Synergistic Effect of Dual Electron-Cocatalysts for Enhanced Photocatalytic Activity: rGO as Electron-Transfer Mediator and Fe(III) as Oxygen-Reduction Active Site

For a high-performance cocatalyst-modified photocatalyst, an effective interfacial separation of photogenerated electron from its corresponding holes and its following reduction reaction at the active sites are highly required. However, it is difficult for a single-component cocatalyst to simultaneously realize the crucial functions. In this study, an effective interfacial transfer of photogenerated electrons and its following rapid oxygen-reduction can be easily realized in a dual electron-cocatalyst modified Fe(III)/rGO-TiO2 photocatalyst, where the rGO nanosheets function as an electron-transfer mediator for the effective transfer of photogenerated electrons from the TiO2 surface while the Fe(III) cocatalyst serves as an electron-reduction active site to promote the following interfacial oxygen reduction. In this case, the rGO nanosheets were firstly loaded on the TiO2 nanoparticle surface by a hydrothermal method and then the Fe(III) cocatalyst was further modified on the rGO nanosheets by an impregnation method to prepare the Fe(III)/rGO-TiO2 photocatalyst. It was found that the dual electron-cocatalyst modified Fe(III)/rGO-TiO2 photocatalyst showed an obviously higher photocatalytic performance than the naked TiO2 and single-cocatalyst modified photocatalysts (such as Fe(III)/TiO2 and rGO-TiO2) owing to the synergistic effect of rGO and Fe(III) bi-cocatalysts. The present work can provide some new insights for the smart design of high-efficiency photocatalytic materials.

for the water splitting). As a consequence, it is very clear that a large amount of interfacial contacting sites with a strongly coupling interface between cocatalyst and photocatalytic materials, and the cocatalyst with a high electron mobility and rapid electron-reduction reaction should be highly required for a high-performance cocatalyst-modified photocatalysts 8,9 . Unfortunately, for a single-component cocatalyst, it is quite difficult to simultaneously realize the above vital functions. For examples, the interfacial contacting area between noble metal nanoparticles (such as Pt, Au, and Ag) and photocatalysts is very limited, while graphene/reduced graphene oxide (rGO) cocatalyst usually shows an obviously lower interfacial reduction rate than the well-known noble metal nanoparticles (Indeed, some researchers have attempted to load some noble metals as the active reaction sites on the graphene surface to improve the interfacial catalytic performance 10,11 ). Therefore, it is highly desirable to further develop new strategy for the optimization of cocatalysts to improve the performance of photocatalytic materials. In addition, considering the expensive cost of various noble metal cocatalysts, it is quite interesting and worthwhile to explore new and cost-effective cocatalysts for the enhanced photocatalytic performance of cocatalyst-modified photocatalysts.
Graphene or rGO is a promising material in various potential applications due to its unique charge carrier mobility properties, large specific surface area and chemical stability [12][13][14][15][16] . In photocatalytic field, it has been widely demonstrated that the photocatalytic performance of various photocatalysts can be greatly improved by the addition of graphene nanosheets [17][18][19][20][21][22][23] . The principal reason is that the graphene or rGO serves as an electron mediator to accelerate the separation efficiency of photogenerated electrons and holes due to its excellent electron mobility. In our recent results, we also found that the rGO nanosheets could be loaded on the TiO 2 surface with a strongly coupling interface by a hydrothermal method, causing an obvious enhancement of its photocatalytic activity 24,25 . In addition, the transition metal Fe(III) and Cu(II) has been widely demonstrated to be a highly efficient electron cocatalyst for the interfacial oxygen reduction during photocatalytic decomposition of organic pollutants [26][27][28][29][30][31][32][33][34][35] . Considering their remarkably different advantages of rGO nanosheets and Fe(III) cocatalyst, it is expected that the photocatalytic performance of TiO 2 can be further improved by the simultaneous modification of rGO nanosheets and Fe(III) cocatalysts.
