Reduced graphene oxide/  -Fe 2 O 3 fibres as active material for supercapacitor

The composite hydrogel, composed of reduced graphene oxide and  -Fe 2 O 3 fibres (rGO/  Fe 2 O 3 ), was successfully prepared by the hydrothermal procedure starting from GO and  Fe 2 O 3 nanofibers. According to the SEM and XRD results,  -Fe 2 O 3 fibres are distributed between rGO sheets increasing the inter-sheet space. The rGO/  -Fe 2 O 3 composite was tested as an active material in supercapacitor by means of cyclic voltammetry, galvanostatic charging/discharging and electrochemical impedance spectroscopy in 0.5 mol dm -3 Na 2 SO 4 . The obtained results confirmed a positive effect of the  -Fe 2 O 3 addition on capacitive properties. Improved capacitive properties of the composite make


Synthesis of graphene oxide (GO)
GO was synthesized from graphite powder according to the Hummers´ procedure. [27] In a typical procedure, 3.0 g of graphite and 1.5 g NaNO 3 was mixed with 69 mL of concentrated sulphuric acid under a constant stirring in an ice bath where the temperature close to 0°C was reached. Afterwards, 9.0 g of potassium permanganate was added gradually to the solution while the temperature was kept below 20°C. After 20 minutes, the flask was moved to a water bath and the solution was stirred at 35C for 30 min. The resulting solution was diluted by gradually adding 120 cm 3 of water and maintaining the temperature at 98 C for 0.5 h. Next, the flask was cooled to room temperature after which the mixture was poured into 420 cm 3 of water with vigorous stirring, and 30% H2O2 was added to the solution in order to quench the reaction and remove the excess oxidant. The resulting mixture was washed with 5% HCl and deionised water by using centrifugation until the pH value approached 7.0. Graphene oxide suspension was prepared by the ultrasonication of graphite oxide in deionised water (40 kHz) during one hour and centrifuged at 4000 rpm to remove any non-exfoliated graphite oxide particles. The GO concentration was 6.7 mg cm -3 .

Synthesis of composite material -(rGO/iron oxide fibres)
Synthesis of a composite material was carried out hydrothermally using iron oxide fibres and GO suspension. Two types of samples were synthesized: (i) reference sample -rGO hydrogel (denoted as rGO1) and (ii) composite sample containing rGO and 1w% of fibres -rGO/-Fe 2 O 3 hydrogel (denoted as rGO2).
For a synthesis of rGO1 sample, 5.97 cm 3 of GO suspension (6.7 mg cm -3 ) and 10 cm 3 of olive leaf extract solution (2.8w%) were mixed and an appropriate amount of deionised H 2 O was added to form a 20 cm 3 reaction mixture in order to obtain a GO concentration of 2 mg cm -3 . The olive leaf extract served as a reducing agent. [28] In the case of rGO2 sample the same procedure was used as for rGO1 except an appropriate amount of fibres was added to the reaction mixture and ultrasonically dispersed during 10 minutes. pH value was 4 in both cases. Finally, the prepared reaction mixtures were put in autoclaves at 50 °C during 1 hour and then gradually heated up to 120 °C and left for 5 hours. Afterward, the autoclaves were gradually cooled to room temperature and the samples, in the form of hydrogels, were washed with deionized water several times and freeze-dried in order to obtain an aerogel. After the synthesis, morphological and electrochemical characterisations of the samples were carried out.

Characterization of samples
The morphology of prepared samples was inspected with a Jeol Ltd. thermal field emission scanning electron microscope (FE-SEM, model JSM-7000F). The FE-SEM was coupled with Oxford Instruments EDS/ INCA 350 (energy dispersive X-ray analyser). XRD diffractometer (model APD 2000) manufactured by Italstructures was used to identify the phase composition. The XRD patterns were recorded with CuKradiation at 40 kV and 30 mA in the 2 range from 8 to 70°.
