III. RESULTS AND DISCUSSION

The synthetic procedures of porous 3DNGFs are illustrated in Fig. 1. Firstly, commercial GP was electrochemically oxidized in a mixed HNO₃–H₂SO₄ solution at a potential of 15 V for 20 min to obtain 3D graphene oxide frameworks (3DGOFs). Then the as-prepared 3DGOFs were soaking in melamine solution and dried in an oven. Finally, the as-obtained 3DGOFs were annealed in N₂ at 900 °C (See Sec. II) to simultaneously achieve N-doping and reduction (recorded as 3DNGFs-900). For comparison, as-prepared 3DGOFs were directly annealed at 900 °C with N₂ as the protective gas to obtain the undoped 3DGFs.

FIG. 1.

Schematic illustration of the synthetic procedures of 3DNGFs.

The digital image [inset of Figs. 2(a) and 2(b)] displays that mechanically robust, interconnected porous monolithic 3D N-doped graphene frameworks (3DNGF) were obtained, and its volume was 6–7 times larger than that of the pristine GP. The morphology and structure of the as-prepared 3DNGFs-900 was first surveyed by scanning electron microscopy [Figs. 2(a) and 2(b)]. Interestingly, in contrast to the compact and smooth surface of GP, the 3DNGFs-900 shows twisted and transparent nanosheets crimped randomly to form a porous network structure, which results in a high surface area. Additionally, the SEM and digital images of the as-prepared 3DGFs sample without nitrogen doping were also presented in Fig. S1. It is showed in the digital images that there are no obvious differences between the 3DGFs and 3DNGFs-900 samples. Similarly, on the surface of 3DGFs, typical wrinkled graphene composed of layered platelets can be easily observed. These results indicate that the morphology of the 3DGFs has not been changed with the doping of nitrogen. To further investigate the quality and detailed microstructure of the 3DNGFs-900, TEM was carried out. Figure S2(a) indicates that the GP sample was smooth and thick that included many graphite flakes. Similarly, HRTEM [Fig. 2(c)] images reveal that it consists of a dozen of stacked graphite plates, which is also reflected in its overlapping SAED diffraction points [Fig. S2(b)]. In contrast, relatively rough and sheet-like structures on the edge 3DNGFs-900 sample can be observed [Fig. S2(c)]. The HTREM [Fig. 2(d)] image results confirm that the sheets are composed of a few layers of graphene, indicating that the bulk graphite has been well exfoliated into graphene sheets. Additionally, the bright spots [Fig. S2(d)] in a hexagonal pattern can be identified in its SAED patterns suggesting the high quality of the obtained graphene products.

FIG. 2.

SEM images of (a) GP, (b) 3DNGFs-900, and the insets are the corresponding digital images. High-resolution TEM (HRTEM) images collected at the edge of (c) GP, and (d) 3DNGFs-900.

