Synthesis and characterisation of heteroatom-doped reduced graphene oxide/bismuth oxide nanocomposites and their application as photoanodes in DSSCs

Abstract

Semiconductor materials have been recently employed in photovoltaic devices, particularly dye-sensitized solar cells (DSSCs), to solve numerous global issues, especially the current energy crisis emanating from the depletion and hazardous nature of conventional energy sources, such as fossil fuels and nuclear energy. However, progress for the past years has been mainly limited by poor electron injection and charge carrier recombination experienced by DSSCs at the photoanode. Thus, novel semiconductor materials such as bismuth oxide (Bi₂O₃) have been investigated as an alternative photoanode material. In this study, Bi₂O₃ was integrated with nitrogen- or boron-doped reduced graphene oxide (N-rGO or B-rGO, respectively) via a hydrothermal approach at a temperature of 200 °C. Various instrumental techniques were used to investigate the morphology, phase structure, thermal stability, and surface area of the resulting nanocomposites. The incorporation of N-rGO or B-rGO into Bi₂O₃ influenced the morphology and structure of the nanocomposite, thereby affecting the conductivity and electrochemical properties of the nanocomposite. B-rGO/Bi₂O₃ exhibited a relatively large surface area (65.5 m² g⁻¹), lower charge transfer resistance (108.4 Ω), higher charge carrier mobility (0.368 cm² V⁻¹ s⁻¹), and higher electrical conductivity (6.31 S cm⁻¹) than N-rGO/Bi₂O₃. This led to the fabrication of B-rGO/Bi₂O₃ photoanode-based DSSCs with superior photovoltaic performance, as revealed by their relatively high power conversion efficiency (PCE) of 2.97%, which outperformed the devices based on N-rGO/Bi₂O₃, rGO/Bi₂O_3, and Bi₂O₃ photoanodes. Therefore, these results demonstrate the promising potential of heteroatom-doped rGO/Bi₂O₃-based nanocomposites as photoanode materials of choice for future DSSCs.

---

1 Introduction

Recent developments in technology have led to a tremendous demand for energy. Thus, renewable energy sources, especially solar energy, which can be converted to electrical energy by photovoltaic devices, particularly dye-sensitized solar cells (DSSCs), have been favoured by many researchers.^1–5 The main component of DSSCs is the photoanode, which is responsible for transferring photogenerated electrons from the excited dye molecules to the collecting electrode. Thus, a high electron-transport rate reduces the charge recombination rate and improves the power conversion efficiency (PCE). The development of DSSCs by means of tailoring the properties of semiconductor nanomaterials to suit photoanode applications has attracted a lot of attention in solar energy research. This is because semiconductor nanomaterials have unique physical, chemical, electronic and optical properties. Semiconductor photoanode nanomaterials with a high surface area can facilitate more dye absorption, effective photon harvesting, and fast charge transport, thereby improving the PCE of DSSCs. Various approaches, such as doping,⁶ surface modification,⁷ and the use of hybrid semiconducting nanomaterials,^8–12 have been proposed to improve the properties of semiconductor materials. Semiconductor nanomaterials such as metal oxides, including titanium dioxide (TiO₂),¹³ zinc oxide (ZnO),¹⁴ zirconium dioxide (ZrO₂),¹⁵ tungsten trioxide (WO₃),¹⁶ and tin(IV) oxide (SnO₂),¹⁷ have been extensively explored as photoanode materials in DSSCs.

To date, there are few studies based on the application of bismuth oxide (Bi₂O₃) as a photoanode in DSSCs, yet Bi₂O₃ has attracted great interest in photocatalysis¹⁸ and supercapacitors¹⁹ due to its unique crystal structure,²⁰ non-toxicity,²¹ and wide bandgap, ranging from 2.47 to 3.55 eV.²² Bi₂O₃ also exhibits a high dielectric permittivity, refractive index, photoconductivity, and oxygen ion conductivity. There are four polymorphs of Bi₂O₃, namely, body-centred cubic (γ), face-centred cubic (δ), tetragonal (β) and monoclinic (α).²³ Among these, α-Bi₂O₃ and δ-Bi₂O₃ are thermodynamically stable at ambient conditions, whereas β-Bi₂O₃ and γ-Bi₂O₃ are metastable, and are stabilised by the addition of impurities or by controlling the reaction conditions.²⁴ An investigation on the effect of β-Bi₂O₃ as a photoanode in DSSCs has been reported by Shaikh et al.²⁵ and resulted in a PCE of 0.078%. However, a slight improvement in the PCE to 1.5% was observed when β-Bi₂O₃ was incorporated with ZnO. Fatima et al.²⁶ reported a PCE of 0.05% for DSSCs with α-Bi₂O₃-based photoanodes. The low PCE was attributed to the presence of trapping states at the interfaces within the photoanode. Thus, more research on Bi₂O₃ nanomaterials for photoanode applications in DSSCs is still required.

Recently, metal oxides have been integrated with an allotrope of carbon, especially graphene, and used as positive electrode material in energy storage devices.^27,28 Graphene has attracted significant interest due to its high charge carrier mobility, large specific surface area, high mechanical strength, high electrical conductivity, and chemical resilience. Also, graphene is reported to display predominant energy storage performance compared with other carbon nanostructures, such as carbon nanotubes and graphite.^29,30 Graphene oxide (GO) is defined as a cutting edge of graphene functionalised with carbonyl and carboxyl groups at the edges; and epoxy and hydroxyl groups in the basal plane.³¹ The oxygen-based functional groups on the surface of GO are responsible for integrating GO with metal oxides, resulting in GO-based metal oxide nanocomposites.

The main focus in the reported studies of Bi₂O₃ was to integrate Bi₂O₃ with reduced graphene oxide (rGO) due to the distinctive mechanical, electrical, and electrochemical properties of rGO.^32–34 This is envisaged to help overcome the drawbacks of pristine rGO, such as its usual tendency to agglomerate or restack to form graphite due to π–π stacking and van der Waals forces when the rGO nanosheets are close to each other,³⁵ resulting in poor electrochemical properties. Nitrogen-doped reduced graphene oxide (N-rGO) and boron-doped reduced graphene oxide (B-rGO) have been reported to have improved electrochemical properties than rGO;³⁶ thus, integrating Bi₂O₃ with N-rGO or B-rGO is expected to yield favourable device performance. However, to the best of our knowledge, no detailed studies have reported the electrochemical properties of heteroatom-doped rGO (N-rGO or B-rGO) integrated with Bi₂O₃ and their application as photoanodes in DSSCs. Hence, in this study, the effect of nitrogen or boron on the physicochemical and electrochemical properties of heteroatom-doped rGO/Bi₂O₃ nanocomposites was investigated for the first time by means of various instrumental techniques, so as to enhance the electrochemical properties of Bi₂O₃, reduce the aggregation of rGO, and subsequently improve the performance of DSSCs.

