Introduction

Given that its exceptionally high surface area (up to 2630 m2/g) and electrical conductivity (106 S/cm), graphene has been considered as the promising electrode material for the supercapacitors1. In addition to graphene, hierarchical nanostructured carbons (HNCs) containing 3D-interconnected macroporous/mesoporous and mesoporous/microporous structures have also attracted considerable attention because the electrode materials with hierarchical nanostructures offer the advantages of high surface area, synergistic interactions and multiple functionalities2,3,4,5,6,7. Similar to other carbonaceous electrode materials such as activated carbon and carbon nanotubes (CNTs), the energy storage mechanism of graphene or HNC based electrodes mainly involves electrical double-layer capacitance. In the developments of graphene based supercapacitors, chemically reduced graphene oxide (rGO) is widely used as the substitute for pristine graphene due to the fact that the chemical exfoliation processes enable the mass production of graphene8. However, the outstanding properties of graphene are significantly weakened in rGO because the re-stacking of rGO via the strong π-π interaction greatly loses the useful surface area and the residual defects in rGO significantly reduce the conductivity of graphene. The aforementioned limitations greatly impair the electrochemical performances of rGO electrodes; as a result, the supercapacitors based on rGO electrodes usually show that the specific capacitances (Cm) range from 100 to 200 F/g in the aqueous electrolytes9,10,11.

Due to the limited capacitance of graphene-based electrodes, much attention has been paid to exploit electrode materials with pseudocapacitive properties such as metal oxide12,13,14,15 and conducting polymers16,17,18,19. Recently, it has been demonstrated that nickel sulfides (NiS) with various nanostructures display high capacitive performance and great long-term stability, suggesting that NiS is a promising supercapacitor electrode material20,21,22. Although metal oxides or sulfides usually show the higher Cm values than graphene-based electrodes, they generally suffer from serious particle agglomeration and poor electrical conductivity, which significantly restricts their applications in supercapacitors. To fully utilize all the potential advantages of graphene and pseudocapacitive materials, much effort has been devoted to the studies of the composites of graphene and pseudocapacitive materials. Many studies have demonstrated that the composite electrodes exhibit the improved capacitive performances compared with the electrodes consisting of the individual components of graphene or pseudocapacitive materials23. For the composite electrode, the enhancements of the electrochemical performances mainly benefit from the introduction of graphene. The graphene in a composite electrode serves as the multifunctional support for anchoring electroactive materials. From the viewpoint of graphene properties, graphene can be considered as an ideal building block in composite electrodes for supercapacitors. Firstly, the high surface area of graphene can induce the uniform growth of nanomaterials on the graphene surface. On the other hand, the electroactive nanomaterials deposited on graphene suppress the aggregation and the re-stacking of graphene, thus increasing the available surface area of the graphene. Secondly, the high conductivity of graphene enables fast electron transport through the underlying graphene layers to electroactive nanomaterials. Finally, the great mechanical flexibility of graphene improves the cycle stability and rate capability of composite electrode for supercapacitors24. Due to the aforementioned synergistic effects of graphene and electroactive nanomaterials, the enhancement of capacitive performance of composite electrode is significant.

Over the past years, an asymmetric supercapacitor (ASC) has aroused more and more attention because it can deliver a remarkably high power density and an ultrahigh energy density. By combining complementary potential windows of the cathode and anode, the ASC can be operated in a wider voltage window and thus exhibits the greater energy density compared with symmetric supercapacitors25. In this study, we report the fabrication of high energy density ASC assembled with highly capacitive graphene-based composite cathode and anode. The composite of Ni3S2 nanoparticles and 3D graphene (Ni3S2/3D-G) synthesized by glucose-assisted hydrothermal method is used as the cathode for ASCs. This novel hybrid electrode exhibits an ultrahigh capacitance of 3296 F/g and good cycling stability at a high current density of 16 A/g. To satisfy the capacitance balance between the cathode and anode, the composite of Fe3O4 nanoparticles and rGO (Fe3O4/rGO) synthesized by a hydrothermal method serves as a highly capacitive anode material. Due to the synergistic effects of conductive rGO and Fe3O4 nanoparticles, the anode shows a high capacitance of 661 F/g at a current density of 1 A/g and a good rate capability. Finally, the ASC assembled with two different composite electrodes can deliver a high energy density of 82.5 Wh/kg at a power density of 930 W/kg.

