Supercapacitors have many advantages over batteries, such as higher specific power and energy, easy manufacturing, and a broader range of uses, which have sparked a lot of research interest in recent years. However, the low specific energy of supercapacitors relative to batteries is one of their main disadvantages[1-3]. Conducting polymers, metal/metal oxides, and various carbonaceous materials, such as graphene, graphene oxide, carbon nanofiber (CNF), carbon nanotube (CNT), and activated carbon, have been employed to boost the specific energy of supercapacitors[3-6].

Currently, a slurry mix of above mentioned carbon compounds with an electrically insulating binder is applied to the current collector and dried in the typical electrode manufacturing process of supercapacitors. The inclusion of insulating binder, increases the electrode resistances, resulting in low capacitance followed by low energy and power. Therefore, the best option could be binder-free electrodes that have good contact with the current collector.

The use of CNT and CNF in supercapacitors has been reported[7, 8]. Some studies have attempted to individually grow CNTs before coating them onto the current collector[9-11], which again causes the above mentioned issues . Chen et al. fabricated CNFs using the electrospinning process to create flexible supercapacitors that exhibit an areal capacitance of 64.2 μF/cm² at 0.5 A/g[12]. Because of its unique structure such as high surface area, one-dimensional property, high porosity, and exceptional mechanical stability, CNF is well known as an ideal framework for reducing manufacturing costs and developing environmentally friendly procedures for making an electrical double-layer capacitor electrode. Our previous work also demonstrated the supercapacitance performance of commercial CNF, with a gravimetric capacitance of 73 F/g at 0.5 A/g[13]. However, there are still limitations in its application because of its relatively high resistance and the complicated synthesis processes.

Commercially available porous conductive substrates (current collectors) with large interfacial areas, shorter ionic diffusion distances, and better charge separation and transport have been used as the base for creating electrodes with three-dimensional (3D) architectures. These substrates include nickel foam (NF), carbon paper, and carbon cloths. Among these, NFs have been utilized more as they offer a large active surface area, adjustable thickness, a highly conductive continuous porous 3D network, and flexibility[12, 14].

CNFs are also hydrophilic, which promotes good adhesion to the surface of the current collector, and it could decrease resistance due to its binder-free nature as well as provide increased capacitance. This makes CNF electrodes an interesting prospect. As a result, CNF could have a good contact with the NF, which could enhance conductivity and flexibility because of the binder-free and synergistic properties of CNF and NF.

In this study, CNF is directly spin-coated on NF to create flexible supercapacitor electrodes. The NF provides good electric conductivity, which results in good ion diffusion paths for the intercalation of ions with the excellent pore structure of the NF. The resulting electrodes provide higher specific energy and power. Our synthesized electrodes provide a specific energy of 8 Wh/kg with a specific power of 400 W/kg, and it retained 84% of the capacitance even after 3000 cycles at 1 A/g with excellent flexibility.

1. Electrode preparation

A piece of commercial NF (> 99.99%, thickness of 1.6 mm, MTI Korea) was cut into rectangles of 3 cm × 1 cm, and then washed with a HCl solution of 3 mol/L by ultrasonic cleansing for 5 min to remove surface impurities, and then another 15 min with each of deionized (DI) water and ethanol washing. The NF strips were then vacuum dried at 65 °C for 24 h and stored in a desiccator prior to use. In this paper, NF denotes the processed nickel foam.

Next, the slurry was prepared as follows: a solution containing 600 mg of CNF (powder, diameter of 100 nm, length of 20–200 μm) and 80 ml of DI water was prepared after 6 h on a magnetic stirring. Another solution containing 90 mg of poly(sodium 4-styrenesulfonate) was stirred for 10 min after mixing with 20 ml of DI water. The two solutions were then combined using a homogenizer for 1 h to make the slurry. The NF strip was spin-coated with the slurry for 30 min at 500 rpm, then the coated strip was dried in a vacuum dryer for 12 h at 50 °C; the new electrode was labeled as CNF@NF. NF was the basis used for comparisons. The weight difference before and after deposition was used to accurately determine the amount of material loaded onto the NF substrate. Consequently, the average loading of the combined materials on the NF substrate was 1.61 mg/cm². A two-electrode system was prepared for symmetric supercapacitors with the separator (Whatman GF/C binder free, 0.26 mm thickness) and a gel electrolyte of polyvinyl alcohol/potassium hydroxide. EC-Lab (Biologic, SP-150/200, France) was used to analyze the electrochemical performance of the CNF@NF sample based on chronocoulometry (CC), cyclic voltammetry (CV), potentia electrochemical impedance spectroscopy (PEIS), and galvanostatic cycling with potential limitation (GCPL) measurements.