2 Experimental

2.1 Materials

Bismuth oxide (Bi₂O₃, 99.99%), 4-nitro-o-phenylenediamine (≥99%), boric anhydride (≥98%), boron standard solution (999.5 mg L⁻¹ ± 20 mg L⁻¹), bismuth standard for ICP traceCert® (1000 mg L⁻¹ Bi in nitric acid), potassium hydroxide (KOH, ≥85%, pellets), Nafion (≤100%), lithium iodide (99.9%), 4-tert-butylpyridine (98%), guanidinium thiocyanate (99%), 1-methyl-3-propylimidazolium iodide (99.99%), acetonitrile (≥99.9%), poly(vinyl acetate) (99.9%), absolute ethanol (99.5%), eosin B (97%) and indium tin oxide (ITO) coated glass slides (15 Ω, 30 × 30 × 0.7 mm) were purchased from Sigma-Aldrich, South Africa.

2.2 Synthesis of the nanocomposite

rGO was synthesised via thermal annealing of GO at a temperature of 600 °C under 10% H₂ in argon,³⁷ while N-rGO or B-rGO was synthesised by thermal treatment of GO and a precursor of 4-nitro-o-phenylenediamine or boric anhydride, respectively, at a temperature of 600 °C.^38,39 Bi₂O₃ (0.2 g) and rGO, N-rGO, or B-rGO (0.5 g) were dispersed in absolute ethanol (60 mL) and sonicated for 2 h (Scheme 1). After sonication, the mixtures were transferred into a Teflon-lined stainless-steel autoclave and thermally treated at a temperature of 200 °C for 24 h. The nanocomposites were purified by washing them with double-distilled water, followed by drying the samples in an oven for 24 h at 60 °C before characterisation.

Scheme 1 Schematic diagram of the conversion of rGO, N-rGO and B-rGO to Bi₂O₃-based nanocomposites.

2.3 Device fabrication

The Bi₂O₃-based nanocomposite (100 mg) was dispersed in absolute ethanol (0.3 mL), sonicated, and deposited onto an ITO glass of the photoanode by using the Doctor Blade method. The photoanode was then annealed at 300 °C for 30 min. After thermal annealing, 0.3 mM eosin B dye (150 μL) was loaded onto the photoanode. The iodide/triiodide gel state electrolyte was stained onto the dye-coated photoanode, followed by assembling the aluminium-coated cathode in a sandwich-like fashion. The iodide/triiodide gel state electrolyte was prepared by firstly synthesizing the liquid electrolyte. The liquid electrolyte was synthesised by mixing lithium iodide (0.3348 g), 4-tert-butylpyridine (1.6900 g), guanidinium thiocyanate (0.2954 g) and 1-methyl-3-propylimidazolium iodide (0.3173 g) and dissolved in acetonitrile (5 mL). The synthesised liquid electrolyte (1.2484) was mixed with poly(vinyl acetate) (0.6022 g) and stirred using a glass rod until a gel state electrolyte was formed, and further kept in a refrigerator at 0 °C.

2.4 Characterisation

Field emission-scanning electron microscopy (FE-SEM, Carl Zeiss Ultra Plus) was used to investigate the surface morphology of the nanocomposites. Transmission electron microscopy (TEM, JEOL JEM, 1010 model) was used to investigate the microstructure of the nanocomposites. The presence of various elements in the nanocomposites was investigated by elemental analysis (Elementar vario EL cube CHNSO elemental analyser), and Bi₂O₃ was quantified via inductively coupled plasma-optical emission spectrometry (ICP-OES, PerkinElmer Optima 5300 DV). The presence of various functional groups was determined by Fourier transform infrared spectroscopy (PerkinElmer Spectrum 100, FTIR spectrophotometer, with an attenuated total reflectance (ATR) accessory).

The phase compositions present in the nanocomposites were evaluated by means of powder X-ray diffraction (XRD, Bruker AXS, Cu K_α radiation source, λ = 0.154 nm). The graphitic nature of the nanocomposites was investigated by a Renishaw inVia Raman microscope at an excitation wavelength of 488 nm. The thermal stability was analysed with a TA Instruments Q series™ thermal analyser DSC/TGA (Q600) instrument. The textural characteristics were determined using a Micromeritics Tristar II 3020 surface area and porosity analyser. The electrical conductivity of the nanocomposites was investigated by a four-point probe (Keithley 2400 source-meter). Hall effect measurements were carried out using an Ecopia Hall effect measurements system, model HMS 3000, equipped with HMS3000 VER 3.15.5 software.

Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were carried out on a VersaSTAT3F potentiostat/galvanostat electrochemical workstation, with ZSimpWin software utilised for EIS data analysis. A counter electrode (platinum wire (Pt)), a reference electrode (Ag/AgCl), and a working electrode (mixture of nanocomposite) were used. A mixture of the nanocomposite and a binder (Nafion) were dispersed in absolute ethanol and cast onto the electrode. The potassium hydroxide redox couple was used as the electrolyte. CV was carried out at scan rates of 5, 25, 50, 75, and 100 mV s⁻¹ in the potential range (−1.0 to 1.0 V). The photovoltaic measurements of the fabricated DSSCs with a photoanode active area of 0.96 cm² were performed with an Oriel Instruments LCS-100 solar simulator accompanied by a Keithley 2420 source meter under one sun illumination (AM 1.5 G, 100 mW cm⁻²).

3 Results and discussion

The subsequent sections elucidate the physicochemical, electrical conductivity, electrochemical properties, and photovoltaic performance of the nanocomposites.

