Earth-abundant and nano-micro composite catalysts of Fe₃O₄@reduced graphene oxide for green and economical mesoscopic photovoltaic devices with high efficiencies up to 9%

Abstract

The ideal liquid–solid heterogeneous electrocatalysis should have not only high catalytic activity but also free electron transport. However, preparing a single catalyst that simultaneously possesses both advantages has proven to be challenging. Herein, we prepared nano–micro composite catalysts (NMCCs) composed of highly dispersed Fe₃O₄ nanoparticles fixed on reduced graphene oxide (RGO) sheets (namely Fe₃O₄@RGO-NMCC) as the counter electrode (CE) in dye-sensitized solar cells (DSCs). Compared with the Fe₃O₄ or RGO CE, the Fe₃O₄@RGO-NMCC CE exhibited improved activity and reversibility for the catalytic reduction of triiodide ions (I₃⁻) to iodide ions (I⁻). Notably, DSCs using rigid and flexible Fe₃O₄@RGO-NMCC CEs achieved high PCEs up to 9% and 8% on fluorine-doped tin oxide (FTO)/glass substrates and flexible polymer substrates, respectively. These values are, to our knowledge, some of the highest reported efficiencies for DSCs based on a flexible Pt-free CE. We ascribed the superior catalytic performance of Fe₃O₄@RGO-NMCC to faster electron hopping between Fe²⁺ and Fe³⁺ and free electron transport by broad RGO sheets. Finally, Fe₃O₄@RGO-NMCC exhibited good stability in the practical application of DSCs because Fe₃O₄ nanoparticles were chemically bonded to the surface of RGO. Our work here will be of great interest for fundamental research and practical applications of Fe₃O₄ in lithium batteries, splitting water and magnetic fields.

---

Introduction

As third-generation photovoltaic devices, dye-sensitized solar cells (DSCs)¹ based on environmentally friendly (i.e., green) raw materials completely free of Cd or Pb have been regarded as very promising solar cells owing to their simple synthesis procedures, high theoretical power conversion efficiency (PCE), and low cost. With continuous efforts from researchers worldwide, the PCEs of DSCs have been improved by 13% (ref. 2) using Pt as the counter electrode (CE).

As key components, CEs of DSCs are used to collect electrons from the external circuit, and more importantly, catalyze the reduction of redox mediators in electrolytes. Pt is the most widely used CE material in DSCs. However, using Pt involves a number of challenges. Firstly, the scarcity and high cost of Pt cannot meet the needs of mass industrial production. Secondly, Pt can be corroded by I⁻/I₃⁻ electrolytes, leading to poor stability of the photovoltaic device.^3,4 Therefore, developing a high-performance, low-cost, corrosion-resistant, non-precious metal catalyst is necessary. In recent years, numerous researchers have developed alternative CE materials, such as inorganic materials,^5,6 carbon materials,^7–10 and conductive polymer materials.^11,12

Of the transition metal oxides studied, iron oxides showed remarkable abundance and notable catalytic activity, with PCEs of 6.89% for Fe₂O₃ nanoparticles¹³ and 7.65% for Fe₃O₄ hierarchical structures.¹⁴ However, grain boundaries in nanoparticle and hierarchical structures are massive and remarkably hinder electron transport.^15,16 Recently, our group further synthesized a composite catalyst of rosin carbon/Fe₃O₄ to enhance the performance of the corresponding DSCs.¹⁷ However, the Fe₃O₄ nanoparticle on rosin carbon substantially aggregates together, resulting in the loss of the active sites.

