244 nm lasers, visible Raman spectroscopy excited by 532 nm laser, X-ray diffraction (XRD), and transmission electron microscopy (TEM). UV Raman spectroscopy is found to be more sensitive to the surface region of TiO2 than visible Raman spectroscopy and XRD because TiO2 strongly absorbs UV light. The anatase phase is detected by UV Raman spectroscopy for the sample calcined at higher temperatures than when it is detected by visible Raman spectroscopy and XRD. The inconsistency in the results from the above three techniques suggests that the anatase phase of TiO2 at the surface region can remain at relatively higher calcination temperatures than that in the bulk during the phase transformation. The TEM results show that small particles agglomerate into big particles when the TiO2 sample is calcined at elevated temperatures and the agglomeration of the TiO2 particles is along with the phase transformation from anatase to rutile. It is suggested that the rutile phase starts to form at the interfaces between the anatase particles in the agglomerated TiO2 particles; namely, the anatase phase in the inner region of the agglomerated TiO2 particles turns out to change into the rutile phase more easily than that in the outer surface region of the agglomerated TiO2 particles. When the anatase particles of TiO2 are covered with highly dispersed La2O3, the phase transformation in both the bulk and surface regions is significantly retarded, owing to avoiding direct contact of the anatase particles and occupying the surface defect sites of the anatase particles by La2O3.

1. Introduction Titania (TiO2) has been widely studied because of its unique optical and chemical properties in catalysis,1 photocatalysis,2 sensitivity to humidity and gas,3,4 nonlinear optics,5 photoluminescence,6 and so on. The two main kinds of crystalline TiO2, anatase and rutile, exhibit different physical and chemical properties. It is well-known that the anatase phase is suitable for catalysts and supports,7 while the rutile phase is used for optical and electronic purposes because of its high dielectric constant and high refractive index.8 It has been well demonstrated that the crystalline phase of TiO2 plays a significant role in catalytic reactions, especially photocatalysis.9-11 Some studies have claimed that the anatase phase was more active than the rutile phase in photocatalysis.9,10 Although at ambient pressure and temperature the rutile phase is more thermodynamically stable than the anatase phase,12 anatase is the common phase rather than rutile because anatase is kinetically stable in nanocrystalline TiO2 at relatively low temperatures.13 It is believed that the anatase phase transforms to the rutile phase over a wide range of temperatures.14 Therefore, understanding and controlling of the crystalline phase and the process of phase transformation of TiO2 are important, though they are difficult. Many studies13-31 have been done to understand the process of the phase transformation of TiO2. Zhang et al.15 proposed that the mechanism of the anatase-rutile phase transformation was temperature-dependent according to the kinetic data from * To whom correspondence should be addressed. Telephone: +86-41184379070. Fax: +86-411-84694447. E-mail: [email protected].

X-ray diffraction (XRD). On the basis of transmission and scanning electron microscopies, Gouma et al.16 suggested that rutile nuclei formed on the surface of coarser anatase particles and the newly transformed rutile particles grew at the expense of neighboring anatase particles. Penn et al.17 suggested that the formation of rutile nuclei at {112} twin interfaces of anatase particles heated hydrothermally. Catalytic performance of TiO2 largely depends on the surface properties, especially the surface phase, because catalytic reaction takes place on the surface. The surface phase of TiO2 should be responsible for its photocatalytic activity because not only the photoinduced reactions take place on the surface32 but also the photoexcited electrons and holes might migrate through the surface region. Therefore, the surface phase of TiO2, which is exposed to the light source, should play a crucial role in photocatalysis. However, the surface phase of TiO2, particularly during the phase transformation, has not been investigated. The challenging questions still remain: is the phase in the surface region the same as that in the bulk region, or how does the phase in the surface region of TiO2 particle change during the phase transformation of its bulk? The difficulty in answering the above questions was mainly due to lacking suitable techniques that can sensitively detect the surface phase of TiO2. UV Raman spectroscopy is found to be more sensitive to the surface phase of a solid sample when the sample absorbs UV light.33 We studied the phase transition of zirconia (ZrO2) from tetragonal phase to monoclinic phase by UV Raman spectroscopy, visible Raman spectroscopy, and XRD.33 These results clearly indicated that the surface phase of ZrO2 is usually different from the bulk phase of ZrO2 and the phase transforma-

10.1021/jp0552473 CCC: $33.50 © 2006 American Chemical Society Published on Web 12/10/2005

