Enhanced photocatalytic dye degradation and hydrogen production ability of Bi25FeO40-rGO nanocomposite and mechanism insight

A comprehensive comparison between BiFeO3-reduced graphene oxide (rGO) nanocomposite and Bi25FeO40-rGO nanocomposite has been performed to investigate their photocatalytic abilities in degradation of Rhodamine B dye and generation of hydrogen by water-splitting. The hydrothermal technique adapted for synthesis of the nanocomposites provides a versatile temperature-controlled phase selection between perovskite BiFeO3 and sillenite Bi25FeO40. Both perovskite and sillenite structured nanocomposites are stable and exhibit considerably higher photocatalytic ability over pure BiFeO3 nanoparticles and commercially available Degussa P25 titania. Notably, Bi25FeO40-rGO nanocomposite has demonstrated superior photocatalytic ability and stability under visible light irradiation than that of BiFeO3-rGO nanocomposite. The possible mechanism behind the superior photocatalytic performance of Bi25FeO40-rGO nanocomposite has been critically discussed.


Results and Discussions
Materials characterization. The crystallographic structure of the synthesized nanomaterials has been investigated by obtaining their X-ray diffraction (XRD) patterns as shown in Fig. 1. The diffraction peaks exhibited by both BiFeO 3 and BiFeO 3 -rGO conform to the single-phase perovskite structure (R3c space group) of BiFeO 3 (JCPDS card No. 86-1518). On the other hand, the diffraction peaks of Bi 25 FeO 40 -rGO are in good agreement with single-phase sillenite structure (I23 space group) of Bi 25 FeO 40  . Reduction of GO can be observed for both nanocomposites as the characteristic peak of GO at (002) is absent in the corresponding patterns. The average particle size of the synthesized nanomaterials has been obtained from particle size histograms extracted from their field emission scanning electron microscopy (FESEM) images. Incorporation of rGO has reduced the average particle size of BiFeO 3 -rGO (40 nm) and Bi 25 FeO 40 -rGO (20 nm) nanocomposites compared to that of pure BiFeO 3 nanoparticles (70 nm) as shown in Supplemental Fig. S1. Comparatively smaller particle size implies larger surface to volume ratio which is a necessary property of an efficient photocatalyst. The mechanism behind inhibition of particle growth due to the inclusion of rGO has been presented elsewhere 38 .
X-ray photoelectron spectroscopy (XPS) has been performed on the synthesized nanomaterials to analyze the formation of different bonds on their surface. Figure 2(a-c) presents the O 1s spectra of BiFeO 3 nanoparticles, BiFeO 3 -rGO, and Bi 25 FeO 40 -rGO nanocomposites. The peaks for both Bi-O and Fe-O bonds occur at around 529 eV and 530 eV respectively for all the samples. This result is consistent with the values reported in the literature [49][50][51] . As shown in Fig. 2(b,c), the presence of a peak around 531 eV can be attributed to the formation of COOR bond, i.e., the formation of a composite with graphene 52 . Figure 2(d-f) shows the C 1s spectra of Bi 25 FeO 40 -rGO, BiFeO 3 -rGO nanocomposites, and GO respectively. As shown in Fig. 2(f), the C-O peak is almost as strong as the C-C peak in GO. However, the intensity of the C-O peak is significantly reduced in both nanocomposites as shown in Fig. 2(d,e) while the intensity of the COOR peak does not change significantly. The dominance of the C-C peak compared to the C=O and COOR peaks implies that GO has been considerably reduced and the honeycomb structure of graphene has also been nearly restored. This partial reduction of GO can be attributed to the nature of reduction, e.g., the hydrothermal process. The hydrothermal process reduces the GO sheets by decomposing the C=O and COOR bonds 53 . Higher thermal energy provided by a higher process temperature favors faster and greater reduction [53][54][55][56][57][58][59] . However, a process temperature that is higher than 180 °C would be unfavorable for the single phase growth of Bi 25 FeO 40 -rGO nanocomposite 38 . Although single phase BiFeO 3 -rGO nanocomposite can be synthesized at a higher temperature, the higher growth rate may cause the particles to grow larger and even agglomerate. In addition to the higher expenses associated with a higher temperature process, we may face further difficulty in the synthesis process as the control over the particle size and morphology might be lost. Besides, the process also removes carbon atoms from the carbon plane in the form of CO 2 and splits the graphene sheet into smaller pieces 59,60 . The dangling carbon atoms at the end of the smaller graphene sheets may be oxidized again and contribute to the COOR bonds present in the surface of the composites.
