Comparison of Fe2TiO5/C photocatalysts synthesized via a nonhydrolytic sol–gel method and solid-state reaction method

Fe2TiO5/C photocatalysts were synthesized by a solid-state reaction method (Fe2TiO5/C(S)) and nonhydrolytic sol–gel (NHSG) method (Fe2TiO5/C(N)), where C was introduced by external carbon and in situ carbon sources, respectively. The Fe2TiO5/C(N) photocatalyst with in situ carbon has much better photocatalytic degradation efficiency than that of Fe2TiO5/C(S) synthesized by doping external carbon. The superiorities of in situ carbon were demonstrated by SEM, EDS, BET and photoelectrochemical analysis. Compared with Fe2TiO5/C(S) using external carbon as a carbon source, Fe2TiO5/C(N) with in situ carbon exhibits more uniform elemental distribution, much larger surface area, higher photocurrent density and lower resistivity of interfacial charge transfer. The results show that the introduction of in situ carbon via the NHSG method more easily promotes the separation of photogenerated electron–hole pairs, owing to the uniformity of the carbon element, thereby improving the photocatalytic activity of the photocatalyst.


Introduction
Since the 21 st century, environmental pollution, especially water pollution, has gradually become a major problem restricting the survival and development of human society. 1 Large amounts of dyes are discharged into rivers without any treatment in the textile industry, posing a great threat to water resources. 2 Much of the recent research has focused on the development of environmental remediation technologies/cleanness of water, 3,4 which is famous for advanced oxidation processes (AOPs). 5,6 It is known that inexhaustible solar energy is the most ideal candidate for energy source. Thus, visible-light photocatalysis has received great attention in environmental remediation, which includes organic photocatalysts 7-9 and inorganic photocatalysts. Xie et al. developed visible-light-induced deoxygenative C2sulfonylation of quinoline N-oxides with sulnic acids and C(sp 2 )-H/O-H cross-dehydrogenative coupling of quinoxalin-2(1H)-ones with alcohols using organic dyes as the catalysts. 7,8 Though researchers have had some achievements in the eld of organic photocatalysis, the inorganic photocatalysts have the merits of high efficiency, simplicity, good repeatability, and easy operation. 10 Therefore, heterogenous photocatalysis using inorganic materials as photocatalysts is oen used to degrade organic pollutants with reactive oxygen species (ROS) and. 11,12 The increasing public awareness on the advantage of heterogenous photocatalysis has promoted the development of a variety of photocatalytic materials. 13,14 Fe 2 TiO 5 is a kind of narrow band gap semiconductor (2.2 eV), which has the strong ability to absorb visible light compared with the traditonal wide band gap semiconductor as diverse as TiO 2 . 15,16 In addtion, Fe 2 TiO 5 has the characteristics of low cost, compared with other noble metallic photocatalyst. 17,18 However, the narrow band gap will lead to the shortcoming of rapid recombination of photoinduce e À /h + pairs. 19 Thus, among the plenty of strategies to solve this problem, [20][21][22] carbon doping is one of the most effective routes to modify photocatalyst. [23][24][25][26] Janus et al. developed carbon-modied TiO 2 photocatalyst by exposure of P25 to the ethanol vapor for the carbon deposition. 23 Zhang et al. fabricated a TiO 2 / C photocatalyst using activated carbon as a support by metal organic chemical vapor deposition. 24 However, the enhancement in the above reports is likely to be the absorption capacity of organic substrates rather than the photocatalytic ability. Besides, the adoption of additional carbon precurors (e.g. sucrose, glucose) to prepare carbon-doped metal oxides requires strict preparation conditions and methods. 25,26 In our recent study, the in situ carbon was rst introduced by NHSG method into the Fe 2 TiO 5 photocatalyst. 27 The Fe 2 TiO 5 /C composite has exhibited superior photocatalytic activity in comparison with Fe 2 TiO 5 . This work aims at exploring the effect of doping methods and uniformity for carbon on the photocatalytic efficiency by the comparison of Fe 2 TiO 5 /C (N) and Fe 2 TiO 5 /C (S) . In addition, three synthetic routes are designed by using different iron precursors to investigate the reason why the Fe 2 TiO 5 can be synthesized at lower temperature by NHSG method.  3 ), basic ferric acetate (FeOH(CH 3 COO) 2 ) and ferric chloride (FeCl 3 ). The preparation process was shown in Fig. 2. Tetra-n-butyl titanite was added to above solution in the molar ratio of Fe : Ti ¼ 1 : 1 with vigorous stirring at 40 C for 1 h to generate the precursor sol. Aerwards, the sol was reuxed with magnetic stirring in oil bath at 80 C for 24 hours with the formation of wet gel. Next, the xerogel was acquired aer dring wet gel was dried at 110 C for 6 hours. The xerogel was then calcined to obtain Fe 2 TiO 5 /C (N) , where C was provided by the carbon-containing precursors.