In this study, a dual electron-cocatalyst modified TiO 2 photocatalyst (Fe(III)/rGO-TiO 2 ) with a high photocatalytic performance was developed by a facile two-step route including the initial synthesis of rGO-TiO 2 composite and the following surface loading of Fe(III) cocatalyst on the rGO nanosheets. In this case of Fe(III)/rGO-TiO 2 photocatalyst, the dual electron cocatalysts of rGO and Fe(III) exhibit two different functions during photocatalytic reaction, namely, the rGO nanosheets as an electron mediator cause the steady capture and rapid transportation of photogenerated electrons from the conduction band (CB) of TiO 2 , while the Fe(III) serves as an effective oxygen-reduction active site for the following interfacial reduction reaction of oxygen. In view of a high specific surface area of rGO nanosheets, the Fe(III) cocatalyst was easily loaded on the rGO surface in the rGO-TiO 2 composite. On the basis of the experimental results, a synergistic effect mechanism of rGO and Fe(III) bi-cocatalysts was proposed to account for its improved photocatalytic performance. To the best of our knowledge, this is the first report about the enhanced photocatalytic performance of TiO 2 by a dual electron-cocatalyst modification of rGO nanosheets and Fe(III) ions. Considering a large-scale production of rGO from the graphite with a low cost, the abundant Fe and C elements in natural resources, and the facile and green synthetic route, the resulting Fe(III)/rGO-TiO 2 photocatalyst can be regarded as one of the most promising photocatalytic materials for the various potential applications. In addition, this work can provide some new insights for the smart design and preparation of inexpensive and high-efficiency photocatalytic materials.

Results
Various cocatalyst-modified photocatalysts such as TiO 2 , Fe(III)/TiO 2 , rGO-TiO 2 and Fe(III)/rGO-TiO 2 can be easily prepared by facile solution routes (Fig. 1). Firstly, rutile TiO 2 precursor (Fig. 1a) was obtained from P25 TiO 2 nanoparticles after high-temperature calcination which can produce a very clear and hydrophilic TiO 2 surface for the following rapid coupling with GO nanosheets. As a consequence, when the calcined TiO 2 particles were dispersed into the GO solution, a homogeneous and stable GO-TiO 2 suspension with a strongly coupling interface was formed immediately. After hydrothermal reduction of GO to rGO, the strongly coupling interface can be well maintained in the resultant rGO-TiO 2 composite (Fig. 1c), as shown in our recent results 24 . In fact, the GO nanosheets were in situ reduced to rGO on the TiO 2 surface via a hydrothermal deoxygenation/dehydration reaction. For the surface modification of Fe(III) cocatalyst on the TiO 2 and rGO-TiO 2 photocatalysts (Fig. 1b,d), a well-known impregnation method of Fe(III) clusters was performed at 80 °C with a mild condition. Considering a low-temperature loading route for the Fe(III) cocatalyst, it is quite believed that the Fe(III) cocatalyst is only loaded on the photocatalyst surface, but not doped into the lattices of TiO 2 and rGO nanosheets.
The controlled preparation of various cocatalyst-modified TiO 2 photocatalysts can firstly be demonstrated by FESEM, TEM and XRD results. Figure 2 shows the FESEM images of the TiO 2 , rGO-TiO 2 , Fe(III)/TiO 2 , and Fe(III)/rGO-TiO 2 . It can be seen that the TiO 2 sample is composed of irregular particles with a size range of 100-600 nm (Fig. 2a), suggesting an obvious growth of the TiO 2 particles during high-temperature calcination (P25 nanoparticles with a size of 20-50 nm). The corresponding XRD result suggests the successful phase transformation of rutile-anatase mixing phase to pure rutile ( Figure S1). After the rutile TiO 2 surface is modified with Fe(III) cocatalyst, the resulting Fe(III)/TiO 2 (Fig. 2c shows a similar morphology as the TiO 2 sample due to a very low amount of Fe(III) cocatalyst (Fe/ TiO 2 = 0.05 wt%). As for the rGO-TiO 2 (Fig. 2b) and Fe(III)/rGO-TiO 2 ( Fig. 2d) samples, they also show very similar particle morphologies with the pure TiO 2 . In addition, it was found that the rGO nanosheets were tightly covered on the TiO 2 particle surface (shown in the arrow), indicating a well interfacial interaction of rGO nanosheets with the TiO 2 nanoparticles, in good agreement with our strategy (Fig. 1). The corresponding TEM images of Fe(III)/rGO-TiO 2 further demonstrated the well coupling of rGO with the TiO 2 particles, as shown in Fig. 2e,f. Moreover, the Fe(III) clusters with a very small size (less than 1 nm) are homogeneously dispersed on the surface of rGO nanosheets (showed by the red arrows in Fig. 2f), suggesting that the impregnation method is an excellent strategy for the loading of Fe(III) cocatalyst. In addition, on the basis of XRD results ( Figure S1), it is found that all the cocatalyst-modified TiO 2 photocatalysts show a comparable diffraction-peak intensity and full width at half-maximum compared with the pure TiO 2 sample, revealing that the crystallization and crystallite size of TiO 2 photocatalyst are not affected by the cocatalyst modification processes owing to their mild conditions. According to the EDX result, the Fe element in the Fe(III)/rGO-TiO 2 composite (Fig. 2d) is calculated to be 0.1 wt%. However, it should be noted that the Fe(III) clusters and its corresponding diffraction peaks cannot be observed in the FESEM image and XRD pattern, respectively, owing to its very limited amount. Considering the well coupling interface of rGO nanosheets and TiO 2 nanoparticles (Fig. 2b,d), it is expected that the photogenerated electrons can be steadily transferred from TiO 2 particles to the rGO nanosheets.
