A New Insight of Graphene oxide-Fe(III) Complex Photochemical Behaviors under Visible Light Irradiation

Graphene oxide (GO) contains not only aromatic carbon lattice but also carboxyl groups which enhanced the aqueous solubility of GO. To study the transformation of GO nanosheets in natural environments, GO aqueous dispersion was mixed with Fe3+ ions to form photoactive complex. Under visible light irradiation, Fe(III) of the complex would be reduced to Fe(II) which could subsequently reduce highly toxic Cr(VI) to Cr3+. The electron of the reduction was contributed by the decarboxylation of carboxyl groups on GO and iron was acting as a catalyst during the photoreduction. On the other hand, the consumption of carboxyl groups may convert GO to rGO which are tend to aggregate since the decreased electrostatic repulsion and the increased π-π attraction. The formed Cr3+ may be electrostatically adsorbed by the rGO sheets and simultaneously precipitated with the aggregated rGO sheets, resulting the effective removal of chromium and GO nanosheets from the aqueous environment. This study may shed a light on understanding the environmental transformation of GO and guide the treatment of Cr(VI).


Transformation of Cr(VI) with the Existence of GO-Fe(III) under Visible Illumination.
shows the time profiles of Cr(VI) reduction under visible light irradiation and dark control with 40 mg/L GO and GO-Fe(III) suspensions at pH 3. After 60 minutes irradiation, a small extent of Cr(VI) reduction occurred in the presence of GO, whereas a nearly 100% conversion rate was obtained in the presence of GO-Fe(III). In a contrast, after stirring for 30 min in the dark, the concentrations of Cr(VI) in the solution were about 35 μ M and 39 μ M in the system of GO-Fe(III) and GO suspensions, respectively, and then kept constant in the next 120 min under dark condition (see Supplementary Figure S5a), which indicated that Cr(VI) didn't generate precipitates with Fe(III) 33 and the adsorption of Cr(VI) by GO-Fe(III) was not the dominant mechanism for the removal of Cr(VI) in the solution. Moreover, these results indicate that the visible light irradiation remarkably accelerated the reduction of Cr(VI) due to the existence of the GO-Fe(III) complex. In addition, there was no noticeable reduction of Cr(VI) without illumination, demonstrating visible light irradiation is an essential reaction condition.
Fe(II) is one of the dominant reductants of Cr(VI) in the environment 34 . The production of Fe 2+ in the GO-Fe(III)/Cr(VI) irradiation system was also detected (Fig. 4b). The amount of generated Fe(II) ions in the solution gradually increased to 45 μ M after 2 h of reaction under visible irradiation. However, without irradiation, the concentration of Fe 2+ in the solution was extremely low and kept invariant (see Supplementary Figure S1c), which further declared that the Cr(VI) reduction are mainly triggered by Fe(II) production. Figure 1b illustrated the transformation of Cr(VI) in the aqueous environment containing GO-Fe(III) under visible light irradiation.
As reported, the photolysis of Fe(III)-carboxylate complexes results in the oxidative degradation of the carboxylate ligand and reduction of the metal center to Fe(II) 17,18 . The rapid generation of Fe(II), as shown in Fig. 4b, should be attributed to the ligand-to-metal charge transfer (LMCT) process of Fe(III)-carboxylate complexes at the edge of GO under irradiation. The high light absorption efficiency of the GO-Fe(III) complex may also promote the photoreduction or Fe(II) generation.
The overall photoreduction process can be mainly segregated into the reduction of Fe(III) to Fe(II) and the subsequent reduction of Cr(VI) accompanied by Fe(II) oxidized to Fe(III). As shown in Fig. 4b, the rate of Fe(II) production was faster than its oxidation by Cr(VI), which is similar with the reported DOM-iron/Cr(VI) reaction system 35,36 . The release of Fe(II) suggested that Fe(II) formed very weak complexes with carboxylate ligand in comparison with Fe(III). After Cr(VI) was completely photoreduced to Cr(III) in 60 mins, the accumulation of Fe 2+ ions was further increased, which indicates that the carboxyl group on the GO has a strong photochemical reactivity when it was combined with Fe 3+ . In addition, during this process, GO was partly reduced along with the carboxyl groups degraded to carbon dioxide. Without Fe(III), the photoreduction of Cr(VI) by pure GO under visible light irradiation could not occur in the first hour. However, with extended irradiation time, about 30% of Cr(VI) was photoreduced in the GO/Cr(VI) system (Fig. 4a), which may be attributed to the light responsive and photoelectric properties of GO [37][38][39] .