928 J. Phys. Chem. B, Vol. 110, No. 2, 2006 tion of ZrO2 starts from its surface region and then gradually develops into its bulk when the ZrO2 with tetragonal phase is calcined at elevated temperatures. These findings lead us to further investigate the phase transformation in the surface region of TiO2 by UV Raman spectroscopy as TiO2 also strongly absorbs UV light. In this study, we compared the Raman spectra of TiO2 calcined at different temperatures with excitation lines in the UV and visible regions. XRD and transmission electron microscopy (TEM) were also recorded to understand the process of phase transformation of TiO2. It was found that the results of UV Raman spectra are different from those of visible Raman spectra and XRD patterns. The anatase phase of TiO2 at the surface region can remain at relatively higher temperatures than that in the bulk at elevated calcination temperatures; namely, the anatase phase in the inner region of the agglomerated TiO2 particles turns out to change into the rutile phase more easily than that in the outer surface region of the agglomerated TiO2 particles. The literature15,17,29 proposed the mechanism that phase transformation of TiO2 might start at the interfaces of contacting anatase particles. If the anatase particles of TiO2 are separated, the phase transformation of TiO2 from anatase to rutile could be retarded or prohibited. Jing et al.34 showed that La3+ did not enter the crystal lattices of TiO2 and was uniformly dispersed onto TiO2 in the form of lanthana (La2O3) particles with small size. To verify the above assumption, this study also prepared the anatase phase of TiO2 sample covered with La2O3 and characterized the above sample by visible Raman spectroscopy and UV Raman spectroscopy. The results of the two types of Raman spectra are in agreement with each other and show that the TiO2 particle covered with La2O3 can retain its anatase phase both in the bulk and in the surface region even after calcination at 900 °C. 2. Experimental Section 2.1. Catalyst Preparation. 2.1.1. Preparation of TiO2. TiO2 was prepared by precipitation method. To 100 mL of anhydrous ethanol was added 20 mL of titanium(IV) n-butoxide (Ti(OBu)4). This solution was added to a mixture solution of deionized water and 100 mL of anhydrous ethanol. The molar ratio of the water/Ti(OBu)4 was 75. After the formed white precipitate was stirred continuously for 24 h, it was filtered and washed twice with deionized water and anhydrous ethanol. Finally, the sample was dried at 100 °C and calcined in air at temperatures from 200 to 800 °C for 4 h, and then cooled to room temperature. 2.1.2. Preparation of La2O3-CoVered TiO2 (La2O3/TiO2). The above TiO2 powder calcined at 500 °C was used as a support. The critical La2O3 loading corresponding to monolayer coverage of La2O3 on the grain surface of TiO2 is 0.27 g/100 m2.35,36 On the basis of the BET surface area of the TiO2 support (54.3 m2/g), the monolayer dispersion capacity can also be expressed as 15 wt % La2O3 of the weight of TiO2. La2O3/TiO2 samples, containing different amounts of La2O3 (0.5-6 wt %) were prepared by a wet impregnation method. The support was impregnated with aqueous solution of various concentrations of lanthanum nitrate (La(NO3)3‚6H2O) and subsequently stirred in a hot water bath until it was dried. After the sample was kept at 110 °C overnight, it was calcined at 900 °C in air for 4 h. A TiO2 sample was prepared by calcining the TiO2 support at 900 °C for 4 h (denoted as TiO2-900) for comparison with the La2O3/TiO2 sample. Pure La2O3 was obtained by calcining La(NO3)3‚6H2O at 550 °C for 4 h. 2.2. Characterization. 2.2.1. UV Raman Spectroscopy. UV Raman spectra were measured at room temperature with a Jobin-

Zhang et al.

Figure 1. Raman spectra of TiO2 calcined at (A) 500 and (B) 800 °C with excitation lines at 532, 325, and 244 nm (λex ) 532, 325, and 244 nm).

Yvon T64000 triple-stage spectrograph with spectral resolution of 2 cm-1. The laser line at 325 nm of a He-Cd laser was used as an exciting source with an output of 25 mW. The power of laser at the sample was about 3.0 mW. The 244 nm line from a Coherent Innova 300 Fred laser was used as another excitation source. The power of the 244 nm line at sample was below 1.0 mW. 2.2.2. Visible Raman Spectroscopy. Visible Raman spectra were recorded at room temperature on a Jobin-Yvon U1000 scanning double monochromator with the spectral resolution of 4 cm-1. The line at 532 nm from a DPSS 532 Model 200 532 nm single-frequency laser was used as the excitation source. 2.2.3. X-ray Powder Diffraction (XRD), TEM, and UltraVioletVisible Diffuse Reflectance Spectroscopy. XRD patterns were obtained on a Rigaku MiniFlex diffractometer with a Cu KR radiation source. Diffraction patterns were collected from 20° to 80° at a speed of 5°/min. TEM was taken on a JEM-2011 TEM for estimating particle size and morphology. UV-vis diffuse reflectance spectra were recorded on a JASCO V-550 UV-vis spectrophotometer. 2.2.4. Brunauer-Emmett-Teller (BET) Specific Surface Area. The BET surface area of the TiO2 support was measured by nitrogen adsorption at 77 K using a Micromeritics ASAP 2000 adsorption analyzer. 3. Results 3.1. Spectral Characteristics of Anatase and Rutile TiO2. The anatase and rutile phases of TiO2 can be sensitively identified by Raman spectroscopy based on their Raman spectra. The anatase phase shows major Raman bands at 144, 197, 399, 515, 519 (superimposed with the 515 cm-1 band), and 639 cm-1.37 These bands can be attributed to the six Raman-active modes of anatase phase with the symmetries of Eg, Eg, B1g, A1g, B1g, and Eg, respectively.37 The typical Raman bands due to rutile phase appear at 143 (superimposed with the 144 cm-1 band due to anatase phase), 235, 447, and 612 cm-1, which can be ascribed to the B1g, two-phonon scattering, Eg, and A1g modes of rutile phase, respectively.38 Additionally, the band at 144 cm-1 is the strongest one for the anatase phase and the band at 143 cm-1 is the weakest one for the rutile phase. Parts A and B, respectively, of Figure 1display the Raman spectra of TiO2 calcined at 500 and 800 °C with excitation lines at 532, 325, and 244 nm. Obviously, both visible Raman spectra and UV Raman spectra show that the TiO2 sample is in the anatase phase (Figure 1A) and rutile phase (Figure 1B).

Phase Transformation of TiO2

J. Phys. Chem. B, Vol. 110, No. 2, 2006 929

Figure 2. UV-vis diffuse reflectance spectra of TiO2 calcined at 500 and 800 °C.