FTIR and Raman spectra can further demonstrate the formation of Fe(III)/rGO-TiO 2 photocatalyst. Figure 3A shows the FTIR spectra of various samples. It is clear that the GO shows many strong absorption peaks corresponding to various oxygen functional groups (C ‖ O, C-OH, C-O-C, C-O-H and C-O) 10,36 . After hydrothermal treatment of the GO, the intensity of all absorption peaks corresponding to oxygen-containing groups has a significant decrease, demonstrating that the hydrothermal treatment is an effective strategy for the successful reduction of GO to rGO. For the rutile TiO 2 nanoparticles, the wide absorption peak at 400-900 cm −1 is attributed to the stretching vibration of Ti-O-Ti bonds in crystalline TiO 2 37 , while the corresponding absorption in the rGO-TiO 2 and Fe(III)/rGO-TiO 2 composites can be attributed to the stretching vibration of Ti-O-Ti and the possible Ti-O-C bonds formed during hydrothermal reaction. Figure 3B shows the Raman spectra of various samples. It is clear that rutile TiO 2 shows strong Raman characteristic peaks at 145 cm −1 (B 1g ), 448 cm −1 (E g ), 613 cm −1 (A 1g ) and 238 cm −1 for second order effect 38 . In addition, the Raman spectra in the inset of Fig. 3B show the characteristic D band at 1347 cm -1 and G band at 1604 cm -1 in the GO, rGO and rGO-modified TiO 2 photocatalysts. It is well reported that the intensity ratio of the D band to the G band usually measures the defects/disorders in GO or graphene, and a small intensity ratio of I D /I G can be assigned to less sp 3 defects/disorders and larger average size of the in-plane graphitic crystallite sp 2 domains 36,39 . Compared with the GO (0.911), the Fe(III)/rGO-TiO 2 composite shows a higher I D /I G ratio (1.099), indicating that the rGO in Fe(III)/rGO-TiO 2 composite contains more sp 3 defects. The possible reason for the increased sp 3 defects can be attributed to the formation of strong interfacial interaction (such as the Ti-O-C bond) between the TiO 2 nanoparticles and rGO nanosheets 39 .