The change of total chromium concentration (denoted as Cr(T)) in the three reaction systems are illustrated in Fig. 4c. In the GO-Fe(III)/Cr(VI) system, the removal percentage of Cr(T) reached nearly 90% in 120 min. Moreover, the Cr(VI) concentration was approaching zero in a faster decrease rate than Cr(T) in 120 min in the GO-Fe(III)/Cr(VI) system. These results indicated that 90% of chromium elements were removed from the solution in 120 min. And the remaining chromium elements in the solution is not in the highly toxic Cr(VI) form. It is highly possible that most of the Cr(VI) complexes were photoreduced to the positively charged Cr 3+ ions which were adsorbed by the generated rGO followed with precipitation to leave the aqueous environment. The quantitative contribution of adsorption of generated Cr 3+ to the removal percentage of Cr(T) by GO-Fe(III) was shown in Figure S5b, suggesting that adsorption contribution of Cr 3+ increased with the light irradiation time.

Characterization of GO-Fe(III) after Photoreaction.
To elucidate the photoreduction mechanisms of the GO-Fe(III) complex, the morphology and functional group variations of GO and GO-Fe(III) after the photoreaction were identified using TEM, XPS and FTIR. The TEM images (Fig. 5a,b) show that after photoreaction both GO and GO-Fe(III) were apparently aggregated with wrinkles which were tightly distributed on the basal planes of GO nanosheets to form grooved regions. And the surface of GO-Fe(III) was relatively rougher compared with that of GO because more oxygen containing functional groups on GO-Fe(III) were removed than GO, inducing stronger π -π interactions between the nanosheets.
The GO-Fe(III) complex maintained consistent reaction efficiency in 4 cycles of photoreduction (see supplementary Figure S6a), which could be attributed to the abundant carboxyl groups and the chemical stability of GO. On the other hand, supplementary Figure S6b shows the visible changes of the aqueous system after four cycles of photoreduction process. In comparison with the solution before the reaction, the color gradually changed from light yellow to black brown after the reaction, and as the number of repetitions increased, the solution darkened. These observations demonstrated that GO was gradually transformed to reduced graphene oxide (rGO) in black color because of the oxygen containing functional groups consuming on GO. Moreover, black precipitates were generated in the reaction solution, when it was leaved for overnight. This should be contributed by the aggregation and precipitation of rGO generated during the photoreduction.
The surface functional groups of GO after the reaction were analyzed using XPS. The C1s spectrum of GO ( Fig. 5c) after the reaction exhibited identical peaks to that of GO before the reaction (Fig. 3e), indicating the oxygen-containing moieties were barely changed during the visible light irradiation. This phenomenon demonstrates that the optical electronic reaction of GO is slow and not obvious.
By contrast, the GO-Fe(III) reaction system after visible light irradiation, the C1s peak distribution (Fig. 5d) dramatically changed from the spectra before reaction (Fig. 3f), i.e., the carboxyl carbon in O-C = O decreased, and the carbon in C-O increased. The O/C ratio decreased, and the C-C/C = C content increased (Table 1), which indicates that the surface oxygen-containing functional groups of GO-Fe(III) were dramatically reduced in the photoreaction process.
Besides, a new peak shows up at 291.4 eV in Fig. 5d, which is assigned to the carbon in C-O-Cr(III). It seems the Cr 3+ ions were coordinated in the precipitation. The Cr2p XPS spectrum of the GO-Fe(III)/Cr(VI) precipitate (Fig. 5e) shows a pair of peaks at 577.1 and 586.8 eV that correspond to Cr(III) 2p3/2 and Cr(III) 2p1/2 40 , respectively, further indicating the existence of Cr(III) in the precipitated rGO sheets. The elemental mapping of GO-Fe(III)/Cr(VI) after photoreaction also indicated the existence of chromium species (see supplementary Figure S7).