3.1 Elemental analysis

The incorporation of Bi₂O₃ into rGO or heteroatom-doped rGO is evident from the presence of bismuth peaks in the EDX spectrum (Fig. 1). ESI† – Fig. S1(a–f) also shows the EDX elemental mapping of the N-rGO/Bi₂O₃ nanocomposite, which confirms that all the constituent elements (carbon, oxygen, nitrogen, and bismuth) were distributed homogeneously in the N-rGO/Bi₂O₃ nanocomposite. This was also observed in rGO/Bi₂O₃ and B-rGO/Bi₂O₃ nanocomposites. The presence of carbon, nitrogen, oxygen, boron, and bismuth in nanocomposites was further confirmed by elemental analysis and ICP-OES (ESI† – Table S1). The Bi₂O₃ content in rGO/Bi₂O₃, N-rGO/Bi₂O₃, and B-rGO/Bi₂O₃ was found to be 20.8, 28.5, and 25.0%, respectively. N-rGO/Bi₂O₃ had a nitrogen content of 3.76%, while B-rGO/Bi₂O₃ had a boron content of 2.56%. ATR-FTIR was further used to investigate the various functional groups present in the nanocomposites.

Fig. 1 EDX spectrum of a measurement conducted at the selected area of N-rGO/Bi₂O₃.

3.2 Surface functional groups

The ATR-FTIR spectrum of Bi₂O₃ is shown in Fig. 2 (a), where a peak at 850 cm⁻¹ is attributed to the symmetrical stretching of Bi–O–Bi, while a peak at 1401 cm⁻¹ is due to the presence of the metal–oxygen vibration (Bi–O).⁴⁰ The heteroatom-doped rGO-Bi₂O₃-based nanocomposites exhibited peaks at around 800–850 cm⁻¹ and 1402 cm⁻¹, associated with the Bi–O–Bi and Bi–O vibrations, respectively. The spectra exhibited a broader peak at around 3000–3400 cm⁻¹ due to the Bi–OH bonds and hydrogen-bonded molecular water species. A weak peak observed at 1665 cm⁻¹ and 1250 cm⁻¹ was assigned to N–H bending and C–N stretching in N-rGO/Bi₂O₃, respectively. The presence of boron in B-rGO/Bi₂O₃ was indicated by a peak at around 1180 cm⁻¹, which is due to the B–C stretching vibration. Quantitative analysis and ATR-FTIR spectroscopy confirmed the incorporation of Bi₂O₃ into heteroatom-doped rGO. Thus, to further study the structures of the nanocomposites, TEM and FE-SEM analyses were conducted.

Fig. 2 (a) ATR-FTIR spectra, (b) powder X-ray diffractograms, (c) Raman spectra and (d) N₂ adsorption–desorption isotherms of the Bi₂O₃-based nanocomposites.

3.3 Microstructure and surface morphology

The TEM images in Fig. 3(a)–(c) show wrinkled graphene sheets with Bi₂O₃ that is uniformly dispersed, indicating the existence of a strong interaction between rGO, N-rGO or B-rGO, and Bi₂O₃. Moreover, the interface between Bi₂O₃ and rGO, N-rGO, or B-rGO shows that the Bi₂O₃ was well anchored on the sheets. This confirms that Bi₂O₃ was successfully incorporated into heteroatom-doped rGO as indicated in the EDX spectrum (Fig. 1). Fig. 3(d) and (e) show SEM images, where agglomeration of Bi₂O₃ is observed in the graphene sheets. N-rGO/Bi₂O₃ exhibited a high degree of agglomeration due to the high Bi₂O₃ content.

Fig. 3 TEM images of (a) rGO/Bi₂O₃, (b) N-rGO/Bi₂O₃, and (c) B-rGO/Bi₂O₃. SEM images of (d) rGO/Bi₂O₃, (e) N-rGO/Bi₂O₃, and (f) B-rGO/Bi₂O₃. HR-TEM images of (g) rGO/Bi₂O₃, (h) N-rGO/Bi₂O₃, and (i) B-rGO/Bi₂O₃.

Consistent with SEM and TEM results, the HRTEM analysis also indicated the successful formation of Bi₂O₃-based nanocomposites. Fig. 3(g) to (i) revealed a decrease in interlayer spacing of the Bi₂O₃-based nanocomposites compared with rGO, N-rGO, or B-rGO (ESI† – Fig. S2) due to the distortion introduced by the inclusion of Bi₂O₃ and the reduction of oxygen functional groups, such as carboxyl, epoxy and hydroxyl groups.

3.4 Phase composition and defects on the graphitic structure

Fig. 2(b) shows the diffractograms of pure Bi₂O₃ and the Bi₂O₃-based nanocomposites. The P-XRD pattern of pure Bi₂O₃ shows five intense peaks at 2θ values of 24.7°, 26.8°, 33.5°, 42.0°, and 52.5°, which correspond to the (102), (120), (121), (122), and (321) planes, respectively, confirming the formation of an α phase. These characteristic peaks are confirmed by the reported standard JCPDS file number 76-1730, which is associated with the monoclinic phase with space group P2₁/c.⁴¹ When comparing the diffractograms of Bi₂O₃ and heteroatom-doped rGO-Bi₂O₃-based nanocomposites, the diffraction peak representing the carbon species expected at 25° was not observed in all nanocomposites because the diffraction peak was obscured by the stronger diffraction peak of α-Bi₂O₃, and due to the low content of rGO, which was below the detection limit.^34,42

The sharp diffraction peak at 26.8° indicates that Bi₂O₃ is in a crystalline state, but this peak was reduced after integrating heteroatom-doped rGO with Bi₂O₃. Bi₂O₃ exhibited a relatively high 2θ peak intensity due to the presence of larger nanoparticles with few grain boundaries and low defect density. Theoretically, larger particles have sharp diffraction peaks, while the diffraction peak width increases as the particle sizes are reduced.⁴³ The most intense diffraction peak (120) was fitted with a Lorentzian distribution function to determine the full-width at half-maximum (FWHM), and the value obtained was employed to calculate the crystallite size of the nanocomposites with Scherrer's formula. The crystallite sizes of Bi₂O₃, rGO/Bi₂O₃, N-rGO/Bi₂O_3, and B-rGO/Bi₂O₃ were 29.2, 20.9, 19.5, and 16.8 nm, respectively. The boron content in B-rGO/Bi₂O₃ caused an increase in structural strain, thus leading to enhanced surface defects. An interlayer spacing of 0.33 nm was obtained for Bi₂O₃, which was comparable to previously reported values.^44,45 The decrease in interlayer spacing after integrating Bi₂O₃ with heteroatom-doped rGO was caused by lattice distortions, attributed to the incorporation of boron or nitrogen and the decrease in oxygen content during the reduction process.

