3.1. Characterization of electrospun carbon nanofibers
Electrospun carbon nanofibers could be fabricated from PAN precursor solutions through the electrospinning method as shown in Fig. 1. Electrospun PAN nanofibers must be thermally-treated in order to convert to carbon nanofiber structures. In the heat treatment, the stabilization process, which is generally referred to as a cyclization and dehydrogenation reaction, was conducted at about 300°C in an air environment to ensure thermal resistance to the electrospun PAN nanofibers [22]. After the stabilization process, carbonization was carried out at a high temperature of 1000°C to remove non-carbon components before converting them to carbon fibers having a carbon content of 95% or more [23, 24]. To further improve the specific surface area of the electrospun carbon nanofibers, a physical activation process with CO2 gas at high temperature was used. Carbon activation makes small, low-volume micro-pores that increase the surface area useful for many applications like chemical adsorption and energy storage [25, 26].
When the electrospun nanofiber was fabricated using a 10 wt.% PAN precursor solution with high viscosity, it was confirmed that the continuous shape of the fiber was produced without any aggregation of fibers due to the high viscosity shown in Fig. 2. During the heat treatment process, the diameters of nanofiber in the activated carbon nanofiber mats were reduced, because of elimination of the non-carbon elements of PAN (Fig. 3). In order to further increase the electrical conductivity of electrospun carbon nanofibers, MWCNTs were dispersed into the PAN polymer. In this study, the interaction between DMF and MWCNTs was increased by using MWCNTs surface-treated with carboxyl groups, which improved the dispersion of MWCNTs in the electrospinning solution [20]. The use of MWCNTs in the electrospinning process not only improved electrical conductivity, but also further increased the specific surface area of the fibers during the activation process [27]. In order to prevent aggregation of the MWCNTs, their addition was fixed at 1 wt.%, which prevented high viscosity or particle aggregation effects. The electrical conductivity of the solution in which MWCNTs were dispersed was increased; therefore, the diameter of PAN-MWCNT electrospun nanofibers decreased. In addition, the reduction of diameter after the carbonization process was also observed.
Figure 4 shows TEM images of the PAN-MWCNT electrospun nanofibers; MWCNTs were in the nanofibers and aligned in the longitudinal direction of the fibers. Dror et al. theoretically proved that electrospinning could help to align the CNTs inside the electrospun fibers along the longitudinal direction [28]. Considering the difference in diameter of MWCNT and PAN nanofibers ranged from tens to hundreds of times, it could be considered that several strands of MWCNTs were aligned together in the nanofibers.
XRD was performed to investigate the macromolecular and crystalline structure of carbon nanofibers produced by electrospinning. As shown in Fig. 5 (a), PAN nanofibers prepared by the electrospinning method showed peaks at 2θ = 17° and 28 ~ 29°, which correspond to the dominant crystal facets: (100) and (110) facets of crystal structures of PAN, respectively [29]. In the case of the heat-treated specimens, however, it can be seen that a very broad peak appears near 2θ = 25 ~ 26° which is attributed to the crystallographic planar structure of graphite [29, 30]. For the PAN-MWCNT electrospun nanofiber, it was shown that relatively higher height of the peak was observed because MWCNTs act like a nucleus, the crystal structure of graphite was formed around the nuclei [31]. According to Bragg’s and Scherrer’s equations, the average interplanar spacing (d(002)), and crystallite size parameter (Lc) could be determined based on the XRD graphs as:
\({d}_{\left(002\right)}=\frac{\lambda }{2sin\theta }\) (Bragg’s Equation) (4)
\({L}_{c}=\frac{0.9\lambda }{\beta cos\theta }\) (Scherrer’s Equation) (5)
where θ is the scattering angle, λ is the wavelength of the X-ray, and β is the width of the diffraction peak measured at half its height.
In an electrospun carbon nanofiber made using PAN-MWCNT nanofiber, the interplanar spacing was greatly reduced, and the size of the crystal structure increased as shown in Fig. 5 (b). This means that in specimens containing MWCNTs, the structure of the graphite layer becomes denser and the crystals develop and form a more ordered structure [29]. Therefore, as mentioned above, MWCNTs act as nuclei that promote the development of graphite layers.