Herein, we prepared nano–micro composite catalysts (NMCCs) composed of highly dispersed Fe₃O₄ nanoparticles fixed on reduced graphene oxide (RGO) sheets (namely, Fe₃O₄@RGO-NMCC) as the CE in DSCs. Compared with the pure Fe₃O₄ and RGO CE, the Fe₃O₄@RGO-NMCC CE exhibited improved activity and reversibility for the catalytic reduction of triiodide (I₃⁻) to iodide (I⁻). Notably, DSCs using rigid and flexible Fe₃O₄@RGO-NMCC CEs achieved high PCEs up to 9% and 8% on fluorine-doped tin oxide (FTO)/glass substrates and flexible polymer substrates, respectively. These values are, to our knowledge, the highest reported efficiencies for DSCs based on a flexible Pt-free CE. We ascribed the superior catalytic performance of Fe₃O₄@RGO-NMCC to faster electron hopping between Fe²⁺ and Fe³⁺ and free electron transport by broad RGO sheets. Finally, Fe₃O₄@RGO-NMCC exhibited good stability in the practical application of DSCs because Fe₃O₄ nanoparticles were chemically bonded to the surface of RGO.

Experimental section

Preparation of Fe₃O₄

Typically, 10 mL 0.5 mol L⁻¹ FeCl₃ in ethylene glycol (EG) solution, 10 mL 0.1 g mL⁻¹ polyethylene glycol (PEG) in EG solution and 10 mL 1.5 mol L⁻¹ NaOH in EG solution were mixed. The mixture was stirred for 4 h at room temperature and then transferred into a Teflon-lined autoclave. After being heated at 200 °C for 15 h, the mixture was cooled to room temperature naturally.

Preparation of RGO

Graphene oxide (GO) was prepared using a modified Hummers method. Typically, flake graphite (1 g), sodium nitrate (1 g), and potassium permanganate (3.0 g) were mixed in 98% sulfuric acid (48 mL) by vigorous agitation in a round-bottom flask at 0 °C for 24 h. Then, the mixture was heated and kept at 35 °C for 30 min. After that, 250 mL distilled water was slowly added into the suspension and stirred for 20 min. Then, 2.5 mL hydrogen peroxide (30%) was added into the mixture. The as-prepared suspension was washed thoroughly with deionized water and centrifuged with 3000 rpm. After that, the supernatant was then centrifuged with 8000 rpm until its pH was ≈7. The obtained precipitates were redispersed in deionized water to obtain a 1.7 wt% GO dispersion. 27 g 1.7 wt% GO dispersion were dispersed in 100 mL EG solution. The mixture was stirred for 4 h at room temperature and then transferred into a Teflon-lined autoclave. After being heated at 200 °C for 15 h, the mixture was cooled to room temperature naturally.

Preparation of Fe₃O₄@RGO-NMCC

Typically, 10 mL 0.5 mol L⁻¹ FeCl₃ in EG solution, 10 mL 0.1 g mL⁻¹ PEG in EG solution and 10 mL 1.5 mol L⁻¹ NaOH in EG solution were mixed with 2.7 g 1.7 wt% GO aqueous solution dispersed in 10 mL EG solution. The mixture was stirred for 4 h at room temperature and then transferred into a Teflon-lined autoclave. After being heated at 200 °C for 15 h, the mixture was cooled to room temperature naturally.

Photoanode preparation and cell fabrication

A layer of 20 nm-sized TiO₂ (P25, Degussa, Germany) (12 μm) was loaded on FTO glass by a printing technique method and used in this study.¹⁰ The obtained film was sintered at 500 °C. After cooling to 90 °C, the TiO₂ films were immersed in a solution of N719 dye (5 × 10⁻⁴ M) in acetonitrile/tert-butyl alcohol (1 : 1 volume ratio) for 20 h. For the TiO₂ photoanode film treated by TiCl₄, the films were immersed in 40 mM TiCl₄ solution at 70 °C for 30 min and then sintered at 500 °C for 30 min. The triiodide/iodide electrolyte for cell testing is composed of LiI (0.03 M), 1-butyl-3-methylimidazolium iodide (0.6 M), I₂ (0.03 M), 4-tert-butyl pyridine (0.5 M), and guanidinium thiocyanate in acetonitrile (0.1 M). DSCs were assembled by using a TiO₂ photoanode with a corresponding counter electrode sandwiching the redox couple in the electrolyte. The symmetrical cells with effective area (0.64 cm²) were measured in the Tafel-polarization test and the EIS experiments.

