Highly Crystallized C-Doped Mesoporous Anatase TiO 2 with Visible Light Photocatalytic Activity

Highly crystallized C-doped mesoporous anatase TiO2 is prepared using a multi-walled carbon nanotube (MWCNT) mat as both a “rigid” pore template and a carbon doping source. SEM and TEM characterization shows that the MWCNT template imposed a pore structure in reverse of that of the MWCNT mat. The pore walls are formed by chain-like interconnected TiO2 nanocrystals with an average diameter about 10 nm, and pores are derived from spaces occupied by MWCNTs before removal. XRD characterization shows that TiO2 is crystallized with a pure anatase phase. XPS characterization reveals that the relative carbon content in the TiO2 is related to the duration of TiO2/MWCNT composite annealing before removal of MWCNT template. Three plateau of adsorption from 450–800 nm in UV–Vis spectra in comparison to that of P25; and display visible light photocatalytic activity for decomposition of methyl orange (MO) in relationship with the carbon content and crystallinity of the anatase TiO2.


Introduction
Since the discovery of its photocatalytic activity more than four decades ago, TiO 2 has received a great deal of attention as a photocatalyst owing to its excellent properties, such as environmental friendliness, chemical stability, and low cost [1][2][3][4][5].Despite extensive effort worldwide, there are still problems which limit the effectiveness of TiO 2 catalyst, including restrictive light absorption (only responsive to ultraviolet light with a wavelength below 387 nm due to its wide band-gap); fast charge-carrier recombination; and low interfacial charge-transfer rate of photogenerated carriers.
Several methods have been reported in the literature for the synthesis of C-doped TiO 2 [21][22][23][24][25], some produced C-doped TiO 2 with sharp UV-Vis absorption edge with variable anatase TiO 2 phase stability and morphology [21][22][23] and some produced materials with poorly crystallized structure lack of a sharp UV-Vis adsorption edge [24,25].Shi et al. [23] prepared C-doped TiO 2 hollow spheres with hierarchical macroporous channels using carbon spheres as both a template and a carbon doping source.
In this report, we prepare a highly crystallized C-doped mesoporous anatase TiO 2 .The synthesis was carried out using a multi-walled carbon nanotube (MWCNT) mat as a "rigid" pore template.MWCNT also served as a source for carbon doping.Attempts were made to vary the carbon doping concentration in the TiO 2 samples.The morphology and structure of C-doped mesoporous TiO 2 were characterized with a scanning electron microscope (SEM), high resolution transmission electron microscope (HRTEM) and X-ray diffractometer (XRD).The relative carbon content in the samples was measured using X-ray photoelectron spectroscopy (XPS).Finally, the effect of C-doping on optical adsorption and photocatalytic properties was evaluated.

Morphology and Structure
Morphology of the samples was followed by SEM imaging at each step of the synthesis.Figure 1 shows SEM images of MWCNT mat, intermediate product-autoclave-treated MWCNT mat infiltrated with butyl titanate, and final product-mesoporous C-doped TiO 2 .As shown in Figure 1a, the MWCNT mat has a uniform network morphology formed by randomly intertwined MWCNTs.The pores in the network are interconnected spaces between MWCNTs.After pore spaces in the MWCNT mat were filled with butyl titanate precursor, MWCNT mat was autoclave-treated at 180 ˝C to convert the precursor to TiO 2 .It can be seen from Figure 2b that TiO 2 was formed uniformly inside the MWCNT network, and the pores were partially filled.The autoclave-treated samples were annealed at 600 ˝C under Ar to crystallize TiO 2 .And duration of the annealing was varied from 1, 2 to 3 h.After annealing, MWCNT template in the samples was removed by heating at 600 ˝C in air under Ar to obtain three TiO 2 samples, denoted as TiO 2 -1, TiO 2 -2 and TiO 2 -3, respectively.The TiO 2 samples took the shape of the MWCNT mat and exhibited a yellowish colour.SEM characterization did not observed any effect of annealing time on the morphology of TiO 2 .Figure 1c shows a SEM image of TiO 2 prepared from TiO 2 /MWCNT annealed for 2 h.It can be seen that MWCNTs have imposed a porous structure in reverse of the MWCNT mat, i.e., pore walls were formed by TiO 2 filling in the empty space of the MWCNT network, and the pore network, to a large extent, inherited the space occupied by MWCNTs in the MWCNT mat.The pore walls are interconnected TiO 2 nanoparticles.To further characterize the morphology of TiO 2 sample, we have carried out HRTEM imaging.Figure 2 shows HRTEM images recorded from TiO 2 -2.HRTEM image (Figure 2a) confirmed SEM observation that the pore walls of TiO 2 were formed by chain-like interconnected TiO 2 nanoparticles.The average diameter of TiO 2 nanoparticles is around 10 nm.As shown in Figure 2b, TiO 2 nanoparticles are highly crystallized single crystallites.The lattice fringe distance measured from Figure 2b is 0.35 nm, corresponding to that of (101) planes of the anatase TiO 2 .
Catalysts 2016, 6, 117 2 of 10 lack of a sharp UV-Vis adsorption edge [24,25].Shi et al. [23] prepared C-doped TiO2 hollow spheres with hierarchical macroporous channels using carbon spheres as both a template and a carbon doping source.In this report, we prepare a highly crystallized C-doped mesoporous anatase TiO2.The synthesis was carried out using a multi-walled carbon nanotube (MWCNT) mat as a "rigid" pore template.MWCNT also served as a source for carbon doping.Attempts were made to vary the carbon doping concentration in the TiO2 samples.The morphology and structure of C-doped mesoporous TiO2 were characterized with a scanning electron microscope (SEM), high resolution transmission electron microscope (HRTEM) and X-ray diffractometer (XRD).The relative carbon content in the samples was measured using X-ray photoelectron spectroscopy (XPS).Finally, the effect of C-doping on optical adsorption and photocatalytic properties was evaluated.

