Graphene Oxide/α-Bi2O3 Composites for Visible-Light Photocatalysis, Chemical Catalysis, and Solar Energy Conversion


  • Dr. Tirtha Som,

    Corresponding author
    1. Institut für Chemie, Humboldt-Universität zu Berlin, Brook-Taylor-Straße 2, 12489 Berlin (Germany)
    • Institut für Chemie, Humboldt-Universität zu Berlin, Brook-Taylor-Straße 2, 12489 Berlin (Germany)

    Search for more papers by this author
  • Gerald V. Troppenz,

    1. Institut für Silizium-Photovoltaik, Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Kekuléstraße 5, 12489 Berlin (Germany)
    Search for more papers by this author
  • Robert Wendt,

    1. Institut für Chemie, Humboldt-Universität zu Berlin, Brook-Taylor-Straße 2, 12489 Berlin (Germany)
    2. Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner-Platz 1, 14109 Berlin (Germany)
    Search for more papers by this author
  • Dr. Markus Wollgarten,

    1. Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Hahn-Meitner-Platz 1, 14109 Berlin (Germany)
    Search for more papers by this author
  • Dr. Jörg Rappich,

    1. Institut für Silizium-Photovoltaik, Helmholtz-Zentrum Berlin für Materialien und Energie GmbH, Kekuléstraße 5, 12489 Berlin (Germany)
    Search for more papers by this author
  • Dr. Franziska Emmerling,

    1. BAM Bundesanstalt für Materialforschung und -prüfung Richard-Willstätter-Straße 11, 12489 Berlin (Germany)
    Search for more papers by this author
  • Prof. Dr. Klaus Rademann

    1. Institut für Chemie, Humboldt-Universität zu Berlin, Brook-Taylor-Straße 2, 12489 Berlin (Germany)
    Search for more papers by this author


The growing challenges of environmental purification by solar photocatalysis, precious-metal-free catalysis, and photocurrent generation in photovoltaic cells receive the utmost global attention. Here we demonstrate a one-pot, green chemical synthesis of a new stable heterostructured, ecofriendly, multifunctional microcomposite that consists of α-Bi2O3 microneedles intercalated with anchored graphene oxide (GO) microsheets (1.0 wt %) for the above-mentioned applications on a large economical scale. The bare α-Bi2O3 microneedles display two times better photocatalytic activities than commercial TiO2 (Degussa-P25), whereas the GO-hybridized composite exhibits approximately four to six times enhanced photocatalytic activities than the neat TiO2 photocatalyst in the degradation of colored aromatic organic dyes (crystal violet and rhodamine 6G) under visible-light irradiation (300 W tungsten lamp). The highly efficient activity is associated with the strong surface adsorption ability of GO for aromatic dye molecules, the high carrier acceptability, and the efficient electron–hole pair separation in Bi2O3 by individual adjoining GO sheets. The introduction of Ag nanoparticles (2.0 wt %) further enhances the photocatalytic performance of the composite over eightfold because of a plasmon-induced electron-transfer process from Ag nanoparticles through the GO sheets into the conduction band of Bi2O3. The new composites are also catalytically active and catalyze the reduction of 4-nitrophenol to 4-aminophenol in the presence of borohydride ions. Photoanodes assembled from GO/α-Bi2O3 and Ag/GO/α-Bi2O3 composites display an improved photocurrent response (power conversion efficiency ∼20 % higher) over those prepared without GO in dye-sensitized solar cells.


The fabrication of hybrid materials with the incorporation of oxidized graphene, such as graphene oxide/semiconductor (GO/SC) composites, have been pursued intensively in recent years because of their promising sustainable applications in environmental conservation,14 catalysis,5, 6 ecofuel H2 generation,79 and light-energy harvesting.1012 Special focus is given to comprehend (i) the unique ability or the mechanistic pathway followed by these materials to photodegrade carcinogenic aromatic industrial effluents13, 14 and other mutagenic organic contaminants,15, 16 (ii) their role in surface catalysis,17, 18 and (iii) their contribution towards the enhancement of existing photovoltaic cell efficiencies.18, 19

GO is constituted structurally of a graphene framework with some oxygen-containing functional groups.2022 The hydroxyl and epoxide groups are bonded covalently at the basal planes, whereas the phenolic OH and carboxylic groups are attached at the edges of the hexagonal carbon framework.2224 The C/O/H ratio in GO may range from 8:2:3 to 8:4:5, which depends on the preparation method and extent of oxidation.25

GO and graphene are chemically interconvertible.20, 21 There are several physicochemical advantages that GO possesses over graphene or chemically reduced graphene oxide (RGO). These properties render GO to be a preferential scaffold for several applications, primarily in photocatalysis, catalysis, optoelectronics, and electrochemistry.2025 The ionizable functional groups present on GO act as binding sites that facilitate the deposition and anchoring of nanoSCs, metal nanoparticles (NPs), and organic (dye) molecules through direct noncovalent, covalent, π–π, and/or van der Waals interactions.2224, 26 GO also offers an incredibly large surface area for adsorption owing to the two accessible (exposed) sites, which are often unavailable in coagulated RGO sheets because of van der Waals interactions between the reduced sheets. The BET surface area reported for RGO is approximately 400–470 m2 g−1,20 whereas the BET surface area reported for GO is in the range ∼600–900 m2 g−1.27

Graphene and RGO are zero-band-gap materials. In GO, the C atoms that are attached covalently to the functional groups are sp3 hybridized. These can be envisaged as locally oxidized regions that disrupt the extended sp2-conjugated honeycomb network. The sp2 C fraction in GO is usually ∼40 % and these can be visualized as the unoxidized regions.2024 Upon reduction (thermal and chemical), the sp2 C fraction can be increased to ∼80 %, and the C/O ratio can be improved to approximately 12:1 with the loss of O-containing functional groups and the restoration of the conductive network.20, 24 Moreover, regulation of the sp2/sp3 carbon proportions in GO enables a continuous tuning of its band gap from approximately 3.5 eV down to 1 eV.28, 29 This provides the opportunity to effectively align the position of the energy levels of GO (valance band) with respect to a variety of available SCs in a composite photocatalytic material. This facilitates the effective charge transfer (electrons and holes) required to drive the photochemical or photo-electrochemical processes. Furthermore, GO also has excellent electron accommodating and shuttling ability. If combined with the SC, GO facilitates immediate charge separation and migration of the photoexcited SC. GO effectively counteracts the high electron–hole pair recombination rate in the bare SC and is expected to enhance the participation of electrons and holes in light-driven processes.

