3.1 Synthesis of ACFF electrodes from textile PAN-based fiber
The purpose of thermal-oxidative stabilization, also known as oxidation, is to provide a structure that can withstand high temperatures and to prevent degradation and melting of the fiber in the subsequent carbonization process. Thermoplastic textile PAN-based fibers are converted into thermosetting fibers due to the cyclization of nitrile groups and the cross-linking of PAN chains [40]. Once stabilized, the PAN-based sample containing carbon and noncarbon elements, such as nitrogen, hydrogen, and oxygen, undergoes a carbonization process. As a result, noncarbon elements are removed and released as volatile gases resulting in weight loss and shrinkage of the sample [41].
Activation is the key process in the synthesis of carbon-based electrodes due to the development of accessible porosity that leads to an improved surface area. During the activation process, existing clogged pores that were developed during carbonization are opened up and enlarged. In addition, new porosity is developed resulting in a well-defined porous structure and an enhanced surface area [42]. Physical activation is an uncomplicated and effective process, although it requires high temperatures (700–1200°C) and some of the carbon atoms are removed during the process resulting in low production yield [43]. The most common examples of oxidizing agents are O2, CO2, and steam. Previous studies reported that CO2 leads to a higher development of narrow microporosity, and hence it was chosen as the major oxidizing agent [44]. As a combination of two oxidizing agents contributes to the development of enhanced porosity, water (steam) was chosen as the minor oxidizing agent. While CO2 leads to the development of new microporosity, water helps to widen the existing micropores. Since water requires lower temperatures in comparison to CO2, water was used in the first stage of the activation process. If the activation time is overextended, the development of mesopores (< 2 - > 50 nm) occurs by burning off the wall bordering the micropores (> 2 nm) [45]. Subsequent to the carbonization-activation process, the ACFF sample had a burn-off of 21%, a total mass loss of 79%, and a shrank of 55% in size.
3.2 Characterization of the ACFF electrodes
According to the IUPAC classification of adsorption Isotherms for gas-solid equilibria [46], the N2 adsorption-desorption isotherm of ACFF electrodes, shown in Fig. 3a, is classified as Type I and indicates a microporous character. Moreover, hysteresis is absent, which suggests that the size of the pores measured on adsorption and desorption branches coincides well. [47]. BET theory was applied to calculate the specific surface area (SBET) of ACFF electrodes from the acquired data of N2 adsorption at 77 K. In accordance with the results, ACFF electrodes have a high SBET value of 1875 m2 g− 1 and a correlation coefficient (r) of 0.999918. The microporosity developed particularly during the activation process contributed to the high value of the specific surface area.
The mobility of the ions into the pores depends on the pore size. Thus, inaccessible pores do not contribute to both double-layer capacitance and energy density [48]. Previous studies suggested that pore size smaller than 0.5 nm is inaccessible to hydrated ions [49]. The technique applied for pore size distribution (Fig. 3b) does not show any information about pores smaller than 1 nm due to the penetration limit of N2. Regardless of the limitation, an ascending line in the region of pore width smaller than 1 nm possibly indicates the presence of said micropores [50]. The pore size distribution curve determined by the DFT method shows a maximum pore width of 3 nm. Furthermore, the size of most pores ranges from 1 to 2 nm confirming the presence of microporosity. As reported in the literature, a small number of pores, wider than 2 nm, increase the power density of SCs [51]. These small mesopores facilitate the access of electrolyte ions into the electrode surface. The average pore size of PAN-based ACF reported in the literature ranges from 4 to 10 nm [52]. The average pore size of the ACFF sample is smaller due to the parameters of the activation process. The development of porosity is influenced by several factors such as the residence time, temperature, original porosity and structure of the carbonized material, pressure and flow rate of the gas, and the type of oxidizing agent [53].
The Raman spectrum for carbon-based materials, amorphous or crystalline, exhibits two well-defined peaks commonly assigned as D and G bands. The D band, also known as the disorder/defect band, exhibits a peak at nearly 1380 cm − 1. The G band, crystalline graphite, exhibits a peak at nearly 1580 cm − 1 due to the C-C bond-stretching of all pairs of sp2 atoms in both rings and chains [54].
