In situ X-ray absorption spectroscopic studies of TiO₂ photocatalytic active sites for degradation of trace CHCl₃ in drinking water

Toxic disinfection byproducts such as trihalo­methanes (e.g. CHCl₃) are often found after chlorination of drinking water. It has been found that photocatalytic degradation of trace CHCl₃ in drinking water generally lacks an expected relationship with the crystalline phase, band-gap energy or the particle sizes of the TiO₂-based photocatalysts used such as nano TiO₂ on SBA-15 (Santa Barbara amorphous-15), TiO₂ clusters (TiO₂–SiO₂) and atomic dispersed Ti [Ti-MCM-41 (Mobil Composition of Matter)]. To engineer capable TiO₂ photocatalysts, a better understanding of their photoactive sites is of great importance and interest. Using in situ X-ray absorption near-edge structure (XANES) spectroscopy, the A₁ (4969 eV), A₂ (4971 eV) and A₃ (4972 eV) sites in TiO₂ can be distinguished as four-, five- and six- coordinated Ti species, respectively. Notably, the A₂ Ti sites that are the main photocatalytic species of TiO₂ are shown to be accountable for about 95% of the photocatalytic degradation of trace CHCl₃ in drinking water (7.2 p.p.m. CHCl₃ gTiO₂⁻¹ h⁻¹). This work reveals that the A₂ Ti species of a TiO₂-based photocatalyst are mainly responsible for the photocatalytic reactivity, especially in photocatalytic degradation of CHCl₃ in drinking water.

1. Introduction

In order to decrease the risk of water-borne diseases, chlorine is used in the disinfection of drinking water. However, toxic disinfection byproducts (DBPs) such as trihalo­methanes (THMs) are frequently formed in the drinking water chlorination processes. The US National Cancer Institute has suggested that chloro­form (one of the THMs) is carcinogenic in rodents (Boorman, 1999). Lin & Hoang (2000) found that the level of THM carcinogen exposure could be as high as 47 µg day⁻¹ by drinking or 30 µg day⁻¹ by inhalation. Conventional methods for reduction of DBPs and precursors include chemical coagulation, activated carbon adsorption, membrane separation, ozonation and combined ozonation/bio-treatments (Pan et al., 2016; Sillanpää et al., 2018; Cai et al., 2018; Abou-Gamra & Ahmed, 2015). Membrane separation and ozonation also are highly efficienct in the reduction of biologically unstable materials. However, those methods may not be economically attractive in engineering applications (Sillanpää et al., 2018; Cai et al., 2018).

TiO₂, which possesses unique characteristics such as high photostability, effective band gap and easy availability, has been widely applied in the photocatalytic degradation of toxic organics, splitting of H₂O, reduction of NO to N₂, and dye-sensitized solar cells (Guo et al., 2018; Kang & Wang, 2013; Yang et al., 2014; Xie et al., 2017; Liao et al., 2014). The photocatalytic activities of TiO₂ can be enhanced with dispersed transition metal promoters such as V, Cr, Fe, Cu, Ni, Mn, Ag or Au (Yang et al., 2017; Wang et al., 2017; Lin et al., 2017; Muñoz-Batista et al., 2014). Nano TiO₂ has been considered a desirable photocatalyst due to its large surface-to-volume ratios (Xu et al., 2019; Chen et al., 2019).

Valence and local structure (<1 nm) of select elements in a complicated solid matrix can be studied by synchrotron X-ray absorption spectroscopy. Molecular structure information such as the coordination number (CN), bond distance and oxidation state of elements can be determined by X-ray absorption [extended X-ray absorption fine structure (EXAFS) and X-ray absorption near-edge structure (XANES)] spectroscopy. Using XANES and EXAFS, it was found that copper oxide clusters in the channels of mesopores (2–6.5 nm) [MCM-41 (Mobil Composition of Matter-41)], micropores (0.54–0.56 nm) [e.g. ZSM-5 (Zeolite Socony Mobil-5) and ZSM-48] and TiO₂-based photocatalysts were very effective in catalysis (Hsiung et al., 2006a, 2021; Chien et al., 2001; Huang et al., 2003). These molecular-scale data turn out to be very useful in revealing the nature of catalytic active species and the reaction paths involved.