A recent study by Kamat et al.30 demonstrated that GO anchored on TiO2 can capture photogenerated electrons from the TiO2. The electrons are transferred by a hopping mechanism to the other side of the GO and can reduce the attached Ag+ ions to Ag NPs.30 GO is also known to effectively facilitate the electron transfer of several heme proteins.31 GO can function both as p- and n-type materials (electron and hole extraction and charge-carrier transport materials), which depends on (i) the dopant introduced,13, 23 (ii) the degree of oxidation (band gap),22, 23, 32, 33 and (iii) appropriate functionalization (charge neutralization).11, 21, 34, 35 This property of GO also promises its application as a material for electrodes.3639 The field-effect mobilities in GO range between 2–200 and 0.5–30 cm2 V−1 s−1 for holes and electrons, respectively, at room temperature.40

With the recent emphasis given to (i) photocatalysis driven by visible light (45 % of sunlight) and (ii) the development of alternative photocatalysts for UV-active TiO2,16 Bi-containing oxides, such as Bi2O2CO3,41, 42 Bi2WO6,43, 44 BiVO4,44 BiFeO3,45 Bi2Ti2O7,46 and Bi2Sn2O7,47 have gained attention because of their appealing optical and electronic properties. Bi2O3, in particular, has emerged as one of the strongest contenders for the most popular TiO2 photocatalyst.48 Bi is nontoxic in its oxide forms and, therefore, finds application in piezoelectric materials,49 biosensors,50 functional glasses, etc. The investigations available on the photocatalytic activities of different phases of Bi2O351, 52 demonstrate its superior performance. In combinations with noble metal NPs, such as Au-loaded α-Bi2O353, 54 and Ag-loaded β-Bi2O3,55 the activity is seen to improve. Recently, the first fabrication of a Bi2O3-based plasmonic photocatalyst (Pt/α-Bi2O3) has been reported.56

Plasmonic photocatalysts exploit the optoelectronic properties of both the plasmonic metal NPs and the SC.57 The interesting light absorption properties of plasmonic photocatalysts in the manipulation of visible radiation have been the subject of numerous recent investigations.5860 In plasmonic photocatalysts, noble metal NPs, such as Ag and Au, which exhibit a surface plasmon resonance (SPR) absorption band at around λ=410 and 520 nm, respectively, (i) undergo plasmonic excitation in presence of visible light and inject the plasmonic electrons into the suitably located conduction band of the SC6165 to generate holes on the metal surface that participate in dye degradation. (ii) The local electric field generated in the immediate vicinity of the metal NPs upon excitation at well-defined wavelengths enhances the rate of inherent electron–hole pair formation on the SC surface and suppresses electron–hole recombination, which can now participate in the oxidation–reduction processes.6165 (iii) The heat produced by the light-absorbing plasmonic NPs further promotes the oxidation of organic pollutants.6165

To the best of our knowledge, there is only one recent report of a graphene/Bi2O3 composite as an electrode material for supercapacitors.66 Therefore, we consider it important to undertake the green chemical synthesis of the new composites: GO/α-Bi2O3 and Ag/GO/α-Bi2O3 and to explore their versatile applications in dye degradation, light-energy conversion, and catalysis. We also attempt to provide a detailed insight into the mechanism of organic dye mineralization by considering the positions of the relative energy levels of the components in these heterostructured composites, the enhanced catalysis, and the underlying response of the improved photocurrent. Two recent feature articles by Tu et al.2 on the versatility of graphene/SC nanocomposites and by Chen and Caruso67 on the diverse applications of SC architectures motivate us further to undertake this multidimensional investigation of these new materials.

Results and Discussion

Surface morphology

The size and morphology of the synthesized photocatalysts were determined by SEM (Figure 1). α-Bi2O3 has a needle-like morphology (Figure 1 a), and the morphology of α-Bi2O3 loaded with Ag NPs (Ag/Bi2O3) is very similar (Figure 1 b). The length of the needles ranges from 5–25 μm, and the diameter is typically approximately 1 μm. The SEM images of the GO-hybridized α-Bi2O3 (GO/Bi2O3) composite clearly show randomly oriented Bi2O3 needles intercalated with transparent GO sheets (Figure 1 c and d). The GO sheets are highlighted (green) to emphasize their explicit intercalation and defined dimensions. The SEM images of the GO-hybridized α-Bi2O3 loaded with Ag NPs (Ag/GO/Bi2O3; Figure 1 e and f) are analogous to those of GO/Bi2O3. The original images that correspond to Figure 1 c–f are shown in the Supporting Information in Figure S1 a–d.

Figure 1.

SEM images of (a) α-Bi2O3 needles, (b) Ag-loaded α-Bi2O3 needles, (c) GO/Bi2O3, (d) GO/Bi2O3 at a higher magnification, (e) Ag/GO/Bi2O3, and (f) Ag/GO/Bi2O3 at a higher magnification. The intercalated GO sheets are highlighted in green.

Microscopic morphology

The high-magnification microscopic morphologies and structures of the Ag/GO/Bi2O3 sample were further investigated by TEM to observe the presence of Ag NPs (Figure 2).

Figure 2.

TEM images of Ag/GO/Bi2O3 that show (a) α-Bi2O3 needles and Ag NPs, (b–d) Ag NPs strongly attached on the surface of the α-Bi2O3 needle, (e) Ag NPs located on the surface of the GO sheets, and (f) near-spherical Ag NPs at a higher magnification.

An extended edge of a Bi2O3 needle is seen with dispersed Ag NPs near it and arranged on its surface (Figure 2 a). Near-spherical Ag NPs are clearly attached strongly to the α-Bi2O3 needle surface (Figure 2 b–d). Ag NPs anchored on crumbled GO sheets are evident in Figure 2 e, and the image shown in Figure 2 f further emphasizes the presence of near-spherical NPs. The size of the Ag NPs varies widely from approximately 2–70 nm (Figure S2). The electron cloud (free conduction band electrons) of Ag NPs of this size range is expected to exhibit collective electronic oscillations relative to the positive metal core induced by an interacting electromagnetic field (SPR). The successful incorporation of plasmonic Ag and its involvement in electron donation to GO and Bi2O3 can be proven by monitoring the GO phonon modes by Raman backscattering experiments (Figure 3).

