1 Introduction

Supercapacitors are one of the energy storage devices which have acquired considerable attention due to their fast charge, high power density, long life cycles, and temperature-independent performance [1, 2]. Recently, these advantages of supercapacitors allow them to be used as ancillary devices of electric and hybrid vehicles for acceleration and regenerative braking [3]. Supercapacitors include electrochemical double layer capacitors (EDLCs), pseudocapacitors, and hybrid capacitors [4]. Among them, EDCLs store the charge by electrostatic interactions on two symmetric electrodes, which are typically porous carbon materials such as activated carbon. Lack of the Faradaic redox reaction of carbon electrodes results in low capacitance and energy density of EDLCs, which limits their wide applications. Since energy storage performance of supercapacitors has a proportional relationship to contact area between electrode and electrolyte, enlarging surface porosity of electrode materials is considered as a good strategy to overcome this shortage.

Cucurbituril (CB) is a nonadecacyclic compound consisting of tetrazocane units fusing glycoluril moieties, which is synthesized from urea, glyoxal, and formaldehyde, and its shape resembles the shape of a pumpkin containing a hollow core [5, 6]. It was found that as the number of glycoluril units increases, the internal pore space of the ring becomes wider and the molecule becomes flexible [7]. Due to its exceptional binding ability and stable pore structures, CB has been investigated in a wide range of applications such as sensors, water treatment, and drug delivery and so on. It was reported that its porous structures are transferred after carbonization, making it an excellent precursor for supercapacitor electrodes [8].

Carbon nanofibers (CNFs) are also good candidate for EDLC electrodes due to their high specific surface properties and good conductivity [9]. There are two synthesis methods: thermal treatment of electrospun nanofibers of precursor polymers, and chemical vapor deposition (CVD) using carbon precursors and metal catalyst [10, 11]. For the former method, the electrochemical performance of CNFs can be optimized by adjusting the chemical structures of precursor polymers, and the conditions for electrospinning and thermal treatments [12, 13]. Electrospinning is the most common way to fabricate nanofibers since the most of polymers and solvents can be applicable. It is also well-known that composite nanofibers can be readily produced by mixing additives in the polymer solutions, and then electrospinning them [14].

Most of the studies on CNFs for supercapacitor electrodes focus on the increase in surface area and porosity using sacrificial pore generating agents such as thermally degradable substance [15]. During the conversion of precursor polymers to carbon under high temperature, they are thermally decomposed creating pores in the CNF surface. It was also reported that precursor polymer blends can convert into porous carbon [16,17,18]. Investigating novel precursor composite or blends is critical to obtain highly porous CNFs with exceptional energy storage performance. Despite the outstanding pore structures of CB, limited research on CB in the field of energy storage has been conducted.

The present study is aimed at producing porous CNF electrode by thermal treatments of PAN/CB composite nanofibers to obtain high energy storage performance. CNFs can make them as free-standing electrodes without the use of binders while the addition of CB can improve the surface porosity of CNFs due to its stable pore structures. Surface morphology and carbon microstructures were investigated by scanning electron microscopy and Raman spectroscopy, respectively. Electrochemical properties were tested by assembling a three-electrode cell with pristine CNFs as a working electrode, and cyclic voltammetry (CV), galvanostatic charge–discharge, and cycling performance were investigated.

2 Experimental

Scheme 1 shows the preparation process of CNFs using PAN and CB. CB [6] consisting of 6 glycoluril subunits was chosen as a pore generator, and synthesized according to literature procedures [19]. Glycoluril was suspended in sulfuric acid, heated at 70 °C, and then formaldehyde dissolved in water was added. While the mixture was stirred at 70–75 °C for 24 h and at 95–100 °C for 12 h, crystalline CB [6] was precipitated. After pouring the reaction mixtures in water, acetone was added to produce precipitate. The mixture was filtered and washed with water and dried under vacuum. The chemical compositions of synthesized CB [6] were investigated using an attenuated total reflection (ATR)-Fourier transform infrared (FTIR) spectroscopy (VERTEX 80 V, Bruker, Ettlingen, Germany) with a germanium ATR crystal.