Characterization

To analyze the composition of the as-synthesized samples, we obtained X-ray diffraction (XRD) patterns using a PANalytical X'Pert diffractometer (Cu Kα radiation at λ = 1.54 Å) sampling at 8° min⁻¹, 40 kV and 100 mA. The as-prepared micro- or nano-structures were characterized and analyzed by scanning electron microscopy (SEM, Nova Nano SEM 450) and transmission electron microscopy (TEM, FEI Tecnai G2 F30) with an accelerating voltage (300 kV). The Mössbauer spectrum of the as-prepared materials was recorded using a Topologic 500A spectrometer and a proportional counter at room temperature. ⁵⁷Co(Rh) moving in a constant acceleration mode was used as the radioactive source. The film thicknesses were measured using a film-thickness measuring device (KLA-Tencor D-100). The photocurrent–voltage performance of DSCs with a 0.16 cm² photoanode film was measured without a metal mask by using a Keithley digital source meter (Keithley 2400, USA) and equipped with a solar simulator (IV5, PV Measurements, Inc., USA). The IPCE of DSCs was measured by using a quantum efficiency/spectral response (SR)/incident photon to current conversion efficiency (IPCE) measurement system (QEX10, PV Measurements, Inc., USA). EIS experiments were performed in dummy cells in the dark using a computer-controlled potentiostat (Zennium Zahner, Germany). Cyclic voltammetry (CV) was carried out in a three-electrode system. The triiodide/iodide electrolyte for CV testing is composed of LiI (2 mM), LiClO₄ (20 mM) and I₂ (0.2 mM).

Results and discussion

In our experiments, Fe₃O₄@RGO-NMCC was prepared using a simple one-pot solvothermal approach as shown in Scheme 1. The details of the preparation can be seen in the Experimental section. SEM images in Fig. S1 and S2† illustrate the agglomerated structure of pure Fe₃O₄ particles and the two-dimensional structure of RGO. The TEM image in Fig. 1a further shows the agglomeration of pure Fe₃O₄ particles resulting in the non-uniform size of the particles or caking within the range of 50 nm to 330 nm (Fig. 1b). The agglomerated structure of pure Fe₃O₄ particles reduced catalytic active sites. In addition, the massive grain boundaries in pure Fe₃O₄ inhibited electron transport.^15,16 These two factors decreased the catalytic activity of pure Fe₃O₄ particles. The transmission electron microscopy (TEM) image of Fe₃O₄@RGO-NMCC (Fig. 1c) showed that Fe₃O₄ nanoparticles were highly dispersed and firmly fixed on the surface of RGO, thus preventing agglomeration. The sizes of Fe₃O₄ nanoparticles on RGO were well within the range of 10 nm to 30 nm, with an average particle size of 19 nm (Fig. 1d). X-ray diffraction patterns of these samples (Fig. 1e) showed that Fe₃O₄@RGO-NMCC displays a crystalline structure similar to that of pure Fe₃O₄, which could be indexed to the standard Fe₃O₄ nanocrystal with a cubic structure (JCPDS01-11-0614). The diffraction peaks of 30.1°, 35.4°, 37.1°, 43.1°, 54.3°, 57.0°, and 62.5° were attributed to the planes of (220), (311), (222), (400), (422), (511), and (440) of Fe₃O₄, respectively. In addition, the peak width of Fe₃O₄@RGO-NMCC was larger compared with that of pure Fe₃O₄, indicating the decrease in the size of Fe₃O₄ on RGO. This finding was consistent with the SEM result. Mössbauer spectroscopy was performed to examine the super-fine structure in pure Fe₃O₄ and Fe₃O₄@RGO. The energy resolution of the Mössbauer spectrum was very high and was used to study the ultrafine interaction between the atomic nucleus and the surrounding environment. The Mössbauer spectrum of pure Fe₃O₄ could be fitted into two fitting curves, which were associated with the tetrahedral [Fe³⁺] ions and the octahedral [Fe²⁺, Fe³⁺] in the inverse spinel structure.¹⁸ In contrast to pure Fe₃O₄, the Mössbauer spectrum of Fe₃O₄@RGO-NMCC was fitted to three fitting curves. Among the three fitting curves, two fitting curves corresponded to the tetrahedral [Fe³⁺] ions and the octahedral [Fe²⁺, Fe³⁺] ions in pure Fe₃O₄. The relatively wide curve fitting was due to the reduced nano-size effect of Fe₃O₄ in Fe₃O₄@RGO-NMCC, which resulted in higher surface energy, enhanced quantum effect, faster electronic hopping between [Fe²⁺] and [Fe³⁺], and stronger relaxation effect (as in dangling bonds). Compared to the large size of pure Fe₃O₄, the higher surface energy of the reduced nano-size effect of Fe₃O₄ in Fe₃O₄@RGO-NMCC originated from more unsaturated bonds of surface atoms, which are active sites to absorb and catalyze “I₃⁻” to “I⁻”. The enhanced quantum effect was attributed to the increased atom density of the surface Fe²⁺ and Fe³⁺, which promoted electronic transport (hopping) in the electro-catalysis reaction “I₃⁻ + 2e⁻ ↔ 3I⁻”. Thus, these structural effects enhanced the catalytic activity of Fe₃O₄@RGO-NMCC.