Figure 3.

Raman backscattering spectra of (a) GO, (b) GO/Bi2O3, and (c) Ag/GO/Bi2O3. The inset shows the positions of the GO G phonon modes for comparison.

Raman spectroscopy

Raman spectroscopy provides a powerful probe to gain information on doping, the number of layers, defects, and other structural and electronic insights in these kinds of materials. The Raman spectra of GO and the different materials that incorporate GO are shown in Figure 3.

The Raman spectrum of GO (Figure 3 a) exhibits two prominent phonon modes. The D phonon mode is located at equation image=1330 cm−1, and the G phonon mode is located at equation image=1586 cm−1. The former arises from a breathing mode of κ-point phonons of A1g symmetry. It is related to disorder in the sp2-hybridized C atoms, functional groups at the edges, lattice distortions, and other defects.6870 The latter indicates the presence of isolated double bonds and arises from first-order scattering of the tangential stretching (E2g) phonon mode of sp2-hybridized C atoms.6870

The Raman spectra of GO/Bi2O3 (Figure 3 b) and Ag/GO/Bi2O3 (Figure 3 c) show additional phonon modes in the low-energy region, which can be assigned to α-Bi2O3 (Figure S3).71 For the GO/Bi2O3 composite, the GO D and G phonon modes are located at equation image=1338 and 1596 cm−1, respectively. After the incorporation of Ag in the Ag/GO/Bi2O3 composite, the GO G phonon mode is located at equation image=1591 cm−1 (the inset of Figure 3 provides a magnification of the G phonon mode frequency region).

In GO, p- and n-type doping cause opposite shifts of the G phonon mode. Electron acceptors or donors, for example, give rise to mid-gap molecular levels with the possibility of tuning the band gap region near the Dirac point.68 These molecules adsorbed onto the GO surface exhibit an effective charge transfer, which is reflected in the frequency shift of the G phonon mode. Electron-withdrawing organic groups (i.e., nitrobenzene) or p-type dopants offer hole-doping, cause the stiffening of the G phonon mode in GO, and shift this mode towards a higher frequency.72 Electron-donating groups in molecules adsorbed to GO (i.e., aniline) or n-type dopants offer electron-doping, cause softening of the G phonon mode, and shift it towards a higher frequency.73

In the GO/Bi2O3 composite, the shift of the G phonon mode from equation image=1586 cm−1 to equation image=1596 cm−1 provides convincing evidence of charge transfer from GO to p-type Bi2O3. Additionally, this blueshift provides clear evidence for the presence of anchoring (chemical bonding and hybridization) between Bi2O3 and GO sheets.69, 70 In the case of the Ag/GO/Bi2O3 composite, the G phonon mode experiences a redshift to equation image=1591 cm−1, which is a lower frequency relative to the GO/Bi2O3 composite. This is because Ag acts as an electron donor to GO. This provides direct evidence of n-type doping by Ag.69, 73, 74 Thus it is reasonable to say that the resultant G phonon mode frequency in the Ag/GO/Bi2O3 composite indicates the formation of heterostructured n/GO/p junction composites. This is further supported by the work functions of Ag (4.2 eV),73 GO (4.70 eV),11 and Bi2O3 (6.23 eV).75 As the work function of Ag is smaller than that of GO, electron transfer from Ag to GO sheets occurs by the creation of Ag/GO/Bi2O3 heterostructures. This is followed by electron transfer from GO to the conduction band of Bi2O3.

The ratio of the intensity of the D and G peaks (ID/IG) is proportional to the average size of the sp2 domains and thus is a measure of the degree of disorder in graphene-based materials. The ID/IG ratio of GO is 1.20. However, this value increases to 1.25 for the GO/α-Bi2O3 composite and is 1.50 for the Ag/GO/α-Bi2O3 composite. This indicates a decrease in the average size of the sp2 domains upon composite formation. The higher ID/IG ratio for the Ag/GO/α-Bi2O3 composite indicates the further introduction of defects and disorder in the GO sheets. This is a result of the deposition or trapping of some Ag NPs in the vacancies of the GO sheets because of the interaction with dangling bonds.

Crystal structure

The powder XRD patterns of the samples provide information about their crystal structure and are shown in Figure 4. The XRD pattern of GO (Figure 4 a) shows a broad prominent peak at approximately 2 θ=10.34°, which corresponds to the (0 0 2) reflection of stacked GO sheets with a calculated interlayer (d) spacing of 0.85 nm. This falls within the range of 0.75–0.94 nm reported for GO materials.13, 27 This calculated layer-to-layer distance is much larger than that obtained for the precursor graphite (0.34 nm), which further advocates the accommodation of various functional groups and H2O molecules between graphite lamellar structures after oxidation.27, 36

Figure 4.

XRD patterns of (a) GO and (b) GO/Bi2O3.

For the GO/Bi2O3 sample (Figure 4 b), all the XRD peaks correspond to that of monoclinic α-Bi2O3 (JCPDS: 41-1149). The absence or disappearance of any GO or RGO peaks is because of their low content. It is also an indication that during the hydrothermal reaction, the crystal growth of Bi2O3 between the layers of GO leads to the easy exfoliation of GO and destroys the regular stacking. Consequently, the Bragg diffraction peaks of GO cannot be recorded. The XRD pattern also illustrates that the presence of the GO sheets does not result in the development of new crystal phases of Bi2O3 or changes to the preferential orientations of the α-Bi2O3 needles.

The XRD pattern of the Ag/GO/Bi2O3 sample is similar to that of GO/Bi2O3 shown in Figure 4 b except for the presence of small quantities of Bi2O2CO3 (Figure S4). This is probably because Bi(NO3)3 in nitric acid solution undergoes hydrolysis to produce soluble BiONO3. The dissolved CO2 present in water also undergoes hydrolysis to CO32− ions that further react with BiONO3 to form Bi2O2CO3 according to Equation (1).76(1)

equation image(1)

Bi2O2CO3 is also a well-known SC photocatalyst, the energy level of the valence band (EVB) and the energy level of the conduction band (ECB) of which lie at 3.31 and 0.41 eV [vs. the normal hydrogen electrode (NHE)],41, 42 respectively. Although present in minor quantities, as evidenced in the XRD pattern (Figure S4), its role in the present photocatalytic system can envisaged to be similar to that of Bi2O3 because of the analogous positions of their energy levels. No diffraction peaks of Ag are observed because of its extremely low content.