XPS technology can provide further information about the formation of Fe(III)/rGO-TiO 2 photocatalyst, as shown in Fig. 3C, D. After hydrothermal reaction of GO nanosheets, the chemical state of carbon atoms in the hexagonal lattice of graphene was changed significantly owing to the deoxygenation reaction, which can be well illustrated in Fig. 3C. The C1s XPS spectrum of the GO nanosheets clearly shows the presence of four types of carbon bonds, namely, the non-oxygenated ring carbons including C-C, C-H, and C‖C (284.8 eV), the C-O groups (286.8 eV), the carbonyl carbon in C‖O (288.0 eV) and the carboxylate carbon in O‖C-OH (288.6 eV). After hydrothermal treatment of the GO, the XPS peak intensity of these carbon-oxygen species in the resultant rGO shows an obvious decrease,  suggesting the effective reduction of GO 40,41 . Considering an identical hydrothermal condition with the rGO nanosheets, it is quite reasonable that the rGO in rGO-TiO 2 and Fe(III)/rGO-TiO 2 system shows a similar surface microstructures. Figure 3D shows the high-resolution XPS spectra of Fe 2p. Compared with the pure TiO 2 samples, new XPS peaks of Fe element at ca. 711.5 eV, which corresponds to the binding energy of Fe 2p 3/2 for the ferric ion 27,42 , are found in the Fe(III)/TiO 2 and Fe(III)/rGO-TiO 2 samples. According to the element component analysis based on the XPS results (Table 1), it is found that the amount of Fe(III) cocatalyst in the Fe(III)/TiO 2 and Fe(III)/rGO-TiO 2 photocatalysts is about 1.17 and 5.38 at.%, respectively. However, according to the atomic absorption spectrometry analysis, the Fe amount in Fe(III)/rGO-TiO 2 was about 0.44 wt%, which suggests that most of the Fe element has been loaded on the surface of graphene nanosheets in rGO-TiO 2 composite. Therefore, the above experimental results strongly suggested that the rGO and Fe(III) cocatalyst has successfully been modified on the surface of TiO 2 photocatalyst. Figure 4 shows the UV-vis spectra of TiO 2 , rGO-TiO 2 , Fe(III)/TiO 2 and Fe(III)/rGO-TiO 2 samples. It is clear that the Fe(III)/TiO 2 shows a similar UV-vis spectrum as the pure TiO 2 owing to a very limited Fe(III) cocatalyst (Fig. 4a,c). After the rGO was loaded on the TiO 2 surface, the absorption edge of rGO-TiO 2 composite (Fig. 4b) shows a slight red-shift to a higher wavelength, and an enhanced visible-light absorption in the range of 400-800 nm can be observed owing to the presence of black rGO nanosheets. When both of the rGO nanosheets and Fe(III) cocatalyst are simultaneously deposited on the TiO 2 surface to form Fe(III)/rGO-TiO 2 photocatalyst (Fig. 4d), it is found that the visible-light absorption in the range of 400-600 nm can be further improved. The enhanced absorption can be attributed to the d-d transition of Fe(III) cocatalyst, clearly revealing a higher amount of Fe(III) cocatalyst in the Fe(III)/rGO-TiO 2 than that in the Fe(III)/TiO 2 (Fig. 4c) owing to the presence of rGO nanosheets, in good agreement with the XPS results. In fact, the colour change from grey colour of rGO-TiO 2 to

Discussion
The photocatalytic performances of TiO 2 , rGO-TiO 2 , Fe(III)/TiO 2 and Fe(III)/rGO-TiO 2 samples were first evaluated by photocatalytic decolorization of MO aqueous solution (Fig. 5). In the dark, no change in the concentration of MO was observed in the presence of different photocatalysts. Furthermore, visible-light illumination in the absence of photocatalysts did not result in the photocatalytic decolorization of MO. Figure 5 shows the corresponding photocatalytic rate constant k of different photocatalysts. For rutile TiO 2 , it exhibits a relative low photocatalytic activity and the k value is about 0.002 min −1 . When the Fe(III) and rGO cocatalysts are grafted on the TiO 2 surface, respectively, both of the resultant Fe(III)/ TiO 2 and rGO-TiO 2 show an obviously improved photocatalytic performance with a k value of 0.007 and 0.009 min −1 , respectively. More specifically, the Fe(III)/rGO-TiO 2 photocatalyst possesses the highest photocatalytic activity (k = 0.013 min −1 ). In fact, the amount of Fe(III) on the rGO surface has a great effect on the photocatalytic performance of Fe(III)/rGO-TiO 2 photocatalyst, and the optimized amount of Fe(III) cocatalyst was about 0.5 wt% ( Figure S2), which is obviously higher than that in the Fe(III)/ TiO 2 photocatalyst (0.05 wt%). To further investigate the photocatalytic ability of cocatalyst-modified TiO 2 , the dimethyl phthalate (DMP) solution was also used to evaluate their photocatalytic performance. It is found that the Fe(III)/rGO-TiO 2 photocatalyst can maintain a stable and efficient photocatalytic performance ( Figure S3).