Results

Material characterization of Ni3S2/3D-G

As shown in Figure S1a and S1b, scanning electron microscopy (SEM) images reveal that the as-grown graphene films adhere to the surface of nickel foam. Due to the difference between the thermal expansion coefficients of nickel and graphene, ripples and wrinkles of the graphene films are observed26. The thickness of the as-grown graphene films on Ni foam was estimated by Raman spectroscopy. As shown in Figure S1c, the peaks at around 1579 and 2660 cm−1 correspond to the G and 2D bands of graphene, respectively. Furthermore, the disorder-induced D band at ca. 1350 cm−1 is not observed, thus indicating the high quality of as-grown graphene. On the basis of the G to 2D peak intensity ratios, the as-grown 3D-G consists of bi- to few-layer graphene sheets27. The non-uniformity is attributed to the polycrystalline nature of the nickel foam, in which individual nickel grains may independently affect the thickness of the graphene film during chemical vapor deposition (CVD)27. The morphology of the 3D-G sample after the hydrothermal reaction was also characterized by SEM. As shown in Figure 1a and 1b, the graphene surface is uniformly and densely decorated with the nanoparticles. In the high-magnification SEM (Figure 1c), it is found that the morphology of a particle is irregular and the diameter of the irregularly shaped particle is ca. 1 μm. The transmission electron microscopy (TEM) image of the composite sample confirm that conductive graphene can serve as the substrate where small nanoparticles are self-assembled into irregularly shaped particles as indicated by the red circles in Figure 1d, which is consistent with our previous study28. Our previous work has verified that the use of glucose and the CNTs with high surface area in the hydrothermal reaction plays a crucial role in the formation of the hierarchical structure. In the hydrothermal reaction, glucose not only mediates the formation of nanoparticles with smaller size but also meanwhile serves as a binder to assist the assembly of nanoparticles and their subsequent growth on the graphene surface29. On the other hand, the graphene with high surface area can serve as a good substrate to avoid the aggregation of deposited material30,31. As evidenced by our control experiment, when 3D-G is absent in the hydrothermal synthesis, the diameters of the as-synthesized particles range from 0.5 to 4 μm (Figure S2). The drastic difference in particle size highlights the important role of 3D-G as an effective substrate for the growth of nanomaterials.

Figure 1
figure 1

Morphology characterizations of Ni3S2/3D-G.

(a)–(c) SEM images and (d) TEM image of Ni3S2/3D-G.

Full size image

To further characterize the identity and structure of the composite sample, X-ray powder diffraction (XRD) and Raman measurements were carried out. As shown Figure 2a, the sharp peak at 2θ = 26.3° indicates that the interlayer spacing inside the few-layered graphene sheet has a structure similar to that of normal graphite32. Except the (002) reflection of graphene, all the identified peaks can be attributed to the Ni3S2 phase (JCPDS card No. 30-0863). Figure 2b shows the Raman spectrum of the composite sample. In the low-frequency region, the three peaks at 308, 328 and 350 cm−1 can be assigned to the Raman characteristic bands of Ni3S233. Furthermore, it is noted that the spectrum profile in the high-frequency region for composite sample is totally different from that of pristine 3D-G (Figure S1c). The presence of amorphous carbon in the composite sample is evidenced by the new peak emerged at ca.1370 cm−1 (D band) and the broad G band at 1568 cm−1. The amorphous carbon is derived from the hydrothermal carbonization of glucose34,35. To analyze the surface composition of the composite sample, we performed X-ray photoelectron spectroscopy (XPS) analysis. Figure 2c and 2d show the XPS results for elemental Ni and C measured from the composite sample. The intense peak at around 856 eV can be deconvoluted into eight components using published fitting information36,37. It is noteworthy that the peak located at 852.6 eV can be assigned to the characteristic peak of Ni3S2. Furthermore, other seven peaks ranging from 854.5 to 861.3 eV suggest the presence of nickel oxyhydroxide (NiOOH) on the surface of Ni3S2. The formation of NiOOH could be due to the oxidation of Ni3S2 in air38. Figure 2d displays that the C 1s spectrum with a feature of a broad band centered at 285.7 eV can be deconvoluted into four peaks. The peaks at 284.6 and 285.7 eV are attributed to sp2- and sp3-hybridized carbon atoms, respectively whereas the peaks at 286.5 and 287.8 eV are carbon atoms bound to one oxygen atom by a single bond and a double bond, respectively39. The aforementioned four different forms of chemical bonding are originated from the amorphous carbon35,40. On the basis of the microscopic and spectroscopic data, we can conclude that the Ni3S2 nanoparticles successfully deposit on graphene substrate and their surface is partially oxidized to NiOOH and is covered by a thin layer of amorphous carbon.

Figure 2
figure 2

Composition characterizations of Ni3S2/3D-G.

(a) XRD pattern (b) Raman spectrum and (c)–(d) XPS spectra of Ni3S2/3D-G.