The Raman spectra of ACFF electrodes are shown in Fig. 4. In all samples, the spectra exhibited the same appearance. There are two very broad peaks located in the range of 1244–1710 cm − 1. The first peak, located at 1342 cm–1, is referred to “D”. The second peak, located at 1598 cm–1 is referred to “G”. The spectra of ACFF electrodes are very similar to that of carbon-based materials.
3.3 Electrochemical performance
Figure 5 shows a comparison of the CV curves measured for aqueous-based and glycerol-based electrolytes at a scan rate of 1m V s− 1. As capacitance is constant over an acknowledged potential window, symmetric and nearly rectangular curves in the cyclic voltammograms are expected for carbon-based EDLC devices [55]. Therefore, in comparison to glycerol-based electrolytes, aqueous-based electrolytes exhibited less distorted rectangular-shaped curves that indicate mostly electrostatic capacitance (a non-faradaic mechanism), good reversibility and low resistance. Furthermore, glycerol-based electrolytes, particularly molar ratios of 1:1 and 2:1, exhibited blunt CV profiles indicating the possibility of higher values for equivalent series resistances (ESR) [56]. The tail observed in the CV curves of KOH:GLY possibly indicates the presence of water content due to the fact that both components of the glycerol-based electrolyte are hygroscopic. Moreover, the tail observed in the CV curves of aqueous-based electrolytes suggests electrochemical decomposition of the electrolyte at high potentials due to water electrolysis [57]. The narrow potential window of aqueous-based electrolyte limits energy density and restricts the amount of energy stored [58].
Among the ACFF-based EDLCs, aqueous KOH electrolyte at a molar concentration of 2 mol L− 1 exhibited the highest gravimetric capacitance value of 129 ± 6 F g− 1 at a scan rate of 1 mV s − 1. Amidst glycerol-based electrolytes, KOH:GLY (2:1) exhibited the highest gravimetric capacitance value of 99 ± 5 F g− 1 possibly due to the long tail, which could affect overall electrochemical performance. KOH:GLY (3:1) exhibits less distorted rectangular-shaped curves and a much shorter tail, which suggest an improved performance and lower resistance.
The absence of humps or redox peaks in the CV curves of ACFF-based EDLCs suggests that there is no occurrence of oxidation-reduction reactions, which is expected from an electrostatic and non-faradaic mechanism [59]. Nevertheless, previous literature reported that carbon-based electrodes possibly exhibit 1–5% of their capacitance as surface-redox pseudocapacitance due to reversible faradaic-redox reactions of active surface functional groups [60]. As explained previously, oxidation time affects the surface chemistry of AC-based materials leading to enhanced capacitance. The occurrence of a minor surface-redox pseudocapacitive behavior could be verified with a three-electrode cell at a low scan rate instead of a two-electrode cell.
Figure 6 shows the Nyquist plot for aqueous-based and glycerol-based electrolytes at an applied potential of 0 V; where -Zim is the imaginary part (capacitive) and Zre is the real part (resistive) [61]. A Nyquist plot exhibits three distinct frequency regions: (1) a high-frequency region that is higher than 10 kHz (low values for Zre), (2) a medium-frequency region, ranging from 10 kHz to 1 Hz, and (3) a low-frequency region that is lower than 1 Hz (high values for Zre) [62].
At high frequencies, the behavior of a supercapacitor is comparable to a resistance and it is related to the bulk solution resistance of the electrolyte [63]. The value of electrolyte resistance (Rs) can be deduced at the point where the plot intercepts the x-axis of the Nyquist plot, also known as the real axis (Zre) [64]. The results are normalized by Ω cm− 2 for comparison.
Aqueous-based electrolytes have low electrolyte resistance in comparison to glycerol-based electrolytes. The lowest value of Rs (0.44 ± 0.04 Ω cm2) can be observed in aqueous KOH electrolyte, at a molar concentration of 2 mol L− 1 (Fig. 6a). Among glycerol-based electrolytes (Fig. 6b), KOH: GLY (3:1) exhibited the lowest value of Rs (17 ± 2 Ω cm2) probably due to the salt concentration as conductivity can be affected by the salt-to-solvent ratio [65].