It is very interesting and worthwhile to investigate the potential photocatalytic mechanism of Fe(III)/ rGO-TiO 2 photocatalyst. The rutile TiO 2 photocatalyst exhibits obvious photocatalytic activity for the decomposition of organic pollutions owing to its strong oxidation power of photogenerated holes (+ 2.63 V, vs. SHE, pH = 7) and suitable reduction potential of photogenerated electrons (− 0.37 V, vs. SHE, pH = 7) ( Figure S4a) 43,44 . However, the photogenerated electrons and holes are easily recombined in the single-component rutile TiO 2 , resulting a low photocatalytic efficiency. When the Fe(III) ions are grafted onto the TiO 2 surface to form Fe(III)/TiO 2 photocatalyst, the photogenerated electrons on the TiO 2 CB can transfer to the Fe(III) cocatalyst to form Fe(II) owing to its more positive potential  of Fe 3+ /Fe 2+ (0.771 V, vs. SHE) 45 than the CB of TiO 2 , which can promote the separation efficiency of photo-generated electrons and holes in bulk TiO 2 ( Figure S4c). More importantly, the as-formed Fe(II) cocatalyst can reduce oxygen rapidly by a possible multi-electron reduction mechanism (4Fe 2+ 45 , causing a significant enhancement of photocatalytic performance. Compared with the Fe(III)/TiO 2 photocatalyst, however, the rGO-TiO 2 shows a completely different improved mechanism ( Figure S4b). In this case, the well coupling interface and a large number of interfacial contacting sites between the rGO nanosheets and TiO 2 particles provide an excellent platform for the rapid and efficient transfer of photogenerated electrons from the TiO 2 CB. As a consequence, the rGO nanosheets work as an effective electron mediator for the rapid capture and steady transportation of photogenerated electrons from the TiO 2 surface, resulting in a lower recombination rate of photogenerated charges in the bulk TiO 2 and a higher photocatalytic activity. When both of the rGO nanosheets and Fe(III) cocatalyst are simultaneously deposited on the surface of TiO 2 , the further improved photocatalytic performance of Fe(III)/rGO-TiO 2 photocatalyst can be well explained by their synergistic effect, as shown in Figure S4d and Fig. 6. In the presence of rGO, the photogenerated electrons of TiO 2 can first transfer to the capture sites of rGO nanosheets (Step (1) in Fig. 6). In this case, the rGO works as an electron sink to rapidly capture the photogenerated electron from the TiO 2 surface, while the photogenerated holes are still remained on the TiO 2 bulk to oxidize the organic substances, resulting in an effective separation of photogenerated charges. Considering a large amount of  interfacial contacting sites with a strongly coupling interface between rGO nanosheets and TiO 2 nanoparticles (Figs 1 and 2), it can be deduced that the rGO has a higher separation rate for the photogenerated charges than the well-known nanoparticle-like cocatalyst such as the Fe(III) cocatalyst in the Fe(III)/TiO 2 . After the capture of photogenerated electrons, the rGO nanosheets then function as an electron mediator to transport the captured electrons to the oxygen-reduction active sites owing to its excellent carrier mobility property (Step (2) in Fig. 6). Finally, considering that rGO nanosheets can serve as an excellent scaffold to homogeneously anchor the Fe(III) cocatalyst, the highly dispersed Fe(III) clusters serve as the effective oxygen-reduction active sites for the highly efficient reduction of oxygen (Step (3) in Fig. 6). Thus, it is very clear that the excellent synergetic effect between Fe(III) and rGO cocatalysts greatly promote the effective separation of photogenerated charges, the steady transportation of photogenerated electrons and its following oxygen reduction, resulting in the improved photocatalytic performance of Fe(III)/rGO-TiO 2 composites. In fact, in addition to the rutile TiO 2 , the synergetic effect between Fe(III) and rGO cocatalysts can also be demonstrated in the well-known P25 nanoparticles (a S BET of ca. 50 m 2 /g, 80% anatase and 20% rutile) under UV-light irradiation ( Figure S5), where the UV-light photocatalytic performance of Fe(III)/rGO-P25 is obviously higher than that of rGO-P25 and P25.