Scheme 1.
scheme 1

Preparation of CNFs derived from PAN/CB

Full size image

Seventeen wt% PAN/CB (2:1) dissolved in dimethyl sulfoxide (DMSO) was electrospun under an applied voltage range from 10 to 15 kV, and PAN only in DMSO was also prepared for comparison. Electrospun nanofibers were stabilized at 280 °C under air for 2 h and carbonized at 800 °C under nitrogen for 30 min to fabricate CNFs. From preliminary study, it concluded that electrochemical properties can be improved as the content of CB increases. CNFs derived from PAN/CB solutions with a ratio of 10:0, 9:1 and 2:1 were prepared and showed the energy densities of 13.4, 19.4 and 25.4 Wh kg−1, respectively. It was also observed that the higher content of CB induces poor solubility and electrospinnability. Based on the electrochemical properties and spinnability, PAN/CB (2:1) was chosen. Carbonized PAN and PAN/CB were represented as cPAN and cPAN/CB, respectively.

Thermal stability was measured by thermogravimetric analysis (TGA) using a DTA-60, DSC-60, Shimadzu, Japan. The surface morphology of electrospun and carbonized PAN and PAN/CB nanofibers was observed using scanning electron microscopy (SEM) via an S-4800, Hitachi, Japan after sputter coating with osmium. Carbon microstructures of cPAN and cPAN/CB were studied by recording Raman spectroscopy using an XPLORA, Horiba, Japan. Surface area and porosity of CNFs were measured using nitrogen adsorption/desorption on an Autosorb iQ and a Quadrasorb SI (Quantachrome, USA). Specific surface area and pore size distribution were calculated using the Brunauer–Emmett–Teller (BET) equation and density functional theory (DFT) methods, respectively.

Electrochemical measurements were performed on an electrochemical workstation (WBCS3000S, Wonatech, KOREA) using a three-electrode cell consisting of Pt as the counter electrode, Ag/AgCl as the reference electrode, and cPAN and cPAN/CB nanofibers were directly used as the working electrode without further treatment. 1 M Na2SO4 was utilized as the electrolyte.

Cyclic voltammograms were recorded between 0 and 1.0 V (vs. Ag/AgCl) at various scan rates ranging from 10 to 100 mV s−1. Galvanostatic charge/discharge (GCD) testing was conducted between 0 and 1.0 V (vs. Ag/AgCl) at different current densities of 0.5–6 A g−1. The specific capacitance (Csp, F g−1) was measured using \(C_{\text{sp}}=\frac{I{\text{d}}t{\text{d}}}{mV}\), where Id is the discharge current (A), td the discharge time (s), m the mass of a working electrode (g), and V the voltage window (V). Energy density (E, Wh kg−1) and power density (P, kW kg−1) were calculated using \(E=\frac{I{\text{d}}t{\text{d}}V}{2m}\) and \(P=\frac{E}{t{\text{d}}}\), respectively. Capacitance retention by GCD testing after 5000 cycles was measured to confirm the cycling stability. Electrochemical impedance spectra (EIS) were evaluated over the frequency range of 100 kHz and 0.1 Hz using a 0.01 V amplitude on a PEC-L01, Peccek, Japan.

3 Results and discussion

CB was synthesized and its chemical structures were studied by FTIR as shown in Fig. 1. It shows distinctive bands at 1717 cm−1 for the stretching vibration of C=O, at 1464 cm−1 for the banding vibration of C–H, and 3005 and 2930 cm−1 for the stretching vibration of C–H of methylene. Significant peak at 3440 cm−1 is assigned to O–H of the water.

Fig. 1
figure 1

FTIR spectrum of synthesized CB

Full size image

PAN and CB were dissolved in DMSO and electrospun, followed by thermal treatments including stabilization under air and carbonization under inert gas. Surface morphology of electrospun and carbonized nanofibers was studied by observing SEM images as shown in Fig. 2. It is seen that both PAN and PAN/CB show continuous and uniform nanofibrous structure with average diameters of 660 nm and 730 nm, respectively. After stabilization at 280 °C and carbonization at 800 °C, they still have nanofibrous structure with reduced diameters of 470 nm for cPAN and 570 nm for cPAN/CB.