Figure 2 shows UV-vis diffuse reflectance spectra of the TiO2 sample calcined at 500 and 800 °C (the TiO2 sample is in the anatase phase and rutile phase, respectively). For the anatase phase, the maximum absorption and the absorption band edge can be estimated to be around 324 and 400 nm, respectively. The maximum absorption and the absorption band edge shift to a little longer wavelength for the rutile phase.39 By comparing the Raman spectra of the anatase (Figure 1A) or rutile phase (Figure 1B) excited by 532, 325, and 244 nm lines, it is found that the relative intensities of characteristic bands due to anatase or rutile phase in the high-frequency region are different. For the anatase phase (Figure 1A), the band at 638 cm-1 is the strongest one in the Raman spectrum with the excitation line at 325 or 532 nm, while the band at 395 cm-1 is the strongest one in the Raman spectrum with the excitation line at 244 nm. For the rutile phase (Figure 1B), the intensities of the bands at 445 and 612 cm-1 are comparable in the visible Raman spectrum. The intensity of the band at 612 cm-1 is stronger than that of the band at 445 cm-1 in the Raman spectrum with the excitation line at 325 nm, and the reverse is true for the Raman spectrum with the excitation line at 244 nm. In addition, for the rutile phase, a band at approximately 826 cm-1 appears in the UV Raman spectra. Some investigations show that the rutile phase of TiO2 exhibits a weak band at 826 cm-1 assigned to the B2g mode.38,40 The fact that the relative intensities of the Raman bands of anatase phase or rutile phase are different for UV Raman spectroscopy and visible Raman spectroscopy are mainly due to the UV resonance Raman effect because the laser lines at 325 and 244 nm are in the electronic absorption region of TiO2 (Figure 2). There is no resonance Raman effect observed for the TiO2 sample excited by visible laser line, because the line at 532 nm is outside the absorption region of TiO2 (Figure 2). Therefore, for the anatase or rutile phase, the Raman spectroscopic characteristics in the visible Raman spectrum are different from those in the UV Raman spectrum. When the UV laser line with different wavelengths is used as the excitation source, the resonance enhancement effect on the Raman bands of anatase or rutile phase is different. For example, for the rutile phase (Figure 1B), the band at 612 cm-1 is easily resonance enhanced when the excitation wavelength is 325 nm. Among all the characteristic bands of the rutile phase, the extent of

Figure 3. (A) Visible Raman spectra of mechanical mixture with 1:1, 1:5, 1:10, 1:15, 5:1, and 10:1 ratios of anatase phase to rutile phase with the excitation line at 532 nm. (B) Plot of area ratios of the visible Raman band at 395 cm-1 for anatase phase to the band at 445 cm-1 for rutile phase (A395 cm-1/A445 cm-1) versus weight ratios of anatase phase to rutile phase (WA/WR).

resonance enhancement of 445 cm-1 is the strongest when the 244 nm laser is used as the excitation source (Figure 1B). 3.2. Semiquantitative Analysis of the Phase Composition of TiO2 by XRD and Raman Spectroscopy. The weight fraction of the rutile phase in the TiO2 sample, WR, can be estimated from the XRD peak intensities using the following formula:41

WR ) 1/[1 + 0.884(Aana/Arut)] where Aana and Arut represent the X-ray integrated intensities of anatase (101) and rutile (110) diffraction peaks, respectively. To estimate the weight fraction of the rutile phase in the TiO2 sample by Raman spectroscopy, pure anatase phase and pure rutile phase of the TiO2 sample, which have been prepared by calcination of TiO2 powder at 500 and 800 °C for 4 h, were mechanically mixed at given weight ratio and ground carefully to mix sufficiently. Figure 3A displays the visible Raman spectra of the mechanical mixture with 1:1, 1:5, 1:10, 1:15, 5:1, and 10:1 ratios of anatase phase to rutile phase. The relationship between the area ratios of the visible Raman band at 395 cm-1 for anatase phase to the band at 445 cm-1 for rutile phase (A395 cm-1/A445 cm-1) and the weight ratios of anatase phase to rutile phase (WA/WR) is plotted in Figure 3B. It can be seen that a linear relationship between the band area ratios and the weight ratios of anatase phase to rutile phase in the mixture is obtained. The rutile content in the Degussa P25, which usually consists of roughly about 80% anatase and 20% rutile phase,42 was estimated by this plot. Our Raman result indicates that the rutile content in the Degussa P25 is about 18.7%, which is close to the known

930 J. Phys. Chem. B, Vol. 110, No. 2, 2006

Zhang et al.

Figure 5. XRD patterns of TiO2 calcined at different temperatures.

Figure 4. (A) UV Raman spectra of mechanical mixture with 1:1, 1:2, 1:4, 1:6, 1:10, and 1:15 ratios of anatase phase to rutile phase with the excitation line at 325 nm. (B) Plot of area ratios of the UV Raman band at 612 cm-1 for rutile phase to the band at 638 cm-1 for anatase phase (A612 cm-1/A638 cm-1) versus weight ratios of rutile phase to anatase phase (WR/WA).

result. Thus, the above linear relationship based on visible Raman spectroscopy can be used to estimate the rutile content in TiO2. Figure 4A presents the UV Raman spectra of the mechanical mixture with 1:1, 1:2, 1:4, 1:6, 1:10, and 1:15 ratios of anatase phase to rutile phase with the excitation line at 325 nm. Figure 4B shows the plot of the area ratios of the UV Raman band at 612 cm-1 for rutile phase to the band at 638 cm-1 for anatase phase (A612 cm-1/A638 cm-1) versus the weight ratios of rutile phase to anatase phase (WR/WA). There is also a linear relationship between the band area ratios and the weight ratios of rutile phase to anatase phase. 3.3. Phase Transformation of TiO2 at Elevated Calcination Temperatures. 3.3.1. XRD Patterns and Visible Raman Spectra of TiO2 Calcined at Different Temperatures. Figure 5 shows the XRD patterns of TiO2 calcined at different temperatures. The “A” and “R” in the figure denote the anatase and rutile phases, respectively. For the sample before calcination, diffraction peaks due to the crystalline phase are not observed, suggesting that the sample is still in the amorphous phase. When the sample was calcined at 200 °C, weak and broad peaks at 2θ ) 25.5°, 37.9°, 48.2°, 53.8°, and 55.0° were observed. These peaks represent the indices of (101), (004), (200), (105), and (211) planes of anatase phase, respectively.43 These results suggest that some portions of the amorphous phase transform into the anatase phase. The diffraction peaks due to anatase phase develop with increasing the temperature of calcination. When the calcination temperature was increased to 500 °C, the diffraction peaks due to anatase phase became narrow and intense in intensity. This indicates that the crystallinity of the anatase phase is further improved.44 When the sample was calcined at 550 °C, weak peaks were observed at 2θ ) 27.6°, 36.1°, 41.2°, and 54.3°, which

