A New High-Current Electrochemical Capacitor Using MnO₂-Coated Vapor-Grown Carbon Fibers

Abstract

Composites of MnO₂ nanosheet arrays vertically grown on the surface of vapor-grown carbon fibers (VGCFs) are fabricated by a low-temperature redox reaction between KMnO₄ and the VGCFs. An assembled AC/0.5 M K₂SO₄/MnO₂@VGCF electrochemical capacitor exhibits a higher specific capacitance, as well as a better rate capability, at a fast-current density compared to the capacitor built on hydrothermally prepared, standalone MnO₂. Electrochemical tests revealed that VGCFs act as a conducting matrix, which effectively improves the conductivity of MnO₂ nanosheets during cycling.

1. Introduction

Electrochemical devices that can store energy have created an interesting world in recent decades. These devices, including fuel cells, batteries and supercapacitors (SCs), possess many advantages, such as high-energy-utilization efficiency, a low output of carbon dioxide and the promotion of renewable energy use [1]. Among them, SCs can compensate the shortage of fuel cells and batteries in power [2]. Therefore, supercapacitor technologies are deemed as a better way to store electricity [3].

So far, a lot of transition metal oxides (TMOs) (e.g., RuO₂ [4], MnO₂ [5], MoO₃ [6], V₂O₅ [7] and Co₃O₄ [8]) have been studied in SCs. By comparing them, MnO₂ possesses the merits of resource abundance, low cost and toxicity, which has been researched widely. Moreover, it has a relatively broad electrochemical window (e.g., 0–1 V vs. SCE) in water-based electrolytes [9]. Generally, composites combing conductive frameworks with a large specific surface area and nanostructured TMOs are considered to be potential electrode materials in SCs. The most used conductive frameworks are templated mesoporous carbon [10], carbon nanotubes (CNTs) [11], carbon nanofoams [12], carbon fiber paper [13] and graphene [14]. Although the MnO₂/carbon composites have been investigated extensively, this work is still distinctly different from previously reported works [15,16,17,18]. For example, the morphology of the MnO₂/VGCF composites presented in this work is completely different when compared to those in the review article [15], which results in a very high sweep rate of 500 mV s⁻¹. As a result, vapor-grown carbon fibers (VGCFs) are deemed a promising candidate because of their abundance, low cost and good electrical conductivity [19]. Additionally, they are a type of multiwalled carbon nanotubes, which possess much more excellent properties (e.g., thermal stability, chemical stability, mechanical strength [20] and a high aspect ratio of approximately 250–2000 (100 μm in length and 50–200 nm in diameter [21])). In terms of synthesis, various methods are used to prepare MnO₂/carbon composites, e.g., redox reaction [22], electro-deposition [23], ball milling [24], thermal decomposition [25], hydrothermal methods [13], chemical co-precipitation [26], sonochemical synthesis [27], microwave irradiation [28] and physical mixing [29]. To the best of our knowledge, the permanganate can be reduced by surface carbon from conductive frameworks to generate MnO₂ nanosheets, which simultaneously deposit on the surface of conductive frameworks. This is the most appealing strategy because of its inherently self-supporting feature [30] and the existence of synergistic effects in physical and chemical phenomena [31]. Li and co-workers also adopted the same strategy by reducing KMnO₄ with CNTs [16]. However, CNTs can be partly cleaved in the process of synthesis because of its smaller diameter, thereby destroying the integrity of the material’s structure. On the contrary, the VGCFs used in this work can maintain their structure integrity, thereby leading to a very large sweep rate. Wang and co-workers synthesized MnO₂/CNTs composites using Mn²⁺ ions to reduce KMnO₄ [17]. It is obvious that the interaction between MnO₂ and CNTs only involves physical adsorption instead of synergistic effects. Furthermore, Wu and co-workers used a similar method to prepare MnO₂/CNTs composites with the assistance of surfactants [18], which made the fabrication process more complicated.