All electrochemical experiments were carried out in an aqueous solution of Na 2 SO 4 at room temperature in a two-electrode cell and for this purpose two symmetric supercapacitors were prepared. The supercapacitor SC1 was assembled using a nickel sheet (1 cm 2 ) and by applying a piece of active material rGO1 hydrogel on the nickel surface. The supercapacitor SC2 was assembled in the same way except that rGO2 hydrogel was used. The hydrogel was obtained by soaking the aerogel in Na 2 SO 4 solution during the night. The electrodes were separated by glass paper fibres wetted with 0.5 mol dm -3 Na 2 SO 4 solutions. The supercapacitor was then placed in an outer casing, in order to prevent electrolyte evaporation, and pressed between two glass plates using clamps.
The galvanostatic charge-discharge (GCD) measurements were performed at a current value of 0.80 or 1.46 A g -1 in the voltage range between 0 and 1.2 V. The cyclic voltammetry (CV) measurements were carried out at a scan rate of 50 mV s -1 in the voltage range between 0 and 1.2 V. The impedance measurements (EIS) were performed in the frequency range between 7 The chemical and phase composition of nanofibers was investigated by EDS and XRD techniques. The EDS spectrum (Fig.1b) of the selected area (Fig 1a) is shown together with the results of element analysis. The nanofibers consisted of 63.95 atomic% of oxygen and 36.05 atomic% of iron which is close to the stoichiometric ratio characteristic for hematite, - The confirmation of the hematite phase is also observed from the diffraction pattern of the nanofibers (Fig 1c). Experimental data were compared with a reference pattern for hematite, JCPDS-ICDD PDF card No. 33-0664. [30] According to the diffractogram, the fibres have

Morphological and structural characteristics of rGO1 and rGO2 samples
The morphology of the reference, rGO1 and composite, rGO2 sample was investigated by SEM and the obtained micrographs are shown in Fig. 2. The rGO1 aerogel is composed of many wrinkled sheets that create a porous structure (Fig. 2a). In Figs. 2b and c the surface of the composite sample rGO2 is visible. As can be noticed, -Fe 2 O 3 fibres are distributed on the surface (Fig. 2b) and between rGO sheets (Fig. 2c). Such distribution of fibres increases the inter-sheet space and decreases the stacking of rGO sheets. [31][32][33] Additionally, a fibril Subramaniyam, [4] and J.G. Lee. [7] In order to determine the influence of -Fe 2 O 3 on structural properties of the rGO2, the prepared samples were analysed with XRD, Figure 3. The obtained results show for rGO1 a broad diffraction peak centred at 24° (corresponding to a d-spacing value or interlayer distance of 3.71Ǻ). This peak was assigned to the c (002) reflection of the graphite derived from the short-range order in stacked graphene sheets indicating a significant reduction of GO after the hydrothermal process. [34,35] Similar diffraction peak was registered in the case of rGO2 what means that stacking also occurred in the case of rGO2. However, the diffraction peak obtained for rGO2 moves to 23°, which corresponds to higher d-spacing value, d = 3.88 Ǻ. Therefore it was concluded that although stacking was not prevented completely, the presence of -Fe 2 O 3 influenced structural properties of rGO2. Additional diffraction peaks, visible in diffractogram of the rGO2, are characteristic for the hematite phase -Fe 2 O 3 (JCPDS no. 33-0664) and for impurities present within the sample. The most probably impurities can be correlated to Na 2 SO 4 (JCPDS 96-101-0523) in which rGO2 sample was stored. Despite intensive washing, Na 2 SO 4 was not completely removed.

Electrochemical performances of rGO1 and rGO2
The electrochemical charge storage characteristics of prepared samples were investigated by assembling supercapacitors SC1 and SC2 with rGO1 and rGO2 active materials, respectively.