The interfacial impedance between the bulk solution (electrolyte) and the electrode can be observed at medium frequencies. A conventional Nyquist plot for an EDLC cell exhibits a single semicircle that represents the ion charge-transfer resistance (Rct), and its diameter is related to the ion mobility in the pores [66]. A single semicircle can be seen in both concentrations of aqueous-based electrolytes. The values of Rct for aqueous KOH electrolyte, at a molar concentration of 1 and 2 mol L− 1 are 0.59 ± 0.06 Ω (0.30 ± 0.03 Ω cm2) and 0.36 ± 0.04 Ω (0.18 ± 0.18 Ω cm2), respectively. The absence of a single semicircle in glycerol-based electrolytes may occur due to the minimal impedance of the electrolyte [67].
Subsequent to the semicircle, a 45-degree line, or slope, at the low frequency region indicates ion diffusion within the pores [68]. While 2 M KOH exhibits a steep slope after the semicircle, 1 M KOH and KOH:GLY (3:1) exhibit gradual slopes. A steep slope suggests that ion penetration into pores occurs efficiently. A gradual slope, as in the case of 1 M KOH and KOH:GLY (3:1), suggests that ion penetration into pores occurs laboriously [69].
A vertical line at the low-frequency region parallel to the imaginary part (y-axis) is expected for an ideal capacitor [70]. The 90-degree line, or tail, suggests that capacitance is constant over the applied frequency range and the electrode surface is entirely impregnated [71], A vertical line can be observed in the Nyquist plot for both concentrations of aqueous-based electrolytes (Fig. 6a), and nearly vertical line can be observed in the glycerol-based electrolyte KOH:GLY (3:1) (Fig. 6b). The Nyquist plots of KOH:GLY with molar ratios of 1:1 and 2:1 deviate from the profile expected from an ideal EDLC. The line (tail) of 2 M KOH leans more towards the -Zim axis (y-axis), which is the imaginary part (capacitive) of the Nyquist plot, indicating that the ACFF-based EDLC in aqueous 2 M KOH has a better capacitive behavior.
While Rs and Rct rely upon the electrolyte solution, the tail depends on both electrode and electrolyte. Due to the low viscosity of its solvent, aqueous-based electrolytes (Fig. 2a) have high ionic conductivity (101 – 102 mS cm− 1) [72] which improves ion mobility and results in smaller values for Rs and Rct [73]. On the contrary, glycerol-based electrolytes (Fig. 2b) have high viscosity and do not provide good contact in the electrode-electrolyte interface. As ion mobility is inversely proportional to the size of the ion, large molecules, such as glycerol molecules, have high resistivity. Thus, large pores in the surface of the electrode are required since inaccessible pores do not contribute to the double layer capacitance [74].
The results are in agreement with the cyclic voltammogram data discussed in the preceding section. Aqueous-based electrolytes exhibited higher values for gravimetric capacitance and nearly CV rectangular-shaped curves indicating lower resistance and better electrochemical performance.
Near-isosceles triangular-shaped curves are expected for carbon-based EDLC devices and suggest primarily an electrostatic (non-faradaic) mechanism [75]. Among the ACFF-based EDLCs, aqueous KOH electrolyte at a molar concentration of 2 mol L− 1 (Fig. 7a) exhibited near-isosceles triangular-shaped curves. Aqueous KOH electrolyte at a molar concentration of 1 mol L− 1 (Fig. 7b) also exhibits triangular-shaped curves, although the nonlinear charging curve reaching a plateau at high potentials may indicate decomposition of the electrolyte possibly due to overcharge [76]. Among the glycerol-based electrolytes, only KOH:GLY with a molar ratio of 3:1 exhibited triangular-shaped curves (Fig. 7d). The discharge curves of both aqueous-based electrolytes and KOH:GLY (3:1) drop linearly indicating that there is no considerable contribution of pseudocapacitance because the charge storage mechanism is manly electrostatically [77]. The GCD curves of KOH:GLY with molar ratios of 1:1 and 2:1 do not drop linearly and deviate from the profile expected from an ideal EDLC (Fig. 7e – 7f).