To further demonstrate the synergistic effect of Fe(III) and rGO cocatalysts for the enhanced photocatalytic performance of Fe(III)/rGO-TiO 2 , a series of controlled experiments are performed (Fig. 7A). Firstly, Fe(III) cocatalyst on the surface of Fe(III)/rGO-TiO 2 photocatalyst are selectively removed by HCl solution (0.1 mol/L), which can be clearly demonstrated by its colour change from greyish yellow to grey (Fig. 7B). In this case, the corresponding k value of the resultant rGO-TiO 2 sample decreases significantly from 0.013 to 0.008 min −1 (Fig. 5e), a value comparable to that of the as-prepared rGO-TiO 2 photocatalyst (Fig. 5b). With further selective removal of rGO by high-temperature calcination in the air (500 °C), the resulting TiO 2 sample shows a further decreased decomposition performance (k = 0.003 min −1 ) (Fig. 5f) and its colour returns to white completely (Fig. 7B). Therefore, the above controlled experimental results strongly support that the synergistic effect of Fe(III) and rGO bi-cocatalysts is the main reason for the enhanced photocatalytic activity of Fe(III)/rGO-TiO 2 .
In summary, a new dual electron-cocatalyst modified Fe(III)/rGO-TiO 2 photocatalyst was successfully prepared by a facile two-step route including the initial synthesis of rGO-TiO 2 composite and the following surface loading of Fe(III) cocatalyst on the rGO nanosheets. It was found that the Fe(III)/ rGO-TiO 2 photocatalyst showed an obviously higher photocatalytic performance than the naked TiO 2 and single-cocatalyst modified photocatalysts (such as Fe(III)/TiO 2 and rGO-TiO 2 ). On the basis of the experimental results, a synergistic effect mechanism of rGO and Fe(III) bi-cocatalysts was proposed to account for its improved photocatalytic performance, namely, the rGO nanosheets function as an electron-transfer mediator to rapidly capture and transfer the photogenerated electron from TiO 2 surface, while the Fe(III) cocatalyst serves as an effectively active site for the following oxygen reduction. The present work will provide new insight for the smart design of new cocatalyst-modified photocatalytic materials with a high performance.

Methods
Chemicals. Commercial Degussa TiO 2 nanoparticles (P25 TiO 2 , a S BET of ca. 50 m 2 /g, 80% anatase and 20% rutile) was pre-treated at 900 o C for 3 h, and the resultant rutile TiO 2 was used as the TiO 2 precursor in this study. Graphene oxide (GO) was synthesized from natural graphite powder (99.95%) by a modified Hummer's method shown in our previous studies, and the brown GO (0.2 wt%) solution was obtained by ultrasonic dispersion of GO (0.5 g) in deionized water (250 mL) for 1 h. All the other reagents (analytical grade) were supplied by Shanghai Chemical Ltd. (P.R. China) and used as received without further purification.
Preparation of rGO-TiO 2 composite. The rGO-TiO 2 composite was obtained via one-step hydrothermal method. According to our results, for the rGO-TiO 2 photocatalyst by using rutile TiO 2 as the precursor, it was found that when the rGO amount (rGO/TiO 2 ) was 0.1 wt%, the prepared rGO-TiO 2 showed the highest photocatalytic performance. Therefore, in this case, the amount of rGO in the rGO-TiO 2 composite was controlled to be 0.1 wt%. For a typical synthesis, 1.6 g of TiO 2 was first dispersed into 15 mL deionized water, and then 0.8 mL of GO solution was added into the above mixing solution. After strong stirring for 2 h, the resultant homogeneous suspension was transformed into Teflon-sealed autoclave and then maintained at 160 o C for 5 h. After being cooled to room temperature naturally, the resulting composite was washed with distilled water for several times, and dried in vacuum at 60 o C for 8 h to obtain the rGO-TiO 2 composites. For comparison, the GO solution was also hydrothermally treated under an identical condition, and the resulting black sample was referred to as rGO.
Preparation of Fe(III)/TiO 2 . The Fe(III)/TiO 2 photocatalyst was synthesized by an impregnation technique similar to our previous studies. According to our experimental results, it was found that when the amount of Fe(III) cocatalyst (compared to TiO 2 ) was controlled to be 0.05 wt%, the resultant Fe(III)/TiO 2 showed the highest photocatalytic activity. Therefore, in this study, the Fe(III)/TiO 2 (Fe(III)/ TiO 2 = 0.05 wt%) was used as the reference sample and was referred to be Fe(III)/TiO 2 . For a typical synthesis, 0.4 g of TiO 2 was dispersed into 10 mL of Fe(NO 3 ) 3 solution (0.36 μ mol/L, pH = 2) under stirring and was then maintained at 80 o C for 1 h. The resulting composite was filtrated, rinsed with distilled water, and dried at 60 o C for 8 h to obtain the Fe(III)/TiO 2 sample.