Morphology and Structure
Morphology of the samples was followed by SEM imaging at each step of the synthesis.Figure 1 shows SEM images of MWCNT mat, intermediate product-autoclave-treated MWCNT mat infiltrated with butyl titanate, and final product-mesoporous C-doped TiO2.As shown in Figure 1a, the MWCNT mat has a uniform network morphology formed by randomly intertwined MWCNTs.The pores in the network are interconnected spaces between MWCNTs.After pore spaces in the MWCNT mat were filled with butyl titanate precursor, MWCNT mat was autoclave-treated at 180 °C to convert the precursor to TiO2.It can be seen from Figure 2b that TiO2 was formed uniformly inside the MWCNT network, and the pores were partially filled.The autoclave-treated samples were annealed at 600 °C under Ar to crystallize TiO2.And duration of the annealing was varied from 1, 2 to 3 hrs.After annealing, MWCNT template in the samples was removed by heating at 600 °C in air under Ar to obtain three TiO2 samples, denoted as TiO2-1, TiO2-2 and TiO2-3, respectively.The TiO2 samples took the shape of the MWCNT mat and exhibited a yellowish colour.SEM characterization did not observed any effect of annealing time on the morphology of TiO2. Figure 1c shows a SEM image of TiO2 prepared from TiO2/MWCNT annealed for 2 hrs.It can be seen that MWCNTs have imposed a porous structure in reverse of the MWCNT mat, i.e. pore walls were formed by TiO2 filling in the empty space of the MWCNT network, and the pore network, to a large extent, inherited the space occupied by MWCNTs in the MWCNT mat.The pore walls are interconnected TiO2 nanoparticles.To further characterize the morphology of TiO2 sample, we have carried out HRTEM imaging.Figure 2 shows HRTEM images recorded from TiO2-2.HRTEM image (Figure 2a) confirmed SEM observation that the pore walls of TiO2 were formed by chain-like interconnected TiO2 nanoparticles.The average diameter of TiO2 nanoparticles is around 10 nm.As shown in Figure 2b, TiO2 nanoparticles are highly crystallized single crystallites.The lattice fringe distance measured from Figure 2b is 0.35 nm, corresponding to that of (101) planes of the anatase TiO2.The crystallographic structure of TiO2 samples was characterized with X-ray diffraction (XRD).As shown in Figure 3, three samples have the same number of diffraction peaks, each at the same 2θ location.The only difference observed is that the patterns of TiO2-2 and TiO2-3 are identical, and slightly sharper than that of TiO2-1.All peaks at 2θ of 25.4°, 37.9°, 48.1°, 54.2°, 55.2°, 62.8°, 69.0°, 71.1°, 75.4° and 83.2° can be assigned to (101), (004), ( 200), ( 105), ( 211), ( 204), ( 116), ( 220), ( 215) and (224) reflections of anatase (JCPDS No. 21-1272), proved that the samples are crystallized in pure anatase phase.The crystallite sizes was calculated from the full width at half maximum (FWHM) of the (101) peak of XRD pattern using Scherrer equation.As listed in Table 1, they are 10.35 nm, 11.85 nm and 11.86 nm for TiO2-1, and TiO2-2 and TiO2-3, respectively, indicating that extending annealing time from 2 to 3 hrs did not affect the size of TiO2 nanocrystals.11.9 102.9 0.66 3.9 1 Calculated TiO2 nanocrystal diameter from XRD;2 Surface area;3 Total pore volume; 4 Calculated C atomic percentage doped in TiO2.The crystallographic structure of TiO 2 samples was characterized with X-ray diffraction (XRD).As shown in Figure 3, three samples have the same number of diffraction peaks, each at the same 2θ location.The only difference observed is that the patterns of TiO 2 -2 and TiO 2 -3 are identical, and slightly sharper than that of TiO 2 -1.All peaks at 2θ of 25.4 ˝, 37.9 ˝, 48.1 ˝, 54.2 ˝, 55.2 ˝, 62.8 ˝, 69.0 ˝, 71.1 ˝, 75.4 ˝and 83.2 ˝can be assigned to (101), ( 004), ( 200), ( 105), ( 211), ( 204), ( 116), ( 220), ( 215) and ( 224) reflections of anatase (JCPDS No. 21-1272), proved that the samples are crystallized in pure anatase phase.The crystallite sizes was calculated from the full width at half maximum (FWHM) of the (101) peak of XRD pattern using Scherrer equation.As listed in Table 1, they are 10.35 nm, 11.85 nm and 11.86 nm for TiO 2 -1, and TiO 2 -2 and TiO 2 -3, respectively, indicating that extending annealing time from 2 to 3 h did not affect the size of TiO 2 nanocrystals.The crystallographic structure of TiO2 samples was characterized with X-ray diffraction (XRD).As shown in Figure 3, three samples have the same number of diffraction peaks, each at the same 2θ location.The only difference observed is that the patterns of TiO2-2 and TiO2-3 are identical, and slightly sharper than that of TiO2-1.All peaks at 2θ of 25.4°, 37.9°, 48.1°, 54.2°, 55.2°, 62.8°, 69.0°, 71.1°, 75.4° and 83.2° can be assigned to (101), ( 004), ( 200), ( 105), ( 211), ( 204), ( 116), ( 220), ( 215) and ( 224) reflections of anatase (JCPDS No. 21-1272), proved that the samples are crystallized in pure anatase phase.The crystallite sizes was calculated from the full width at half maximum (FWHM) of the (101) peak of XRD pattern using Scherrer equation.As listed in Table 1, they are 10.35 nm, 11.85 nm and 11.86 nm for TiO2-1, and TiO2-2 and TiO2-3, respectively, indicating that extending annealing time from 2 to 3 hrs did not affect the size of TiO2 nanocrystals.11.9 102.9 0.66 3.9