Full size image

Electrochemical performances of Ni3S2/3D-G Electrode

Our previous studies have proved that the consecutive cyclic voltammetry (CV) scanning in a KOH solution induces the morphological change of the composite electrode synthesized by glucose-assisted hydrothermal method, which further enhances the capacitive performance of the electrode28. This activation process involving 400 CV cycles at 50 mV/s in 1 M KOH is also performed in the current work. Figure 3a shows the CV curves of the electrodes of 3D-G, as-synthesized Ni3S2/3D-G and the activated Ni3S2/3D-G measured at a scan rate of 5 mV/s in 1 M KOH. The CV curve of the Ni3S2/3D-G electrode exhibits a pair of redox peaks in the potential range from 0 to 0.6 V and the redox peaks at around 0.45 and 0.23 V can be attributed to the reversible redox reactions of Ni(II) ↔ Ni(III)20,21. Furthermore, the redox peak current of the Ni3S2/3D-G electrode is much larger than that of 3D-G electrode. The area surrounded by the CV curve is dramatically enhanced by the combination of Ni3S2 nanoparticles with 3D-G. This result suggests that the large capacitance of the composite electrode is mainly derived from the pseudocapacitance of the electrochemically active Ni3S2 nanoparticles instead of double-layer capacitance from the 3D-G. It is noteworthy that the Ni3S2/3D-G electrode after the activation process shows a larger area under the CV curve compared with as-synthesized one. The Cm value of the composite electrode can be estimated by integrating the area under the reduction peak of a stable CV and then dividing by the scan rate (v), the mass of the active material (m) in the electrode and the potential window (Va − Vc) according to the following equation.

After the activation process, the Cm value of the composite electrode is increased from 1120 to 2362 F/g and the increasing percentage of the Cm value is ca. 111%. To further emphasize the effect of combining Ni3S2 nanoparticles with 3D-G on electrode performance, the comparisons between Ni3S2 electrode and Ni3S2/3D-G are made. Figure 3b and Figure S2c show the CV curves of the activated Ni3S2/3D-G and Ni3S2 electrodes at the different scan rates, respectively. It is noteworthy that the Cm values of the activated Ni3S2/3D-G electrode obtained at 5, 10, 20 and 50 mV/s are 2362, 2250, 2068 and 1859 F/g, respectively and are ca. 3.7 times higher than those of the Ni3S2 electrode. For example, at a scan rate of 50 mV/s, the Cm value of the activated Ni3S2/3D-G electrode is 1859 F/g whereas the pure Ni3S2 electrode shows a much lower capacitance of 494 F/g. The relatively large Cm values of the composite electrode could be attributed to the fact that the small sizes of Ni3S2 nanoparticles on 3D-G allow nearly full utilization of the electrode materials41,42. On the other hand, the inset in Figure 3b shows that the cathodic peak current increases linearly with the square root of the scan rate, indicating that the redox reaction at electrode/electrolyte interface belongs to diffusion-controlled process and therefore the charge-transfer reaction in the interface is relatively rapid. According to the slopes shown in the insets of Figure 3b and the Figure S2c, the diffusion coefficient of electrolyte ions for the activated Ni3S2/3D-G electrode is 6.5 times higher than that of the Ni3S2 electrode. This result suggests that the Ni3S2/3D-G electrode has the efficient and fast ion diffusion due to the open pore system and nano-sized morphology of the composite14,43. In Figure 3c, the galvanostatic charge/discharge curves of the activated Ni3S2/3D-G electrode obtained at different current densities exhibit a non-linear profile as expected for the pseudocapacitive electrode. The Cm value can be alternatively estimated from the discharge section according to the equation (2):

where I, Δt, ΔV and m represent discharge current, discharge time, potential window and the mass of Ni3S2 in the composite, respectively. The Cm values of the activated Ni3S2/3D-G electrode are calculated to be 4598, 4284, 3897, 3296 and 2261 F/g at 2, 4, 8, 16 and 32 A/g, respectively. It is noteworthy that the Cm value of the activated Ni3S2/3D-G electrode is much higher than the electrodes constructed by NiS nanoflakes (664 F/g at 4 A/g)20, NiS hollow spheres (927 F/g at 4.08 A/g)21, NiS nanoparticles (893 F/g at 5 A/g)22, Ni3S2/CNT composite (514 F/g at 4 A/g)44, NiS/graphene oxide (800 F/g at 1 A/g)45 and Ni3S2/rGO (1424 F/g at 0.75 A/g)46. To get more information about the electrochemical stability of the composite electrode, the consecutive charge–discharge cycles were performed at a high current density of 16 A/g. As shown in Figure 3d, the Cm retention of Ni3S2/3D-G is 74% after 1000 cycles, which shows the reasonable cycling stability. The decrease in the Cm value of Ni3S2/3D-G after a long-term test can be attributed the exfoliation of Ni3S2/3D-G from Ni-foam as evidenced by SEM observations (Figure S3). Furthermore, the composite electrode exhibits nearly 100% Coulombic efficiency over the entire cycling test.

Figure 3
figure 3

Electrochemical performances of Ni3S2/3D-G electrode.