In this work, a co-axial electrode material with a flower-like structure is introduced for the electrochemical capacitor by assembling MnO₂ nanosheet arrays onto the VGCFs’ surface without surfactant assistance, which is depicted in Scheme 1. The axial materials include VGCFs, which possess a large surface area and serve as a conductive agent, thereby reducing the mass ratio of inactive materials in electrodes. Additionally, the VGCFs are treated in an acidic solution before use, which makes the surface more hydrophilic due to the presence of –COOH and –OH groups. Moreover, some impurities can also be removed thoroughly. The structural characterization and electrochemical performances of the MnO₂@VGCF composites are discussed and compared to the MnO₂ materials synthesized in a similar condition, but without the use of the VGCFs, known as standalone MnO₂.

2. Experimental Section

2.1. Chemicals and Materials

Potassium permanganate (KMnO₄, 99%), potassium sulfate (K₂SO₄, 99%), manganese sulfate (MnSO₄·H₂O, 99%), ammonium persulfate ((NH₄)₂S₂O₈, 98%) and ammonium sulfate ((NH₄)₂SO₄, 99%) were all purchased from Sigma-Aldrich, Shanghai, China. Acetylene black (Kappa 100, 99.9%) and poly (tetrafluoroethylene) (PTFE, 60%) solution were purchased from Canrd, Guangdong, China. Vapor-grown carbon fibers (VGCFs, diameters 120~150 nm) were purchased from Showa Denko, Japan.

2.2. Synthesis of the MnO₂ Nanorods and the MnO₂@VGCFs

6 M HNO₃ was used to treat commercial VGCFs and refluxed under 60 °C for two hours, thereby promoting the formation of –COOH and –OH hydrophilic groups on the VGCF surface and disposing of impurities. After being washed, the VGCFs were filtered and collected. The MnO₂ nanosheets were vertically deposited on the VGCF surface via redox reaction, following a modified method from the literature [32]. A schematic illustration is shown in Scheme 1. Briefly, 80 mg of VGCFs was placed in 100 mL of water containing 5 mmol of H₂SO₄ in a 0.25 L flask. Subsequently, 1 mmol of KMnO₄ was gradually added. After 20 min of stirring, the above mixture was kept under 85 °C for 3 h. The as-synthesized powder was filtrated and collected.

For comparison, the MnO₂ nanorods were synthesized via a hydrothermal method. In detail, the reaction solution was mixed with MnSO₄, (NH₄)₂S₂O₈ and (NH₄)₂SO₄ in a 1:1:4 molar ratio. Then, a 35 mL mixture was poured into a 0.05 L autoclave. Afterwards, the autoclave was kept under 140 °C for 1 h. Finally, the obtained powder was collected, washed and dried overnight at 90 °C.

2.3. Fabrication of Cathodes

The cathodes were fabricated by mixing the sample (MnO₂@VGCF or standalone MnO₂), poly (tetrafluoroethylene) (PTFE) and acetylene black. Their mass ratio was set as 75:10:15. The resulting flakes were beaten into small discs, each being about 2 mg in mass, 0.15 mm in thickness and 0.64 cm² in area. At last, these discs were kept at 120 °C for 12 h. The activated carbon (AC) anode was fabricated in an identical way. The aqueous electrolyte was the 0.5 M K₂SO₄ solution.

2.4. Characterization Techniques

SEM was performed on a FEI Nova NanoSEM NPE207 (America), and the acceleration voltage was 20 kV. Bruker D8 Discover (Germany) was applied to measure XRD data, the parameter of wavelength was set as 1.54 Å and the Cu X-ray tube was used in a X-ray diffractometer. An STA209 PC (Germany) analyzer was conducted on oxygen flow to obtain thermogravimetry (TG) data. An electrochemical workstation (CHI604 C, Chenhua Ltd. Co., Shanghai, China) was used to test electrochemical impedance spectroscopy (EIS) and cyclic voltammetry (CV). In a three-electrode system, galvanostatic charge/discharge (GCD) was carried out from 0 to 1 V versus SCE. A two-electrode supercapacitor was used for the investigation, consisting of the cathode (MnO₂@VGCFs or MnO₂) and the AC anode, filled with 0.5 M of K₂SO₄ aqueous electrolyte.

