Alcohol solvents evaporation-induced self-assembly synthesis of mesoporous TiO_2−x−y C _x N _y nanoparticles toward visible-light driven photocatalytic activity

Abstract

A one-step solvent evaporation-induced self-assembly (SEISA) process was demonstrated to prepare carbon and nitrogen co-doping mesoporous TiO₂ nanoparticles (MesoTiO_2−x−y C _x N _y -S) using an ionic liquid as carbon and nitrogen sources as well as mesoporous template. After the evaporation of different solvents (methanol, ethanol, and isopropanol) and subsequent calcinations at 773 K, the obtained MesoTiO_2−x−y C _x N _y -S samples were systematically characterized by a variety of spectroscopic and analytical techniques, including small- and large-angle X-ray diffraction (XRD), Raman, transmission electron microscopy (TEM), N₂ adsorption–desorption isotherms, Fourier transform infrared (FTIR), and X-ray photoelectron (XPS) spectroscopies. The results indicate that the solvents play an essential role on the chemical microstructure, doping elemental states, and photocatalytic performance of catalysts. The MesoTiO_2−x−y C _x N _y -I samples have the lowest band gap of ca. 2.75 eV and strongest absorbance of visible light in the range of 400–600 nm. Among the MesoTiO_2−x−y C _x N _y -S photocatalysts, the MesoTiO_2−x−y C _x N _y -M catalysts show superior photocatalytic activity of hydrogen generation in methanol aqueous solution under visible light irradiation as compared to MesoTiO_2−x−y C _x N _y -E, MesoTiO_2−x−y C _x N _y -I, and commercial Degussa TiO₂. This result could be attributed to the moderate C,N co-doping amounts on their developed mesoporous texture (pore size = 8.0 nm) and high surface area (107 m² g⁻¹) of TiO₂ (crystallite size = 9.9 nm) in the MesoTiO_2−x−y C _x N _y -M catalysts.

Introduction

In the recent years, much effort has been devoted to modification of TiO₂ in order to allow the utilization of a wider fraction of visible light (Ould-Chikh et al. 2014; Venieri et al. 2014; Zhang et al. 2014). Until now, the adjusting of the electronic structure of TiO₂ materials is an effective approach, for example, substituting lattice oxygen with nonmetal elements such as C, N, S, F, Cl, or I (Han et al. 2014; Huang and Wey 2013; Umadevi et al. 2014; Vereb et al. 2013; Wang et al. 2009). It was reported that co-doped TiO₂ with two or more nonmetal elements, such as C–N (Cong et al. 2006; Dolat et al. 2012; Ruzybayev and Shah 2014; Wang and Lim 2011; Wu and Ju 2014; Xu et al. 2011; Yang et al. 2008; Zhang and Song 2009), S–N (Todorova et al. 2013), N–F (Chen et al. 2009) and C–S (Scarisoreanu et al. 2014) can further exhibit synergistic effect of increasing the visible-light activities. Among them, C and N co-doping onto TiO₂ has shown to have superior photocatalytic performance under visible light (Chen et al. 2007a) because of its large specific surface area and developed porosity favors solar energy conversion. In the earlier reports, C,N co-doping can be attained by mechano-chemical, ammonolysis, and sputtering processes (Hojamberdiev et al. 2012; Noguchi et al. 2005; Yin et al. 2007). These methods may be performed at a high temperature which can unavoidably destruct the mesoporosity of the mesoporous TiO₂, lead to phase transition and lower their photocatalytic performance. To keep their developed porosity, a lot of low temperature processes, such as sol–gel (Xu et al. 2011), hydrothermal, and solvothermal methods (Cong et al. 2006; Dolat et al. 2012; Wang and Lim 2011) have been proposed for preparing mesoporous TiO₂. Among these techniques, hydrolysis, nitriding, and crystallization processes were controlled by the mixture of solvent, surfactant template, and chemical sources of Ti, C, and N, which can determine the mesoporosity of the mesoporous TiO₂.

Solvent evaporation-induced self-assembly (SEISA) has been reported to be a good method to synthesize mesoporous TiO₂ thin films (Martínez-Ferrero et al. 2007) because it is easy to handle the synthesis system (solvents, surfactant templates, and Ti precursors). In addition, the possible recycle of evaporated solvent makes it an environmental friendly synthetic route to scale up. The SEISA was proven to be successful in preparation of N-doped mesoporous TiO₂ thin films by postnitriding mesoporous TiO₂ photoelectrodes (Martínez-Ferrero et al. 2007; Soni et al. 2008). However, limited studies were found in the synthesis of C and N co-doped mesoporous TiO₂ powders using SEISA process.