Photocatalytic degradation of trace THMs or DBPs by nano TiO₂ was shown to be not very effective (Diaz-Angulo et al., 2019; El-Mragui et al., 2019). On the other hand, photocatalytic generation of H₂ (from H₂O) on atomic dispersed Ti and Zr in the mesoporous molecular sieve (MCM-41) (Ti-MCM-41 and Zr-MCM-41) can be enhanced by 17 and 80 times, respectively (if compared with the bulk states) (Liu et al., 2003). It seems that nano, cluster (<4 nm in diameter) or even atomic dispersed TiO₂ may lead to unique enhancements in photocatalytic reactions. However, speciation of photocatalytic active sites in TiO₂ has not been well studied particularly during reactions (Hsiung et al., 2006b, 2008). In order to prepare effective TiO₂ photocatalysts, a molecular-scale understanding of their photoactive sites is essential. Thus, the main objective of the present work was to study the chemical structure of Ti species with different TiO₂ sizes (i.e. nano, cluster and atomic dispersion) in relation to the photocatalytic degradation of trace CHCl₃ in drinking water by in situ XANES and EXAFS.

2. Experimental

The nano TiO₂–SiO₂ (Ti/Si atomic ratio = 3/7) photocatalyst was prepared by the sol-gel method (Beck et al., 2001). Typically, 12.8 g of titanium (IV) n-butoxide (Acros, 99%) and 24.3 g of tetra­ethyl­orthosilicate (C₈H₂₀OSi, TEOS) (Merck) were well mixed (at the pH value of 7.0) in an ethanol and distilled water (1:2) solution (45 ml). The TiO₂–SiO₂ solution was heated slowly until the sol-gel became xerogel, and then separated and calcined at 673 K for 1 h. The nano TiO₂ photocatalysts were synthesized using a similar procedure.

The TiO₂ supported on SBA-15 (Santa Barbara amorphous-15), that has a 2D hexagonal mesoporous structure with long straight channels parallel to the silica skeletons, photocatalysts were prepared by co-condensation of silica, titanium oxide and neutral-charge poly(ethyl­ene oxide)-poly(propyl­ene oxide)-poly(ethyl­ene oxide) [EO₂₀PO₇₀EO₂₀ (P123)] (Aldrich, 99%) nonionic surfactant. Briefly, 0.21 g of titanium n-butoxide [Ti(OBu)₄, Aldrich, 97%] and 15 g of sulfuric acid (1.2 M, Merck) were well mixed until the solution became clear; a solution containing 2.75 g of sodium silica [14% NaOH and 27% SiO₂ (Aldrich)], 1.25 g of P123 (Aldrich, 99%) and 200 g of distilled water was added. The pH value of the solution was adjusted to 5.0. After separation, the photocatalyst was calcined at 773 K for 6 h.

The Ti-inserted (framework) MCM-41 (Ti-MCM-41) photocatalysts were synthesized by the liquid crystal templating method. Fumed silica (Sigma) and sodium silicate solution were used as silica sources for the preparation of MCM-41. Hexa­decyl­tri­methyl ammonium bromide (CTAB) (Sigma, purity >99%) and tetra­methyl ammonium hydroxide [TMAOH, 25% (Sigma–Aldrich)] were used as a template and mineralizer, respectively. The mole ratio of H₂O, CTAB, TMAOH and Si in the mixture was 86:0.27:0.58:1. Titanium n-butoxide (Acros, 99%) was added to the MCM-41 mother liquid for the synthesis of Ti-MCM-41. A diluted sulfuric acid solution was used to adjust the pH values of the solution to between 11 and 12. The mixture was then heated in a Teflon-lined autoclave at 375 K for 24 h. The as-synthesized MCM-41 was filtered, washed with distilled water, dried and calcined at 773 K for 5 h.

The chemical structure of the photocatalysts was characterized by X-ray powder diffraction (XRD) spectroscopy (D8 Advance, Bruker) with Cu Kα (1.542 Å) radiation. Samples were scanned from 10 to 50° (2θ) at a scan rate of 5° min⁻¹. Diffuse reflectance ultraviolet/visible spectra (DR UV–vis) of the photocatalysts were also determined (between 200 and 700 nm) on a UV–vis spectrophotometer (Hitachi U-3010) at a scan speed of 120 nm min⁻¹.