Figure 6. Visible Raman spectra of TiO2 calcined at different temperatures with the excitation line at 532 nm.

correspond to the indices of (110), (101), (111), and (211) planes of rutile phase.43 This indicates that the anatase phase starts to transform into the rutile phase at 550 °C. The diffraction peaks of anatase phase gradually diminish in intensity and the diffraction patterns of rutile phase become predominant with the calcination temperatures from 580 to 700 °C. These results clearly show that the phase transformation from anatase to rutile progressively proceeds at the elevated temperatures. The diffraction peaks assigned to anatase phase disappear at 750 °C, indicating that the anatase phase completely changes into the rutile phase. The diffraction peaks of rutile phase became quite strong and sharp after the sample was calcined at 800 °C, owing to the high crystallinity of the rutile phase. Figure 6 displays the visible Raman spectra of TiO2 calcined at different temperatures. For the sample before calcination, two broad bands at about 430 and 605 cm-1 are observed, indicating that the sample is in the amorphous phase.19 For the sample calcined at 200 °C, a Raman band at 143 cm-1 is observed and the high-frequency region shows interference from the fluorescence background, which might come from organic species. After calcination at 300 °C, other characteristic bands of anatase phase appear at 195, 395, 515, and 638 cm-1, but some portions of the sample may exist in the amorphous phase because there is still a broad background in Figure 6. It was found that, when the sample was calcined at 400 °C, the fluorescence disappeared, possibly because the organic

Phase Transformation of TiO2

Figure 7. Weight percentage of the rutile phase in the TiO2 sample calcined at different temperatures estimated by visible Raman spectroscopy and XRD.

residues were removed by the oxidation. The bands of anatase phase increased in intensity and decreased in line width when the sample was calcined at 500 °C. This result suggests that the crystallinity of the anatase phase is greatly improved,18 which is confirmed by XRD (Figure 5). The enlarged section of Figure 6 shows the Raman spectrum of the high-frequency region of the sample calcined at 500 °C. Besides the bands at 395, 515, and 638 cm-1, two very weak bands at 320 and 796 cm-1 are observed. These two bands can be assigned to a two-phonon scattering band and a first overtone of B1g at 396 cm-1, respectively.37 It is noteworthy that a very weak band appears at 445 cm-1 due to rutile phase for the sample calcined at 550 °C. This indicates that the anatase phase starts to change into the rutile phase at 550 °C. This result is in good agreement with that of XRD patterns (Figure 5). The weight percentage of the rutile phase in the samples calcined at different temperatures was estimated by visible Raman spectroscopy and XRD (shown in Figure 7). As seen from Figure 7, the rutile content estimated from visible Raman spectrum and XRD pattern of the sample calcined at 550 °C is 4.2% and 5.7%, respectively. It can be seen that the rutile contents estimated by visible Raman spectroscopy and XRD are also in accordance with each other. When the sample was calcined at 580 °C, other two characteristic bands were observed at 235 and 612 cm-1 due to Raman-active modes of rutile phase. Figure 7 shows the rutile content is 13.6% and 10.9% based on the visible Raman spectrum and XRD pattern of the sample calcined at 580 °C. The intensities of the bands of rutile phase (235, 445, and 612 cm-1) increased steadily while those of the bands of anatase phase (195, 395, 515, and 638 cm-1) decreased when the calcination temperatures were elevated from 600 to 680 °C (Figure 6). These results suggest that the TiO2 sample undergoes the phase transformation from anatase to rutile gradually. The rutile content was estimated for the samples calcined from 600 to 680 °C based on the visible Raman spectra. The results show that the content of the rutile phase is increased from 33.1% to 91.2% respectively for the samples calcined at 600 and 680 °C (Figure 7). The XRD results corresponding to the above two samples indicate that the rutile content is changed from 32.9% to 90.7% (Figure 7). These results clearly show that the rutile content in the sample estimated by visible Raman spectroscopy is agreement with that estimated by XRD. The Raman spectrum of the sample calcined at 700 °C shows mainly the characteristic bands of rutile phase, but the very weak bands of anatase phase are still observed (Figure 6). When the sample was calcined at 750 °C, the bands of anatase phase disappeared and only the bands due to rutile phase (143, 235, 445, and 612 cm-1) were observed. These results indicate that the anatase phase completely transforms into the rutile phase