Ionic liquids (ILs) have many distinctive properties, including negligible vapor pressures, superior thermal stability, tunable solubility for both organic and inorganic molecules, and much synthetic flexibility (Dupont and Scholten 2010; Ma et al. 2010). Recently, the utilization of ILs in synthesis of metal nanoparticles, supported catalysts and functional material composites has been widely investigated. Also, ILs used as the templates for micro and mesostructured materials have been regarded as a promising substitute for structural directing agents in nanocasting. For example, various ILs such as [C₄mim][PF₆], [C₄mim][BF₄], [C₄mim][CF₃SO₃], and [C₆mim][PF₆] have been employed as the templates for the preparation of mesoporous TiO₂ (Chen et al. 2007b; Han et al. 2010; Li et al. 2014; Sajjad et al. 2013) that not only had a high surface area but also a controllable pore size distribution, which improved the photoactivities. In the previous work (Liu and Syu 2013), we reported a facile procedure to synthesize C,N co-doped mesoporous TiO₂ nanoparticles using an IL as carbon, nitrogen sources, and mesoporous templates. In this work, to further study the solvent effects on the developed mesoporosity of C,N co-doped TiO₂ during SEISA process, a certain amounts of ILs (1-Ethyl-3-methylimidazolium chloride, EMIC) was added into various alcohol solutions (methanol, ethanol, and isopropanol) of titanium (IV) isopropoxide under continuous stirring, followed by treating a SEISA process and calcinations at 773 K. The resulting MesoTiO_2−x−y C _x N _y -S materials were characterized by various spectroscopic and analytical techniques. The effects of different alcohols on the SEISA synthesis of mesostructured C,N co-doped TiO₂ nanoparticles were evaluated by photocatalytic activities in water splitting using visible light.

Experimental method

Photocatalysts preparation

First, ca. 3.0 g of 1-Ethyl-3-methylimidazolium chloride (EMIC, Sigma) were added in a 40 mL of various solvents (methanol, ethanol, and isopropanol), followed by adding 3.4 g of 35 wt% hydrochloric acid and continuously stirring until dissolution. Second, ca. 4 mL water containing 1.2 g citric acid was added into the above solution under vigorous stirring for 2 h at room temperature. Consequently, 5.6 g titanium (IV) isopropoxide (TTIP, ACROS) was dissolved into 40 mL of respective solvents (methanol, ethanol, and isopropanol) and then slowly added to the above solution. The mixture solution was performed with the SEISA technique at 343 K for 48 h to attain the solid product. Finally, the resultant solid was heated to 773 K with a heating rate of 1 K min⁻¹ and maintained at the same temperature for 5 h under the air flow to remove template. The obtained C,N co-doped mesoporous TiO₂ nanoparticles were denoted as MesoTiO_2−x−y C _x N _y -S where S represents the different solvents mentioned above.

Characterization methods

To identify the crystallite size and phase of samples, the powder X-ray diffraction (XRD) patterns were obtained with a PANalytical (X’Pert PRO) instrument using Cu-Kα radiation (λ = 0.1541 nm) in a 2θ range of 20–80°. Raman measurements were operated with a Lambda Dimension-P2 spectrometer coupled with a microscope in a reflectance mode using a 785 nm excitation laser source and a spectral resolution of 0.3 cm⁻¹. Small-angle X-ray scattering (SAXS) of all samples was measured on a Nanostar U system (Bruker, AXS Gmbh). The high-resolution transmission electron microscopy (TEM) images were performed on an electron microscope (JEOL TEM-3010) with an electron acceleration voltage of 200 kV. Nitrogen adsorption isotherms were conducted at 77 K using a Micromeritics ASAP 2020 system and the corresponding specific surface areas were calculated from nitrogen adsorption data in the relative pressure (P/P₀) range from 0.05 to 0.2 via Brunauer-Emmett-Teller (BET) equation. The pore volumes and pore size distribution curves were derived from the adsorption branch using Barrett-Joyner-Halenda (BJH) model. Fourier transform infrared (FTIR) spectra were obtained on a Bio-rad 165 spectrometer with 4 cm⁻¹ resolution using KBr pellets at room temperature. X-ray photoelectron spectra (XPS) were collected on a spectrometer (PHI 5000 VersaProbe) in an ultrahigh vacuum condition (10⁻¹⁰ Torr) at room temperature. The binding energies in XPS spectra for the synthesized samples have been referenced to the C 1 s core level at 284.5 eV of adventitious carbon. A JASCO V-670 UV–vis spectrometer with an integrating sphere apparatus was used to record UV–vis diffuse reflectance spectra of samples.