Surface Area and Pore Size Distribution
1 Calculated TiO2 nanocrystal diameter from XRD; 2 Surface area; 3 Total pore volume; 4 Calculated C atomic percentage doped in TiO2.Table 1.Particle size, specific surface area, total pore volume and C content of the samples.11.9 102.9 0.66 3.9

Surface Area and Pore Size Distribution
Figure 4a shows nitrogen adsorption-desorption isotherms curves of three samples.All curves showed a hysteresis loop, characteristic of a type H2 isotherm according to International Union of Pure and Applied Chemistry (IUPAC) classification [26], suggesting mesoporous nature of the pore structure.The surface areas listed in Table 1 were calculated from the low-pressure portion of the adsorption isotherm using the Brunauer-Emmett-Teller (BET) method.The surface area of TiO 2 -1 is 129.1 m 2 g ´1, which decreased to 102.3 and 102.9 cm 3 g ´1 for TiO 2 -2 and TiO 2 -3, respectively.
Figure 4a shows nitrogen adsorption-desorption isotherms curves of three samples.All curves showed a hysteresis loop, characteristic of a type H2 isotherm according to International Union of Pure and Applied Chemistry (IUPAC) classification [26], suggesting mesoporous nature of the pore structure.The surface areas listed in Table 1 were calculated from the low-pressure portion of the adsorption isotherm using the Brunauer-Emmett-Teller (BET) method.The surface area of TiO2-1 is 129.1 m 2 g −1 , which decreased to 102.3 and 102.9 cm 3 g −1 for TiO2-2 and TiO2-3, respectively.
The pore size distribution analysis and cumulative pore volume calculation were carried out using the Barrett-Joyner-Halenda (BJH) approach.The pore size distributions for all three samples are similar.The result for TiO2-2 is plotted in Figure 4b.It can be seen that pore sizes are distributed in the range of 6 to 60 nm, peaked around 24 nm which is about twice of the diameters of MWCNTs.The total pore volume of TiO2-1 is 0.85 cm 3 g −1 , which decreased to 0.64 and 0.66 cm 3 g −1 forTiO2-2, and TiO2-3, respectively.