2.5. Calculation of Energy and Power Densities

The discharge capacitance C (F g⁻¹), power density P (W kg⁻¹) and energy density E (Wh kg⁻¹) were calculated from CV curves and GCD profiles as seen below:

$$C=\frac{\int IdU}{2vm\Delta U}$$

(1)

$$E=\frac{It{U}_a}{m}$$

(2)

$$P=\frac{I{U}_a}{m}$$

(3)

where I refers to the applied current, ΔU stands for the electrochemical window, m represents the total masses of the active materials in both the cathode (e.g., MnO₂@VGCF or MnO₂) and the anode (e.g., activated carbon), v typifies the sweep rate, 2 is a factor because both negative and positive scans are included, t stands for the time of one discharge cycle and U_a is the average voltage, which equates to (U₁ + U₂)/2; herein, U₁ is the initial voltage and U₂ is the cut-off voltage of the electrochemical window.

3. Results and Discussion

Usually, the following stoichiometric reaction formula is obeyed to synthesize MnO₂, which can deposit different carbonaceous substrates by redox reaction in situ [33].

$$4{\mathrm{KMnO}}_4+3\mathrm{C}+{\mathrm{H}}_2\mathrm{O}=4{\mathrm{MnO}}_2+2{\mathrm{KHCO}}_3+{\mathrm{K}}_2{\mathrm{CO}}_3$$

(4)

In this work, the VGCFs were utilized as both a reducing agent and as a substrate for MnO₂ growth. Figure 1 exhibits the scanning electron microscopy (SEM) of the MnO₂@VGCF composites and the standalone MnO₂ nanorods. This microscopy reveals that the as-prepared MnO₂@VGCFs (Figure 1a,b) existed in the shape of a 1D co-axial microstructure, along with a width of about 600 nm and a length of 5–10 μm. This microstructure was assembled from the MnO₂ nanosheet arrays vertically grown on the VGCF surface. This heterostructure makes the MnO₂ more electrochemically active because of its inherent connection with the VGCFs while leaving a straightforward ingress/egress of the electrolyte ions, thereby enhancing the surface pseudo-capacitance and electric double-layer capacitance (EDLC) formation. As for the standalone MnO₂ (Figure 1c), it is apparent that these MnO₂ nanorods are about 300–500 nm and 20 nm in length and width, respectively.

Figure 2 displays X-ray diffraction (XRD) patterns related to the MnO₂@VGCF composites and the standalone MnO₂. All diffraction characteristic peaks in Figure 2a were referred to as the hexagonal birnessite MnO₂, which is in good agreement with the standard card (JCPDS 18-0802) [34]. The intensity of the (002) plane from the composites is much stronger compared to other diffraction peaks, signifying the presence of graphite from the VGCFs at about 2θ = 26° [35]. Figure 2b exhibits the characteristic of δ-MnO₂ (potassium birnessite) that is consistent with the standard card (JCPDS 15-0604) [36]. Additionally, an impurity broad peak was detected at 2θ = 22°, which was ascribed to the (101) plane of γ-MnO₂ (ramsdellite, JCPDS 44-0142) [37].

Figure 3 exhibits the TG curves of the MnO₂@VGCF composites. Free water and crystal-water loss could be read at under 400 °C. Subsequently, the combustion of the VGCFs occurred from 400 to 550 °C. The mass loss related to the reduction in MnO₂→1/2Mn₂O₃ + 1/2O₂ was observed between 550–650 °C [38,39], which is similar to that of the standalone MnO₂. From the TGA results, the mass percentage of the MnO₂ in the MnO₂/VGCF composites was calculated to be about 65%.

Figure 4a displays the cyclic voltammetry (CV) of the MnO₂@VGCF composites in the 0.5 M K₂SO₄ electrolyte between 0–1 V (vs. SCE). It is obvious that the MnO₂@VGCFs composites showed an ideal symmetric rectangular shape: that is to say that this is typical pseudo-capacitive behavior. Interestingly, the rectangular shape was still symmetric at a high scan rate of 500 mV s⁻¹, signifying that the composites possess good rate capabilities [40]. Figure 4b–d demonstrate the CV curves of the MnO₂ in 0.5 M K₂SO₄ electrolytes at 10, 100 and 500 mV s⁻¹, respectively. At the slow sweep rate of 10 and 100 mV s⁻¹, the MnO₂ nanorods showed small but distinct redox peaks, which is possible because of the reversible ingress/egress of the potassium ions into the MnO₂ [41,42]. However, in the MnO₂@VGCF composites, there were no apparent redox peaks. This may be explained by the heterostructure, which makes the deposited MnO₂ nanosheets more electrochemically active and makes the intercalation of the electrolyte ions easier, thereby promoting pseudo-capacitance formation. This result is unanimous with what has been discussed in the literature [35,43].