The crystallite size and graphitic nature of the nanocomposites were further investigated by Raman spectroscopy. The crystallite size (L_a) was calculated using eqn (1) reported by Mallet-Ladeira et al.:⁴⁶

HWHM = 71 − 5.2L_a (1)

where HWHM is the half-width at a half maximum, which is the half of the full width at half maximum (FWHM) when the function is symmetric. The crystallite sizes of Bi₂O₃-based nanocomposites were smaller than that of Bi₂O₃, indicating that thermal treatment and the type of heteroatom-doping in rGO affected the crystallite size due to the formation of small crystals. The crystallite sizes from Raman analysis are slightly larger than those from P-XRD analysis; however, they displayed a similar decreasing trend (Table 1).

Table 1 P-XRD parameters and crystallinity analysis of the nanocomposites

Sample	P-XRD			Raman
	2θ/degree	Crystallite size/nm	Interlayer spacing/nm	ID/IG	Crystallite size/nm
Bi2O3	26.8	29.2	0.33	—	34.4
rGO/Bi2O3	27.1	20.9	0.32	1.50	29.6
N-rGO/Bi2O3	27.9	19.5	0.30	1.79	28.1
B-rGO/Bi2O3	28.0	16.8	0.30	1.82	27.3

---

The graphitic nature of rGO, heteroatom-doped rGO, and Bi₂O₃-based nanocomposites was further investigated by Raman spectroscopy. The Raman spectrum of rGO (ESI† – Fig. S3 and Table S2) reveals a D-band at 1350 cm⁻¹ and a G-band at 1594 cm⁻¹, whereby the peak area ratio (I_D/I_G) shows the disorder and the GO structure's graphitic symmetry, respectively. After nitrogen- or boron-doping, the G-bands revealed a minimal shift in the wavenumber. This shift indicated that incorporating nitrogen or boron atoms into the GO lattice prompted an increase in the disordered structure relative to rGO. According to Nanda et al.,⁴⁷ the shift in the G-band of graphene represents the vibrational mode of sp²-hybridized carbon atoms. Thus, the change in the G-band position suggests the change in the number of layers in graphene. The G-band shift towards a higher wavenumber indicates a decrease in the number of graphene layers.

Fig. 2(c) shows that the intensity of the G-band of rGO/Bi₂O₃ is highest, corresponding to the sp² structure of the graphite sheet.⁴⁸ The low intensities of the D- and G-band of N-rGO/Bi₂O₃ suggest a strong interaction between N-rGO and Bi₂O₃. This has also been evidenced by the presence of the highest Bi₂O₃ content in N-rGO/Bi₂O₃ (ESI† – Table S1) compared with rGO/Bi₂O₃ and B-rGO/Bi₂O₃. Thus, Raman analysis demonstrated the chemical bonding between Bi₂O₃ and rGO, N-rGO, or B-rGO sheets, in agreement with elemental analysis.

The I_D/I_G ratio of rGO was lower (0.71 – ESI† (Table S2)) due to the reduction of the oxygen functionalities and the recovery of sp²-hybridized C–C bonds. However, the I_D/I_G ratios of N-rGO and B-rGO were higher than for rGO due to structural defects introduced by nitrogen or boron atoms implanted at the radicalized graphene sites. B-rGO displayed an I_D/I_G ratio of 1.18, which was higher than that of N-rGO (1.09), revealing the presence of a higher degree of disorder in B-rGO than N-rGO. Bi₂O₃-based nanocomposites showed higher I_D/I_G ratios than rGO, N-rGO and B-rGO, demonstrating that these nanocomposites possess more defects, with B-rGO/Bi₂O₃ exhibiting the highest I_D/I_G ratio (1.82).

3.5 Surface area and porosity

The specific surface area and pore size of the Bi₂O₃-based nanocomposites were investigated by the nitrogen adsorption/desorption method (Table 2). The B-rGO/Bi₂O₃ nanocomposite was found to have the largest specific surface area of 65.5 m² g⁻¹. The relatively low surface area of 53.7 m² g⁻¹ for rGO/Bi₂O₃ was attributed to the blockage of the rGO lattice by the Bi₂O₃ particles. The larger surface area of nanomaterials is beneficial for enhancing dye loading in the photoanode of DSSCs. Therefore, nanomaterials with a small surface area result in low dye loading, thereby reducing the short-circuit current density (J_sc) value and the PCE of DSSCs.⁴⁹ Thus, to enhance the PCE of DSSCs, the photoanode nanomaterials ought to have a larger surface area. All the nanocomposites exhibited pore sizes below 23 nm, which belong to mesoporous material. Such mesoporous materials with pore sizes between 2 and 50 nm have been widely reported to improve the infiltration of materials in electrolytes.^50,51

Table 2 Textural characterisation of the nanocomposites

Sample	Surface area/m^2 g^−1	Pore size/nm
Bi2O3	4.6	33.52
rGO/Bi2O3	53.7	20.37
N-rGO/Bi2O3	58.2	22.15
B-rGO/Bi2O3	65.5	17.80

---

Fig. 2(d) exhibited type IV adsorption–desorption isotherms with an H3 hysteresis loop in the range of 0.45 − 1.0 P/P₀. The H3 hysteresis loop indicates the presence of narrower pores, channel-like pores, and a pore network-linking effect.⁵² This again suggests that the synthesised nanocomposites are mesoporous materials. The relative pressure tends to increase (P/P₀ > 0.85), and the shape of the adsorption isotherm rises, indicating an increase in the amount of N₂-adsorption and the presence of capillary condensation in the mesoporous material. This is attributed to the rGO nanomaterial having a larger pore size and being capable of adsorbing a large amount of N₂.