Photocatalytic measurements

The photocatalytic activities of photocatalysts were examined in terms of hydrogen production from water splitting under visible light irradiation. The reaction was carried out in a top-irradiation photoreactor with an optical Quartz window (Liu and Syu 2012). Briefly, ca. 0.2 g photocatalyst was dispersed homogeneously with a magnetic stirrer in 155 mL of aqueous solution containing 40 mL methanol in the reaction vessel. Prior to the irradiation, the suspension was purged with extra pure argon to remove air. Then, the solution was illuminated by a 450 W Xe lamp (Newport, Model 66142) with a cut-off filter (λ ≥ 420 nm) under continuously magnetic stirring. The temperature of the whole reactor was kept at 298 K with a circulation of cold water. The hydrogen evolution was frequently measured by online gas chromatography (Bruker 430-GC) equipped with a thermal conductivity detector (TCD) and a packed column (molecular sieve 5A, 99.999 % N₂ carrier).

Results and discussion

As shown in Fig. 1, the large-angle XRD patterns of the MesoTiO_2-x-y C _x N _y -S samples all indicated well-resolved peaks at 2θ = 25.3°, 37.9°, 48.3°, 53.9°, 55.2°, 62.9°, 69.1°, 70.3°, and 75.2°, respectively, which are the features of (101), (004), (200), (105), (211), (204), (116), (220), and (215) diffraction peaks. This indicates the existence of well crystalline anatase phase (JCPDS, CAS number 21-1272) of all TiO₂ nanoparticles without phase transition after heat treatments. As can be seen in Table 1, the crystallite sizes (d _p) of all samples were calculated using Scherrer formula based on (101) diffraction peak. It was observed that the d _p was in the order of MesoTiO_2−x−y C _x N _y -I > MesoTiO_2−x−y C _x N _y -E > MesoTiO_2−x−y C _x N _y -M, which implied that the solvents with shorter alcohol chain lengths could hinder the grain growth. Additionally, the intensity of diffraction peak (101) of MesoTiO_2−x−y C _x N _y -M samples was found to be slightly decreased, indicating the restricted crystallization of TiO₂ nanoparticles affected by methanol in the SEISA synthesis. Additional Raman spectroscopy displayed in Fig. 2 was used to further confirm the crystalline phase of TiO₂. Six Raman peaks were observed for MesoTiO_2−x−y C _x N _y -S samples at 144 cm⁻¹ (E_g), 197 cm⁻¹ (E_g), 399 cm⁻¹ (B_1g), 513 cm⁻¹ (A_1g), 519 cm⁻¹ (B_1g), and 639 cm⁻¹ (E_g), suggesting the pure anatase phase of TiO₂ which is in accordance with the XRD results. Also, a slight shift toward high wavenumber region and broadening of E_g bands at 144 cm⁻¹ are observed for MesoTiO_2−x−y C _x N _y -M while methanol was used as solvent during SEISA process, which may be due to the formation of oxygen deficiency (Parker and Siegel 1990) and small particle size (Yang et al. 2010) of TiO₂ nanoparticles. This finding is consistent with the above-mentioned XRD results.

Fig. 1

Large-angle XRD patterns of MesoTiO_2−x−y C _x N _y -S samples

Table 1 Textural properties of MesoTiO_2−x−y C _x N _y -S samples

Fig. 2

Raman spectra of MesoTiO_2−x−y C _x N _y -S samples

The small-angle powder X-ray scattering patterns (see Fig. 3) of MesoTiO_2−x−y C _x N _y -M, MesoTiO_2−x−y C _x N _y -E and MesoTiO_2−x−y C _x N _y -I samples show only one feature at 2θ ≒ 0.35°, 0.20°, and 0.25°, indicating the existence of wormlike mesopores with maximum distribution of the nearest-neighbor particle-center-to-particle-center distances of 25.2, 44.1, and 35.3 nm, respectively. Among all the MesoTiO_2−x−y C _x N _y -S samples, MesoTiO_2−x−y C _x N _y -M possesses the obvious diffraction peak. However, the weaker and unclear peaks can be observed for MesoTiO_2−x−y C _x N _y -E and MesoTiO_2−x−y C _x N _y -I, indicating the less ordering of mesoporous structure was formed upon the presence of ethanol and isopropanol during SEISA. This result implies that the degree of self-assembly may be affected by the solvents existing in synthetic mixture.