Scheme 1 Schematic diagrams to illustrate simple methods for preparing Fe₃O₄@RGO (reduced graphene oxide) nano–micro composite catalysts by a one-pot solvothermal approach and fabrication of flexible Fe₃O₄@RGO counter electrodes by aerial spraying.

Fig. 1 (a) TEM images of the as-prepared pure Fe₃O₄ agglomerated nanoparticles; (b) size distribution of pure Fe₃O₄ agglomerated nanoparticles. (c) TEM images of Fe₃O₄@RGO nano–micro composite catalysts; (d) size distribution of highly dispersed Fe₃O₄ nanoparticles in Fe₃O₄@RGO nano–micro composite catalysts; (e) XRD patterns of the as-prepared GO, RGO, Fe₃O₄, Fe₃O₄@RGO powder; (f) the Mössbauer spectrum of the as-prepared pure Fe₃O₄ agglomerated nanoparticles; (g) the Mössbauer spectrum of the as-prepared Fe₃O₄@RGO nano–micro composite catalysts.

The thickness of the counter electrode influences the catalytic activity. Different thicknesses of CEs were prepared by spraying different volumes of Fe₃O₄@RGO-NMCC dispersion onto the FTO glass substrate. Fig. S3† shows the current density (J)–voltage (V) characteristics of DSCs based on different thicknesses of Fe₃O₄@RGO-NMCC CEs. The detailed photovoltaic parameters are shown in Table S1.† The photovoltaic devices with approximately 16 μm thick Fe₃O₄@RGO-NMCC CEs have the best performance. As reference, the same thicknesses of Fe₃O₄ and RGO were then fabricated into CEs by spraying onto the FTO glass substrate. First, the photovoltaic performance was obtained by characterizing DSCs based on these three CEs under AM1.5, 100 mW cm⁻² simulated illumination. For each CE, we fabricated four DSC devices and obtained the mean photovoltaic performance. The detailed photovoltaic parameters are summarized in Table 1. Fig. 2a shows the best photocurrent density with respect to voltage (J–V curve) for the DSCs in each group. The DSCs based on Fe₃O₄@RGO-NMCC CEs presented a PCE of 6.76%, the highest among the three CEs. Compared with DSCs based on Fe₃O₄@RGO-NMCC CEs, open-circuit voltage (V_oc) and short-circuit current density (J_sc) slightly decreased for DSCs based on Fe₃O₄ and RGO CEs. Thus, the main factor leading to low PCEs (from 6.76% to 1.92% and 3.71%) was derived from the deterioration of the fill factors of DSCs based on Fe₃O₄ and RGO CEs (from 0.62 to 0.39 and 0.22). Among the factors determining photovoltaic parameters of the DSCs, large internal resistance and low catalytic activity caused a decrease in fill factor and PCE. A large interparticle boundary in Fe₃O₄ adversely affected the electron transfer, further increasing the internal resistance of devices, and the less active sites in the RGO should be the reason of the low fill factor of the photovoltaic device. To improve the performance of DSCs based on Fe₃O₄@RGO-NMCC CEs, TiO₂ photoanodes were optimized by TiCl₄ treatment. This optimization resulted in a PCE of 9.04%, nearly approaching 9.46% obtained from DSCs based on pyrolytic Pt CEs (Fig. 2b).