Photocatalytic performance

The photocatalytic activities of the synthesized composites were determined by evaluating their activity in the decolorization and mineralization of the model non-biodegradable organic pollutants crystal violet (CV) and rhodamine 6G (Rh6G) under visible-light irradiation (300 W tungsten incandescent lamp), and the results of the time-dependent degradation are shown in Figure 5 a and b.

Figure 5.

Comparison of the photocatalytic activity of the different samples for the degradation of (a) CV and (b) Rh6G as a function of visible-light irradiation, in which C0 and Ct are the initial and actual concentration of the dyes at different reaction times (t), respectively.

Commercial TiO2 (Degussa P25) was chosen as the reference photocatalyst to compare the performance. The change in intensity of the main absorbance peak (A0) with time (At) at 590 nm for CV and 525 nm for Rh6G were monitored under visible-light irradiation. A typical sequence of the gradual disappearance of the characteristic absorption band of CV and Rh6G because of the photocatalytic degradation is provided in Figure S5.

Most photocatalytic reactions follow the Langmuir–Hinshelwood adsorption model,77 which can be simplified into a pseudo-first-order expression [Eq. (2)]:

equation image(2) )

in which C0 and Ct are the initial equilibrium concentration of the adsorbed dye and the concentration of the dye after illumination time, t, respectively, which is proportional to the normalized absorbance A0 and At, and k is the apparent rate constant. If we use regression-fitting techniques, the linear plots of ln(C0/Ct) versus irradiation time t are attained. The calculated rate (k), the corresponding correlation coefficient (R), and the reaction rate ratio are summarized in Table 1. The photocatalytic reaction rate of GO/α-Bi2O3 composite is four to six times greater than that of TiO2 P25 in the degradation of Rh6G and CV, respectively. With the incorporation of Ag, the photocatalytic reaction rate can be increased by up to eightfold over TiO2 P25. The possible mechanism of the photocatalytic degradation of organic dyes by considering the energy level diagrams of GO, α-Bi2O3, and Ag NPs is elucidated in Figure 6.

Table 1. Degradation rate constants of CV and Rh6G with different catalysts.


Linear equation

k [min−1]



for CV


TiO2 P25

ln(C0/Ct)=0.00690 t+0.14823





ln(C0/Ct)=0.01488 t−0.07138





ln(C0/Ct)=0.01550 t−0.00197





ln(C0/Ct)=0.03876 t+0.05252





ln(C0/Ct)=0.05640 t−0.01719




for Rh6G


TiO2 P25

ln(C0/Ct)=0.00413 t−0.08690





ln(C0/Ct)=0.00211 t+0.00363





ln(C0/Ct)=0.00316 t+0.02476





ln(C0/Ct)=0.01542 t+0.11998





ln(C0/Ct)=0.01578 t+0.05591




Figure 6.

Energy-level diagram in relation to the redox potentials and schematic illustration of the photocatalytic organic dye degradation for the Ag/GO/Bi2O3 composite that shows the three different possible simultaneous light-induced excitation pathways i) dye excitation, ii) Bi2O3 excitation, and iii) Ag plasmon excitation. This is subsequently followed by a) electron–hole formation and transfer, b) reactive species formation, and c) dye degradation.

Upon illumination, two possible electronic pathways may occur: (i) the SC is excited, and the electron is promoted to the conduction band (CB) to create a hole in the valence band (VB) [Eq. (3)] and (ii) the dye is excited [Eq. (4)].(3), (4)

equation image(3)
equation image(4)

The enhanced photocatalytic activity of Bi2O3 can be attributed to the favorably located energy levels of Bi2O3 that result in the easy formation of highly reactive hydroxyl radicals (.OH) that participate in the light-driven reactions. The O2 reduction potential for the single-electron reduction process depicted in Equation (5) is −0.56 V vs. NHE, whereas for the multielectron reduction process depicted in Equation (6) it is +0.40 V vs. NHE.52, 56 For Bi2O3, the ECB lies at +0.33 V vs. NHE, which is exactly in the potential range for the multielectron reduction of O2. Thus the electrons in the CB of Bi2O3 can be transferred effectively to the adsorbed O2 to form .OH. On the contrary, the band gap (Eg) of TiO2 is 3.0–3.2 eV and its ECB lies at −0.3 V vs. NHE. So, for TiO2, O2 is likely to be reduced by a single-electron process. Moreover, the holes in the valence band (h+VB) of Bi2O3 have enough positive potential to generate hydroxyl radicals according to Equation (7).(5), (6), (7)

equation image(5)
equation image(6)
equation image(7)

The higher efficiency of α-Bi2O3 may also be a result of the distorted Bi[BOND]O polyhedron structure induced by the 6 s2 lone pair of Bi3+. An asymmetric coordination environment facilitates the effective surface migration of carriers and retards exciton recombination.7880 Monoclinic BiVO4 (Eg=2.41 eV) displays a higher photocatalytic activity than tetrahedral BiVO4 (Eg=2.34 eV).80

Apart from the band gap, other important factors that govern the photocatalytic activity of the SC are the availability and number of surface active sites57 and oxygen vacancies.81 TiO2-P25 shows a lower activity than Bi2O3 for the degradation of CV. This is probably because of the loss of surface active sites of TiO2-P25 owing to surface poisoning effects after prolonged treatment (adsorption) with CV. On the contrary, we consider the degradation of Rh6G to result in the loss of surface active sites of Bi2O3, which reduces its overall efficiency relative to TiO2 in Rh6G degradation. The preferential adsorption of specific dyes over selective oxides is governed by various relative factors such as surface area, particle size of the adsorbent, pH, temperature, degree of ionization of the dye molecules in water, and surface charges,82 that is, the relative concentration of H+ and .OH on the surface of the SCs TiO2 and Bi2O3 in the present case.