The in situ Ti K-edge (4966 eV) XANES spectra of the photocatalysts were collected at 298 K on the wiggler beamline of the Taiwan National Synchrotron Radiation Research Center. The absorption spectra were collected in ion chambers that were filled with helium and nitro­gen mixed gases. The beam energy was calibrated to the adsorption edge of a Ti foil at the energy of 4966 eV. The electron storage ring was operated at an energy of 1.5 GeV (ring current = 120–200 A). The isolated EXAFS data were normalized to the edge jump and converted to the wavenumber scale. Fitting of data to model compounds was performed using FEFFIT (UWXAFS 3.0) in combination with the multiple scattering code FEFF 8.0 simulation programs (Chiu et al., 2011). FEFFIT was used to determine the best-fitting results with the minimum Debye–Waller factors (<0.01 Ų). The Fourier transform was performed on k³-weighted EXAFS oscillations in the range 3.8–11.7 Å⁻¹.

Photocatalytic degradation of CHCl₃ (500 p.p.b.) was performed on a homemade in situ EXAFS cell (Kang & Wang, 2013). Typically, 0.15 g of the TiO₂ photocatalyst was suspended in H₂O containing trace CHCl₃ (500 p.p.b.). A 300 W high-pressure Xe arc lamp (Newport, model 6258) was used as the light source (250–800 nm) while the synchrotron X-ray absorption spectra of the TiO₂ photocatalysts were determined. Concentrations of CHCl₃ were determined using a gas chromatograph (model 3400, Varian) equipped with an electrolytic conductivity detector (model 100, Tracor) and a purge/trap system (model LCS-2000, Tekmar).

3. Results and discussion

The XRD patterns of the TiO₂-based photocatalysts are shown in Fig. 1. The Ti-MCM-41 and TiO₂–SiO₂ photocatalysts have broadened peaks between 17 and 31° (2θ), suggesting the absence of crystalline TiO₂, which may be associated with the existence of very small TiO₂ nanoparticles or highly atomic dispersed Ti therein. For the TiO₂/SBA-15 and nano TiO₂, anatase phases [(101), (004) and (200)] are observed with crystalline sizes of about 4.6 and 8.5 nm, respectively (calculated using the Scherrer equation).

Figure 1 XRD patterns of (a) Ti-MCM-41, (b) TiO2–SiO2, (c) TiO2/SBA-15 and (d) nano TiO2.

Fig. 2 shows the DR UV–vis spectra of the photocatalysts. The band-gap energies for the TiO₂–SiO₂ and Ti-MCM-41 are 3.38 and 3.42 eV, respectively. Beck et al. (2001) also found blue shifts for the band-gap absorption edge with highly dispersed Ti in the SiO₂ framework. The absorption feature (<260 nm) for the Ti-MCM-41 can be attributed to the ligand-to-metal charge transfer between Ti^IV and the amorphous SiO₂ framework. It seems that Ti is well dispersed in the Ti-MCM-41 photocatalyst.

Figure 2 DR UV–vis spectra of (a) Ti-MCM-41, (b) TiO2–SiO2, (c) nano TiO2 and (d) TiO2/SBA-15.

The EXAFS spectra best-fitted by the Fourier transformation and k³-weighted for the TiO₂-based photocatalysts in the ranges 0–4 and 4–11 Å, respectively, are shown in Fig. 3. The speciation parameters for the Ti in those photocatalysts are also listed in Table 1. The bond distances for the Ti—(O)—Si species (in the second shells) in the Ti-MCM-41 and TiO₂–SiO₂ photocatalysts are 3.13 and 3.31 Å, respectively. It is worth noting that the Ti-MCM-41 has a greater CN for the Ti—(O)—Si bonding than does the TiO₂–SiO₂, demonstrating that more Ti atoms are dispersed in the amorphous SiO₂ framework for the Ti-MCM-41, and that the TiO₂–SiO₂ one has more TiO₂ aggregates on the SiO₂ surfaces with a metal–support interaction related to its relatively low CN. It seems that the TiO₂–SiO₂ contains sub-nano TiO₂ (supported on the SiO₂) that is not observed by XRD, which may behave like a TiO₂ cluster. The Ti—O—Si species are, in contrast, not found in the TiO₂/SBA-15. It seems that the TiO₂ nanoparticles may be formed in the hexagonally ordered mesoporous channels (an opening of 5.7 nm) of the SBA-15. Note that the averaged diameter of the TiO₂ nanoparticles in the SBA-15 is approximately 4.6 nm (see Table 2).