3.6 Thermal stability

The TGA analysis in Fig. 4(a) shows the thermal stability studies represented as TGA thermograms, and ESI† (Table S3) shows the decomposition temperatures and residual content of the nanocomposites. All the nanocomposites exhibited few weight losses around the decomposition temperature of 100 °C, which was attributed to the removal of moisture adsorbed in the interlayers of the nanocomposites. Furthermore, the weight loss that occurred at 200–400 °C is due to the loss of amorphous carbon and other functional groups with the release of CO_x species. Such decrease in weight in the 200–400 °C range correlates with TEM and SEM analysis (Fig. 3), inferring the presence of amorphous materials. The thermograms indicated that the order of the nanocomposites' tolerance to heat is B-rGO/Bi₂O₃ < N-rGO/Bi₂O₃ < rGO/Bi₂O₃, i.e., with decomposition temperatures of 437, 490, and 518 °C, respectively. The thermal stability trends agree with the graphitic nature (lower crystallinity) (Raman analysis – Table 1) of the nanocomposites. B-rGO/Bi₂O_3, with the highest fraction of layered graphene, exhibited the lowest thermal stability of 437 °C and the lowest residual content of Bi₂O₃ of 38%. The highest amount of residue (57%) for N-rGO/Bi₂O₃ could be attributed to some remnant Bi₂O₃ that remains entrapped in the nanocomposite. The elemental analysis also indicates a high Bi₂O₃ content in N-rGO/Bi₂O₃ (ESI† – Table S1).

Fig. 4 (a) Thermal stability studies represented as TGA thermograms, (b) current–voltage characteristics, (c) cyclic voltammograms and (d) Nyquist plots of the Bi₂O₃-based nanocomposites.

3.7 Electrical properties

Four-probe and Hall measurements were used to investigate the electrical conductivity and charge carrier mobility, respectively. Hall effect studies showed that Bi₂O₃, rGO/Bi₂O_3, and N-rGO/Bi₂O₃ have n-type charge carriers, while B-rGO/Bi₂O₃ has p-type charge carriers. Such n- and p-type charge carrier characteristics are due to the change in the electronic structure of the nanocomposite. This suggests that N-rGO-Bi₂O₃ is a good material for electron transport, while B-rGO/Bi₂O₃ is a good hole transporter. The I–V characteristics (Fig. 4(b) and Table 3) revealed that the maximum conductivity (6.31 S cm⁻¹) was obtained from B-rGO/Bi₂O₃ with an improved charge carrier mobility (0.368 cm² V⁻¹ s⁻¹). The improvement in charge carrier mobility is attributed to the decrease in the scattering probability of the charge carriers. The size of boron ions is smaller than that of bismuth ions (the ionic radius of B³⁺ is 0.23 Å and Bi³⁺ is 1.17 Å). Thus, there is less scattering of electrons, thereby increasing the charge carrier mobility. B-rGO/Bi₂O₃ had significantly improved electrical conductivity, and this is attributed to its relatively large surface area and high charge carrier mobility.

Table 3 Electrical properties of the Bi₂O₃-based nanocomposites

Sample	Conductivity/S cm^−1	Carrier mobility/cm^2 V^−1 s^−1
Bi2O3	2.10 × 10^−2	0.080
rGO/Bi2O3	3.09	0.267
N-rGO/Bi2O3	3.84	0.295
B-rGO/Bi2O3	6.31	0.368

---

3.8 Electrochemical properties

3.8.1 Electrode potential characteristics. The cyclic voltammograms of the samples are shown in Fig. 4(c), where a comparison of Bi₂O₃ and the nanocomposites was obtained at a scanning rate of 100 mV s⁻¹. The variation of scan rate, i.e., 5, 25, 50, 75, and 100 mV s⁻¹ in the −1.0 to 1.0 V potential range (ESI† – Fig. S4), showed that the cyclic voltammograms maintained certain curve stability,^48,53 indicating that the nanocomposites were mainly electric double layer capacitive. Bi₂O₃ exhibited a small rectangular area, indicating a small capacitance, whereas the Bi₂O₃-based nanocomposites exhibited extensive background and peak currents. This shows that the effective area of the electrodes was improved, and the electron exchange rate was accelerated by the conductive properties of the Bi₂O₃-based nanocomposites.

The presence of oxygen-functional groups in Bi₂O₃ decreases its electrical conductivity and deteriorates in the aqueous electrolyte. During the integration of Bi₂O₃ with rGO or heteroatom-doped rGO, the oxygen-functional groups were reduced during thermal treatment and the high-pressure environment of the hydrothermal system. Therefore, reducing oxygen-functional groups increased the degree of wetting between the electrode and the electrolyte, thus resulting in an enhanced electrical conductivity, consequently improving the capacitance. A similar observation has been reported by Yang et al.,⁵⁴ who reported an increase in the capacitance of highly orientated Bi₂O₃/rGO nanocomposites. The electrochemical activity and electrochemical reversibility are also enhanced due to the introduction of defects and heteroatom-containing groups at the electrode surface, which accelerates the charge transfer rate across the electrode. This also indicates that Bi₂O₃-based nanocomposites have ideal capacitor characteristics. The reversible charge–discharge characteristics in Bi₂O₃-based nanocomposites can probably be attributed to the electrical conductivity of the randomly distributed β-Bi₂O₃ that is among the α- and δ-phases.⁵⁵ Thus, the nanocomposites were further investigated with EIS.

3.8.2 Interfacial charger transfer characteristics. For impedance measurements, the Bi₂O₃ and Bi₂O₃-based nanocomposite electrodes were characterised via Nyquist plots (Fig. 4(d)). The Nyquist plots showed a semicircle corresponding to the faradaic charge transfer resistance (R_ct) (Table 4) and a linear region, indicating a pure capacitive behaviour and a diffusion-limited process, respectively.^56,57 The Bi₂O₃ electrode exhibited a small semicircle in the high-frequency region due to lower resistance. A lower impedance, observed for rGO/Bi₂O₃ compared to Bi₂O_3, led to an improved electron transfer rate. All the heteroatom-doped rGO/Bi₂O₃-based nanocomposites exhibited no semicircle, indicating that the resistance is negligible, thus, suggesting a remarkable charge reversal with a change in voltage. This was attributed to the high electrical conductivity, large specific surface areas, and high nitrogen/boron-doping of the nanocomposites, which provided more sites for enhanced catalytic activity. For instance, in the case of B-rGO/Bi₂O₃, the lack of semicircle was attributed to the largest surface area of B-rGO and its sp³ network, which behaves as an electron transport medium from the conduction band of Bi₂O₃ to B-rGO. Therefore, the presence of B-rGO in the nanocomposite decreases the charge recombination rate, thus resulting in an enhanced lifetime of the charge carriers.⁵⁸