(a) CV curves of the 3D-G and Ni3S2/3D-G electrodes measured at a scan rate of 5 mV/s. (b) CV curves of the activated Ni3S2/3D-G electrode recorded at the different scan rates (5, 10, 20 and 50 mV/s) in 1 M KOH. The inset shows that the variation of the current density of cathodic peak as a function of the square root of scan rate. (c) Charge-discharge curves of the activated Ni3S2/3D-G electrode at different current densities (2, 4, 8, 16 and 32 A/g) in 1 M KOH. (d) The variations in specific capacitance and Coulombic efficiency of the activated Ni3S2/3D-G electrode as a function of cycle number at the current density of 16 A/g.

Full size image

To understand the origin of the increase in capacitance of Ni3S2/3D-G electrode after the activation process, we analyzed the electrode morphology and its surface composition. As shown in Figure 4a, the SEM image of the activated Ni3S2/3D-G electrode reveals that the electrode surface become very rough. Furthermore, Figure 4b and 4c show the XPS results for elemental Ni and C measured from the activated composite electrode. It is noteworthy that the Ni3S2 peak at 852.6 eV disappears and only the characteristic peaks of NiOOH are observed. Furthermore, no signal of the elemental S can be detected in the surface of activated electrode as shown in Figure S4. These results suggest that the surface composition of electrode is converted from Ni3S2 into NiOOH and thus the activated composite electrode is referred to as NiOOH/Ni3S2/3D-G in the subsequent discussion. The formation of NiOOH during the activation process is consistent with the previous study of using Ni3S2 nanorods/Ni foam composite electrode as a catalyst for the oxygen evolution reaction in a alkaline solution36. Furthermore, the C1s spectrum profile of the NiOOH/Ni3S2/3D-G electrode is significantly different from that of the as-synthesized Ni3S2/3D-G. It is noteworthy that the peak attributed to sp3-hybridized carbon atoms (at 285.7 eV) nearly disappears as shown in Figure 4c. Moreover, the atomic ratio of Ni to C in the NiOOH/Ni3S2/3D-G electrode is 0.227:1 whereas that in the Ni3S2/3D-G electrode is 0.07:1. Due to the decrease in the peak intensity of sp3-hybridized carbon atoms and the increase in Ni content, it is suggested that the amorphous carbon is partially removed from the electrode surface via the consecutive CV scanning in a KOH solution. The change in electrode morphology and the capacitance enhancement after the activation process can be attributed to the combined effect of the removal of amorphous carbon and the formation of NiOOH. The partial removal of amorphous carbon from the electrode surface leads to the increase in the exposure of the active sites in the composite electrode. Furthermore, due to the fact that NiOOH and Ni3S2 co-exist in the electrode, these electroactive materials can both give the contributions to the electrode capacitance. As a result, the possible pseudocapacitive reaction mechanisms are described by the following equations20,21,47:

To further understand the mechanism behind the excellent capacitive performance of the NiOOH/Ni3S2/3D-G electrode, the electrochemical impedance spectroscopy (EIS) measurements were performed at a frequency range from 10 kHz to 0.01 Hz. As shown in Figure 4d, each Nyquist plot consists of a semicircle in the high frequency region and a straight line in the low frequency region. The intercept at real axis represents the combined resistance (Rs) including ionic resistance of electrolyte, intrinsic resistance of substrate and contact resistance at the interface of active material/current collector whereas the diameter of a semicircle corresponds to the charge transfer resistance (Rct) at the electrode/electrolyte interface25. The Rs and Rct values can be evaluated by fitting the EIS spectra based on the equivalent circuit proposed in the inset of Figure 4d. The Rct values of the Ni3S2/3D-G electrode and NiOOH/Ni3S2/3D-G electrode are 844 and 315 m Ω, respectively, being much smaller than that of the pure Ni3S2 electrode (947 mΩ). Therefore, the composite electrode shows faster reaction kinetics than the Ni3S2 electrode. Furthermore, the combination of Ni3S2 with highly conductive graphene leads to the decrease in Rs value. For example, the Rs value of the NiOOH/Ni3S2/3D-G electrode is 1.23 Ω whereas the pure Ni3S2 electrode shows a higher resistance of 1.78 Ω. The aforementioned EIS studies verify that the synergistic effect of the electrochemically active nanoparticles and the conductive 3D-G is responsible for enhancing the capacitive performance of the NiOOH/Ni3S2/3D-G electrode.

Figure 4
figure 4

Surface analysis of the activated Ni3S2/3D-G electrode.

(a) SEM image and (b)–(c) XPS spectra of the activated Ni3S2/3D-G electrode. (d) Nyquist plots of the pure Ni3S2, as-synthesized Ni3S2/3D-G and NiOOH/Ni3S2/3D-G electrodes. The inset shows the equivalent circuit used to simulate the Nyquist plots. In the equivalent circuit, CPE represents the constant phase element accounting for a double-layer capacitance, ZW is the Warburg impedance and CF is the Faradaic pseudocapacitor.