Table 1
Speciation parameters of titanium in the photocatalysts for degradation of trace CHCl₃ in drinking water

	Shell	Bond distance (Å)	Coordination number	Debye–Waller factor (Å^2)
Ti-MCM-41	Ti—Oeq	1.75	0.9	0.0058
	Ti—Oax	1.99	2.6	0.0029
	Ti—(O)—Si	3.13	2.9	0.0103
TiO2–SiO2	Ti—Oeq	1.77	1.9	0.0023
	Ti—Oax	2.32	2.5	0.0030
	Ti—(O)—Si	3.31	1.1	0.0027
Nano TiO2	Ti—Oeq	1.92	1.5	0.0004
	Ti—Oax	2.03	3.1	0.0045
	Ti—(O)—Ti	3.03	3.5	0.0049
TiO2/SBA-15	Ti—Oeq	1.90	2.0	0.0038
	Ti—Oax	1.99	3.3	0.0044
	Ti—(O)—Ti	3.03	2.1	0.0043

Table 2
Characterization and reactivity of the photocatalysts for degradation of trace CHCl₃ in drinking water

TiO₂ chemical structure was determined by XRD. Crystalline size was calculated using the Scherrer equation. Band-gap absorption edge was determined by DR UV–vis. Photocatalytic degradation of CHCl₃ was determined in a total UV–vis reflectance system under UV–vis irradiation for 4 h [initial CHCl₃ concentration = 500 p.p.b. (in H₂O)]. Ti species were determined by XANES [A₁: TiO₄; A₂: (Ti=O)O₄; A₃: TiO₆]. Fractions of the Ti species (i.e. A₁, A₂ and A₃) related to the photocatalytic degradation of CHCl₃ determined by the linear regression analysis are well fitted for A₂ (R² > 0.9) while the R² values for A₁ and A₃ are 0.37 and 0.64, respectively.

Photocatalyst	TiO2 chemical structure	Crystalline size (nm)	Band-gap absorption edge (eV)	Photocatalytic degradation of CHCl3 (p.p.m. gTiO2^−1 h^−1)	Ti species (%)
					A1	A2	A3
Ti-MCM-41	Amorphous	–	3.42	7.2	7	49	44
TiO2–SiO2	Amorphous	–	3.38	0.9	25	33	42
Nano TiO2	Anatase	8.5	3.25	0.7	22	32	46
TiO2/SBA-15	Anatase	4.6	3.30	5.6	19	40	41

Figure 3 Ti K-edge XANES spectra of (a) TiO2–SiO2, (b) nano TiO2, (c) TiO2/SBA-15 and (d) Ti-MCM-41.

Photocatalytic degradation of trace CHCl₃ (500 p.p.b. in drinking water) by the nano, cluster and atomic dispersed TiO₂ photocatalysts under UV–vis radiation (250–800 nm) is shown in Fig. 4. An enhanced photocatalytic degradation of trace CHCl₃ by the Ti-MCM-41 is observed. After a 4 h UV–vis radiation, accumulated 28.8 p.p.m. CHCl₃ gTiO₂⁻¹ are photocatalytically degraded by Ti-MCM-41. The nano TiO₂ shows relatively low activity for photocatalytic degradation of CHCl₃ when compared with the Ti-MCM-41. Note that the TiO₂–SiO₂ containing TiO₂ clusters also shows much less activity for the photocatalytic degradation than the TiO₂/SBA-15 that has larger TiO₂ nanoparticles (an average size of 4.6 nm). It appears that the reactivity with respect to photocatalytic degradation of trace CHCl₃ in drinking water may not be related to the crystalline phase, band-gap absorption edge or particle size of the TiO₂-based photocatalysts.

Figure 4 Time dependence for photocatalytic degradation of trace CHCl3 in H2O effected by (a) Ti-MCM-41, (b) TiO2/SBA-15, (c) TiO2–SiO2 and (d) nano TiO2.