Fig. 2
figure 2

Surface morphology of a as-spun PAN, b as-spun PAN/CB, c cPAN, and d cPAN/CB nanofibers

Full size image

The thermal stability of synthesized CB, and electrospun PAN and PAN/CB-6 nanofibers was observed using TGA measurements with a heating rate of 10 °C min−1 from room temperature to 800 °C, as shown in Fig. 3. It seems the pyrolysis of PAN starts around 300 °C, and the oxidative stabilization temperature was generally based on it [20]. During stabilization, linear PAN chains are transformed into oriented cyclized ladder structures caused by thermal cross-linking, and thus the stabilization step determines the carbon yield and the graphitic structures resultant CNFs [21]. For CB, the weight loss around 250 °C may occur due to the loss of hydrated and coordinated water molecules [7, 22]. PAN/CB shows significant weight loss around 250 °C resulting from the decomposition of CB, which is also observed in a TGA curve of CB only. It is also seen that 50% of PAN/CB-6 remains at 800 °C, and the intermolecular interaction between nitrile groups in PAN and carbamide groups in CB may result in high thermal stability of PAN/CB.

Fig. 3
figure 3

Thermogravimetric analysis of a PAN, b CB and c PAN/CB

Full size image

The Microstructure of CNFs derived from PAN and PAN/CB was studied by measuring Raman spectroscopy, as shown in Fig. 4. Both cPAN and cPAN/CB indicated two significant peaks near 1585 and 1350 cm−1, correlating with E2g2 graphitic crystallites (G-band) and disordered structure of carbon (D-band), respectively. This indicates the successful conversion of nanofibers into carbon via thermal treatments. The ratio of peak intensities (ID/IG), calculated using area under each curve, are 3.36 for cPAN and 3.56 for cPAN/CB, indicating predominantly disordered carbon derived from polymer precursors. Carbon materials derived from polymer precursors generally showed the high content of amorphous carbon. It was reported that the ratio of peak intensities is 2.9 for CNFs from PAN and 3.3 for CNFs from polybenzimidazole [20, 23, 24]. It was also known that carbonized CB is mainly composed of amorphous carbon phase [8].

Fig. 4
figure 4

Raman spectra of cPAN, and cPAN/CB

Full size image

Specific surface area and porosity were measured by nitrogen adsorption/desorption, as shown in Fig. 5, and specific surface area and pore volumes were calculated, as shown in Table 1. It is clearly seen that cPAN/CB exhibits higher surface area and pore volume than cPAN. It is also noticeable that cPAN/CB has significant increased meso-pore (with diameters in the range of 2–50 nm) volume of 0.140 cm3g−1, which is 26.9% of total pores. This means that the decomposition and carbonization of CB induce the creation of meso-pores on the CNF surface, and large pores of carbon electrodes can facilitate the access of electrolyte ions and improve the interfacial contact of ions between electrode/electrolyte [25].

Fig. 5
figure 5

Nitrogen adsorption/desorption isotherms of CNFs

Full size image
Table 1 Structural parameters of cPAN and cPAN/CB
Full size table

Electrochemical properties of CNFs were measured in a three-electrode system with 1 M Na2SO4, and CNFs were used as a working electrode without adding any binders. Figure 6 exhibits cyclic voltammograms (CVs) of CNF electrodes, and both cPAN and cPAN/CB show typical CVs for carbon electrode in EDLCs. It is obviously seen that cPAN/CB has higher current densities than cPAN over the whole voltage range (Fig. 7).

Fig. 6 
figure 6

Cyclic voltammograms of a cPAN, and b cPAN/CB electrodes

Full size image
Fig. 7
figure 7

Galvanostatic discharge curves of a cPAN, and b cPAN/CB electrodes

Full size image

Galvanostatic charge–discharge testing was performed in the current density range from 0.5 to 6 A/g, the energy storage performance such as specific capacitance, and energy and power densities were calculated from the discharge curves from 1 to 0 V. Specific capacitances of cPAN and cPAN/CB are 96.5 and 183.5 F g−1 at 0.5 A/g, respectively. It is also seen that energy densities of cPAN and cPAN/CB are 13.4 and 25.4 Wh kg−1 at 0.5 A g−1, respectively, while power densities show the identical value of 1.8 kW kg−1 at 6 A g−1. Higher surface area and porosity caused by the existence of CB may result in significant higher energy storage performance.