At the faster sweep rate of 500 mV s⁻¹ (Figure 4d), these redox peaks in the MnO₂ nanorods diminished. In addition, the CV curves demonstrated an obvious polarization that deviated from pure capacitive behavior, which might be because the MnO₂ nanorods cannot afford a very high current [5].

The influences of sweep rates on the capacitance for standalone MnO₂ nanorods and MnO₂@VGCF composites are demonstrated in Figure 4e. By integrating the areas of the CV curves, the specific capacitances can be found. In regions of slow scan rates, the capacitance values of the standalone MnO₂ nanorods were greater than those of the MnO₂@VGCF composites. This finding can be explained as follows. At slow scan rates, the standalone MnO₂ nanorods produced an additional battery-type capacity contribution besides the absorption/desorption reaction, resulting in a higher capacitance than that of the MnO₂@VGCF composites. After increasing the scan rate, the effective contribution of the redox reaction was controlled to some extent, owing to poor conductivity, causing the falloff of the capacitance of the standalone MnO₂ nanorods. However, the capacitance of the MnO₂@VGCF composites only slightly declined. This can be accredited to the VGCFs’ conductive additive, which can load large currents during discharging/charging processes and improve the rate capability of deposited MnO₂ nanosheets. As a consequence, the MnO₂@VGCF composites showed a larger capacitance compared to the MnO₂ nanorods at fast sweep rates.

For further understanding, the impedance of the MnO₂ nanorods and the MnO₂@VGCF composites after 20 cycles were measured from 10⁵ to 10⁻¹ Hz at their open circuit potentials. An estimation of the impedance data was carried out with ZView software version 2.0. The tested impedance data were analyzed using a complex nonlinear least-squares fitting (NLSF) method [28,44] based on the equivalent circuit (Figure 4f). The experimental and fitting data are displayed in Figure 5a,b, and the values of the elements in the equivalent circuit are listed in

Table 1. Fitting results of the electrochemical impedance.

Types	MnO2	MnO2@VGCFs
Rs	1.78	1.87
CPE-T	5.43 × 10−3	4.46 × 10−3
CPE-P	0.74	0.78
Rct	3.49	1.44
L	2.0 × 10−6	1.0 × 10−7
Wo-R	8.16 × 10−3	8.34 × 10−4
Wo-T	7.09 × 10−4	1.85 × 10−4
Wo-P	0.45	0.44

Notes: R_s (solution resistance, Unit: Ω); CPE-T (constant phase element, Unit: S); CPE-P (exponent, no unit); R_ct (charge-transfer resistance, Unit: Ω); L (inductance, Unit: H); W_o-R (Warburg resistance, Unit: Ω); W_o-T (Warburg conductance, Unit: S); W_o-P (exponent, no unit).

. The intercept at the real part (Z’) axis at the highest frequency stands for the solution resistance (R_s). The R_s of the two electrodes were almost identical. R_ct represents the charge-transfer resistance caused by Faradaic reactions of the active materials. The MnO₂ electrode presented an R_ct value of 3.49 Ω, whereas the MnO₂@VGCF electrode showed a lower value of R_ct of 1.44 Ω, thanks to the high electronic conductivity of the VGCFs. This is also a reason for the good CV performances of the MnO₂@VGCF composites at a large sweep rate of 500 mV s⁻¹. Inductance is the property of an electric circuit element that demonstrates opposition to the change of current flowing in it. From the viewpoint of common terminology in electric circuits, an inductor is defined as a two-terminal element, consisting of a winding of many (mostly denoted as N) turns to introduce inductance into an electric circuit [45]. In this work, an inductor (L) was introduced in series with a constant phase element (Q_CPE), and then paralleled with the charge-transfer resistor (R_ct) in the equivalent circuit, which refers to the negative values of the imaginary part in the high frequency. The close fitting of the EIS data (Figure 5a,b) on the basis of the equivalent circuit (Figure 4f) implies that this equivalent circuit model can successfully represent the electrochemical processes of the active materials. The values of the inductance for the MnO₂ and the MnO₂@VGCF electrodes were 2 × 10⁻⁶ and 1 × 10⁻⁷ H, respectively. These values suggest that the as-prepared materials possessed metallic properties [46]. In the region of low frequency, the linear part was attributed to the frequency dependence of K⁺ ions transport/diffusion in the host, which is normally called Warburg diffusion [45]. Therefore, the Warburg impedance [47] is:

$$W_o=\frac{\Delta V}{\Delta I}=R\frac{1}{l}{(\frac{D}{i\omega })}^P\tan \mathrm{h}[l·{\left(\frac{i\omega }{D}\right)}^P]$$