Since XPS analyzes the bond properties of the surface (within 3-5 nm) of a sample rather than its bulk, an analysis like FTIR spectroscopy would be more useful for studying the presence of different functional groups in the nanocomposites. Hence, FTIR spectroscopy has also been performed on GO and the nanocomposites to verify the reduction of GO into rGO conclusively. The components and functional groups have been identified from the FTIR spectra as shown in Fig. 3. The vibrational band ranging from 1053 to 1220 cm −1 can be attributed to the presence of different types of C-O bonds. In addition, the vibrational bands at 1718 cm −1 and 3441 cm −1 can be assigned to the presence of C=O and −OH bonds respectively [61][62][63][64][65] . As evident from Fig. 3, the peaks for each of these oxidized species are considerably smaller in the spectra of the nanocomposites compared to that of GO. Notably, the peak due to C=O and COOR bonds is significantly smaller than that of GO for both SciENtific REPORTS | (2018) 8:11090 | DOI:10.1038/s41598-018-29402-w nanocomposites. Hence, the FTIR spectra prove that even the C=O and COOR bonds have also been considerably reduced. FTIR characterization, in conjunction with the XPS characterization, leads us to the conclusion that GO has been reduced to a significant extent during the formation of the nanocomposites.
Photocatalytic Degradation Activity and Stability. BiFeO Fig. 4 In addition to the superior degradation ability, a photocatalyst is expected to be stable under the reaction conditions to be considered for practical applications. Since the stability of BiFeO 3 as a photocatalyst is well known, only BiFeO 3 -rGO and Bi 25 FeO 40 -rGO nanocomposites were subject to the recyclability test. As observed in Fig. 4(d), BiFeO 3 -rGO exhibits a small loss in degradation ability during each successive cycle. However, Bi 25 FeO 40 -rGO shows excellent stability with negligible loss in photocatalytic ability after 3 consecutive cycles. Since Bi 25 FeO 40 -rGO nanocomposite has shown a comparatively higher photocatalytic degradation ability and better stability than BiFeO 3 -rGO nanocomposite, we believe that it can be considered for different practical applications such as environmental pollutant degradation, hydrogen production via water-splitting, etc. Photocatalytic Hydrogen Production Test Analysis. A photocatalytic hydrogen production experiment has also been performed using the synthesized nanomaterials. Figure 5 presents the amount of hydrogen gas (ml/g) generated after 4 hours of visible light irradiation. Photocatalytic hydrogen production of BiFeO 3 , BiFeO 3 -rGO, and Bi 25 FeO 40 -rGO has also been compared to that of a standard photocatalyst, Degussa P25 titania nanoparticles 66 . Figure 5 presents the amount of hydrogen produced by BiFeO 3 , BiFeO 3 -rGO, Bi 25 FeO 40 -rGO and Degussa P25 under the same experimental conditions. In addition to the high photocatalytic degradation ability, Bi 25 FeO 40 -rGO nanocomposite demonstrates the highest volume of hydrogen production which is almost 3.5 times greater than that of Degussa P25 nanoparticles.  Photocatalytic Mechanism. The optical band gap is a crucial parameter for gaining an insight into the photocatalytic ability of a material. However, the potentials of the valence band maxima (VBM) and conduction band minima (CBM) also play significant roles in catalyzing the redox reactions in the electrolyte system. Absorption of a photon with energy greater than or equal to the band gap of a material excites an electron from the valence band to its conduction band, i.e., creates an electron-hole pair. The reduction half-reaction commences when the photogenerated electron has a potential that exceeds the reduction potential of the redox couple. Similarly, the photogenerated hole is required to possess a potential that exceeds the redox potential of the oxidation half-reaction. Efficient photocatalytic degradation ability of the nanocomposites can be better understood by analyzing the redox reactions in the electrolyte system that govern the degradation of RhB dye. Photocatalytic degradation of RhB is, in fact, a collection of specific redox reactions. Among these redox reactions, two are particularly important 33 . First, the photogenerated electron can react with the surface adsorbed O 2 (redox potential: −0.16 V vs. NHE) to form − O 2 , which will further react with RhB to cause degradation 33 . Second, the photogenerated holes can react with the OH − ionized from the water molecules to produce ⋅OH (redox potential: 2.38 V vs. NHE). This ⋅OH can further oxidize the RhB molecules. Therefore, a photocatalyst needs to possess a CBM < −0.16 V to drive the first reaction and VBM > 2.38 V to drive the second one efficiently.