Table 4 Resistance values for the nanocomposites

Sample	Rct/Ω
Bi2O3	201.1
rGO/Bi2O3	169.8
N-rGO/Bi2O3	125.7
B-rGO/Bi2O3	108.4

---

The series resistance is different for different electrodes used in Fig. 4(d) due to the presence of various oxygen-functional groups in the Bi₂O₃-based nanocomposites. This could also be attributed to the constriction phenomenon of the thick porous film, which had a film thickness of 2 μm that might not be thick enough for electrodes to allow better current collection. Similar observations have been previously reported by Tezyk et al.⁵⁹

The R_ct values (Table 4) were calculated from the EIS data. The R_ct values of the Bi₂O₃-based nanocomposites were lower than that of Bi₂O₃ due to the free charge transfer in the nanocomposites. The relatively small R_ct values for the nanocomposites indicate that the modification of Bi₂O₃ with heteroatom-doped rGO drastically reduces the electrode–electrolyte interfacial resistance. The introduction of heteroatom-doped rGO into Bi₂O₃ lowered the R_ct value, which indicates the efficient separation of photogenerated electron–hole pairs and faster interfacial charge transfer. Therefore, heteroatom-doped rGO acts as the conducting channels inside the Bi₂O₃ matrix. These results affirm that the effective incorporation of heteroatom-doped rGO sheets improves the electron transport and electrical conductivity of the nanocomposites, resulting in a significant enhancement in electrochemical performance.

3.9 Photovoltaic performance of the fabricated DSSCs

DSSCs were fabricated using pure Bi₂O₃, rGO/Bi₂O₃, N-rGO/Bi₂O_3, and B-rGO/Bi₂O₃ as photoanodes; and the open-circuit voltage (V_oc), J_sc, fill factor (FF) and PCE were determined. The photovoltaic properties of DSSCs are listed in Table 5, and the J–V curves are shown in Fig. 5. Bi₂O₃-based DSSCs exhibited a higher V_oc than rGO/Bi₂O₃-based DSSCs due to back electron transfer (from Bi₂O₃ to the redox couple in the electrolyte), which is associated with electrons located in a shallow conduction band edge, thus resulting in a short lifetime. The low PCE in the Bi₂O₃-based DSSC is due to poor dye absorption and back recombination at the photoanode–electrolyte interface. A similar observation was reported by Fatima et al.,²⁶ who obtained a PCE of 0.05% for the Bi₂O₃-based DSSC. Poor dye absorption causes a decrease in light absorption, which reduces the photogeneration of charge carriers, and hence, lowers the current density, resulting in poor device performance. Thus, optimisation by surface modification of Bi₂O₃ has been used to increase dye absorption and reduce back recombination, thereby improving the performance of DSSCs.

Table 5 Photovoltaic performance of DSSCs with Bi₂O₃-based nanocomposite photoanodes

Photoanode	Voc/V	Jsc/mA cm^−2	FF/%	PCE
Bi2O3	0.53	6.0	20.63	0.42 ± 0.04
rGO/Bi2O3	0.48	9.8	43.9	1.68 ± 0.01
N-rGO/Bi2O3	0.55	9.2	47.4	1.97 ± 0.02
B-rGO/Bi2O3	0.59	10.0	50.2	2.79 ± 0.01

---

Fig. 5 J–V characteristics of the Bi₂O₃-based nanocomposites.

When Bi₂O₃-based nanocomposites were used as a photoanode, the DSSCs exhibited a higher PCE than the Bi₂O₃-based DSSCs. From the EIS analysis and electrical properties of Bi₂O₃-based nanocomposites, it can be suggested that the improvement in PCE is attributed to the low charge transfer resistance and high electrical conductivity of these nanocomposites. When carbon atoms in the graphene lattice are substituted by nitrogen or boron atoms, the conductivity of N-rGO/Bi₂O₃ or B-rGO/Bi₂O₃ increases remarkably, and the excited electrons are transferred faster in the delocalized π structure of N-rGO/Bi₂O₃-based photoanodes or B-rGO/Bi₂O₃-based photoanodes than rGO/Bi₂O₃-based photoanodes.

B-rGO/Bi₂O₃-based DSSCs exhibited the highest PCE due to enhanced transportation and collection of photogenerated charge carriers. The high PCE is attributed to the hole (h⁺) leaving the valence band of B-rGO and quickly moving to the valence band of Bi₂O₃, which facilitates electron and hole (e⁻/h⁺) separation. B-rGO has been reported to have a lower bandgap;⁶⁰ thus, the lower bandgap of B-rGO is responsible for providing a photovoltaic effect in Bi₂O₃ with a high bandgap,⁶¹ enhancing charge separation and extending the energy of photoexcitation in DSSCs. Moreover, the electron mobility rate in B-rGO/Bi₂O₃ is higher than in rGO/Bi₂O₃ and N-rGO/Bi₂O₃, enabling the efficient transportation of electrons and reducing charge recombination. Less charge recombination means more electron density and a shift of the Fermi level, resulting in an increased V_oc. This is consistent with EIS data, in which the B-rGO/Bi₂O₃ nanocomposite exhibited the lowest R_ct value. This results in fast electron collection and low charge carrier recombination, leading to the fabrication of a B-rGO/Bi₂O₃-based DSSC with the highest PCE of 2.79%.

The introduction of rGO or heteroatom-doped rGO into Bi₂O₃ did not only improve the PCE, but also enhanced the J_sc and V_oc values. In the case of the rGO/Bi₂O₃-based DSSC and N-rGO/Bi₂O₃-based DSSC, low V_oc values of 0.48 and 0.55 V, respectively (Table 5 and Fig. 5), were attributed to faster dye regeneration, which reduces the lifetime of the oxidised dye molecule to suppress charge recombination, thus reducing the V_oc. The presence of a low Bi₂O₃ content in rGO/Bi₂O₃ increased the J_sc but lowered the V_oc, which substantially reduced the PCE of DSSCs. Although N-rGO/Bi₂O₃ showed a higher electrical conductivity and charge carrier mobility than rGO/Bi₂O₃, its performance in DSSCs as a photoanode resulted in a low J_sc due to charge recombination, which is attributed to various bonding configurations of nitrogen (pyrrolic-N, pyridinic-N, and graphitic-N) present on the GO surface.