Full size image

Material and electrochemical characterizations of Fe3O4/rGO electrode

To evaluate the capacitive performance of the Ni3S2/3D-G electrode in a full-cell configuration, it is attempted to fabricate an ASC device. Because the Ni3S2/3D-G electrode exhibits the exceptionally high Cm value, the reasonable matching of cathode/anode becomes extremely important for obtaining an ASC with a high energy density. As a result, exploring the anode material with high capacitance becomes a huge challenge. In the current study, we synthesize the Fe3O4/rGO composite as the material for anode. Figure 5a and 5b show low- and high magnification TEM images of Fe3O4/rGO composite, respectively. It is evident that the conductive rGO can serve as the substrate where Fe3O4 nanoparticles with the averaged diameter of 5 nm densely deposit. No apparent aggregation of Fe3O4 nanoparticles on the rGO is observed; in contrast, Fe3O4 nanoparticles aggregate to large particles in the absence of graphene (Figure S5a). To further characterize the composition and structure of the composite sample, energy dispersive X-ray spectroscopy (EDS) and XRD measurements were carried out. As shown in Figure 5c, the EDS analysis of the composite sample reveals the presence of C, O, Si and Fe elements. Furthermore, the XRD pattern of the Fe3O4/rGO composite shown in Figure 5d corresponds to the magnetite Fe3O4 (JCPDS card No. 72-2302). On the basis of the aforementioned data, we can conclude that the Fe3O4/rGO composite is successfully synthesized.

Figure 5
figure 5

Material characterizations of Fe3O4/rGO composite.

(a)–(b) TEM images, (c) EDS spectrum and (d) XRD pattern of Fe3O4/rGO.

Full size image

To further emphasize the effect of combining Fe3O4 nanoparticles with rGO on electrode performance, the comparisons among pure Fe3O4, pure rGO and Fe3O4/rGO electrodes are made. Figure 6a shows the CV curves of the electrodes of Fe3O4, rGO and Fe3O4/rGO measured at a scan rate of 20 mV/s in 1 M KOH. The CV curve of the rGO electrode measured in the potential range from 0 to −1.2 V exhibits nearly rectangular-like shape, which clearly manifests the electrical double layer capacitive behavior. In contrast, the CV curves of Fe3O4/rGO and pure Fe3O4 electrodes exhibit a pair of redox peaks, which corresponds to the reversible reactions of Fe(II) ↔ Fe(III)48. It is noteworthy that the Fe3O4/rGO electrode shows a larger area under the CV curve than two pure electrodes and thus possesses the superior capacitive property. The Cm values of the Fe3O4/rGO electrode obtained at 5, 10, 20 and 50 mV/s are 447, 461, 412 and 303 F/g, respectively (Figure S6). A large specific capacitance of the composite electrode is mainly derived from the pseudocapacitance of the electrochemically active Fe3O4 nanoparticles instead of double-layer capacitance from the rGO. In Figure 6b, the galvanostatic charge/discharge curve of the rGO electrode measured at the current density of 1 A/g is symmetric and linear whereas the Fe3O4 and Fe3O4/rGO electrodes exhibit the non-linear curves. The difference in curve profile is attributed to the distinct energy storage mechanisms. The Cm values measured at 1 A/g for rGO, Fe3O4 and Fe3O4/rGO electrodes are 228, 325.7 and 661.3 F/g, respectively. The Fe3O4/rGO electrode exhibits the higher capacitance than pure rGO and Fe3O4 electrodes due to the fact that perfect combination of Fe3O4 and rGO not only prevents the agglomeration of nanoparticles but also enables fast electron transport through the underlying rGO layers to Fe3O4 nanoparticles48,49. Figure 6c and Figure S5c show the galvanostatic charge/discharge curves of the Fe3O4/rGO and Fe3O4 electrodes obtained at different current densities, respectively. The Cm values of the Fe3O4/rGO electrode are calculated to be 661.3, 536.8, 475.6 and 430.4 F/g at 1, 2, 4 and 8 A/g, respectively. Furthermore, the Fe3O4/rGO electrode shows the better rate capability than pure Fe3O4 electrode as shown in Figure 6d. It is noteworthy that the Fe3O4/rGO electrode in the current work has a higher Cm value compared with the previously reported Fe3O4/rGO electrodes50,51,52,53.

Figure 6
figure 6

Electrochemical performances of Fe3O4/rGO electrode.

(a) CV curves (measured at 20 mV/s) and (b) Charge-discharge curves (measured at 1 A/g) of the Fe3O4, rGO and Fe3O4/rGO electrodes. (c) Charge-discharge curves of the Fe3O4/rGO electrode at different current densities (1, 2, 4 and 8 A/g) in 1 M KOH. (d) Capacitance retentions of the Fe3O4 and Fe3O4/rGO electrodes as a function of discharge current densities.