Photoelectrochemical measurement
The photoelectrochemical properties of the prepared samples were measured on a CHI760E electrochemical station using a standard three-electrode cell. A Pt sheet was employed as the counter electrode. A saturated calomel electrode and the asprepared photocatalyst acted as the reference electrode and working electrode, respectively. The working electrode is prepared by the following process: rst, 10 mg as-prepared powder were evenly dispersed in 2 mL ethanol solution by ultrasound for 30 minutes, then 30 mL suspension sampled was dropped onto the ITO-coated glass and dried at room temperature for photoelectrochemical measurement. Thereinto, the transient photocurrent response measurement was carried out in 0.5 M Na 2 SO 4 aqueous solution under a 300 W xenon light irradiation/using a 300 W Xe lamp as illumination source. Electrochemical Impedance Spectroscopy (EIS) was conducted in 0.5 M K 3 [Fe(CN) 6 ] solution chosen as electrolyte solution and was carried out under static condition with scanning frequencies from 100 kHz to 0.01 Hz and amplitude of 5 mV.

Evaluation of visible light photocatalytic activity
The visible-light catalytic activities of the obtained Fe 2 TiO 5 /C prepared by two methods were investigated using MB as the degradation object at ambient temperature under visible light irradiation. 0.25 g of as-prepared sample was dispersed in 100 mL (50 mg L À1 ) dye solution. Before the photoreaction started, the suspension kept stirring in dark for 30 min to reach the adsorption-desorption equilibrium between the photocatalyst and the MB aqueous solution. The suspension under magnetic stirring was illuminated during exposure to visible light (>400 nm) using a 300 W Xe lamp as light source. 5 mL aliquots were withdrawn at certain time intervals for analysis. The photocatalytic degradation performance was evaluated by measuring the residual concentration of the simulating pollutant as a function of light time with an ultraviolet-visible Lambda850 spectrophotometer (PerkinElmer Instrument Company, America).

Characterizations
DSC-TG/DTA-TG analysis was conducted on a simultaneous thermal analyzer (NETZSCH STA449C). The crystal phases of the as-prepared products were characterized by a D8 Advance X-ray diffractometer at 40 kV and 30 mA using CuKa radiation source. The morphological observations were carried out by a eldemission scanning electron microscope (JEOL JSM-6700E) equipped with an energy-dispersive X-ray spectroscopy (EDS). For identifying the chemical bonds of xerogel, Fourier transform infrared spectroscopy (FTIR) were recorded between 400 and 4000 cm À1 using a Nicolet 5700 spectrometer. Raman spectra was recorded from 100 cm À1 to 2000 cm À1 on a Renishaw Laser Confocal Raman Spectrometer(in Via, England). Brunauer-Emmett-Teller (BET) specic surface area of as-synthesized samples were evaluated by isothermal N 2 adsorption with an ASAP 2020 surface area analyzer of Micromeritics.