2. Characterization

Using field emission scanning electron microscopy (SEM, S-4300, Hitachi, Japan), the surface structure and morphology of the samples were studied. Energy dispersive X-ray spectroscopy (EDS; S-4300; JEOL) was used to analyze the elemental composition data as well as elemental mapping of the samples. CV measurement was conducted to determine the electrochemical surface reaction at the electrodes during the charging and discharging process, which was conducted at various sweep rates (10, 20, 50, 100, and 200 mV/s) and applied from 0 to 0.8 V potential. The charge–discharge behavior of the materials was studied by GCPL at various current densities of 1 to 5 A/g with a potential up to 0.8 V. Results from PEIS in the 100 mHz to 500 kHz frequency range were obtained, and CC was used to study the sample’s adsorption capacity and active electrochemical surface area. The electrochemical surface area can be expressed mathematically from Anson’s plot[13-15].

1. Physicochemical characterization

The SEM images of NF and CNF@NF are shown in Fig. 1. Figure 1(a) and (b) shows the NF prior to CNF coating, which exhibited smooth surfaces and a 3D macroporous network structure. The typical NF pore size is 0.25 ± 0.02 mm. CNF was homogeneously coated on the NF surface as shown in Fig. 1(c). A detailed fibrous structure with a straight and smooth surface of the CNF is shown in Fig. 1(d). The one-dimensional structure of the CNF was clearly seen in the CNF@NF electrode. EDS mapping of NF and CNF@NF, as shown in Fig. 1(e) and (f), revealed the good distribution of Ni and carbon after CNF coating on NF. The elemental data were acquired from EDS and presented in Table 1. NF consists of 17 wt% of carbon, 10 wt% of oxygen, and 73 wt% of nickel, while CNF@NF consists of 48 wt% of carbon, 25 wt% of oxygen, and 27 wt% of nickel.

Table 1 Elemental data obtained from EDS.

Samples	Element	wt%	at%	C/Ni
NF	C	17	44	0.23
	O	10	19
	Ni	73	37
CNF@NF	C	48	66	1.77
	O	25	26
	Ni	27	8

Figure 1. (Color online) Scanning electron microscopy result of (a,b) NF and (c,d) CNF@NF, and energy dispersive X-ray spectroscopy mapping of (e) NF and (f) CNF@NF.

2. Electrochemical performance

The electrochemical surface area of NF and CNF@NF was estimated by CC measurement[15, 16]. The surface area, which is proportional to the slope, was obtained as 0.014 cm² for NF and 0.035 cm² for CNF@NF (Fig. 2(a)). The CNF@NF electrode has more than two times larger surface area compared to the NF electrode. The electrochemical active surface area is directly proportional to the adsorption capacity of the supercapacitor[13], with CNF@NF having a higher adsorption capacity compared to NF.

Figure 2. (Color online) (a) CC, (b) PEIS, (c) CV results of samples at 50 mV/s, CV results at various scan rates of (d) CNF @NF and (e) NF with (f) the resultant specific capacitance as a function of the scan rate.

The performance of supercapacitors is also closely related to their electrochemical impedance. The Nyquist plots of the NF and CNF@NF are shown in Fig. 2(b), where |Zre| is a function of −|Zim|[17, 18]. The charge-transfer resistance was determined based on the impedance results and a fitting software[18-20] as follows: 5.5 Ω for NF and 3.6 Ω for CNF@NF. This confirms that CNF@NF has a lower charge-transfer resistance compared with NF.

Figure 2(c) shows the CV results of the electrodes at 50 mV/s scan rate. The smaller CV area and lower currents are found in NF. After CNF coating on NF, the CV area and currents of the coated NF increased because of its larger surface area and lower resistance. The detailed CV results of the electrodes measured at different scan rates ranging from 10 to 200 mV/s are shown in Fig. 2(d) and (e). The specific capacitance determined from the CV curve[21, 22] is presented in Fig. 2(f), in which the specific capacitance of CNF@NF is obviously larger than that of NF. CNF@NF produced the largest capacitance at 83 F/g while the corresponding capacitance of NF was 43 F/g, which shows the significant difference in capacitance between the two.