Full size image

Electrochemical performances of the ASC based on Ni3S2/3D-G and Fe3O4/rGO composite electrodes

Since we successfully develop the highly capacitive cathode and anode, the ASC is fabricated by using the NiOOH/Ni3S2/3D-G as the cathode and Fe3O4/rGO as the anode (denoted as NiOOH/Ni3S2/3D-G//Fe3O4/rGO). To obtain a well electrochemical performance for ASC, the charge balance between the cathode and the anode should be satisfied. Therefore, on the basis of the Cm values and potential windows of two composite electrodes, the optimum loading mass ratio of Ni3S2 to Fe3O4/rGO is 0.26:1. As shown in Figure 7a, the CV curves of NiOOH/Ni3S2/3D-G//Fe3O4/rGO device measured at different scan rates exhibit the redox humps, which may be attributed to the contribution from the pseudocapacitive electrodes. It is noteworthy that the Cm values of the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device obtained at 5, 10, 20 and 50 mV/s are 233, 214, 192 and 167 F/g, respectively. The galvanostatic charge/discharge curves of the ASC were measured at the current densities from 1 to 16 A/g in a potential window of 0–1.6 V. As shown in Figure 7b, the perfectly linear charge/discharge curves of electrical double layer capacitors are not obtained in the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device, which indicates the pseudocapacitive nature of the supercapacitor. Furthermore, the Cm values of NiOOH/Ni3S2/3D-G//Fe3O4/rGO device are calculated to be 170, 158, 148, 136 and 118 F/g at 1, 2, 4, 8 and 16 A/g, respectively. Even at a high current density of 16 A/g, ca. 70% capacitance retention is observed. In Figure 7c, Ragone plot related to power densities (P) and energy densities (E) is further used to evaluate the performance of NiOOH/Ni3S2/3D-G//Fe3O4/rGO device. The P and E values of the ASC were calculated by the following equations:

where Cm is calculated based on the CV curve of the ASC device, ΔV is the operating voltage of the cell and Δt is the discharge time. The NiOOH/Ni3S2/3D-G//Fe3O4/rGO device can deliver a high energy density of 82.5 Wh/kg at a power density of 930 W/kg and still maintain 63.1 Wh/kg at a high power density of 6.9 kW/kg. It is noteworthy that the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device exhibits the superior energy density to the ASCs based on pseudocapacitive cathodes and anodes such as H-CoOx@Ni(OH)2//RGO@Fe3O4 (45.3 Wh/kg at 1010 W/kg)53, MnO2//FeOOH (24 Wh/kg at 450 W/kg)54, MnO2//Fe3O4 (8.1 Wh/kg at 10.2 kW/kg)55, Ni(OH)2//ZnFe2O4 (14 Wh/kg at 209 W/kg)56, LiMn2O4//MnFe2O4 (10 Wh/kg at 300 W/kg)57, LiMn2O4//MnO2/CNT (56 Wh/kg at 300 W/kg)58. To test the electrochemical stability of the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device, consecutive charge–discharge cycles were performed at a current density of 1 A/g. However, ca. 74% decrease in capacitance is observed after 2000 charge/discharge cycles (Figure 7d). The instability of NiOOH/Ni3S2/3D-G//Fe3O4/rGO device mainly originates from the decay of Fe3O4/rGO electrode. After 2000 charge/discharge cycles, the Cm value of Fe3O4/rGO electrode is decreased to 106 F/g as shown in Figure S7. As a result, the charge balance between the cathode and the anode is not maintained, which further impairs the capacitive performance of NiOOH/Ni3S2/3D-G//Fe3O4/rGO device. Very recently, Long et al. have reported that the high capacitance of graphene supported iron nanosheets is mainly attributed to electrochemical transformation from Fe to Fe3O4 under alkaline condition48. In their cycling test, the composite anode shows a significant decrease in capacitance. This instability could be mainly associated with the aggregation of Fe3O4 particles during electrochemical cycles, resulting in the loss of bonding between Fe3O4 and the graphene sheets and the subsequent loss of the electrochemical activity. Because our TEM observation on Fe3O4/rGO electrode after 2000 charge/discharge cycles is consistent with the scenario reported by Long et al. (Figure S7c), we use gel electrolyte to improve the electrochemical stability of Fe3O4/rGO electrode59. The encapsulation of Fe3O4/rGO electrode within gel electrolyte should be able to prevent the separation of Fe3O4 nanoparticles from rGO substrate. We thus examine the cycling performance of Fe3O4/rGO electrode in a gel electrolyte and compare it to the values obtained in an aqueous electrolyte. As shown in Figure S8c, the capacitance of Fe3O4/rGO electrode decreases continuously in a 1 M KOH solution and only 38% of the initial capacitance is retained after 1000 cycles. In contrast, the capacitance of Fe3O4/rGO electrode decreases slowly in the gel electrolyte and more than 70% of the initial capacitance is retained after 1000 cycles. Although Fe3O4/rGO electrode in a gel electrolyte shows a lower Cm value (287.4 F/g at 1 A/g, Figure S8b), its cycling stability is significantly improved. Searching the ideal composition of gel electrolyte where electrodes without sacrificing their electrochemical performances and the assembly of ASC device with gel electrolyte, requiring intense research efforts, are currently under investigation in our group.