The FTIR spectra (Fig. 5f) further reveal the considerable changes in the various oxygen-containing functional groups on GO-Fe(III) after the reaction. In Fig. 5f, the peaks indicative of the carboxyl groups in C = O and the O = C-O-Fe both diminished after each reaction cycle, demonstrating that the oxygen-containing functional groups of GO-Fe(III) are gradually removed in the Fe(III)/Fe(II) cyclic reaction under visible light irradiation. In supplementary Figure S8, the C = O of the carboxyl groups on GO is converted into the epoxy C-O which is excited by light illumination. Simultaneously, the FTIR peak for the alkoxy in C-O-C was slightly decreased because the thermal stability of the alkoxy C-O-C on the surface is poor, and the alkoxy can be easily removed under light irradiation. In short, the GO sheets were reduced to rGO after the photoreaction with the existence of Fe 3+ . The generated rGO sheets adsorbed with Cr(III) spontaneously aggregated and precipitated to leave the aqueous solution.

Influential Factors of Cr(VI) Photoreduction by the GO-Fe(III) Complex. The effects of dissolved
oxygen to the photoreduction of Cr(VI) by GO-Fe(III) were examined (Fig. 6a). The Cr(VI) reduction kinetics are similar in N 2 and in an air-saturated solution, which indicates that oxygen is not competitive with Cr(VI) for accepting electrons. This observation can be readily explained from a thermodynamic perspective (see Fig. 6a). The excited GO (E 0 (GO/ RGO) = − 0.85 V vs NHE) 41 preferentially donates an electron to the structural Fe(III) 43 . So the photo-induced electron would preferentially transfer to Fe(III) which has higher redox potential than O 2 . Then, the generated Fe(II) may reduce Cr(VI) to Cr(III). This observation indicates that when the GO-Fe(III) complex is under sunlight illumination, the reduction of Cr(VI) can be easily achieved, even under anaerobic conditions. The effects of the GO proportion in the GO-Fe(III) complex on the photoreduction process were examined under visible light irradiation (Fig. 6b). The reaction rate constants of 5 wt% GO-Fe(III), 1 wt% GO-Fe(III), and 0.1 wt% GO-Fe(III) are 4.81 × 10 −2 min −1 (R 2 = 0.98), 2.87 × 10 −2 min −1 (R 2 = 0.97), and 2.81 × 10 −2 min −1 (R 2 = 0.98), respectively. These results implied that the increased content of GO can accelerate the electron transfer from GO-Fe(III) to Cr(VI). Although with lower GO loading, the time for complete reduction of Cr(VI) was The Cr(VI) reduction rates at different pH values were shown in supplementary Figure S9a. Under acidic conditions, the reduction rate increased with decreasing pH value. The GO-Fe(III) complex is more stable at low pH; and with the pH increase, the ferric ions gradually generate precipitates. The removal rate of Cr(VI) is also affected by the chromium speciation which is pH-dependent. At low pH, the predominant Cr(VI) specie is HCrO 4 − , which shifts to CrO 4 2− with the increase of solution pH. HCrO 4 − and CrO 4 2− display different reactivities 44 . The high removal efficiency at low pH is attributed to the highly protonated and positively charged surface of GO-Fe(III), which favors the approach of the negatively charged HCrO 4 − via electrostatic attraction. Concurrently, the competition of OH − with chromate ions reduces the Cr(VI) removal. The solution pH also affects the thermodynamic driving force of Cr(VI) photoreduction 45,46 . Therefore, the acid or faintly acid environment would be in favor of the Cr(VI) reduction.
The photoreduction was conducted with Cr(VI) in a serial of initial concentrations (see Supplementary Figure S9b). The photoreduction kinetics of Cr(VI) with an initial concentration of 20-100 μ M well obeyed the first-order kinetic model with a good linear correlation (R 2 > 0.9) (insert in Figure S7b). It has been reported that, at low Cr(VI) concentrations, the rate of Cr(VI) reduction is proportional to the initial Cr(VI) concentration 47 . At high Cr(VI) concentrations, the rate of Cr(VI) reduction begins to level off due to the site-saturation behavior caused by the precipitation of soluble solid such as the Cr(III) complex or the depletion of available Fe(II) from GO-Fe(III) in a heterogeneous reaction. Thus, the removal of Cr(VI) by a certain amount of the GO-Fe(III) complex under sunlight irradiation is faster in an aquatic environment containing less Cr(VI).