Fe 2 TiO 5 /C synthesized by solid-state reaction method
The DSC and TG curves of the Fe 2 O 3 and TiO 2 mixture, shown in Fig. 3a, exhibit two major peaks and one mass regions from 50 C to 1100 C, respectively. Fig. 3b displays the XRD patterns of samples calcined at varying temperatures by solid-state reaction method. The exothermic peak centered at 303 C can be ascribed to the combustion of residual organics in the mixtures, corresponding to 0.586% mass loss in Fig. 3a. The endothermic peak located at 879 C in the DSC curve is related with the phase transition, 28 owing to the rst occurrence of Fe 2 TiO 5 phase from 800 C to 900 C in the XRD patterns. The solid-state reaction is shown in formula (1). Moreover, it should be pointed out that some diffraction peaks indexed to hematite Fe 2 O 3 (JCPDS 33-0664) still exist at 1100 C as shown in Fig. 3b, which indicates the difficulty in obtaining pure phase Fe 2 TiO 5 for solid-state reaction method. In addtion, it undoubtedly  increases industrial preparation costs traced to the higher energy by solid-state reaction method. Ultimately, Fe 2 TiO 5 /C (S) is obtained by adding graphite into Fe 2 TiO 5 prepared by solidstate reaction method, as shown in the XRD pattens (Fig. 4a). In order to test the carbonaceous content in sample, thermogravimetry (TG) cuve was obtained at O 2 atmosphere, as shown in Fig. 4d. According to the mass of the sample tested and the nal weight loss in the test process, the carbonaceous content of Fe 2 TiO 5 /C (S) is calculated to be 8.23 wt%. NHSG method is a liquid-phase synthesis method in organic solvents which overcomes the disparity in the reaction rates of the different sol-gel precursors in the traditional hydrolytic solgel method (HSG). In addtion, NHSG route is easier to achieve uniform mixing at the atomic level, promoting the preparation of homogeneous mixed metal oxide system. 29 Especially, metal organic compounds and organic solvents including a plenty of carbon-containing groups are usually used as raw materials, providing the necessary carbon source for the synthesis of in situ carbon doped Fe 2 TiO 5 . There are usually three routes for the synthesis of heterometallic compounds by NHSG method. 30 (1) Alkyl chloride elimination route: where R represents C 2 H 2n+1 (alkyl) group.
(3) Ester elimination route:  where Z refers to an CH 3 CO (acetyl) group.
Here, in order to verify the feasibility of the above three routes in the synthesis of Fe 2 TiO 5 , FeCl 3 , FeOH(CH 3 COO) 2 and Fe(OEt) 3 were used as iron precursors, respectively. Fig. 5a presents the DTA-TG curves of xerogel using FeCl 3 as iron precursor. The rst two exothermic peaks at 299 C and 456 C are ascribed to the combustion of residual carbon and the transition of anatase to rutile, corresponding to a weight loss of 42.72%. The last exothermic peak centered at 889 C shows the reaction of Fe 2 O 3 and TiO 2 to form Fe 2 TiO 5 accompanied almost no mass loss, which is also cormed by the later XRD results. Fig. 5b shows the XRD patterns of samples calcined at different temperatures using FeCl 3 as iron precursor by NHSG method. It can be seen that the weak Fe 2 TiO 5 diffraction peak appeared until 900 C. The phase transition process during calcination is shown below.  Fig. 6a presents the DTA-TG curves of xerogel using FeOH(CH 3 COO) 2 as iron precursor. A sharp exothermic peak in the DTA curve at 294 C is related with the carbonization of organic residues and the volatilization with burning of part residual organic group, 27 corresponding to a mass loss of 35.28% in the TG curve. In addtion, the phase transition from amorphous Fe 2 TiO 5 to crystallized Fe 2 TiO 5 contributes to the small exotherm at 519 C (shown in formula (4)), which is supported by XRD results in Fig. 6b.  Fig. 7 shows DTA-TG curves of xerogel and XRD patterns of samples calcined at varying temperatures via NHSG method using Fe(OEt) 3 as iron precursor. In the rst stage, a 17.05% weight decrease due to with the carbonization of organic residues and the volatilization with burning of part residual organic group, corresponding to an exothermic reaction at 216 C. In the second stage between 500 C and 600 C, an exothermic peak at 584 C originates from the formation of crystalline Fe 2 TiO 5 modication from the amorphous form. XRD analysis showed that Fe 2 TiO 5 crystallization started at lower temperatures. The whole reaction process during heatment can be summarized in formula (5).