For CNF electrodes synthesized via electrospinning, most of studies focus on enlarging surface porosity by adding pore generating materials. Polymers with poor thermal stability can be good candidates since they are decomposed during thermal treatment creating pores on the CNF surface. Poly-l-lactic acid (PLLA), poly (ethylene glycol) (PEG), and poly (methyl methacrylate) (PMMA) have been reported as pore generating materials, and CNFs from PAN showed high energy storage performance due to the enlarged surface area and porosity [26,27,28]. CB was chosen as pore generating materials since its stable cavity structures can generate pores on the CNF surface via thermal treatment. It is known that the cavity size of CB can be variable by controlling the number of glycoluril subunits [29, 30]. It was found that the width of cavities is ranged from 4.4 to 8.8 Å from CB [5] and CB [8], which can produce different size of pores when they are carbonized. It is known that an electrolyte is one of the critical parts to determine the energy storage performance by controlling the voltage window. Ionic liquid electrolytes show wide voltage window compared to aqueous ones (up to 1 V) [31, 32]. However, the bulkiness of their ions limits the access to the electrode surfaces, and it becomes more important enlarging surface area and porosity of electrode when ionic liquids are used as an electrolyte. It seems worthy to study the size of pores generated by various CBs and to match with size of ionic liquid ions, which can result in the optimization of energy storage performance.

It has been reported that high crystallinity of PAN is unfavorable to the cyclization during the stabilization step, in which the linear polymer chains are converted to an aromatic ladder structure [33]. Vinylimidazole and itaconic acid have been reported as effective comonomer candidates to disrupt the strong dipole–dipole interactions between nitrile groups in PAN and to reduce the crystallinity [34,35,36,37]. Small organic molecules such as CB can also play a role of a plasticizer to enhance the movement of PAN chains. It can be concluded that the introduction of CB is effective not only for increasing surface area and porosity, but for reducing crystallinity of PAN, resulting in superior energy storage performance the resultant CNFs.

EDLCs store their charge physically at the interface between electrodes and electrolyte without irreversible redox reactions, and thus exhibit long cycle life. Cycling stability of cPAN and cPAN/CB was evaluated by measuring specific capacitance at 1 A/g for every 250 cycles, as shown in Fig. 8. cPAN and cPAN/CB exhibit high capacitance retentions of 92.0% and 91.6% after 5000 cycles, respectively. A high degree of reversibility demonstrates the stable cyclic performance of the CNF electrodes.

Fig. 8
figure 8

Cyclic stability of cPAN, and cPAN/CB electrodes

Full size image

Figure 9 shows EIS of cPAN and cPAN/CB electrodes. The size and shape of the semicircle at high frequency depend on the adsorption kinetics of ions at electrode pores, and on the resistances of materials and mass transfer while the slope of the plots in the low frequency region is related to the rate of the electrochemical double layer formation inside of the electrode pores [38]. cPAN/CB exhibits significantly lower impedance than cPAN, indicating the incorporation of CB effectively improves electric conductivity by enlarging the contact area and facilitating the charge transfer. Wu et al. reported that presence of pyrrolic-nitrogen and quaternary-nitrogen in carbonized CB, which are conductive to the electron conduction [8]. Thus, it is also expected that a higher content of nitrogen enables cPAN/CB to contain more conductive carbon matrix.

Fig. 9
figure 9

EIS of cPAN, and cPAN/CB electrodes

Full size image

4 Conclusion

CB was introduced as a pore generating material to produce porous CNF electrodes in supercapacitors. PAN/CB (2:1) solution in DMSO was prepared, electrospun, and thermally treated. Thermal stability of PAN/CB was confirmed by TGA, and the conversion of PAN/CB into CNFs was done by Raman spectroscopy. cPAN/CB showed significantly high specific surface area and meso-pore volumes compared to cPAN. Due to its superior surface properties, cPAN/CB exhibited exceptional energy storage performance such as a specific capacitance of 183.5 F/g and an energy density of 25.4 Wh/kg at 0.5 A/g with stable cyclic stability during 5000 cycles. It was also found that cPAN/CB has significantly lower impedance than cPAN, confirming PAN/CB composite is excellent precursor candidate as porous carbon for supercapacitor electrodes.