Carbon Content
Surface elemental compositions of the samples were analyzed by X-ray photoelectron spectroscopy (XPS).Spectra of all three samples were similar.Figure 5a shows a full scan spectrum of TiO2-2.All peaks belong to TiO2 and carbon, suggesting no detectable impurities except doped carbon are present in the samples.Two strong peaks at 527 and 456 eV can be attributed to O1s and Ti2p excitations, respectively.The weak peak at 284 eV belongs to C1s.In order to analyze bonding nature of carbon atoms and estimate carbon content in the samples, focused scan was carried out around 284 eV. Figure 5b shows the high resolution XPS spectrum of TiO2-2.The spectrum can be fitted into two peaks at 284.7 eV and 288.4 eV.The peak at 284.7 eV can be assigned to residue residual carbon with sp2 hybridization in the sample.The peak at 288.4 eV is believed to be from C atoms which substituted Ti in the TiO2 lattice [27][28][29].The relative carbon content in the samples were calculated from XPS spectra, which are ~2.3%,~2.8% and 3.9% for TiO2-1, TiO2-2 and TiO2-3, respectively (Table 1), increased as annealing time for TiO2/MWCNT increased from 1 to 3 hrs.The pore size distribution analysis and cumulative pore volume calculation were carried out using the Barrett-Joyner-Halenda (BJH) approach.The pore size distributions for all three samples are similar.The result for TiO 2 -2 is plotted in Figure 4b.It can be seen that pore sizes are distributed in the range of 6 to 60 nm, peaked around 24 nm which is about twice of the diameters of MWCNTs.The total pore volume of TiO 2 -1 is 0.85 cm 3 g ´1, which decreased to 0.64 and 0.66 cm 3 g ´1 forTiO 2 -2, and TiO 2 -3, respectively.

Carbon Content
Surface elemental compositions of the samples were analyzed by X-ray photoelectron spectroscopy (XPS).Spectra of all three samples were similar.Figure 5a shows a full scan spectrum of TiO 2 -2.All peaks belong to TiO 2 and carbon, suggesting no detectable impurities except doped carbon are present in the samples.Two strong peaks at 527 and 456 eV can be attributed to O1s and Ti2p excitations, respectively.The weak peak at 284 eV belongs to C1s.In order to analyze bonding nature of carbon atoms and estimate carbon content in the samples, focused scan was carried out around 284 eV. Figure 5b shows the high resolution XPS spectrum of TiO 2 -2.The spectrum can be fitted into two peaks at 284.7 eV and 288.4 eV.The peak at 284.7 eV can be assigned to residue residual carbon with sp2 hybridization in the sample.The peak at 288.4 eV is believed to be from C atoms which substituted Ti in the TiO 2 lattice [27][28][29].The relative carbon content in the samples were calculated from XPS spectra, which are ~2.3%,~2.8% and 3.9% for TiO 2 -1, TiO 2 -2 and TiO 2 -3, respectively (Table 1), increased as annealing time for TiO 2 /MWCNT increased from 1 to 3 h.

UV-Vis Adsorption
Figure 6a shows UV-Vis absorption spectroscopy of three samples.For comparison, the spectrum measured from Degussa P25 is also plotted.All of three TiO2 samples show an adsorption edge red shift of ~30 nm and raised adsorption plateau from 400 to 800 nm in comparison to the spectrum of P25.The band gaps were calculated using a reported method [6], which are 3.02, 3.00 and 2.96 eV for TiO2-1, TiO2-2 and TiO2-3, respectively, about 0.2 eV reduction from 3.20 eV of typical band gap of the anatase TiO2.Photoluminescence (PL) spectra of the samples were measured with an excitation wavelength of 230 nm.As shown in Figure 6b, the C-doping has suppressed emissions across entire spectra from band gap emission (Strong peak) to emissions related to surface states and defects.The band gap emission peak (Inset) position of TiO2-2 is at wavelength slightly higher than that of P25, in consistent with observation from UV-Vis adsorption.