In the absence of GO, the photosensitized dye* can inject photoexcited electrons directly into the CB of Bi2O3, which leads to reactive radical formation and results in a self-degradation pathway under visible-light irradiation. The holes can also combine with the dye to form unstable radical species [Eq. (8)]. Thus, the dye undergoes oxidation followed by degradation.(8)

equation image(8)

With the introduction of Ag, the photocatalytic activity of the composite is improved compared to bare Bi2O3 needles. Although Ag is widely recognized as an electron scavenger, in the case of the Ag/Bi2O3 composite, Ag cannot be a cocatalyst with its normal electron-scavenging function. In this case, the Ag NPs act as visible-light harvesting nanoantennas, which are small enough to exhibit plasmon photoexcitation and act as electron-generating centers. Metal particles such as Ag and Au have a Fermi level (EF) at 0.4 V vs. NHE.83 The band structure of the Bi2O3 supports the transfer of photogenerated plasmonic electrons from Ag to the CB of Bi2O3 (0.33 eV). Thus, the Ag/Bi2O3 composites support the model of a plasmonic photocatalyst. This is similar to the Pt/Bi2O3 system reported by Li et al.56 The photons absorbed by the Ag NPs are economically separated into electrons and holes. The electrons are transferred to the CB of Bi2O3, and the holes diffuse on the surface of the Ag NPs where they take part in the decomposition of the organic matter. Moreover, the deposition of minute crystalline Ag NPs on the surface of these Bi2O3 needles is expected to cause small changes in the surface microstructure of Ag/Bi2O3, improve the crystallinity, increase the number of available active sites, and thereby enhance dye adsorption.

The remarkable enhancement of the photocatalytic activity with the introduction of GO sheets is a result of the synergistic effects between Bi2O3 and GO. It has been reported that the ECB of GO does not change with the degree of oxidation. Thus GO samples with different oxidation levels have an ECB at approximately −0.75 V (vs. NHE).4, 84, 85 From the indirect transition energy plot of our GO samples, the band gap was determined to be in the range of 1.48–2.28 eV (Figure S6 and band gap determination for GO in the Supporting Information). Therefore, the EVB maximum is at 1.48 V (vs. NHE). The band structure of GO is constructed (Figure 6) based on the above ECB and EVB levels.

The GO sheets function as adsorption mats that provide a high surface area and result in enhanced dye adsorption. The strong affinity of GO to the aromatic dye molecules is because of π–π interactions as well as ionic and dipolar interactions. In addition, GO can be considered to possess some ionized carboxylic acid groups. Therefore, cationic dye molecules can also be adsorbed by electrostatic forcefield interactions.26, 86

The GO sheets act as electron scavengers that accept photoexcited electrons from the excited dye, which are delocalized over their π-orbitals. There is convincing evidence in the literature to support electron transfer from photochemically generated dye radicals to GO.8789 Thus, a high electron density develops over the GO sheets. The electrons are then transferred to the CB of Bi2O3. This transfer process, if GO and Bi2O3 are in contact, is supported by the Raman spectra (Figure 3) and the literature on work-function values. A recent density functional calculation study has shown that charge transfer from excited graphene to the CB of rutile titania is possible and results in hole-doping in graphene.90 This behavior may be viewed as similar to that of TiO2/carbon nanotube (CNT) hybrids in which electrons from the sensitizer CNTs are injected into the CB of TiO2.91

In the case of Ag/GO/Bi2O3, GO also acts as an electron-shuttle to transfer plasmonically excited electrons from Ag to Bi2O3. This information can be directly derived from our Raman spectra (Figure 3). In a very recent report on the photocatalytic activity of CNT/Ag/AgBr composite, CNTs were shown to exhibit a similar electron-shuttling and transfer function to promote electron–hole pair separation.64

The resultant high concentration of electrons finally available in the CB of Bi2O3 favors multielectron reduction processes. Some electrons available in the CB of GO may also participate in the single-electron reduction process of O2. The holes generated in the VB of Bi2O3 upon visible-light excitation are transferred readily to the VB of GO. Consequently, the lifetimes of the holes are increased and they can participate in the dye-mineralization process. Moreover, the high concentrations of the adsorbed dye molecules, preferentially over the GO surface, can increase the rate of reaction with the aid of photogenerated reactive species.

Structural modification after photocatalysis

FTIR spectroscopic studies provide convincing information on the structure of the photocatalysts. The FTIR spectrum of GO sheets is shown in Figure 7 a, which confirms the presence of different oxygen-containing functionalities that are responsible for anchoring different particles of Bi2O3 and Ag. For GO, the intense absorption band at equation image=3391 cm−1 is because of the hydroxyl O[BOND]H stretching vibration. The band at equation image=1732 cm−1 indicates the C[DOUBLE BOND]O stretching vibration of the COOH groups. The band at equation image=1618 cm−1 arises as a result of skeletal vibrations from unoxidized graphitic domains. The band at equation image=1406 cm−1 originates from the tertiary C[BOND]OH stretching vibration, and the band at equation image=1030 cm−1 indicates a C[BOND]O stretching vibration.92 The set of bands below equation image=600 cm−1 arises from Bi[BOND]O vibrations in Bi2O3 (Figure 7 b).93, 94

Figure 7.

FTIR spectra of (a) GO, (b) α-Bi2O3 needles, (c) Ag/GO/Bi2O3, (d) dried Ag/GO/Bi2O3 photocatalyst recovered after the photodegradation of CV, and (e) dried Ag/GO/Bi2O3 photocatalyst recovered after the photodegradation of Rh6G.

The GO sheets act like fishing nets to absorb large quantities of the dye from the solution. Therefore, to confirm that the reduction of the pollutant concentration is essentially from photocatalytic degradation, FTIR spectra of the photocatalyst Ag/GO/Bi2O3 were determined before (Figure 7 c) and after the photocatalytic reactions (Figure 7 d and e).

The FTIR spectrum of the Ag/GO/Bi2O3 hybrid composite Figure 7 c contains bands that arise from both Bi[BOND]O vibrations of Bi2O3 and bands from the vibrations of the different functional groups of GO. The small band at equation image=846 cm−1 in the spectrum of Ag/GO/Bi2O3 may be a result of the presence of a small amount of Bi2O2CO3 formed during the synthesis of the material as mentioned above [Eq. (1)]. This ν2 mode CO32− band at equation image=846 cm−1 is intensified in the spectrum of Ag/GO/Bi2O3 collected after the photocatalytic reactions (Figure 7 d and e). In addition, strong bands appear at equation image=1380 and 1472 cm−1 (Figure 7 d and e), which can be assigned to the ν3 vibrational mode of the CO32− group.41, 93, 95

CO2 is generated as one of the end products of the mineralization of the organic compounds.15, 42, 77 In the presence of .OH, which is also generated during the photocatalytic process, CO2 forms bicarbonate species in solution that further react with Bi2O3 with prolonged irradiation to form (BiO)4CO3(OH)2 and finally Bi2O2CO3. These are possibly deposited on the surface sites of the Bi2O3. The transformation from Bi2O3 to Bi2O2CO3 in water that contains CO2/HCO3 has been observed by Huang et al.93 Moreover, photoinduced structural transformations of oxide SCs into carbonates during the photodegradation of methyl orange azo dye have been observed by Yang et al.96 Marinho et al.97 observed the transformation of Bi2O2CO3 back to Bi2O3 upon calcination. Thus, the time, irradiation, and pH-dependent structural modification of the photocatalyst, the influence of intermediate reaction products during the photocatalytic reaction and, calcination-induced activation and recyclability, which are significant with regard to the performance of the Bi2O3-based photocatalysts, demand further intensive investigation.