Three different techniques, cyclic voltammetry (CV), galvanostatic charge/discharge (GCD), and electrochemical impedance spectroscopy (EIS) measurements were used. All measurements were done in a two-electrode system. Fig. 4 presents the CV curves of the SC1 supercapacitor (a, b) and the SC2 supercapacitor (c, d) recorded at a scan rate of 50 mV s -1 before (a, c) and after (b, d) charge/discharge tests. It is well known that an ideal capacitor should retain a constant current during CV scans according to the relation I(t) = C (dU/dt), where I is the current, U is the voltage and t is the time. [36][37][38] Although CV curves of both supercapacitors are characterized by a good capacitive, almost rectangular CV shape, SC2 shows higher specific currents (Fig. 4cd) and a smaller deviation from the rectangular shape compared to SC1 (Fig. 4ab). The polarisation of each electrode was carried out by changing the polarity that resulted in two cyclic voltamograms, one from 0 V to 1.2 V and the other from 0 V to -1.2 V. The same response was obtained by applying a positive or a negative charge at each electrode, as it is evident from CV responses. This behaviour is quite reasonable since the symmetric supercapacitors were tested. However, after 1000 cycles of GCD (Fig. 5), a different behaviour was observed. CV responses at positive voltage values, i.e. responses obtained by using the same electrode polarity that was used for GCD tests (from 0 to 1.2 V), are significantly different before (Fig. 4a, c) and after GCD test (Fig. 4b, d). CV responses at negative voltage values, i.e. responses obtained by using the opposite polarity than those used for GCD tests (from 0 to -1.2 V), are almost the same before and after GCD. From the obtained results it can be concluded that the degradation of active material during GCD test was different for each electrode. Similar effect was obtained for supercapacitor SC1 and supercapacitor SC2 indicating that the presence of -Fe 2 O 3 did not significantly influence the stability of the active material.
The specific capacitance (C s /Fg -1 ) values of the prepared electrodes were calculated using equation (1) and the results are shown in Table 1.
In equation (1), Q/C is the sum of anodic and cathodic charges obtained by the integration of the cyclic voltammograms in the voltage range ∆U/V, and m/g is the mass of the active material at one electrode.
From Table 1 it is evident that C s in the negative voltage range has not been changed while C s values in the positive voltage range were different before and after 1000 cycles of charging/discharging, which supports the previous conclusion. It is also evident that higher C s values are registered for SC2 compared to SC1. It can be explained by the influence of -Fe 2 O 3 fibres on structural properties of the composite material. As it was shown before, -Fe 2 O 3 fibres are incorporated between rGO layers (see Fig. 2bc) decreasing an agglomeration of rGO and increasing its surface area available for a charge/discharge process. Consequently, specific capacitance values are increased. The response characteristic for pseudocapacitive redox reaction of -Fe 2 O 3 was not noticed in our experiment and therefore we believe that a pseudocapacitive redox reaction of -Fe 2 O 3 does not take place.
The long-term stability of the prepared samples was investigated by the GCD test during 1000 cycles and the results are shown in Fig. 5. GCD profiles were recorded at selected current densities in the voltage range from 0 to 1.2 V. Both GCD profiles exhibit three different regions for charging and three different regions for discharging processes: [36] (i) stabilization period in the first 5 s at the open circuit potential (OCP), (ii) IR drop caused by the internal system resistance, (iii) charging/discharging profile. During the discharging process, a linear relationship between voltage and time is obtained, while the relationship obtained during the charging process is not linear and this deviation is more pronounced for the SC1 supercapacitor. The linear relationship between voltage and time indicates a double layer characteristic and good capacitive behaviour of the investigated system. The difference between these two capacitors can be explained by the influence of -Fe 2 O 3 fibres on rGO capacitive properties. The linear GCD response is in accordance with the CV results (Fig. 4) and it also indicates better capacitive behaviour of the SC2 supercapacitor.