Fig. 3

Small-angle XRD patterns of MesoTiO_2−x−y C _x N _y -S samples

As displayed in Fig. 4a, the representative TEM images of MesoTiO_2−x−y C _x N _y -M show low and high electron density regions throughout the samples which are due to the formation of TiO₂ nanoparticles in a mesoscopic disordering during SEISA. A similar result (Sivaranjani and Gopinath 2011) was reported previously that the wormhole-like mesoporosity of N-doped TiO₂ synthesized via solution combustion method was observed. The mesoporous structures formed in the inter-nanoparticles would increase the light scattering inside the pore channels and thus enhance the light harvesting (Pan et al. 2008). Moreover, the high-resolution TEM image in Fig. 4b indicates that the majority of lattice fringes are attributed to (101) plane, i.e., d₁₀₁ ≒ 0.35 nm of anatase phase, in agreement with the result obtained from XRD and Raman analysis. As can be seen in the inset of Fig. 4b, selected area electron diffraction (SAED) pattern of the MesoTiO_2−x−y C _x N _y -M demonstrates well-resolved diffuse electron diffraction rings with their respective d-spacings of (101), (004), (200), (211), (204), (220), and (215), again in a good accordance with the anatase phase of spherical TiO₂ nanocrystals (JCPDS, CAS number 21-1272).

Fig. 4

a Low magnification, b high magnification TEM images of MesoTiO_2−x−y C _x N _y -M samples. Inset corresponding selected area electron diffraction pattern

Figure 5a shows the N₂ adsorption–desorption isotherms of MesoTiO_2-x-y C _x N _y -S samples. According to the IUPAC classification, all the samples are type-IV isotherms with well-developed hysteresis loops, implying the typical signatures of mesoporosity. Additionally, large uniform mesopores with a cage-like pore structure connected by small size windows are observed in these samples (Das et al. 2009) since the loop in the P/P₀ range of 0.4–0.9 is an intermediate between typical H1 and H2-type hysteresis. Compared with MesoTiO_2−x−y C _x N _y -M, the hysteresis loops of MesoTiO_2−x−y C _x N _y -E, and MesoTiO_2−x−y C _x N _y -I, the shifting to the region of higher relative pressure suggests the increasing pore sizes. The pore size distributions determined from adsorption branches of isotherms by the BJH method of the MesoTiO_2−x−y C _x N _y -S are shown in Fig. 6. The pore sizes ranges of MesoTiO_2−x−y C _x N _y -M and MesoTiO_2−x−y C _x N _y -E samples are between 3 to 10 nm, indicating the narrow distributions of the mesopores. Nonetheless, the broad distribution of pore range is observed for MesoTiO_2−x−y C _x N _y -I, suggesting the perturbation of mesoporous ordering upon the usage of isopropanol during SEISA. The properties of MesoTiO_2−x−y C _x N _y -S samples obtained from N₂ adsorption–desorption isotherms such as BET surface area, pore volume, and pore size are summarized in Table 1. The MesoTiO_2−x−y C _x N _y -M samples were found to possess BET surface areas of ca. 107 m² g⁻¹ and a uniform pore size distribution of 8.0 nm. The much lower sorption curves and the broader pore size distribution were observed for MesoTiO_2−x−y C _x N _y -E and MesoTiO_2−x−y C _x N _y -I, resulting in their lower surface areas (ca. 76 and 56 m² g⁻¹, respectively, in Table 1) and pore volumes, which may be caused by the more disorder mesostructures and larger crystallite size (d _p), while the samples were prepared from ethanol and isopropanol solvents during SEISA.