Visible Light Photcatalytic Properties
The photocatalytic activities of all samples were evaluated by the decoloration of MO solution under visible light irradiation (λ > 420 nm) without investigating the degradation intermediates in detail.The results are plotted in Figure 7. Before the reaction, the solution including MO and the catalyst was stirred in dark for one hour to establish the adsorption equilibrium.In order to ensure accuracy, the testing for each sample was repeated three times.The data presented are mean values and total error is < 5%.Under the same conditions, P25 showed no visible light photocatalytic activity.It can be seen from Figure 7a that the relative concentration of MO decreased as the reaction time

UV-Vis Adsorption
Figure 6a shows UV-Vis absorption spectroscopy of three samples.For comparison, the spectrum measured from Degussa P25 is also plotted.All of three TiO 2 samples show an adsorption edge red shift of ~30 nm and raised adsorption plateau from 400 to 800 nm in comparison to the spectrum of P25.The band gaps were calculated using a reported method [6], which are 3.02, 3.00 and 2.96 eV for TiO 2 -1, TiO 2 -2 and TiO 2 -3, respectively, about 0.2 eV reduction from 3.20 eV of typical band gap of the anatase TiO 2 .Photoluminescence (PL) spectra of the samples were measured with an excitation wavelength of 230 nm.As shown in Figure 6b, the C-doping has suppressed emissions across entire spectra from band gap emission (Strong peak) to emissions related to surface states and defects.The band gap emission peak (Inset) position of TiO 2 -2 is at wavelength slightly higher than that of P25, in consistent with observation from UV-Vis adsorption.

UV-Vis Adsorption
Figure 6a shows UV-Vis absorption spectroscopy of three samples.For comparison, the spectrum measured from Degussa P25 is also plotted.All of three TiO2 samples show an adsorption edge red shift of ~30 nm and raised adsorption plateau from 400 to 800 nm in comparison to the spectrum of P25.The band gaps were calculated using a reported method [6], which are 3.02, 3.00 and 2.96 eV for TiO2-1, TiO2-2 and TiO2-3, respectively, about 0.2 eV reduction from 3.20 eV of typical band gap of the anatase TiO2.Photoluminescence (PL) spectra of the samples were measured with an excitation wavelength of 230 nm.As shown in Figure 6b, the C-doping has suppressed emissions across entire spectra from band gap emission (Strong peak) to emissions related to surface states and defects.The band gap emission peak (Inset) position of TiO2-2 is at wavelength slightly higher than that of P25, in consistent with observation from UV-Vis adsorption.

Visible Light Photcatalytic Properties
The photocatalytic activities of all samples were evaluated by the decoloration of MO solution under visible light irradiation (λ > 420 nm) without investigating the degradation intermediates in detail.The results are plotted in Figure 7. Before the reaction, the solution including MO and the catalyst was stirred in dark for one hour to establish the adsorption equilibrium.In order to ensure accuracy, the testing for each sample was repeated three times.The data presented are mean values and total error is < 5%.Under the same conditions, P25 showed no visible light photocatalytic activity.It can be seen from Figure 7a that the relative concentration of MO decreased as the reaction time

Visible Light Photcatalytic Properties
The photocatalytic activities of all samples were evaluated by the decoloration of MO solution under visible light irradiation (λ > 420 nm) without investigating the degradation intermediates in detail.The results are plotted in Figure 7. Before the reaction, the solution including MO and the catalyst was stirred in dark for one hour to establish the adsorption equilibrium.In order to ensure accuracy, the testing for each sample was repeated three times.The data presented are mean values