Comparison of Optical
These prerequisites necessitate an investigation of the potentials of the VBM and CBM of the synthesized photocatalysts. The Fermi energy of the pure BiFeO 3 and Bi 25 FeO 40 can be calculated theoretically from the equation, E F = χ − E C 72,73 . Here, χ is the absolute electronegativity of the semiconductor which can be found from the geometric mean of the absolute electronegativity of the constituent elements, and E C is the energy of a free electron in hydrogen scale which is approximately 4.5 eV. While usually, Fermi energy lies midway between the CBM and VBM of a semiconductor, first-principles studies of several sillenites show that the energy of their VBM coincides with the Fermi energy 74 . Hence, the energies of CBM and VBM of BiFeO 3 were obtained by simply adding and subtracting 0.5E g to its Fermi energy whereas for Bi 25 75 . We propose that similar interactions may have taken place in the BiFeO 3 -rGO and Bi 25 FeO 40 -rGO nanocomposites. Furthermore, this shift in the band positions can be viewed from a different perspective. Figure 6(a) shows the formation of a heterojunction between graphene and a semiconductor where Fermi energies of graphene and the semiconductor have been denoted as E FG and E FS respectively. If E FG > E FS , electrons will be transferred to the semiconductor from graphene during the formation of the heterojunction until the system reaches the equilibrium Fermi energy, E F 76 . This electron transfer at the heterointerface will give rise to the formation of a surface dipole 77-79 as shown in Fig. 6(a). For a bulk semiconductor, the electric field that arises from the surface dipole will cause band bending that will uniformly change the potentials of vacuum, CBM, and VBM. However, for a nanostructured semiconductor, nearly flatband conditions are observed at the heterojunction since its dimensions are not large enough to change the potentials of CBM  80,81 . Consequently, the electric field brings the vacuum level closer to both CBM and VBM. This shift in the potential of vacuum level will reduce the CBM of the nanostructured semiconductor as shown in Fig. 6(a). Since Fermi energy of pristine graphene (0 V vs. NHE) 70,71 is greater than the Fermi energies of both BiFeO 3 and Bi 25 FeO 40 , the model described above predicts that upward band-shifting should take place during the formation of both nanocomposites. This upward shift makes the VBM < 2.38 V for both nanocomposites, rendering them unable to perform the second redox reaction of RhB degradation using the photogenerated holes. Therefore, the high photocatalytic degradation performance by the nanocomposites should be attributed to the first reaction. Due to the upward band-shifting, CBM potentials of both nanocomposites become sufficiently negative (CBM < −0.16 V) which drives the photogenerated electrons in performing the first redox reaction of RhB degradation. Figure 6(b) consolidates the estimated CBM and VBM potentials of all the synthesized nanomaterials as well as the redox potentials of different redox half-reactions of our interest. Notably, negative CBM potential is one of the most important conditions for efficient photocatalytic hydrogen production via water-splitting. As observed in Fig. 5, our experimental results from photocatalytic hydrogen production test also support the prediction made by our model as the nanocomposites demonstrate considerably higher hydrogen production compared to that of pure BiFeO 3 nanoparticles.