Catalytic performance

Noble metal NPs possess excellent catalytic properties and are known to catalyze the reduction of 4-nitrophenol (4-NP) to 4-aminophenol (4-AmP) in the presence of NaBH4 at 0 °C.98 As SC NPs are generally inactive towards this reaction, therefore, the presence of Ag (metal) doping in Ag/Bi2O3 and Ag/GO/Bi2O3 composites is expected to be indicated by a positive catalytic response.

Under neutral or acidic conditions, 4-NP exhibits an absorption peak at λ=317 nm. The addition of NaBH4 results in the formation of 4-nitrophenolate ions that result in the redshift of the absorption peak to approximately λ=400 nm. As the metal NPs catalyze the formation of 4-AmP, the intensity of the 4-nitrophenolate absorption peak at λ=400 nm gradually decreases and a new peak emerges at λ=298 nm, which corresponds to 4-AmP.71, 98

However, in the present case, NaBH4 reduces the Bi2O3 needles to metallic Bi0 needles and the GO to reduced graphene oxide (RGO) or graphene. In our previous study, we demonstrated for the first time that Bi particles (microhexagons) display positive catalytic activity towards this reaction.71 Consequently, the catalytic activities were recorded for all samples. This activity is a result of the in situ chemical reduction of α-Bi2O3, Ag/Bi2O3, GO/Bi2O3, and Ag/GO/Bi2O3 to Bi0, Ag0/Bi0, RGO/Bi0, and Ag0/RGO/Bi0 composites by NaBH4. A representative time-dependent absorption spectrum that shows the conversion of 4-NP into 4-AmP is shown in Figure S7.

The rates of the catalytic activities were evaluated for these pseudo-first-order reactions by plotting ln(At/A0) versus time (t), in which At and A0 are the absorbances at λ=400 nm at times t and 0 (Figure 8). The rate or activity is found to increase in the order: Bi0 <Ag0/Bi0<RGO/Bi0<Ag0/RGO/Bi0 (by 1<2.5<7<9.7-fold; inset in Figure 8). Bi and Ag form a low-melting eutectic mixture with 2.5 % Ag.99 Therefore, in the formation of Ag-Bi, some alloy even with a very low concentration of Ag (2 %) with the surface segregation of Ag atoms under the present experimental conditions cannot be overruled. Interestingly, even after reduction, the specific morphologies of the composites are retained, as seen from the SEM image presented in Figure S8.

Figure 8.

Plot of ln(At/A0) versus time and evaluation of the rate constants for the catalytic reduction of 4-NP by NaBH4 in the presence of different composites, in which A0 is the initial normalized absorbance and At is the absorbance after time (t). The inset shows a comparative bar graph to elucidate the different rates for different samples.

The high activities of the RGO-doped composites are because the intercalated graphene sheets act as adsorption mats for 4-NP through π–π stacking interactions. This enables a high concentration of 4-NP around Bi and Ag particles for effective reduction. In the absence of graphene mats, the reaction rate is slower as it depends on the probability of the collision of the 4-NP molecules with the metal particles. 4-NP molecules that do not collide with the metal particles can pass back into solution.100, 101

This reaction gives an indication that although noble metal particles (Ag) are catalytically more active than heavy metal particles (Bi), the introduction of graphene has a dramatic effect. Composite materials of Bi nanometal and its intermetallic alloys (Bi-Ag) incorporated on RGO sheets have superior activity and deserve further investigation because of their potential to replace noble metal catalysts.

Performance as photoanodes

Dye-sensitized solar cells (DSSCs) have emerged as the most economical assembly for indoor photovoltaic applications and educational purposes. To study the performance of these composite materials as photoanodes in DSSCs, we used TiO2 as a secondary binding agent (Experimental Section). The advantages of TiO2 that make it suitable for DSSC applications are (i) the position of its ECB with respect to the redox potential of the well-known I3/I system (0.4 V vs. NHE), which enables the achievement of high open-circuit photovoltage (Voc) values.102 The Voc is related to the energy difference between the quasi-Fermi level of the electrons in the SC and the chemical potential of the redox mediator in the electrolyte (depicted schematically in Figure 10). A high Voc value can help to achieve a high fill factor (FF) and thereby a high photoconversion efficiency (η). (ii) TiO2 has a higher isoelectric point (IEP) than Bi2O3. The pH(IEP) of TiO2 is approximately 5.2,103 whereas that of Bi2O3 is approximately 3.2.104 The IEP is the pH value at which the metal oxide surface carries no net electric charge and is an important surface property that determines the dye–metal oxide interface in the DSSC. The metal oxide surfaces bear a net positive charge at a pH below the IEP and a net negative charge at a pH above the IEP. Particles with high pH(IEP) values show high dye-loading abilities.105 Bi2O3 needs a longer time for hibiscus red anthocyanin dye loading than TiO2.

The photoinduced photocurrent density–voltage (JV) curves are presented in Figure 9. These are used to evaluate the energy conversion properties of the DSSCs. The photovoltaic properties of the different photoanodes are tabulated in Table 2.

Figure 9.

JV curves for different photoanodes. In all three cases of Bi2O3, GO/Bi2O3, and Ag/GO/Bi2O3, the remaining 10 % (by weight) was commercial TiO2 paste, which acts as an auxiliary binding agent (see photoanode preparation in the Experimental Section).

Table 2. Photovoltaic properties of the four different photoanodes.