J. Phys. Chem. B, Vol. 110, No. 2, 2006 931 and are consistent with the results from XRD (Figure 5). When the temperature was increased to 800 °C, the characteristic bands due to rutile phase increased in intensity further. Both the results of XRD and visible Raman spectra (Figures 5 and 6) show that the anatase phase appears at around 200 °C and perfect anatase phase is formed after calcination at temperatures of 400-500 °C. The rutile phase starts to form at 550 °C, and the anatase phase completely transforms into the rutile phase at 750 °C. The signals of visible Raman spectra come mainly from the bulk region of TiO2 because the TiO2 sample is transparent in the visible region (Figure 2).33 XRD is known as a bulk-sensitive method. Therefore, it is essentially in agreement between the results of visible Raman spectra and XRD patterns. 3.3.2. UV Raman Spectra of TiO2 Calcined at Different Temperatures. UV-vis diffuse reflectance spectra (Figure 2) clearly show that TiO2 has strong electronic absorption in the UV region. Thus, the UV Raman spectra excited by a UV laser line contain more signal from the surface skin region than the bulk of the TiO2 sample because the signal from the bulk is attenuated sharply due to the strong absorption.33 Therefore, if a UV laser line in the absorption region of TiO2 is used as the excitation source of Raman spectroscopy, the information from UV Raman spectra is often different from that of visible Raman spectra. The laser line at 325 nm was selected as the excitation source of the UV Raman spectra. The UV Raman spectra and the content of the rutile phase of the TiO2 sample calcined at different temperatures are shown in parts A and B, respectively, of Figure 8. When the sample was calcined at 200 or 300 °C, the Raman band at 143 cm-1 with a shoulder band at 195 cm-1 and three broad bands at 395, 515, and 638 cm-1 were observed, indicating that the anatase phase is formed in the sample. However, the low intensity and the broad band indicate that the amorphous phase still remains in the sample. It can be seen that the fluorescence in the high-frequency region can be avoided when the UV laser line is used as the excitation line. However, the corresponding visible Raman spectra (Figure 6) show interference from the fluorescence. All bands assigned to anatase phase become sharp and strong after calcination at 500 °C (Figure 8A). These results are in agreement with those of XRD and visible Raman spectra (Figures 5 and 6). The UV Raman spectra of the sample with the calcination temperatures from 550 to 680 °C are essentially the same as those of the sample calcined at 500 °C (Figure 8A). However, according to the XRD patterns and visible Raman spectra (Figures 5 and 6), the anatase phase starts to transform into the rutile phase at only 550 °C and the anatase phase gradually changes into the rutile phase in the temperature range of 550-680 °C. After calcination at 700 °C, a new band at 612 cm-1 and two weak bands at 235 and 445 cm-1 due to rutile phase appear while the intensities of the bands of anatase phase begin to decrease (Figure 8A). On the basis of the UV Raman spectrum and XRD pattern of the sample calcined at 700 °C, the rutile content in the sample is 56.1% and 97.0%, respectively (Figure 8B). It is found that the rutile content estimated by UV Raman spectroscopy is far less than that estimated by XRD. When the sample was calcined at 750 °C, the intensities of the bands due to rutile phase increased, but the intensities of the bands due to anatase phase were still strong in the UV Raman spectra (Figure 8A). The UV Raman spectrum of the sample calcined at 750 °C indicates that the rutile content is 84.3% (Figure 8B). However, the results of XRD and visible

932 J. Phys. Chem. B, Vol. 110, No. 2, 2006

Zhang et al.

Figure 9. UV Raman spectra of TiO2 calcined at different temperatures with the excitation line at 244 nm.

Figure 8. (A) UV Raman spectra of TiO2 calcined at different temperatures with the excitation line at 325 nm. (B) Rutile content in the TiO2 sample calcined at different temperatures estimated from the UV Raman spectra with the excitation line at 325 nm. The rutile content in the TiO2 sample calcined at different temperatures estimated by XRD is added for comparison.

Raman spectrum (Figures 5 and 6) suggest that the anatase phase totally transformed into the rutile phase after the sample was calcined at 750 °C. The characteristic bands due to anatase phase disappear and the sample is in the rutile phase after calcination at 800 °C (Figure 8A). Obviously, there are distinct differences between the results from the UV Raman spectra, visible Raman spectra, and XRD patterns. It seems that the anatase phase remains at relatively higher temperatures when detected by UV Raman spectroscopy than by XRD and visible Raman spectroscopy. Another UV laser line at 244 nm was also selected as the excitation source of UV Raman spectroscopy in order to get further insights into the phase transformation of TiO2. The results of the UV Raman spectra of TiO2 calcined at different temperatures with the excitation line at 244 nm are presented in Figure 9. When the sample was calcined at 200 °C, four broad bands were observed at 143, 395, 515, and 638 cm-1, which clearly indicate that the anatase phase exists in the sample. The intensities of the Raman bands due to anatase phase (143, 395, 515, and 638 cm-1) become strong after calcination at 500 °C. The UV Raman spectra hardly change for the sample calcined at different temperatures even up to 680 °C. The characteristic bands (445 and 612 cm-1) of rutile phase appear only when the calcination temperature exceeds 700 °C. This result is in good agreement with that from the UV Raman spectrum of the sample calcined at 700 °C with 325 nm excitation (Figure 8A). The intensities of the bands due to anatase phase (395, 515, and 638 cm-1) decrease while those of bands assigned to rutile phase (445 and 612 cm-1) increase after calcination at 750 °C

(Figure 9). When the sample was calcined at 800 °C, the Raman bands due to anatase phase disappeared, while the bands of rutile phase developed. This result indicates that the sample calcined at 800 °C is in the rutile phase. It is interesting to note that the results of the UV Raman spectra with the excitation lines at 325 and 244 nm are in agreement with each other but are different from those of XRD patterns and visible Raman spectra. 3.3.3. TEM of the TiO2 Sample Calcined at Different Temperatures. TEM was used to characterize the microstructure of the TiO2 sample calcined at 500, 600, and 800 °C (shown in Figure 10). Most particles in the sample calcined at 500 °C exhibit diameters in a range between 10 and 30 nm (Figure 10a). On the other hand, remarkable agglomeration is observed for the TiO2 sample calcined at 500 °C. The particle size increases after calcination at 600 °C (Figure 10b). According to the results of XRD and visible Raman spectra (Figures 5 and 6), the sample undergoes the phase transformation from anatase to rutile gradually in the temperature range of 550680 °C. These results imply that the phase transformation and growth of the particle size are interrelated. Many researchers45-49 reported similar phenomena. Kumar et al.45 attributed this particle size growth to the higher atomic mobility because of bond breakage during the phase transformation. When the calcination temperature was increased to 800 °C, TiO2 particles further grew and the particle size could be as large as about 200 nm (Figure 10c). 3.4. Visible Raman Spectra and UV Raman Spectra of La2O3/TiO2 with Increasing La2O3 Loading. The literature15,17,29 proposed the mechanism that the phase transformation of TiO2 might start at the interfaces of contacting anatase particles. If direct contact between anatase particles of TiO2 is avoided, the phase transformation of TiO2 from anatase to rutile could be retarded or prohibited. This assumption may be verified by covering the surface of anatase TiO2 with an additive. In this work, we used La2O3 dispersed on anatase TiO2 to prevent the anatase particles from contacting directly because it was reported that La2O3 could be highly dispersed on anatase TiO2.34-36 Figure 11A displays the visible Raman spectra of La2O3/TiO2 with increasing La2O3 loading. TiO2 support is in the anatase phase because only characteristic bands (143, 195, 395, 515, and 638 cm-1) due to anatase phase are observed. When the TiO2 support was calcined at 900 °C (TiO2-900), the Raman spectrum gave the characteristic bands of rutile phase, indicating that the TiO2-900 sample was in the rutile phase. When the