The Mechanism of Cr(VI) Photoreduction and Removal by GO-Fe(III). Based on the photoreduction
kinetics, influential factors, and structural characteristics before and after irradiation, the photoreduction mechanism and Cr(VI) removal by GO-Fe(III) are initially proposed in Fig. 7. The proposed mechanism is related to the synergistic interfacial effects of the oxygen-containing groups on GO and Fe(III), which include the photoreduction, adsorption and precipitation. First, Fe 3+ is adsorbed by the carboxyl groups on the GO nanosheet surface to form the GO-Fe(III) complex (process a). Then, the photoreaction occurs at the solid-liquid interface, where the active sites of the Fe(III) complex with the surface carboxyl groups of GO, frequently converting Fe(III) to Fe(II) by GO-Fe(III) sensitization under visible light irradiation (process b), which is accompanied with the oxidative degradation of carboxyl groups to CO 2 (process c). The generated Fe(II) can be present in the binding state   Concurrently, Fe 2+ is restored to Fe 3+ in the subsequent reduction reaction (process e), and the produced Cr 3+ ions are adsorbed by GO or rGO (process f), which decreases the total chromium concentration of the solution.
The photoreaction mechanism of GO-Fe(III) is similar to that of ubiquitous complexes of dissolved organic matter (DOM) with Fe(III) (i.e., DOM-Fe(III)) in a natural environment 35,36,48 , but the environmental behavior of GO-Fe(III) is different from the interfacial photoreaction of Fe(III)/Fe(II) and Cr(VI)/Cr(III) since the GO sheets are also photoreduced to rGO (process g). The enhanced π -π attraction between layers and Cr(III) adsorption further promote the aggregation, resulting co-precipitation of rGO and Cr(III). On the other hand, GO itself has slight photosensitization because of the quinonyl groups 49,50 , which could slowly transfer the photo-electron to the adsorbed Cr(VI) realizing its minor reduction (process h). Presumably, in a complicated aqueous environment containing GO nanosheets, Fe(III) and highly toxic Cr(VI), the GO sheets may coordinate with Fe(III) to form complex which will reduce Cr(VI) to less toxic positively charged Cr 3+ ions and simultaneously be converted to rGO under visible light irradiation. Subsequently, the rGO sheets will adsorb Cr 3+ ions and spontaneously aggregate to form precipitations. In short, both GO and toxic Cr(VI) can be removed from the aqueous environment with the existence of iron ions as the catalyst under light irradiation.

Conclusions
The widespread applications of GO in diverse fields opens the possibility for GO to inevitably enter into natural aquatic environments. The environmental factors (such as light, transition metals, heavy metals, dissolved oxygen, and pH) would influence the transportation and transformation of GO in the environment. Our work suggests that GO will form photoactive complex with Fe(III) in the aqueous environment. The GO-Fe(III) complex may effectively reduce Cr(VI) to Cr(III) under visible light irradiation because of the decarboxylation on GO. The losing of carboxyl groups on GO generates rGO which is also able to adsorb Cr(III) but easier to aggregate and precipitate. The transformation of GO with the help of iron ions under visible light indicated a promising route to convert and remove the highly toxic Cr(VI) pollutant from the aqueous environment.
Graphene oxide (GO) was synthesized from graphite flakes using the modified Hummers method 51 . The GO-Fe(III) complex was prepared by mixing the GO solution (pH = 2, adjusted by HCl) with FeCl 3 (1 mol/L) under stirring for 24 h. Then, the mixed solution was centrifuged at 10000 rpm and washed with deionized water several times until the upper solution contained no ferric ions. The resultant precipitate was dispersed into water as photoreaction materials and dried at 60 °C for structural characterization. Different GO-loading complexes were synthesized by changing the ratio of GO to 0.1 wt%, 1 wt% and 5 wt%.  Scientific RepoRts | 7:40711 | DOI: 10.1038/srep40711 of 3.00 ± 0.05, except for experiments that considered the effect of pH values. The initial pH was adjusted using diluted HCl and NaOH. Before illumination, GO/Cr(VI) and GO-Fe(III)/Cr(VI) were stirred for 30 min in the dark to obtain the adsorption/desorption equilibrium. Deaerated suspensions were prepared by purging nitrogen gas for at least 30 min before irradiation and throughout the entire experimental process. At regular time intervals, approximately 2 mL of the reaction solutions were sampled and filtered through a 0.22 μ m filter for further detection. Duplicate runs were performed for each experiment. All experiments were conducted in a 60 mL Pyrex vessel with magnetic stirring. The irradiation source was a 500 W halogen lamp with a UV cutoff filter (λ > 420 nm) in a XPA-7 type photochemical reactor (Xujiang Electromechanical Plant, Nanjing, China). The temperature was maintained at room temperature by circulating cooling water.