Jsc [mA cm−2]

Voc [mV]


η [%]

  1. [a] See Experimental Section and calculation of FF and η values.

commercial TiO2 paste (reference)





α-Bi2O3:TiO2 (9:1)





GO/α-Bi2O3:TiO2 (9:1)





Ag/GO/α-Bi2O3:TiO2 (9:1)





The DSSC with the Bi2O3-containing electrode shows a short-circuit photocurrent density (Jsc) of 0.064 mA cm−2 and an overall energy conversion efficiency (η) of 0.013 %. The introduction of GO leads to an enhancement in both Jsc and η by 14 and 20 %, respectively, with respect to the TiO2 photoanode (Table 2). The presence of 2 wt % Ag NPs in the Ag/GO/Bi2O3 electrodes further increases Jsc and η by ∼4 and ∼6 %, respectively, in comparison with the GO/Bi2O3 anode. The main reason for the rise in the η value is because of the increase in the FF and Jsc. It is, therefore, interesting to note that both anodes, GO/Bi2O3 and Ag/GO/Bi2O3, display better performances than the commercial TiO2 anode.

The fundamental working mechanism of our DSSCs is presented in Figure 10. It is expected that in the anode, the electrons are injected from the LUMO of the sensitized anthocyanin dye into the CB of TiO2, followed by electron injection into the CB of Bi2O3. The electrons are transported through the SC layers by diffusion to reach the conducting indium tin oxide (ITO) glass substrate and thereby to the outer circuit. This electron flow is assisted by the partially reduced GO sheets. At the carbon-coated counter electrode, the oxidized species of the redox electrolyte, that is, I3 ions in the iodide/triiodide complex, are reduced to I. The I ion now donates an electron to the oxidized dye (D+) and regenerates the dye molecule. These processes cycle to result in a continuous current flow through the external circuit as long as light is incident on the cell.

Figure 10.

Schematic representation of the fundamental working principle of our DSSC. The inset shows the absorption spectrum of the anthocyanin dye. After excitation of the dye (HOMO→LUMO), the electrons can be transferred sequentially to the ECB of TiO2, to the ECB of Bi2O3, and to the anode. This process is assisted by the partially reduced GO sheets. The transfer continues to the cathode and finally to the redox couple, which thus completes the cycle with an idealized 100 % quantum yield. R1, R2, R3, and R4 are the competing electron recombination processes that dramatically reduce the obtainable efficiency.

The possibility of the improvement of Jsc and performance by the incorporation of graphene-based structures is noticeable. Firstly, GO causes an increased absorption of dye molecules. More dye molecules are expected to harvest more light energy. Secondly, GO provides efficient electron transfer pathways to the ITO surface. Thus the electrons are rapidly collected before they are recombined. The resistance of GO sheets is usually approximately 1×105 Ω □−1.106 TiO2 films may exhibit resistances of approximately 107 Ω □−1,107 and Bi2O3 films have a resistance of approximately 106 Ω □−1.108

There is considerable scope for the improvement of the Jsc in DSSCs by designing thin layers and structures that reduce the electron diffusion length and improve carrier mobilities, which depend on the film morphologies.109 The carefully designed thin layers would result in less charge depletion and back-electron transfer (recombination processes R1, R2, R3, and R4 in Figure 10). Finally, the use of competent dyes110 to improve the interfacial contact between TiO2 and Bi2O3 and their crystallinity, and decrease the resistivity of GO by their selective conversion into graphene111, 112 intercalated sheets is expected to improve their behavior.


We have presented a facile, single-step, green synthetic strategy to prepare multifunctional graphene oxide (GO)/α-Bi2O3 and Ag/GO/α-Bi2O3 hybrid composites. Our strategy circumvents the use of any toxic precursors. We have fully characterized these heterostructured composites by SEM, XRD, and Raman spectroscopy and exploited their applications towards photocatalysis, solar energy harvesting, and catalysis.

The photocatalytic activities of Bi2O3 in the degradation of organic pollutants (crystal violet and rhodamine 6G) are enhanced by four to six times if hybridized with only 1.0 wt % of graphene-based structures compared to commercial TiO2-P25. With the incorporation of Ag (2.0 wt % Ag), the photocatalytic efficiency further increases by over eightfold. We discuss the photocatalytic degradation mechanism in view of plasmonic photocatalysis. We propose that the photogenerated electron–hole pairs are formed on the surface of the minute Ag particles owing to plasmonic excitation under visible-light irradiation. These electrons are instantly transferred to the favorably located energy levels of the conduction band in Bi2O3. The small amount of GO sheets is highly beneficial for the photogenerated electron transfer from Ag to Bi2O3 as evidenced by Raman spectra. The GO sheets also allow the strong adsorption of aromatic dye molecules and, therefore, enhance the photocatalytic activity immensely.

We have further extended the applications of these hybrid composites in (i) catalysis and (ii) dye sensitized solar cells (DSSCs). GO/Bi2O3 composites can be reduced to RGO/Bi0 composites by NaBH4. Reduced graphene oxide (RGO)/Bi0 shows catalytic performance in the conversion of 4-aminophenol to 4-nitrophenol, which is increased approximately seven times compared to that of metallic Bi0 with a similar needle morphology. This enhancement is possibly a result of a synergistic effect. The catalytic performances of Ag0/Bi0 and Ag0/RGO/Bi0 composites have also been compared, and a very small amount of noble metal (2 wt % of Ag) in Ag0/RGO/Bi0 is the most efficient of these. Both the short-circuit photocurrent density and overall energy conversion of DSSCs with GO/Bi2O3 and Ag/GO/Bi2O3 electrodes increase by ∼20 % compared with commercial TiO2 electrodes. The electronic interaction between GO and Bi2O3 is considered to be mainly responsible for the enhanced photocurrent conversion.

For future research, we essentially emphasize the advantages of using Bi2O3-based ecofriendly materials as an efficient alternative to TiO2-P25 with regard to visible-light-driven photocatalysis. The possibility to use Bi2O3 as a plasmonic photocatalyst in combination with metal nanoparticles such as Ag is also highlighted. The band positions of Bi2O3 favor the fabrication of plasmonic photocatalysts. At the same time, we have demonstrated the possibility to use Bi2O3 as an electrode material in DSSCs in combination with TiO2. The photocurrent generation and power conversion efficiency are enhanced by 14 and 20 %, respectively, upon incorporation of GO. Finally, we draw the attention of readers towards the partial replacement of noble metal catalysts. Catalysis with heavy metal (Bi)/graphene composites has been shown to possess significant potential in the emerging research area of “reduced usage of noble metals in catalysis”. Overall, both GO and graphene have been demonstrated to act as an efficient promoter in photocatalysts, photoanodes, and catalysts.