Preparation of Fe(III)/rGO-TiO 2 . The Fe(III)/rGO-TiO 2 photocatalyst was prepared under an identical experimental condition with the Fe(III)/TiO 2 photocatalyst by using rGO-TiO 2 as the precursor. In addition, the amount of Fe(III) in the Fe(III)/rGO-TiO 2 composites was controlled to be 0.05, 0.1, 0.2, 0.5 and 1.0 wt%, and the resulting sample can be referred to as Fe(III)/rGO-TiO 2 (X), where X represents 0.05, 0.1, 0.2, 0.5, and 1.0, respectively. According to the results by atomic absorption spectrometry, when the amount of Fe(III) precursor was controlled to be 0.05, 0.1, 0.2, 0.5 and 1.0 wt%, and the real amount of Fe(III) in the resultant Fe(III)/rGO-TiO 2 composites was 0.048, 0.093, 0.14, 0.33 and 0.69 wt%, respectively. However, as showed in this study ( Figure S2), when the X was 0.5, the prepared Fe(III)/rGO-TiO 2 photocatalyst showed the highest photocatalytic performance. Thus, the Fe(III)/rGO-TiO 2 (0.5) was used as the reference sample and was referred to be Fe(III)/rGO-TiO 2 in this work.
The selective removal of Fe(III) and rGO. For the selective removal of Fe(III) and rGO cocatalysts from the TiO 2 surface, the Fe(III)/rGO-TiO 2 sample was firstly dispersed into a HCl solution (0.1 mol/L) and then was calcined at 500 o C for 2 h in the air.
Characterization. X-ray diffraction (XRD) patterns were obtained on a D/MAXRBX-ray diffractometer (Rigaku, Japan). Morphological analysis was performed with an S-4800 field emission scanning electron microscope (FESEM) (Hitachi, Japan) with an acceleration voltage of 10 kV. TEM/HRTEM image was conducted using a JEM-2100F transmitting electron microscope. Raman spectra were collected using an INVIA spectrophotometer (Renishaw, UK). Fourier Transform Infrared spectra (FTIR) were acquired using a Nexus FT-IR spectrophotometer (Thermo Nicolet, America). X-ray photoelectron spectroscopy (XPS) measurements were done on a KRATOA XSAM800 XPS system with Mg Kα source. All the binding energies were referenced to the C1 s peak at 284.8 eV for the surface adventitious carbon. The amount of Fe(III) in Fe(III)/rGO-TiO 2 was performed on an atomic absorption spectrometry (GBC AVANTA-M, Australia). UV-vis absorption spectra were obtained using a UV-visible spectrophotometer (UV-1240, SHIMADZU, Japan).
Photocatalytic activity. The evaluation of photocatalytic activity of the prepared samples for the photocatalytic decomposition of methyl orange (MO) and dimethyl phthalate (DMP) solutions was evaluated at ambient temperature. Experimental details were shown as follows: 0.1 g of the sample was dispersed into 10 mL of MO solution (20 mg/L) or DMP solution (10 mg/L) in a disk with a diameter of ca. 5 cm. The solution was allowed to reach an adsorption-desorption equilibrium among the photocatalyst, MO (or DMP) and water before irradiation. A 350-W xenon lamp equipped with a UV-cutoff filter (providing visible light > 400 nm wavelength) was used as a visible-light source. The average light intensity striking the liquid surface of the reaction solution was about 40 mW/cm 2 . The concentration of the MO (or DMP) was determined by an UV-visible spectrophotometer (UV-1240, SHIMADZU, Japan). As for the MO (or DMP) aqueous solution with low concentration, its photocatalytic decolorization is a pseudo-first-order reaction and its kinetics may be expressed as ln(c/c 0 ) = − kt, where k is the apparent rate constant, and c 0 and c are the MO (or DMP) concentrations at initial state and after irradiation for t min, respectively. For the repeated photocatalytic performance, the photocatalysts were first separated by centrifugation, washed with distilled water, and were then re-dispersed into the MO (or DMP) solutions.