Galvanostatic charge–discharge experiments were conducted to further investigate the electrochemical characteristics. The discharge slope (dV/dt) of the GCPL in Fig. 3 is used to estimate the capacitance which is equal to the current divided by the discharge slope[23]. Figure 3(a) shows the comparison of the GCPL curves of the electrodes at a current density of 1 A/g. The GCPL results indicate that CNF@NF has a longer discharge time than NF. The specific capacitances obtained from the GCPL results were 40 and 87 F/g for NF and CNF@NF, respectively. The GCPL data at the current densities from 1 to 5 A/g are shown in Fig. 3(b) and (c) for CNF@NF and NF, respectively, and the resultant specific capacitance as a function of the current density is shown in Fig. 3(d). It is concluded that CNF@NF has a higher capacitance at all current densities compared to NF.

Figure 3. (Color online) GCPL result of the electrodes (a) at 1 A/g and at various current densities for (b) CNF@NF and (c) NF with (d) the resultant specific capacitance as a function of the current density.

The electrochemical properties of the flexible electrodes were further studied by CV, PEIS, and GCPL using a gel electrolyte in the scan rate and current density range. Even at higher scan rates, the CV curves (Fig. 4(a)) retain their rectangular shape, and the absence of redox peaks indicates the strong charging and discharging rate capabilities of the CNF@NF electrode. In the Nyquist plot (Fig. 4(b)) of the CNF@NF electrode, the small semicircle corresponds to the charge-transfer resistance of 2.4 Ω. GCPL measurements were conducted at different current densities to obtain quantitative capacitance and more accurately characterize the capacitive behavior of the electrodes throughout the charging and discharging process. The results show that the curves at the current densities have triangular charging and discharging slopes (Fig. 4(c)), and the resultant capacitance of 89 F/g was obtained from the flexible supercapacitor of CNF@NF (Fig. 4(d)). The Ragone plot is also shown in Fig. 4(e). The specific energy of 8 Wh/kg was obtained with 400 W/kg of specific power at 1 A/g current density. The inset in the figure shows that our assembled CNF@NF device, as the flexible supercapacitor, when linked to a commercial LED after charging, actually lights an LED. Capacitance retention and coulombic efficiency were also studied (Fig. 4(f)). The discharge curve at 1 A/g up to 3000 cycles was also used to test the capacitance retention and coulombic efficiency of the CNF@NF device, and 84% and 85% in capacitance retention and coulombic efficiency were achieved, respectively. The mechanical stability and flexibility of the CNF@NF device determined by CV measurement at 50 mV/s, as shown in Fig. 5, show stability and flexibility at different bending angles (Fig. 5(a)). Figure 5(b) shows the preparation and bending phases of the device. The capacitance performance of the CNF@NF device at different current densities is shown in Table 2.

Table 2 Summary of capacitance performance for CNF@NF device.

Current density	Δt (s)	Cs (F/g)	S. E (Wh/kg)	S. P (W/kg)
1	71	89	8	200
2	30	73	6.5	400
3	17.5	65	5.7	1200
4	10.5	51	4.5	1600
5	7	43	3.8	2000

Δt: Discharge time

CS: Specific capacitance

S. E: Specific energy

S. P: Specific power

Figure 4. (Color online) (a) CV, (b) PEIS, (c) GCPL, (d) specific capacitance as a function of current density, (e) Ragon plot, and (f) cyclic stability test of the CNF@NF device.

Figure 5. (Color online) (a) CV results of the flexible CNF@NF device of different bending angles at 50 mV/s of the scan rate with (b) the corresponding photograph of the CNF@NF device.

CNF is directly spin-coated on NF to create flexible supercapacitor electrodes, and the manufacturing method is relatively easy and simple without the need for any additional binders, surface preparation, or external catalysts. The combination of the intrinsic properties of NF and CNF leads to a specific capacitance of 89 F/g and maximum specific energy of 8 Wh/kg at a power of 400 W/kg at 1 A/g. The CNF on the NF electrodes showed low resistance and high electrochemical surface area. Even after 3000 charge–discharge cycles, the electrodes exhibited 84% stability from the initial capacitance. The binder-free CNF-coated NF is an interesting flexible supercapacitor electrode because of its simple and easy manufacturing method without the need for an external catalyst, expensive noble metals, and even complex thermal or chemical processes.