Table 1 Photovoltaic parameters of the DSCs without TiCl₄ treatment for each group (four DSC devices)

CE	V oc (V)	J sc (mA cm^−2)	FF	PCE/%
Fe3O4/RGO	0.76 ± 0.01	14.4 ± 0.1	0.62 ± 0.01	6.76 ± 0.05
Fe3O4	0.67 ± 0.02	13.2 ± 0.2	0.22 ± 0.02	1.92 ± 0.03
RGO	0.74 ± 0.02	13.0 ± 0.1	0.39 ± 0.01	3.71 ± 0.05
Pt	0.76 ± 0.01	12.5 ± 0.2	0.74 ± 0.01	7.00 ± 0.04

---

Fig. 2 Photovoltaic and electrochemical property characterization of CEs fabricated on the FTO substrate: (a) J–V curves of DSCs based on these three CEs under AM1.5, 100 mW cm⁻² simulated illumination. (b) J–V curves of DSCs optimized by TiCl₄ treatment. (c) Tafel polarization curves of the I₃⁻/I⁻ symmetrical cells fabricated with two identical CEs (CE/electrolyte/CE); (d) CV curves of various CEs in the liquid electrolyte (I₃⁻/I⁻) system of DSCs.

The cyclic voltammetry (CV) curve and Tafel polarization curve were obtained to study the catalytic performance and electron transfer in these CEs. Fig. 2c shows the cyclic voltammograms of the I⁻/I₃⁻ redox couple based on Fe₃O₄, RGO, Fe₃O₄@RGO-NMCC, and Pt. The peak current of Fe₃O₄ CEs in CV was extremely small, indicating that a massive interparticle boundary adversely affected electron transport. This poor electron transport was bound to cause the absence of redox peaks for the I⁻/I₃⁻ redox couple. This finding confirmed the results of our J–V curve analysis. Three curves based on RGO, Fe₃O₄@RGO-NMCC, and Pt CEs exhibited two pairs of redox peaks. The redox peak at low potential was attributed to the reaction: I₃⁻ + 2e⁻ ↔ 3I⁻. The separation between the anodic and cathodic peaks (ΔE) was inversely related to the rate of the above redox reaction and regeneration rate of the I⁻/I₃⁻ redox couple. The ΔE value for Fe₃O₄@RGO-NMCC CEs (104 mV) was significantly smaller than that for RGO (280 mV) and Pt (126 mV), implying notable catalytic behaviors for the reduction of I₃⁻ or I₂ to I⁻. Although the Fe₃O₄@RGO CE had higher cathodic peak current density and smaller ΔE than the Pt reference, its redox potential (E_R) (0.23 V vs. Ag/AgCl) is lower than that of the Pt CE (0.29 V vs. Ag/AgCl), which caused the Fe₃O₄@RGO CE to deliver a low PCE of 9.04% compared to that of the Pt reference (9.46%) in Fig. 2b. The peak current of RGO in CV was relatively large, indicating that electron transfer readily occurred. However, the redox peak was not prominent, indicating few surface catalytic sites. Using symmetrical cells consisting of two identical CEs, Tafel polarization curves were measured as shown in Fig. 2d. Among the four CEs, Fe₃O₄ CEs showed the least limited exchange current density, which again demonstrated that massive interparticle boundary detrimentally affected electron transfer. For Fe₃O₄@RGO-NMCC symmetrical cells, the charge transfer in the Tafel zone was remarkably higher than that of Fe₃O₄ and RGO, indicating superior catalytic activity. This finding was in good agreement with the CV measurements. Thus, the Fe₃O₄@RGO-NMCC not only increased the catalytic activity but also accelerated electron transfer between the interfaces, which in turn increased the performance of the corresponding DSCs.