From the galvanostatic discharge curves gravimetric specific capacitances (C s ) of investigated samples were calculated according to the equation (2): [23,36] where C s is the specific capacitance of the cell, I is the constant current, t is the discharging time, m is the total weight of one electrode active material, and U is the voltage drop during the discharging process. The obtained C s values are shown in Table 2. As can be seen the Specific energy (E s ), and specific power (P s ), as key parameters for fabricating energy storage devices, were calculated according to the equations: [36,39] E s = 0.5•C s •U 2 (3) where ESR is the electrical series resistance, which represents total internal resistance of a capacitor that includes electrolyte, contact, active material and separator resistances. [5,39,40] ESR was calculated from the voltage drop of the galvanostatic discharge curve (IR drop). ESR values for the SC1 ranged from 34  to 42 , while for the SC2 values ranged from 35 to40 According to the obtained values it can be concluded that there is no significant difference between these two supercapacitors. All calculated parameters are listed in Table 2.
The calculated specific energy values are similar for both supercapacitors, although higher C s values were registered for the SC2. However, a more significant voltage drop due to the open circuit discharge in the case of SC2 (see Fig. 5b) has affected the reduction in the amount of stored specific energy and specific power of the supercapacitor. The energy and power values calculated in this work are in accordance with the literature [41][42][43] and they indicate high power capability of the prepared supercapacitors.
Electrochemical impedance spectroscopy (EIS) was used for in situ study of current collector/active material and active material/electrolyte interfaces of investigated supercapacitors. Impedance spectra of the SC1 and SC2 supercapacitors, recorded in 0.5 mol dm -3 Na 2 SO 4 , are presented in the form of Nyquist and Bode plots in Fig. 6. As can be seen from Fig. 6, the appearance of parasitic inductive impedance (positive imaginary part of impedance) can be noticed at the highest frequencies. At higher frequencies, a capacitive loop for both samples is observed. This capacitive loop can be described by the charge transfer at the current collector or the formation of the interfacial region between the current collector and active material. [39,40,44] At high-to-medium frequencies a diffusional response due to the effect of electrode porosity ( = -45°) followed by a capacitive response at lower frequencies is visible. Similar impedance responses were obtained for the supercapacitors based on carbon nanofibers/carbon nanosheets, [45] FeOx/rGO [46] and Fe 2 O 3 /rGO [12] systems.
The investigated systems are well described by the electrical equivalent circuit (EEC), shown as an inset in Fig. 6, which consists of the following elements: inductor (L) in series with electrolyte resistance (R HF ) followed by (Q 1 R 1 W o ) and constant phase element (Q 2 ). R 1 is the contact resistance and Q 1 is the double layer charging at the current collector/active material interface. W o is the finite-length Warburg element for diffusion of electrolyte ions into the pores of the active material (rGO1 or rGO2). Q 2 represents the double layer charging at the active material/electrolyte interface. Because the measured capacitive response is not generally ideal due to certain heterogeneity of the electrode surface, [47] a constant phase element (CPE) has been introduced for fitting the spectra, instead of an ideal capacitance element. Its impedance can be defined by Z(CPE) = [Q(j) n ] 1 , where Q is a constant,  is the angular frequency, and n is the CPE power. The factor n is an adjustable parameter, which has values between -1 and 1; a value of -1 is characteristic for an inductance, a value of 1 corresponds to a capacitor, a value of 0 corresponds to a resistor. [48] The values of impedance parameters, obtained by the performed analysis, are listed in Table 3. As evidenced from Table 3 and Fig. 6, a very good agreement between the experimental (symbols) and modelled data (solid lines) was obtained using the EEC presented as the inset in Fig. 6.
According to the results after charging/discharging of the SC1 and SC2 supercapacitors the most significant change was obtained for Q 1 , R 1 and W o elements. Since these elements represent the current collector/active material interface, changes at the current collector (Ni) surface and/or in the structure of active materials during these GCD tests can be assumed.