Phase Transformation of TiO2

J. Phys. Chem. B, Vol. 110, No. 2, 2006 933

Figure 10. TEM images of TiO2 calcined at (a) 500, (b) 600, and (c) 800 °C.

Figure 11. (A) Visible Raman spectra of La2O3/TiO2 with increasing La2O3 loading. (B) UV Raman spectra of La2O3/TiO2 with increasing La2O3 loading.

sample with 0.5 wt % La2O3 loading was calcined at 900 °C, the Raman spectrum drastically changed, as compared to the TiO2-900 sample. Only characteristic bands of anatase phase were observed, suggesting that the TiO2 sample retains its anatase phase when La2O3 loading is 0.5 wt % while the TiO2900 sample is in the rutile phase. The spectra of the samples with La2O3 loadings from 1 to 6 wt % are almost the same as those for the sample with 0.5 wt % La2O3 loading, except for a small decrease in intensity. Figure 11B shows the UV Raman spectra of La2O3/TiO2 with increasing La2O3 loading. The Raman spectrum of the TiO2 support and the TiO2-900 sample gives the characteristic bands of anatase phase and rutile phase, respectively. When La2O3 loading is varied from 0.5 to 6 wt %, the results of the UV Raman spectra show that all the samples are in the anatase phase. These results are in accordance with those of visible Raman spectra (Figure 11A). The results from XRD patterns (the XRD patterns for the above samples are not shown) agree well with those of visible Raman spectra and UV Raman spectra. For both visible Raman spectra and UV Raman spectra, no bands due to crystalline phase of La2O3 are observed for all the La2O3/TiO2 samples, showing that La2O3 is highly dispersed on the surface of the anatase phase of TiO2 particles.35 4. Discussion It has been clearly shown that the results of visible Raman spectra (Figure 6) are in good agreement with the results from XRD patterns (Figure 5) but are different from those of UV Raman spectra (Figures 8A and 9) for TiO2 calcined at elevated temperatures. The inconsistency in the results from the three techniques can be explained by the fact that UV Raman

spectroscopy provides the information mainly from the surface region while visible Raman spectroscopy and XRD provide information from the bulk of TiO2. The discrepant results of UV Raman spectra, visible Raman spectra, and XRD patterns are attributed to their different detectable depths for TiO2 particles.33 The disagreements of UV Raman spectra, visible Raman spectra, and XRD patterns suggest that the crystal phase in the surface region is different from that in the bulk during the phase transformation of TiO2. Busca et al.20 characterized the Degussa P25 using Fourier transform Raman spectroscopy and XRD technology. They estimated the rutile-to-anatase ratio by Raman spectroscopy and XRD, and found that the ratio estimated by Raman spectroscopy for the Degussa P25 was smaller than that evaluated by XRD. Therefore, they assumed that the rutile phase in the Degussa P25 was more concentrated in the bulk because Raman spectroscopy excited by a near-IR laser line should be more surface-sensitive than XRD. Different from their investigations, our experimental results give direct UV Raman evidence to show that the phases in the surface region are generally different from that in the bulk region of TiO2, particularly when TiO2 is in the transition stage of the phase transformation. As presented above, the anatase phase can remain at relatively higher temperatures as observed by UV Raman spectroscopy than by visible Raman spectroscopy and XRD. These facts lead us to the conclusion that the phase transformation of TiO2 takes place from its bulk region and then extends to its surface region. To further understand the process of the phase transformation of TiO2, we used the TEM technique to observe the microstructure of the sample calcined at different temperatures (Figure 10). TEM results show that the samples are composed of aggregated particles after the calcination. According to TEM measurements, Penn et al.17 suggested that structural elements with rutilelike character can be produced at a subset of anatase interfaces, and these might serve as rutile nucleation sites. Lee et al.29 investigated the growth and transformation in nanometersize TiO2 powders by in situ TEM. The nucleation of rutile was found to occur at the amorphous interface of anatase particles where there are strain and disorder. On the basis of the experimental observations and combined with the results from the literature, we suggest that the phase transformation of TiO2 starts from the interfaces among the anatase particles of the agglomerated TiO2 particles. On the basis of kinetic data from XRD, Zhang et al.15 also proposed that interface nucleation dominated the transformation for nanocrystalline anatase samples with denser particle packing below 620 °C, or in the temperature range of 620-680 °C. If the direct contact between anatase particles is avoided, the phase transformation could be retarded or prohibited because the rutile phase nucleates at the interfaces of contacting anatase

934 J. Phys. Chem. B, Vol. 110, No. 2, 2006

Figure 12. Proposed scheme for the phase transformation of TiO2 with increasing calcination temperature.