The surface area and the porosity of the material are always related to high specific capacitance. Therefore, the nitrogen adsorption–desorption measurement was performed to evaluate the variation of the surface area and the porous structure of the 3DNGFs-900 sample. As shown in Fig. 3(a), both the nitrogen adsorption–desorption isotherms of GP and 3DNGFs-900 show a typical type IV hysteresis loop, which is characteristic of mesoporous materials.^{Reference Li, Liang, Li and He38,Reference Wen, Zhang, Yan, Zhang and Shi39} The hysteresis loop appeared at the lower relative pressure of 0.4–0.8 which indicates the presence of mesopores, while at a higher relative pressure of 0.8–1, it is attributed to macropores.^{Reference Wen, Zhang, Yan, Zhang and Shi39} As observed in Fig. 3(a), there are noticeable differences in the shape of the nitrogen sorption isotherms between the GP and 3DNGFs-900 samples. For the 3DNGFs-900 sample, the isotherms become increasingly steeper than that of the GP in the range of the relative pressure beyond 0.45 and up to 1. These phenomena indicate that numerous mesopores and macropores were created by the electrochemical oxidation process. The calculated BET surface area of 3DNGFs-900 achieved 181.98 m²/g, which is 3 times larger than that of the pristine GP (59.75 m²/g), demonstrating that the electrochemical exfoliation could remarkably increase the specific surface area. Significantly, as shown in the pore volume and pore size distribution diagram [Inset in Fig. 3(a)], a drastic increase of micropores (<2 nm) and mesopores (2–50 nm) can easily be recognized for the 3DGNGFs-900 sample, with the total volume reached 0.102 m⁻³/g, which is 4.8 times larger than that of GP (0.021 m⁻³/g). The presence of the micropores (<2 nm) could be helpful for the improvement of double-layer capacitance, while mesopores (2–50 nm) would be benefit for ion transport and electrolyte diffusion. Besides, we also compared the adsorption–desorption isotherms and pore volume/pore width distribution plots (Fig. S3) of the 3DGFs and 3DNGFs-900 samples. As shown in Fig. S3, there is little difference between the adsorption–desorption isotherms of 3DGFs and 3DNGFs-900 samples, in which they both showed a typical type IV hysteresis loop, indicating the presence of micropores and mesopores created in the electrochemical oxidization process. Meanwhile, the pore volume/pore width distribution plots show slight differences between the results of the two samples, with an increase in mesopore (pore width at about 50–60 nm) volume for the nitrogen-doped sample. Furthermore, the calculated BET surface area for 3DGFs was 173.60 m²/g, which is a little lower than that of the nitrogen-doped sample. These may ascribe to the reason that the doping of nitrogen can prevent the stacking of graphene layers by the incorporation of nitrogen functional groups and thus result in higher specific surface area and pore volume, which can ensure more sites for charge storage during the charge/discharge process.^{Reference Sun, Wang, Tian, Tan, Xie, Shi, Li and Fu40,Reference Xu, Pan, Liu, Lu and Sun41}

FIG. 3.

(a) Nitrogen adsorption–desorption isotherms of GP and 3DNGFs-900 at 77 K. The inset is the pore-volume distribution of GP and 3DNGFs-900. (b) Raman spectra of GP, 3DGFs, and 3DNGFs-900 samples. (c) C 1s core-level XPS spectra of upper: 3DNGFs-900 and bottom: GP. (d) N 1s core-level XPS spectra of the 3DNGFs-900 sample.

The crystal structure of GP, 3DGOFs, and 3DGNFs-900 were analyzed by XRD studies, and the results were presented in Fig. S4. For the GP sample, a sharp diffraction peak can be observed at approximately 26°, which is attributed to the (002) crystal plane of graphite (JCPDS No. 41-1487). Compared with pristine GP, 3DGOFs exhibit an additional peak at about 11° attributing to the (002) crystalline plane of graphene oxide, indicating the successful oxidation of the starting graphite (Fig. S4).^{Reference Chen, Zeng, Zhao, Yu, Cheng, Lu and Tong42} Meanwhile, the strong diffraction peak at 26° still existed, which means that the graphite is not fully oxidized in the 3DGOFs sample.^{Reference Chen, Zeng, Zhao, Yu, Cheng, Lu and Tong42} Differently, the exfoliated 3DGOFs and 3DNGFs-900 display a broader diffraction peak at 26° than GP, which implies the presence of the stacked graphene layers.^{Reference Chen and Xue43} Consequently, XRD results again proved that the 3DNGFs have been successful prepared by the electrochemical oxidation process.

Raman spectroscopy was used to identify the structure and doping levels of GP, 3DGFs, and 3DNGFs-900 samples. As presented in Fig. 3(b), the peak located at around 1597 cm⁻¹ can be easily identified for all the three samples, which is attributed to the G band for the E_2g vibration mode of sp² carbon domains.^{Reference Wang, Wang, Wu, Xia, Zhao and Jia29} In comparison to the pristine GP, an additional peak located at 1346 cm⁻¹ (D band) can be observed for both 3DGFs and 3DNGFs-900 samples, due to the breathing mode of k-point phonons of A_1g symmetry. The appearance of D band is attributed to local defects and/or disorders.^{Reference Zhang, Zhao, Ji, Stoller, Lai, Murali, Mcdonnell, Cleveger, Wallace and Ruoff31,Reference Zhu, Murali, Cai, Li, Suk, Potts and Ruoff44} The intensity ratio of peak D and G (I _D/I _G) increases from 0.546 for 3DGFs to 0.690 for 3DGNFs-900, suggesting that more defects are formed in the 3DNGFs-900 sample. These defects are attributed to the doping of the nitrogen atom.^{Reference Wang, Wang, Wu, Xia, Zhao and Jia29} Besides, there is a slightly positive shift in the position of the G band for 3DNGFs-900 when compared with 3DGFs, which can be ascribed to the formation of doped graphene.^{Reference Zhang, Zhao, Ji, Stoller, Lai, Murali, Mcdonnell, Cleveger, Wallace and Ruoff31} It is believed that the defects arising from doping in the graphene layer may create additional active sites for charge storage.