Nonhydrolytic polycondensation intermediate
Furthermore, Fourier transform infrared spectrometer (FTIR spectrometer) was used to identify the chemical bonds and functional groups of xerogels with different iron precursors. From the FTIR spectrum of I-FeCl 3 , the absorption peak appear at the range of 500-600 cm À1 , belonging to Ti-O-Ti stretching vibrations. 31,32 At the same time, the absence of Fe-O-Ti bond (629 cm À1 ) suggests the failure of reaction route (1), which is because that FeCl 3 , as Lewis acid, can promote homogeneous polycondensation to form Ti-O-Ti linkages along with the elimination of C 4 H 9 OC 4 H 9 , as shown in Fig. 8. 33 At the same time, the characteristic peak of ether appears at 1100 cm À1 verify the existence of C 4 H 9 OC 4 H 9 . 34 Moreover, the FTIR spectrum of xerogel using FeOH(CH 3 COO) 2 as iron precursor is shown in Fig. 8-II. The FTIR spectrum of II-FeOH(CH 3 COO) 2 showed two obvious absorption peaks at 1428 cm À1 and 1546 cm À1 , which is attributed to the symmetric stretching vibration of and the anti-symmetric stretching vibration of the O-C-O bond, respectively. 35 In addition, it is worth noting that Fe-O-Ti bond appears in the III-FeOH(CH 3 COO) 2 , 36 which demonstrates the occurrence of heterogeneous polycondensation in gel using FeOH(CH 3 COO) 2 as iron precursor, as shown in reaction eqn (7). Meanwhile, in the FTIR spectrum of II-Fe(OEt) 3 , a weak absorption peak located at 629 cm À1 is ascribed to the stretching vibration of Fe-O-Ti bond, indicating the occurrence of reaction eqn (8). The peak at 1100 cm À1 shows stretching vibrations of C-O-C bond in C 4 H 9 OC 2 H 5 . 37 The 1525 cm À1 band is attributed to C-H vibrational modes. 38 The existence of C-H bonds indicates that the xerogel contains some residual organic functional groups, on account of the incomplete polycondensation substitution reaction. 39 It is helpful for the formation of in situ doped C during post-heating treatment (Fig. 9). In conclusion, Fe 2 TiO 5 can be synthesized at the lower temperature (800 C) using FeOH(CH 3 COO) 2 and Fe(OEt) 3 as iron precursors on the account of uniform mixing at the atomic level and the formation of Fe-O-Ti linkages. However, the C element mass fractions of FeOH(CH 3 COO) 2 and Fe(OEt) 3 were calculated to be 25.14% and 37.70% accroding to the formula (8), thus, Fe(OEt) 3 containing more carbon-based groups can provide more in situ carbon for the synthesis of Fe 2 TiO 5 /C. In the following, Fe 2 TiO 5 /C prepared with Fe(OEt) 3 as the iron precursor via NHSG method was compared with the Fe 2 TiO 5 /C prepared by the solid-state reaction method.
where u refers to element mass fraction, A r , M r and N stand for the relative atomic mass, relative molecular mass and atomic number, respectively.   This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 43762-43772 | 43767 Paper RSC Advances

Microstructure analysis
Fig. 10a-c show the SEM images of Fe 2 TiO 5 /C prepared by solid-phase method with different magnications. It can be seen that the sample has irregular bulk distribution morphology and the two-dimensional graphite sheets as carbon source only were only stacked in local area in SEM image. In order to evaluate the uniformity of elemental distribution, EDS elemental mapping analysis of Fe 2 TiO 5 /C prepared by solid-state reaction method is presented in Fig. 10d. The mapping result shows that the C element is not evenly distributed in sample, which matches well with the corresponding SEM image. Fig. 10e and f display the SEM image and elemental mapping of Fe 2 TiO 5 /C (N) , respectively. Fe 2 TiO 5 /C (N) composite is composed of a large number of hollow spherical particles. C, Fe, O, Ti elements disperse evenly in the sample, indicating the successful incorporation of C element into Fe 2 TiO 5 hollow spherical structure. In addition, the hollow spherical structure permits light to be reected multiple times inside the hollow spheres, thereby promoting the utilization of the incident light. 40 Form and content of in situ carbon introduced by NHSG Fig. 11a shows the Raman spectroscopy of Fe 2 TiO 5 /C (N) . From 100-1000 cm À1 , several broad bands could be attributed to Fe 2 TiO 5 (154, 303, 407, 564, 731 cm À1 ). The band at 1384 cm À1 is associated with the vibrations of carbon atoms with dangling bonds for the in-plane terminations of disordered graphite and is labeled as the D-band, and the band at 1589 cm-1 (G-band) (corresponding to the E 2g mode) was closely related to the vibration in all sp 2 bonded carbon atoms in a 2-dimensional hexagonal lattice. The intensity ratio of the D to G band (I D /I G ) is calculated to be 1.457 according to the integral area values of D and G peaks obtained by peak tting, indicating the in situ carbons contain a lot of defects. Defects are easier to capture photogenerated electrons, thereby inhibiting the recombination of photogenerated electrons/holes. 41 According to TG curve in Fig. 11b, the carbonaceous content was calculated into 8.17 wt%, which is very close to carbonaceous content of Fe 2 TiO 5 /C (s) (8.23 wt%).