Fig. 5

N₂ adsorption/desorption isotherm of MesoTiO_2−x−y C _x N _y -S samples

Fig. 6

The BJH pore size distribution of MesoTiO_2−x−y C _x N _y -S samples

The FTIR spectra of the MesoTiO_2−x−y C _x N _y -S are shown in Fig. 7. The broad peaks observed at ca. 3,000–3,700 and 1,640 cm⁻¹ could be attributed to surface adsorbed water and hydroxyl groups; a major factor influencing the photocatalytic activity (Ding et al. 2000) which is generally motivated by hydroxyl radicals produced via the step involving the attack of photogenerated holes on surface bound water or hydroxyl groups (Dhananjeyan et al. 2000). A weak feature at ca. 1,400 cm⁻¹ could be attributed to the CO or NH₄ ⁺ groups which imply that carbon and nitrogen were successfully incorporated into the structural lattice (Xu et al. 2010). The signals in the range of 500–1,250 cm⁻¹ are characteristic of the formation of O–Ti–O lattices (Chen et al. 2005). In addition, two peaks at 2,920 and 2,851 cm⁻¹ corresponding to asymmetric and symmetric C–H stretching vibrations, respectively, were observed in the MesoTiO_2−x−y C _x N _y -S samples.

Fig. 7

FTIR spectra of various MesoTiO_2−x−y C _x N _y -S samples

The chemical states of carbon and nitrogen atoms doped in the MesoTiO_2−x−y C _x N _y -S samples were examined by XPS analysis. The high-resolution C 1 s XPS spectra of MesoTiO_2−x−y C _x N _y -S are shown in Fig. 8. All the MesoTiO_2−x−y C _x N _y -S samples possess a peak at 284.5 eV which is attributed to the presence of adventitious carbon. An additional peak at ca. 285.7 eV which can be ascribed to C–O bonds was observed in MesoTiO_2−x−y C _x N _y -S, indicating that carbonated species may be formed or carbon atoms may substitute titania atoms in the form of Ti–O-C bonds on the TiO₂ surface (Song et al. 2013; Zhao et al. 2010). Moreover, the MesoTiO_2−x−y C _x N _y -I sample has the highest contents of carbon (6.6 at%, see Table 1) doped in the titania surface among the MesoTiO_2−x−y C _x N _y -S. Based on the earlier reports (Wang et al. 2012; Zhao et al. 2010), formation of carbonated species on TiO₂ surface might act as the photosensitizer to improve the visible-light absorption ability. It is worth to note that the peak at ca. 281 eV from Ti–C bonds was not found, suggesting that carbon atoms may not replace O atoms in the TiO₂ lattice.

Fig. 8

XPS spectra of MesoTiO_2−x−y C _x N _y -S at the C 1 s region

Furthermore, the states of N doped on MesoTiO_2−x−y C _x N _y -S samples were studied using high-resolution and curve-fitting of N 1 s XPS spectra, as shown in Fig. 9. A broad peak ranging from 396 to 402 eV was observed in the MesoTiO_2−x−y C _x N _y -S at N 1 s region. For the MesoTiO_2−x−y C _x N _y -S samples, the overlay of two deconvoluted peaks at 398.4 and 400.2 eV was observed, indicating no formation of N–Ti–N bonds (396 eV) by replacement of the oxygen atoms in the TiO₂ crystal lattice (Manole et al. 2014). The N 1 s peaks at 398.4 eV can be attributed to the anionic N⁻ in the O–Ti–N linkages (Etacheri et al. 2010), as the electronegativity of N atoms is smaller than O atoms, the electron density around N is decreased as compared to that in a TiN crystal. However, the N 1 s peak at 400.2 eV can be ascribed to oxidized nitrogen, such as Ti–O–N or Ti–N–O linkages in the MesoTiO_2−x−y C _x N _y -S samples (Etacheri et al. 2010; Naik et al. 2011). In addition, the weak feature at 402.2 eV has been known to represent molecular adsorbed N species (Tian et al. 2010). From XPS analysis, the corresponding percentages of N dopants are estimated to be 3.1–4.8 at%, which were successfully doped on MesoTiO_2−x−y C _x N _y -S samples, as indicated in Table 1. UV–vis diffuse reflectance (UV–vis DR) spectra of commercial TiO₂ (P25) and MesoTiO_2−x−y C _x N _y -S samples are shown in Fig. 10a. It can be seen that a broad background absorption in the visible light region is observed for MesoTiO_2−x−y C _x N _y -M, which is not present in the case of P25. Upon using ethanol and isopropanol as solvents during SEISA, further increases in the intensity of visible light can be found. In addition, MesoTiO_2−x−y C _x N _y -S samples possess a remarkable red-shift absorbance threshold as compared to P25, implying that the increased visible-light absorption edges and intensities of MesoTiO_2−x−y C _x N _y -S can be due to the synergistic effect of carbon and nitrogen species co-doped on the TiO₂ (Chen et al. 2007a). The plot of (αhν)² vs. the energy of the absorbed light (Fig. 10b) can be used to evaluate the band gaps of the synthetic samples which are indirect semiconductors. As a result, their corresponding energy band gaps are obtained and summarized in Table 1. All the MesoTiO_2−x−y C _x N _y -S samples have lower band gap energies as compared to that of P25 (ca. 3.2 eV). Among them, the MesoTiO_2−x−y C _x N _y -I samples are found to have the lowest band gap of ca. 2.75 eV and strongest absorbance of visible light in the range of 400–600 nm, which are favorable for the photocatalytic evolution of hydrogen under visible light. This finding indicates that more carbon and nitrogen atoms could be doped into the lattice of titania and thus leads to the higher visible-light response, while the isopropanol was used during SEISA.