Generally, the carbon fiber fabrics used in fiber-reinforced composites could not achieve a sufficient energy storage performance due to the very low specific surface area as shown in Table 1. However, the specific surface area of electrospun CNF and CNF-MWCNT were increased about 50 times as compared with carbon fiber fabrics, and that means electrospinning could be one of the most effective ways to solve the low surface area problem. Through the activation process, the micropores structures were developed by gasification of the pore walls and the specific surface area increased by about several thousand times than that of non-activated carbon nanofibers. The generation of micropores on the surface of the carbon nanofibers resulted in a decrease in the mean pore diameter. The amount of micropores on the surface of the activated carbon nanofibers (ACNF) in which MWCNTs (ACNF-MWCNT) were embedded increased; therefore, the specific surface area of ACNF-MWCNT was found to be twice as high as that of ACNF.
Table 1 Surface properties of as-received CF fabrics and electrospun CNF and ACNF with MWCNTs.
3.2. Electrical and mechanical properties of structural supercapacitors with electrospun carbon nanofibers
Using the electrospun carbon nanofiber mat, structural supercapacitors were fabricated by vacuum resin infusion method as shown in Fig. 6. The carbon fiber fabric was used as a backbone which could guarantee good mechanical properties, and electrospun carbon nanofiber was laid beneath the fabric as an electrode of the supercapacitor. Two layers of glass fiber fabric were used as a separator that could prevent the contact of the electrodes. In order to increase the ionic conductivity of the multifunctional matrix for the structural supercapacitors, 10 wt% of ionic liquid in which a salt was dissolved was added to the epoxy resin. PEGDGE-based epoxy, which is a relatively soft epoxy that can help to move the ions in the capacitors, was used as a base material of the multifunctional matrix [13, 32].
The electrochemical properties of the structural supercapacitors fabricated with electrospun carbon nanofiber were investigated using cyclic voltammetry and charge–discharge measurements. Transient current behavior was measured as a voltage of 1 V was applied for 60 s through the structural supercapacitors. Based on the equivalent circuit of the supercapacitors, the transient response could be fitted using Eq. (1), and the capacitance determined using the properties derived from equation Rs, Rp, and τ. Rs, also referred to as the equivalent series resistance (ESR), was caused by the electrical resistance of the electrodes and the ionic resistance of the electrolyte. Rp is generated by the electrical resistance of the two electrodes constituting the supercapacitor or electrical contact in an electrochemical reaction. Since Rp is connected to the capacitor in parallel, it causes leakage current in the charging process [15, 33].
Figure 7 shows the transient current response and capacitance of the structural supercapacitors with electrospun nanofibers. As shown in Table 2, a lower equivalent series resistance was exhibited in the structural supercapacitors fabricated with electrospun carbon nanofiber mat. A number of factors could affect ESR, such as the surface area, the electrical conductivity, and the interfacial resistance of the electrode; in this case, ESR decreased because of the increased active surface area with electrospun carbon nanofibers [15]. The lower ESR values contributed to an increase of about 4-fold of the specific capacitances of the structural supercapacitors where the electrospun carbon nanofiber structure was inserted. However, the MWCNTs could not affect the specific capacitance and energy storage ability of the supercapacitors with CNF-MWCNT electrodes. Due to the micropores created by the activation process, ESR further decreased in the supercapacitors made by activated electrospun carbon nanofiber electrodes because of the increased ionic accessibility of the electrode. Apart from the reduced series resistance, the parallel resistance for activated carbon nanofiber electrode also decreased. For ideal supercapacitors, there should be nearly zero values of Rs and infinite Rp in order to store electrical energy without any leakage. However, for the activated electrospun carbon nanofiber electrode, the value of Rp became lower than that of a non-activated electrode. This means that there was a short-circuit in the structural supercapacitors where the ACNF electrode was applied. This may be caused by the small-sized particles released from the carbon nanofiber electrodes during the vacuum resin infusion process, which could create a bypass for the current and lead to a short-circuit with the electrode on the opposite side due to migration or convection through the separator. In fact, it could be observed in transient current measurement for charge–discharge methods. The current graph was floated from the bottom while the supercapacitors were charged with a 1 V voltage (60 ~ 120 s). That was the current stored in the electrode during charging leaked through the bypass formed by isolated carbon particles from the electrode. As a result, although the ACNF and ACNF-MWCNT could improve the specific capacitance and energy density caused by low series resistances, low parallel resistances led to a decrease of specific power density of the structural supercapacitors, as shown in Fig. 8.