The development of flexible electrodes is the main research direction of current and future portable and curved electronic devices.^19–21 Flexible Fe₃O₄@RGO-NMCC CEs were prepared on a polyethylene naphthalate/indium tin oxide plastic substrate, employing flexible Pt CEs (prepared by chemical reduction) as the reference. The J–V curves of DSCs based on flexible CEs are shown in Fig. 3a. DSCs based on flexible Fe₃O₄@RGO-NMCC CEs gave a PCE of 8.0%, which was higher than that of DSCs based on flexible Pt (7.35%), which could be attributed to the higher photocurrent density of DSCs based on flexible Fe₃O₄@RGO-NMCC CEs. Fig. 3b shows the incident-photon-to-current conversion efficiency for DSCs based on flexible Fe₃O₄@RGO-NMCC and Pt CEs. The integrated currents for DSCs based on flexible Fe₃O₄@RGO-NMCC and Pt CEs were 15.30 and 11.76 mA cm⁻², respectively, agreeing with the J–V results. As for a rigid DSC system, the Fe₃O₄@RGO CE delivered a low PCE of 9.04% compared to that of the Pt reference (9.46%) in Fig. 2b, whereas the DSCs based on flexible Fe₃O₄@RGO CEs gave a larger PCE than Pt. The reason for these results was attributed to the different preparing method of Pt on the rigid FTO/glass and flexible PEN/ITO substrate. The Pt on rigid FTO/glass was prepared by a high-temperature pyrolytic method, whereas Pt on the flexible PEN/ITO substrate was prepared by a chemical reduction method because of the instability of the flexible plastic substrate at high temperatures (>150 °C). To confirm superior performance of the flexible Fe₃O₄@RGO CE to flexible Pt on the PEN/ITO substrate, Tafel-polarization and CV and electrochemical impedance spectroscopy (EIS) were carried out to reveal the catalytic activity and electron transport of flexible CEs.

Fig. 3 Photovoltaic and electrochemical property characterization of CEs fabricated on the flexible PEN/ITO substrate: (a) J–V curves of DSCs based on different flexible CEs; (b) IPCE of DSCs based on different flexible CEs; (c) Tafel polarization curves of the I₃⁻/I⁻ symmetrical cells; (d) CV curves of various flexible CEs in the I₃⁻/I⁻ electrolyte; (e) electrochemical impedance spectroscopy of the I₃⁻/I⁻ symmetrical cells; (f) PCE variations after 50 times bending randomly selected two flexible CEs around one pen with a diameter of 1 cm.

The exchange current density (J₀) (Fig. 3c) for flexible Fe₃O₄@RGO-NMCC CEs (1.61 mA cm⁻²) was significantly higher than that for Pt (1.36 mA cm⁻²), implying its faster electron transfer than that of flexible Pt CEs. The ΔE value for flexible Fe₃O₄@RGO-NMCC (282 mV) CEs was significantly less than that for Pt (544 mV), implying its superior catalytic behavior to flexible Pt CEs. EIS was used to reveal the inherent interface resistance, with the results shown in Fig. 3e. Meanwhile, an equivalent circuit diagram (Fig. 3e, inset) is provided for fitting Nyquist plots with the Z-view software. Each plot comprised two irregular semicircles, with the first one originating from the charge transfer resistance (R_ct) at the CE/electrolyte interface. By contrast, the second semicircle arises from the Nernst diffusion impedance (Z_N) of I₃⁻/I⁻ within the electrolyte. Usually, R_ct occurs in the high frequency region, whereas the Z_N appears in the low frequency region. In addition, the value intercepted on the real axis of the Nyquist plot was attributed to the series resistance (R_s). The fitting results of Nyquist plots are listed in Table 2, which showed that the R_ct of DSCs based on flexible Fe₃O₄@RGO-NMCC CEs (3.85 Ω) was smaller than that of the flexible Pt CE (4.50 Ω). Smaller charge transfer resistance based on flexible Fe₃O₄@RGO-NMCC CEs facilitated electron transfer. Suitable electron transfers are highly relevant to their structural advantages, e.g., electron hopping between Fe²⁺ and Fe³⁺. Electrons were transported freely along a broad two-dimensional conductive surface based on RGO. Thus, the superior catalytic activity and faster electron transport of flexible Fe₃O₄@RGO CEs enhanced the photocurrent density of the corresponding DSCs. Bending tests (Fig. 3f) indicated that the PCE of DSCs based on flexible Fe₃O₄@RGO-NMCC CEs could maintain 80% of their initial PCE after bending 50 times. However, the performance of DSCs based on flexible Pt CEs was notably decreased after bending 50 times. As we all know, a two-dimensional material or its thin film has better mechanical flexibility than that of nanoparticles. We ascribed the deterioration in the performance of DSCs based on the flexible Pt CE after bending to its mechanical brittleness (the crack defects on the flexible Pt CE produced by repeated bending). Considering their photovoltaic and anti-bending performance, flexible Fe₃O₄@RGO-NMCC CEs are ideal. The stability test of DSCs based on Fe₃O₄@RGO-NMCC CEs was conducted as shown in Fig. 4. After 2000 h, DSCs based on Fe₃O₄@RGO-NMCC CEs maintained 60% of their initial PCE.