Fig. 9

XPS spectra of MesoTiO_2−x−y C _x N _y -S at the N 1 s region

Fig. 10

a DR UV–vis spectra and b band gap evaluation from the plots of (αhν)² vs. the energy of the absorbed light for P25 and MesoTiO_2−x−y C _x N _y -S samples

The visible-light responsive activities of P25 and various MesoTiO_2−x−y C _x N _y -S catalysts were evaluated by photocatalytic evolution of hydrogen from suspension of catalysts in aqueous methanol solution. In general, photocatalytic activity of TiO₂ can be affected by various factors such as specific surface area, crystal size, crystallinity, pore and phase structures, surface properties, and band gap values (Das et al. 2010; Kim and Kang 2012). The previous results (Das et al. 2010; Ouzzine et al. 2014) indicated that higher surface area of TiO₂ improves the absorption of more water molecules on the surface and the higher crystalline but smaller particle size results in an efficient utilization of charge carriers, while the photo-generated carriers diffuse rapidly to the surface and reacts quickly with the adsorbed water molecules. As can be seen in Fig. 11, the H₂ generation amounts of P25 and MesoTiO_2−x−y C _x N _y -S increase with increasing irradiation time. The photocatalytic activities of MesoTiO_2−x−y C _x N _y -S are highly enhanced while compared to commercial Degussa TiO₂ (P25) which is due to the little absorption in the visible light and low surface area. Among all the MesoTiO_2−x−y C _x N _y -S examined, the MesoTiO_2−x−y C _x N _y -M catalysts prepared with methanol as solvents during SEISA, which have distinctive properties including relatively high surface areas (107 m² g⁻¹, see Table 1), smaller crystallite size (9.9 nm, see Table 1), mesoporous structure, and pure anatase crystalline phase, were found to exhibit the superior visible-light photocatalytic activity. The most evident red shift (see Fig. 10) caused by C,N co-doping is observed for MesoTiO_2−x−y C _x N _y -I catalyst which has the worst photocatalytic performance among the MesoTiO_2−x−y C _x N _y -S. This finding implies that the absorbance of visible-light in the MesoTiO_2−x−y C _x N _y -S could not be the dominant factor for H₂ evolution from water splitting. From the results discussed above, C,N atoms co-doped on TiO₂ with mesoporous structure and high surface area is responsible for surpassing photocatalytic activity under visible light.

Fig. 11

Photocatalytic H₂ evolution as a function of time catalyzed by MesoTiO_2−x−y C _x N _y -S and commercial P25 photocatalysts

Conclusions

In a conclusion, C,N co-doped on mesostructured photocatalysts synthesized via a simple SEISA method with different kinds of solvents was proposed. The resultant MesoTiO_2−x−y C _x N _y -S samples were thoroughly characterized by a variety of spectroscopic and analytical techniques and subsequently used as photocatalysts for hydrogen evolution from water splitting under visible light. Although MesoTiO_2−x−y C _x N _y -S photocatalysts (S = E and I) have stronger absorption of visible light in the region of 400–600 nm, photocatalytic generation of hydrogen (78.4 μmol g⁻¹ h⁻¹) for MesoTiO_2−x−y C _x N _y -M was highly enhanced over those of MesoTiO_2−x−y C _x N _y -E, MesoTiO_2−x−y C _x N _y -I and the commercial P25. This improvement of visible-light-responsive (>420 nm) photocatalysis in the splitting of water may be due to the suitable C,N co-doping on well-developed mesoporous architecture with large surface area (107 m² g⁻¹), which is almost double that of the MesoTiO_2−x−y C _x N _y -I.