particles. We prepared the anatase particles covered with highly dispersed La2O3, and the results of visible Raman spectra and UV Raman spectra of the La2O3/TiO2 samples (Figure 11) indicate that the samples are stabilized at their anatase phase both in bulk and in the surface region even after the calcination at 900 °C. It is interesting to note that the impregnation of only 0.5 wt % La2O3 can inhibit the phase transformation. The explanation of this interesting result could be as follows: the defect sites on the surface of the anatase particles are assumed to play an important role in the phase transformation of TiO2. When the defect sites of the anatase particle react with a neighboring anatase particle with or without defect sites, the rutile phase formation may start at these sites. La2O3 easily reacts with the defect sites of the anatase particles, and only 0.5 wt % La2O3 can occupy or deactivate all the defect sites of the anatase particles because usually the surface defect sites concentration is relatively low. The easy migration of surface atoms of anatase and the nucleation of rutile phase most possibly take place at the surface defect sites. Therefore, only 0.5 wt % La2O3 can effectively inhibit the phase transformation of anatase. In addition to occupying the defect sites, the highly dispersed La2O3 on the surface of the anatase particles effectively prevents direct contact of the anatase particles for the sample with high La2O3 loadings. Therefore, the La2O3/TiO2 sample can retain its anatase phase even when the calcination temperature is up to 900 °C, owing to the above two roles played by La2O3. A proposed scheme for the phase transformation of TiO2 with increasing calcination temperature is illustrated in Figure 12. The TiO2 particles with anatase phase intimately contact each other. Thus, the surface and the bulk region of the TiO2 sample actually refer to respectively the outer surface region and the inner region of agglomerated TiO2 particles. The interfaces of contacting anatase particles, which are only present in the inner region of agglomerated particles, provide the nucleation sites of the rutile phase. Therefore, the rutile phase is first detected by XRD and visible Raman spectroscopy for the sample calcined at 550-680 °C. Once phase transformation takes place, the particle size increases rapidly. The agglomeration of the TiO2 particles is along with the phase transformation from anatase

Zhang et al. to rutile. The rutile phase needs a fairly high temperature to progressively develop into the whole conglomeration composed of the coalescence of some neighboring particles because the phase transformation is a diffusing process. Thus, the outer surface region of the agglomerated particles without directly interacting with other particles maintains in the anatase phase when the calcination temperature is below 680 °C. Accordingly, UV Raman spectroscopy detects only the anatase phase in the outer region of agglomerated particles in the temperature range of 550-680 °C. When the calcination temperature is higher than 700 °C, the anatase phase in the outer surface region of agglomerated particles begins to change into the rutile phase. Therefore, the mixed phases of anatase and rutile are observed by UV Raman spectroscopy but the inner region of agglomerated particles is nearly in the rutile phase. Both XRD and visible Raman spectra show that the inner region of agglomerated particles is in the rutile phase when the sample is calcined at 750 °C. However, the phase transformation is not yet complete because the outer surface region is still in the mixed phases of anatase and rutile. After calcination at 800 °C, the anatase phase in the outer surface region completely transforms to the rutile phase; the whole agglomerated particles are in the rutile phase. Following the above reasoning, for the agglomerated particles of TiO2, the rutile phase nucleates at the interfaces of contacting anatase particles. However, from the point view of a single anatase crystal particle, the rutile phase starts to form still at the surface of TiO2, where it contacts other particles. It is reasonably assumed that the phase transformation of single particles might start from the surface, where there is no direct interaction with other particles, but it needs a high temperature. Zhang et al.15 indicated that thermal fluctuation of Ti and O atoms in anatase is not strong enough to generate rutile nuclei on the surfaces or in the bulk of the anatase particles at lower temperatures. When the small anatase particles agglomerate into large particles, interface nucleation is easier, as compared to nucleation at the surface, where there is no contact with other particles. Zhang et al.15 indicated that the activation energy for surface nucleation is expected to be higher than that for interface nucleation. Therefore, we observed the phenomenon that the crystalline phase in the outer surface region of agglomerated TiO2 particles is different from that in the inner region of agglomerated TiO2 particles. 5. Conclusions UV Raman spectroscopy is found to be more surface-sensitive than visible Raman spectroscopy and XRD for TiO2 because of strong absorption of TiO2 in the UV region. According to the visible Raman spectra and XRD patterns, the phase transformation from anatase to rutile takes place at 550 °C and the anatase phase completely transfers to the rutile phase when the sample is calcined at temperatures up to 750 °C. On the basis of the UV Raman spectra, the rutile phase is observed only when the calcination temperature exceeds 700 °C, and the anatase phase can still be detected at 750 °C. The disagreements of UV Raman spectra, visible Raman spectra, and XRD patterns suggest that the crystalline phase in the surface region is usually different from that in the bulk during the phase transformation of TiO2. Furthermore, the anatase phase in the surface region can remain at relatively higher temperatures than it can in the bulk region of TiO2. The results of TEM show that the agglomeration of the TiO2 particles and growth of particle size are along with the phase transformation. The anatase phase of