Discussion
We have used an MWCNT mat with a uniform network structure as both a "rigid" pore template and a carbon doping source to prepare C-doped anatase TiO2 with high crystallinity, high surface area and narrow pore size distribution.As a pore template, the MWCNT mat is unique.Its high temperature stability under Ar offers an opportunity to resolve a contradiction between high crystallinity and high surface area faced in the synthesis of porous materials.TiO2 was first prepared inside the network of MWCNT mat by autoclave treatment of butyl titanate infiltrated MWCNT mat, then the TiO2/MWCNT samples were subject to heating at 600 °C at different duration to crystallize TiO2.After crystallization, MWCNT template was removed by heating in the air to obtain C-doped TiO2.XRD characterization showed that TiO2 samples were crystallized in pure anatase phase.SEM and TEM results revealed that the MWCNT mat imposed upon TiO2 a porous structure that was largely the inverse of the MWCNT network, i.e. the pore walls and pore network of TiO2, to a large extent, took the shapes of empty spaces and the space occupied by MWCNTs in the MWCNT mat template, respectively.The HRTEM image (Figure 2) showed that pore walls comprised chain-like interconnected TiO2 nanocrystals, each of which is a single crystal with average diameter ~10 nm.The size of the TiO2 nanocrystals was confirmed by XRD analysis, which increased from 10.4 nm of TiO2-1 to 11.9 nm of TiO2-2 owing to increasing in annealing time of TiO2/MWCNT from 1 to 2 hrs prior to removal of MWCNT template.Further increasing annealing time to 3 hrs did not promote TiO2 crystallite size growth; the size of TiO2-3 (annealed 3 hrs) remained the same, 11.9 nm of TiO2-2.The formation of such small highly crystallized TiO2 nanoparticles with uniform particle size distribution can be explained by separation and space restriction imposed to the TiO2 by MWCNTs during annealing.The surface area and pore volume analysis results are in consistent with size changes of nano TiO2 crystallites.The surface area and total pore volume decreased from 129.1 m 2 g −1 and 0.85 cm 3 g −1 for TiO2-1 to 102.3 m 2 g −1 and 0.64 cm 3 g −1 , respectively.Both values remained almost unchanged from TiO2-2 to TiO2-3.

Discussion
We have used an MWCNT mat with a uniform network structure as both a "rigid" pore template and a carbon doping source to prepare C-doped anatase TiO 2 with high crystallinity, high surface area and narrow pore size distribution.As a pore template, the MWCNT mat is unique.Its high temperature stability under Ar offers an opportunity to resolve a contradiction between high crystallinity and high surface area faced in the synthesis of porous materials.TiO 2 was first prepared inside the network of MWCNT mat by autoclave treatment of butyl titanate infiltrated MWCNT mat, then the TiO 2 /MWCNT samples were subject to heating at 600 ˝C at different duration to crystallize TiO 2 .After crystallization, MWCNT template was removed by heating in the air to obtain C-doped TiO 2 .XRD characterization showed that TiO 2 samples were crystallized in pure anatase phase.SEM and TEM results revealed that the MWCNT mat imposed upon TiO 2 a porous structure that was largely the inverse of the MWCNT network, i.e., the pore walls and pore network of TiO 2 , to a large extent, took the shapes of empty spaces and the space occupied by MWCNTs in the MWCNT mat template, respectively.The HRTEM image (Figure 2) showed that pore walls comprised chain-like interconnected TiO 2 nanocrystals, each of which is a single crystal with average diameter ~10 nm.The size of the TiO 2 nanocrystals was confirmed by XRD analysis, which increased from 10.4 nm of TiO 2 -1 to 11.9 nm of TiO 2 -2 owing to increasing in annealing time of TiO 2 /MWCNT from 1 to 2 h prior to removal of MWCNT template.Further increasing annealing time to 3 h did not promote TiO 2 crystallite size growth; the size of TiO 2 -3 (annealed 3 h) remained the same, 11.9 nm of TiO 2 -2.The formation of such small highly crystallized TiO 2 nanoparticles with uniform particle size distribution can be explained by separation and space restriction imposed to the TiO 2 by MWCNTs during annealing.The surface area and pore volume analysis results are in consistent with size changes of nano TiO 2 crystallites.The surface area and total pore volume decreased from 129.1 m 2 g ´1 and 0.85 cm 3 g ´1 for TiO 2 -1 to 102.3 m 2 g ´1 and 0.64 cm 3 g ´1, respectively.Both values remained almost unchanged from TiO 2 -2 to TiO 2 -3.
XPS analysis showed that, in addition to the role of the porogen, MWCNT also played the role of carbon dopant.It apparent that carbon concentration in TiO 2 is related to the duration of the TiO 2 /MWCNT annealing prior to removal of MWCNT template.The carbon concentration increased from 2.3% of TiO 2 -1 to 2.8% of TiO 2 -2.Though structure characterization (particle size, surface area and total pore volume) showed that TiO 2 -2 and TiO 2 -3 are almost identical, their carbon contents are different, 2.8% of TiO 2 -2 vs. 3.9% of TiO 2 -3.UV-Vis adsorption spectra of three samples showed a very sharp adsorption transition edge comparable to that of flame-synthesized P25, a reflection of high crystallinity of the samples proved by HRTEM image and XRD.The bad gap reduction induced by C-doping is small, about 0.2 eV, which is consistent with reported density functional theory (DFT) calculation [27] which predicted a small band gap reduction for C-doping at Ti sites.In addition to UV-Vis adsorption transition edge red shift, C-doping also resulted in an adsorption plateau in the visible range from 450 to 800 nm.Although light adsorption increased across the entire measured spectrum region, radiative emission intensities decreased greatly by the C-doping, suggesting a strong suppressing effect of doped C atoms on radiative recombination of photo generated charge carriers, and consequentially an improvement in photocatalytic activity.This effect has been reported for several dopants, including C and N atoms [6,30].Photocatalytic results of MO decomposition showed that all three samples are visible light photocatalytic active.The reaction kinetics followed a first order mechanism with rate constants of 0.52, 1.0 and 0.73 h ´1 for TiO 2 -1, TiO 2 -2 and TiO 2 -3, respectively.It is apparent that the activities are related to C carbon content and crystallinity of the samples [29].