(5)

$$\mathrm{R}=\frac{k_{B}Tl}{{\left(ne\right)}^{2}AD{C}_o}$$

(6)

where ω is angular frequency, i is equal to $\sqrt{-1},$k_B is Boltzmann constant, T is time, l is diffusion length, n is number of transfer electron, A is area of electrode, D is diffusion coefficient, C_o is ion concentration.

The constant phase element (Q_CPE) represents the non-uniform charge distribution at the grain interfaces, and describes the deformed nature of the semicircles in Nyquist plots, which is written below [48]:

$$Q_{CPE}={\left(Y_o{\left(j\omega \right)}^P\right)}^{-1}$$

(7)

where Y_o is determined by a combination of some properties of both electro-active parts and surfaces. The values of the parameter Y_o can be obtained from the fitting results, which equate to those of the CPE-T and W_o-T for the CPE element and Warburg element, respectively. The P is frequency power, and when P = 0, 1 and −1, this refers to a resistor, capacitor and inductor, respectively; when P = 0.5, it represents Warburg diffusion.

Figure 6a manifests the constant current charge/discharge profiles of an assembled supercapacitor, which consists of the MnO₂@VGCFs’ cathode and the AC anode using a 0.5 M K₂SO₄ electrolyte. The mass ratios of AC to MnO₂ and MnO₂@VGCF were fixed at 1:1.25. The electrochemical window for the supercapacitor was between 0 and 1.8 V. The GCD profiles are, on the whole, symmetric, which also indirectly proves ideal pseudo-capacitive properties. The Ragone plots of two aqueous SCs are shown in Figure 6b. The total masses of the two electrodes were adopted for calculations. The energy density of the MnO₂ nanorods was 16 Wh kg⁻¹ at 985 W kg⁻¹, diminishing to 7 Wh kg⁻¹ at 8600 W kg⁻¹. On the contrary, the energy density of the MnO₂@VGCF composites was 22.6 Wh kg⁻¹ at 1000 W kg⁻¹. Moreover, it retained a superior rate capability, accompanied with 10.7 Wh kg⁻¹ even at 19,200 W kg⁻¹. These data are much better than that of the MnO₂ nanorods. Distinctly, this good rate capability is credited to the VGCFs’ co-axial structure, which is well matched with the CV results. Additionally, the rate property of the composites is better than that of previously published results of MnO₂ [49,50], which is possible because the mobility of hydrated potassium ions is higher than that of hydrated sodium ions in aqueous solution [5,51].

The cycling performance of the assembled AC/0.5 M K₂SO₄/MnO₂@VGCF SCs at 1000 mA g⁻¹ is displayed in Figure 7. The discharge capacitance of the MnO₂@VGCF composites was 50 F/g, which is larger when compared with the standalone MnO₂ nanorods. After 10,000 cycles, there was no apparent capacitance fading. In contrast, the MnO₂ nanorods exhibited a slight capacitance fading in the K₂SO₄ aqueous solution. The loss percentage was about 4% after 10,000 cycles. Consequently, the VGCFs’ co-axial structure for the MnO₂@VGCFs composites presented good cycling and high specific capacitance at a high current density, which is compossible with the beforehand tests of rate behavior.

4. Conclusions

In this work, we prepared a composite of MnO₂ nanosheet arrays vertically grown on the VGCF surface via a low-temperature redox reaction between KMnO₄ and the VGCFs. The as-synthesized MnO₂@VGCF composites demonstrated a larger capacitance, superior rate behavior and cycle performance compared with the standalone MnO₂ synthesized by the hydrothermal method, which suggests that VGCFs work well as a conductive additive for aqueous electrochemical capacitors.