The J_sc is simultaneously affected by electron transfer efficiency, light scattering, and dye absorption. Thus, the improvement of J_sc is due to the increase in the absorption of dye molecules in DSSCs. The large surface area of the nanocomposite is beneficial for enhancing dye loading in the photoanode of DSSCs. Therefore, the dye molecules are capable of harvesting more light energy for the effective photogeneration of charge carriers. B-rGO/Bi₂O₃ had the largest surface area for efficient dye-loading, which increased the J_sc and promoted a higher PCE in DSSCs.

4 Conclusion

In summary, heteroatom-doped rGO/Bi₂O₃-based nanocomposites were successfully synthesised and applied as photoanodes in DSSCs. P-XRD analysis showed that pristine Bi₂O₃ is in the monoclinic (α) form, and this polymorph was maintained even after introducing heteroatom-doped rGO. The heteroatom-doped rGO/Bi₂O₃-based nanocomposites exhibited better catalytic activity than pure Bi₂O₃. Favourable properties, such as large surface area, small pore size, low charge transfer resistance, high charge carrier mobility, and high electrical conductivity, were observed in the heteroatom-doped rGO/Bi₂O₃-based nanocomposites. This demonstrated the suitability of the prepared Bi₂O₃-based nanocomposites as potential photoanode materials for DSSCs. The highest PCE of 2.79% was achieved for the B-rGO/Bi₂O₃-based DSSC, which was attributed to the enhanced channels for electron transfer and electrolyte diffusion. Therefore, the synthesised heteroatom-doped rGO/Bi₂O₃-based nanocomposites are ideal materials for the photoanode in DSSCs.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors would like to thank the National Research Foundation (NRF, Grant numbers – 101357, 116505 and 103979), Moses Kotane Institute, Eskom Tertiary Education Support Programme (TESP), UKZN Nanotechnology Platform, and the University of KwaZulu-Natal for financial support and facilities.