Table 2 Photovoltaic parameters of the DSCs based on flexible CEs and EIS parameters of the symmetrical cells based on flexible CEs

Flexible CEs	V oc (V)	J sc (mA cm^−2)	J sc (mA cm^−2) from IPCE	FF	PCE (%)	R s (Ω)	R ct (Ω)
PEN/ITO/Fe3O4@RGO	0.75 ± 0.01	15.63 ± 0.2	15.30 ± 0.2	0.69 ± 0.01	8.0 ± 0.02	19.6 ± 0.5	3.85 ± 0.5
PEN/ITO/Pt	0.80 ± 0.02	12.10 ± 0.1	11.76 ± 0.2	0.76 ± 0.01	7.35 ± 0.01	14.6 ± 0.5	4.50 ± 0.6

---

Fig. 4 Stability of DSCs based on Fe₃O₄@RGO-NMCC CEs on the FTO substrate (25 °C, 30% humidity, encapsulation, AM 1.5, 100 mW cm⁻²).

Conclusions

In summary, we report the generation of nano–micro composite catalysts made of highly dispersed Fe₃O₄ nanoparticles fixed on RGO sheets as the CE in DSCs using a simple one-pot solvothermal approach. Compared with pure Fe₃O₄ and RGO CEs, the Fe₃O₄@RGO-NMCC CE exhibited superior electrocatalysis for the catalytic reduction of the I⁻/I₃⁻ redox couple. Notably, DSCs using rigid and flexible Fe₃O₄@RGO-NMCC CEs achieved high PCEs reaching 9% and 8% on FTO/glass substrates and flexible polymer substrates, respectively. These values are the highest reported efficiencies for DSCs based on a flexible Pt-free CE. We ascribed the notable catalytic performance of Fe₃O₄@RGO-NMCC to faster electron hopping between Fe²⁺ and Fe³⁺ and free electron transport by broad electron transport-sheets of RGO. Finally, Fe₃O₄@RGO-NMCC exhibited good anti-bending and stability in the practical application of DSCs.

Acknowledgements

This work was financially supported by the Shandong Province Natural Science Foundation (Grant No. BS2015NJ013), Science and Technology Innovation Foundation for the Uinversity or College Students (Grant No. SF2014002), and Research Fund for the Doctoral Program of Liaocheng University (Grant No. 31805).