X-ray photoelectron spectroscopy (XPS) survey was further used to measure the nitrogen content and examine chemical changes caused by nitrogen doping. Figure 3(c) compares the high resolution C 1s spectra for GP and 3DNGFs-900. It is obvious that the 3DNGFs-900 sample shows a broader peak than that of GP during the same binding energies. The broad peak can be deconvoluted into three peeks, within which the two peaks at the binding energies of 284.8 and 286.1 eV are attributed to sp² hybridized C–C and residual C–OH, respectively, which are the same with that of GP.^{Reference Xiao, Ding, Chen, Wang, Lu and Wang45} Significantly, when compared with GP, an additional peak at 285.2 eV arises for the 3DNGFs-900 sample due to the doping of nitrogen atoms,^{Reference Xiao, Ding, Chen, Wang, Lu and Wang45} demonstrating the successful doping of nitrogen species. The introduction of nitrogen species in the 3DNGFs-900 sample was further probed by high-resolution N 1s spectra [Fig. 3(d)]. As shown in Fig. 3(d), the N 1s core-level XPS spectra can be fitted with three synthetic peaks, revealing the presence of three types of nitrogen functionalities. The first peak is located at a lower binding energy of 398.2 eV correspond to pyridinic (N-6), which is defined as the nitrogen atom in a six-membered ring and contributes one p-electron to the graphene layers. The peak at 399.3 eV can be assigned to pyrrolic/pyridone (N-5) nitrogen, with two p-electrons contributing to the graphene layers. The last peak at a binding energy of 401.4 eV is attributed to quaternary (N-Q) nitrogen, with a relatively more positively charged nitrogen incorporated in a graphene layer when compared to N-6.^{Reference Wang, Shi, Fan, Ge, Yin and Hu22,Reference Xu, Pan, Liu, Lu and Sun41,Reference Zhu, Murali, Cai, Li, Suk, Potts and Ruoff44,Reference Gorgulho, Gonçalves, Pereira and Figueiredo46} The detailed relative concentrations of nitrogen species obtained by fitting N 1s core-level XPS spectra of the 3DNGFs-900 sample are shown in Table I. It is suggested that quaternary (N-Q) nitrogen is the dominant component (59.09%) of nitrogen for the 3DNGFs-900 sample. Previous studies have proposed that N-Q species could lead to an improvement in the electronic conductivity of graphene and consequently enhanceme the capacitance activity. Differently, weak N 1s signals were observed in high-resolution N 1s spectra for pristine GP and the 3DGFs samples, indicating that the 3DNGFs-900 sample was indeed N-doped graphene (Fig. S5).

TABLE I.

Relative concentrations of nitrogen species obtained by fitting N 1s core-level XPS spectra.