Since pure BiFeO 3 has a positive CBM potential, it is intuitively confusing to notice that it can also produce hydrogen by water-splitting. A positive CBM potential indicates that only the states at the minima of the conduction band have positive potential. However, there should be several other allowed states in the conduction band with negative potential. When photons with appropriate energy are incident on a material, the electrons of valence band will gain enough energy to reach those states with negative potential and participate in the production of hydrogen. However, these photons, of course, must have energies that are higher than the band gap energy. Therefore, these energies may not lie in the visible range of the solar spectrum. Since the intensity of the incident photons of sunlight decreases considerably with the increase in photon energy beyond visible range, the  number of available photons for such transitions is considerably smaller. Besides, these allowed transitions may even be indirect and therefore have smaller transition probability than the near band gap transitions. Therefore, while pure BiFeO 3 and several other semiconductors with positive CBM potential may be able to produce hydrogen via water-splitting, their efficiency will be significantly smaller than that of the semiconductors with negative CBM potential. This hypothesis is also evident when we compare the results of photocatalytic degradation of RhB and photocatalytic hydrogen generation. While both BiFeO 3 -rGO and Bi 25 FeO 40 -rGO nanocomposites exhibit a similar trend of photocatalytic activities during both experiments, BiFeO 3 shows a comparatively poorer performance in hydrogen production than that of RhB degradation. This inconsistent performance of BiFeO 3 can be better understood if we take a keen look at its VBM potential(2.61 V). Since BiFeO 3 has a VBM potential that exceeds 2.38 V, the holes generated by incident photons with energy greater than its band gap of 2.1 eV are sufficient for photodegradation of RhB. However, as mentioned above, photons with much higher energy than 2.1 eV are necessary to commence the production of hydrogen. This inherent shortcoming of BiFeO 3 nanoparticles eventually reduces their efficiency in photocatalytic hydrogen production.
Due to the high electronegativity of oxygen atoms, many metal oxides with narrow band gaps such as α-Fe 2 O 3 , BiVO 4 , WO 3 , etc. have positive CBM potentials that suppress their ability in water-splitting. Our proposed model predicts that formation of nanocomposites with rGO may shift their CBM potentials to negative values and trigger the suppressed photocatalytic hydrogen production ability of these metal oxides. This model unveils a novel mechanism behind the superior photocatalytic abilities of semiconductor-rGO nanocomposites.
Effective suppression of recombination processes is another important characteristic of a good photocatalyst. BiFeO 3 is a ferroelectric material that possesses a spontaneous inherent electric field. This electric field can spatially separate the photogenerated electrons and holes to reduce the probability of recombination 82 . A previous first-principles investigation revealed that the recombination processes in a sillenite are suppressed due to an intraband transition of the photogenerated hole 74 . We believe that this suppression of recombination processes would also apply to Bi 25 FeO 40 . Besides, the formation of a composite with rGO causes the photogenerated electrons to move to the rGO sheet where the superior electron mobility of the restored honeycomb structure can help the electron transport to the redox couples [26][27][28]38,79 . As shown in Fig. S3, steady-state photoluminescence (PL) spectra of the photocatalyst samples justify our hypothesis as the nanocomposites show considerably smaller PL peak intensities compared to that of pure BiFeO 3 nanoparticles. Smaller PL peak intensity implies that a smaller number of photogenerated electrons and holes are radiatively recombined. Therefore, we believe that the photogenerated electron-hole pairs are more effectively separated when a composite with rGO is formed.