Specic surface area analysis
It is known to all that the smaller specic surface area (SSA BET ) is regarded as the other limitations in semiconductor photocatalysis because the photocatalysis occurs at the surface of photocatalyst. Sample with larger specic surface area can offer more active sites, thereby promoting the photocatalytic reaction. The SSA BET of Fe 2 TiO 5 /C (S) and Fe 2 TiO 5 /C (N) is calculated to be 3.5100 m 2 g À1 and 195.5153 m 2 g À1 , respectively. The SSA BET value of the latter is almost 56 times that of the former, indicating Fe 2 TiO 5 /C (N) is more conducive to facilitating the masstransport and charge-transfer. 42 Photocatalytic activity analysis However, a little change in the UV absorption spectra in the upper right corner in Fig. 12 and the inconspicuous enhancement from 7.902% to 8.871%, reveals that it is unsatisfactory to promote the catalytic performance of samples by adding additonal carbon. Consequently, the uniform in situ carbon is introduced into Fe 2 TiO 5 by containg-carbon precursors via NHSG method for the purpose of improving photocatalytic activity. The Fe 2 TiO 5 /C (N) photocatalyst has the higher photodegradation rate (88.2%) in comparison with the Fe 2 TiO 5 / C (S) photocatalyst (8.871%) exposed to visible light for 30 min in Fig. 12b. And the dye can be almost completely decomposed by the Fe 2 TiO 5 /C (N) under the visible light irradiation within 70 min, indicating that the photocatalyst has excellent visible light photocatalytic performance.

Separation of photoexcited electrons and holes
Photocurrent response analysis has been widely recognized as an effective means to evaluate the separation ability of photogenerated carriers. When light energy is used to excite the material, the valence band electrons are excited and transition to the conduction band. Under the strong electric eld, the conduction band electrons will move directionally to form a current, that is, a photo-generated current. As shown in Fig. 13a, the photocurrent-time curves (I-T curves) were performed for several on-off cycles of visible-light irradiation (l > 420 nm). The photocurrent of Fe 2 TiO 5 /C (N) electrode is signicantly higher than that of Fe 2 TiO 5 /C (S) . In general, when the optical radiation energy is absorbed by the semiconductor material to generate a photocurrent, a higher photocurrent response indicates better charge separation performance. 43 Therefore, the NHSG method is more conducive to promoting the separation of photo-generated carriers, owing to the excellent electron transportation ability of the uniform in situ carbon introduced by NHSG, which is in good agreement with the photocatalytic activity test results shown in Fig. 12. Electrochemical Impedance Spectroscopy (EIS) offers another strong evidence for exploration of interfacial charge transfer behaviors. During the electrochemical impedance test  This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 43762-43772 | 43769 of photocatalytic materials, the electrochemical impedance spectroscopy (EIS) obtained generally consists of an "arc" and a "tail", as shown Fig. 13b. The "arc" in the high-frequency and low-resistance region is mainly dominated by the charge transfer. The "tail" in the low-frequency high-resistance region is mainly controlled by the mass transfer. Therefore, R ct (charge transfer resistance) is generally determined by the arc radius. It is clearly shown that the arc radius on EIS Nyquist plot of Fe 2 TiO 5 /C (N) with in situ carbon is much smaller than that of Fe 2 TiO 5 /C (S) prepared by adding extra carbon, which indicates the Fe 2 TiO 5 /C (N) presents higher charge separation rate. The smaller arc radius corresponds to the lower R ct value and weaker impedance facilitating the separation of photoinduced electron-hole pairs, which is consistence with the results of photocurrent response analysis. 44

Conclusion
This work makes a systematical comparison of Fe 2 TiO 5 /C photocatalysts synthesized by NHSG method and solid-state reaction method. In conclusion, the Fe 2 TiO 5 /C (N) with in situ carbon as carbon source has higher catalytic efficiency under visiblelight irradiation than that of Fe 2 TiO 5 /C (S) with the extra carbon. The in situ carbon plays the important role in enhancing the visible-light catalytic ability. Most in situ carbon evenly distributed in the sample which greatly facilitates electron transport and promotes the separation of photogenerated electrons and holes. Photoelectrochemical measurements have been utilized to evaluate the separation ability of photogenerated carriers. Compared with Fe 2 TiO 5 /C (s) with addition carbon, Fe 2 TiO 5 /C (N) with in situ carbon has more photocurrent and weaker resistance, indicating the effective separation of photogenerated electrons and holes in the photocatalytic system of Fe 2 TiO 5 /C (N) .

Conflicts of interest
There are no conicts to declare.