Introduction

As the depletion of fossil fuels and environmental pollution leads to the accelerated development of renewable and clean energy conversion/storage systems, two types of electrochemical devices such as batteries and electrochemical capacitors (ECs) have been paid more attention. Batteries with high energy densities suffer from slow power delivery, whereas ECs exhibit high power but low energy densities. Hence both high energy and high power without sacrificing cycle life will be demanded for future electrochemical energy storage applications. This demand can be met by the ECs if their energy densities can be enhanced efficiently1,2,3,4,5,6,7. High power density, long cycle life and environment-friendly, compared to traditional batteries, make ECs promising energy storage devices in a wide range of applications, such as hybrid electric vehicles, mobile electronic devices, large industrial equipments and space or military devices, wherein relatively light weight and small size are desirable. However, the disadvantage of ECs including low energy densities has been identified as a major challenge for the capacitive storage science2,4,8.

Electrochemical performance of the ECs depends on both electrode and electrolyte. For electric double layer capacitors (EDLCs) employed high-surface-area carbon materials as the electrode in aqueous or organic electrolytes9,10,11, electric charges physically stored on the carbon particles in porous electrode layers are unfortunately limited, resulting in a low specific capacitance, around 150 Fg−112,13. Compared to carbon-based materials used in EDLCs, low-cost pseudo-capacitive active materials such as Ni(OH)2, NiO, Co(OH)2, Co3O4, MnO2 et al. have much higher theoretical capacitance (above 1000 Fg−1) because of the presence of faradaic redox reactions in aqueous electrolytes. However, faradaic redox reactions also exhibit a low rate capability and low cycling stability due to the poor conductivity4,14,15. With nanostructured carbon used as support for direct growth of active materials, nanostructured three-dimensional electrode architectures have been realized to tailor ionic and electronic transport and increase efficient utilization of pseudo-capacitive materials16,17,18,19. Although the combination of EDLC with pseudocapacitor electrode is able to improve the electrochemical performance of ECs effectively, the energy density cannot be improved significantly via only relying on the active electrode materials.

Very recently, there have been a few reports that redox additives are introduced into the electrolyte for carbon EDLCs to substantially enhance the capacitance via redox reactions of the additives between the electrode and electrolyte. For example, Wu et al20. have introduced phenylenediamine as a redox mediator into KOH electrolyte for the carbon-based supercapacitor and a much higher specific capacitance (605.2 Fg−1) than that (144.0 Fg−1) with conventional KOH electrolyte has been achieved. Roldan et al21. have reported that via adding electrochemically active compound quinone/hydroquinone (Q/HQ) into the H2SO4 electrolyte, an ultrahigh specific capacitance of 5017 Fg−1 has been obtained for carbon-based supercapacitor. The significantly improved capacitance, upon introduction of HQ into the electrolyte, attributes to that the faradaic reactions of the Q/HQ couples () contribute pseudocapacitance to the electrode. However, the working condition for realizing the reactions of these quinoid compounds is that [H+] > [Q] in electrolyte as unbuffered media. This working condition is not easily maintained, leading to a poor electrochemical stability. Alshareef et al22. have added the Q/HQ into H2SO4 electrolyte and found that the specific capacitance of rGO/PAni is enhanced. However, the specific capacitance of rGO/PAni-Q/HQ decreases to 64% of initial value after 50000 cycling at a relatively small current density 10 Ag−1, which is close to that of rGO/PAni, meaning that the capacitance from Q/HQ disappears gradually. This demonstrates that the capacitance contributed by Q/HQ is not constant and stable. Senthilkumar et al23. have improved the capacitance for activated carbon materials through adding either KI or KBr into 1 M H2SO4 or Na2SO4 electrolyte. However, there is also a problem with the electrochemical stability for only relying on the pseudo-capacitive contribution from additive for EDLCs.