Conclusions
We have demonstrated a low temperature versatile hydrothermal technique for synthesizing highly efficient visible light photocatalysts of bismuth ferrite-rGO nanocomposite family whose phase can be controlled by tuning the synthesis temperature. Merging rGO sheets with each of the pristine BiFeO 3 and Bi 25 FeO 40 nanoparticles has resulted in better morphology, superior optical absorption, and favorable band edge shifts. The improvements associated with inclusion of rGO have also been reflected on the photocatalytic ability of the nanocomposites as they demonstrated considerably better degradation of RhB dye as well as greater hydrogen production by water-splitting over both pure BiFeO 3 nanoparticles and a standard photocatalyst, Degussa P25 titania. Furthermore, Bi 25 FeO 40 -rGO nanocomposite has shown 3.5 times greater photocatalytic hydrogen production compared to that of Degussa P25 titania nanoparticles. Moreover, a theoretical model has been proposed to explain the band-shifting phenomenon which may prove to be significant in understanding the superior photocatalytic performance of different metal oxide-rGO nanocomposites. Characterizations Techniques. Structural characterization of the prepared samples was performed by obtaining their X-ray diffraction (XRD) patterns using a diffractometer (PANalytical Empyrean) with a Cu X-ray source (Wavelength, λ: K α1 = 1.540598 Å and K α2 = 1.544426 Å). Field emission scanning electron microscopy (FESEM) imaging of the samples was performed using a scanning electron microscope (XL30SFEG; Philips, Netherlands). The FESEM images were analyzed to determine the morphology of the nanomaterials. ImageJ software was used to analyze the size of the particles. X-ray photoelectron spectroscopy (XPS) was used to investigate the surface chemical states of synthesized nanomaterials. Fourier transform infrared (FTIR) spectroscopy (Shimadzu, IRSpirit-T) was also performed in the wavenumber range of 1000 cm −1 to 4000 cm −1 for studying the presence of different functional groups. Steady-state photoluminescence (PL) spectroscopy (Hitachi, F-700 FL spectrophotometer) was used to investigate the photogenerated carrier recombination of the nanomaterials. The excitation wavelength used for PL spectroscopy was 400 nm.

Synthesis of BiFeO
SciENtific REPORTS | (2018) 8:11090 | DOI:10.1038/s41598-018-29402-w Photocatalytic degradation test. A photocatalytic degradation test was performed on Rhodamine B (RhB) dye using the synthesized nanomaterials as photocatalyst following the typical procedures. A solution was prepared by dissolving 10 mg RhB in 50 ml of distilled water. An aliquot of 5 ml was extracted from the solution to measure its absorbance using a UV-vis spectrophotometer. The intensity of the absorbance peak (553 nm) can be considered proportional to the amount of RhB present in the solution. This method was used each time when it was required to determine the remnant amount of RhB in the solution. Then the solution was magnetically stirred in dark conditions for 30 minutes to obtain a homogeneous solution. 40 mg of photocatalyst was then added to the solution and again stirred for 30 minutes to ensure adsorption-desorption equilibrium between photocatalyst and RhB. The photocatalysis process was initiated by irradiating the suspension with a Mercury-Xenon lamp (Hamamatsu L8288, 500 W). The UV portion of the light was filtered out with a UV cut-off filter (<400 nm). The degradation of RhB dye was monitored by carrying out absorbance measurements of the suspension at every 30 minutes interval. However, the photocatalyst nanomaterials were precipitated from the suspension using a centrifuge machine at 5500 rpm for 10 minutes before measuring its absorbance. Prior to recycling the photocatalyst, the remnant photocatalyst nanomaterials in the suspension were separated by centrifugation, washed with distilled water and dried afterward.
Photocatalytic hydrogen generation test. Photocatalytic hydrogen production test was performed in a slurry-type photochemical reactor. 20 mg of photocatalyst sample was dispersed into 30 ml deionized water, and the suspension was stirred magnetically to obtain homogeneity. The inert atmosphere required for photocatalytic hydrogen production was ensured by purging the system with argon gas for 30 minutes. The suspension was irradiated with visible light by a Mercury-Xenon lamp (Hamamatsu L8288, 500 W) coupled with a UV cut-off filter (<400 nm). The generated gas was extracted at an interval of 30 minutes for 4 hours and evaluated by a gas chromatography (GC) device equipped with a thermal conductivity detector (TCD) and a gas analyzer. The amount of generated hydrogen gas was determined in ml H 2 /g catalyst.