In order to investigate the effect of electrospun carbon nanofibers on mechanical properties, tensile testing was performed with an electrospun carbon nanofiber mat which was inserted between the interfaces of the carbon and glass fiber fabrics as it was done with the structural supercapacitors. The tensile testing was conducted with 10 samples under each condition, and the results are shown with calculated error bars. Figure 9 shows the result of tensile testing of multifunctional supercapacitors according to the ASTM D882 Standard. The tensile modulus and strength were not significantly affected even after electrospun carbon nanofiber structures were inserted. Because tensile properties are fiber-dominant, the insertion of electrospun nanofibers could not have an effect on the tensile strength and modulus because of its low mechanical properties. However, the insertion of carbon nanofiber electrodes containing MWCNTs slightly improved the tensile modulus and strength compared to specimens that did not contain MWCNTs.
To examine the effect of electrospun carbon nanofibers on the interfacial mechanical properties, which are matrix-dominant, a DCB test was conducted in accordance with ASTM D5528. Figure 10 show the load–displacement and delamination resistance curves for non-interleaved and interleaved DCB specimens. The critical load, Pc, and fracture toughness, GIc, of the specimens with electrospun carbon nanofiber mats inserted into them were higher than that of the non-inserted specimen. This is because carbon nanofibers inserted between carbon fiber fabrics cause interlocking effects such as “hooks and loops” in Velcro, and improve interfacial strength. In addition, the insertion of the electrospun carbon nanofiber has an effect of preventing stress concentration caused by the difference of mechanical properties between the carbon fiber and matrix [34]. The critical load and interlaminar fracture toughness were decreased for the specimens in which carbon nanofiber mat was inserted. The micropores of the carbon nanofiber that were developed during the activation process act as defects, and thus it is difficult to achieve a sufficient interlocking effect.
Table 2 Electrochemical properties of structural supercapacitors with different electrode materials
3.3. Multifunctionality
Structural supercapacitors should carry a mechanical load whilst simultaneously storing electrical energy. In order to determine whether both performance criteria are met simultaneously, an appropriate standard should be present. O’Brien et al. suggested an equation that provides a fundamental design requirement for mass-saving multifunctional designs [12]. Based on this equation, there is criteria for energy storage efficiency and structural efficiency of a conventional system. However, because determining the criteria is vague and controversial, the multifunctionality of the overall system can be calculated differently depending on the criteria. Therefore, Fig. 11 presents mechanical properties and energy density to verify only the tendency of the multifunctionality for each case. In the graphs, moving from the bottom left to the top right in the graph could be determined to be a multifunctional supercapacitor. According to Fig. 11, the structural supercapacitor fabricated using activated carbon nanofiber with MWCNTs was the most multifunctional structure when tensile modulus was used as a criterion of mechanical properties. These trends were observed in Table 3. A CNF-MWCNT-based structural supercapacitor had an even higher tensile modulus than an ACNF-MWCNT electrode, but it had an insufficient energy density to store electrical energy. In the case of interlaminar fracture toughness, ACNF-MWCNT electrode-based supercapacitors could also be determined to be a multifunctional composite. Therefore, there is the possibility of application of carbon nanofibers as an electrode in multifunctional composites fabricated using a simple and convenient method. This work is ‘proof-of-concept’ of structural supercapacitors, and shows the most effective way to improve the multifunctionality of the composites.