Phase Transformation of TiO2 TiO2 sample covered with highly dispersed La2O3 can retain its anatase phase both in bulk and in surface region even after calcination at 900 °C because the direct contact of anatase particles of TiO2 is avoided and the surface defect sites of anatase particles are occupied by La2O3. It is proposed that the phase transformation of TiO2 starts from the interfaces between the anatase particles in the agglomerated TiO2 particles. Acknowledgment. This work was financially supported by the National Natural Science Foundation of China (NSFC, Grants 20273069, 90210036), the National Basic Research Program of China (Grant 2003CB615806), and the National Key Basic Research and Development Program (Grant 2003CB214500). References and Notes (1) Coulter, L. E.; Sault, A. G. J. Catal. 1995, 154, 56. (2) Fujishima, A.; Honda, K. Nature 1972, 238, 37. (3) Yeh, Y. C.; Tseng, T. T.; Chang, D. A. J. Am. Ceram. Soc. 1989, 72, 1472. (4) Ketrom, L. Am. Ceram. Soc. Bull. 1989, 68, 860. (5) Gratzel, M. Nature 1992, 353, 737. (6) Liu, Y.; Claus, R. O. J. Am. Chem. Soc. 1997, 119, 5273. (7) Foger, K.; Anderson, J. R. Appl. Catal. 1986, 23, 139. (8) Zhang, F.; Zheng, Z.; Ding, X.; Mao, Y.; Chen, Z.; Yang, S.; Liu, X. J. Vac. Sci. Technol., A 1997, 15, 1824. (9) Karakitsou, K. E.; Verykios, X. E. J. Phys. Chem. 1993, 97, 1184. (10) Zhu, J.; Zheng, W.; He, B.; Zhang, J.; Anpo, M. J. Mol. Catal., A 2004, 216, 35. (11) Ding, Z.; Liu, G. Q.; Greenfield, P. F. J. Phys. Chem. B 2000, 104, 4815. (12) Muscat, J.; Swamy, V.; Harrison, N. M. Phys. ReV. B 2002, 65, 224112. (13) Ovenstone, J.; Yanagisawa, K. Chem. Mater. 1999, 11, 2770. (14) Shannon, R. D.; Pask, J. A. J. Am. Ceram. Soc. 1965, 48, 391. (15) Zhang, H. Z.; Banfield, J. F. J. Mater. Res. 2000, 15, 437. (16) Gouma, P. I.; Mills, M. J. J. Am. Ceram. Soc. 2001, 84, 619. (17) Penn, R. L.; Banfield, J. F. Am. Mineral. 1999, 84, 871. (18) Ma, W.; Lu, Z.; Zhang, M. Appl. Phys. A 1998, 66, 621. (19) Zhang, Y. H.; Chan, C. K.; Porter, J. F.; Guo, W. J. Mater. Res. 1998, 13, 2602. (20) Busca, G.; Ramis, G.; Amores, J. M. G.; Escribano, V. S.; Piaggio, P. J. Chem. Soc., Faraday Trans. 1994, 90, 3181.

J. Phys. Chem. B, Vol. 110, No. 2, 2006 935 (21) Hwu, Y.; Yao, Y. D.; Cheng, N. F.; Tung, C. Y.; Lin, H. M. Nanostruct. Mater. 1997, 9, 355. (22) Okada, K.; Yamamoto, N.; Kameshima, Y.; Yasumori, A. J. Am. Ceram. Soc. 2001, 84, 1591. (23) Yoshinaka, M.; Hirota, K.; Yamaguchi, O. J. Am. Ceram. Soc. 1997, 80, 2749. (24) Yang, J.; Mei, S.; Ferreira, J. M. F. J. Am. Ceram. Soc. 2000, 83, 1361. (25) Ozaki, S.; Iida, Y. J. Am. Ceram. Soc. 1961, 44, 120. (26) Navrotsky, A.; Kleppa, O. J. J. Am. Ceram. Soc. 1967, 50, 626. (27) Mitsuhashi, T.; Kleppa, O. J. J. Am. Ceram. Soc. 1979, 62, 356. (28) Ahonen, P. P.; Kauppinen, E. I.; Joubert, J. C.; Deschanvres, J. L.; Van Tendeloo, G. J. Mater. Res. 1999, 14, 3938. (29) Lee, G. H.; Zuo, J.-M. J. Am. Ceram. Soc. 2004, 87, 473. (30) Ranade, M. R.; Navrotsky, A.; Zhang, H. Z.; Banfield, J. F.; Elder, S. H.; Zaban, A.; Borse, P. H.; Kulkarni, S. K.; Doran, G. S.; Whitfield, H. J. Proc. Natl. Acad. Sci. U.S.A. 2002, 99 (Suppl. 2), 6476. (31) Zhang, H. Z.; Banfield, J. F. J. Phys. Chem. B 2000, 104, 3481. (32) Linsebigler, A. L.; Lu, G. Q.; Yates, J. T., Jr. Chem. ReV. 1995, 95, 735. (33) (a) Li, M. J.; Feng, Z. C.; Xiong, G.; Ying, P. L.; Xin, Q.; Li, C. J. Phys. Chem. B 2001, 105, 8107. (b) Li, C.; Li, M. J. J. Raman Spectrosc. 2002, 33, 301. (34) Jing, L. Q.; Sun, X. J.; Xin, B. F.; Wang, B. Q.; Cai, W. M.; Fu, H. G. J. Solid State Chem. 2004, 177, 3375. (35) Xie, Y. C.; Tang, Y. Q. AdV. Catal. 1990, 37, 1. (36) Gopalan, R.; Lin, Y. S. Ind. Eng. Chem. Res. 1995, 34, 1189. (37) Ohsaka, T.; Izumi, F.; Fujiki, Y. J. Raman Spectrosc. 1978, 7, 321. (38) Chaves, A.; Katiyan, K. S.; Porto, S. P. S. Phys. ReV. 1974, 10, 3522. (39) Bickley, R. I.; Gonzalez-Carreno, T.; Lees, J. S.; Palmisano, L.; Tilley, R. D. J. Solid State Chem. 1991, 92, 178. (40) West, A. R. Solid State Chemistry and Its Applications; John Wiley & Sons: New York, 1984; p 174. (41) Gribb, A. A.; Banfield, J. F. Am. Mineral. 1997, 82, 717. (42) Yan, M. C.; Chen, F.; Zhang, J. L.; Anpo, M. J. Phys. Chem. B 2005, 109, 8673. (43) Sreethawong, T.; Suzuki, Y.; Yoshikawa, S. J. Solid State Chem. 2005, 178, 329. (44) Tsai, S.-J.; Cheng, S. Catal. Today 1997, 33, 227. (45) Kumar, K. P. Scr. Metall. Mater. 1995, 32, 873. (46) Hague, D. C.; Mayo, M. J. Nanostruct. Mater. 1993, 3, 61. (47) Banfield, J. F.; Bischoff, B. L.; Anderson, M. A. Chem. Geol. 1993, 110, 211. (48) Mackenzie, K. J. D. Trans. J. Br. Ceram. 1975, 74, 29. (49) Kumar, K.-N. P.; Keizer, K. A.; Burggraaf, J.; Okubo, T.; Nagamoto, H.; Morooka, S. Nature 1992, 58, 48.