Materials and Methods
Pristine MWCNTs were synthesized by chemical vapor deposition (CVD) using Co/Fe-Al 2 O 3 as the catalyst, ethylene as the carbon source [31].In order to remove amorphous carbon from the surface of MWCNTs and introduce functional groups (-OH and -COOH) to the defective sites, the as-prepared MWCNTs were soaked in a mixture solution of H 2 SO 4 and (NH 4 ) 2 S 2 O 8 (AR, Sinopharm Chemical Reagent, Shanghai, China) with a mole ratio of 1:1 for 7 days, then collected by filtration and washed with excess water to neutral.MWCNT dense mats were prepared by a vacuum filtration procedure [32].Typically, 0.1 g of functional MWCNTs was dispersed in 100 mL deionized water assisted by high speed mechanical shearing (20,000 rpm), then MWCNTs were collected on the top of a polyvinylidene fluoride (PVDF) membrane by vacuum filtration, washed and dried at 100 ˝C for 1 h to obtain a freestanding mat.
The C-doped TiO 2 film was prepared using MWCNT mat as the template.Typically, MWCNT mats were inserted into 70 mL solution containing butyl titanate and ethanol (AR, Sinopharm Chemical Reagent, Shanghai, China) at a volume ratio of 1:6.After soaking in the solution for 2 h, the MWCNT mats saturated with butyl titanate ethanol solution was transferred into a stainless steel autoclave (Xian Often Instrumen Equipment, Xi 1 an, China) lined with polytetrafluoroethylene (PTFE) (150 mL with an inner diameter of 50 mm) and heated at 180 ˝C for 24 h.After it was cooled down to room temperature, the mats were taken out and rinsed with ethanol for several times to remove the materials on external surface of the mat and then dried at 60 ˝C in argon atmosphere to obtain an intermediate product denoted as TiO 2 /MWCNT.Then, TiO 2 /MWCNT samples were annealed at 600 ˝C under flowing argon (400 sccm) for various time (1, 2, 3 h).Finally, the MWCNT template was removed by heat-treated at the same temperature under flowing mixture gases of Ar and O 2 (4:1) for one hour to yield C-doped TiO 2 .The samples were denoted as TiO 2 -1, TiO 2 -2 and TiO 2 -3 according to the annealing time, respectively.
Scanning electron microscopy (SEM) images were obtained using a FEI Quanta 250 SEM (FEI, Hillsboro, OR, USA).Transmission electron microscope (TEM) study was carried out using a JEM-2100 HT TEM (JEOL, Tokyo Japan).The X-ray diffraction (XRD) patterns were recorded with a D2 Phaser X-ray diffractometer (Bruker, Madison, WI, USA) at room temperature using Cu Kα radiation.The nitrogen adsorption isotherms of the samples were measured using an Autosorb-iQ analyzer (Quantachrome, Boynton Beach, FL, USA).Before the measurement, all samples were degassed at a temperature of 100 ˝C for 6 h.Specific surface area was calculated by BET method using linear portion of adsorption branch of the isotherms.Pore size and pore volume analysis were carried out with BJH method using desorption branch of the isotherm.X-ray photoelectron spectroscopy (XPS) analysis was carried out with an AXIS Ultra OLD X-ray Photoelectron Spectrometer (KRATOS ANALYTICAL, Manchester, UK) operated at 150 W with Al Kα irradiation.XPS data was analyzed using XPS peak 4.1 software packages (KRATOS ANALYTICAL, Manchester, UK).UV-visible absorption spectroscopies were recorded by a V-670 spectrophotometer (Jasco, Tokyo, Japan) equipped with an integrating sphere, and the baseline correction was carried out using a standard sample of barium sulfate.Photoluminescence (PL) spectra were measured on a FLS980 spectrometer (Edinburgh Instruments, Edinburgh, UK) at room temperature using 230 nm excitation.
The photocatalytic activity was evaluated with a photo reaction system [27][28][29] using methyl orange (MO) as a model pollutant.A 1000 W Xe lamp with a 420 nm glass filter (removing the UV irradiation below 420 nm), positioned in the center of a water-cooled quartz jacket, was used to provide visible light irradiation and a 50 mL cylindrical tube reactor placed at the side of quartz jacket was used as the reactor.The distance between lamp and reactive bottle was 40 mm.For a typical reaction, 50 mg TiO 2 photocatalyst powder was added into the reactor containing 50 mL of 10 mg/L MO solution to form a suspension by magnetic stirring.Then, the suspension was irradiated with visible light.During the irradiation, the temperature of the reaction solution was maintained at 30 ˝C ˘0.5 ˝C by water cooling and the suspension was stirred continuously.At a given time interval, 3 mL of suspension was taken out and immediately centrifuged to eliminate solid particles.The absorbance of supernatant was measured by a spectrophotometer at the maximum absorbance peak of MO, 465 nm.