Figure 7
figure 7

Electrochemical performances of an ASC.

(a) CV curves of the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device collected at the different scan rates (5, 10, 20 and 50 mV/s). (b) Galvanostatic charge/discharge curves of the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device recorded at different current densities (1, 2, 4, 8 and 16 A/g). (c) Ragone plot of the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device. (d) The capacitance retention of the NiOOH/Ni3S2/3D-G//Fe3O4/rGO device as a function of cycle number at a current density of 1 A/g.

Full size image

Discussion

We have successfully synthesized hierarchically structured Ni3S2/3D-G composites using glucose-assisted hydrothermal method. The Ni3S2 nanoparticles are densely and uniformly grown on the surface of 3D-G serving as the porous and conductive substrate. It is found that the use of glucose in hydrothermal reaction is essential to the formation of this hybrid structure. As evidenced by SEM observations, the morphology of Ni3S2/3D-G becomes very rough after the composite electrode is activated by the consecutive CV scanning in a 1 M KOH solution. On the basis of XPS data, we have confirmed that the change in electrode morphology and the capacitance enhancement after the activation process can be attributed to the combined effect of the partial removal of amorphous carbon and the formation of NiOOH. Therefore, the partial removal of amorphous carbon from the electrode surface leads to the increase in the exposure of the active sites in the composite electrode, which further enhances the capacitive performance of composite electrode. Furthermore, the systematic CV measurements have verified that the NiOOH/Ni3S2/3D-G electrode shows the higher capacitive performance and better ion diffusion compared with the pure Ni3S2 electrode. The enhancement of capacitive performance of NiOOH/Ni3S2/3D-G can be attributed to the facts that the highly conductive graphene ensures rapid charge transfer and the small sizes of Ni3S2 nanoparticles on 3D-G allow nearly full utilization of the electrode materials. On the other hand, the improved ion diffusion can be ascribed to that the open pore system of the composite electrode makes electrolyte fully access to Ni3S2. Furthermore, EIS measurements reveal that the NiOOH/Ni3S2/3D-G has the lower Rs and Rct values than the pure Ni3S2 electrode, which further suggests that 3D porous and conductive graphene network is beneficial to electrochemical performance of supercapacitor electrode. Due to the synergistic effects of the electrochemically active nanoparticles and conductive 3D-G, the NiOOH/Ni3S2/3D-G electrode exhibits the high capacitance of 3296 F/g and great cycling stability at a current density of 16 A/g.

To obtain the reasonable matching of cathode/anode in an ASC device, we have synthesized highly capacitive Fe3O4/rGO composite using a hydrothermal method. Fe3O4 nanoparticles with the averaged diameter of 5 nm are densely and homogeneously dispersed onto the rGO surface. The Fe3O4/rGO electrode exhibits the high capacitance of 661 F/g at 1 A/g and great rate capability. More importantly, an ASC fabricated by using the NiOOH/Ni3S2/3D-G as the cathode and Fe3O4/rGO as the anode exhibits an ultrahigh energy density. The NiOOH/Ni3S2/3D-G//Fe3O4/rGO device can be reversibly charged/discharged at a maximum cell voltage of 1.6 V and a maximum capacitance of 233 F/g can be achieved at a scan rate of 5 mV/s, which delivers a maximum energy density of 82.5 Wh/kg at a power density of 930 W/kg. Such excellent capacitive performance suggests that NiOOH/Ni3S2/3D-G//Fe3O4/rGO device shows a great potential application as an inexpensive energy storage system.

Methods

Synthesis of 3D-G

The 3D-G was synthesized by CVD using nickel foam (1 cm2) as an interconnected 3D scaffold. The nickel foams were heated to 1000°C in a horizontal tube furnace under 10% H2 (400 s.c.c.m.) and annealed for 30 min to remove a thin surface oxide layer. Then, 10% H2 (100 s.c.c.m.) and CH4 (100 s.c.c.m.) were introduced into the reaction tube. The growth of 3D-G was performed for 30 min under the pressure of 10 torr. Finally, the sample was cooled to room temperature under 10% H2 (100 s.c.c.m.). Typically, the weight of 3D-G on Ni foam ranged from 0.6 to 0.8 mg.