As for the pseudocapacitors, is it possible to improve the pseudocapacitance via introducing redox additives into electrolyte? So far, there have been few reports on this investigation except that Su et al24. have improved the capacitance of Co–Al layered double hydroxide (LDH) electrode through the addition of either K3Fe(CN)6 or K4Fe(CN)6 into KOH solution. The Co–Al LDH in mixed either KOH/K3Fe(CN)6 or KOH/K4Fe(CN)6 electrolytes exhibits an increased capacitance up to 712 and 317 Fg−1 at a current density of 2 Ag−1, respectively, which is higher than that only in KOH solution (226 Fg−1). This enhancement in capacitance can be ascribed to that Fe(CN)64−/Fe(CN)63− promotes the electron gain and loss of Co ions during charging/discharging. However, no redox peaks from Fe(CN)64−/Fe(CN)63− are observed. Also, unfortunately, the electrochemical stability for Co-Al LDH after introducing either K3Fe(CN)6 or K4Fe(CN)6 into electrolyte is not discussed, although there is a problem with the electrochemical stability for Co-Al LDH too.

As mentioned above, although more attention has been paid to investigating either electrode or electrolyte for enhancing the capacitance, no any efforts have been done to improve both the solid electrode and liquid electrolyte simultaneously for pseudocapacitors. In previous work, our obtained binder-free Co(OH)2/GNS/Ni foam electrode exhibits a high rate capability and long-term cycle life, in which GNS provides a high electrical conductivity and high specific surface area, allowing rapid and effective charge transfer and electron transport. Hence GNS play an important role in improving the pseudocapacitance of Co(OH)2, including enhancing its rate capability and electrochemical stability25. In this work, we aim at further improving the pseudo-capacitive behavior for Co(OH)2/GNS electrode through adding K3Fe(CN)6 into KOH electrolyte, based on that Fe(CN)63−/Fe(CN)64− redox pair has a high solubility and electrochemical activity. Our findings are that Fe(CN)63−/Fe(CN)64− not only promotes electron gain and loss of Co ions, but also directly gives a contribution to the pseudocapacitance for Co(OH)2/GNS-KOH/K3Fe(CN)6. Our designed and prepared Co(OH)2/GNS-KOH/K3Fe(CN)6 shows an unusually high specific capacitance (7514 Fg−1 at 16 Ag−1 in mixed 1 M KOH and 0.08 M K3Fe(CN)6), more than 100% coulombic efficiency and long-term cycling stability (the capacitance retention is 75% after 20000 continuous charge-discharge cycles in mixed 1 M KOH and 0.04 M K3Fe(CN)6).

Results

The XRD measurements for Co(OH)2/GNS before electrochemical test and after 2000 charge-discharge cycles, in the mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution, at the current density of 80 Ag−1, show that XRD patterns of Co(OH)2/GNS before electrochemical test and after 2000 cycles are almost the same and no new diffraction peaks are detected after 2000 cycles (Supporting material 1). Also, the morphologies for original Co(OH)2/GNS and Co(OH)2/GNS after 2000 cycles are nearly the same (supporting material 2). XRD and SEM measurements confirm that the novel electrode-electrolyte system is stable.

In order to compare the redox reactions for Co(OH)2/GNS before and after adding K3Fe(CN)6 into KOH electrolyte, the cyclic voltammograms (CV) measurements have been performed. Figure 1(a)–(c) display for three systems: (1) Co(OH)2/GNS electrode in 1 M KOH aqueous solution (Co(OH)2/GNS-KOH); (2) platinum plate electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 aqueous solution (Pt-KOH/K3Fe(CN)6); (3) Co(OH)2/GNS electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 aqueous solution (Co(OH)2/GNS-KOH/K3Fe(CN)6), at different scan rates (2, 5, 10, 25 and 50 mVs−1) in the potential range of −0.2–0.5 V, respectively. Here, system 1 and 2 is chosen for comparison. For Co(OH)2 active material as electrode, the oxidation and reduction reactions observe the following faradaic reactions:

For system 1 with Co(OH)2/GNS as electrode in 1 M KOH aqueous solution, as shown in Fig. 1(a), a reversible charge-transfer process is observed: the charging process corresponds to the oxidation of Co(OH)2 to CoOOH, while the discharging process corresponds to the reduction of CoOOH to Co(OH)2. The oxidation and reduction peaks appear at 0.033 V and 0.022 V, respectively and the peak potential separation (ΔEP) is 0.011 V at a scan rate of 2 mVs−1. The oxidation peak upshifts and the reduction peak downshifts slightly with increasing scan rate. In addition, the peak current of anodic oxidation almost equals to that of the cathodic reduction for each curve. Hence the Co(OH)2/GNS electrode exhibits a high electrochemical activity and good charge/discharge reversibility. This capacitance behavior can be explained by that GNS provides a high electrical conductivity and high specific surface area, allowing rapid and effective electron and ion transport, while Co(OH)2 nanosheets grown on GNS realizes a fast charge transfer, providing a high electrochemical activity and redox reversibility. For system 2 with the platinum plate as electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 aqueous solution, as shown in Fig. 1(b), a pair of reversible redox peaks are observed, where the oxidation peak is related to the charging process of K4Fe(CN)6 to K3Fe(CN)6 and reduced peak is from the reverse process, corresponding to the following reactions:

The characteristic CV shape of K3Fe(CN)6 is not significantly influenced as the scan rate is increased. The redox reaction on platinum electrode is a fast and reversible electrochemical process, indicating that K3Fe(CN)6 has a high electrochemical activity. For system 3 with the Co(OH)2/GNS as electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 aqueous solution, as shown in Fig. 1(c), two symmetric anodic/cathodic pairs superimposed on a broad redox background are recognized, indicating that reversible redox reactions of Co(OH)2 solid electrode and K3Fe(CN)6 liquid electrolyte occur simultaneously and independently, according to the equations (1) and (2).

Figure 1
figure 1

Cyclic voltammograms at different scan rates (2, 5, 10, 25 and 50 mVs−1) for system: Co(OH)2/GNS electrode in 1 M KOH solution (a), platinum electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution (b) and Co(OH)2/GNS electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution (c).