The equivalent series resistance of the cell (ESR), also known as internal resistance, consists of the bulk solution resistance of the electrolyte (Rs), the electrode-electrolyte interfacial resistance (Rct), the resistance of intra-particle pores, and the contact resistance between the current collector and the electrode [78]. For a better electrochemical performance, a low ESR value is preferred.
Voltage drop (IR-drop), which is the resistance of the cell, was used to deduce the value of ESR. At a current density of 0.15 A g− 1, the occurrence of IR-drop is noticeable in all glycerol-based electrolytes (Fig. 7d-7f). Among all glycerol-based electrolytes, KOH: GLY (3:1) exhibited the lowest value for ESR (Table 1) apparently due to the salt concentration as electrolyte conductivity can be affected by the salt-to-solvent ratio.
Table 1
Value of ESR in glycerol-based electrolytes at a current density of 0.15 A g-1
|
|
ESR (Ω)
|
ESR (Ω cm2)
|
Glycerol-based
|
KOH:GLY (3:1)
|
76 ± 4
|
38 ± 2
|
KOH:GLY (2:1)
|
193 ± 10
|
97 ± 5
|
KOH:GLY (1:1)
|
390 ± 20
|
195 ± 10
|
The occurrence of IR-drop in aqueous-based electrolytes is not noticeable in small current densities, such as 0.15 A g− 1. Due to the nature of aqueous-based electrolytes (high conductivity and low viscosity) larger current densities were required. The occurrence of IR-drop in aqueous 2 M KOH is only noticeable at current densities ≥ 8 A g− 1 (Fig. 8A). Even in larger current densities, aqueous 2 M KOH exhibits triangular-shaped curves, a discharge curve that drops linearly, and low internal resistance (0.83 ± 0.04 Ω cm2) indicating satisfactory electrochemical performance. Although aqueous 1 M KOH performed better than glycerol-based electrolytes, the occurrence of IR-drop is noticeable at current densities > 1 A g− 1 (Fig. 8b); an inferior performance in comparison to 2 M KOH.
The results are in agreement with the Nyquist plot discussed in the preceding section. Aqueous-based electrolytes have high ionic conductivity and wettability which enhance electrode-electrolyte interaction and consequently decrease ESR values.
Table 2
Gravimetric energy density (Es) and specific power (Ps) of ACFF-based EDLCs
Type of electrolyte
|
Es
(Wh kg− 1)
|
Ps
(kW kg− 1)
|
Aqueous-based
|
1 M KOH
|
14 ± 1
|
3.5 ± 0.3
|
2 M KOH
|
18 ± 1
|
17.9 ± 1.8
|
Glycerol-based
|
KOH:GLY (1:1)
|
48 ± 2
|
0.1 ± 0.1
|
KOH:GLY (2:1)
|
61 ± 3
|
0.2 ± 0.1
|
KOH:GLY (3:1)
|
50 ± 3
|
0.4 ± 0.1
|
Previous literature reported that the gravimetric energy density, also known as specific energy, for EDLCs commonly ranges from 5–10 Wh Kg− 1 [79]. As can be seen in Table 2, the energy density of all ACFF-based EDLCs exhibited superior performance. According to Eq. 3, gravimetric energy density is proportional to gravimetric capacitance (Cs) and the square of the voltage (V). As a result, increasing either or both of them enhances the energy density of EDLCs and expands the range of SCs applications. Although aqueous-based electrolytes exhibited higher values for gravimetric capacitance (Fig. 5), glycerol-based electrolytes exhibited higher values for energy density due to the electrochemical stability window of the hybrid electrolyte, which is twice greater.
ESR limits power density and the energy cannot be delivered completely due to the distance between the electrolyte and the electrode [80]. As can be seen in Table 2, aqueous-based electrolytes exhibited higher values of specific power (Ps) in comparison to glycerol-based electrolytes due to low internal resistance. In order to improve the specific power of glycerol-based electrolytes, decreasing the values of ESR and improving the ionic conductivity is suggested.