Photoreduction of Cr(VI) by GO-Fe
The Cr(VI) concentration was determined using the diphenylcarbazide (DPC) method at a wavelength of 540 nm in a Shimadzu UV-2550 spectrophotometer 52 . Briefly, a 1 mL sample was added to 1 mL of buffer solution (V(H 2 SO 4 ):V(H 3 PO 4 ):V(H 2 O) = 1:1:2), and 2 mL chromogenic reagent (0.2 g diphenylcarbazide mixture with 100 mL acetone solution V (acetone:H 2 O) = 1:1) was subsequently added. To prevent Fe(II) from interfering with the determination of the Cr(VI) concentration, the Cr(VI) concentration was calibrated using a blank control. Dissolved Fe 2+ was quantified using a 1,10-phenanthroline method at a wavelength of 510 nm in a Shimadzu UV-2550 spectrophotometer 53 . Briefly, the premixture was prepared by combining 0.5 mL of sodium acetate/acetic acid buffer (pH = 5.7), 0.25 mL of ammonium fluoride solution (0.4 mol/L), 0.25 mL of 1,10-phenanthroline solution (20 mmol/L), and 1 mL of sample solution. The Fe(II) molar concentration (μ mol/L) was calculated as follows: [Fe(II)] = (2A i1 − A i2 )/0.011, where A i1 is the absorbance of the sample after a chromogenic reagent was added at 510 nm and A i2 is the absorbance of the sample at 510 nm. The Cr(VI) reduction efficiency at a given time (t) was calculated as follows: reduction efficiency (%) = (C 0 − C t )/C 0 × 100%. A pseudo first-order model was applied to describe the kinetics of Cr(VI) photoreduction by GO-Fe(III). The first-order constant k (min −1 ) was determined according to the following equation: ln(C 0 /C t ) = kt, where C t is the concentration of Cr(VI) at time t, C 0 is the initial concentration of Cr(VI), k is the reduction rate constant, and t is the irradiation time. The total Cr and Fe concentrations in the supernatant were determined using atomic absorption spectrometry (Perkin Elmer Analyst 700). The variation of the surface functional groups and morphologies of the GO-Fe(III) complex were observed using XPS, FTIR and TEM.

Characterization of GO-Fe(III).
The surface morphologies of GO and GO-Fe(III) were characterized using SU-8000 scanning electron microscopy (SEM) (Hitachi, Tokyo) and FEI Tecnai G 2 F20 S-TWIN Transmission Electron Microscopy (TEM) (FEI, America). Thermogravimetric analyses (TGA) was conducted on a SDT Q600 V8.2 Build 100 apparatus at a heating rate of 5 °C min −1 from room temperature to 850 °C in air and N 2 flow. Atomic force microscopy (AFM) images of GO and GO-Fe(III) on a freshly cleaved mica surface were obtained using a Nanoscope III in tapping mode with a NSC14/no Al probe (Dimension Icon, Veeco). After sonication was performed for 5 min, a droplet of the sample dispersion (~0.01 mg/mL) was cast onto a freshly cleaved mica surface. The sample was maintained at room temperature overnight to allow evaporation of the water. The surface functional groups were observed by Fourier transform infrared spectroscopy (FTIR) and X-ray photoelectron spectroscopy (XPS). The FTIR spectra were recorded in the 4000-400 cm −1 region with a resolution of 4 cm −1 using a Bruker Vector 22 FTIR spectrometer. The XPS experiments were performed on an Escalab 250 Xi with a resolution below 0.5 eV, and the C1s, Fe2p and Cr2p peak spectra were analyzed using XPS Peak 4.1 software.