The K3Fe(CN)6-dependent capacitive properties can be determined from the charge-discharge potential vs. time curve. Figure 2(a) and 2(b) show the charge-discharge potential vs. time curve for system 1 or 3 at high current density of 16 or 32 Ag−1 in the potential range of −0.1–0.45 V. From Fig. 2(a), the potential plateaus at about 0.05 V in charging process and 0.00 V in discharging processes correspond to the oxidation of Co(OH)2 and the reduction of CoOOH, respectively. From Fig. 2(b), besides the potential plateaus at about 0.025 V in charging process and −0.025 V in discharging processes, corresponding to the redox reaction of Co(OH)2, there are other potential plateaus at about 0.25 V in charging process and 0.18 V in discharging process, corresponding to the redox reaction of K3Fe(CN)6. The potential plateaus result from a lot of exchanges occurred between electrons, corresponding to the anodic and cathodic peaks in the CV curves. The specific capacitance and coulombic efficiency for the two systems can be determined based on the equation (3) and (4), respectively. From the galvanostatic charge-discharge curve at 16 Ag−1 (32 Ag−1), the specific capacitance for system 1 is 567.3 Fg−1 (517.8 Fg−1), while that for system 3 is 2434.9 Fg−1 (1733.0 Fg−1). The coulombic efficiency for system 1 is 93.2% (90.6%) and 163.8% (121.6%) for system 3. The high specific capacitance for system 3 is due to that both Co(OH)2 and K3Fe(CN)6 directly do contribute the pseudocapacitance to the system 3, while only Co(OH)2 gives a contribution of pseudocapacitance to system 1. As for the high coulombic efficiency for system 3, it can be explained by that for the initial additive of Fe(CN)63−, according to the equation (2), in discharging process, only part of Fe(CN)63− gains an electron and is reduced into Fe(CN)64−, while in charging process, Fe(CN)64− loses an electron and is completely oxidized into Fe(CN)63−. Since the released charges include both discharging and no discharging Fe(CN)63−, which can be much more than reserved charges in charging process. This is the reason why the coulombic efficiency is larger than 100%. For example, the electric charge with a quantity of 77.2 C for system 3 can be completely released from 0.04 M K3Fe(CN)6 in 20 mL solution, if all K3Fe(CN)6 is reduced into K4Fe(CN)6. In fact, in a single charge-discharge process for K3Fe(CN)6 electrolyte, the electric charge with a quantity of 0.024 C is charged accompanied by the change of K4Fe(CN)6 into K3Fe(CN)6 and 0.027 C is discharged accompanied by the change of K3Fe(CN)6 into K4Fe(CN)6 for system 2 (both positive and negative electrodes consist of platinum plate). After each cycling, the K3Fe(CN)6 concentration in KOH electrolyte changes slightly (only 0.04‰ of K3Fe(CN)6 after each cycling is consumed) compared with that at the initial state, demonstrating that KOH/K3Fe(CN)6 electrolyte is similar to a constant buffer solution. Hence the high coulombic efficiency (more than 100%) can be maintained for a long time. In particular, if the pseudocapacitance mainly comes from Co(OH)2, rather than K3Fe(CN)6, the K3Fe(CN)6 will be consumed slowly, resulting in a long electrochemical stability for system 3. However, the coulombic efficiency decreases with increasing cycle number, which can be attributed to the consumption of Fe(CN)63− in each discharging process. The high coulombic efficiency is essential for a battery-type supercapacitor device with a high energy density. In the practical application, the high coulombic efficiency can be recovered via only charging the electrolyte containing K4Fe(CN)6. As the electrolyte is applied at a high constant potential, Fe(CN)64− will be oxidized completely. As a result, all K4Fe(CN)6 can be transformed into K3Fe(CN)6 and the high coulombic efficiency can be recovered (Supporting materials 3).

Figure 2
figure 2

Galvanostatic charge-discharge curves at different current densities for Co(OH)2/GNS electrode in 1 M KOH solution (a) and Co(OH)2/GNS electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution (b).

The electrochemical stability is a very important factor for determining the capacitive properties of pseudocapacitors with introduction of additive into electrolyte. The cycling stability for either system 1 or 3 is examined by continuous charge-discharge experiments for 2000 cycles at a high current density of 80 Ag−1, as shown in Fig. 3(a). Both systems exhibit a high cycling stability, with retention of 94.4% and 91.1% initial capacitance after 2000 cycles. This indicates that either system 1 or 3 is stable and the addition of K3Fe(CN)6 into KOH electrolyte does not obviously affect the stability of the electrode material, which is consistent with the results from XRD and SEM. The cycling stability of system 3 is also performed at 80 Ag−1 for 20000 continuous charge-discharge cycles, as shown in Fig. 3(b), from which the system maintains a high specific capacitance of 610.9 Fg−1 after 20000 cycles and retention of 75.0% the initial specific capacitance (814.5 Fg−1), indicating that the novel Co(OH)2/GNS-KOH/K3Fe(CN)6 system exhibits a high electrochemical activity and excellent cycling stability.

Figure 3
figure 3

Cycling performance of Co(OH)2/GNS electrode in 1 M KOH and Co(OH)2/GNS electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution for 2000 cycles (a) and Co(OH)2/GNS electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution for 20000 cycles (b), measured by using the galvanostatic charge-discharge technique at a current density of 80 Ag−1.