To test its capacitive activities, the 3DNGFs-900 electrode was measured in a 1 M H₂SO₄ aqueous electrolyte (See Sec. II) in a conventional three-electrode cell with a saturated calomel electrode (SCE) as the reference electrode and a Pt wire as the counter electrode. For comparison, 3DGFs without N-doping and the pristine GP electrode were also measured in the same condition. The cyclic voltammetry (CV) curves of the GP, 3DGFs, and 3DNGFs-900 electrodes at different scan rates with a potential window of 0–1 V were shown in Figs. S6 and 4(a), respectively. All the three electrodes exhibit rectangular shape curves, which are characteristic of the double-layer capacitance. Besides, a pair of redox peeks with good symmetry can be clearly observed, indicating that they have pseudocapacitive characteristics. The redox peaks correspond to the oxidation and reduction of residual oxygen-containing functional groups.^{Reference Yang, Peng, Shu, Lv, He and Long47–Reference Chen, Zhang, Zhang, Yu and Ma49} This phenomenon is consistent with the XPS analysis results. A representative comparison of the CV curves for the GP, 3DGFs, and 3DNGFs-900 electrodes collected at the scan rate of 20 mV/s are shown in Fig. 4(b). Remarkably, the redox peak current densities of the 3DNGFs-900 sample are substantially higher than those of the 3DGFs and GP, indicating that the presence of nitrogen heteroatoms can considerably contribute to an additional pseudocapacitance. The CV plot area for the 3DGFs sample is larger than that of GP, indicating that the 3D architecture and the nature of graphene layers are benefits for the improvement of the capacitance performance. Moreover, it increases further after the doping of the heteroatom nitrogen. The CV curve of the 3DNGFs-900 sample shows a much larger area than that of the 3DGFs [Fig. 4(b)] electrode, which could probable due to the doping of nitrogen improve the wettability on the 3DNGFs surface thus enhancing the charge-storage capability of the electrode. Furthermore, for the 3DNGFs-900 electrode, when the scan rate increases from 5 to 100 mV/s, the CV curve still retains in a rectangular shape, demonstrating its excellent rate capability. The specific capacitance [Fig. 4(c)] calculated for GP, 3DGFs, and 3DNGFs-900 electrodes based on the CV curves as a function of the scan rate showed that the 3DNGFs-900 electrode reached a maximum specific capacitance of 1402.1 mF/cm² at a scan rate of 5 mV/s, which is substantially higher than that of GP (359.5 mF/cm²), and 3DGFs (672.8 mF/cm²), confirming that the as-mentioned method can effectively produce 3DNGFs with high capacities properties. Remarkably, the capacitance of the 3DNGFs electrode is substantially higher than or comparable to some recently reported bulk 3D graphene-based materials, such as, graphene-gel/nickel foam^{Reference Zhai, Wang, Yu, Xie, Liang, Li, Xiao, Tang, Wu, Lu and Tong50} (0.070 F/cm², 10 mV/s), MnO₂/graphene-gel/nickel foam^{Reference Zhai, Wang, Yu, Xie, Liang, Li, Xiao, Tang, Wu, Lu and Tong50} (13.6 mg/cm², 3.18 F/cm², 1 mV/s), 3D graphene/flexible graphite-paper^{Reference Ramadoss, Yoon, Kwak, Kim, Ryu and Jang51} (15.6 mF/cm², 5 mV/s), and 3D-rGO/carbonized cotton fabric (87.53 mF/cm² at 2 mV/s).^{Reference Zhou, Ye, Wan and Jia52}

FIG. 4.

CV curves of the (a) 3DNGFs-900 electrode at scan rates of 5–100 mV/s. (b) CV curves of the for GP, 3DGFs and 3DNGFs-900 electrodes at scan rates of 20 mV/s. (c) Specific capacitance calculated for GP, 3DGFs, and 3DNGFs-900 electrodes based on the CV curves as a function of the scan rate.