Notes and references

1. B. Oregan, M. Grat and zel, Nature, 1991, 353, 737–740.
2. S. Mathew, A. Yella, P. Gao, R. Humphry-Baker, B. F. E. Curchod, N. Ashari-Astani, I. Tavernelli, U. Rothlisberger, M. K. Nazeeruddin and M. Graetzel, Nat. Chem., 2014, 6, 242–247.
3. M. J. Ju, J. C. Kim, H.-J. Choi, I. T. Choi, S. G. Kim, K. Lim, J. Ko, J.-J. Lee, I.-Y. Jeon, J.-B. Baek and H. K. Kim, ACS Nano, 2013, 7, 5243–5250.
4. M. J. Ju, I.-Y. Jeon, K. Lim, J. C. Kim, H.-J. Choi, I. T. Choi, Y. K. Eom, Y. J. Kwon, J. Ko and J.-J. Lee, Adv. Mater., 2014, 26, 3055–3062.
5. M. K. Wang, A. M. Anghel, B. Marsan, N. L. C. Ha, N. Pootrakulchote, S. M. Zakeeruddin and M. Gratzel, J. Am. Chem. Soc., 2009, 131, 15976–15977.
6. M. Wu, X. Lin, Y. Wang, L. Wang, W. Guo, D. Qu, X. Peng, A. Hagfeldt, M. Graetzel and T. Ma, J. Am. Chem. Soc., 2012, 134, 3419–3428.
7. A. Kay and M. Gratzel, Sol. Energy Mater. Sol. Cells, 1996, 44, 99–117.
8. M. Wu, X. Lin, T. Wang, J. Qiu and T. Ma, Energy Environ. Sci., 2011, 4, 2308–2315.
9. M. J. Ju, I.-Y. Jeon, K. Lim, J. C. Kim, H.-J. Choi, I. T. Choi, Y. K. Eom, Y. J. Kwon, J. Ko, J.-J. Lee, J.-B. Baek and H. K. Kim, Energy Environ. Sci., 2014, 7, 1044–1052.
10. X. Meng, C. Yu, X. Song, Y. Liu, S. Liang, Z. Liu, C. Hao and J. Qiu, Adv. Energy Mater., 2015, 5 DOI:10.1002/aenm.201500180.
11. T. Yohannes and O. Inganas, Sol. Energy Mater. Sol. Cells, 1998, 51, 193–202.
12. Q. Li, J. Wu, Q. Tang, Z. Lan, P. Li, J. Lin and L. Fan, Electrochem. Commun., 2008, 10, 1299–1302.
13. Y. Hou, D. Wang, X. H. Yang, W. Q. Fang, B. Zhang, H. F. Wang, G. Z. Lu, P. Hu, H. J. Zhao and H. G. Yang, Nat. Commun., 2013, 4, 1583.
14. L. Wang, Y. Shi, H. Zhang, X. Bai, Y. Wang and T. Ma, J. Mater. Chem. A, 2014, 2, 15279–15283.
15. Q. Yu, L. A. Jauregui, W. Wu, R. Colby, J. Tian, Z. Su, H. Cao, Z. Liu, D. Pandey, D. Wei, T. F. Chung, P. Peng, N. P. Guisinger, E. A. Stach, J. Bao, S.-S. Pei and Y. P. Chen, Nat. Mater., 2011, 10, 443–449.
16. D. Dimos, P. Chaudhari and J. Mannhart, Phys. Rev. B: Condens. Matter Mater. Phys., 1990, 41, 4038.
17. L. Wang, Y. Shi, Y. Wang, H. Zhang, H. Zhou, Y. Wei, S. Tao and T. Ma, Chem. Commun., 2014, 50, 1701–1703.
18. S. W. Lee, S. J. Kim, I. B. Shim, S. Bae and C. S. Kim, IEEE Trans. Magn., 2005, 41, 4114–4116.
19. H. Zhou, Y. Shi, D. Qin, J. An, L. Chu, C. Wang, Y. Wang, W. Guo, L. Wang and T. Ma, J. Mater. Chem. A, 2013, 1, 3932–3937.
20. H. Zhou, Y. Shi, K. Wang, Q. Dong, X. Bai, Y. Xing, Y. Du and T. Ma, J. Phys. Chem. C, 2015, 119, 4600–4605.
21. L. Li, Z. Wu, S. Yuan and X.-B. Zhang, Energy Environ. Sci., 2014, 7, 2101–2122.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ta06525a

---