The comparison in electrical conductivity between system 1 and 3, determined by the electrochemical impedance spectroscopy (EIS) measurements, is shown in Fig. 4, from which system 3 possesses a smaller interfacial charge-transfer resistance (Rct, calculated from the span of the single semi-circle along the x-axis from the high to low frequency region) and lower internal resistance (Ri, the intersecting point with the x-axis in the range of high frequency), compared to system 1. Such a pattern of the plots can be fitted by that from an equivalent circuit in the inset of Fig. 4, in which Rct = 0.1  and Ri = 300.9  for system 3, while Rct = 10.0  and Ri = 500.2  for system 126,27. The lower resistance for system 3 can afford a facile ionic and electronic transfer to ensure a high capacitive performance. Clearly, the electrochemical performances for system 3 are enhanced remarkably after introducing the redox mediator of K3Fe(CN)6 into KOH electrolyte.

Figure 4
figure 4

Nyquist plots of Co(OH)2/GNS electrode in 1 M KOH (a) and Co(OH)2/GNS electrode in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution (b), in which the inset shows the equivalent circuit.

Finally, the effect of the K3Fe(CN)6 concentration in KOH electrolyte on the pesudocapacitive properties for system 3 is explored and the CV curves for system 3 with a concentration K3Fe(CN)6 of 0.00, 0.01, 0.02, 0.04, 0.06 and 0.08 M, respectively, at a scan rate of 25 mVs−1, are shown in Fig. 5(a). The peak current from Co(OH)2 decreases slightly first, then increases and finally almost keeps constant and that from K3Fe(CN)6 increases obviously with increasing the K3Fe(CN)6 concentration (Supporting materials 4), indicating the capacitance contributed by Co(OH)2 decreases slightly first, then increases and finally almost keeps constant. In contrast, the capacitance contributed by K3Fe(CN)6 increases obviously. The peak currents for the oxidation of Co(OH)2 and reduction of CoOOH are depressed, resulting from the adsorption of Fe(CN)63−/Fe(CN)64− on the Co(OH)2/GNS electrode surface, which suppresses the OH diffusion. This suppression leads to a shortage of OH, decreasing the reaction rate of Co(OH)2/CoOOH. On the other hand, upon further increasing the K3Fe(CN)6 concentration, more Fe(CN)63−/Fe(CN)64− ions play the role of “electron shuttle” in the charge/discharge processes of Co(OH)2/CoOOH, promoting a high activity of Co(OH)2/CoOOH24. This promotion results in an increase in peak current. As above two effects are balanced, the peak current keeps almost constant. In addition, anodic and cathodic peak currents corresponding to redox reaction of Co(OH)2/CoOOH downshift, which is also caused by electrostatic adsorption, the more Fe(CN)63−/Fe(CN)64− ions are adsorbed on the Co(OH)2 electrode surface, the larger the resistance to OH diffusion is. In Fig. 5(b), as the K3Fe(CN)6 concentration in system 3 is 0.01, 0.02, 0.04, 0.06 and 0.08 M, respectively, the specific capacitance, evaluated at a high current density of 16 Ag−1 from charge-discharge curves, will be 852.4, 1093.8, 2434.9, 4983.3 and 7514.2 Fg−1, respectively, corresponding to the coulombic efficiency of 103.9%, 125.8%, 163.8%, 285.5% and 541.4%, respectively (charge and discharge time are listed in Table 1). The system can be charged quickly and discharged slowly, which means that the promise can be offered to realize a battery-type supercapacitor. In addition, if the high-rate discharge ability (HRD) of the electrode is defined as the ratio of discharge capacitance at 32 Ag−1 to that at 16 Ag−1, the HRD for system 3 containing K3Fe(CN)6 with a concentration of 0.01, 0.02, 0.04, 0.06 and 0.08 M, respectively, will be 86.0%, 80.9%, 71.2%, 52.8% and 30.4%, respectively. The typical data about the specific capacitance and HRD for system 3 containing various of K3Fe(CN)6 concentrations are shown in Fig. 5(c). When the K3Fe(CN)6 concentration is lower, the contribution of the redox reaction from K3Fe(CN)6 is lower, resulting in a low specific capacitance. However, high concentration and charge-discharge current density will cause high concentration polarization, leading to a low rate property and poor electrochemical stability. From above, the K3Fe(CN)6 concentration has a significant influence on the specific capacitance, coulombic efficiency and rate property. High K3Fe(CN)6 concentration facilitates a high specific capacitance and coulombic efficiency, but worsens a rate property and electrochemical stability. Therefore, only the K3Fe(CN)6 concentration is proper, the energy density, power density, coulombic efficiency and long-term cycle life can be compromised. For one system, a high specific capacitance indicates a high energy density, while a high power density associates with a large current charge-discharge characteristic. Through well controlling the K3Fe(CN)6 concentration, a novel system with high energy density, power density, coulombic efficiency and long-term cycle life can be realized.