The electrochemical performance of the 3DNGFs-900 electrode was further evaluated by GCD measurements. Figure S7 displays the GCD curves of different electrodes at various current densities, from 4 to 12 mA/cm². Figure 5(a) shows a representative comparison of the GCD curves for the GP, 3DGFs, and 3DNGFs-900 electrodes collected at the current density of 4 mA/cm². When compared with that of GP electrodes, a longer discharge time was observed for the 3DGFs electrode, indicating an improvement of capacitance of the 3DGFs electrode due to the superior properties of graphene and the unique structure of the well-connected microporous 3D structure. The results are consistent with that of the CV investigation. Additionally, compared with the undoped 3DGFs, a much longer discharge time can be observed for the 3DNGFs-900 electrode, indicating a substantial enhancement of capacitance of the doped electrode. Moreover, the relatively symmetric GCD profiles for 3DNGFs-900 show their reversible pseudocapacitive behavior as well as rapid charge/discharge properties. Corresponding specific capacitance for different electrodes calculated as a function of current densities are showed in Fig. 5(b). The areal capacitance values of 3DGFs-900 at current densities of 12, 10, 8, 6, and 4 mA/cm² are 1322.2, 1473.8, 1679.5, 2005.9, and 2250.3 mF/cm², respectively, which are substantially higher than that of the GP and 3DGFs. Significantly, the 3DNGFs-900 electrode exhibited a highest specific capacitance of 2250.4 mF/cm² at a current density of 4 mA/cm², which is 6.3 times higher than that of GP (354.3 mF/cm²) and 3 times higher than the values of the 3DGFs (736.9 mF/cm²). These results convincingly supported the fact that the as-mentioned method can effectively produce 3DNGFs with high capacitance performances. Additionally, as observed in Fig. 5(a), the 3DNGFs-900 electrode displays a minimum iR drop of 0.007 V, revealing its superior electric conductivity. Also, the superior conductivity of the 3DNGFs-900 electrode was verified by the EIS measurements. As shown in Fig. 5(c), the Nyquist plots of all three samples show a small semicircle in the high frequency and a typical straight line in the low frequency region, which indicates that the charge-transfer is a fast process and the rate of the electrochemical process is governed by mass transfer. Notably, the 3DGFs and 3DNGFs-900 electrodes exhibit a more upward line than GP in the low frequency region, demonstrating the high capacitive behavior of the 3D architecture. A corresponding equivalent circuit [Upper inset of Fig. 5(c)] was also used to fit the Nyquist plots, in which R _s is the ohmic resistance, which consists of the electrolyte resistance and the resistance of the electrical connections to the electrode. The semicircle is fitted with a parallel equivalent circuit made up of R _ct and the constant phase element CPE₁, in which R _ct is the charge-transfer resistance. The fitting results are shown in Table II. Obviously, the R _s and R _ct of 3DGFs was 1.35 and 4.89 Ω, which is much smaller than that of GP (2.63 and 10.22 Ω), respectively, indicating that the electrode conductivity can be substantially improved by the electrochemical exfoliation and calcination processes. Significantly, for the 3DNGFs-900 sample with nitrogen doping, the R _ct further decrease to 0.42 Ω, which is more than one order of magnitude lower than that of the pristine GP sample, suggesting that the sample’s electrical conductivity was markedly enhanced after nitrogen doping. Furthermore, the straight lines in the low-frequency region are fitted by the equivalence element CPE₂ to evaluate the capacitive behavior of different electrodes. The impedance of a CPE is given by the following equation:

(3)$${Z_{{\rm{CPE}}}} = {1 \over {{Y_0}{{\left( {j\omega } \right)}^n}}}\quad ,$$

where ω is the frequency, j is the imaginary number, and Y ₀ and n (0 ≤ n ≤1) are the frequency independent parameters. The meaning of Y ₀ is depending on the value of n. When n = 1, Y ₀ represents a pure capacitance; when n = 0, Y ₀ is a pure resistor, and when n = 0.5, Y ₀ is a Warburg impedance.^{Reference Chen, Zeng, Zhao, Yu, Cheng, Lu and Tong42} As shown in Table II, the 3DNGFs-900 sample shows the largest Y ₀ and n, again demonstrating that the 3DNGFs-900 exhibits the best capacitive behavior and the lowest diffusion resistance.

FIG. 5.

(a) GCD curves of GP, 3DGFs and 3DNGFs-900 electrodes collected at the current density of 4 mA/cm². (b) Specific capacitance of the GP, 3DGFs and 3DNGFs-900 electrodes calculated from GCD curves as a function of current density. (c) Nyquist plots of the GP, 3DGFs, and 3DNGFs-900 electrodes (upper inset: the corresponding equivalent circuit diagram). (d) Cycling performance of the 3DNGFs-900 electrode collected at 100 mV/s for 10,000 cycles. The inset shows the first, 5000th, and 10,000th CV curves of the 3DNGFs-900 electrode.

TABLE II.

Values of the parameters obtained from the Nyquist plots of different electrodes.