Experimental Section

Synthesis of GO

The precursor GO was synthesized by a modified Hummers’ method. Graphite (0.5 g) was oxidized with NaNO3(0.5 g), 98 % H2SO4 (23 mL; J. T. Baker), and KMnO4 (3 g) in an ice bath. Water (46 mL) was added slowly, and the mixture was kept at 95 °C for 1 h under continuous stirring. The unreacted KMnO4 was removed by the addition of H2O2 (10 mL). The oxidized graphite was purified by washing with 10 % HCl (J. T. Baker) and a solvent mixture (H2O/EtOH 1:5) and dried at 60 °C under vacuum. Exfoliation was accomplished by sonicating an aqueous solution of GO for 120 min.

Synthesis of hybrid composites

The α-Bi2O3 needles and GO/α-Bi2O3 composite were prepared by the hydrolysis of a solution of Bi(NO3)3⋅5 H2O (0.5 m; Aldrich) in HNO3 (1 m; J. T. Baker) in combination with an acidic solution of exfoliated GO (1 wt %) in NaOH (2 M) solution at 80 °C [Eq. (9)].(9)

equation image(9)

The Ag/α-Bi2O3 and Ag/GO/α-Bi2O3 composites were prepared by heating the undoped composites in AgNO3 solution (2 wt % Ag) at 80 °C under vigorous stirring for 2 h [Eq. (10)].(10)

equation image(10)

This forms the underlying principle of the O2 evolution reaction by a SC photocatalyst from an aqueous AgNO3 solution.113 The Ag-undoped nanocomposites were repeatedly washed prior use with diluted HNO3 (1 %) solution followed by water to make them NaOH free. This is because adhered NaOH may undergo a double-decomposition reaction with AgNO3 to produce AgOH and thereby Ag2O upon warming [Eqs. (11) and (12)].(11), (12)

equation image(11)
equation image(12)

Ag2O is photocatalyst that is known to decompose to metallic Ag only at temperatures ≥330 °C or under irradiation [Eq. (13)].(13)

equation image(13)


The UV/Vis absorption spectra were recorded by using an HR2000 UV/Vis spectrophotometer (Ocean Optics, Germany). SEM images were obtained by using a TM-1000 tabletop-SEM (Hitachi, Japan). Samples for TEM observation were prepared by dropping the products on a carbon-coated copper grid after ultrasonic dispersion in ethanol/water (1:1) and allowed them to dry in air before analysis. TEM measurements were by using a LIBRA 200FE (Carl Zeiss AG, Germany) instrument operated at 200 kV. XRD measurements were performed by using a D8 Discover diffractometer (Bruker AXS, Karlsruhe, Germany) with CuKα1 radiation (α=1.5374 Å) with an operating current of 40 mA, a voltage of 40 kV, and a scanning rate of 5 s per step. Micro-Raman measurements were obtained by using a Princeton Instruments spectrometer (Roper Scientific, Germany) equipped with a low-power laser (5 mW) red 633 nm monochromatic excitation source. FTIR spectra were recorded by using a VERTEX 70 spectrometer (Bruker, Germany) with an attenuated total reflectance (ATR) diamond unit.

Photocatalytic activity evaluation set-up

To determine the photocatalytic activity, photocatalyst (0.2 g) was added to an aqueous solution (50 mL) of the dyes CV (Aldrich) and Rh6G (Aldrich), which each had a concentration of 1×10−5 mol L−1, contained in a closed borosilicate container. Prior to irradiation, the suspensions were magnetically stirred in the dark for 30 min to ensure adsorption/desorption equilibrium. Thereafter, the borosilicate glass photoreactors were exposed to a 300 W tungsten incandescent lamp as the illumination source. The spectral range of a tungsten filament lamp is 350–2500 nm and consists mainly of a large part of the visible spectrum. The dye solutions were maintained under ice-cold conditions to prevent evaporation or thermal decomposition. Commercial TiO2-P25 (20 % rutile and 80 % anatase) nanopowder (Degussa Co., Ltd., Germany) was used a benchmark (reference) photocatalyst.

Catalytic performance evaluation

To evaluate the catalytic activity, an aqueous solution of 4-nitrophenol (10 mL, 1.0×10−4M) and each of the samples (0.0075 g) were mixed followed by the addition of aqueous NaBH4 solution (5 mL, 6.0×10−2M) to the reaction mixture under ice-cold conditions and time-dependent absorption spectra were recorded.

Photoanode preparation

To prepare the photoanodes, excess quantities of the composites with a commercial TiO2 paste (Man Solar) in a weight ratio of 9:1 we mixed mechanically. The commercial TiO2 paste contained a transparent conducting polymer and a binding agent that help to improve the film quality and stability. After the films were coated onto ITO glass surfaces (surface resistivity R=28–30 Ω cm−2), they were heated at 350 °C for 40 min. Commercial carbon paste (R=110–120 Ω cm−2) coated onto an ITO surface was used as the cathode. A natural flavonoid anthocyanin dye extracted from hibiscus flowers, which can absorb visible light in the range of λ=475–650 nm with a characteristic absorption peak at λmax=540 nm, was used directly as a sensitizer. The I3/I couple in acetonitrile was used as the electrolyte. The active areas of the photoanodes were approximately 6 cm2. The active area of illumination was maintained at (1000±10) W m−2 (by using a PL-110SM Voltcraft solar radiation measuring instrument, Conrad Electronics SE Hirschau) with an artificial indoor lamp. The power conversion efficiency of the DSSC (η) is the product of three terms: the Jsc, the Voc, and the FF divided by the incoming incident solar power (Pin). The η values are evaluated according to Equation (14):

equation image(14)

The FF value is defined as the ratio of maximum power output (Jmax×Vmax) and the product of Jsc and Voc [Eq. (15)].(15)

equation image(15)

If we maximize any of the numerator terms in Equation (14), η will be increased. From the cross-sectional SEM images, the thickness of the films prepared from the composite materials was verified to be approximately (12±2) μm.


T.S. thanks the Alexander von Humboldt Foundation, Bonn, Germany for a postdoctoral fellowship and research grants. We thank Dr. Kallol Ray, Junior Research Group Leader at the Humboldt University, Berlin, within the Uni-Cat cluster of excellence for his support with the photocatalytic experimental set-up and Dr. Gregor Meier, Inorganic Chemistry and Homogeneous Catalysis Group, Humboldt University, Berlin for his help with FTIR measurements.