Table 1 Charge and discharge time for Co(OH)2/GNS electrode in 1 M KOH containing different concentrations of K3Fe(CN)6 mixed solution, calculated from galvanostatic charge-discharge curves at a current density of 16 Ag−1
Figure 5
figure 5

Cyclic voltammograms at a scan rate of 25 mVs−1 (a), Galvanostatic charge-discharge curves at a current density of 16 Ag−1 (b), specific capacitance and high-rate discharge ability vs. different K3Fe(CN)6 concentrations (c) for the system of Co(OH)2/GNS electrode in 1 M KOH containing different concentrations of K3Fe(CN)6 mixed solution.

Discussion

In summary, a simple and effective method has been implemented to enhance the electrochemical performance for Co(OH)2/GNS electrode through introducing K3Fe(CN)6 into the conventional KOH electrolyte. As both solid electrode and liquid electrolyte can contribute to the pseudocapacitance simultaneously, an ultrahigh capacitive performance has been realized. The system exhibits not only an ultrahigh specific capacitance and coulombic efficiency (a maximum specific capacitance of 7514.2 Fg−1 and coulombic efficiency of 541.4% at a high current density of 16 Ag−1 in the mixed 1 M KOH and 0.08 M K3Fe(CN)6 solution), but also an excellent cycling stability (the capacitance retention is 91.1% after 2000 continuous charge-discharge cycles and 75.0% after 20000 continuous cycles at 80 Ag−1 in mixed 1 M KOH and 0.04 M K3Fe(CN)6 solution). Upon increasing the K3Fe(CN)6 concentration, both specific capacitance and coulombic efficiency increase, but the rate property is worsened. The energy density and power density for the system can be compromised via well controlling the K3Fe(CN)6 concentration according to the practical application and the high coulombic efficiency (more than 100%) can be recovered by charging the electrolyte. This encouraging investigation shows great potential in developing a battery-type supercapacitor with a high energy density, power density, coulombic efficiency and long-term cycling stability.

Methods

Vertically oriented GNS were synthesized by radio-frequency plasma-enhanced chemical vapour deposition (RF-PECVD) on nickel-foam used as substrates for cathodic electrodeposition of Co(OH)2 nanosheets in Co(NO3)2 aqueous solution, the detailed description about the preparation can be found in ref. 25. The microstructure and surface morphology of the electrode materials were characterized by X-ray diffraction (XRD) (RIGAKU D/MAX2500) and field emission scanning electron microscopy (FE-SEM) (JEOL JSM-6700F). Electrochemical measurements were carried out via a computer-controlled electrochemical working station (CHI760D or PARSTAT2273) with a conventional three-electrode electrochemical cell. The electrolytes were 1 M KOH aqueous solution with or without K3Fe(CN)6 with different concentrations. The working electrode was the prepared Co(OH)2/GNS/Ni foam and a platinum plate and a saturated calomel electrode (SCE) were used as counter electrode and reference electrode, respectively. The specific capacitance of the system is calculated from galvanostatic charge-discharge curves according to equation (3):

where I (A) is discharging current, Δt (s) is discharging time, ΔV (V) is potential drop during discharging process and m (g) is mass of the Co(OH)2. Another important parameter coulombic efficiency (η) is evaluated from equation (4):

where tD and tC are discharging time and charging time, respectively.