We also studied the long-term cycling stability of the 3DNGFs-900 electrode by a CV test at a scan rate of 100 mV/s. As shown in Fig. 5(d), a slow increase of specific capacitance can be observed in the first initial 2500 cycles, which is probably due to the activation of the electrode material during the charge and discharge processes. After that, there is less capacity decay over 10,000 cycles (97.5% capacitance retention at the 10 000^th cycle), indicating the remarkable electrochemical stability of the 3DNGFs-900 electrode. These results above convincingly supported that the as-prepared 3DNGFs possesse excellent performances when used as in energy storage applications. The superior electrochemical performance of the 3DNGFs-900 electrode is due to the following reasons: (i) the abundant well-connected micropores and mesopores created in the oxidation process provide a large surface area and guarantee fast ion transport and electrolyte diffusion, resulting in an improvement of double-layer capacitance. (ii) The highly conductive nature of graphene enhances the electrical conductivity of the material. (iii) The doping of nitrogen can enhance the wettability and conductivity of graphene, helping to reduce R _s and R _ct of the electrode and thus enables rapid charge transport via the network. (iv) Nitrogen doping can produce defects in the graphene layer and create additional active sites for pseudocapacitance.

For better understand the relationship between the electrochemical properties of 3DNGFs and the doping levels of nitrogen, we prepared 3DNGFs at different annealing temperatures, respectively, (700 °C, 800 °C, 900 °C, and 1000 °C, denoted as 3DNGFs-700, 3DNGFs-800, 3DNGFs-900, and 3DNGFs-1000, respectively). As shown in Raman spectra [Fig. 6(a)], the 3DNGFs-900 sample displayed the highest ratio of I _D:I _G, indicating it processes most defects/edges than other samples, meaning much more additional active sites for charge storage. Simultaneously, XPS results showed that the nitrogen atom content in 3DNGFs decreased with the increasing annealing temperatures (Fig. S8 and Table I). The high-resolution N 1s spectrum of 3DNGFs-700, 3DNGFs-800, 3DNGFs-900, and 3DNGFs-1000 are shown in Fig. S8. The N 1s core-level XPS spectrum of 3DNGFs obtained at a relatively low annealing temperature of 700 and 800 °C can be fitted with four synthetic peaks, with the binding energies at 398.2, 399.3, 401.4, and 404.8 eV corresponding to pyridinic (N-6), pyrrolic/pyridone (N-5), quaternary (N-Q), and oxidized pyridinic nitrogen (N-X), respectively. Table II gives the concentrations of nitrogen species for different samples. It can be seen that the contents of the N-X species decrease as the annealing temperature decreases and disappear when the annealing temperature reached 900 °C. Meanwhile, the contents of quaternary nitrogen (N-Q) increase with the annealing temperature increase from 700 to 900 °C and then decrease over 900 °C. Interestingly, in all samples, N-Q accounted for the largest proportion, especially for the 3DNGFs-900 sample. It is reported that among various nitrogen species, the quaternary nitrogen (N-Q) species in the graphene layers could enhance the electronic conductivity of the carbon materials, which is important for the capacitance.^{Reference Wang, Wei, Nie and Zhang24,Reference Xiao, Ding, Chen, Wang, Lu and Wang45}

FIG. 6.

(a) Raman spectra of different samples. (b) Specific capacitance calculated for different electrodes based on the CV curves as a function of scan rate. (c) Specific capacitance of the different electrodes calculated from GCD curves as a function of current density. (d) Nyquist plots of the different electrodes.

Furthermore, CV and GCD results [Figs. 6(b), 6(c), S9, and S10] also showed that the specific capacitance of the electrode increased with the increase of annealing temperature from 700 to 900 °C and then decreased over 900 °C. In addition, the EIS results [Fig. 6(d) and Table I], confirmed that the 3DNGFs-900 electrode possessed the lowest charge-transfer resistance and the best capacitive behavior than other 3DNGFs electrodes. Therefore, the 3DNGFs electrode obtained at a temperature of 900 °C has the best performance. The probable reasons may be that: (i) the highest value of I _D:I _G in Raman spectra indicating 3DNGFs-900 sample processes most defect/edge, which enables it exhibiting the best capacitance performance; (ii) the 3DNGFs-900 sample processes the highest proportion of N-Q among all the nitrogen species, indicating that most of the incorporated nitrogen can play an effective role in the electrochemical performance of the material; (iii) the introduction of proper content of nitrogen doping in the 3DNGFs-900 sample can enhance the electrical conductivity and thus enable it to deliver the lowest R _s and R _ct and guarantee the best capacitance behaviors.^{Reference Xiao, Ding, Chen, Wang, Lu and Wang45}