Synthesis of Ni3S2/3D-G and Fe3O4/rGO

For the synthesis of Ni3S2/3D-G, glucose (90 mg) was added to the mixture solution of ethanol (9 ml) and distilled water (1 ml). The mixed solution was sonciated for 20 min to make a homogeneous solution and then nickel chloride (48.6 mg), thiourea (19 mg) and 1 ml of ammonia were added to the solution. The solution was transferred to a Teflon-lined stainless steel autoclave (20 ml capacity) and then a piece of 3D-G on nickel foam (1 cm2) was immersed into the autoclave. After sealing, the autoclave was heated to 180°C for 12 h and then cooled to room temperature. Finally, the resulting Ni3S2/3D-G was washed with distilled water and dried in air. In a separate experiment, the Ni3S2 powder was synthesized also using glucose-assisted hydrothermal method. Except the addition of 3D-G, the synthetic procedures for Ni3S2 powder were the same as those for Ni3S2/3D-G.

For the preparation of the material for anode, Fe3O4/rGO composite was synthesized based on the previous literature60. In brief, FeCl3.6H2O (5 mg) and FeCl2.4H2O (2 mg) was added to the GO dispersion of 10 mL (a concentration of 0.25 mg/mL) and the mixed solution was stirred overnight at 70°C to make a homogeneous solution. Then 1.5 mL NH3 solution was added into the solution and the mixed solution was transferred to a Teflon-lined stainless steel autoclave (20 ml capacity). After sealing, the autoclave was heated to 150°C for 2 h and then cooled to room temperature. Finally, the resulting product was washed with distilled water and dried in air. In the separate experiment, the Fe3O4 powder was synthesized by following the synthetic procedures of Fe3O4/rGO except the addition of graphene oxide. Furthermore, rGO powder is synthesized by a simple hydrothermal method. In brief, graphene oxide (20 mg) synthesized by a modified Hummer's method was added to the mixture solution of ammonia (1.5 ml) and distilled water (10 ml)8. The mixed solution was transferred to a Teflon-lined stainless steel autoclave (20 ml capacity) and then the autoclave was heated to 150°C for 2 h. Finally, the resulting product was washed with distilled water and dried in air.

Material characterizations

The surface morphology of Ni3S2/3D-G was characterized using a field emission scanning electron microscope (JSM-7000F). Field-emission transmission electron microscope (JEOL JEM-2100F, operated at 200 kV with a point-to-point resolution of 0.19 nm) equipped with an energy dispersive spectrometer was used to obtain the information on the microstructures and the chemical compositions. XRD patterns of the composites were obtained from Philips X'Pert Pro MPD. The Raman measurements were performed using a confocal Raman spectroscopic system (Horiba Jobin Yvon) with 532 nm excitation laser and a 50× objective. XPS measurements were carried out with a PHI VersaProbe XPS Microprobe with monochromatic Al Kα X-ray radiation.

Electrochemical measurements

All electrochemical measurements were performed in a standard three-electrode cell. The composite electrode served as the working electrode, while a Pt wire and an Ag/AgCl (sat.KCl) electrode were used as the counter and a reference electrode, respectively. CV and galvanostatic charge–discharge curves were measured in a 1 M KOH solution using a CHI 635A electrochemical workstation. EIS measurements were conducted with an Eco Chemie Autolab PGSTAT-128N potentiostat equipped with the FRA2 frequency response analyzer module and GPES/FRA software. An AC voltage with 5 mV amplitude in a frequency range of 0.01 Hz–10 kHz was applied. For the electrode preparations of Ni3S2, rGO, Fe3O4 or Fe3O4/rGO, the individual powder was mixed with carbon black and polyvinylidene difluoride at a weight ratio of 8:1:1. After thorough mixing, the slurry was pressed onto Ni foam (1 cm2) and then dried at 60°C in vacuum for 1 day. The as-fabricated electrode was characterized in a 1 M KOH solution using CV and EIS.

Preparation of KOH/PVA gel electrolyte

The powder of polyvinyl alcohol (PVA, 1 g) was dissolved in 10 mL of 1 M KOH solution at 80°C with stirring. After the solution was cooled to room temperature, Fe3O4/rGO electrode, a Pt wire and an Ag/AgCl (sat.KCl) electrode were immersed in the solution. Finally, the solution became a clear and transparent gel after it was dried at 40°C in air.

Fabrication of ASC devices

To fabricate the ASCs, Fe3O4/rGO and Ni3S2/3D-G were used as the anode and the cathode, respectively. Two electrodes where the total mass loading of active materials was 1.3 mg were separated by a filter paper. The cell was encapsulated by flexible plastic film with two pieces of platinum wires connected to the edges of the two electrodes. Finally, a 1 M KOH solution was injected into the cell to rinse the filter paper. The electrochemical properties of the ASC devices were measured using a CHI 635A electrochemical workstation.