To reveal the key factor that influences the photocatalytic degradation reactivity of the TiO₂-based photocatalysts, Table 2 summarizes the characterization of the TiO₂-based photocatalysts, the related photocatalytic degradation reactivity and the photoactive sites involved. It seems that the photocatalysts, i.e. Ti-MCM-41 and TiO₂/SBA-15, possessing meso-pore channel systems have greater photocatalytic degradation conversions than the other ones that have nano and cluster TiO₂ on SiO₂. Apparently, the mesopores in the MCM-41 and SBA-15 have a large enough pore opening for free diffusion of small CHCl₃ molecules to the TiO₂ photoactive sites. Thus, there may exist other factors such as different loadings of TiO₂ and hardness or toxic ions in real drinking water that may also disturb the photocatalysis.

In Fig. 5, A₁ (4969 eV), A₂ (4971 eV) and A₃ (4972 eV) sites can be resolved in the pre-edge XANES spectra [between the absorption threshold (4966 eV) and jump (4984 eV)] of the photocatalysts. The A₁, A₂ and A₃ species in TiO₂ may be attributed to four- (TiO₄), five- [(Ti=O)O₄] and six- (TiO₆) coordinated Ti species, respectively (Hsiung et al., 2006a, 2008). The B and C features (at 4974 and 4980 eV, respectively) are due to the interactions of the central Ti 4p orbitals hybridized with the near Ti or O atoms. The Ti K post-edge at 4980–5020 eV can be assigned to 3s-to-np dipole-allowed transitions. The insertion of Ti atoms into the SiO₂ matrix may cause distortion of the Ti–O octahedron structure. It is worth noting that the broadened white line absorption (4987 eV) for Ti-MCM-41 and TiO₂–SiO₂ is observed, suggesting the existence of highly dispersed Ti atoms in the SiO₂ matrix.

Figure 5 Ti K-edge XANES spectra of (a) TiO2–SiO2, (b) nano TiO2, (c) TiO2/SBA-15 and (d) Ti-MCM-41.

The Ti species (A₁, A₂ and A₃) in the photocatalysts were distinguished (in the XANES spectra). As shown in Table 2, the well dispersed Ti in Ti-MCM-41 has a much better performance for photocatalytic degradation of trace CHCl₃ in drinking water. The A₂ Ti species (fivefold-coordinated Ti^IV) possesses a square-pyramid structure, consisting of a double bond (Ti=O) and four Ti—O bonds, which is predominant generally on the surfaces of the very small (<10 nm) TiO₂ nanoparticles. Fractions of the Ti active species related to the photocatalytic degradation of CHCl₃ determined by the linear regression analysis are well fitted (R² > 0.9) while the R² values for A₁ and A3 are 0.37 and 0.64, respectively. The A₂ Ti species make a major contribution (75% approximately) to the photocatalytic degradation of trace CHCl₃ in H₂O. Note that about 95% of the photocatalytic conversion (7.2 p.p.m. CHCl₃ gTiO₂⁻¹ h⁻¹) is associated with the A₂ Ti active species in Ti-MCM-41.

4. Conclusions

The A₁ (TiO₄), A₂ [(Ti=O)O₄] and A₃ (TiO₆) Ti species in TiO₂ were observed and distinguished by component fitted pre-edge XANES spectra of the nano TiO₂ (TiO₂/SBA-15), TiO₂ cluster (TiO₂-SiO₂) and highly atomic dispersed Ti (Ti-MCM-41) photocatalysts. It appears that the photocatalytic degradation of trace CHCl₃ in drinking water may not be related to the crystalline phase, band-gap absorption edge or the particle size of the TiO₂-based photocatalysts. Nevertheless, the A₂ Ti species in the TiO₂-based photocatalysts may make a major contribution (75% approximately) to the overall photocatalytic performance. In the photocatalytic degradation of trace CHCl_3, for example, the higher A₂ Ti species in the atomic dispersed Ti (Ti-MCM-41) may account for 95% of the accumulated photocatalytic conversion. This work reveals that the A₂ Ti species of a TiO₂-based photocatalyst are associated significantly with its photocatalytic reactivity, especially in photocatalytic degradation of CHCl₃.