References

1. G. Boschloo, Front. Chem., 2019, 7, 1–9.
2. Q. Huaulmé, V. M. Mwalukuku, D. Joly, J. Liotier, Y. Kervella, P. Maldivi, S. Narbey, F. Oswald, A. J. Riquelme, J. A. Anta and R. Demadrille, Nat. Energy, 2020, 5, 468–477.
3. G. George, R. S. Yendaluru and A. Mary Ealias, Energy Sources, Part A, 2020, 1, 1–15.
4. J. Castro-Gutiérrez, A. Celzard and V. Fierro, Front. Mater., 2020, 7, 1–25.
5. Y. Ma, D. Chen, Z. Fang, Y. Zheng, W. Li, S. Xu, X. Lu, G. Shao, Q. Liu and W. Yang, Proc. Natl. Acad. Sci. U.S.A., 2021, 118, e2105610118–e2105610218.
6. R. Krishnapriya, C. Nizamudeen, B. Saini, M. S. Mozumder, R. K. Sharma and A. H. I. Mourad, Sci. Rep., 2021, 11, 16265–16277.
7. H. Ozawa, Y. Okuyama and H. Arakawa, Dalton Trans., 2012, 41, 5137–5139.
8. M. U. Rahman, M. Wei, F. Xie and M. Khan, Catalysts, 2019, 9, 273–274.
9. P. M. Pataniya, D. Late and C. K. Sumesh, ACS Appl. Energy Mater., 2021, 4, 755–762.
10. P. M. Pataniya, V. Patel and C. K. Sumesh, ACS Appl. Energy Mater., 2021, 4, 7891–7899.
11. S. Yousaf, S. Zulfiqar, M. I. Din, P. O. Agboola, M. F. Aly Aboud, M. F. Warsi and I. Shakir, J. Mater. Res. Technol., 2021, 12, 999–1009.
12. M. Adeel, M. Saeed, I. Khan, M. Muneer and N. Akram, ACS Omega, 2021, 6, 1426–1435.
13. T. Solaiyammal and P. Murugakoothan, Mater. Sci. Energy Technol., 2019, 2, 171–180.
14. R. Chauhan, A. Kumar, G. G. Umarji, U. P. Mulik and D. P. Amalnerkar, J. Solid State Electrochem., 2015, 19, 161–168.
15. M. Waghmare, N. Beedri, A. Ubale and H. Pathan, Eng. Sci., 2019, 6, 36–43.
16. X. Chen, J. Ye, S. Ouyang, T. Kako, Z. Li and Z. Zou, ACS Nano, 2011, 5, 4310–4318.
17. A. Birkel, Y.-G. Lee, D. Koll, X. Van Meerbeek, S. Frank, M. J. Choi, Y. S. Kang, K. Char and W. Tremel, Energy Environ. Sci., 2012, 5, 5392–5400.
18. K. Yang, R. Li, C. Zhu and J. Pei, J. Mater. Res., 2021, 36, 2936–2949.
19. C. Li, P. He, F. Dong, H. Liu, L. Jia, D. Liu, L. Du, H. Liu, S. Wang and Y. Zhang, Mater. Lett., 2019, 245, 29–32.
20. Y. Qiu, H. Fan, X. Chang, H. Dang, Q. Luo and Z. Cheng, Appl. Surf. Sci., 2018, 434, 16–20.
21. J. M. Bothwell, S. W. Krabbe and R. S. Mohan, Chem. Soc. Rev., 2011, 40, 4649–4707.
22. Y. Qiu, M. Yang, H. Fan, Y. Zuo, Y. Shao, Y. Xu, X. Yang and S. Yang, CrystEngComm, 2011, 13, 1843–1850.
23. M. Ciszewski, A. Mianowski, P. Szatkowski, G. Nawrat and J. Adamek, Ionics, 2015, 21, 557–563.
24. M. Schlesinger, S. Schulze, M. Hietschold and M. Mehring, Dalton Trans., 2013, 42, 1047–1056.
25. S. Shaikh, G. Rahman, R. S. Mane and O.-S. Joo, Electrochim. Acta, 2013, 111, 593–600.
26. M. J. Jabeen Fatima, C. V. Niveditha and S. Sindhu, RSC Adv., 2015, 5, 78299–78305.
27. P. Fernández-Ibáñez, M. I. Polo-López, S. Malato, S. Wadhwa, J. W. J. Hamilton, P. S. M. Dunlop, R. D’Sa, E. Magee, K. O'Shea, D. D. Dionysiou and J. A. Byrne, Chem. Eng. J., 2015, 261, 36–44.
28. E. Kusiak-Nejman, A. Wanag, Ł. Kowalczyk, J. Kapica-Kozar, C. Colbeau-Justin, M. G. Mendez Medrano and A. W. Morawski, Catal. Today, 2017, 287, 189–195.
29. J. P. Mensing, C. Poochai, S. Kerdpocha, C. Sriprachuabwong, A. Wisitsoraat and A. Tuantranont, Adv. Nat. Sci.: Nanosci. Nanotechnol., 2017, 8, 033001–033008.
30. M. I. A. Abdel Maksoud, R. A. Fahim, A. E. Shalan, M. Abd Elkodous, S. O. Olojede, A. I. Osman, C. Farrell, A. H. Al-Muhtaseb, A. S. Awed, A. H. Ashour and D. W. Rooney, Environ. Chem. Lett., 2021, 19, 375–439.
31. C. K. Chua and M. Pumera, Chem. Soc. Rev., 2014, 43, 291–312.
32. Z. Bo, Z. Wen, H. Kim, G. Lu, K. Yu and J. Chen, Carbon, 2012, 50, 4379–4387.
33. F. M. Ascencio Aguirre and R. Herrera Becerra, Appl. Phys. A, 2015, 119, 909–915.
34. X. Liu, L. Pan, T. Lv, Z. Sun and C. Q. Sun, J. Colloid Interface Sci., 2013, 408, 145–150.
35. T. Kuila, S. Bose, A. K. Mishra, P. Khanra, N. H. Kim and J. H. Lee, Prog. Mater. Sci., 2012, 57, 1061–1105.
36. D.-Y. Yeom, W. Jeon, N. D. K. Tu, S. Y. Yeo, S.-S. Lee, B. J. Sung, H. Chang, J. A. Lim and H. Kim, Sci. Rep., 2015, 5, 9817–9836.
37. S. Pei and H.-M. Cheng, Carbon, 2012, 50, 3210–3228.
38. N. P. D. Ngidi, M. A. Ollengo and V. O. Nyamori, Materials, 2019, 12, 3376–3401.
39. N. P. D. Ngidi, M. A. Ollengo and V. O. Nyamori, New J. Chem., 2020, 44, 16864–16876.
40. Y. Astuti, B. M. Listyani, L. Suyati and A. Darmawan, Indones. J. Chem., 2021, 21, 108–117.
41. M. Malligavathy and D. P. Padiyan, Adv. Mater. Processes, 2017, 2, 51–55.
42. H. Zhang, X. Lv, Y. Li, Y. Wang and J. Li, ACS Nano, 2010, 4, 380–386.
43. Z. A. Zulkifli, K. A. Razak and W. N. W. A. Rahman, AIP Conf. Proc., 2017, 1901, 020011–020016.
44. M. Jalalah, M. Faisal, H. Bouzid, J.-G. Park, S. A. Al-Sayari and A. A. Ismail, J. Ind. Eng. Chem., 2015, 30, 183–189.
45. E. Pargoletti, S. Mostoni, G. Rassu, V. Pifferi, D. Meroni, L. Falciola, E. Davoli, M. Marelli and G. Cappelletti, Environ. Sci. Pollut. Res., 2017, 24, 8287–8296.
46. P. Mallet-Ladeira, P. Puech, C. Toulouse, M. Cazayous, N. Ratel-Ramond, P. Weisbecker, G. L. Vignoles and M. Monthioux, Carbon, 2014, 80, 629–639.
47. S. S. Nanda, M. J. Kim, K. S. Yeom, S. S. A. An, H. Ju and D. K. Yi, TrAC, Trends Anal. Chem., 2016, 80, 125–131.
48. W.-D. Yang and L. Yu-Jiang, J. Electr. Eng., 2019, 70, 101–106.
49. K. Song, I. Jang, D. Song, Y. S. Kang and S.-G. Oh, Sol. Energy, 2014, 105, 218–224.
50. G. Gryglewicz, J. Machnikowski, E. Lorenc-Grabowska, G. Lota and E. Frackowiak, Electrochim. Acta, 2005, 50, 1197–1206.
51. S. Liu, Y. Wang and Z. Ma, Int. J. Electrochem. Sci., 2018, 13, 12256–12265.
52. K. S. Sing and R. T. Williams, Adsorp. Sci. Technol., 2004, 22, 773–782.
53. H. Shen, Y. Zhang, X. Song, Y. Liu, H. Wang, H. Duan and X. Kong, J. Alloys Compd., 2019, 770, 926–933.
54. W.-D. Yang and Y.-J. Lin, Int. J. Electrochem. Sci., 2020, 15, 1915–1929.
55. X. Yang, X. Lian, S. Liu, G. Wang, C. Jiang, J. Tian, J. Chen and R. Wang, J. Phys. D: Appl. Phys., 2012, 46, 035103–035110.
56. T. Zhan, X. Wang, X. Li, Y. Song and W. Hou, Sens. Actuators, B, 2016, 228, 101–108.
57. Z. Wei, Y. Wang and J. Zhang, Sci. Rep., 2018, 8, 6929–6937.
58. X. Bai, X. Zhang, Z. Hua, W. Ma, Z. Dai, X. Huang and H. Gu, J. Alloys Compd., 2014, 599, 10–18.
59. V. Tezyk, C. Rossignol, N. Sergent, E. Djurado, J. Laurencin and E. Siebert, Electrochim. Acta, 2019, 304, 312–322.
60. M. Junaid, M. H. M. Khir, G. Witjaksono, N. Tansu, M. S. M. Saheed, P. Kumar, Z. Ullah, A. Yar and F. Usman, Molecules, 2020, 25, 3646–3665.
61. J. Li, B. Z. Wu and Z. X. Zhou, Micro Nano Lett., 2018, 13, 1443–1446.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ra08888b

‡ On leave from Bindura University of Science Education, Department of Engineering and Physics, Private Bag 1020, Bindura, Zimbabwe.

---