Conclusions
Highly crystallized C-doped mesoporous anatase TiO 2 is prepared using a multi-walled carbon nanotube (MWCNT) mat as both a "rigid" pore template and a carbon doping source.As a porogen, MWCNT is unique; due to its high temperature stability, it allows crystallizing TiO 2 at 600 ˝C with minimum sintering.SEM and TEM characterization shows that MWCNT template imposed a pore structure in reverse of that of the MWCNT mat.The pore walls are formed by chain-like, interconnected TiO 2 nanocrystals with an average diameter about 10 nm, and pores are derived from spaces occupied by MWCNTs before removal.XRD characterization shows that TiO 2 is crystallized with a pure anatase phase.Furthermore, structural characterization showed that increasing the annealing time of TiO 2 /MWCNT from 1 to 2 h prior to MWCNT template removal led to a surface area and total pore value drop, from 129.1 m 2 g ´1 and 0.85 m 3 g ´1 of TiO 2 -1 to 102 m 2 g ´1 and 0.64 m 3 g ´1, and particle size increasing from 10.4 nm of TiO 2 -1 to 11.9 nm of TiO 2 -2; however, when annealing time was further increased to 3 h for TiO 2 -3, the surface area, total pore volume and particle size remained the same as those of TiO 2 -2.XPS characterization revealed that the relative carbon content in the TiO 2 is related to the duration of TiO 2 /MWCNT composite annealing before removal of MWCNT template, increasing from 2.3% of TiO 2 -1 to 2.8% of TiO 2 -2 and 3.9% of TiO 2 -3.Three samples show a ~30 nm red shift and an additional plateau of adsorption from 450-800 nm in UV-Vis spectra in comparison to that of P25.All of them are visible light photocatalytic active.The activity is related to the C-content and crystallinity of the samples.

Table 1 .
Particle size, specific surface area, total pore volume and C content of the samples.

Figure 7 .
Figure 7. Visible light photocatalytic activity of decomposition of methyl orange (MO): (a) Concentration dependent on time; (b) ln(Co/C) dependent on time.

Figure 7 .
Figure 7. Visible light photocatalytic activity of decomposition of methyl orange (MO): (a) Concentration dependent on time; (b) ln(C o /C) dependent on time.

Table 1 .
Particle size, specific surface area, total pore volume and C content of the samples.